[PATCH 0/5] Builtin FSMonitor Part 2

[PATCH 0/5] Builtin FSMonitor Part 2

[Jeff Hostetler via GitGitGadget]

Here is Part 2 of my Builtin FSMonitor series. Part 2 requires
jh/builtin-fsmonitor-part1 be present.

Part 2 contains:

 * Client-side code to connect to a (yet to be developed) FSMonitor daemon
   over simple IPC (using Named Pipes or Unix Domain Sockets).
 * Updated config settings to allow the client to decide whether to use
   Simple IPC or the existing FSMonitor Hook API.

The original FSMonitor feature "e05336bdda (Merge branch 'bp/fsmonitor',
2017-11-21)" consisted of two parts:

 1. Client code: 1.1. Called by commands like git status 1.2. Uses the Hook
    API to spawn and talk to a shell or perl script (usually) 1.3. Receives
    a list of recently modified files and/or directories 1.4. Updates flags
    and/or loops in various algorithms to avoid unnecessary lstat() calls on
    unchanged ones.

 2. Hook code: 2.1. A hook process, spawned by each client command, talks to
    a long-running FS monitoring process. 2.2. A long-running FS monitoring
    process, usually Watchman. 2.3. The hook process acts as a relay to
    translate requests responses from Watchman back to the client command.

The Builtin FSMonitor feature extends step 1.2 and allows clients to
communicate directly with a FS daemon process using Simple IPC. The daemon
response is then processed as before in step 1.3.

Part 2 contains these client-side changes for IPC. A builtin FSMonitor
daemon will be added in a later patch series.

Jeff Hostetler (5):
  fsmonitor: enhance existing comments
  fsmonitor-ipc: create client routines for git-fsmonitor--daemon
  fsmonitor: config settings are repository-specific
  fsmonitor: use IPC to query the builtin FSMonitor daemon
  fsmonitor: update fsmonitor config documentation

 Documentation/config/core.txt      |  56 ++++++---
 Documentation/git-update-index.txt |  27 +++--
 Documentation/githooks.txt         |   3 +-
 Makefile                           |   2 +
 builtin/update-index.c             |  19 +++-
 cache.h                            |   1 -
 config.c                           |  14 ---
 config.h                           |   1 -
 environment.c                      |   1 -
 fsmonitor-ipc.c                    | 176 +++++++++++++++++++++++++++++
 fsmonitor-ipc.h                    |  48 ++++++++
 fsmonitor-settings.c               |  97 ++++++++++++++++
 fsmonitor-settings.h               |  21 ++++
 fsmonitor.c                        | 130 +++++++++++++++------
 fsmonitor.h                        |  18 ++-
 repo-settings.c                    |   2 +
 repository.h                       |   3 +
 t/README                           |   4 +-
 18 files changed, 538 insertions(+), 85 deletions(-)
 create mode 100644 fsmonitor-ipc.c
 create mode 100644 fsmonitor-ipc.h
 create mode 100644 fsmonitor-settings.c
 create mode 100644 fsmonitor-settings.h


base-commit: 8b7c11b8668b4e774f81a9f0b4c30144b818f1d1
Published-As: https://github.com/gitgitgadget/git/releases/tag/pr-1041%2Fjeffhostetler%2Fbuiltin-fsmonitor-part2-v1
Fetch-It-Via: git fetch https://github.com/gitgitgadget/git pr-1041/jeffhostetler/builtin-fsmonitor-part2-v1
Pull-Request: https://github.com/gitgitgadget/git/pull/1041
-- 
gitgitgadget

[PATCH 1/5] fsmonitor: enhance existing comments

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 fsmonitor.c | 37 ++++++++++++++++++++++++++++++-------
 1 file changed, 30 insertions(+), 7 deletions(-)

diff --git a/fsmonitor.c b/fsmonitor.c
index ab9bfc60b34..ec4c46407c5 100644
--- a/fsmonitor.c
+++ b/fsmonitor.c
@@ -301,9 +301,25 @@ void refresh_fsmonitor(struct index_state *istate)
 			core_fsmonitor, query_success ? "success" : "failure");
 	}
 
-	/* a fsmonitor process can return '/' to indicate all entries are invalid */
+	/*
+	 * The response from FSMonitor (excluding the header token) is
+	 * either:
+	 *
+	 * [a] a (possibly empty) list of NUL delimited relative
+	 *     pathnames of changed paths.  This list can contain
+	 *     files and directories.  Directories have a trailing
+	 *     slash.
+	 *
+	 * [b] a single '/' to indicate the provider had no
+	 *     information and that we should consider everything
+	 *     invalid.  We call this a trivial response.
+	 */
 	if (query_success && query_result.buf[bol] != '/') {
-		/* Mark all entries returned by the monitor as dirty */
+		/*
+		 * Mark all pathnames returned by the monitor as dirty.
+		 *
+		 * This updates both the cache-entries and the untracked-cache.
+		 */
 		buf = query_result.buf;
 		for (i = bol; i < query_result.len; i++) {
 			if (buf[i] != '\0')
@@ -318,11 +334,15 @@ void refresh_fsmonitor(struct index_state *istate)
 		if (istate->untracked)
 			istate->untracked->use_fsmonitor = 1;
 	} else {
-
-		/* We only want to run the post index changed hook if we've actually changed entries, so keep track
-		 * if we actually changed entries or not */
+		/*
+		 * We received a trivial response, so invalidate everything.
+		 *
+		 * We only want to run the post index changed hook if
+		 * we've actually changed entries, so keep track if we
+		 * actually changed entries or not.
+		 */
 		int is_cache_changed = 0;
-		/* Mark all entries invalid */
+
 		for (i = 0; i < istate->cache_nr; i++) {
 			if (istate->cache[i]->ce_flags & CE_FSMONITOR_VALID) {
 				is_cache_changed = 1;
@@ -330,7 +350,10 @@ void refresh_fsmonitor(struct index_state *istate)
 			}
 		}
 
-		/* If we're going to check every file, ensure we save the results */
+		/*
+		 * If we're going to check every file, ensure we save
+		 * the results.
+		 */
 		if (is_cache_changed)
 			istate->cache_changed |= FSMONITOR_CHANGED;
 
-- 
gitgitgadget

[PATCH 2/5] fsmonitor-ipc: create client routines for git-fsmonitor--daemon

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Create fsmonitor_ipc__*() client routines to spawn the built-in file
system monitor daemon and send it an IPC request using the `Simple
IPC` API.

Stub in empty fsmonitor_ipc__*() functions for unsupported platforms.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 Makefile        |   1 +
 fsmonitor-ipc.c | 176 ++++++++++++++++++++++++++++++++++++++++++++++++
 fsmonitor-ipc.h |  48 +++++++++++++
 3 files changed, 225 insertions(+)
 create mode 100644 fsmonitor-ipc.c
 create mode 100644 fsmonitor-ipc.h

diff --git a/Makefile b/Makefile
index 429c276058d..79e72ca9977 100644
--- a/Makefile
+++ b/Makefile
@@ -901,6 +901,7 @@ LIB_OBJS += fetch-pack.o
 LIB_OBJS += fmt-merge-msg.o
 LIB_OBJS += fsck.o
 LIB_OBJS += fsmonitor.o
+LIB_OBJS += fsmonitor-ipc.o
 LIB_OBJS += gettext.o
 LIB_OBJS += gpg-interface.o
 LIB_OBJS += graph.o
diff --git a/fsmonitor-ipc.c b/fsmonitor-ipc.c
new file mode 100644
index 00000000000..ccc32d2a17e
--- /dev/null
+++ b/fsmonitor-ipc.c
@@ -0,0 +1,176 @@
+#include "cache.h"
+#include "fsmonitor.h"
+#include "simple-ipc.h"
+#include "fsmonitor-ipc.h"
+#include "run-command.h"
+#include "strbuf.h"
+#include "trace2.h"
+
+#ifdef HAVE_FSMONITOR_DAEMON_BACKEND
+
+int fsmonitor_ipc__is_supported(void)
+{
+	return 1;
+}
+
+GIT_PATH_FUNC(fsmonitor_ipc__get_path, "fsmonitor--daemon.ipc")
+
+enum ipc_active_state fsmonitor_ipc__get_state(void)
+{
+	return ipc_get_active_state(fsmonitor_ipc__get_path());
+}
+
+static int spawn_daemon(void)
+{
+	const char *args[] = { "fsmonitor--daemon", "start", NULL };
+
+	return run_command_v_opt_tr2(args, RUN_COMMAND_NO_STDIN | RUN_GIT_CMD,
+				    "fsmonitor");
+}
+
+int fsmonitor_ipc__send_query(const char *since_token,
+			      struct strbuf *answer)
+{
+	int ret = -1;
+	int tried_to_spawn = 0;
+	enum ipc_active_state state = IPC_STATE__OTHER_ERROR;
+	struct ipc_client_connection *connection = NULL;
+	struct ipc_client_connect_options options
+		= IPC_CLIENT_CONNECT_OPTIONS_INIT;
+	const char *tok = since_token ? since_token : "";
+	size_t tok_len = since_token ? strlen(since_token) : 0;
+
+	options.wait_if_busy = 1;
+	options.wait_if_not_found = 0;
+
+	trace2_region_enter("fsm_client", "query", NULL);
+	trace2_data_string("fsm_client", NULL, "query/command", tok);
+
+try_again:
+	state = ipc_client_try_connect(fsmonitor_ipc__get_path(), &options,
+				       &connection);
+
+	switch (state) {
+	case IPC_STATE__LISTENING:
+		ret = ipc_client_send_command_to_connection(
+			connection, tok, tok_len, answer);
+		ipc_client_close_connection(connection);
+
+		trace2_data_intmax("fsm_client", NULL,
+				   "query/response-length", answer->len);
+
+		if (fsmonitor_is_trivial_response(answer))
+			trace2_data_intmax("fsm_client", NULL,
+					   "query/trivial-response", 1);
+
+		goto done;
+
+	case IPC_STATE__NOT_LISTENING:
+	case IPC_STATE__PATH_NOT_FOUND:
+		if (tried_to_spawn)
+			goto done;
+
+		tried_to_spawn++;
+		if (spawn_daemon())
+			goto done;
+
+		/*
+		 * Try again, but this time give the daemon a chance to
+		 * actually create the pipe/socket.
+		 *
+		 * Granted, the daemon just started so it can't possibly have
+		 * any FS cached yet, so we'll always get a trivial answer.
+		 * BUT the answer should include a new token that can serve
+		 * as the basis for subsequent requests.
+		 */
+		options.wait_if_not_found = 1;
+		goto try_again;
+
+	case IPC_STATE__INVALID_PATH:
+		ret = error(_("fsmonitor_ipc__send_query: invalid path '%s'"),
+			    fsmonitor_ipc__get_path());
+		goto done;
+
+	case IPC_STATE__OTHER_ERROR:
+	default:
+		ret = error(_("fsmonitor_ipc__send_query: unspecified error on '%s'"),
+			    fsmonitor_ipc__get_path());
+		goto done;
+	}
+
+done:
+	trace2_region_leave("fsm_client", "query", NULL);
+
+	return ret;
+}
+
+int fsmonitor_ipc__send_command(const char *command,
+				struct strbuf *answer)
+{
+	struct ipc_client_connection *connection = NULL;
+	struct ipc_client_connect_options options
+		= IPC_CLIENT_CONNECT_OPTIONS_INIT;
+	int ret;
+	enum ipc_active_state state;
+	const char *c = command ? command : "";
+	size_t c_len = command ? strlen(command) : 0;
+
+	strbuf_reset(answer);
+
+	options.wait_if_busy = 1;
+	options.wait_if_not_found = 0;
+
+	state = ipc_client_try_connect(fsmonitor_ipc__get_path(), &options,
+				       &connection);
+	if (state != IPC_STATE__LISTENING) {
+		die("fsmonitor--daemon is not running");
+		return -1;
+	}
+
+	ret = ipc_client_send_command_to_connection(connection, c, c_len,
+						    answer);
+	ipc_client_close_connection(connection);
+
+	if (ret == -1) {
+		die("could not send '%s' command to fsmonitor--daemon", c);
+		return -1;
+	}
+
+	return 0;
+}
+
+#else
+
+/*
+ * A trivial implementation of the fsmonitor_ipc__ API for unsupported
+ * platforms.
+ */
+
+int fsmonitor_ipc__is_supported(void)
+{
+	return 0;
+}
+
+const char *fsmonitor_ipc__get_path(void)
+{
+	return NULL;
+}
+
+enum ipc_active_state fsmonitor_ipc__get_state(void)
+{
+	return IPC_STATE__OTHER_ERROR;
+}
+
+int fsmonitor_ipc__send_query(const char *since_token,
+			      struct strbuf *answer)
+{
+	return -1;
+}
+
+int fsmonitor_ipc__send_command(const char *command,
+				struct strbuf *answer)
+{
+	return -1;
+}
+
+#endif
diff --git a/fsmonitor-ipc.h b/fsmonitor-ipc.h
new file mode 100644
index 00000000000..b6a7067c3af
--- /dev/null
+++ b/fsmonitor-ipc.h
@@ -0,0 +1,48 @@
+#ifndef FSMONITOR_IPC_H
+#define FSMONITOR_IPC_H
+
+#include "simple-ipc.h"
+
+/*
+ * Returns true if built-in file system monitor daemon is defined
+ * for this platform.
+ */
+int fsmonitor_ipc__is_supported(void);
+
+/*
+ * Returns the pathname to the IPC named pipe or Unix domain socket
+ * where a `git-fsmonitor--daemon` process will listen.  This is a
+ * per-worktree value.
+ *
+ * Returns NULL if the daemon is not supported on this platform.
+ */
+const char *fsmonitor_ipc__get_path(void);
+
+/*
+ * Try to determine whether there is a `git-fsmonitor--daemon` process
+ * listening on the IPC pipe/socket.
+ */
+enum ipc_active_state fsmonitor_ipc__get_state(void);
+
+/*
+ * Connect to a `git-fsmonitor--daemon` process via simple-ipc
+ * and ask for the set of changed files since the given token.
+ *
+ * Spawn a daemon process in the background if necessary.
+ *
+ * Returns -1 on error; 0 on success.
+ */
+int fsmonitor_ipc__send_query(const char *since_token,
+			      struct strbuf *answer);
+
+/*
+ * Connect to a `git-fsmonitor--daemon` process via simple-ipc and
+ * send a command verb.  If no daemon is available, we DO NOT try to
+ * start one.
+ *
+ * Returns -1 on error; 0 on success.
+ */
+int fsmonitor_ipc__send_command(const char *command,
+				struct strbuf *answer);
+
+#endif /* FSMONITOR_IPC_H */
-- 
gitgitgadget

[PATCH 3/5] fsmonitor: config settings are repository-specific

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Move FSMonitor config settings to a new `struct fsmonitor_settings`
structure.  Add a lazily-loaded pointer to `struct repo_settings`.
Create `fsm_settings__get_*()` getters to lazily look up fsmonitor-
related config settings.

Get rid of the `core_fsmonitor` global variable, and add support for
the new `core.useBuiltinFSMonitor` config setting.  Move config code
to lookup the existing `core.fsmonitor` value to `fsmonitor-settings.[ch]`.

The `core_fsmonitor` global variable was used to store the pathname to
the FSMonitor hook and it was used as a boolean to see if FSMonitor
was enabled.  This dual usage will lead to confusion when we add
support for a builtin FSMonitor based on IPC, since the builtin
FSMonitor doesn't need the hook pathname.

Replace the boolean usage with an `enum fsmonitor_mode` to represent
the state of FSMonitor.  And only set the pathname when in HOOK mode.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 Makefile               |  1 +
 builtin/update-index.c | 19 +++++++--
 cache.h                |  1 -
 config.c               | 14 ------
 config.h               |  1 -
 environment.c          |  1 -
 fsmonitor-settings.c   | 97 ++++++++++++++++++++++++++++++++++++++++++
 fsmonitor-settings.h   | 21 +++++++++
 fsmonitor.c            | 63 ++++++++++++++++-----------
 fsmonitor.h            | 18 ++++++--
 repo-settings.c        |  2 +
 repository.h           |  3 ++
 t/README               |  4 +-
 13 files changed, 194 insertions(+), 51 deletions(-)
 create mode 100644 fsmonitor-settings.c
 create mode 100644 fsmonitor-settings.h

diff --git a/Makefile b/Makefile
index 79e72ca9977..8b5d241df72 100644
--- a/Makefile
+++ b/Makefile
@@ -902,6 +902,7 @@ LIB_OBJS += fmt-merge-msg.o
 LIB_OBJS += fsck.o
 LIB_OBJS += fsmonitor.o
 LIB_OBJS += fsmonitor-ipc.o
+LIB_OBJS += fsmonitor-settings.o
 LIB_OBJS += gettext.o
 LIB_OBJS += gpg-interface.o
 LIB_OBJS += graph.o
diff --git a/builtin/update-index.c b/builtin/update-index.c
index 187203e8bb5..79db3ff37e2 100644
--- a/builtin/update-index.c
+++ b/builtin/update-index.c
@@ -1214,14 +1214,25 @@ int cmd_update_index(int argc, const char **argv, const char *prefix)
 	}
 
 	if (fsmonitor > 0) {
-		if (git_config_get_fsmonitor() == 0)
+		enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
+
+		if (fsm_mode == FSMONITOR_MODE_DISABLED) {
+			warning(_("core.useBuiltinFSMonitor is unset; "
+				"set it if you really want to enable the "
+				"builtin fsmonitor"));
 			warning(_("core.fsmonitor is unset; "
-				"set it if you really want to "
-				"enable fsmonitor"));
+				"set it if you really want to enable the "
+				"hook-based fsmonitor"));
+		}
 		add_fsmonitor(&the_index);
 		report(_("fsmonitor enabled"));
 	} else if (!fsmonitor) {
-		if (git_config_get_fsmonitor() == 1)
+		enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
+		if (fsm_mode == FSMONITOR_MODE_IPC)
+			warning(_("core.useBuiltinFSMonitor is set; "
+				"remove it if you really want to "
+				"disable fsmonitor"));
+		if (fsm_mode == FSMONITOR_MODE_HOOK)
 			warning(_("core.fsmonitor is set; "
 				"remove it if you really want to "
 				"disable fsmonitor"));
diff --git a/cache.h b/cache.h
index d23de693680..fa83dc2aa84 100644
--- a/cache.h
+++ b/cache.h
@@ -990,7 +990,6 @@ extern int core_preload_index;
 extern int precomposed_unicode;
 extern int protect_hfs;
 extern int protect_ntfs;
-extern const char *core_fsmonitor;
 
 extern int core_apply_sparse_checkout;
 extern int core_sparse_checkout_cone;
diff --git a/config.c b/config.c
index cb4a8058bff..7389ddf18f6 100644
--- a/config.c
+++ b/config.c
@@ -2517,20 +2517,6 @@ int git_config_get_max_percent_split_change(void)
 	return -1; /* default value */
 }
 
-int git_config_get_fsmonitor(void)
-{
-	if (git_config_get_pathname("core.fsmonitor", &core_fsmonitor))
-		core_fsmonitor = getenv("GIT_TEST_FSMONITOR");
-
-	if (core_fsmonitor && !*core_fsmonitor)
-		core_fsmonitor = NULL;
-
-	if (core_fsmonitor)
-		return 1;
-
-	return 0;
-}
-
 int git_config_get_index_threads(int *dest)
 {
 	int is_bool, val;
diff --git a/config.h b/config.h
index a2200f31115..110c426b082 100644
--- a/config.h
+++ b/config.h
@@ -609,7 +609,6 @@ int git_config_get_index_threads(int *dest);
 int git_config_get_untracked_cache(void);
 int git_config_get_split_index(void);
 int git_config_get_max_percent_split_change(void);
-int git_config_get_fsmonitor(void);
 
 /* This dies if the configured or default date is in the future */
 int git_config_get_expiry(const char *key, const char **output);
diff --git a/environment.c b/environment.c
index d6b22ede7ea..e6b66315284 100644
--- a/environment.c
+++ b/environment.c
@@ -84,7 +84,6 @@ int protect_hfs = PROTECT_HFS_DEFAULT;
 #define PROTECT_NTFS_DEFAULT 1
 #endif
 int protect_ntfs = PROTECT_NTFS_DEFAULT;
-const char *core_fsmonitor;
 
 /*
  * The character that begins a commented line in user-editable file
diff --git a/fsmonitor-settings.c b/fsmonitor-settings.c
new file mode 100644
index 00000000000..2770266f5ee
--- /dev/null
+++ b/fsmonitor-settings.c
@@ -0,0 +1,97 @@
+#include "cache.h"
+#include "config.h"
+#include "repository.h"
+#include "fsmonitor-settings.h"
+
+/*
+ * We keep this structure defintion private and have getters
+ * for all fields so that we can lazy load it as needed.
+ */
+struct fsmonitor_settings {
+	enum fsmonitor_mode mode;
+	char *hook_path;
+};
+
+void fsm_settings__set_ipc(struct repository *r)
+{
+	struct fsmonitor_settings *s = r->settings.fsmonitor;
+
+	s->mode = FSMONITOR_MODE_IPC;
+}
+
+void fsm_settings__set_hook(struct repository *r, const char *path)
+{
+	struct fsmonitor_settings *s = r->settings.fsmonitor;
+
+	s->mode = FSMONITOR_MODE_HOOK;
+	s->hook_path = strdup(path);
+}
+
+void fsm_settings__set_disabled(struct repository *r)
+{
+	struct fsmonitor_settings *s = r->settings.fsmonitor;
+
+	s->mode = FSMONITOR_MODE_DISABLED;
+	FREE_AND_NULL(s->hook_path);
+}
+
+static int check_for_ipc(struct repository *r)
+{
+	int value;
+
+	if (!repo_config_get_bool(r, "core.usebuiltinfsmonitor", &value) &&
+	    value) {
+		fsm_settings__set_ipc(r);
+		return 1;
+	}
+
+	return 0;
+}
+
+static int check_for_hook(struct repository *r)
+{
+	const char *const_str;
+
+	if (repo_config_get_pathname(r, "core.fsmonitor", &const_str))
+		const_str = getenv("GIT_TEST_FSMONITOR");
+
+	if (const_str && *const_str) {
+		fsm_settings__set_hook(r, const_str);
+		return 1;
+	}
+
+	return 0;
+}
+
+static void lookup_fsmonitor_settings(struct repository *r)
+{
+	struct fsmonitor_settings *s;
+
+	CALLOC_ARRAY(s, 1);
+
+	r->settings.fsmonitor = s;
+
+	if (check_for_ipc(r))
+		return;
+
+	if (check_for_hook(r))
+		return;
+
+	fsm_settings__set_disabled(r);
+}
+
+enum fsmonitor_mode fsm_settings__get_mode(struct repository *r)
+{
+	if (!r->settings.fsmonitor)
+		lookup_fsmonitor_settings(r);
+
+	return r->settings.fsmonitor->mode;
+}
+
+const char *fsm_settings__get_hook_path(struct repository *r)
+{
+	if (!r->settings.fsmonitor)
+		lookup_fsmonitor_settings(r);
+
+	return r->settings.fsmonitor->hook_path;
+}
diff --git a/fsmonitor-settings.h b/fsmonitor-settings.h
new file mode 100644
index 00000000000..50b29234616
--- /dev/null
+++ b/fsmonitor-settings.h
@@ -0,0 +1,21 @@
+#ifndef FSMONITOR_SETTINGS_H
+#define FSMONITOR_SETTINGS_H
+
+struct repository;
+
+enum fsmonitor_mode {
+	FSMONITOR_MODE_DISABLED = 0,
+	FSMONITOR_MODE_HOOK = 1, /* core.fsmonitor */
+	FSMONITOR_MODE_IPC = 2,  /* core.useBuiltinFSMonitor */
+};
+
+void fsm_settings__set_ipc(struct repository *r);
+void fsm_settings__set_hook(struct repository *r, const char *path);
+void fsm_settings__set_disabled(struct repository *r);
+
+enum fsmonitor_mode fsm_settings__get_mode(struct repository *r);
+const char *fsm_settings__get_hook_path(struct repository *r);
+
+struct fsmonitor_settings;
+
+#endif /* FSMONITOR_SETTINGS_H */
diff --git a/fsmonitor.c b/fsmonitor.c
index ec4c46407c5..63174630c0e 100644
--- a/fsmonitor.c
+++ b/fsmonitor.c
@@ -3,6 +3,7 @@
 #include "dir.h"
 #include "ewah/ewok.h"
 #include "fsmonitor.h"
+#include "fsmonitor-ipc.h"
 #include "run-command.h"
 #include "strbuf.h"
 
@@ -148,15 +149,18 @@ void write_fsmonitor_extension(struct strbuf *sb, struct index_state *istate)
 /*
  * Call the query-fsmonitor hook passing the last update token of the saved results.
  */
-static int query_fsmonitor(int version, const char *last_update, struct strbuf *query_result)
+static int query_fsmonitor_hook(struct repository *r,
+				int version,
+				const char *last_update,
+				struct strbuf *query_result)
 {
 	struct child_process cp = CHILD_PROCESS_INIT;
 	int result;
 
-	if (!core_fsmonitor)
+	if (fsm_settings__get_mode(r) != FSMONITOR_MODE_HOOK)
 		return -1;
 
-	strvec_push(&cp.args, core_fsmonitor);
+	strvec_push(&cp.args, fsm_settings__get_hook_path(r));
 	strvec_pushf(&cp.args, "%d", version);
 	strvec_pushf(&cp.args, "%s", last_update);
 	cp.use_shell = 1;
@@ -238,17 +242,28 @@ void refresh_fsmonitor(struct index_state *istate)
 	struct strbuf last_update_token = STRBUF_INIT;
 	char *buf;
 	unsigned int i;
+	struct repository *r = istate->repo ? istate->repo : the_repository;
+	enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
 
-	if (!core_fsmonitor || istate->fsmonitor_has_run_once)
+	if (fsm_mode <= FSMONITOR_MODE_DISABLED ||
+	    istate->fsmonitor_has_run_once)
 		return;
 
-	hook_version = fsmonitor_hook_version();
-
 	istate->fsmonitor_has_run_once = 1;
 
 	trace_printf_key(&trace_fsmonitor, "refresh fsmonitor");
+
+	if (fsm_mode == FSMONITOR_MODE_IPC) {
+		/* TODO */
+		return;
+	}
+
+	assert(fsm_mode == FSMONITOR_MODE_HOOK);
+
+	hook_version = fsmonitor_hook_version();
+
 	/*
-	 * This could be racy so save the date/time now and query_fsmonitor
+	 * This could be racy so save the date/time now and query_fsmonitor_hook
 	 * should be inclusive to ensure we don't miss potential changes.
 	 */
 	last_update = getnanotime();
@@ -256,13 +271,14 @@ void refresh_fsmonitor(struct index_state *istate)
 		strbuf_addf(&last_update_token, "%"PRIu64"", last_update);
 
 	/*
-	 * If we have a last update token, call query_fsmonitor for the set of
+	 * If we have a last update token, call query_fsmonitor_hook for the set of
 	 * changes since that token, else assume everything is possibly dirty
 	 * and check it all.
 	 */
 	if (istate->fsmonitor_last_update) {
 		if (hook_version == -1 || hook_version == HOOK_INTERFACE_VERSION2) {
-			query_success = !query_fsmonitor(HOOK_INTERFACE_VERSION2,
+			query_success = !query_fsmonitor_hook(
+				r, HOOK_INTERFACE_VERSION2,
 				istate->fsmonitor_last_update, &query_result);
 
 			if (query_success) {
@@ -292,13 +308,17 @@ void refresh_fsmonitor(struct index_state *istate)
 		}
 
 		if (hook_version == HOOK_INTERFACE_VERSION1) {
-			query_success = !query_fsmonitor(HOOK_INTERFACE_VERSION1,
+			query_success = !query_fsmonitor_hook(
+				r, HOOK_INTERFACE_VERSION1,
 				istate->fsmonitor_last_update, &query_result);
 		}
 
-		trace_performance_since(last_update, "fsmonitor process '%s'", core_fsmonitor);
-		trace_printf_key(&trace_fsmonitor, "fsmonitor process '%s' returned %s",
-			core_fsmonitor, query_success ? "success" : "failure");
+		trace_performance_since(last_update, "fsmonitor process '%s'",
+					fsm_settings__get_hook_path(r));
+		trace_printf_key(&trace_fsmonitor,
+				 "fsmonitor process '%s' returned %s",
+				 fsm_settings__get_hook_path(r),
+				 query_success ? "success" : "failure");
 	}
 
 	/*
@@ -434,7 +454,8 @@ void remove_fsmonitor(struct index_state *istate)
 void tweak_fsmonitor(struct index_state *istate)
 {
 	unsigned int i;
-	int fsmonitor_enabled = git_config_get_fsmonitor();
+	struct repository *r = istate->repo ? istate->repo : the_repository;
+	int fsmonitor_enabled = (fsm_settings__get_mode(r) > FSMONITOR_MODE_DISABLED);
 
 	if (istate->fsmonitor_dirty) {
 		if (fsmonitor_enabled) {
@@ -454,16 +475,8 @@ void tweak_fsmonitor(struct index_state *istate)
 		istate->fsmonitor_dirty = NULL;
 	}
 
-	switch (fsmonitor_enabled) {
-	case -1: /* keep: do nothing */
-		break;
-	case 0: /* false */
-		remove_fsmonitor(istate);
-		break;
-	case 1: /* true */
+	if (fsmonitor_enabled)
 		add_fsmonitor(istate);
-		break;
-	default: /* unknown value: do nothing */
-		break;
-	}
+	else
+		remove_fsmonitor(istate);
 }
diff --git a/fsmonitor.h b/fsmonitor.h
index f20d72631d7..f9201411aa7 100644
--- a/fsmonitor.h
+++ b/fsmonitor.h
@@ -3,6 +3,7 @@
 
 #include "cache.h"
 #include "dir.h"
+#include "fsmonitor-settings.h"
 
 extern struct trace_key trace_fsmonitor;
 
@@ -57,7 +58,11 @@ int fsmonitor_is_trivial_response(const struct strbuf *query_result);
  */
 static inline int is_fsmonitor_refreshed(const struct index_state *istate)
 {
-	return !core_fsmonitor || istate->fsmonitor_has_run_once;
+	struct repository *r = istate->repo ? istate->repo : the_repository;
+	enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
+
+	return fsm_mode <= FSMONITOR_MODE_DISABLED ||
+		istate->fsmonitor_has_run_once;
 }
 
 /*
@@ -67,7 +72,11 @@ static inline int is_fsmonitor_refreshed(const struct index_state *istate)
  */
 static inline void mark_fsmonitor_valid(struct index_state *istate, struct cache_entry *ce)
 {
-	if (core_fsmonitor && !(ce->ce_flags & CE_FSMONITOR_VALID)) {
+	struct repository *r = istate->repo ? istate->repo : the_repository;
+	enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
+
+	if (fsm_mode > FSMONITOR_MODE_DISABLED &&
+	    !(ce->ce_flags & CE_FSMONITOR_VALID)) {
 		istate->cache_changed = 1;
 		ce->ce_flags |= CE_FSMONITOR_VALID;
 		trace_printf_key(&trace_fsmonitor, "mark_fsmonitor_clean '%s'", ce->name);
@@ -83,7 +92,10 @@ static inline void mark_fsmonitor_valid(struct index_state *istate, struct cache
  */
 static inline void mark_fsmonitor_invalid(struct index_state *istate, struct cache_entry *ce)
 {
-	if (core_fsmonitor) {
+	struct repository *r = istate->repo ? istate->repo : the_repository;
+	enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
+
+	if (fsm_mode > FSMONITOR_MODE_DISABLED) {
 		ce->ce_flags &= ~CE_FSMONITOR_VALID;
 		untracked_cache_invalidate_path(istate, ce->name, 1);
 		trace_printf_key(&trace_fsmonitor, "mark_fsmonitor_invalid '%s'", ce->name);
diff --git a/repo-settings.c b/repo-settings.c
index 0cfe8b787db..e69dad1e776 100644
--- a/repo-settings.c
+++ b/repo-settings.c
@@ -26,6 +26,8 @@ void prepare_repo_settings(struct repository *r)
 	UPDATE_DEFAULT_BOOL(r->settings.commit_graph_read_changed_paths, 1);
 	UPDATE_DEFAULT_BOOL(r->settings.gc_write_commit_graph, 1);
 
+	r->settings.fsmonitor = NULL; /* lazy loaded */
+
 	if (!repo_config_get_int(r, "index.version", &value))
 		r->settings.index_version = value;
 	if (!repo_config_get_maybe_bool(r, "core.untrackedcache", &value)) {
diff --git a/repository.h b/repository.h
index 3740c93bc0f..fdc0a818b83 100644
--- a/repository.h
+++ b/repository.h
@@ -4,6 +4,7 @@
 #include "path.h"
 
 struct config_set;
+struct fsmonitor_settings;
 struct git_hash_algo;
 struct index_state;
 struct lock_file;
@@ -35,6 +36,8 @@ struct repo_settings {
 	int gc_write_commit_graph;
 	int fetch_write_commit_graph;
 
+	struct fsmonitor_settings *fsmonitor; /* lazy loaded */
+
 	int index_version;
 	enum untracked_cache_setting core_untracked_cache;
 
diff --git a/t/README b/t/README
index 9e701223020..7be3662e038 100644
--- a/t/README
+++ b/t/README
@@ -398,8 +398,8 @@ every 'git commit-graph write', as if the `--changed-paths` option was
 passed in.
 
 GIT_TEST_FSMONITOR=$PWD/t7519/fsmonitor-all exercises the fsmonitor
-code path for utilizing a file system monitor to speed up detecting
-new or changed files.
+code path for utilizing a (hook based) file system monitor to speed up
+detecting new or changed files.
 
 GIT_TEST_INDEX_VERSION=<n> exercises the index read/write code path
 for the index version specified.  Can be set to any valid version
-- 
gitgitgadget

[PATCH 4/5] fsmonitor: use IPC to query the builtin FSMonitor daemon

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Use simple IPC to directly communicate with the new builtin file
system monitor daemon when `core.useBuiltinFSMonitor` is set.

The `core.fsmonitor` setting has already been defined as a HOOK
pathname.  Historically, this has been set to a HOOK script that will
talk with Watchman.  For compatibility reasons, we do not want to
overload that definition (and cause problems if users have multiple
versions of Git installed).

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 fsmonitor.c | 34 ++++++++++++++++++++++++++++++++--
 1 file changed, 32 insertions(+), 2 deletions(-)

diff --git a/fsmonitor.c b/fsmonitor.c
index 63174630c0e..695fb0ce4e7 100644
--- a/fsmonitor.c
+++ b/fsmonitor.c
@@ -254,8 +254,37 @@ void refresh_fsmonitor(struct index_state *istate)
 	trace_printf_key(&trace_fsmonitor, "refresh fsmonitor");
 
 	if (fsm_mode == FSMONITOR_MODE_IPC) {
-		/* TODO */
-		return;
+		query_success = !fsmonitor_ipc__send_query(
+			istate->fsmonitor_last_update ?
+			istate->fsmonitor_last_update : "builtin:fake",
+			&query_result);
+		if (query_success) {
+			/*
+			 * The response contains a series of nul terminated
+			 * strings.  The first is the new token.
+			 *
+			 * Use `char *buf` as an interlude to trick the CI
+			 * static analysis to let us use `strbuf_addstr()`
+			 * here (and only copy the token) rather than
+			 * `strbuf_addbuf()`.
+			 */
+			buf = query_result.buf;
+			strbuf_addstr(&last_update_token, buf);
+			bol = last_update_token.len + 1;
+		} else {
+			/*
+			 * The builtin daemon is not available on this
+			 * platform -OR- we failed to get a response.
+			 *
+			 * Generate a fake token (rather than a V1
+			 * timestamp) for the index extension.  (If
+			 * they switch back to the hook API, we don't
+			 * want ambiguous state.)
+			 */
+			strbuf_addstr(&last_update_token, "builtin:fake");
+		}
+
+		goto apply_results;
 	}
 
 	assert(fsm_mode == FSMONITOR_MODE_HOOK);
@@ -321,6 +350,7 @@ void refresh_fsmonitor(struct index_state *istate)
 				 query_success ? "success" : "failure");
 	}
 
+apply_results:
 	/*
 	 * The response from FSMonitor (excluding the header token) is
 	 * either:
-- 
gitgitgadget

[PATCH 5/5] fsmonitor: update fsmonitor config documentation

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Update references to `core.fsmonitor` and `core.fsmonitorHookVersion` and
pointers to `Watchman` to mention the new `core.useBuiltinFSMonitor`
value.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 Documentation/config/core.txt      | 56 ++++++++++++++++++++++--------
 Documentation/git-update-index.txt | 27 +++++++-------
 Documentation/githooks.txt         |  3 +-
 3 files changed, 59 insertions(+), 27 deletions(-)

diff --git a/Documentation/config/core.txt b/Documentation/config/core.txt
index c04f62a54a1..4f6e519bc02 100644
--- a/Documentation/config/core.txt
+++ b/Documentation/config/core.txt
@@ -62,22 +62,50 @@ core.protectNTFS::
 	Defaults to `true` on Windows, and `false` elsewhere.
 
 core.fsmonitor::
-	If set, the value of this variable is used as a command which
-	will identify all files that may have changed since the
-	requested date/time. This information is used to speed up git by
-	avoiding unnecessary processing of files that have not changed.
-	See the "fsmonitor-watchman" section of linkgit:githooks[5].
+	If set, this variable contains the pathname of the "fsmonitor"
+	hook command.
++
+This hook command is used to identify all files that may have changed
+since the requested date/time. This information is used to speed up
+git by avoiding unnecessary scanning of files that have not changed.
++
+See the "fsmonitor-watchman" section of linkgit:githooks[5].
++
+Note: The value of this config setting is ignored if the
+built-in file system monitor is enabled (see `core.useBuiltinFSMonitor`).
 
 core.fsmonitorHookVersion::
-	Sets the version of hook that is to be used when calling fsmonitor.
-	There are currently versions 1 and 2. When this is not set,
-	version 2 will be tried first and if it fails then version 1
-	will be tried. Version 1 uses a timestamp as input to determine
-	which files have changes since that time but some monitors
-	like watchman have race conditions when used with a timestamp.
-	Version 2 uses an opaque string so that the monitor can return
-	something that can be used to determine what files have changed
-	without race conditions.
+	Sets the protocol version to be used when invoking the
+	"fsmonitor" hook.
++
+There are currently versions 1 and 2. When this is not set,
+version 2 will be tried first and if it fails then version 1
+will be tried. Version 1 uses a timestamp as input to determine
+which files have changes since that time but some monitors
+like Watchman have race conditions when used with a timestamp.
+Version 2 uses an opaque string so that the monitor can return
+something that can be used to determine what files have changed
+without race conditions.
++
+Note: The value of this config setting is ignored if the
+built-in file system monitor is enabled (see `core.useBuiltinFSMonitor`).
+
+core.useBuiltinFSMonitor::
+	If set to true, enable the built-in file system monitor
+	daemon for this working directory (linkgit:git-fsmonitor--daemon[1]).
++
+Like hook-based file system monitors, the built-in file system monitor
+can speed up Git commands that need to refresh the Git index
+(e.g. `git status`) in a working directory with many files.  The
+built-in monitor eliminates the need to install and maintain an
+external third-party tool.
++
+The built-in file system monitor is currently available only on a
+limited set of supported platforms.  Currently, this includes Windows
+and MacOS.
++
+Note: if this config setting is set to `true`, the values of
+`core.fsmonitor` and `core.fsmonitorHookVersion` are ignored.
 
 core.trustctime::
 	If false, the ctime differences between the index and the
diff --git a/Documentation/git-update-index.txt b/Documentation/git-update-index.txt
index 2853f168d97..c7c31b3fcf9 100644
--- a/Documentation/git-update-index.txt
+++ b/Documentation/git-update-index.txt
@@ -498,7 +498,9 @@ FILE SYSTEM MONITOR
 This feature is intended to speed up git operations for repos that have
 large working directories.
 
-It enables git to work together with a file system monitor (see the
+It enables git to work together with a file system monitor (see
+linkgit:git-fsmonitor--daemon[1]
+and the
 "fsmonitor-watchman" section of linkgit:githooks[5]) that can
 inform it as to what files have been modified. This enables git to avoid
 having to lstat() every file to find modified files.
@@ -508,17 +510,18 @@ performance by avoiding the cost of scanning the entire working directory
 looking for new files.
 
 If you want to enable (or disable) this feature, it is easier to use
-the `core.fsmonitor` configuration variable (see
-linkgit:git-config[1]) than using the `--fsmonitor` option to
-`git update-index` in each repository, especially if you want to do so
-across all repositories you use, because you can set the configuration
-variable in your `$HOME/.gitconfig` just once and have it affect all
-repositories you touch.
-
-When the `core.fsmonitor` configuration variable is changed, the
-file system monitor is added to or removed from the index the next time
-a command reads the index. When `--[no-]fsmonitor` are used, the file
-system monitor is immediately added to or removed from the index.
+the `core.fsmonitor` or `core.useBuiltinFSMonitor` configuration
+variable (see linkgit:git-config[1]) than using the `--fsmonitor`
+option to `git update-index` in each repository, especially if you
+want to do so across all repositories you use, because you can set the
+configuration variable in your `$HOME/.gitconfig` just once and have
+it affect all repositories you touch.
+
+When the `core.fsmonitor` or `core.useBuiltinFSMonitor` configuration
+variable is changed, the file system monitor is added to or removed
+from the index the next time a command reads the index. When
+`--[no-]fsmonitor` are used, the file system monitor is immediately
+added to or removed from the index.
 
 CONFIGURATION
 -------------
diff --git a/Documentation/githooks.txt b/Documentation/githooks.txt
index b51959ff941..b7d5e926f7b 100644
--- a/Documentation/githooks.txt
+++ b/Documentation/githooks.txt
@@ -593,7 +593,8 @@ fsmonitor-watchman
 
 This hook is invoked when the configuration option `core.fsmonitor` is
 set to `.git/hooks/fsmonitor-watchman` or `.git/hooks/fsmonitor-watchmanv2`
-depending on the version of the hook to use.
+depending on the version of the hook to use, unless overridden via
+`core.useBuiltinFSMonitor` (see linkgit:git-config[1]).
 
 Version 1 takes two arguments, a version (1) and the time in elapsed
 nanoseconds since midnight, January 1, 1970.
-- 
gitgitgadget

Re: [PATCH 1/5] fsmonitor: enhance existing comments

[Bagas Sanjaya]

On 17/09/21 02.54, Jeff Hostetler via GitGitGadget wrote:
> -		/* If we're going to check every file, ensure we save the results */
> +		/*
> +		 * If we're going to check every file, ensure we save
> +		 * the results.
> +		 */

Why did you split the comment above?

-- 
An old man doll... just what I always wanted! - Clara

Re: [PATCH 1/5] fsmonitor: enhance existing comments

[Junio C Hamano]

Bagas Sanjaya <bagasdotme@gmail.com> writes:

> On 17/09/21 02.54, Jeff Hostetler via GitGitGadget wrote:
>> -		/* If we're going to check every file, ensure we save the results */
>> +		/*
>> +		 * If we're going to check every file, ensure we save
>> +		 * the results.
>> +		 */
>
> Why did you split the comment above?

I would guess that the reason why it is done is because the original
line is overly long it (extends to 84 columns, if I am counting
correctly).

Re: [PATCH 1/5] fsmonitor: enhance existing comments

[Jeff Hostetler]

On 9/17/21 2:44 AM, Junio C Hamano wrote:
> Bagas Sanjaya <bagasdotme@gmail.com> writes:
> 
>> On 17/09/21 02.54, Jeff Hostetler via GitGitGadget wrote:
>>> -		/* If we're going to check every file, ensure we save the results */
>>> +		/*
>>> +		 * If we're going to check every file, ensure we save
>>> +		 * the results.
>>> +		 */
>>
>> Why did you split the comment above?
> 
> I would guess that the reason why it is done is because the original
> line is overly long it (extends to 84 columns, if I am counting
> correctly).
> 

Yes, I just wrapped it because it was too long and the commit
was focused on improving other nearby comments (and no code),
so it seemed like a good opportunity to cleanup this one too.

Jeff

[PATCH v2 0/5] Builtin FSMonitor Part 2

[Jeff Hostetler via GitGitGadget]

Here is V2 of Part 2 of my Builtin FSMonitor series.

This version has been rebased onto "next" because it collided with a change
"repo-settings.c" from "ab/repo-settings-cleanup". And the third commit was
updated to resolve those conflicts.

It also requires "jh/builtin-fsmonitor-part1" which is now in "next".

Part 2 contains:

Client-side code to connect to a (yet to be developed) FSMonitor daemon over
Simple IPC (using Named Pipes or Unix Domain Sockets).

Updated config settings to allow the client to decide whether to use Simple
IPC or the existing FSMonitor Hook API. This includes retiring the existing
"core_fsmonitor" global variable that had confusing double-duty.

Part 2 essentially enables client-side code to talk to different types of
providers of FS change data. Processing of that data is unchanged.

A builtin FSMonitor daemon will be added in a later patch series.

Jeff Hostetler (5):
  fsmonitor: enhance existing comments
  fsmonitor-ipc: create client routines for git-fsmonitor--daemon
  fsmonitor: config settings are repository-specific
  fsmonitor: use IPC to query the builtin FSMonitor daemon
  fsmonitor: update fsmonitor config documentation

 Documentation/config/core.txt      |  56 ++++++---
 Documentation/git-update-index.txt |  27 +++--
 Documentation/githooks.txt         |   3 +-
 Makefile                           |   2 +
 builtin/update-index.c             |  19 +++-
 cache.h                            |   1 -
 config.c                           |  14 ---
 config.h                           |   1 -
 environment.c                      |   1 -
 fsmonitor-ipc.c                    | 176 +++++++++++++++++++++++++++++
 fsmonitor-ipc.h                    |  48 ++++++++
 fsmonitor-settings.c               |  97 ++++++++++++++++
 fsmonitor-settings.h               |  21 ++++
 fsmonitor.c                        | 130 +++++++++++++++------
 fsmonitor.h                        |  18 ++-
 repo-settings.c                    |   2 +
 repository.h                       |   3 +
 t/README                           |   4 +-
 18 files changed, 538 insertions(+), 85 deletions(-)
 create mode 100644 fsmonitor-ipc.c
 create mode 100644 fsmonitor-ipc.h
 create mode 100644 fsmonitor-settings.c
 create mode 100644 fsmonitor-settings.h


base-commit: 6e70778dc91e2139466c15ff15a02a22a2ada2d1
Published-As: https://github.com/gitgitgadget/git/releases/tag/pr-1041%2Fjeffhostetler%2Fbuiltin-fsmonitor-part2-v2
Fetch-It-Via: git fetch https://github.com/gitgitgadget/git pr-1041/jeffhostetler/builtin-fsmonitor-part2-v2
Pull-Request: https://github.com/gitgitgadget/git/pull/1041

Range-diff vs v1:

 1:  6be687ba39d = 1:  cb25eeaf72d fsmonitor: enhance existing comments
 2:  b584f133d28 = 2:  df81a63acee fsmonitor-ipc: create client routines for git-fsmonitor--daemon
 3:  f1266c99adf ! 3:  7d5a353e74d fsmonitor: config settings are repository-specific
     @@ config.c: int git_config_get_max_percent_split_change(void)
       	int is_bool, val;
      
       ## config.h ##
     -@@ config.h: int git_config_get_index_threads(int *dest);
     - int git_config_get_untracked_cache(void);
     +@@ config.h: int git_config_get_pathname(const char *key, const char **dest);
     + int git_config_get_index_threads(int *dest);
       int git_config_get_split_index(void);
       int git_config_get_max_percent_split_change(void);
      -int git_config_get_fsmonitor(void);
     @@ fsmonitor.h: static inline void mark_fsmonitor_valid(struct index_state *istate,
      
       ## repo-settings.c ##
      @@ repo-settings.c: void prepare_repo_settings(struct repository *r)
     - 	UPDATE_DEFAULT_BOOL(r->settings.commit_graph_read_changed_paths, 1);
     - 	UPDATE_DEFAULT_BOOL(r->settings.gc_write_commit_graph, 1);
     + 	if (r->settings.initialized++)
     + 		return;
       
      +	r->settings.fsmonitor = NULL; /* lazy loaded */
      +
     - 	if (!repo_config_get_int(r, "index.version", &value))
     - 		r->settings.index_version = value;
     - 	if (!repo_config_get_maybe_bool(r, "core.untrackedcache", &value)) {
     + 	/* Defaults */
     + 	r->settings.index_version = -1;
     + 	r->settings.core_untracked_cache = UNTRACKED_CACHE_KEEP;
      
       ## repository.h ##
      @@
     @@ repository.h
       struct index_state;
       struct lock_file;
      @@ repository.h: struct repo_settings {
     - 	int gc_write_commit_graph;
     - 	int fetch_write_commit_graph;
     + 	int command_requires_full_index;
     + 	int sparse_index;
       
      +	struct fsmonitor_settings *fsmonitor; /* lazy loaded */
      +
 4:  f2e5da9e5c6 = 4:  8608c8718d8 fsmonitor: use IPC to query the builtin FSMonitor daemon
 5:  fb5251e47ae = 5:  7c22ce53377 fsmonitor: update fsmonitor config documentation

-- 
gitgitgadget

[PATCH v2 1/5] fsmonitor: enhance existing comments

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 fsmonitor.c | 37 ++++++++++++++++++++++++++++++-------
 1 file changed, 30 insertions(+), 7 deletions(-)

diff --git a/fsmonitor.c b/fsmonitor.c
index ab9bfc60b34..ec4c46407c5 100644
--- a/fsmonitor.c
+++ b/fsmonitor.c
@@ -301,9 +301,25 @@ void refresh_fsmonitor(struct index_state *istate)
 			core_fsmonitor, query_success ? "success" : "failure");
 	}
 
-	/* a fsmonitor process can return '/' to indicate all entries are invalid */
+	/*
+	 * The response from FSMonitor (excluding the header token) is
+	 * either:
+	 *
+	 * [a] a (possibly empty) list of NUL delimited relative
+	 *     pathnames of changed paths.  This list can contain
+	 *     files and directories.  Directories have a trailing
+	 *     slash.
+	 *
+	 * [b] a single '/' to indicate the provider had no
+	 *     information and that we should consider everything
+	 *     invalid.  We call this a trivial response.
+	 */
 	if (query_success && query_result.buf[bol] != '/') {
-		/* Mark all entries returned by the monitor as dirty */
+		/*
+		 * Mark all pathnames returned by the monitor as dirty.
+		 *
+		 * This updates both the cache-entries and the untracked-cache.
+		 */
 		buf = query_result.buf;
 		for (i = bol; i < query_result.len; i++) {
 			if (buf[i] != '\0')
@@ -318,11 +334,15 @@ void refresh_fsmonitor(struct index_state *istate)
 		if (istate->untracked)
 			istate->untracked->use_fsmonitor = 1;
 	} else {
-
-		/* We only want to run the post index changed hook if we've actually changed entries, so keep track
-		 * if we actually changed entries or not */
+		/*
+		 * We received a trivial response, so invalidate everything.
+		 *
+		 * We only want to run the post index changed hook if
+		 * we've actually changed entries, so keep track if we
+		 * actually changed entries or not.
+		 */
 		int is_cache_changed = 0;
-		/* Mark all entries invalid */
+
 		for (i = 0; i < istate->cache_nr; i++) {
 			if (istate->cache[i]->ce_flags & CE_FSMONITOR_VALID) {
 				is_cache_changed = 1;
@@ -330,7 +350,10 @@ void refresh_fsmonitor(struct index_state *istate)
 			}
 		}
 
-		/* If we're going to check every file, ensure we save the results */
+		/*
+		 * If we're going to check every file, ensure we save
+		 * the results.
+		 */
 		if (is_cache_changed)
 			istate->cache_changed |= FSMONITOR_CHANGED;
 
-- 
gitgitgadget

[PATCH v2 2/5] fsmonitor-ipc: create client routines for git-fsmonitor--daemon

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Create fsmonitor_ipc__*() client routines to spawn the built-in file
system monitor daemon and send it an IPC request using the `Simple
IPC` API.

Stub in empty fsmonitor_ipc__*() functions for unsupported platforms.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 Makefile        |   1 +
 fsmonitor-ipc.c | 176 ++++++++++++++++++++++++++++++++++++++++++++++++
 fsmonitor-ipc.h |  48 +++++++++++++
 3 files changed, 225 insertions(+)
 create mode 100644 fsmonitor-ipc.c
 create mode 100644 fsmonitor-ipc.h

diff --git a/Makefile b/Makefile
index 381bed2c1d2..d51fd8b33ce 100644
--- a/Makefile
+++ b/Makefile
@@ -897,6 +897,7 @@ LIB_OBJS += fetch-pack.o
 LIB_OBJS += fmt-merge-msg.o
 LIB_OBJS += fsck.o
 LIB_OBJS += fsmonitor.o
+LIB_OBJS += fsmonitor-ipc.o
 LIB_OBJS += gettext.o
 LIB_OBJS += gpg-interface.o
 LIB_OBJS += graph.o
diff --git a/fsmonitor-ipc.c b/fsmonitor-ipc.c
new file mode 100644
index 00000000000..ccc32d2a17e
--- /dev/null
+++ b/fsmonitor-ipc.c
@@ -0,0 +1,176 @@
+#include "cache.h"
+#include "fsmonitor.h"
+#include "simple-ipc.h"
+#include "fsmonitor-ipc.h"
+#include "run-command.h"
+#include "strbuf.h"
+#include "trace2.h"
+
+#ifdef HAVE_FSMONITOR_DAEMON_BACKEND
+
+int fsmonitor_ipc__is_supported(void)
+{
+	return 1;
+}
+
+GIT_PATH_FUNC(fsmonitor_ipc__get_path, "fsmonitor--daemon.ipc")
+
+enum ipc_active_state fsmonitor_ipc__get_state(void)
+{
+	return ipc_get_active_state(fsmonitor_ipc__get_path());
+}
+
+static int spawn_daemon(void)
+{
+	const char *args[] = { "fsmonitor--daemon", "start", NULL };
+
+	return run_command_v_opt_tr2(args, RUN_COMMAND_NO_STDIN | RUN_GIT_CMD,
+				    "fsmonitor");
+}
+
+int fsmonitor_ipc__send_query(const char *since_token,
+			      struct strbuf *answer)
+{
+	int ret = -1;
+	int tried_to_spawn = 0;
+	enum ipc_active_state state = IPC_STATE__OTHER_ERROR;
+	struct ipc_client_connection *connection = NULL;
+	struct ipc_client_connect_options options
+		= IPC_CLIENT_CONNECT_OPTIONS_INIT;
+	const char *tok = since_token ? since_token : "";
+	size_t tok_len = since_token ? strlen(since_token) : 0;
+
+	options.wait_if_busy = 1;
+	options.wait_if_not_found = 0;
+
+	trace2_region_enter("fsm_client", "query", NULL);
+	trace2_data_string("fsm_client", NULL, "query/command", tok);
+
+try_again:
+	state = ipc_client_try_connect(fsmonitor_ipc__get_path(), &options,
+				       &connection);
+
+	switch (state) {
+	case IPC_STATE__LISTENING:
+		ret = ipc_client_send_command_to_connection(
+			connection, tok, tok_len, answer);
+		ipc_client_close_connection(connection);
+
+		trace2_data_intmax("fsm_client", NULL,
+				   "query/response-length", answer->len);
+
+		if (fsmonitor_is_trivial_response(answer))
+			trace2_data_intmax("fsm_client", NULL,
+					   "query/trivial-response", 1);
+
+		goto done;
+
+	case IPC_STATE__NOT_LISTENING:
+	case IPC_STATE__PATH_NOT_FOUND:
+		if (tried_to_spawn)
+			goto done;
+
+		tried_to_spawn++;
+		if (spawn_daemon())
+			goto done;
+
+		/*
+		 * Try again, but this time give the daemon a chance to
+		 * actually create the pipe/socket.
+		 *
+		 * Granted, the daemon just started so it can't possibly have
+		 * any FS cached yet, so we'll always get a trivial answer.
+		 * BUT the answer should include a new token that can serve
+		 * as the basis for subsequent requests.
+		 */
+		options.wait_if_not_found = 1;
+		goto try_again;
+
+	case IPC_STATE__INVALID_PATH:
+		ret = error(_("fsmonitor_ipc__send_query: invalid path '%s'"),
+			    fsmonitor_ipc__get_path());
+		goto done;
+
+	case IPC_STATE__OTHER_ERROR:
+	default:
+		ret = error(_("fsmonitor_ipc__send_query: unspecified error on '%s'"),
+			    fsmonitor_ipc__get_path());
+		goto done;
+	}
+
+done:
+	trace2_region_leave("fsm_client", "query", NULL);
+
+	return ret;
+}
+
+int fsmonitor_ipc__send_command(const char *command,
+				struct strbuf *answer)
+{
+	struct ipc_client_connection *connection = NULL;
+	struct ipc_client_connect_options options
+		= IPC_CLIENT_CONNECT_OPTIONS_INIT;
+	int ret;
+	enum ipc_active_state state;
+	const char *c = command ? command : "";
+	size_t c_len = command ? strlen(command) : 0;
+
+	strbuf_reset(answer);
+
+	options.wait_if_busy = 1;
+	options.wait_if_not_found = 0;
+
+	state = ipc_client_try_connect(fsmonitor_ipc__get_path(), &options,
+				       &connection);
+	if (state != IPC_STATE__LISTENING) {
+		die("fsmonitor--daemon is not running");
+		return -1;
+	}
+
+	ret = ipc_client_send_command_to_connection(connection, c, c_len,
+						    answer);
+	ipc_client_close_connection(connection);
+
+	if (ret == -1) {
+		die("could not send '%s' command to fsmonitor--daemon", c);
+		return -1;
+	}
+
+	return 0;
+}
+
+#else
+
+/*
+ * A trivial implementation of the fsmonitor_ipc__ API for unsupported
+ * platforms.
+ */
+
+int fsmonitor_ipc__is_supported(void)
+{
+	return 0;
+}
+
+const char *fsmonitor_ipc__get_path(void)
+{
+	return NULL;
+}
+
+enum ipc_active_state fsmonitor_ipc__get_state(void)
+{
+	return IPC_STATE__OTHER_ERROR;
+}
+
+int fsmonitor_ipc__send_query(const char *since_token,
+			      struct strbuf *answer)
+{
+	return -1;
+}
+
+int fsmonitor_ipc__send_command(const char *command,
+				struct strbuf *answer)
+{
+	return -1;
+}
+
+#endif
diff --git a/fsmonitor-ipc.h b/fsmonitor-ipc.h
new file mode 100644
index 00000000000..b6a7067c3af
--- /dev/null
+++ b/fsmonitor-ipc.h
@@ -0,0 +1,48 @@
+#ifndef FSMONITOR_IPC_H
+#define FSMONITOR_IPC_H
+
+#include "simple-ipc.h"
+
+/*
+ * Returns true if built-in file system monitor daemon is defined
+ * for this platform.
+ */
+int fsmonitor_ipc__is_supported(void);
+
+/*
+ * Returns the pathname to the IPC named pipe or Unix domain socket
+ * where a `git-fsmonitor--daemon` process will listen.  This is a
+ * per-worktree value.
+ *
+ * Returns NULL if the daemon is not supported on this platform.
+ */
+const char *fsmonitor_ipc__get_path(void);
+
+/*
+ * Try to determine whether there is a `git-fsmonitor--daemon` process
+ * listening on the IPC pipe/socket.
+ */
+enum ipc_active_state fsmonitor_ipc__get_state(void);
+
+/*
+ * Connect to a `git-fsmonitor--daemon` process via simple-ipc
+ * and ask for the set of changed files since the given token.
+ *
+ * Spawn a daemon process in the background if necessary.
+ *
+ * Returns -1 on error; 0 on success.
+ */
+int fsmonitor_ipc__send_query(const char *since_token,
+			      struct strbuf *answer);
+
+/*
+ * Connect to a `git-fsmonitor--daemon` process via simple-ipc and
+ * send a command verb.  If no daemon is available, we DO NOT try to
+ * start one.
+ *
+ * Returns -1 on error; 0 on success.
+ */
+int fsmonitor_ipc__send_command(const char *command,
+				struct strbuf *answer);
+
+#endif /* FSMONITOR_IPC_H */
-- 
gitgitgadget

[PATCH v2 3/5] fsmonitor: config settings are repository-specific

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Move FSMonitor config settings to a new `struct fsmonitor_settings`
structure.  Add a lazily-loaded pointer to `struct repo_settings`.
Create `fsm_settings__get_*()` getters to lazily look up fsmonitor-
related config settings.

Get rid of the `core_fsmonitor` global variable, and add support for
the new `core.useBuiltinFSMonitor` config setting.  Move config code
to lookup the existing `core.fsmonitor` value to `fsmonitor-settings.[ch]`.

The `core_fsmonitor` global variable was used to store the pathname to
the FSMonitor hook and it was used as a boolean to see if FSMonitor
was enabled.  This dual usage will lead to confusion when we add
support for a builtin FSMonitor based on IPC, since the builtin
FSMonitor doesn't need the hook pathname.

Replace the boolean usage with an `enum fsmonitor_mode` to represent
the state of FSMonitor.  And only set the pathname when in HOOK mode.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 Makefile               |  1 +
 builtin/update-index.c | 19 +++++++--
 cache.h                |  1 -
 config.c               | 14 ------
 config.h               |  1 -
 environment.c          |  1 -
 fsmonitor-settings.c   | 97 ++++++++++++++++++++++++++++++++++++++++++
 fsmonitor-settings.h   | 21 +++++++++
 fsmonitor.c            | 63 ++++++++++++++++-----------
 fsmonitor.h            | 18 ++++++--
 repo-settings.c        |  2 +
 repository.h           |  3 ++
 t/README               |  4 +-
 13 files changed, 194 insertions(+), 51 deletions(-)
 create mode 100644 fsmonitor-settings.c
 create mode 100644 fsmonitor-settings.h

diff --git a/Makefile b/Makefile
index d51fd8b33ce..29ed7c4aba6 100644
--- a/Makefile
+++ b/Makefile
@@ -898,6 +898,7 @@ LIB_OBJS += fmt-merge-msg.o
 LIB_OBJS += fsck.o
 LIB_OBJS += fsmonitor.o
 LIB_OBJS += fsmonitor-ipc.o
+LIB_OBJS += fsmonitor-settings.o
 LIB_OBJS += gettext.o
 LIB_OBJS += gpg-interface.o
 LIB_OBJS += graph.o
diff --git a/builtin/update-index.c b/builtin/update-index.c
index 187203e8bb5..79db3ff37e2 100644
--- a/builtin/update-index.c
+++ b/builtin/update-index.c
@@ -1214,14 +1214,25 @@ int cmd_update_index(int argc, const char **argv, const char *prefix)
 	}
 
 	if (fsmonitor > 0) {
-		if (git_config_get_fsmonitor() == 0)
+		enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
+
+		if (fsm_mode == FSMONITOR_MODE_DISABLED) {
+			warning(_("core.useBuiltinFSMonitor is unset; "
+				"set it if you really want to enable the "
+				"builtin fsmonitor"));
 			warning(_("core.fsmonitor is unset; "
-				"set it if you really want to "
-				"enable fsmonitor"));
+				"set it if you really want to enable the "
+				"hook-based fsmonitor"));
+		}
 		add_fsmonitor(&the_index);
 		report(_("fsmonitor enabled"));
 	} else if (!fsmonitor) {
-		if (git_config_get_fsmonitor() == 1)
+		enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
+		if (fsm_mode == FSMONITOR_MODE_IPC)
+			warning(_("core.useBuiltinFSMonitor is set; "
+				"remove it if you really want to "
+				"disable fsmonitor"));
+		if (fsm_mode == FSMONITOR_MODE_HOOK)
 			warning(_("core.fsmonitor is set; "
 				"remove it if you really want to "
 				"disable fsmonitor"));
diff --git a/cache.h b/cache.h
index d092820c943..8f4e3c8bd1d 100644
--- a/cache.h
+++ b/cache.h
@@ -989,7 +989,6 @@ extern int core_preload_index;
 extern int precomposed_unicode;
 extern int protect_hfs;
 extern int protect_ntfs;
-extern const char *core_fsmonitor;
 
 extern int core_apply_sparse_checkout;
 extern int core_sparse_checkout_cone;
diff --git a/config.c b/config.c
index 2dcbe901b6b..6b6e9cacac3 100644
--- a/config.c
+++ b/config.c
@@ -2502,20 +2502,6 @@ int git_config_get_max_percent_split_change(void)
 	return -1; /* default value */
 }
 
-int git_config_get_fsmonitor(void)
-{
-	if (git_config_get_pathname("core.fsmonitor", &core_fsmonitor))
-		core_fsmonitor = getenv("GIT_TEST_FSMONITOR");
-
-	if (core_fsmonitor && !*core_fsmonitor)
-		core_fsmonitor = NULL;
-
-	if (core_fsmonitor)
-		return 1;
-
-	return 0;
-}
-
 int git_config_get_index_threads(int *dest)
 {
 	int is_bool, val;
diff --git a/config.h b/config.h
index f119de01309..69d733824a0 100644
--- a/config.h
+++ b/config.h
@@ -610,7 +610,6 @@ int git_config_get_pathname(const char *key, const char **dest);
 int git_config_get_index_threads(int *dest);
 int git_config_get_split_index(void);
 int git_config_get_max_percent_split_change(void);
-int git_config_get_fsmonitor(void);
 
 /* This dies if the configured or default date is in the future */
 int git_config_get_expiry(const char *key, const char **output);
diff --git a/environment.c b/environment.c
index 9da7f3c1a19..68f90632245 100644
--- a/environment.c
+++ b/environment.c
@@ -82,7 +82,6 @@ int protect_hfs = PROTECT_HFS_DEFAULT;
 #define PROTECT_NTFS_DEFAULT 1
 #endif
 int protect_ntfs = PROTECT_NTFS_DEFAULT;
-const char *core_fsmonitor;
 
 /*
  * The character that begins a commented line in user-editable file
diff --git a/fsmonitor-settings.c b/fsmonitor-settings.c
new file mode 100644
index 00000000000..2770266f5ee
--- /dev/null
+++ b/fsmonitor-settings.c
@@ -0,0 +1,97 @@
+#include "cache.h"
+#include "config.h"
+#include "repository.h"
+#include "fsmonitor-settings.h"
+
+/*
+ * We keep this structure defintion private and have getters
+ * for all fields so that we can lazy load it as needed.
+ */
+struct fsmonitor_settings {
+	enum fsmonitor_mode mode;
+	char *hook_path;
+};
+
+void fsm_settings__set_ipc(struct repository *r)
+{
+	struct fsmonitor_settings *s = r->settings.fsmonitor;
+
+	s->mode = FSMONITOR_MODE_IPC;
+}
+
+void fsm_settings__set_hook(struct repository *r, const char *path)
+{
+	struct fsmonitor_settings *s = r->settings.fsmonitor;
+
+	s->mode = FSMONITOR_MODE_HOOK;
+	s->hook_path = strdup(path);
+}
+
+void fsm_settings__set_disabled(struct repository *r)
+{
+	struct fsmonitor_settings *s = r->settings.fsmonitor;
+
+	s->mode = FSMONITOR_MODE_DISABLED;
+	FREE_AND_NULL(s->hook_path);
+}
+
+static int check_for_ipc(struct repository *r)
+{
+	int value;
+
+	if (!repo_config_get_bool(r, "core.usebuiltinfsmonitor", &value) &&
+	    value) {
+		fsm_settings__set_ipc(r);
+		return 1;
+	}
+
+	return 0;
+}
+
+static int check_for_hook(struct repository *r)
+{
+	const char *const_str;
+
+	if (repo_config_get_pathname(r, "core.fsmonitor", &const_str))
+		const_str = getenv("GIT_TEST_FSMONITOR");
+
+	if (const_str && *const_str) {
+		fsm_settings__set_hook(r, const_str);
+		return 1;
+	}
+
+	return 0;
+}
+
+static void lookup_fsmonitor_settings(struct repository *r)
+{
+	struct fsmonitor_settings *s;
+
+	CALLOC_ARRAY(s, 1);
+
+	r->settings.fsmonitor = s;
+
+	if (check_for_ipc(r))
+		return;
+
+	if (check_for_hook(r))
+		return;
+
+	fsm_settings__set_disabled(r);
+}
+
+enum fsmonitor_mode fsm_settings__get_mode(struct repository *r)
+{
+	if (!r->settings.fsmonitor)
+		lookup_fsmonitor_settings(r);
+
+	return r->settings.fsmonitor->mode;
+}
+
+const char *fsm_settings__get_hook_path(struct repository *r)
+{
+	if (!r->settings.fsmonitor)
+		lookup_fsmonitor_settings(r);
+
+	return r->settings.fsmonitor->hook_path;
+}
diff --git a/fsmonitor-settings.h b/fsmonitor-settings.h
new file mode 100644
index 00000000000..50b29234616
--- /dev/null
+++ b/fsmonitor-settings.h
@@ -0,0 +1,21 @@
+#ifndef FSMONITOR_SETTINGS_H
+#define FSMONITOR_SETTINGS_H
+
+struct repository;
+
+enum fsmonitor_mode {
+	FSMONITOR_MODE_DISABLED = 0,
+	FSMONITOR_MODE_HOOK = 1, /* core.fsmonitor */
+	FSMONITOR_MODE_IPC = 2,  /* core.useBuiltinFSMonitor */
+};
+
+void fsm_settings__set_ipc(struct repository *r);
+void fsm_settings__set_hook(struct repository *r, const char *path);
+void fsm_settings__set_disabled(struct repository *r);
+
+enum fsmonitor_mode fsm_settings__get_mode(struct repository *r);
+const char *fsm_settings__get_hook_path(struct repository *r);
+
+struct fsmonitor_settings;
+
+#endif /* FSMONITOR_SETTINGS_H */
diff --git a/fsmonitor.c b/fsmonitor.c
index ec4c46407c5..63174630c0e 100644
--- a/fsmonitor.c
+++ b/fsmonitor.c
@@ -3,6 +3,7 @@
 #include "dir.h"
 #include "ewah/ewok.h"
 #include "fsmonitor.h"
+#include "fsmonitor-ipc.h"
 #include "run-command.h"
 #include "strbuf.h"
 
@@ -148,15 +149,18 @@ void write_fsmonitor_extension(struct strbuf *sb, struct index_state *istate)
 /*
  * Call the query-fsmonitor hook passing the last update token of the saved results.
  */
-static int query_fsmonitor(int version, const char *last_update, struct strbuf *query_result)
+static int query_fsmonitor_hook(struct repository *r,
+				int version,
+				const char *last_update,
+				struct strbuf *query_result)
 {
 	struct child_process cp = CHILD_PROCESS_INIT;
 	int result;
 
-	if (!core_fsmonitor)
+	if (fsm_settings__get_mode(r) != FSMONITOR_MODE_HOOK)
 		return -1;
 
-	strvec_push(&cp.args, core_fsmonitor);
+	strvec_push(&cp.args, fsm_settings__get_hook_path(r));
 	strvec_pushf(&cp.args, "%d", version);
 	strvec_pushf(&cp.args, "%s", last_update);
 	cp.use_shell = 1;
@@ -238,17 +242,28 @@ void refresh_fsmonitor(struct index_state *istate)
 	struct strbuf last_update_token = STRBUF_INIT;
 	char *buf;
 	unsigned int i;
+	struct repository *r = istate->repo ? istate->repo : the_repository;
+	enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
 
-	if (!core_fsmonitor || istate->fsmonitor_has_run_once)
+	if (fsm_mode <= FSMONITOR_MODE_DISABLED ||
+	    istate->fsmonitor_has_run_once)
 		return;
 
-	hook_version = fsmonitor_hook_version();
-
 	istate->fsmonitor_has_run_once = 1;
 
 	trace_printf_key(&trace_fsmonitor, "refresh fsmonitor");
+
+	if (fsm_mode == FSMONITOR_MODE_IPC) {
+		/* TODO */
+		return;
+	}
+
+	assert(fsm_mode == FSMONITOR_MODE_HOOK);
+
+	hook_version = fsmonitor_hook_version();
+
 	/*
-	 * This could be racy so save the date/time now and query_fsmonitor
+	 * This could be racy so save the date/time now and query_fsmonitor_hook
 	 * should be inclusive to ensure we don't miss potential changes.
 	 */
 	last_update = getnanotime();
@@ -256,13 +271,14 @@ void refresh_fsmonitor(struct index_state *istate)
 		strbuf_addf(&last_update_token, "%"PRIu64"", last_update);
 
 	/*
-	 * If we have a last update token, call query_fsmonitor for the set of
+	 * If we have a last update token, call query_fsmonitor_hook for the set of
 	 * changes since that token, else assume everything is possibly dirty
 	 * and check it all.
 	 */
 	if (istate->fsmonitor_last_update) {
 		if (hook_version == -1 || hook_version == HOOK_INTERFACE_VERSION2) {
-			query_success = !query_fsmonitor(HOOK_INTERFACE_VERSION2,
+			query_success = !query_fsmonitor_hook(
+				r, HOOK_INTERFACE_VERSION2,
 				istate->fsmonitor_last_update, &query_result);
 
 			if (query_success) {
@@ -292,13 +308,17 @@ void refresh_fsmonitor(struct index_state *istate)
 		}
 
 		if (hook_version == HOOK_INTERFACE_VERSION1) {
-			query_success = !query_fsmonitor(HOOK_INTERFACE_VERSION1,
+			query_success = !query_fsmonitor_hook(
+				r, HOOK_INTERFACE_VERSION1,
 				istate->fsmonitor_last_update, &query_result);
 		}
 
-		trace_performance_since(last_update, "fsmonitor process '%s'", core_fsmonitor);
-		trace_printf_key(&trace_fsmonitor, "fsmonitor process '%s' returned %s",
-			core_fsmonitor, query_success ? "success" : "failure");
+		trace_performance_since(last_update, "fsmonitor process '%s'",
+					fsm_settings__get_hook_path(r));
+		trace_printf_key(&trace_fsmonitor,
+				 "fsmonitor process '%s' returned %s",
+				 fsm_settings__get_hook_path(r),
+				 query_success ? "success" : "failure");
 	}
 
 	/*
@@ -434,7 +454,8 @@ void remove_fsmonitor(struct index_state *istate)
 void tweak_fsmonitor(struct index_state *istate)
 {
 	unsigned int i;
-	int fsmonitor_enabled = git_config_get_fsmonitor();
+	struct repository *r = istate->repo ? istate->repo : the_repository;
+	int fsmonitor_enabled = (fsm_settings__get_mode(r) > FSMONITOR_MODE_DISABLED);
 
 	if (istate->fsmonitor_dirty) {
 		if (fsmonitor_enabled) {
@@ -454,16 +475,8 @@ void tweak_fsmonitor(struct index_state *istate)
 		istate->fsmonitor_dirty = NULL;
 	}
 
-	switch (fsmonitor_enabled) {
-	case -1: /* keep: do nothing */
-		break;
-	case 0: /* false */
-		remove_fsmonitor(istate);
-		break;
-	case 1: /* true */
+	if (fsmonitor_enabled)
 		add_fsmonitor(istate);
-		break;
-	default: /* unknown value: do nothing */
-		break;
-	}
+	else
+		remove_fsmonitor(istate);
 }
diff --git a/fsmonitor.h b/fsmonitor.h
index f20d72631d7..f9201411aa7 100644
--- a/fsmonitor.h
+++ b/fsmonitor.h
@@ -3,6 +3,7 @@
 
 #include "cache.h"
 #include "dir.h"
+#include "fsmonitor-settings.h"
 
 extern struct trace_key trace_fsmonitor;
 
@@ -57,7 +58,11 @@ int fsmonitor_is_trivial_response(const struct strbuf *query_result);
  */
 static inline int is_fsmonitor_refreshed(const struct index_state *istate)
 {
-	return !core_fsmonitor || istate->fsmonitor_has_run_once;
+	struct repository *r = istate->repo ? istate->repo : the_repository;
+	enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
+
+	return fsm_mode <= FSMONITOR_MODE_DISABLED ||
+		istate->fsmonitor_has_run_once;
 }
 
 /*
@@ -67,7 +72,11 @@ static inline int is_fsmonitor_refreshed(const struct index_state *istate)
  */
 static inline void mark_fsmonitor_valid(struct index_state *istate, struct cache_entry *ce)
 {
-	if (core_fsmonitor && !(ce->ce_flags & CE_FSMONITOR_VALID)) {
+	struct repository *r = istate->repo ? istate->repo : the_repository;
+	enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
+
+	if (fsm_mode > FSMONITOR_MODE_DISABLED &&
+	    !(ce->ce_flags & CE_FSMONITOR_VALID)) {
 		istate->cache_changed = 1;
 		ce->ce_flags |= CE_FSMONITOR_VALID;
 		trace_printf_key(&trace_fsmonitor, "mark_fsmonitor_clean '%s'", ce->name);
@@ -83,7 +92,10 @@ static inline void mark_fsmonitor_valid(struct index_state *istate, struct cache
  */
 static inline void mark_fsmonitor_invalid(struct index_state *istate, struct cache_entry *ce)
 {
-	if (core_fsmonitor) {
+	struct repository *r = istate->repo ? istate->repo : the_repository;
+	enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
+
+	if (fsm_mode > FSMONITOR_MODE_DISABLED) {
 		ce->ce_flags &= ~CE_FSMONITOR_VALID;
 		untracked_cache_invalidate_path(istate, ce->name, 1);
 		trace_printf_key(&trace_fsmonitor, "mark_fsmonitor_invalid '%s'", ce->name);
diff --git a/repo-settings.c b/repo-settings.c
index b93e91a212e..0ca85f54bb8 100644
--- a/repo-settings.c
+++ b/repo-settings.c
@@ -20,6 +20,8 @@ void prepare_repo_settings(struct repository *r)
 	if (r->settings.initialized++)
 		return;
 
+	r->settings.fsmonitor = NULL; /* lazy loaded */
+
 	/* Defaults */
 	r->settings.index_version = -1;
 	r->settings.core_untracked_cache = UNTRACKED_CACHE_KEEP;
diff --git a/repository.h b/repository.h
index a057653981c..89a1873ade7 100644
--- a/repository.h
+++ b/repository.h
@@ -4,6 +4,7 @@
 #include "path.h"
 
 struct config_set;
+struct fsmonitor_settings;
 struct git_hash_algo;
 struct index_state;
 struct lock_file;
@@ -34,6 +35,8 @@ struct repo_settings {
 	int command_requires_full_index;
 	int sparse_index;
 
+	struct fsmonitor_settings *fsmonitor; /* lazy loaded */
+
 	int index_version;
 	enum untracked_cache_setting core_untracked_cache;
 
diff --git a/t/README b/t/README
index b92155a822e..6dc4a1d10cf 100644
--- a/t/README
+++ b/t/README
@@ -405,8 +405,8 @@ every 'git commit-graph write', as if the `--changed-paths` option was
 passed in.
 
 GIT_TEST_FSMONITOR=$PWD/t7519/fsmonitor-all exercises the fsmonitor
-code path for utilizing a file system monitor to speed up detecting
-new or changed files.
+code path for utilizing a (hook based) file system monitor to speed up
+detecting new or changed files.
 
 GIT_TEST_INDEX_VERSION=<n> exercises the index read/write code path
 for the index version specified.  Can be set to any valid version
-- 
gitgitgadget

[PATCH v2 4/5] fsmonitor: use IPC to query the builtin FSMonitor daemon

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Use simple IPC to directly communicate with the new builtin file
system monitor daemon when `core.useBuiltinFSMonitor` is set.

The `core.fsmonitor` setting has already been defined as a HOOK
pathname.  Historically, this has been set to a HOOK script that will
talk with Watchman.  For compatibility reasons, we do not want to
overload that definition (and cause problems if users have multiple
versions of Git installed).

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 fsmonitor.c | 34 ++++++++++++++++++++++++++++++++--
 1 file changed, 32 insertions(+), 2 deletions(-)

diff --git a/fsmonitor.c b/fsmonitor.c
index 63174630c0e..695fb0ce4e7 100644
--- a/fsmonitor.c
+++ b/fsmonitor.c
@@ -254,8 +254,37 @@ void refresh_fsmonitor(struct index_state *istate)
 	trace_printf_key(&trace_fsmonitor, "refresh fsmonitor");
 
 	if (fsm_mode == FSMONITOR_MODE_IPC) {
-		/* TODO */
-		return;
+		query_success = !fsmonitor_ipc__send_query(
+			istate->fsmonitor_last_update ?
+			istate->fsmonitor_last_update : "builtin:fake",
+			&query_result);
+		if (query_success) {
+			/*
+			 * The response contains a series of nul terminated
+			 * strings.  The first is the new token.
+			 *
+			 * Use `char *buf` as an interlude to trick the CI
+			 * static analysis to let us use `strbuf_addstr()`
+			 * here (and only copy the token) rather than
+			 * `strbuf_addbuf()`.
+			 */
+			buf = query_result.buf;
+			strbuf_addstr(&last_update_token, buf);
+			bol = last_update_token.len + 1;
+		} else {
+			/*
+			 * The builtin daemon is not available on this
+			 * platform -OR- we failed to get a response.
+			 *
+			 * Generate a fake token (rather than a V1
+			 * timestamp) for the index extension.  (If
+			 * they switch back to the hook API, we don't
+			 * want ambiguous state.)
+			 */
+			strbuf_addstr(&last_update_token, "builtin:fake");
+		}
+
+		goto apply_results;
 	}
 
 	assert(fsm_mode == FSMONITOR_MODE_HOOK);
@@ -321,6 +350,7 @@ void refresh_fsmonitor(struct index_state *istate)
 				 query_success ? "success" : "failure");
 	}
 
+apply_results:
 	/*
 	 * The response from FSMonitor (excluding the header token) is
 	 * either:
-- 
gitgitgadget

[PATCH v2 5/5] fsmonitor: update fsmonitor config documentation

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Update references to `core.fsmonitor` and `core.fsmonitorHookVersion` and
pointers to `Watchman` to mention the new `core.useBuiltinFSMonitor`
value.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 Documentation/config/core.txt      | 56 ++++++++++++++++++++++--------
 Documentation/git-update-index.txt | 27 +++++++-------
 Documentation/githooks.txt         |  3 +-
 3 files changed, 59 insertions(+), 27 deletions(-)

diff --git a/Documentation/config/core.txt b/Documentation/config/core.txt
index c04f62a54a1..4f6e519bc02 100644
--- a/Documentation/config/core.txt
+++ b/Documentation/config/core.txt
@@ -62,22 +62,50 @@ core.protectNTFS::
 	Defaults to `true` on Windows, and `false` elsewhere.
 
 core.fsmonitor::
-	If set, the value of this variable is used as a command which
-	will identify all files that may have changed since the
-	requested date/time. This information is used to speed up git by
-	avoiding unnecessary processing of files that have not changed.
-	See the "fsmonitor-watchman" section of linkgit:githooks[5].
+	If set, this variable contains the pathname of the "fsmonitor"
+	hook command.
++
+This hook command is used to identify all files that may have changed
+since the requested date/time. This information is used to speed up
+git by avoiding unnecessary scanning of files that have not changed.
++
+See the "fsmonitor-watchman" section of linkgit:githooks[5].
++
+Note: The value of this config setting is ignored if the
+built-in file system monitor is enabled (see `core.useBuiltinFSMonitor`).
 
 core.fsmonitorHookVersion::
-	Sets the version of hook that is to be used when calling fsmonitor.
-	There are currently versions 1 and 2. When this is not set,
-	version 2 will be tried first and if it fails then version 1
-	will be tried. Version 1 uses a timestamp as input to determine
-	which files have changes since that time but some monitors
-	like watchman have race conditions when used with a timestamp.
-	Version 2 uses an opaque string so that the monitor can return
-	something that can be used to determine what files have changed
-	without race conditions.
+	Sets the protocol version to be used when invoking the
+	"fsmonitor" hook.
++
+There are currently versions 1 and 2. When this is not set,
+version 2 will be tried first and if it fails then version 1
+will be tried. Version 1 uses a timestamp as input to determine
+which files have changes since that time but some monitors
+like Watchman have race conditions when used with a timestamp.
+Version 2 uses an opaque string so that the monitor can return
+something that can be used to determine what files have changed
+without race conditions.
++
+Note: The value of this config setting is ignored if the
+built-in file system monitor is enabled (see `core.useBuiltinFSMonitor`).
+
+core.useBuiltinFSMonitor::
+	If set to true, enable the built-in file system monitor
+	daemon for this working directory (linkgit:git-fsmonitor--daemon[1]).
++
+Like hook-based file system monitors, the built-in file system monitor
+can speed up Git commands that need to refresh the Git index
+(e.g. `git status`) in a working directory with many files.  The
+built-in monitor eliminates the need to install and maintain an
+external third-party tool.
++
+The built-in file system monitor is currently available only on a
+limited set of supported platforms.  Currently, this includes Windows
+and MacOS.
++
+Note: if this config setting is set to `true`, the values of
+`core.fsmonitor` and `core.fsmonitorHookVersion` are ignored.
 
 core.trustctime::
 	If false, the ctime differences between the index and the
diff --git a/Documentation/git-update-index.txt b/Documentation/git-update-index.txt
index 2853f168d97..c7c31b3fcf9 100644
--- a/Documentation/git-update-index.txt
+++ b/Documentation/git-update-index.txt
@@ -498,7 +498,9 @@ FILE SYSTEM MONITOR
 This feature is intended to speed up git operations for repos that have
 large working directories.
 
-It enables git to work together with a file system monitor (see the
+It enables git to work together with a file system monitor (see
+linkgit:git-fsmonitor--daemon[1]
+and the
 "fsmonitor-watchman" section of linkgit:githooks[5]) that can
 inform it as to what files have been modified. This enables git to avoid
 having to lstat() every file to find modified files.
@@ -508,17 +510,18 @@ performance by avoiding the cost of scanning the entire working directory
 looking for new files.
 
 If you want to enable (or disable) this feature, it is easier to use
-the `core.fsmonitor` configuration variable (see
-linkgit:git-config[1]) than using the `--fsmonitor` option to
-`git update-index` in each repository, especially if you want to do so
-across all repositories you use, because you can set the configuration
-variable in your `$HOME/.gitconfig` just once and have it affect all
-repositories you touch.
-
-When the `core.fsmonitor` configuration variable is changed, the
-file system monitor is added to or removed from the index the next time
-a command reads the index. When `--[no-]fsmonitor` are used, the file
-system monitor is immediately added to or removed from the index.
+the `core.fsmonitor` or `core.useBuiltinFSMonitor` configuration
+variable (see linkgit:git-config[1]) than using the `--fsmonitor`
+option to `git update-index` in each repository, especially if you
+want to do so across all repositories you use, because you can set the
+configuration variable in your `$HOME/.gitconfig` just once and have
+it affect all repositories you touch.
+
+When the `core.fsmonitor` or `core.useBuiltinFSMonitor` configuration
+variable is changed, the file system monitor is added to or removed
+from the index the next time a command reads the index. When
+`--[no-]fsmonitor` are used, the file system monitor is immediately
+added to or removed from the index.
 
 CONFIGURATION
 -------------
diff --git a/Documentation/githooks.txt b/Documentation/githooks.txt
index b51959ff941..b7d5e926f7b 100644
--- a/Documentation/githooks.txt
+++ b/Documentation/githooks.txt
@@ -593,7 +593,8 @@ fsmonitor-watchman
 
 This hook is invoked when the configuration option `core.fsmonitor` is
 set to `.git/hooks/fsmonitor-watchman` or `.git/hooks/fsmonitor-watchmanv2`
-depending on the version of the hook to use.
+depending on the version of the hook to use, unless overridden via
+`core.useBuiltinFSMonitor` (see linkgit:git-config[1]).
 
 Version 1 takes two arguments, a version (1) and the time in elapsed
 nanoseconds since midnight, January 1, 1970.
-- 
gitgitgadget

Re: [PATCH v2 3/5] fsmonitor: config settings are repository-specific

[Ævar Arnfjörð Bjarmason]

On Thu, Oct 07 2021, Jeff Hostetler via GitGitGadget wrote:

Good to see this move forward!

This bit:

> --- a/repo-settings.c
> +++ b/repo-settings.c
> @@ -20,6 +20,8 @@ void prepare_repo_settings(struct repository *r)
>  	if (r->settings.initialized++)
>  		return;
>  
> +	r->settings.fsmonitor = NULL; /* lazy loaded */
> +
>  	/* Defaults */
>  	r->settings.index_version = -1;
>  	r->settings.core_untracked_cache = UNTRACKED_CACHE_KEEP;
> diff --git a/repository.h b/repository.h

Is carried forward from v1, but with 3050b6dfc75 (repo-settings.c:
simplify the setup, 2021-09-21) isn't needed. It's init'd to 0/NULL
already, but was -1 before.

So this hunk can go, and its presence makes for confusing reading
without that history, is it set before somehow? No, just working around
older code that's no longer there.

But also: For untracked_cache_setting and fetch_negotiation_setting
we've got an embedded enum in the struct, but this...

> index a057653981c..89a1873ade7 100644
> --- a/repository.h
> +++ b/repository.h
> @@ -4,6 +4,7 @@
>  #include "path.h"
>  
>  struct config_set;
> +struct fsmonitor_settings;
>  struct git_hash_algo;
>  struct index_state;
>  struct lock_file;
> @@ -34,6 +35,8 @@ struct repo_settings {
>  	int command_requires_full_index;
>  	int sparse_index;
>  
> +	struct fsmonitor_settings *fsmonitor; /* lazy loaded */
> +
>  	int index_version;
>  	enum untracked_cache_setting core_untracked_cache;
>  

Is a pointer to a struct that has an "enum fsmonitor_mode mode", and the
code in fsmonitor-settings.c seems to be repeating the patterns we had
in repo-settings.c pre-3050b6dfc75, e.g. checking whether a bool config
variable exists *and* is true, v.s. checking if it exists (presumably an
explicit false wants to override something).

I haven't looked carefully, but between that & the "char *hook_path"
being something that'll need to be made to use Emily's hook config
series sooner than later, can't we read/setup the initial config in
"repo_cfg_bool"?

The relevant commit message just says:

    Move FSMonitor config settings to a new `struct fsmonitor_settings`
    structure.  Add a lazily-loaded pointer to `struct repo_settings`.
    Create `fsm_settings__get_*()` getters to lazily look up fsmonitor-
    related config settings.[...]

Which I think can be paraphrased as "Add scaffolding to repo-settings.c
but do config loading differently than everything there (lazily),
because...", except the "because" is missing :)

Re: [PATCH v2 3/5] fsmonitor: config settings are repository-specific

[Jeff Hostetler]

On 10/7/21 12:59 PM, Ævar Arnfjörð Bjarmason wrote:
> 
> On Thu, Oct 07 2021, Jeff Hostetler via GitGitGadget wrote:
> 
> Good to see this move forward!
> 
> This bit:
> 
>> --- a/repo-settings.c
>> +++ b/repo-settings.c
>> @@ -20,6 +20,8 @@ void prepare_repo_settings(struct repository *r)
>>   	if (r->settings.initialized++)
>>   		return;
>>   
>> +	r->settings.fsmonitor = NULL; /* lazy loaded */
>> +
>>   	/* Defaults */
>>   	r->settings.index_version = -1;
>>   	r->settings.core_untracked_cache = UNTRACKED_CACHE_KEEP;
>> diff --git a/repository.h b/repository.h
> 
> Is carried forward from v1, but with 3050b6dfc75 (repo-settings.c:
> simplify the setup, 2021-09-21) isn't needed. It's init'd to 0/NULL
> already, but was -1 before.
> 
> So this hunk can go, and its presence makes for confusing reading
> without that history, is it set before somehow? No, just working around
> older code that's no longer there.

Um, yeah, if the structure is now zero-filled on init
rather than the weird -1, then we don't need this here.
I'll remove in my next version.  Thanks!


> 
> But also: For untracked_cache_setting and fetch_negotiation_setting
> we've got an embedded enum in the struct, but this...
> 
>> index a057653981c..89a1873ade7 100644
>> --- a/repository.h
>> +++ b/repository.h
>> @@ -4,6 +4,7 @@
>>   #include "path.h"
>>   
>>   struct config_set;
>> +struct fsmonitor_settings;
>>   struct git_hash_algo;
>>   struct index_state;
>>   struct lock_file;
>> @@ -34,6 +35,8 @@ struct repo_settings {
>>   	int command_requires_full_index;
>>   	int sparse_index;
>>   
>> +	struct fsmonitor_settings *fsmonitor; /* lazy loaded */
>> +
>>   	int index_version;
>>   	enum untracked_cache_setting core_untracked_cache;
>>   
> 
> Is a pointer to a struct that has an "enum fsmonitor_mode mode", and the
> code in fsmonitor-settings.c seems to be repeating the patterns we had
> in repo-settings.c pre-3050b6dfc75, e.g. checking whether a bool config
> variable exists *and* is true, v.s. checking if it exists (presumably an
> explicit false wants to override something).

My usage is a little more complicated because the historical value
had double duty.  And I only want to set the hook value when not
using the builtin and when the repo itself is compatible with being
monitored.  Later in the series we'll disallow remote worktrees,
for example.  I wanted to hide some of those details in my
fsmonitor-settings.c and keep them out of repo-settings.c.  And
to avoid paying for that when not needed, I made is a lazy-load
and hid the values behind an opaque type with access functions.


> 
> I haven't looked carefully, but between that & the "char *hook_path"
> being something that'll need to be made to use Emily's hook config
> series sooner than later, can't we read/setup the initial config in
> "repo_cfg_bool"?

It's been a while since I looked at Emily's series. I'll have to
revisit that later.  I do wonder if the existing fsmonitor hook
is in the same class as the other hooks or whether it is a one-off
and should be separate, but again I need to review that work first.

> 
> The relevant commit message just says:
> 
>      Move FSMonitor config settings to a new `struct fsmonitor_settings`
>      structure.  Add a lazily-loaded pointer to `struct repo_settings`.
>      Create `fsm_settings__get_*()` getters to lazily look up fsmonitor-
>      related config settings.[...]
> 
> Which I think can be paraphrased as "Add scaffolding to repo-settings.c
> but do config loading differently than everything there (lazily),
> because...", except the "because" is missing :)
> 

Let me reword the commit message and try to explain what I was
going for.  Thanks!

Jeff

[PATCH v3 0/5] Builtin FSMonitor Part 2

[Jeff Hostetler via GitGitGadget]

Here is V3 of Part 2 of my Builtin FSMonitor series. Like V2, it is built
upon "next" because it requires "ab/repo-settings-cleanup" and
"jh/builtin-fsmonitor-part1" series.

V3 removes the explicit initialization of r->repo_settings->fsmonitor in
repo-settings.c as requested. It also includes a more detailed commit
message for the 3 commit to explain the rationale for putting fsmonitor
settings in its own source file rather than adding it repo-settings.c

There was a comment on the V2 series about integrating the fsmonitor hook
path with the on-going "hook config" effort that I did not address. I think
it would be best to do that in a follow-on if it makes sense to do so.

cc: Bagas Sanjaya bagasdotme@gmail.com cc: Jeff Hostetler
git@jeffhostetler.com

Jeff Hostetler (5):
  fsmonitor: enhance existing comments
  fsmonitor-ipc: create client routines for git-fsmonitor--daemon
  fsmonitor: config settings are repository-specific
  fsmonitor: use IPC to query the builtin FSMonitor daemon
  fsmonitor: update fsmonitor config documentation

 Documentation/config/core.txt      |  56 ++++++---
 Documentation/git-update-index.txt |  27 +++--
 Documentation/githooks.txt         |   3 +-
 Makefile                           |   2 +
 builtin/update-index.c             |  19 +++-
 cache.h                            |   1 -
 config.c                           |  14 ---
 config.h                           |   1 -
 environment.c                      |   1 -
 fsmonitor-ipc.c                    | 176 +++++++++++++++++++++++++++++
 fsmonitor-ipc.h                    |  48 ++++++++
 fsmonitor-settings.c               |  97 ++++++++++++++++
 fsmonitor-settings.h               |  21 ++++
 fsmonitor.c                        | 130 +++++++++++++++------
 fsmonitor.h                        |  18 ++-
 repository.h                       |   3 +
 t/README                           |   4 +-
 17 files changed, 536 insertions(+), 85 deletions(-)
 create mode 100644 fsmonitor-ipc.c
 create mode 100644 fsmonitor-ipc.h
 create mode 100644 fsmonitor-settings.c
 create mode 100644 fsmonitor-settings.h


base-commit: 6e70778dc91e2139466c15ff15a02a22a2ada2d1
Published-As: https://github.com/gitgitgadget/git/releases/tag/pr-1041%2Fjeffhostetler%2Fbuiltin-fsmonitor-part2-v3
Fetch-It-Via: git fetch https://github.com/gitgitgadget/git pr-1041/jeffhostetler/builtin-fsmonitor-part2-v3
Pull-Request: https://github.com/gitgitgadget/git/pull/1041

Range-diff vs v2:

 1:  cb25eeaf72d = 1:  cb25eeaf72d fsmonitor: enhance existing comments
 2:  df81a63acee = 2:  df81a63acee fsmonitor-ipc: create client routines for git-fsmonitor--daemon
 3:  7d5a353e74d ! 3:  a1d606aa622 fsmonitor: config settings are repository-specific
     @@ Metadata
       ## Commit message ##
          fsmonitor: config settings are repository-specific
      
     -    Move FSMonitor config settings to a new `struct fsmonitor_settings`
     -    structure.  Add a lazily-loaded pointer to `struct repo_settings`.
     +    Move fsmonitor config settings to a new and opaque
     +    `struct fsmonitor_settings` structure.  Add a lazily-loaded pointer
     +    to this into `struct repo_settings`
     +
     +    Create an `enum fsmonitor_mode` type in `struct fsmonitor_settings` to
     +    represent the state of fsmonitor.  This lets us represent which, if
     +    any, fsmonitor provider (hook or IPC) is enabled.
     +
          Create `fsm_settings__get_*()` getters to lazily look up fsmonitor-
          related config settings.
      
     -    Get rid of the `core_fsmonitor` global variable, and add support for
     -    the new `core.useBuiltinFSMonitor` config setting.  Move config code
     -    to lookup the existing `core.fsmonitor` value to `fsmonitor-settings.[ch]`.
     +    Add support for the new `core.useBuiltinFSMonitor` config setting.
     +
     +    Get rid of the `core_fsmonitor` global variable.  Move the code to
     +    lookup the existing `core.fsmonitor` config value into the fsmonitor
     +    settings.
     +
     +    Create a hook pathname variable in `struct fsmonitor-settings` and
     +    only set it when in hook mode.
      
     -    The `core_fsmonitor` global variable was used to store the pathname to
     -    the FSMonitor hook and it was used as a boolean to see if FSMonitor
     -    was enabled.  This dual usage will lead to confusion when we add
     -    support for a builtin FSMonitor based on IPC, since the builtin
     -    FSMonitor doesn't need the hook pathname.
     +    The existing `core_fsmonitor` global variable was used to store the
     +    pathname to the fsmonitor hook *and* it was used as a boolean to see
     +    if fsmonitor was enabled.  This dual usage and global visibility leads
     +    to confusion when we add the IPC-based provider.  So lets hide the
     +    details in fsmonitor-settings.c and let it decide which provider to
     +    use in the case of multiple settings.  This avoids cluttering up
     +    repo-settings.c with these private details.
      
     -    Replace the boolean usage with an `enum fsmonitor_mode` to represent
     -    the state of FSMonitor.  And only set the pathname when in HOOK mode.
     +    A future commit in builtin-fsmonitor series will add the ability to
     +    disqualify worktrees for various reasons, such as being mounted from a
     +    remote volume, where fsmonitor should not be started.  Having the
     +    config settings hidden in fsmonitor-settings.c allows such worktree
     +    restrictions to override the config values used.
      
          Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
      
     @@ fsmonitor.h: static inline void mark_fsmonitor_valid(struct index_state *istate,
       		untracked_cache_invalidate_path(istate, ce->name, 1);
       		trace_printf_key(&trace_fsmonitor, "mark_fsmonitor_invalid '%s'", ce->name);
      
     - ## repo-settings.c ##
     -@@ repo-settings.c: void prepare_repo_settings(struct repository *r)
     - 	if (r->settings.initialized++)
     - 		return;
     - 
     -+	r->settings.fsmonitor = NULL; /* lazy loaded */
     -+
     - 	/* Defaults */
     - 	r->settings.index_version = -1;
     - 	r->settings.core_untracked_cache = UNTRACKED_CACHE_KEEP;
     -
       ## repository.h ##
      @@
       #include "path.h"
 4:  8608c8718d8 = 4:  4d8d812be08 fsmonitor: use IPC to query the builtin FSMonitor daemon
 5:  7c22ce53377 = 5:  45a86cef8d7 fsmonitor: update fsmonitor config documentation

-- 
gitgitgadget

[PATCH v3 1/5] fsmonitor: enhance existing comments

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 fsmonitor.c | 37 ++++++++++++++++++++++++++++++-------
 1 file changed, 30 insertions(+), 7 deletions(-)

diff --git a/fsmonitor.c b/fsmonitor.c
index ab9bfc60b34..ec4c46407c5 100644
--- a/fsmonitor.c
+++ b/fsmonitor.c
@@ -301,9 +301,25 @@ void refresh_fsmonitor(struct index_state *istate)
 			core_fsmonitor, query_success ? "success" : "failure");
 	}
 
-	/* a fsmonitor process can return '/' to indicate all entries are invalid */
+	/*
+	 * The response from FSMonitor (excluding the header token) is
+	 * either:
+	 *
+	 * [a] a (possibly empty) list of NUL delimited relative
+	 *     pathnames of changed paths.  This list can contain
+	 *     files and directories.  Directories have a trailing
+	 *     slash.
+	 *
+	 * [b] a single '/' to indicate the provider had no
+	 *     information and that we should consider everything
+	 *     invalid.  We call this a trivial response.
+	 */
 	if (query_success && query_result.buf[bol] != '/') {
-		/* Mark all entries returned by the monitor as dirty */
+		/*
+		 * Mark all pathnames returned by the monitor as dirty.
+		 *
+		 * This updates both the cache-entries and the untracked-cache.
+		 */
 		buf = query_result.buf;
 		for (i = bol; i < query_result.len; i++) {
 			if (buf[i] != '\0')
@@ -318,11 +334,15 @@ void refresh_fsmonitor(struct index_state *istate)
 		if (istate->untracked)
 			istate->untracked->use_fsmonitor = 1;
 	} else {
-
-		/* We only want to run the post index changed hook if we've actually changed entries, so keep track
-		 * if we actually changed entries or not */
+		/*
+		 * We received a trivial response, so invalidate everything.
+		 *
+		 * We only want to run the post index changed hook if
+		 * we've actually changed entries, so keep track if we
+		 * actually changed entries or not.
+		 */
 		int is_cache_changed = 0;
-		/* Mark all entries invalid */
+
 		for (i = 0; i < istate->cache_nr; i++) {
 			if (istate->cache[i]->ce_flags & CE_FSMONITOR_VALID) {
 				is_cache_changed = 1;
@@ -330,7 +350,10 @@ void refresh_fsmonitor(struct index_state *istate)
 			}
 		}
 
-		/* If we're going to check every file, ensure we save the results */
+		/*
+		 * If we're going to check every file, ensure we save
+		 * the results.
+		 */
 		if (is_cache_changed)
 			istate->cache_changed |= FSMONITOR_CHANGED;
 
-- 
gitgitgadget

[PATCH v3 2/5] fsmonitor-ipc: create client routines for git-fsmonitor--daemon

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Create fsmonitor_ipc__*() client routines to spawn the built-in file
system monitor daemon and send it an IPC request using the `Simple
IPC` API.

Stub in empty fsmonitor_ipc__*() functions for unsupported platforms.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 Makefile        |   1 +
 fsmonitor-ipc.c | 176 ++++++++++++++++++++++++++++++++++++++++++++++++
 fsmonitor-ipc.h |  48 +++++++++++++
 3 files changed, 225 insertions(+)
 create mode 100644 fsmonitor-ipc.c
 create mode 100644 fsmonitor-ipc.h

diff --git a/Makefile b/Makefile
index 381bed2c1d2..d51fd8b33ce 100644
--- a/Makefile
+++ b/Makefile
@@ -897,6 +897,7 @@ LIB_OBJS += fetch-pack.o
 LIB_OBJS += fmt-merge-msg.o
 LIB_OBJS += fsck.o
 LIB_OBJS += fsmonitor.o
+LIB_OBJS += fsmonitor-ipc.o
 LIB_OBJS += gettext.o
 LIB_OBJS += gpg-interface.o
 LIB_OBJS += graph.o
diff --git a/fsmonitor-ipc.c b/fsmonitor-ipc.c
new file mode 100644
index 00000000000..ccc32d2a17e
--- /dev/null
+++ b/fsmonitor-ipc.c
@@ -0,0 +1,176 @@
+#include "cache.h"
+#include "fsmonitor.h"
+#include "simple-ipc.h"
+#include "fsmonitor-ipc.h"
+#include "run-command.h"
+#include "strbuf.h"
+#include "trace2.h"
+
+#ifdef HAVE_FSMONITOR_DAEMON_BACKEND
+
+int fsmonitor_ipc__is_supported(void)
+{
+	return 1;
+}
+
+GIT_PATH_FUNC(fsmonitor_ipc__get_path, "fsmonitor--daemon.ipc")
+
+enum ipc_active_state fsmonitor_ipc__get_state(void)
+{
+	return ipc_get_active_state(fsmonitor_ipc__get_path());
+}
+
+static int spawn_daemon(void)
+{
+	const char *args[] = { "fsmonitor--daemon", "start", NULL };
+
+	return run_command_v_opt_tr2(args, RUN_COMMAND_NO_STDIN | RUN_GIT_CMD,
+				    "fsmonitor");
+}
+
+int fsmonitor_ipc__send_query(const char *since_token,
+			      struct strbuf *answer)
+{
+	int ret = -1;
+	int tried_to_spawn = 0;
+	enum ipc_active_state state = IPC_STATE__OTHER_ERROR;
+	struct ipc_client_connection *connection = NULL;
+	struct ipc_client_connect_options options
+		= IPC_CLIENT_CONNECT_OPTIONS_INIT;
+	const char *tok = since_token ? since_token : "";
+	size_t tok_len = since_token ? strlen(since_token) : 0;
+
+	options.wait_if_busy = 1;
+	options.wait_if_not_found = 0;
+
+	trace2_region_enter("fsm_client", "query", NULL);
+	trace2_data_string("fsm_client", NULL, "query/command", tok);
+
+try_again:
+	state = ipc_client_try_connect(fsmonitor_ipc__get_path(), &options,
+				       &connection);
+
+	switch (state) {
+	case IPC_STATE__LISTENING:
+		ret = ipc_client_send_command_to_connection(
+			connection, tok, tok_len, answer);
+		ipc_client_close_connection(connection);
+
+		trace2_data_intmax("fsm_client", NULL,
+				   "query/response-length", answer->len);
+
+		if (fsmonitor_is_trivial_response(answer))
+			trace2_data_intmax("fsm_client", NULL,
+					   "query/trivial-response", 1);
+
+		goto done;
+
+	case IPC_STATE__NOT_LISTENING:
+	case IPC_STATE__PATH_NOT_FOUND:
+		if (tried_to_spawn)
+			goto done;
+
+		tried_to_spawn++;
+		if (spawn_daemon())
+			goto done;
+
+		/*
+		 * Try again, but this time give the daemon a chance to
+		 * actually create the pipe/socket.
+		 *
+		 * Granted, the daemon just started so it can't possibly have
+		 * any FS cached yet, so we'll always get a trivial answer.
+		 * BUT the answer should include a new token that can serve
+		 * as the basis for subsequent requests.
+		 */
+		options.wait_if_not_found = 1;
+		goto try_again;
+
+	case IPC_STATE__INVALID_PATH:
+		ret = error(_("fsmonitor_ipc__send_query: invalid path '%s'"),
+			    fsmonitor_ipc__get_path());
+		goto done;
+
+	case IPC_STATE__OTHER_ERROR:
+	default:
+		ret = error(_("fsmonitor_ipc__send_query: unspecified error on '%s'"),
+			    fsmonitor_ipc__get_path());
+		goto done;
+	}
+
+done:
+	trace2_region_leave("fsm_client", "query", NULL);
+
+	return ret;
+}
+
+int fsmonitor_ipc__send_command(const char *command,
+				struct strbuf *answer)
+{
+	struct ipc_client_connection *connection = NULL;
+	struct ipc_client_connect_options options
+		= IPC_CLIENT_CONNECT_OPTIONS_INIT;
+	int ret;
+	enum ipc_active_state state;
+	const char *c = command ? command : "";
+	size_t c_len = command ? strlen(command) : 0;
+
+	strbuf_reset(answer);
+
+	options.wait_if_busy = 1;
+	options.wait_if_not_found = 0;
+
+	state = ipc_client_try_connect(fsmonitor_ipc__get_path(), &options,
+				       &connection);
+	if (state != IPC_STATE__LISTENING) {
+		die("fsmonitor--daemon is not running");
+		return -1;
+	}
+
+	ret = ipc_client_send_command_to_connection(connection, c, c_len,
+						    answer);
+	ipc_client_close_connection(connection);
+
+	if (ret == -1) {
+		die("could not send '%s' command to fsmonitor--daemon", c);
+		return -1;
+	}
+
+	return 0;
+}
+
+#else
+
+/*
+ * A trivial implementation of the fsmonitor_ipc__ API for unsupported
+ * platforms.
+ */
+
+int fsmonitor_ipc__is_supported(void)
+{
+	return 0;
+}
+
+const char *fsmonitor_ipc__get_path(void)
+{
+	return NULL;
+}
+
+enum ipc_active_state fsmonitor_ipc__get_state(void)
+{
+	return IPC_STATE__OTHER_ERROR;
+}
+
+int fsmonitor_ipc__send_query(const char *since_token,
+			      struct strbuf *answer)
+{
+	return -1;
+}
+
+int fsmonitor_ipc__send_command(const char *command,
+				struct strbuf *answer)
+{
+	return -1;
+}
+
+#endif
diff --git a/fsmonitor-ipc.h b/fsmonitor-ipc.h
new file mode 100644
index 00000000000..b6a7067c3af
--- /dev/null
+++ b/fsmonitor-ipc.h
@@ -0,0 +1,48 @@
+#ifndef FSMONITOR_IPC_H
+#define FSMONITOR_IPC_H
+
+#include "simple-ipc.h"
+
+/*
+ * Returns true if built-in file system monitor daemon is defined
+ * for this platform.
+ */
+int fsmonitor_ipc__is_supported(void);
+
+/*
+ * Returns the pathname to the IPC named pipe or Unix domain socket
+ * where a `git-fsmonitor--daemon` process will listen.  This is a
+ * per-worktree value.
+ *
+ * Returns NULL if the daemon is not supported on this platform.
+ */
+const char *fsmonitor_ipc__get_path(void);
+
+/*
+ * Try to determine whether there is a `git-fsmonitor--daemon` process
+ * listening on the IPC pipe/socket.
+ */
+enum ipc_active_state fsmonitor_ipc__get_state(void);
+
+/*
+ * Connect to a `git-fsmonitor--daemon` process via simple-ipc
+ * and ask for the set of changed files since the given token.
+ *
+ * Spawn a daemon process in the background if necessary.
+ *
+ * Returns -1 on error; 0 on success.
+ */
+int fsmonitor_ipc__send_query(const char *since_token,
+			      struct strbuf *answer);
+
+/*
+ * Connect to a `git-fsmonitor--daemon` process via simple-ipc and
+ * send a command verb.  If no daemon is available, we DO NOT try to
+ * start one.
+ *
+ * Returns -1 on error; 0 on success.
+ */
+int fsmonitor_ipc__send_command(const char *command,
+				struct strbuf *answer);
+
+#endif /* FSMONITOR_IPC_H */
-- 
gitgitgadget

[PATCH v3 3/5] fsmonitor: config settings are repository-specific

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Move fsmonitor config settings to a new and opaque
`struct fsmonitor_settings` structure.  Add a lazily-loaded pointer
to this into `struct repo_settings`

Create an `enum fsmonitor_mode` type in `struct fsmonitor_settings` to
represent the state of fsmonitor.  This lets us represent which, if
any, fsmonitor provider (hook or IPC) is enabled.

Create `fsm_settings__get_*()` getters to lazily look up fsmonitor-
related config settings.

Add support for the new `core.useBuiltinFSMonitor` config setting.

Get rid of the `core_fsmonitor` global variable.  Move the code to
lookup the existing `core.fsmonitor` config value into the fsmonitor
settings.

Create a hook pathname variable in `struct fsmonitor-settings` and
only set it when in hook mode.

The existing `core_fsmonitor` global variable was used to store the
pathname to the fsmonitor hook *and* it was used as a boolean to see
if fsmonitor was enabled.  This dual usage and global visibility leads
to confusion when we add the IPC-based provider.  So lets hide the
details in fsmonitor-settings.c and let it decide which provider to
use in the case of multiple settings.  This avoids cluttering up
repo-settings.c with these private details.

A future commit in builtin-fsmonitor series will add the ability to
disqualify worktrees for various reasons, such as being mounted from a
remote volume, where fsmonitor should not be started.  Having the
config settings hidden in fsmonitor-settings.c allows such worktree
restrictions to override the config values used.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 Makefile               |  1 +
 builtin/update-index.c | 19 +++++++--
 cache.h                |  1 -
 config.c               | 14 ------
 config.h               |  1 -
 environment.c          |  1 -
 fsmonitor-settings.c   | 97 ++++++++++++++++++++++++++++++++++++++++++
 fsmonitor-settings.h   | 21 +++++++++
 fsmonitor.c            | 63 ++++++++++++++++-----------
 fsmonitor.h            | 18 ++++++--
 repository.h           |  3 ++
 t/README               |  4 +-
 12 files changed, 192 insertions(+), 51 deletions(-)
 create mode 100644 fsmonitor-settings.c
 create mode 100644 fsmonitor-settings.h

diff --git a/Makefile b/Makefile
index d51fd8b33ce..29ed7c4aba6 100644
--- a/Makefile
+++ b/Makefile
@@ -898,6 +898,7 @@ LIB_OBJS += fmt-merge-msg.o
 LIB_OBJS += fsck.o
 LIB_OBJS += fsmonitor.o
 LIB_OBJS += fsmonitor-ipc.o
+LIB_OBJS += fsmonitor-settings.o
 LIB_OBJS += gettext.o
 LIB_OBJS += gpg-interface.o
 LIB_OBJS += graph.o
diff --git a/builtin/update-index.c b/builtin/update-index.c
index 187203e8bb5..79db3ff37e2 100644
--- a/builtin/update-index.c
+++ b/builtin/update-index.c
@@ -1214,14 +1214,25 @@ int cmd_update_index(int argc, const char **argv, const char *prefix)
 	}
 
 	if (fsmonitor > 0) {
-		if (git_config_get_fsmonitor() == 0)
+		enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
+
+		if (fsm_mode == FSMONITOR_MODE_DISABLED) {
+			warning(_("core.useBuiltinFSMonitor is unset; "
+				"set it if you really want to enable the "
+				"builtin fsmonitor"));
 			warning(_("core.fsmonitor is unset; "
-				"set it if you really want to "
-				"enable fsmonitor"));
+				"set it if you really want to enable the "
+				"hook-based fsmonitor"));
+		}
 		add_fsmonitor(&the_index);
 		report(_("fsmonitor enabled"));
 	} else if (!fsmonitor) {
-		if (git_config_get_fsmonitor() == 1)
+		enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
+		if (fsm_mode == FSMONITOR_MODE_IPC)
+			warning(_("core.useBuiltinFSMonitor is set; "
+				"remove it if you really want to "
+				"disable fsmonitor"));
+		if (fsm_mode == FSMONITOR_MODE_HOOK)
 			warning(_("core.fsmonitor is set; "
 				"remove it if you really want to "
 				"disable fsmonitor"));
diff --git a/cache.h b/cache.h
index d092820c943..8f4e3c8bd1d 100644
--- a/cache.h
+++ b/cache.h
@@ -989,7 +989,6 @@ extern int core_preload_index;
 extern int precomposed_unicode;
 extern int protect_hfs;
 extern int protect_ntfs;
-extern const char *core_fsmonitor;
 
 extern int core_apply_sparse_checkout;
 extern int core_sparse_checkout_cone;
diff --git a/config.c b/config.c
index 2dcbe901b6b..6b6e9cacac3 100644
--- a/config.c
+++ b/config.c
@@ -2502,20 +2502,6 @@ int git_config_get_max_percent_split_change(void)
 	return -1; /* default value */
 }
 
-int git_config_get_fsmonitor(void)
-{
-	if (git_config_get_pathname("core.fsmonitor", &core_fsmonitor))
-		core_fsmonitor = getenv("GIT_TEST_FSMONITOR");
-
-	if (core_fsmonitor && !*core_fsmonitor)
-		core_fsmonitor = NULL;
-
-	if (core_fsmonitor)
-		return 1;
-
-	return 0;
-}
-
 int git_config_get_index_threads(int *dest)
 {
 	int is_bool, val;
diff --git a/config.h b/config.h
index f119de01309..69d733824a0 100644
--- a/config.h
+++ b/config.h
@@ -610,7 +610,6 @@ int git_config_get_pathname(const char *key, const char **dest);
 int git_config_get_index_threads(int *dest);
 int git_config_get_split_index(void);
 int git_config_get_max_percent_split_change(void);
-int git_config_get_fsmonitor(void);
 
 /* This dies if the configured or default date is in the future */
 int git_config_get_expiry(const char *key, const char **output);
diff --git a/environment.c b/environment.c
index 9da7f3c1a19..68f90632245 100644
--- a/environment.c
+++ b/environment.c
@@ -82,7 +82,6 @@ int protect_hfs = PROTECT_HFS_DEFAULT;
 #define PROTECT_NTFS_DEFAULT 1
 #endif
 int protect_ntfs = PROTECT_NTFS_DEFAULT;
-const char *core_fsmonitor;
 
 /*
  * The character that begins a commented line in user-editable file
diff --git a/fsmonitor-settings.c b/fsmonitor-settings.c
new file mode 100644
index 00000000000..2770266f5ee
--- /dev/null
+++ b/fsmonitor-settings.c
@@ -0,0 +1,97 @@
+#include "cache.h"
+#include "config.h"
+#include "repository.h"
+#include "fsmonitor-settings.h"
+
+/*
+ * We keep this structure defintion private and have getters
+ * for all fields so that we can lazy load it as needed.
+ */
+struct fsmonitor_settings {
+	enum fsmonitor_mode mode;
+	char *hook_path;
+};
+
+void fsm_settings__set_ipc(struct repository *r)
+{
+	struct fsmonitor_settings *s = r->settings.fsmonitor;
+
+	s->mode = FSMONITOR_MODE_IPC;
+}
+
+void fsm_settings__set_hook(struct repository *r, const char *path)
+{
+	struct fsmonitor_settings *s = r->settings.fsmonitor;
+
+	s->mode = FSMONITOR_MODE_HOOK;
+	s->hook_path = strdup(path);
+}
+
+void fsm_settings__set_disabled(struct repository *r)
+{
+	struct fsmonitor_settings *s = r->settings.fsmonitor;
+
+	s->mode = FSMONITOR_MODE_DISABLED;
+	FREE_AND_NULL(s->hook_path);
+}
+
+static int check_for_ipc(struct repository *r)
+{
+	int value;
+
+	if (!repo_config_get_bool(r, "core.usebuiltinfsmonitor", &value) &&
+	    value) {
+		fsm_settings__set_ipc(r);
+		return 1;
+	}
+
+	return 0;
+}
+
+static int check_for_hook(struct repository *r)
+{
+	const char *const_str;
+
+	if (repo_config_get_pathname(r, "core.fsmonitor", &const_str))
+		const_str = getenv("GIT_TEST_FSMONITOR");
+
+	if (const_str && *const_str) {
+		fsm_settings__set_hook(r, const_str);
+		return 1;
+	}
+
+	return 0;
+}
+
+static void lookup_fsmonitor_settings(struct repository *r)
+{
+	struct fsmonitor_settings *s;
+
+	CALLOC_ARRAY(s, 1);
+
+	r->settings.fsmonitor = s;
+
+	if (check_for_ipc(r))
+		return;
+
+	if (check_for_hook(r))
+		return;
+
+	fsm_settings__set_disabled(r);
+}
+
+enum fsmonitor_mode fsm_settings__get_mode(struct repository *r)
+{
+	if (!r->settings.fsmonitor)
+		lookup_fsmonitor_settings(r);
+
+	return r->settings.fsmonitor->mode;
+}
+
+const char *fsm_settings__get_hook_path(struct repository *r)
+{
+	if (!r->settings.fsmonitor)
+		lookup_fsmonitor_settings(r);
+
+	return r->settings.fsmonitor->hook_path;
+}
diff --git a/fsmonitor-settings.h b/fsmonitor-settings.h
new file mode 100644
index 00000000000..50b29234616
--- /dev/null
+++ b/fsmonitor-settings.h
@@ -0,0 +1,21 @@
+#ifndef FSMONITOR_SETTINGS_H
+#define FSMONITOR_SETTINGS_H
+
+struct repository;
+
+enum fsmonitor_mode {
+	FSMONITOR_MODE_DISABLED = 0,
+	FSMONITOR_MODE_HOOK = 1, /* core.fsmonitor */
+	FSMONITOR_MODE_IPC = 2,  /* core.useBuiltinFSMonitor */
+};
+
+void fsm_settings__set_ipc(struct repository *r);
+void fsm_settings__set_hook(struct repository *r, const char *path);
+void fsm_settings__set_disabled(struct repository *r);
+
+enum fsmonitor_mode fsm_settings__get_mode(struct repository *r);
+const char *fsm_settings__get_hook_path(struct repository *r);
+
+struct fsmonitor_settings;
+
+#endif /* FSMONITOR_SETTINGS_H */
diff --git a/fsmonitor.c b/fsmonitor.c
index ec4c46407c5..63174630c0e 100644
--- a/fsmonitor.c
+++ b/fsmonitor.c
@@ -3,6 +3,7 @@
 #include "dir.h"
 #include "ewah/ewok.h"
 #include "fsmonitor.h"
+#include "fsmonitor-ipc.h"
 #include "run-command.h"
 #include "strbuf.h"
 
@@ -148,15 +149,18 @@ void write_fsmonitor_extension(struct strbuf *sb, struct index_state *istate)
 /*
  * Call the query-fsmonitor hook passing the last update token of the saved results.
  */
-static int query_fsmonitor(int version, const char *last_update, struct strbuf *query_result)
+static int query_fsmonitor_hook(struct repository *r,
+				int version,
+				const char *last_update,
+				struct strbuf *query_result)
 {
 	struct child_process cp = CHILD_PROCESS_INIT;
 	int result;
 
-	if (!core_fsmonitor)
+	if (fsm_settings__get_mode(r) != FSMONITOR_MODE_HOOK)
 		return -1;
 
-	strvec_push(&cp.args, core_fsmonitor);
+	strvec_push(&cp.args, fsm_settings__get_hook_path(r));
 	strvec_pushf(&cp.args, "%d", version);
 	strvec_pushf(&cp.args, "%s", last_update);
 	cp.use_shell = 1;
@@ -238,17 +242,28 @@ void refresh_fsmonitor(struct index_state *istate)
 	struct strbuf last_update_token = STRBUF_INIT;
 	char *buf;
 	unsigned int i;
+	struct repository *r = istate->repo ? istate->repo : the_repository;
+	enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
 
-	if (!core_fsmonitor || istate->fsmonitor_has_run_once)
+	if (fsm_mode <= FSMONITOR_MODE_DISABLED ||
+	    istate->fsmonitor_has_run_once)
 		return;
 
-	hook_version = fsmonitor_hook_version();
-
 	istate->fsmonitor_has_run_once = 1;
 
 	trace_printf_key(&trace_fsmonitor, "refresh fsmonitor");
+
+	if (fsm_mode == FSMONITOR_MODE_IPC) {
+		/* TODO */
+		return;
+	}
+
+	assert(fsm_mode == FSMONITOR_MODE_HOOK);
+
+	hook_version = fsmonitor_hook_version();
+
 	/*
-	 * This could be racy so save the date/time now and query_fsmonitor
+	 * This could be racy so save the date/time now and query_fsmonitor_hook
 	 * should be inclusive to ensure we don't miss potential changes.
 	 */
 	last_update = getnanotime();
@@ -256,13 +271,14 @@ void refresh_fsmonitor(struct index_state *istate)
 		strbuf_addf(&last_update_token, "%"PRIu64"", last_update);
 
 	/*
-	 * If we have a last update token, call query_fsmonitor for the set of
+	 * If we have a last update token, call query_fsmonitor_hook for the set of
 	 * changes since that token, else assume everything is possibly dirty
 	 * and check it all.
 	 */
 	if (istate->fsmonitor_last_update) {
 		if (hook_version == -1 || hook_version == HOOK_INTERFACE_VERSION2) {
-			query_success = !query_fsmonitor(HOOK_INTERFACE_VERSION2,
+			query_success = !query_fsmonitor_hook(
+				r, HOOK_INTERFACE_VERSION2,
 				istate->fsmonitor_last_update, &query_result);
 
 			if (query_success) {
@@ -292,13 +308,17 @@ void refresh_fsmonitor(struct index_state *istate)
 		}
 
 		if (hook_version == HOOK_INTERFACE_VERSION1) {
-			query_success = !query_fsmonitor(HOOK_INTERFACE_VERSION1,
+			query_success = !query_fsmonitor_hook(
+				r, HOOK_INTERFACE_VERSION1,
 				istate->fsmonitor_last_update, &query_result);
 		}
 
-		trace_performance_since(last_update, "fsmonitor process '%s'", core_fsmonitor);
-		trace_printf_key(&trace_fsmonitor, "fsmonitor process '%s' returned %s",
-			core_fsmonitor, query_success ? "success" : "failure");
+		trace_performance_since(last_update, "fsmonitor process '%s'",
+					fsm_settings__get_hook_path(r));
+		trace_printf_key(&trace_fsmonitor,
+				 "fsmonitor process '%s' returned %s",
+				 fsm_settings__get_hook_path(r),
+				 query_success ? "success" : "failure");
 	}
 
 	/*
@@ -434,7 +454,8 @@ void remove_fsmonitor(struct index_state *istate)
 void tweak_fsmonitor(struct index_state *istate)
 {
 	unsigned int i;
-	int fsmonitor_enabled = git_config_get_fsmonitor();
+	struct repository *r = istate->repo ? istate->repo : the_repository;
+	int fsmonitor_enabled = (fsm_settings__get_mode(r) > FSMONITOR_MODE_DISABLED);
 
 	if (istate->fsmonitor_dirty) {
 		if (fsmonitor_enabled) {
@@ -454,16 +475,8 @@ void tweak_fsmonitor(struct index_state *istate)
 		istate->fsmonitor_dirty = NULL;
 	}
 
-	switch (fsmonitor_enabled) {
-	case -1: /* keep: do nothing */
-		break;
-	case 0: /* false */
-		remove_fsmonitor(istate);
-		break;
-	case 1: /* true */
+	if (fsmonitor_enabled)
 		add_fsmonitor(istate);
-		break;
-	default: /* unknown value: do nothing */
-		break;
-	}
+	else
+		remove_fsmonitor(istate);
 }
diff --git a/fsmonitor.h b/fsmonitor.h
index f20d72631d7..f9201411aa7 100644
--- a/fsmonitor.h
+++ b/fsmonitor.h
@@ -3,6 +3,7 @@
 
 #include "cache.h"
 #include "dir.h"
+#include "fsmonitor-settings.h"
 
 extern struct trace_key trace_fsmonitor;
 
@@ -57,7 +58,11 @@ int fsmonitor_is_trivial_response(const struct strbuf *query_result);
  */
 static inline int is_fsmonitor_refreshed(const struct index_state *istate)
 {
-	return !core_fsmonitor || istate->fsmonitor_has_run_once;
+	struct repository *r = istate->repo ? istate->repo : the_repository;
+	enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
+
+	return fsm_mode <= FSMONITOR_MODE_DISABLED ||
+		istate->fsmonitor_has_run_once;
 }
 
 /*
@@ -67,7 +72,11 @@ static inline int is_fsmonitor_refreshed(const struct index_state *istate)
  */
 static inline void mark_fsmonitor_valid(struct index_state *istate, struct cache_entry *ce)
 {
-	if (core_fsmonitor && !(ce->ce_flags & CE_FSMONITOR_VALID)) {
+	struct repository *r = istate->repo ? istate->repo : the_repository;
+	enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
+
+	if (fsm_mode > FSMONITOR_MODE_DISABLED &&
+	    !(ce->ce_flags & CE_FSMONITOR_VALID)) {
 		istate->cache_changed = 1;
 		ce->ce_flags |= CE_FSMONITOR_VALID;
 		trace_printf_key(&trace_fsmonitor, "mark_fsmonitor_clean '%s'", ce->name);
@@ -83,7 +92,10 @@ static inline void mark_fsmonitor_valid(struct index_state *istate, struct cache
  */
 static inline void mark_fsmonitor_invalid(struct index_state *istate, struct cache_entry *ce)
 {
-	if (core_fsmonitor) {
+	struct repository *r = istate->repo ? istate->repo : the_repository;
+	enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
+
+	if (fsm_mode > FSMONITOR_MODE_DISABLED) {
 		ce->ce_flags &= ~CE_FSMONITOR_VALID;
 		untracked_cache_invalidate_path(istate, ce->name, 1);
 		trace_printf_key(&trace_fsmonitor, "mark_fsmonitor_invalid '%s'", ce->name);
diff --git a/repository.h b/repository.h
index a057653981c..89a1873ade7 100644
--- a/repository.h
+++ b/repository.h
@@ -4,6 +4,7 @@
 #include "path.h"
 
 struct config_set;
+struct fsmonitor_settings;
 struct git_hash_algo;
 struct index_state;
 struct lock_file;
@@ -34,6 +35,8 @@ struct repo_settings {
 	int command_requires_full_index;
 	int sparse_index;
 
+	struct fsmonitor_settings *fsmonitor; /* lazy loaded */
+
 	int index_version;
 	enum untracked_cache_setting core_untracked_cache;
 
diff --git a/t/README b/t/README
index b92155a822e..6dc4a1d10cf 100644
--- a/t/README
+++ b/t/README
@@ -405,8 +405,8 @@ every 'git commit-graph write', as if the `--changed-paths` option was
 passed in.
 
 GIT_TEST_FSMONITOR=$PWD/t7519/fsmonitor-all exercises the fsmonitor
-code path for utilizing a file system monitor to speed up detecting
-new or changed files.
+code path for utilizing a (hook based) file system monitor to speed up
+detecting new or changed files.
 
 GIT_TEST_INDEX_VERSION=<n> exercises the index read/write code path
 for the index version specified.  Can be set to any valid version
-- 
gitgitgadget

[PATCH v3 4/5] fsmonitor: use IPC to query the builtin FSMonitor daemon

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Use simple IPC to directly communicate with the new builtin file
system monitor daemon when `core.useBuiltinFSMonitor` is set.

The `core.fsmonitor` setting has already been defined as a HOOK
pathname.  Historically, this has been set to a HOOK script that will
talk with Watchman.  For compatibility reasons, we do not want to
overload that definition (and cause problems if users have multiple
versions of Git installed).

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 fsmonitor.c | 34 ++++++++++++++++++++++++++++++++--
 1 file changed, 32 insertions(+), 2 deletions(-)

diff --git a/fsmonitor.c b/fsmonitor.c
index 63174630c0e..695fb0ce4e7 100644
--- a/fsmonitor.c
+++ b/fsmonitor.c
@@ -254,8 +254,37 @@ void refresh_fsmonitor(struct index_state *istate)
 	trace_printf_key(&trace_fsmonitor, "refresh fsmonitor");
 
 	if (fsm_mode == FSMONITOR_MODE_IPC) {
-		/* TODO */
-		return;
+		query_success = !fsmonitor_ipc__send_query(
+			istate->fsmonitor_last_update ?
+			istate->fsmonitor_last_update : "builtin:fake",
+			&query_result);
+		if (query_success) {
+			/*
+			 * The response contains a series of nul terminated
+			 * strings.  The first is the new token.
+			 *
+			 * Use `char *buf` as an interlude to trick the CI
+			 * static analysis to let us use `strbuf_addstr()`
+			 * here (and only copy the token) rather than
+			 * `strbuf_addbuf()`.
+			 */
+			buf = query_result.buf;
+			strbuf_addstr(&last_update_token, buf);
+			bol = last_update_token.len + 1;
+		} else {
+			/*
+			 * The builtin daemon is not available on this
+			 * platform -OR- we failed to get a response.
+			 *
+			 * Generate a fake token (rather than a V1
+			 * timestamp) for the index extension.  (If
+			 * they switch back to the hook API, we don't
+			 * want ambiguous state.)
+			 */
+			strbuf_addstr(&last_update_token, "builtin:fake");
+		}
+
+		goto apply_results;
 	}
 
 	assert(fsm_mode == FSMONITOR_MODE_HOOK);
@@ -321,6 +350,7 @@ void refresh_fsmonitor(struct index_state *istate)
 				 query_success ? "success" : "failure");
 	}
 
+apply_results:
 	/*
 	 * The response from FSMonitor (excluding the header token) is
 	 * either:
-- 
gitgitgadget

[PATCH v3 5/5] fsmonitor: update fsmonitor config documentation

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Update references to `core.fsmonitor` and `core.fsmonitorHookVersion` and
pointers to `Watchman` to mention the new `core.useBuiltinFSMonitor`
value.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 Documentation/config/core.txt      | 56 ++++++++++++++++++++++--------
 Documentation/git-update-index.txt | 27 +++++++-------
 Documentation/githooks.txt         |  3 +-
 3 files changed, 59 insertions(+), 27 deletions(-)

diff --git a/Documentation/config/core.txt b/Documentation/config/core.txt
index c04f62a54a1..4f6e519bc02 100644
--- a/Documentation/config/core.txt
+++ b/Documentation/config/core.txt
@@ -62,22 +62,50 @@ core.protectNTFS::
 	Defaults to `true` on Windows, and `false` elsewhere.
 
 core.fsmonitor::
-	If set, the value of this variable is used as a command which
-	will identify all files that may have changed since the
-	requested date/time. This information is used to speed up git by
-	avoiding unnecessary processing of files that have not changed.
-	See the "fsmonitor-watchman" section of linkgit:githooks[5].
+	If set, this variable contains the pathname of the "fsmonitor"
+	hook command.
++
+This hook command is used to identify all files that may have changed
+since the requested date/time. This information is used to speed up
+git by avoiding unnecessary scanning of files that have not changed.
++
+See the "fsmonitor-watchman" section of linkgit:githooks[5].
++
+Note: The value of this config setting is ignored if the
+built-in file system monitor is enabled (see `core.useBuiltinFSMonitor`).
 
 core.fsmonitorHookVersion::
-	Sets the version of hook that is to be used when calling fsmonitor.
-	There are currently versions 1 and 2. When this is not set,
-	version 2 will be tried first and if it fails then version 1
-	will be tried. Version 1 uses a timestamp as input to determine
-	which files have changes since that time but some monitors
-	like watchman have race conditions when used with a timestamp.
-	Version 2 uses an opaque string so that the monitor can return
-	something that can be used to determine what files have changed
-	without race conditions.
+	Sets the protocol version to be used when invoking the
+	"fsmonitor" hook.
++
+There are currently versions 1 and 2. When this is not set,
+version 2 will be tried first and if it fails then version 1
+will be tried. Version 1 uses a timestamp as input to determine
+which files have changes since that time but some monitors
+like Watchman have race conditions when used with a timestamp.
+Version 2 uses an opaque string so that the monitor can return
+something that can be used to determine what files have changed
+without race conditions.
++
+Note: The value of this config setting is ignored if the
+built-in file system monitor is enabled (see `core.useBuiltinFSMonitor`).
+
+core.useBuiltinFSMonitor::
+	If set to true, enable the built-in file system monitor
+	daemon for this working directory (linkgit:git-fsmonitor--daemon[1]).
++
+Like hook-based file system monitors, the built-in file system monitor
+can speed up Git commands that need to refresh the Git index
+(e.g. `git status`) in a working directory with many files.  The
+built-in monitor eliminates the need to install and maintain an
+external third-party tool.
++
+The built-in file system monitor is currently available only on a
+limited set of supported platforms.  Currently, this includes Windows
+and MacOS.
++
+Note: if this config setting is set to `true`, the values of
+`core.fsmonitor` and `core.fsmonitorHookVersion` are ignored.
 
 core.trustctime::
 	If false, the ctime differences between the index and the
diff --git a/Documentation/git-update-index.txt b/Documentation/git-update-index.txt
index 2853f168d97..c7c31b3fcf9 100644
--- a/Documentation/git-update-index.txt
+++ b/Documentation/git-update-index.txt
@@ -498,7 +498,9 @@ FILE SYSTEM MONITOR
 This feature is intended to speed up git operations for repos that have
 large working directories.
 
-It enables git to work together with a file system monitor (see the
+It enables git to work together with a file system monitor (see
+linkgit:git-fsmonitor--daemon[1]
+and the
 "fsmonitor-watchman" section of linkgit:githooks[5]) that can
 inform it as to what files have been modified. This enables git to avoid
 having to lstat() every file to find modified files.
@@ -508,17 +510,18 @@ performance by avoiding the cost of scanning the entire working directory
 looking for new files.
 
 If you want to enable (or disable) this feature, it is easier to use
-the `core.fsmonitor` configuration variable (see
-linkgit:git-config[1]) than using the `--fsmonitor` option to
-`git update-index` in each repository, especially if you want to do so
-across all repositories you use, because you can set the configuration
-variable in your `$HOME/.gitconfig` just once and have it affect all
-repositories you touch.
-
-When the `core.fsmonitor` configuration variable is changed, the
-file system monitor is added to or removed from the index the next time
-a command reads the index. When `--[no-]fsmonitor` are used, the file
-system monitor is immediately added to or removed from the index.
+the `core.fsmonitor` or `core.useBuiltinFSMonitor` configuration
+variable (see linkgit:git-config[1]) than using the `--fsmonitor`
+option to `git update-index` in each repository, especially if you
+want to do so across all repositories you use, because you can set the
+configuration variable in your `$HOME/.gitconfig` just once and have
+it affect all repositories you touch.
+
+When the `core.fsmonitor` or `core.useBuiltinFSMonitor` configuration
+variable is changed, the file system monitor is added to or removed
+from the index the next time a command reads the index. When
+`--[no-]fsmonitor` are used, the file system monitor is immediately
+added to or removed from the index.
 
 CONFIGURATION
 -------------
diff --git a/Documentation/githooks.txt b/Documentation/githooks.txt
index b51959ff941..b7d5e926f7b 100644
--- a/Documentation/githooks.txt
+++ b/Documentation/githooks.txt
@@ -593,7 +593,8 @@ fsmonitor-watchman
 
 This hook is invoked when the configuration option `core.fsmonitor` is
 set to `.git/hooks/fsmonitor-watchman` or `.git/hooks/fsmonitor-watchmanv2`
-depending on the version of the hook to use.
+depending on the version of the hook to use, unless overridden via
+`core.useBuiltinFSMonitor` (see linkgit:git-config[1]).
 
 Version 1 takes two arguments, a version (1) and the time in elapsed
 nanoseconds since midnight, January 1, 1970.
-- 
gitgitgadget

Re: [PATCH v3 0/5] Builtin FSMonitor Part 2

[Ævar Arnfjörð Bjarmason]

On Wed, Oct 13 2021, Jeff Hostetler via GitGitGadget wrote:

> Here is V3 of Part 2 of my Builtin FSMonitor series. Like V2, it is built
> upon "next" because it requires "ab/repo-settings-cleanup" and
> "jh/builtin-fsmonitor-part1" series.

FYI: Both of those have landed, so a next iteration of this can be built
on "master".

> V3 removes the explicit initialization of r->repo_settings->fsmonitor in
> repo-settings.c as requested. It also includes a more detailed commit
> message for the 3 commit to explain the rationale for putting fsmonitor
> settings in its own source file rather than adding it repo-settings.c

This series breaks with the "make check-docs" target, which as an aside
is broken because of recent changes of mine, so CI didn't catch this
(I'll submit a series to fix it):

    config/core.txt:95: error: git-fsmonitor--daemon[1]: link outside of our own docs, shown with 'HERE' below:
    config/core.txt:95:     'daemon for this working directory (linkgit:git-fsmonitor--daemon[1]' <-- HERE
    git-update-index.txt:502: error: git-fsmonitor--daemon[1]: link outside of our own docs, shown with 'HERE' below:
    git-update-index.txt:502:       'linkgit:git-fsmonitor--daemon[1]' <-- HERE

But that broken-ness points to a more general issue, which is that it's
not just a broken link, but that docs in this series are referring to a
manpage that doesn't exist yet.

I was going to check out some of the semantics of
core.useBuiltinfsMonitor that I commented on in earlier rounds, but I
see that there's no tests for it, probably for similar reasons as there
not being a git-fsmonitor--daemon yet: It's all (hopefully) in some
yet-to-come series.

I'm all for this being split up in steps from the 30-something patches
it was before, but if we're making forward-references to docs that don't
exist yet, adding ~100 line *.c files that seemingly aren't used at all
yet etc., the split-up seems to be a bit too aggressive.

E.g. the below diff seems to have all tests passing, so
core.usebuiltinfsmonitor is unused still? There's also large amounts of
code ifdef'd away under HAVE_FSMONITOR_DAEMON_BACKEND, but no other
in-tree reference to it?

AFAICT that's going to be used in the future. So seemingly something
that'll only be used in the series after this? I.e. this seems to have
stopped at around part 7/28 of a previous submission, but digging in the
archive e.g. HAVE_FSMONITOR_DAEMON_BACKEND semes to be first used in
step 11/28 of that[1].

Sorry to be party pooper again, I really did try to review this, but
just found that I couldn't, since so much of it seems to be a
forward-reference to some future state.

Isn't there a way to re-arrange this so that the "teach daemon XYZ
command" around steps 13-14/28 of your previous series comes before
references in the docs to the daemon that doesn't exist yet (5/5 here),
or there's perhaps some of the config scaffolding, but
e.g. "core.usebuiltinfsmonitor" comes along with the later change that
uses it?

Or just make this "part 2" series larger than 5 patches, so that it
manages to tell some holistic story, and the various bits are used at
the end?

Even so it would be a lot easier to follow if e.g. a patch using a new
"core.usebuiltinfsmonitor" variable doesn't come at say step 11/20, with
that variable having been added at a 3/20. I.e. maybe 3/20 can have the
scaffolding, but the part that adds the "core.usebuiltinfsmonitor"
support either immediately precedes its use, or is squashed into the
relevant commit that needs it. Ditto docs for so-far-not-existing things
etc.

Thanks! And hopefully this helps.

1. https://lore.kernel.org/git/49f9e2e3d49ce6e7b56839bf44535f271216abeb.1621691828.git.gitgitgadget@gmail.com/

diff --git a/fsmonitor-settings.c b/fsmonitor-settings.c
index 2770266f5ee..2c479d4f36b 100644
--- a/fsmonitor-settings.c
+++ b/fsmonitor-settings.c
@@ -35,19 +35,6 @@ void fsm_settings__set_disabled(struct repository *r)
 	FREE_AND_NULL(s->hook_path);
 }
 
-static int check_for_ipc(struct repository *r)
-{
-	int value;
-
-	if (!repo_config_get_bool(r, "core.usebuiltinfsmonitor", &value) &&
-	    value) {
-		fsm_settings__set_ipc(r);
-		return 1;
-	}
-
-	return 0;
-}
-
 static int check_for_hook(struct repository *r)
 {
 	const char *const_str;
@@ -71,9 +58,6 @@ static void lookup_fsmonitor_settings(struct repository *r)
 
 	r->settings.fsmonitor = s;
 
-	if (check_for_ipc(r))
-		return;
-
 	if (check_for_hook(r))
 		return;
 

Re: [PATCH v3 0/5] Builtin FSMonitor Part 2

[Jeff Hostetler]

On 10/15/21 8:03 AM, Ævar Arnfjörð Bjarmason wrote:
> 
> On Wed, Oct 13 2021, Jeff Hostetler via GitGitGadget wrote:
> 
>> Here is V3 of Part 2 of my Builtin FSMonitor series. Like V2, it is built
>> upon "next" because it requires "ab/repo-settings-cleanup" and
>> "jh/builtin-fsmonitor-part1" series.
> 
> FYI: Both of those have landed, so a next iteration of this can be built
> on "master".
> 
>> V3 removes the explicit initialization of r->repo_settings->fsmonitor in
>> repo-settings.c as requested. It also includes a more detailed commit
>> message for the 3 commit to explain the rationale for putting fsmonitor
>> settings in its own source file rather than adding it repo-settings.c
> 
> This series breaks with the "make check-docs" target, which as an aside
> is broken because of recent changes of mine, so CI didn't catch this
> (I'll submit a series to fix it):
> 
>      config/core.txt:95: error: git-fsmonitor--daemon[1]: link outside of our own docs, shown with 'HERE' below:
>      config/core.txt:95:     'daemon for this working directory (linkgit:git-fsmonitor--daemon[1]' <-- HERE
>      git-update-index.txt:502: error: git-fsmonitor--daemon[1]: link outside of our own docs, shown with 'HERE' below:
>      git-update-index.txt:502:       'linkgit:git-fsmonitor--daemon[1]' <-- HERE
> 
> But that broken-ness points to a more general issue, which is that it's
> not just a broken link, but that docs in this series are referring to a
> manpage that doesn't exist yet.
> 
> I was going to check out some of the semantics of
> core.useBuiltinfsMonitor that I commented on in earlier rounds, but I
> see that there's no tests for it, probably for similar reasons as there
> not being a git-fsmonitor--daemon yet: It's all (hopefully) in some
> yet-to-come series.
> 
> I'm all for this being split up in steps from the 30-something patches
> it was before, but if we're making forward-references to docs that don't
> exist yet, adding ~100 line *.c files that seemingly aren't used at all
> yet etc., the split-up seems to be a bit too aggressive.
> 
> E.g. the below diff seems to have all tests passing, so
> core.usebuiltinfsmonitor is unused still? There's also large amounts of
> code ifdef'd away under HAVE_FSMONITOR_DAEMON_BACKEND, but no other
> in-tree reference to it?
> 
> AFAICT that's going to be used in the future. So seemingly something
> that'll only be used in the series after this? I.e. this seems to have
> stopped at around part 7/28 of a previous submission, but digging in the
> archive e.g. HAVE_FSMONITOR_DAEMON_BACKEND semes to be first used in
> step 11/28 of that[1].
> 
> Sorry to be party pooper again, I really did try to review this, but
> just found that I couldn't, since so much of it seems to be a
> forward-reference to some future state.
> 
> Isn't there a way to re-arrange this so that the "teach daemon XYZ
> command" around steps 13-14/28 of your previous series comes before
> references in the docs to the daemon that doesn't exist yet (5/5 here),
> or there's perhaps some of the config scaffolding, but
> e.g. "core.usebuiltinfsmonitor" comes along with the later change that
> uses it?
> 
> Or just make this "part 2" series larger than 5 patches, so that it
> manages to tell some holistic story, and the various bits are used at
> the end?
> 
> Even so it would be a lot easier to follow if e.g. a patch using a new
> "core.usebuiltinfsmonitor" variable doesn't come at say step 11/20, with
> that variable having been added at a 3/20. I.e. maybe 3/20 can have the
> scaffolding, but the part that adds the "core.usebuiltinfsmonitor"
> support either immediately precedes its use, or is squashed into the
> relevant commit that needs it. Ditto docs for so-far-not-existing things
> etc.
> 
> Thanks! And hopefully this helps.
> 
> 1. https://lore.kernel.org/git/49f9e2e3d49ce6e7b56839bf44535f271216abeb.1621691828.git.gitgitgadget@gmail.com/
> 
> diff --git a/fsmonitor-settings.c b/fsmonitor-settings.c
> index 2770266f5ee..2c479d4f36b 100644
> --- a/fsmonitor-settings.c
> +++ b/fsmonitor-settings.c
> @@ -35,19 +35,6 @@ void fsm_settings__set_disabled(struct repository *r)
>   	FREE_AND_NULL(s->hook_path);
>   }
>   
> -static int check_for_ipc(struct repository *r)
> -{
> -	int value;
> -
> -	if (!repo_config_get_bool(r, "core.usebuiltinfsmonitor", &value) &&
> -	    value) {
> -		fsm_settings__set_ipc(r);
> -		return 1;
> -	}
> -
> -	return 0;
> -}
> -
>   static int check_for_hook(struct repository *r)
>   {
>   	const char *const_str;
> @@ -71,9 +58,6 @@ static void lookup_fsmonitor_settings(struct repository *r)
>   
>   	r->settings.fsmonitor = s;
>   
> -	if (check_for_ipc(r))
> -		return;
> -
>   	if (check_for_hook(r))
>   		return;
>   
> 

Because of the size -- I'm up to ~60 commits now -- I was trying
to break it up into logical pieces.  And yes, there's a bit of a
chicken-n-egg problem with introducing pieces.  I was trying to make
Part 2 be the client side -- everything needed to talk to a daemon
(even if it doesn't exist yet) and then have Part 3 be the daemon
proper (and build it up without worrying about when the corresponding
client peer code appears because the client side is already present).
The daemon in Part 3 would be a little rough around the edges, but
sufficient to do end-to-end testing with both t/ and t/perf tests.
Then Part 4 would clean up some of the details and add more tests.

That was the idea anyway.

I'll take a stab at refactoring this and send a new Part 2 that
contains both the client and MVP daemon parts.  This will let
let basic end-to-end testing happen, fix the doc breakage, and
hopefully make it easier to see the big picture in a single patch
series.  This should be about 30 commits.  Once that has settled
a bit, I'll followup with a new Part 3 that contains the refinements
and additional tests.


Thanks,
Jeff

[PATCH v4 00/29] Builtin FSMonitor Part 2

[Jeff Hostetler via GitGitGadget]

Here is V4 of Part 2 of my Builtin FSMonitor series. I have rebased it onto
the current "master" branch to address conflicts and errors with recent
changes already in master.

In response to comments on V3, I have pulled in more commits from complete
series. Part 2 now includes both the client-side code to talk to via a
daemon via IPC and an MVP implementation of the fsmonitor--daemon itself.

This version of the daemon is functional and can be tested using t/t7527 and
t/perf/p7519.

A followup Part 3 will contain additional refinements to the daemon and
additional tests. I drew the line here between Part 2 and 3 to make it
easier to review.

cc: Bagas Sanjaya bagasdotme@gmail.com cc: Ævar Arnfjörð Bjarmason
avarab@gmail.com cc: Jeff Hostetler git@jeffhostetler.com

Jeff Hostetler (29):
  fsmonitor: enhance existing comments
  fsmonitor-ipc: create client routines for git-fsmonitor--daemon
  fsmonitor: config settings are repository-specific
  fsmonitor: use IPC to query the builtin FSMonitor daemon
  fsmonitor: document builtin fsmonitor
  fsmonitor--daemon: add a built-in fsmonitor daemon
  fsmonitor--daemon: implement 'stop' and 'status' commands
  compat/fsmonitor/fsm-listen-win32: stub in backend for Windows
  compat/fsmonitor/fsm-listen-darwin: stub in backend for Darwin
  fsmonitor--daemon: implement 'run' command
  fsmonitor--daemon: implement 'start' command
  fsmonitor--daemon: add pathname classification
  fsmonitor--daemon: define token-ids
  fsmonitor--daemon: create token-based changed path cache
  compat/fsmonitor/fsm-listen-win32: implement FSMonitor backend on
    Windows
  compat/fsmonitor/fsm-listen-darwin: add macos header files for FSEvent
  compat/fsmonitor/fsm-listen-darwin: implement FSEvent listener on
    MacOS
  fsmonitor--daemon: implement handle_client callback
  help: include fsmonitor--daemon feature flag in version info
  t/helper/fsmonitor-client: create IPC client to talk to FSMonitor
    Daemon
  t7527: create test for fsmonitor--daemon
  t/perf: avoid copying builtin fsmonitor files into test repo
  t/helper/test-chmtime: skip directories on Windows
  t/perf/p7519: speed up test on Windows
  t/perf/p7519: add fsmonitor--daemon test cases
  fsmonitor--daemon: periodically truncate list of modified files
  fsmonitor--daemon: use a cookie file to sync with file system
  fsmonitor: force update index after large responses
  t7527: test status with untracked-cache and fsmonitor--daemon

 .gitignore                              |    1 +
 Documentation/config/core.txt           |   56 +-
 Documentation/git-fsmonitor--daemon.txt |   75 ++
 Documentation/git-update-index.txt      |   27 +-
 Documentation/githooks.txt              |    3 +-
 Makefile                                |   17 +
 builtin.h                               |    1 +
 builtin/fsmonitor--daemon.c             | 1454 +++++++++++++++++++++++
 builtin/update-index.c                  |   19 +-
 cache.h                                 |    1 -
 compat/fsmonitor/fsm-listen-darwin.c    |  496 ++++++++
 compat/fsmonitor/fsm-listen-win32.c     |  586 +++++++++
 compat/fsmonitor/fsm-listen.h           |   49 +
 config.c                                |   14 -
 config.h                                |    1 -
 config.mak.uname                        |   20 +
 contrib/buildsystems/CMakeLists.txt     |   10 +
 environment.c                           |    1 -
 fsmonitor--daemon.h                     |  140 +++
 fsmonitor-ipc.c                         |  176 +++
 fsmonitor-ipc.h                         |   48 +
 fsmonitor-settings.c                    |   97 ++
 fsmonitor-settings.h                    |   21 +
 fsmonitor.c                             |  187 ++-
 fsmonitor.h                             |   18 +-
 git.c                                   |    1 +
 help.c                                  |    4 +
 repo-settings.c                         |    1 +
 repository.h                            |    3 +
 t/README                                |    4 +-
 t/helper/test-chmtime.c                 |   15 +
 t/helper/test-fsmonitor-client.c        |  121 ++
 t/helper/test-tool.c                    |    1 +
 t/helper/test-tool.h                    |    1 +
 t/perf/p7519-fsmonitor.sh               |   61 +-
 t/perf/perf-lib.sh                      |    2 +-
 t/t7527-builtin-fsmonitor.sh            |  607 ++++++++++
 t/test-lib.sh                           |    6 +
 38 files changed, 4247 insertions(+), 98 deletions(-)
 create mode 100644 Documentation/git-fsmonitor--daemon.txt
 create mode 100644 builtin/fsmonitor--daemon.c
 create mode 100644 compat/fsmonitor/fsm-listen-darwin.c
 create mode 100644 compat/fsmonitor/fsm-listen-win32.c
 create mode 100644 compat/fsmonitor/fsm-listen.h
 create mode 100644 fsmonitor--daemon.h
 create mode 100644 fsmonitor-ipc.c
 create mode 100644 fsmonitor-ipc.h
 create mode 100644 fsmonitor-settings.c
 create mode 100644 fsmonitor-settings.h
 create mode 100644 t/helper/test-fsmonitor-client.c
 create mode 100755 t/t7527-builtin-fsmonitor.sh


base-commit: 9d530dc0024503ab4218fe6c4395b8a0aa245478
Published-As: https://github.com/gitgitgadget/git/releases/tag/pr-1041%2Fjeffhostetler%2Fbuiltin-fsmonitor-part2-v4
Fetch-It-Via: git fetch https://github.com/gitgitgadget/git pr-1041/jeffhostetler/builtin-fsmonitor-part2-v4
Pull-Request: https://github.com/gitgitgadget/git/pull/1041

Range-diff vs v3:

  1:  cb25eeaf72d =  1:  ecc40795fa2 fsmonitor: enhance existing comments
  2:  df81a63acee =  2:  82f17692128 fsmonitor-ipc: create client routines for git-fsmonitor--daemon
  3:  a1d606aa622 =  3:  882789b4dfe fsmonitor: config settings are repository-specific
  4:  4d8d812be08 =  4:  de82c726182 fsmonitor: use IPC to query the builtin FSMonitor daemon
  5:  45a86cef8d7 !  5:  d365704d551 fsmonitor: update fsmonitor config documentation
     @@ Metadata
      Author: Jeff Hostetler <jeffhost@microsoft.com>
      
       ## Commit message ##
     -    fsmonitor: update fsmonitor config documentation
     +    fsmonitor: document builtin fsmonitor
     +
     +    Document the new `core.useBuiltinFSMonitor` config value.
      
          Update references to `core.fsmonitor` and `core.fsmonitorHookVersion` and
     -    pointers to `Watchman` to mention the new `core.useBuiltinFSMonitor`
     -    value.
     +    pointers to `Watchman` to refer to it.
     +
     +    Create `git-fsmonitor--daemon` manual page and describe its features.
      
          Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
      
     @@ Documentation/config/core.txt: core.protectNTFS::
       core.trustctime::
       	If false, the ctime differences between the index and the
      
     + ## Documentation/git-fsmonitor--daemon.txt (new) ##
     +@@
     ++git-fsmonitor--daemon(1)
     ++========================
     ++
     ++NAME
     ++----
     ++git-fsmonitor--daemon - A Built-in File System Monitor
     ++
     ++SYNOPSIS
     ++--------
     ++[verse]
     ++'git fsmonitor--daemon' start
     ++'git fsmonitor--daemon' run
     ++'git fsmonitor--daemon' stop
     ++'git fsmonitor--daemon' status
     ++
     ++DESCRIPTION
     ++-----------
     ++
     ++A daemon to watch the working directory for file and directory
     ++changes using platform-specific file system notification facilities.
     ++
     ++This daemon communicates directly with commands like `git status`
     ++using the link:technical/api-simple-ipc.html[simple IPC] interface
     ++instead of the slower linkgit:githooks[5] interface.
     ++
     ++This daemon is built into Git so that no third-party tools are
     ++required.
     ++
     ++OPTIONS
     ++-------
     ++
     ++start::
     ++	Starts a daemon in the background.
     ++
     ++run::
     ++	Runs a daemon in the foreground.
     ++
     ++stop::
     ++	Stops the daemon running in the current working
     ++	directory, if present.
     ++
     ++status::
     ++	Exits with zero status if a daemon is watching the
     ++	current working directory.
     ++
     ++REMARKS
     ++-------
     ++
     ++This daemon is a long running process used to watch a single working
     ++directory and maintain a list of the recently changed files and
     ++directories.  Performance of commands such as `git status` can be
     ++increased if they just ask for a summary of changes to the working
     ++directory and can avoid scanning the disk.
     ++
     ++When `core.useBuiltinFSMonitor` is set to `true` (see
     ++linkgit:git-config[1]) commands, such as `git status`, will ask the
     ++daemon for changes and automatically start it (if necessary).
     ++
     ++For more information see the "File System Monitor" section in
     ++linkgit:git-update-index[1].
     ++
     ++CAVEATS
     ++-------
     ++
     ++The fsmonitor daemon does not currently know about submodules and does
     ++not know to filter out file system events that happen within a
     ++submodule.  If fsmonitor daemon is watching a super repo and a file is
     ++modified within the working directory of a submodule, it will report
     ++the change (as happening against the super repo).  However, the client
     ++will properly ignore these extra events, so performance may be affected
     ++but it will not cause an incorrect result.
     ++
     ++GIT
     ++---
     ++Part of the linkgit:git[1] suite
     +
       ## Documentation/git-update-index.txt ##
      @@ Documentation/git-update-index.txt: FILE SYSTEM MONITOR
       This feature is intended to speed up git operations for repos that have
  -:  ----------- >  6:  78e682fc530 fsmonitor--daemon: add a built-in fsmonitor daemon
  -:  ----------- >  7:  ea64b5c9753 fsmonitor--daemon: implement 'stop' and 'status' commands
  -:  ----------- >  8:  5a40b33a00c compat/fsmonitor/fsm-listen-win32: stub in backend for Windows
  -:  ----------- >  9:  ed5819e29f8 compat/fsmonitor/fsm-listen-darwin: stub in backend for Darwin
  -:  ----------- > 10:  d3ac973a5f1 fsmonitor--daemon: implement 'run' command
  -:  ----------- > 11:  d08c28b549c fsmonitor--daemon: implement 'start' command
  -:  ----------- > 12:  6fa71fdc825 fsmonitor--daemon: add pathname classification
  -:  ----------- > 13:  65821da5b03 fsmonitor--daemon: define token-ids
  -:  ----------- > 14:  429c48a5bad fsmonitor--daemon: create token-based changed path cache
  -:  ----------- > 15:  b04c460c619 compat/fsmonitor/fsm-listen-win32: implement FSMonitor backend on Windows
  -:  ----------- > 16:  862bbfcc32e compat/fsmonitor/fsm-listen-darwin: add macos header files for FSEvent
  -:  ----------- > 17:  40d9a816b52 compat/fsmonitor/fsm-listen-darwin: implement FSEvent listener on MacOS
  -:  ----------- > 18:  241962894f1 fsmonitor--daemon: implement handle_client callback
  -:  ----------- > 19:  704d37d2033 help: include fsmonitor--daemon feature flag in version info
  -:  ----------- > 20:  de6c72a9ce0 t/helper/fsmonitor-client: create IPC client to talk to FSMonitor Daemon
  -:  ----------- > 21:  eedaa787c2e t7527: create test for fsmonitor--daemon
  -:  ----------- > 22:  4e96e0667ba t/perf: avoid copying builtin fsmonitor files into test repo
  -:  ----------- > 23:  de9c015d78c t/helper/test-chmtime: skip directories on Windows
  -:  ----------- > 24:  1c2eccacff6 t/perf/p7519: speed up test on Windows
  -:  ----------- > 25:  236b5966257 t/perf/p7519: add fsmonitor--daemon test cases
  -:  ----------- > 26:  54710a4830d fsmonitor--daemon: periodically truncate list of modified files
  -:  ----------- > 27:  1e2bd77fcea fsmonitor--daemon: use a cookie file to sync with file system
  -:  ----------- > 28:  30e61b6d1ad fsmonitor: force update index after large responses
  -:  ----------- > 29:  507020bbef0 t7527: test status with untracked-cache and fsmonitor--daemon

-- 
gitgitgadget

[PATCH v4 01/29] fsmonitor: enhance existing comments

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 fsmonitor.c | 37 ++++++++++++++++++++++++++++++-------
 1 file changed, 30 insertions(+), 7 deletions(-)

diff --git a/fsmonitor.c b/fsmonitor.c
index ab9bfc60b34..ec4c46407c5 100644
--- a/fsmonitor.c
+++ b/fsmonitor.c
@@ -301,9 +301,25 @@ void refresh_fsmonitor(struct index_state *istate)
 			core_fsmonitor, query_success ? "success" : "failure");
 	}
 
-	/* a fsmonitor process can return '/' to indicate all entries are invalid */
+	/*
+	 * The response from FSMonitor (excluding the header token) is
+	 * either:
+	 *
+	 * [a] a (possibly empty) list of NUL delimited relative
+	 *     pathnames of changed paths.  This list can contain
+	 *     files and directories.  Directories have a trailing
+	 *     slash.
+	 *
+	 * [b] a single '/' to indicate the provider had no
+	 *     information and that we should consider everything
+	 *     invalid.  We call this a trivial response.
+	 */
 	if (query_success && query_result.buf[bol] != '/') {
-		/* Mark all entries returned by the monitor as dirty */
+		/*
+		 * Mark all pathnames returned by the monitor as dirty.
+		 *
+		 * This updates both the cache-entries and the untracked-cache.
+		 */
 		buf = query_result.buf;
 		for (i = bol; i < query_result.len; i++) {
 			if (buf[i] != '\0')
@@ -318,11 +334,15 @@ void refresh_fsmonitor(struct index_state *istate)
 		if (istate->untracked)
 			istate->untracked->use_fsmonitor = 1;
 	} else {
-
-		/* We only want to run the post index changed hook if we've actually changed entries, so keep track
-		 * if we actually changed entries or not */
+		/*
+		 * We received a trivial response, so invalidate everything.
+		 *
+		 * We only want to run the post index changed hook if
+		 * we've actually changed entries, so keep track if we
+		 * actually changed entries or not.
+		 */
 		int is_cache_changed = 0;
-		/* Mark all entries invalid */
+
 		for (i = 0; i < istate->cache_nr; i++) {
 			if (istate->cache[i]->ce_flags & CE_FSMONITOR_VALID) {
 				is_cache_changed = 1;
@@ -330,7 +350,10 @@ void refresh_fsmonitor(struct index_state *istate)
 			}
 		}
 
-		/* If we're going to check every file, ensure we save the results */
+		/*
+		 * If we're going to check every file, ensure we save
+		 * the results.
+		 */
 		if (is_cache_changed)
 			istate->cache_changed |= FSMONITOR_CHANGED;
 
-- 
gitgitgadget

[PATCH v4 02/29] fsmonitor-ipc: create client routines for git-fsmonitor--daemon

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Create fsmonitor_ipc__*() client routines to spawn the built-in file
system monitor daemon and send it an IPC request using the `Simple
IPC` API.

Stub in empty fsmonitor_ipc__*() functions for unsupported platforms.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 Makefile        |   1 +
 fsmonitor-ipc.c | 176 ++++++++++++++++++++++++++++++++++++++++++++++++
 fsmonitor-ipc.h |  48 +++++++++++++
 3 files changed, 225 insertions(+)
 create mode 100644 fsmonitor-ipc.c
 create mode 100644 fsmonitor-ipc.h

diff --git a/Makefile b/Makefile
index 381bed2c1d2..d51fd8b33ce 100644
--- a/Makefile
+++ b/Makefile
@@ -897,6 +897,7 @@ LIB_OBJS += fetch-pack.o
 LIB_OBJS += fmt-merge-msg.o
 LIB_OBJS += fsck.o
 LIB_OBJS += fsmonitor.o
+LIB_OBJS += fsmonitor-ipc.o
 LIB_OBJS += gettext.o
 LIB_OBJS += gpg-interface.o
 LIB_OBJS += graph.o
diff --git a/fsmonitor-ipc.c b/fsmonitor-ipc.c
new file mode 100644
index 00000000000..ccc32d2a17e
--- /dev/null
+++ b/fsmonitor-ipc.c
@@ -0,0 +1,176 @@
+#include "cache.h"
+#include "fsmonitor.h"
+#include "simple-ipc.h"
+#include "fsmonitor-ipc.h"
+#include "run-command.h"
+#include "strbuf.h"
+#include "trace2.h"
+
+#ifdef HAVE_FSMONITOR_DAEMON_BACKEND
+
+int fsmonitor_ipc__is_supported(void)
+{
+	return 1;
+}
+
+GIT_PATH_FUNC(fsmonitor_ipc__get_path, "fsmonitor--daemon.ipc")
+
+enum ipc_active_state fsmonitor_ipc__get_state(void)
+{
+	return ipc_get_active_state(fsmonitor_ipc__get_path());
+}
+
+static int spawn_daemon(void)
+{
+	const char *args[] = { "fsmonitor--daemon", "start", NULL };
+
+	return run_command_v_opt_tr2(args, RUN_COMMAND_NO_STDIN | RUN_GIT_CMD,
+				    "fsmonitor");
+}
+
+int fsmonitor_ipc__send_query(const char *since_token,
+			      struct strbuf *answer)
+{
+	int ret = -1;
+	int tried_to_spawn = 0;
+	enum ipc_active_state state = IPC_STATE__OTHER_ERROR;
+	struct ipc_client_connection *connection = NULL;
+	struct ipc_client_connect_options options
+		= IPC_CLIENT_CONNECT_OPTIONS_INIT;
+	const char *tok = since_token ? since_token : "";
+	size_t tok_len = since_token ? strlen(since_token) : 0;
+
+	options.wait_if_busy = 1;
+	options.wait_if_not_found = 0;
+
+	trace2_region_enter("fsm_client", "query", NULL);
+	trace2_data_string("fsm_client", NULL, "query/command", tok);
+
+try_again:
+	state = ipc_client_try_connect(fsmonitor_ipc__get_path(), &options,
+				       &connection);
+
+	switch (state) {
+	case IPC_STATE__LISTENING:
+		ret = ipc_client_send_command_to_connection(
+			connection, tok, tok_len, answer);
+		ipc_client_close_connection(connection);
+
+		trace2_data_intmax("fsm_client", NULL,
+				   "query/response-length", answer->len);
+
+		if (fsmonitor_is_trivial_response(answer))
+			trace2_data_intmax("fsm_client", NULL,
+					   "query/trivial-response", 1);
+
+		goto done;
+
+	case IPC_STATE__NOT_LISTENING:
+	case IPC_STATE__PATH_NOT_FOUND:
+		if (tried_to_spawn)
+			goto done;
+
+		tried_to_spawn++;
+		if (spawn_daemon())
+			goto done;
+
+		/*
+		 * Try again, but this time give the daemon a chance to
+		 * actually create the pipe/socket.
+		 *
+		 * Granted, the daemon just started so it can't possibly have
+		 * any FS cached yet, so we'll always get a trivial answer.
+		 * BUT the answer should include a new token that can serve
+		 * as the basis for subsequent requests.
+		 */
+		options.wait_if_not_found = 1;
+		goto try_again;
+
+	case IPC_STATE__INVALID_PATH:
+		ret = error(_("fsmonitor_ipc__send_query: invalid path '%s'"),
+			    fsmonitor_ipc__get_path());
+		goto done;
+
+	case IPC_STATE__OTHER_ERROR:
+	default:
+		ret = error(_("fsmonitor_ipc__send_query: unspecified error on '%s'"),
+			    fsmonitor_ipc__get_path());
+		goto done;
+	}
+
+done:
+	trace2_region_leave("fsm_client", "query", NULL);
+
+	return ret;
+}
+
+int fsmonitor_ipc__send_command(const char *command,
+				struct strbuf *answer)
+{
+	struct ipc_client_connection *connection = NULL;
+	struct ipc_client_connect_options options
+		= IPC_CLIENT_CONNECT_OPTIONS_INIT;
+	int ret;
+	enum ipc_active_state state;
+	const char *c = command ? command : "";
+	size_t c_len = command ? strlen(command) : 0;
+
+	strbuf_reset(answer);
+
+	options.wait_if_busy = 1;
+	options.wait_if_not_found = 0;
+
+	state = ipc_client_try_connect(fsmonitor_ipc__get_path(), &options,
+				       &connection);
+	if (state != IPC_STATE__LISTENING) {
+		die("fsmonitor--daemon is not running");
+		return -1;
+	}
+
+	ret = ipc_client_send_command_to_connection(connection, c, c_len,
+						    answer);
+	ipc_client_close_connection(connection);
+
+	if (ret == -1) {
+		die("could not send '%s' command to fsmonitor--daemon", c);
+		return -1;
+	}
+
+	return 0;
+}
+
+#else
+
+/*
+ * A trivial implementation of the fsmonitor_ipc__ API for unsupported
+ * platforms.
+ */
+
+int fsmonitor_ipc__is_supported(void)
+{
+	return 0;
+}
+
+const char *fsmonitor_ipc__get_path(void)
+{
+	return NULL;
+}
+
+enum ipc_active_state fsmonitor_ipc__get_state(void)
+{
+	return IPC_STATE__OTHER_ERROR;
+}
+
+int fsmonitor_ipc__send_query(const char *since_token,
+			      struct strbuf *answer)
+{
+	return -1;
+}
+
+int fsmonitor_ipc__send_command(const char *command,
+				struct strbuf *answer)
+{
+	return -1;
+}
+
+#endif
diff --git a/fsmonitor-ipc.h b/fsmonitor-ipc.h
new file mode 100644
index 00000000000..b6a7067c3af
--- /dev/null
+++ b/fsmonitor-ipc.h
@@ -0,0 +1,48 @@
+#ifndef FSMONITOR_IPC_H
+#define FSMONITOR_IPC_H
+
+#include "simple-ipc.h"
+
+/*
+ * Returns true if built-in file system monitor daemon is defined
+ * for this platform.
+ */
+int fsmonitor_ipc__is_supported(void);
+
+/*
+ * Returns the pathname to the IPC named pipe or Unix domain socket
+ * where a `git-fsmonitor--daemon` process will listen.  This is a
+ * per-worktree value.
+ *
+ * Returns NULL if the daemon is not supported on this platform.
+ */
+const char *fsmonitor_ipc__get_path(void);
+
+/*
+ * Try to determine whether there is a `git-fsmonitor--daemon` process
+ * listening on the IPC pipe/socket.
+ */
+enum ipc_active_state fsmonitor_ipc__get_state(void);
+
+/*
+ * Connect to a `git-fsmonitor--daemon` process via simple-ipc
+ * and ask for the set of changed files since the given token.
+ *
+ * Spawn a daemon process in the background if necessary.
+ *
+ * Returns -1 on error; 0 on success.
+ */
+int fsmonitor_ipc__send_query(const char *since_token,
+			      struct strbuf *answer);
+
+/*
+ * Connect to a `git-fsmonitor--daemon` process via simple-ipc and
+ * send a command verb.  If no daemon is available, we DO NOT try to
+ * start one.
+ *
+ * Returns -1 on error; 0 on success.
+ */
+int fsmonitor_ipc__send_command(const char *command,
+				struct strbuf *answer);
+
+#endif /* FSMONITOR_IPC_H */
-- 
gitgitgadget

[PATCH v4 03/29] fsmonitor: config settings are repository-specific

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Move fsmonitor config settings to a new and opaque
`struct fsmonitor_settings` structure.  Add a lazily-loaded pointer
to this into `struct repo_settings`

Create an `enum fsmonitor_mode` type in `struct fsmonitor_settings` to
represent the state of fsmonitor.  This lets us represent which, if
any, fsmonitor provider (hook or IPC) is enabled.

Create `fsm_settings__get_*()` getters to lazily look up fsmonitor-
related config settings.

Add support for the new `core.useBuiltinFSMonitor` config setting.

Get rid of the `core_fsmonitor` global variable.  Move the code to
lookup the existing `core.fsmonitor` config value into the fsmonitor
settings.

Create a hook pathname variable in `struct fsmonitor-settings` and
only set it when in hook mode.

The existing `core_fsmonitor` global variable was used to store the
pathname to the fsmonitor hook *and* it was used as a boolean to see
if fsmonitor was enabled.  This dual usage and global visibility leads
to confusion when we add the IPC-based provider.  So lets hide the
details in fsmonitor-settings.c and let it decide which provider to
use in the case of multiple settings.  This avoids cluttering up
repo-settings.c with these private details.

A future commit in builtin-fsmonitor series will add the ability to
disqualify worktrees for various reasons, such as being mounted from a
remote volume, where fsmonitor should not be started.  Having the
config settings hidden in fsmonitor-settings.c allows such worktree
restrictions to override the config values used.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 Makefile               |  1 +
 builtin/update-index.c | 19 +++++++--
 cache.h                |  1 -
 config.c               | 14 ------
 config.h               |  1 -
 environment.c          |  1 -
 fsmonitor-settings.c   | 97 ++++++++++++++++++++++++++++++++++++++++++
 fsmonitor-settings.h   | 21 +++++++++
 fsmonitor.c            | 63 ++++++++++++++++-----------
 fsmonitor.h            | 18 ++++++--
 repository.h           |  3 ++
 t/README               |  4 +-
 12 files changed, 192 insertions(+), 51 deletions(-)
 create mode 100644 fsmonitor-settings.c
 create mode 100644 fsmonitor-settings.h

diff --git a/Makefile b/Makefile
index d51fd8b33ce..29ed7c4aba6 100644
--- a/Makefile
+++ b/Makefile
@@ -898,6 +898,7 @@ LIB_OBJS += fmt-merge-msg.o
 LIB_OBJS += fsck.o
 LIB_OBJS += fsmonitor.o
 LIB_OBJS += fsmonitor-ipc.o
+LIB_OBJS += fsmonitor-settings.o
 LIB_OBJS += gettext.o
 LIB_OBJS += gpg-interface.o
 LIB_OBJS += graph.o
diff --git a/builtin/update-index.c b/builtin/update-index.c
index 187203e8bb5..79db3ff37e2 100644
--- a/builtin/update-index.c
+++ b/builtin/update-index.c
@@ -1214,14 +1214,25 @@ int cmd_update_index(int argc, const char **argv, const char *prefix)
 	}
 
 	if (fsmonitor > 0) {
-		if (git_config_get_fsmonitor() == 0)
+		enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
+
+		if (fsm_mode == FSMONITOR_MODE_DISABLED) {
+			warning(_("core.useBuiltinFSMonitor is unset; "
+				"set it if you really want to enable the "
+				"builtin fsmonitor"));
 			warning(_("core.fsmonitor is unset; "
-				"set it if you really want to "
-				"enable fsmonitor"));
+				"set it if you really want to enable the "
+				"hook-based fsmonitor"));
+		}
 		add_fsmonitor(&the_index);
 		report(_("fsmonitor enabled"));
 	} else if (!fsmonitor) {
-		if (git_config_get_fsmonitor() == 1)
+		enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
+		if (fsm_mode == FSMONITOR_MODE_IPC)
+			warning(_("core.useBuiltinFSMonitor is set; "
+				"remove it if you really want to "
+				"disable fsmonitor"));
+		if (fsm_mode == FSMONITOR_MODE_HOOK)
 			warning(_("core.fsmonitor is set; "
 				"remove it if you really want to "
 				"disable fsmonitor"));
diff --git a/cache.h b/cache.h
index d092820c943..8f4e3c8bd1d 100644
--- a/cache.h
+++ b/cache.h
@@ -989,7 +989,6 @@ extern int core_preload_index;
 extern int precomposed_unicode;
 extern int protect_hfs;
 extern int protect_ntfs;
-extern const char *core_fsmonitor;
 
 extern int core_apply_sparse_checkout;
 extern int core_sparse_checkout_cone;
diff --git a/config.c b/config.c
index 2dcbe901b6b..6b6e9cacac3 100644
--- a/config.c
+++ b/config.c
@@ -2502,20 +2502,6 @@ int git_config_get_max_percent_split_change(void)
 	return -1; /* default value */
 }
 
-int git_config_get_fsmonitor(void)
-{
-	if (git_config_get_pathname("core.fsmonitor", &core_fsmonitor))
-		core_fsmonitor = getenv("GIT_TEST_FSMONITOR");
-
-	if (core_fsmonitor && !*core_fsmonitor)
-		core_fsmonitor = NULL;
-
-	if (core_fsmonitor)
-		return 1;
-
-	return 0;
-}
-
 int git_config_get_index_threads(int *dest)
 {
 	int is_bool, val;
diff --git a/config.h b/config.h
index f119de01309..69d733824a0 100644
--- a/config.h
+++ b/config.h
@@ -610,7 +610,6 @@ int git_config_get_pathname(const char *key, const char **dest);
 int git_config_get_index_threads(int *dest);
 int git_config_get_split_index(void);
 int git_config_get_max_percent_split_change(void);
-int git_config_get_fsmonitor(void);
 
 /* This dies if the configured or default date is in the future */
 int git_config_get_expiry(const char *key, const char **output);
diff --git a/environment.c b/environment.c
index 9da7f3c1a19..68f90632245 100644
--- a/environment.c
+++ b/environment.c
@@ -82,7 +82,6 @@ int protect_hfs = PROTECT_HFS_DEFAULT;
 #define PROTECT_NTFS_DEFAULT 1
 #endif
 int protect_ntfs = PROTECT_NTFS_DEFAULT;
-const char *core_fsmonitor;
 
 /*
  * The character that begins a commented line in user-editable file
diff --git a/fsmonitor-settings.c b/fsmonitor-settings.c
new file mode 100644
index 00000000000..2770266f5ee
--- /dev/null
+++ b/fsmonitor-settings.c
@@ -0,0 +1,97 @@
+#include "cache.h"
+#include "config.h"
+#include "repository.h"
+#include "fsmonitor-settings.h"
+
+/*
+ * We keep this structure defintion private and have getters
+ * for all fields so that we can lazy load it as needed.
+ */
+struct fsmonitor_settings {
+	enum fsmonitor_mode mode;
+	char *hook_path;
+};
+
+void fsm_settings__set_ipc(struct repository *r)
+{
+	struct fsmonitor_settings *s = r->settings.fsmonitor;
+
+	s->mode = FSMONITOR_MODE_IPC;
+}
+
+void fsm_settings__set_hook(struct repository *r, const char *path)
+{
+	struct fsmonitor_settings *s = r->settings.fsmonitor;
+
+	s->mode = FSMONITOR_MODE_HOOK;
+	s->hook_path = strdup(path);
+}
+
+void fsm_settings__set_disabled(struct repository *r)
+{
+	struct fsmonitor_settings *s = r->settings.fsmonitor;
+
+	s->mode = FSMONITOR_MODE_DISABLED;
+	FREE_AND_NULL(s->hook_path);
+}
+
+static int check_for_ipc(struct repository *r)
+{
+	int value;
+
+	if (!repo_config_get_bool(r, "core.usebuiltinfsmonitor", &value) &&
+	    value) {
+		fsm_settings__set_ipc(r);
+		return 1;
+	}
+
+	return 0;
+}
+
+static int check_for_hook(struct repository *r)
+{
+	const char *const_str;
+
+	if (repo_config_get_pathname(r, "core.fsmonitor", &const_str))
+		const_str = getenv("GIT_TEST_FSMONITOR");
+
+	if (const_str && *const_str) {
+		fsm_settings__set_hook(r, const_str);
+		return 1;
+	}
+
+	return 0;
+}
+
+static void lookup_fsmonitor_settings(struct repository *r)
+{
+	struct fsmonitor_settings *s;
+
+	CALLOC_ARRAY(s, 1);
+
+	r->settings.fsmonitor = s;
+
+	if (check_for_ipc(r))
+		return;
+
+	if (check_for_hook(r))
+		return;
+
+	fsm_settings__set_disabled(r);
+}
+
+enum fsmonitor_mode fsm_settings__get_mode(struct repository *r)
+{
+	if (!r->settings.fsmonitor)
+		lookup_fsmonitor_settings(r);
+
+	return r->settings.fsmonitor->mode;
+}
+
+const char *fsm_settings__get_hook_path(struct repository *r)
+{
+	if (!r->settings.fsmonitor)
+		lookup_fsmonitor_settings(r);
+
+	return r->settings.fsmonitor->hook_path;
+}
diff --git a/fsmonitor-settings.h b/fsmonitor-settings.h
new file mode 100644
index 00000000000..50b29234616
--- /dev/null
+++ b/fsmonitor-settings.h
@@ -0,0 +1,21 @@
+#ifndef FSMONITOR_SETTINGS_H
+#define FSMONITOR_SETTINGS_H
+
+struct repository;
+
+enum fsmonitor_mode {
+	FSMONITOR_MODE_DISABLED = 0,
+	FSMONITOR_MODE_HOOK = 1, /* core.fsmonitor */
+	FSMONITOR_MODE_IPC = 2,  /* core.useBuiltinFSMonitor */
+};
+
+void fsm_settings__set_ipc(struct repository *r);
+void fsm_settings__set_hook(struct repository *r, const char *path);
+void fsm_settings__set_disabled(struct repository *r);
+
+enum fsmonitor_mode fsm_settings__get_mode(struct repository *r);
+const char *fsm_settings__get_hook_path(struct repository *r);
+
+struct fsmonitor_settings;
+
+#endif /* FSMONITOR_SETTINGS_H */
diff --git a/fsmonitor.c b/fsmonitor.c
index ec4c46407c5..63174630c0e 100644
--- a/fsmonitor.c
+++ b/fsmonitor.c
@@ -3,6 +3,7 @@
 #include "dir.h"
 #include "ewah/ewok.h"
 #include "fsmonitor.h"
+#include "fsmonitor-ipc.h"
 #include "run-command.h"
 #include "strbuf.h"
 
@@ -148,15 +149,18 @@ void write_fsmonitor_extension(struct strbuf *sb, struct index_state *istate)
 /*
  * Call the query-fsmonitor hook passing the last update token of the saved results.
  */
-static int query_fsmonitor(int version, const char *last_update, struct strbuf *query_result)
+static int query_fsmonitor_hook(struct repository *r,
+				int version,
+				const char *last_update,
+				struct strbuf *query_result)
 {
 	struct child_process cp = CHILD_PROCESS_INIT;
 	int result;
 
-	if (!core_fsmonitor)
+	if (fsm_settings__get_mode(r) != FSMONITOR_MODE_HOOK)
 		return -1;
 
-	strvec_push(&cp.args, core_fsmonitor);
+	strvec_push(&cp.args, fsm_settings__get_hook_path(r));
 	strvec_pushf(&cp.args, "%d", version);
 	strvec_pushf(&cp.args, "%s", last_update);
 	cp.use_shell = 1;
@@ -238,17 +242,28 @@ void refresh_fsmonitor(struct index_state *istate)
 	struct strbuf last_update_token = STRBUF_INIT;
 	char *buf;
 	unsigned int i;
+	struct repository *r = istate->repo ? istate->repo : the_repository;
+	enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
 
-	if (!core_fsmonitor || istate->fsmonitor_has_run_once)
+	if (fsm_mode <= FSMONITOR_MODE_DISABLED ||
+	    istate->fsmonitor_has_run_once)
 		return;
 
-	hook_version = fsmonitor_hook_version();
-
 	istate->fsmonitor_has_run_once = 1;
 
 	trace_printf_key(&trace_fsmonitor, "refresh fsmonitor");
+
+	if (fsm_mode == FSMONITOR_MODE_IPC) {
+		/* TODO */
+		return;
+	}
+
+	assert(fsm_mode == FSMONITOR_MODE_HOOK);
+
+	hook_version = fsmonitor_hook_version();
+
 	/*
-	 * This could be racy so save the date/time now and query_fsmonitor
+	 * This could be racy so save the date/time now and query_fsmonitor_hook
 	 * should be inclusive to ensure we don't miss potential changes.
 	 */
 	last_update = getnanotime();
@@ -256,13 +271,14 @@ void refresh_fsmonitor(struct index_state *istate)
 		strbuf_addf(&last_update_token, "%"PRIu64"", last_update);
 
 	/*
-	 * If we have a last update token, call query_fsmonitor for the set of
+	 * If we have a last update token, call query_fsmonitor_hook for the set of
 	 * changes since that token, else assume everything is possibly dirty
 	 * and check it all.
 	 */
 	if (istate->fsmonitor_last_update) {
 		if (hook_version == -1 || hook_version == HOOK_INTERFACE_VERSION2) {
-			query_success = !query_fsmonitor(HOOK_INTERFACE_VERSION2,
+			query_success = !query_fsmonitor_hook(
+				r, HOOK_INTERFACE_VERSION2,
 				istate->fsmonitor_last_update, &query_result);
 
 			if (query_success) {
@@ -292,13 +308,17 @@ void refresh_fsmonitor(struct index_state *istate)
 		}
 
 		if (hook_version == HOOK_INTERFACE_VERSION1) {
-			query_success = !query_fsmonitor(HOOK_INTERFACE_VERSION1,
+			query_success = !query_fsmonitor_hook(
+				r, HOOK_INTERFACE_VERSION1,
 				istate->fsmonitor_last_update, &query_result);
 		}
 
-		trace_performance_since(last_update, "fsmonitor process '%s'", core_fsmonitor);
-		trace_printf_key(&trace_fsmonitor, "fsmonitor process '%s' returned %s",
-			core_fsmonitor, query_success ? "success" : "failure");
+		trace_performance_since(last_update, "fsmonitor process '%s'",
+					fsm_settings__get_hook_path(r));
+		trace_printf_key(&trace_fsmonitor,
+				 "fsmonitor process '%s' returned %s",
+				 fsm_settings__get_hook_path(r),
+				 query_success ? "success" : "failure");
 	}
 
 	/*
@@ -434,7 +454,8 @@ void remove_fsmonitor(struct index_state *istate)
 void tweak_fsmonitor(struct index_state *istate)
 {
 	unsigned int i;
-	int fsmonitor_enabled = git_config_get_fsmonitor();
+	struct repository *r = istate->repo ? istate->repo : the_repository;
+	int fsmonitor_enabled = (fsm_settings__get_mode(r) > FSMONITOR_MODE_DISABLED);
 
 	if (istate->fsmonitor_dirty) {
 		if (fsmonitor_enabled) {
@@ -454,16 +475,8 @@ void tweak_fsmonitor(struct index_state *istate)
 		istate->fsmonitor_dirty = NULL;
 	}
 
-	switch (fsmonitor_enabled) {
-	case -1: /* keep: do nothing */
-		break;
-	case 0: /* false */
-		remove_fsmonitor(istate);
-		break;
-	case 1: /* true */
+	if (fsmonitor_enabled)
 		add_fsmonitor(istate);
-		break;
-	default: /* unknown value: do nothing */
-		break;
-	}
+	else
+		remove_fsmonitor(istate);
 }
diff --git a/fsmonitor.h b/fsmonitor.h
index f20d72631d7..f9201411aa7 100644
--- a/fsmonitor.h
+++ b/fsmonitor.h
@@ -3,6 +3,7 @@
 
 #include "cache.h"
 #include "dir.h"
+#include "fsmonitor-settings.h"
 
 extern struct trace_key trace_fsmonitor;
 
@@ -57,7 +58,11 @@ int fsmonitor_is_trivial_response(const struct strbuf *query_result);
  */
 static inline int is_fsmonitor_refreshed(const struct index_state *istate)
 {
-	return !core_fsmonitor || istate->fsmonitor_has_run_once;
+	struct repository *r = istate->repo ? istate->repo : the_repository;
+	enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
+
+	return fsm_mode <= FSMONITOR_MODE_DISABLED ||
+		istate->fsmonitor_has_run_once;
 }
 
 /*
@@ -67,7 +72,11 @@ static inline int is_fsmonitor_refreshed(const struct index_state *istate)
  */
 static inline void mark_fsmonitor_valid(struct index_state *istate, struct cache_entry *ce)
 {
-	if (core_fsmonitor && !(ce->ce_flags & CE_FSMONITOR_VALID)) {
+	struct repository *r = istate->repo ? istate->repo : the_repository;
+	enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
+
+	if (fsm_mode > FSMONITOR_MODE_DISABLED &&
+	    !(ce->ce_flags & CE_FSMONITOR_VALID)) {
 		istate->cache_changed = 1;
 		ce->ce_flags |= CE_FSMONITOR_VALID;
 		trace_printf_key(&trace_fsmonitor, "mark_fsmonitor_clean '%s'", ce->name);
@@ -83,7 +92,10 @@ static inline void mark_fsmonitor_valid(struct index_state *istate, struct cache
  */
 static inline void mark_fsmonitor_invalid(struct index_state *istate, struct cache_entry *ce)
 {
-	if (core_fsmonitor) {
+	struct repository *r = istate->repo ? istate->repo : the_repository;
+	enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
+
+	if (fsm_mode > FSMONITOR_MODE_DISABLED) {
 		ce->ce_flags &= ~CE_FSMONITOR_VALID;
 		untracked_cache_invalidate_path(istate, ce->name, 1);
 		trace_printf_key(&trace_fsmonitor, "mark_fsmonitor_invalid '%s'", ce->name);
diff --git a/repository.h b/repository.h
index a057653981c..89a1873ade7 100644
--- a/repository.h
+++ b/repository.h
@@ -4,6 +4,7 @@
 #include "path.h"
 
 struct config_set;
+struct fsmonitor_settings;
 struct git_hash_algo;
 struct index_state;
 struct lock_file;
@@ -34,6 +35,8 @@ struct repo_settings {
 	int command_requires_full_index;
 	int sparse_index;
 
+	struct fsmonitor_settings *fsmonitor; /* lazy loaded */
+
 	int index_version;
 	enum untracked_cache_setting core_untracked_cache;
 
diff --git a/t/README b/t/README
index b92155a822e..6dc4a1d10cf 100644
--- a/t/README
+++ b/t/README
@@ -405,8 +405,8 @@ every 'git commit-graph write', as if the `--changed-paths` option was
 passed in.
 
 GIT_TEST_FSMONITOR=$PWD/t7519/fsmonitor-all exercises the fsmonitor
-code path for utilizing a file system monitor to speed up detecting
-new or changed files.
+code path for utilizing a (hook based) file system monitor to speed up
+detecting new or changed files.
 
 GIT_TEST_INDEX_VERSION=<n> exercises the index read/write code path
 for the index version specified.  Can be set to any valid version
-- 
gitgitgadget

[PATCH v4 04/29] fsmonitor: use IPC to query the builtin FSMonitor daemon

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Use simple IPC to directly communicate with the new builtin file
system monitor daemon when `core.useBuiltinFSMonitor` is set.

The `core.fsmonitor` setting has already been defined as a HOOK
pathname.  Historically, this has been set to a HOOK script that will
talk with Watchman.  For compatibility reasons, we do not want to
overload that definition (and cause problems if users have multiple
versions of Git installed).

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 fsmonitor.c | 34 ++++++++++++++++++++++++++++++++--
 1 file changed, 32 insertions(+), 2 deletions(-)

diff --git a/fsmonitor.c b/fsmonitor.c
index 63174630c0e..695fb0ce4e7 100644
--- a/fsmonitor.c
+++ b/fsmonitor.c
@@ -254,8 +254,37 @@ void refresh_fsmonitor(struct index_state *istate)
 	trace_printf_key(&trace_fsmonitor, "refresh fsmonitor");
 
 	if (fsm_mode == FSMONITOR_MODE_IPC) {
-		/* TODO */
-		return;
+		query_success = !fsmonitor_ipc__send_query(
+			istate->fsmonitor_last_update ?
+			istate->fsmonitor_last_update : "builtin:fake",
+			&query_result);
+		if (query_success) {
+			/*
+			 * The response contains a series of nul terminated
+			 * strings.  The first is the new token.
+			 *
+			 * Use `char *buf` as an interlude to trick the CI
+			 * static analysis to let us use `strbuf_addstr()`
+			 * here (and only copy the token) rather than
+			 * `strbuf_addbuf()`.
+			 */
+			buf = query_result.buf;
+			strbuf_addstr(&last_update_token, buf);
+			bol = last_update_token.len + 1;
+		} else {
+			/*
+			 * The builtin daemon is not available on this
+			 * platform -OR- we failed to get a response.
+			 *
+			 * Generate a fake token (rather than a V1
+			 * timestamp) for the index extension.  (If
+			 * they switch back to the hook API, we don't
+			 * want ambiguous state.)
+			 */
+			strbuf_addstr(&last_update_token, "builtin:fake");
+		}
+
+		goto apply_results;
 	}
 
 	assert(fsm_mode == FSMONITOR_MODE_HOOK);
@@ -321,6 +350,7 @@ void refresh_fsmonitor(struct index_state *istate)
 				 query_success ? "success" : "failure");
 	}
 
+apply_results:
 	/*
 	 * The response from FSMonitor (excluding the header token) is
 	 * either:
-- 
gitgitgadget

[PATCH v4 05/29] fsmonitor: document builtin fsmonitor

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Document the new `core.useBuiltinFSMonitor` config value.

Update references to `core.fsmonitor` and `core.fsmonitorHookVersion` and
pointers to `Watchman` to refer to it.

Create `git-fsmonitor--daemon` manual page and describe its features.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 Documentation/config/core.txt           | 56 +++++++++++++-----
 Documentation/git-fsmonitor--daemon.txt | 75 +++++++++++++++++++++++++
 Documentation/git-update-index.txt      | 27 +++++----
 Documentation/githooks.txt              |  3 +-
 4 files changed, 134 insertions(+), 27 deletions(-)
 create mode 100644 Documentation/git-fsmonitor--daemon.txt

diff --git a/Documentation/config/core.txt b/Documentation/config/core.txt
index c04f62a54a1..4f6e519bc02 100644
--- a/Documentation/config/core.txt
+++ b/Documentation/config/core.txt
@@ -62,22 +62,50 @@ core.protectNTFS::
 	Defaults to `true` on Windows, and `false` elsewhere.
 
 core.fsmonitor::
-	If set, the value of this variable is used as a command which
-	will identify all files that may have changed since the
-	requested date/time. This information is used to speed up git by
-	avoiding unnecessary processing of files that have not changed.
-	See the "fsmonitor-watchman" section of linkgit:githooks[5].
+	If set, this variable contains the pathname of the "fsmonitor"
+	hook command.
++
+This hook command is used to identify all files that may have changed
+since the requested date/time. This information is used to speed up
+git by avoiding unnecessary scanning of files that have not changed.
++
+See the "fsmonitor-watchman" section of linkgit:githooks[5].
++
+Note: The value of this config setting is ignored if the
+built-in file system monitor is enabled (see `core.useBuiltinFSMonitor`).
 
 core.fsmonitorHookVersion::
-	Sets the version of hook that is to be used when calling fsmonitor.
-	There are currently versions 1 and 2. When this is not set,
-	version 2 will be tried first and if it fails then version 1
-	will be tried. Version 1 uses a timestamp as input to determine
-	which files have changes since that time but some monitors
-	like watchman have race conditions when used with a timestamp.
-	Version 2 uses an opaque string so that the monitor can return
-	something that can be used to determine what files have changed
-	without race conditions.
+	Sets the protocol version to be used when invoking the
+	"fsmonitor" hook.
++
+There are currently versions 1 and 2. When this is not set,
+version 2 will be tried first and if it fails then version 1
+will be tried. Version 1 uses a timestamp as input to determine
+which files have changes since that time but some monitors
+like Watchman have race conditions when used with a timestamp.
+Version 2 uses an opaque string so that the monitor can return
+something that can be used to determine what files have changed
+without race conditions.
++
+Note: The value of this config setting is ignored if the
+built-in file system monitor is enabled (see `core.useBuiltinFSMonitor`).
+
+core.useBuiltinFSMonitor::
+	If set to true, enable the built-in file system monitor
+	daemon for this working directory (linkgit:git-fsmonitor--daemon[1]).
++
+Like hook-based file system monitors, the built-in file system monitor
+can speed up Git commands that need to refresh the Git index
+(e.g. `git status`) in a working directory with many files.  The
+built-in monitor eliminates the need to install and maintain an
+external third-party tool.
++
+The built-in file system monitor is currently available only on a
+limited set of supported platforms.  Currently, this includes Windows
+and MacOS.
++
+Note: if this config setting is set to `true`, the values of
+`core.fsmonitor` and `core.fsmonitorHookVersion` are ignored.
 
 core.trustctime::
 	If false, the ctime differences between the index and the
diff --git a/Documentation/git-fsmonitor--daemon.txt b/Documentation/git-fsmonitor--daemon.txt
new file mode 100644
index 00000000000..154e7684daa
--- /dev/null
+++ b/Documentation/git-fsmonitor--daemon.txt
@@ -0,0 +1,75 @@
+git-fsmonitor--daemon(1)
+========================
+
+NAME
+----
+git-fsmonitor--daemon - A Built-in File System Monitor
+
+SYNOPSIS
+--------
+[verse]
+'git fsmonitor--daemon' start
+'git fsmonitor--daemon' run
+'git fsmonitor--daemon' stop
+'git fsmonitor--daemon' status
+
+DESCRIPTION
+-----------
+
+A daemon to watch the working directory for file and directory
+changes using platform-specific file system notification facilities.
+
+This daemon communicates directly with commands like `git status`
+using the link:technical/api-simple-ipc.html[simple IPC] interface
+instead of the slower linkgit:githooks[5] interface.
+
+This daemon is built into Git so that no third-party tools are
+required.
+
+OPTIONS
+-------
+
+start::
+	Starts a daemon in the background.
+
+run::
+	Runs a daemon in the foreground.
+
+stop::
+	Stops the daemon running in the current working
+	directory, if present.
+
+status::
+	Exits with zero status if a daemon is watching the
+	current working directory.
+
+REMARKS
+-------
+
+This daemon is a long running process used to watch a single working
+directory and maintain a list of the recently changed files and
+directories.  Performance of commands such as `git status` can be
+increased if they just ask for a summary of changes to the working
+directory and can avoid scanning the disk.
+
+When `core.useBuiltinFSMonitor` is set to `true` (see
+linkgit:git-config[1]) commands, such as `git status`, will ask the
+daemon for changes and automatically start it (if necessary).
+
+For more information see the "File System Monitor" section in
+linkgit:git-update-index[1].
+
+CAVEATS
+-------
+
+The fsmonitor daemon does not currently know about submodules and does
+not know to filter out file system events that happen within a
+submodule.  If fsmonitor daemon is watching a super repo and a file is
+modified within the working directory of a submodule, it will report
+the change (as happening against the super repo).  However, the client
+will properly ignore these extra events, so performance may be affected
+but it will not cause an incorrect result.
+
+GIT
+---
+Part of the linkgit:git[1] suite
diff --git a/Documentation/git-update-index.txt b/Documentation/git-update-index.txt
index 2853f168d97..c7c31b3fcf9 100644
--- a/Documentation/git-update-index.txt
+++ b/Documentation/git-update-index.txt
@@ -498,7 +498,9 @@ FILE SYSTEM MONITOR
 This feature is intended to speed up git operations for repos that have
 large working directories.
 
-It enables git to work together with a file system monitor (see the
+It enables git to work together with a file system monitor (see
+linkgit:git-fsmonitor--daemon[1]
+and the
 "fsmonitor-watchman" section of linkgit:githooks[5]) that can
 inform it as to what files have been modified. This enables git to avoid
 having to lstat() every file to find modified files.
@@ -508,17 +510,18 @@ performance by avoiding the cost of scanning the entire working directory
 looking for new files.
 
 If you want to enable (or disable) this feature, it is easier to use
-the `core.fsmonitor` configuration variable (see
-linkgit:git-config[1]) than using the `--fsmonitor` option to
-`git update-index` in each repository, especially if you want to do so
-across all repositories you use, because you can set the configuration
-variable in your `$HOME/.gitconfig` just once and have it affect all
-repositories you touch.
-
-When the `core.fsmonitor` configuration variable is changed, the
-file system monitor is added to or removed from the index the next time
-a command reads the index. When `--[no-]fsmonitor` are used, the file
-system monitor is immediately added to or removed from the index.
+the `core.fsmonitor` or `core.useBuiltinFSMonitor` configuration
+variable (see linkgit:git-config[1]) than using the `--fsmonitor`
+option to `git update-index` in each repository, especially if you
+want to do so across all repositories you use, because you can set the
+configuration variable in your `$HOME/.gitconfig` just once and have
+it affect all repositories you touch.
+
+When the `core.fsmonitor` or `core.useBuiltinFSMonitor` configuration
+variable is changed, the file system monitor is added to or removed
+from the index the next time a command reads the index. When
+`--[no-]fsmonitor` are used, the file system monitor is immediately
+added to or removed from the index.
 
 CONFIGURATION
 -------------
diff --git a/Documentation/githooks.txt b/Documentation/githooks.txt
index b51959ff941..b7d5e926f7b 100644
--- a/Documentation/githooks.txt
+++ b/Documentation/githooks.txt
@@ -593,7 +593,8 @@ fsmonitor-watchman
 
 This hook is invoked when the configuration option `core.fsmonitor` is
 set to `.git/hooks/fsmonitor-watchman` or `.git/hooks/fsmonitor-watchmanv2`
-depending on the version of the hook to use.
+depending on the version of the hook to use, unless overridden via
+`core.useBuiltinFSMonitor` (see linkgit:git-config[1]).
 
 Version 1 takes two arguments, a version (1) and the time in elapsed
 nanoseconds since midnight, January 1, 1970.
-- 
gitgitgadget

[PATCH v4 06/29] fsmonitor--daemon: add a built-in fsmonitor daemon

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Create a built-in file system monitoring daemon that can be used by
the existing `fsmonitor` feature (protocol API and index extension)
to improve the performance of various Git commands, such as `status`.

The `fsmonitor--daemon` feature builds upon the `Simple IPC` API and
provides an alternative to hook access to existing fsmonitors such
as `watchman`.

This commit merely adds the new command without any functionality.

Co-authored-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 .gitignore                  |  1 +
 Makefile                    |  1 +
 builtin.h                   |  1 +
 builtin/fsmonitor--daemon.c | 46 +++++++++++++++++++++++++++++++++++++
 git.c                       |  1 +
 5 files changed, 50 insertions(+)
 create mode 100644 builtin/fsmonitor--daemon.c

diff --git a/.gitignore b/.gitignore
index 054249b20a8..e863add420a 100644
--- a/.gitignore
+++ b/.gitignore
@@ -72,6 +72,7 @@
 /git-format-patch
 /git-fsck
 /git-fsck-objects
+/git-fsmonitor--daemon
 /git-gc
 /git-get-tar-commit-id
 /git-grep
diff --git a/Makefile b/Makefile
index 29ed7c4aba6..95ac224bc8d 100644
--- a/Makefile
+++ b/Makefile
@@ -1104,6 +1104,7 @@ BUILTIN_OBJS += builtin/fmt-merge-msg.o
 BUILTIN_OBJS += builtin/for-each-ref.o
 BUILTIN_OBJS += builtin/for-each-repo.o
 BUILTIN_OBJS += builtin/fsck.o
+BUILTIN_OBJS += builtin/fsmonitor--daemon.o
 BUILTIN_OBJS += builtin/gc.o
 BUILTIN_OBJS += builtin/get-tar-commit-id.o
 BUILTIN_OBJS += builtin/grep.o
diff --git a/builtin.h b/builtin.h
index 8a58743ed63..837b9ede598 100644
--- a/builtin.h
+++ b/builtin.h
@@ -159,6 +159,7 @@ int cmd_for_each_ref(int argc, const char **argv, const char *prefix);
 int cmd_for_each_repo(int argc, const char **argv, const char *prefix);
 int cmd_format_patch(int argc, const char **argv, const char *prefix);
 int cmd_fsck(int argc, const char **argv, const char *prefix);
+int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix);
 int cmd_gc(int argc, const char **argv, const char *prefix);
 int cmd_get_tar_commit_id(int argc, const char **argv, const char *prefix);
 int cmd_grep(int argc, const char **argv, const char *prefix);
diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
new file mode 100644
index 00000000000..f0498793379
--- /dev/null
+++ b/builtin/fsmonitor--daemon.c
@@ -0,0 +1,46 @@
+#include "builtin.h"
+#include "config.h"
+#include "parse-options.h"
+#include "fsmonitor.h"
+#include "fsmonitor-ipc.h"
+#include "simple-ipc.h"
+#include "khash.h"
+
+static const char * const builtin_fsmonitor__daemon_usage[] = {
+	NULL
+};
+
+#ifdef HAVE_FSMONITOR_DAEMON_BACKEND
+
+int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
+{
+	const char *subcmd;
+
+	struct option options[] = {
+		OPT_END()
+	};
+
+	git_config(git_default_config, NULL);
+
+	argc = parse_options(argc, argv, prefix, options,
+			     builtin_fsmonitor__daemon_usage, 0);
+	if (argc != 1)
+		usage_with_options(builtin_fsmonitor__daemon_usage, options);
+	subcmd = argv[0];
+
+	die(_("Unhandled subcommand '%s'"), subcmd);
+}
+
+#else
+int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
+{
+	struct option options[] = {
+		OPT_END()
+	};
+
+	if (argc == 2 && !strcmp(argv[1], "-h"))
+		usage_with_options(builtin_fsmonitor__daemon_usage, options);
+
+	die(_("fsmonitor--daemon not supported on this platform"));
+}
+#endif
diff --git a/git.c b/git.c
index 5ff21be21f3..28c8af6b795 100644
--- a/git.c
+++ b/git.c
@@ -533,6 +533,7 @@ static struct cmd_struct commands[] = {
 	{ "format-patch", cmd_format_patch, RUN_SETUP },
 	{ "fsck", cmd_fsck, RUN_SETUP },
 	{ "fsck-objects", cmd_fsck, RUN_SETUP },
+	{ "fsmonitor--daemon", cmd_fsmonitor__daemon, RUN_SETUP },
 	{ "gc", cmd_gc, RUN_SETUP },
 	{ "get-tar-commit-id", cmd_get_tar_commit_id, NO_PARSEOPT },
 	{ "grep", cmd_grep, RUN_SETUP_GENTLY },
-- 
gitgitgadget

[PATCH v4 07/29] fsmonitor--daemon: implement 'stop' and 'status' commands

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Implement `stop` and `status` client commands to control and query the
status of a `fsmonitor--daemon` server process (and implicitly start a
server process if necessary).

Later commits will implement the actual server and monitor the file
system.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 builtin/fsmonitor--daemon.c | 51 +++++++++++++++++++++++++++++++++++++
 1 file changed, 51 insertions(+)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index f0498793379..5e3178b8bdd 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -7,10 +7,55 @@
 #include "khash.h"
 
 static const char * const builtin_fsmonitor__daemon_usage[] = {
+	N_("git fsmonitor--daemon stop"),
+	N_("git fsmonitor--daemon status"),
 	NULL
 };
 
 #ifdef HAVE_FSMONITOR_DAEMON_BACKEND
+/*
+ * Acting as a CLIENT.
+ *
+ * Send a "quit" command to the `git-fsmonitor--daemon` (if running)
+ * and wait for it to shutdown.
+ */
+static int do_as_client__send_stop(void)
+{
+	struct strbuf answer = STRBUF_INIT;
+	int ret;
+
+	ret = fsmonitor_ipc__send_command("quit", &answer);
+
+	/* The quit command does not return any response data. */
+	strbuf_release(&answer);
+
+	if (ret)
+		return ret;
+
+	trace2_region_enter("fsm_client", "polling-for-daemon-exit", NULL);
+	while (fsmonitor_ipc__get_state() == IPC_STATE__LISTENING)
+		sleep_millisec(50);
+	trace2_region_leave("fsm_client", "polling-for-daemon-exit", NULL);
+
+	return 0;
+}
+
+static int do_as_client__status(void)
+{
+	enum ipc_active_state state = fsmonitor_ipc__get_state();
+
+	switch (state) {
+	case IPC_STATE__LISTENING:
+		printf(_("fsmonitor-daemon is watching '%s'\n"),
+		       the_repository->worktree);
+		return 0;
+
+	default:
+		printf(_("fsmonitor-daemon is not watching '%s'\n"),
+		       the_repository->worktree);
+		return 1;
+	}
+}
 
 int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
 {
@@ -28,6 +73,12 @@ int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
 		usage_with_options(builtin_fsmonitor__daemon_usage, options);
 	subcmd = argv[0];
 
+	if (!strcmp(subcmd, "stop"))
+		return !!do_as_client__send_stop();
+
+	if (!strcmp(subcmd, "status"))
+		return !!do_as_client__status();
+
 	die(_("Unhandled subcommand '%s'"), subcmd);
 }
 
-- 
gitgitgadget

[PATCH v4 08/29] compat/fsmonitor/fsm-listen-win32: stub in backend for Windows

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Stub in empty filesystem listener backend for fsmonitor--daemon on Windows.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 Makefile                            | 13 ++++++++
 compat/fsmonitor/fsm-listen-win32.c | 21 +++++++++++++
 compat/fsmonitor/fsm-listen.h       | 49 +++++++++++++++++++++++++++++
 config.mak.uname                    | 10 ++++++
 contrib/buildsystems/CMakeLists.txt |  7 +++++
 repo-settings.c                     |  1 +
 6 files changed, 101 insertions(+)
 create mode 100644 compat/fsmonitor/fsm-listen-win32.c
 create mode 100644 compat/fsmonitor/fsm-listen.h

diff --git a/Makefile b/Makefile
index 95ac224bc8d..8a26696c9ed 100644
--- a/Makefile
+++ b/Makefile
@@ -465,6 +465,11 @@ all::
 # directory, and the JSON compilation database 'compile_commands.json' will be
 # created at the root of the repository.
 #
+# If your platform supports a built-in fsmonitor backend, set
+# FSMONITOR_DAEMON_BACKEND to the "<name>" of the corresponding
+# `compat/fsmonitor/fsm-listen-<name>.c` that implements the
+# `fsm_listen__*()` routines.
+#
 # Define DEVELOPER to enable more compiler warnings. Compiler version
 # and family are auto detected, but could be overridden by defining
 # COMPILER_FEATURES (see config.mak.dev). You can still set
@@ -1930,6 +1935,11 @@ ifdef NEED_ACCESS_ROOT_HANDLER
 	COMPAT_OBJS += compat/access.o
 endif
 
+ifdef FSMONITOR_DAEMON_BACKEND
+	COMPAT_CFLAGS += -DHAVE_FSMONITOR_DAEMON_BACKEND
+	COMPAT_OBJS += compat/fsmonitor/fsm-listen-$(FSMONITOR_DAEMON_BACKEND).o
+endif
+
 ifeq ($(TCLTK_PATH),)
 NO_TCLTK = NoThanks
 endif
@@ -2805,6 +2815,9 @@ GIT-BUILD-OPTIONS: FORCE
 	@echo DC_SHA1=\''$(subst ','\'',$(subst ','\'',$(DC_SHA1)))'\' >>$@+
 	@echo SANITIZE_LEAK=\''$(subst ','\'',$(subst ','\'',$(SANITIZE_LEAK)))'\' >>$@+
 	@echo X=\'$(X)\' >>$@+
+ifdef FSMONITOR_DAEMON_BACKEND
+	@echo FSMONITOR_DAEMON_BACKEND=\''$(subst ','\'',$(subst ','\'',$(FSMONITOR_DAEMON_BACKEND)))'\' >>$@+
+endif
 ifdef TEST_OUTPUT_DIRECTORY
 	@echo TEST_OUTPUT_DIRECTORY=\''$(subst ','\'',$(subst ','\'',$(TEST_OUTPUT_DIRECTORY)))'\' >>$@+
 endif
diff --git a/compat/fsmonitor/fsm-listen-win32.c b/compat/fsmonitor/fsm-listen-win32.c
new file mode 100644
index 00000000000..916cbea254f
--- /dev/null
+++ b/compat/fsmonitor/fsm-listen-win32.c
@@ -0,0 +1,21 @@
+#include "cache.h"
+#include "config.h"
+#include "fsmonitor.h"
+#include "fsm-listen.h"
+
+void fsm_listen__stop_async(struct fsmonitor_daemon_state *state)
+{
+}
+
+void fsm_listen__loop(struct fsmonitor_daemon_state *state)
+{
+}
+
+int fsm_listen__ctor(struct fsmonitor_daemon_state *state)
+{
+	return -1;
+}
+
+void fsm_listen__dtor(struct fsmonitor_daemon_state *state)
+{
+}
diff --git a/compat/fsmonitor/fsm-listen.h b/compat/fsmonitor/fsm-listen.h
new file mode 100644
index 00000000000..f0539349baf
--- /dev/null
+++ b/compat/fsmonitor/fsm-listen.h
@@ -0,0 +1,49 @@
+#ifndef FSM_LISTEN_H
+#define FSM_LISTEN_H
+
+/* This needs to be implemented by each backend */
+
+#ifdef HAVE_FSMONITOR_DAEMON_BACKEND
+
+struct fsmonitor_daemon_state;
+
+/*
+ * Initialize platform-specific data for the fsmonitor listener thread.
+ * This will be called from the main thread PRIOR to staring the
+ * fsmonitor_fs_listener thread.
+ *
+ * Returns 0 if successful.
+ * Returns -1 otherwise.
+ */
+int fsm_listen__ctor(struct fsmonitor_daemon_state *state);
+
+/*
+ * Cleanup platform-specific data for the fsmonitor listener thread.
+ * This will be called from the main thread AFTER joining the listener.
+ */
+void fsm_listen__dtor(struct fsmonitor_daemon_state *state);
+
+/*
+ * The main body of the platform-specific event loop to watch for
+ * filesystem events.  This will run in the fsmonitor_fs_listen thread.
+ *
+ * It should call `ipc_server_stop_async()` if the listener thread
+ * prematurely terminates (because of a filesystem error or if it
+ * detects that the .git directory has been deleted).  (It should NOT
+ * do so if the listener thread receives a normal shutdown signal from
+ * the IPC layer.)
+ *
+ * It should set `state->error_code` to -1 if the daemon should exit
+ * with an error.
+ */
+void fsm_listen__loop(struct fsmonitor_daemon_state *state);
+
+/*
+ * Gently request that the fsmonitor listener thread shutdown.
+ * It does not wait for it to stop.  The caller should do a JOIN
+ * to wait for it.
+ */
+void fsm_listen__stop_async(struct fsmonitor_daemon_state *state);
+
+#endif /* HAVE_FSMONITOR_DAEMON_BACKEND */
+#endif /* FSM_LISTEN_H */
diff --git a/config.mak.uname b/config.mak.uname
index 3236a4918a3..e6a900c6fd4 100644
--- a/config.mak.uname
+++ b/config.mak.uname
@@ -426,6 +426,11 @@ ifeq ($(uname_S),Windows)
 	# so we don't need this:
 	#
 	#   SNPRINTF_RETURNS_BOGUS = YesPlease
+
+	# The builtin FSMonitor requires Named Pipes and Threads on Windows.
+	# These are always available, so we do not have to conditionally
+	# support it.
+	FSMONITOR_DAEMON_BACKEND = win32
 	NO_SVN_TESTS = YesPlease
 	RUNTIME_PREFIX = YesPlease
 	HAVE_WPGMPTR = YesWeDo
@@ -609,6 +614,11 @@ ifeq ($(uname_S),MINGW)
 	NO_STRTOUMAX = YesPlease
 	NO_MKDTEMP = YesPlease
 	NO_SVN_TESTS = YesPlease
+
+	# The builtin FSMonitor requires Named Pipes and Threads on Windows.
+	# These are always available, so we do not have to conditionally
+	# support it.
+	FSMONITOR_DAEMON_BACKEND = win32
 	RUNTIME_PREFIX = YesPlease
 	HAVE_WPGMPTR = YesWeDo
 	NO_ST_BLOCKS_IN_STRUCT_STAT = YesPlease
diff --git a/contrib/buildsystems/CMakeLists.txt b/contrib/buildsystems/CMakeLists.txt
index fd1399c440f..24a26743dcd 100644
--- a/contrib/buildsystems/CMakeLists.txt
+++ b/contrib/buildsystems/CMakeLists.txt
@@ -285,6 +285,13 @@ else()
 	endif()
 endif()
 
+if(SUPPORTS_SIMPLE_IPC)
+	if(CMAKE_SYSTEM_NAME STREQUAL "Windows")
+		add_compile_definitions(HAVE_FSMONITOR_DAEMON_BACKEND)
+		list(APPEND compat_SOURCES compat/fsmonitor/fsm-listen-win32.c)
+	endif()
+endif()
+
 set(EXE_EXTENSION ${CMAKE_EXECUTABLE_SUFFIX})
 
 #header checks
diff --git a/repo-settings.c b/repo-settings.c
index b93e91a212e..3ed763ad9c6 100644
--- a/repo-settings.c
+++ b/repo-settings.c
@@ -2,6 +2,7 @@
 #include "config.h"
 #include "repository.h"
 #include "midx.h"
+#include "compat/fsmonitor/fsm-listen.h"
 
 static void repo_cfg_bool(struct repository *r, const char *key, int *dest,
 			  int def)
-- 
gitgitgadget

[PATCH v4 09/29] compat/fsmonitor/fsm-listen-darwin: stub in backend for Darwin

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Stub in empty implementation of fsmonitor--daemon
backend for Darwin (aka MacOS).

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 compat/fsmonitor/fsm-listen-darwin.c | 20 ++++++++++++++++++++
 config.mak.uname                     | 10 ++++++++++
 contrib/buildsystems/CMakeLists.txt  |  3 +++
 3 files changed, 33 insertions(+)
 create mode 100644 compat/fsmonitor/fsm-listen-darwin.c

diff --git a/compat/fsmonitor/fsm-listen-darwin.c b/compat/fsmonitor/fsm-listen-darwin.c
new file mode 100644
index 00000000000..c84e3344ab9
--- /dev/null
+++ b/compat/fsmonitor/fsm-listen-darwin.c
@@ -0,0 +1,20 @@
+#include "cache.h"
+#include "fsmonitor.h"
+#include "fsm-listen.h"
+
+int fsm_listen__ctor(struct fsmonitor_daemon_state *state)
+{
+	return -1;
+}
+
+void fsm_listen__dtor(struct fsmonitor_daemon_state *state)
+{
+}
+
+void fsm_listen__stop_async(struct fsmonitor_daemon_state *state)
+{
+}
+
+void fsm_listen__loop(struct fsmonitor_daemon_state *state)
+{
+}
diff --git a/config.mak.uname b/config.mak.uname
index e6a900c6fd4..81f21d9e201 100644
--- a/config.mak.uname
+++ b/config.mak.uname
@@ -153,6 +153,16 @@ ifeq ($(uname_S),Darwin)
 			MSGFMT = /usr/local/opt/gettext/bin/msgfmt
 		endif
 	endif
+
+	# The builtin FSMonitor on MacOS builds upon Simple-IPC.  Both require
+	# Unix domain sockets and PThreads.
+	ifndef NO_PTHREADS
+	ifndef NO_UNIX_SOCKETS
+	FSMONITOR_DAEMON_BACKEND = darwin
+	endif
+	endif
+
+	BASIC_LDFLAGS += -framework CoreServices
 endif
 ifeq ($(uname_S),SunOS)
 	NEEDS_SOCKET = YesPlease
diff --git a/contrib/buildsystems/CMakeLists.txt b/contrib/buildsystems/CMakeLists.txt
index 24a26743dcd..41395705ac6 100644
--- a/contrib/buildsystems/CMakeLists.txt
+++ b/contrib/buildsystems/CMakeLists.txt
@@ -289,6 +289,9 @@ if(SUPPORTS_SIMPLE_IPC)
 	if(CMAKE_SYSTEM_NAME STREQUAL "Windows")
 		add_compile_definitions(HAVE_FSMONITOR_DAEMON_BACKEND)
 		list(APPEND compat_SOURCES compat/fsmonitor/fsm-listen-win32.c)
+	elseif(CMAKE_SYSTEM_NAME STREQUAL "Darwin")
+		add_compile_definitions(HAVE_FSMONITOR_DAEMON_BACKEND)
+		list(APPEND compat_SOURCES compat/fsmonitor/fsm-listen-darwin.c)
 	endif()
 endif()
 
-- 
gitgitgadget

[PATCH v4 10/29] fsmonitor--daemon: implement 'run' command

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Implement `run` command to try to begin listening for file system events.

This version defines the thread structure with a single fsmonitor_fs_listen
thread to watch for file system events and a simple IPC thread pool to
watch for connection from Git clients over a well-known named pipe or
Unix domain socket.

This commit does not actually do anything yet because the platform
backends are still just stubs.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 builtin/fsmonitor--daemon.c | 213 +++++++++++++++++++++++++++++++++++-
 fsmonitor--daemon.h         |  34 ++++++
 2 files changed, 246 insertions(+), 1 deletion(-)
 create mode 100644 fsmonitor--daemon.h

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index 5e3178b8bdd..b5ebd1eca33 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -3,16 +3,39 @@
 #include "parse-options.h"
 #include "fsmonitor.h"
 #include "fsmonitor-ipc.h"
+#include "compat/fsmonitor/fsm-listen.h"
+#include "fsmonitor--daemon.h"
 #include "simple-ipc.h"
 #include "khash.h"
 
 static const char * const builtin_fsmonitor__daemon_usage[] = {
+	N_("git fsmonitor--daemon run [<options>]"),
 	N_("git fsmonitor--daemon stop"),
 	N_("git fsmonitor--daemon status"),
 	NULL
 };
 
 #ifdef HAVE_FSMONITOR_DAEMON_BACKEND
+/*
+ * Global state loaded from config.
+ */
+#define FSMONITOR__IPC_THREADS "fsmonitor.ipcthreads"
+static int fsmonitor__ipc_threads = 8;
+
+static int fsmonitor_config(const char *var, const char *value, void *cb)
+{
+	if (!strcmp(var, FSMONITOR__IPC_THREADS)) {
+		int i = git_config_int(var, value);
+		if (i < 1)
+			return error(_("value of '%s' out of range: %d"),
+				     FSMONITOR__IPC_THREADS, i);
+		fsmonitor__ipc_threads = i;
+		return 0;
+	}
+
+	return git_default_config(var, value, cb);
+}
+
 /*
  * Acting as a CLIENT.
  *
@@ -57,15 +80,196 @@ static int do_as_client__status(void)
 	}
 }
 
+static ipc_server_application_cb handle_client;
+
+static int handle_client(void *data,
+			 const char *command, size_t command_len,
+			 ipc_server_reply_cb *reply,
+			 struct ipc_server_reply_data *reply_data)
+{
+	/* struct fsmonitor_daemon_state *state = data; */
+	int result;
+
+	/*
+	 * The Simple IPC API now supports {char*, len} arguments, but
+	 * FSMonitor always uses proper null-terminated strings, so
+	 * we can ignore the command_len argument.  (Trust, but verify.)
+	 */
+	if (command_len != strlen(command))
+		BUG("FSMonitor assumes text messages");
+
+	trace2_region_enter("fsmonitor", "handle_client", the_repository);
+	trace2_data_string("fsmonitor", the_repository, "request", command);
+
+	result = 0; /* TODO Do something here. */
+
+	trace2_region_leave("fsmonitor", "handle_client", the_repository);
+
+	return result;
+}
+
+static void *fsm_listen__thread_proc(void *_state)
+{
+	struct fsmonitor_daemon_state *state = _state;
+
+	trace2_thread_start("fsm-listen");
+
+	trace_printf_key(&trace_fsmonitor, "Watching: worktree '%s'",
+			 state->path_worktree_watch.buf);
+	if (state->nr_paths_watching > 1)
+		trace_printf_key(&trace_fsmonitor, "Watching: gitdir '%s'",
+				 state->path_gitdir_watch.buf);
+
+	fsm_listen__loop(state);
+
+	trace2_thread_exit();
+	return NULL;
+}
+
+static int fsmonitor_run_daemon_1(struct fsmonitor_daemon_state *state)
+{
+	struct ipc_server_opts ipc_opts = {
+		.nr_threads = fsmonitor__ipc_threads,
+
+		/*
+		 * We know that there are no other active threads yet,
+		 * so we can let the IPC layer temporarily chdir() if
+		 * it needs to when creating the server side of the
+		 * Unix domain socket.
+		 */
+		.uds_disallow_chdir = 0
+	};
+
+	/*
+	 * Start the IPC thread pool before the we've started the file
+	 * system event listener thread so that we have the IPC handle
+	 * before we need it.
+	 */
+	if (ipc_server_run_async(&state->ipc_server_data,
+				 fsmonitor_ipc__get_path(), &ipc_opts,
+				 handle_client, state))
+		return error_errno(
+			_("could not start IPC thread pool on '%s'"),
+			fsmonitor_ipc__get_path());
+
+	/*
+	 * Start the fsmonitor listener thread to collect filesystem
+	 * events.
+	 */
+	if (pthread_create(&state->listener_thread, NULL,
+			   fsm_listen__thread_proc, state) < 0) {
+		ipc_server_stop_async(state->ipc_server_data);
+		ipc_server_await(state->ipc_server_data);
+
+		return error(_("could not start fsmonitor listener thread"));
+	}
+
+	/*
+	 * The daemon is now fully functional in background threads.
+	 * Wait for the IPC thread pool to shutdown (whether by client
+	 * request or from filesystem activity).
+	 */
+	ipc_server_await(state->ipc_server_data);
+
+	/*
+	 * The fsmonitor listener thread may have received a shutdown
+	 * event from the IPC thread pool, but it doesn't hurt to tell
+	 * it again.  And wait for it to shutdown.
+	 */
+	fsm_listen__stop_async(state);
+	pthread_join(state->listener_thread, NULL);
+
+	return state->error_code;
+}
+
+static int fsmonitor_run_daemon(void)
+{
+	struct fsmonitor_daemon_state state;
+	int err;
+
+	memset(&state, 0, sizeof(state));
+
+	pthread_mutex_init(&state.main_lock, NULL);
+	state.error_code = 0;
+	state.current_token_data = NULL;
+
+	/* Prepare to (recursively) watch the <worktree-root> directory. */
+	strbuf_init(&state.path_worktree_watch, 0);
+	strbuf_addstr(&state.path_worktree_watch, absolute_path(get_git_work_tree()));
+	state.nr_paths_watching = 1;
+
+	/*
+	 * We create and delete cookie files somewhere inside the .git
+	 * directory to help us keep sync with the file system.  If
+	 * ".git" is not a directory, then <gitdir> is not inside the
+	 * cone of <worktree-root>, so set up a second watch to watch
+	 * the <gitdir> so that we get events for the cookie files.
+	 */
+	strbuf_init(&state.path_gitdir_watch, 0);
+	strbuf_addbuf(&state.path_gitdir_watch, &state.path_worktree_watch);
+	strbuf_addstr(&state.path_gitdir_watch, "/.git");
+	if (!is_directory(state.path_gitdir_watch.buf)) {
+		strbuf_reset(&state.path_gitdir_watch);
+		strbuf_addstr(&state.path_gitdir_watch, absolute_path(get_git_dir()));
+		state.nr_paths_watching = 2;
+	}
+
+	/*
+	 * Confirm that we can create platform-specific resources for the
+	 * filesystem listener before we bother starting all the threads.
+	 */
+	if (fsm_listen__ctor(&state)) {
+		err = error(_("could not initialize listener thread"));
+		goto done;
+	}
+
+	err = fsmonitor_run_daemon_1(&state);
+
+done:
+	pthread_mutex_destroy(&state.main_lock);
+	fsm_listen__dtor(&state);
+
+	ipc_server_free(state.ipc_server_data);
+
+	strbuf_release(&state.path_worktree_watch);
+	strbuf_release(&state.path_gitdir_watch);
+
+	return err;
+}
+
+static int try_to_run_foreground_daemon(void)
+{
+	/*
+	 * Technically, we don't need to probe for an existing daemon
+	 * process, since we could just call `fsmonitor_run_daemon()`
+	 * and let it fail if the pipe/socket is busy.
+	 *
+	 * However, this method gives us a nicer error message for a
+	 * common error case.
+	 */
+	if (fsmonitor_ipc__get_state() == IPC_STATE__LISTENING)
+		die("fsmonitor--daemon is already running '%s'",
+		    the_repository->worktree);
+
+	printf(_("running fsmonitor-daemon in '%s'\n"),
+	       the_repository->worktree);
+	fflush(stdout);
+
+	return !!fsmonitor_run_daemon();
+}
+
 int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
 {
 	const char *subcmd;
 
 	struct option options[] = {
+		OPT_INTEGER(0, "ipc-threads",
+			    &fsmonitor__ipc_threads,
+			    N_("use <n> ipc worker threads")),
 		OPT_END()
 	};
 
-	git_config(git_default_config, NULL);
+	git_config(fsmonitor_config, NULL);
 
 	argc = parse_options(argc, argv, prefix, options,
 			     builtin_fsmonitor__daemon_usage, 0);
@@ -73,6 +277,13 @@ int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
 		usage_with_options(builtin_fsmonitor__daemon_usage, options);
 	subcmd = argv[0];
 
+	if (fsmonitor__ipc_threads < 1)
+		die(_("invalid 'ipc-threads' value (%d)"),
+		    fsmonitor__ipc_threads);
+
+	if (!strcmp(subcmd, "run"))
+		return !!try_to_run_foreground_daemon();
+
 	if (!strcmp(subcmd, "stop"))
 		return !!do_as_client__send_stop();
 
diff --git a/fsmonitor--daemon.h b/fsmonitor--daemon.h
new file mode 100644
index 00000000000..3009c1a83de
--- /dev/null
+++ b/fsmonitor--daemon.h
@@ -0,0 +1,34 @@
+#ifndef FSMONITOR_DAEMON_H
+#define FSMONITOR_DAEMON_H
+
+#ifdef HAVE_FSMONITOR_DAEMON_BACKEND
+
+#include "cache.h"
+#include "dir.h"
+#include "run-command.h"
+#include "simple-ipc.h"
+#include "thread-utils.h"
+
+struct fsmonitor_batch;
+struct fsmonitor_token_data;
+
+struct fsmonitor_daemon_backend_data; /* opaque platform-specific data */
+
+struct fsmonitor_daemon_state {
+	pthread_t listener_thread;
+	pthread_mutex_t main_lock;
+
+	struct strbuf path_worktree_watch;
+	struct strbuf path_gitdir_watch;
+	int nr_paths_watching;
+
+	struct fsmonitor_token_data *current_token_data;
+
+	int error_code;
+	struct fsmonitor_daemon_backend_data *backend_data;
+
+	struct ipc_server_data *ipc_server_data;
+};
+
+#endif /* HAVE_FSMONITOR_DAEMON_BACKEND */
+#endif /* FSMONITOR_DAEMON_H */
-- 
gitgitgadget

[PATCH v4 11/29] fsmonitor--daemon: implement 'start' command

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Implement 'git fsmonitor--daemon start' command.  This command starts
an instance of 'git fsmonitor--daemon run' in the background using
the new 'start_bg_command()' function.

We avoid the fork-and-call technique on Unix systems in favor of a
fork-and-exec technique.  This gives us more uniform Trace2 child-*
events.  It also makes our usage more consistent with Windows usage.

On Windows, teach 'git fsmonitor--daemon run' to optionally call
'FreeConsole()' to release handles to the inherited Win32 console
(despite being passed invalid handles for stdin/out/err).  Without
this, command prompts and powershell terminal windows could hang
in "exit" until the last background child process exited or released
their Win32 console handle.  (This was not seen with git-bash shells
because they don't have a Win32 console attached to them.)

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 builtin/fsmonitor--daemon.c | 107 +++++++++++++++++++++++++++++++++++-
 1 file changed, 105 insertions(+), 2 deletions(-)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index b5ebd1eca33..8988440b7c1 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -9,6 +9,7 @@
 #include "khash.h"
 
 static const char * const builtin_fsmonitor__daemon_usage[] = {
+	N_("git fsmonitor--daemon start [<options>]"),
 	N_("git fsmonitor--daemon run [<options>]"),
 	N_("git fsmonitor--daemon stop"),
 	N_("git fsmonitor--daemon status"),
@@ -22,6 +23,9 @@ static const char * const builtin_fsmonitor__daemon_usage[] = {
 #define FSMONITOR__IPC_THREADS "fsmonitor.ipcthreads"
 static int fsmonitor__ipc_threads = 8;
 
+#define FSMONITOR__START_TIMEOUT "fsmonitor.starttimeout"
+static int fsmonitor__start_timeout_sec = 60;
+
 static int fsmonitor_config(const char *var, const char *value, void *cb)
 {
 	if (!strcmp(var, FSMONITOR__IPC_THREADS)) {
@@ -33,6 +37,15 @@ static int fsmonitor_config(const char *var, const char *value, void *cb)
 		return 0;
 	}
 
+	if (!strcmp(var, FSMONITOR__START_TIMEOUT)) {
+		int i = git_config_int(var, value);
+		if (i < 0)
+			return error(_("value of '%s' out of range: %d"),
+				     FSMONITOR__START_TIMEOUT, i);
+		fsmonitor__start_timeout_sec = i;
+		return 0;
+	}
+
 	return git_default_config(var, value, cb);
 }
 
@@ -237,7 +250,7 @@ done:
 	return err;
 }
 
-static int try_to_run_foreground_daemon(void)
+static int try_to_run_foreground_daemon(int free_console)
 {
 	/*
 	 * Technically, we don't need to probe for an existing daemon
@@ -255,17 +268,104 @@ static int try_to_run_foreground_daemon(void)
 	       the_repository->worktree);
 	fflush(stdout);
 
+#ifdef GIT_WINDOWS_NATIVE
+	if (free_console)
+		FreeConsole();
+#endif
+
 	return !!fsmonitor_run_daemon();
 }
 
+static start_bg_wait_cb bg_wait_cb;
+
+static int bg_wait_cb(const struct child_process *cp, void *cb_data)
+{
+	enum ipc_active_state s = fsmonitor_ipc__get_state();
+
+	switch (s) {
+	case IPC_STATE__LISTENING:
+		/* child is "ready" */
+		return 0;
+
+	case IPC_STATE__NOT_LISTENING:
+	case IPC_STATE__PATH_NOT_FOUND:
+		/* give child more time */
+		return 1;
+
+	default:
+	case IPC_STATE__INVALID_PATH:
+	case IPC_STATE__OTHER_ERROR:
+		/* all the time in world won't help */
+		return -1;
+	}
+}
+
+static int try_to_start_background_daemon(void)
+{
+	struct child_process cp = CHILD_PROCESS_INIT;
+	enum start_bg_result sbgr;
+
+	/*
+	 * Before we try to create a background daemon process, see
+	 * if a daemon process is already listening.  This makes it
+	 * easier for us to report an already-listening error to the
+	 * console, since our spawn/daemon can only report the success
+	 * of creating the background process (and not whether it
+	 * immediately exited).
+	 */
+	if (fsmonitor_ipc__get_state() == IPC_STATE__LISTENING)
+		die("fsmonitor--daemon is already running '%s'",
+		    the_repository->worktree);
+
+	printf(_("starting fsmonitor-daemon in '%s'\n"),
+	       the_repository->worktree);
+	fflush(stdout);
+
+	cp.git_cmd = 1;
+
+	strvec_push(&cp.args, "fsmonitor--daemon");
+	strvec_push(&cp.args, "run");
+	strvec_push(&cp.args, "--free-console");
+	strvec_pushf(&cp.args, "--ipc-threads=%d", fsmonitor__ipc_threads);
+
+	cp.no_stdin = 1;
+	cp.no_stdout = 1;
+	cp.no_stderr = 1;
+
+	sbgr = start_bg_command(&cp, bg_wait_cb, NULL,
+				fsmonitor__start_timeout_sec);
+
+	switch (sbgr) {
+	case SBGR_READY:
+		return 0;
+
+	default:
+	case SBGR_ERROR:
+	case SBGR_CB_ERROR:
+		return error("daemon failed to start");
+
+	case SBGR_TIMEOUT:
+		return error("daemon not online yet");
+
+	case SBGR_DIED:
+		return error("daemon terminated");
+	}
+}
+
 int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
 {
 	const char *subcmd;
+	int free_console = 0;
 
 	struct option options[] = {
+		OPT_BOOL(0, "free-console", &free_console, N_("free console")),
 		OPT_INTEGER(0, "ipc-threads",
 			    &fsmonitor__ipc_threads,
 			    N_("use <n> ipc worker threads")),
+		OPT_INTEGER(0, "start-timeout",
+			    &fsmonitor__start_timeout_sec,
+			    N_("Max seconds to wait for background daemon startup")),
+
 		OPT_END()
 	};
 
@@ -281,8 +381,11 @@ int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
 		die(_("invalid 'ipc-threads' value (%d)"),
 		    fsmonitor__ipc_threads);
 
+	if (!strcmp(subcmd, "start"))
+		return !!try_to_start_background_daemon();
+
 	if (!strcmp(subcmd, "run"))
-		return !!try_to_run_foreground_daemon();
+		return !!try_to_run_foreground_daemon(free_console);
 
 	if (!strcmp(subcmd, "stop"))
 		return !!do_as_client__send_stop();
-- 
gitgitgadget

[PATCH v4 12/29] fsmonitor--daemon: add pathname classification

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach fsmonitor--daemon to classify relative and absolute
pathnames and decide how they should be handled.  This will
be used by the platform-specific backend to respond to each
filesystem event.

When we register for filesystem notifications on a directory,
we get events for everything (recursively) in the directory.
We want to report to clients changes to tracked and untracked
paths within the working directory.  We do not want to report
changes within the .git directory, for example.

This classification will be used in a later commit by the
different backends to classify paths as events are received.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 builtin/fsmonitor--daemon.c | 81 +++++++++++++++++++++++++++++++++++++
 fsmonitor--daemon.h         | 61 ++++++++++++++++++++++++++++
 2 files changed, 142 insertions(+)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index 8988440b7c1..1618f1aa27f 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -121,6 +121,87 @@ static int handle_client(void *data,
 	return result;
 }
 
+#define FSMONITOR_COOKIE_PREFIX ".fsmonitor-daemon-"
+
+enum fsmonitor_path_type fsmonitor_classify_path_workdir_relative(
+	const char *rel)
+{
+	if (fspathncmp(rel, ".git", 4))
+		return IS_WORKDIR_PATH;
+	rel += 4;
+
+	if (!*rel)
+		return IS_DOT_GIT;
+	if (*rel != '/')
+		return IS_WORKDIR_PATH; /* e.g. .gitignore */
+	rel++;
+
+	if (!fspathncmp(rel, FSMONITOR_COOKIE_PREFIX,
+			strlen(FSMONITOR_COOKIE_PREFIX)))
+		return IS_INSIDE_DOT_GIT_WITH_COOKIE_PREFIX;
+
+	return IS_INSIDE_DOT_GIT;
+}
+
+enum fsmonitor_path_type fsmonitor_classify_path_gitdir_relative(
+	const char *rel)
+{
+	if (!fspathncmp(rel, FSMONITOR_COOKIE_PREFIX,
+			strlen(FSMONITOR_COOKIE_PREFIX)))
+		return IS_INSIDE_GITDIR_WITH_COOKIE_PREFIX;
+
+	return IS_INSIDE_GITDIR;
+}
+
+static enum fsmonitor_path_type try_classify_workdir_abs_path(
+	struct fsmonitor_daemon_state *state,
+	const char *path)
+{
+	const char *rel;
+
+	if (fspathncmp(path, state->path_worktree_watch.buf,
+		       state->path_worktree_watch.len))
+		return IS_OUTSIDE_CONE;
+
+	rel = path + state->path_worktree_watch.len;
+
+	if (!*rel)
+		return IS_WORKDIR_PATH; /* it is the root dir exactly */
+	if (*rel != '/')
+		return IS_OUTSIDE_CONE;
+	rel++;
+
+	return fsmonitor_classify_path_workdir_relative(rel);
+}
+
+enum fsmonitor_path_type fsmonitor_classify_path_absolute(
+	struct fsmonitor_daemon_state *state,
+	const char *path)
+{
+	const char *rel;
+	enum fsmonitor_path_type t;
+
+	t = try_classify_workdir_abs_path(state, path);
+	if (state->nr_paths_watching == 1)
+		return t;
+	if (t != IS_OUTSIDE_CONE)
+		return t;
+
+	if (fspathncmp(path, state->path_gitdir_watch.buf,
+		       state->path_gitdir_watch.len))
+		return IS_OUTSIDE_CONE;
+
+	rel = path + state->path_gitdir_watch.len;
+
+	if (!*rel)
+		return IS_GITDIR; /* it is the <gitdir> exactly */
+	if (*rel != '/')
+		return IS_OUTSIDE_CONE;
+	rel++;
+
+	return fsmonitor_classify_path_gitdir_relative(rel);
+}
+
 static void *fsm_listen__thread_proc(void *_state)
 {
 	struct fsmonitor_daemon_state *state = _state;
diff --git a/fsmonitor--daemon.h b/fsmonitor--daemon.h
index 3009c1a83de..7bbb3a27a1c 100644
--- a/fsmonitor--daemon.h
+++ b/fsmonitor--daemon.h
@@ -30,5 +30,66 @@ struct fsmonitor_daemon_state {
 	struct ipc_server_data *ipc_server_data;
 };
 
+/*
+ * Pathname classifications.
+ *
+ * The daemon classifies the pathnames that it receives from file
+ * system notification events into the following categories and uses
+ * that to decide whether clients are told about them.  (And to watch
+ * for file system synchronization events.)
+ *
+ * The client should only care about paths within the working
+ * directory proper (inside the working directory and not ".git" nor
+ * inside of ".git/").  That is, the client has read the index and is
+ * asking for a list of any paths in the working directory that have
+ * been modified since the last token.  The client does not care about
+ * file system changes within the .git directory (such as new loose
+ * objects or packfiles).  So the client will only receive paths that
+ * are classified as IS_WORKDIR_PATH.
+ *
+ * The daemon uses the IS_DOT_GIT and IS_GITDIR internally to mean the
+ * exact ".git" directory or GITDIR.  If the daemon receives a delete
+ * event for either of these directories, it will automatically
+ * shutdown, for example.
+ *
+ * Note that the daemon DOES NOT explicitly watch nor special case the
+ * ".git/index" file.  The daemon does not read the index and does not
+ * have any internal index-relative state.  The daemon only collects
+ * the set of modified paths within the working directory.
+ */
+enum fsmonitor_path_type {
+	IS_WORKDIR_PATH = 0,
+
+	IS_DOT_GIT,
+	IS_INSIDE_DOT_GIT,
+	IS_INSIDE_DOT_GIT_WITH_COOKIE_PREFIX,
+
+	IS_GITDIR,
+	IS_INSIDE_GITDIR,
+	IS_INSIDE_GITDIR_WITH_COOKIE_PREFIX,
+
+	IS_OUTSIDE_CONE,
+};
+
+/*
+ * Classify a pathname relative to the root of the working directory.
+ */
+enum fsmonitor_path_type fsmonitor_classify_path_workdir_relative(
+	const char *relative_path);
+
+/*
+ * Classify a pathname relative to a <gitdir> that is external to the
+ * worktree directory.
+ */
+enum fsmonitor_path_type fsmonitor_classify_path_gitdir_relative(
+	const char *relative_path);
+
+/*
+ * Classify an absolute pathname received from a filesystem event.
+ */
+enum fsmonitor_path_type fsmonitor_classify_path_absolute(
+	struct fsmonitor_daemon_state *state,
+	const char *path);
+
 #endif /* HAVE_FSMONITOR_DAEMON_BACKEND */
 #endif /* FSMONITOR_DAEMON_H */
-- 
gitgitgadget

[PATCH v4 13/29] fsmonitor--daemon: define token-ids

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach fsmonitor--daemon to create token-ids and define the
overall token naming scheme.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 builtin/fsmonitor--daemon.c | 116 +++++++++++++++++++++++++++++++++++-
 1 file changed, 115 insertions(+), 1 deletion(-)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index 1618f1aa27f..2997c2cba8c 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -93,6 +93,120 @@ static int do_as_client__status(void)
 	}
 }
 
+/*
+ * Requests to and from a FSMonitor Protocol V2 provider use an opaque
+ * "token" as a virtual timestamp.  Clients can request a summary of all
+ * created/deleted/modified files relative to a token.  In the response,
+ * clients receive a new token for the next (relative) request.
+ *
+ *
+ * Token Format
+ * ============
+ *
+ * The contents of the token are private and provider-specific.
+ *
+ * For the built-in fsmonitor--daemon, we define a token as follows:
+ *
+ *     "builtin" ":" <token_id> ":" <sequence_nr>
+ *
+ * The "builtin" prefix is used as a namespace to avoid conflicts
+ * with other providers (such as Watchman).
+ *
+ * The <token_id> is an arbitrary OPAQUE string, such as a GUID,
+ * UUID, or {timestamp,pid}.  It is used to group all filesystem
+ * events that happened while the daemon was monitoring (and in-sync
+ * with the filesystem).
+ *
+ *     Unlike FSMonitor Protocol V1, it is not defined as a timestamp
+ *     and does not define less-than/greater-than relationships.
+ *     (There are too many race conditions to rely on file system
+ *     event timestamps.)
+ *
+ * The <sequence_nr> is a simple integer incremented whenever the
+ * daemon needs to make its state public.  For example, if 1000 file
+ * system events come in, but no clients have requested the data,
+ * the daemon can continue to accumulate file changes in the same
+ * bin and does not need to advance the sequence number.  However,
+ * as soon as a client does arrive, the daemon needs to start a new
+ * bin and increment the sequence number.
+ *
+ *     The sequence number serves as the boundary between 2 sets
+ *     of bins -- the older ones that the client has already seen
+ *     and the newer ones that it hasn't.
+ *
+ * When a new <token_id> is created, the <sequence_nr> is reset to
+ * zero.
+ *
+ *
+ * About Token Ids
+ * ===============
+ *
+ * A new token_id is created:
+ *
+ * [1] each time the daemon is started.
+ *
+ * [2] any time that the daemon must re-sync with the filesystem
+ *     (such as when the kernel drops or we miss events on a very
+ *     active volume).
+ *
+ * [3] in response to a client "flush" command (for dropped event
+ *     testing).
+ *
+ * When a new token_id is created, the daemon is free to discard all
+ * cached filesystem events associated with any previous token_ids.
+ * Events associated with a non-current token_id will never be sent
+ * to a client.  A token_id change implicitly means that the daemon
+ * has gap in its event history.
+ *
+ * Therefore, clients that present a token with a stale (non-current)
+ * token_id will always be given a trivial response.
+ */
+struct fsmonitor_token_data {
+	struct strbuf token_id;
+	struct fsmonitor_batch *batch_head;
+	struct fsmonitor_batch *batch_tail;
+	uint64_t client_ref_count;
+};
+
+static struct fsmonitor_token_data *fsmonitor_new_token_data(void)
+{
+	static int test_env_value = -1;
+	static uint64_t flush_count = 0;
+	struct fsmonitor_token_data *token;
+
+	CALLOC_ARRAY(token, 1);
+
+	strbuf_init(&token->token_id, 0);
+	token->batch_head = NULL;
+	token->batch_tail = NULL;
+	token->client_ref_count = 0;
+
+	if (test_env_value < 0)
+		test_env_value = git_env_bool("GIT_TEST_FSMONITOR_TOKEN", 0);
+
+	if (!test_env_value) {
+		struct timeval tv;
+		struct tm tm;
+		time_t secs;
+
+		gettimeofday(&tv, NULL);
+		secs = tv.tv_sec;
+		gmtime_r(&secs, &tm);
+
+		strbuf_addf(&token->token_id,
+			    "%"PRIu64".%d.%4d%02d%02dT%02d%02d%02d.%06ldZ",
+			    flush_count++,
+			    getpid(),
+			    tm.tm_year + 1900, tm.tm_mon + 1, tm.tm_mday,
+			    tm.tm_hour, tm.tm_min, tm.tm_sec,
+			    (long)tv.tv_usec);
+	} else {
+		strbuf_addf(&token->token_id, "test_%08x", test_env_value++);
+	}
+
+	return token;
+}
+
 static ipc_server_application_cb handle_client;
 
 static int handle_client(void *data,
@@ -285,7 +399,7 @@ static int fsmonitor_run_daemon(void)
 
 	pthread_mutex_init(&state.main_lock, NULL);
 	state.error_code = 0;
-	state.current_token_data = NULL;
+	state.current_token_data = fsmonitor_new_token_data();
 
 	/* Prepare to (recursively) watch the <worktree-root> directory. */
 	strbuf_init(&state.path_worktree_watch, 0);
-- 
gitgitgadget

[PATCH v4 14/29] fsmonitor--daemon: create token-based changed path cache

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach fsmonitor--daemon to build a list of changed paths and associate
them with a token-id.  This will be used by the platform-specific
backends to accumulate changed paths in response to filesystem events.

The platform-specific file system listener thread receives file system
events containing one or more changed pathnames (with whatever bucketing
or grouping that is convenient for the file system).  These paths are
accumulated (without locking) by the file system layer into a `fsmonitor_batch`.

When the file system layer has drained the kernel event queue, it will
"publish" them to our token queue and make them visible to concurrent
client worker threads.  The token layer is free to combine and/or de-dup
paths within these batches for efficient presentation to clients.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 builtin/fsmonitor--daemon.c | 230 +++++++++++++++++++++++++++++++++++-
 fsmonitor--daemon.h         |  40 +++++++
 2 files changed, 268 insertions(+), 2 deletions(-)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index 2997c2cba8c..34603f23053 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -168,17 +168,27 @@ struct fsmonitor_token_data {
 	uint64_t client_ref_count;
 };
 
+struct fsmonitor_batch {
+	struct fsmonitor_batch *next;
+	uint64_t batch_seq_nr;
+	const char **interned_paths;
+	size_t nr, alloc;
+	time_t pinned_time;
+};
+
 static struct fsmonitor_token_data *fsmonitor_new_token_data(void)
 {
 	static int test_env_value = -1;
 	static uint64_t flush_count = 0;
 	struct fsmonitor_token_data *token;
+	struct fsmonitor_batch *batch;
 
 	CALLOC_ARRAY(token, 1);
+	batch = fsmonitor_batch__new();
 
 	strbuf_init(&token->token_id, 0);
-	token->batch_head = NULL;
-	token->batch_tail = NULL;
+	token->batch_head = batch;
+	token->batch_tail = batch;
 	token->client_ref_count = 0;
 
 	if (test_env_value < 0)
@@ -204,9 +214,143 @@ static struct fsmonitor_token_data *fsmonitor_new_token_data(void)
 		strbuf_addf(&token->token_id, "test_%08x", test_env_value++);
 	}
 
+	/*
+	 * We created a new <token_id> and are starting a new series
+	 * of tokens with a zero <seq_nr>.
+	 *
+	 * Since clients cannot guess our new (non test) <token_id>
+	 * they will always receive a trivial response (because of the
+	 * mismatch on the <token_id>).  The trivial response will
+	 * tell them our new <token_id> so that subsequent requests
+	 * will be relative to our new series.  (And when sending that
+	 * response, we pin the current head of the batch list.)
+	 *
+	 * Even if the client correctly guesses the <token_id>, their
+	 * request of "builtin:<token_id>:0" asks for all changes MORE
+	 * RECENT than batch/bin 0.
+	 *
+	 * This implies that it is a waste to accumulate paths in the
+	 * initial batch/bin (because they will never be transmitted).
+	 *
+	 * So the daemon could be running for days and watching the
+	 * file system, but doesn't need to actually accumulate any
+	 * paths UNTIL we need to set a reference point for a later
+	 * relative request.
+	 *
+	 * However, it is very useful for testing to always have a
+	 * reference point set.  Pin batch 0 to force early file system
+	 * events to accumulate.
+	 */
+	if (test_env_value)
+		batch->pinned_time = time(NULL);
+
 	return token;
 }
 
+struct fsmonitor_batch *fsmonitor_batch__new(void)
+{
+	struct fsmonitor_batch *batch;
+
+	CALLOC_ARRAY(batch, 1);
+
+	return batch;
+}
+
+void fsmonitor_batch__free_list(struct fsmonitor_batch *batch)
+{
+	while (batch) {
+		struct fsmonitor_batch *next = batch->next;
+
+		/*
+		 * The actual strings within the array of this batch
+		 * are interned, so we don't own them.  We only own
+		 * the array.
+		 */
+		free(batch->interned_paths);
+		free(batch);
+
+		batch = next;
+	}
+}
+
+void fsmonitor_batch__add_path(struct fsmonitor_batch *batch,
+			       const char *path)
+{
+	const char *interned_path = strintern(path);
+
+	trace_printf_key(&trace_fsmonitor, "event: %s", interned_path);
+
+	ALLOC_GROW(batch->interned_paths, batch->nr + 1, batch->alloc);
+	batch->interned_paths[batch->nr++] = interned_path;
+}
+
+static void fsmonitor_batch__combine(struct fsmonitor_batch *batch_dest,
+				     const struct fsmonitor_batch *batch_src)
+{
+	size_t k;
+
+	ALLOC_GROW(batch_dest->interned_paths,
+		   batch_dest->nr + batch_src->nr + 1,
+		   batch_dest->alloc);
+
+	for (k = 0; k < batch_src->nr; k++)
+		batch_dest->interned_paths[batch_dest->nr++] =
+			batch_src->interned_paths[k];
+}
+
+static void fsmonitor_free_token_data(struct fsmonitor_token_data *token)
+{
+	if (!token)
+		return;
+
+	assert(token->client_ref_count == 0);
+
+	strbuf_release(&token->token_id);
+
+	fsmonitor_batch__free_list(token->batch_head);
+
+	free(token);
+}
+
+/*
+ * Flush all of our cached data about the filesystem.  Call this if we
+ * lose sync with the filesystem and miss some notification events.
+ *
+ * [1] If we are missing events, then we no longer have a complete
+ *     history of the directory (relative to our current start token).
+ *     We should create a new token and start fresh (as if we just
+ *     booted up).
+ *
+ * If there are no concurrent threads readering the current token data
+ * series, we can free it now.  Otherwise, let the last reader free
+ * it.
+ *
+ * Either way, the old token data series is no longer associated with
+ * our state data.
+ */
+static void with_lock__do_force_resync(struct fsmonitor_daemon_state *state)
+{
+	/* assert current thread holding state->main_lock */
+
+	struct fsmonitor_token_data *free_me = NULL;
+	struct fsmonitor_token_data *new_one = NULL;
+
+	new_one = fsmonitor_new_token_data();
+
+	if (state->current_token_data->client_ref_count == 0)
+		free_me = state->current_token_data;
+	state->current_token_data = new_one;
+
+	fsmonitor_free_token_data(free_me);
+}
+
+void fsmonitor_force_resync(struct fsmonitor_daemon_state *state)
+{
+	pthread_mutex_lock(&state->main_lock);
+	with_lock__do_force_resync(state);
+	pthread_mutex_unlock(&state->main_lock);
+}
+
 static ipc_server_application_cb handle_client;
 
 static int handle_client(void *data,
@@ -316,6 +460,81 @@ enum fsmonitor_path_type fsmonitor_classify_path_absolute(
 	return fsmonitor_classify_path_gitdir_relative(rel);
 }
 
+/*
+ * We try to combine small batches at the front of the batch-list to avoid
+ * having a long list.  This hopefully makes it a little easier when we want
+ * to truncate and maintain the list.  However, we don't want the paths array
+ * to just keep growing and growing with realloc, so we insert an arbitrary
+ * limit.
+ */
+#define MY_COMBINE_LIMIT (1024)
+
+void fsmonitor_publish(struct fsmonitor_daemon_state *state,
+		       struct fsmonitor_batch *batch,
+		       const struct string_list *cookie_names)
+{
+	if (!batch && !cookie_names->nr)
+		return;
+
+	pthread_mutex_lock(&state->main_lock);
+
+	if (batch) {
+		struct fsmonitor_batch *head;
+
+		head = state->current_token_data->batch_head;
+		if (!head) {
+			BUG("token does not have batch");
+		} else if (head->pinned_time) {
+			/*
+			 * We cannot alter the current batch list
+			 * because:
+			 *
+			 * [a] it is being transmitted to at least one
+			 * client and the handle_client() thread has a
+			 * ref-count, but not a lock on the batch list
+			 * starting with this item.
+			 *
+			 * [b] it has been transmitted in the past to
+			 * at least one client such that future
+			 * requests are relative to this head batch.
+			 *
+			 * So, we can only prepend a new batch onto
+			 * the front of the list.
+			 */
+			batch->batch_seq_nr = head->batch_seq_nr + 1;
+			batch->next = head;
+			state->current_token_data->batch_head = batch;
+		} else if (!head->batch_seq_nr) {
+			/*
+			 * Batch 0 is unpinned.  See the note in
+			 * `fsmonitor_new_token_data()` about why we
+			 * don't need to accumulate these paths.
+			 */
+			fsmonitor_batch__free_list(batch);
+		} else if (head->nr + batch->nr > MY_COMBINE_LIMIT) {
+			/*
+			 * The head batch in the list has never been
+			 * transmitted to a client, but folding the
+			 * contents of the new batch onto it would
+			 * exceed our arbitrary limit, so just prepend
+			 * the new batch onto the list.
+			 */
+			batch->batch_seq_nr = head->batch_seq_nr + 1;
+			batch->next = head;
+			state->current_token_data->batch_head = batch;
+		} else {
+			/*
+			 * We are free to add the paths in the given
+			 * batch onto the end of the current head batch.
+			 */
+			fsmonitor_batch__combine(head, batch);
+			fsmonitor_batch__free_list(batch);
+		}
+	}
+
+	pthread_mutex_unlock(&state->main_lock);
+}
+
 static void *fsm_listen__thread_proc(void *_state)
 {
 	struct fsmonitor_daemon_state *state = _state;
@@ -330,6 +549,13 @@ static void *fsm_listen__thread_proc(void *_state)
 
 	fsm_listen__loop(state);
 
+	pthread_mutex_lock(&state->main_lock);
+	if (state->current_token_data &&
+	    state->current_token_data->client_ref_count == 0)
+		fsmonitor_free_token_data(state->current_token_data);
+	state->current_token_data = NULL;
+	pthread_mutex_unlock(&state->main_lock);
+
 	trace2_thread_exit();
 	return NULL;
 }
diff --git a/fsmonitor--daemon.h b/fsmonitor--daemon.h
index 7bbb3a27a1c..20a815d80f8 100644
--- a/fsmonitor--daemon.h
+++ b/fsmonitor--daemon.h
@@ -12,6 +12,27 @@
 struct fsmonitor_batch;
 struct fsmonitor_token_data;
 
+/*
+ * Create a new batch of path(s).  The returned batch is considered
+ * private and not linked into the fsmonitor daemon state.  The caller
+ * should fill this batch with one or more paths and then publish it.
+ */
+struct fsmonitor_batch *fsmonitor_batch__new(void);
+
+/*
+ * Free the list of batches starting with this one.
+ */
+void fsmonitor_batch__free_list(struct fsmonitor_batch *batch);
+
+/*
+ * Add this path to this batch of modified files.
+ *
+ * The batch should be private and NOT (yet) linked into the fsmonitor
+ * daemon state and therefore not yet visible to worker threads and so
+ * no locking is required.
+ */
+void fsmonitor_batch__add_path(struct fsmonitor_batch *batch, const char *path);
+
 struct fsmonitor_daemon_backend_data; /* opaque platform-specific data */
 
 struct fsmonitor_daemon_state {
@@ -91,5 +112,24 @@ enum fsmonitor_path_type fsmonitor_classify_path_absolute(
 	struct fsmonitor_daemon_state *state,
 	const char *path);
 
+/*
+ * Prepend the this batch of path(s) onto the list of batches associated
+ * with the current token.  This makes the batch visible to worker threads.
+ *
+ * The caller no longer owns the batch and must not free it.
+ *
+ * Wake up the client threads waiting on these cookies.
+ */
+void fsmonitor_publish(struct fsmonitor_daemon_state *state,
+		       struct fsmonitor_batch *batch,
+		       const struct string_list *cookie_names);
+
+/*
+ * If the platform-specific layer loses sync with the filesystem,
+ * it should call this to invalidate cached data and abort waiting
+ * threads.
+ */
+void fsmonitor_force_resync(struct fsmonitor_daemon_state *state);
+
 #endif /* HAVE_FSMONITOR_DAEMON_BACKEND */
 #endif /* FSMONITOR_DAEMON_H */
-- 
gitgitgadget

[PATCH v4 15/29] compat/fsmonitor/fsm-listen-win32: implement FSMonitor backend on Windows

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach the win32 backend to register a watch on the working tree
root directory (recursively).  Also watch the <gitdir> if it is
not inside the working tree.  And to collect path change notifications
into batches and publish.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 compat/fsmonitor/fsm-listen-win32.c | 565 ++++++++++++++++++++++++++++
 1 file changed, 565 insertions(+)

diff --git a/compat/fsmonitor/fsm-listen-win32.c b/compat/fsmonitor/fsm-listen-win32.c
index 916cbea254f..c2d11acbc1e 100644
--- a/compat/fsmonitor/fsm-listen-win32.c
+++ b/compat/fsmonitor/fsm-listen-win32.c
@@ -2,20 +2,585 @@
 #include "config.h"
 #include "fsmonitor.h"
 #include "fsm-listen.h"
+#include "fsmonitor--daemon.h"
+
+/*
+ * The documentation of ReadDirectoryChangesW() states that the maximum
+ * buffer size is 64K when the monitored directory is remote.
+ *
+ * Larger buffers may be used when the monitored directory is local and
+ * will help us receive events faster from the kernel and avoid dropped
+ * events.
+ *
+ * So we try to use a very large buffer and silently fallback to 64K if
+ * we get an error.
+ */
+#define MAX_RDCW_BUF_FALLBACK (65536)
+#define MAX_RDCW_BUF          (65536 * 8)
+
+struct one_watch
+{
+	char buffer[MAX_RDCW_BUF];
+	DWORD buf_len;
+	DWORD count;
+
+	struct strbuf path;
+	HANDLE hDir;
+	HANDLE hEvent;
+	OVERLAPPED overlapped;
+
+	/*
+	 * Is there an active ReadDirectoryChangesW() call pending.  If so, we
+	 * need to later call GetOverlappedResult() and possibly CancelIoEx().
+	 */
+	BOOL is_active;
+};
+
+struct fsmonitor_daemon_backend_data
+{
+	struct one_watch *watch_worktree;
+	struct one_watch *watch_gitdir;
+
+	HANDLE hEventShutdown;
+
+	HANDLE hListener[3]; /* we don't own these handles */
+#define LISTENER_SHUTDOWN 0
+#define LISTENER_HAVE_DATA_WORKTREE 1
+#define LISTENER_HAVE_DATA_GITDIR 2
+	int nr_listener_handles;
+};
+
+/*
+ * Convert the WCHAR path from the notification into UTF8 and
+ * then normalize it.
+ */
+static int normalize_path_in_utf8(FILE_NOTIFY_INFORMATION *info,
+				  struct strbuf *normalized_path)
+{
+	int reserve;
+	int len = 0;
+
+	strbuf_reset(normalized_path);
+	if (!info->FileNameLength)
+		goto normalize;
+
+	/*
+	 * Pre-reserve enough space in the UTF8 buffer for
+	 * each Unicode WCHAR character to be mapped into a
+	 * sequence of 2 UTF8 characters.  That should let us
+	 * avoid ERROR_INSUFFICIENT_BUFFER 99.9+% of the time.
+	 */
+	reserve = info->FileNameLength + 1;
+	strbuf_grow(normalized_path, reserve);
+
+	for (;;) {
+		len = WideCharToMultiByte(CP_UTF8, 0, info->FileName,
+					  info->FileNameLength / sizeof(WCHAR),
+					  normalized_path->buf,
+					  strbuf_avail(normalized_path) - 1,
+					  NULL, NULL);
+		if (len > 0)
+			goto normalize;
+		if (GetLastError() != ERROR_INSUFFICIENT_BUFFER) {
+			error("[GLE %ld] could not convert path to UTF-8: '%.*ls'",
+			      GetLastError(),
+			      (int)(info->FileNameLength / sizeof(WCHAR)),
+			      info->FileName);
+			return -1;
+		}
+
+		strbuf_grow(normalized_path,
+			    strbuf_avail(normalized_path) + reserve);
+	}
+
+normalize:
+	strbuf_setlen(normalized_path, len);
+	return strbuf_normalize_path(normalized_path);
+}
 
 void fsm_listen__stop_async(struct fsmonitor_daemon_state *state)
 {
+	SetEvent(state->backend_data->hListener[LISTENER_SHUTDOWN]);
+}
+
+static struct one_watch *create_watch(struct fsmonitor_daemon_state *state,
+				      const char *path)
+{
+	struct one_watch *watch = NULL;
+	DWORD desired_access = FILE_LIST_DIRECTORY;
+	DWORD share_mode =
+		FILE_SHARE_WRITE | FILE_SHARE_READ | FILE_SHARE_DELETE;
+	HANDLE hDir;
+	wchar_t wpath[MAX_PATH];
+
+	if (xutftowcs_path(wpath, path) < 0) {
+		error(_("could not convert to wide characters: '%s'"), path);
+		return NULL;
+	}
+
+	hDir = CreateFileW(wpath,
+			   desired_access, share_mode, NULL, OPEN_EXISTING,
+			   FILE_FLAG_BACKUP_SEMANTICS | FILE_FLAG_OVERLAPPED,
+			   NULL);
+	if (hDir == INVALID_HANDLE_VALUE) {
+		error(_("[GLE %ld] could not watch '%s'"),
+		      GetLastError(), path);
+		return NULL;
+	}
+
+	CALLOC_ARRAY(watch, 1);
+
+	watch->buf_len = sizeof(watch->buffer); /* assume full MAX_RDCW_BUF */
+
+	strbuf_init(&watch->path, 0);
+	strbuf_addstr(&watch->path, path);
+
+	watch->hDir = hDir;
+	watch->hEvent = CreateEvent(NULL, TRUE, FALSE, NULL);
+
+	return watch;
+}
+
+static void destroy_watch(struct one_watch *watch)
+{
+	if (!watch)
+		return;
+
+	strbuf_release(&watch->path);
+	if (watch->hDir != INVALID_HANDLE_VALUE)
+		CloseHandle(watch->hDir);
+	if (watch->hEvent != INVALID_HANDLE_VALUE)
+		CloseHandle(watch->hEvent);
+
+	free(watch);
+}
+
+static int start_rdcw_watch(struct fsmonitor_daemon_backend_data *data,
+			    struct one_watch *watch)
+{
+	DWORD dwNotifyFilter =
+		FILE_NOTIFY_CHANGE_FILE_NAME |
+		FILE_NOTIFY_CHANGE_DIR_NAME |
+		FILE_NOTIFY_CHANGE_ATTRIBUTES |
+		FILE_NOTIFY_CHANGE_SIZE |
+		FILE_NOTIFY_CHANGE_LAST_WRITE |
+		FILE_NOTIFY_CHANGE_CREATION;
+
+	ResetEvent(watch->hEvent);
+
+	memset(&watch->overlapped, 0, sizeof(watch->overlapped));
+	watch->overlapped.hEvent = watch->hEvent;
+
+	/*
+	 * Queue an async call using Overlapped IO.  This returns immediately.
+	 * Our event handle will be signalled when the real result is available.
+	 *
+	 * The return value here just means that we successfully queued it.
+	 * We won't know if the Read...() actually produces data until later.
+	 */
+	watch->is_active = ReadDirectoryChangesW(
+		watch->hDir, watch->buffer, watch->buf_len, TRUE,
+		dwNotifyFilter, &watch->count, &watch->overlapped, NULL);
+
+	if (watch->is_active)
+		return 0;
+
+	error("ReadDirectoryChangedW failed on '%s' [GLE %ld]",
+	      watch->path.buf, GetLastError());
+	return -1;
+}
+
+static int recv_rdcw_watch(struct one_watch *watch)
+{
+	DWORD gle;
+
+	watch->is_active = FALSE;
+
+	/*
+	 * The overlapped result is ready.  If the Read...() was successful
+	 * we finally receive the actual result into our buffer.
+	 */
+	if (GetOverlappedResult(watch->hDir, &watch->overlapped, &watch->count,
+				TRUE))
+		return 0;
+
+	gle = GetLastError();
+	if (gle == ERROR_INVALID_PARAMETER &&
+	    /*
+	     * The kernel throws an invalid parameter error when our
+	     * buffer is too big and we are pointed at a remote
+	     * directory (and possibly for other reasons).  Quietly
+	     * set it down and try again.
+	     *
+	     * See note about MAX_RDCW_BUF at the top.
+	     */
+	    watch->buf_len > MAX_RDCW_BUF_FALLBACK) {
+		watch->buf_len = MAX_RDCW_BUF_FALLBACK;
+		return -2;
+	}
+
+	/*
+	 * NEEDSWORK: If an external <gitdir> is deleted, the above
+	 * returns an error.  I'm not sure that there's anything that
+	 * we can do here other than failing -- the <worktree>/.git
+	 * link file would be broken anyway.  We might try to check
+	 * for that and return a better error message, but I'm not
+	 * sure it is worth it.
+	 */
+
+	error("GetOverlappedResult failed on '%s' [GLE %ld]",
+	      watch->path.buf, gle);
+	return -1;
+}
+
+static void cancel_rdcw_watch(struct one_watch *watch)
+{
+	DWORD count;
+
+	if (!watch || !watch->is_active)
+		return;
+
+	/*
+	 * The calls to ReadDirectoryChangesW() and GetOverlappedResult()
+	 * form a "pair" (my term) where we queue an IO and promise to
+	 * hang around and wait for the kernel to give us the result.
+	 *
+	 * If for some reason after we queue the IO, we have to quit
+	 * or otherwise not stick around for the second half, we must
+	 * tell the kernel to abort the IO.  This prevents the kernel
+	 * from writing to our buffer and/or signalling our event
+	 * after we free them.
+	 *
+	 * (Ask me how much fun it was to track that one down).
+	 */
+	CancelIoEx(watch->hDir, &watch->overlapped);
+	GetOverlappedResult(watch->hDir, &watch->overlapped, &count, TRUE);
+	watch->is_active = FALSE;
+}
+
+/*
+ * Process filesystem events that happen anywhere (recursively) under the
+ * <worktree> root directory.  For a normal working directory, this includes
+ * both version controlled files and the contents of the .git/ directory.
+ *
+ * If <worktree>/.git is a file, then we only see events for the file
+ * itself.
+ */
+static int process_worktree_events(struct fsmonitor_daemon_state *state)
+{
+	struct fsmonitor_daemon_backend_data *data = state->backend_data;
+	struct one_watch *watch = data->watch_worktree;
+	struct strbuf path = STRBUF_INIT;
+	struct string_list cookie_list = STRING_LIST_INIT_DUP;
+	struct fsmonitor_batch *batch = NULL;
+	const char *p = watch->buffer;
+
+	/*
+	 * If the kernel gets more events than will fit in the kernel
+	 * buffer associated with our RDCW handle, it drops them and
+	 * returns a count of zero.
+	 *
+	 * Yes, the call returns WITHOUT error and with length zero.
+	 * This is the documented behavior.  (My testing has confirmed
+	 * that it also sets the last error to ERROR_NOTIFY_ENUM_DIR,
+	 * but we do not rely on that since the function did not
+	 * return an error and it is not documented.)
+	 *
+	 * (The "overflow" case is not ambiguous with the "no data" case
+	 * because we did an INFINITE wait.)
+	 *
+	 * This means we have a gap in coverage.  Tell the daemon layer
+	 * to resync.
+	 */
+	if (!watch->count) {
+		trace2_data_string("fsmonitor", NULL, "fsm-listen/kernel",
+				   "overflow");
+		fsmonitor_force_resync(state);
+		return LISTENER_HAVE_DATA_WORKTREE;
+	}
+
+	/*
+	 * On Windows, `info` contains an "array" of paths that are
+	 * relative to the root of whichever directory handle received
+	 * the event.
+	 */
+	for (;;) {
+		FILE_NOTIFY_INFORMATION *info = (void *)p;
+		const char *slash;
+		enum fsmonitor_path_type t;
+
+		strbuf_reset(&path);
+		if (normalize_path_in_utf8(info, &path) == -1)
+			goto skip_this_path;
+
+		t = fsmonitor_classify_path_workdir_relative(path.buf);
+
+		switch (t) {
+		case IS_INSIDE_DOT_GIT_WITH_COOKIE_PREFIX:
+			/* special case cookie files within .git */
+
+			/* Use just the filename of the cookie file. */
+			slash = find_last_dir_sep(path.buf);
+			string_list_append(&cookie_list,
+					   slash ? slash + 1 : path.buf);
+			break;
+
+		case IS_INSIDE_DOT_GIT:
+			/* ignore everything inside of "<worktree>/.git/" */
+			break;
+
+		case IS_DOT_GIT:
+			/* "<worktree>/.git" was deleted (or renamed away) */
+			if ((info->Action == FILE_ACTION_REMOVED) ||
+			    (info->Action == FILE_ACTION_RENAMED_OLD_NAME)) {
+				trace2_data_string("fsmonitor", NULL,
+						   "fsm-listen/dotgit",
+						   "removed");
+				goto force_shutdown;
+			}
+			break;
+
+		case IS_WORKDIR_PATH:
+			/* queue normal pathname */
+			if (!batch)
+				batch = fsmonitor_batch__new();
+			fsmonitor_batch__add_path(batch, path.buf);
+			break;
+
+		case IS_GITDIR:
+		case IS_INSIDE_GITDIR:
+		case IS_INSIDE_GITDIR_WITH_COOKIE_PREFIX:
+		default:
+			BUG("unexpected path classification '%d' for '%s'",
+			    t, path.buf);
+		}
+
+skip_this_path:
+		if (!info->NextEntryOffset)
+			break;
+		p += info->NextEntryOffset;
+	}
+
+	fsmonitor_publish(state, batch, &cookie_list);
+	batch = NULL;
+	string_list_clear(&cookie_list, 0);
+	strbuf_release(&path);
+	return LISTENER_HAVE_DATA_WORKTREE;
+
+force_shutdown:
+	fsmonitor_batch__free_list(batch);
+	string_list_clear(&cookie_list, 0);
+	strbuf_release(&path);
+	return LISTENER_SHUTDOWN;
+}
+
+/*
+ * Process filesystem events that happened anywhere (recursively) under the
+ * external <gitdir> (such as non-primary worktrees or submodules).
+ * We only care about cookie files that our client threads created here.
+ *
+ * Note that we DO NOT get filesystem events on the external <gitdir>
+ * itself (it is not inside something that we are watching).  In particular,
+ * we do not get an event if the external <gitdir> is deleted.
+ */
+static int process_gitdir_events(struct fsmonitor_daemon_state *state)
+{
+	struct fsmonitor_daemon_backend_data *data = state->backend_data;
+	struct one_watch *watch = data->watch_gitdir;
+	struct strbuf path = STRBUF_INIT;
+	struct string_list cookie_list = STRING_LIST_INIT_DUP;
+	const char *p = watch->buffer;
+
+	if (!watch->count) {
+		trace2_data_string("fsmonitor", NULL, "fsm-listen/kernel",
+				   "overflow");
+		fsmonitor_force_resync(state);
+		return LISTENER_HAVE_DATA_GITDIR;
+	}
+
+	for (;;) {
+		FILE_NOTIFY_INFORMATION *info = (void *)p;
+		const char *slash;
+		enum fsmonitor_path_type t;
+
+		strbuf_reset(&path);
+		if (normalize_path_in_utf8(info, &path) == -1)
+			goto skip_this_path;
+
+		t = fsmonitor_classify_path_gitdir_relative(path.buf);
+
+		switch (t) {
+		case IS_INSIDE_GITDIR_WITH_COOKIE_PREFIX:
+			/* special case cookie files within gitdir */
+
+			/* Use just the filename of the cookie file. */
+			slash = find_last_dir_sep(path.buf);
+			string_list_append(&cookie_list,
+					   slash ? slash + 1 : path.buf);
+			break;
+
+		case IS_INSIDE_GITDIR:
+			goto skip_this_path;
+
+		default:
+			BUG("unexpected path classification '%d' for '%s'",
+			    t, path.buf);
+		}
+
+skip_this_path:
+		if (!info->NextEntryOffset)
+			break;
+		p += info->NextEntryOffset;
+	}
+
+	fsmonitor_publish(state, NULL, &cookie_list);
+	string_list_clear(&cookie_list, 0);
+	strbuf_release(&path);
+	return LISTENER_HAVE_DATA_GITDIR;
 }
 
 void fsm_listen__loop(struct fsmonitor_daemon_state *state)
 {
+	struct fsmonitor_daemon_backend_data *data = state->backend_data;
+	DWORD dwWait;
+	int result;
+
+	state->error_code = 0;
+
+	if (start_rdcw_watch(data, data->watch_worktree) == -1)
+		goto force_error_stop;
+
+	if (data->watch_gitdir &&
+	    start_rdcw_watch(data, data->watch_gitdir) == -1)
+		goto force_error_stop;
+
+	for (;;) {
+		dwWait = WaitForMultipleObjects(data->nr_listener_handles,
+						data->hListener,
+						FALSE, INFINITE);
+
+		if (dwWait == WAIT_OBJECT_0 + LISTENER_HAVE_DATA_WORKTREE) {
+			result = recv_rdcw_watch(data->watch_worktree);
+			if (result == -1) {
+				/* hard error */
+				goto force_error_stop;
+			}
+			if (result == -2) {
+				/* retryable error */
+				if (start_rdcw_watch(data, data->watch_worktree) == -1)
+					goto force_error_stop;
+				continue;
+			}
+
+			/* have data */
+			if (process_worktree_events(state) == LISTENER_SHUTDOWN)
+				goto force_shutdown;
+			if (start_rdcw_watch(data, data->watch_worktree) == -1)
+				goto force_error_stop;
+			continue;
+		}
+
+		if (dwWait == WAIT_OBJECT_0 + LISTENER_HAVE_DATA_GITDIR) {
+			result = recv_rdcw_watch(data->watch_gitdir);
+			if (result == -1) {
+				/* hard error */
+				goto force_error_stop;
+			}
+			if (result == -2) {
+				/* retryable error */
+				if (start_rdcw_watch(data, data->watch_gitdir) == -1)
+					goto force_error_stop;
+				continue;
+			}
+
+			/* have data */
+			if (process_gitdir_events(state) == LISTENER_SHUTDOWN)
+				goto force_shutdown;
+			if (start_rdcw_watch(data, data->watch_gitdir) == -1)
+				goto force_error_stop;
+			continue;
+		}
+
+		if (dwWait == WAIT_OBJECT_0 + LISTENER_SHUTDOWN)
+			goto clean_shutdown;
+
+		error(_("could not read directory changes [GLE %ld]"),
+		      GetLastError());
+		goto force_error_stop;
+	}
+
+force_error_stop:
+	state->error_code = -1;
+
+force_shutdown:
+	/*
+	 * Tell the IPC thead pool to stop (which completes the await
+	 * in the main thread (which will also signal this thread (if
+	 * we are still alive))).
+	 */
+	ipc_server_stop_async(state->ipc_server_data);
+
+clean_shutdown:
+	cancel_rdcw_watch(data->watch_worktree);
+	cancel_rdcw_watch(data->watch_gitdir);
 }
 
 int fsm_listen__ctor(struct fsmonitor_daemon_state *state)
 {
+	struct fsmonitor_daemon_backend_data *data;
+
+	CALLOC_ARRAY(data, 1);
+
+	data->hEventShutdown = CreateEvent(NULL, TRUE, FALSE, NULL);
+
+	data->watch_worktree = create_watch(state,
+					    state->path_worktree_watch.buf);
+	if (!data->watch_worktree)
+		goto failed;
+
+	if (state->nr_paths_watching > 1) {
+		data->watch_gitdir = create_watch(state,
+						  state->path_gitdir_watch.buf);
+		if (!data->watch_gitdir)
+			goto failed;
+	}
+
+	data->hListener[LISTENER_SHUTDOWN] = data->hEventShutdown;
+	data->nr_listener_handles++;
+
+	data->hListener[LISTENER_HAVE_DATA_WORKTREE] =
+		data->watch_worktree->hEvent;
+	data->nr_listener_handles++;
+
+	if (data->watch_gitdir) {
+		data->hListener[LISTENER_HAVE_DATA_GITDIR] =
+			data->watch_gitdir->hEvent;
+		data->nr_listener_handles++;
+	}
+
+	state->backend_data = data;
+	return 0;
+
+failed:
+	CloseHandle(data->hEventShutdown);
+	destroy_watch(data->watch_worktree);
+	destroy_watch(data->watch_gitdir);
+
 	return -1;
 }
 
 void fsm_listen__dtor(struct fsmonitor_daemon_state *state)
 {
+	struct fsmonitor_daemon_backend_data *data;
+
+	if (!state || !state->backend_data)
+		return;
+
+	data = state->backend_data;
+
+	CloseHandle(data->hEventShutdown);
+	destroy_watch(data->watch_worktree);
+	destroy_watch(data->watch_gitdir);
+
+	FREE_AND_NULL(state->backend_data);
 }
-- 
gitgitgadget

[PATCH v4 16/29] compat/fsmonitor/fsm-listen-darwin: add macos header files for FSEvent

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Include MacOS system declarations to allow us to use FSEvent and
CoreFoundation APIs.  We need GCC and clang versions because of
compiler and header file conflicts.

While it is quite possible to #include Apple's CoreServices.h when
compiling C source code with clang, trying to build it with GCC
currently fails with this error:

In file included
   from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/Library/Frameworks/Security.framework/Headers/AuthSession.h:32,
   from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/Library/Frameworks/Security.framework/Headers/Security.h:42,
   from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/Library/Frameworks/CoreServices.framework/Frameworks/OSServices.framework/Headers/CSIdentity.h:43,
   from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/Library/Frameworks/CoreServices.framework/Frameworks/OSServices.framework/Headers/OSServices.h:29,
   from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/Library/Frameworks/CoreServices.framework/Frameworks/LaunchServices.framework/Headers/IconsCore.h:23,
   from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/Library/Frameworks/CoreServices.framework/Frameworks/LaunchServices.framework/Headers/LaunchServices.h:23,
   from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/Library/Frameworks/CoreServices.framework/Headers/CoreServices.h:45,
     /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/Library/Frameworks/Security.framework/Headers/Authorization.h:193:7: error: variably modified 'bytes' at file scope
       193 | char bytes[kAuthorizationExternalFormLength];
           |      ^~~~~

The underlying reason is that GCC (rightfully) objects that an `enum`
value such as `kAuthorizationExternalFormLength` is not a constant
(because it is not, the preprocessor has no knowledge of it, only the
actual C compiler does) and can therefore not be used to define the size
of a C array.

This is a known problem and tracked in GCC's bug tracker:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93082

In the meantime, let's not block things and go the slightly ugly route
of declaring/defining the FSEvents constants, data structures and
functions that we need, so that we can avoid above-mentioned issue.

Let's do this _only_ for GCC, though, so that the CI/PR builds (which
build both with clang and with GCC) can guarantee that we _are_ using
the correct data types.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 compat/fsmonitor/fsm-listen-darwin.c | 96 ++++++++++++++++++++++++++++
 1 file changed, 96 insertions(+)

diff --git a/compat/fsmonitor/fsm-listen-darwin.c b/compat/fsmonitor/fsm-listen-darwin.c
index c84e3344ab9..f424253d3eb 100644
--- a/compat/fsmonitor/fsm-listen-darwin.c
+++ b/compat/fsmonitor/fsm-listen-darwin.c
@@ -1,3 +1,99 @@
+#if defined(__GNUC__)
+/*
+ * It is possible to #include CoreFoundation/CoreFoundation.h when compiling
+ * with clang, but not with GCC as of time of writing.
+ *
+ * See https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93082 for details.
+ */
+typedef unsigned int FSEventStreamCreateFlags;
+#define kFSEventStreamEventFlagNone               0x00000000
+#define kFSEventStreamEventFlagMustScanSubDirs    0x00000001
+#define kFSEventStreamEventFlagUserDropped        0x00000002
+#define kFSEventStreamEventFlagKernelDropped      0x00000004
+#define kFSEventStreamEventFlagEventIdsWrapped    0x00000008
+#define kFSEventStreamEventFlagHistoryDone        0x00000010
+#define kFSEventStreamEventFlagRootChanged        0x00000020
+#define kFSEventStreamEventFlagMount              0x00000040
+#define kFSEventStreamEventFlagUnmount            0x00000080
+#define kFSEventStreamEventFlagItemCreated        0x00000100
+#define kFSEventStreamEventFlagItemRemoved        0x00000200
+#define kFSEventStreamEventFlagItemInodeMetaMod   0x00000400
+#define kFSEventStreamEventFlagItemRenamed        0x00000800
+#define kFSEventStreamEventFlagItemModified       0x00001000
+#define kFSEventStreamEventFlagItemFinderInfoMod  0x00002000
+#define kFSEventStreamEventFlagItemChangeOwner    0x00004000
+#define kFSEventStreamEventFlagItemXattrMod       0x00008000
+#define kFSEventStreamEventFlagItemIsFile         0x00010000
+#define kFSEventStreamEventFlagItemIsDir          0x00020000
+#define kFSEventStreamEventFlagItemIsSymlink      0x00040000
+#define kFSEventStreamEventFlagOwnEvent           0x00080000
+#define kFSEventStreamEventFlagItemIsHardlink     0x00100000
+#define kFSEventStreamEventFlagItemIsLastHardlink 0x00200000
+#define kFSEventStreamEventFlagItemCloned         0x00400000
+
+typedef struct __FSEventStream *FSEventStreamRef;
+typedef const FSEventStreamRef ConstFSEventStreamRef;
+
+typedef unsigned int CFStringEncoding;
+#define kCFStringEncodingUTF8 0x08000100
+
+typedef const struct __CFString *CFStringRef;
+typedef const struct __CFArray *CFArrayRef;
+typedef const struct __CFRunLoop *CFRunLoopRef;
+
+struct FSEventStreamContext {
+    long long version;
+    void *cb_data, *retain, *release, *copy_description;
+};
+
+typedef struct FSEventStreamContext FSEventStreamContext;
+typedef unsigned int FSEventStreamEventFlags;
+#define kFSEventStreamCreateFlagNoDefer 0x02
+#define kFSEventStreamCreateFlagWatchRoot 0x04
+#define kFSEventStreamCreateFlagFileEvents 0x10
+
+typedef unsigned long long FSEventStreamEventId;
+#define kFSEventStreamEventIdSinceNow 0xFFFFFFFFFFFFFFFFULL
+
+typedef void (*FSEventStreamCallback)(ConstFSEventStreamRef streamRef,
+				      void *context,
+				      __SIZE_TYPE__ num_of_events,
+				      void *event_paths,
+				      const FSEventStreamEventFlags event_flags[],
+				      const FSEventStreamEventId event_ids[]);
+typedef double CFTimeInterval;
+FSEventStreamRef FSEventStreamCreate(void *allocator,
+				     FSEventStreamCallback callback,
+				     FSEventStreamContext *context,
+				     CFArrayRef paths_to_watch,
+				     FSEventStreamEventId since_when,
+				     CFTimeInterval latency,
+				     FSEventStreamCreateFlags flags);
+CFStringRef CFStringCreateWithCString(void *allocator, const char *string,
+				      CFStringEncoding encoding);
+CFArrayRef CFArrayCreate(void *allocator, const void **items, long long count,
+			 void *callbacks);
+void CFRunLoopRun(void);
+void CFRunLoopStop(CFRunLoopRef run_loop);
+CFRunLoopRef CFRunLoopGetCurrent(void);
+extern CFStringRef kCFRunLoopDefaultMode;
+void FSEventStreamScheduleWithRunLoop(FSEventStreamRef stream,
+				      CFRunLoopRef run_loop,
+				      CFStringRef run_loop_mode);
+unsigned char FSEventStreamStart(FSEventStreamRef stream);
+void FSEventStreamStop(FSEventStreamRef stream);
+void FSEventStreamInvalidate(FSEventStreamRef stream);
+void FSEventStreamRelease(FSEventStreamRef stream);
+#else
+/*
+ * Let Apple's headers declare `isalnum()` first, before
+ * Git's headers override it via a constant
+ */
+#include <string.h>
+#include <CoreFoundation/CoreFoundation.h>
+#include <CoreServices/CoreServices.h>
+#endif
+
 #include "cache.h"
 #include "fsmonitor.h"
 #include "fsm-listen.h"
-- 
gitgitgadget

[PATCH v4 17/29] compat/fsmonitor/fsm-listen-darwin: implement FSEvent listener on MacOS

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Implement file system event listener on MacOS using FSEvent,
CoreFoundation, and CoreServices.

Co-authored-by: Kevin Willford <Kevin.Willford@microsoft.com>
Co-authored-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 compat/fsmonitor/fsm-listen-darwin.c | 382 ++++++++++++++++++++++++++-
 1 file changed, 381 insertions(+), 1 deletion(-)

diff --git a/compat/fsmonitor/fsm-listen-darwin.c b/compat/fsmonitor/fsm-listen-darwin.c
index f424253d3eb..2aefdc14d89 100644
--- a/compat/fsmonitor/fsm-listen-darwin.c
+++ b/compat/fsmonitor/fsm-listen-darwin.c
@@ -1,4 +1,4 @@
-#if defined(__GNUC__)
+#ifndef __clang__
 /*
  * It is possible to #include CoreFoundation/CoreFoundation.h when compiling
  * with clang, but not with GCC as of time of writing.
@@ -97,20 +97,400 @@ void FSEventStreamRelease(FSEventStreamRef stream);
 #include "cache.h"
 #include "fsmonitor.h"
 #include "fsm-listen.h"
+#include "fsmonitor--daemon.h"
+
+struct fsmonitor_daemon_backend_data
+{
+	CFStringRef cfsr_worktree_path;
+	CFStringRef cfsr_gitdir_path;
+
+	CFArrayRef cfar_paths_to_watch;
+	int nr_paths_watching;
+
+	FSEventStreamRef stream;
+
+	CFRunLoopRef rl;
+
+	enum shutdown_style {
+		SHUTDOWN_EVENT = 0,
+		FORCE_SHUTDOWN,
+		FORCE_ERROR_STOP,
+	} shutdown_style;
+
+	unsigned int stream_scheduled:1;
+	unsigned int stream_started:1;
+};
+
+static void log_flags_set(const char *path, const FSEventStreamEventFlags flag)
+{
+	struct strbuf msg = STRBUF_INIT;
+
+	if (flag & kFSEventStreamEventFlagMustScanSubDirs)
+		strbuf_addstr(&msg, "MustScanSubDirs|");
+	if (flag & kFSEventStreamEventFlagUserDropped)
+		strbuf_addstr(&msg, "UserDropped|");
+	if (flag & kFSEventStreamEventFlagKernelDropped)
+		strbuf_addstr(&msg, "KernelDropped|");
+	if (flag & kFSEventStreamEventFlagEventIdsWrapped)
+		strbuf_addstr(&msg, "EventIdsWrapped|");
+	if (flag & kFSEventStreamEventFlagHistoryDone)
+		strbuf_addstr(&msg, "HistoryDone|");
+	if (flag & kFSEventStreamEventFlagRootChanged)
+		strbuf_addstr(&msg, "RootChanged|");
+	if (flag & kFSEventStreamEventFlagMount)
+		strbuf_addstr(&msg, "Mount|");
+	if (flag & kFSEventStreamEventFlagUnmount)
+		strbuf_addstr(&msg, "Unmount|");
+	if (flag & kFSEventStreamEventFlagItemChangeOwner)
+		strbuf_addstr(&msg, "ItemChangeOwner|");
+	if (flag & kFSEventStreamEventFlagItemCreated)
+		strbuf_addstr(&msg, "ItemCreated|");
+	if (flag & kFSEventStreamEventFlagItemFinderInfoMod)
+		strbuf_addstr(&msg, "ItemFinderInfoMod|");
+	if (flag & kFSEventStreamEventFlagItemInodeMetaMod)
+		strbuf_addstr(&msg, "ItemInodeMetaMod|");
+	if (flag & kFSEventStreamEventFlagItemIsDir)
+		strbuf_addstr(&msg, "ItemIsDir|");
+	if (flag & kFSEventStreamEventFlagItemIsFile)
+		strbuf_addstr(&msg, "ItemIsFile|");
+	if (flag & kFSEventStreamEventFlagItemIsHardlink)
+		strbuf_addstr(&msg, "ItemIsHardlink|");
+	if (flag & kFSEventStreamEventFlagItemIsLastHardlink)
+		strbuf_addstr(&msg, "ItemIsLastHardlink|");
+	if (flag & kFSEventStreamEventFlagItemIsSymlink)
+		strbuf_addstr(&msg, "ItemIsSymlink|");
+	if (flag & kFSEventStreamEventFlagItemModified)
+		strbuf_addstr(&msg, "ItemModified|");
+	if (flag & kFSEventStreamEventFlagItemRemoved)
+		strbuf_addstr(&msg, "ItemRemoved|");
+	if (flag & kFSEventStreamEventFlagItemRenamed)
+		strbuf_addstr(&msg, "ItemRenamed|");
+	if (flag & kFSEventStreamEventFlagItemXattrMod)
+		strbuf_addstr(&msg, "ItemXattrMod|");
+	if (flag & kFSEventStreamEventFlagOwnEvent)
+		strbuf_addstr(&msg, "OwnEvent|");
+	if (flag & kFSEventStreamEventFlagItemCloned)
+		strbuf_addstr(&msg, "ItemCloned|");
+
+	trace_printf_key(&trace_fsmonitor, "fsevent: '%s', flags=%u %s",
+			 path, flag, msg.buf);
+
+	strbuf_release(&msg);
+}
+
+static int ef_is_root_delete(const FSEventStreamEventFlags ef)
+{
+	return (ef & kFSEventStreamEventFlagItemIsDir &&
+		ef & kFSEventStreamEventFlagItemRemoved);
+}
+
+static int ef_is_root_renamed(const FSEventStreamEventFlags ef)
+{
+	return (ef & kFSEventStreamEventFlagItemIsDir &&
+		ef & kFSEventStreamEventFlagItemRenamed);
+}
+
+static int ef_is_dropped(const FSEventStreamEventFlags ef)
+{
+	return (ef & kFSEventStreamEventFlagMustScanSubDirs ||
+		ef & kFSEventStreamEventFlagKernelDropped ||
+		ef & kFSEventStreamEventFlagUserDropped);
+}
+
+static void fsevent_callback(ConstFSEventStreamRef streamRef,
+			     void *ctx,
+			     size_t num_of_events,
+			     void *event_paths,
+			     const FSEventStreamEventFlags event_flags[],
+			     const FSEventStreamEventId event_ids[])
+{
+	struct fsmonitor_daemon_state *state = ctx;
+	struct fsmonitor_daemon_backend_data *data = state->backend_data;
+	char **paths = (char **)event_paths;
+	struct fsmonitor_batch *batch = NULL;
+	struct string_list cookie_list = STRING_LIST_INIT_DUP;
+	const char *path_k;
+	const char *slash;
+	int k;
+	struct strbuf tmp = STRBUF_INIT;
+
+	/*
+	 * Build a list of all filesystem changes into a private/local
+	 * list and without holding any locks.
+	 */
+	for (k = 0; k < num_of_events; k++) {
+		/*
+		 * On Mac, we receive an array of absolute paths.
+		 */
+		path_k = paths[k];
+
+		/*
+		 * If you want to debug FSEvents, log them to GIT_TRACE_FSMONITOR.
+		 * Please don't log them to Trace2.
+		 *
+		 * trace_printf_key(&trace_fsmonitor, "Path: '%s'", path_k);
+		 */
+
+		/*
+		 * If event[k] is marked as dropped, we assume that we have
+		 * lost sync with the filesystem and should flush our cached
+		 * data.  We need to:
+		 *
+		 * [1] Abort/wake any client threads waiting for a cookie and
+		 *     flush the cached state data (the current token), and
+		 *     create a new token.
+		 *
+		 * [2] Discard the batch that we were locally building (since
+		 *     they are conceptually relative to the just flushed
+		 *     token).
+		 */
+		if (ef_is_dropped(event_flags[k])) {
+			if (trace_pass_fl(&trace_fsmonitor))
+				log_flags_set(path_k, event_flags[k]);
+
+			fsmonitor_force_resync(state);
+			fsmonitor_batch__free_list(batch);
+			string_list_clear(&cookie_list, 0);
+
+			/*
+			 * We assume that any events that we received
+			 * in this callback after this dropped event
+			 * may still be valid, so we continue rather
+			 * than break.  (And just in case there is a
+			 * delete of ".git" hiding in there.)
+			 */
+			continue;
+		}
+
+		switch (fsmonitor_classify_path_absolute(state, path_k)) {
+
+		case IS_INSIDE_DOT_GIT_WITH_COOKIE_PREFIX:
+		case IS_INSIDE_GITDIR_WITH_COOKIE_PREFIX:
+			/* special case cookie files within .git or gitdir */
+
+			/* Use just the filename of the cookie file. */
+			slash = find_last_dir_sep(path_k);
+			string_list_append(&cookie_list,
+					   slash ? slash + 1 : path_k);
+			break;
+
+		case IS_INSIDE_DOT_GIT:
+		case IS_INSIDE_GITDIR:
+			/* ignore all other paths inside of .git or gitdir */
+			break;
+
+		case IS_DOT_GIT:
+		case IS_GITDIR:
+			/*
+			 * If .git directory is deleted or renamed away,
+			 * we have to quit.
+			 */
+			if (ef_is_root_delete(event_flags[k])) {
+				trace_printf_key(&trace_fsmonitor,
+						 "event: gitdir removed");
+				goto force_shutdown;
+			}
+			if (ef_is_root_renamed(event_flags[k])) {
+				trace_printf_key(&trace_fsmonitor,
+						 "event: gitdir renamed");
+				goto force_shutdown;
+			}
+			break;
+
+		case IS_WORKDIR_PATH:
+			/* try to queue normal pathnames */
+
+			if (trace_pass_fl(&trace_fsmonitor))
+				log_flags_set(path_k, event_flags[k]);
+
+			/*
+			 * Because of the implicit "binning" (the
+			 * kernel calls us at a given frequency) and
+			 * de-duping (the kernel is free to combine
+			 * multiple events for a given pathname), an
+			 * individual fsevent could be marked as both
+			 * a file and directory.  Add it to the queue
+			 * with both spellings so that the client will
+			 * know how much to invalidate/refresh.
+			 */
+
+			if (event_flags[k] & kFSEventStreamEventFlagItemIsFile) {
+				const char *rel = path_k +
+					state->path_worktree_watch.len + 1;
+
+				if (!batch)
+					batch = fsmonitor_batch__new();
+				fsmonitor_batch__add_path(batch, rel);
+			}
+
+			if (event_flags[k] & kFSEventStreamEventFlagItemIsDir) {
+				const char *rel = path_k +
+					state->path_worktree_watch.len + 1;
+
+				strbuf_reset(&tmp);
+				strbuf_addstr(&tmp, rel);
+				strbuf_addch(&tmp, '/');
+
+				if (!batch)
+					batch = fsmonitor_batch__new();
+				fsmonitor_batch__add_path(batch, tmp.buf);
+			}
+
+			break;
+
+		case IS_OUTSIDE_CONE:
+		default:
+			trace_printf_key(&trace_fsmonitor,
+					 "ignoring '%s'", path_k);
+			break;
+		}
+	}
+
+	fsmonitor_publish(state, batch, &cookie_list);
+	string_list_clear(&cookie_list, 0);
+	strbuf_release(&tmp);
+	return;
+
+force_shutdown:
+	fsmonitor_batch__free_list(batch);
+	string_list_clear(&cookie_list, 0);
+
+	data->shutdown_style = FORCE_SHUTDOWN;
+	CFRunLoopStop(data->rl);
+	strbuf_release(&tmp);
+	return;
+}
+
+/*
+ * NEEDSWORK: Investigate the proper value for the `latency` argument
+ * in the call to `FSEventStreamCreate()`.  I'm not sure that this
+ * needs to be a config setting or just something that we tune after
+ * some testing.
+ *
+ * With a latency of 0.1, I was seeing lots of dropped events during
+ * the "touch 100000" files test within t/perf/p7519, but with a
+ * latency of 0.001 I did not see any dropped events.  So the
+ * "correct" value may be somewhere in between.
+ *
+ * https://developer.apple.com/documentation/coreservices/1443980-fseventstreamcreate
+ */
 
 int fsm_listen__ctor(struct fsmonitor_daemon_state *state)
 {
+	FSEventStreamCreateFlags flags = kFSEventStreamCreateFlagNoDefer |
+		kFSEventStreamCreateFlagWatchRoot |
+		kFSEventStreamCreateFlagFileEvents;
+	FSEventStreamContext ctx = {
+		0,
+		state,
+		NULL,
+		NULL,
+		NULL
+	};
+	struct fsmonitor_daemon_backend_data *data;
+	const void *dir_array[2];
+
+	CALLOC_ARRAY(data, 1);
+	state->backend_data = data;
+
+	data->cfsr_worktree_path = CFStringCreateWithCString(
+		NULL, state->path_worktree_watch.buf, kCFStringEncodingUTF8);
+	dir_array[data->nr_paths_watching++] = data->cfsr_worktree_path;
+
+	if (state->nr_paths_watching > 1) {
+		data->cfsr_gitdir_path = CFStringCreateWithCString(
+			NULL, state->path_gitdir_watch.buf,
+			kCFStringEncodingUTF8);
+		dir_array[data->nr_paths_watching++] = data->cfsr_gitdir_path;
+	}
+
+	data->cfar_paths_to_watch = CFArrayCreate(NULL, dir_array,
+						  data->nr_paths_watching,
+						  NULL);
+	data->stream = FSEventStreamCreate(NULL, fsevent_callback, &ctx,
+					   data->cfar_paths_to_watch,
+					   kFSEventStreamEventIdSinceNow,
+					   0.001, flags);
+	if (data->stream == NULL)
+		goto failed;
+
+	/*
+	 * `data->rl` needs to be set inside the listener thread.
+	 */
+
+	return 0;
+
+failed:
+	error("Unable to create FSEventStream.");
+
+	FREE_AND_NULL(state->backend_data);
 	return -1;
 }
 
 void fsm_listen__dtor(struct fsmonitor_daemon_state *state)
 {
+	struct fsmonitor_daemon_backend_data *data;
+
+	if (!state || !state->backend_data)
+		return;
+
+	data = state->backend_data;
+
+	if (data->stream) {
+		if (data->stream_started)
+			FSEventStreamStop(data->stream);
+		if (data->stream_scheduled)
+			FSEventStreamInvalidate(data->stream);
+		FSEventStreamRelease(data->stream);
+	}
+
+	FREE_AND_NULL(state->backend_data);
 }
 
 void fsm_listen__stop_async(struct fsmonitor_daemon_state *state)
 {
+	struct fsmonitor_daemon_backend_data *data;
+
+	data = state->backend_data;
+	data->shutdown_style = SHUTDOWN_EVENT;
+
+	CFRunLoopStop(data->rl);
 }
 
 void fsm_listen__loop(struct fsmonitor_daemon_state *state)
 {
+	struct fsmonitor_daemon_backend_data *data;
+
+	data = state->backend_data;
+
+	data->rl = CFRunLoopGetCurrent();
+
+	FSEventStreamScheduleWithRunLoop(data->stream, data->rl, kCFRunLoopDefaultMode);
+	data->stream_scheduled = 1;
+
+	if (!FSEventStreamStart(data->stream)) {
+		error("Failed to start the FSEventStream");
+		goto force_error_stop_without_loop;
+	}
+	data->stream_started = 1;
+
+	CFRunLoopRun();
+
+	switch (data->shutdown_style) {
+	case FORCE_ERROR_STOP:
+		state->error_code = -1;
+		/* fall thru */
+	case FORCE_SHUTDOWN:
+		ipc_server_stop_async(state->ipc_server_data);
+		/* fall thru */
+	case SHUTDOWN_EVENT:
+	default:
+		break;
+	}
+	return;
+
+force_error_stop_without_loop:
+	state->error_code = -1;
+	ipc_server_stop_async(state->ipc_server_data);
+	return;
 }
-- 
gitgitgadget

[PATCH v4 18/29] fsmonitor--daemon: implement handle_client callback

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach fsmonitor--daemon to respond to IPC requests from client
Git processes and respond with a list of modified pathnames
relative to the provided token.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 builtin/fsmonitor--daemon.c | 312 +++++++++++++++++++++++++++++++++++-
 1 file changed, 310 insertions(+), 2 deletions(-)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index 34603f23053..4c88171e06c 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -7,6 +7,7 @@
 #include "fsmonitor--daemon.h"
 #include "simple-ipc.h"
 #include "khash.h"
+#include "pkt-line.h"
 
 static const char * const builtin_fsmonitor__daemon_usage[] = {
 	N_("git fsmonitor--daemon start [<options>]"),
@@ -351,6 +352,311 @@ void fsmonitor_force_resync(struct fsmonitor_daemon_state *state)
 	pthread_mutex_unlock(&state->main_lock);
 }
 
+/*
+ * Format an opaque token string to send to the client.
+ */
+static void with_lock__format_response_token(
+	struct strbuf *response_token,
+	const struct strbuf *response_token_id,
+	const struct fsmonitor_batch *batch)
+{
+	/* assert current thread holding state->main_lock */
+
+	strbuf_reset(response_token);
+	strbuf_addf(response_token, "builtin:%s:%"PRIu64,
+		    response_token_id->buf, batch->batch_seq_nr);
+}
+
+/*
+ * Parse an opaque token from the client.
+ * Returns -1 on error.
+ */
+static int fsmonitor_parse_client_token(const char *buf_token,
+					struct strbuf *requested_token_id,
+					uint64_t *seq_nr)
+{
+	const char *p;
+	char *p_end;
+
+	strbuf_reset(requested_token_id);
+	*seq_nr = 0;
+
+	if (!skip_prefix(buf_token, "builtin:", &p))
+		return -1;
+
+	while (*p && *p != ':')
+		strbuf_addch(requested_token_id, *p++);
+	if (!*p++)
+		return -1;
+
+	*seq_nr = (uint64_t)strtoumax(p, &p_end, 10);
+	if (*p_end)
+		return -1;
+
+	return 0;
+}
+
+KHASH_INIT(str, const char *, int, 0, kh_str_hash_func, kh_str_hash_equal)
+
+static int do_handle_client(struct fsmonitor_daemon_state *state,
+			    const char *command,
+			    ipc_server_reply_cb *reply,
+			    struct ipc_server_reply_data *reply_data)
+{
+	struct fsmonitor_token_data *token_data = NULL;
+	struct strbuf response_token = STRBUF_INIT;
+	struct strbuf requested_token_id = STRBUF_INIT;
+	struct strbuf payload = STRBUF_INIT;
+	uint64_t requested_oldest_seq_nr = 0;
+	uint64_t total_response_len = 0;
+	const char *p;
+	const struct fsmonitor_batch *batch_head;
+	const struct fsmonitor_batch *batch;
+	intmax_t count = 0, duplicates = 0;
+	kh_str_t *shown;
+	int hash_ret;
+	int do_trivial = 0;
+	int do_flush = 0;
+
+	/*
+	 * We expect `command` to be of the form:
+	 *
+	 * <command> := quit NUL
+	 *            | flush NUL
+	 *            | <V1-time-since-epoch-ns> NUL
+	 *            | <V2-opaque-fsmonitor-token> NUL
+	 */
+
+	if (!strcmp(command, "quit")) {
+		/*
+		 * A client has requested over the socket/pipe that the
+		 * daemon shutdown.
+		 *
+		 * Tell the IPC thread pool to shutdown (which completes
+		 * the await in the main thread (which can stop the
+		 * fsmonitor listener thread)).
+		 *
+		 * There is no reply to the client.
+		 */
+		return SIMPLE_IPC_QUIT;
+
+	} else if (!strcmp(command, "flush")) {
+		/*
+		 * Flush all of our cached data and generate a new token
+		 * just like if we lost sync with the filesystem.
+		 *
+		 * Then send a trivial response using the new token.
+		 */
+		do_flush = 1;
+		do_trivial = 1;
+
+	} else if (!skip_prefix(command, "builtin:", &p)) {
+		/* assume V1 timestamp or garbage */
+
+		char *p_end;
+
+		strtoumax(command, &p_end, 10);
+		trace_printf_key(&trace_fsmonitor,
+				 ((*p_end) ?
+				  "fsmonitor: invalid command line '%s'" :
+				  "fsmonitor: unsupported V1 protocol '%s'"),
+				 command);
+		do_trivial = 1;
+
+	} else {
+		/* We have "builtin:*" */
+		if (fsmonitor_parse_client_token(command, &requested_token_id,
+						 &requested_oldest_seq_nr)) {
+			trace_printf_key(&trace_fsmonitor,
+					 "fsmonitor: invalid V2 protocol token '%s'",
+					 command);
+			do_trivial = 1;
+
+		} else {
+			/*
+			 * We have a V2 valid token:
+			 *     "builtin:<token_id>:<seq_nr>"
+			 */
+		}
+	}
+
+	pthread_mutex_lock(&state->main_lock);
+
+	if (!state->current_token_data)
+		BUG("fsmonitor state does not have a current token");
+
+	if (do_flush)
+		with_lock__do_force_resync(state);
+
+	/*
+	 * We mark the current head of the batch list as "pinned" so
+	 * that the listener thread will treat this item as read-only
+	 * (and prevent any more paths from being added to it) from
+	 * now on.
+	 */
+	token_data = state->current_token_data;
+	batch_head = token_data->batch_head;
+	((struct fsmonitor_batch *)batch_head)->pinned_time = time(NULL);
+
+	/*
+	 * FSMonitor Protocol V2 requires that we send a response header
+	 * with a "new current token" and then all of the paths that changed
+	 * since the "requested token".  We send the seq_nr of the just-pinned
+	 * head batch so that future requests from a client will be relative
+	 * to it.
+	 */
+	with_lock__format_response_token(&response_token,
+					 &token_data->token_id, batch_head);
+
+	reply(reply_data, response_token.buf, response_token.len + 1);
+	total_response_len += response_token.len + 1;
+
+	trace2_data_string("fsmonitor", the_repository, "response/token",
+			   response_token.buf);
+	trace_printf_key(&trace_fsmonitor, "response token: %s",
+			 response_token.buf);
+
+	if (!do_trivial) {
+		if (strcmp(requested_token_id.buf, token_data->token_id.buf)) {
+			/*
+			 * The client last spoke to a different daemon
+			 * instance -OR- the daemon had to resync with
+			 * the filesystem (and lost events), so reject.
+			 */
+			trace2_data_string("fsmonitor", the_repository,
+					   "response/token", "different");
+			do_trivial = 1;
+
+		} else if (requested_oldest_seq_nr <
+			   token_data->batch_tail->batch_seq_nr) {
+			/*
+			 * The client wants older events than we have for
+			 * this token_id.  This means that the end of our
+			 * batch list was truncated and we cannot give the
+			 * client a complete snapshot relative to their
+			 * request.
+			 */
+			trace_printf_key(&trace_fsmonitor,
+					 "client requested truncated data");
+			do_trivial = 1;
+		}
+	}
+
+	if (do_trivial) {
+		pthread_mutex_unlock(&state->main_lock);
+
+		reply(reply_data, "/", 2);
+
+		trace2_data_intmax("fsmonitor", the_repository,
+				   "response/trivial", 1);
+
+		strbuf_release(&response_token);
+		strbuf_release(&requested_token_id);
+		return 0;
+	}
+
+	/*
+	 * We're going to hold onto a pointer to the current
+	 * token-data while we walk the list of batches of files.
+	 * During this time, we will NOT be under the lock.
+	 * So we ref-count it.
+	 *
+	 * This allows the listener thread to continue prepending
+	 * new batches of items to the token-data (which we'll ignore).
+	 *
+	 * AND it allows the listener thread to do a token-reset
+	 * (and install a new `current_token_data`).
+	 */
+	token_data->client_ref_count++;
+
+	pthread_mutex_unlock(&state->main_lock);
+
+	/*
+	 * The client request is relative to the token that they sent,
+	 * so walk the batch list backwards from the current head back
+	 * to the batch (sequence number) they named.
+	 *
+	 * We use khash to de-dup the list of pathnames.
+	 *
+	 * NEEDSWORK: each batch contains a list of interned strings,
+	 * so we only need to do pointer comparisons here to build the
+	 * hash table.  Currently, we're still comparing the string
+	 * values.
+	 */
+	shown = kh_init_str();
+	for (batch = batch_head;
+	     batch && batch->batch_seq_nr > requested_oldest_seq_nr;
+	     batch = batch->next) {
+		size_t k;
+
+		for (k = 0; k < batch->nr; k++) {
+			const char *s = batch->interned_paths[k];
+			size_t s_len;
+
+			if (kh_get_str(shown, s) != kh_end(shown))
+				duplicates++;
+			else {
+				kh_put_str(shown, s, &hash_ret);
+
+				trace_printf_key(&trace_fsmonitor,
+						 "send[%"PRIuMAX"]: %s",
+						 count, s);
+
+				/* Each path gets written with a trailing NUL */
+				s_len = strlen(s) + 1;
+
+				if (payload.len + s_len >=
+				    LARGE_PACKET_DATA_MAX) {
+					reply(reply_data, payload.buf,
+					      payload.len);
+					total_response_len += payload.len;
+					strbuf_reset(&payload);
+				}
+
+				strbuf_add(&payload, s, s_len);
+				count++;
+			}
+		}
+	}
+
+	if (payload.len) {
+		reply(reply_data, payload.buf, payload.len);
+		total_response_len += payload.len;
+	}
+
+	kh_release_str(shown);
+
+	pthread_mutex_lock(&state->main_lock);
+
+	if (token_data->client_ref_count > 0)
+		token_data->client_ref_count--;
+
+	if (token_data->client_ref_count == 0) {
+		if (token_data != state->current_token_data) {
+			/*
+			 * The listener thread did a token-reset while we were
+			 * walking the batch list.  Therefore, this token is
+			 * stale and can be discarded completely.  If we are
+			 * the last reader thread using this token, we own
+			 * that work.
+			 */
+			fsmonitor_free_token_data(token_data);
+		}
+	}
+
+	pthread_mutex_unlock(&state->main_lock);
+
+	trace2_data_intmax("fsmonitor", the_repository, "response/length", total_response_len);
+	trace2_data_intmax("fsmonitor", the_repository, "response/count/files", count);
+	trace2_data_intmax("fsmonitor", the_repository, "response/count/duplicates", duplicates);
+
+	strbuf_release(&response_token);
+	strbuf_release(&requested_token_id);
+	strbuf_release(&payload);
+
+	return 0;
+}
+
 static ipc_server_application_cb handle_client;
 
 static int handle_client(void *data,
@@ -358,7 +664,7 @@ static int handle_client(void *data,
 			 ipc_server_reply_cb *reply,
 			 struct ipc_server_reply_data *reply_data)
 {
-	/* struct fsmonitor_daemon_state *state = data; */
+	struct fsmonitor_daemon_state *state = data;
 	int result;
 
 	/*
@@ -369,10 +675,12 @@ static int handle_client(void *data,
 	if (command_len != strlen(command))
 		BUG("FSMonitor assumes text messages");
 
+	trace_printf_key(&trace_fsmonitor, "requested token: %s", command);
+
 	trace2_region_enter("fsmonitor", "handle_client", the_repository);
 	trace2_data_string("fsmonitor", the_repository, "request", command);
 
-	result = 0; /* TODO Do something here. */
+	result = do_handle_client(state, command, reply, reply_data);
 
 	trace2_region_leave("fsmonitor", "handle_client", the_repository);
 
-- 
gitgitgadget

[PATCH v4 19/29] help: include fsmonitor--daemon feature flag in version info

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Add the "feature: fsmonitor--daemon" message to the output of
`git version --build-options`.

The builtin FSMonitor is only available on certain platforms and
even then only when certain Makefile flags are enabled, so print
a message in the verbose version output when it is available.

This can be used by test scripts for prereq testing.  Granted, tests
could just try `git fsmonitor--daemon status` and look for a 128 exit
code or grep for a "not supported" message on stderr, but this is
rather obscure.

The main advantage is that the feature message will automatically
appear in bug reports and other support requests.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 help.c        | 4 ++++
 t/test-lib.sh | 6 ++++++
 2 files changed, 10 insertions(+)

diff --git a/help.c b/help.c
index 973e47cdc30..708eed5d9ad 100644
--- a/help.c
+++ b/help.c
@@ -12,6 +12,7 @@
 #include "refs.h"
 #include "parse-options.h"
 #include "prompt.h"
+#include "fsmonitor-ipc.h"
 
 struct category_description {
 	uint32_t category;
@@ -695,6 +696,9 @@ void get_version_info(struct strbuf *buf, int show_build_options)
 		strbuf_addf(buf, "sizeof-size_t: %d\n", (int)sizeof(size_t));
 		strbuf_addf(buf, "shell-path: %s\n", SHELL_PATH);
 		/* NEEDSWORK: also save and output GIT-BUILD_OPTIONS? */
+
+		if (fsmonitor_ipc__is_supported())
+			strbuf_addstr(buf, "feature: fsmonitor--daemon\n");
 	}
 }
 
diff --git a/t/test-lib.sh b/t/test-lib.sh
index 151da80c561..4013ef8906e 100644
--- a/t/test-lib.sh
+++ b/t/test-lib.sh
@@ -1736,3 +1736,9 @@ test_lazy_prereq SHA1 '
 # Tests that verify the scheduler integration must set this locally
 # to avoid errors.
 GIT_TEST_MAINT_SCHEDULER="none:exit 1"
+
+# Does this platform support `git fsmonitor--daemon`
+#
+test_lazy_prereq FSMONITOR_DAEMON '
+	git version --build-options | grep "feature:" | grep "fsmonitor--daemon"
+'
-- 
gitgitgadget

[PATCH v4 20/29] t/helper/fsmonitor-client: create IPC client to talk to FSMonitor Daemon

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Create an IPC client to send query and flush commands to the daemon.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 Makefile                         |   1 +
 t/helper/test-fsmonitor-client.c | 121 +++++++++++++++++++++++++++++++
 t/helper/test-tool.c             |   1 +
 t/helper/test-tool.h             |   1 +
 4 files changed, 124 insertions(+)
 create mode 100644 t/helper/test-fsmonitor-client.c

diff --git a/Makefile b/Makefile
index 8a26696c9ed..5ad0cef69f4 100644
--- a/Makefile
+++ b/Makefile
@@ -710,6 +710,7 @@ TEST_BUILTINS_OBJS += test-dump-split-index.o
 TEST_BUILTINS_OBJS += test-dump-untracked-cache.o
 TEST_BUILTINS_OBJS += test-example-decorate.o
 TEST_BUILTINS_OBJS += test-fast-rebase.o
+TEST_BUILTINS_OBJS += test-fsmonitor-client.o
 TEST_BUILTINS_OBJS += test-genrandom.o
 TEST_BUILTINS_OBJS += test-genzeros.o
 TEST_BUILTINS_OBJS += test-getcwd.o
diff --git a/t/helper/test-fsmonitor-client.c b/t/helper/test-fsmonitor-client.c
new file mode 100644
index 00000000000..f7a5b3a32fa
--- /dev/null
+++ b/t/helper/test-fsmonitor-client.c
@@ -0,0 +1,121 @@
+/*
+ * test-fsmonitor-client.c: client code to send commands/requests to
+ * a `git fsmonitor--daemon` daemon.
+ */
+
+#include "test-tool.h"
+#include "cache.h"
+#include "parse-options.h"
+#include "fsmonitor-ipc.h"
+
+#ifndef HAVE_FSMONITOR_DAEMON_BACKEND
+int cmd__fsmonitor_client(int argc, const char **argv)
+{
+	die("fsmonitor--daemon not available on this platform");
+}
+#else
+
+/*
+ * Read the `.git/index` to get the last token written to the
+ * FSMonitor Index Extension.
+ */
+static const char *get_token_from_index(void)
+{
+	struct index_state *istate = the_repository->index;
+
+	if (do_read_index(istate, the_repository->index_file, 0) < 0)
+		die("unable to read index file");
+	if (!istate->fsmonitor_last_update)
+		die("index file does not have fsmonitor extension");
+
+	return istate->fsmonitor_last_update;
+}
+
+/*
+ * Send an IPC query to a `git-fsmonitor--daemon` daemon and
+ * ask for the changes since the given token or from the last
+ * token in the index extension.
+ *
+ * This will implicitly start a daemon process if necessary.  The
+ * daemon process will persist after we exit.
+ */
+static int do_send_query(const char *token)
+{
+	struct strbuf answer = STRBUF_INIT;
+	int ret;
+
+	if (!token || !*token)
+		token = get_token_from_index();
+
+	ret = fsmonitor_ipc__send_query(token, &answer);
+	if (ret < 0)
+		die(_("could not query fsmonitor--daemon"));
+
+	write_in_full(1, answer.buf, answer.len);
+	strbuf_release(&answer);
+
+	return 0;
+}
+
+/*
+ * Send a "flush" command to the `git-fsmonitor--daemon` (if running)
+ * and tell it to flush its cache.
+ *
+ * This feature is primarily used by the test suite to simulate a loss of
+ * sync with the filesystem where we miss kernel events.
+ */
+static int do_send_flush(void)
+{
+	struct strbuf answer = STRBUF_INIT;
+	int ret;
+
+	ret = fsmonitor_ipc__send_command("flush", &answer);
+	if (ret)
+		return ret;
+
+	write_in_full(1, answer.buf, answer.len);
+	strbuf_release(&answer);
+
+	return 0;
+}
+
+int cmd__fsmonitor_client(int argc, const char **argv)
+{
+	const char *subcmd;
+	const char *token = NULL;
+
+	const char * const fsmonitor_client_usage[] = {
+		N_("test-helper fsmonitor-client query [<token>]"),
+		N_("test-helper fsmonitor-client flush"),
+		NULL,
+	};
+
+	struct option options[] = {
+		OPT_STRING(0, "token", &token, N_("token"),
+			   N_("command token to send to the server")),
+		OPT_END()
+	};
+
+	if (argc < 2)
+		usage_with_options(fsmonitor_client_usage, options);
+
+	if (argc == 2 && !strcmp(argv[1], "-h"))
+		usage_with_options(fsmonitor_client_usage, options);
+
+	subcmd = argv[1];
+	argv--;
+	argc++;
+
+	argc = parse_options(argc, argv, NULL, options, fsmonitor_client_usage, 0);
+
+	setup_git_directory();
+
+	if (!strcmp(subcmd, "query"))
+		return !!do_send_query(token);
+
+	if (!strcmp(subcmd, "flush"))
+		return !!do_send_flush();
+
+	die("Unhandled subcommand: '%s'", subcmd);
+}
+#endif
diff --git a/t/helper/test-tool.c b/t/helper/test-tool.c
index 3ce5585e53a..39119844693 100644
--- a/t/helper/test-tool.c
+++ b/t/helper/test-tool.c
@@ -31,6 +31,7 @@ static struct test_cmd cmds[] = {
 	{ "dump-untracked-cache", cmd__dump_untracked_cache },
 	{ "example-decorate", cmd__example_decorate },
 	{ "fast-rebase", cmd__fast_rebase },
+	{ "fsmonitor-client", cmd__fsmonitor_client },
 	{ "genrandom", cmd__genrandom },
 	{ "genzeros", cmd__genzeros },
 	{ "getcwd", cmd__getcwd },
diff --git a/t/helper/test-tool.h b/t/helper/test-tool.h
index 9f0f5228508..0f1074a32a6 100644
--- a/t/helper/test-tool.h
+++ b/t/helper/test-tool.h
@@ -21,6 +21,7 @@ int cmd__dump_split_index(int argc, const char **argv);
 int cmd__dump_untracked_cache(int argc, const char **argv);
 int cmd__example_decorate(int argc, const char **argv);
 int cmd__fast_rebase(int argc, const char **argv);
+int cmd__fsmonitor_client(int argc, const char **argv);
 int cmd__genrandom(int argc, const char **argv);
 int cmd__genzeros(int argc, const char **argv);
 int cmd__getcwd(int argc, const char **argv);
-- 
gitgitgadget

[PATCH v4 21/29] t7527: create test for fsmonitor--daemon

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 t/t7527-builtin-fsmonitor.sh | 511 +++++++++++++++++++++++++++++++++++
 1 file changed, 511 insertions(+)
 create mode 100755 t/t7527-builtin-fsmonitor.sh

diff --git a/t/t7527-builtin-fsmonitor.sh b/t/t7527-builtin-fsmonitor.sh
new file mode 100755
index 00000000000..eea9ca1a309
--- /dev/null
+++ b/t/t7527-builtin-fsmonitor.sh
@@ -0,0 +1,511 @@
+#!/bin/sh
+
+test_description='built-in file system watcher'
+
+. ./test-lib.sh
+
+if ! test_have_prereq FSMONITOR_DAEMON
+then
+	skip_all="fsmonitor--daemon is not supported on this platform"
+	test_done
+fi
+
+stop_daemon_delete_repo () {
+	r=$1
+	git -C $r fsmonitor--daemon stop >/dev/null 2>/dev/null
+	rm -rf $1
+	return 0
+}
+
+start_daemon () {
+	case "$#" in
+		1) r="-C $1";;
+		*) r="";
+	esac
+
+	git $r fsmonitor--daemon start || return $?
+	git $r fsmonitor--daemon status || return $?
+
+	return 0
+}
+
+# Is a Trace2 data event present with the given catetory and key?
+# We do not care what the value is.
+#
+have_t2_data_event () {
+	c=$1
+	k=$2
+
+	grep -e '"event":"data".*"category":"'"$c"'".*"key":"'"$k"'"'
+}
+
+test_expect_success 'explicit daemon start and stop' '
+	test_when_finished "stop_daemon_delete_repo test_explicit" &&
+
+	git init test_explicit &&
+	start_daemon test_explicit &&
+
+	git -C test_explicit fsmonitor--daemon stop &&
+	test_must_fail git -C test_explicit fsmonitor--daemon status
+'
+
+test_expect_success 'implicit daemon start' '
+	test_when_finished "stop_daemon_delete_repo test_implicit" &&
+
+	git init test_implicit &&
+	test_must_fail git -C test_implicit fsmonitor--daemon status &&
+
+	# query will implicitly start the daemon.
+	#
+	# for test-script simplicity, we send a V1 timestamp rather than
+	# a V2 token.  either way, the daemon response to any query contains
+	# a new V2 token.  (the daemon may complain that we sent a V1 request,
+	# but this test case is only concerned with whether the daemon was
+	# implicitly started.)
+
+	GIT_TRACE2_EVENT="$(pwd)/.git/trace" \
+		test-tool -C test_implicit fsmonitor-client query --token 0 >actual &&
+	nul_to_q <actual >actual.filtered &&
+	grep "builtin:" actual.filtered &&
+
+	# confirm that a daemon was started in the background.
+	#
+	# since the mechanism for starting the background daemon is platform
+	# dependent, just confirm that the foreground command received a
+	# response from the daemon.
+
+	have_t2_data_event fsm_client query/response-length <.git/trace &&
+
+	git -C test_implicit fsmonitor--daemon status &&
+	git -C test_implicit fsmonitor--daemon stop &&
+	test_must_fail git -C test_implicit fsmonitor--daemon status
+'
+
+test_expect_success 'implicit daemon stop (delete .git)' '
+	test_when_finished "stop_daemon_delete_repo test_implicit_1" &&
+
+	git init test_implicit_1 &&
+
+	start_daemon test_implicit_1 &&
+
+	# deleting the .git directory will implicitly stop the daemon.
+	rm -rf test_implicit_1/.git &&
+
+	# [1] Create an empty .git directory so that the following Git
+	#     command will stay relative to the `-C` directory.
+	#
+	#     Without this, the Git command will override the requested
+	#     -C argument and crawl out to the containing Git source tree.
+	#     This would make the test result dependent upon whether we
+	#     were using fsmonitor on our development worktree.
+	#
+	sleep 1 &&
+	mkdir test_implicit_1/.git &&
+
+	test_must_fail git -C test_implicit_1 fsmonitor--daemon status
+'
+
+test_expect_success 'implicit daemon stop (rename .git)' '
+	test_when_finished "stop_daemon_delete_repo test_implicit_2" &&
+
+	git init test_implicit_2 &&
+
+	start_daemon test_implicit_2 &&
+
+	# renaming the .git directory will implicitly stop the daemon.
+	mv test_implicit_2/.git test_implicit_2/.xxx &&
+
+	# See [1] above.
+	#
+	sleep 1 &&
+	mkdir test_implicit_2/.git &&
+
+	test_must_fail git -C test_implicit_2 fsmonitor--daemon status
+'
+
+test_expect_success 'cannot start multiple daemons' '
+	test_when_finished "stop_daemon_delete_repo test_multiple" &&
+
+	git init test_multiple &&
+
+	start_daemon test_multiple &&
+
+	test_must_fail git -C test_multiple fsmonitor--daemon start 2>actual &&
+	grep "fsmonitor--daemon is already running" actual &&
+
+	git -C test_multiple fsmonitor--daemon stop &&
+	test_must_fail git -C test_multiple fsmonitor--daemon status
+'
+
+# These tests use the main repo in the trash directory
+
+test_expect_success 'setup' '
+	>tracked &&
+	>modified &&
+	>delete &&
+	>rename &&
+	mkdir dir1 &&
+	>dir1/tracked &&
+	>dir1/modified &&
+	>dir1/delete &&
+	>dir1/rename &&
+	mkdir dir2 &&
+	>dir2/tracked &&
+	>dir2/modified &&
+	>dir2/delete &&
+	>dir2/rename &&
+	mkdir dirtorename &&
+	>dirtorename/a &&
+	>dirtorename/b &&
+
+	cat >.gitignore <<-\EOF &&
+	.gitignore
+	expect*
+	actual*
+	EOF
+
+	git -c core.useBuiltinFSMonitor= add . &&
+	test_tick &&
+	git -c core.useBuiltinFSMonitor= commit -m initial &&
+
+	git config core.useBuiltinFSMonitor true
+'
+
+# The test already explicitly stopped (or tried to stop) the daemon.
+# This is here in case something else fails first.
+#
+redundant_stop_daemon () {
+	git fsmonitor--daemon stop
+	return 0
+}
+
+test_expect_success 'update-index implicitly starts daemon' '
+	test_when_finished redundant_stop_daemon &&
+
+	test_must_fail git fsmonitor--daemon status &&
+
+	GIT_TRACE2_EVENT="$(pwd)/.git/trace_implicit_1" \
+		git update-index --fsmonitor &&
+
+	git fsmonitor--daemon status &&
+	test_might_fail git fsmonitor--daemon stop &&
+
+	# Confirm that the trace2 log contains a record of the
+	# daemon starting.
+	test_subcommand git fsmonitor--daemon start <.git/trace_implicit_1
+'
+
+test_expect_success 'status implicitly starts daemon' '
+	test_when_finished redundant_stop_daemon &&
+
+	test_must_fail git fsmonitor--daemon status &&
+
+	GIT_TRACE2_EVENT="$(pwd)/.git/trace_implicit_2" \
+		git status >actual &&
+
+	git fsmonitor--daemon status &&
+	test_might_fail git fsmonitor--daemon stop &&
+
+	# Confirm that the trace2 log contains a record of the
+	# daemon starting.
+	test_subcommand git fsmonitor--daemon start <.git/trace_implicit_2
+'
+
+edit_files() {
+	echo 1 >modified
+	echo 2 >dir1/modified
+	echo 3 >dir2/modified
+	>dir1/untracked
+}
+
+delete_files() {
+	rm -f delete
+	rm -f dir1/delete
+	rm -f dir2/delete
+}
+
+create_files() {
+	echo 1 >new
+	echo 2 >dir1/new
+	echo 3 >dir2/new
+}
+
+rename_files() {
+	mv rename renamed
+	mv dir1/rename dir1/renamed
+	mv dir2/rename dir2/renamed
+}
+
+file_to_directory() {
+	rm -f delete
+	mkdir delete
+	echo 1 >delete/new
+}
+
+directory_to_file() {
+	rm -rf dir1
+	echo 1 >dir1
+}
+
+verify_status() {
+	git status >actual &&
+	GIT_INDEX_FILE=.git/fresh-index git read-tree master &&
+	GIT_INDEX_FILE=.git/fresh-index git -c core.useBuiltinFSMonitor= status >expect &&
+	test_cmp expect actual &&
+	echo HELLO AFTER &&
+	cat .git/trace &&
+	echo HELLO AFTER
+}
+
+# The next few test cases confirm that our fsmonitor daemon sees each type
+# of OS filesystem notification that we care about.  At this layer we just
+# ensure we are getting the OS notifications and do not try to confirm what
+# is reported by `git status`.
+#
+# We run a simple query after modifying the filesystem just to introduce
+# a bit of a delay so that the trace logging from the daemon has time to
+# get flushed to disk.
+#
+# We `reset` and `clean` at the bottom of each test (and before stopping the
+# daemon) because these commands might implicitly restart the daemon.
+
+clean_up_repo_and_stop_daemon () {
+	git reset --hard HEAD
+	git clean -fd
+	git fsmonitor--daemon stop
+	rm -f .git/trace
+}
+
+test_expect_success 'edit some files' '
+	test_when_finished clean_up_repo_and_stop_daemon &&
+
+	(
+		GIT_TRACE_FSMONITOR="$(pwd)/.git/trace" &&
+		export GIT_TRACE_FSMONITOR &&
+
+		start_daemon
+	) &&
+
+	edit_files &&
+
+	test-tool fsmonitor-client query --token 0 >/dev/null 2>&1 &&
+
+	grep "^event: dir1/modified$"  .git/trace &&
+	grep "^event: dir2/modified$"  .git/trace &&
+	grep "^event: modified$"       .git/trace &&
+	grep "^event: dir1/untracked$" .git/trace
+'
+
+test_expect_success 'create some files' '
+	test_when_finished clean_up_repo_and_stop_daemon &&
+
+	(
+		GIT_TRACE_FSMONITOR="$(pwd)/.git/trace" &&
+		export GIT_TRACE_FSMONITOR &&
+
+		start_daemon
+	) &&
+
+	create_files &&
+
+	test-tool fsmonitor-client query --token 0 >/dev/null 2>&1 &&
+
+	grep "^event: dir1/new$" .git/trace &&
+	grep "^event: dir2/new$" .git/trace &&
+	grep "^event: new$"      .git/trace
+'
+
+test_expect_success 'delete some files' '
+	test_when_finished clean_up_repo_and_stop_daemon &&
+
+	(
+		GIT_TRACE_FSMONITOR="$(pwd)/.git/trace" &&
+		export GIT_TRACE_FSMONITOR &&
+
+		start_daemon
+	) &&
+
+	delete_files &&
+
+	test-tool fsmonitor-client query --token 0 >/dev/null 2>&1 &&
+
+	grep "^event: dir1/delete$" .git/trace &&
+	grep "^event: dir2/delete$" .git/trace &&
+	grep "^event: delete$"      .git/trace
+'
+
+test_expect_success 'rename some files' '
+	test_when_finished clean_up_repo_and_stop_daemon &&
+
+	(
+		GIT_TRACE_FSMONITOR="$(pwd)/.git/trace" &&
+		export GIT_TRACE_FSMONITOR &&
+
+		start_daemon
+	) &&
+
+	rename_files &&
+
+	test-tool fsmonitor-client query --token 0 >/dev/null 2>&1 &&
+
+	grep "^event: dir1/rename$"  .git/trace &&
+	grep "^event: dir2/rename$"  .git/trace &&
+	grep "^event: rename$"       .git/trace &&
+	grep "^event: dir1/renamed$" .git/trace &&
+	grep "^event: dir2/renamed$" .git/trace &&
+	grep "^event: renamed$"      .git/trace
+'
+
+test_expect_success 'rename directory' '
+	test_when_finished clean_up_repo_and_stop_daemon &&
+
+	(
+		GIT_TRACE_FSMONITOR="$(pwd)/.git/trace" &&
+		export GIT_TRACE_FSMONITOR &&
+
+		start_daemon
+	) &&
+
+	mv dirtorename dirrenamed &&
+
+	test-tool fsmonitor-client query --token 0 >/dev/null 2>&1 &&
+
+	grep "^event: dirtorename/*$" .git/trace &&
+	grep "^event: dirrenamed/*$"  .git/trace
+'
+
+test_expect_success 'file changes to directory' '
+	test_when_finished clean_up_repo_and_stop_daemon &&
+
+	(
+		GIT_TRACE_FSMONITOR="$(pwd)/.git/trace" &&
+		export GIT_TRACE_FSMONITOR &&
+
+		start_daemon
+	) &&
+
+	file_to_directory &&
+
+	test-tool fsmonitor-client query --token 0 >/dev/null 2>&1 &&
+
+	grep "^event: delete$"     .git/trace &&
+	grep "^event: delete/new$" .git/trace
+'
+
+test_expect_success 'directory changes to a file' '
+	test_when_finished clean_up_repo_and_stop_daemon &&
+
+	(
+		GIT_TRACE_FSMONITOR="$(pwd)/.git/trace" &&
+		export GIT_TRACE_FSMONITOR &&
+
+		start_daemon
+	) &&
+
+	directory_to_file &&
+
+	test-tool fsmonitor-client query --token 0 >/dev/null 2>&1 &&
+
+	grep "^event: dir1$" .git/trace
+'
+
+# The next few test cases exercise the token-resync code.  When filesystem
+# drops events (because of filesystem velocity or because the daemon isn't
+# polling fast enough), we need to discard the cached data (relative to the
+# current token) and start collecting events under a new token.
+#
+# the 'test-tool fsmonitor-client flush' command can be used to send a
+# "flush" message to a running daemon and ask it to do a flush/resync.
+
+test_expect_success 'flush cached data' '
+	test_when_finished "stop_daemon_delete_repo test_flush" &&
+
+	git init test_flush &&
+
+	(
+		GIT_TEST_FSMONITOR_TOKEN=true &&
+		export GIT_TEST_FSMONITOR_TOKEN &&
+
+		GIT_TRACE_FSMONITOR="$(pwd)/.git/trace_daemon" &&
+		export GIT_TRACE_FSMONITOR &&
+
+		start_daemon test_flush
+	) &&
+
+	# The daemon should have an initial token with no events in _0 and
+	# then a few (probably platform-specific number of) events in _1.
+	# These should both have the same <token_id>.
+
+	test-tool -C test_flush fsmonitor-client query --token "builtin:test_00000001:0" >actual_0 &&
+	nul_to_q <actual_0 >actual_q0 &&
+
+	touch test_flush/file_1 &&
+	touch test_flush/file_2 &&
+
+	test-tool -C test_flush fsmonitor-client query --token "builtin:test_00000001:0" >actual_1 &&
+	nul_to_q <actual_1 >actual_q1 &&
+
+	grep "file_1" actual_q1 &&
+
+	# Force a flush.  This will change the <token_id>, reset the <seq_nr>, and
+	# flush the file data.  Then create some events and ensure that the file
+	# again appears in the cache.  It should have the new <token_id>.
+
+	test-tool -C test_flush fsmonitor-client flush >flush_0 &&
+	nul_to_q <flush_0 >flush_q0 &&
+	grep "^builtin:test_00000002:0Q/Q$" flush_q0 &&
+
+	test-tool -C test_flush fsmonitor-client query --token "builtin:test_00000002:0" >actual_2 &&
+	nul_to_q <actual_2 >actual_q2 &&
+
+	grep "^builtin:test_00000002:0Q$" actual_q2 &&
+
+	touch test_flush/file_3 &&
+
+	test-tool -C test_flush fsmonitor-client query --token "builtin:test_00000002:0" >actual_3 &&
+	nul_to_q <actual_3 >actual_q3 &&
+
+	grep "file_3" actual_q3
+'
+
+# The next few test cases create repos where the .git directory is NOT
+# inside the one of the working directory.  That is, where .git is a file
+# that points to a directory elsewhere.  This happens for submodules and
+# non-primary worktrees.
+
+test_expect_success 'setup worktree base' '
+	git init wt-base &&
+	echo 1 >wt-base/file1 &&
+	git -C wt-base add file1 &&
+	git -C wt-base commit -m "c1"
+'
+
+test_expect_success 'worktree with .git file' '
+	git -C wt-base worktree add ../wt-secondary &&
+
+	(
+		GIT_TRACE2_PERF="$(pwd)/trace2_wt_secondary" &&
+		export GIT_TRACE2_PERF &&
+
+		GIT_TRACE_FSMONITOR="$(pwd)/trace_wt_secondary" &&
+		export GIT_TRACE_FSMONITOR &&
+
+		start_daemon wt-secondary
+	) &&
+
+	git -C wt-secondary fsmonitor--daemon stop &&
+	test_must_fail git -C wt-secondary fsmonitor--daemon status
+'
+
+# NEEDSWORK: Repeat one of the "edit" tests on wt-secondary and
+# confirm that we get the same events and behavior -- that is, that
+# fsmonitor--daemon correctly watches BOTH the working directory and
+# the external GITDIR directory and behaves the same as when ".git"
+# is a directory inside the working directory.
+
+test_expect_success 'cleanup worktrees' '
+	stop_daemon_delete_repo wt-secondary &&
+	stop_daemon_delete_repo wt-base
+'
+
+test_done
-- 
gitgitgadget

[PATCH v4 22/29] t/perf: avoid copying builtin fsmonitor files into test repo

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Do not copy any of the various fsmonitor--daemon files from the .git
directory of the (GIT_PREF_REPO or GIT_PERF_LARGE_REPO) source repo
into the test's trash directory.

When perf tests start, they copy the contents of the source repo into
the test's trash directory.  If fsmonitor is running in the source repo,
there may be control files, such as the IPC socket and/or fsmonitor
cookie files.  These should not be copied into the test repo.

Unix domain sockets cannot be copied in the manner used by the test
setup, so if present, the test setup fails.

Cookie files are harmless, but we should avoid them.

The builtin fsmonitor keeps all such control files/sockets in
.git/fsmonitor--daemon*, so it is simple to exclude them.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 t/perf/perf-lib.sh | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/t/perf/perf-lib.sh b/t/perf/perf-lib.sh
index f5ed092ee59..dc29df12fd5 100644
--- a/t/perf/perf-lib.sh
+++ b/t/perf/perf-lib.sh
@@ -74,7 +74,7 @@ test_perf_copy_repo_contents () {
 	for stuff in "$1"/*
 	do
 		case "$stuff" in
-		*/objects|*/hooks|*/config|*/commondir|*/gitdir|*/worktrees)
+		*/objects|*/hooks|*/config|*/commondir|*/gitdir|*/worktrees|*/fsmonitor--daemon*)
 			;;
 		*)
 			cp -R "$stuff" "$repo/.git/" || exit 1
-- 
gitgitgadget

[PATCH v4 23/29] t/helper/test-chmtime: skip directories on Windows

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach `test-tool.exe chmtime` to ignore errors when setting the mtime
on a directory on Windows.

NEEDSWORK: The Windows version of `utime()` (aka `mingw_utime()`) does
not properly handle directories because it uses `_wopen()`.  It should
be converted to using `CreateFileW()` and backup semantics at a minimum.
Since I'm already in the middle of a large patch series, I did not want
to destabilize other callers of `utime()` right now.  The problem has
only been observed in the t/perf/p7519 test when the test repo contains
an empty directory on disk.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 t/helper/test-chmtime.c | 15 +++++++++++++++
 1 file changed, 15 insertions(+)

diff --git a/t/helper/test-chmtime.c b/t/helper/test-chmtime.c
index 524b55ca496..dc28890a183 100644
--- a/t/helper/test-chmtime.c
+++ b/t/helper/test-chmtime.c
@@ -134,6 +134,21 @@ int cmd__chmtime(int argc, const char **argv)
 		}
 
 		if (utb.modtime != sb.st_mtime && utime(argv[i], &utb) < 0) {
+#ifdef GIT_WINDOWS_NATIVE
+			if (S_ISDIR(sb.st_mode)) {
+				/*
+				 * NEEDSWORK: The Windows version of `utime()`
+				 * (aka `mingw_utime()`) does not correctly
+				 * handle directory arguments, since it uses
+				 * `_wopen()`.  Ignore it for now since this
+				 * is just a test.
+				 */
+				fprintf(stderr,
+					("Failed to modify time on directory %s. "
+					 "Skipping\n"), argv[i]);
+				continue;
+			}
+#endif
 			fprintf(stderr, "Failed to modify time on %s: %s\n",
 			        argv[i], strerror(errno));
 			return 1;
-- 
gitgitgadget

[PATCH v4 24/29] t/perf/p7519: speed up test on Windows

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Change p7519 to use `test_seq` and `xargs` rather than a `for` loop
to touch thousands of files.  This takes minutes off of test runs
on Windows because of process creation overhead.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 t/perf/p7519-fsmonitor.sh | 24 ++++++++++++++++--------
 1 file changed, 16 insertions(+), 8 deletions(-)

diff --git a/t/perf/p7519-fsmonitor.sh b/t/perf/p7519-fsmonitor.sh
index 5eb5044a103..171644ffc90 100755
--- a/t/perf/p7519-fsmonitor.sh
+++ b/t/perf/p7519-fsmonitor.sh
@@ -98,6 +98,13 @@ trace_stop() {
 	fi
 }
 
+touch_files() {
+	n=$1
+	d="$n"_files
+
+	(cd $d ; test_seq 1 $n | xargs touch )
+}
+
 test_expect_success "one time repo setup" '
 	# set untrackedCache depending on the environment
 	if test -n "$GIT_PERF_7519_UNTRACKED_CACHE"
@@ -119,10 +126,11 @@ test_expect_success "one time repo setup" '
 	fi &&
 
 	mkdir 1_file 10_files 100_files 1000_files 10000_files &&
-	for i in $(test_seq 1 10); do touch 10_files/$i; done &&
-	for i in $(test_seq 1 100); do touch 100_files/$i; done &&
-	for i in $(test_seq 1 1000); do touch 1000_files/$i; done &&
-	for i in $(test_seq 1 10000); do touch 10000_files/$i; done &&
+	touch_files 1 &&
+	touch_files 10 &&
+	touch_files 100 &&
+	touch_files 1000 &&
+	touch_files 10000 &&
 	git add 1_file 10_files 100_files 1000_files 10000_files &&
 	git commit -qm "Add files" &&
 
@@ -199,15 +207,15 @@ test_fsmonitor_suite() {
 
 	# Update the mtimes on upto 100k files to make status think
 	# that they are dirty.  For simplicity, omit any files with
-	# LFs (i.e. anything that ls-files thinks it needs to dquote).
-	# Then fully backslash-quote the paths to capture any
-	# whitespace so that they pass thru xargs properly.
+	# LFs (i.e. anything that ls-files thinks it needs to dquote)
+	# and any files with whitespace so that they pass thru xargs
+	# properly.
 	#
 	test_perf_w_drop_caches "status (dirty) ($DESC)" '
 		git ls-files | \
 			head -100000 | \
 			grep -v \" | \
-			sed '\''s/\(.\)/\\\1/g'\'' | \
+			egrep -v " ." | \
 			xargs test-tool chmtime -300 &&
 		git status
 	'
-- 
gitgitgadget

[PATCH v4 25/29] t/perf/p7519: add fsmonitor--daemon test cases

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Repeat all of the fsmonitor perf tests using `git fsmonitor--daemon` and
the "Simple IPC" interface.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 t/perf/p7519-fsmonitor.sh | 37 ++++++++++++++++++++++++++++++++++---
 1 file changed, 34 insertions(+), 3 deletions(-)

diff --git a/t/perf/p7519-fsmonitor.sh b/t/perf/p7519-fsmonitor.sh
index 171644ffc90..e70252ed65a 100755
--- a/t/perf/p7519-fsmonitor.sh
+++ b/t/perf/p7519-fsmonitor.sh
@@ -24,7 +24,8 @@ test_description="Test core.fsmonitor"
 # GIT_PERF_7519_SPLIT_INDEX: used to configure core.splitIndex
 # GIT_PERF_7519_FSMONITOR: used to configure core.fsMonitor. May be an
 #   absolute path to an integration. May be a space delimited list of
-#   absolute paths to integrations.
+#   absolute paths to integrations.  (This hook or list of hooks does not
+#   include the built-in fsmonitor--daemon.)
 #
 # The big win for using fsmonitor is the elimination of the need to scan the
 # working directory looking for changed and untracked files. If the file
@@ -143,10 +144,16 @@ test_expect_success "one time repo setup" '
 
 setup_for_fsmonitor() {
 	# set INTEGRATION_SCRIPT depending on the environment
-	if test -n "$INTEGRATION_PATH"
+	if test -n "$USE_FSMONITOR_DAEMON"
 	then
+		git config core.useBuiltinFSMonitor true &&
+		INTEGRATION_SCRIPT=false
+	elif test -n "$INTEGRATION_PATH"
+	then
+		git config core.useBuiltinFSMonitor false &&
 		INTEGRATION_SCRIPT="$INTEGRATION_PATH"
 	else
+		git config core.useBuiltinFSMonitor false &&
 		#
 		# Choose integration script based on existence of Watchman.
 		# Fall back to an empty integration script.
@@ -182,7 +189,10 @@ test_perf_w_drop_caches () {
 }
 
 test_fsmonitor_suite() {
-	if test -n "$INTEGRATION_SCRIPT"; then
+	if test -n "$USE_FSMONITOR_DAEMON"
+	then
+		DESC="builtin fsmonitor--daemon"
+	elif test -n "$INTEGRATION_SCRIPT"; then
 		DESC="fsmonitor=$(basename $INTEGRATION_SCRIPT)"
 	else
 		DESC="fsmonitor=disabled"
@@ -293,4 +303,25 @@ test_expect_success "setup without fsmonitor" '
 test_fsmonitor_suite
 trace_stop
 
+#
+# Run a full set of perf tests using the built-in fsmonitor--daemon.
+# It does not use the Hook API, so it has a different setup.
+# Explicitly start the daemon here and before we start client commands
+# so that we can later add custom tracing.
+#
+if test_have_prereq FSMONITOR_DAEMON
+then
+	USE_FSMONITOR_DAEMON=t
+
+	trace_start fsmonitor--daemon--server
+	git fsmonitor--daemon start
+
+	trace_start fsmonitor--daemon--client
+	test_expect_success "setup for fsmonitor--daemon" 'setup_for_fsmonitor'
+	test_fsmonitor_suite
+
+	git fsmonitor--daemon stop
+	trace_stop
+fi
+
 test_done
-- 
gitgitgadget

[PATCH v4 26/29] fsmonitor--daemon: periodically truncate list of modified files

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach fsmonitor--daemon to periodically truncate the list of
modified files to save some memory.

Clients will ask for the set of changes relative to a token that they
found in the FSMN index extension in the index.  (This token is like a
point in time, but different).  Clients will then update the index to
contain the response token (so that subsequent commands will be
relative to this new token).

Therefore, the daemon can gradually truncate the in-memory list of
changed paths as they become obsolete (older than the previous token).
Since we may have multiple clients making concurrent requests with a
skew of tokens and clients may be racing to the talk to the daemon,
we lazily truncate the list.

We introduce a 5 minute delay and truncate batches 5 minutes after
they are considered obsolete.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 builtin/fsmonitor--daemon.c | 88 +++++++++++++++++++++++++++++++++++++
 1 file changed, 88 insertions(+)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index 4c88171e06c..962b24569e1 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -299,6 +299,75 @@ static void fsmonitor_batch__combine(struct fsmonitor_batch *batch_dest,
 			batch_src->interned_paths[k];
 }
 
+/*
+ * To keep the batch list from growing unbounded in response to filesystem
+ * activity, we try to truncate old batches from the end of the list as
+ * they become irrelevant.
+ *
+ * We assume that the .git/index will be updated with the most recent token
+ * any time the index is updated.  And future commands will only ask for
+ * recent changes *since* that new token.  So as tokens advance into the
+ * future, older batch items will never be requested/needed.  So we can
+ * truncate them without loss of functionality.
+ *
+ * However, multiple commands may be talking to the daemon concurrently
+ * or perform a slow command, so a little "token skew" is possible.
+ * Therefore, we want this to be a little bit lazy and have a generous
+ * delay.
+ *
+ * The current reader thread walked backwards in time from `token->batch_head`
+ * back to `batch_marker` somewhere in the middle of the batch list.
+ *
+ * Let's walk backwards in time from that marker an arbitrary delay
+ * and truncate the list there.  Note that these timestamps are completely
+ * artificial (based on when we pinned the batch item) and not on any
+ * filesystem activity.
+ *
+ * Return the obsolete portion of the list after we have removed it from
+ * the official list so that the caller can free it after leaving the lock.
+ */
+#define MY_TIME_DELAY_SECONDS (5 * 60) /* seconds */
+
+static struct fsmonitor_batch *with_lock__truncate_old_batches(
+	struct fsmonitor_daemon_state *state,
+	const struct fsmonitor_batch *batch_marker)
+{
+	/* assert current thread holding state->main_lock */
+
+	const struct fsmonitor_batch *batch;
+	struct fsmonitor_batch *remainder;
+
+	if (!batch_marker)
+		return NULL;
+
+	trace_printf_key(&trace_fsmonitor, "Truncate: mark (%"PRIu64",%"PRIu64")",
+			 batch_marker->batch_seq_nr,
+			 (uint64_t)batch_marker->pinned_time);
+
+	for (batch = batch_marker; batch; batch = batch->next) {
+		time_t t;
+
+		if (!batch->pinned_time) /* an overflow batch */
+			continue;
+
+		t = batch->pinned_time + MY_TIME_DELAY_SECONDS;
+		if (t > batch_marker->pinned_time) /* too close to marker */
+			continue;
+
+		goto truncate_past_here;
+	}
+
+	return NULL;
+
+truncate_past_here:
+	state->current_token_data->batch_tail = (struct fsmonitor_batch *)batch;
+
+	remainder = ((struct fsmonitor_batch *)batch)->next;
+	((struct fsmonitor_batch *)batch)->next = NULL;
+
+	return remainder;
+}
+
 static void fsmonitor_free_token_data(struct fsmonitor_token_data *token)
 {
 	if (!token)
@@ -412,6 +481,7 @@ static int do_handle_client(struct fsmonitor_daemon_state *state,
 	const char *p;
 	const struct fsmonitor_batch *batch_head;
 	const struct fsmonitor_batch *batch;
+	struct fsmonitor_batch *remainder = NULL;
 	intmax_t count = 0, duplicates = 0;
 	kh_str_t *shown;
 	int hash_ret;
@@ -641,11 +711,29 @@ static int do_handle_client(struct fsmonitor_daemon_state *state,
 			 * that work.
 			 */
 			fsmonitor_free_token_data(token_data);
+		} else if (batch) {
+			/*
+			 * We are holding the lock and are the only
+			 * reader of the ref-counted portion of the
+			 * list, so we get the honor of seeing if the
+			 * list can be truncated to save memory.
+			 *
+			 * The main loop did not walk to the end of the
+			 * list, so this batch is the first item in the
+			 * batch-list that is older than the requested
+			 * end-point sequence number.  See if the tail
+			 * end of the list is obsolete.
+			 */
+			remainder = with_lock__truncate_old_batches(state,
+								    batch);
 		}
 	}
 
 	pthread_mutex_unlock(&state->main_lock);
 
+	if (remainder)
+		fsmonitor_batch__free_list(remainder);
+
 	trace2_data_intmax("fsmonitor", the_repository, "response/length", total_response_len);
 	trace2_data_intmax("fsmonitor", the_repository, "response/count/files", count);
 	trace2_data_intmax("fsmonitor", the_repository, "response/count/duplicates", duplicates);
-- 
gitgitgadget

[PATCH v4 27/29] fsmonitor--daemon: use a cookie file to sync with file system

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach fsmonitor--daemon client threads to create a cookie file
inside the .git directory and then wait until FS events for the
cookie are observed by the FS listener thread.

This helps address the racy nature of file system events by
blocking the client response until the kernel has drained any
event backlog.

This is especially important on MacOS where kernel events are
only issued with a limited frequency.  See the `latency` argument
of `FSeventStreamCreate()`.  The kernel only signals every `latency`
seconds, but does not guarantee that the kernel queue is completely
drained, so we may have to wait more than one interval.  If we
increase the frequency, the system is more likely to drop events.
We avoid these issues by having each client thread create a unique
cookie file and then wait until it is seen in the event stream.

Co-authored-by: Kevin Willford <Kevin.Willford@microsoft.com>
Co-authored-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 builtin/fsmonitor--daemon.c | 228 +++++++++++++++++++++++++++++++++++-
 fsmonitor--daemon.h         |   5 +
 2 files changed, 232 insertions(+), 1 deletion(-)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index 962b24569e1..6011fe42ee0 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -94,6 +94,149 @@ static int do_as_client__status(void)
 	}
 }
 
+enum fsmonitor_cookie_item_result {
+	FCIR_ERROR = -1, /* could not create cookie file ? */
+	FCIR_INIT = 0,
+	FCIR_SEEN,
+	FCIR_ABORT,
+};
+
+struct fsmonitor_cookie_item {
+	struct hashmap_entry entry;
+	const char *name;
+	enum fsmonitor_cookie_item_result result;
+};
+
+static int cookies_cmp(const void *data, const struct hashmap_entry *he1,
+		     const struct hashmap_entry *he2, const void *keydata)
+{
+	const struct fsmonitor_cookie_item *a =
+		container_of(he1, const struct fsmonitor_cookie_item, entry);
+	const struct fsmonitor_cookie_item *b =
+		container_of(he2, const struct fsmonitor_cookie_item, entry);
+
+	return strcmp(a->name, keydata ? keydata : b->name);
+}
+
+static enum fsmonitor_cookie_item_result with_lock__wait_for_cookie(
+	struct fsmonitor_daemon_state *state)
+{
+	/* assert current thread holding state->main_lock */
+
+	int fd;
+	struct fsmonitor_cookie_item *cookie;
+	struct strbuf cookie_pathname = STRBUF_INIT;
+	struct strbuf cookie_filename = STRBUF_INIT;
+	enum fsmonitor_cookie_item_result result;
+	int my_cookie_seq;
+
+	CALLOC_ARRAY(cookie, 1);
+
+	my_cookie_seq = state->cookie_seq++;
+
+	strbuf_addf(&cookie_filename, "%i-%i", getpid(), my_cookie_seq);
+
+	strbuf_addbuf(&cookie_pathname, &state->path_cookie_prefix);
+	strbuf_addbuf(&cookie_pathname, &cookie_filename);
+
+	cookie->name = strbuf_detach(&cookie_filename, NULL);
+	cookie->result = FCIR_INIT;
+	hashmap_entry_init(&cookie->entry, strhash(cookie->name));
+
+	hashmap_add(&state->cookies, &cookie->entry);
+
+	trace_printf_key(&trace_fsmonitor, "cookie-wait: '%s' '%s'",
+			 cookie->name, cookie_pathname.buf);
+
+	/*
+	 * Create the cookie file on disk and then wait for a notification
+	 * that the listener thread has seen it.
+	 */
+	fd = open(cookie_pathname.buf, O_WRONLY | O_CREAT | O_EXCL, 0600);
+	if (fd >= 0) {
+		close(fd);
+		unlink(cookie_pathname.buf);
+
+		/*
+		 * NEEDSWORK: This is an infinite wait (well, unless another
+		 * thread sends us an abort).  I'd like to change this to
+		 * use `pthread_cond_timedwait()` and return an error/timeout
+		 * and let the caller do the trivial response thing.
+		 */
+		while (cookie->result == FCIR_INIT)
+			pthread_cond_wait(&state->cookies_cond,
+					  &state->main_lock);
+	} else {
+		error_errno(_("could not create fsmonitor cookie '%s'"),
+			    cookie->name);
+
+		cookie->result = FCIR_ERROR;
+	}
+
+	hashmap_remove(&state->cookies, &cookie->entry, NULL);
+
+	result = cookie->result;
+
+	free((char*)cookie->name);
+	free(cookie);
+	strbuf_release(&cookie_pathname);
+
+	return result;
+}
+
+/*
+ * Mark these cookies as _SEEN and wake up the corresponding client threads.
+ */
+static void with_lock__mark_cookies_seen(struct fsmonitor_daemon_state *state,
+					 const struct string_list *cookie_names)
+{
+	/* assert current thread holding state->main_lock */
+
+	int k;
+	int nr_seen = 0;
+
+	for (k = 0; k < cookie_names->nr; k++) {
+		struct fsmonitor_cookie_item key;
+		struct fsmonitor_cookie_item *cookie;
+
+		key.name = cookie_names->items[k].string;
+		hashmap_entry_init(&key.entry, strhash(key.name));
+
+		cookie = hashmap_get_entry(&state->cookies, &key, entry, NULL);
+		if (cookie) {
+			trace_printf_key(&trace_fsmonitor, "cookie-seen: '%s'",
+					 cookie->name);
+			cookie->result = FCIR_SEEN;
+			nr_seen++;
+		}
+	}
+
+	if (nr_seen)
+		pthread_cond_broadcast(&state->cookies_cond);
+}
+
+/*
+ * Set _ABORT on all pending cookies and wake up all client threads.
+ */
+static void with_lock__abort_all_cookies(struct fsmonitor_daemon_state *state)
+{
+	/* assert current thread holding state->main_lock */
+
+	struct hashmap_iter iter;
+	struct fsmonitor_cookie_item *cookie;
+	int nr_aborted = 0;
+
+	hashmap_for_each_entry(&state->cookies, &iter, cookie, entry) {
+		trace_printf_key(&trace_fsmonitor, "cookie-abort: '%s'",
+				 cookie->name);
+		cookie->result = FCIR_ABORT;
+		nr_aborted++;
+	}
+
+	if (nr_aborted)
+		pthread_cond_broadcast(&state->cookies_cond);
+}
+
 /*
  * Requests to and from a FSMonitor Protocol V2 provider use an opaque
  * "token" as a virtual timestamp.  Clients can request a summary of all
@@ -391,6 +534,9 @@ static void fsmonitor_free_token_data(struct fsmonitor_token_data *token)
  *     We should create a new token and start fresh (as if we just
  *     booted up).
  *
+ * [2] Some of those lost events may have been for cookie files.  We
+ *     should assume the worst and abort them rather letting them starve.
+ *
  * If there are no concurrent threads readering the current token data
  * series, we can free it now.  Otherwise, let the last reader free
  * it.
@@ -412,6 +558,8 @@ static void with_lock__do_force_resync(struct fsmonitor_daemon_state *state)
 	state->current_token_data = new_one;
 
 	fsmonitor_free_token_data(free_me);
+
+	with_lock__abort_all_cookies(state);
 }
 
 void fsmonitor_force_resync(struct fsmonitor_daemon_state *state)
@@ -487,6 +635,8 @@ static int do_handle_client(struct fsmonitor_daemon_state *state,
 	int hash_ret;
 	int do_trivial = 0;
 	int do_flush = 0;
+	int do_cookie = 0;
+	enum fsmonitor_cookie_item_result cookie_result;
 
 	/*
 	 * We expect `command` to be of the form:
@@ -547,6 +697,7 @@ static int do_handle_client(struct fsmonitor_daemon_state *state,
 			 * We have a V2 valid token:
 			 *     "builtin:<token_id>:<seq_nr>"
 			 */
+			do_cookie = 1;
 		}
 	}
 
@@ -555,6 +706,30 @@ static int do_handle_client(struct fsmonitor_daemon_state *state,
 	if (!state->current_token_data)
 		BUG("fsmonitor state does not have a current token");
 
+	/*
+	 * Write a cookie file inside the directory being watched in
+	 * an effort to flush out existing filesystem events that we
+	 * actually care about.  Suspend this client thread until we
+	 * see the filesystem events for this cookie file.
+	 *
+	 * Creating the cookie lets us guarantee that our FS listener
+	 * thread has drained the kernel queue and we are caught up
+	 * with the kernel.
+	 *
+	 * If we cannot create the cookie (or otherwise guarantee that
+	 * we are caught up), we send a trivial response.  We have to
+	 * assume that there might be some very, very recent activity
+	 * on the FS still in flight.
+	 */
+	if (do_cookie) {
+		cookie_result = with_lock__wait_for_cookie(state);
+		if (cookie_result != FCIR_SEEN) {
+			error(_("fsmonitor: cookie_result '%d' != SEEN"),
+			      cookie_result);
+			do_trivial = 1;
+		}
+	}
+
 	if (do_flush)
 		with_lock__do_force_resync(state);
 
@@ -775,7 +950,9 @@ static int handle_client(void *data,
 	return result;
 }
 
-#define FSMONITOR_COOKIE_PREFIX ".fsmonitor-daemon-"
+#define FSMONITOR_DIR           "fsmonitor--daemon"
+#define FSMONITOR_COOKIE_DIR    "cookies"
+#define FSMONITOR_COOKIE_PREFIX (FSMONITOR_DIR "/" FSMONITOR_COOKIE_DIR "/")
 
 enum fsmonitor_path_type fsmonitor_classify_path_workdir_relative(
 	const char *rel)
@@ -928,6 +1105,9 @@ void fsmonitor_publish(struct fsmonitor_daemon_state *state,
 		}
 	}
 
+	if (cookie_names->nr)
+		with_lock__mark_cookies_seen(state, cookie_names);
+
 	pthread_mutex_unlock(&state->main_lock);
 }
 
@@ -1019,7 +1199,9 @@ static int fsmonitor_run_daemon(void)
 
 	memset(&state, 0, sizeof(state));
 
+	hashmap_init(&state.cookies, cookies_cmp, NULL, 0);
 	pthread_mutex_init(&state.main_lock, NULL);
+	pthread_cond_init(&state.cookies_cond, NULL);
 	state.error_code = 0;
 	state.current_token_data = fsmonitor_new_token_data();
 
@@ -1044,6 +1226,44 @@ static int fsmonitor_run_daemon(void)
 		state.nr_paths_watching = 2;
 	}
 
+	/*
+	 * We will write filesystem syncing cookie files into
+	 * <gitdir>/<fsmonitor-dir>/<cookie-dir>/<pid>-<seq>.
+	 *
+	 * The extra layers of subdirectories here keep us from
+	 * changing the mtime on ".git/" or ".git/foo/" when we create
+	 * or delete cookie files.
+	 *
+	 * There have been problems with some IDEs that do a
+	 * non-recursive watch of the ".git/" directory and run a
+	 * series of commands any time something happens.
+	 *
+	 * For example, if we place our cookie files directly in
+	 * ".git/" or ".git/foo/" then a `git status` (or similar
+	 * command) from the IDE will cause a cookie file to be
+	 * created in one of those dirs.  This causes the mtime of
+	 * those dirs to change.  This triggers the IDE's watch
+	 * notification.  This triggers the IDE to run those commands
+	 * again.  And the process repeats and the machine never goes
+	 * idle.
+	 *
+	 * Adding the extra layers of subdirectories prevents the
+	 * mtime of ".git/" and ".git/foo" from changing when a
+	 * cookie file is created.
+	 */
+	strbuf_init(&state.path_cookie_prefix, 0);
+	strbuf_addbuf(&state.path_cookie_prefix, &state.path_gitdir_watch);
+
+	strbuf_addch(&state.path_cookie_prefix, '/');
+	strbuf_addstr(&state.path_cookie_prefix, FSMONITOR_DIR);
+	mkdir(state.path_cookie_prefix.buf, 0777);
+
+	strbuf_addch(&state.path_cookie_prefix, '/');
+	strbuf_addstr(&state.path_cookie_prefix, FSMONITOR_COOKIE_DIR);
+	mkdir(state.path_cookie_prefix.buf, 0777);
+
+	strbuf_addch(&state.path_cookie_prefix, '/');
+
 	/*
 	 * Confirm that we can create platform-specific resources for the
 	 * filesystem listener before we bother starting all the threads.
@@ -1056,6 +1276,7 @@ static int fsmonitor_run_daemon(void)
 	err = fsmonitor_run_daemon_1(&state);
 
 done:
+	pthread_cond_destroy(&state.cookies_cond);
 	pthread_mutex_destroy(&state.main_lock);
 	fsm_listen__dtor(&state);
 
@@ -1063,6 +1284,11 @@ done:
 
 	strbuf_release(&state.path_worktree_watch);
 	strbuf_release(&state.path_gitdir_watch);
+	strbuf_release(&state.path_cookie_prefix);
+
+	/*
+	 * NEEDSWORK: Consider "rm -rf <gitdir>/<fsmonitor-dir>"
+	 */
 
 	return err;
 }
diff --git a/fsmonitor--daemon.h b/fsmonitor--daemon.h
index 20a815d80f8..c16ef095688 100644
--- a/fsmonitor--daemon.h
+++ b/fsmonitor--daemon.h
@@ -45,6 +45,11 @@ struct fsmonitor_daemon_state {
 
 	struct fsmonitor_token_data *current_token_data;
 
+	struct strbuf path_cookie_prefix;
+	pthread_cond_t cookies_cond;
+	int cookie_seq;
+	struct hashmap cookies;
+
 	int error_code;
 	struct fsmonitor_daemon_backend_data *backend_data;
 
-- 
gitgitgadget

[PATCH v4 28/29] fsmonitor: force update index after large responses

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Measure the time taken to apply the FSMonitor query result
to the index and the untracked-cache.

Set the `FSMONITOR_CHANGED` bit on `istate->cache_changed` when
FSMonitor returns a very large repsonse to ensure that the index is
written to disk.

Normally, when the FSMonitor response includes a tracked file, the
index is always updated.  Similarly, the index might be updated when
the response alters the untracked-cache (when enabled).  However, in
cases where neither of those cause the index to be considered changed,
the FSMonitor response is wasted.  Subsequent Git commands will make
requests with the same token and receive the same response.

If that response is very large, performance may suffer.  It would be
more efficient to force update the index now (and the token in the
index extension) in order to reduce the size of the response received
by future commands.

This was observed on Windows after a large checkout.  On Windows, the
kernel emits events for the files that are changed as they are
changed.  However, it might delay events for the containing
directories until the system is more idle (or someone scans the
directory (so it seems)).  The first status following a checkout would
get the list of files.  The subsequent status commands would get the
list of directories as the events trickled out.  But they would never
catch up because the token was not advanced because the index wasn't
updated.

This list of directories caused `wt_status_collect_untracked()` to
unnecessarily spend time actually scanning them during each command.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 fsmonitor.c | 57 ++++++++++++++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 56 insertions(+), 1 deletion(-)

diff --git a/fsmonitor.c b/fsmonitor.c
index 695fb0ce4e7..2befde1ffd2 100644
--- a/fsmonitor.c
+++ b/fsmonitor.c
@@ -233,6 +233,45 @@ static void fsmonitor_refresh_callback(struct index_state *istate, char *name)
 	untracked_cache_invalidate_path(istate, name, 0);
 }
 
+/*
+ * The number of pathnames that we need to receive from FSMonitor
+ * before we force the index to be updated.
+ *
+ * Note that any pathname within the set of received paths MAY cause
+ * cache-entry or istate flag bits to be updated and thus cause the
+ * index to be updated on disk.
+ *
+ * However, the response may contain many paths (such as ignored
+ * paths) that will not update any flag bits.  And thus not force the
+ * index to be updated.  (This is fine and normal.)  It also means
+ * that the token will not be updated in the FSMonitor index
+ * extension.  So the next Git command will find the same token in the
+ * index, make the same token-relative request, and receive the same
+ * response (plus any newly changed paths).  If this response is large
+ * (and continues to grow), performance could be impacted.
+ *
+ * For example, if the user runs a build and it writes 100K object
+ * files but doesn't modify any source files, the index would not need
+ * to be updated.  The FSMonitor response (after the build and
+ * relative to a pre-build token) might be 5MB.  Each subsequent Git
+ * command will receive that same 100K/5MB response until something
+ * causes the index to be updated.  And `refresh_fsmonitor()` will
+ * have to iterate over those 100K paths each time.
+ *
+ * Performance could be improved if we optionally force update the
+ * index after a very large response and get an updated token into
+ * the FSMonitor index extension.  This should allow subsequent
+ * commands to get smaller and more current responses.
+ *
+ * The value chosen here does not need to be precise.  The index
+ * will be updated automatically the first time the user touches
+ * a tracked file and causes a command like `git status` to
+ * update an mtime to be updated and/or set a flag bit.
+ *
+ * NEEDSWORK: Does this need to be a config value?
+ */
+static int fsmonitor_force_update_threshold = 100;
+
 void refresh_fsmonitor(struct index_state *istate)
 {
 	struct strbuf query_result = STRBUF_INIT;
@@ -364,25 +403,39 @@ apply_results:
 	 *     information and that we should consider everything
 	 *     invalid.  We call this a trivial response.
 	 */
+	trace2_region_enter("fsmonitor", "apply_results", istate->repo);
+
 	if (query_success && query_result.buf[bol] != '/') {
 		/*
 		 * Mark all pathnames returned by the monitor as dirty.
 		 *
 		 * This updates both the cache-entries and the untracked-cache.
 		 */
+		int count = 0;
+
 		buf = query_result.buf;
 		for (i = bol; i < query_result.len; i++) {
 			if (buf[i] != '\0')
 				continue;
 			fsmonitor_refresh_callback(istate, buf + bol);
 			bol = i + 1;
+			count++;
 		}
-		if (bol < query_result.len)
+		if (bol < query_result.len) {
 			fsmonitor_refresh_callback(istate, buf + bol);
+			count++;
+		}
 
 		/* Now mark the untracked cache for fsmonitor usage */
 		if (istate->untracked)
 			istate->untracked->use_fsmonitor = 1;
+
+		if (count > fsmonitor_force_update_threshold)
+			istate->cache_changed |= FSMONITOR_CHANGED;
+
+		trace2_data_intmax("fsmonitor", istate->repo, "apply_count",
+				   count);
+
 	} else {
 		/*
 		 * We received a trivial response, so invalidate everything.
@@ -410,6 +463,8 @@ apply_results:
 		if (istate->untracked)
 			istate->untracked->use_fsmonitor = 0;
 	}
+	trace2_region_leave("fsmonitor", "apply_results", istate->repo);
+
 	strbuf_release(&query_result);
 
 	/* Now that we've updated istate, save the last_update_token */
-- 
gitgitgadget

[PATCH v4 29/29] t7527: test status with untracked-cache and fsmonitor--daemon

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Create 2x2 test matrix with the untracked-cache and fsmonitor--daemon
features and a series of edits and verify that status output is
identical.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 t/t7527-builtin-fsmonitor.sh | 96 ++++++++++++++++++++++++++++++++++++
 1 file changed, 96 insertions(+)

diff --git a/t/t7527-builtin-fsmonitor.sh b/t/t7527-builtin-fsmonitor.sh
index eea9ca1a309..e62ec9aa3ca 100755
--- a/t/t7527-builtin-fsmonitor.sh
+++ b/t/t7527-builtin-fsmonitor.sh
@@ -162,6 +162,8 @@ test_expect_success 'setup' '
 	.gitignore
 	expect*
 	actual*
+	flush*
+	trace*
 	EOF
 
 	git -c core.useBuiltinFSMonitor= add . &&
@@ -508,4 +510,98 @@ test_expect_success 'cleanup worktrees' '
 	stop_daemon_delete_repo wt-base
 '
 
+# The next few tests perform arbitrary/contrived file operations and
+# confirm that status is correct.  That is, that the data (or lack of
+# data) from fsmonitor doesn't cause incorrect results.  And doesn't
+# cause incorrect results when the untracked-cache is enabled.
+
+test_lazy_prereq UNTRACKED_CACHE '
+	{ git update-index --test-untracked-cache; ret=$?; } &&
+	test $ret -ne 1
+'
+
+test_expect_success 'Matrix: setup for untracked-cache,fsmonitor matrix' '
+	test_might_fail git config --unset core.useBuiltinFSMonitor &&
+	git update-index --no-fsmonitor &&
+	test_might_fail git fsmonitor--daemon stop
+'
+
+matrix_clean_up_repo () {
+	git reset --hard HEAD
+	git clean -fd
+}
+
+matrix_try () {
+	uc=$1
+	fsm=$2
+	fn=$3
+
+	test_expect_success "Matrix[uc:$uc][fsm:$fsm] $fn" '
+		matrix_clean_up_repo &&
+		$fn &&
+		if test $uc = false -a $fsm = false
+		then
+			git status --porcelain=v1 >.git/expect.$fn
+		else
+			git status --porcelain=v1 >.git/actual.$fn
+			test_cmp .git/expect.$fn .git/actual.$fn
+		fi
+	'
+
+	return $?
+}
+
+uc_values="false"
+test_have_prereq UNTRACKED_CACHE && uc_values="false true"
+for uc_val in $uc_values
+do
+	if test $uc_val = false
+	then
+		test_expect_success "Matrix[uc:$uc_val] disable untracked cache" '
+			git config core.untrackedcache false &&
+			git update-index --no-untracked-cache
+		'
+	else
+		test_expect_success "Matrix[uc:$uc_val] enable untracked cache" '
+			git config core.untrackedcache true &&
+			git update-index --untracked-cache
+		'
+	fi
+
+	fsm_values="false true"
+	for fsm_val in $fsm_values
+	do
+		if test $fsm_val = false
+		then
+			test_expect_success "Matrix[uc:$uc_val][fsm:$fsm_val] disable fsmonitor" '
+				test_might_fail git config --unset core.useBuiltinFSMonitor &&
+				git update-index --no-fsmonitor &&
+				test_might_fail git fsmonitor--daemon stop 2>/dev/null
+			'
+		else
+			test_expect_success "Matrix[uc:$uc_val][fsm:$fsm_val] enable fsmonitor" '
+				git config core.useBuiltinFSMonitor true &&
+				git fsmonitor--daemon start &&
+				git update-index --fsmonitor
+			'
+		fi
+
+		matrix_try $uc_val $fsm_val edit_files
+		matrix_try $uc_val $fsm_val delete_files
+		matrix_try $uc_val $fsm_val create_files
+		matrix_try $uc_val $fsm_val rename_files
+		matrix_try $uc_val $fsm_val file_to_directory
+		matrix_try $uc_val $fsm_val directory_to_file
+
+		if test $fsm_val = true
+		then
+			test_expect_success "Matrix[uc:$uc_val][fsm:$fsm_val] disable fsmonitor at end" '
+				test_might_fail git config --unset core.useBuiltinFSMonitor &&
+				git update-index --no-fsmonitor &&
+				test_might_fail git fsmonitor--daemon stop 2>/dev/null
+			'
+		fi
+	done
+done
+
 test_done
-- 
gitgitgadget

Re: [PATCH v4 01/29] fsmonitor: enhance existing comments

[Junio C Hamano]

"Jeff Hostetler via GitGitGadget" <gitgitgadget@gmail.com> writes:

> From: Jeff Hostetler <jeffhost@microsoft.com>
>
> Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
> ---
>  fsmonitor.c | 37 ++++++++++++++++++++++++++++++-------
>  1 file changed, 30 insertions(+), 7 deletions(-)
>
> diff --git a/fsmonitor.c b/fsmonitor.c
> index ab9bfc60b34..ec4c46407c5 100644
> --- a/fsmonitor.c
> +++ b/fsmonitor.c
> @@ -301,9 +301,25 @@ void refresh_fsmonitor(struct index_state *istate)
>  			core_fsmonitor, query_success ? "success" : "failure");
>  	}
>  
> -	/* a fsmonitor process can return '/' to indicate all entries are invalid */
> +	/*
> +	 * The response from FSMonitor (excluding the header token) is
> +	 * either:
> +	 *
> +	 * [a] a (possibly empty) list of NUL delimited relative
> +	 *     pathnames of changed paths.  This list can contain
> +	 *     files and directories.  Directories have a trailing
> +	 *     slash.
> +	 *
> +	 * [b] a single '/' to indicate the provider had no
> +	 *     information and that we should consider everything
> +	 *     invalid.  We call this a trivial response.
> +	 */
>  	if (query_success && query_result.buf[bol] != '/') {
> -		/* Mark all entries returned by the monitor as dirty */
> +		/*
> +		 * Mark all pathnames returned by the monitor as dirty.
> +		 *
> +		 * This updates both the cache-entries and the untracked-cache.
> +		 */

Not a problem this patch introduces, but we only checked that the
query result begins with a slash, not "we did receive a trivial
response", but the "else" clause of this statement pretends as if we
did.

It is a shame that we do have fsmonitor_is_trivial_response()
function defined, but its interface is not capable of helping us
here.

Or is fsmonitor_is_trivial_response() already good to do this, and
reliance of [bol] this code has is the source of confusion?  I
notice that when we have last update token and makes a call to
query_fsmonitor() with HOOK_INTERFACE_VERSION1, nobody updates bol
(hence it stays 0), and with HOOK_INTERFACE_VERSION2, bol is at the
NUL that terminates the initial string of the query result, after
which presumably has either '/' NUL (trivial) or list of paths.

I am not sure about the VERSION1 case, but at least in the VERSION2
case, making sure that the last three bytes are NUL slash NUL like
fsmonitor_is_trivial_response() does and the half check the above is
doing (i.e. the byte after the NUL is slash, without making sure
about the length of the whole response or what follows the slash is
NUL), seems "close enough" (meaning: the check in this code is a
sloppy attempt to reinvent what _is_trivial_response() function
already does).

So, would it make sense to rewrite the condition to

	if (query_success &&
	    !fsmonitor_is_trivial_response(&query_result)) {

here?  Or perhaps

	if (query_success &&
	    !(query_result.len == bol + 3 &&
	      query_result[bol] == '/' && !query_result[bol+1])) {

which would be open coding the _is_trivial_response() function.

>  		buf = query_result.buf;
>  		for (i = bol; i < query_result.len; i++) {
>  			if (buf[i] != '\0')
> @@ -318,11 +334,15 @@ void refresh_fsmonitor(struct index_state *istate)
>  		if (istate->untracked)
>  			istate->untracked->use_fsmonitor = 1;
>  	} else {
> -
> -		/* We only want to run the post index changed hook if we've actually changed entries, so keep track
> -		 * if we actually changed entries or not */
> +		/*
> +		 * We received a trivial response, so invalidate everything.
> +		 *


> +		 * We only want to run the post index changed hook if
> +		 * we've actually changed entries, so keep track if we
> +		 * actually changed entries or not.
> +		 */
>  		int is_cache_changed = 0;
> -		/* Mark all entries invalid */
> +
>  		for (i = 0; i < istate->cache_nr; i++) {
>  			if (istate->cache[i]->ce_flags & CE_FSMONITOR_VALID) {
>  				is_cache_changed = 1;
> @@ -330,7 +350,10 @@ void refresh_fsmonitor(struct index_state *istate)
>  			}
>  		}
>  
> -		/* If we're going to check every file, ensure we save the results */
> +		/*
> +		 * If we're going to check every file, ensure we save
> +		 * the results.
> +		 */
>  		if (is_cache_changed)
>  			istate->cache_changed |= FSMONITOR_CHANGED;

Re: [PATCH v4 03/29] fsmonitor: config settings are repository-specific

[Junio C Hamano]

"Jeff Hostetler via GitGitGadget" <gitgitgadget@gmail.com> writes:

>  	if (fsmonitor > 0) {
> -		if (git_config_get_fsmonitor() == 0)
> +		enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
> +
> +		if (fsm_mode == FSMONITOR_MODE_DISABLED) {
> +			warning(_("core.useBuiltinFSMonitor is unset; "
> +				"set it if you really want to enable the "
> +				"builtin fsmonitor"));
>  			warning(_("core.fsmonitor is unset; "
> -				"set it if you really want to "
> -				"enable fsmonitor"));
> +				"set it if you really want to enable the "
> +				"hook-based fsmonitor"));
> +		}
>  		add_fsmonitor(&the_index);
>  		report(_("fsmonitor enabled"));
>  	} else if (!fsmonitor) {
> -		if (git_config_get_fsmonitor() == 1)
> +		enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
> +		if (fsm_mode == FSMONITOR_MODE_IPC)
> +			warning(_("core.useBuiltinFSMonitor is set; "
> +				"remove it if you really want to "
> +				"disable fsmonitor"));
> +		if (fsm_mode == FSMONITOR_MODE_HOOK)
>  			warning(_("core.fsmonitor is set; "
>  				"remove it if you really want to "
>  				"disable fsmonitor"));

Hmph.  This does not change the behaviour per-se, but what are we
trying to achieve with these warning messages?  

The user uses --fsmonitor or --no-fsmonitor option from the command
line, presumably as a one-shot "this time I'd operate the command
differently from the configured way", so it seems unlikely that the
user is doing so because "... really want to enable/disable".  The
"report()" calls in these if/else cases seem sufficient reminder of
what is going on---perhaps these warnings should be made silenceable
by turning them into advice messages?

> -int git_config_get_fsmonitor(void)
> -{
> -	if (git_config_get_pathname("core.fsmonitor", &core_fsmonitor))
> -		core_fsmonitor = getenv("GIT_TEST_FSMONITOR");
> -
> -	if (core_fsmonitor && !*core_fsmonitor)
> -		core_fsmonitor = NULL;
> -
> -	if (core_fsmonitor)
> -		return 1;
> -
> -	return 0;
> -}

This used to be how we got the configuration.

> --- a/config.h
> +++ b/config.h
> @@ -610,7 +610,6 @@ int git_config_get_pathname(const char *key, const char **dest);
>  int git_config_get_index_threads(int *dest);
>  int git_config_get_split_index(void);
>  int git_config_get_max_percent_split_change(void);
> -int git_config_get_fsmonitor(void);

And that is removed so any in-flight topic that adds new caller will
be caught by the compiler.  OK.

> diff --git a/environment.c b/environment.c
> index 9da7f3c1a19..68f90632245 100644
> --- a/environment.c
> +++ b/environment.c
> @@ -82,7 +82,6 @@ int protect_hfs = PROTECT_HFS_DEFAULT;
>  #define PROTECT_NTFS_DEFAULT 1
>  #endif
>  int protect_ntfs = PROTECT_NTFS_DEFAULT;
> -const char *core_fsmonitor;

So is this.

All nice.

> +static void lookup_fsmonitor_settings(struct repository *r)
> +{
> +	struct fsmonitor_settings *s;
> +
> +	CALLOC_ARRAY(s, 1);
> +
> +	r->settings.fsmonitor = s;
> +
> +	if (check_for_ipc(r))
> +		return;
> +
> +	if (check_for_hook(r))
> +		return;
> +
> +	fsm_settings__set_disabled(r);
> +}
> +
> +enum fsmonitor_mode fsm_settings__get_mode(struct repository *r)
> +{
> +	if (!r->settings.fsmonitor)
> +		lookup_fsmonitor_settings(r);

OK, and these "lookup" calls are what make this field "lazily
loaded".  A helper

static inline void lazily_load_fsmonitor_settings(struct repository *r)
{
	if (!r->settings.fsmonitor)
		lookup_fsmonitor_settings(r);
}

might be handy.  Also an assert to ensure nobody calls lookup() on a
repository that already has lazily loaded the settings would be
necessary.

	static void lookup_fsmonitor_settings(struct repository *r)
	{
		if (r->settings.fsmonitor)
			BUG("...");
		CALLOC_ARRAY(r->settings.fsmonitor, 1);

> +enum fsmonitor_mode {
> +	FSMONITOR_MODE_DISABLED = 0,
> +	FSMONITOR_MODE_HOOK = 1, /* core.fsmonitor */
> +	FSMONITOR_MODE_IPC = 2,  /* core.useBuiltinFSMonitor */
> +};

Please remind me why we need a new separate variable, instead of
turning the core.fsmonitor variable into an extended bool <false,
true, builtin>?  The compatibility issues during transition is the
same either way.  Old clients will ignore the request silently when
you set core.useBuiltinFSMonitor, or they will barf if you set
core.fsmonitor to 'builtin', so in a sense, extending the existing
variable may be a safer option.

> diff --git a/repository.h b/repository.h
> index a057653981c..89a1873ade7 100644
> --- a/repository.h
> +++ b/repository.h
> @@ -4,6 +4,7 @@
>  #include "path.h"
>  
>  struct config_set;
> +struct fsmonitor_settings;
>  struct git_hash_algo;
>  struct index_state;
>  struct lock_file;
> @@ -34,6 +35,8 @@ struct repo_settings {
>  	int command_requires_full_index;
>  	int sparse_index;
>  
> +	struct fsmonitor_settings *fsmonitor; /* lazy loaded */

"lazily" loaded, I think.

>  GIT_TEST_FSMONITOR=$PWD/t7519/fsmonitor-all exercises the fsmonitor
> -code path for utilizing a file system monitor to speed up detecting
> -new or changed files.
> +code path for utilizing a (hook based) file system monitor to speed up
> +detecting new or changed files.

Nice attention to the detail here.

Thanks.

Re: [PATCH v4 03/29] fsmonitor: config settings are repository-specific

[Junio C Hamano]

Junio C Hamano <gitster@pobox.com> writes:

>> +enum fsmonitor_mode {
>> +	FSMONITOR_MODE_DISABLED = 0,
>> +	FSMONITOR_MODE_HOOK = 1, /* core.fsmonitor */
>> +	FSMONITOR_MODE_IPC = 2,  /* core.useBuiltinFSMonitor */
>> +};
>
> Please remind me why we need a new separate variable, instead of
> turning the core.fsmonitor variable into an extended bool <false,
> true, builtin>?

Ah, I see.

The vocabulary of the value for the existing variable is between
"unset means disabled" and "the path-to-hook means enabled", so
unless we forbid a bareword path "builtin" (which I do not think is
such a bad idea, by the way), it becomes a bit fuzzy what a
non-empty token means.

In any case, the "set to path to enable, leave unset to leave
disabled" is a cumbersome to use and may want to be rethought.  It
is unclear how one would override a configured path-to-hook, for
example.

Considering that we need to reserve a special word, say, "disabled",
that has to be distinguishable from a normal "here is a path to the
hook script" ANYWAY, in order to allow such a "last one wins"
configuration override (or "git -c core.fsmonitor=disabled cmd"), it
starts to sound more and more reasonable to reserve yet another word
"builtin" as a special value of core.fsmonitor, without having to
introduce a new configuration variable, no?

Re: [PATCH v4 29/29] t7527: test status with untracked-cache and fsmonitor--daemon

[Eric Sunshine]

On Thu, Oct 21, 2021 at 10:26 AM Jeff Hostetler via GitGitGadget
<gitgitgadget@gmail.com> wrote:
> Create 2x2 test matrix with the untracked-cache and fsmonitor--daemon
> features and a series of edits and verify that status output is
> identical.
>
> Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
> ---
> diff --git a/t/t7527-builtin-fsmonitor.sh b/t/t7527-builtin-fsmonitor.sh
> @@ -508,4 +510,98 @@ test_expect_success 'cleanup worktrees' '
> +test_lazy_prereq UNTRACKED_CACHE '
> +       { git update-index --test-untracked-cache; ret=$?; } &&
> +       test $ret -ne 1
> +'

I may be missing something obvious, but can't this be expressed more simply as:

    test_lazy_prereq UNTRACKED_CACHE '
        git update-index --test-untracked-cache
        test $? -ne 1
    '

How significant is it to test specifically against 1? If not, then
even simpler would be:

    test_lazy_prereq UNTRACKED_CACHE '
        git update-index --test-untracked-cache
    '

which is also more robust since it won't be fooled by die() or crashes.

> +test_expect_success 'Matrix: setup for untracked-cache,fsmonitor matrix' '
> +       test_might_fail git config --unset core.useBuiltinFSMonitor &&

More idiomatic:

    test_unconfig core.useBuiltinFSMonitor &&

> +       git update-index --no-fsmonitor &&
> +       test_might_fail git fsmonitor--daemon stop
> +'
> +
> +matrix_clean_up_repo () {
> +       git reset --hard HEAD
> +       git clean -fd
> +}

Since calls to this function are part of the &&-chain in tests, it
probably would be a good idea to maintain the &&-chain within the body
of the function, as well.

> +matrix_try () {
> +       test_expect_success "Matrix[uc:$uc][fsm:$fsm] $fn" '
> +               matrix_clean_up_repo &&
> +               $fn &&
> +               if test $uc = false -a $fsm = false

We avoid -a and -o with `test` and instead chain them with &&:

    if test $uc = false && test $fsm = false

Documentation/CodingGuidelines mentions this. Also see [1] & [2].

[1]: https://lore.kernel.org/git/xmqqa6qkb5fi.fsf@gitster.g/
[2]: https://lore.kernel.org/git/CAPig+cQFFsLeE921WpzTxVnBMnNRiKs4N=hUQ2UQi1VznNEQwg@mail.gmail.com/

> +               then
> +                       git status --porcelain=v1 >.git/expect.$fn
> +               else
> +                       git status --porcelain=v1 >.git/actual.$fn
> +                       test_cmp .git/expect.$fn .git/actual.$fn
> +               fi
> +       '

Broken &&-chain in the `else` arm.

> +       return $?
> +}

No callers care about the return value of this function, so the
`return $?` can be dropped.

> +uc_values="false"
> +test_have_prereq UNTRACKED_CACHE && uc_values="false true"
> +for uc_val in $uc_values
> +do
> +       if test $uc_val = false
> +       then
> +               test_expect_success "Matrix[uc:$uc_val] disable untracked cache" '
> +                       git config core.untrackedcache false &&
> +                       git update-index --no-untracked-cache
> +               '
> +       else
> +               test_expect_success "Matrix[uc:$uc_val] enable untracked cache" '
> +                       git config core.untrackedcache true &&
> +                       git update-index --untracked-cache
> +               '
> +       fi
> +
> +       fsm_values="false true"
> +       for fsm_val in $fsm_values
> +       do
> +               if test $fsm_val = false
> +               then
> +                       test_expect_success "Matrix[uc:$uc_val][fsm:$fsm_val] disable fsmonitor" '
> +                               test_might_fail git config --unset core.useBuiltinFSMonitor &&

Ditto: test_unconfig()

> +                               git update-index --no-fsmonitor &&
> +                               test_might_fail git fsmonitor--daemon stop 2>/dev/null
> +                       '

stderr is redirected within tests anyhow, so we normally don't
suppress it manually like this (especially since it may come in handy
when debugging a failing test).

> +               else
> +                       test_expect_success "Matrix[uc:$uc_val][fsm:$fsm_val] enable fsmonitor" '
> +                               git config core.useBuiltinFSMonitor true &&
> +                               git fsmonitor--daemon start &&
> +                               git update-index --fsmonitor
> +                       '
> +               fi
> +
> +               matrix_try $uc_val $fsm_val edit_files
> +               matrix_try $uc_val $fsm_val delete_files
> +               matrix_try $uc_val $fsm_val create_files
> +               matrix_try $uc_val $fsm_val rename_files
> +               matrix_try $uc_val $fsm_val file_to_directory
> +               matrix_try $uc_val $fsm_val directory_to_file
> +
> +               if test $fsm_val = true
> +               then
> +                       test_expect_success "Matrix[uc:$uc_val][fsm:$fsm_val] disable fsmonitor at end" '
> +                               test_might_fail git config --unset core.useBuiltinFSMonitor &&

Ditto: test_unconfig()

> +                               git update-index --no-fsmonitor &&
> +                               test_might_fail git fsmonitor--daemon stop 2>/dev/null

Ditto: stderr

> +                       '
> +               fi
> +       done
> +done

Re: [PATCH v4 01/29] fsmonitor: enhance existing comments

[Jeff Hostetler]

On 10/21/21 4:40 PM, Junio C Hamano wrote:
> "Jeff Hostetler via GitGitGadget" <gitgitgadget@gmail.com> writes:
> 
>> From: Jeff Hostetler <jeffhost@microsoft.com>
>>
>> Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
>> ---
>>   fsmonitor.c | 37 ++++++++++++++++++++++++++++++-------
>>   1 file changed, 30 insertions(+), 7 deletions(-)
>>
>> diff --git a/fsmonitor.c b/fsmonitor.c
>> index ab9bfc60b34..ec4c46407c5 100644
>> --- a/fsmonitor.c
>> +++ b/fsmonitor.c
>> @@ -301,9 +301,25 @@ void refresh_fsmonitor(struct index_state *istate)
>>   			core_fsmonitor, query_success ? "success" : "failure");
>>   	}
>>   
>> -	/* a fsmonitor process can return '/' to indicate all entries are invalid */
>> +	/*
>> +	 * The response from FSMonitor (excluding the header token) is
>> +	 * either:
>> +	 *
>> +	 * [a] a (possibly empty) list of NUL delimited relative
>> +	 *     pathnames of changed paths.  This list can contain
>> +	 *     files and directories.  Directories have a trailing
>> +	 *     slash.
>> +	 *
>> +	 * [b] a single '/' to indicate the provider had no
>> +	 *     information and that we should consider everything
>> +	 *     invalid.  We call this a trivial response.
>> +	 */
>>   	if (query_success && query_result.buf[bol] != '/') {
>> -		/* Mark all entries returned by the monitor as dirty */
>> +		/*
>> +		 * Mark all pathnames returned by the monitor as dirty.
>> +		 *
>> +		 * This updates both the cache-entries and the untracked-cache.
>> +		 */
> 
> Not a problem this patch introduces, but we only checked that the
> query result begins with a slash, not "we did receive a trivial
> response", but the "else" clause of this statement pretends as if we
> did.
> 
> It is a shame that we do have fsmonitor_is_trivial_response()
> function defined, but its interface is not capable of helping us
> here.
> 
> Or is fsmonitor_is_trivial_response() already good to do this, and
> reliance of [bol] this code has is the source of confusion?  I
> notice that when we have last update token and makes a call to
> query_fsmonitor() with HOOK_INTERFACE_VERSION1, nobody updates bol
> (hence it stays 0), and with HOOK_INTERFACE_VERSION2, bol is at the
> NUL that terminates the initial string of the query result, after
> which presumably has either '/' NUL (trivial) or list of paths.
> 
> I am not sure about the VERSION1 case, but at least in the VERSION2
> case, making sure that the last three bytes are NUL slash NUL like
> fsmonitor_is_trivial_response() does and the half check the above is
> doing (i.e. the byte after the NUL is slash, without making sure
> about the length of the whole response or what follows the slash is
> NUL), seems "close enough" (meaning: the check in this code is a
> sloppy attempt to reinvent what _is_trivial_response() function
> already does).
> 
> So, would it make sense to rewrite the condition to
> 
> 	if (query_success &&
> 	    !fsmonitor_is_trivial_response(&query_result)) {
> 
> here?  Or perhaps
> 
> 	if (query_success &&
> 	    !(query_result.len == bol + 3 &&
> 	      query_result[bol] == '/' && !query_result[bol+1])) {
> 
> which would be open coding the _is_trivial_response() function.
> 

Yes, there is an opportunity here to clean this up.

The original code was a little sloppy -- just testing buf[bol] for
slash and assuming that that was sufficient.  Technically, no response
should begin with slash (since the set of reported modified files are
relative to the root directory), so my _is_trivial_ function could be
simplified a little.

I'll refactor this.

Thanks
Jeff

Re: [PATCH v4 03/29] fsmonitor: config settings are repository-specific

[Jeff Hostetler]

On 10/21/21 5:05 PM, Junio C Hamano wrote:
> "Jeff Hostetler via GitGitGadget" <gitgitgadget@gmail.com> writes:
> 
>>   	if (fsmonitor > 0) {
>> -		if (git_config_get_fsmonitor() == 0)
>> +		enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
>> +
>> +		if (fsm_mode == FSMONITOR_MODE_DISABLED) {
>> +			warning(_("core.useBuiltinFSMonitor is unset; "
>> +				"set it if you really want to enable the "
>> +				"builtin fsmonitor"));
>>   			warning(_("core.fsmonitor is unset; "
>> -				"set it if you really want to "
>> -				"enable fsmonitor"));
>> +				"set it if you really want to enable the "
>> +				"hook-based fsmonitor"));
>> +		}
>>   		add_fsmonitor(&the_index);
>>   		report(_("fsmonitor enabled"));
>>   	} else if (!fsmonitor) {
>> -		if (git_config_get_fsmonitor() == 1)
>> +		enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
>> +		if (fsm_mode == FSMONITOR_MODE_IPC)
>> +			warning(_("core.useBuiltinFSMonitor is set; "
>> +				"remove it if you really want to "
>> +				"disable fsmonitor"));
>> +		if (fsm_mode == FSMONITOR_MODE_HOOK)
>>   			warning(_("core.fsmonitor is set; "
>>   				"remove it if you really want to "
>>   				"disable fsmonitor"));
> 
> Hmph.  This does not change the behaviour per-se, but what are we
> trying to achieve with these warning messages?
> 
> The user uses --fsmonitor or --no-fsmonitor option from the command
> line, presumably as a one-shot "this time I'd operate the command
> differently from the configured way", so it seems unlikely that the
> user is doing so because "... really want to enable/disable".  The
> "report()" calls in these if/else cases seem sufficient reminder of
> what is going on---perhaps these warnings should be made silenceable
> by turning them into advice messages?
>

The original code had the basic warning for `core.fsmonitor`
which becomes ambiguous/confusing when we have two different
config values.

I was really wanting to just get rid of the warnings.  They
only appear if the users passes a `--[no-]fsmonitor` on the
command line to temporarily override the config setting.

I'll see about making them advice messages.


>> -int git_config_get_fsmonitor(void)
>> -{
>> -	if (git_config_get_pathname("core.fsmonitor", &core_fsmonitor))
>> -		core_fsmonitor = getenv("GIT_TEST_FSMONITOR");
>> -
>> -	if (core_fsmonitor && !*core_fsmonitor)
>> -		core_fsmonitor = NULL;
>> -
>> -	if (core_fsmonitor)
>> -		return 1;
>> -
>> -	return 0;
>> -}
> 
> This used to be how we got the configuration.
> 
>> --- a/config.h
>> +++ b/config.h
>> @@ -610,7 +610,6 @@ int git_config_get_pathname(const char *key, const char **dest);
>>   int git_config_get_index_threads(int *dest);
>>   int git_config_get_split_index(void);
>>   int git_config_get_max_percent_split_change(void);
>> -int git_config_get_fsmonitor(void);
> 
> And that is removed so any in-flight topic that adds new caller will
> be caught by the compiler.  OK.
> 
>> diff --git a/environment.c b/environment.c
>> index 9da7f3c1a19..68f90632245 100644
>> --- a/environment.c
>> +++ b/environment.c
>> @@ -82,7 +82,6 @@ int protect_hfs = PROTECT_HFS_DEFAULT;
>>   #define PROTECT_NTFS_DEFAULT 1
>>   #endif
>>   int protect_ntfs = PROTECT_NTFS_DEFAULT;
>> -const char *core_fsmonitor;
> 
> So is this.
> 
> All nice.
> 
>> +static void lookup_fsmonitor_settings(struct repository *r)
>> +{
>> +	struct fsmonitor_settings *s;
>> +
>> +	CALLOC_ARRAY(s, 1);
>> +
>> +	r->settings.fsmonitor = s;
>> +
>> +	if (check_for_ipc(r))
>> +		return;
>> +
>> +	if (check_for_hook(r))
>> +		return;
>> +
>> +	fsm_settings__set_disabled(r);
>> +}
>> +
>> +enum fsmonitor_mode fsm_settings__get_mode(struct repository *r)
>> +{
>> +	if (!r->settings.fsmonitor)
>> +		lookup_fsmonitor_settings(r);
> 
> OK, and these "lookup" calls are what make this field "lazily
> loaded".  A helper
> 
> static inline void lazily_load_fsmonitor_settings(struct repository *r)
> {
> 	if (!r->settings.fsmonitor)
> 		lookup_fsmonitor_settings(r);
> }
> 
> might be handy.  Also an assert to ensure nobody calls lookup() on a
> repository that already has lazily loaded the settings would be
> necessary.
> 
> 	static void lookup_fsmonitor_settings(struct repository *r)
> 	{
> 		if (r->settings.fsmonitor)
> 			BUG("...");
> 		CALLOC_ARRAY(r->settings.fsmonitor, 1);

good point.


> 
>> +enum fsmonitor_mode {
>> +	FSMONITOR_MODE_DISABLED = 0,
>> +	FSMONITOR_MODE_HOOK = 1, /* core.fsmonitor */
>> +	FSMONITOR_MODE_IPC = 2,  /* core.useBuiltinFSMonitor */
>> +};
> 
> Please remind me why we need a new separate variable, instead of
> turning the core.fsmonitor variable into an extended bool <false,
> true, builtin>?  The compatibility issues during transition is the
> same either way.  Old clients will ignore the request silently when
> you set core.useBuiltinFSMonitor, or they will barf if you set
> core.fsmonitor to 'builtin', so in a sense, extending the existing
> variable may be a safer option.
> 
>> diff --git a/repository.h b/repository.h
>> index a057653981c..89a1873ade7 100644
>> --- a/repository.h
>> +++ b/repository.h
>> @@ -4,6 +4,7 @@
>>   #include "path.h"
>>   
>>   struct config_set;
>> +struct fsmonitor_settings;
>>   struct git_hash_algo;
>>   struct index_state;
>>   struct lock_file;
>> @@ -34,6 +35,8 @@ struct repo_settings {
>>   	int command_requires_full_index;
>>   	int sparse_index;
>>   
>> +	struct fsmonitor_settings *fsmonitor; /* lazy loaded */
> 
> "lazily" loaded, I think.
> 
>>   GIT_TEST_FSMONITOR=$PWD/t7519/fsmonitor-all exercises the fsmonitor
>> -code path for utilizing a file system monitor to speed up detecting
>> -new or changed files.
>> +code path for utilizing a (hook based) file system monitor to speed up
>> +detecting new or changed files.
> 
> Nice attention to the detail here.
> 
> Thanks.
> 

Re: [PATCH v4 03/29] fsmonitor: config settings are repository-specific

[Jeff Hostetler]

On 10/21/21 5:16 PM, Junio C Hamano wrote:
> Junio C Hamano <gitster@pobox.com> writes:
> 
>>> +enum fsmonitor_mode {
>>> +	FSMONITOR_MODE_DISABLED = 0,
>>> +	FSMONITOR_MODE_HOOK = 1, /* core.fsmonitor */
>>> +	FSMONITOR_MODE_IPC = 2,  /* core.useBuiltinFSMonitor */
>>> +};
>>
>> Please remind me why we need a new separate variable, instead of
>> turning the core.fsmonitor variable into an extended bool <false,
>> true, builtin>?
> 
> Ah, I see.
> 
> The vocabulary of the value for the existing variable is between
> "unset means disabled" and "the path-to-hook means enabled", so
> unless we forbid a bareword path "builtin" (which I do not think is
> such a bad idea, by the way), it becomes a bit fuzzy what a
> non-empty token means.
> 
> In any case, the "set to path to enable, leave unset to leave
> disabled" is a cumbersome to use and may want to be rethought.  It
> is unclear how one would override a configured path-to-hook, for
> example.
> 
> Considering that we need to reserve a special word, say, "disabled",
> that has to be distinguishable from a normal "here is a path to the
> hook script" ANYWAY, in order to allow such a "last one wins"
> configuration override (or "git -c core.fsmonitor=disabled cmd"), it
> starts to sound more and more reasonable to reserve yet another word
> "builtin" as a special value of core.fsmonitor, without having to
> introduce a new configuration variable, no?
> 

For a while we were using a ":builtin:" reserved value (which isn't
a valid pathname on Windows) but thought it better to split it into
two different config values to avoid the confusion (since it is a
valid path on Mac/Linux).  But having 2 config vars is also confusing.

And yes, I'm not sure there is a way for a local fsmonitor hook
config to override a global hook value unless we add a "disabled"
or "off" value.

Let me revisit this.  We could have:

     [] unset, <any of the standard boolean false values>, "disabled"
     [] <hook-path>
     [] "builtin", "ipc"

and just make the literals special reserved values.


I've not kept up on the configurable hooks series, but I have to
wonder if this usage (pathname or reserved word) will cause problems
with the changes being planned for hooks.  The existing fsmonitor
usage doesn't use find_hook() nor run_hook*(), so I don't expect
an immediate conflict.  But again, I haven't kept up with that
effort.

Jeff

Re: [PATCH v4 29/29] t7527: test status with untracked-cache and fsmonitor--daemon

[Jeff Hostetler]

On 10/22/21 1:23 AM, Eric Sunshine wrote:
> On Thu, Oct 21, 2021 at 10:26 AM Jeff Hostetler via GitGitGadget
> <gitgitgadget@gmail.com> wrote:
>> Create 2x2 test matrix with the untracked-cache and fsmonitor--daemon
>> features and a series of edits and verify that status output is
>> identical.
>>
>> Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
>> ---
>> diff --git a/t/t7527-builtin-fsmonitor.sh b/t/t7527-builtin-fsmonitor.sh
>> @@ -508,4 +510,98 @@ test_expect_success 'cleanup worktrees' '
>> +test_lazy_prereq UNTRACKED_CACHE '
>> +       { git update-index --test-untracked-cache; ret=$?; } &&
>> +       test $ret -ne 1
>> +'
> 
> I may be missing something obvious, but can't this be expressed more simply as:
> 
>      test_lazy_prereq UNTRACKED_CACHE '
>          git update-index --test-untracked-cache
>          test $? -ne 1
>      '
> 
> How significant is it to test specifically against 1? If not, then
> even simpler would be:
> 
>      test_lazy_prereq UNTRACKED_CACHE '
>          git update-index --test-untracked-cache
>      '
> 
> which is also more robust since it won't be fooled by die() or crashes.
> 

IIRC I stole that from t7063.  It didn't occur to me to try to
simplify it.


>> +test_expect_success 'Matrix: setup for untracked-cache,fsmonitor matrix' '
>> +       test_might_fail git config --unset core.useBuiltinFSMonitor &&
> 
> More idiomatic:
> 
>      test_unconfig core.useBuiltinFSMonitor &&
> 

Good to know, thanks!
I'll take a look at this and the rest of your comments below.

thanks
Jeff


>> +       git update-index --no-fsmonitor &&
>> +       test_might_fail git fsmonitor--daemon stop
>> +'
>> +
>> +matrix_clean_up_repo () {
>> +       git reset --hard HEAD
>> +       git clean -fd
>> +}
> 
> Since calls to this function are part of the &&-chain in tests, it
> probably would be a good idea to maintain the &&-chain within the body
> of the function, as well.
> 
>> +matrix_try () {
>> +       test_expect_success "Matrix[uc:$uc][fsm:$fsm] $fn" '
>> +               matrix_clean_up_repo &&
>> +               $fn &&
>> +               if test $uc = false -a $fsm = false
> 
> We avoid -a and -o with `test` and instead chain them with &&:
> 
>      if test $uc = false && test $fsm = false
> 
> Documentation/CodingGuidelines mentions this. Also see [1] & [2].
> 
> [1]: https://lore.kernel.org/git/xmqqa6qkb5fi.fsf@gitster.g/
> [2]: https://lore.kernel.org/git/CAPig+cQFFsLeE921WpzTxVnBMnNRiKs4N=hUQ2UQi1VznNEQwg@mail.gmail.com/
> 
>> +               then
>> +                       git status --porcelain=v1 >.git/expect.$fn
>> +               else
>> +                       git status --porcelain=v1 >.git/actual.$fn
>> +                       test_cmp .git/expect.$fn .git/actual.$fn
>> +               fi
>> +       '
> 
> Broken &&-chain in the `else` arm.
> 
>> +       return $?
>> +}
> 
> No callers care about the return value of this function, so the
> `return $?` can be dropped.
> 
>> +uc_values="false"
>> +test_have_prereq UNTRACKED_CACHE && uc_values="false true"
>> +for uc_val in $uc_values
>> +do
>> +       if test $uc_val = false
>> +       then
>> +               test_expect_success "Matrix[uc:$uc_val] disable untracked cache" '
>> +                       git config core.untrackedcache false &&
>> +                       git update-index --no-untracked-cache
>> +               '
>> +       else
>> +               test_expect_success "Matrix[uc:$uc_val] enable untracked cache" '
>> +                       git config core.untrackedcache true &&
>> +                       git update-index --untracked-cache
>> +               '
>> +       fi
>> +
>> +       fsm_values="false true"
>> +       for fsm_val in $fsm_values
>> +       do
>> +               if test $fsm_val = false
>> +               then
>> +                       test_expect_success "Matrix[uc:$uc_val][fsm:$fsm_val] disable fsmonitor" '
>> +                               test_might_fail git config --unset core.useBuiltinFSMonitor &&
> 
> Ditto: test_unconfig()
> 
>> +                               git update-index --no-fsmonitor &&
>> +                               test_might_fail git fsmonitor--daemon stop 2>/dev/null
>> +                       '
> 
> stderr is redirected within tests anyhow, so we normally don't
> suppress it manually like this (especially since it may come in handy
> when debugging a failing test).
> 
>> +               else
>> +                       test_expect_success "Matrix[uc:$uc_val][fsm:$fsm_val] enable fsmonitor" '
>> +                               git config core.useBuiltinFSMonitor true &&
>> +                               git fsmonitor--daemon start &&
>> +                               git update-index --fsmonitor
>> +                       '
>> +               fi
>> +
>> +               matrix_try $uc_val $fsm_val edit_files
>> +               matrix_try $uc_val $fsm_val delete_files
>> +               matrix_try $uc_val $fsm_val create_files
>> +               matrix_try $uc_val $fsm_val rename_files
>> +               matrix_try $uc_val $fsm_val file_to_directory
>> +               matrix_try $uc_val $fsm_val directory_to_file
>> +
>> +               if test $fsm_val = true
>> +               then
>> +                       test_expect_success "Matrix[uc:$uc_val][fsm:$fsm_val] disable fsmonitor at end" '
>> +                               test_might_fail git config --unset core.useBuiltinFSMonitor &&
> 
> Ditto: test_unconfig()
> 
>> +                               git update-index --no-fsmonitor &&
>> +                               test_might_fail git fsmonitor--daemon stop 2>/dev/null
> 
> Ditto: stderr
> 
>> +                       '
>> +               fi
>> +       done
>> +done

Re: [PATCH v4 21/29] t7527: create test for fsmonitor--daemon

[Junio C Hamano]

While I was looking at the differences between next..seen to catch
new topics that try to sneak style violations into the codebase, I
found this one (there was another in the same series).

----- >8 --------- >8 --------- >8 --------- >8 --------- >8 -----
Subject: [PATCH] fixup! t7527: create test for fsmonitor--daemon

Fix style violation of a shell-function definition

Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 t/t7527-builtin-fsmonitor.sh | 14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)

diff --git a/t/t7527-builtin-fsmonitor.sh b/t/t7527-builtin-fsmonitor.sh
index e62ec9aa3c..313c4fad34 100755
--- a/t/t7527-builtin-fsmonitor.sh
+++ b/t/t7527-builtin-fsmonitor.sh
@@ -213,43 +213,43 @@ test_expect_success 'status implicitly starts daemon' '
 	test_subcommand git fsmonitor--daemon start <.git/trace_implicit_2
 '
 
-edit_files() {
+edit_files () {
 	echo 1 >modified
 	echo 2 >dir1/modified
 	echo 3 >dir2/modified
 	>dir1/untracked
 }
 
-delete_files() {
+delete_files () {
 	rm -f delete
 	rm -f dir1/delete
 	rm -f dir2/delete
 }
 
-create_files() {
+create_files () {
 	echo 1 >new
 	echo 2 >dir1/new
 	echo 3 >dir2/new
 }
 
-rename_files() {
+rename_files () {
 	mv rename renamed
 	mv dir1/rename dir1/renamed
 	mv dir2/rename dir2/renamed
 }
 
-file_to_directory() {
+file_to_directory () {
 	rm -f delete
 	mkdir delete
 	echo 1 >delete/new
 }
 
-directory_to_file() {
+directory_to_file () {
 	rm -rf dir1
 	echo 1 >dir1
 }
 
-verify_status() {
+verify_status () {
 	git status >actual &&
 	GIT_INDEX_FILE=.git/fresh-index git read-tree master &&
 	GIT_INDEX_FILE=.git/fresh-index git -c core.useBuiltinFSMonitor= status >expect &&
-- 
2.34.1-568-g69e9fd72b5

Re: [PATCH v4 24/29] t/perf/p7519: speed up test on Windows

[Junio C Hamano]

Ditto.

----- >8 --------- >8 --------- >8 --------- >8 --------- >8 -----
Subject: [PATCH] fixup! t/perf/p7519: speed up test on Windows

Fix style violation of a shell-function definition introduced by
this step, and correct existing ones while at it.

Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 t/perf/p7519-fsmonitor.sh | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/t/perf/p7519-fsmonitor.sh b/t/perf/p7519-fsmonitor.sh
index e70252ed65..9e08fc169a 100755
--- a/t/perf/p7519-fsmonitor.sh
+++ b/t/perf/p7519-fsmonitor.sh
@@ -73,7 +73,7 @@ then
 	fi
 fi
 
-trace_start() {
+trace_start () {
 	if test -n "$GIT_PERF_7519_TRACE"
 	then
 		name="$1"
@@ -92,14 +92,14 @@ trace_start() {
 	fi
 }
 
-trace_stop() {
+trace_stop () {
 	if test -n "$GIT_PERF_7519_TRACE"
 	then
 		unset GIT_TRACE2_PERF
 	fi
 }
 
-touch_files() {
+touch_files () {
 	n=$1
 	d="$n"_files
 
@@ -142,7 +142,7 @@ test_expect_success "one time repo setup" '
 	fi
 '
 
-setup_for_fsmonitor() {
+setup_for_fsmonitor () {
 	# set INTEGRATION_SCRIPT depending on the environment
 	if test -n "$USE_FSMONITOR_DAEMON"
 	then
@@ -188,7 +188,7 @@ test_perf_w_drop_caches () {
 	test_perf "$@"
 }
 
-test_fsmonitor_suite() {
+test_fsmonitor_suite () {
 	if test -n "$USE_FSMONITOR_DAEMON"
 	then
 		DESC="builtin fsmonitor--daemon"
-- 
2.34.1-568-g69e9fd72b5

[PATCH v5 00/30] Builtin FSMonitor Part 2

[Jeff Hostetler via GitGitGadget]

Here is V5 of Part 2 of my Builtin FSMonitor series. I apologize for the
delay since V4 that I submitted back in October. (Insert the usual $DayJob
excuse...)

I have rebased this branch onto the current "master" branch.

In this version I removed the core.useBuiltinFSMonitor config setting and
instead extended the existing core.fsmonitor. Previously it was defined as
containing the pathname to the FSMonitor hook script/application. Now it can
also have a boolean value to select the builtin FSMonitor daemon. This does
simplify a few things in the documentation and in the code.

There were other changes, but they are mostly just general cleanup and
coding style items. I also squashed in Junio's Dec 25 commits related to
coding style in the test scripts.

This version contains the client code and a MVP version of the daemon. It
can be be tested using t/t7527 and t/perf/p7519.

A followup Part 3 will contain additional refinements to the daemon and
additional tests. I drew the line here between Part 2 and 3 to make it
easier to review.

Jeff Hostetler (30):
  fsmonitor: enhance existing comments, clarify trivial response
    handling
  fsmonitor-ipc: create client routines for git-fsmonitor--daemon
  fsmonitor: config settings are repository-specific
  fsmonitor: use IPC to query the builtin FSMonitor daemon
  fsmonitor: document builtin fsmonitor
  fsmonitor--daemon: add a built-in fsmonitor daemon
  fsmonitor--daemon: implement 'stop' and 'status' commands
  compat/fsmonitor/fsm-listen-win32: stub in backend for Windows
  compat/fsmonitor/fsm-listen-darwin: stub in backend for Darwin
  fsmonitor--daemon: implement 'run' command
  fsmonitor--daemon: implement 'start' command
  fsmonitor--daemon: add pathname classification
  fsmonitor--daemon: define token-ids
  fsmonitor--daemon: create token-based changed path cache
  compat/fsmonitor/fsm-listen-win32: implement FSMonitor backend on
    Windows
  compat/fsmonitor/fsm-listen-darwin: add macos header files for FSEvent
  compat/fsmonitor/fsm-listen-darwin: implement FSEvent listener on
    MacOS
  fsmonitor--daemon: implement handle_client callback
  help: include fsmonitor--daemon feature flag in version info
  t/helper/fsmonitor-client: create IPC client to talk to FSMonitor
    Daemon
  t7527: create test for fsmonitor--daemon
  t/perf: avoid copying builtin fsmonitor files into test repo
  t/helper/test-chmtime: skip directories on Windows
  t/perf/p7519: speed up test on Windows
  t/perf/p7519: add fsmonitor--daemon test cases
  fsmonitor--daemon: periodically truncate list of modified files
  fsmonitor--daemon: use a cookie file to sync with file system
  fsmonitor: force update index after large responses
  t7527: test status with untracked-cache and fsmonitor--daemon
  update-index: convert fsmonitor warnings to advise

 .gitignore                              |    1 +
 Documentation/config/core.txt           |   46 +-
 Documentation/git-fsmonitor--daemon.txt |   75 ++
 Documentation/git-update-index.txt      |    8 +-
 Makefile                                |   17 +
 builtin.h                               |    1 +
 builtin/fsmonitor--daemon.c             | 1454 +++++++++++++++++++++++
 builtin/update-index.c                  |   19 +-
 cache.h                                 |    1 -
 compat/fsmonitor/fsm-listen-darwin.c    |  496 ++++++++
 compat/fsmonitor/fsm-listen-win32.c     |  586 +++++++++
 compat/fsmonitor/fsm-listen.h           |   49 +
 config.c                                |   14 -
 config.h                                |    1 -
 config.mak.uname                        |   20 +
 contrib/buildsystems/CMakeLists.txt     |   10 +
 environment.c                           |    1 -
 fsmonitor--daemon.h                     |  140 +++
 fsmonitor-ipc.c                         |  171 +++
 fsmonitor-ipc.h                         |   48 +
 fsmonitor-settings.c                    |  100 ++
 fsmonitor-settings.h                    |   21 +
 fsmonitor.c                             |  218 +++-
 fsmonitor.h                             |   18 +-
 git.c                                   |    1 +
 help.c                                  |    4 +
 repo-settings.c                         |    1 +
 repository.h                            |    3 +
 t/README                                |    4 +-
 t/helper/test-chmtime.c                 |   15 +
 t/helper/test-fsmonitor-client.c        |  121 ++
 t/helper/test-tool.c                    |    1 +
 t/helper/test-tool.h                    |    1 +
 t/perf/p7519-fsmonitor.sh               |   61 +-
 t/perf/perf-lib.sh                      |    2 +-
 t/t7527-builtin-fsmonitor.sh            |  604 ++++++++++
 t/test-lib.sh                           |    6 +
 37 files changed, 4229 insertions(+), 110 deletions(-)
 create mode 100644 Documentation/git-fsmonitor--daemon.txt
 create mode 100644 builtin/fsmonitor--daemon.c
 create mode 100644 compat/fsmonitor/fsm-listen-darwin.c
 create mode 100644 compat/fsmonitor/fsm-listen-win32.c
 create mode 100644 compat/fsmonitor/fsm-listen.h
 create mode 100644 fsmonitor--daemon.h
 create mode 100644 fsmonitor-ipc.c
 create mode 100644 fsmonitor-ipc.h
 create mode 100644 fsmonitor-settings.c
 create mode 100644 fsmonitor-settings.h
 create mode 100644 t/helper/test-fsmonitor-client.c
 create mode 100755 t/t7527-builtin-fsmonitor.sh


base-commit: 2b9c1209706bc2ef0ab09fb0bdc7d405e225ce8b
Published-As: https://github.com/gitgitgadget/git/releases/tag/pr-1041%2Fjeffhostetler%2Fbuiltin-fsmonitor-part2-v5
Fetch-It-Via: git fetch https://github.com/gitgitgadget/git pr-1041/jeffhostetler/builtin-fsmonitor-part2-v5
Pull-Request: https://github.com/gitgitgadget/git/pull/1041

Range-diff vs v4:

  1:  ecc40795fa2 !  1:  a5ecabb4d02 fsmonitor: enhance existing comments
     @@ Metadata
      Author: Jeff Hostetler <jeffhost@microsoft.com>
      
       ## Commit message ##
     -    fsmonitor: enhance existing comments
     +    fsmonitor: enhance existing comments, clarify trivial response handling
      
          Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
      
       ## fsmonitor.c ##
     +@@ fsmonitor.c: static int query_fsmonitor(int version, const char *last_update, struct strbuf *
     + 
     + 	if (result)
     + 		trace2_data_intmax("fsm_hook", NULL, "query/failed", result);
     +-	else {
     ++	else
     + 		trace2_data_intmax("fsm_hook", NULL, "query/response-length",
     + 				   query_result->len);
     + 
     +-		if (fsmonitor_is_trivial_response(query_result))
     +-			trace2_data_intmax("fsm_hook", NULL,
     +-					   "query/trivial-response", 1);
     +-	}
     +-
     + 	trace2_region_leave("fsm_hook", "query", NULL);
     + 
     + 	return result;
     + }
     + 
     +-int fsmonitor_is_trivial_response(const struct strbuf *query_result)
     +-{
     +-	static char trivial_response[3] = { '\0', '/', '\0' };
     +-
     +-	return query_result->len >= 3 &&
     +-		!memcmp(trivial_response,
     +-			&query_result->buf[query_result->len - 3], 3);
     +-}
     +-
     + static void fsmonitor_refresh_callback(struct index_state *istate, char *name)
     + {
     + 	int i, len = strlen(name);
      @@ fsmonitor.c: void refresh_fsmonitor(struct index_state *istate)
     + 	struct strbuf last_update_token = STRBUF_INIT;
     + 	char *buf;
     + 	unsigned int i;
     ++	int is_trivial = 0;
     + 
     + 	if (!core_fsmonitor || istate->fsmonitor_has_run_once)
     + 		return;
     +@@ fsmonitor.c: void refresh_fsmonitor(struct index_state *istate)
     + 					query_success = 0;
     + 				} else {
     + 					bol = last_update_token.len + 1;
     ++					is_trivial = query_result.buf[bol] == '/';
     + 				}
     + 			} else if (hook_version < 0) {
     + 				hook_version = HOOK_INTERFACE_VERSION1;
     +@@ fsmonitor.c: void refresh_fsmonitor(struct index_state *istate)
     + 		if (hook_version == HOOK_INTERFACE_VERSION1) {
     + 			query_success = !query_fsmonitor(HOOK_INTERFACE_VERSION1,
     + 				istate->fsmonitor_last_update, &query_result);
     ++			if (query_success)
     ++				is_trivial = query_result.buf[0] == '/';
     + 		}
     + 
     ++		if (is_trivial)
     ++			trace2_data_intmax("fsm_hook", NULL,
     ++					   "query/trivial-response", 1);
     ++
     + 		trace_performance_since(last_update, "fsmonitor process '%s'", core_fsmonitor);
     + 		trace_printf_key(&trace_fsmonitor, "fsmonitor process '%s' returned %s",
       			core_fsmonitor, query_success ? "success" : "failure");
       	}
       
      -	/* a fsmonitor process can return '/' to indicate all entries are invalid */
     +-	if (query_success && query_result.buf[bol] != '/') {
     +-		/* Mark all entries returned by the monitor as dirty */
      +	/*
      +	 * The response from FSMonitor (excluding the header token) is
      +	 * either:
     @@ fsmonitor.c: void refresh_fsmonitor(struct index_state *istate)
      +	 *     information and that we should consider everything
      +	 *     invalid.  We call this a trivial response.
      +	 */
     - 	if (query_success && query_result.buf[bol] != '/') {
     --		/* Mark all entries returned by the monitor as dirty */
     ++	if (query_success && !is_trivial) {
      +		/*
      +		 * Mark all pathnames returned by the monitor as dirty.
      +		 *
     @@ fsmonitor.c: void refresh_fsmonitor(struct index_state *istate)
      -		/* We only want to run the post index changed hook if we've actually changed entries, so keep track
      -		 * if we actually changed entries or not */
      +		/*
     -+		 * We received a trivial response, so invalidate everything.
     ++		 * We failed to get a response or received a trivial response,
     ++		 * so invalidate everything.
      +		 *
      +		 * We only want to run the post index changed hook if
      +		 * we've actually changed entries, so keep track if we
  2:  82f17692128 !  2:  365964b7664 fsmonitor-ipc: create client routines for git-fsmonitor--daemon
     @@ fsmonitor-ipc.c (new)
      +
      +		trace2_data_intmax("fsm_client", NULL,
      +				   "query/response-length", answer->len);
     -+
     -+		if (fsmonitor_is_trivial_response(answer))
     -+			trace2_data_intmax("fsm_client", NULL,
     -+					   "query/trivial-response", 1);
     -+
      +		goto done;
      +
      +	case IPC_STATE__NOT_LISTENING:
  3:  882789b4dfe !  3:  384516ce1a1 fsmonitor: config settings are repository-specific
     @@ Commit message
          Create `fsm_settings__get_*()` getters to lazily look up fsmonitor-
          related config settings.
      
     -    Add support for the new `core.useBuiltinFSMonitor` config setting.
     -
          Get rid of the `core_fsmonitor` global variable.  Move the code to
          lookup the existing `core.fsmonitor` config value into the fsmonitor
          settings.
     @@ Commit message
          Create a hook pathname variable in `struct fsmonitor-settings` and
          only set it when in hook mode.
      
     +    Extend the definition of `core.fsmonitor` to be either a boolean
     +    or a hook pathname.  When true, the builtin FSMonitor is used.
     +    When false or unset, no FSMonitor (neither builtin nor hook) is
     +    used.
     +
          The existing `core_fsmonitor` global variable was used to store the
          pathname to the fsmonitor hook *and* it was used as a boolean to see
          if fsmonitor was enabled.  This dual usage and global visibility leads
     @@ builtin/update-index.c: int cmd_update_index(int argc, const char **argv, const
       	if (fsmonitor > 0) {
      -		if (git_config_get_fsmonitor() == 0)
      +		enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
     -+
      +		if (fsm_mode == FSMONITOR_MODE_DISABLED) {
     -+			warning(_("core.useBuiltinFSMonitor is unset; "
     -+				"set it if you really want to enable the "
     -+				"builtin fsmonitor"));
       			warning(_("core.fsmonitor is unset; "
     --				"set it if you really want to "
     --				"enable fsmonitor"));
     -+				"set it if you really want to enable the "
     -+				"hook-based fsmonitor"));
     + 				"set it if you really want to "
     + 				"enable fsmonitor"));
      +		}
       		add_fsmonitor(&the_index);
       		report(_("fsmonitor enabled"));
       	} else if (!fsmonitor) {
      -		if (git_config_get_fsmonitor() == 1)
      +		enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
     -+		if (fsm_mode == FSMONITOR_MODE_IPC)
     -+			warning(_("core.useBuiltinFSMonitor is set; "
     -+				"remove it if you really want to "
     -+				"disable fsmonitor"));
     -+		if (fsm_mode == FSMONITOR_MODE_HOOK)
     ++		if (fsm_mode > FSMONITOR_MODE_DISABLED)
       			warning(_("core.fsmonitor is set; "
       				"remove it if you really want to "
       				"disable fsmonitor"));
     @@ fsmonitor-settings.c (new)
      +	char *hook_path;
      +};
      +
     -+void fsm_settings__set_ipc(struct repository *r)
     ++static void lookup_fsmonitor_settings(struct repository *r)
      +{
     -+	struct fsmonitor_settings *s = r->settings.fsmonitor;
     ++	struct fsmonitor_settings *s;
     ++	const char *const_str;
     ++	int bool_value;
      +
     -+	s->mode = FSMONITOR_MODE_IPC;
     -+}
     ++	if (r->settings.fsmonitor)
     ++		return;
      +
     -+void fsm_settings__set_hook(struct repository *r, const char *path)
     -+{
     -+	struct fsmonitor_settings *s = r->settings.fsmonitor;
     ++	CALLOC_ARRAY(s, 1);
      +
     -+	s->mode = FSMONITOR_MODE_HOOK;
     -+	s->hook_path = strdup(path);
     -+}
     ++	r->settings.fsmonitor = s;
      +
     -+void fsm_settings__set_disabled(struct repository *r)
     -+{
     -+	struct fsmonitor_settings *s = r->settings.fsmonitor;
     ++	fsm_settings__set_disabled(r);
      +
     -+	s->mode = FSMONITOR_MODE_DISABLED;
     -+	FREE_AND_NULL(s->hook_path);
     -+}
     ++	/*
     ++	 * Overload the existing "core.fsmonitor" config setting (which
     ++	 * has historically been either unset or a hook pathname) to
     ++	 * now allow a boolean value to enable the builtin FSMonitor
     ++	 * or to turn everything off.  (This does imply that you can't
     ++	 * use a hook script named "true" or "false", but that's OK.)
     ++	 */
     ++	switch (repo_config_get_maybe_bool(r, "core.fsmonitor", &bool_value)) {
     ++
     ++	case 0: /* config value was set to <bool> */
     ++		if (bool_value)
     ++			fsm_settings__set_ipc(r);
     ++		return;
      +
     -+static int check_for_ipc(struct repository *r)
     -+{
     -+	int value;
     ++	case 1: /* config value was unset */
     ++		const_str = getenv("GIT_TEST_FSMONITOR");
     ++		break;
      +
     -+	if (!repo_config_get_bool(r, "core.usebuiltinfsmonitor", &value) &&
     -+	    value) {
     -+		fsm_settings__set_ipc(r);
     -+		return 1;
     ++	case -1: /* config value set to an arbitrary string */
     ++		if (repo_config_get_pathname(r, "core.fsmonitor", &const_str))
     ++			return; /* should not happen */
     ++		break;
     ++
     ++	default: /* should not happen */
     ++		return;
      +	}
      +
     -+	return 0;
     ++	if (!const_str || !*const_str)
     ++		return;
     ++
     ++	fsm_settings__set_hook(r, const_str);
      +}
      +
     -+static int check_for_hook(struct repository *r)
     ++enum fsmonitor_mode fsm_settings__get_mode(struct repository *r)
      +{
     -+	const char *const_str;
     ++	lookup_fsmonitor_settings(r);
      +
     -+	if (repo_config_get_pathname(r, "core.fsmonitor", &const_str))
     -+		const_str = getenv("GIT_TEST_FSMONITOR");
     -+
     -+	if (const_str && *const_str) {
     -+		fsm_settings__set_hook(r, const_str);
     -+		return 1;
     -+	}
     -+
     -+	return 0;
     ++	return r->settings.fsmonitor->mode;
      +}
      +
     -+static void lookup_fsmonitor_settings(struct repository *r)
     ++const char *fsm_settings__get_hook_path(struct repository *r)
      +{
     -+	struct fsmonitor_settings *s;
     -+
     -+	CALLOC_ARRAY(s, 1);
     ++	lookup_fsmonitor_settings(r);
      +
     -+	r->settings.fsmonitor = s;
     -+
     -+	if (check_for_ipc(r))
     -+		return;
     ++	return r->settings.fsmonitor->hook_path;
     ++}
      +
     -+	if (check_for_hook(r))
     -+		return;
     ++void fsm_settings__set_ipc(struct repository *r)
     ++{
     ++	lookup_fsmonitor_settings(r);
      +
     -+	fsm_settings__set_disabled(r);
     ++	r->settings.fsmonitor->mode = FSMONITOR_MODE_IPC;
     ++	FREE_AND_NULL(r->settings.fsmonitor->hook_path);
      +}
      +
     -+enum fsmonitor_mode fsm_settings__get_mode(struct repository *r)
     ++void fsm_settings__set_hook(struct repository *r, const char *path)
      +{
     -+	if (!r->settings.fsmonitor)
     -+		lookup_fsmonitor_settings(r);
     ++	lookup_fsmonitor_settings(r);
      +
     -+	return r->settings.fsmonitor->mode;
     ++	r->settings.fsmonitor->mode = FSMONITOR_MODE_HOOK;
     ++	FREE_AND_NULL(r->settings.fsmonitor->hook_path);
     ++	r->settings.fsmonitor->hook_path = strdup(path);
      +}
      +
     -+const char *fsm_settings__get_hook_path(struct repository *r)
     ++void fsm_settings__set_disabled(struct repository *r)
      +{
     -+	if (!r->settings.fsmonitor)
     -+		lookup_fsmonitor_settings(r);
     ++	lookup_fsmonitor_settings(r);
      +
     -+	return r->settings.fsmonitor->hook_path;
     ++	r->settings.fsmonitor->mode = FSMONITOR_MODE_DISABLED;
     ++	FREE_AND_NULL(r->settings.fsmonitor->hook_path);
      +}
      
       ## fsmonitor-settings.h (new) ##
     @@ fsmonitor-settings.h (new)
      +
      +enum fsmonitor_mode {
      +	FSMONITOR_MODE_DISABLED = 0,
     -+	FSMONITOR_MODE_HOOK = 1, /* core.fsmonitor */
     -+	FSMONITOR_MODE_IPC = 2,  /* core.useBuiltinFSMonitor */
     ++	FSMONITOR_MODE_HOOK = 1, /* core.fsmonitor=<hook_path> */
     ++	FSMONITOR_MODE_IPC = 2,  /* core.fsmonitor=<true> */
      +};
      +
      +void fsm_settings__set_ipc(struct repository *r);
     @@ fsmonitor.c: void write_fsmonitor_extension(struct strbuf *sb, struct index_stat
       	strvec_pushf(&cp.args, "%s", last_update);
       	cp.use_shell = 1;
      @@ fsmonitor.c: void refresh_fsmonitor(struct index_state *istate)
     - 	struct strbuf last_update_token = STRBUF_INIT;
       	char *buf;
       	unsigned int i;
     + 	int is_trivial = 0;
      +	struct repository *r = istate->repo ? istate->repo : the_repository;
      +	enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
       
     @@ fsmonitor.c: void refresh_fsmonitor(struct index_state *istate)
      +			query_success = !query_fsmonitor_hook(
      +				r, HOOK_INTERFACE_VERSION1,
       				istate->fsmonitor_last_update, &query_result);
     - 		}
     + 			if (query_success)
     + 				is_trivial = query_result.buf[0] == '/';
     +@@ fsmonitor.c: void refresh_fsmonitor(struct index_state *istate)
     + 			trace2_data_intmax("fsm_hook", NULL,
     + 					   "query/trivial-response", 1);
       
      -		trace_performance_since(last_update, "fsmonitor process '%s'", core_fsmonitor);
      -		trace_printf_key(&trace_fsmonitor, "fsmonitor process '%s' returned %s",
     @@ repository.h: struct repo_settings {
       	int command_requires_full_index;
       	int sparse_index;
       
     -+	struct fsmonitor_settings *fsmonitor; /* lazy loaded */
     ++	struct fsmonitor_settings *fsmonitor; /* lazily loaded */
      +
       	int index_version;
       	enum untracked_cache_setting core_untracked_cache;
     @@ t/README: every 'git commit-graph write', as if the `--changed-paths` option was
       GIT_TEST_FSMONITOR=$PWD/t7519/fsmonitor-all exercises the fsmonitor
      -code path for utilizing a file system monitor to speed up detecting
      -new or changed files.
     -+code path for utilizing a (hook based) file system monitor to speed up
     ++code paths for utilizing a (hook based) file system monitor to speed up
      +detecting new or changed files.
       
       GIT_TEST_INDEX_VERSION=<n> exercises the index read/write code path
  4:  de82c726182 !  4:  8e738a83bc5 fsmonitor: use IPC to query the builtin FSMonitor daemon
     @@ Commit message
          fsmonitor: use IPC to query the builtin FSMonitor daemon
      
          Use simple IPC to directly communicate with the new builtin file
     -    system monitor daemon when `core.useBuiltinFSMonitor` is set.
     -
     -    The `core.fsmonitor` setting has already been defined as a HOOK
     -    pathname.  Historically, this has been set to a HOOK script that will
     -    talk with Watchman.  For compatibility reasons, we do not want to
     -    overload that definition (and cause problems if users have multiple
     -    versions of Git installed).
     +    system monitor daemon when `core.fsmonitor` is set to true.
      
          Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
          Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
     @@ fsmonitor.c: void refresh_fsmonitor(struct index_state *istate)
      +			buf = query_result.buf;
      +			strbuf_addstr(&last_update_token, buf);
      +			bol = last_update_token.len + 1;
     ++			is_trivial = query_result.buf[bol] == '/';
     ++			if (is_trivial)
     ++				trace2_data_intmax("fsm_client", NULL,
     ++						   "query/trivial-response", 1);
      +		} else {
      +			/*
      +			 * The builtin daemon is not available on this
  5:  d365704d551 !  5:  49e4c146e02 fsmonitor: document builtin fsmonitor
     @@ Metadata
       ## Commit message ##
          fsmonitor: document builtin fsmonitor
      
     -    Document the new `core.useBuiltinFSMonitor` config value.
     +    Document how `core.fsmonitor` can be set to a boolean to enable
     +    or disable the builtin FSMonitor.
      
          Update references to `core.fsmonitor` and `core.fsmonitorHookVersion` and
          pointers to `Watchman` to refer to it.
     @@ Documentation/config/core.txt: core.protectNTFS::
      -	requested date/time. This information is used to speed up git by
      -	avoiding unnecessary processing of files that have not changed.
      -	See the "fsmonitor-watchman" section of linkgit:githooks[5].
     -+	If set, this variable contains the pathname of the "fsmonitor"
     ++	If set to true, enable the built-in file system monitor
     ++	daemon for this working directory (linkgit:git-fsmonitor--daemon[1]).
     +++
     ++Like hook-based file system monitors, the built-in file system monitor
     ++can speed up Git commands that need to refresh the Git index
     ++(e.g. `git status`) in a working directory with many files.  The
     ++built-in monitor eliminates the need to install and maintain an
     ++external third-party tool.
     +++
     ++The built-in file system monitor is currently available only on a
     ++limited set of supported platforms.  Currently, this includes Windows
     ++and MacOS.
     +++
     ++	Otherwise, this variable contains the pathname of the "fsmonitor"
      +	hook command.
      ++
      +This hook command is used to identify all files that may have changed
     @@ Documentation/config/core.txt: core.protectNTFS::
      +git by avoiding unnecessary scanning of files that have not changed.
      ++
      +See the "fsmonitor-watchman" section of linkgit:githooks[5].
     -++
     -+Note: The value of this config setting is ignored if the
     -+built-in file system monitor is enabled (see `core.useBuiltinFSMonitor`).
       
       core.fsmonitorHookVersion::
      -	Sets the version of hook that is to be used when calling fsmonitor.
     @@ Documentation/config/core.txt: core.protectNTFS::
      +Version 2 uses an opaque string so that the monitor can return
      +something that can be used to determine what files have changed
      +without race conditions.
     -++
     -+Note: The value of this config setting is ignored if the
     -+built-in file system monitor is enabled (see `core.useBuiltinFSMonitor`).
     -+
     -+core.useBuiltinFSMonitor::
     -+	If set to true, enable the built-in file system monitor
     -+	daemon for this working directory (linkgit:git-fsmonitor--daemon[1]).
     -++
     -+Like hook-based file system monitors, the built-in file system monitor
     -+can speed up Git commands that need to refresh the Git index
     -+(e.g. `git status`) in a working directory with many files.  The
     -+built-in monitor eliminates the need to install and maintain an
     -+external third-party tool.
     -++
     -+The built-in file system monitor is currently available only on a
     -+limited set of supported platforms.  Currently, this includes Windows
     -+and MacOS.
     -++
     -+Note: if this config setting is set to `true`, the values of
     -+`core.fsmonitor` and `core.fsmonitorHookVersion` are ignored.
       
       core.trustctime::
       	If false, the ctime differences between the index and the
     @@ Documentation/git-fsmonitor--daemon.txt (new)
      +increased if they just ask for a summary of changes to the working
      +directory and can avoid scanning the disk.
      +
     -+When `core.useBuiltinFSMonitor` is set to `true` (see
     -+linkgit:git-config[1]) commands, such as `git status`, will ask the
     -+daemon for changes and automatically start it (if necessary).
     ++When `core.fsmonitor` is set to `true` (see linkgit:git-config[1])
     ++commands, such as `git status`, will ask the daemon for changes and
     ++automatically start it (if necessary).
      +
      +For more information see the "File System Monitor" section in
      +linkgit:git-update-index[1].
     @@ Documentation/git-update-index.txt: FILE SYSTEM MONITOR
       "fsmonitor-watchman" section of linkgit:githooks[5]) that can
       inform it as to what files have been modified. This enables git to avoid
       having to lstat() every file to find modified files.
     -@@ Documentation/git-update-index.txt: performance by avoiding the cost of scanning the entire working directory
     - looking for new files.
     +@@ Documentation/git-update-index.txt: looking for new files.
       
       If you want to enable (or disable) this feature, it is easier to use
     --the `core.fsmonitor` configuration variable (see
     + the `core.fsmonitor` configuration variable (see
      -linkgit:git-config[1]) than using the `--fsmonitor` option to
      -`git update-index` in each repository, especially if you want to do so
     --across all repositories you use, because you can set the configuration
     --variable in your `$HOME/.gitconfig` just once and have it affect all
     --repositories you touch.
     --
     --When the `core.fsmonitor` configuration variable is changed, the
     --file system monitor is added to or removed from the index the next time
     --a command reads the index. When `--[no-]fsmonitor` are used, the file
     --system monitor is immediately added to or removed from the index.
     -+the `core.fsmonitor` or `core.useBuiltinFSMonitor` configuration
     -+variable (see linkgit:git-config[1]) than using the `--fsmonitor`
     -+option to `git update-index` in each repository, especially if you
     -+want to do so across all repositories you use, because you can set the
     -+configuration variable in your `$HOME/.gitconfig` just once and have
     -+it affect all repositories you touch.
     -+
     -+When the `core.fsmonitor` or `core.useBuiltinFSMonitor` configuration
     -+variable is changed, the file system monitor is added to or removed
     -+from the index the next time a command reads the index. When
     -+`--[no-]fsmonitor` are used, the file system monitor is immediately
     -+added to or removed from the index.
     - 
     - CONFIGURATION
     - -------------
     -
     - ## Documentation/githooks.txt ##
     -@@ Documentation/githooks.txt: fsmonitor-watchman
     - 
     - This hook is invoked when the configuration option `core.fsmonitor` is
     - set to `.git/hooks/fsmonitor-watchman` or `.git/hooks/fsmonitor-watchmanv2`
     --depending on the version of the hook to use.
     -+depending on the version of the hook to use, unless overridden via
     -+`core.useBuiltinFSMonitor` (see linkgit:git-config[1]).
     - 
     - Version 1 takes two arguments, a version (1) and the time in elapsed
     - nanoseconds since midnight, January 1, 1970.
     ++linkgit:git-config[1]) than using the `--fsmonitor` option to `git
     ++update-index` in each repository, especially if you want to do so
     + across all repositories you use, because you can set the configuration
     + variable in your `$HOME/.gitconfig` just once and have it affect all
     + repositories you touch.
  6:  78e682fc530 =  6:  bdd7334da31 fsmonitor--daemon: add a built-in fsmonitor daemon
  7:  ea64b5c9753 =  7:  9f00ada3dd3 fsmonitor--daemon: implement 'stop' and 'status' commands
  8:  5a40b33a00c =  8:  d6819bdad66 compat/fsmonitor/fsm-listen-win32: stub in backend for Windows
  9:  ed5819e29f8 =  9:  7de3d01cccc compat/fsmonitor/fsm-listen-darwin: stub in backend for Darwin
 10:  d3ac973a5f1 = 10:  6fe5a2bc79e fsmonitor--daemon: implement 'run' command
 11:  d08c28b549c = 11:  69fc0998286 fsmonitor--daemon: implement 'start' command
 12:  6fa71fdc825 = 12:  21c099c5197 fsmonitor--daemon: add pathname classification
 13:  65821da5b03 = 13:  72979c35ceb fsmonitor--daemon: define token-ids
 14:  429c48a5bad ! 14:  8d1db644409 fsmonitor--daemon: create token-based changed path cache
     @@ Commit message
          backends to accumulate changed paths in response to filesystem events.
      
          The platform-specific file system listener thread receives file system
     -    events containing one or more changed pathnames (with whatever bucketing
     -    or grouping that is convenient for the file system).  These paths are
     -    accumulated (without locking) by the file system layer into a `fsmonitor_batch`.
     +    events containing one or more changed pathnames (with whatever
     +    bucketing or grouping that is convenient for the file system).  These
     +    paths are accumulated (without locking) by the file system layer into
     +    a `fsmonitor_batch`.
      
          When the file system layer has drained the kernel event queue, it will
          "publish" them to our token queue and make them visible to concurrent
 15:  b04c460c619 = 15:  98c5adf8ca0 compat/fsmonitor/fsm-listen-win32: implement FSMonitor backend on Windows
 16:  862bbfcc32e ! 16:  a3b881315fa compat/fsmonitor/fsm-listen-darwin: add macos header files for FSEvent
     @@ Commit message
          currently fails with this error:
      
          In file included
     -       from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/Library/Frameworks/Security.framework/Headers/AuthSession.h:32,
     -       from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/Library/Frameworks/Security.framework/Headers/Security.h:42,
     -       from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/Library/Frameworks/CoreServices.framework/Frameworks/OSServices.framework/Headers/CSIdentity.h:43,
     -       from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/Library/Frameworks/CoreServices.framework/Frameworks/OSServices.framework/Headers/OSServices.h:29,
     -       from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/Library/Frameworks/CoreServices.framework/Frameworks/LaunchServices.framework/Headers/IconsCore.h:23,
     -       from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/Library/Frameworks/CoreServices.framework/Frameworks/LaunchServices.framework/Headers/LaunchServices.h:23,
     -       from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/Library/Frameworks/CoreServices.framework/Headers/CoreServices.h:45,
     -         /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/Library/Frameworks/Security.framework/Headers/Authorization.h:193:7: error: variably modified 'bytes' at file scope
     +       from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
     +       ...Library/Frameworks/Security.framework/Headers/AuthSession.h:32,
     +       from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
     +       ...Library/Frameworks/Security.framework/Headers/Security.h:42,
     +       from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
     +       ...Library/Frameworks/CoreServices.framework/Frameworks/...
     +       ...OSServices.framework/Headers/CSIdentity.h:43,
     +       from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
     +       ...Library/Frameworks/CoreServices.framework/Frameworks/...
     +       ...OSServices.framework/Headers/OSServices.h:29,
     +       from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
     +       ...Library/Frameworks/CoreServices.framework/Frameworks/...
     +       ...LaunchServices.framework/Headers/IconsCore.h:23,
     +       from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
     +       ...Library/Frameworks/CoreServices.framework/Frameworks/...
     +       ...LaunchServices.framework/Headers/LaunchServices.h:23,
     +       from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
     +       ...Library/Frameworks/CoreServices.framework/Headers/CoreServices.h:45,
     +
     +         /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
     +         ...Library/Frameworks/Security.framework/Headers/Authorization.h:193:7:
     +         error: variably modified 'bytes' at file scope
                 193 | char bytes[kAuthorizationExternalFormLength];
                     |      ^~~~~
      
 17:  40d9a816b52 = 17:  162e357db72 compat/fsmonitor/fsm-listen-darwin: implement FSEvent listener on MacOS
 18:  241962894f1 = 18:  3de1c43beaf fsmonitor--daemon: implement handle_client callback
 19:  704d37d2033 = 19:  3517c4a3c13 help: include fsmonitor--daemon feature flag in version info
 20:  de6c72a9ce0 ! 20:  4ffc2ddf516 t/helper/fsmonitor-client: create IPC client to talk to FSMonitor Daemon
     @@ t/helper/test-tool.c: static struct test_cmd cmds[] = {
       	{ "getcwd", cmd__getcwd },
      
       ## t/helper/test-tool.h ##
     -@@ t/helper/test-tool.h: int cmd__dump_split_index(int argc, const char **argv);
     - int cmd__dump_untracked_cache(int argc, const char **argv);
     +@@ t/helper/test-tool.h: int cmd__dump_untracked_cache(int argc, const char **argv);
     + int cmd__dump_reftable(int argc, const char **argv);
       int cmd__example_decorate(int argc, const char **argv);
       int cmd__fast_rebase(int argc, const char **argv);
      +int cmd__fsmonitor_client(int argc, const char **argv);
 21:  eedaa787c2e ! 21:  d93310a7c64 t7527: create test for fsmonitor--daemon
     @@ t/t7527-builtin-fsmonitor.sh (new)
      +	actual*
      +	EOF
      +
     -+	git -c core.useBuiltinFSMonitor= add . &&
     ++	git -c core.fsmonitor=false add . &&
      +	test_tick &&
     -+	git -c core.useBuiltinFSMonitor= commit -m initial &&
     ++	git -c core.fsmonitor=false commit -m initial &&
      +
     -+	git config core.useBuiltinFSMonitor true
     ++	git config core.fsmonitor true
      +'
      +
      +# The test already explicitly stopped (or tried to stop) the daemon.
     @@ t/t7527-builtin-fsmonitor.sh (new)
      +	test_subcommand git fsmonitor--daemon start <.git/trace_implicit_2
      +'
      +
     -+edit_files() {
     ++edit_files () {
      +	echo 1 >modified
      +	echo 2 >dir1/modified
      +	echo 3 >dir2/modified
      +	>dir1/untracked
      +}
      +
     -+delete_files() {
     ++delete_files () {
      +	rm -f delete
      +	rm -f dir1/delete
      +	rm -f dir2/delete
      +}
      +
     -+create_files() {
     ++create_files () {
      +	echo 1 >new
      +	echo 2 >dir1/new
      +	echo 3 >dir2/new
      +}
      +
     -+rename_files() {
     ++rename_files () {
      +	mv rename renamed
      +	mv dir1/rename dir1/renamed
      +	mv dir2/rename dir2/renamed
      +}
      +
     -+file_to_directory() {
     ++file_to_directory () {
      +	rm -f delete
      +	mkdir delete
      +	echo 1 >delete/new
      +}
      +
     -+directory_to_file() {
     ++directory_to_file () {
      +	rm -rf dir1
      +	echo 1 >dir1
      +}
      +
     -+verify_status() {
     ++verify_status () {
      +	git status >actual &&
      +	GIT_INDEX_FILE=.git/fresh-index git read-tree master &&
     -+	GIT_INDEX_FILE=.git/fresh-index git -c core.useBuiltinFSMonitor= status >expect &&
     ++	GIT_INDEX_FILE=.git/fresh-index git -c core.fsmonitor=false status >expect &&
      +	test_cmp expect actual &&
      +	echo HELLO AFTER &&
      +	cat .git/trace &&
 22:  4e96e0667ba = 22:  27e47108908 t/perf: avoid copying builtin fsmonitor files into test repo
 23:  de9c015d78c = 23:  6cba1d950b0 t/helper/test-chmtime: skip directories on Windows
 24:  1c2eccacff6 ! 24:  fcf843a0d42 t/perf/p7519: speed up test on Windows
     @@ Commit message
          Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
      
       ## t/perf/p7519-fsmonitor.sh ##
     -@@ t/perf/p7519-fsmonitor.sh: trace_stop() {
     +@@ t/perf/p7519-fsmonitor.sh: then
     + 	fi
     + fi
     + 
     +-trace_start() {
     ++trace_start () {
     + 	if test -n "$GIT_PERF_7519_TRACE"
     + 	then
     + 		name="$1"
     +@@ t/perf/p7519-fsmonitor.sh: trace_start() {
     + 	fi
     + }
     + 
     +-trace_stop() {
     ++trace_stop () {
     + 	if test -n "$GIT_PERF_7519_TRACE"
     + 	then
     + 		unset GIT_TRACE2_PERF
       	fi
       }
       
     -+touch_files() {
     ++touch_files () {
      +	n=$1
      +	d="$n"_files
      +
     @@ t/perf/p7519-fsmonitor.sh: test_expect_success "one time repo setup" '
       	fi &&
       
       	mkdir 1_file 10_files 100_files 1000_files 10000_files &&
     --	for i in $(test_seq 1 10); do touch 10_files/$i; done &&
     --	for i in $(test_seq 1 100); do touch 100_files/$i; done &&
     --	for i in $(test_seq 1 1000); do touch 1000_files/$i; done &&
     --	for i in $(test_seq 1 10000); do touch 10000_files/$i; done &&
     +-	for i in $(test_seq 1 10); do touch 10_files/$i || return 1; done &&
     +-	for i in $(test_seq 1 100); do touch 100_files/$i || return 1; done &&
     +-	for i in $(test_seq 1 1000); do touch 1000_files/$i || return 1; done &&
     +-	for i in $(test_seq 1 10000); do touch 10000_files/$i || return 1; done &&
      +	touch_files 1 &&
      +	touch_files 10 &&
      +	touch_files 100 &&
     @@ t/perf/p7519-fsmonitor.sh: test_expect_success "one time repo setup" '
       	git add 1_file 10_files 100_files 1000_files 10000_files &&
       	git commit -qm "Add files" &&
       
     +@@ t/perf/p7519-fsmonitor.sh: test_expect_success "one time repo setup" '
     + 	fi
     + '
     + 
     +-setup_for_fsmonitor() {
     ++setup_for_fsmonitor () {
     + 	# set INTEGRATION_SCRIPT depending on the environment
     + 	if test -n "$INTEGRATION_PATH"
     + 	then
     +@@ t/perf/p7519-fsmonitor.sh: test_perf_w_drop_caches () {
     + 	test_perf "$@"
     + }
     + 
     +-test_fsmonitor_suite() {
     ++test_fsmonitor_suite () {
     + 	if test -n "$INTEGRATION_SCRIPT"; then
     + 		DESC="fsmonitor=$(basename $INTEGRATION_SCRIPT)"
     + 	else
      @@ t/perf/p7519-fsmonitor.sh: test_fsmonitor_suite() {
       
       	# Update the mtimes on upto 100k files to make status think
 25:  236b5966257 ! 25:  198f47bda5a t/perf/p7519: add fsmonitor--daemon test cases
     @@ Commit message
          Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
      
       ## t/perf/p7519-fsmonitor.sh ##
     -@@ t/perf/p7519-fsmonitor.sh: test_description="Test core.fsmonitor"
     - # GIT_PERF_7519_SPLIT_INDEX: used to configure core.splitIndex
     - # GIT_PERF_7519_FSMONITOR: used to configure core.fsMonitor. May be an
     - #   absolute path to an integration. May be a space delimited list of
     --#   absolute paths to integrations.
     -+#   absolute paths to integrations.  (This hook or list of hooks does not
     -+#   include the built-in fsmonitor--daemon.)
     - #
     - # The big win for using fsmonitor is the elimination of the need to scan the
     - # working directory looking for changed and untracked files. If the file
     -@@ t/perf/p7519-fsmonitor.sh: test_expect_success "one time repo setup" '
     - 
     - setup_for_fsmonitor() {
     - 	# set INTEGRATION_SCRIPT depending on the environment
     --	if test -n "$INTEGRATION_PATH"
     -+	if test -n "$USE_FSMONITOR_DAEMON"
     - 	then
     -+		git config core.useBuiltinFSMonitor true &&
     -+		INTEGRATION_SCRIPT=false
     -+	elif test -n "$INTEGRATION_PATH"
     -+	then
     -+		git config core.useBuiltinFSMonitor false &&
     - 		INTEGRATION_SCRIPT="$INTEGRATION_PATH"
     - 	else
     -+		git config core.useBuiltinFSMonitor false &&
     - 		#
     - 		# Choose integration script based on existence of Watchman.
     - 		# Fall back to an empty integration script.
      @@ t/perf/p7519-fsmonitor.sh: test_perf_w_drop_caches () {
       }
       
     - test_fsmonitor_suite() {
     + test_fsmonitor_suite () {
      -	if test -n "$INTEGRATION_SCRIPT"; then
      +	if test -n "$USE_FSMONITOR_DAEMON"
      +	then
     @@ t/perf/p7519-fsmonitor.sh: test_expect_success "setup without fsmonitor" '
      +	git fsmonitor--daemon start
      +
      +	trace_start fsmonitor--daemon--client
     -+	test_expect_success "setup for fsmonitor--daemon" 'setup_for_fsmonitor'
     ++
     ++	git config core.fsmonitor true
     ++	git update-index --fsmonitor
     ++
      +	test_fsmonitor_suite
      +
      +	git fsmonitor--daemon stop
 26:  54710a4830d = 26:  19993c130d2 fsmonitor--daemon: periodically truncate list of modified files
 27:  1e2bd77fcea = 27:  f47a763dc26 fsmonitor--daemon: use a cookie file to sync with file system
 28:  30e61b6d1ad ! 28:  aec44a21afd fsmonitor: force update index after large responses
     @@ fsmonitor.c: apply_results:
       	 */
      +	trace2_region_enter("fsmonitor", "apply_results", istate->repo);
      +
     - 	if (query_success && query_result.buf[bol] != '/') {
     + 	if (query_success && !is_trivial) {
       		/*
       		 * Mark all pathnames returned by the monitor as dirty.
       		 *
     @@ fsmonitor.c: apply_results:
      +
       	} else {
       		/*
     - 		 * We received a trivial response, so invalidate everything.
     + 		 * We failed to get a response or received a trivial response,
      @@ fsmonitor.c: apply_results:
       		if (istate->untracked)
       			istate->untracked->use_fsmonitor = 0;
 29:  507020bbef0 ! 29:  d6039987df8 t7527: test status with untracked-cache and fsmonitor--daemon
     @@ t/t7527-builtin-fsmonitor.sh: test_expect_success 'setup' '
      +	trace*
       	EOF
       
     - 	git -c core.useBuiltinFSMonitor= add . &&
     + 	git -c core.fsmonitor=false add . &&
      @@ t/t7527-builtin-fsmonitor.sh: test_expect_success 'cleanup worktrees' '
       	stop_daemon_delete_repo wt-base
       '
     @@ t/t7527-builtin-fsmonitor.sh: test_expect_success 'cleanup worktrees' '
      +# cause incorrect results when the untracked-cache is enabled.
      +
      +test_lazy_prereq UNTRACKED_CACHE '
     -+	{ git update-index --test-untracked-cache; ret=$?; } &&
     -+	test $ret -ne 1
     ++	git update-index --test-untracked-cache
      +'
      +
      +test_expect_success 'Matrix: setup for untracked-cache,fsmonitor matrix' '
     -+	test_might_fail git config --unset core.useBuiltinFSMonitor &&
     ++	test_unconfig core.fsmonitor &&
      +	git update-index --no-fsmonitor &&
      +	test_might_fail git fsmonitor--daemon stop
      +'
      +
      +matrix_clean_up_repo () {
     -+	git reset --hard HEAD
     ++	git reset --hard HEAD &&
      +	git clean -fd
      +}
      +
     @@ t/t7527-builtin-fsmonitor.sh: test_expect_success 'cleanup worktrees' '
      +	test_expect_success "Matrix[uc:$uc][fsm:$fsm] $fn" '
      +		matrix_clean_up_repo &&
      +		$fn &&
     -+		if test $uc = false -a $fsm = false
     ++		if test $uc = false && test $fsm = false
      +		then
      +			git status --porcelain=v1 >.git/expect.$fn
      +		else
     -+			git status --porcelain=v1 >.git/actual.$fn
     ++			git status --porcelain=v1 >.git/actual.$fn &&
      +			test_cmp .git/expect.$fn .git/actual.$fn
      +		fi
      +	'
     -+
     -+	return $?
      +}
      +
      +uc_values="false"
     @@ t/t7527-builtin-fsmonitor.sh: test_expect_success 'cleanup worktrees' '
      +		if test $fsm_val = false
      +		then
      +			test_expect_success "Matrix[uc:$uc_val][fsm:$fsm_val] disable fsmonitor" '
     -+				test_might_fail git config --unset core.useBuiltinFSMonitor &&
     ++				test_unconfig core.fsmonitor &&
      +				git update-index --no-fsmonitor &&
     -+				test_might_fail git fsmonitor--daemon stop 2>/dev/null
     ++				test_might_fail git fsmonitor--daemon stop
      +			'
      +		else
      +			test_expect_success "Matrix[uc:$uc_val][fsm:$fsm_val] enable fsmonitor" '
     -+				git config core.useBuiltinFSMonitor true &&
     ++				git config core.fsmonitor true &&
      +				git fsmonitor--daemon start &&
      +				git update-index --fsmonitor
      +			'
     @@ t/t7527-builtin-fsmonitor.sh: test_expect_success 'cleanup worktrees' '
      +		if test $fsm_val = true
      +		then
      +			test_expect_success "Matrix[uc:$uc_val][fsm:$fsm_val] disable fsmonitor at end" '
     -+				test_might_fail git config --unset core.useBuiltinFSMonitor &&
     ++				test_unconfig core.fsmonitor &&
      +				git update-index --no-fsmonitor &&
     -+				test_might_fail git fsmonitor--daemon stop 2>/dev/null
     ++				test_might_fail git fsmonitor--daemon stop
      +			'
      +		fi
      +	done
  -:  ----------- > 30:  5117fbdfc63 update-index: convert fsmonitor warnings to advise

-- 
gitgitgadget

[PATCH v5 01/30] fsmonitor: enhance existing comments, clarify trivial response handling

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 fsmonitor.c | 64 ++++++++++++++++++++++++++++++++++-------------------
 1 file changed, 41 insertions(+), 23 deletions(-)

diff --git a/fsmonitor.c b/fsmonitor.c
index ab9bfc60b34..448d0ee33f5 100644
--- a/fsmonitor.c
+++ b/fsmonitor.c
@@ -168,29 +168,15 @@ static int query_fsmonitor(int version, const char *last_update, struct strbuf *
 
 	if (result)
 		trace2_data_intmax("fsm_hook", NULL, "query/failed", result);
-	else {
+	else
 		trace2_data_intmax("fsm_hook", NULL, "query/response-length",
 				   query_result->len);
 
-		if (fsmonitor_is_trivial_response(query_result))
-			trace2_data_intmax("fsm_hook", NULL,
-					   "query/trivial-response", 1);
-	}
-
 	trace2_region_leave("fsm_hook", "query", NULL);
 
 	return result;
 }
 
-int fsmonitor_is_trivial_response(const struct strbuf *query_result)
-{
-	static char trivial_response[3] = { '\0', '/', '\0' };
-
-	return query_result->len >= 3 &&
-		!memcmp(trivial_response,
-			&query_result->buf[query_result->len - 3], 3);
-}
-
 static void fsmonitor_refresh_callback(struct index_state *istate, char *name)
 {
 	int i, len = strlen(name);
@@ -238,6 +224,7 @@ void refresh_fsmonitor(struct index_state *istate)
 	struct strbuf last_update_token = STRBUF_INIT;
 	char *buf;
 	unsigned int i;
+	int is_trivial = 0;
 
 	if (!core_fsmonitor || istate->fsmonitor_has_run_once)
 		return;
@@ -283,6 +270,7 @@ void refresh_fsmonitor(struct index_state *istate)
 					query_success = 0;
 				} else {
 					bol = last_update_token.len + 1;
+					is_trivial = query_result.buf[bol] == '/';
 				}
 			} else if (hook_version < 0) {
 				hook_version = HOOK_INTERFACE_VERSION1;
@@ -294,16 +282,38 @@ void refresh_fsmonitor(struct index_state *istate)
 		if (hook_version == HOOK_INTERFACE_VERSION1) {
 			query_success = !query_fsmonitor(HOOK_INTERFACE_VERSION1,
 				istate->fsmonitor_last_update, &query_result);
+			if (query_success)
+				is_trivial = query_result.buf[0] == '/';
 		}
 
+		if (is_trivial)
+			trace2_data_intmax("fsm_hook", NULL,
+					   "query/trivial-response", 1);
+
 		trace_performance_since(last_update, "fsmonitor process '%s'", core_fsmonitor);
 		trace_printf_key(&trace_fsmonitor, "fsmonitor process '%s' returned %s",
 			core_fsmonitor, query_success ? "success" : "failure");
 	}
 
-	/* a fsmonitor process can return '/' to indicate all entries are invalid */
-	if (query_success && query_result.buf[bol] != '/') {
-		/* Mark all entries returned by the monitor as dirty */
+	/*
+	 * The response from FSMonitor (excluding the header token) is
+	 * either:
+	 *
+	 * [a] a (possibly empty) list of NUL delimited relative
+	 *     pathnames of changed paths.  This list can contain
+	 *     files and directories.  Directories have a trailing
+	 *     slash.
+	 *
+	 * [b] a single '/' to indicate the provider had no
+	 *     information and that we should consider everything
+	 *     invalid.  We call this a trivial response.
+	 */
+	if (query_success && !is_trivial) {
+		/*
+		 * Mark all pathnames returned by the monitor as dirty.
+		 *
+		 * This updates both the cache-entries and the untracked-cache.
+		 */
 		buf = query_result.buf;
 		for (i = bol; i < query_result.len; i++) {
 			if (buf[i] != '\0')
@@ -318,11 +328,16 @@ void refresh_fsmonitor(struct index_state *istate)
 		if (istate->untracked)
 			istate->untracked->use_fsmonitor = 1;
 	} else {
-
-		/* We only want to run the post index changed hook if we've actually changed entries, so keep track
-		 * if we actually changed entries or not */
+		/*
+		 * We failed to get a response or received a trivial response,
+		 * so invalidate everything.
+		 *
+		 * We only want to run the post index changed hook if
+		 * we've actually changed entries, so keep track if we
+		 * actually changed entries or not.
+		 */
 		int is_cache_changed = 0;
-		/* Mark all entries invalid */
+
 		for (i = 0; i < istate->cache_nr; i++) {
 			if (istate->cache[i]->ce_flags & CE_FSMONITOR_VALID) {
 				is_cache_changed = 1;
@@ -330,7 +345,10 @@ void refresh_fsmonitor(struct index_state *istate)
 			}
 		}
 
-		/* If we're going to check every file, ensure we save the results */
+		/*
+		 * If we're going to check every file, ensure we save
+		 * the results.
+		 */
 		if (is_cache_changed)
 			istate->cache_changed |= FSMONITOR_CHANGED;
 
-- 
gitgitgadget

[PATCH v5 02/30] fsmonitor-ipc: create client routines for git-fsmonitor--daemon

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Create fsmonitor_ipc__*() client routines to spawn the built-in file
system monitor daemon and send it an IPC request using the `Simple
IPC` API.

Stub in empty fsmonitor_ipc__*() functions for unsupported platforms.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 Makefile        |   1 +
 fsmonitor-ipc.c | 171 ++++++++++++++++++++++++++++++++++++++++++++++++
 fsmonitor-ipc.h |  48 ++++++++++++++
 3 files changed, 220 insertions(+)
 create mode 100644 fsmonitor-ipc.c
 create mode 100644 fsmonitor-ipc.h

diff --git a/Makefile b/Makefile
index 8e07003840b..dee3ba4a0b1 100644
--- a/Makefile
+++ b/Makefile
@@ -899,6 +899,7 @@ LIB_OBJS += fetch-pack.o
 LIB_OBJS += fmt-merge-msg.o
 LIB_OBJS += fsck.o
 LIB_OBJS += fsmonitor.o
+LIB_OBJS += fsmonitor-ipc.o
 LIB_OBJS += gettext.o
 LIB_OBJS += gpg-interface.o
 LIB_OBJS += graph.o
diff --git a/fsmonitor-ipc.c b/fsmonitor-ipc.c
new file mode 100644
index 00000000000..01c8c25bf50
--- /dev/null
+++ b/fsmonitor-ipc.c
@@ -0,0 +1,171 @@
+#include "cache.h"
+#include "fsmonitor.h"
+#include "simple-ipc.h"
+#include "fsmonitor-ipc.h"
+#include "run-command.h"
+#include "strbuf.h"
+#include "trace2.h"
+
+#ifdef HAVE_FSMONITOR_DAEMON_BACKEND
+
+int fsmonitor_ipc__is_supported(void)
+{
+	return 1;
+}
+
+GIT_PATH_FUNC(fsmonitor_ipc__get_path, "fsmonitor--daemon.ipc")
+
+enum ipc_active_state fsmonitor_ipc__get_state(void)
+{
+	return ipc_get_active_state(fsmonitor_ipc__get_path());
+}
+
+static int spawn_daemon(void)
+{
+	const char *args[] = { "fsmonitor--daemon", "start", NULL };
+
+	return run_command_v_opt_tr2(args, RUN_COMMAND_NO_STDIN | RUN_GIT_CMD,
+				    "fsmonitor");
+}
+
+int fsmonitor_ipc__send_query(const char *since_token,
+			      struct strbuf *answer)
+{
+	int ret = -1;
+	int tried_to_spawn = 0;
+	enum ipc_active_state state = IPC_STATE__OTHER_ERROR;
+	struct ipc_client_connection *connection = NULL;
+	struct ipc_client_connect_options options
+		= IPC_CLIENT_CONNECT_OPTIONS_INIT;
+	const char *tok = since_token ? since_token : "";
+	size_t tok_len = since_token ? strlen(since_token) : 0;
+
+	options.wait_if_busy = 1;
+	options.wait_if_not_found = 0;
+
+	trace2_region_enter("fsm_client", "query", NULL);
+	trace2_data_string("fsm_client", NULL, "query/command", tok);
+
+try_again:
+	state = ipc_client_try_connect(fsmonitor_ipc__get_path(), &options,
+				       &connection);
+
+	switch (state) {
+	case IPC_STATE__LISTENING:
+		ret = ipc_client_send_command_to_connection(
+			connection, tok, tok_len, answer);
+		ipc_client_close_connection(connection);
+
+		trace2_data_intmax("fsm_client", NULL,
+				   "query/response-length", answer->len);
+		goto done;
+
+	case IPC_STATE__NOT_LISTENING:
+	case IPC_STATE__PATH_NOT_FOUND:
+		if (tried_to_spawn)
+			goto done;
+
+		tried_to_spawn++;
+		if (spawn_daemon())
+			goto done;
+
+		/*
+		 * Try again, but this time give the daemon a chance to
+		 * actually create the pipe/socket.
+		 *
+		 * Granted, the daemon just started so it can't possibly have
+		 * any FS cached yet, so we'll always get a trivial answer.
+		 * BUT the answer should include a new token that can serve
+		 * as the basis for subsequent requests.
+		 */
+		options.wait_if_not_found = 1;
+		goto try_again;
+
+	case IPC_STATE__INVALID_PATH:
+		ret = error(_("fsmonitor_ipc__send_query: invalid path '%s'"),
+			    fsmonitor_ipc__get_path());
+		goto done;
+
+	case IPC_STATE__OTHER_ERROR:
+	default:
+		ret = error(_("fsmonitor_ipc__send_query: unspecified error on '%s'"),
+			    fsmonitor_ipc__get_path());
+		goto done;
+	}
+
+done:
+	trace2_region_leave("fsm_client", "query", NULL);
+
+	return ret;
+}
+
+int fsmonitor_ipc__send_command(const char *command,
+				struct strbuf *answer)
+{
+	struct ipc_client_connection *connection = NULL;
+	struct ipc_client_connect_options options
+		= IPC_CLIENT_CONNECT_OPTIONS_INIT;
+	int ret;
+	enum ipc_active_state state;
+	const char *c = command ? command : "";
+	size_t c_len = command ? strlen(command) : 0;
+
+	strbuf_reset(answer);
+
+	options.wait_if_busy = 1;
+	options.wait_if_not_found = 0;
+
+	state = ipc_client_try_connect(fsmonitor_ipc__get_path(), &options,
+				       &connection);
+	if (state != IPC_STATE__LISTENING) {
+		die("fsmonitor--daemon is not running");
+		return -1;
+	}
+
+	ret = ipc_client_send_command_to_connection(connection, c, c_len,
+						    answer);
+	ipc_client_close_connection(connection);
+
+	if (ret == -1) {
+		die("could not send '%s' command to fsmonitor--daemon", c);
+		return -1;
+	}
+
+	return 0;
+}
+
+#else
+
+/*
+ * A trivial implementation of the fsmonitor_ipc__ API for unsupported
+ * platforms.
+ */
+
+int fsmonitor_ipc__is_supported(void)
+{
+	return 0;
+}
+
+const char *fsmonitor_ipc__get_path(void)
+{
+	return NULL;
+}
+
+enum ipc_active_state fsmonitor_ipc__get_state(void)
+{
+	return IPC_STATE__OTHER_ERROR;
+}
+
+int fsmonitor_ipc__send_query(const char *since_token,
+			      struct strbuf *answer)
+{
+	return -1;
+}
+
+int fsmonitor_ipc__send_command(const char *command,
+				struct strbuf *answer)
+{
+	return -1;
+}
+
+#endif
diff --git a/fsmonitor-ipc.h b/fsmonitor-ipc.h
new file mode 100644
index 00000000000..b6a7067c3af
--- /dev/null
+++ b/fsmonitor-ipc.h
@@ -0,0 +1,48 @@
+#ifndef FSMONITOR_IPC_H
+#define FSMONITOR_IPC_H
+
+#include "simple-ipc.h"
+
+/*
+ * Returns true if built-in file system monitor daemon is defined
+ * for this platform.
+ */
+int fsmonitor_ipc__is_supported(void);
+
+/*
+ * Returns the pathname to the IPC named pipe or Unix domain socket
+ * where a `git-fsmonitor--daemon` process will listen.  This is a
+ * per-worktree value.
+ *
+ * Returns NULL if the daemon is not supported on this platform.
+ */
+const char *fsmonitor_ipc__get_path(void);
+
+/*
+ * Try to determine whether there is a `git-fsmonitor--daemon` process
+ * listening on the IPC pipe/socket.
+ */
+enum ipc_active_state fsmonitor_ipc__get_state(void);
+
+/*
+ * Connect to a `git-fsmonitor--daemon` process via simple-ipc
+ * and ask for the set of changed files since the given token.
+ *
+ * Spawn a daemon process in the background if necessary.
+ *
+ * Returns -1 on error; 0 on success.
+ */
+int fsmonitor_ipc__send_query(const char *since_token,
+			      struct strbuf *answer);
+
+/*
+ * Connect to a `git-fsmonitor--daemon` process via simple-ipc and
+ * send a command verb.  If no daemon is available, we DO NOT try to
+ * start one.
+ *
+ * Returns -1 on error; 0 on success.
+ */
+int fsmonitor_ipc__send_command(const char *command,
+				struct strbuf *answer);
+
+#endif /* FSMONITOR_IPC_H */
-- 
gitgitgadget

[PATCH v5 03/30] fsmonitor: config settings are repository-specific

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Move fsmonitor config settings to a new and opaque
`struct fsmonitor_settings` structure.  Add a lazily-loaded pointer
to this into `struct repo_settings`

Create an `enum fsmonitor_mode` type in `struct fsmonitor_settings` to
represent the state of fsmonitor.  This lets us represent which, if
any, fsmonitor provider (hook or IPC) is enabled.

Create `fsm_settings__get_*()` getters to lazily look up fsmonitor-
related config settings.

Get rid of the `core_fsmonitor` global variable.  Move the code to
lookup the existing `core.fsmonitor` config value into the fsmonitor
settings.

Create a hook pathname variable in `struct fsmonitor-settings` and
only set it when in hook mode.

Extend the definition of `core.fsmonitor` to be either a boolean
or a hook pathname.  When true, the builtin FSMonitor is used.
When false or unset, no FSMonitor (neither builtin nor hook) is
used.

The existing `core_fsmonitor` global variable was used to store the
pathname to the fsmonitor hook *and* it was used as a boolean to see
if fsmonitor was enabled.  This dual usage and global visibility leads
to confusion when we add the IPC-based provider.  So lets hide the
details in fsmonitor-settings.c and let it decide which provider to
use in the case of multiple settings.  This avoids cluttering up
repo-settings.c with these private details.

A future commit in builtin-fsmonitor series will add the ability to
disqualify worktrees for various reasons, such as being mounted from a
remote volume, where fsmonitor should not be started.  Having the
config settings hidden in fsmonitor-settings.c allows such worktree
restrictions to override the config values used.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 Makefile               |   1 +
 builtin/update-index.c |   7 ++-
 cache.h                |   1 -
 config.c               |  14 ------
 config.h               |   1 -
 environment.c          |   1 -
 fsmonitor-settings.c   | 100 +++++++++++++++++++++++++++++++++++++++++
 fsmonitor-settings.h   |  21 +++++++++
 fsmonitor.c            |  63 +++++++++++++++-----------
 fsmonitor.h            |  18 ++++++--
 repository.h           |   3 ++
 t/README               |   4 +-
 12 files changed, 185 insertions(+), 49 deletions(-)
 create mode 100644 fsmonitor-settings.c
 create mode 100644 fsmonitor-settings.h

diff --git a/Makefile b/Makefile
index dee3ba4a0b1..9943f0f7c11 100644
--- a/Makefile
+++ b/Makefile
@@ -900,6 +900,7 @@ LIB_OBJS += fmt-merge-msg.o
 LIB_OBJS += fsck.o
 LIB_OBJS += fsmonitor.o
 LIB_OBJS += fsmonitor-ipc.o
+LIB_OBJS += fsmonitor-settings.o
 LIB_OBJS += gettext.o
 LIB_OBJS += gpg-interface.o
 LIB_OBJS += graph.o
diff --git a/builtin/update-index.c b/builtin/update-index.c
index 7e0a0d9bf80..fed24ea1fb6 100644
--- a/builtin/update-index.c
+++ b/builtin/update-index.c
@@ -1225,14 +1225,17 @@ int cmd_update_index(int argc, const char **argv, const char *prefix)
 	}
 
 	if (fsmonitor > 0) {
-		if (git_config_get_fsmonitor() == 0)
+		enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
+		if (fsm_mode == FSMONITOR_MODE_DISABLED) {
 			warning(_("core.fsmonitor is unset; "
 				"set it if you really want to "
 				"enable fsmonitor"));
+		}
 		add_fsmonitor(&the_index);
 		report(_("fsmonitor enabled"));
 	} else if (!fsmonitor) {
-		if (git_config_get_fsmonitor() == 1)
+		enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
+		if (fsm_mode > FSMONITOR_MODE_DISABLED)
 			warning(_("core.fsmonitor is set; "
 				"remove it if you really want to "
 				"disable fsmonitor"));
diff --git a/cache.h b/cache.h
index 4148b6322d5..c67c426a9dd 100644
--- a/cache.h
+++ b/cache.h
@@ -1000,7 +1000,6 @@ extern int core_preload_index;
 extern int precomposed_unicode;
 extern int protect_hfs;
 extern int protect_ntfs;
-extern const char *core_fsmonitor;
 
 extern int core_apply_sparse_checkout;
 extern int core_sparse_checkout_cone;
diff --git a/config.c b/config.c
index e0c03d154c9..0494d8256d0 100644
--- a/config.c
+++ b/config.c
@@ -2624,20 +2624,6 @@ int git_config_get_max_percent_split_change(void)
 	return -1; /* default value */
 }
 
-int git_config_get_fsmonitor(void)
-{
-	if (git_config_get_pathname("core.fsmonitor", &core_fsmonitor))
-		core_fsmonitor = getenv("GIT_TEST_FSMONITOR");
-
-	if (core_fsmonitor && !*core_fsmonitor)
-		core_fsmonitor = NULL;
-
-	if (core_fsmonitor)
-		return 1;
-
-	return 0;
-}
-
 int git_config_get_index_threads(int *dest)
 {
 	int is_bool, val;
diff --git a/config.h b/config.h
index ab0106d2875..03294d340d1 100644
--- a/config.h
+++ b/config.h
@@ -590,7 +590,6 @@ int git_config_get_pathname(const char *key, const char **dest);
 int git_config_get_index_threads(int *dest);
 int git_config_get_split_index(void);
 int git_config_get_max_percent_split_change(void);
-int git_config_get_fsmonitor(void);
 
 /* This dies if the configured or default date is in the future */
 int git_config_get_expiry(const char *key, const char **output);
diff --git a/environment.c b/environment.c
index fd0501e77a5..00682e638d7 100644
--- a/environment.c
+++ b/environment.c
@@ -84,7 +84,6 @@ int protect_hfs = PROTECT_HFS_DEFAULT;
 #define PROTECT_NTFS_DEFAULT 1
 #endif
 int protect_ntfs = PROTECT_NTFS_DEFAULT;
-const char *core_fsmonitor;
 
 /*
  * The character that begins a commented line in user-editable file
diff --git a/fsmonitor-settings.c b/fsmonitor-settings.c
new file mode 100644
index 00000000000..eb4d661c81e
--- /dev/null
+++ b/fsmonitor-settings.c
@@ -0,0 +1,100 @@
+#include "cache.h"
+#include "config.h"
+#include "repository.h"
+#include "fsmonitor-settings.h"
+
+/*
+ * We keep this structure defintion private and have getters
+ * for all fields so that we can lazy load it as needed.
+ */
+struct fsmonitor_settings {
+	enum fsmonitor_mode mode;
+	char *hook_path;
+};
+
+static void lookup_fsmonitor_settings(struct repository *r)
+{
+	struct fsmonitor_settings *s;
+	const char *const_str;
+	int bool_value;
+
+	if (r->settings.fsmonitor)
+		return;
+
+	CALLOC_ARRAY(s, 1);
+
+	r->settings.fsmonitor = s;
+
+	fsm_settings__set_disabled(r);
+
+	/*
+	 * Overload the existing "core.fsmonitor" config setting (which
+	 * has historically been either unset or a hook pathname) to
+	 * now allow a boolean value to enable the builtin FSMonitor
+	 * or to turn everything off.  (This does imply that you can't
+	 * use a hook script named "true" or "false", but that's OK.)
+	 */
+	switch (repo_config_get_maybe_bool(r, "core.fsmonitor", &bool_value)) {
+
+	case 0: /* config value was set to <bool> */
+		if (bool_value)
+			fsm_settings__set_ipc(r);
+		return;
+
+	case 1: /* config value was unset */
+		const_str = getenv("GIT_TEST_FSMONITOR");
+		break;
+
+	case -1: /* config value set to an arbitrary string */
+		if (repo_config_get_pathname(r, "core.fsmonitor", &const_str))
+			return; /* should not happen */
+		break;
+
+	default: /* should not happen */
+		return;
+	}
+
+	if (!const_str || !*const_str)
+		return;
+
+	fsm_settings__set_hook(r, const_str);
+}
+
+enum fsmonitor_mode fsm_settings__get_mode(struct repository *r)
+{
+	lookup_fsmonitor_settings(r);
+
+	return r->settings.fsmonitor->mode;
+}
+
+const char *fsm_settings__get_hook_path(struct repository *r)
+{
+	lookup_fsmonitor_settings(r);
+
+	return r->settings.fsmonitor->hook_path;
+}
+
+void fsm_settings__set_ipc(struct repository *r)
+{
+	lookup_fsmonitor_settings(r);
+
+	r->settings.fsmonitor->mode = FSMONITOR_MODE_IPC;
+	FREE_AND_NULL(r->settings.fsmonitor->hook_path);
+}
+
+void fsm_settings__set_hook(struct repository *r, const char *path)
+{
+	lookup_fsmonitor_settings(r);
+
+	r->settings.fsmonitor->mode = FSMONITOR_MODE_HOOK;
+	FREE_AND_NULL(r->settings.fsmonitor->hook_path);
+	r->settings.fsmonitor->hook_path = strdup(path);
+}
+
+void fsm_settings__set_disabled(struct repository *r)
+{
+	lookup_fsmonitor_settings(r);
+
+	r->settings.fsmonitor->mode = FSMONITOR_MODE_DISABLED;
+	FREE_AND_NULL(r->settings.fsmonitor->hook_path);
+}
diff --git a/fsmonitor-settings.h b/fsmonitor-settings.h
new file mode 100644
index 00000000000..a4c5d7b4889
--- /dev/null
+++ b/fsmonitor-settings.h
@@ -0,0 +1,21 @@
+#ifndef FSMONITOR_SETTINGS_H
+#define FSMONITOR_SETTINGS_H
+
+struct repository;
+
+enum fsmonitor_mode {
+	FSMONITOR_MODE_DISABLED = 0,
+	FSMONITOR_MODE_HOOK = 1, /* core.fsmonitor=<hook_path> */
+	FSMONITOR_MODE_IPC = 2,  /* core.fsmonitor=<true> */
+};
+
+void fsm_settings__set_ipc(struct repository *r);
+void fsm_settings__set_hook(struct repository *r, const char *path);
+void fsm_settings__set_disabled(struct repository *r);
+
+enum fsmonitor_mode fsm_settings__get_mode(struct repository *r);
+const char *fsm_settings__get_hook_path(struct repository *r);
+
+struct fsmonitor_settings;
+
+#endif /* FSMONITOR_SETTINGS_H */
diff --git a/fsmonitor.c b/fsmonitor.c
index 448d0ee33f5..17cba68235a 100644
--- a/fsmonitor.c
+++ b/fsmonitor.c
@@ -3,6 +3,7 @@
 #include "dir.h"
 #include "ewah/ewok.h"
 #include "fsmonitor.h"
+#include "fsmonitor-ipc.h"
 #include "run-command.h"
 #include "strbuf.h"
 
@@ -148,15 +149,18 @@ void write_fsmonitor_extension(struct strbuf *sb, struct index_state *istate)
 /*
  * Call the query-fsmonitor hook passing the last update token of the saved results.
  */
-static int query_fsmonitor(int version, const char *last_update, struct strbuf *query_result)
+static int query_fsmonitor_hook(struct repository *r,
+				int version,
+				const char *last_update,
+				struct strbuf *query_result)
 {
 	struct child_process cp = CHILD_PROCESS_INIT;
 	int result;
 
-	if (!core_fsmonitor)
+	if (fsm_settings__get_mode(r) != FSMONITOR_MODE_HOOK)
 		return -1;
 
-	strvec_push(&cp.args, core_fsmonitor);
+	strvec_push(&cp.args, fsm_settings__get_hook_path(r));
 	strvec_pushf(&cp.args, "%d", version);
 	strvec_pushf(&cp.args, "%s", last_update);
 	cp.use_shell = 1;
@@ -225,17 +229,28 @@ void refresh_fsmonitor(struct index_state *istate)
 	char *buf;
 	unsigned int i;
 	int is_trivial = 0;
+	struct repository *r = istate->repo ? istate->repo : the_repository;
+	enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
 
-	if (!core_fsmonitor || istate->fsmonitor_has_run_once)
+	if (fsm_mode <= FSMONITOR_MODE_DISABLED ||
+	    istate->fsmonitor_has_run_once)
 		return;
 
-	hook_version = fsmonitor_hook_version();
-
 	istate->fsmonitor_has_run_once = 1;
 
 	trace_printf_key(&trace_fsmonitor, "refresh fsmonitor");
+
+	if (fsm_mode == FSMONITOR_MODE_IPC) {
+		/* TODO */
+		return;
+	}
+
+	assert(fsm_mode == FSMONITOR_MODE_HOOK);
+
+	hook_version = fsmonitor_hook_version();
+
 	/*
-	 * This could be racy so save the date/time now and query_fsmonitor
+	 * This could be racy so save the date/time now and query_fsmonitor_hook
 	 * should be inclusive to ensure we don't miss potential changes.
 	 */
 	last_update = getnanotime();
@@ -243,13 +258,14 @@ void refresh_fsmonitor(struct index_state *istate)
 		strbuf_addf(&last_update_token, "%"PRIu64"", last_update);
 
 	/*
-	 * If we have a last update token, call query_fsmonitor for the set of
+	 * If we have a last update token, call query_fsmonitor_hook for the set of
 	 * changes since that token, else assume everything is possibly dirty
 	 * and check it all.
 	 */
 	if (istate->fsmonitor_last_update) {
 		if (hook_version == -1 || hook_version == HOOK_INTERFACE_VERSION2) {
-			query_success = !query_fsmonitor(HOOK_INTERFACE_VERSION2,
+			query_success = !query_fsmonitor_hook(
+				r, HOOK_INTERFACE_VERSION2,
 				istate->fsmonitor_last_update, &query_result);
 
 			if (query_success) {
@@ -280,7 +296,8 @@ void refresh_fsmonitor(struct index_state *istate)
 		}
 
 		if (hook_version == HOOK_INTERFACE_VERSION1) {
-			query_success = !query_fsmonitor(HOOK_INTERFACE_VERSION1,
+			query_success = !query_fsmonitor_hook(
+				r, HOOK_INTERFACE_VERSION1,
 				istate->fsmonitor_last_update, &query_result);
 			if (query_success)
 				is_trivial = query_result.buf[0] == '/';
@@ -290,9 +307,12 @@ void refresh_fsmonitor(struct index_state *istate)
 			trace2_data_intmax("fsm_hook", NULL,
 					   "query/trivial-response", 1);
 
-		trace_performance_since(last_update, "fsmonitor process '%s'", core_fsmonitor);
-		trace_printf_key(&trace_fsmonitor, "fsmonitor process '%s' returned %s",
-			core_fsmonitor, query_success ? "success" : "failure");
+		trace_performance_since(last_update, "fsmonitor process '%s'",
+					fsm_settings__get_hook_path(r));
+		trace_printf_key(&trace_fsmonitor,
+				 "fsmonitor process '%s' returned %s",
+				 fsm_settings__get_hook_path(r),
+				 query_success ? "success" : "failure");
 	}
 
 	/*
@@ -429,7 +449,8 @@ void remove_fsmonitor(struct index_state *istate)
 void tweak_fsmonitor(struct index_state *istate)
 {
 	unsigned int i;
-	int fsmonitor_enabled = git_config_get_fsmonitor();
+	struct repository *r = istate->repo ? istate->repo : the_repository;
+	int fsmonitor_enabled = (fsm_settings__get_mode(r) > FSMONITOR_MODE_DISABLED);
 
 	if (istate->fsmonitor_dirty) {
 		if (fsmonitor_enabled) {
@@ -449,16 +470,8 @@ void tweak_fsmonitor(struct index_state *istate)
 		istate->fsmonitor_dirty = NULL;
 	}
 
-	switch (fsmonitor_enabled) {
-	case -1: /* keep: do nothing */
-		break;
-	case 0: /* false */
-		remove_fsmonitor(istate);
-		break;
-	case 1: /* true */
+	if (fsmonitor_enabled)
 		add_fsmonitor(istate);
-		break;
-	default: /* unknown value: do nothing */
-		break;
-	}
+	else
+		remove_fsmonitor(istate);
 }
diff --git a/fsmonitor.h b/fsmonitor.h
index f20d72631d7..f9201411aa7 100644
--- a/fsmonitor.h
+++ b/fsmonitor.h
@@ -3,6 +3,7 @@
 
 #include "cache.h"
 #include "dir.h"
+#include "fsmonitor-settings.h"
 
 extern struct trace_key trace_fsmonitor;
 
@@ -57,7 +58,11 @@ int fsmonitor_is_trivial_response(const struct strbuf *query_result);
  */
 static inline int is_fsmonitor_refreshed(const struct index_state *istate)
 {
-	return !core_fsmonitor || istate->fsmonitor_has_run_once;
+	struct repository *r = istate->repo ? istate->repo : the_repository;
+	enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
+
+	return fsm_mode <= FSMONITOR_MODE_DISABLED ||
+		istate->fsmonitor_has_run_once;
 }
 
 /*
@@ -67,7 +72,11 @@ static inline int is_fsmonitor_refreshed(const struct index_state *istate)
  */
 static inline void mark_fsmonitor_valid(struct index_state *istate, struct cache_entry *ce)
 {
-	if (core_fsmonitor && !(ce->ce_flags & CE_FSMONITOR_VALID)) {
+	struct repository *r = istate->repo ? istate->repo : the_repository;
+	enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
+
+	if (fsm_mode > FSMONITOR_MODE_DISABLED &&
+	    !(ce->ce_flags & CE_FSMONITOR_VALID)) {
 		istate->cache_changed = 1;
 		ce->ce_flags |= CE_FSMONITOR_VALID;
 		trace_printf_key(&trace_fsmonitor, "mark_fsmonitor_clean '%s'", ce->name);
@@ -83,7 +92,10 @@ static inline void mark_fsmonitor_valid(struct index_state *istate, struct cache
  */
 static inline void mark_fsmonitor_invalid(struct index_state *istate, struct cache_entry *ce)
 {
-	if (core_fsmonitor) {
+	struct repository *r = istate->repo ? istate->repo : the_repository;
+	enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
+
+	if (fsm_mode > FSMONITOR_MODE_DISABLED) {
 		ce->ce_flags &= ~CE_FSMONITOR_VALID;
 		untracked_cache_invalidate_path(istate, ce->name, 1);
 		trace_printf_key(&trace_fsmonitor, "mark_fsmonitor_invalid '%s'", ce->name);
diff --git a/repository.h b/repository.h
index 2b5cf97f31e..6708207cd4c 100644
--- a/repository.h
+++ b/repository.h
@@ -4,6 +4,7 @@
 #include "path.h"
 
 struct config_set;
+struct fsmonitor_settings;
 struct git_hash_algo;
 struct index_state;
 struct lock_file;
@@ -35,6 +36,8 @@ struct repo_settings {
 	int command_requires_full_index;
 	int sparse_index;
 
+	struct fsmonitor_settings *fsmonitor; /* lazily loaded */
+
 	int index_version;
 	enum untracked_cache_setting core_untracked_cache;
 
diff --git a/t/README b/t/README
index f48e0542cdc..9ffea1d3147 100644
--- a/t/README
+++ b/t/README
@@ -405,8 +405,8 @@ every 'git commit-graph write', as if the `--changed-paths` option was
 passed in.
 
 GIT_TEST_FSMONITOR=$PWD/t7519/fsmonitor-all exercises the fsmonitor
-code path for utilizing a file system monitor to speed up detecting
-new or changed files.
+code paths for utilizing a (hook based) file system monitor to speed up
+detecting new or changed files.
 
 GIT_TEST_INDEX_VERSION=<n> exercises the index read/write code path
 for the index version specified.  Can be set to any valid version
-- 
gitgitgadget

[PATCH v5 04/30] fsmonitor: use IPC to query the builtin FSMonitor daemon

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Use simple IPC to directly communicate with the new builtin file
system monitor daemon when `core.fsmonitor` is set to true.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 fsmonitor.c | 38 ++++++++++++++++++++++++++++++++++++--
 1 file changed, 36 insertions(+), 2 deletions(-)

diff --git a/fsmonitor.c b/fsmonitor.c
index 17cba68235a..4287aad6bbb 100644
--- a/fsmonitor.c
+++ b/fsmonitor.c
@@ -241,8 +241,41 @@ void refresh_fsmonitor(struct index_state *istate)
 	trace_printf_key(&trace_fsmonitor, "refresh fsmonitor");
 
 	if (fsm_mode == FSMONITOR_MODE_IPC) {
-		/* TODO */
-		return;
+		query_success = !fsmonitor_ipc__send_query(
+			istate->fsmonitor_last_update ?
+			istate->fsmonitor_last_update : "builtin:fake",
+			&query_result);
+		if (query_success) {
+			/*
+			 * The response contains a series of nul terminated
+			 * strings.  The first is the new token.
+			 *
+			 * Use `char *buf` as an interlude to trick the CI
+			 * static analysis to let us use `strbuf_addstr()`
+			 * here (and only copy the token) rather than
+			 * `strbuf_addbuf()`.
+			 */
+			buf = query_result.buf;
+			strbuf_addstr(&last_update_token, buf);
+			bol = last_update_token.len + 1;
+			is_trivial = query_result.buf[bol] == '/';
+			if (is_trivial)
+				trace2_data_intmax("fsm_client", NULL,
+						   "query/trivial-response", 1);
+		} else {
+			/*
+			 * The builtin daemon is not available on this
+			 * platform -OR- we failed to get a response.
+			 *
+			 * Generate a fake token (rather than a V1
+			 * timestamp) for the index extension.  (If
+			 * they switch back to the hook API, we don't
+			 * want ambiguous state.)
+			 */
+			strbuf_addstr(&last_update_token, "builtin:fake");
+		}
+
+		goto apply_results;
 	}
 
 	assert(fsm_mode == FSMONITOR_MODE_HOOK);
@@ -315,6 +348,7 @@ void refresh_fsmonitor(struct index_state *istate)
 				 query_success ? "success" : "failure");
 	}
 
+apply_results:
 	/*
 	 * The response from FSMonitor (excluding the header token) is
 	 * either:
-- 
gitgitgadget

[PATCH v5 05/30] fsmonitor: document builtin fsmonitor

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Document how `core.fsmonitor` can be set to a boolean to enable
or disable the builtin FSMonitor.

Update references to `core.fsmonitor` and `core.fsmonitorHookVersion` and
pointers to `Watchman` to refer to it.

Create `git-fsmonitor--daemon` manual page and describe its features.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 Documentation/config/core.txt           | 46 ++++++++++-----
 Documentation/git-fsmonitor--daemon.txt | 75 +++++++++++++++++++++++++
 Documentation/git-update-index.txt      |  8 ++-
 3 files changed, 112 insertions(+), 17 deletions(-)
 create mode 100644 Documentation/git-fsmonitor--daemon.txt

diff --git a/Documentation/config/core.txt b/Documentation/config/core.txt
index c04f62a54a1..86059dc7b88 100644
--- a/Documentation/config/core.txt
+++ b/Documentation/config/core.txt
@@ -62,22 +62,40 @@ core.protectNTFS::
 	Defaults to `true` on Windows, and `false` elsewhere.
 
 core.fsmonitor::
-	If set, the value of this variable is used as a command which
-	will identify all files that may have changed since the
-	requested date/time. This information is used to speed up git by
-	avoiding unnecessary processing of files that have not changed.
-	See the "fsmonitor-watchman" section of linkgit:githooks[5].
+	If set to true, enable the built-in file system monitor
+	daemon for this working directory (linkgit:git-fsmonitor--daemon[1]).
++
+Like hook-based file system monitors, the built-in file system monitor
+can speed up Git commands that need to refresh the Git index
+(e.g. `git status`) in a working directory with many files.  The
+built-in monitor eliminates the need to install and maintain an
+external third-party tool.
++
+The built-in file system monitor is currently available only on a
+limited set of supported platforms.  Currently, this includes Windows
+and MacOS.
++
+	Otherwise, this variable contains the pathname of the "fsmonitor"
+	hook command.
++
+This hook command is used to identify all files that may have changed
+since the requested date/time. This information is used to speed up
+git by avoiding unnecessary scanning of files that have not changed.
++
+See the "fsmonitor-watchman" section of linkgit:githooks[5].
 
 core.fsmonitorHookVersion::
-	Sets the version of hook that is to be used when calling fsmonitor.
-	There are currently versions 1 and 2. When this is not set,
-	version 2 will be tried first and if it fails then version 1
-	will be tried. Version 1 uses a timestamp as input to determine
-	which files have changes since that time but some monitors
-	like watchman have race conditions when used with a timestamp.
-	Version 2 uses an opaque string so that the monitor can return
-	something that can be used to determine what files have changed
-	without race conditions.
+	Sets the protocol version to be used when invoking the
+	"fsmonitor" hook.
++
+There are currently versions 1 and 2. When this is not set,
+version 2 will be tried first and if it fails then version 1
+will be tried. Version 1 uses a timestamp as input to determine
+which files have changes since that time but some monitors
+like Watchman have race conditions when used with a timestamp.
+Version 2 uses an opaque string so that the monitor can return
+something that can be used to determine what files have changed
+without race conditions.
 
 core.trustctime::
 	If false, the ctime differences between the index and the
diff --git a/Documentation/git-fsmonitor--daemon.txt b/Documentation/git-fsmonitor--daemon.txt
new file mode 100644
index 00000000000..0fedf5a4565
--- /dev/null
+++ b/Documentation/git-fsmonitor--daemon.txt
@@ -0,0 +1,75 @@
+git-fsmonitor--daemon(1)
+========================
+
+NAME
+----
+git-fsmonitor--daemon - A Built-in File System Monitor
+
+SYNOPSIS
+--------
+[verse]
+'git fsmonitor--daemon' start
+'git fsmonitor--daemon' run
+'git fsmonitor--daemon' stop
+'git fsmonitor--daemon' status
+
+DESCRIPTION
+-----------
+
+A daemon to watch the working directory for file and directory
+changes using platform-specific file system notification facilities.
+
+This daemon communicates directly with commands like `git status`
+using the link:technical/api-simple-ipc.html[simple IPC] interface
+instead of the slower linkgit:githooks[5] interface.
+
+This daemon is built into Git so that no third-party tools are
+required.
+
+OPTIONS
+-------
+
+start::
+	Starts a daemon in the background.
+
+run::
+	Runs a daemon in the foreground.
+
+stop::
+	Stops the daemon running in the current working
+	directory, if present.
+
+status::
+	Exits with zero status if a daemon is watching the
+	current working directory.
+
+REMARKS
+-------
+
+This daemon is a long running process used to watch a single working
+directory and maintain a list of the recently changed files and
+directories.  Performance of commands such as `git status` can be
+increased if they just ask for a summary of changes to the working
+directory and can avoid scanning the disk.
+
+When `core.fsmonitor` is set to `true` (see linkgit:git-config[1])
+commands, such as `git status`, will ask the daemon for changes and
+automatically start it (if necessary).
+
+For more information see the "File System Monitor" section in
+linkgit:git-update-index[1].
+
+CAVEATS
+-------
+
+The fsmonitor daemon does not currently know about submodules and does
+not know to filter out file system events that happen within a
+submodule.  If fsmonitor daemon is watching a super repo and a file is
+modified within the working directory of a submodule, it will report
+the change (as happening against the super repo).  However, the client
+will properly ignore these extra events, so performance may be affected
+but it will not cause an incorrect result.
+
+GIT
+---
+Part of the linkgit:git[1] suite
diff --git a/Documentation/git-update-index.txt b/Documentation/git-update-index.txt
index 2853f168d97..53ea48a04e2 100644
--- a/Documentation/git-update-index.txt
+++ b/Documentation/git-update-index.txt
@@ -498,7 +498,9 @@ FILE SYSTEM MONITOR
 This feature is intended to speed up git operations for repos that have
 large working directories.
 
-It enables git to work together with a file system monitor (see the
+It enables git to work together with a file system monitor (see
+linkgit:git-fsmonitor--daemon[1]
+and the
 "fsmonitor-watchman" section of linkgit:githooks[5]) that can
 inform it as to what files have been modified. This enables git to avoid
 having to lstat() every file to find modified files.
@@ -509,8 +511,8 @@ looking for new files.
 
 If you want to enable (or disable) this feature, it is easier to use
 the `core.fsmonitor` configuration variable (see
-linkgit:git-config[1]) than using the `--fsmonitor` option to
-`git update-index` in each repository, especially if you want to do so
+linkgit:git-config[1]) than using the `--fsmonitor` option to `git
+update-index` in each repository, especially if you want to do so
 across all repositories you use, because you can set the configuration
 variable in your `$HOME/.gitconfig` just once and have it affect all
 repositories you touch.
-- 
gitgitgadget

[PATCH v5 06/30] fsmonitor--daemon: add a built-in fsmonitor daemon

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Create a built-in file system monitoring daemon that can be used by
the existing `fsmonitor` feature (protocol API and index extension)
to improve the performance of various Git commands, such as `status`.

The `fsmonitor--daemon` feature builds upon the `Simple IPC` API and
provides an alternative to hook access to existing fsmonitors such
as `watchman`.

This commit merely adds the new command without any functionality.

Co-authored-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 .gitignore                  |  1 +
 Makefile                    |  1 +
 builtin.h                   |  1 +
 builtin/fsmonitor--daemon.c | 46 +++++++++++++++++++++++++++++++++++++
 git.c                       |  1 +
 5 files changed, 50 insertions(+)
 create mode 100644 builtin/fsmonitor--daemon.c

diff --git a/.gitignore b/.gitignore
index f817c509ec0..e81de1063a4 100644
--- a/.gitignore
+++ b/.gitignore
@@ -72,6 +72,7 @@
 /git-format-patch
 /git-fsck
 /git-fsck-objects
+/git-fsmonitor--daemon
 /git-gc
 /git-get-tar-commit-id
 /git-grep
diff --git a/Makefile b/Makefile
index 9943f0f7c11..3b7a3f88b50 100644
--- a/Makefile
+++ b/Makefile
@@ -1106,6 +1106,7 @@ BUILTIN_OBJS += builtin/fmt-merge-msg.o
 BUILTIN_OBJS += builtin/for-each-ref.o
 BUILTIN_OBJS += builtin/for-each-repo.o
 BUILTIN_OBJS += builtin/fsck.o
+BUILTIN_OBJS += builtin/fsmonitor--daemon.o
 BUILTIN_OBJS += builtin/gc.o
 BUILTIN_OBJS += builtin/get-tar-commit-id.o
 BUILTIN_OBJS += builtin/grep.o
diff --git a/builtin.h b/builtin.h
index 83379f3832c..40e9ecc8485 100644
--- a/builtin.h
+++ b/builtin.h
@@ -159,6 +159,7 @@ int cmd_for_each_ref(int argc, const char **argv, const char *prefix);
 int cmd_for_each_repo(int argc, const char **argv, const char *prefix);
 int cmd_format_patch(int argc, const char **argv, const char *prefix);
 int cmd_fsck(int argc, const char **argv, const char *prefix);
+int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix);
 int cmd_gc(int argc, const char **argv, const char *prefix);
 int cmd_get_tar_commit_id(int argc, const char **argv, const char *prefix);
 int cmd_grep(int argc, const char **argv, const char *prefix);
diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
new file mode 100644
index 00000000000..f0498793379
--- /dev/null
+++ b/builtin/fsmonitor--daemon.c
@@ -0,0 +1,46 @@
+#include "builtin.h"
+#include "config.h"
+#include "parse-options.h"
+#include "fsmonitor.h"
+#include "fsmonitor-ipc.h"
+#include "simple-ipc.h"
+#include "khash.h"
+
+static const char * const builtin_fsmonitor__daemon_usage[] = {
+	NULL
+};
+
+#ifdef HAVE_FSMONITOR_DAEMON_BACKEND
+
+int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
+{
+	const char *subcmd;
+
+	struct option options[] = {
+		OPT_END()
+	};
+
+	git_config(git_default_config, NULL);
+
+	argc = parse_options(argc, argv, prefix, options,
+			     builtin_fsmonitor__daemon_usage, 0);
+	if (argc != 1)
+		usage_with_options(builtin_fsmonitor__daemon_usage, options);
+	subcmd = argv[0];
+
+	die(_("Unhandled subcommand '%s'"), subcmd);
+}
+
+#else
+int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
+{
+	struct option options[] = {
+		OPT_END()
+	};
+
+	if (argc == 2 && !strcmp(argv[1], "-h"))
+		usage_with_options(builtin_fsmonitor__daemon_usage, options);
+
+	die(_("fsmonitor--daemon not supported on this platform"));
+}
+#endif
diff --git a/git.c b/git.c
index 340665d4a04..a8b44d9b587 100644
--- a/git.c
+++ b/git.c
@@ -536,6 +536,7 @@ static struct cmd_struct commands[] = {
 	{ "format-patch", cmd_format_patch, RUN_SETUP },
 	{ "fsck", cmd_fsck, RUN_SETUP },
 	{ "fsck-objects", cmd_fsck, RUN_SETUP },
+	{ "fsmonitor--daemon", cmd_fsmonitor__daemon, RUN_SETUP },
 	{ "gc", cmd_gc, RUN_SETUP },
 	{ "get-tar-commit-id", cmd_get_tar_commit_id, NO_PARSEOPT },
 	{ "grep", cmd_grep, RUN_SETUP_GENTLY },
-- 
gitgitgadget

[PATCH v5 07/30] fsmonitor--daemon: implement 'stop' and 'status' commands

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Implement `stop` and `status` client commands to control and query the
status of a `fsmonitor--daemon` server process (and implicitly start a
server process if necessary).

Later commits will implement the actual server and monitor the file
system.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 builtin/fsmonitor--daemon.c | 51 +++++++++++++++++++++++++++++++++++++
 1 file changed, 51 insertions(+)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index f0498793379..5e3178b8bdd 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -7,10 +7,55 @@
 #include "khash.h"
 
 static const char * const builtin_fsmonitor__daemon_usage[] = {
+	N_("git fsmonitor--daemon stop"),
+	N_("git fsmonitor--daemon status"),
 	NULL
 };
 
 #ifdef HAVE_FSMONITOR_DAEMON_BACKEND
+/*
+ * Acting as a CLIENT.
+ *
+ * Send a "quit" command to the `git-fsmonitor--daemon` (if running)
+ * and wait for it to shutdown.
+ */
+static int do_as_client__send_stop(void)
+{
+	struct strbuf answer = STRBUF_INIT;
+	int ret;
+
+	ret = fsmonitor_ipc__send_command("quit", &answer);
+
+	/* The quit command does not return any response data. */
+	strbuf_release(&answer);
+
+	if (ret)
+		return ret;
+
+	trace2_region_enter("fsm_client", "polling-for-daemon-exit", NULL);
+	while (fsmonitor_ipc__get_state() == IPC_STATE__LISTENING)
+		sleep_millisec(50);
+	trace2_region_leave("fsm_client", "polling-for-daemon-exit", NULL);
+
+	return 0;
+}
+
+static int do_as_client__status(void)
+{
+	enum ipc_active_state state = fsmonitor_ipc__get_state();
+
+	switch (state) {
+	case IPC_STATE__LISTENING:
+		printf(_("fsmonitor-daemon is watching '%s'\n"),
+		       the_repository->worktree);
+		return 0;
+
+	default:
+		printf(_("fsmonitor-daemon is not watching '%s'\n"),
+		       the_repository->worktree);
+		return 1;
+	}
+}
 
 int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
 {
@@ -28,6 +73,12 @@ int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
 		usage_with_options(builtin_fsmonitor__daemon_usage, options);
 	subcmd = argv[0];
 
+	if (!strcmp(subcmd, "stop"))
+		return !!do_as_client__send_stop();
+
+	if (!strcmp(subcmd, "status"))
+		return !!do_as_client__status();
+
 	die(_("Unhandled subcommand '%s'"), subcmd);
 }
 
-- 
gitgitgadget

[PATCH v5 08/30] compat/fsmonitor/fsm-listen-win32: stub in backend for Windows

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Stub in empty filesystem listener backend for fsmonitor--daemon on Windows.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 Makefile                            | 13 ++++++++
 compat/fsmonitor/fsm-listen-win32.c | 21 +++++++++++++
 compat/fsmonitor/fsm-listen.h       | 49 +++++++++++++++++++++++++++++
 config.mak.uname                    | 10 ++++++
 contrib/buildsystems/CMakeLists.txt |  7 +++++
 repo-settings.c                     |  1 +
 6 files changed, 101 insertions(+)
 create mode 100644 compat/fsmonitor/fsm-listen-win32.c
 create mode 100644 compat/fsmonitor/fsm-listen.h

diff --git a/Makefile b/Makefile
index 3b7a3f88b50..b98f4899ac0 100644
--- a/Makefile
+++ b/Makefile
@@ -464,6 +464,11 @@ all::
 # directory, and the JSON compilation database 'compile_commands.json' will be
 # created at the root of the repository.
 #
+# If your platform supports a built-in fsmonitor backend, set
+# FSMONITOR_DAEMON_BACKEND to the "<name>" of the corresponding
+# `compat/fsmonitor/fsm-listen-<name>.c` that implements the
+# `fsm_listen__*()` routines.
+#
 # Define DEVELOPER to enable more compiler warnings. Compiler version
 # and family are auto detected, but could be overridden by defining
 # COMPILER_FEATURES (see config.mak.dev). You can still set
@@ -1939,6 +1944,11 @@ ifdef NEED_ACCESS_ROOT_HANDLER
 	COMPAT_OBJS += compat/access.o
 endif
 
+ifdef FSMONITOR_DAEMON_BACKEND
+	COMPAT_CFLAGS += -DHAVE_FSMONITOR_DAEMON_BACKEND
+	COMPAT_OBJS += compat/fsmonitor/fsm-listen-$(FSMONITOR_DAEMON_BACKEND).o
+endif
+
 ifeq ($(TCLTK_PATH),)
 NO_TCLTK = NoThanks
 endif
@@ -2858,6 +2868,9 @@ GIT-BUILD-OPTIONS: FORCE
 	@echo DC_SHA1=\''$(subst ','\'',$(subst ','\'',$(DC_SHA1)))'\' >>$@+
 	@echo SANITIZE_LEAK=\''$(subst ','\'',$(subst ','\'',$(SANITIZE_LEAK)))'\' >>$@+
 	@echo X=\'$(X)\' >>$@+
+ifdef FSMONITOR_DAEMON_BACKEND
+	@echo FSMONITOR_DAEMON_BACKEND=\''$(subst ','\'',$(subst ','\'',$(FSMONITOR_DAEMON_BACKEND)))'\' >>$@+
+endif
 ifdef TEST_OUTPUT_DIRECTORY
 	@echo TEST_OUTPUT_DIRECTORY=\''$(subst ','\'',$(subst ','\'',$(TEST_OUTPUT_DIRECTORY)))'\' >>$@+
 endif
diff --git a/compat/fsmonitor/fsm-listen-win32.c b/compat/fsmonitor/fsm-listen-win32.c
new file mode 100644
index 00000000000..916cbea254f
--- /dev/null
+++ b/compat/fsmonitor/fsm-listen-win32.c
@@ -0,0 +1,21 @@
+#include "cache.h"
+#include "config.h"
+#include "fsmonitor.h"
+#include "fsm-listen.h"
+
+void fsm_listen__stop_async(struct fsmonitor_daemon_state *state)
+{
+}
+
+void fsm_listen__loop(struct fsmonitor_daemon_state *state)
+{
+}
+
+int fsm_listen__ctor(struct fsmonitor_daemon_state *state)
+{
+	return -1;
+}
+
+void fsm_listen__dtor(struct fsmonitor_daemon_state *state)
+{
+}
diff --git a/compat/fsmonitor/fsm-listen.h b/compat/fsmonitor/fsm-listen.h
new file mode 100644
index 00000000000..f0539349baf
--- /dev/null
+++ b/compat/fsmonitor/fsm-listen.h
@@ -0,0 +1,49 @@
+#ifndef FSM_LISTEN_H
+#define FSM_LISTEN_H
+
+/* This needs to be implemented by each backend */
+
+#ifdef HAVE_FSMONITOR_DAEMON_BACKEND
+
+struct fsmonitor_daemon_state;
+
+/*
+ * Initialize platform-specific data for the fsmonitor listener thread.
+ * This will be called from the main thread PRIOR to staring the
+ * fsmonitor_fs_listener thread.
+ *
+ * Returns 0 if successful.
+ * Returns -1 otherwise.
+ */
+int fsm_listen__ctor(struct fsmonitor_daemon_state *state);
+
+/*
+ * Cleanup platform-specific data for the fsmonitor listener thread.
+ * This will be called from the main thread AFTER joining the listener.
+ */
+void fsm_listen__dtor(struct fsmonitor_daemon_state *state);
+
+/*
+ * The main body of the platform-specific event loop to watch for
+ * filesystem events.  This will run in the fsmonitor_fs_listen thread.
+ *
+ * It should call `ipc_server_stop_async()` if the listener thread
+ * prematurely terminates (because of a filesystem error or if it
+ * detects that the .git directory has been deleted).  (It should NOT
+ * do so if the listener thread receives a normal shutdown signal from
+ * the IPC layer.)
+ *
+ * It should set `state->error_code` to -1 if the daemon should exit
+ * with an error.
+ */
+void fsm_listen__loop(struct fsmonitor_daemon_state *state);
+
+/*
+ * Gently request that the fsmonitor listener thread shutdown.
+ * It does not wait for it to stop.  The caller should do a JOIN
+ * to wait for it.
+ */
+void fsm_listen__stop_async(struct fsmonitor_daemon_state *state);
+
+#endif /* HAVE_FSMONITOR_DAEMON_BACKEND */
+#endif /* FSM_LISTEN_H */
diff --git a/config.mak.uname b/config.mak.uname
index c48db45106c..9c88e1b244e 100644
--- a/config.mak.uname
+++ b/config.mak.uname
@@ -434,6 +434,11 @@ ifeq ($(uname_S),Windows)
 	# so we don't need this:
 	#
 	#   SNPRINTF_RETURNS_BOGUS = YesPlease
+
+	# The builtin FSMonitor requires Named Pipes and Threads on Windows.
+	# These are always available, so we do not have to conditionally
+	# support it.
+	FSMONITOR_DAEMON_BACKEND = win32
 	NO_SVN_TESTS = YesPlease
 	RUNTIME_PREFIX = YesPlease
 	HAVE_WPGMPTR = YesWeDo
@@ -619,6 +624,11 @@ ifeq ($(uname_S),MINGW)
 	NO_STRTOUMAX = YesPlease
 	NO_MKDTEMP = YesPlease
 	NO_SVN_TESTS = YesPlease
+
+	# The builtin FSMonitor requires Named Pipes and Threads on Windows.
+	# These are always available, so we do not have to conditionally
+	# support it.
+	FSMONITOR_DAEMON_BACKEND = win32
 	RUNTIME_PREFIX = YesPlease
 	HAVE_WPGMPTR = YesWeDo
 	NO_ST_BLOCKS_IN_STRUCT_STAT = YesPlease
diff --git a/contrib/buildsystems/CMakeLists.txt b/contrib/buildsystems/CMakeLists.txt
index 5100f56bb37..1525a81109c 100644
--- a/contrib/buildsystems/CMakeLists.txt
+++ b/contrib/buildsystems/CMakeLists.txt
@@ -285,6 +285,13 @@ else()
 	endif()
 endif()
 
+if(SUPPORTS_SIMPLE_IPC)
+	if(CMAKE_SYSTEM_NAME STREQUAL "Windows")
+		add_compile_definitions(HAVE_FSMONITOR_DAEMON_BACKEND)
+		list(APPEND compat_SOURCES compat/fsmonitor/fsm-listen-win32.c)
+	endif()
+endif()
+
 set(EXE_EXTENSION ${CMAKE_EXECUTABLE_SUFFIX})
 
 #header checks
diff --git a/repo-settings.c b/repo-settings.c
index 00ca5571a1a..373a341ad2b 100644
--- a/repo-settings.c
+++ b/repo-settings.c
@@ -2,6 +2,7 @@
 #include "config.h"
 #include "repository.h"
 #include "midx.h"
+#include "compat/fsmonitor/fsm-listen.h"
 
 static void repo_cfg_bool(struct repository *r, const char *key, int *dest,
 			  int def)
-- 
gitgitgadget

[PATCH v5 09/30] compat/fsmonitor/fsm-listen-darwin: stub in backend for Darwin

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Stub in empty implementation of fsmonitor--daemon
backend for Darwin (aka MacOS).

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 compat/fsmonitor/fsm-listen-darwin.c | 20 ++++++++++++++++++++
 config.mak.uname                     | 10 ++++++++++
 contrib/buildsystems/CMakeLists.txt  |  3 +++
 3 files changed, 33 insertions(+)
 create mode 100644 compat/fsmonitor/fsm-listen-darwin.c

diff --git a/compat/fsmonitor/fsm-listen-darwin.c b/compat/fsmonitor/fsm-listen-darwin.c
new file mode 100644
index 00000000000..c84e3344ab9
--- /dev/null
+++ b/compat/fsmonitor/fsm-listen-darwin.c
@@ -0,0 +1,20 @@
+#include "cache.h"
+#include "fsmonitor.h"
+#include "fsm-listen.h"
+
+int fsm_listen__ctor(struct fsmonitor_daemon_state *state)
+{
+	return -1;
+}
+
+void fsm_listen__dtor(struct fsmonitor_daemon_state *state)
+{
+}
+
+void fsm_listen__stop_async(struct fsmonitor_daemon_state *state)
+{
+}
+
+void fsm_listen__loop(struct fsmonitor_daemon_state *state)
+{
+}
diff --git a/config.mak.uname b/config.mak.uname
index 9c88e1b244e..d65a5eb807c 100644
--- a/config.mak.uname
+++ b/config.mak.uname
@@ -157,6 +157,16 @@ ifeq ($(uname_S),Darwin)
 			MSGFMT = /usr/local/opt/gettext/bin/msgfmt
 		endif
 	endif
+
+	# The builtin FSMonitor on MacOS builds upon Simple-IPC.  Both require
+	# Unix domain sockets and PThreads.
+	ifndef NO_PTHREADS
+	ifndef NO_UNIX_SOCKETS
+	FSMONITOR_DAEMON_BACKEND = darwin
+	endif
+	endif
+
+	BASIC_LDFLAGS += -framework CoreServices
 endif
 ifeq ($(uname_S),SunOS)
 	NEEDS_SOCKET = YesPlease
diff --git a/contrib/buildsystems/CMakeLists.txt b/contrib/buildsystems/CMakeLists.txt
index 1525a81109c..7da31c38f48 100644
--- a/contrib/buildsystems/CMakeLists.txt
+++ b/contrib/buildsystems/CMakeLists.txt
@@ -289,6 +289,9 @@ if(SUPPORTS_SIMPLE_IPC)
 	if(CMAKE_SYSTEM_NAME STREQUAL "Windows")
 		add_compile_definitions(HAVE_FSMONITOR_DAEMON_BACKEND)
 		list(APPEND compat_SOURCES compat/fsmonitor/fsm-listen-win32.c)
+	elseif(CMAKE_SYSTEM_NAME STREQUAL "Darwin")
+		add_compile_definitions(HAVE_FSMONITOR_DAEMON_BACKEND)
+		list(APPEND compat_SOURCES compat/fsmonitor/fsm-listen-darwin.c)
 	endif()
 endif()
 
-- 
gitgitgadget

[PATCH v5 10/30] fsmonitor--daemon: implement 'run' command

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Implement `run` command to try to begin listening for file system events.

This version defines the thread structure with a single fsmonitor_fs_listen
thread to watch for file system events and a simple IPC thread pool to
watch for connection from Git clients over a well-known named pipe or
Unix domain socket.

This commit does not actually do anything yet because the platform
backends are still just stubs.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 builtin/fsmonitor--daemon.c | 213 +++++++++++++++++++++++++++++++++++-
 fsmonitor--daemon.h         |  34 ++++++
 2 files changed, 246 insertions(+), 1 deletion(-)
 create mode 100644 fsmonitor--daemon.h

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index 5e3178b8bdd..b5ebd1eca33 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -3,16 +3,39 @@
 #include "parse-options.h"
 #include "fsmonitor.h"
 #include "fsmonitor-ipc.h"
+#include "compat/fsmonitor/fsm-listen.h"
+#include "fsmonitor--daemon.h"
 #include "simple-ipc.h"
 #include "khash.h"
 
 static const char * const builtin_fsmonitor__daemon_usage[] = {
+	N_("git fsmonitor--daemon run [<options>]"),
 	N_("git fsmonitor--daemon stop"),
 	N_("git fsmonitor--daemon status"),
 	NULL
 };
 
 #ifdef HAVE_FSMONITOR_DAEMON_BACKEND
+/*
+ * Global state loaded from config.
+ */
+#define FSMONITOR__IPC_THREADS "fsmonitor.ipcthreads"
+static int fsmonitor__ipc_threads = 8;
+
+static int fsmonitor_config(const char *var, const char *value, void *cb)
+{
+	if (!strcmp(var, FSMONITOR__IPC_THREADS)) {
+		int i = git_config_int(var, value);
+		if (i < 1)
+			return error(_("value of '%s' out of range: %d"),
+				     FSMONITOR__IPC_THREADS, i);
+		fsmonitor__ipc_threads = i;
+		return 0;
+	}
+
+	return git_default_config(var, value, cb);
+}
+
 /*
  * Acting as a CLIENT.
  *
@@ -57,15 +80,196 @@ static int do_as_client__status(void)
 	}
 }
 
+static ipc_server_application_cb handle_client;
+
+static int handle_client(void *data,
+			 const char *command, size_t command_len,
+			 ipc_server_reply_cb *reply,
+			 struct ipc_server_reply_data *reply_data)
+{
+	/* struct fsmonitor_daemon_state *state = data; */
+	int result;
+
+	/*
+	 * The Simple IPC API now supports {char*, len} arguments, but
+	 * FSMonitor always uses proper null-terminated strings, so
+	 * we can ignore the command_len argument.  (Trust, but verify.)
+	 */
+	if (command_len != strlen(command))
+		BUG("FSMonitor assumes text messages");
+
+	trace2_region_enter("fsmonitor", "handle_client", the_repository);
+	trace2_data_string("fsmonitor", the_repository, "request", command);
+
+	result = 0; /* TODO Do something here. */
+
+	trace2_region_leave("fsmonitor", "handle_client", the_repository);
+
+	return result;
+}
+
+static void *fsm_listen__thread_proc(void *_state)
+{
+	struct fsmonitor_daemon_state *state = _state;
+
+	trace2_thread_start("fsm-listen");
+
+	trace_printf_key(&trace_fsmonitor, "Watching: worktree '%s'",
+			 state->path_worktree_watch.buf);
+	if (state->nr_paths_watching > 1)
+		trace_printf_key(&trace_fsmonitor, "Watching: gitdir '%s'",
+				 state->path_gitdir_watch.buf);
+
+	fsm_listen__loop(state);
+
+	trace2_thread_exit();
+	return NULL;
+}
+
+static int fsmonitor_run_daemon_1(struct fsmonitor_daemon_state *state)
+{
+	struct ipc_server_opts ipc_opts = {
+		.nr_threads = fsmonitor__ipc_threads,
+
+		/*
+		 * We know that there are no other active threads yet,
+		 * so we can let the IPC layer temporarily chdir() if
+		 * it needs to when creating the server side of the
+		 * Unix domain socket.
+		 */
+		.uds_disallow_chdir = 0
+	};
+
+	/*
+	 * Start the IPC thread pool before the we've started the file
+	 * system event listener thread so that we have the IPC handle
+	 * before we need it.
+	 */
+	if (ipc_server_run_async(&state->ipc_server_data,
+				 fsmonitor_ipc__get_path(), &ipc_opts,
+				 handle_client, state))
+		return error_errno(
+			_("could not start IPC thread pool on '%s'"),
+			fsmonitor_ipc__get_path());
+
+	/*
+	 * Start the fsmonitor listener thread to collect filesystem
+	 * events.
+	 */
+	if (pthread_create(&state->listener_thread, NULL,
+			   fsm_listen__thread_proc, state) < 0) {
+		ipc_server_stop_async(state->ipc_server_data);
+		ipc_server_await(state->ipc_server_data);
+
+		return error(_("could not start fsmonitor listener thread"));
+	}
+
+	/*
+	 * The daemon is now fully functional in background threads.
+	 * Wait for the IPC thread pool to shutdown (whether by client
+	 * request or from filesystem activity).
+	 */
+	ipc_server_await(state->ipc_server_data);
+
+	/*
+	 * The fsmonitor listener thread may have received a shutdown
+	 * event from the IPC thread pool, but it doesn't hurt to tell
+	 * it again.  And wait for it to shutdown.
+	 */
+	fsm_listen__stop_async(state);
+	pthread_join(state->listener_thread, NULL);
+
+	return state->error_code;
+}
+
+static int fsmonitor_run_daemon(void)
+{
+	struct fsmonitor_daemon_state state;
+	int err;
+
+	memset(&state, 0, sizeof(state));
+
+	pthread_mutex_init(&state.main_lock, NULL);
+	state.error_code = 0;
+	state.current_token_data = NULL;
+
+	/* Prepare to (recursively) watch the <worktree-root> directory. */
+	strbuf_init(&state.path_worktree_watch, 0);
+	strbuf_addstr(&state.path_worktree_watch, absolute_path(get_git_work_tree()));
+	state.nr_paths_watching = 1;
+
+	/*
+	 * We create and delete cookie files somewhere inside the .git
+	 * directory to help us keep sync with the file system.  If
+	 * ".git" is not a directory, then <gitdir> is not inside the
+	 * cone of <worktree-root>, so set up a second watch to watch
+	 * the <gitdir> so that we get events for the cookie files.
+	 */
+	strbuf_init(&state.path_gitdir_watch, 0);
+	strbuf_addbuf(&state.path_gitdir_watch, &state.path_worktree_watch);
+	strbuf_addstr(&state.path_gitdir_watch, "/.git");
+	if (!is_directory(state.path_gitdir_watch.buf)) {
+		strbuf_reset(&state.path_gitdir_watch);
+		strbuf_addstr(&state.path_gitdir_watch, absolute_path(get_git_dir()));
+		state.nr_paths_watching = 2;
+	}
+
+	/*
+	 * Confirm that we can create platform-specific resources for the
+	 * filesystem listener before we bother starting all the threads.
+	 */
+	if (fsm_listen__ctor(&state)) {
+		err = error(_("could not initialize listener thread"));
+		goto done;
+	}
+
+	err = fsmonitor_run_daemon_1(&state);
+
+done:
+	pthread_mutex_destroy(&state.main_lock);
+	fsm_listen__dtor(&state);
+
+	ipc_server_free(state.ipc_server_data);
+
+	strbuf_release(&state.path_worktree_watch);
+	strbuf_release(&state.path_gitdir_watch);
+
+	return err;
+}
+
+static int try_to_run_foreground_daemon(void)
+{
+	/*
+	 * Technically, we don't need to probe for an existing daemon
+	 * process, since we could just call `fsmonitor_run_daemon()`
+	 * and let it fail if the pipe/socket is busy.
+	 *
+	 * However, this method gives us a nicer error message for a
+	 * common error case.
+	 */
+	if (fsmonitor_ipc__get_state() == IPC_STATE__LISTENING)
+		die("fsmonitor--daemon is already running '%s'",
+		    the_repository->worktree);
+
+	printf(_("running fsmonitor-daemon in '%s'\n"),
+	       the_repository->worktree);
+	fflush(stdout);
+
+	return !!fsmonitor_run_daemon();
+}
+
 int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
 {
 	const char *subcmd;
 
 	struct option options[] = {
+		OPT_INTEGER(0, "ipc-threads",
+			    &fsmonitor__ipc_threads,
+			    N_("use <n> ipc worker threads")),
 		OPT_END()
 	};
 
-	git_config(git_default_config, NULL);
+	git_config(fsmonitor_config, NULL);
 
 	argc = parse_options(argc, argv, prefix, options,
 			     builtin_fsmonitor__daemon_usage, 0);
@@ -73,6 +277,13 @@ int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
 		usage_with_options(builtin_fsmonitor__daemon_usage, options);
 	subcmd = argv[0];
 
+	if (fsmonitor__ipc_threads < 1)
+		die(_("invalid 'ipc-threads' value (%d)"),
+		    fsmonitor__ipc_threads);
+
+	if (!strcmp(subcmd, "run"))
+		return !!try_to_run_foreground_daemon();
+
 	if (!strcmp(subcmd, "stop"))
 		return !!do_as_client__send_stop();
 
diff --git a/fsmonitor--daemon.h b/fsmonitor--daemon.h
new file mode 100644
index 00000000000..3009c1a83de
--- /dev/null
+++ b/fsmonitor--daemon.h
@@ -0,0 +1,34 @@
+#ifndef FSMONITOR_DAEMON_H
+#define FSMONITOR_DAEMON_H
+
+#ifdef HAVE_FSMONITOR_DAEMON_BACKEND
+
+#include "cache.h"
+#include "dir.h"
+#include "run-command.h"
+#include "simple-ipc.h"
+#include "thread-utils.h"
+
+struct fsmonitor_batch;
+struct fsmonitor_token_data;
+
+struct fsmonitor_daemon_backend_data; /* opaque platform-specific data */
+
+struct fsmonitor_daemon_state {
+	pthread_t listener_thread;
+	pthread_mutex_t main_lock;
+
+	struct strbuf path_worktree_watch;
+	struct strbuf path_gitdir_watch;
+	int nr_paths_watching;
+
+	struct fsmonitor_token_data *current_token_data;
+
+	int error_code;
+	struct fsmonitor_daemon_backend_data *backend_data;
+
+	struct ipc_server_data *ipc_server_data;
+};
+
+#endif /* HAVE_FSMONITOR_DAEMON_BACKEND */
+#endif /* FSMONITOR_DAEMON_H */
-- 
gitgitgadget

[PATCH v5 11/30] fsmonitor--daemon: implement 'start' command

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Implement 'git fsmonitor--daemon start' command.  This command starts
an instance of 'git fsmonitor--daemon run' in the background using
the new 'start_bg_command()' function.

We avoid the fork-and-call technique on Unix systems in favor of a
fork-and-exec technique.  This gives us more uniform Trace2 child-*
events.  It also makes our usage more consistent with Windows usage.

On Windows, teach 'git fsmonitor--daemon run' to optionally call
'FreeConsole()' to release handles to the inherited Win32 console
(despite being passed invalid handles for stdin/out/err).  Without
this, command prompts and powershell terminal windows could hang
in "exit" until the last background child process exited or released
their Win32 console handle.  (This was not seen with git-bash shells
because they don't have a Win32 console attached to them.)

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 builtin/fsmonitor--daemon.c | 107 +++++++++++++++++++++++++++++++++++-
 1 file changed, 105 insertions(+), 2 deletions(-)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index b5ebd1eca33..8988440b7c1 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -9,6 +9,7 @@
 #include "khash.h"
 
 static const char * const builtin_fsmonitor__daemon_usage[] = {
+	N_("git fsmonitor--daemon start [<options>]"),
 	N_("git fsmonitor--daemon run [<options>]"),
 	N_("git fsmonitor--daemon stop"),
 	N_("git fsmonitor--daemon status"),
@@ -22,6 +23,9 @@ static const char * const builtin_fsmonitor__daemon_usage[] = {
 #define FSMONITOR__IPC_THREADS "fsmonitor.ipcthreads"
 static int fsmonitor__ipc_threads = 8;
 
+#define FSMONITOR__START_TIMEOUT "fsmonitor.starttimeout"
+static int fsmonitor__start_timeout_sec = 60;
+
 static int fsmonitor_config(const char *var, const char *value, void *cb)
 {
 	if (!strcmp(var, FSMONITOR__IPC_THREADS)) {
@@ -33,6 +37,15 @@ static int fsmonitor_config(const char *var, const char *value, void *cb)
 		return 0;
 	}
 
+	if (!strcmp(var, FSMONITOR__START_TIMEOUT)) {
+		int i = git_config_int(var, value);
+		if (i < 0)
+			return error(_("value of '%s' out of range: %d"),
+				     FSMONITOR__START_TIMEOUT, i);
+		fsmonitor__start_timeout_sec = i;
+		return 0;
+	}
+
 	return git_default_config(var, value, cb);
 }
 
@@ -237,7 +250,7 @@ done:
 	return err;
 }
 
-static int try_to_run_foreground_daemon(void)
+static int try_to_run_foreground_daemon(int free_console)
 {
 	/*
 	 * Technically, we don't need to probe for an existing daemon
@@ -255,17 +268,104 @@ static int try_to_run_foreground_daemon(void)
 	       the_repository->worktree);
 	fflush(stdout);
 
+#ifdef GIT_WINDOWS_NATIVE
+	if (free_console)
+		FreeConsole();
+#endif
+
 	return !!fsmonitor_run_daemon();
 }
 
+static start_bg_wait_cb bg_wait_cb;
+
+static int bg_wait_cb(const struct child_process *cp, void *cb_data)
+{
+	enum ipc_active_state s = fsmonitor_ipc__get_state();
+
+	switch (s) {
+	case IPC_STATE__LISTENING:
+		/* child is "ready" */
+		return 0;
+
+	case IPC_STATE__NOT_LISTENING:
+	case IPC_STATE__PATH_NOT_FOUND:
+		/* give child more time */
+		return 1;
+
+	default:
+	case IPC_STATE__INVALID_PATH:
+	case IPC_STATE__OTHER_ERROR:
+		/* all the time in world won't help */
+		return -1;
+	}
+}
+
+static int try_to_start_background_daemon(void)
+{
+	struct child_process cp = CHILD_PROCESS_INIT;
+	enum start_bg_result sbgr;
+
+	/*
+	 * Before we try to create a background daemon process, see
+	 * if a daemon process is already listening.  This makes it
+	 * easier for us to report an already-listening error to the
+	 * console, since our spawn/daemon can only report the success
+	 * of creating the background process (and not whether it
+	 * immediately exited).
+	 */
+	if (fsmonitor_ipc__get_state() == IPC_STATE__LISTENING)
+		die("fsmonitor--daemon is already running '%s'",
+		    the_repository->worktree);
+
+	printf(_("starting fsmonitor-daemon in '%s'\n"),
+	       the_repository->worktree);
+	fflush(stdout);
+
+	cp.git_cmd = 1;
+
+	strvec_push(&cp.args, "fsmonitor--daemon");
+	strvec_push(&cp.args, "run");
+	strvec_push(&cp.args, "--free-console");
+	strvec_pushf(&cp.args, "--ipc-threads=%d", fsmonitor__ipc_threads);
+
+	cp.no_stdin = 1;
+	cp.no_stdout = 1;
+	cp.no_stderr = 1;
+
+	sbgr = start_bg_command(&cp, bg_wait_cb, NULL,
+				fsmonitor__start_timeout_sec);
+
+	switch (sbgr) {
+	case SBGR_READY:
+		return 0;
+
+	default:
+	case SBGR_ERROR:
+	case SBGR_CB_ERROR:
+		return error("daemon failed to start");
+
+	case SBGR_TIMEOUT:
+		return error("daemon not online yet");
+
+	case SBGR_DIED:
+		return error("daemon terminated");
+	}
+}
+
 int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
 {
 	const char *subcmd;
+	int free_console = 0;
 
 	struct option options[] = {
+		OPT_BOOL(0, "free-console", &free_console, N_("free console")),
 		OPT_INTEGER(0, "ipc-threads",
 			    &fsmonitor__ipc_threads,
 			    N_("use <n> ipc worker threads")),
+		OPT_INTEGER(0, "start-timeout",
+			    &fsmonitor__start_timeout_sec,
+			    N_("Max seconds to wait for background daemon startup")),
+
 		OPT_END()
 	};
 
@@ -281,8 +381,11 @@ int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
 		die(_("invalid 'ipc-threads' value (%d)"),
 		    fsmonitor__ipc_threads);
 
+	if (!strcmp(subcmd, "start"))
+		return !!try_to_start_background_daemon();
+
 	if (!strcmp(subcmd, "run"))
-		return !!try_to_run_foreground_daemon();
+		return !!try_to_run_foreground_daemon(free_console);
 
 	if (!strcmp(subcmd, "stop"))
 		return !!do_as_client__send_stop();
-- 
gitgitgadget

[PATCH v5 12/30] fsmonitor--daemon: add pathname classification

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach fsmonitor--daemon to classify relative and absolute
pathnames and decide how they should be handled.  This will
be used by the platform-specific backend to respond to each
filesystem event.

When we register for filesystem notifications on a directory,
we get events for everything (recursively) in the directory.
We want to report to clients changes to tracked and untracked
paths within the working directory.  We do not want to report
changes within the .git directory, for example.

This classification will be used in a later commit by the
different backends to classify paths as events are received.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 builtin/fsmonitor--daemon.c | 81 +++++++++++++++++++++++++++++++++++++
 fsmonitor--daemon.h         | 61 ++++++++++++++++++++++++++++
 2 files changed, 142 insertions(+)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index 8988440b7c1..1618f1aa27f 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -121,6 +121,87 @@ static int handle_client(void *data,
 	return result;
 }
 
+#define FSMONITOR_COOKIE_PREFIX ".fsmonitor-daemon-"
+
+enum fsmonitor_path_type fsmonitor_classify_path_workdir_relative(
+	const char *rel)
+{
+	if (fspathncmp(rel, ".git", 4))
+		return IS_WORKDIR_PATH;
+	rel += 4;
+
+	if (!*rel)
+		return IS_DOT_GIT;
+	if (*rel != '/')
+		return IS_WORKDIR_PATH; /* e.g. .gitignore */
+	rel++;
+
+	if (!fspathncmp(rel, FSMONITOR_COOKIE_PREFIX,
+			strlen(FSMONITOR_COOKIE_PREFIX)))
+		return IS_INSIDE_DOT_GIT_WITH_COOKIE_PREFIX;
+
+	return IS_INSIDE_DOT_GIT;
+}
+
+enum fsmonitor_path_type fsmonitor_classify_path_gitdir_relative(
+	const char *rel)
+{
+	if (!fspathncmp(rel, FSMONITOR_COOKIE_PREFIX,
+			strlen(FSMONITOR_COOKIE_PREFIX)))
+		return IS_INSIDE_GITDIR_WITH_COOKIE_PREFIX;
+
+	return IS_INSIDE_GITDIR;
+}
+
+static enum fsmonitor_path_type try_classify_workdir_abs_path(
+	struct fsmonitor_daemon_state *state,
+	const char *path)
+{
+	const char *rel;
+
+	if (fspathncmp(path, state->path_worktree_watch.buf,
+		       state->path_worktree_watch.len))
+		return IS_OUTSIDE_CONE;
+
+	rel = path + state->path_worktree_watch.len;
+
+	if (!*rel)
+		return IS_WORKDIR_PATH; /* it is the root dir exactly */
+	if (*rel != '/')
+		return IS_OUTSIDE_CONE;
+	rel++;
+
+	return fsmonitor_classify_path_workdir_relative(rel);
+}
+
+enum fsmonitor_path_type fsmonitor_classify_path_absolute(
+	struct fsmonitor_daemon_state *state,
+	const char *path)
+{
+	const char *rel;
+	enum fsmonitor_path_type t;
+
+	t = try_classify_workdir_abs_path(state, path);
+	if (state->nr_paths_watching == 1)
+		return t;
+	if (t != IS_OUTSIDE_CONE)
+		return t;
+
+	if (fspathncmp(path, state->path_gitdir_watch.buf,
+		       state->path_gitdir_watch.len))
+		return IS_OUTSIDE_CONE;
+
+	rel = path + state->path_gitdir_watch.len;
+
+	if (!*rel)
+		return IS_GITDIR; /* it is the <gitdir> exactly */
+	if (*rel != '/')
+		return IS_OUTSIDE_CONE;
+	rel++;
+
+	return fsmonitor_classify_path_gitdir_relative(rel);
+}
+
 static void *fsm_listen__thread_proc(void *_state)
 {
 	struct fsmonitor_daemon_state *state = _state;
diff --git a/fsmonitor--daemon.h b/fsmonitor--daemon.h
index 3009c1a83de..7bbb3a27a1c 100644
--- a/fsmonitor--daemon.h
+++ b/fsmonitor--daemon.h
@@ -30,5 +30,66 @@ struct fsmonitor_daemon_state {
 	struct ipc_server_data *ipc_server_data;
 };
 
+/*
+ * Pathname classifications.
+ *
+ * The daemon classifies the pathnames that it receives from file
+ * system notification events into the following categories and uses
+ * that to decide whether clients are told about them.  (And to watch
+ * for file system synchronization events.)
+ *
+ * The client should only care about paths within the working
+ * directory proper (inside the working directory and not ".git" nor
+ * inside of ".git/").  That is, the client has read the index and is
+ * asking for a list of any paths in the working directory that have
+ * been modified since the last token.  The client does not care about
+ * file system changes within the .git directory (such as new loose
+ * objects or packfiles).  So the client will only receive paths that
+ * are classified as IS_WORKDIR_PATH.
+ *
+ * The daemon uses the IS_DOT_GIT and IS_GITDIR internally to mean the
+ * exact ".git" directory or GITDIR.  If the daemon receives a delete
+ * event for either of these directories, it will automatically
+ * shutdown, for example.
+ *
+ * Note that the daemon DOES NOT explicitly watch nor special case the
+ * ".git/index" file.  The daemon does not read the index and does not
+ * have any internal index-relative state.  The daemon only collects
+ * the set of modified paths within the working directory.
+ */
+enum fsmonitor_path_type {
+	IS_WORKDIR_PATH = 0,
+
+	IS_DOT_GIT,
+	IS_INSIDE_DOT_GIT,
+	IS_INSIDE_DOT_GIT_WITH_COOKIE_PREFIX,
+
+	IS_GITDIR,
+	IS_INSIDE_GITDIR,
+	IS_INSIDE_GITDIR_WITH_COOKIE_PREFIX,
+
+	IS_OUTSIDE_CONE,
+};
+
+/*
+ * Classify a pathname relative to the root of the working directory.
+ */
+enum fsmonitor_path_type fsmonitor_classify_path_workdir_relative(
+	const char *relative_path);
+
+/*
+ * Classify a pathname relative to a <gitdir> that is external to the
+ * worktree directory.
+ */
+enum fsmonitor_path_type fsmonitor_classify_path_gitdir_relative(
+	const char *relative_path);
+
+/*
+ * Classify an absolute pathname received from a filesystem event.
+ */
+enum fsmonitor_path_type fsmonitor_classify_path_absolute(
+	struct fsmonitor_daemon_state *state,
+	const char *path);
+
 #endif /* HAVE_FSMONITOR_DAEMON_BACKEND */
 #endif /* FSMONITOR_DAEMON_H */
-- 
gitgitgadget

[PATCH v5 13/30] fsmonitor--daemon: define token-ids

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach fsmonitor--daemon to create token-ids and define the
overall token naming scheme.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 builtin/fsmonitor--daemon.c | 116 +++++++++++++++++++++++++++++++++++-
 1 file changed, 115 insertions(+), 1 deletion(-)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index 1618f1aa27f..2997c2cba8c 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -93,6 +93,120 @@ static int do_as_client__status(void)
 	}
 }
 
+/*
+ * Requests to and from a FSMonitor Protocol V2 provider use an opaque
+ * "token" as a virtual timestamp.  Clients can request a summary of all
+ * created/deleted/modified files relative to a token.  In the response,
+ * clients receive a new token for the next (relative) request.
+ *
+ *
+ * Token Format
+ * ============
+ *
+ * The contents of the token are private and provider-specific.
+ *
+ * For the built-in fsmonitor--daemon, we define a token as follows:
+ *
+ *     "builtin" ":" <token_id> ":" <sequence_nr>
+ *
+ * The "builtin" prefix is used as a namespace to avoid conflicts
+ * with other providers (such as Watchman).
+ *
+ * The <token_id> is an arbitrary OPAQUE string, such as a GUID,
+ * UUID, or {timestamp,pid}.  It is used to group all filesystem
+ * events that happened while the daemon was monitoring (and in-sync
+ * with the filesystem).
+ *
+ *     Unlike FSMonitor Protocol V1, it is not defined as a timestamp
+ *     and does not define less-than/greater-than relationships.
+ *     (There are too many race conditions to rely on file system
+ *     event timestamps.)
+ *
+ * The <sequence_nr> is a simple integer incremented whenever the
+ * daemon needs to make its state public.  For example, if 1000 file
+ * system events come in, but no clients have requested the data,
+ * the daemon can continue to accumulate file changes in the same
+ * bin and does not need to advance the sequence number.  However,
+ * as soon as a client does arrive, the daemon needs to start a new
+ * bin and increment the sequence number.
+ *
+ *     The sequence number serves as the boundary between 2 sets
+ *     of bins -- the older ones that the client has already seen
+ *     and the newer ones that it hasn't.
+ *
+ * When a new <token_id> is created, the <sequence_nr> is reset to
+ * zero.
+ *
+ *
+ * About Token Ids
+ * ===============
+ *
+ * A new token_id is created:
+ *
+ * [1] each time the daemon is started.
+ *
+ * [2] any time that the daemon must re-sync with the filesystem
+ *     (such as when the kernel drops or we miss events on a very
+ *     active volume).
+ *
+ * [3] in response to a client "flush" command (for dropped event
+ *     testing).
+ *
+ * When a new token_id is created, the daemon is free to discard all
+ * cached filesystem events associated with any previous token_ids.
+ * Events associated with a non-current token_id will never be sent
+ * to a client.  A token_id change implicitly means that the daemon
+ * has gap in its event history.
+ *
+ * Therefore, clients that present a token with a stale (non-current)
+ * token_id will always be given a trivial response.
+ */
+struct fsmonitor_token_data {
+	struct strbuf token_id;
+	struct fsmonitor_batch *batch_head;
+	struct fsmonitor_batch *batch_tail;
+	uint64_t client_ref_count;
+};
+
+static struct fsmonitor_token_data *fsmonitor_new_token_data(void)
+{
+	static int test_env_value = -1;
+	static uint64_t flush_count = 0;
+	struct fsmonitor_token_data *token;
+
+	CALLOC_ARRAY(token, 1);
+
+	strbuf_init(&token->token_id, 0);
+	token->batch_head = NULL;
+	token->batch_tail = NULL;
+	token->client_ref_count = 0;
+
+	if (test_env_value < 0)
+		test_env_value = git_env_bool("GIT_TEST_FSMONITOR_TOKEN", 0);
+
+	if (!test_env_value) {
+		struct timeval tv;
+		struct tm tm;
+		time_t secs;
+
+		gettimeofday(&tv, NULL);
+		secs = tv.tv_sec;
+		gmtime_r(&secs, &tm);
+
+		strbuf_addf(&token->token_id,
+			    "%"PRIu64".%d.%4d%02d%02dT%02d%02d%02d.%06ldZ",
+			    flush_count++,
+			    getpid(),
+			    tm.tm_year + 1900, tm.tm_mon + 1, tm.tm_mday,
+			    tm.tm_hour, tm.tm_min, tm.tm_sec,
+			    (long)tv.tv_usec);
+	} else {
+		strbuf_addf(&token->token_id, "test_%08x", test_env_value++);
+	}
+
+	return token;
+}
+
 static ipc_server_application_cb handle_client;
 
 static int handle_client(void *data,
@@ -285,7 +399,7 @@ static int fsmonitor_run_daemon(void)
 
 	pthread_mutex_init(&state.main_lock, NULL);
 	state.error_code = 0;
-	state.current_token_data = NULL;
+	state.current_token_data = fsmonitor_new_token_data();
 
 	/* Prepare to (recursively) watch the <worktree-root> directory. */
 	strbuf_init(&state.path_worktree_watch, 0);
-- 
gitgitgadget

[PATCH v5 14/30] fsmonitor--daemon: create token-based changed path cache

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach fsmonitor--daemon to build a list of changed paths and associate
them with a token-id.  This will be used by the platform-specific
backends to accumulate changed paths in response to filesystem events.

The platform-specific file system listener thread receives file system
events containing one or more changed pathnames (with whatever
bucketing or grouping that is convenient for the file system).  These
paths are accumulated (without locking) by the file system layer into
a `fsmonitor_batch`.

When the file system layer has drained the kernel event queue, it will
"publish" them to our token queue and make them visible to concurrent
client worker threads.  The token layer is free to combine and/or de-dup
paths within these batches for efficient presentation to clients.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 builtin/fsmonitor--daemon.c | 230 +++++++++++++++++++++++++++++++++++-
 fsmonitor--daemon.h         |  40 +++++++
 2 files changed, 268 insertions(+), 2 deletions(-)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index 2997c2cba8c..34603f23053 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -168,17 +168,27 @@ struct fsmonitor_token_data {
 	uint64_t client_ref_count;
 };
 
+struct fsmonitor_batch {
+	struct fsmonitor_batch *next;
+	uint64_t batch_seq_nr;
+	const char **interned_paths;
+	size_t nr, alloc;
+	time_t pinned_time;
+};
+
 static struct fsmonitor_token_data *fsmonitor_new_token_data(void)
 {
 	static int test_env_value = -1;
 	static uint64_t flush_count = 0;
 	struct fsmonitor_token_data *token;
+	struct fsmonitor_batch *batch;
 
 	CALLOC_ARRAY(token, 1);
+	batch = fsmonitor_batch__new();
 
 	strbuf_init(&token->token_id, 0);
-	token->batch_head = NULL;
-	token->batch_tail = NULL;
+	token->batch_head = batch;
+	token->batch_tail = batch;
 	token->client_ref_count = 0;
 
 	if (test_env_value < 0)
@@ -204,9 +214,143 @@ static struct fsmonitor_token_data *fsmonitor_new_token_data(void)
 		strbuf_addf(&token->token_id, "test_%08x", test_env_value++);
 	}
 
+	/*
+	 * We created a new <token_id> and are starting a new series
+	 * of tokens with a zero <seq_nr>.
+	 *
+	 * Since clients cannot guess our new (non test) <token_id>
+	 * they will always receive a trivial response (because of the
+	 * mismatch on the <token_id>).  The trivial response will
+	 * tell them our new <token_id> so that subsequent requests
+	 * will be relative to our new series.  (And when sending that
+	 * response, we pin the current head of the batch list.)
+	 *
+	 * Even if the client correctly guesses the <token_id>, their
+	 * request of "builtin:<token_id>:0" asks for all changes MORE
+	 * RECENT than batch/bin 0.
+	 *
+	 * This implies that it is a waste to accumulate paths in the
+	 * initial batch/bin (because they will never be transmitted).
+	 *
+	 * So the daemon could be running for days and watching the
+	 * file system, but doesn't need to actually accumulate any
+	 * paths UNTIL we need to set a reference point for a later
+	 * relative request.
+	 *
+	 * However, it is very useful for testing to always have a
+	 * reference point set.  Pin batch 0 to force early file system
+	 * events to accumulate.
+	 */
+	if (test_env_value)
+		batch->pinned_time = time(NULL);
+
 	return token;
 }
 
+struct fsmonitor_batch *fsmonitor_batch__new(void)
+{
+	struct fsmonitor_batch *batch;
+
+	CALLOC_ARRAY(batch, 1);
+
+	return batch;
+}
+
+void fsmonitor_batch__free_list(struct fsmonitor_batch *batch)
+{
+	while (batch) {
+		struct fsmonitor_batch *next = batch->next;
+
+		/*
+		 * The actual strings within the array of this batch
+		 * are interned, so we don't own them.  We only own
+		 * the array.
+		 */
+		free(batch->interned_paths);
+		free(batch);
+
+		batch = next;
+	}
+}
+
+void fsmonitor_batch__add_path(struct fsmonitor_batch *batch,
+			       const char *path)
+{
+	const char *interned_path = strintern(path);
+
+	trace_printf_key(&trace_fsmonitor, "event: %s", interned_path);
+
+	ALLOC_GROW(batch->interned_paths, batch->nr + 1, batch->alloc);
+	batch->interned_paths[batch->nr++] = interned_path;
+}
+
+static void fsmonitor_batch__combine(struct fsmonitor_batch *batch_dest,
+				     const struct fsmonitor_batch *batch_src)
+{
+	size_t k;
+
+	ALLOC_GROW(batch_dest->interned_paths,
+		   batch_dest->nr + batch_src->nr + 1,
+		   batch_dest->alloc);
+
+	for (k = 0; k < batch_src->nr; k++)
+		batch_dest->interned_paths[batch_dest->nr++] =
+			batch_src->interned_paths[k];
+}
+
+static void fsmonitor_free_token_data(struct fsmonitor_token_data *token)
+{
+	if (!token)
+		return;
+
+	assert(token->client_ref_count == 0);
+
+	strbuf_release(&token->token_id);
+
+	fsmonitor_batch__free_list(token->batch_head);
+
+	free(token);
+}
+
+/*
+ * Flush all of our cached data about the filesystem.  Call this if we
+ * lose sync with the filesystem and miss some notification events.
+ *
+ * [1] If we are missing events, then we no longer have a complete
+ *     history of the directory (relative to our current start token).
+ *     We should create a new token and start fresh (as if we just
+ *     booted up).
+ *
+ * If there are no concurrent threads readering the current token data
+ * series, we can free it now.  Otherwise, let the last reader free
+ * it.
+ *
+ * Either way, the old token data series is no longer associated with
+ * our state data.
+ */
+static void with_lock__do_force_resync(struct fsmonitor_daemon_state *state)
+{
+	/* assert current thread holding state->main_lock */
+
+	struct fsmonitor_token_data *free_me = NULL;
+	struct fsmonitor_token_data *new_one = NULL;
+
+	new_one = fsmonitor_new_token_data();
+
+	if (state->current_token_data->client_ref_count == 0)
+		free_me = state->current_token_data;
+	state->current_token_data = new_one;
+
+	fsmonitor_free_token_data(free_me);
+}
+
+void fsmonitor_force_resync(struct fsmonitor_daemon_state *state)
+{
+	pthread_mutex_lock(&state->main_lock);
+	with_lock__do_force_resync(state);
+	pthread_mutex_unlock(&state->main_lock);
+}
+
 static ipc_server_application_cb handle_client;
 
 static int handle_client(void *data,
@@ -316,6 +460,81 @@ enum fsmonitor_path_type fsmonitor_classify_path_absolute(
 	return fsmonitor_classify_path_gitdir_relative(rel);
 }
 
+/*
+ * We try to combine small batches at the front of the batch-list to avoid
+ * having a long list.  This hopefully makes it a little easier when we want
+ * to truncate and maintain the list.  However, we don't want the paths array
+ * to just keep growing and growing with realloc, so we insert an arbitrary
+ * limit.
+ */
+#define MY_COMBINE_LIMIT (1024)
+
+void fsmonitor_publish(struct fsmonitor_daemon_state *state,
+		       struct fsmonitor_batch *batch,
+		       const struct string_list *cookie_names)
+{
+	if (!batch && !cookie_names->nr)
+		return;
+
+	pthread_mutex_lock(&state->main_lock);
+
+	if (batch) {
+		struct fsmonitor_batch *head;
+
+		head = state->current_token_data->batch_head;
+		if (!head) {
+			BUG("token does not have batch");
+		} else if (head->pinned_time) {
+			/*
+			 * We cannot alter the current batch list
+			 * because:
+			 *
+			 * [a] it is being transmitted to at least one
+			 * client and the handle_client() thread has a
+			 * ref-count, but not a lock on the batch list
+			 * starting with this item.
+			 *
+			 * [b] it has been transmitted in the past to
+			 * at least one client such that future
+			 * requests are relative to this head batch.
+			 *
+			 * So, we can only prepend a new batch onto
+			 * the front of the list.
+			 */
+			batch->batch_seq_nr = head->batch_seq_nr + 1;
+			batch->next = head;
+			state->current_token_data->batch_head = batch;
+		} else if (!head->batch_seq_nr) {
+			/*
+			 * Batch 0 is unpinned.  See the note in
+			 * `fsmonitor_new_token_data()` about why we
+			 * don't need to accumulate these paths.
+			 */
+			fsmonitor_batch__free_list(batch);
+		} else if (head->nr + batch->nr > MY_COMBINE_LIMIT) {
+			/*
+			 * The head batch in the list has never been
+			 * transmitted to a client, but folding the
+			 * contents of the new batch onto it would
+			 * exceed our arbitrary limit, so just prepend
+			 * the new batch onto the list.
+			 */
+			batch->batch_seq_nr = head->batch_seq_nr + 1;
+			batch->next = head;
+			state->current_token_data->batch_head = batch;
+		} else {
+			/*
+			 * We are free to add the paths in the given
+			 * batch onto the end of the current head batch.
+			 */
+			fsmonitor_batch__combine(head, batch);
+			fsmonitor_batch__free_list(batch);
+		}
+	}
+
+	pthread_mutex_unlock(&state->main_lock);
+}
+
 static void *fsm_listen__thread_proc(void *_state)
 {
 	struct fsmonitor_daemon_state *state = _state;
@@ -330,6 +549,13 @@ static void *fsm_listen__thread_proc(void *_state)
 
 	fsm_listen__loop(state);
 
+	pthread_mutex_lock(&state->main_lock);
+	if (state->current_token_data &&
+	    state->current_token_data->client_ref_count == 0)
+		fsmonitor_free_token_data(state->current_token_data);
+	state->current_token_data = NULL;
+	pthread_mutex_unlock(&state->main_lock);
+
 	trace2_thread_exit();
 	return NULL;
 }
diff --git a/fsmonitor--daemon.h b/fsmonitor--daemon.h
index 7bbb3a27a1c..20a815d80f8 100644
--- a/fsmonitor--daemon.h
+++ b/fsmonitor--daemon.h
@@ -12,6 +12,27 @@
 struct fsmonitor_batch;
 struct fsmonitor_token_data;
 
+/*
+ * Create a new batch of path(s).  The returned batch is considered
+ * private and not linked into the fsmonitor daemon state.  The caller
+ * should fill this batch with one or more paths and then publish it.
+ */
+struct fsmonitor_batch *fsmonitor_batch__new(void);
+
+/*
+ * Free the list of batches starting with this one.
+ */
+void fsmonitor_batch__free_list(struct fsmonitor_batch *batch);
+
+/*
+ * Add this path to this batch of modified files.
+ *
+ * The batch should be private and NOT (yet) linked into the fsmonitor
+ * daemon state and therefore not yet visible to worker threads and so
+ * no locking is required.
+ */
+void fsmonitor_batch__add_path(struct fsmonitor_batch *batch, const char *path);
+
 struct fsmonitor_daemon_backend_data; /* opaque platform-specific data */
 
 struct fsmonitor_daemon_state {
@@ -91,5 +112,24 @@ enum fsmonitor_path_type fsmonitor_classify_path_absolute(
 	struct fsmonitor_daemon_state *state,
 	const char *path);
 
+/*
+ * Prepend the this batch of path(s) onto the list of batches associated
+ * with the current token.  This makes the batch visible to worker threads.
+ *
+ * The caller no longer owns the batch and must not free it.
+ *
+ * Wake up the client threads waiting on these cookies.
+ */
+void fsmonitor_publish(struct fsmonitor_daemon_state *state,
+		       struct fsmonitor_batch *batch,
+		       const struct string_list *cookie_names);
+
+/*
+ * If the platform-specific layer loses sync with the filesystem,
+ * it should call this to invalidate cached data and abort waiting
+ * threads.
+ */
+void fsmonitor_force_resync(struct fsmonitor_daemon_state *state);
+
 #endif /* HAVE_FSMONITOR_DAEMON_BACKEND */
 #endif /* FSMONITOR_DAEMON_H */
-- 
gitgitgadget

[PATCH v5 15/30] compat/fsmonitor/fsm-listen-win32: implement FSMonitor backend on Windows

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach the win32 backend to register a watch on the working tree
root directory (recursively).  Also watch the <gitdir> if it is
not inside the working tree.  And to collect path change notifications
into batches and publish.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 compat/fsmonitor/fsm-listen-win32.c | 565 ++++++++++++++++++++++++++++
 1 file changed, 565 insertions(+)

diff --git a/compat/fsmonitor/fsm-listen-win32.c b/compat/fsmonitor/fsm-listen-win32.c
index 916cbea254f..c2d11acbc1e 100644
--- a/compat/fsmonitor/fsm-listen-win32.c
+++ b/compat/fsmonitor/fsm-listen-win32.c
@@ -2,20 +2,585 @@
 #include "config.h"
 #include "fsmonitor.h"
 #include "fsm-listen.h"
+#include "fsmonitor--daemon.h"
+
+/*
+ * The documentation of ReadDirectoryChangesW() states that the maximum
+ * buffer size is 64K when the monitored directory is remote.
+ *
+ * Larger buffers may be used when the monitored directory is local and
+ * will help us receive events faster from the kernel and avoid dropped
+ * events.
+ *
+ * So we try to use a very large buffer and silently fallback to 64K if
+ * we get an error.
+ */
+#define MAX_RDCW_BUF_FALLBACK (65536)
+#define MAX_RDCW_BUF          (65536 * 8)
+
+struct one_watch
+{
+	char buffer[MAX_RDCW_BUF];
+	DWORD buf_len;
+	DWORD count;
+
+	struct strbuf path;
+	HANDLE hDir;
+	HANDLE hEvent;
+	OVERLAPPED overlapped;
+
+	/*
+	 * Is there an active ReadDirectoryChangesW() call pending.  If so, we
+	 * need to later call GetOverlappedResult() and possibly CancelIoEx().
+	 */
+	BOOL is_active;
+};
+
+struct fsmonitor_daemon_backend_data
+{
+	struct one_watch *watch_worktree;
+	struct one_watch *watch_gitdir;
+
+	HANDLE hEventShutdown;
+
+	HANDLE hListener[3]; /* we don't own these handles */
+#define LISTENER_SHUTDOWN 0
+#define LISTENER_HAVE_DATA_WORKTREE 1
+#define LISTENER_HAVE_DATA_GITDIR 2
+	int nr_listener_handles;
+};
+
+/*
+ * Convert the WCHAR path from the notification into UTF8 and
+ * then normalize it.
+ */
+static int normalize_path_in_utf8(FILE_NOTIFY_INFORMATION *info,
+				  struct strbuf *normalized_path)
+{
+	int reserve;
+	int len = 0;
+
+	strbuf_reset(normalized_path);
+	if (!info->FileNameLength)
+		goto normalize;
+
+	/*
+	 * Pre-reserve enough space in the UTF8 buffer for
+	 * each Unicode WCHAR character to be mapped into a
+	 * sequence of 2 UTF8 characters.  That should let us
+	 * avoid ERROR_INSUFFICIENT_BUFFER 99.9+% of the time.
+	 */
+	reserve = info->FileNameLength + 1;
+	strbuf_grow(normalized_path, reserve);
+
+	for (;;) {
+		len = WideCharToMultiByte(CP_UTF8, 0, info->FileName,
+					  info->FileNameLength / sizeof(WCHAR),
+					  normalized_path->buf,
+					  strbuf_avail(normalized_path) - 1,
+					  NULL, NULL);
+		if (len > 0)
+			goto normalize;
+		if (GetLastError() != ERROR_INSUFFICIENT_BUFFER) {
+			error("[GLE %ld] could not convert path to UTF-8: '%.*ls'",
+			      GetLastError(),
+			      (int)(info->FileNameLength / sizeof(WCHAR)),
+			      info->FileName);
+			return -1;
+		}
+
+		strbuf_grow(normalized_path,
+			    strbuf_avail(normalized_path) + reserve);
+	}
+
+normalize:
+	strbuf_setlen(normalized_path, len);
+	return strbuf_normalize_path(normalized_path);
+}
 
 void fsm_listen__stop_async(struct fsmonitor_daemon_state *state)
 {
+	SetEvent(state->backend_data->hListener[LISTENER_SHUTDOWN]);
+}
+
+static struct one_watch *create_watch(struct fsmonitor_daemon_state *state,
+				      const char *path)
+{
+	struct one_watch *watch = NULL;
+	DWORD desired_access = FILE_LIST_DIRECTORY;
+	DWORD share_mode =
+		FILE_SHARE_WRITE | FILE_SHARE_READ | FILE_SHARE_DELETE;
+	HANDLE hDir;
+	wchar_t wpath[MAX_PATH];
+
+	if (xutftowcs_path(wpath, path) < 0) {
+		error(_("could not convert to wide characters: '%s'"), path);
+		return NULL;
+	}
+
+	hDir = CreateFileW(wpath,
+			   desired_access, share_mode, NULL, OPEN_EXISTING,
+			   FILE_FLAG_BACKUP_SEMANTICS | FILE_FLAG_OVERLAPPED,
+			   NULL);
+	if (hDir == INVALID_HANDLE_VALUE) {
+		error(_("[GLE %ld] could not watch '%s'"),
+		      GetLastError(), path);
+		return NULL;
+	}
+
+	CALLOC_ARRAY(watch, 1);
+
+	watch->buf_len = sizeof(watch->buffer); /* assume full MAX_RDCW_BUF */
+
+	strbuf_init(&watch->path, 0);
+	strbuf_addstr(&watch->path, path);
+
+	watch->hDir = hDir;
+	watch->hEvent = CreateEvent(NULL, TRUE, FALSE, NULL);
+
+	return watch;
+}
+
+static void destroy_watch(struct one_watch *watch)
+{
+	if (!watch)
+		return;
+
+	strbuf_release(&watch->path);
+	if (watch->hDir != INVALID_HANDLE_VALUE)
+		CloseHandle(watch->hDir);
+	if (watch->hEvent != INVALID_HANDLE_VALUE)
+		CloseHandle(watch->hEvent);
+
+	free(watch);
+}
+
+static int start_rdcw_watch(struct fsmonitor_daemon_backend_data *data,
+			    struct one_watch *watch)
+{
+	DWORD dwNotifyFilter =
+		FILE_NOTIFY_CHANGE_FILE_NAME |
+		FILE_NOTIFY_CHANGE_DIR_NAME |
+		FILE_NOTIFY_CHANGE_ATTRIBUTES |
+		FILE_NOTIFY_CHANGE_SIZE |
+		FILE_NOTIFY_CHANGE_LAST_WRITE |
+		FILE_NOTIFY_CHANGE_CREATION;
+
+	ResetEvent(watch->hEvent);
+
+	memset(&watch->overlapped, 0, sizeof(watch->overlapped));
+	watch->overlapped.hEvent = watch->hEvent;
+
+	/*
+	 * Queue an async call using Overlapped IO.  This returns immediately.
+	 * Our event handle will be signalled when the real result is available.
+	 *
+	 * The return value here just means that we successfully queued it.
+	 * We won't know if the Read...() actually produces data until later.
+	 */
+	watch->is_active = ReadDirectoryChangesW(
+		watch->hDir, watch->buffer, watch->buf_len, TRUE,
+		dwNotifyFilter, &watch->count, &watch->overlapped, NULL);
+
+	if (watch->is_active)
+		return 0;
+
+	error("ReadDirectoryChangedW failed on '%s' [GLE %ld]",
+	      watch->path.buf, GetLastError());
+	return -1;
+}
+
+static int recv_rdcw_watch(struct one_watch *watch)
+{
+	DWORD gle;
+
+	watch->is_active = FALSE;
+
+	/*
+	 * The overlapped result is ready.  If the Read...() was successful
+	 * we finally receive the actual result into our buffer.
+	 */
+	if (GetOverlappedResult(watch->hDir, &watch->overlapped, &watch->count,
+				TRUE))
+		return 0;
+
+	gle = GetLastError();
+	if (gle == ERROR_INVALID_PARAMETER &&
+	    /*
+	     * The kernel throws an invalid parameter error when our
+	     * buffer is too big and we are pointed at a remote
+	     * directory (and possibly for other reasons).  Quietly
+	     * set it down and try again.
+	     *
+	     * See note about MAX_RDCW_BUF at the top.
+	     */
+	    watch->buf_len > MAX_RDCW_BUF_FALLBACK) {
+		watch->buf_len = MAX_RDCW_BUF_FALLBACK;
+		return -2;
+	}
+
+	/*
+	 * NEEDSWORK: If an external <gitdir> is deleted, the above
+	 * returns an error.  I'm not sure that there's anything that
+	 * we can do here other than failing -- the <worktree>/.git
+	 * link file would be broken anyway.  We might try to check
+	 * for that and return a better error message, but I'm not
+	 * sure it is worth it.
+	 */
+
+	error("GetOverlappedResult failed on '%s' [GLE %ld]",
+	      watch->path.buf, gle);
+	return -1;
+}
+
+static void cancel_rdcw_watch(struct one_watch *watch)
+{
+	DWORD count;
+
+	if (!watch || !watch->is_active)
+		return;
+
+	/*
+	 * The calls to ReadDirectoryChangesW() and GetOverlappedResult()
+	 * form a "pair" (my term) where we queue an IO and promise to
+	 * hang around and wait for the kernel to give us the result.
+	 *
+	 * If for some reason after we queue the IO, we have to quit
+	 * or otherwise not stick around for the second half, we must
+	 * tell the kernel to abort the IO.  This prevents the kernel
+	 * from writing to our buffer and/or signalling our event
+	 * after we free them.
+	 *
+	 * (Ask me how much fun it was to track that one down).
+	 */
+	CancelIoEx(watch->hDir, &watch->overlapped);
+	GetOverlappedResult(watch->hDir, &watch->overlapped, &count, TRUE);
+	watch->is_active = FALSE;
+}
+
+/*
+ * Process filesystem events that happen anywhere (recursively) under the
+ * <worktree> root directory.  For a normal working directory, this includes
+ * both version controlled files and the contents of the .git/ directory.
+ *
+ * If <worktree>/.git is a file, then we only see events for the file
+ * itself.
+ */
+static int process_worktree_events(struct fsmonitor_daemon_state *state)
+{
+	struct fsmonitor_daemon_backend_data *data = state->backend_data;
+	struct one_watch *watch = data->watch_worktree;
+	struct strbuf path = STRBUF_INIT;
+	struct string_list cookie_list = STRING_LIST_INIT_DUP;
+	struct fsmonitor_batch *batch = NULL;
+	const char *p = watch->buffer;
+
+	/*
+	 * If the kernel gets more events than will fit in the kernel
+	 * buffer associated with our RDCW handle, it drops them and
+	 * returns a count of zero.
+	 *
+	 * Yes, the call returns WITHOUT error and with length zero.
+	 * This is the documented behavior.  (My testing has confirmed
+	 * that it also sets the last error to ERROR_NOTIFY_ENUM_DIR,
+	 * but we do not rely on that since the function did not
+	 * return an error and it is not documented.)
+	 *
+	 * (The "overflow" case is not ambiguous with the "no data" case
+	 * because we did an INFINITE wait.)
+	 *
+	 * This means we have a gap in coverage.  Tell the daemon layer
+	 * to resync.
+	 */
+	if (!watch->count) {
+		trace2_data_string("fsmonitor", NULL, "fsm-listen/kernel",
+				   "overflow");
+		fsmonitor_force_resync(state);
+		return LISTENER_HAVE_DATA_WORKTREE;
+	}
+
+	/*
+	 * On Windows, `info` contains an "array" of paths that are
+	 * relative to the root of whichever directory handle received
+	 * the event.
+	 */
+	for (;;) {
+		FILE_NOTIFY_INFORMATION *info = (void *)p;
+		const char *slash;
+		enum fsmonitor_path_type t;
+
+		strbuf_reset(&path);
+		if (normalize_path_in_utf8(info, &path) == -1)
+			goto skip_this_path;
+
+		t = fsmonitor_classify_path_workdir_relative(path.buf);
+
+		switch (t) {
+		case IS_INSIDE_DOT_GIT_WITH_COOKIE_PREFIX:
+			/* special case cookie files within .git */
+
+			/* Use just the filename of the cookie file. */
+			slash = find_last_dir_sep(path.buf);
+			string_list_append(&cookie_list,
+					   slash ? slash + 1 : path.buf);
+			break;
+
+		case IS_INSIDE_DOT_GIT:
+			/* ignore everything inside of "<worktree>/.git/" */
+			break;
+
+		case IS_DOT_GIT:
+			/* "<worktree>/.git" was deleted (or renamed away) */
+			if ((info->Action == FILE_ACTION_REMOVED) ||
+			    (info->Action == FILE_ACTION_RENAMED_OLD_NAME)) {
+				trace2_data_string("fsmonitor", NULL,
+						   "fsm-listen/dotgit",
+						   "removed");
+				goto force_shutdown;
+			}
+			break;
+
+		case IS_WORKDIR_PATH:
+			/* queue normal pathname */
+			if (!batch)
+				batch = fsmonitor_batch__new();
+			fsmonitor_batch__add_path(batch, path.buf);
+			break;
+
+		case IS_GITDIR:
+		case IS_INSIDE_GITDIR:
+		case IS_INSIDE_GITDIR_WITH_COOKIE_PREFIX:
+		default:
+			BUG("unexpected path classification '%d' for '%s'",
+			    t, path.buf);
+		}
+
+skip_this_path:
+		if (!info->NextEntryOffset)
+			break;
+		p += info->NextEntryOffset;
+	}
+
+	fsmonitor_publish(state, batch, &cookie_list);
+	batch = NULL;
+	string_list_clear(&cookie_list, 0);
+	strbuf_release(&path);
+	return LISTENER_HAVE_DATA_WORKTREE;
+
+force_shutdown:
+	fsmonitor_batch__free_list(batch);
+	string_list_clear(&cookie_list, 0);
+	strbuf_release(&path);
+	return LISTENER_SHUTDOWN;
+}
+
+/*
+ * Process filesystem events that happened anywhere (recursively) under the
+ * external <gitdir> (such as non-primary worktrees or submodules).
+ * We only care about cookie files that our client threads created here.
+ *
+ * Note that we DO NOT get filesystem events on the external <gitdir>
+ * itself (it is not inside something that we are watching).  In particular,
+ * we do not get an event if the external <gitdir> is deleted.
+ */
+static int process_gitdir_events(struct fsmonitor_daemon_state *state)
+{
+	struct fsmonitor_daemon_backend_data *data = state->backend_data;
+	struct one_watch *watch = data->watch_gitdir;
+	struct strbuf path = STRBUF_INIT;
+	struct string_list cookie_list = STRING_LIST_INIT_DUP;
+	const char *p = watch->buffer;
+
+	if (!watch->count) {
+		trace2_data_string("fsmonitor", NULL, "fsm-listen/kernel",
+				   "overflow");
+		fsmonitor_force_resync(state);
+		return LISTENER_HAVE_DATA_GITDIR;
+	}
+
+	for (;;) {
+		FILE_NOTIFY_INFORMATION *info = (void *)p;
+		const char *slash;
+		enum fsmonitor_path_type t;
+
+		strbuf_reset(&path);
+		if (normalize_path_in_utf8(info, &path) == -1)
+			goto skip_this_path;
+
+		t = fsmonitor_classify_path_gitdir_relative(path.buf);
+
+		switch (t) {
+		case IS_INSIDE_GITDIR_WITH_COOKIE_PREFIX:
+			/* special case cookie files within gitdir */
+
+			/* Use just the filename of the cookie file. */
+			slash = find_last_dir_sep(path.buf);
+			string_list_append(&cookie_list,
+					   slash ? slash + 1 : path.buf);
+			break;
+
+		case IS_INSIDE_GITDIR:
+			goto skip_this_path;
+
+		default:
+			BUG("unexpected path classification '%d' for '%s'",
+			    t, path.buf);
+		}
+
+skip_this_path:
+		if (!info->NextEntryOffset)
+			break;
+		p += info->NextEntryOffset;
+	}
+
+	fsmonitor_publish(state, NULL, &cookie_list);
+	string_list_clear(&cookie_list, 0);
+	strbuf_release(&path);
+	return LISTENER_HAVE_DATA_GITDIR;
 }
 
 void fsm_listen__loop(struct fsmonitor_daemon_state *state)
 {
+	struct fsmonitor_daemon_backend_data *data = state->backend_data;
+	DWORD dwWait;
+	int result;
+
+	state->error_code = 0;
+
+	if (start_rdcw_watch(data, data->watch_worktree) == -1)
+		goto force_error_stop;
+
+	if (data->watch_gitdir &&
+	    start_rdcw_watch(data, data->watch_gitdir) == -1)
+		goto force_error_stop;
+
+	for (;;) {
+		dwWait = WaitForMultipleObjects(data->nr_listener_handles,
+						data->hListener,
+						FALSE, INFINITE);
+
+		if (dwWait == WAIT_OBJECT_0 + LISTENER_HAVE_DATA_WORKTREE) {
+			result = recv_rdcw_watch(data->watch_worktree);
+			if (result == -1) {
+				/* hard error */
+				goto force_error_stop;
+			}
+			if (result == -2) {
+				/* retryable error */
+				if (start_rdcw_watch(data, data->watch_worktree) == -1)
+					goto force_error_stop;
+				continue;
+			}
+
+			/* have data */
+			if (process_worktree_events(state) == LISTENER_SHUTDOWN)
+				goto force_shutdown;
+			if (start_rdcw_watch(data, data->watch_worktree) == -1)
+				goto force_error_stop;
+			continue;
+		}
+
+		if (dwWait == WAIT_OBJECT_0 + LISTENER_HAVE_DATA_GITDIR) {
+			result = recv_rdcw_watch(data->watch_gitdir);
+			if (result == -1) {
+				/* hard error */
+				goto force_error_stop;
+			}
+			if (result == -2) {
+				/* retryable error */
+				if (start_rdcw_watch(data, data->watch_gitdir) == -1)
+					goto force_error_stop;
+				continue;
+			}
+
+			/* have data */
+			if (process_gitdir_events(state) == LISTENER_SHUTDOWN)
+				goto force_shutdown;
+			if (start_rdcw_watch(data, data->watch_gitdir) == -1)
+				goto force_error_stop;
+			continue;
+		}
+
+		if (dwWait == WAIT_OBJECT_0 + LISTENER_SHUTDOWN)
+			goto clean_shutdown;
+
+		error(_("could not read directory changes [GLE %ld]"),
+		      GetLastError());
+		goto force_error_stop;
+	}
+
+force_error_stop:
+	state->error_code = -1;
+
+force_shutdown:
+	/*
+	 * Tell the IPC thead pool to stop (which completes the await
+	 * in the main thread (which will also signal this thread (if
+	 * we are still alive))).
+	 */
+	ipc_server_stop_async(state->ipc_server_data);
+
+clean_shutdown:
+	cancel_rdcw_watch(data->watch_worktree);
+	cancel_rdcw_watch(data->watch_gitdir);
 }
 
 int fsm_listen__ctor(struct fsmonitor_daemon_state *state)
 {
+	struct fsmonitor_daemon_backend_data *data;
+
+	CALLOC_ARRAY(data, 1);
+
+	data->hEventShutdown = CreateEvent(NULL, TRUE, FALSE, NULL);
+
+	data->watch_worktree = create_watch(state,
+					    state->path_worktree_watch.buf);
+	if (!data->watch_worktree)
+		goto failed;
+
+	if (state->nr_paths_watching > 1) {
+		data->watch_gitdir = create_watch(state,
+						  state->path_gitdir_watch.buf);
+		if (!data->watch_gitdir)
+			goto failed;
+	}
+
+	data->hListener[LISTENER_SHUTDOWN] = data->hEventShutdown;
+	data->nr_listener_handles++;
+
+	data->hListener[LISTENER_HAVE_DATA_WORKTREE] =
+		data->watch_worktree->hEvent;
+	data->nr_listener_handles++;
+
+	if (data->watch_gitdir) {
+		data->hListener[LISTENER_HAVE_DATA_GITDIR] =
+			data->watch_gitdir->hEvent;
+		data->nr_listener_handles++;
+	}
+
+	state->backend_data = data;
+	return 0;
+
+failed:
+	CloseHandle(data->hEventShutdown);
+	destroy_watch(data->watch_worktree);
+	destroy_watch(data->watch_gitdir);
+
 	return -1;
 }
 
 void fsm_listen__dtor(struct fsmonitor_daemon_state *state)
 {
+	struct fsmonitor_daemon_backend_data *data;
+
+	if (!state || !state->backend_data)
+		return;
+
+	data = state->backend_data;
+
+	CloseHandle(data->hEventShutdown);
+	destroy_watch(data->watch_worktree);
+	destroy_watch(data->watch_gitdir);
+
+	FREE_AND_NULL(state->backend_data);
 }
-- 
gitgitgadget

[PATCH v5 16/30] compat/fsmonitor/fsm-listen-darwin: add macos header files for FSEvent

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Include MacOS system declarations to allow us to use FSEvent and
CoreFoundation APIs.  We need GCC and clang versions because of
compiler and header file conflicts.

While it is quite possible to #include Apple's CoreServices.h when
compiling C source code with clang, trying to build it with GCC
currently fails with this error:

In file included
   from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
   ...Library/Frameworks/Security.framework/Headers/AuthSession.h:32,
   from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
   ...Library/Frameworks/Security.framework/Headers/Security.h:42,
   from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
   ...Library/Frameworks/CoreServices.framework/Frameworks/...
   ...OSServices.framework/Headers/CSIdentity.h:43,
   from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
   ...Library/Frameworks/CoreServices.framework/Frameworks/...
   ...OSServices.framework/Headers/OSServices.h:29,
   from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
   ...Library/Frameworks/CoreServices.framework/Frameworks/...
   ...LaunchServices.framework/Headers/IconsCore.h:23,
   from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
   ...Library/Frameworks/CoreServices.framework/Frameworks/...
   ...LaunchServices.framework/Headers/LaunchServices.h:23,
   from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
   ...Library/Frameworks/CoreServices.framework/Headers/CoreServices.h:45,

     /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
     ...Library/Frameworks/Security.framework/Headers/Authorization.h:193:7:
     error: variably modified 'bytes' at file scope
       193 | char bytes[kAuthorizationExternalFormLength];
           |      ^~~~~

The underlying reason is that GCC (rightfully) objects that an `enum`
value such as `kAuthorizationExternalFormLength` is not a constant
(because it is not, the preprocessor has no knowledge of it, only the
actual C compiler does) and can therefore not be used to define the size
of a C array.

This is a known problem and tracked in GCC's bug tracker:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93082

In the meantime, let's not block things and go the slightly ugly route
of declaring/defining the FSEvents constants, data structures and
functions that we need, so that we can avoid above-mentioned issue.

Let's do this _only_ for GCC, though, so that the CI/PR builds (which
build both with clang and with GCC) can guarantee that we _are_ using
the correct data types.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 compat/fsmonitor/fsm-listen-darwin.c | 96 ++++++++++++++++++++++++++++
 1 file changed, 96 insertions(+)

diff --git a/compat/fsmonitor/fsm-listen-darwin.c b/compat/fsmonitor/fsm-listen-darwin.c
index c84e3344ab9..f424253d3eb 100644
--- a/compat/fsmonitor/fsm-listen-darwin.c
+++ b/compat/fsmonitor/fsm-listen-darwin.c
@@ -1,3 +1,99 @@
+#if defined(__GNUC__)
+/*
+ * It is possible to #include CoreFoundation/CoreFoundation.h when compiling
+ * with clang, but not with GCC as of time of writing.
+ *
+ * See https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93082 for details.
+ */
+typedef unsigned int FSEventStreamCreateFlags;
+#define kFSEventStreamEventFlagNone               0x00000000
+#define kFSEventStreamEventFlagMustScanSubDirs    0x00000001
+#define kFSEventStreamEventFlagUserDropped        0x00000002
+#define kFSEventStreamEventFlagKernelDropped      0x00000004
+#define kFSEventStreamEventFlagEventIdsWrapped    0x00000008
+#define kFSEventStreamEventFlagHistoryDone        0x00000010
+#define kFSEventStreamEventFlagRootChanged        0x00000020
+#define kFSEventStreamEventFlagMount              0x00000040
+#define kFSEventStreamEventFlagUnmount            0x00000080
+#define kFSEventStreamEventFlagItemCreated        0x00000100
+#define kFSEventStreamEventFlagItemRemoved        0x00000200
+#define kFSEventStreamEventFlagItemInodeMetaMod   0x00000400
+#define kFSEventStreamEventFlagItemRenamed        0x00000800
+#define kFSEventStreamEventFlagItemModified       0x00001000
+#define kFSEventStreamEventFlagItemFinderInfoMod  0x00002000
+#define kFSEventStreamEventFlagItemChangeOwner    0x00004000
+#define kFSEventStreamEventFlagItemXattrMod       0x00008000
+#define kFSEventStreamEventFlagItemIsFile         0x00010000
+#define kFSEventStreamEventFlagItemIsDir          0x00020000
+#define kFSEventStreamEventFlagItemIsSymlink      0x00040000
+#define kFSEventStreamEventFlagOwnEvent           0x00080000
+#define kFSEventStreamEventFlagItemIsHardlink     0x00100000
+#define kFSEventStreamEventFlagItemIsLastHardlink 0x00200000
+#define kFSEventStreamEventFlagItemCloned         0x00400000
+
+typedef struct __FSEventStream *FSEventStreamRef;
+typedef const FSEventStreamRef ConstFSEventStreamRef;
+
+typedef unsigned int CFStringEncoding;
+#define kCFStringEncodingUTF8 0x08000100
+
+typedef const struct __CFString *CFStringRef;
+typedef const struct __CFArray *CFArrayRef;
+typedef const struct __CFRunLoop *CFRunLoopRef;
+
+struct FSEventStreamContext {
+    long long version;
+    void *cb_data, *retain, *release, *copy_description;
+};
+
+typedef struct FSEventStreamContext FSEventStreamContext;
+typedef unsigned int FSEventStreamEventFlags;
+#define kFSEventStreamCreateFlagNoDefer 0x02
+#define kFSEventStreamCreateFlagWatchRoot 0x04
+#define kFSEventStreamCreateFlagFileEvents 0x10
+
+typedef unsigned long long FSEventStreamEventId;
+#define kFSEventStreamEventIdSinceNow 0xFFFFFFFFFFFFFFFFULL
+
+typedef void (*FSEventStreamCallback)(ConstFSEventStreamRef streamRef,
+				      void *context,
+				      __SIZE_TYPE__ num_of_events,
+				      void *event_paths,
+				      const FSEventStreamEventFlags event_flags[],
+				      const FSEventStreamEventId event_ids[]);
+typedef double CFTimeInterval;
+FSEventStreamRef FSEventStreamCreate(void *allocator,
+				     FSEventStreamCallback callback,
+				     FSEventStreamContext *context,
+				     CFArrayRef paths_to_watch,
+				     FSEventStreamEventId since_when,
+				     CFTimeInterval latency,
+				     FSEventStreamCreateFlags flags);
+CFStringRef CFStringCreateWithCString(void *allocator, const char *string,
+				      CFStringEncoding encoding);
+CFArrayRef CFArrayCreate(void *allocator, const void **items, long long count,
+			 void *callbacks);
+void CFRunLoopRun(void);
+void CFRunLoopStop(CFRunLoopRef run_loop);
+CFRunLoopRef CFRunLoopGetCurrent(void);
+extern CFStringRef kCFRunLoopDefaultMode;
+void FSEventStreamScheduleWithRunLoop(FSEventStreamRef stream,
+				      CFRunLoopRef run_loop,
+				      CFStringRef run_loop_mode);
+unsigned char FSEventStreamStart(FSEventStreamRef stream);
+void FSEventStreamStop(FSEventStreamRef stream);
+void FSEventStreamInvalidate(FSEventStreamRef stream);
+void FSEventStreamRelease(FSEventStreamRef stream);
+#else
+/*
+ * Let Apple's headers declare `isalnum()` first, before
+ * Git's headers override it via a constant
+ */
+#include <string.h>
+#include <CoreFoundation/CoreFoundation.h>
+#include <CoreServices/CoreServices.h>
+#endif
+
 #include "cache.h"
 #include "fsmonitor.h"
 #include "fsm-listen.h"
-- 
gitgitgadget

[PATCH v5 17/30] compat/fsmonitor/fsm-listen-darwin: implement FSEvent listener on MacOS

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Implement file system event listener on MacOS using FSEvent,
CoreFoundation, and CoreServices.

Co-authored-by: Kevin Willford <Kevin.Willford@microsoft.com>
Co-authored-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 compat/fsmonitor/fsm-listen-darwin.c | 382 ++++++++++++++++++++++++++-
 1 file changed, 381 insertions(+), 1 deletion(-)

diff --git a/compat/fsmonitor/fsm-listen-darwin.c b/compat/fsmonitor/fsm-listen-darwin.c
index f424253d3eb..2aefdc14d89 100644
--- a/compat/fsmonitor/fsm-listen-darwin.c
+++ b/compat/fsmonitor/fsm-listen-darwin.c
@@ -1,4 +1,4 @@
-#if defined(__GNUC__)
+#ifndef __clang__
 /*
  * It is possible to #include CoreFoundation/CoreFoundation.h when compiling
  * with clang, but not with GCC as of time of writing.
@@ -97,20 +97,400 @@ void FSEventStreamRelease(FSEventStreamRef stream);
 #include "cache.h"
 #include "fsmonitor.h"
 #include "fsm-listen.h"
+#include "fsmonitor--daemon.h"
+
+struct fsmonitor_daemon_backend_data
+{
+	CFStringRef cfsr_worktree_path;
+	CFStringRef cfsr_gitdir_path;
+
+	CFArrayRef cfar_paths_to_watch;
+	int nr_paths_watching;
+
+	FSEventStreamRef stream;
+
+	CFRunLoopRef rl;
+
+	enum shutdown_style {
+		SHUTDOWN_EVENT = 0,
+		FORCE_SHUTDOWN,
+		FORCE_ERROR_STOP,
+	} shutdown_style;
+
+	unsigned int stream_scheduled:1;
+	unsigned int stream_started:1;
+};
+
+static void log_flags_set(const char *path, const FSEventStreamEventFlags flag)
+{
+	struct strbuf msg = STRBUF_INIT;
+
+	if (flag & kFSEventStreamEventFlagMustScanSubDirs)
+		strbuf_addstr(&msg, "MustScanSubDirs|");
+	if (flag & kFSEventStreamEventFlagUserDropped)
+		strbuf_addstr(&msg, "UserDropped|");
+	if (flag & kFSEventStreamEventFlagKernelDropped)
+		strbuf_addstr(&msg, "KernelDropped|");
+	if (flag & kFSEventStreamEventFlagEventIdsWrapped)
+		strbuf_addstr(&msg, "EventIdsWrapped|");
+	if (flag & kFSEventStreamEventFlagHistoryDone)
+		strbuf_addstr(&msg, "HistoryDone|");
+	if (flag & kFSEventStreamEventFlagRootChanged)
+		strbuf_addstr(&msg, "RootChanged|");
+	if (flag & kFSEventStreamEventFlagMount)
+		strbuf_addstr(&msg, "Mount|");
+	if (flag & kFSEventStreamEventFlagUnmount)
+		strbuf_addstr(&msg, "Unmount|");
+	if (flag & kFSEventStreamEventFlagItemChangeOwner)
+		strbuf_addstr(&msg, "ItemChangeOwner|");
+	if (flag & kFSEventStreamEventFlagItemCreated)
+		strbuf_addstr(&msg, "ItemCreated|");
+	if (flag & kFSEventStreamEventFlagItemFinderInfoMod)
+		strbuf_addstr(&msg, "ItemFinderInfoMod|");
+	if (flag & kFSEventStreamEventFlagItemInodeMetaMod)
+		strbuf_addstr(&msg, "ItemInodeMetaMod|");
+	if (flag & kFSEventStreamEventFlagItemIsDir)
+		strbuf_addstr(&msg, "ItemIsDir|");
+	if (flag & kFSEventStreamEventFlagItemIsFile)
+		strbuf_addstr(&msg, "ItemIsFile|");
+	if (flag & kFSEventStreamEventFlagItemIsHardlink)
+		strbuf_addstr(&msg, "ItemIsHardlink|");
+	if (flag & kFSEventStreamEventFlagItemIsLastHardlink)
+		strbuf_addstr(&msg, "ItemIsLastHardlink|");
+	if (flag & kFSEventStreamEventFlagItemIsSymlink)
+		strbuf_addstr(&msg, "ItemIsSymlink|");
+	if (flag & kFSEventStreamEventFlagItemModified)
+		strbuf_addstr(&msg, "ItemModified|");
+	if (flag & kFSEventStreamEventFlagItemRemoved)
+		strbuf_addstr(&msg, "ItemRemoved|");
+	if (flag & kFSEventStreamEventFlagItemRenamed)
+		strbuf_addstr(&msg, "ItemRenamed|");
+	if (flag & kFSEventStreamEventFlagItemXattrMod)
+		strbuf_addstr(&msg, "ItemXattrMod|");
+	if (flag & kFSEventStreamEventFlagOwnEvent)
+		strbuf_addstr(&msg, "OwnEvent|");
+	if (flag & kFSEventStreamEventFlagItemCloned)
+		strbuf_addstr(&msg, "ItemCloned|");
+
+	trace_printf_key(&trace_fsmonitor, "fsevent: '%s', flags=%u %s",
+			 path, flag, msg.buf);
+
+	strbuf_release(&msg);
+}
+
+static int ef_is_root_delete(const FSEventStreamEventFlags ef)
+{
+	return (ef & kFSEventStreamEventFlagItemIsDir &&
+		ef & kFSEventStreamEventFlagItemRemoved);
+}
+
+static int ef_is_root_renamed(const FSEventStreamEventFlags ef)
+{
+	return (ef & kFSEventStreamEventFlagItemIsDir &&
+		ef & kFSEventStreamEventFlagItemRenamed);
+}
+
+static int ef_is_dropped(const FSEventStreamEventFlags ef)
+{
+	return (ef & kFSEventStreamEventFlagMustScanSubDirs ||
+		ef & kFSEventStreamEventFlagKernelDropped ||
+		ef & kFSEventStreamEventFlagUserDropped);
+}
+
+static void fsevent_callback(ConstFSEventStreamRef streamRef,
+			     void *ctx,
+			     size_t num_of_events,
+			     void *event_paths,
+			     const FSEventStreamEventFlags event_flags[],
+			     const FSEventStreamEventId event_ids[])
+{
+	struct fsmonitor_daemon_state *state = ctx;
+	struct fsmonitor_daemon_backend_data *data = state->backend_data;
+	char **paths = (char **)event_paths;
+	struct fsmonitor_batch *batch = NULL;
+	struct string_list cookie_list = STRING_LIST_INIT_DUP;
+	const char *path_k;
+	const char *slash;
+	int k;
+	struct strbuf tmp = STRBUF_INIT;
+
+	/*
+	 * Build a list of all filesystem changes into a private/local
+	 * list and without holding any locks.
+	 */
+	for (k = 0; k < num_of_events; k++) {
+		/*
+		 * On Mac, we receive an array of absolute paths.
+		 */
+		path_k = paths[k];
+
+		/*
+		 * If you want to debug FSEvents, log them to GIT_TRACE_FSMONITOR.
+		 * Please don't log them to Trace2.
+		 *
+		 * trace_printf_key(&trace_fsmonitor, "Path: '%s'", path_k);
+		 */
+
+		/*
+		 * If event[k] is marked as dropped, we assume that we have
+		 * lost sync with the filesystem and should flush our cached
+		 * data.  We need to:
+		 *
+		 * [1] Abort/wake any client threads waiting for a cookie and
+		 *     flush the cached state data (the current token), and
+		 *     create a new token.
+		 *
+		 * [2] Discard the batch that we were locally building (since
+		 *     they are conceptually relative to the just flushed
+		 *     token).
+		 */
+		if (ef_is_dropped(event_flags[k])) {
+			if (trace_pass_fl(&trace_fsmonitor))
+				log_flags_set(path_k, event_flags[k]);
+
+			fsmonitor_force_resync(state);
+			fsmonitor_batch__free_list(batch);
+			string_list_clear(&cookie_list, 0);
+
+			/*
+			 * We assume that any events that we received
+			 * in this callback after this dropped event
+			 * may still be valid, so we continue rather
+			 * than break.  (And just in case there is a
+			 * delete of ".git" hiding in there.)
+			 */
+			continue;
+		}
+
+		switch (fsmonitor_classify_path_absolute(state, path_k)) {
+
+		case IS_INSIDE_DOT_GIT_WITH_COOKIE_PREFIX:
+		case IS_INSIDE_GITDIR_WITH_COOKIE_PREFIX:
+			/* special case cookie files within .git or gitdir */
+
+			/* Use just the filename of the cookie file. */
+			slash = find_last_dir_sep(path_k);
+			string_list_append(&cookie_list,
+					   slash ? slash + 1 : path_k);
+			break;
+
+		case IS_INSIDE_DOT_GIT:
+		case IS_INSIDE_GITDIR:
+			/* ignore all other paths inside of .git or gitdir */
+			break;
+
+		case IS_DOT_GIT:
+		case IS_GITDIR:
+			/*
+			 * If .git directory is deleted or renamed away,
+			 * we have to quit.
+			 */
+			if (ef_is_root_delete(event_flags[k])) {
+				trace_printf_key(&trace_fsmonitor,
+						 "event: gitdir removed");
+				goto force_shutdown;
+			}
+			if (ef_is_root_renamed(event_flags[k])) {
+				trace_printf_key(&trace_fsmonitor,
+						 "event: gitdir renamed");
+				goto force_shutdown;
+			}
+			break;
+
+		case IS_WORKDIR_PATH:
+			/* try to queue normal pathnames */
+
+			if (trace_pass_fl(&trace_fsmonitor))
+				log_flags_set(path_k, event_flags[k]);
+
+			/*
+			 * Because of the implicit "binning" (the
+			 * kernel calls us at a given frequency) and
+			 * de-duping (the kernel is free to combine
+			 * multiple events for a given pathname), an
+			 * individual fsevent could be marked as both
+			 * a file and directory.  Add it to the queue
+			 * with both spellings so that the client will
+			 * know how much to invalidate/refresh.
+			 */
+
+			if (event_flags[k] & kFSEventStreamEventFlagItemIsFile) {
+				const char *rel = path_k +
+					state->path_worktree_watch.len + 1;
+
+				if (!batch)
+					batch = fsmonitor_batch__new();
+				fsmonitor_batch__add_path(batch, rel);
+			}
+
+			if (event_flags[k] & kFSEventStreamEventFlagItemIsDir) {
+				const char *rel = path_k +
+					state->path_worktree_watch.len + 1;
+
+				strbuf_reset(&tmp);
+				strbuf_addstr(&tmp, rel);
+				strbuf_addch(&tmp, '/');
+
+				if (!batch)
+					batch = fsmonitor_batch__new();
+				fsmonitor_batch__add_path(batch, tmp.buf);
+			}
+
+			break;
+
+		case IS_OUTSIDE_CONE:
+		default:
+			trace_printf_key(&trace_fsmonitor,
+					 "ignoring '%s'", path_k);
+			break;
+		}
+	}
+
+	fsmonitor_publish(state, batch, &cookie_list);
+	string_list_clear(&cookie_list, 0);
+	strbuf_release(&tmp);
+	return;
+
+force_shutdown:
+	fsmonitor_batch__free_list(batch);
+	string_list_clear(&cookie_list, 0);
+
+	data->shutdown_style = FORCE_SHUTDOWN;
+	CFRunLoopStop(data->rl);
+	strbuf_release(&tmp);
+	return;
+}
+
+/*
+ * NEEDSWORK: Investigate the proper value for the `latency` argument
+ * in the call to `FSEventStreamCreate()`.  I'm not sure that this
+ * needs to be a config setting or just something that we tune after
+ * some testing.
+ *
+ * With a latency of 0.1, I was seeing lots of dropped events during
+ * the "touch 100000" files test within t/perf/p7519, but with a
+ * latency of 0.001 I did not see any dropped events.  So the
+ * "correct" value may be somewhere in between.
+ *
+ * https://developer.apple.com/documentation/coreservices/1443980-fseventstreamcreate
+ */
 
 int fsm_listen__ctor(struct fsmonitor_daemon_state *state)
 {
+	FSEventStreamCreateFlags flags = kFSEventStreamCreateFlagNoDefer |
+		kFSEventStreamCreateFlagWatchRoot |
+		kFSEventStreamCreateFlagFileEvents;
+	FSEventStreamContext ctx = {
+		0,
+		state,
+		NULL,
+		NULL,
+		NULL
+	};
+	struct fsmonitor_daemon_backend_data *data;
+	const void *dir_array[2];
+
+	CALLOC_ARRAY(data, 1);
+	state->backend_data = data;
+
+	data->cfsr_worktree_path = CFStringCreateWithCString(
+		NULL, state->path_worktree_watch.buf, kCFStringEncodingUTF8);
+	dir_array[data->nr_paths_watching++] = data->cfsr_worktree_path;
+
+	if (state->nr_paths_watching > 1) {
+		data->cfsr_gitdir_path = CFStringCreateWithCString(
+			NULL, state->path_gitdir_watch.buf,
+			kCFStringEncodingUTF8);
+		dir_array[data->nr_paths_watching++] = data->cfsr_gitdir_path;
+	}
+
+	data->cfar_paths_to_watch = CFArrayCreate(NULL, dir_array,
+						  data->nr_paths_watching,
+						  NULL);
+	data->stream = FSEventStreamCreate(NULL, fsevent_callback, &ctx,
+					   data->cfar_paths_to_watch,
+					   kFSEventStreamEventIdSinceNow,
+					   0.001, flags);
+	if (data->stream == NULL)
+		goto failed;
+
+	/*
+	 * `data->rl` needs to be set inside the listener thread.
+	 */
+
+	return 0;
+
+failed:
+	error("Unable to create FSEventStream.");
+
+	FREE_AND_NULL(state->backend_data);
 	return -1;
 }
 
 void fsm_listen__dtor(struct fsmonitor_daemon_state *state)
 {
+	struct fsmonitor_daemon_backend_data *data;
+
+	if (!state || !state->backend_data)
+		return;
+
+	data = state->backend_data;
+
+	if (data->stream) {
+		if (data->stream_started)
+			FSEventStreamStop(data->stream);
+		if (data->stream_scheduled)
+			FSEventStreamInvalidate(data->stream);
+		FSEventStreamRelease(data->stream);
+	}
+
+	FREE_AND_NULL(state->backend_data);
 }
 
 void fsm_listen__stop_async(struct fsmonitor_daemon_state *state)
 {
+	struct fsmonitor_daemon_backend_data *data;
+
+	data = state->backend_data;
+	data->shutdown_style = SHUTDOWN_EVENT;
+
+	CFRunLoopStop(data->rl);
 }
 
 void fsm_listen__loop(struct fsmonitor_daemon_state *state)
 {
+	struct fsmonitor_daemon_backend_data *data;
+
+	data = state->backend_data;
+
+	data->rl = CFRunLoopGetCurrent();
+
+	FSEventStreamScheduleWithRunLoop(data->stream, data->rl, kCFRunLoopDefaultMode);
+	data->stream_scheduled = 1;
+
+	if (!FSEventStreamStart(data->stream)) {
+		error("Failed to start the FSEventStream");
+		goto force_error_stop_without_loop;
+	}
+	data->stream_started = 1;
+
+	CFRunLoopRun();
+
+	switch (data->shutdown_style) {
+	case FORCE_ERROR_STOP:
+		state->error_code = -1;
+		/* fall thru */
+	case FORCE_SHUTDOWN:
+		ipc_server_stop_async(state->ipc_server_data);
+		/* fall thru */
+	case SHUTDOWN_EVENT:
+	default:
+		break;
+	}
+	return;
+
+force_error_stop_without_loop:
+	state->error_code = -1;
+	ipc_server_stop_async(state->ipc_server_data);
+	return;
 }
-- 
gitgitgadget

[PATCH v5 18/30] fsmonitor--daemon: implement handle_client callback

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach fsmonitor--daemon to respond to IPC requests from client
Git processes and respond with a list of modified pathnames
relative to the provided token.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 builtin/fsmonitor--daemon.c | 312 +++++++++++++++++++++++++++++++++++-
 1 file changed, 310 insertions(+), 2 deletions(-)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index 34603f23053..4c88171e06c 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -7,6 +7,7 @@
 #include "fsmonitor--daemon.h"
 #include "simple-ipc.h"
 #include "khash.h"
+#include "pkt-line.h"
 
 static const char * const builtin_fsmonitor__daemon_usage[] = {
 	N_("git fsmonitor--daemon start [<options>]"),
@@ -351,6 +352,311 @@ void fsmonitor_force_resync(struct fsmonitor_daemon_state *state)
 	pthread_mutex_unlock(&state->main_lock);
 }
 
+/*
+ * Format an opaque token string to send to the client.
+ */
+static void with_lock__format_response_token(
+	struct strbuf *response_token,
+	const struct strbuf *response_token_id,
+	const struct fsmonitor_batch *batch)
+{
+	/* assert current thread holding state->main_lock */
+
+	strbuf_reset(response_token);
+	strbuf_addf(response_token, "builtin:%s:%"PRIu64,
+		    response_token_id->buf, batch->batch_seq_nr);
+}
+
+/*
+ * Parse an opaque token from the client.
+ * Returns -1 on error.
+ */
+static int fsmonitor_parse_client_token(const char *buf_token,
+					struct strbuf *requested_token_id,
+					uint64_t *seq_nr)
+{
+	const char *p;
+	char *p_end;
+
+	strbuf_reset(requested_token_id);
+	*seq_nr = 0;
+
+	if (!skip_prefix(buf_token, "builtin:", &p))
+		return -1;
+
+	while (*p && *p != ':')
+		strbuf_addch(requested_token_id, *p++);
+	if (!*p++)
+		return -1;
+
+	*seq_nr = (uint64_t)strtoumax(p, &p_end, 10);
+	if (*p_end)
+		return -1;
+
+	return 0;
+}
+
+KHASH_INIT(str, const char *, int, 0, kh_str_hash_func, kh_str_hash_equal)
+
+static int do_handle_client(struct fsmonitor_daemon_state *state,
+			    const char *command,
+			    ipc_server_reply_cb *reply,
+			    struct ipc_server_reply_data *reply_data)
+{
+	struct fsmonitor_token_data *token_data = NULL;
+	struct strbuf response_token = STRBUF_INIT;
+	struct strbuf requested_token_id = STRBUF_INIT;
+	struct strbuf payload = STRBUF_INIT;
+	uint64_t requested_oldest_seq_nr = 0;
+	uint64_t total_response_len = 0;
+	const char *p;
+	const struct fsmonitor_batch *batch_head;
+	const struct fsmonitor_batch *batch;
+	intmax_t count = 0, duplicates = 0;
+	kh_str_t *shown;
+	int hash_ret;
+	int do_trivial = 0;
+	int do_flush = 0;
+
+	/*
+	 * We expect `command` to be of the form:
+	 *
+	 * <command> := quit NUL
+	 *            | flush NUL
+	 *            | <V1-time-since-epoch-ns> NUL
+	 *            | <V2-opaque-fsmonitor-token> NUL
+	 */
+
+	if (!strcmp(command, "quit")) {
+		/*
+		 * A client has requested over the socket/pipe that the
+		 * daemon shutdown.
+		 *
+		 * Tell the IPC thread pool to shutdown (which completes
+		 * the await in the main thread (which can stop the
+		 * fsmonitor listener thread)).
+		 *
+		 * There is no reply to the client.
+		 */
+		return SIMPLE_IPC_QUIT;
+
+	} else if (!strcmp(command, "flush")) {
+		/*
+		 * Flush all of our cached data and generate a new token
+		 * just like if we lost sync with the filesystem.
+		 *
+		 * Then send a trivial response using the new token.
+		 */
+		do_flush = 1;
+		do_trivial = 1;
+
+	} else if (!skip_prefix(command, "builtin:", &p)) {
+		/* assume V1 timestamp or garbage */
+
+		char *p_end;
+
+		strtoumax(command, &p_end, 10);
+		trace_printf_key(&trace_fsmonitor,
+				 ((*p_end) ?
+				  "fsmonitor: invalid command line '%s'" :
+				  "fsmonitor: unsupported V1 protocol '%s'"),
+				 command);
+		do_trivial = 1;
+
+	} else {
+		/* We have "builtin:*" */
+		if (fsmonitor_parse_client_token(command, &requested_token_id,
+						 &requested_oldest_seq_nr)) {
+			trace_printf_key(&trace_fsmonitor,
+					 "fsmonitor: invalid V2 protocol token '%s'",
+					 command);
+			do_trivial = 1;
+
+		} else {
+			/*
+			 * We have a V2 valid token:
+			 *     "builtin:<token_id>:<seq_nr>"
+			 */
+		}
+	}
+
+	pthread_mutex_lock(&state->main_lock);
+
+	if (!state->current_token_data)
+		BUG("fsmonitor state does not have a current token");
+
+	if (do_flush)
+		with_lock__do_force_resync(state);
+
+	/*
+	 * We mark the current head of the batch list as "pinned" so
+	 * that the listener thread will treat this item as read-only
+	 * (and prevent any more paths from being added to it) from
+	 * now on.
+	 */
+	token_data = state->current_token_data;
+	batch_head = token_data->batch_head;
+	((struct fsmonitor_batch *)batch_head)->pinned_time = time(NULL);
+
+	/*
+	 * FSMonitor Protocol V2 requires that we send a response header
+	 * with a "new current token" and then all of the paths that changed
+	 * since the "requested token".  We send the seq_nr of the just-pinned
+	 * head batch so that future requests from a client will be relative
+	 * to it.
+	 */
+	with_lock__format_response_token(&response_token,
+					 &token_data->token_id, batch_head);
+
+	reply(reply_data, response_token.buf, response_token.len + 1);
+	total_response_len += response_token.len + 1;
+
+	trace2_data_string("fsmonitor", the_repository, "response/token",
+			   response_token.buf);
+	trace_printf_key(&trace_fsmonitor, "response token: %s",
+			 response_token.buf);
+
+	if (!do_trivial) {
+		if (strcmp(requested_token_id.buf, token_data->token_id.buf)) {
+			/*
+			 * The client last spoke to a different daemon
+			 * instance -OR- the daemon had to resync with
+			 * the filesystem (and lost events), so reject.
+			 */
+			trace2_data_string("fsmonitor", the_repository,
+					   "response/token", "different");
+			do_trivial = 1;
+
+		} else if (requested_oldest_seq_nr <
+			   token_data->batch_tail->batch_seq_nr) {
+			/*
+			 * The client wants older events than we have for
+			 * this token_id.  This means that the end of our
+			 * batch list was truncated and we cannot give the
+			 * client a complete snapshot relative to their
+			 * request.
+			 */
+			trace_printf_key(&trace_fsmonitor,
+					 "client requested truncated data");
+			do_trivial = 1;
+		}
+	}
+
+	if (do_trivial) {
+		pthread_mutex_unlock(&state->main_lock);
+
+		reply(reply_data, "/", 2);
+
+		trace2_data_intmax("fsmonitor", the_repository,
+				   "response/trivial", 1);
+
+		strbuf_release(&response_token);
+		strbuf_release(&requested_token_id);
+		return 0;
+	}
+
+	/*
+	 * We're going to hold onto a pointer to the current
+	 * token-data while we walk the list of batches of files.
+	 * During this time, we will NOT be under the lock.
+	 * So we ref-count it.
+	 *
+	 * This allows the listener thread to continue prepending
+	 * new batches of items to the token-data (which we'll ignore).
+	 *
+	 * AND it allows the listener thread to do a token-reset
+	 * (and install a new `current_token_data`).
+	 */
+	token_data->client_ref_count++;
+
+	pthread_mutex_unlock(&state->main_lock);
+
+	/*
+	 * The client request is relative to the token that they sent,
+	 * so walk the batch list backwards from the current head back
+	 * to the batch (sequence number) they named.
+	 *
+	 * We use khash to de-dup the list of pathnames.
+	 *
+	 * NEEDSWORK: each batch contains a list of interned strings,
+	 * so we only need to do pointer comparisons here to build the
+	 * hash table.  Currently, we're still comparing the string
+	 * values.
+	 */
+	shown = kh_init_str();
+	for (batch = batch_head;
+	     batch && batch->batch_seq_nr > requested_oldest_seq_nr;
+	     batch = batch->next) {
+		size_t k;
+
+		for (k = 0; k < batch->nr; k++) {
+			const char *s = batch->interned_paths[k];
+			size_t s_len;
+
+			if (kh_get_str(shown, s) != kh_end(shown))
+				duplicates++;
+			else {
+				kh_put_str(shown, s, &hash_ret);
+
+				trace_printf_key(&trace_fsmonitor,
+						 "send[%"PRIuMAX"]: %s",
+						 count, s);
+
+				/* Each path gets written with a trailing NUL */
+				s_len = strlen(s) + 1;
+
+				if (payload.len + s_len >=
+				    LARGE_PACKET_DATA_MAX) {
+					reply(reply_data, payload.buf,
+					      payload.len);
+					total_response_len += payload.len;
+					strbuf_reset(&payload);
+				}
+
+				strbuf_add(&payload, s, s_len);
+				count++;
+			}
+		}
+	}
+
+	if (payload.len) {
+		reply(reply_data, payload.buf, payload.len);
+		total_response_len += payload.len;
+	}
+
+	kh_release_str(shown);
+
+	pthread_mutex_lock(&state->main_lock);
+
+	if (token_data->client_ref_count > 0)
+		token_data->client_ref_count--;
+
+	if (token_data->client_ref_count == 0) {
+		if (token_data != state->current_token_data) {
+			/*
+			 * The listener thread did a token-reset while we were
+			 * walking the batch list.  Therefore, this token is
+			 * stale and can be discarded completely.  If we are
+			 * the last reader thread using this token, we own
+			 * that work.
+			 */
+			fsmonitor_free_token_data(token_data);
+		}
+	}
+
+	pthread_mutex_unlock(&state->main_lock);
+
+	trace2_data_intmax("fsmonitor", the_repository, "response/length", total_response_len);
+	trace2_data_intmax("fsmonitor", the_repository, "response/count/files", count);
+	trace2_data_intmax("fsmonitor", the_repository, "response/count/duplicates", duplicates);
+
+	strbuf_release(&response_token);
+	strbuf_release(&requested_token_id);
+	strbuf_release(&payload);
+
+	return 0;
+}
+
 static ipc_server_application_cb handle_client;
 
 static int handle_client(void *data,
@@ -358,7 +664,7 @@ static int handle_client(void *data,
 			 ipc_server_reply_cb *reply,
 			 struct ipc_server_reply_data *reply_data)
 {
-	/* struct fsmonitor_daemon_state *state = data; */
+	struct fsmonitor_daemon_state *state = data;
 	int result;
 
 	/*
@@ -369,10 +675,12 @@ static int handle_client(void *data,
 	if (command_len != strlen(command))
 		BUG("FSMonitor assumes text messages");
 
+	trace_printf_key(&trace_fsmonitor, "requested token: %s", command);
+
 	trace2_region_enter("fsmonitor", "handle_client", the_repository);
 	trace2_data_string("fsmonitor", the_repository, "request", command);
 
-	result = 0; /* TODO Do something here. */
+	result = do_handle_client(state, command, reply, reply_data);
 
 	trace2_region_leave("fsmonitor", "handle_client", the_repository);
 
-- 
gitgitgadget

[PATCH v5 19/30] help: include fsmonitor--daemon feature flag in version info

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Add the "feature: fsmonitor--daemon" message to the output of
`git version --build-options`.

The builtin FSMonitor is only available on certain platforms and
even then only when certain Makefile flags are enabled, so print
a message in the verbose version output when it is available.

This can be used by test scripts for prereq testing.  Granted, tests
could just try `git fsmonitor--daemon status` and look for a 128 exit
code or grep for a "not supported" message on stderr, but this is
rather obscure.

The main advantage is that the feature message will automatically
appear in bug reports and other support requests.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 help.c        | 4 ++++
 t/test-lib.sh | 6 ++++++
 2 files changed, 10 insertions(+)

diff --git a/help.c b/help.c
index 71444906ddf..9112a51e84b 100644
--- a/help.c
+++ b/help.c
@@ -12,6 +12,7 @@
 #include "refs.h"
 #include "parse-options.h"
 #include "prompt.h"
+#include "fsmonitor-ipc.h"
 
 struct category_description {
 	uint32_t category;
@@ -695,6 +696,9 @@ void get_version_info(struct strbuf *buf, int show_build_options)
 		strbuf_addf(buf, "sizeof-size_t: %d\n", (int)sizeof(size_t));
 		strbuf_addf(buf, "shell-path: %s\n", SHELL_PATH);
 		/* NEEDSWORK: also save and output GIT-BUILD_OPTIONS? */
+
+		if (fsmonitor_ipc__is_supported())
+			strbuf_addstr(buf, "feature: fsmonitor--daemon\n");
 	}
 }
 
diff --git a/t/test-lib.sh b/t/test-lib.sh
index e4716b0b867..46cd596e7f5 100644
--- a/t/test-lib.sh
+++ b/t/test-lib.sh
@@ -1799,3 +1799,9 @@ test_lazy_prereq SHA1 '
 # Tests that verify the scheduler integration must set this locally
 # to avoid errors.
 GIT_TEST_MAINT_SCHEDULER="none:exit 1"
+
+# Does this platform support `git fsmonitor--daemon`
+#
+test_lazy_prereq FSMONITOR_DAEMON '
+	git version --build-options | grep "feature:" | grep "fsmonitor--daemon"
+'
-- 
gitgitgadget

[PATCH v5 20/30] t/helper/fsmonitor-client: create IPC client to talk to FSMonitor Daemon

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Create an IPC client to send query and flush commands to the daemon.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 Makefile                         |   1 +
 t/helper/test-fsmonitor-client.c | 121 +++++++++++++++++++++++++++++++
 t/helper/test-tool.c             |   1 +
 t/helper/test-tool.h             |   1 +
 4 files changed, 124 insertions(+)
 create mode 100644 t/helper/test-fsmonitor-client.c

diff --git a/Makefile b/Makefile
index b98f4899ac0..6846d91c37b 100644
--- a/Makefile
+++ b/Makefile
@@ -709,6 +709,7 @@ TEST_BUILTINS_OBJS += test-dump-split-index.o
 TEST_BUILTINS_OBJS += test-dump-untracked-cache.o
 TEST_BUILTINS_OBJS += test-example-decorate.o
 TEST_BUILTINS_OBJS += test-fast-rebase.o
+TEST_BUILTINS_OBJS += test-fsmonitor-client.o
 TEST_BUILTINS_OBJS += test-genrandom.o
 TEST_BUILTINS_OBJS += test-genzeros.o
 TEST_BUILTINS_OBJS += test-getcwd.o
diff --git a/t/helper/test-fsmonitor-client.c b/t/helper/test-fsmonitor-client.c
new file mode 100644
index 00000000000..f7a5b3a32fa
--- /dev/null
+++ b/t/helper/test-fsmonitor-client.c
@@ -0,0 +1,121 @@
+/*
+ * test-fsmonitor-client.c: client code to send commands/requests to
+ * a `git fsmonitor--daemon` daemon.
+ */
+
+#include "test-tool.h"
+#include "cache.h"
+#include "parse-options.h"
+#include "fsmonitor-ipc.h"
+
+#ifndef HAVE_FSMONITOR_DAEMON_BACKEND
+int cmd__fsmonitor_client(int argc, const char **argv)
+{
+	die("fsmonitor--daemon not available on this platform");
+}
+#else
+
+/*
+ * Read the `.git/index` to get the last token written to the
+ * FSMonitor Index Extension.
+ */
+static const char *get_token_from_index(void)
+{
+	struct index_state *istate = the_repository->index;
+
+	if (do_read_index(istate, the_repository->index_file, 0) < 0)
+		die("unable to read index file");
+	if (!istate->fsmonitor_last_update)
+		die("index file does not have fsmonitor extension");
+
+	return istate->fsmonitor_last_update;
+}
+
+/*
+ * Send an IPC query to a `git-fsmonitor--daemon` daemon and
+ * ask for the changes since the given token or from the last
+ * token in the index extension.
+ *
+ * This will implicitly start a daemon process if necessary.  The
+ * daemon process will persist after we exit.
+ */
+static int do_send_query(const char *token)
+{
+	struct strbuf answer = STRBUF_INIT;
+	int ret;
+
+	if (!token || !*token)
+		token = get_token_from_index();
+
+	ret = fsmonitor_ipc__send_query(token, &answer);
+	if (ret < 0)
+		die(_("could not query fsmonitor--daemon"));
+
+	write_in_full(1, answer.buf, answer.len);
+	strbuf_release(&answer);
+
+	return 0;
+}
+
+/*
+ * Send a "flush" command to the `git-fsmonitor--daemon` (if running)
+ * and tell it to flush its cache.
+ *
+ * This feature is primarily used by the test suite to simulate a loss of
+ * sync with the filesystem where we miss kernel events.
+ */
+static int do_send_flush(void)
+{
+	struct strbuf answer = STRBUF_INIT;
+	int ret;
+
+	ret = fsmonitor_ipc__send_command("flush", &answer);
+	if (ret)
+		return ret;
+
+	write_in_full(1, answer.buf, answer.len);
+	strbuf_release(&answer);
+
+	return 0;
+}
+
+int cmd__fsmonitor_client(int argc, const char **argv)
+{
+	const char *subcmd;
+	const char *token = NULL;
+
+	const char * const fsmonitor_client_usage[] = {
+		N_("test-helper fsmonitor-client query [<token>]"),
+		N_("test-helper fsmonitor-client flush"),
+		NULL,
+	};
+
+	struct option options[] = {
+		OPT_STRING(0, "token", &token, N_("token"),
+			   N_("command token to send to the server")),
+		OPT_END()
+	};
+
+	if (argc < 2)
+		usage_with_options(fsmonitor_client_usage, options);
+
+	if (argc == 2 && !strcmp(argv[1], "-h"))
+		usage_with_options(fsmonitor_client_usage, options);
+
+	subcmd = argv[1];
+	argv--;
+	argc++;
+
+	argc = parse_options(argc, argv, NULL, options, fsmonitor_client_usage, 0);
+
+	setup_git_directory();
+
+	if (!strcmp(subcmd, "query"))
+		return !!do_send_query(token);
+
+	if (!strcmp(subcmd, "flush"))
+		return !!do_send_flush();
+
+	die("Unhandled subcommand: '%s'", subcmd);
+}
+#endif
diff --git a/t/helper/test-tool.c b/t/helper/test-tool.c
index 338a57b104d..a4c5647b2f1 100644
--- a/t/helper/test-tool.c
+++ b/t/helper/test-tool.c
@@ -31,6 +31,7 @@ static struct test_cmd cmds[] = {
 	{ "dump-untracked-cache", cmd__dump_untracked_cache },
 	{ "example-decorate", cmd__example_decorate },
 	{ "fast-rebase", cmd__fast_rebase },
+	{ "fsmonitor-client", cmd__fsmonitor_client },
 	{ "genrandom", cmd__genrandom },
 	{ "genzeros", cmd__genzeros },
 	{ "getcwd", cmd__getcwd },
diff --git a/t/helper/test-tool.h b/t/helper/test-tool.h
index 48cee1f4a2d..fa4c936310c 100644
--- a/t/helper/test-tool.h
+++ b/t/helper/test-tool.h
@@ -22,6 +22,7 @@ int cmd__dump_untracked_cache(int argc, const char **argv);
 int cmd__dump_reftable(int argc, const char **argv);
 int cmd__example_decorate(int argc, const char **argv);
 int cmd__fast_rebase(int argc, const char **argv);
+int cmd__fsmonitor_client(int argc, const char **argv);
 int cmd__genrandom(int argc, const char **argv);
 int cmd__genzeros(int argc, const char **argv);
 int cmd__getcwd(int argc, const char **argv);
-- 
gitgitgadget

[PATCH v5 21/30] t7527: create test for fsmonitor--daemon

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 t/t7527-builtin-fsmonitor.sh | 511 +++++++++++++++++++++++++++++++++++
 1 file changed, 511 insertions(+)
 create mode 100755 t/t7527-builtin-fsmonitor.sh

diff --git a/t/t7527-builtin-fsmonitor.sh b/t/t7527-builtin-fsmonitor.sh
new file mode 100755
index 00000000000..5f7b8e54233
--- /dev/null
+++ b/t/t7527-builtin-fsmonitor.sh
@@ -0,0 +1,511 @@
+#!/bin/sh
+
+test_description='built-in file system watcher'
+
+. ./test-lib.sh
+
+if ! test_have_prereq FSMONITOR_DAEMON
+then
+	skip_all="fsmonitor--daemon is not supported on this platform"
+	test_done
+fi
+
+stop_daemon_delete_repo () {
+	r=$1
+	git -C $r fsmonitor--daemon stop >/dev/null 2>/dev/null
+	rm -rf $1
+	return 0
+}
+
+start_daemon () {
+	case "$#" in
+		1) r="-C $1";;
+		*) r="";
+	esac
+
+	git $r fsmonitor--daemon start || return $?
+	git $r fsmonitor--daemon status || return $?
+
+	return 0
+}
+
+# Is a Trace2 data event present with the given catetory and key?
+# We do not care what the value is.
+#
+have_t2_data_event () {
+	c=$1
+	k=$2
+
+	grep -e '"event":"data".*"category":"'"$c"'".*"key":"'"$k"'"'
+}
+
+test_expect_success 'explicit daemon start and stop' '
+	test_when_finished "stop_daemon_delete_repo test_explicit" &&
+
+	git init test_explicit &&
+	start_daemon test_explicit &&
+
+	git -C test_explicit fsmonitor--daemon stop &&
+	test_must_fail git -C test_explicit fsmonitor--daemon status
+'
+
+test_expect_success 'implicit daemon start' '
+	test_when_finished "stop_daemon_delete_repo test_implicit" &&
+
+	git init test_implicit &&
+	test_must_fail git -C test_implicit fsmonitor--daemon status &&
+
+	# query will implicitly start the daemon.
+	#
+	# for test-script simplicity, we send a V1 timestamp rather than
+	# a V2 token.  either way, the daemon response to any query contains
+	# a new V2 token.  (the daemon may complain that we sent a V1 request,
+	# but this test case is only concerned with whether the daemon was
+	# implicitly started.)
+
+	GIT_TRACE2_EVENT="$(pwd)/.git/trace" \
+		test-tool -C test_implicit fsmonitor-client query --token 0 >actual &&
+	nul_to_q <actual >actual.filtered &&
+	grep "builtin:" actual.filtered &&
+
+	# confirm that a daemon was started in the background.
+	#
+	# since the mechanism for starting the background daemon is platform
+	# dependent, just confirm that the foreground command received a
+	# response from the daemon.
+
+	have_t2_data_event fsm_client query/response-length <.git/trace &&
+
+	git -C test_implicit fsmonitor--daemon status &&
+	git -C test_implicit fsmonitor--daemon stop &&
+	test_must_fail git -C test_implicit fsmonitor--daemon status
+'
+
+test_expect_success 'implicit daemon stop (delete .git)' '
+	test_when_finished "stop_daemon_delete_repo test_implicit_1" &&
+
+	git init test_implicit_1 &&
+
+	start_daemon test_implicit_1 &&
+
+	# deleting the .git directory will implicitly stop the daemon.
+	rm -rf test_implicit_1/.git &&
+
+	# [1] Create an empty .git directory so that the following Git
+	#     command will stay relative to the `-C` directory.
+	#
+	#     Without this, the Git command will override the requested
+	#     -C argument and crawl out to the containing Git source tree.
+	#     This would make the test result dependent upon whether we
+	#     were using fsmonitor on our development worktree.
+	#
+	sleep 1 &&
+	mkdir test_implicit_1/.git &&
+
+	test_must_fail git -C test_implicit_1 fsmonitor--daemon status
+'
+
+test_expect_success 'implicit daemon stop (rename .git)' '
+	test_when_finished "stop_daemon_delete_repo test_implicit_2" &&
+
+	git init test_implicit_2 &&
+
+	start_daemon test_implicit_2 &&
+
+	# renaming the .git directory will implicitly stop the daemon.
+	mv test_implicit_2/.git test_implicit_2/.xxx &&
+
+	# See [1] above.
+	#
+	sleep 1 &&
+	mkdir test_implicit_2/.git &&
+
+	test_must_fail git -C test_implicit_2 fsmonitor--daemon status
+'
+
+test_expect_success 'cannot start multiple daemons' '
+	test_when_finished "stop_daemon_delete_repo test_multiple" &&
+
+	git init test_multiple &&
+
+	start_daemon test_multiple &&
+
+	test_must_fail git -C test_multiple fsmonitor--daemon start 2>actual &&
+	grep "fsmonitor--daemon is already running" actual &&
+
+	git -C test_multiple fsmonitor--daemon stop &&
+	test_must_fail git -C test_multiple fsmonitor--daemon status
+'
+
+# These tests use the main repo in the trash directory
+
+test_expect_success 'setup' '
+	>tracked &&
+	>modified &&
+	>delete &&
+	>rename &&
+	mkdir dir1 &&
+	>dir1/tracked &&
+	>dir1/modified &&
+	>dir1/delete &&
+	>dir1/rename &&
+	mkdir dir2 &&
+	>dir2/tracked &&
+	>dir2/modified &&
+	>dir2/delete &&
+	>dir2/rename &&
+	mkdir dirtorename &&
+	>dirtorename/a &&
+	>dirtorename/b &&
+
+	cat >.gitignore <<-\EOF &&
+	.gitignore
+	expect*
+	actual*
+	EOF
+
+	git -c core.fsmonitor=false add . &&
+	test_tick &&
+	git -c core.fsmonitor=false commit -m initial &&
+
+	git config core.fsmonitor true
+'
+
+# The test already explicitly stopped (or tried to stop) the daemon.
+# This is here in case something else fails first.
+#
+redundant_stop_daemon () {
+	git fsmonitor--daemon stop
+	return 0
+}
+
+test_expect_success 'update-index implicitly starts daemon' '
+	test_when_finished redundant_stop_daemon &&
+
+	test_must_fail git fsmonitor--daemon status &&
+
+	GIT_TRACE2_EVENT="$(pwd)/.git/trace_implicit_1" \
+		git update-index --fsmonitor &&
+
+	git fsmonitor--daemon status &&
+	test_might_fail git fsmonitor--daemon stop &&
+
+	# Confirm that the trace2 log contains a record of the
+	# daemon starting.
+	test_subcommand git fsmonitor--daemon start <.git/trace_implicit_1
+'
+
+test_expect_success 'status implicitly starts daemon' '
+	test_when_finished redundant_stop_daemon &&
+
+	test_must_fail git fsmonitor--daemon status &&
+
+	GIT_TRACE2_EVENT="$(pwd)/.git/trace_implicit_2" \
+		git status >actual &&
+
+	git fsmonitor--daemon status &&
+	test_might_fail git fsmonitor--daemon stop &&
+
+	# Confirm that the trace2 log contains a record of the
+	# daemon starting.
+	test_subcommand git fsmonitor--daemon start <.git/trace_implicit_2
+'
+
+edit_files () {
+	echo 1 >modified
+	echo 2 >dir1/modified
+	echo 3 >dir2/modified
+	>dir1/untracked
+}
+
+delete_files () {
+	rm -f delete
+	rm -f dir1/delete
+	rm -f dir2/delete
+}
+
+create_files () {
+	echo 1 >new
+	echo 2 >dir1/new
+	echo 3 >dir2/new
+}
+
+rename_files () {
+	mv rename renamed
+	mv dir1/rename dir1/renamed
+	mv dir2/rename dir2/renamed
+}
+
+file_to_directory () {
+	rm -f delete
+	mkdir delete
+	echo 1 >delete/new
+}
+
+directory_to_file () {
+	rm -rf dir1
+	echo 1 >dir1
+}
+
+verify_status () {
+	git status >actual &&
+	GIT_INDEX_FILE=.git/fresh-index git read-tree master &&
+	GIT_INDEX_FILE=.git/fresh-index git -c core.fsmonitor=false status >expect &&
+	test_cmp expect actual &&
+	echo HELLO AFTER &&
+	cat .git/trace &&
+	echo HELLO AFTER
+}
+
+# The next few test cases confirm that our fsmonitor daemon sees each type
+# of OS filesystem notification that we care about.  At this layer we just
+# ensure we are getting the OS notifications and do not try to confirm what
+# is reported by `git status`.
+#
+# We run a simple query after modifying the filesystem just to introduce
+# a bit of a delay so that the trace logging from the daemon has time to
+# get flushed to disk.
+#
+# We `reset` and `clean` at the bottom of each test (and before stopping the
+# daemon) because these commands might implicitly restart the daemon.
+
+clean_up_repo_and_stop_daemon () {
+	git reset --hard HEAD
+	git clean -fd
+	git fsmonitor--daemon stop
+	rm -f .git/trace
+}
+
+test_expect_success 'edit some files' '
+	test_when_finished clean_up_repo_and_stop_daemon &&
+
+	(
+		GIT_TRACE_FSMONITOR="$(pwd)/.git/trace" &&
+		export GIT_TRACE_FSMONITOR &&
+
+		start_daemon
+	) &&
+
+	edit_files &&
+
+	test-tool fsmonitor-client query --token 0 >/dev/null 2>&1 &&
+
+	grep "^event: dir1/modified$"  .git/trace &&
+	grep "^event: dir2/modified$"  .git/trace &&
+	grep "^event: modified$"       .git/trace &&
+	grep "^event: dir1/untracked$" .git/trace
+'
+
+test_expect_success 'create some files' '
+	test_when_finished clean_up_repo_and_stop_daemon &&
+
+	(
+		GIT_TRACE_FSMONITOR="$(pwd)/.git/trace" &&
+		export GIT_TRACE_FSMONITOR &&
+
+		start_daemon
+	) &&
+
+	create_files &&
+
+	test-tool fsmonitor-client query --token 0 >/dev/null 2>&1 &&
+
+	grep "^event: dir1/new$" .git/trace &&
+	grep "^event: dir2/new$" .git/trace &&
+	grep "^event: new$"      .git/trace
+'
+
+test_expect_success 'delete some files' '
+	test_when_finished clean_up_repo_and_stop_daemon &&
+
+	(
+		GIT_TRACE_FSMONITOR="$(pwd)/.git/trace" &&
+		export GIT_TRACE_FSMONITOR &&
+
+		start_daemon
+	) &&
+
+	delete_files &&
+
+	test-tool fsmonitor-client query --token 0 >/dev/null 2>&1 &&
+
+	grep "^event: dir1/delete$" .git/trace &&
+	grep "^event: dir2/delete$" .git/trace &&
+	grep "^event: delete$"      .git/trace
+'
+
+test_expect_success 'rename some files' '
+	test_when_finished clean_up_repo_and_stop_daemon &&
+
+	(
+		GIT_TRACE_FSMONITOR="$(pwd)/.git/trace" &&
+		export GIT_TRACE_FSMONITOR &&
+
+		start_daemon
+	) &&
+
+	rename_files &&
+
+	test-tool fsmonitor-client query --token 0 >/dev/null 2>&1 &&
+
+	grep "^event: dir1/rename$"  .git/trace &&
+	grep "^event: dir2/rename$"  .git/trace &&
+	grep "^event: rename$"       .git/trace &&
+	grep "^event: dir1/renamed$" .git/trace &&
+	grep "^event: dir2/renamed$" .git/trace &&
+	grep "^event: renamed$"      .git/trace
+'
+
+test_expect_success 'rename directory' '
+	test_when_finished clean_up_repo_and_stop_daemon &&
+
+	(
+		GIT_TRACE_FSMONITOR="$(pwd)/.git/trace" &&
+		export GIT_TRACE_FSMONITOR &&
+
+		start_daemon
+	) &&
+
+	mv dirtorename dirrenamed &&
+
+	test-tool fsmonitor-client query --token 0 >/dev/null 2>&1 &&
+
+	grep "^event: dirtorename/*$" .git/trace &&
+	grep "^event: dirrenamed/*$"  .git/trace
+'
+
+test_expect_success 'file changes to directory' '
+	test_when_finished clean_up_repo_and_stop_daemon &&
+
+	(
+		GIT_TRACE_FSMONITOR="$(pwd)/.git/trace" &&
+		export GIT_TRACE_FSMONITOR &&
+
+		start_daemon
+	) &&
+
+	file_to_directory &&
+
+	test-tool fsmonitor-client query --token 0 >/dev/null 2>&1 &&
+
+	grep "^event: delete$"     .git/trace &&
+	grep "^event: delete/new$" .git/trace
+'
+
+test_expect_success 'directory changes to a file' '
+	test_when_finished clean_up_repo_and_stop_daemon &&
+
+	(
+		GIT_TRACE_FSMONITOR="$(pwd)/.git/trace" &&
+		export GIT_TRACE_FSMONITOR &&
+
+		start_daemon
+	) &&
+
+	directory_to_file &&
+
+	test-tool fsmonitor-client query --token 0 >/dev/null 2>&1 &&
+
+	grep "^event: dir1$" .git/trace
+'
+
+# The next few test cases exercise the token-resync code.  When filesystem
+# drops events (because of filesystem velocity or because the daemon isn't
+# polling fast enough), we need to discard the cached data (relative to the
+# current token) and start collecting events under a new token.
+#
+# the 'test-tool fsmonitor-client flush' command can be used to send a
+# "flush" message to a running daemon and ask it to do a flush/resync.
+
+test_expect_success 'flush cached data' '
+	test_when_finished "stop_daemon_delete_repo test_flush" &&
+
+	git init test_flush &&
+
+	(
+		GIT_TEST_FSMONITOR_TOKEN=true &&
+		export GIT_TEST_FSMONITOR_TOKEN &&
+
+		GIT_TRACE_FSMONITOR="$(pwd)/.git/trace_daemon" &&
+		export GIT_TRACE_FSMONITOR &&
+
+		start_daemon test_flush
+	) &&
+
+	# The daemon should have an initial token with no events in _0 and
+	# then a few (probably platform-specific number of) events in _1.
+	# These should both have the same <token_id>.
+
+	test-tool -C test_flush fsmonitor-client query --token "builtin:test_00000001:0" >actual_0 &&
+	nul_to_q <actual_0 >actual_q0 &&
+
+	touch test_flush/file_1 &&
+	touch test_flush/file_2 &&
+
+	test-tool -C test_flush fsmonitor-client query --token "builtin:test_00000001:0" >actual_1 &&
+	nul_to_q <actual_1 >actual_q1 &&
+
+	grep "file_1" actual_q1 &&
+
+	# Force a flush.  This will change the <token_id>, reset the <seq_nr>, and
+	# flush the file data.  Then create some events and ensure that the file
+	# again appears in the cache.  It should have the new <token_id>.
+
+	test-tool -C test_flush fsmonitor-client flush >flush_0 &&
+	nul_to_q <flush_0 >flush_q0 &&
+	grep "^builtin:test_00000002:0Q/Q$" flush_q0 &&
+
+	test-tool -C test_flush fsmonitor-client query --token "builtin:test_00000002:0" >actual_2 &&
+	nul_to_q <actual_2 >actual_q2 &&
+
+	grep "^builtin:test_00000002:0Q$" actual_q2 &&
+
+	touch test_flush/file_3 &&
+
+	test-tool -C test_flush fsmonitor-client query --token "builtin:test_00000002:0" >actual_3 &&
+	nul_to_q <actual_3 >actual_q3 &&
+
+	grep "file_3" actual_q3
+'
+
+# The next few test cases create repos where the .git directory is NOT
+# inside the one of the working directory.  That is, where .git is a file
+# that points to a directory elsewhere.  This happens for submodules and
+# non-primary worktrees.
+
+test_expect_success 'setup worktree base' '
+	git init wt-base &&
+	echo 1 >wt-base/file1 &&
+	git -C wt-base add file1 &&
+	git -C wt-base commit -m "c1"
+'
+
+test_expect_success 'worktree with .git file' '
+	git -C wt-base worktree add ../wt-secondary &&
+
+	(
+		GIT_TRACE2_PERF="$(pwd)/trace2_wt_secondary" &&
+		export GIT_TRACE2_PERF &&
+
+		GIT_TRACE_FSMONITOR="$(pwd)/trace_wt_secondary" &&
+		export GIT_TRACE_FSMONITOR &&
+
+		start_daemon wt-secondary
+	) &&
+
+	git -C wt-secondary fsmonitor--daemon stop &&
+	test_must_fail git -C wt-secondary fsmonitor--daemon status
+'
+
+# NEEDSWORK: Repeat one of the "edit" tests on wt-secondary and
+# confirm that we get the same events and behavior -- that is, that
+# fsmonitor--daemon correctly watches BOTH the working directory and
+# the external GITDIR directory and behaves the same as when ".git"
+# is a directory inside the working directory.
+
+test_expect_success 'cleanup worktrees' '
+	stop_daemon_delete_repo wt-secondary &&
+	stop_daemon_delete_repo wt-base
+'
+
+test_done
-- 
gitgitgadget

[PATCH v5 22/30] t/perf: avoid copying builtin fsmonitor files into test repo

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Do not copy any of the various fsmonitor--daemon files from the .git
directory of the (GIT_PREF_REPO or GIT_PERF_LARGE_REPO) source repo
into the test's trash directory.

When perf tests start, they copy the contents of the source repo into
the test's trash directory.  If fsmonitor is running in the source repo,
there may be control files, such as the IPC socket and/or fsmonitor
cookie files.  These should not be copied into the test repo.

Unix domain sockets cannot be copied in the manner used by the test
setup, so if present, the test setup fails.

Cookie files are harmless, but we should avoid them.

The builtin fsmonitor keeps all such control files/sockets in
.git/fsmonitor--daemon*, so it is simple to exclude them.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 t/perf/perf-lib.sh | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/t/perf/perf-lib.sh b/t/perf/perf-lib.sh
index 407252bac70..932105cd12c 100644
--- a/t/perf/perf-lib.sh
+++ b/t/perf/perf-lib.sh
@@ -78,7 +78,7 @@ test_perf_copy_repo_contents () {
 	for stuff in "$1"/*
 	do
 		case "$stuff" in
-		*/objects|*/hooks|*/config|*/commondir|*/gitdir|*/worktrees)
+		*/objects|*/hooks|*/config|*/commondir|*/gitdir|*/worktrees|*/fsmonitor--daemon*)
 			;;
 		*)
 			cp -R "$stuff" "$repo/.git/" || exit 1
-- 
gitgitgadget

[PATCH v5 23/30] t/helper/test-chmtime: skip directories on Windows

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach `test-tool.exe chmtime` to ignore errors when setting the mtime
on a directory on Windows.

NEEDSWORK: The Windows version of `utime()` (aka `mingw_utime()`) does
not properly handle directories because it uses `_wopen()`.  It should
be converted to using `CreateFileW()` and backup semantics at a minimum.
Since I'm already in the middle of a large patch series, I did not want
to destabilize other callers of `utime()` right now.  The problem has
only been observed in the t/perf/p7519 test when the test repo contains
an empty directory on disk.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 t/helper/test-chmtime.c | 15 +++++++++++++++
 1 file changed, 15 insertions(+)

diff --git a/t/helper/test-chmtime.c b/t/helper/test-chmtime.c
index 524b55ca496..dc28890a183 100644
--- a/t/helper/test-chmtime.c
+++ b/t/helper/test-chmtime.c
@@ -134,6 +134,21 @@ int cmd__chmtime(int argc, const char **argv)
 		}
 
 		if (utb.modtime != sb.st_mtime && utime(argv[i], &utb) < 0) {
+#ifdef GIT_WINDOWS_NATIVE
+			if (S_ISDIR(sb.st_mode)) {
+				/*
+				 * NEEDSWORK: The Windows version of `utime()`
+				 * (aka `mingw_utime()`) does not correctly
+				 * handle directory arguments, since it uses
+				 * `_wopen()`.  Ignore it for now since this
+				 * is just a test.
+				 */
+				fprintf(stderr,
+					("Failed to modify time on directory %s. "
+					 "Skipping\n"), argv[i]);
+				continue;
+			}
+#endif
 			fprintf(stderr, "Failed to modify time on %s: %s\n",
 			        argv[i], strerror(errno));
 			return 1;
-- 
gitgitgadget

[PATCH v5 24/30] t/perf/p7519: speed up test on Windows

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Change p7519 to use `test_seq` and `xargs` rather than a `for` loop
to touch thousands of files.  This takes minutes off of test runs
on Windows because of process creation overhead.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 t/perf/p7519-fsmonitor.sh | 32 ++++++++++++++++++++------------
 1 file changed, 20 insertions(+), 12 deletions(-)

diff --git a/t/perf/p7519-fsmonitor.sh b/t/perf/p7519-fsmonitor.sh
index c8be58f3c76..054fc8d5d1d 100755
--- a/t/perf/p7519-fsmonitor.sh
+++ b/t/perf/p7519-fsmonitor.sh
@@ -72,7 +72,7 @@ then
 	fi
 fi
 
-trace_start() {
+trace_start () {
 	if test -n "$GIT_PERF_7519_TRACE"
 	then
 		name="$1"
@@ -91,13 +91,20 @@ trace_start() {
 	fi
 }
 
-trace_stop() {
+trace_stop () {
 	if test -n "$GIT_PERF_7519_TRACE"
 	then
 		unset GIT_TRACE2_PERF
 	fi
 }
 
+touch_files () {
+	n=$1
+	d="$n"_files
+
+	(cd $d ; test_seq 1 $n | xargs touch )
+}
+
 test_expect_success "one time repo setup" '
 	# set untrackedCache depending on the environment
 	if test -n "$GIT_PERF_7519_UNTRACKED_CACHE"
@@ -119,10 +126,11 @@ test_expect_success "one time repo setup" '
 	fi &&
 
 	mkdir 1_file 10_files 100_files 1000_files 10000_files &&
-	for i in $(test_seq 1 10); do touch 10_files/$i || return 1; done &&
-	for i in $(test_seq 1 100); do touch 100_files/$i || return 1; done &&
-	for i in $(test_seq 1 1000); do touch 1000_files/$i || return 1; done &&
-	for i in $(test_seq 1 10000); do touch 10000_files/$i || return 1; done &&
+	touch_files 1 &&
+	touch_files 10 &&
+	touch_files 100 &&
+	touch_files 1000 &&
+	touch_files 10000 &&
 	git add 1_file 10_files 100_files 1000_files 10000_files &&
 	git commit -qm "Add files" &&
 
@@ -133,7 +141,7 @@ test_expect_success "one time repo setup" '
 	fi
 '
 
-setup_for_fsmonitor() {
+setup_for_fsmonitor () {
 	# set INTEGRATION_SCRIPT depending on the environment
 	if test -n "$INTEGRATION_PATH"
 	then
@@ -173,7 +181,7 @@ test_perf_w_drop_caches () {
 	test_perf "$@"
 }
 
-test_fsmonitor_suite() {
+test_fsmonitor_suite () {
 	if test -n "$INTEGRATION_SCRIPT"; then
 		DESC="fsmonitor=$(basename $INTEGRATION_SCRIPT)"
 	else
@@ -199,15 +207,15 @@ test_fsmonitor_suite() {
 
 	# Update the mtimes on upto 100k files to make status think
 	# that they are dirty.  For simplicity, omit any files with
-	# LFs (i.e. anything that ls-files thinks it needs to dquote).
-	# Then fully backslash-quote the paths to capture any
-	# whitespace so that they pass thru xargs properly.
+	# LFs (i.e. anything that ls-files thinks it needs to dquote)
+	# and any files with whitespace so that they pass thru xargs
+	# properly.
 	#
 	test_perf_w_drop_caches "status (dirty) ($DESC)" '
 		git ls-files | \
 			head -100000 | \
 			grep -v \" | \
-			sed '\''s/\(.\)/\\\1/g'\'' | \
+			egrep -v " ." | \
 			xargs test-tool chmtime -300 &&
 		git status
 	'
-- 
gitgitgadget

[PATCH v5 25/30] t/perf/p7519: add fsmonitor--daemon test cases

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Repeat all of the fsmonitor perf tests using `git fsmonitor--daemon` and
the "Simple IPC" interface.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 t/perf/p7519-fsmonitor.sh | 29 ++++++++++++++++++++++++++++-
 1 file changed, 28 insertions(+), 1 deletion(-)

diff --git a/t/perf/p7519-fsmonitor.sh b/t/perf/p7519-fsmonitor.sh
index 054fc8d5d1d..9a2288a622d 100755
--- a/t/perf/p7519-fsmonitor.sh
+++ b/t/perf/p7519-fsmonitor.sh
@@ -182,7 +182,10 @@ test_perf_w_drop_caches () {
 }
 
 test_fsmonitor_suite () {
-	if test -n "$INTEGRATION_SCRIPT"; then
+	if test -n "$USE_FSMONITOR_DAEMON"
+	then
+		DESC="builtin fsmonitor--daemon"
+	elif test -n "$INTEGRATION_SCRIPT"; then
 		DESC="fsmonitor=$(basename $INTEGRATION_SCRIPT)"
 	else
 		DESC="fsmonitor=disabled"
@@ -293,4 +296,28 @@ test_expect_success "setup without fsmonitor" '
 test_fsmonitor_suite
 trace_stop
 
+#
+# Run a full set of perf tests using the built-in fsmonitor--daemon.
+# It does not use the Hook API, so it has a different setup.
+# Explicitly start the daemon here and before we start client commands
+# so that we can later add custom tracing.
+#
+if test_have_prereq FSMONITOR_DAEMON
+then
+	USE_FSMONITOR_DAEMON=t
+
+	trace_start fsmonitor--daemon--server
+	git fsmonitor--daemon start
+
+	trace_start fsmonitor--daemon--client
+
+	git config core.fsmonitor true
+	git update-index --fsmonitor
+
+	test_fsmonitor_suite
+
+	git fsmonitor--daemon stop
+	trace_stop
+fi
+
 test_done
-- 
gitgitgadget

[PATCH v5 26/30] fsmonitor--daemon: periodically truncate list of modified files

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach fsmonitor--daemon to periodically truncate the list of
modified files to save some memory.

Clients will ask for the set of changes relative to a token that they
found in the FSMN index extension in the index.  (This token is like a
point in time, but different).  Clients will then update the index to
contain the response token (so that subsequent commands will be
relative to this new token).

Therefore, the daemon can gradually truncate the in-memory list of
changed paths as they become obsolete (older than the previous token).
Since we may have multiple clients making concurrent requests with a
skew of tokens and clients may be racing to the talk to the daemon,
we lazily truncate the list.

We introduce a 5 minute delay and truncate batches 5 minutes after
they are considered obsolete.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 builtin/fsmonitor--daemon.c | 88 +++++++++++++++++++++++++++++++++++++
 1 file changed, 88 insertions(+)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index 4c88171e06c..962b24569e1 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -299,6 +299,75 @@ static void fsmonitor_batch__combine(struct fsmonitor_batch *batch_dest,
 			batch_src->interned_paths[k];
 }
 
+/*
+ * To keep the batch list from growing unbounded in response to filesystem
+ * activity, we try to truncate old batches from the end of the list as
+ * they become irrelevant.
+ *
+ * We assume that the .git/index will be updated with the most recent token
+ * any time the index is updated.  And future commands will only ask for
+ * recent changes *since* that new token.  So as tokens advance into the
+ * future, older batch items will never be requested/needed.  So we can
+ * truncate them without loss of functionality.
+ *
+ * However, multiple commands may be talking to the daemon concurrently
+ * or perform a slow command, so a little "token skew" is possible.
+ * Therefore, we want this to be a little bit lazy and have a generous
+ * delay.
+ *
+ * The current reader thread walked backwards in time from `token->batch_head`
+ * back to `batch_marker` somewhere in the middle of the batch list.
+ *
+ * Let's walk backwards in time from that marker an arbitrary delay
+ * and truncate the list there.  Note that these timestamps are completely
+ * artificial (based on when we pinned the batch item) and not on any
+ * filesystem activity.
+ *
+ * Return the obsolete portion of the list after we have removed it from
+ * the official list so that the caller can free it after leaving the lock.
+ */
+#define MY_TIME_DELAY_SECONDS (5 * 60) /* seconds */
+
+static struct fsmonitor_batch *with_lock__truncate_old_batches(
+	struct fsmonitor_daemon_state *state,
+	const struct fsmonitor_batch *batch_marker)
+{
+	/* assert current thread holding state->main_lock */
+
+	const struct fsmonitor_batch *batch;
+	struct fsmonitor_batch *remainder;
+
+	if (!batch_marker)
+		return NULL;
+
+	trace_printf_key(&trace_fsmonitor, "Truncate: mark (%"PRIu64",%"PRIu64")",
+			 batch_marker->batch_seq_nr,
+			 (uint64_t)batch_marker->pinned_time);
+
+	for (batch = batch_marker; batch; batch = batch->next) {
+		time_t t;
+
+		if (!batch->pinned_time) /* an overflow batch */
+			continue;
+
+		t = batch->pinned_time + MY_TIME_DELAY_SECONDS;
+		if (t > batch_marker->pinned_time) /* too close to marker */
+			continue;
+
+		goto truncate_past_here;
+	}
+
+	return NULL;
+
+truncate_past_here:
+	state->current_token_data->batch_tail = (struct fsmonitor_batch *)batch;
+
+	remainder = ((struct fsmonitor_batch *)batch)->next;
+	((struct fsmonitor_batch *)batch)->next = NULL;
+
+	return remainder;
+}
+
 static void fsmonitor_free_token_data(struct fsmonitor_token_data *token)
 {
 	if (!token)
@@ -412,6 +481,7 @@ static int do_handle_client(struct fsmonitor_daemon_state *state,
 	const char *p;
 	const struct fsmonitor_batch *batch_head;
 	const struct fsmonitor_batch *batch;
+	struct fsmonitor_batch *remainder = NULL;
 	intmax_t count = 0, duplicates = 0;
 	kh_str_t *shown;
 	int hash_ret;
@@ -641,11 +711,29 @@ static int do_handle_client(struct fsmonitor_daemon_state *state,
 			 * that work.
 			 */
 			fsmonitor_free_token_data(token_data);
+		} else if (batch) {
+			/*
+			 * We are holding the lock and are the only
+			 * reader of the ref-counted portion of the
+			 * list, so we get the honor of seeing if the
+			 * list can be truncated to save memory.
+			 *
+			 * The main loop did not walk to the end of the
+			 * list, so this batch is the first item in the
+			 * batch-list that is older than the requested
+			 * end-point sequence number.  See if the tail
+			 * end of the list is obsolete.
+			 */
+			remainder = with_lock__truncate_old_batches(state,
+								    batch);
 		}
 	}
 
 	pthread_mutex_unlock(&state->main_lock);
 
+	if (remainder)
+		fsmonitor_batch__free_list(remainder);
+
 	trace2_data_intmax("fsmonitor", the_repository, "response/length", total_response_len);
 	trace2_data_intmax("fsmonitor", the_repository, "response/count/files", count);
 	trace2_data_intmax("fsmonitor", the_repository, "response/count/duplicates", duplicates);
-- 
gitgitgadget

[PATCH v5 27/30] fsmonitor--daemon: use a cookie file to sync with file system

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach fsmonitor--daemon client threads to create a cookie file
inside the .git directory and then wait until FS events for the
cookie are observed by the FS listener thread.

This helps address the racy nature of file system events by
blocking the client response until the kernel has drained any
event backlog.

This is especially important on MacOS where kernel events are
only issued with a limited frequency.  See the `latency` argument
of `FSeventStreamCreate()`.  The kernel only signals every `latency`
seconds, but does not guarantee that the kernel queue is completely
drained, so we may have to wait more than one interval.  If we
increase the frequency, the system is more likely to drop events.
We avoid these issues by having each client thread create a unique
cookie file and then wait until it is seen in the event stream.

Co-authored-by: Kevin Willford <Kevin.Willford@microsoft.com>
Co-authored-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 builtin/fsmonitor--daemon.c | 228 +++++++++++++++++++++++++++++++++++-
 fsmonitor--daemon.h         |   5 +
 2 files changed, 232 insertions(+), 1 deletion(-)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index 962b24569e1..6011fe42ee0 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -94,6 +94,149 @@ static int do_as_client__status(void)
 	}
 }
 
+enum fsmonitor_cookie_item_result {
+	FCIR_ERROR = -1, /* could not create cookie file ? */
+	FCIR_INIT = 0,
+	FCIR_SEEN,
+	FCIR_ABORT,
+};
+
+struct fsmonitor_cookie_item {
+	struct hashmap_entry entry;
+	const char *name;
+	enum fsmonitor_cookie_item_result result;
+};
+
+static int cookies_cmp(const void *data, const struct hashmap_entry *he1,
+		     const struct hashmap_entry *he2, const void *keydata)
+{
+	const struct fsmonitor_cookie_item *a =
+		container_of(he1, const struct fsmonitor_cookie_item, entry);
+	const struct fsmonitor_cookie_item *b =
+		container_of(he2, const struct fsmonitor_cookie_item, entry);
+
+	return strcmp(a->name, keydata ? keydata : b->name);
+}
+
+static enum fsmonitor_cookie_item_result with_lock__wait_for_cookie(
+	struct fsmonitor_daemon_state *state)
+{
+	/* assert current thread holding state->main_lock */
+
+	int fd;
+	struct fsmonitor_cookie_item *cookie;
+	struct strbuf cookie_pathname = STRBUF_INIT;
+	struct strbuf cookie_filename = STRBUF_INIT;
+	enum fsmonitor_cookie_item_result result;
+	int my_cookie_seq;
+
+	CALLOC_ARRAY(cookie, 1);
+
+	my_cookie_seq = state->cookie_seq++;
+
+	strbuf_addf(&cookie_filename, "%i-%i", getpid(), my_cookie_seq);
+
+	strbuf_addbuf(&cookie_pathname, &state->path_cookie_prefix);
+	strbuf_addbuf(&cookie_pathname, &cookie_filename);
+
+	cookie->name = strbuf_detach(&cookie_filename, NULL);
+	cookie->result = FCIR_INIT;
+	hashmap_entry_init(&cookie->entry, strhash(cookie->name));
+
+	hashmap_add(&state->cookies, &cookie->entry);
+
+	trace_printf_key(&trace_fsmonitor, "cookie-wait: '%s' '%s'",
+			 cookie->name, cookie_pathname.buf);
+
+	/*
+	 * Create the cookie file on disk and then wait for a notification
+	 * that the listener thread has seen it.
+	 */
+	fd = open(cookie_pathname.buf, O_WRONLY | O_CREAT | O_EXCL, 0600);
+	if (fd >= 0) {
+		close(fd);
+		unlink(cookie_pathname.buf);
+
+		/*
+		 * NEEDSWORK: This is an infinite wait (well, unless another
+		 * thread sends us an abort).  I'd like to change this to
+		 * use `pthread_cond_timedwait()` and return an error/timeout
+		 * and let the caller do the trivial response thing.
+		 */
+		while (cookie->result == FCIR_INIT)
+			pthread_cond_wait(&state->cookies_cond,
+					  &state->main_lock);
+	} else {
+		error_errno(_("could not create fsmonitor cookie '%s'"),
+			    cookie->name);
+
+		cookie->result = FCIR_ERROR;
+	}
+
+	hashmap_remove(&state->cookies, &cookie->entry, NULL);
+
+	result = cookie->result;
+
+	free((char*)cookie->name);
+	free(cookie);
+	strbuf_release(&cookie_pathname);
+
+	return result;
+}
+
+/*
+ * Mark these cookies as _SEEN and wake up the corresponding client threads.
+ */
+static void with_lock__mark_cookies_seen(struct fsmonitor_daemon_state *state,
+					 const struct string_list *cookie_names)
+{
+	/* assert current thread holding state->main_lock */
+
+	int k;
+	int nr_seen = 0;
+
+	for (k = 0; k < cookie_names->nr; k++) {
+		struct fsmonitor_cookie_item key;
+		struct fsmonitor_cookie_item *cookie;
+
+		key.name = cookie_names->items[k].string;
+		hashmap_entry_init(&key.entry, strhash(key.name));
+
+		cookie = hashmap_get_entry(&state->cookies, &key, entry, NULL);
+		if (cookie) {
+			trace_printf_key(&trace_fsmonitor, "cookie-seen: '%s'",
+					 cookie->name);
+			cookie->result = FCIR_SEEN;
+			nr_seen++;
+		}
+	}
+
+	if (nr_seen)
+		pthread_cond_broadcast(&state->cookies_cond);
+}
+
+/*
+ * Set _ABORT on all pending cookies and wake up all client threads.
+ */
+static void with_lock__abort_all_cookies(struct fsmonitor_daemon_state *state)
+{
+	/* assert current thread holding state->main_lock */
+
+	struct hashmap_iter iter;
+	struct fsmonitor_cookie_item *cookie;
+	int nr_aborted = 0;
+
+	hashmap_for_each_entry(&state->cookies, &iter, cookie, entry) {
+		trace_printf_key(&trace_fsmonitor, "cookie-abort: '%s'",
+				 cookie->name);
+		cookie->result = FCIR_ABORT;
+		nr_aborted++;
+	}
+
+	if (nr_aborted)
+		pthread_cond_broadcast(&state->cookies_cond);
+}
+
 /*
  * Requests to and from a FSMonitor Protocol V2 provider use an opaque
  * "token" as a virtual timestamp.  Clients can request a summary of all
@@ -391,6 +534,9 @@ static void fsmonitor_free_token_data(struct fsmonitor_token_data *token)
  *     We should create a new token and start fresh (as if we just
  *     booted up).
  *
+ * [2] Some of those lost events may have been for cookie files.  We
+ *     should assume the worst and abort them rather letting them starve.
+ *
  * If there are no concurrent threads readering the current token data
  * series, we can free it now.  Otherwise, let the last reader free
  * it.
@@ -412,6 +558,8 @@ static void with_lock__do_force_resync(struct fsmonitor_daemon_state *state)
 	state->current_token_data = new_one;
 
 	fsmonitor_free_token_data(free_me);
+
+	with_lock__abort_all_cookies(state);
 }
 
 void fsmonitor_force_resync(struct fsmonitor_daemon_state *state)
@@ -487,6 +635,8 @@ static int do_handle_client(struct fsmonitor_daemon_state *state,
 	int hash_ret;
 	int do_trivial = 0;
 	int do_flush = 0;
+	int do_cookie = 0;
+	enum fsmonitor_cookie_item_result cookie_result;
 
 	/*
 	 * We expect `command` to be of the form:
@@ -547,6 +697,7 @@ static int do_handle_client(struct fsmonitor_daemon_state *state,
 			 * We have a V2 valid token:
 			 *     "builtin:<token_id>:<seq_nr>"
 			 */
+			do_cookie = 1;
 		}
 	}
 
@@ -555,6 +706,30 @@ static int do_handle_client(struct fsmonitor_daemon_state *state,
 	if (!state->current_token_data)
 		BUG("fsmonitor state does not have a current token");
 
+	/*
+	 * Write a cookie file inside the directory being watched in
+	 * an effort to flush out existing filesystem events that we
+	 * actually care about.  Suspend this client thread until we
+	 * see the filesystem events for this cookie file.
+	 *
+	 * Creating the cookie lets us guarantee that our FS listener
+	 * thread has drained the kernel queue and we are caught up
+	 * with the kernel.
+	 *
+	 * If we cannot create the cookie (or otherwise guarantee that
+	 * we are caught up), we send a trivial response.  We have to
+	 * assume that there might be some very, very recent activity
+	 * on the FS still in flight.
+	 */
+	if (do_cookie) {
+		cookie_result = with_lock__wait_for_cookie(state);
+		if (cookie_result != FCIR_SEEN) {
+			error(_("fsmonitor: cookie_result '%d' != SEEN"),
+			      cookie_result);
+			do_trivial = 1;
+		}
+	}
+
 	if (do_flush)
 		with_lock__do_force_resync(state);
 
@@ -775,7 +950,9 @@ static int handle_client(void *data,
 	return result;
 }
 
-#define FSMONITOR_COOKIE_PREFIX ".fsmonitor-daemon-"
+#define FSMONITOR_DIR           "fsmonitor--daemon"
+#define FSMONITOR_COOKIE_DIR    "cookies"
+#define FSMONITOR_COOKIE_PREFIX (FSMONITOR_DIR "/" FSMONITOR_COOKIE_DIR "/")
 
 enum fsmonitor_path_type fsmonitor_classify_path_workdir_relative(
 	const char *rel)
@@ -928,6 +1105,9 @@ void fsmonitor_publish(struct fsmonitor_daemon_state *state,
 		}
 	}
 
+	if (cookie_names->nr)
+		with_lock__mark_cookies_seen(state, cookie_names);
+
 	pthread_mutex_unlock(&state->main_lock);
 }
 
@@ -1019,7 +1199,9 @@ static int fsmonitor_run_daemon(void)
 
 	memset(&state, 0, sizeof(state));
 
+	hashmap_init(&state.cookies, cookies_cmp, NULL, 0);
 	pthread_mutex_init(&state.main_lock, NULL);
+	pthread_cond_init(&state.cookies_cond, NULL);
 	state.error_code = 0;
 	state.current_token_data = fsmonitor_new_token_data();
 
@@ -1044,6 +1226,44 @@ static int fsmonitor_run_daemon(void)
 		state.nr_paths_watching = 2;
 	}
 
+	/*
+	 * We will write filesystem syncing cookie files into
+	 * <gitdir>/<fsmonitor-dir>/<cookie-dir>/<pid>-<seq>.
+	 *
+	 * The extra layers of subdirectories here keep us from
+	 * changing the mtime on ".git/" or ".git/foo/" when we create
+	 * or delete cookie files.
+	 *
+	 * There have been problems with some IDEs that do a
+	 * non-recursive watch of the ".git/" directory and run a
+	 * series of commands any time something happens.
+	 *
+	 * For example, if we place our cookie files directly in
+	 * ".git/" or ".git/foo/" then a `git status` (or similar
+	 * command) from the IDE will cause a cookie file to be
+	 * created in one of those dirs.  This causes the mtime of
+	 * those dirs to change.  This triggers the IDE's watch
+	 * notification.  This triggers the IDE to run those commands
+	 * again.  And the process repeats and the machine never goes
+	 * idle.
+	 *
+	 * Adding the extra layers of subdirectories prevents the
+	 * mtime of ".git/" and ".git/foo" from changing when a
+	 * cookie file is created.
+	 */
+	strbuf_init(&state.path_cookie_prefix, 0);
+	strbuf_addbuf(&state.path_cookie_prefix, &state.path_gitdir_watch);
+
+	strbuf_addch(&state.path_cookie_prefix, '/');
+	strbuf_addstr(&state.path_cookie_prefix, FSMONITOR_DIR);
+	mkdir(state.path_cookie_prefix.buf, 0777);
+
+	strbuf_addch(&state.path_cookie_prefix, '/');
+	strbuf_addstr(&state.path_cookie_prefix, FSMONITOR_COOKIE_DIR);
+	mkdir(state.path_cookie_prefix.buf, 0777);
+
+	strbuf_addch(&state.path_cookie_prefix, '/');
+
 	/*
 	 * Confirm that we can create platform-specific resources for the
 	 * filesystem listener before we bother starting all the threads.
@@ -1056,6 +1276,7 @@ static int fsmonitor_run_daemon(void)
 	err = fsmonitor_run_daemon_1(&state);
 
 done:
+	pthread_cond_destroy(&state.cookies_cond);
 	pthread_mutex_destroy(&state.main_lock);
 	fsm_listen__dtor(&state);
 
@@ -1063,6 +1284,11 @@ done:
 
 	strbuf_release(&state.path_worktree_watch);
 	strbuf_release(&state.path_gitdir_watch);
+	strbuf_release(&state.path_cookie_prefix);
+
+	/*
+	 * NEEDSWORK: Consider "rm -rf <gitdir>/<fsmonitor-dir>"
+	 */
 
 	return err;
 }
diff --git a/fsmonitor--daemon.h b/fsmonitor--daemon.h
index 20a815d80f8..c16ef095688 100644
--- a/fsmonitor--daemon.h
+++ b/fsmonitor--daemon.h
@@ -45,6 +45,11 @@ struct fsmonitor_daemon_state {
 
 	struct fsmonitor_token_data *current_token_data;
 
+	struct strbuf path_cookie_prefix;
+	pthread_cond_t cookies_cond;
+	int cookie_seq;
+	struct hashmap cookies;
+
 	int error_code;
 	struct fsmonitor_daemon_backend_data *backend_data;
 
-- 
gitgitgadget

[PATCH v5 28/30] fsmonitor: force update index after large responses

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Measure the time taken to apply the FSMonitor query result
to the index and the untracked-cache.

Set the `FSMONITOR_CHANGED` bit on `istate->cache_changed` when
FSMonitor returns a very large repsonse to ensure that the index is
written to disk.

Normally, when the FSMonitor response includes a tracked file, the
index is always updated.  Similarly, the index might be updated when
the response alters the untracked-cache (when enabled).  However, in
cases where neither of those cause the index to be considered changed,
the FSMonitor response is wasted.  Subsequent Git commands will make
requests with the same token and receive the same response.

If that response is very large, performance may suffer.  It would be
more efficient to force update the index now (and the token in the
index extension) in order to reduce the size of the response received
by future commands.

This was observed on Windows after a large checkout.  On Windows, the
kernel emits events for the files that are changed as they are
changed.  However, it might delay events for the containing
directories until the system is more idle (or someone scans the
directory (so it seems)).  The first status following a checkout would
get the list of files.  The subsequent status commands would get the
list of directories as the events trickled out.  But they would never
catch up because the token was not advanced because the index wasn't
updated.

This list of directories caused `wt_status_collect_untracked()` to
unnecessarily spend time actually scanning them during each command.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 fsmonitor.c | 57 ++++++++++++++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 56 insertions(+), 1 deletion(-)

diff --git a/fsmonitor.c b/fsmonitor.c
index 4287aad6bbb..8e3499d0667 100644
--- a/fsmonitor.c
+++ b/fsmonitor.c
@@ -219,6 +219,45 @@ static void fsmonitor_refresh_callback(struct index_state *istate, char *name)
 	untracked_cache_invalidate_path(istate, name, 0);
 }
 
+/*
+ * The number of pathnames that we need to receive from FSMonitor
+ * before we force the index to be updated.
+ *
+ * Note that any pathname within the set of received paths MAY cause
+ * cache-entry or istate flag bits to be updated and thus cause the
+ * index to be updated on disk.
+ *
+ * However, the response may contain many paths (such as ignored
+ * paths) that will not update any flag bits.  And thus not force the
+ * index to be updated.  (This is fine and normal.)  It also means
+ * that the token will not be updated in the FSMonitor index
+ * extension.  So the next Git command will find the same token in the
+ * index, make the same token-relative request, and receive the same
+ * response (plus any newly changed paths).  If this response is large
+ * (and continues to grow), performance could be impacted.
+ *
+ * For example, if the user runs a build and it writes 100K object
+ * files but doesn't modify any source files, the index would not need
+ * to be updated.  The FSMonitor response (after the build and
+ * relative to a pre-build token) might be 5MB.  Each subsequent Git
+ * command will receive that same 100K/5MB response until something
+ * causes the index to be updated.  And `refresh_fsmonitor()` will
+ * have to iterate over those 100K paths each time.
+ *
+ * Performance could be improved if we optionally force update the
+ * index after a very large response and get an updated token into
+ * the FSMonitor index extension.  This should allow subsequent
+ * commands to get smaller and more current responses.
+ *
+ * The value chosen here does not need to be precise.  The index
+ * will be updated automatically the first time the user touches
+ * a tracked file and causes a command like `git status` to
+ * update an mtime to be updated and/or set a flag bit.
+ *
+ * NEEDSWORK: Does this need to be a config value?
+ */
+static int fsmonitor_force_update_threshold = 100;
+
 void refresh_fsmonitor(struct index_state *istate)
 {
 	struct strbuf query_result = STRBUF_INIT;
@@ -362,25 +401,39 @@ apply_results:
 	 *     information and that we should consider everything
 	 *     invalid.  We call this a trivial response.
 	 */
+	trace2_region_enter("fsmonitor", "apply_results", istate->repo);
+
 	if (query_success && !is_trivial) {
 		/*
 		 * Mark all pathnames returned by the monitor as dirty.
 		 *
 		 * This updates both the cache-entries and the untracked-cache.
 		 */
+		int count = 0;
+
 		buf = query_result.buf;
 		for (i = bol; i < query_result.len; i++) {
 			if (buf[i] != '\0')
 				continue;
 			fsmonitor_refresh_callback(istate, buf + bol);
 			bol = i + 1;
+			count++;
 		}
-		if (bol < query_result.len)
+		if (bol < query_result.len) {
 			fsmonitor_refresh_callback(istate, buf + bol);
+			count++;
+		}
 
 		/* Now mark the untracked cache for fsmonitor usage */
 		if (istate->untracked)
 			istate->untracked->use_fsmonitor = 1;
+
+		if (count > fsmonitor_force_update_threshold)
+			istate->cache_changed |= FSMONITOR_CHANGED;
+
+		trace2_data_intmax("fsmonitor", istate->repo, "apply_count",
+				   count);
+
 	} else {
 		/*
 		 * We failed to get a response or received a trivial response,
@@ -409,6 +462,8 @@ apply_results:
 		if (istate->untracked)
 			istate->untracked->use_fsmonitor = 0;
 	}
+	trace2_region_leave("fsmonitor", "apply_results", istate->repo);
+
 	strbuf_release(&query_result);
 
 	/* Now that we've updated istate, save the last_update_token */
-- 
gitgitgadget

[PATCH v5 29/30] t7527: test status with untracked-cache and fsmonitor--daemon

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Create 2x2 test matrix with the untracked-cache and fsmonitor--daemon
features and a series of edits and verify that status output is
identical.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 t/t7527-builtin-fsmonitor.sh | 93 ++++++++++++++++++++++++++++++++++++
 1 file changed, 93 insertions(+)

diff --git a/t/t7527-builtin-fsmonitor.sh b/t/t7527-builtin-fsmonitor.sh
index 5f7b8e54233..0ccbfb9616f 100755
--- a/t/t7527-builtin-fsmonitor.sh
+++ b/t/t7527-builtin-fsmonitor.sh
@@ -162,6 +162,8 @@ test_expect_success 'setup' '
 	.gitignore
 	expect*
 	actual*
+	flush*
+	trace*
 	EOF
 
 	git -c core.fsmonitor=false add . &&
@@ -508,4 +510,95 @@ test_expect_success 'cleanup worktrees' '
 	stop_daemon_delete_repo wt-base
 '
 
+# The next few tests perform arbitrary/contrived file operations and
+# confirm that status is correct.  That is, that the data (or lack of
+# data) from fsmonitor doesn't cause incorrect results.  And doesn't
+# cause incorrect results when the untracked-cache is enabled.
+
+test_lazy_prereq UNTRACKED_CACHE '
+	git update-index --test-untracked-cache
+'
+
+test_expect_success 'Matrix: setup for untracked-cache,fsmonitor matrix' '
+	test_unconfig core.fsmonitor &&
+	git update-index --no-fsmonitor &&
+	test_might_fail git fsmonitor--daemon stop
+'
+
+matrix_clean_up_repo () {
+	git reset --hard HEAD &&
+	git clean -fd
+}
+
+matrix_try () {
+	uc=$1
+	fsm=$2
+	fn=$3
+
+	test_expect_success "Matrix[uc:$uc][fsm:$fsm] $fn" '
+		matrix_clean_up_repo &&
+		$fn &&
+		if test $uc = false && test $fsm = false
+		then
+			git status --porcelain=v1 >.git/expect.$fn
+		else
+			git status --porcelain=v1 >.git/actual.$fn &&
+			test_cmp .git/expect.$fn .git/actual.$fn
+		fi
+	'
+}
+
+uc_values="false"
+test_have_prereq UNTRACKED_CACHE && uc_values="false true"
+for uc_val in $uc_values
+do
+	if test $uc_val = false
+	then
+		test_expect_success "Matrix[uc:$uc_val] disable untracked cache" '
+			git config core.untrackedcache false &&
+			git update-index --no-untracked-cache
+		'
+	else
+		test_expect_success "Matrix[uc:$uc_val] enable untracked cache" '
+			git config core.untrackedcache true &&
+			git update-index --untracked-cache
+		'
+	fi
+
+	fsm_values="false true"
+	for fsm_val in $fsm_values
+	do
+		if test $fsm_val = false
+		then
+			test_expect_success "Matrix[uc:$uc_val][fsm:$fsm_val] disable fsmonitor" '
+				test_unconfig core.fsmonitor &&
+				git update-index --no-fsmonitor &&
+				test_might_fail git fsmonitor--daemon stop
+			'
+		else
+			test_expect_success "Matrix[uc:$uc_val][fsm:$fsm_val] enable fsmonitor" '
+				git config core.fsmonitor true &&
+				git fsmonitor--daemon start &&
+				git update-index --fsmonitor
+			'
+		fi
+
+		matrix_try $uc_val $fsm_val edit_files
+		matrix_try $uc_val $fsm_val delete_files
+		matrix_try $uc_val $fsm_val create_files
+		matrix_try $uc_val $fsm_val rename_files
+		matrix_try $uc_val $fsm_val file_to_directory
+		matrix_try $uc_val $fsm_val directory_to_file
+
+		if test $fsm_val = true
+		then
+			test_expect_success "Matrix[uc:$uc_val][fsm:$fsm_val] disable fsmonitor at end" '
+				test_unconfig core.fsmonitor &&
+				git update-index --no-fsmonitor &&
+				test_might_fail git fsmonitor--daemon stop
+			'
+		fi
+	done
+done
+
 test_done
-- 
gitgitgadget

[PATCH v5 30/30] update-index: convert fsmonitor warnings to advise

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 builtin/update-index.c | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)

diff --git a/builtin/update-index.c b/builtin/update-index.c
index fed24ea1fb6..f94a89d30c8 100644
--- a/builtin/update-index.c
+++ b/builtin/update-index.c
@@ -1227,18 +1227,18 @@ int cmd_update_index(int argc, const char **argv, const char *prefix)
 	if (fsmonitor > 0) {
 		enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
 		if (fsm_mode == FSMONITOR_MODE_DISABLED) {
-			warning(_("core.fsmonitor is unset; "
-				"set it if you really want to "
-				"enable fsmonitor"));
+			advise(_("core.fsmonitor is unset; "
+				 "set it if you really want to "
+				 "enable fsmonitor"));
 		}
 		add_fsmonitor(&the_index);
 		report(_("fsmonitor enabled"));
 	} else if (!fsmonitor) {
 		enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
 		if (fsm_mode > FSMONITOR_MODE_DISABLED)
-			warning(_("core.fsmonitor is set; "
-				"remove it if you really want to "
-				"disable fsmonitor"));
+			advise(_("core.fsmonitor is set; "
+				 "remove it if you really want to "
+				 "disable fsmonitor"));
 		remove_fsmonitor(&the_index);
 		report(_("fsmonitor disabled"));
 	}
-- 
gitgitgadget

Re: [PATCH v5 24/30] t/perf/p7519: speed up test on Windows

[Junio C Hamano]

"Jeff Hostetler via GitGitGadget" <gitgitgadget@gmail.com> writes:

> +touch_files () {
> +	n=$1
> +	d="$n"_files
> +
> +	(cd $d ; test_seq 1 $n | xargs touch )
> +}
> +
>  test_expect_success "one time repo setup" '
>  	# set untrackedCache depending on the environment
>  	if test -n "$GIT_PERF_7519_UNTRACKED_CACHE"
> @@ -119,10 +126,11 @@ test_expect_success "one time repo setup" '
>  	fi &&
>  
>  	mkdir 1_file 10_files 100_files 1000_files 10000_files &&
> -	for i in $(test_seq 1 10); do touch 10_files/$i || return 1; done &&
> -	for i in $(test_seq 1 100); do touch 100_files/$i || return 1; done &&
> -	for i in $(test_seq 1 1000); do touch 1000_files/$i || return 1; done &&
> -	for i in $(test_seq 1 10000); do touch 10000_files/$i || return 1; done &&
> +	touch_files 1 &&

This causes touch_files to chdir to 1_files and run "touch 1" in
there, but because there is no such directory (we have 1_file/
directory, but not 1_files/ directory), it would fail.

> +	touch_files 10 &&
> +	touch_files 100 &&
> +	touch_files 1000 &&
> +	touch_files 10000 &&

Apparently nobody has run this perf script recently since part #2
was posted.

>  	git add 1_file 10_files 100_files 1000_files 10000_files &&

The original introduced at bb7cc7e7 (t/perf/fsmonitor: separate one
time repo initialization, 2020-10-26) created an empty directory 1_file
and without creating anything in it, ran "git add" on it.

If we are not doing anything to 1_file directory anyway, perhaps we
can get rid of it to avoid the breakage in "make perf"?

If we have a chance to reroll this series, we can squash in
something like this, perhaps (it does not deserve to be a separate
step).

--- >8 ---
Subject: [PATCH] p7519: leave 1_file directory empty

The step "t/perf/p7519: speed up test on Windows" in the topic
builtin-fsmonitor-part-2 (not in 'next' yet) attempts to create one
file in 1_files directory, but the original introduced at bb7cc7e7
(t/perf/fsmonitor: separate one time repo initialization,
2020-10-26):

 (1) created 1_file directory,

 (2) left the directory empty, and

 (3) a later test expected (and still expects) that there is nothing
     in the directory.

Revert the behaviour back to what the original wanted to do.

Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 t/perf/p7519-fsmonitor.sh | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/t/perf/p7519-fsmonitor.sh b/t/perf/p7519-fsmonitor.sh
index 9a2288a622..a1c552129c 100755
--- a/t/perf/p7519-fsmonitor.sh
+++ b/t/perf/p7519-fsmonitor.sh
@@ -126,7 +126,7 @@ test_expect_success "one time repo setup" '
 	fi &&
 
 	mkdir 1_file 10_files 100_files 1000_files 10000_files &&
-	touch_files 1 &&
+	: 1_file directory should be left empty &&
 	touch_files 10 &&
 	touch_files 100 &&
 	touch_files 1000 &&
-- 
2.35.1-193-g45fe28c951

Re: [PATCH v5 00/30] Builtin FSMonitor Part 2

[Johannes Schindelin]

Hi Jeff,

On Fri, 11 Feb 2022, Jeff Hostetler via GitGitGadget wrote:

> Here is V5 of Part 2 of my Builtin FSMonitor series. I apologize for the
> delay since V4 that I submitted back in October. (Insert the usual $DayJob
> excuse...)
>
> I have rebased this branch onto the current "master" branch.

Thank you for your tireless work on this. I do see that it requires a ton
of effort on your part and just wanted to let you know that I appreciate
it very much!

> In this version I removed the core.useBuiltinFSMonitor config setting and
> instead extended the existing core.fsmonitor.

I am somewhat surprised that a reviewer suggested this, as it breaks the
common paradigm we use to allow using several Git versions on the same
worktree.

Imagine, for example, that you run a Git version that understands
`core.fsmonitor=true` to imply the built-in FSMonitor, while you _also_
use an IDE that bundles a slightly older Git version that mistakes the
`true` for meaning the executable `true` (which is not a FSMonitor at all,
but its exit code suggests that everything's fine and dandy). The result
would be that the IDE does not see _any_ updates anymore, but nothing
would suggest that anything is wrong.

We can probably warn users about this, and we can also work around the
fact that Git for Windows already uses `core.useBuiltinFSMonitor`, but it
makes me somewhat uneasy nevertheless.

Thank you,
Dscho

Re: [PATCH v5 02/30] fsmonitor-ipc: create client routines for git-fsmonitor--daemon

[Johannes Schindelin]

Hi Jeff,

On Fri, 11 Feb 2022, Jeff Hostetler via GitGitGadget wrote:

> From: Jeff Hostetler <jeffhost@microsoft.com>
>
> Create fsmonitor_ipc__*() client routines to spawn the built-in file
> system monitor daemon and send it an IPC request using the `Simple
> IPC` API.
>
> Stub in empty fsmonitor_ipc__*() functions for unsupported platforms.

This looks very straight-forward, thanks to your design of the Simple IPC.

Just one thing:

> diff --git a/fsmonitor-ipc.c b/fsmonitor-ipc.c
> new file mode 100644
> index 00000000000..01c8c25bf50
> --- /dev/null
> +++ b/fsmonitor-ipc.c
> @@ -0,0 +1,171 @@
> +#include "cache.h"
> +#include "fsmonitor.h"
> +#include "simple-ipc.h"
> +#include "fsmonitor-ipc.h"
> +#include "run-command.h"
> +#include "strbuf.h"
> +#include "trace2.h"
> +
> +#ifdef HAVE_FSMONITOR_DAEMON_BACKEND
> +
> [...]
> +
> +#else
> +
> +/*
> + * A trivial implementation of the fsmonitor_ipc__ API for unsupported
> + * platforms.
> + */
> +
> +int fsmonitor_ipc__is_supported(void)
> +{
> +	return 0;
> +}
> +
> +const char *fsmonitor_ipc__get_path(void)
> +{
> +	return NULL;
> +}
> +
> +enum ipc_active_state fsmonitor_ipc__get_state(void)
> +{
> +	return IPC_STATE__OTHER_ERROR;
> +}
> +
> +int fsmonitor_ipc__send_query(const char *since_token,
> +			      struct strbuf *answer)
> +{
> +	return -1;
> +}
> +
> +int fsmonitor_ipc__send_command(const char *command,
> +				struct strbuf *answer)
> +{
> +	return -1;
> +}
> +
> +#endif

Not a big deal (read: I won't mind if you leave the code as-is), but I
usually find it easier to read the shorter, more trivial arm of
conditional if/else arms first. In this instance, the dummy implementation
for platforms that are not (yet) supported looks more trivial to me.

Thanks,
Dscho

Re: [PATCH v5 03/30] fsmonitor: config settings are repository-specific

[Johannes Schindelin]

Hi Jeff,


On Fri, 11 Feb 2022, Jeff Hostetler via GitGitGadget wrote:

> From: Jeff Hostetler <jeffhost@microsoft.com>
>
> Move fsmonitor config settings to a new and opaque
> `struct fsmonitor_settings` structure.  Add a lazily-loaded pointer
> to this into `struct repo_settings`
>
> Create an `enum fsmonitor_mode` type in `struct fsmonitor_settings` to
> represent the state of fsmonitor.  This lets us represent which, if
> any, fsmonitor provider (hook or IPC) is enabled.
>
> Create `fsm_settings__get_*()` getters to lazily look up fsmonitor-
> related config settings.
>
> Get rid of the `core_fsmonitor` global variable.  Move the code to
> lookup the existing `core.fsmonitor` config value into the fsmonitor
> settings.
>
> Create a hook pathname variable in `struct fsmonitor-settings` and
> only set it when in hook mode.
>
> Extend the definition of `core.fsmonitor` to be either a boolean
> or a hook pathname.  When true, the builtin FSMonitor is used.
> When false or unset, no FSMonitor (neither builtin nor hook) is
> used.
>
> The existing `core_fsmonitor` global variable was used to store the
> pathname to the fsmonitor hook *and* it was used as a boolean to see
> if fsmonitor was enabled.  This dual usage and global visibility leads
> to confusion when we add the IPC-based provider.  So lets hide the
> details in fsmonitor-settings.c and let it decide which provider to
> use in the case of multiple settings.  This avoids cluttering up
> repo-settings.c with these private details.
>
> A future commit in builtin-fsmonitor series will add the ability to
> disqualify worktrees for various reasons, such as being mounted from a
> remote volume, where fsmonitor should not be started.  Having the
> config settings hidden in fsmonitor-settings.c allows such worktree
> restrictions to override the config values used.

Apart from my forward-compatibility concern regarding interpreting
`core.fsmonitor` as a Boolean, this looks good. Just one thing:

> diff --git a/fsmonitor.h b/fsmonitor.h
> index f20d72631d7..f9201411aa7 100644
> --- a/fsmonitor.h
> +++ b/fsmonitor.h
> @@ -3,6 +3,7 @@
>
>  #include "cache.h"
>  #include "dir.h"
> +#include "fsmonitor-settings.h"
>
>  extern struct trace_key trace_fsmonitor;
>
> @@ -57,7 +58,11 @@ int fsmonitor_is_trivial_response(const struct strbuf *query_result);
>   */
>  static inline int is_fsmonitor_refreshed(const struct index_state *istate)
>  {
> -	return !core_fsmonitor || istate->fsmonitor_has_run_once;
> +	struct repository *r = istate->repo ? istate->repo : the_repository;

I see this repeated a few times. Would it maybe make sense to change the
signature of the `fsm_settings__*()` functions to accept an index instead
of a repository?

Ciao,
Dscho

Re: [PATCH v5 10/30] fsmonitor--daemon: implement 'run' command

[Johannes Schindelin]

Hi Jeff,

On Fri, 11 Feb 2022, Jeff Hostetler via GitGitGadget wrote:

> +static int try_to_run_foreground_daemon(void)
> +{
> +	/*
> +	 * Technically, we don't need to probe for an existing daemon
> +	 * process, since we could just call `fsmonitor_run_daemon()`
> +	 * and let it fail if the pipe/socket is busy.
> +	 *
> +	 * However, this method gives us a nicer error message for a
> +	 * common error case.
> +	 */
> +	if (fsmonitor_ipc__get_state() == IPC_STATE__LISTENING)
> +		die("fsmonitor--daemon is already running '%s'",
> +		    the_repository->worktree);
> +
> +	printf(_("running fsmonitor-daemon in '%s'\n"),
> +	       the_repository->worktree);
> +	fflush(stdout);

Do we want to print this to `stderr` instead?

Ciao,
Dscho

Re: [PATCH v5 11/30] fsmonitor--daemon: implement 'start' command

[Johannes Schindelin]

Hi Jeff,

On Fri, 11 Feb 2022, Jeff Hostetler via GitGitGadget wrote:

> +static int try_to_start_background_daemon(void)
> +{
> +	struct child_process cp = CHILD_PROCESS_INIT;
> +	enum start_bg_result sbgr;
> +
> +	/*
> +	 * Before we try to create a background daemon process, see
> +	 * if a daemon process is already listening.  This makes it
> +	 * easier for us to report an already-listening error to the
> +	 * console, since our spawn/daemon can only report the success
> +	 * of creating the background process (and not whether it
> +	 * immediately exited).
> +	 */
> +	if (fsmonitor_ipc__get_state() == IPC_STATE__LISTENING)
> +		die("fsmonitor--daemon is already running '%s'",
> +		    the_repository->worktree);
> +
> +	printf(_("starting fsmonitor-daemon in '%s'\n"),
> +	       the_repository->worktree);
> +	fflush(stdout);

Just like for the patch before, my question whether `stdout` or `stderr`
is preferable here?

> +	cp.git_cmd = 1;
> +
> +	strvec_push(&cp.args, "fsmonitor--daemon");
> +	strvec_push(&cp.args, "run");
> +	strvec_push(&cp.args, "--free-console");

We could call this function `--detached` or `--detach`, too, to indicate
the intention.

I am fine with the code as-is, just wanted to make sure that these
questions are on record ;-)

Ciao,
Dscho

Re: [PATCH v5 24/30] t/perf/p7519: speed up test on Windows

[Jeff Hostetler]

On 2/16/22 8:15 PM, Junio C Hamano wrote:
> "Jeff Hostetler via GitGitGadget" <gitgitgadget@gmail.com> writes:
> 
>> +touch_files () {
>> +	n=$1
>> +	d="$n"_files
>> +
>> +	(cd $d ; test_seq 1 $n | xargs touch )
>> +}
>> +
>>   test_expect_success "one time repo setup" '
>>   	# set untrackedCache depending on the environment
>>   	if test -n "$GIT_PERF_7519_UNTRACKED_CACHE"
>> @@ -119,10 +126,11 @@ test_expect_success "one time repo setup" '
>>   	fi &&
>>   
>>   	mkdir 1_file 10_files 100_files 1000_files 10000_files &&
>> -	for i in $(test_seq 1 10); do touch 10_files/$i || return 1; done &&
>> -	for i in $(test_seq 1 100); do touch 100_files/$i || return 1; done &&
>> -	for i in $(test_seq 1 1000); do touch 1000_files/$i || return 1; done &&
>> -	for i in $(test_seq 1 10000); do touch 10000_files/$i || return 1; done &&
>> +	touch_files 1 &&
> 
> This causes touch_files to chdir to 1_files and run "touch 1" in
> there, but because there is no such directory (we have 1_file/
> directory, but not 1_files/ directory), it would fail.
> 
>> +	touch_files 10 &&
>> +	touch_files 100 &&
>> +	touch_files 1000 &&
>> +	touch_files 10000 &&
> 
> Apparently nobody has run this perf script recently since part #2
> was posted.
> 
>>   	git add 1_file 10_files 100_files 1000_files 10000_files &&
> 
> The original introduced at bb7cc7e7 (t/perf/fsmonitor: separate one
> time repo initialization, 2020-10-26) created an empty directory 1_file
> and without creating anything in it, ran "git add" on it.
> 
> If we are not doing anything to 1_file directory anyway, perhaps we
> can get rid of it to avoid the breakage in "make perf"?
> 
> If we have a chance to reroll this series, we can squash in
> something like this, perhaps (it does not deserve to be a separate
> step).

Good catch!

It looks to me like there was an oversight/typo in the original
89afd5f5ad (t/perf: add fsmonitor perf test for git diff, 2020-10-20).
They created the "1_file" directory and didn't put anything in it.
Then later when they test it, they say "0_files" in the test name
and "1_file" in the "git diff" command.

I don't think it's worth keeping an empty directory here, since there
won't be anything in the index after the add and since the directory
is empty the untracked cache won't have anything to scan.

My version (without the typo) would have created 1 file in the directory
but I don't think that's worth keeping either, since we create thousands
of files in steps right after it.

I'll make a note to remove it.

Jeff



> 
> --- >8 ---
> Subject: [PATCH] p7519: leave 1_file directory empty
> 
> The step "t/perf/p7519: speed up test on Windows" in the topic
> builtin-fsmonitor-part-2 (not in 'next' yet) attempts to create one
> file in 1_files directory, but the original introduced at bb7cc7e7
> (t/perf/fsmonitor: separate one time repo initialization,
> 2020-10-26):
> 
>   (1) created 1_file directory,
> 
>   (2) left the directory empty, and
> 
>   (3) a later test expected (and still expects) that there is nothing
>       in the directory.
> 
> Revert the behaviour back to what the original wanted to do.
> 
> Signed-off-by: Junio C Hamano <gitster@pobox.com>
> ---
>   t/perf/p7519-fsmonitor.sh | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/t/perf/p7519-fsmonitor.sh b/t/perf/p7519-fsmonitor.sh
> index 9a2288a622..a1c552129c 100755
> --- a/t/perf/p7519-fsmonitor.sh
> +++ b/t/perf/p7519-fsmonitor.sh
> @@ -126,7 +126,7 @@ test_expect_success "one time repo setup" '
>   	fi &&
>   
>   	mkdir 1_file 10_files 100_files 1000_files 10000_files &&
> -	touch_files 1 &&
> +	: 1_file directory should be left empty &&
>   	touch_files 10 &&
>   	touch_files 100 &&
>   	touch_files 1000 &&
> 

Re: [PATCH v5 03/30] fsmonitor: config settings are repository-specific

[Jeff Hostetler]

On 2/17/22 11:27 AM, Johannes Schindelin wrote:
> Hi Jeff,
> 
> 
> On Fri, 11 Feb 2022, Jeff Hostetler via GitGitGadget wrote:
> 
>> From: Jeff Hostetler <jeffhost@microsoft.com>
>>
>> Move fsmonitor config settings to a new and opaque
>> `struct fsmonitor_settings` structure.  Add a lazily-loaded pointer
>> to this into `struct repo_settings`
>>
>> Create an `enum fsmonitor_mode` type in `struct fsmonitor_settings` to
>> represent the state of fsmonitor.  This lets us represent which, if
>> any, fsmonitor provider (hook or IPC) is enabled.
>>
>> Create `fsm_settings__get_*()` getters to lazily look up fsmonitor-
>> related config settings.
>>
>> Get rid of the `core_fsmonitor` global variable.  Move the code to
>> lookup the existing `core.fsmonitor` config value into the fsmonitor
>> settings.
>>
>> Create a hook pathname variable in `struct fsmonitor-settings` and
>> only set it when in hook mode.
>>
>> Extend the definition of `core.fsmonitor` to be either a boolean
>> or a hook pathname.  When true, the builtin FSMonitor is used.
>> When false or unset, no FSMonitor (neither builtin nor hook) is
>> used.
>>
>> The existing `core_fsmonitor` global variable was used to store the
>> pathname to the fsmonitor hook *and* it was used as a boolean to see
>> if fsmonitor was enabled.  This dual usage and global visibility leads
>> to confusion when we add the IPC-based provider.  So lets hide the
>> details in fsmonitor-settings.c and let it decide which provider to
>> use in the case of multiple settings.  This avoids cluttering up
>> repo-settings.c with these private details.
>>
>> A future commit in builtin-fsmonitor series will add the ability to
>> disqualify worktrees for various reasons, such as being mounted from a
>> remote volume, where fsmonitor should not be started.  Having the
>> config settings hidden in fsmonitor-settings.c allows such worktree
>> restrictions to override the config values used.
> 
> Apart from my forward-compatibility concern regarding interpreting
> `core.fsmonitor` as a Boolean, this looks good. Just one thing:
> 
>> diff --git a/fsmonitor.h b/fsmonitor.h
>> index f20d72631d7..f9201411aa7 100644
>> --- a/fsmonitor.h
>> +++ b/fsmonitor.h
>> @@ -3,6 +3,7 @@
>>
>>   #include "cache.h"
>>   #include "dir.h"
>> +#include "fsmonitor-settings.h"
>>
>>   extern struct trace_key trace_fsmonitor;
>>
>> @@ -57,7 +58,11 @@ int fsmonitor_is_trivial_response(const struct strbuf *query_result);
>>    */
>>   static inline int is_fsmonitor_refreshed(const struct index_state *istate)
>>   {
>> -	return !core_fsmonitor || istate->fsmonitor_has_run_once;
>> +	struct repository *r = istate->repo ? istate->repo : the_repository;
> 
> I see this repeated a few times. Would it maybe make sense to change the
> signature of the `fsm_settings__*()` functions to accept an index instead
> of a repository?

I think is just me being paranoid -- testing istate->repo for null
and assuming the_repository if necessary.  I'm wondering if it is
always safe to just do

	fsm_mode = fsm_settings__get_mode(istate->repo);

(or maybe put the null check inside the fsm_settings__*() functions.


changing the signature of those fsm_* functions seems wrong since they
are associated with a repo and not an index.

Jeff

Re: [PATCH v5 10/30] fsmonitor--daemon: implement 'run' command

[Jeff Hostetler]

On 2/17/22 11:46 AM, Johannes Schindelin wrote:
> Hi Jeff,
> 
> On Fri, 11 Feb 2022, Jeff Hostetler via GitGitGadget wrote:
> 
>> +static int try_to_run_foreground_daemon(void)
>> +{
>> +	/*
>> +	 * Technically, we don't need to probe for an existing daemon
>> +	 * process, since we could just call `fsmonitor_run_daemon()`
>> +	 * and let it fail if the pipe/socket is busy.
>> +	 *
>> +	 * However, this method gives us a nicer error message for a
>> +	 * common error case.
>> +	 */
>> +	if (fsmonitor_ipc__get_state() == IPC_STATE__LISTENING)
>> +		die("fsmonitor--daemon is already running '%s'",
>> +		    the_repository->worktree);
>> +
>> +	printf(_("running fsmonitor-daemon in '%s'\n"),
>> +	       the_repository->worktree);
>> +	fflush(stdout);
> 
> Do we want to print this to `stderr` instead?


Yes.  I have a patch in part 3 that moves it to stderr
and only emits it if a config setting is present.

fsmonitor--daemon: print start message only if fsmonitor.announceStartup

Wondering if I should squash that in here instead.

Jeff

Re: [PATCH v5 00/30] Builtin FSMonitor Part 2

[Junio C Hamano]

Johannes Schindelin <Johannes.Schindelin@gmx.de> writes:

>> In this version I removed the core.useBuiltinFSMonitor config setting and
>> instead extended the existing core.fsmonitor.
>
> I am somewhat surprised that a reviewer suggested this, as it breaks the
> common paradigm we use to allow using several Git versions on the same
> worktree.

I do not think sharing the same repository with different versions
of Git was considered as a possible source of problems during the
review discussion.

https://lore.kernel.org/git/74282d08-aaeb-0a1e-cad3-1de17d59b4d1@jeffhostetler.com/

I am not saying that we should not consider it; I am just stating
the fact that there was nobody who raised as a potential issue
during the discussion that lead to the cited message.

Re: [PATCH v5 00/30] Builtin FSMonitor Part 2

[Jeff Hostetler]

On 2/17/22 11:06 AM, Johannes Schindelin wrote:
> Hi Jeff,
> 
> On Fri, 11 Feb 2022, Jeff Hostetler via GitGitGadget wrote:
> 
>> Here is V5 of Part 2 of my Builtin FSMonitor series. I apologize for the
>> delay since V4 that I submitted back in October. (Insert the usual $DayJob
>> excuse...)
>>
>> I have rebased this branch onto the current "master" branch.
> 
> Thank you for your tireless work on this. I do see that it requires a ton
> of effort on your part and just wanted to let you know that I appreciate
> it very much!
> 
>> In this version I removed the core.useBuiltinFSMonitor config setting and
>> instead extended the existing core.fsmonitor.
> 
> I am somewhat surprised that a reviewer suggested this, as it breaks the
> common paradigm we use to allow using several Git versions on the same
> worktree.
> 
> Imagine, for example, that you run a Git version that understands
> `core.fsmonitor=true` to imply the built-in FSMonitor, while you _also_
> use an IDE that bundles a slightly older Git version that mistakes the
> `true` for meaning the executable `true` (which is not a FSMonitor at all,
> but its exit code suggests that everything's fine and dandy). The result
> would be that the IDE does not see _any_ updates anymore, but nothing
> would suggest that anything is wrong.
> 
> We can probably warn users about this, and we can also work around the
> fact that Git for Windows already uses `core.useBuiltinFSMonitor`, but it
> makes me somewhat uneasy nevertheless.
> 
> Thank you,
> Dscho
> 

This is a valid concern and I should have thought to mention it when
the suggestion came up on the list.  Yes, extending `core.fsmonitor` to
take a boolean or a path could confuse older clients (like ones bundled
with an IDE, like VS).

My assumption was that since we shipped `core.useBuiltinFSMonitor`
in GFW with an experimental label, that normal users would not be
using it at all and especially not from their IDEs, so it wouldn't
matter.  And experimental features are just that -- experimental
and subject to change.

But your point is valid -- if someone does have the odd hook called
"true" or "1", they'll get an unexpected result.

Jeff

Re: [PATCH v5 12/30] fsmonitor--daemon: add pathname classification

[Johannes Schindelin]

Hi Jeff,

On Fri, 11 Feb 2022, Jeff Hostetler via GitGitGadget wrote:

> +enum fsmonitor_path_type fsmonitor_classify_path_workdir_relative(
> +	const char *rel)
> +{
> +	if (fspathncmp(rel, ".git", 4))
> +		return IS_WORKDIR_PATH;
> +	rel += 4;
> +
> +	if (!*rel)
> +		return IS_DOT_GIT;
> +	if (*rel != '/')
> +		return IS_WORKDIR_PATH; /* e.g. .gitignore */
> +	rel++;
> +
> +	if (!fspathncmp(rel, FSMONITOR_COOKIE_PREFIX,
> +			strlen(FSMONITOR_COOKIE_PREFIX)))
> +		return IS_INSIDE_DOT_GIT_WITH_COOKIE_PREFIX;
> +
> +	return IS_INSIDE_DOT_GIT;
> +}
> +
> +enum fsmonitor_path_type fsmonitor_classify_path_gitdir_relative(
> +	const char *rel)
> +{
> +	if (!fspathncmp(rel, FSMONITOR_COOKIE_PREFIX,
> +			strlen(FSMONITOR_COOKIE_PREFIX)))
> +		return IS_INSIDE_GITDIR_WITH_COOKIE_PREFIX;
> +
> +	return IS_INSIDE_GITDIR;

At first, I was puzzled why this is not `IS_INSIDE_DOT_GIT` as above, but
then...

> [...]
> diff --git a/fsmonitor--daemon.h b/fsmonitor--daemon.h
> index 3009c1a83de..7bbb3a27a1c 100644
> --- a/fsmonitor--daemon.h
> +++ b/fsmonitor--daemon.h
> @@ -30,5 +30,66 @@ struct fsmonitor_daemon_state {
>  	struct ipc_server_data *ipc_server_data;
>  };
>
> +/*
> + * Pathname classifications.
> + *
> + * The daemon classifies the pathnames that it receives from file
> + * system notification events into the following categories and uses
> + * that to decide whether clients are told about them.  (And to watch
> + * for file system synchronization events.)
> + *
> + * The client should only care about paths within the working
> + * directory proper (inside the working directory and not ".git" nor
> + * inside of ".git/").  That is, the client has read the index and is
> + * asking for a list of any paths in the working directory that have
> + * been modified since the last token.  The client does not care about
> + * file system changes within the .git directory (such as new loose
> + * objects or packfiles).  So the client will only receive paths that
> + * are classified as IS_WORKDIR_PATH.
> + *
> + * The daemon uses the IS_DOT_GIT and IS_GITDIR internally to mean the
> + * exact ".git" directory or GITDIR.  If the daemon receives a delete

... I read this, and it started to click.

However, may I suggest to point out that ".git" can also be a _file_, not
only a directory? Otherwise it would not make sense to distinguish between
the `.git` directory/file and `GITDIR`, I'd say.

The rest of the patch looks good to me.

Thanks!
Dscho

Re: [PATCH v5 15/30] compat/fsmonitor/fsm-listen-win32: implement FSMonitor backend on Windows

[Johannes Schindelin]

Hi Jeff,

merely a comment:

On Fri, 11 Feb 2022, Jeff Hostetler via GitGitGadget wrote:

> +/*
> + * Convert the WCHAR path from the notification into UTF8 and
> + * then normalize it.
> + */
> +static int normalize_path_in_utf8(FILE_NOTIFY_INFORMATION *info,
> +				  struct strbuf *normalized_path)
> +{
> +	int reserve;
> +	int len = 0;
> +
> +	strbuf_reset(normalized_path);
> +	if (!info->FileNameLength)
> +		goto normalize;
> +
> +	/*
> +	 * Pre-reserve enough space in the UTF8 buffer for
> +	 * each Unicode WCHAR character to be mapped into a
> +	 * sequence of 2 UTF8 characters.  That should let us
> +	 * avoid ERROR_INSUFFICIENT_BUFFER 99.9+% of the time.
> +	 */
> +	reserve = info->FileNameLength + 1;
> +	strbuf_grow(normalized_path, reserve);
> +
> +	for (;;) {
> +		len = WideCharToMultiByte(CP_UTF8, 0, info->FileName,
> +					  info->FileNameLength / sizeof(WCHAR),
> +					  normalized_path->buf,
> +					  strbuf_avail(normalized_path) - 1,
> +					  NULL, NULL);
> +		if (len > 0)
> +			goto normalize;
> +		if (GetLastError() != ERROR_INSUFFICIENT_BUFFER) {
> +			error("[GLE %ld] could not convert path to UTF-8: '%.*ls'",
> +			      GetLastError(),
> +			      (int)(info->FileNameLength / sizeof(WCHAR)),
> +			      info->FileName);
> +			return -1;
> +		}
> +
> +		strbuf_grow(normalized_path,
> +			    strbuf_avail(normalized_path) + reserve);
> +	}
> +
> +normalize:
> +	strbuf_setlen(normalized_path, len);
> +	return strbuf_normalize_path(normalized_path);
> +}

There are Unicode pages that require quite a few more bytes per wide
character (IIRC it can blow up to six bytes), but it should be good enough
for now, and we can always revisit this easily enough at a later stage.

I really like this patch, as it makes the complexities of handling two
watches and the complexities of handling overlapped (and asynchronous)
results look easy.

Thank you!
Dscho

Re: [PATCH v5 16/30] compat/fsmonitor/fsm-listen-darwin: add macos header files for FSEvent

[Johannes Schindelin]

Hi Jeff,

just a quick note:

On Fri, 11 Feb 2022, Jeff Hostetler via GitGitGadget wrote:

> [...]
> The underlying reason is that GCC (rightfully) objects that an `enum`
> value such as `kAuthorizationExternalFormLength` is not a constant
> (because it is not, the preprocessor has no knowledge of it, only the
> actual C compiler does) and can therefore not be used to define the size
> of a C array.
>
> This is a known problem and tracked in GCC's bug tracker:
> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93082

I just checked and the bug is still marked as "NEW". Therefore we still
need this patch.

Ciao,
Dscho

Re: [PATCH v5 17/30] compat/fsmonitor/fsm-listen-darwin: implement FSEvent listener on MacOS

[Johannes Schindelin]

Hi Jeff,

On Fri, 11 Feb 2022, Jeff Hostetler via GitGitGadget wrote:

> diff --git a/compat/fsmonitor/fsm-listen-darwin.c b/compat/fsmonitor/fsm-listen-darwin.c
> index f424253d3eb..2aefdc14d89 100644
> --- a/compat/fsmonitor/fsm-listen-darwin.c
> +++ b/compat/fsmonitor/fsm-listen-darwin.c
> @@ -1,4 +1,4 @@
> -#if defined(__GNUC__)
> +#ifndef __clang__

This probably wants to be squashed into the previous patch, and while at
it, the corresponding commit message should probably be adjusted:

	We need GCC and clang versions because of compiler and header file
	conflicts.

This should probably say something along the lines "This is only needed
for GCC, clang is fine" instead.

> +static void log_flags_set(const char *path, const FSEventStreamEventFlags flag)
> +{
> +	struct strbuf msg = STRBUF_INIT;
> +
> +	if (flag & kFSEventStreamEventFlagMustScanSubDirs)
> +		strbuf_addstr(&msg, "MustScanSubDirs|");
> +	if (flag & kFSEventStreamEventFlagUserDropped)
> +		strbuf_addstr(&msg, "UserDropped|");
> +	if (flag & kFSEventStreamEventFlagKernelDropped)
> +		strbuf_addstr(&msg, "KernelDropped|");
> +	if (flag & kFSEventStreamEventFlagEventIdsWrapped)
> +		strbuf_addstr(&msg, "EventIdsWrapped|");
> +	if (flag & kFSEventStreamEventFlagHistoryDone)
> +		strbuf_addstr(&msg, "HistoryDone|");
> +	if (flag & kFSEventStreamEventFlagRootChanged)
> +		strbuf_addstr(&msg, "RootChanged|");
> +	if (flag & kFSEventStreamEventFlagMount)
> +		strbuf_addstr(&msg, "Mount|");
> +	if (flag & kFSEventStreamEventFlagUnmount)
> +		strbuf_addstr(&msg, "Unmount|");
> +	if (flag & kFSEventStreamEventFlagItemChangeOwner)
> +		strbuf_addstr(&msg, "ItemChangeOwner|");
> +	if (flag & kFSEventStreamEventFlagItemCreated)
> +		strbuf_addstr(&msg, "ItemCreated|");
> +	if (flag & kFSEventStreamEventFlagItemFinderInfoMod)
> +		strbuf_addstr(&msg, "ItemFinderInfoMod|");
> +	if (flag & kFSEventStreamEventFlagItemInodeMetaMod)
> +		strbuf_addstr(&msg, "ItemInodeMetaMod|");
> +	if (flag & kFSEventStreamEventFlagItemIsDir)
> +		strbuf_addstr(&msg, "ItemIsDir|");
> +	if (flag & kFSEventStreamEventFlagItemIsFile)
> +		strbuf_addstr(&msg, "ItemIsFile|");
> +	if (flag & kFSEventStreamEventFlagItemIsHardlink)
> +		strbuf_addstr(&msg, "ItemIsHardlink|");
> +	if (flag & kFSEventStreamEventFlagItemIsLastHardlink)
> +		strbuf_addstr(&msg, "ItemIsLastHardlink|");
> +	if (flag & kFSEventStreamEventFlagItemIsSymlink)
> +		strbuf_addstr(&msg, "ItemIsSymlink|");
> +	if (flag & kFSEventStreamEventFlagItemModified)
> +		strbuf_addstr(&msg, "ItemModified|");
> +	if (flag & kFSEventStreamEventFlagItemRemoved)
> +		strbuf_addstr(&msg, "ItemRemoved|");
> +	if (flag & kFSEventStreamEventFlagItemRenamed)
> +		strbuf_addstr(&msg, "ItemRenamed|");
> +	if (flag & kFSEventStreamEventFlagItemXattrMod)
> +		strbuf_addstr(&msg, "ItemXattrMod|");
> +	if (flag & kFSEventStreamEventFlagOwnEvent)
> +		strbuf_addstr(&msg, "OwnEvent|");
> +	if (flag & kFSEventStreamEventFlagItemCloned)
> +		strbuf_addstr(&msg, "ItemCloned|");

I cannot think of any more elegant way to do this, either, but I wish
there was a way...

> +/*
> + * NEEDSWORK: Investigate the proper value for the `latency` argument
> + * in the call to `FSEventStreamCreate()`.  I'm not sure that this
> + * needs to be a config setting or just something that we tune after
> + * some testing.

Since this was written, we had ample time to investigate the proper value,
and I think you actually did. Therefore this comment is probably no longer
needed.

> + *
> + * With a latency of 0.1, I was seeing lots of dropped events during
> + * the "touch 100000" files test within t/perf/p7519, but with a
> + * latency of 0.001 I did not see any dropped events.  So the
> + * "correct" value may be somewhere in between.
> + *
> + * https://developer.apple.com/documentation/coreservices/1443980-fseventstreamcreate
> + */

The rest of the patch looks good to me, and since we saw this code working
nicely in practice, I have high confidence in it.

Ciao,
Dscho

Re: [PATCH v5 11/30] fsmonitor--daemon: implement 'start' command

[Johannes Schindelin]

Hi Jeff,

something I missed in my review, but which causes failures in `seen`
because of the interplay with `ac/usage-string-fixups`:

On Fri, 11 Feb 2022, Jeff Hostetler via GitGitGadget wrote:

>  int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
>  {
>  	const char *subcmd;
> +	int free_console = 0;
>
>  	struct option options[] = {
> +		OPT_BOOL(0, "free-console", &free_console, N_("free console")),
>  		OPT_INTEGER(0, "ipc-threads",
>  			    &fsmonitor__ipc_threads,
>  			    N_("use <n> ipc worker threads")),
> +		OPT_INTEGER(0, "start-timeout",
> +			    &fsmonitor__start_timeout_sec,
> +			    N_("Max seconds to wait for background daemon startup")),

Git is about to be stricter about these option usage strings: they are no
longer allowed to start with an upper-case letter. This diff fixes it for me:

-- snip --
From: Johannes Schindelin <johannes.schindelin@gmx.de>
Date: Thu, 24 Feb 2022 15:48:01 +0100
Subject: [PATCH] fixup??? fsmonitor--daemon: implement 'start' command

There is a patch series in `seen` that errors out on option usage
strings starting with a capital letter. Let's avoid that preemptively.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
---
 builtin/fsmonitor--daemon.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index 591433e897df..775e4de5584d 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -1504,7 +1504,7 @@ int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
 			    N_("use <n> ipc worker threads")),
 		OPT_INTEGER(0, "start-timeout",
 			    &fsmonitor__start_timeout_sec,
-			    N_("Max seconds to wait for background daemon startup")),
+			    N_("max seconds to wait for background daemon startup")),

 		OPT_END()
 	};
--
2.35.1.windows.2
-- snap --

Could I ask you to squash this in before you send a new iteration?

Thank you,
Dscho


> +
>  		OPT_END()
>  	};
>

Re: [PATCH v5 10/30] fsmonitor--daemon: implement 'run' command

[Johannes Schindelin]

Hi Jeff,

On Thu, 17 Feb 2022, Jeff Hostetler wrote:

> On 2/17/22 11:46 AM, Johannes Schindelin wrote:
>
> > On Fri, 11 Feb 2022, Jeff Hostetler via GitGitGadget wrote:
> >
> > > +static int try_to_run_foreground_daemon(void)
> > > +{
> > > +	/*
> > > +	 * Technically, we don't need to probe for an existing daemon
> > > +	 * process, since we could just call `fsmonitor_run_daemon()`
> > > +	 * and let it fail if the pipe/socket is busy.
> > > +	 *
> > > +	 * However, this method gives us a nicer error message for a
> > > +	 * common error case.
> > > +	 */
> > > +	if (fsmonitor_ipc__get_state() == IPC_STATE__LISTENING)
> > > +		die("fsmonitor--daemon is already running '%s'",
> > > +		    the_repository->worktree);
> > > +
> > > +	printf(_("running fsmonitor-daemon in '%s'\n"),
> > > +	       the_repository->worktree);
> > > +	fflush(stdout);
> >
> > Do we want to print this to `stderr` instead?
>
> Yes.  I have a patch in part 3 that moves it to stderr
> and only emits it if a config setting is present.
>
> fsmonitor--daemon: print start message only if fsmonitor.announceStartup
>
> Wondering if I should squash that in here instead.

I would like that.

Thanks,
Dscho

Re: [PATCH v5 19/30] help: include fsmonitor--daemon feature flag in version info

[Johannes Schindelin]

Hi Jeff,

On Fri, 11 Feb 2022, Jeff Hostetler via GitGitGadget wrote:

> From: Jeff Hostetler <jeffhost@microsoft.com>
>
> Add the "feature: fsmonitor--daemon" message to the output of
> `git version --build-options`.
>
> The builtin FSMonitor is only available on certain platforms and
> even then only when certain Makefile flags are enabled, so print
> a message in the verbose version output when it is available.
>
> This can be used by test scripts for prereq testing.  Granted, tests
> could just try `git fsmonitor--daemon status` and look for a 128 exit
> code or grep for a "not supported" message on stderr, but this is
> rather obscure.
>
> The main advantage is that the feature message will automatically
> appear in bug reports and other support requests.

It was also used by Scalar before we started the journey to integrate it
into core Git ;-)

Maybe include that, for history buffs like me?

Ciao,
Dscho

Re: [PATCH v5 00/30] Builtin FSMonitor Part 2

[Johannes Schindelin]

Hi Junio,

On Thu, 17 Feb 2022, Junio C Hamano wrote:

> Johannes Schindelin <Johannes.Schindelin@gmx.de> writes:
>
> >> In this version I removed the core.useBuiltinFSMonitor config setting and
> >> instead extended the existing core.fsmonitor.
> >
> > I am somewhat surprised that a reviewer suggested this, as it breaks the
> > common paradigm we use to allow using several Git versions on the same
> > worktree.
>
> I do not think sharing the same repository with different versions
> of Git was considered as a possible source of problems during the
> review discussion.
>
> https://lore.kernel.org/git/74282d08-aaeb-0a1e-cad3-1de17d59b4d1@jeffhostetler.com/
>
> I am not saying that we should not consider it; I am just stating
> the fact that there was nobody who raised as a potential issue
> during the discussion that lead to the cited message.

Just to make sure: I did not intend to insult anyone (and in hindsight I
wish that I had made that clearer).

Ciao,
Dscho

Re: [PATCH v5 03/30] fsmonitor: config settings are repository-specific

[Johannes Schindelin]

Hi Jeff,

On Thu, 17 Feb 2022, Jeff Hostetler wrote:

> On 2/17/22 11:27 AM, Johannes Schindelin wrote:
>
> > On Fri, 11 Feb 2022, Jeff Hostetler via GitGitGadget wrote:
> >
> > > diff --git a/fsmonitor.h b/fsmonitor.h
> > > index f20d72631d7..f9201411aa7 100644
> > > --- a/fsmonitor.h
> > > +++ b/fsmonitor.h
> > > @@ -3,6 +3,7 @@
> > >
> > >   #include "cache.h"
> > >   #include "dir.h"
> > > +#include "fsmonitor-settings.h"
> > >
> > >   extern struct trace_key trace_fsmonitor;
> > >
> > > @@ -57,7 +58,11 @@ int fsmonitor_is_trivial_response(const struct strbuf
> > > *query_result);
> > >   */
> > >   static inline int is_fsmonitor_refreshed(const struct index_state
> > >   *istate)
> > >   {
> > > -	return !core_fsmonitor || istate->fsmonitor_has_run_once;
> > > +	struct repository *r = istate->repo ? istate->repo : the_repository;
> >
> > I see this repeated a few times. Would it maybe make sense to change
> > the signature of the `fsm_settings__*()` functions to accept an index
> > instead of a repository?
>
> I think is just me being paranoid -- testing istate->repo for null
> and assuming the_repository if necessary.

Indeed, this might be a bit too careful here. There is a pretty strong
assumption built into the Git index that there is a repository, after all.

> I'm wondering if it is always safe to just do
>
> 	fsm_mode = fsm_settings__get_mode(istate->repo);
>
> (or maybe put the null check inside the fsm_settings__*() functions.

I would like to try. The test suite should give us enough confidence that
we have dashed all Ts and dotted all Is on the question whether
`istate->repo` is always non-NULL.

> changing the signature of those fsm_* functions seems wrong since they
> are associated with a repo and not an index.

True.

Thank you for the sanity check,
Dscho

Re: [PATCH v5 00/30] Builtin FSMonitor Part 2

[Johannes Schindelin]

Hi Jeff,

On Tue, 22 Feb 2022, Jeff Hostetler wrote:

> On 2/17/22 11:06 AM, Johannes Schindelin wrote:
>
> > On Fri, 11 Feb 2022, Jeff Hostetler via GitGitGadget wrote:
> >
> > > In this version I removed the core.useBuiltinFSMonitor config
> > > setting and instead extended the existing core.fsmonitor.
> >
> > I am somewhat surprised that a reviewer suggested this, as it breaks
> > the common paradigm we use to allow using several Git versions on the
> > same worktree.
> >
> > Imagine, for example, that you run a Git version that understands
> > `core.fsmonitor=true` to imply the built-in FSMonitor, while you
> > _also_ use an IDE that bundles a slightly older Git version that
> > mistakes the `true` for meaning the executable `true` (which is not a
> > FSMonitor at all, but its exit code suggests that everything's fine
> > and dandy). The result would be that the IDE does not see _any_
> > updates anymore, but nothing would suggest that anything is wrong.
> >
> > We can probably warn users about this, and we can also work around the
> > fact that Git for Windows already uses `core.useBuiltinFSMonitor`, but
> > it makes me somewhat uneasy nevertheless.
>
> This is a valid concern and I should have thought to mention it when
> the suggestion came up on the list.  Yes, extending `core.fsmonitor` to
> take a boolean or a path could confuse older clients (like ones bundled
> with an IDE, like VS).
>
> My assumption was that since we shipped `core.useBuiltinFSMonitor`
> in GFW with an experimental label, that normal users would not be
> using it at all and especially not from their IDEs, so it wouldn't
> matter.  And experimental features are just that -- experimental
> and subject to change.
>
> But your point is valid -- if someone does have the odd hook called
> "true" or "1", they'll get an unexpected result.

I wondered about that for a while, and put that to a test last night. I
set `core.fsmonitor = true` and then modified a file and ran `git status`.
Something I did not expect happened: it picked up on the modified file!

It also printed out a warning:

	warning: Empty last update token.

This is the reason why it works: by default, current Git versions assume
that the FSMonitor hook understands the FSMonitor protocol v2, which
starts by the client sending out a token, receiving a new token and then
the paths of the files/directories/symlinks to inspect. Since the program
`true` does _not_ write that token, Git warns that it did not receive a
token and continues as if no FSMonitor had been configured.

So that's good news!

The less good news is that prior to v2.26.0, Git did not support v2 of the
FSMonitor protocol, but only v1. And v1 does not expect such a token. Git
versions between v2.16.0 and v2.26.0 will interpret a successful run of
the `true` executable with an empty output to mean that no files have been
modified.

And indeed, in my tests, after making sure that the Git index had been
refreshed explicitly and then modifying a file and then running `git
status` with v2.16.0, Git did not pick up on the modification.

That's the less good news.

At first I thought that we're pretty safe because nobody should use older
Git versions and enable FSMonitor because FSMonitor protocol v1 is known
to be subject to racy behavior. But then, Git users sometimes do not
completely control which Git versions they use. Take for example Visual
Studio users who also use the Git Bash to work on their worktree. While
their Git Bash might be reasonably recent, Visual Studio comes with its
own embedded Git version. Therefore, a user might want to play with the
built-in FSMonitor in Git Bash, find that it dramatically speeds up
everything (as it does for me, thank you so much!), and not realize that
the Git executable used by Visual Studio totally misinterprets
`core.fsmonitor` to refer to `/usr/bin/true.exe` and then miss any
modifications.

As long as the embedded Git version is at least v2.26.0, Visual Studio
will at least work correctly (because it ignores `true.exe`'s output and
continue as if no FSMonitor had been configured). But as soon as an older
version is used, Git would work incorrectly, without any indication what
is going wrong.

I tried to come up with alternatives (because I _really_ dislike being a
reviewer who only points out what's wrong without any constructive
suggestion how to do it better), and the best alternatives I came up were:

- stick with `core.useBuiltinFSMonitor` as before, or

- use a special value of `core.fsmonitor` that simply is not a valid
  executable name. In 2019, when I worked on the original precursor of the
  built-in FSMonitor (before I had to drop working on FSMonitor
  because of all the security work that went into v2.24.1), I had picked
  `:builtin:` because colons are illegal on Windows, but of _course_ they
  are legal everywhere else. But one thing is not possible, even on Linux:
  to have a trailing slash in an executable name. So something like
  `/builtin-fsmonitor/` would work.

However, after seeing how nicely your latest iteration cleans up the code
by simply interpreting a Boolean value to refer to the built-in FSMonitor,
I _really_ would like to make it work.

Maybe we can declare that it is "safe enough" to rely on new enough Git
versions to be used by users who use multiple Git versions on the same
worktree? They should use _at least_ v2.26.1 anyway, because that one
fixed a rather important vulnerability (CVE-2020-5260)? At least for
Visual Studio, this is true: it ships with Git version 2.33.0.windows.2.

What do you think? Can we somehow make `core.fsmonitor = true` work?

Ciao,
Dscho

Re: [PATCH v5 26/30] fsmonitor--daemon: periodically truncate list of modified files

[Johannes Schindelin]

Hi Jeff,

On Fri, 11 Feb 2022, Jeff Hostetler via GitGitGadget wrote:

> From: Jeff Hostetler <jeffhost@microsoft.com>
>
> Teach fsmonitor--daemon to periodically truncate the list of
> modified files to save some memory.
>
> Clients will ask for the set of changes relative to a token that they
> found in the FSMN index extension in the index.  (This token is like a
> point in time, but different).  Clients will then update the index to
> contain the response token (so that subsequent commands will be
> relative to this new token).
>
> Therefore, the daemon can gradually truncate the in-memory list of
> changed paths as they become obsolete (older than the previous token).
> Since we may have multiple clients making concurrent requests with a
> skew of tokens and clients may be racing to the talk to the daemon,
> we lazily truncate the list.
>
> We introduce a 5 minute delay and truncate batches 5 minutes after
> they are considered obsolete.

I tried to poke holes into this strategy for a while, but failed whichever
way I looked. Meaning: I am now convinced that this strategy is sound.

Thanks,
Dscho

Re: [PATCH v5 27/30] fsmonitor--daemon: use a cookie file to sync with file system

[Johannes Schindelin]

Hi Jeff,

On Fri, 11 Feb 2022, Jeff Hostetler via GitGitGadget wrote:

> From: Jeff Hostetler <jeffhost@microsoft.com>
>
> Teach fsmonitor--daemon client threads to create a cookie file
> inside the .git directory and then wait until FS events for the
> cookie are observed by the FS listener thread.
>
> This helps address the racy nature of file system events by
> blocking the client response until the kernel has drained any
> event backlog.
>
> This is especially important on MacOS where kernel events are
> only issued with a limited frequency.  See the `latency` argument
> of `FSeventStreamCreate()`.  The kernel only signals every `latency`
> seconds, but does not guarantee that the kernel queue is completely
> drained, so we may have to wait more than one interval.  If we
> increase the frequency, the system is more likely to drop events.
> We avoid these issues by having each client thread create a unique
> cookie file and then wait until it is seen in the event stream.

It took a couple iterations of this cookie file business to become
robust... ;-)

About these NEEDSWORKs:

> +	/*
> +	 * Create the cookie file on disk and then wait for a notification
> +	 * that the listener thread has seen it.
> +	 */
> +	fd = open(cookie_pathname.buf, O_WRONLY | O_CREAT | O_EXCL, 0600);
> +	if (fd >= 0) {
> +		close(fd);
> +		unlink(cookie_pathname.buf);
> +
> +		/*
> +		 * NEEDSWORK: This is an infinite wait (well, unless another
> +		 * thread sends us an abort).  I'd like to change this to
> +		 * use `pthread_cond_timedwait()` and return an error/timeout
> +		 * and let the caller do the trivial response thing.
> +		 */
> +		while (cookie->result == FCIR_INIT)
> +			pthread_cond_wait(&state->cookies_cond,
> +					  &state->main_lock);

It would probably make sense to do this at some stage, but since we have
code that has seen quite a bit of real-world testing, I am in favor of
postponing this change to a later date.

> @@ -1063,6 +1284,11 @@ done:
>
>  	strbuf_release(&state.path_worktree_watch);
>  	strbuf_release(&state.path_gitdir_watch);
> +	strbuf_release(&state.path_cookie_prefix);
> +
> +	/*
> +	 * NEEDSWORK: Consider "rm -rf <gitdir>/<fsmonitor-dir>"
> +	 */
>
>  	return err;
>  }

In this instance, I think we can just drop the `NEEDSWORK`: it makes sense
to keep around these directories rather than destroying and re-creating
them over and over again.

Ciao,
Dscho

Re: [PATCH v5 28/30] fsmonitor: force update index after large responses

[Johannes Schindelin]

Hi Jeff,

On Fri, 11 Feb 2022, Jeff Hostetler via GitGitGadget wrote:

> [...]
> + *
> + * NEEDSWORK: Does this need to be a config value?
> + */
> +static int fsmonitor_force_update_threshold = 100;
> +
>  void refresh_fsmonitor(struct index_state *istate)
>  {
>  	struct strbuf query_result = STRBUF_INIT;

We gave this ample opportunity to rise to a need, and no need arose, to
introduce a config option.

Therefore I am in favor of simply dropping the comment.

Ciao,
Dscho

Re: [PATCH v5 00/30] Builtin FSMonitor Part 2

[Junio C Hamano]

Johannes Schindelin <Johannes.Schindelin@gmx.de> writes:

> Just to make sure: I did not intend to insult anyone (and in hindsight I
> wish that I had made that clearer).

It is OK that you are wiser in hindsight.  We all are, and we try to
do better the next time ;-)

Thanks for reminding of the topic.

As a general principle, when introducing a new feature that achieves
the same goal in a new and improved way, it is safer to introduce it
in such a way that users of older implementations that lack the
feature cannot choose it by mistake.

One way to do so is to we reuse the same configuration knob by
adding a new settings value that older implementations would choke
on, the users will be forced to ensure that the older and proven way
is used consistently everywhere, until every tool the user uses are
ready.

For this instance, I think it is OK to split and allow two to
operate on the same data at the same time, because I believe that
both old and new implementation will leave a permanent difference to
the on-disk data that cannot later be reused by the other [*].

But it is an exception than a norm when adding a new thing that
extends an existing feature (as opposed to inventing totally a new
thing that won't overlap with any existing one).  As a general
principle, it is much safer to make sure it breaks (and have users
hold off) when the new setting is given to an old implementation.

    * Side note.  For example, if we introduce the index-v5 feature
    by not reusing the index.version but with index.usev5 variable,
    new implementations that know about the knob would write out v5
    data that existing implementations will not work with.

Also, from the point of view of the longer-term maintenance, of
course not having to deal with orthogonal looking different
configuration variables where newer ones override the older ones
will induce more pain over time.

Re: [PATCH v5 00/30] Builtin FSMonitor Part 2

[Jeff Hostetler]

On 2/24/22 11:22 AM, Johannes Schindelin wrote:
> Hi Jeff,
> 
> On Tue, 22 Feb 2022, Jeff Hostetler wrote:
> 
>> On 2/17/22 11:06 AM, Johannes Schindelin wrote:
>>
>>> On Fri, 11 Feb 2022, Jeff Hostetler via GitGitGadget wrote:
>>>
>>>> In this version I removed the core.useBuiltinFSMonitor config
>>>> setting and instead extended the existing core.fsmonitor.
>>>
>>> I am somewhat surprised that a reviewer suggested this, as it breaks
>>> the common paradigm we use to allow using several Git versions on the
>>> same worktree.
>>>
>>> Imagine, for example, that you run a Git version that understands
>>> `core.fsmonitor=true` to imply the built-in FSMonitor, while you
>>> _also_ use an IDE that bundles a slightly older Git version that
>>> mistakes the `true` for meaning the executable `true` (which is not a
>>> FSMonitor at all, but its exit code suggests that everything's fine
>>> and dandy). The result would be that the IDE does not see _any_
>>> updates anymore, but nothing would suggest that anything is wrong.
>>>
>>> We can probably warn users about this, and we can also work around the
>>> fact that Git for Windows already uses `core.useBuiltinFSMonitor`, but
>>> it makes me somewhat uneasy nevertheless.
>>
>> This is a valid concern and I should have thought to mention it when
>> the suggestion came up on the list.  Yes, extending `core.fsmonitor` to
>> take a boolean or a path could confuse older clients (like ones bundled
>> with an IDE, like VS).
>>
>> My assumption was that since we shipped `core.useBuiltinFSMonitor`
>> in GFW with an experimental label, that normal users would not be
>> using it at all and especially not from their IDEs, so it wouldn't
>> matter.  And experimental features are just that -- experimental
>> and subject to change.
>>
>> But your point is valid -- if someone does have the odd hook called
>> "true" or "1", they'll get an unexpected result.
> 
> I wondered about that for a while, and put that to a test last night. I
> set `core.fsmonitor = true` and then modified a file and ran `git status`.
> Something I did not expect happened: it picked up on the modified file!
> 
> It also printed out a warning:
> 
> 	warning: Empty last update token.
> 
> This is the reason why it works: by default, current Git versions assume
> that the FSMonitor hook understands the FSMonitor protocol v2, which
> starts by the client sending out a token, receiving a new token and then
> the paths of the files/directories/symlinks to inspect. Since the program
> `true` does _not_ write that token, Git warns that it did not receive a
> token and continues as if no FSMonitor had been configured.
> 
> So that's good news!
> 
> The less good news is that prior to v2.26.0, Git did not support v2 of the
> FSMonitor protocol, but only v1. And v1 does not expect such a token. Git
> versions between v2.16.0 and v2.26.0 will interpret a successful run of
> the `true` executable with an empty output to mean that no files have been
> modified.
> 
> And indeed, in my tests, after making sure that the Git index had been
> refreshed explicitly and then modifying a file and then running `git
> status` with v2.16.0, Git did not pick up on the modification.
> 
> That's the less good news.
> 
> At first I thought that we're pretty safe because nobody should use older
> Git versions and enable FSMonitor because FSMonitor protocol v1 is known
> to be subject to racy behavior. But then, Git users sometimes do not
> completely control which Git versions they use. Take for example Visual
> Studio users who also use the Git Bash to work on their worktree. While
> their Git Bash might be reasonably recent, Visual Studio comes with its
> own embedded Git version. Therefore, a user might want to play with the
> built-in FSMonitor in Git Bash, find that it dramatically speeds up
> everything (as it does for me, thank you so much!), and not realize that
> the Git executable used by Visual Studio totally misinterprets
> `core.fsmonitor` to refer to `/usr/bin/true.exe` and then miss any
> modifications.
> 
> As long as the embedded Git version is at least v2.26.0, Visual Studio
> will at least work correctly (because it ignores `true.exe`'s output and
> continue as if no FSMonitor had been configured). But as soon as an older
> version is used, Git would work incorrectly, without any indication what
> is going wrong.
> 
> I tried to come up with alternatives (because I _really_ dislike being a
> reviewer who only points out what's wrong without any constructive
> suggestion how to do it better), and the best alternatives I came up were:
> 
> - stick with `core.useBuiltinFSMonitor` as before, or
> 
> - use a special value of `core.fsmonitor` that simply is not a valid
>    executable name. In 2019, when I worked on the original precursor of the
>    built-in FSMonitor (before I had to drop working on FSMonitor
>    because of all the security work that went into v2.24.1), I had picked
>    `:builtin:` because colons are illegal on Windows, but of _course_ they
>    are legal everywhere else. But one thing is not possible, even on Linux:
>    to have a trailing slash in an executable name. So something like
>    `/builtin-fsmonitor/` would work.
> 
> However, after seeing how nicely your latest iteration cleans up the code
> by simply interpreting a Boolean value to refer to the built-in FSMonitor,
> I _really_ would like to make it work.
> 
> Maybe we can declare that it is "safe enough" to rely on new enough Git
> versions to be used by users who use multiple Git versions on the same
> worktree? They should use _at least_ v2.26.1 anyway, because that one
> fixed a rather important vulnerability (CVE-2020-5260)? At least for
> Visual Studio, this is true: it ships with Git version 2.33.0.windows.2.
> 
> What do you think? Can we somehow make `core.fsmonitor = true` work?
> 
> Ciao,
> Dscho
> 

Thanks for the testing and background information here!

I agree.  I would like to keep the current

     "core.fsmonitor = <bool> | <path>"

usage that I have in V5.

It cleaned up things very nicely and it got rid of the somewhat awkward
usage of having "core.useBuiltinFSMonitor" override the existing
"core.fsmonitor" setting.

It is unfortunate that it might cause a breakage for users who are
*also* running a Git version between 2.16 ... 2.26.  I have to wonder
if it wouldn't be better to spend our energy documenting that users
should upgrade, rather than trying to support interop with them.

The 2.16 ... 2.26 versions don't have the security fix referenced
above.

The 2.16 ... 2.26 versions are based upon the V1 FSMonitor protocol
and don't have fixes for several serious bugs or races in the original
implementation:

* 398a3b0899 (fsmonitor: force a refresh after the index was discarded, 
2019-05-07)
* 679f2f9fdd (unpack-trees: skip stat on fsmonitor-valid files, 2019-11-20)
* dfaed02862 (fsmonitor: update documentation for hook version and 
watchman hooks, 2020-01-23)

That last fix added the V2 FSMonitor protocol.  This is used by both
the hook- and the IPC-based providers.

So anything still using the V1 FSMonitor protocol is going to unreliable
and racy and users should not use it, so I don't think it is worth the 
effort to complicate our current solution to maintain compatibility.
(I hate to say that, but they just shouldn't be using V1 any more.)


On a slight tangent, the current code (before my patch series) does
support a "core.fsmonitorhookversion" to allow the client to talk to
a V1 or V2 provider explicitly (vs the default of trying V2 and then
trying V1).  The IPC implementation does not use this config setting,
but I could see adding something to emit a warning if it was set to
1 when using the builtin FSMonitor.  This might help users who are
*also* running a Git version between 2.26 and 2.35 to understand the
fallback after the true.exe warning that Johannes described.


On another slight tangent, I'm wondering if we want to officially
deprecate the V1 hook code and/or remove support for it from the code.


Jeff

Re: [PATCH v5 00/30] Builtin FSMonitor Part 2

[Johannes Schindelin]

Hi Jeff,

On Thu, 24 Feb 2022, Jeff Hostetler wrote:

> On 2/24/22 11:22 AM, Johannes Schindelin wrote:
>
> > On Tue, 22 Feb 2022, Jeff Hostetler wrote:
> >
> > > On 2/17/22 11:06 AM, Johannes Schindelin wrote:
> > >
> > > > On Fri, 11 Feb 2022, Jeff Hostetler via GitGitGadget wrote:
> > > >
> > > > > In this version I removed the core.useBuiltinFSMonitor config
> > > > > setting and instead extended the existing core.fsmonitor.
> > > >
> > > > I am somewhat surprised that a reviewer suggested this, as it
> > > > breaks the common paradigm we use to allow using several Git
> > > > versions on the same worktree.
> > [...]
> >
> > I wondered about that for a while, and put that to a test last night.
> > I set `core.fsmonitor = true` and then modified a file and ran `git
> > status`. Something I did not expect happened: it picked up on the
> > modified file!
> >
> > [... describing how FSMonitor protocol v1 is affected...]
> >
> > However, after seeing how nicely your latest iteration cleans up the
> > code by simply interpreting a Boolean value to refer to the built-in
> > FSMonitor, I _really_ would like to make it work.
> >
> > Maybe we can declare that it is "safe enough" to rely on new enough
> > Git versions to be used by users who use multiple Git versions on the
> > same worktree? They should use _at least_ v2.26.1 anyway, because that
> > one fixed a rather important vulnerability (CVE-2020-5260)? At least
> > for Visual Studio, this is true: it ships with Git version
> > 2.33.0.windows.2.
> >
> > What do you think? Can we somehow make `core.fsmonitor = true` work?
>
> [...]
>
> I agree.  I would like to keep the current
>
>     "core.fsmonitor = <bool> | <path>"
>
> usage that I have in V5.
>
> It cleaned up things very nicely and it got rid of the somewhat awkward
> usage of having "core.useBuiltinFSMonitor" override the existing
> "core.fsmonitor" setting.

Yes!

> It is unfortunate that it might cause a breakage for users who are
> *also* running a Git version between 2.16 ... 2.26.  I have to wonder
> if it wouldn't be better to spend our energy documenting that users
> should upgrade, rather than trying to support interop with them.

That's a good point: I guess if you added a comment to the documentation
of `core.fsmonitor = true`, that should be good enough.

> [...]
>
> So anything still using the V1 FSMonitor protocol is going to unreliable
> and racy and users should not use it, so I don't think it is worth the effort
> to complicate our current solution to maintain compatibility.
> (I hate to say that, but they just shouldn't be using V1 any more.)

That's a really good point.

> On a slight tangent, the current code (before my patch series) does
> support a "core.fsmonitorhookversion" to allow the client to talk to
> a V1 or V2 provider explicitly (vs the default of trying V2 and then
> trying V1).  The IPC implementation does not use this config setting,
> but I could see adding something to emit a warning if it was set to
> 1 when using the builtin FSMonitor.  This might help users who are
> *also* running a Git version between 2.26 and 2.35 to understand the
> fallback after the true.exe warning that Johannes described.

How about making it an error instead? That should really be helpful: if
`core.fsmonitor = true` and `core.fsmonitorHookVersion = 1`, just error
out. That way, users will more likely fall into the pit of success.

> On another slight tangent, I'm wondering if we want to officially
> deprecate the V1 hook code and/or remove support for it from the code.

Oooh! That's _also_ a good point.

Maybe we can keep this deprecation out of this here patch series, though.
It would be good to get this finished and into `next`, I think.

I spent some time (in two separate thrusts) to review the entire patch
series, and hope that my feedback was useful to you.

In particular with your idea to document the incompatibilities of
`core.fsmonitor = true` with Git v2.16.0..v2.26.0, I am really eager to
see (the next iteration of) this patch series advance to the `next`
branch, and then into an official Git version so that more users can
benefit from it.

Thank you for all your work on this,
Dscho

Re: [PATCH v5 06/30] fsmonitor--daemon: add a built-in fsmonitor daemon

[Ævar Arnfjörð Bjarmason]

On Fri, Feb 11 2022, Jeff Hostetler via GitGitGadget wrote:

> From: Jeff Hostetler <jeffhost@microsoft.com>
>
> Create a built-in file system monitoring daemon that can be used by
> the existing `fsmonitor` feature (protocol API and index extension)
> to improve the performance of various Git commands, such as `status`.
>
> The `fsmonitor--daemon` feature builds upon the `Simple IPC` API and
> provides an alternative to hook access to existing fsmonitors such
> as `watchman`.
>
> This commit merely adds the new command without any functionality.
>
> Co-authored-by: Johannes Schindelin <johannes.schindelin@gmx.de>
> Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
> ---
>  .gitignore                  |  1 +
>  Makefile                    |  1 +
>  builtin.h                   |  1 +
>  builtin/fsmonitor--daemon.c | 46 +++++++++++++++++++++++++++++++++++++
>  git.c                       |  1 +
>  5 files changed, 50 insertions(+)
>  create mode 100644 builtin/fsmonitor--daemon.c
>
> diff --git a/.gitignore b/.gitignore
> index f817c509ec0..e81de1063a4 100644
> --- a/.gitignore
> +++ b/.gitignore
> @@ -72,6 +72,7 @@
>  /git-format-patch
>  /git-fsck
>  /git-fsck-objects
> +/git-fsmonitor--daemon
>  /git-gc
>  /git-get-tar-commit-id
>  /git-grep
> diff --git a/Makefile b/Makefile
> index 9943f0f7c11..3b7a3f88b50 100644
> --- a/Makefile
> +++ b/Makefile
> @@ -1106,6 +1106,7 @@ BUILTIN_OBJS += builtin/fmt-merge-msg.o
>  BUILTIN_OBJS += builtin/for-each-ref.o
>  BUILTIN_OBJS += builtin/for-each-repo.o
>  BUILTIN_OBJS += builtin/fsck.o
> +BUILTIN_OBJS += builtin/fsmonitor--daemon.o
>  BUILTIN_OBJS += builtin/gc.o
>  BUILTIN_OBJS += builtin/get-tar-commit-id.o
>  BUILTIN_OBJS += builtin/grep.o
> diff --git a/builtin.h b/builtin.h
> index 83379f3832c..40e9ecc8485 100644
> --- a/builtin.h
> +++ b/builtin.h
> @@ -159,6 +159,7 @@ int cmd_for_each_ref(int argc, const char **argv, const char *prefix);
>  int cmd_for_each_repo(int argc, const char **argv, const char *prefix);
>  int cmd_format_patch(int argc, const char **argv, const char *prefix);
>  int cmd_fsck(int argc, const char **argv, const char *prefix);
> +int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix);
>  int cmd_gc(int argc, const char **argv, const char *prefix);
>  int cmd_get_tar_commit_id(int argc, const char **argv, const char *prefix);
>  int cmd_grep(int argc, const char **argv, const char *prefix);
> diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
> new file mode 100644
> index 00000000000..f0498793379
> --- /dev/null
> +++ b/builtin/fsmonitor--daemon.c
> @@ -0,0 +1,46 @@
> +#include "builtin.h"
> +#include "config.h"
> +#include "parse-options.h"
> +#include "fsmonitor.h"
> +#include "fsmonitor-ipc.h"
> +#include "simple-ipc.h"
> +#include "khash.h"
> +
> +static const char * const builtin_fsmonitor__daemon_usage[] = {
> +	NULL
> +};
> +
> +#ifdef HAVE_FSMONITOR_DAEMON_BACKEND
> +
> +int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
> +{
> +	const char *subcmd;
> +
> +	struct option options[] = {
> +		OPT_END()
> +	};
> +
> +	git_config(git_default_config, NULL);
> +
> +	argc = parse_options(argc, argv, prefix, options,
> +			     builtin_fsmonitor__daemon_usage, 0);
> +	if (argc != 1)
> +		usage_with_options(builtin_fsmonitor__daemon_usage, options);
> +	subcmd = argv[0];
> +
> +	die(_("Unhandled subcommand '%s'"), subcmd);
> +}
> +
> +#else
> +int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
> +{
> +	struct option options[] = {
> +		OPT_END()
> +	};
> +
> +	if (argc == 2 && !strcmp(argv[1], "-h"))
> +		usage_with_options(builtin_fsmonitor__daemon_usage, options);
> +
> +	die(_("fsmonitor--daemon not supported on this platform"));
> +}
> +#endif
> diff --git a/git.c b/git.c
> index 340665d4a04..a8b44d9b587 100644
> --- a/git.c
> +++ b/git.c
> @@ -536,6 +536,7 @@ static struct cmd_struct commands[] = {
>  	{ "format-patch", cmd_format_patch, RUN_SETUP },
>  	{ "fsck", cmd_fsck, RUN_SETUP },
>  	{ "fsck-objects", cmd_fsck, RUN_SETUP },
> +	{ "fsmonitor--daemon", cmd_fsmonitor__daemon, RUN_SETUP },
>  	{ "gc", cmd_gc, RUN_SETUP },
>  	{ "get-tar-commit-id", cmd_get_tar_commit_id, NO_PARSEOPT },
>  	{ "grep", cmd_grep, RUN_SETUP_GENTLY },


I brought this up in another thread in how this series interacts with
another, but this patch below on top of "seen" would allow you to catch
parse_options() BUGs on Linux, even if you don't have a no-OSX
non-Windows backend yet:
	
	diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
	index 591433e897d..62c0b1d486b 100644
	--- a/builtin/fsmonitor--daemon.c
	+++ b/builtin/fsmonitor--daemon.c
	@@ -18,7 +18,6 @@ static const char * const builtin_fsmonitor__daemon_usage[] = {
	 	NULL
	 };
	 
	-#ifdef HAVE_FSMONITOR_DAEMON_BACKEND
	 /*
	  * Global state loaded from config.
	  */
	@@ -63,6 +62,7 @@ static int fsmonitor_config(const char *var, const char *value, void *cb)
	 
	 	return git_default_config(var, value, cb);
	 }
	+#ifdef HAVE_FSMONITOR_DAEMON_BACKEND
	 
	 /*
	  * Acting as a CLIENT.
	@@ -1492,6 +1492,8 @@ static int try_to_start_background_daemon(void)
	 	}
	 }
	 
	+#endif
	+
	 int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
	 {
	 	const char *subcmd;
	@@ -1532,6 +1534,7 @@ int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
	 		return -1;
	 	}
	 
	+#ifdef HAVE_FSMONITOR_DAEMON_BACKEND
	 	if (!strcmp(subcmd, "start"))
	 		return !!try_to_start_background_daemon();
	 
	@@ -1543,20 +1546,8 @@ int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
	 
	 	if (!strcmp(subcmd, "status"))
	 		return !!do_as_client__status();
	-
	 	die(_("Unhandled subcommand '%s'"), subcmd);
	-}
	-
	 #else
	-int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
	-{
	-	struct option options[] = {
	-		OPT_END()
	-	};
	-
	-	if (argc == 2 && !strcmp(argv[1], "-h"))
	-		usage_with_options(builtin_fsmonitor__daemon_usage, options);
	-
	 	die(_("fsmonitor--daemon not supported on this platform"));
	-}
	 #endif
	+}

I.e. we can be a less zealous when setting the ifdef boundaries, and
it's actually less code as well.

Re: [PATCH v5 23/30] t/helper/test-chmtime: skip directories on Windows

[Tao Klerks]

Random follow-up on this: in message/patch
76b6216281e3463821e650495f3090c677905f73.1646041236.git.gitgitgadget@gmail.com
I propose a fix to this issue. If accepted, that fix (and related
changes to rely
more heavily on chmtime in t7063) will cause *this* change to cause test
failures.

If that patch is accepted, this commit will simply need to be dropped
as far as I understand.



On Fri, Feb 11, 2022 at 9:57 PM Jeff Hostetler via GitGitGadget
<gitgitgadget@gmail.com> wrote:
>
> From: Jeff Hostetler <jeffhost@microsoft.com>
>
> Teach `test-tool.exe chmtime` to ignore errors when setting the mtime
> on a directory on Windows.
>
> NEEDSWORK: The Windows version of `utime()` (aka `mingw_utime()`) does
> not properly handle directories because it uses `_wopen()`.  It should
> be converted to using `CreateFileW()` and backup semantics at a minimum.
> Since I'm already in the middle of a large patch series, I did not want
> to destabilize other callers of `utime()` right now.  The problem has
> only been observed in the t/perf/p7519 test when the test repo contains
> an empty directory on disk.
>
> Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
> ---
>  t/helper/test-chmtime.c | 15 +++++++++++++++
>  1 file changed, 15 insertions(+)
>
> diff --git a/t/helper/test-chmtime.c b/t/helper/test-chmtime.c
> index 524b55ca496..dc28890a183 100644
> --- a/t/helper/test-chmtime.c
> +++ b/t/helper/test-chmtime.c
> @@ -134,6 +134,21 @@ int cmd__chmtime(int argc, const char **argv)
>                 }
>
>                 if (utb.modtime != sb.st_mtime && utime(argv[i], &utb) < 0) {
> +#ifdef GIT_WINDOWS_NATIVE
> +                       if (S_ISDIR(sb.st_mode)) {
> +                               /*
> +                                * NEEDSWORK: The Windows version of `utime()`
> +                                * (aka `mingw_utime()`) does not correctly
> +                                * handle directory arguments, since it uses
> +                                * `_wopen()`.  Ignore it for now since this
> +                                * is just a test.
> +                                */
> +                               fprintf(stderr,
> +                                       ("Failed to modify time on directory %s. "
> +                                        "Skipping\n"), argv[i]);
> +                               continue;
> +                       }
> +#endif
>                         fprintf(stderr, "Failed to modify time on %s: %s\n",
>                                 argv[i], strerror(errno));
>                         return 1;
> --
> gitgitgadget
>

Re: [PATCH v5 23/30] t/helper/test-chmtime: skip directories on Windows

[Jeff Hostetler]

On 2/28/22 4:43 AM, Tao Klerks wrote:
> Random follow-up on this: in message/patch
> 76b6216281e3463821e650495f3090c677905f73.1646041236.git.gitgitgadget@gmail.com
> I propose a fix to this issue. If accepted, that fix (and related
> changes to rely
> more heavily on chmtime in t7063) will cause *this* change to cause test
> failures.
> 
> If that patch is accepted, this commit will simply need to be dropped
> as far as I understand.
> 

Thanks for the note.  Yes, with your change we should be able to
drop or revert this commit.

Thanks
Jeff

Re: [PATCH v5 23/30] t/helper/test-chmtime: skip directories on Windows

[Junio C Hamano]

Jeff Hostetler <git@jeffhostetler.com> writes:

> On 2/28/22 4:43 AM, Tao Klerks wrote:
>> Random follow-up on this: in message/patch
>> 76b6216281e3463821e650495f3090c677905f73.1646041236.git.gitgitgadget@gmail.com
>> I propose a fix to this issue. If accepted, that fix (and related
>> changes to rely
>> more heavily on chmtime in t7063) will cause *this* change to cause test
>> failures.
>> If that patch is accepted, this commit will simply need to be
>> dropped
>> as far as I understand.
>> 
>
> Thanks for the note.  Yes, with your change we should be able to
> drop or revert this commit.

That sounds nice.  The fewer differences among platforms we have to
maintain, the easier.

Re: [PATCH v5 06/30] fsmonitor--daemon: add a built-in fsmonitor daemon

[Jeff Hostetler]

On 2/25/22 5:46 PM, Ævar Arnfjörð Bjarmason wrote:
> 
> On Fri, Feb 11 2022, Jeff Hostetler via GitGitGadget wrote:
> 
>> From: Jeff Hostetler <jeffhost@microsoft.com>
>>
>> Create a built-in file system monitoring daemon that can be used by
>> the existing `fsmonitor` feature (protocol API and index extension)
>> to improve the performance of various Git commands, such as `status`.
>>
>> The `fsmonitor--daemon` feature builds upon the `Simple IPC` API and
>> provides an alternative to hook access to existing fsmonitors such
>> as `watchman`.
>>
>> This commit merely adds the new command without any functionality.
>>
[...]
>> +
>> +#ifdef HAVE_FSMONITOR_DAEMON_BACKEND
>> +
>> +int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
>> +{
>> +	const char *subcmd;
>> +
>> +	struct option options[] = {
>> +		OPT_END()
>> +	};
>> +
>> +	git_config(git_default_config, NULL);
>> +
>> +	argc = parse_options(argc, argv, prefix, options,
>> +			     builtin_fsmonitor__daemon_usage, 0);
>> +	if (argc != 1)
>> +		usage_with_options(builtin_fsmonitor__daemon_usage, options);
>> +	subcmd = argv[0];
>> +
>> +	die(_("Unhandled subcommand '%s'"), subcmd);
>> +}
>> +
>> +#else
>> +int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
>> +{
>> +	struct option options[] = {
>> +		OPT_END()
>> +	};
>> +
>> +	if (argc == 2 && !strcmp(argv[1], "-h"))
>> +		usage_with_options(builtin_fsmonitor__daemon_usage, options);
>> +
>> +	die(_("fsmonitor--daemon not supported on this platform"));
>> +}
>> +#endif
> 
> 
> I brought this up in another thread in how this series interacts with
> another, but this patch below on top of "seen" would allow you to catch
> parse_options() BUGs on Linux, even if you don't have a no-OSX
> non-Windows backend yet:
> 	
> 	diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
> 	index 591433e897d..62c0b1d486b 100644
> 	--- a/builtin/fsmonitor--daemon.c
> 	+++ b/builtin/fsmonitor--daemon.c
> 	@@ -18,7 +18,6 @@ static const char * const builtin_fsmonitor__daemon_usage[] = {
> 	 	NULL
> 	 };
> 	
> 	-#ifdef HAVE_FSMONITOR_DAEMON_BACKEND
> 	 /*
> 	  * Global state loaded from config.
> 	  */
> 	@@ -63,6 +62,7 @@ static int fsmonitor_config(const char *var, const char *value, void *cb)
> 	
> 	 	return git_default_config(var, value, cb);
> 	 }
> 	+#ifdef HAVE_FSMONITOR_DAEMON_BACKEND
> 	
> 	 /*
> 	  * Acting as a CLIENT.
> 	@@ -1492,6 +1492,8 @@ static int try_to_start_background_daemon(void)
> 	 	}
> 	 }
> 	
> 	+#endif
> 	+
> 	 int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
> 	 {
> 	 	const char *subcmd;
> 	@@ -1532,6 +1534,7 @@ int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
> 	 		return -1;
> 	 	}
> 	
> 	+#ifdef HAVE_FSMONITOR_DAEMON_BACKEND
> 	 	if (!strcmp(subcmd, "start"))
> 	 		return !!try_to_start_background_daemon();
> 	
> 	@@ -1543,20 +1546,8 @@ int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
> 	
> 	 	if (!strcmp(subcmd, "status"))
> 	 		return !!do_as_client__status();
> 	-
> 	 	die(_("Unhandled subcommand '%s'"), subcmd);
> 	-}
> 	-
> 	 #else
> 	-int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
> 	-{
> 	-	struct option options[] = {
> 	-		OPT_END()
> 	-	};
> 	-
> 	-	if (argc == 2 && !strcmp(argv[1], "-h"))
> 	-		usage_with_options(builtin_fsmonitor__daemon_usage, options);
> 	-
> 	 	die(_("fsmonitor--daemon not supported on this platform"));
> 	-}
> 	 #endif
> 	+}
> 
> I.e. we can be a less zealous when setting the ifdef boundaries, and
> it's actually less code as well.
> 

Yes, it would be possible to distribute the ifdef throughout the file
and avoid duplicating the function declaration, but I'm not sure that
that adds any clarity or readability.

In my version, I have a stub version of the cmd_fsmonitor__daemon()
function and it is very clear that it does nothing when the feature
is not supported on a platform.  The rest of the source file is
concerned with supporting the feature.  And no interweaving of ifdefs
throughout the file is required.

Your version sets us up for future problems inside the body of the
cmd_ function.  For example, any static function called in the
supported portion of the function would also need to be ifdef'd
(as you have indicated).  But any local variables needed by the
supported portion would need to be declared at the top of the
function and also ifdef'd -- or we'd need to indent the entire body
of the supported portion inside another level of { }.  None of this
adds clarity.  (Just to avoid an 11 line stub function.)

Finally, I'm not sure how much value there is in being able to catch
parse_options() BUGs on Linux (or any other yet-to-be-supported
platform).  The daemon isn't supported and dies immediately. I'm not
sure that forcing the user to properly compose any arguments before
we just call die() is helpful.

So, I'd rather leave this as is.

Thanks,
Jeff

Re: [PATCH v5 06/30] fsmonitor--daemon: add a built-in fsmonitor daemon

[Junio C Hamano]

Jeff Hostetler <git@jeffhostetler.com> writes:

> Finally, I'm not sure how much value there is in being able to catch
> parse_options() BUGs on Linux (or any other yet-to-be-supported
> platform).

I don't either.  Even though I am not 100% happy with the current
implementation of the embedded sanity checker in parse-options API,
as long as it is made available on all platforms, developers of a
platform specific part can also take advantage of it to catch any
issues while they develop, without waiting for Linux (or other
platforms) users to help them do so.

[PATCH v6 00/30] Builtin FSMonitor Part 2

[Jeff Hostetler via GitGitGadget]

Here is V6 of Part 2 of my builtin FSmonitor series.

This version contains mostly cleanup based on feedback from V5. Of note:

 * I squashed in the 1_file fix for p7519.
 * I squashed in a commit from part 3 to optionally print the "running
   daemon on..." message on stderr.
 * I added a note to the documentation about incompatible changes around
   core.fsmonitor.
 * Removed/rephrased some obsolete NEEDSWORK items.

Tao has an ongoing parallel series to fix test-chmtime on Windows.
https://lore.kernel.org/all/pull.1166.git.1646041236.gitgitgadget@gmail.com/

If that lands first, we should be able to drop my 't/helper/test-chmtime:
skip directories on Windows' commit.

A followup Part 3 will contain additional refinements to the daemon and
additional tests. I drew the line here between Part 2 and 3 to make it
easier to review.

Jeff Hostetler (30):
  fsmonitor: enhance existing comments, clarify trivial response
    handling
  fsmonitor-ipc: create client routines for git-fsmonitor--daemon
  fsmonitor: config settings are repository-specific
  fsmonitor: use IPC to query the builtin FSMonitor daemon
  fsmonitor: document builtin fsmonitor
  fsmonitor--daemon: add a built-in fsmonitor daemon
  fsmonitor--daemon: implement 'stop' and 'status' commands
  compat/fsmonitor/fsm-listen-win32: stub in backend for Windows
  compat/fsmonitor/fsm-listen-darwin: stub in backend for Darwin
  fsmonitor--daemon: implement 'run' command
  fsmonitor--daemon: implement 'start' command
  fsmonitor--daemon: add pathname classification
  fsmonitor--daemon: define token-ids
  fsmonitor--daemon: create token-based changed path cache
  compat/fsmonitor/fsm-listen-win32: implement FSMonitor backend on
    Windows
  compat/fsmonitor/fsm-listen-darwin: add MacOS header files for FSEvent
  compat/fsmonitor/fsm-listen-darwin: implement FSEvent listener on
    MacOS
  fsmonitor--daemon: implement handle_client callback
  help: include fsmonitor--daemon feature flag in version info
  t/helper/fsmonitor-client: create IPC client to talk to FSMonitor
    Daemon
  t7527: create test for fsmonitor--daemon
  t/perf: avoid copying builtin fsmonitor files into test repo
  t/helper/test-chmtime: skip directories on Windows
  t/perf/p7519: speed up test on Windows
  t/perf/p7519: add fsmonitor--daemon test cases
  fsmonitor--daemon: periodically truncate list of modified files
  fsmonitor--daemon: use a cookie file to sync with file system
  fsmonitor: force update index after large responses
  t7527: test status with untracked-cache and fsmonitor--daemon
  update-index: convert fsmonitor warnings to advise

 .gitignore                              |    1 +
 Documentation/config/core.txt           |   60 +-
 Documentation/git-fsmonitor--daemon.txt |   75 ++
 Documentation/git-update-index.txt      |    8 +-
 Makefile                                |   17 +
 builtin.h                               |    1 +
 builtin/fsmonitor--daemon.c             | 1473 +++++++++++++++++++++++
 builtin/update-index.c                  |   19 +-
 cache.h                                 |    1 -
 compat/fsmonitor/fsm-listen-darwin.c    |  499 ++++++++
 compat/fsmonitor/fsm-listen-win32.c     |  586 +++++++++
 compat/fsmonitor/fsm-listen.h           |   49 +
 config.c                                |   14 -
 config.h                                |    1 -
 config.mak.uname                        |   20 +
 contrib/buildsystems/CMakeLists.txt     |   10 +
 environment.c                           |    1 -
 fsmonitor--daemon.h                     |  166 +++
 fsmonitor-ipc.c                         |  171 +++
 fsmonitor-ipc.h                         |   48 +
 fsmonitor-settings.c                    |  115 ++
 fsmonitor-settings.h                    |   21 +
 fsmonitor.c                             |  216 +++-
 fsmonitor.h                             |   15 +-
 git.c                                   |    1 +
 help.c                                  |    4 +
 repo-settings.c                         |    1 +
 repository.h                            |    3 +
 t/README                                |    4 +-
 t/helper/test-chmtime.c                 |   15 +
 t/helper/test-fsmonitor-client.c        |  121 ++
 t/helper/test-tool.c                    |    1 +
 t/helper/test-tool.h                    |    1 +
 t/perf/p7519-fsmonitor.sh               |   61 +-
 t/perf/perf-lib.sh                      |    2 +-
 t/t7527-builtin-fsmonitor.sh            |  604 ++++++++++
 t/test-lib.sh                           |    6 +
 37 files changed, 4301 insertions(+), 110 deletions(-)
 create mode 100644 Documentation/git-fsmonitor--daemon.txt
 create mode 100644 builtin/fsmonitor--daemon.c
 create mode 100644 compat/fsmonitor/fsm-listen-darwin.c
 create mode 100644 compat/fsmonitor/fsm-listen-win32.c
 create mode 100644 compat/fsmonitor/fsm-listen.h
 create mode 100644 fsmonitor--daemon.h
 create mode 100644 fsmonitor-ipc.c
 create mode 100644 fsmonitor-ipc.h
 create mode 100644 fsmonitor-settings.c
 create mode 100644 fsmonitor-settings.h
 create mode 100644 t/helper/test-fsmonitor-client.c
 create mode 100755 t/t7527-builtin-fsmonitor.sh


base-commit: 715d08a9e51251ad8290b181b6ac3b9e1f9719d7
Published-As: https://github.com/gitgitgadget/git/releases/tag/pr-1041%2Fjeffhostetler%2Fbuiltin-fsmonitor-part2-v6
Fetch-It-Via: git fetch https://github.com/gitgitgadget/git pr-1041/jeffhostetler/builtin-fsmonitor-part2-v6
Pull-Request: https://github.com/gitgitgadget/git/pull/1041

Range-diff vs v5:

  1:  a5ecabb4d02 !  1:  10a34082fcc fsmonitor: enhance existing comments, clarify trivial response handling
     @@ Commit message
          fsmonitor: enhance existing comments, clarify trivial response handling
      
          Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
     +    Signed-off-by: Junio C Hamano <gitster@pobox.com>
      
       ## fsmonitor.c ##
      @@ fsmonitor.c: static int query_fsmonitor(int version, const char *last_update, struct strbuf *
  2:  365964b7664 !  2:  e1c946af2c5 fsmonitor-ipc: create client routines for git-fsmonitor--daemon
     @@ Commit message
          Stub in empty fsmonitor_ipc__*() functions for unsupported platforms.
      
          Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
     +    Signed-off-by: Junio C Hamano <gitster@pobox.com>
      
       ## Makefile ##
      @@ Makefile: LIB_OBJS += fetch-pack.o
     @@ fsmonitor-ipc.c (new)
      +#include "strbuf.h"
      +#include "trace2.h"
      +
     -+#ifdef HAVE_FSMONITOR_DAEMON_BACKEND
     ++#ifndef HAVE_FSMONITOR_DAEMON_BACKEND
     ++
     ++/*
     ++ * A trivial implementation of the fsmonitor_ipc__ API for unsupported
     ++ * platforms.
     ++ */
     ++
     ++int fsmonitor_ipc__is_supported(void)
     ++{
     ++	return 0;
     ++}
     ++
     ++const char *fsmonitor_ipc__get_path(void)
     ++{
     ++	return NULL;
     ++}
     ++
     ++enum ipc_active_state fsmonitor_ipc__get_state(void)
     ++{
     ++	return IPC_STATE__OTHER_ERROR;
     ++}
     ++
     ++int fsmonitor_ipc__send_query(const char *since_token,
     ++			      struct strbuf *answer)
     ++{
     ++	return -1;
     ++}
     ++
     ++int fsmonitor_ipc__send_command(const char *command,
     ++				struct strbuf *answer)
     ++{
     ++	return -1;
     ++}
     ++
     ++#else
      +
      +int fsmonitor_ipc__is_supported(void)
      +{
     @@ fsmonitor-ipc.c (new)
      +	return 0;
      +}
      +
     -+#else
     -+
     -+/*
     -+ * A trivial implementation of the fsmonitor_ipc__ API for unsupported
     -+ * platforms.
     -+ */
     -+
     -+int fsmonitor_ipc__is_supported(void)
     -+{
     -+	return 0;
     -+}
     -+
     -+const char *fsmonitor_ipc__get_path(void)
     -+{
     -+	return NULL;
     -+}
     -+
     -+enum ipc_active_state fsmonitor_ipc__get_state(void)
     -+{
     -+	return IPC_STATE__OTHER_ERROR;
     -+}
     -+
     -+int fsmonitor_ipc__send_query(const char *since_token,
     -+			      struct strbuf *answer)
     -+{
     -+	return -1;
     -+}
     -+
     -+int fsmonitor_ipc__send_command(const char *command,
     -+				struct strbuf *answer)
     -+{
     -+	return -1;
     -+}
     -+
      +#endif
      
       ## fsmonitor-ipc.h (new) ##
  3:  384516ce1a1 !  3:  ae622a517cf fsmonitor: config settings are repository-specific
     @@ Commit message
          restrictions to override the config values used.
      
          Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
     +    Signed-off-by: Junio C Hamano <gitster@pobox.com>
      
       ## Makefile ##
      @@ Makefile: LIB_OBJS += fmt-merge-msg.o
     @@ fsmonitor-settings.c (new)
      +
      +enum fsmonitor_mode fsm_settings__get_mode(struct repository *r)
      +{
     ++	if (!r)
     ++		r = the_repository;
     ++
      +	lookup_fsmonitor_settings(r);
      +
      +	return r->settings.fsmonitor->mode;
     @@ fsmonitor-settings.c (new)
      +
      +const char *fsm_settings__get_hook_path(struct repository *r)
      +{
     ++	if (!r)
     ++		r = the_repository;
     ++
      +	lookup_fsmonitor_settings(r);
      +
      +	return r->settings.fsmonitor->hook_path;
     @@ fsmonitor-settings.c (new)
      +
      +void fsm_settings__set_ipc(struct repository *r)
      +{
     ++	if (!r)
     ++		r = the_repository;
     ++
      +	lookup_fsmonitor_settings(r);
      +
      +	r->settings.fsmonitor->mode = FSMONITOR_MODE_IPC;
     @@ fsmonitor-settings.c (new)
      +
      +void fsm_settings__set_hook(struct repository *r, const char *path)
      +{
     ++	if (!r)
     ++		r = the_repository;
     ++
      +	lookup_fsmonitor_settings(r);
      +
      +	r->settings.fsmonitor->mode = FSMONITOR_MODE_HOOK;
     @@ fsmonitor-settings.c (new)
      +
      +void fsm_settings__set_disabled(struct repository *r)
      +{
     ++	if (!r)
     ++		r = the_repository;
     ++
      +	lookup_fsmonitor_settings(r);
      +
      +	r->settings.fsmonitor->mode = FSMONITOR_MODE_DISABLED;
     @@ fsmonitor.c: void remove_fsmonitor(struct index_state *istate)
       {
       	unsigned int i;
      -	int fsmonitor_enabled = git_config_get_fsmonitor();
     -+	struct repository *r = istate->repo ? istate->repo : the_repository;
     -+	int fsmonitor_enabled = (fsm_settings__get_mode(r) > FSMONITOR_MODE_DISABLED);
     ++	int fsmonitor_enabled = (fsm_settings__get_mode(istate->repo)
     ++				 > FSMONITOR_MODE_DISABLED);
       
       	if (istate->fsmonitor_dirty) {
       		if (fsmonitor_enabled) {
     @@ fsmonitor.h: int fsmonitor_is_trivial_response(const struct strbuf *query_result
       static inline int is_fsmonitor_refreshed(const struct index_state *istate)
       {
      -	return !core_fsmonitor || istate->fsmonitor_has_run_once;
     -+	struct repository *r = istate->repo ? istate->repo : the_repository;
     -+	enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
     ++	enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(istate->repo);
      +
      +	return fsm_mode <= FSMONITOR_MODE_DISABLED ||
      +		istate->fsmonitor_has_run_once;
     @@ fsmonitor.h: static inline int is_fsmonitor_refreshed(const struct index_state *
       static inline void mark_fsmonitor_valid(struct index_state *istate, struct cache_entry *ce)
       {
      -	if (core_fsmonitor && !(ce->ce_flags & CE_FSMONITOR_VALID)) {
     -+	struct repository *r = istate->repo ? istate->repo : the_repository;
     -+	enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
     ++	enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(istate->repo);
      +
      +	if (fsm_mode > FSMONITOR_MODE_DISABLED &&
      +	    !(ce->ce_flags & CE_FSMONITOR_VALID)) {
     @@ fsmonitor.h: static inline void mark_fsmonitor_valid(struct index_state *istate,
       static inline void mark_fsmonitor_invalid(struct index_state *istate, struct cache_entry *ce)
       {
      -	if (core_fsmonitor) {
     -+	struct repository *r = istate->repo ? istate->repo : the_repository;
     -+	enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
     ++	enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(istate->repo);
      +
      +	if (fsm_mode > FSMONITOR_MODE_DISABLED) {
       		ce->ce_flags &= ~CE_FSMONITOR_VALID;
  4:  8e738a83bc5 !  4:  55974867da5 fsmonitor: use IPC to query the builtin FSMonitor daemon
     @@ Commit message
      
          Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
          Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
     +    Signed-off-by: Junio C Hamano <gitster@pobox.com>
      
       ## fsmonitor.c ##
      @@ fsmonitor.c: void refresh_fsmonitor(struct index_state *istate)
  5:  49e4c146e02 !  5:  ce42d5bbaf6 fsmonitor: document builtin fsmonitor
     @@ Commit message
          Create `git-fsmonitor--daemon` manual page and describe its features.
      
          Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
     +    Signed-off-by: Junio C Hamano <gitster@pobox.com>
      
       ## Documentation/config/core.txt ##
      @@ Documentation/config/core.txt: core.protectNTFS::
     @@ Documentation/config/core.txt: core.protectNTFS::
      +git by avoiding unnecessary scanning of files that have not changed.
      ++
      +See the "fsmonitor-watchman" section of linkgit:githooks[5].
     +++
     ++Note that if you concurrently use multiple versions of Git, such
     ++as one version on the command line and another version in an IDE
     ++tool, that the definition of `core.fsmonitor` was extended to
     ++allow boolean values in addition to hook pathnames.  Git versions
     ++2.35.1 and prior will not understand the boolean values and will
     ++consider the "true" or "false" values as hook pathnames to be
     ++invoked.  Git versions 2.26 thru 2.35.1 default to hook protocol
     ++V2 and will fall back to no fsmonitor (full scan).  Git versions
     ++prior to 2.26 default to hook protocol V1 and will silently
     ++assume there were no changes to report (no scan), so status
     ++commands may report incomplete results.  For this reason, it is
     ++best to upgrade all of your Git versions before using the built-in
     ++file system monitor.
       
       core.fsmonitorHookVersion::
      -	Sets the version of hook that is to be used when calling fsmonitor.
  6:  bdd7334da31 !  6:  9ce938c69b5 fsmonitor--daemon: add a built-in fsmonitor daemon
     @@ Commit message
      
          Co-authored-by: Johannes Schindelin <johannes.schindelin@gmx.de>
          Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
     +    Signed-off-by: Junio C Hamano <gitster@pobox.com>
      
       ## .gitignore ##
      @@
  7:  9f00ada3dd3 !  7:  3bd7e6a051e fsmonitor--daemon: implement 'stop' and 'status' commands
     @@ Commit message
          system.
      
          Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
     +    Signed-off-by: Junio C Hamano <gitster@pobox.com>
      
       ## builtin/fsmonitor--daemon.c ##
      @@
  8:  d6819bdad66 !  8:  92d3e54da7f compat/fsmonitor/fsm-listen-win32: stub in backend for Windows
     @@ Commit message
          Stub in empty filesystem listener backend for fsmonitor--daemon on Windows.
      
          Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
     +    Signed-off-by: Junio C Hamano <gitster@pobox.com>
      
       ## Makefile ##
      @@ Makefile: all::
  9:  7de3d01cccc !  9:  f1219221270 compat/fsmonitor/fsm-listen-darwin: stub in backend for Darwin
     @@ Commit message
          backend for Darwin (aka MacOS).
      
          Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
     +    Signed-off-by: Junio C Hamano <gitster@pobox.com>
      
       ## compat/fsmonitor/fsm-listen-darwin.c (new) ##
      @@
 10:  6fe5a2bc79e ! 10:  a3f94627cf6 fsmonitor--daemon: implement 'run' command
     @@ Commit message
          backends are still just stubs.
      
          Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
     +    Signed-off-by: Junio C Hamano <gitster@pobox.com>
      
       ## builtin/fsmonitor--daemon.c ##
      @@
     @@ builtin/fsmonitor--daemon.c
      +#define FSMONITOR__IPC_THREADS "fsmonitor.ipcthreads"
      +static int fsmonitor__ipc_threads = 8;
      +
     ++#define FSMONITOR__ANNOUNCE_STARTUP "fsmonitor.announcestartup"
     ++static int fsmonitor__announce_startup = 0;
     ++
      +static int fsmonitor_config(const char *var, const char *value, void *cb)
      +{
      +	if (!strcmp(var, FSMONITOR__IPC_THREADS)) {
     @@ builtin/fsmonitor--daemon.c
      +		return 0;
      +	}
      +
     ++	if (!strcmp(var, FSMONITOR__ANNOUNCE_STARTUP)) {
     ++		int is_bool;
     ++		int i = git_config_bool_or_int(var, value, &is_bool);
     ++		if (i < 0)
     ++			return error(_("value of '%s' not bool or int: %d"),
     ++				     var, i);
     ++		fsmonitor__announce_startup = i;
     ++		return 0;
     ++	}
     ++
      +	return git_default_config(var, value, cb);
      +}
      +
     @@ builtin/fsmonitor--daemon.c: static int do_as_client__status(void)
      +		die("fsmonitor--daemon is already running '%s'",
      +		    the_repository->worktree);
      +
     -+	printf(_("running fsmonitor-daemon in '%s'\n"),
     -+	       the_repository->worktree);
     -+	fflush(stdout);
     ++	if (fsmonitor__announce_startup) {
     ++		fprintf(stderr, _("running fsmonitor-daemon in '%s'\n"),
     ++			the_repository->worktree);
     ++		fflush(stderr);
     ++	}
      +
      +	return !!fsmonitor_run_daemon();
      +}
 11:  69fc0998286 ! 11:  8de40b0fe8b fsmonitor--daemon: implement 'start' command
     @@ Commit message
          because they don't have a Win32 console attached to them.)
      
          Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
     +    Signed-off-by: Junio C Hamano <gitster@pobox.com>
      
       ## builtin/fsmonitor--daemon.c ##
      @@
     @@ builtin/fsmonitor--daemon.c: static const char * const builtin_fsmonitor__daemon
      +#define FSMONITOR__START_TIMEOUT "fsmonitor.starttimeout"
      +static int fsmonitor__start_timeout_sec = 60;
      +
     - static int fsmonitor_config(const char *var, const char *value, void *cb)
     - {
     - 	if (!strcmp(var, FSMONITOR__IPC_THREADS)) {
     + #define FSMONITOR__ANNOUNCE_STARTUP "fsmonitor.announcestartup"
     + static int fsmonitor__announce_startup = 0;
     + 
      @@ builtin/fsmonitor--daemon.c: static int fsmonitor_config(const char *var, const char *value, void *cb)
       		return 0;
       	}
     @@ builtin/fsmonitor--daemon.c: static int fsmonitor_config(const char *var, const
      +		return 0;
      +	}
      +
     - 	return git_default_config(var, value, cb);
     - }
     - 
     + 	if (!strcmp(var, FSMONITOR__ANNOUNCE_STARTUP)) {
     + 		int is_bool;
     + 		int i = git_config_bool_or_int(var, value, &is_bool);
      @@ builtin/fsmonitor--daemon.c: done:
       	return err;
       }
       
      -static int try_to_run_foreground_daemon(void)
     -+static int try_to_run_foreground_daemon(int free_console)
     ++static int try_to_run_foreground_daemon(int detach_console)
       {
       	/*
       	 * Technically, we don't need to probe for an existing daemon
      @@ builtin/fsmonitor--daemon.c: static int try_to_run_foreground_daemon(void)
     - 	       the_repository->worktree);
     - 	fflush(stdout);
     + 		fflush(stderr);
     + 	}
       
      +#ifdef GIT_WINDOWS_NATIVE
     -+	if (free_console)
     ++	if (detach_console)
      +		FreeConsole();
      +#endif
      +
     @@ builtin/fsmonitor--daemon.c: static int try_to_run_foreground_daemon(void)
      +		die("fsmonitor--daemon is already running '%s'",
      +		    the_repository->worktree);
      +
     -+	printf(_("starting fsmonitor-daemon in '%s'\n"),
     -+	       the_repository->worktree);
     -+	fflush(stdout);
     ++	if (fsmonitor__announce_startup) {
     ++		fprintf(stderr, _("starting fsmonitor-daemon in '%s'\n"),
     ++			the_repository->worktree);
     ++		fflush(stderr);
     ++	}
      +
      +	cp.git_cmd = 1;
      +
      +	strvec_push(&cp.args, "fsmonitor--daemon");
      +	strvec_push(&cp.args, "run");
     -+	strvec_push(&cp.args, "--free-console");
     ++	strvec_push(&cp.args, "--detach");
      +	strvec_pushf(&cp.args, "--ipc-threads=%d", fsmonitor__ipc_threads);
      +
      +	cp.no_stdin = 1;
     @@ builtin/fsmonitor--daemon.c: static int try_to_run_foreground_daemon(void)
       int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
       {
       	const char *subcmd;
     -+	int free_console = 0;
     ++	int detach_console = 0;
       
       	struct option options[] = {
     -+		OPT_BOOL(0, "free-console", &free_console, N_("free console")),
     ++		OPT_BOOL(0, "detach", &detach_console, N_("detach from console")),
       		OPT_INTEGER(0, "ipc-threads",
       			    &fsmonitor__ipc_threads,
       			    N_("use <n> ipc worker threads")),
      +		OPT_INTEGER(0, "start-timeout",
      +			    &fsmonitor__start_timeout_sec,
     -+			    N_("Max seconds to wait for background daemon startup")),
     ++			    N_("max seconds to wait for background daemon startup")),
      +
       		OPT_END()
       	};
     @@ builtin/fsmonitor--daemon.c: int cmd_fsmonitor__daemon(int argc, const char **ar
      +
       	if (!strcmp(subcmd, "run"))
      -		return !!try_to_run_foreground_daemon();
     -+		return !!try_to_run_foreground_daemon(free_console);
     ++		return !!try_to_run_foreground_daemon(detach_console);
       
       	if (!strcmp(subcmd, "stop"))
       		return !!do_as_client__send_stop();
 12:  21c099c5197 ! 12:  e5d419b1ea0 fsmonitor--daemon: add pathname classification
     @@ Commit message
          When we register for filesystem notifications on a directory,
          we get events for everything (recursively) in the directory.
          We want to report to clients changes to tracked and untracked
     -    paths within the working directory.  We do not want to report
     -    changes within the .git directory, for example.
     +    paths within the working directory proper.  We do not want to
     +    report changes within the .git directory, for example.
      
          This classification will be used in a later commit by the
          different backends to classify paths as events are received.
      
          Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
     +    Signed-off-by: Junio C Hamano <gitster@pobox.com>
      
       ## builtin/fsmonitor--daemon.c ##
      @@ builtin/fsmonitor--daemon.c: static int handle_client(void *data,
     @@ fsmonitor--daemon.h: struct fsmonitor_daemon_state {
      + * that to decide whether clients are told about them.  (And to watch
      + * for file system synchronization events.)
      + *
     ++ * The daemon only collects and reports on the set of modified paths
     ++ * within the working directory (proper).
     ++ *
      + * The client should only care about paths within the working
      + * directory proper (inside the working directory and not ".git" nor
      + * inside of ".git/").  That is, the client has read the index and is
      + * asking for a list of any paths in the working directory that have
      + * been modified since the last token.  The client does not care about
     -+ * file system changes within the .git directory (such as new loose
     ++ * file system changes within the ".git/" directory (such as new loose
      + * objects or packfiles).  So the client will only receive paths that
      + * are classified as IS_WORKDIR_PATH.
      + *
     ++ * Note that ".git" is usually a directory and is therefore inside
     ++ * the cone of the FS watch that we have on the working directory root,
     ++ * so we will also get FS events for disk activity on and within ".git/"
     ++ * that we need to respond to or filter from the client.
     ++ *
     ++ * But Git also allows ".git" to be a *file* that points to a GITDIR
     ++ * outside of the working directory.  When this happens, we need to
     ++ * create FS watches on both the working directory root *and* on the
     ++ * (external) GITDIR root.  (The latter is required because we put
     ++ * cookie files inside it and use them to sync with the FS event
     ++ * stream.)
     ++ *
     ++ * Note that in the context of this discussion, I'm using "GITDIR"
     ++ * to only mean an external GITDIR referenced by a ".git" file.
     ++ *
     ++ * The platform FS event backends will receive watch-specific
     ++ * relative paths (except for those OS's that always emit absolute
     ++ * paths).  We use the following enum and routines to classify each
     ++ * path so that we know how to handle it.  There is a slight asymmetry
     ++ * here because ".git/" is inside the working directory and the
     ++ * (external) GITDIR is not, and therefore how we handle events may
     ++ * vary slightly, so I have different enums for "IS...DOT_GIT..." and
     ++ * "IS...GITDIR...".
     ++ *
      + * The daemon uses the IS_DOT_GIT and IS_GITDIR internally to mean the
     -+ * exact ".git" directory or GITDIR.  If the daemon receives a delete
     -+ * event for either of these directories, it will automatically
     -+ * shutdown, for example.
     ++ * exact ".git" file/directory or GITDIR directory.  If the daemon
     ++ * receives a delete event for either of these paths, it will
     ++ * automatically shutdown, for example.
      + *
      + * Note that the daemon DOES NOT explicitly watch nor special case the
     -+ * ".git/index" file.  The daemon does not read the index and does not
     -+ * have any internal index-relative state.  The daemon only collects
     -+ * the set of modified paths within the working directory.
     ++ * index.  The daemon does not read the index nor have any internal
     ++ * index-relative state, so there are no "IS...INDEX..." enum values.
      + */
      +enum fsmonitor_path_type {
      +	IS_WORKDIR_PATH = 0,
 13:  72979c35ceb ! 13:  c2f3668a866 fsmonitor--daemon: define token-ids
     @@ Commit message
          overall token naming scheme.
      
          Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
     +    Signed-off-by: Junio C Hamano <gitster@pobox.com>
      
       ## builtin/fsmonitor--daemon.c ##
      @@ builtin/fsmonitor--daemon.c: static int do_as_client__status(void)
 14:  8d1db644409 ! 14:  bd492d13fb1 fsmonitor--daemon: create token-based changed path cache
     @@ Commit message
          paths within these batches for efficient presentation to clients.
      
          Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
     +    Signed-off-by: Junio C Hamano <gitster@pobox.com>
      
       ## builtin/fsmonitor--daemon.c ##
      @@ builtin/fsmonitor--daemon.c: struct fsmonitor_token_data {
     @@ builtin/fsmonitor--daemon.c: static struct fsmonitor_token_data *fsmonitor_new_t
      + *     We should create a new token and start fresh (as if we just
      + *     booted up).
      + *
     -+ * If there are no concurrent threads readering the current token data
     ++ * If there are no concurrent threads reading the current token data
      + * series, we can free it now.  Otherwise, let the last reader free
      + * it.
      + *
 15:  98c5adf8ca0 ! 15:  6b5b3d08227 compat/fsmonitor/fsm-listen-win32: implement FSMonitor backend on Windows
     @@ Commit message
          into batches and publish.
      
          Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
     +    Signed-off-by: Junio C Hamano <gitster@pobox.com>
      
       ## compat/fsmonitor/fsm-listen-win32.c ##
      @@
 16:  a3b881315fa ! 16:  cdef9730b3f compat/fsmonitor/fsm-listen-darwin: add macos header files for FSEvent
     @@ Metadata
      Author: Jeff Hostetler <jeffhost@microsoft.com>
      
       ## Commit message ##
     -    compat/fsmonitor/fsm-listen-darwin: add macos header files for FSEvent
     +    compat/fsmonitor/fsm-listen-darwin: add MacOS header files for FSEvent
      
          Include MacOS system declarations to allow us to use FSEvent and
     -    CoreFoundation APIs.  We need GCC and clang versions because of
     -    compiler and header file conflicts.
     +    CoreFoundation APIs.  We need different versions of the declarations
     +    for GCC vs. clang because of compiler and header file conflicts.
      
          While it is quite possible to #include Apple's CoreServices.h when
          compiling C source code with clang, trying to build it with GCC
     @@ Commit message
      
          Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
          Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
     +    Signed-off-by: Junio C Hamano <gitster@pobox.com>
      
       ## compat/fsmonitor/fsm-listen-darwin.c ##
      @@
     -+#if defined(__GNUC__)
     ++#ifndef __clang__
      +/*
      + * It is possible to #include CoreFoundation/CoreFoundation.h when compiling
      + * with clang, but not with GCC as of time of writing.
 17:  162e357db72 ! 17:  aabfebd3a41 compat/fsmonitor/fsm-listen-darwin: implement FSEvent listener on MacOS
     @@ Commit message
          Co-authored-by: Kevin Willford <Kevin.Willford@microsoft.com>
          Co-authored-by: Johannes Schindelin <johannes.schindelin@gmx.de>
          Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
     +    Signed-off-by: Junio C Hamano <gitster@pobox.com>
      
       ## compat/fsmonitor/fsm-listen-darwin.c ##
     -@@
     --#if defined(__GNUC__)
     -+#ifndef __clang__
     - /*
     -  * It is possible to #include CoreFoundation/CoreFoundation.h when compiling
     -  * with clang, but not with GCC as of time of writing.
      @@ compat/fsmonitor/fsm-listen-darwin.c: void FSEventStreamRelease(FSEventStreamRef stream);
       #include "cache.h"
       #include "fsmonitor.h"
     @@ compat/fsmonitor/fsm-listen-darwin.c: void FSEventStreamRelease(FSEventStreamRef
      +}
      +
      +/*
     -+ * NEEDSWORK: Investigate the proper value for the `latency` argument
     -+ * in the call to `FSEventStreamCreate()`.  I'm not sure that this
     -+ * needs to be a config setting or just something that we tune after
     -+ * some testing.
     ++ * In the call to `FSEventStreamCreate()` to setup our watch, the
     ++ * `latency` argument determines the frequency of calls to our callback
     ++ * with new FS events.  Too slow and events get dropped; too fast and
     ++ * we burn CPU unnecessarily.  Since it is rather obscure, I don't
     ++ * think this needs to be a config setting.  I've done extensive
     ++ * testing on my systems and chosen the value below.  It gives good
     ++ * results and I've not seen any dropped events.
      + *
      + * With a latency of 0.1, I was seeing lots of dropped events during
      + * the "touch 100000" files test within t/perf/p7519, but with a
     -+ * latency of 0.001 I did not see any dropped events.  So the
     -+ * "correct" value may be somewhere in between.
     ++ * latency of 0.001 I did not see any dropped events.  So I'm going
     ++ * to assume that this is the "correct" value.
      + *
      + * https://developer.apple.com/documentation/coreservices/1443980-fseventstreamcreate
      + */
 18:  3de1c43beaf ! 18:  0896e72e2b8 fsmonitor--daemon: implement handle_client callback
     @@ Commit message
          relative to the provided token.
      
          Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
     +    Signed-off-by: Junio C Hamano <gitster@pobox.com>
      
       ## builtin/fsmonitor--daemon.c ##
      @@
 19:  3517c4a3c13 ! 19:  bc240a9e665 help: include fsmonitor--daemon feature flag in version info
     @@ Commit message
      
          This can be used by test scripts for prereq testing.  Granted, tests
          could just try `git fsmonitor--daemon status` and look for a 128 exit
     -    code or grep for a "not supported" message on stderr, but this is
     -    rather obscure.
     +    code or grep for a "not supported" message on stderr, but these
     +    methods are rather obscure.
      
          The main advantage is that the feature message will automatically
          appear in bug reports and other support requests.
      
     +    This concept was also used during the development of Scalar for
     +    similar reasons.
     +
          Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
     +    Signed-off-by: Junio C Hamano <gitster@pobox.com>
      
       ## help.c ##
      @@
 20:  4ffc2ddf516 ! 20:  2b563c240eb t/helper/fsmonitor-client: create IPC client to talk to FSMonitor Daemon
     @@ Commit message
          Create an IPC client to send query and flush commands to the daemon.
      
          Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
     +    Signed-off-by: Junio C Hamano <gitster@pobox.com>
      
       ## Makefile ##
      @@ Makefile: TEST_BUILTINS_OBJS += test-dump-split-index.o
 21:  d93310a7c64 ! 21:  db8809da089 t7527: create test for fsmonitor--daemon
     @@ Commit message
          t7527: create test for fsmonitor--daemon
      
          Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
     +    Signed-off-by: Junio C Hamano <gitster@pobox.com>
      
       ## t/t7527-builtin-fsmonitor.sh (new) ##
      @@
 22:  27e47108908 ! 22:  15bd5aaff36 t/perf: avoid copying builtin fsmonitor files into test repo
     @@ Commit message
          .git/fsmonitor--daemon*, so it is simple to exclude them.
      
          Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
     +    Signed-off-by: Junio C Hamano <gitster@pobox.com>
      
       ## t/perf/perf-lib.sh ##
      @@ t/perf/perf-lib.sh: test_perf_copy_repo_contents () {
 23:  6cba1d950b0 ! 23:  4cb97918d20 t/helper/test-chmtime: skip directories on Windows
     @@ Commit message
          an empty directory on disk.
      
          Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
     +    Signed-off-by: Junio C Hamano <gitster@pobox.com>
      
       ## t/helper/test-chmtime.c ##
      @@ t/helper/test-chmtime.c: int cmd__chmtime(int argc, const char **argv)
 24:  fcf843a0d42 ! 24:  a70748b4640 t/perf/p7519: speed up test on Windows
     @@ Commit message
          on Windows because of process creation overhead.
      
          Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
     +    Signed-off-by: Junio C Hamano <gitster@pobox.com>
      
       ## t/perf/p7519-fsmonitor.sh ##
      @@ t/perf/p7519-fsmonitor.sh: then
     @@ t/perf/p7519-fsmonitor.sh: test_expect_success "one time repo setup" '
      -	for i in $(test_seq 1 100); do touch 100_files/$i || return 1; done &&
      -	for i in $(test_seq 1 1000); do touch 1000_files/$i || return 1; done &&
      -	for i in $(test_seq 1 10000); do touch 10000_files/$i || return 1; done &&
     -+	touch_files 1 &&
     ++	: 1_file directory should be left empty &&
      +	touch_files 10 &&
      +	touch_files 100 &&
      +	touch_files 1000 &&
 25:  198f47bda5a ! 25:  b0c9c9c7a40 t/perf/p7519: add fsmonitor--daemon test cases
     @@ Commit message
          the "Simple IPC" interface.
      
          Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
     +    Signed-off-by: Junio C Hamano <gitster@pobox.com>
      
       ## t/perf/p7519-fsmonitor.sh ##
      @@ t/perf/p7519-fsmonitor.sh: test_perf_w_drop_caches () {
 26:  19993c130d2 ! 26:  64a5b741670 fsmonitor--daemon: periodically truncate list of modified files
     @@ Commit message
          they are considered obsolete.
      
          Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
     +    Signed-off-by: Junio C Hamano <gitster@pobox.com>
      
       ## builtin/fsmonitor--daemon.c ##
      @@ builtin/fsmonitor--daemon.c: static void fsmonitor_batch__combine(struct fsmonitor_batch *batch_dest,
 27:  f47a763dc26 ! 27:  5b06eb5d0e6 fsmonitor--daemon: use a cookie file to sync with file system
     @@ Commit message
          of `FSeventStreamCreate()`.  The kernel only signals every `latency`
          seconds, but does not guarantee that the kernel queue is completely
          drained, so we may have to wait more than one interval.  If we
     -    increase the frequency, the system is more likely to drop events.
     +    increase the latency, the system is more likely to drop events.
          We avoid these issues by having each client thread create a unique
          cookie file and then wait until it is seen in the event stream.
      
          Co-authored-by: Kevin Willford <Kevin.Willford@microsoft.com>
          Co-authored-by: Johannes Schindelin <johannes.schindelin@gmx.de>
          Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
     +    Signed-off-by: Junio C Hamano <gitster@pobox.com>
      
       ## builtin/fsmonitor--daemon.c ##
      @@ builtin/fsmonitor--daemon.c: static int do_as_client__status(void)
     @@ builtin/fsmonitor--daemon.c: static int do_as_client__status(void)
      +		unlink(cookie_pathname.buf);
      +
      +		/*
     -+		 * NEEDSWORK: This is an infinite wait (well, unless another
     ++		 * Technically, this is an infinite wait (well, unless another
      +		 * thread sends us an abort).  I'd like to change this to
      +		 * use `pthread_cond_timedwait()` and return an error/timeout
     -+		 * and let the caller do the trivial response thing.
     ++		 * and let the caller do the trivial response thing, but we
     ++		 * don't have that routine in our thread-utils.
     ++		 *
     ++		 * After extensive beta testing I'm not really worried about
     ++		 * this.  Also note that the above open() and unlink() calls
     ++		 * will cause at least two FS events on that path, so the odds
     ++		 * of getting stuck are pretty slim.
      +		 */
      +		while (cookie->result == FCIR_INIT)
      +			pthread_cond_wait(&state->cookies_cond,
     @@ builtin/fsmonitor--daemon.c: static void fsmonitor_free_token_data(struct fsmoni
      + * [2] Some of those lost events may have been for cookie files.  We
      + *     should assume the worst and abort them rather letting them starve.
      + *
     -  * If there are no concurrent threads readering the current token data
     +  * If there are no concurrent threads reading the current token data
        * series, we can free it now.  Otherwise, let the last reader free
        * it.
      @@ builtin/fsmonitor--daemon.c: static void with_lock__do_force_resync(struct fsmonitor_daemon_state *state)
     @@ builtin/fsmonitor--daemon.c: done:
       	strbuf_release(&state.path_worktree_watch);
       	strbuf_release(&state.path_gitdir_watch);
      +	strbuf_release(&state.path_cookie_prefix);
     -+
     -+	/*
     -+	 * NEEDSWORK: Consider "rm -rf <gitdir>/<fsmonitor-dir>"
     -+	 */
       
       	return err;
       }
 28:  aec44a21afd ! 28:  1fd5439de03 fsmonitor: force update index after large responses
     @@ Commit message
          unnecessarily spend time actually scanning them during each command.
      
          Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
     +    Signed-off-by: Junio C Hamano <gitster@pobox.com>
      
       ## fsmonitor.c ##
      @@ fsmonitor.c: static void fsmonitor_refresh_callback(struct index_state *istate, char *name)
     @@ fsmonitor.c: static void fsmonitor_refresh_callback(struct index_state *istate,
      + * will be updated automatically the first time the user touches
      + * a tracked file and causes a command like `git status` to
      + * update an mtime to be updated and/or set a flag bit.
     -+ *
     -+ * NEEDSWORK: Does this need to be a config value?
      + */
      +static int fsmonitor_force_update_threshold = 100;
      +
 29:  d6039987df8 ! 29:  6fc1430285f t7527: test status with untracked-cache and fsmonitor--daemon
     @@ Commit message
          identical.
      
          Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
     +    Signed-off-by: Junio C Hamano <gitster@pobox.com>
      
       ## t/t7527-builtin-fsmonitor.sh ##
      @@ t/t7527-builtin-fsmonitor.sh: test_expect_success 'setup' '
 30:  5117fbdfc63 ! 30:  b915b95cc2f update-index: convert fsmonitor warnings to advise
     @@ Commit message
          update-index: convert fsmonitor warnings to advise
      
          Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
     +    Signed-off-by: Junio C Hamano <gitster@pobox.com>
      
       ## builtin/update-index.c ##
      @@ builtin/update-index.c: int cmd_update_index(int argc, const char **argv, const char *prefix)

-- 
gitgitgadget

[PATCH v6 01/30] fsmonitor: enhance existing comments, clarify trivial response handling

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 fsmonitor.c | 64 ++++++++++++++++++++++++++++++++++-------------------
 1 file changed, 41 insertions(+), 23 deletions(-)

diff --git a/fsmonitor.c b/fsmonitor.c
index ab9bfc60b34..448d0ee33f5 100644
--- a/fsmonitor.c
+++ b/fsmonitor.c
@@ -168,29 +168,15 @@ static int query_fsmonitor(int version, const char *last_update, struct strbuf *
 
 	if (result)
 		trace2_data_intmax("fsm_hook", NULL, "query/failed", result);
-	else {
+	else
 		trace2_data_intmax("fsm_hook", NULL, "query/response-length",
 				   query_result->len);
 
-		if (fsmonitor_is_trivial_response(query_result))
-			trace2_data_intmax("fsm_hook", NULL,
-					   "query/trivial-response", 1);
-	}
-
 	trace2_region_leave("fsm_hook", "query", NULL);
 
 	return result;
 }
 
-int fsmonitor_is_trivial_response(const struct strbuf *query_result)
-{
-	static char trivial_response[3] = { '\0', '/', '\0' };
-
-	return query_result->len >= 3 &&
-		!memcmp(trivial_response,
-			&query_result->buf[query_result->len - 3], 3);
-}
-
 static void fsmonitor_refresh_callback(struct index_state *istate, char *name)
 {
 	int i, len = strlen(name);
@@ -238,6 +224,7 @@ void refresh_fsmonitor(struct index_state *istate)
 	struct strbuf last_update_token = STRBUF_INIT;
 	char *buf;
 	unsigned int i;
+	int is_trivial = 0;
 
 	if (!core_fsmonitor || istate->fsmonitor_has_run_once)
 		return;
@@ -283,6 +270,7 @@ void refresh_fsmonitor(struct index_state *istate)
 					query_success = 0;
 				} else {
 					bol = last_update_token.len + 1;
+					is_trivial = query_result.buf[bol] == '/';
 				}
 			} else if (hook_version < 0) {
 				hook_version = HOOK_INTERFACE_VERSION1;
@@ -294,16 +282,38 @@ void refresh_fsmonitor(struct index_state *istate)
 		if (hook_version == HOOK_INTERFACE_VERSION1) {
 			query_success = !query_fsmonitor(HOOK_INTERFACE_VERSION1,
 				istate->fsmonitor_last_update, &query_result);
+			if (query_success)
+				is_trivial = query_result.buf[0] == '/';
 		}
 
+		if (is_trivial)
+			trace2_data_intmax("fsm_hook", NULL,
+					   "query/trivial-response", 1);
+
 		trace_performance_since(last_update, "fsmonitor process '%s'", core_fsmonitor);
 		trace_printf_key(&trace_fsmonitor, "fsmonitor process '%s' returned %s",
 			core_fsmonitor, query_success ? "success" : "failure");
 	}
 
-	/* a fsmonitor process can return '/' to indicate all entries are invalid */
-	if (query_success && query_result.buf[bol] != '/') {
-		/* Mark all entries returned by the monitor as dirty */
+	/*
+	 * The response from FSMonitor (excluding the header token) is
+	 * either:
+	 *
+	 * [a] a (possibly empty) list of NUL delimited relative
+	 *     pathnames of changed paths.  This list can contain
+	 *     files and directories.  Directories have a trailing
+	 *     slash.
+	 *
+	 * [b] a single '/' to indicate the provider had no
+	 *     information and that we should consider everything
+	 *     invalid.  We call this a trivial response.
+	 */
+	if (query_success && !is_trivial) {
+		/*
+		 * Mark all pathnames returned by the monitor as dirty.
+		 *
+		 * This updates both the cache-entries and the untracked-cache.
+		 */
 		buf = query_result.buf;
 		for (i = bol; i < query_result.len; i++) {
 			if (buf[i] != '\0')
@@ -318,11 +328,16 @@ void refresh_fsmonitor(struct index_state *istate)
 		if (istate->untracked)
 			istate->untracked->use_fsmonitor = 1;
 	} else {
-
-		/* We only want to run the post index changed hook if we've actually changed entries, so keep track
-		 * if we actually changed entries or not */
+		/*
+		 * We failed to get a response or received a trivial response,
+		 * so invalidate everything.
+		 *
+		 * We only want to run the post index changed hook if
+		 * we've actually changed entries, so keep track if we
+		 * actually changed entries or not.
+		 */
 		int is_cache_changed = 0;
-		/* Mark all entries invalid */
+
 		for (i = 0; i < istate->cache_nr; i++) {
 			if (istate->cache[i]->ce_flags & CE_FSMONITOR_VALID) {
 				is_cache_changed = 1;
@@ -330,7 +345,10 @@ void refresh_fsmonitor(struct index_state *istate)
 			}
 		}
 
-		/* If we're going to check every file, ensure we save the results */
+		/*
+		 * If we're going to check every file, ensure we save
+		 * the results.
+		 */
 		if (is_cache_changed)
 			istate->cache_changed |= FSMONITOR_CHANGED;
 
-- 
gitgitgadget

[PATCH v6 02/30] fsmonitor-ipc: create client routines for git-fsmonitor--daemon

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Create fsmonitor_ipc__*() client routines to spawn the built-in file
system monitor daemon and send it an IPC request using the `Simple
IPC` API.

Stub in empty fsmonitor_ipc__*() functions for unsupported platforms.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 Makefile        |   1 +
 fsmonitor-ipc.c | 171 ++++++++++++++++++++++++++++++++++++++++++++++++
 fsmonitor-ipc.h |  48 ++++++++++++++
 3 files changed, 220 insertions(+)
 create mode 100644 fsmonitor-ipc.c
 create mode 100644 fsmonitor-ipc.h

diff --git a/Makefile b/Makefile
index 6f0b4b775fe..a19d850e716 100644
--- a/Makefile
+++ b/Makefile
@@ -907,6 +907,7 @@ LIB_OBJS += fetch-pack.o
 LIB_OBJS += fmt-merge-msg.o
 LIB_OBJS += fsck.o
 LIB_OBJS += fsmonitor.o
+LIB_OBJS += fsmonitor-ipc.o
 LIB_OBJS += gettext.o
 LIB_OBJS += gpg-interface.o
 LIB_OBJS += graph.o
diff --git a/fsmonitor-ipc.c b/fsmonitor-ipc.c
new file mode 100644
index 00000000000..91a535f1219
--- /dev/null
+++ b/fsmonitor-ipc.c
@@ -0,0 +1,171 @@
+#include "cache.h"
+#include "fsmonitor.h"
+#include "simple-ipc.h"
+#include "fsmonitor-ipc.h"
+#include "run-command.h"
+#include "strbuf.h"
+#include "trace2.h"
+
+#ifndef HAVE_FSMONITOR_DAEMON_BACKEND
+
+/*
+ * A trivial implementation of the fsmonitor_ipc__ API for unsupported
+ * platforms.
+ */
+
+int fsmonitor_ipc__is_supported(void)
+{
+	return 0;
+}
+
+const char *fsmonitor_ipc__get_path(void)
+{
+	return NULL;
+}
+
+enum ipc_active_state fsmonitor_ipc__get_state(void)
+{
+	return IPC_STATE__OTHER_ERROR;
+}
+
+int fsmonitor_ipc__send_query(const char *since_token,
+			      struct strbuf *answer)
+{
+	return -1;
+}
+
+int fsmonitor_ipc__send_command(const char *command,
+				struct strbuf *answer)
+{
+	return -1;
+}
+
+#else
+
+int fsmonitor_ipc__is_supported(void)
+{
+	return 1;
+}
+
+GIT_PATH_FUNC(fsmonitor_ipc__get_path, "fsmonitor--daemon.ipc")
+
+enum ipc_active_state fsmonitor_ipc__get_state(void)
+{
+	return ipc_get_active_state(fsmonitor_ipc__get_path());
+}
+
+static int spawn_daemon(void)
+{
+	const char *args[] = { "fsmonitor--daemon", "start", NULL };
+
+	return run_command_v_opt_tr2(args, RUN_COMMAND_NO_STDIN | RUN_GIT_CMD,
+				    "fsmonitor");
+}
+
+int fsmonitor_ipc__send_query(const char *since_token,
+			      struct strbuf *answer)
+{
+	int ret = -1;
+	int tried_to_spawn = 0;
+	enum ipc_active_state state = IPC_STATE__OTHER_ERROR;
+	struct ipc_client_connection *connection = NULL;
+	struct ipc_client_connect_options options
+		= IPC_CLIENT_CONNECT_OPTIONS_INIT;
+	const char *tok = since_token ? since_token : "";
+	size_t tok_len = since_token ? strlen(since_token) : 0;
+
+	options.wait_if_busy = 1;
+	options.wait_if_not_found = 0;
+
+	trace2_region_enter("fsm_client", "query", NULL);
+	trace2_data_string("fsm_client", NULL, "query/command", tok);
+
+try_again:
+	state = ipc_client_try_connect(fsmonitor_ipc__get_path(), &options,
+				       &connection);
+
+	switch (state) {
+	case IPC_STATE__LISTENING:
+		ret = ipc_client_send_command_to_connection(
+			connection, tok, tok_len, answer);
+		ipc_client_close_connection(connection);
+
+		trace2_data_intmax("fsm_client", NULL,
+				   "query/response-length", answer->len);
+		goto done;
+
+	case IPC_STATE__NOT_LISTENING:
+	case IPC_STATE__PATH_NOT_FOUND:
+		if (tried_to_spawn)
+			goto done;
+
+		tried_to_spawn++;
+		if (spawn_daemon())
+			goto done;
+
+		/*
+		 * Try again, but this time give the daemon a chance to
+		 * actually create the pipe/socket.
+		 *
+		 * Granted, the daemon just started so it can't possibly have
+		 * any FS cached yet, so we'll always get a trivial answer.
+		 * BUT the answer should include a new token that can serve
+		 * as the basis for subsequent requests.
+		 */
+		options.wait_if_not_found = 1;
+		goto try_again;
+
+	case IPC_STATE__INVALID_PATH:
+		ret = error(_("fsmonitor_ipc__send_query: invalid path '%s'"),
+			    fsmonitor_ipc__get_path());
+		goto done;
+
+	case IPC_STATE__OTHER_ERROR:
+	default:
+		ret = error(_("fsmonitor_ipc__send_query: unspecified error on '%s'"),
+			    fsmonitor_ipc__get_path());
+		goto done;
+	}
+
+done:
+	trace2_region_leave("fsm_client", "query", NULL);
+
+	return ret;
+}
+
+int fsmonitor_ipc__send_command(const char *command,
+				struct strbuf *answer)
+{
+	struct ipc_client_connection *connection = NULL;
+	struct ipc_client_connect_options options
+		= IPC_CLIENT_CONNECT_OPTIONS_INIT;
+	int ret;
+	enum ipc_active_state state;
+	const char *c = command ? command : "";
+	size_t c_len = command ? strlen(command) : 0;
+
+	strbuf_reset(answer);
+
+	options.wait_if_busy = 1;
+	options.wait_if_not_found = 0;
+
+	state = ipc_client_try_connect(fsmonitor_ipc__get_path(), &options,
+				       &connection);
+	if (state != IPC_STATE__LISTENING) {
+		die("fsmonitor--daemon is not running");
+		return -1;
+	}
+
+	ret = ipc_client_send_command_to_connection(connection, c, c_len,
+						    answer);
+	ipc_client_close_connection(connection);
+
+	if (ret == -1) {
+		die("could not send '%s' command to fsmonitor--daemon", c);
+		return -1;
+	}
+
+	return 0;
+}
+
+#endif
diff --git a/fsmonitor-ipc.h b/fsmonitor-ipc.h
new file mode 100644
index 00000000000..b6a7067c3af
--- /dev/null
+++ b/fsmonitor-ipc.h
@@ -0,0 +1,48 @@
+#ifndef FSMONITOR_IPC_H
+#define FSMONITOR_IPC_H
+
+#include "simple-ipc.h"
+
+/*
+ * Returns true if built-in file system monitor daemon is defined
+ * for this platform.
+ */
+int fsmonitor_ipc__is_supported(void);
+
+/*
+ * Returns the pathname to the IPC named pipe or Unix domain socket
+ * where a `git-fsmonitor--daemon` process will listen.  This is a
+ * per-worktree value.
+ *
+ * Returns NULL if the daemon is not supported on this platform.
+ */
+const char *fsmonitor_ipc__get_path(void);
+
+/*
+ * Try to determine whether there is a `git-fsmonitor--daemon` process
+ * listening on the IPC pipe/socket.
+ */
+enum ipc_active_state fsmonitor_ipc__get_state(void);
+
+/*
+ * Connect to a `git-fsmonitor--daemon` process via simple-ipc
+ * and ask for the set of changed files since the given token.
+ *
+ * Spawn a daemon process in the background if necessary.
+ *
+ * Returns -1 on error; 0 on success.
+ */
+int fsmonitor_ipc__send_query(const char *since_token,
+			      struct strbuf *answer);
+
+/*
+ * Connect to a `git-fsmonitor--daemon` process via simple-ipc and
+ * send a command verb.  If no daemon is available, we DO NOT try to
+ * start one.
+ *
+ * Returns -1 on error; 0 on success.
+ */
+int fsmonitor_ipc__send_command(const char *command,
+				struct strbuf *answer);
+
+#endif /* FSMONITOR_IPC_H */
-- 
gitgitgadget

[PATCH v6 03/30] fsmonitor: config settings are repository-specific

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Move fsmonitor config settings to a new and opaque
`struct fsmonitor_settings` structure.  Add a lazily-loaded pointer
to this into `struct repo_settings`

Create an `enum fsmonitor_mode` type in `struct fsmonitor_settings` to
represent the state of fsmonitor.  This lets us represent which, if
any, fsmonitor provider (hook or IPC) is enabled.

Create `fsm_settings__get_*()` getters to lazily look up fsmonitor-
related config settings.

Get rid of the `core_fsmonitor` global variable.  Move the code to
lookup the existing `core.fsmonitor` config value into the fsmonitor
settings.

Create a hook pathname variable in `struct fsmonitor-settings` and
only set it when in hook mode.

Extend the definition of `core.fsmonitor` to be either a boolean
or a hook pathname.  When true, the builtin FSMonitor is used.
When false or unset, no FSMonitor (neither builtin nor hook) is
used.

The existing `core_fsmonitor` global variable was used to store the
pathname to the fsmonitor hook *and* it was used as a boolean to see
if fsmonitor was enabled.  This dual usage and global visibility leads
to confusion when we add the IPC-based provider.  So lets hide the
details in fsmonitor-settings.c and let it decide which provider to
use in the case of multiple settings.  This avoids cluttering up
repo-settings.c with these private details.

A future commit in builtin-fsmonitor series will add the ability to
disqualify worktrees for various reasons, such as being mounted from a
remote volume, where fsmonitor should not be started.  Having the
config settings hidden in fsmonitor-settings.c allows such worktree
restrictions to override the config values used.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 Makefile               |   1 +
 builtin/update-index.c |   7 ++-
 cache.h                |   1 -
 config.c               |  14 -----
 config.h               |   1 -
 environment.c          |   1 -
 fsmonitor-settings.c   | 115 +++++++++++++++++++++++++++++++++++++++++
 fsmonitor-settings.h   |  21 ++++++++
 fsmonitor.c            |  63 +++++++++++++---------
 fsmonitor.h            |  15 ++++--
 repository.h           |   3 ++
 t/README               |   4 +-
 12 files changed, 197 insertions(+), 49 deletions(-)
 create mode 100644 fsmonitor-settings.c
 create mode 100644 fsmonitor-settings.h

diff --git a/Makefile b/Makefile
index a19d850e716..707a56d4c11 100644
--- a/Makefile
+++ b/Makefile
@@ -908,6 +908,7 @@ LIB_OBJS += fmt-merge-msg.o
 LIB_OBJS += fsck.o
 LIB_OBJS += fsmonitor.o
 LIB_OBJS += fsmonitor-ipc.o
+LIB_OBJS += fsmonitor-settings.o
 LIB_OBJS += gettext.o
 LIB_OBJS += gpg-interface.o
 LIB_OBJS += graph.o
diff --git a/builtin/update-index.c b/builtin/update-index.c
index aafe7eeac2a..876112abb21 100644
--- a/builtin/update-index.c
+++ b/builtin/update-index.c
@@ -1236,14 +1236,17 @@ int cmd_update_index(int argc, const char **argv, const char *prefix)
 	}
 
 	if (fsmonitor > 0) {
-		if (git_config_get_fsmonitor() == 0)
+		enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
+		if (fsm_mode == FSMONITOR_MODE_DISABLED) {
 			warning(_("core.fsmonitor is unset; "
 				"set it if you really want to "
 				"enable fsmonitor"));
+		}
 		add_fsmonitor(&the_index);
 		report(_("fsmonitor enabled"));
 	} else if (!fsmonitor) {
-		if (git_config_get_fsmonitor() == 1)
+		enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
+		if (fsm_mode > FSMONITOR_MODE_DISABLED)
 			warning(_("core.fsmonitor is set; "
 				"remove it if you really want to "
 				"disable fsmonitor"));
diff --git a/cache.h b/cache.h
index 04d4d2db25c..aaf334e2aa4 100644
--- a/cache.h
+++ b/cache.h
@@ -999,7 +999,6 @@ extern int core_preload_index;
 extern int precomposed_unicode;
 extern int protect_hfs;
 extern int protect_ntfs;
-extern const char *core_fsmonitor;
 
 extern int core_apply_sparse_checkout;
 extern int core_sparse_checkout_cone;
diff --git a/config.c b/config.c
index 383b1a4885b..3f9b0739a78 100644
--- a/config.c
+++ b/config.c
@@ -2626,20 +2626,6 @@ int git_config_get_max_percent_split_change(void)
 	return -1; /* default value */
 }
 
-int git_config_get_fsmonitor(void)
-{
-	if (git_config_get_pathname("core.fsmonitor", &core_fsmonitor))
-		core_fsmonitor = getenv("GIT_TEST_FSMONITOR");
-
-	if (core_fsmonitor && !*core_fsmonitor)
-		core_fsmonitor = NULL;
-
-	if (core_fsmonitor)
-		return 1;
-
-	return 0;
-}
-
 int git_config_get_index_threads(int *dest)
 {
 	int is_bool, val;
diff --git a/config.h b/config.h
index bb49baf1ee0..7654f61c634 100644
--- a/config.h
+++ b/config.h
@@ -597,7 +597,6 @@ int git_config_get_pathname(const char *key, const char **dest);
 int git_config_get_index_threads(int *dest);
 int git_config_get_split_index(void);
 int git_config_get_max_percent_split_change(void);
-int git_config_get_fsmonitor(void);
 
 /* This dies if the configured or default date is in the future */
 int git_config_get_expiry(const char *key, const char **output);
diff --git a/environment.c b/environment.c
index fd0501e77a5..00682e638d7 100644
--- a/environment.c
+++ b/environment.c
@@ -84,7 +84,6 @@ int protect_hfs = PROTECT_HFS_DEFAULT;
 #define PROTECT_NTFS_DEFAULT 1
 #endif
 int protect_ntfs = PROTECT_NTFS_DEFAULT;
-const char *core_fsmonitor;
 
 /*
  * The character that begins a commented line in user-editable file
diff --git a/fsmonitor-settings.c b/fsmonitor-settings.c
new file mode 100644
index 00000000000..3b54e7a51f6
--- /dev/null
+++ b/fsmonitor-settings.c
@@ -0,0 +1,115 @@
+#include "cache.h"
+#include "config.h"
+#include "repository.h"
+#include "fsmonitor-settings.h"
+
+/*
+ * We keep this structure defintion private and have getters
+ * for all fields so that we can lazy load it as needed.
+ */
+struct fsmonitor_settings {
+	enum fsmonitor_mode mode;
+	char *hook_path;
+};
+
+static void lookup_fsmonitor_settings(struct repository *r)
+{
+	struct fsmonitor_settings *s;
+	const char *const_str;
+	int bool_value;
+
+	if (r->settings.fsmonitor)
+		return;
+
+	CALLOC_ARRAY(s, 1);
+
+	r->settings.fsmonitor = s;
+
+	fsm_settings__set_disabled(r);
+
+	/*
+	 * Overload the existing "core.fsmonitor" config setting (which
+	 * has historically been either unset or a hook pathname) to
+	 * now allow a boolean value to enable the builtin FSMonitor
+	 * or to turn everything off.  (This does imply that you can't
+	 * use a hook script named "true" or "false", but that's OK.)
+	 */
+	switch (repo_config_get_maybe_bool(r, "core.fsmonitor", &bool_value)) {
+
+	case 0: /* config value was set to <bool> */
+		if (bool_value)
+			fsm_settings__set_ipc(r);
+		return;
+
+	case 1: /* config value was unset */
+		const_str = getenv("GIT_TEST_FSMONITOR");
+		break;
+
+	case -1: /* config value set to an arbitrary string */
+		if (repo_config_get_pathname(r, "core.fsmonitor", &const_str))
+			return; /* should not happen */
+		break;
+
+	default: /* should not happen */
+		return;
+	}
+
+	if (!const_str || !*const_str)
+		return;
+
+	fsm_settings__set_hook(r, const_str);
+}
+
+enum fsmonitor_mode fsm_settings__get_mode(struct repository *r)
+{
+	if (!r)
+		r = the_repository;
+
+	lookup_fsmonitor_settings(r);
+
+	return r->settings.fsmonitor->mode;
+}
+
+const char *fsm_settings__get_hook_path(struct repository *r)
+{
+	if (!r)
+		r = the_repository;
+
+	lookup_fsmonitor_settings(r);
+
+	return r->settings.fsmonitor->hook_path;
+}
+
+void fsm_settings__set_ipc(struct repository *r)
+{
+	if (!r)
+		r = the_repository;
+
+	lookup_fsmonitor_settings(r);
+
+	r->settings.fsmonitor->mode = FSMONITOR_MODE_IPC;
+	FREE_AND_NULL(r->settings.fsmonitor->hook_path);
+}
+
+void fsm_settings__set_hook(struct repository *r, const char *path)
+{
+	if (!r)
+		r = the_repository;
+
+	lookup_fsmonitor_settings(r);
+
+	r->settings.fsmonitor->mode = FSMONITOR_MODE_HOOK;
+	FREE_AND_NULL(r->settings.fsmonitor->hook_path);
+	r->settings.fsmonitor->hook_path = strdup(path);
+}
+
+void fsm_settings__set_disabled(struct repository *r)
+{
+	if (!r)
+		r = the_repository;
+
+	lookup_fsmonitor_settings(r);
+
+	r->settings.fsmonitor->mode = FSMONITOR_MODE_DISABLED;
+	FREE_AND_NULL(r->settings.fsmonitor->hook_path);
+}
diff --git a/fsmonitor-settings.h b/fsmonitor-settings.h
new file mode 100644
index 00000000000..a4c5d7b4889
--- /dev/null
+++ b/fsmonitor-settings.h
@@ -0,0 +1,21 @@
+#ifndef FSMONITOR_SETTINGS_H
+#define FSMONITOR_SETTINGS_H
+
+struct repository;
+
+enum fsmonitor_mode {
+	FSMONITOR_MODE_DISABLED = 0,
+	FSMONITOR_MODE_HOOK = 1, /* core.fsmonitor=<hook_path> */
+	FSMONITOR_MODE_IPC = 2,  /* core.fsmonitor=<true> */
+};
+
+void fsm_settings__set_ipc(struct repository *r);
+void fsm_settings__set_hook(struct repository *r, const char *path);
+void fsm_settings__set_disabled(struct repository *r);
+
+enum fsmonitor_mode fsm_settings__get_mode(struct repository *r);
+const char *fsm_settings__get_hook_path(struct repository *r);
+
+struct fsmonitor_settings;
+
+#endif /* FSMONITOR_SETTINGS_H */
diff --git a/fsmonitor.c b/fsmonitor.c
index 448d0ee33f5..0e961b74d82 100644
--- a/fsmonitor.c
+++ b/fsmonitor.c
@@ -3,6 +3,7 @@
 #include "dir.h"
 #include "ewah/ewok.h"
 #include "fsmonitor.h"
+#include "fsmonitor-ipc.h"
 #include "run-command.h"
 #include "strbuf.h"
 
@@ -148,15 +149,18 @@ void write_fsmonitor_extension(struct strbuf *sb, struct index_state *istate)
 /*
  * Call the query-fsmonitor hook passing the last update token of the saved results.
  */
-static int query_fsmonitor(int version, const char *last_update, struct strbuf *query_result)
+static int query_fsmonitor_hook(struct repository *r,
+				int version,
+				const char *last_update,
+				struct strbuf *query_result)
 {
 	struct child_process cp = CHILD_PROCESS_INIT;
 	int result;
 
-	if (!core_fsmonitor)
+	if (fsm_settings__get_mode(r) != FSMONITOR_MODE_HOOK)
 		return -1;
 
-	strvec_push(&cp.args, core_fsmonitor);
+	strvec_push(&cp.args, fsm_settings__get_hook_path(r));
 	strvec_pushf(&cp.args, "%d", version);
 	strvec_pushf(&cp.args, "%s", last_update);
 	cp.use_shell = 1;
@@ -225,17 +229,28 @@ void refresh_fsmonitor(struct index_state *istate)
 	char *buf;
 	unsigned int i;
 	int is_trivial = 0;
+	struct repository *r = istate->repo ? istate->repo : the_repository;
+	enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
 
-	if (!core_fsmonitor || istate->fsmonitor_has_run_once)
+	if (fsm_mode <= FSMONITOR_MODE_DISABLED ||
+	    istate->fsmonitor_has_run_once)
 		return;
 
-	hook_version = fsmonitor_hook_version();
-
 	istate->fsmonitor_has_run_once = 1;
 
 	trace_printf_key(&trace_fsmonitor, "refresh fsmonitor");
+
+	if (fsm_mode == FSMONITOR_MODE_IPC) {
+		/* TODO */
+		return;
+	}
+
+	assert(fsm_mode == FSMONITOR_MODE_HOOK);
+
+	hook_version = fsmonitor_hook_version();
+
 	/*
-	 * This could be racy so save the date/time now and query_fsmonitor
+	 * This could be racy so save the date/time now and query_fsmonitor_hook
 	 * should be inclusive to ensure we don't miss potential changes.
 	 */
 	last_update = getnanotime();
@@ -243,13 +258,14 @@ void refresh_fsmonitor(struct index_state *istate)
 		strbuf_addf(&last_update_token, "%"PRIu64"", last_update);
 
 	/*
-	 * If we have a last update token, call query_fsmonitor for the set of
+	 * If we have a last update token, call query_fsmonitor_hook for the set of
 	 * changes since that token, else assume everything is possibly dirty
 	 * and check it all.
 	 */
 	if (istate->fsmonitor_last_update) {
 		if (hook_version == -1 || hook_version == HOOK_INTERFACE_VERSION2) {
-			query_success = !query_fsmonitor(HOOK_INTERFACE_VERSION2,
+			query_success = !query_fsmonitor_hook(
+				r, HOOK_INTERFACE_VERSION2,
 				istate->fsmonitor_last_update, &query_result);
 
 			if (query_success) {
@@ -280,7 +296,8 @@ void refresh_fsmonitor(struct index_state *istate)
 		}
 
 		if (hook_version == HOOK_INTERFACE_VERSION1) {
-			query_success = !query_fsmonitor(HOOK_INTERFACE_VERSION1,
+			query_success = !query_fsmonitor_hook(
+				r, HOOK_INTERFACE_VERSION1,
 				istate->fsmonitor_last_update, &query_result);
 			if (query_success)
 				is_trivial = query_result.buf[0] == '/';
@@ -290,9 +307,12 @@ void refresh_fsmonitor(struct index_state *istate)
 			trace2_data_intmax("fsm_hook", NULL,
 					   "query/trivial-response", 1);
 
-		trace_performance_since(last_update, "fsmonitor process '%s'", core_fsmonitor);
-		trace_printf_key(&trace_fsmonitor, "fsmonitor process '%s' returned %s",
-			core_fsmonitor, query_success ? "success" : "failure");
+		trace_performance_since(last_update, "fsmonitor process '%s'",
+					fsm_settings__get_hook_path(r));
+		trace_printf_key(&trace_fsmonitor,
+				 "fsmonitor process '%s' returned %s",
+				 fsm_settings__get_hook_path(r),
+				 query_success ? "success" : "failure");
 	}
 
 	/*
@@ -429,7 +449,8 @@ void remove_fsmonitor(struct index_state *istate)
 void tweak_fsmonitor(struct index_state *istate)
 {
 	unsigned int i;
-	int fsmonitor_enabled = git_config_get_fsmonitor();
+	int fsmonitor_enabled = (fsm_settings__get_mode(istate->repo)
+				 > FSMONITOR_MODE_DISABLED);
 
 	if (istate->fsmonitor_dirty) {
 		if (fsmonitor_enabled) {
@@ -449,16 +470,8 @@ void tweak_fsmonitor(struct index_state *istate)
 		istate->fsmonitor_dirty = NULL;
 	}
 
-	switch (fsmonitor_enabled) {
-	case -1: /* keep: do nothing */
-		break;
-	case 0: /* false */
-		remove_fsmonitor(istate);
-		break;
-	case 1: /* true */
+	if (fsmonitor_enabled)
 		add_fsmonitor(istate);
-		break;
-	default: /* unknown value: do nothing */
-		break;
-	}
+	else
+		remove_fsmonitor(istate);
 }
diff --git a/fsmonitor.h b/fsmonitor.h
index f20d72631d7..3f41f653691 100644
--- a/fsmonitor.h
+++ b/fsmonitor.h
@@ -3,6 +3,7 @@
 
 #include "cache.h"
 #include "dir.h"
+#include "fsmonitor-settings.h"
 
 extern struct trace_key trace_fsmonitor;
 
@@ -57,7 +58,10 @@ int fsmonitor_is_trivial_response(const struct strbuf *query_result);
  */
 static inline int is_fsmonitor_refreshed(const struct index_state *istate)
 {
-	return !core_fsmonitor || istate->fsmonitor_has_run_once;
+	enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(istate->repo);
+
+	return fsm_mode <= FSMONITOR_MODE_DISABLED ||
+		istate->fsmonitor_has_run_once;
 }
 
 /*
@@ -67,7 +71,10 @@ static inline int is_fsmonitor_refreshed(const struct index_state *istate)
  */
 static inline void mark_fsmonitor_valid(struct index_state *istate, struct cache_entry *ce)
 {
-	if (core_fsmonitor && !(ce->ce_flags & CE_FSMONITOR_VALID)) {
+	enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(istate->repo);
+
+	if (fsm_mode > FSMONITOR_MODE_DISABLED &&
+	    !(ce->ce_flags & CE_FSMONITOR_VALID)) {
 		istate->cache_changed = 1;
 		ce->ce_flags |= CE_FSMONITOR_VALID;
 		trace_printf_key(&trace_fsmonitor, "mark_fsmonitor_clean '%s'", ce->name);
@@ -83,7 +90,9 @@ static inline void mark_fsmonitor_valid(struct index_state *istate, struct cache
  */
 static inline void mark_fsmonitor_invalid(struct index_state *istate, struct cache_entry *ce)
 {
-	if (core_fsmonitor) {
+	enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(istate->repo);
+
+	if (fsm_mode > FSMONITOR_MODE_DISABLED) {
 		ce->ce_flags &= ~CE_FSMONITOR_VALID;
 		untracked_cache_invalidate_path(istate, ce->name, 1);
 		trace_printf_key(&trace_fsmonitor, "mark_fsmonitor_invalid '%s'", ce->name);
diff --git a/repository.h b/repository.h
index ca837cb9e91..9bbb4659cc8 100644
--- a/repository.h
+++ b/repository.h
@@ -4,6 +4,7 @@
 #include "path.h"
 
 struct config_set;
+struct fsmonitor_settings;
 struct git_hash_algo;
 struct index_state;
 struct lock_file;
@@ -35,6 +36,8 @@ struct repo_settings {
 	int command_requires_full_index;
 	int sparse_index;
 
+	struct fsmonitor_settings *fsmonitor; /* lazily loaded */
+
 	int index_version;
 	enum untracked_cache_setting core_untracked_cache;
 
diff --git a/t/README b/t/README
index f48e0542cdc..9ffea1d3147 100644
--- a/t/README
+++ b/t/README
@@ -405,8 +405,8 @@ every 'git commit-graph write', as if the `--changed-paths` option was
 passed in.
 
 GIT_TEST_FSMONITOR=$PWD/t7519/fsmonitor-all exercises the fsmonitor
-code path for utilizing a file system monitor to speed up detecting
-new or changed files.
+code paths for utilizing a (hook based) file system monitor to speed up
+detecting new or changed files.
 
 GIT_TEST_INDEX_VERSION=<n> exercises the index read/write code path
 for the index version specified.  Can be set to any valid version
-- 
gitgitgadget

[PATCH v6 04/30] fsmonitor: use IPC to query the builtin FSMonitor daemon

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Use simple IPC to directly communicate with the new builtin file
system monitor daemon when `core.fsmonitor` is set to true.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 fsmonitor.c | 38 ++++++++++++++++++++++++++++++++++++--
 1 file changed, 36 insertions(+), 2 deletions(-)

diff --git a/fsmonitor.c b/fsmonitor.c
index 0e961b74d82..a38b5710eb3 100644
--- a/fsmonitor.c
+++ b/fsmonitor.c
@@ -241,8 +241,41 @@ void refresh_fsmonitor(struct index_state *istate)
 	trace_printf_key(&trace_fsmonitor, "refresh fsmonitor");
 
 	if (fsm_mode == FSMONITOR_MODE_IPC) {
-		/* TODO */
-		return;
+		query_success = !fsmonitor_ipc__send_query(
+			istate->fsmonitor_last_update ?
+			istate->fsmonitor_last_update : "builtin:fake",
+			&query_result);
+		if (query_success) {
+			/*
+			 * The response contains a series of nul terminated
+			 * strings.  The first is the new token.
+			 *
+			 * Use `char *buf` as an interlude to trick the CI
+			 * static analysis to let us use `strbuf_addstr()`
+			 * here (and only copy the token) rather than
+			 * `strbuf_addbuf()`.
+			 */
+			buf = query_result.buf;
+			strbuf_addstr(&last_update_token, buf);
+			bol = last_update_token.len + 1;
+			is_trivial = query_result.buf[bol] == '/';
+			if (is_trivial)
+				trace2_data_intmax("fsm_client", NULL,
+						   "query/trivial-response", 1);
+		} else {
+			/*
+			 * The builtin daemon is not available on this
+			 * platform -OR- we failed to get a response.
+			 *
+			 * Generate a fake token (rather than a V1
+			 * timestamp) for the index extension.  (If
+			 * they switch back to the hook API, we don't
+			 * want ambiguous state.)
+			 */
+			strbuf_addstr(&last_update_token, "builtin:fake");
+		}
+
+		goto apply_results;
 	}
 
 	assert(fsm_mode == FSMONITOR_MODE_HOOK);
@@ -315,6 +348,7 @@ void refresh_fsmonitor(struct index_state *istate)
 				 query_success ? "success" : "failure");
 	}
 
+apply_results:
 	/*
 	 * The response from FSMonitor (excluding the header token) is
 	 * either:
-- 
gitgitgadget

[PATCH v6 05/30] fsmonitor: document builtin fsmonitor

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Document how `core.fsmonitor` can be set to a boolean to enable
or disable the builtin FSMonitor.

Update references to `core.fsmonitor` and `core.fsmonitorHookVersion` and
pointers to `Watchman` to refer to it.

Create `git-fsmonitor--daemon` manual page and describe its features.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 Documentation/config/core.txt           | 60 +++++++++++++++-----
 Documentation/git-fsmonitor--daemon.txt | 75 +++++++++++++++++++++++++
 Documentation/git-update-index.txt      |  8 ++-
 3 files changed, 126 insertions(+), 17 deletions(-)
 create mode 100644 Documentation/git-fsmonitor--daemon.txt

diff --git a/Documentation/config/core.txt b/Documentation/config/core.txt
index c04f62a54a1..6303c36c7ed 100644
--- a/Documentation/config/core.txt
+++ b/Documentation/config/core.txt
@@ -62,22 +62,54 @@ core.protectNTFS::
 	Defaults to `true` on Windows, and `false` elsewhere.
 
 core.fsmonitor::
-	If set, the value of this variable is used as a command which
-	will identify all files that may have changed since the
-	requested date/time. This information is used to speed up git by
-	avoiding unnecessary processing of files that have not changed.
-	See the "fsmonitor-watchman" section of linkgit:githooks[5].
+	If set to true, enable the built-in file system monitor
+	daemon for this working directory (linkgit:git-fsmonitor--daemon[1]).
++
+Like hook-based file system monitors, the built-in file system monitor
+can speed up Git commands that need to refresh the Git index
+(e.g. `git status`) in a working directory with many files.  The
+built-in monitor eliminates the need to install and maintain an
+external third-party tool.
++
+The built-in file system monitor is currently available only on a
+limited set of supported platforms.  Currently, this includes Windows
+and MacOS.
++
+	Otherwise, this variable contains the pathname of the "fsmonitor"
+	hook command.
++
+This hook command is used to identify all files that may have changed
+since the requested date/time. This information is used to speed up
+git by avoiding unnecessary scanning of files that have not changed.
++
+See the "fsmonitor-watchman" section of linkgit:githooks[5].
++
+Note that if you concurrently use multiple versions of Git, such
+as one version on the command line and another version in an IDE
+tool, that the definition of `core.fsmonitor` was extended to
+allow boolean values in addition to hook pathnames.  Git versions
+2.35.1 and prior will not understand the boolean values and will
+consider the "true" or "false" values as hook pathnames to be
+invoked.  Git versions 2.26 thru 2.35.1 default to hook protocol
+V2 and will fall back to no fsmonitor (full scan).  Git versions
+prior to 2.26 default to hook protocol V1 and will silently
+assume there were no changes to report (no scan), so status
+commands may report incomplete results.  For this reason, it is
+best to upgrade all of your Git versions before using the built-in
+file system monitor.
 
 core.fsmonitorHookVersion::
-	Sets the version of hook that is to be used when calling fsmonitor.
-	There are currently versions 1 and 2. When this is not set,
-	version 2 will be tried first and if it fails then version 1
-	will be tried. Version 1 uses a timestamp as input to determine
-	which files have changes since that time but some monitors
-	like watchman have race conditions when used with a timestamp.
-	Version 2 uses an opaque string so that the monitor can return
-	something that can be used to determine what files have changed
-	without race conditions.
+	Sets the protocol version to be used when invoking the
+	"fsmonitor" hook.
++
+There are currently versions 1 and 2. When this is not set,
+version 2 will be tried first and if it fails then version 1
+will be tried. Version 1 uses a timestamp as input to determine
+which files have changes since that time but some monitors
+like Watchman have race conditions when used with a timestamp.
+Version 2 uses an opaque string so that the monitor can return
+something that can be used to determine what files have changed
+without race conditions.
 
 core.trustctime::
 	If false, the ctime differences between the index and the
diff --git a/Documentation/git-fsmonitor--daemon.txt b/Documentation/git-fsmonitor--daemon.txt
new file mode 100644
index 00000000000..0fedf5a4565
--- /dev/null
+++ b/Documentation/git-fsmonitor--daemon.txt
@@ -0,0 +1,75 @@
+git-fsmonitor--daemon(1)
+========================
+
+NAME
+----
+git-fsmonitor--daemon - A Built-in File System Monitor
+
+SYNOPSIS
+--------
+[verse]
+'git fsmonitor--daemon' start
+'git fsmonitor--daemon' run
+'git fsmonitor--daemon' stop
+'git fsmonitor--daemon' status
+
+DESCRIPTION
+-----------
+
+A daemon to watch the working directory for file and directory
+changes using platform-specific file system notification facilities.
+
+This daemon communicates directly with commands like `git status`
+using the link:technical/api-simple-ipc.html[simple IPC] interface
+instead of the slower linkgit:githooks[5] interface.
+
+This daemon is built into Git so that no third-party tools are
+required.
+
+OPTIONS
+-------
+
+start::
+	Starts a daemon in the background.
+
+run::
+	Runs a daemon in the foreground.
+
+stop::
+	Stops the daemon running in the current working
+	directory, if present.
+
+status::
+	Exits with zero status if a daemon is watching the
+	current working directory.
+
+REMARKS
+-------
+
+This daemon is a long running process used to watch a single working
+directory and maintain a list of the recently changed files and
+directories.  Performance of commands such as `git status` can be
+increased if they just ask for a summary of changes to the working
+directory and can avoid scanning the disk.
+
+When `core.fsmonitor` is set to `true` (see linkgit:git-config[1])
+commands, such as `git status`, will ask the daemon for changes and
+automatically start it (if necessary).
+
+For more information see the "File System Monitor" section in
+linkgit:git-update-index[1].
+
+CAVEATS
+-------
+
+The fsmonitor daemon does not currently know about submodules and does
+not know to filter out file system events that happen within a
+submodule.  If fsmonitor daemon is watching a super repo and a file is
+modified within the working directory of a submodule, it will report
+the change (as happening against the super repo).  However, the client
+will properly ignore these extra events, so performance may be affected
+but it will not cause an incorrect result.
+
+GIT
+---
+Part of the linkgit:git[1] suite
diff --git a/Documentation/git-update-index.txt b/Documentation/git-update-index.txt
index 2853f168d97..53ea48a04e2 100644
--- a/Documentation/git-update-index.txt
+++ b/Documentation/git-update-index.txt
@@ -498,7 +498,9 @@ FILE SYSTEM MONITOR
 This feature is intended to speed up git operations for repos that have
 large working directories.
 
-It enables git to work together with a file system monitor (see the
+It enables git to work together with a file system monitor (see
+linkgit:git-fsmonitor--daemon[1]
+and the
 "fsmonitor-watchman" section of linkgit:githooks[5]) that can
 inform it as to what files have been modified. This enables git to avoid
 having to lstat() every file to find modified files.
@@ -509,8 +511,8 @@ looking for new files.
 
 If you want to enable (or disable) this feature, it is easier to use
 the `core.fsmonitor` configuration variable (see
-linkgit:git-config[1]) than using the `--fsmonitor` option to
-`git update-index` in each repository, especially if you want to do so
+linkgit:git-config[1]) than using the `--fsmonitor` option to `git
+update-index` in each repository, especially if you want to do so
 across all repositories you use, because you can set the configuration
 variable in your `$HOME/.gitconfig` just once and have it affect all
 repositories you touch.
-- 
gitgitgadget

[PATCH v6 06/30] fsmonitor--daemon: add a built-in fsmonitor daemon

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Create a built-in file system monitoring daemon that can be used by
the existing `fsmonitor` feature (protocol API and index extension)
to improve the performance of various Git commands, such as `status`.

The `fsmonitor--daemon` feature builds upon the `Simple IPC` API and
provides an alternative to hook access to existing fsmonitors such
as `watchman`.

This commit merely adds the new command without any functionality.

Co-authored-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 .gitignore                  |  1 +
 Makefile                    |  1 +
 builtin.h                   |  1 +
 builtin/fsmonitor--daemon.c | 46 +++++++++++++++++++++++++++++++++++++
 git.c                       |  1 +
 5 files changed, 50 insertions(+)
 create mode 100644 builtin/fsmonitor--daemon.c

diff --git a/.gitignore b/.gitignore
index f817c509ec0..e81de1063a4 100644
--- a/.gitignore
+++ b/.gitignore
@@ -72,6 +72,7 @@
 /git-format-patch
 /git-fsck
 /git-fsck-objects
+/git-fsmonitor--daemon
 /git-gc
 /git-get-tar-commit-id
 /git-grep
diff --git a/Makefile b/Makefile
index 707a56d4c11..5af1d5b112e 100644
--- a/Makefile
+++ b/Makefile
@@ -1114,6 +1114,7 @@ BUILTIN_OBJS += builtin/fmt-merge-msg.o
 BUILTIN_OBJS += builtin/for-each-ref.o
 BUILTIN_OBJS += builtin/for-each-repo.o
 BUILTIN_OBJS += builtin/fsck.o
+BUILTIN_OBJS += builtin/fsmonitor--daemon.o
 BUILTIN_OBJS += builtin/gc.o
 BUILTIN_OBJS += builtin/get-tar-commit-id.o
 BUILTIN_OBJS += builtin/grep.o
diff --git a/builtin.h b/builtin.h
index 83379f3832c..40e9ecc8485 100644
--- a/builtin.h
+++ b/builtin.h
@@ -159,6 +159,7 @@ int cmd_for_each_ref(int argc, const char **argv, const char *prefix);
 int cmd_for_each_repo(int argc, const char **argv, const char *prefix);
 int cmd_format_patch(int argc, const char **argv, const char *prefix);
 int cmd_fsck(int argc, const char **argv, const char *prefix);
+int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix);
 int cmd_gc(int argc, const char **argv, const char *prefix);
 int cmd_get_tar_commit_id(int argc, const char **argv, const char *prefix);
 int cmd_grep(int argc, const char **argv, const char *prefix);
diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
new file mode 100644
index 00000000000..f0498793379
--- /dev/null
+++ b/builtin/fsmonitor--daemon.c
@@ -0,0 +1,46 @@
+#include "builtin.h"
+#include "config.h"
+#include "parse-options.h"
+#include "fsmonitor.h"
+#include "fsmonitor-ipc.h"
+#include "simple-ipc.h"
+#include "khash.h"
+
+static const char * const builtin_fsmonitor__daemon_usage[] = {
+	NULL
+};
+
+#ifdef HAVE_FSMONITOR_DAEMON_BACKEND
+
+int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
+{
+	const char *subcmd;
+
+	struct option options[] = {
+		OPT_END()
+	};
+
+	git_config(git_default_config, NULL);
+
+	argc = parse_options(argc, argv, prefix, options,
+			     builtin_fsmonitor__daemon_usage, 0);
+	if (argc != 1)
+		usage_with_options(builtin_fsmonitor__daemon_usage, options);
+	subcmd = argv[0];
+
+	die(_("Unhandled subcommand '%s'"), subcmd);
+}
+
+#else
+int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
+{
+	struct option options[] = {
+		OPT_END()
+	};
+
+	if (argc == 2 && !strcmp(argv[1], "-h"))
+		usage_with_options(builtin_fsmonitor__daemon_usage, options);
+
+	die(_("fsmonitor--daemon not supported on this platform"));
+}
+#endif
diff --git a/git.c b/git.c
index a25940d72e8..3d8e48cf555 100644
--- a/git.c
+++ b/git.c
@@ -537,6 +537,7 @@ static struct cmd_struct commands[] = {
 	{ "format-patch", cmd_format_patch, RUN_SETUP },
 	{ "fsck", cmd_fsck, RUN_SETUP },
 	{ "fsck-objects", cmd_fsck, RUN_SETUP },
+	{ "fsmonitor--daemon", cmd_fsmonitor__daemon, RUN_SETUP },
 	{ "gc", cmd_gc, RUN_SETUP },
 	{ "get-tar-commit-id", cmd_get_tar_commit_id, NO_PARSEOPT },
 	{ "grep", cmd_grep, RUN_SETUP_GENTLY },
-- 
gitgitgadget

[PATCH v6 07/30] fsmonitor--daemon: implement 'stop' and 'status' commands

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Implement `stop` and `status` client commands to control and query the
status of a `fsmonitor--daemon` server process (and implicitly start a
server process if necessary).

Later commits will implement the actual server and monitor the file
system.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 builtin/fsmonitor--daemon.c | 51 +++++++++++++++++++++++++++++++++++++
 1 file changed, 51 insertions(+)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index f0498793379..5e3178b8bdd 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -7,10 +7,55 @@
 #include "khash.h"
 
 static const char * const builtin_fsmonitor__daemon_usage[] = {
+	N_("git fsmonitor--daemon stop"),
+	N_("git fsmonitor--daemon status"),
 	NULL
 };
 
 #ifdef HAVE_FSMONITOR_DAEMON_BACKEND
+/*
+ * Acting as a CLIENT.
+ *
+ * Send a "quit" command to the `git-fsmonitor--daemon` (if running)
+ * and wait for it to shutdown.
+ */
+static int do_as_client__send_stop(void)
+{
+	struct strbuf answer = STRBUF_INIT;
+	int ret;
+
+	ret = fsmonitor_ipc__send_command("quit", &answer);
+
+	/* The quit command does not return any response data. */
+	strbuf_release(&answer);
+
+	if (ret)
+		return ret;
+
+	trace2_region_enter("fsm_client", "polling-for-daemon-exit", NULL);
+	while (fsmonitor_ipc__get_state() == IPC_STATE__LISTENING)
+		sleep_millisec(50);
+	trace2_region_leave("fsm_client", "polling-for-daemon-exit", NULL);
+
+	return 0;
+}
+
+static int do_as_client__status(void)
+{
+	enum ipc_active_state state = fsmonitor_ipc__get_state();
+
+	switch (state) {
+	case IPC_STATE__LISTENING:
+		printf(_("fsmonitor-daemon is watching '%s'\n"),
+		       the_repository->worktree);
+		return 0;
+
+	default:
+		printf(_("fsmonitor-daemon is not watching '%s'\n"),
+		       the_repository->worktree);
+		return 1;
+	}
+}
 
 int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
 {
@@ -28,6 +73,12 @@ int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
 		usage_with_options(builtin_fsmonitor__daemon_usage, options);
 	subcmd = argv[0];
 
+	if (!strcmp(subcmd, "stop"))
+		return !!do_as_client__send_stop();
+
+	if (!strcmp(subcmd, "status"))
+		return !!do_as_client__status();
+
 	die(_("Unhandled subcommand '%s'"), subcmd);
 }
 
-- 
gitgitgadget

[PATCH v6 08/30] compat/fsmonitor/fsm-listen-win32: stub in backend for Windows

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Stub in empty filesystem listener backend for fsmonitor--daemon on Windows.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 Makefile                            | 13 ++++++++
 compat/fsmonitor/fsm-listen-win32.c | 21 +++++++++++++
 compat/fsmonitor/fsm-listen.h       | 49 +++++++++++++++++++++++++++++
 config.mak.uname                    | 10 ++++++
 contrib/buildsystems/CMakeLists.txt |  7 +++++
 repo-settings.c                     |  1 +
 6 files changed, 101 insertions(+)
 create mode 100644 compat/fsmonitor/fsm-listen-win32.c
 create mode 100644 compat/fsmonitor/fsm-listen.h

diff --git a/Makefile b/Makefile
index 5af1d5b112e..26567d4f772 100644
--- a/Makefile
+++ b/Makefile
@@ -470,6 +470,11 @@ all::
 # directory, and the JSON compilation database 'compile_commands.json' will be
 # created at the root of the repository.
 #
+# If your platform supports a built-in fsmonitor backend, set
+# FSMONITOR_DAEMON_BACKEND to the "<name>" of the corresponding
+# `compat/fsmonitor/fsm-listen-<name>.c` that implements the
+# `fsm_listen__*()` routines.
+#
 # Define DEVELOPER to enable more compiler warnings. Compiler version
 # and family are auto detected, but could be overridden by defining
 # COMPILER_FEATURES (see config.mak.dev). You can still set
@@ -1968,6 +1973,11 @@ ifdef NEED_ACCESS_ROOT_HANDLER
 	COMPAT_OBJS += compat/access.o
 endif
 
+ifdef FSMONITOR_DAEMON_BACKEND
+	COMPAT_CFLAGS += -DHAVE_FSMONITOR_DAEMON_BACKEND
+	COMPAT_OBJS += compat/fsmonitor/fsm-listen-$(FSMONITOR_DAEMON_BACKEND).o
+endif
+
 ifeq ($(TCLTK_PATH),)
 NO_TCLTK = NoThanks
 endif
@@ -2887,6 +2897,9 @@ GIT-BUILD-OPTIONS: FORCE
 	@echo DC_SHA1=\''$(subst ','\'',$(subst ','\'',$(DC_SHA1)))'\' >>$@+
 	@echo SANITIZE_LEAK=\''$(subst ','\'',$(subst ','\'',$(SANITIZE_LEAK)))'\' >>$@+
 	@echo X=\'$(X)\' >>$@+
+ifdef FSMONITOR_DAEMON_BACKEND
+	@echo FSMONITOR_DAEMON_BACKEND=\''$(subst ','\'',$(subst ','\'',$(FSMONITOR_DAEMON_BACKEND)))'\' >>$@+
+endif
 ifdef TEST_OUTPUT_DIRECTORY
 	@echo TEST_OUTPUT_DIRECTORY=\''$(subst ','\'',$(subst ','\'',$(TEST_OUTPUT_DIRECTORY)))'\' >>$@+
 endif
diff --git a/compat/fsmonitor/fsm-listen-win32.c b/compat/fsmonitor/fsm-listen-win32.c
new file mode 100644
index 00000000000..916cbea254f
--- /dev/null
+++ b/compat/fsmonitor/fsm-listen-win32.c
@@ -0,0 +1,21 @@
+#include "cache.h"
+#include "config.h"
+#include "fsmonitor.h"
+#include "fsm-listen.h"
+
+void fsm_listen__stop_async(struct fsmonitor_daemon_state *state)
+{
+}
+
+void fsm_listen__loop(struct fsmonitor_daemon_state *state)
+{
+}
+
+int fsm_listen__ctor(struct fsmonitor_daemon_state *state)
+{
+	return -1;
+}
+
+void fsm_listen__dtor(struct fsmonitor_daemon_state *state)
+{
+}
diff --git a/compat/fsmonitor/fsm-listen.h b/compat/fsmonitor/fsm-listen.h
new file mode 100644
index 00000000000..f0539349baf
--- /dev/null
+++ b/compat/fsmonitor/fsm-listen.h
@@ -0,0 +1,49 @@
+#ifndef FSM_LISTEN_H
+#define FSM_LISTEN_H
+
+/* This needs to be implemented by each backend */
+
+#ifdef HAVE_FSMONITOR_DAEMON_BACKEND
+
+struct fsmonitor_daemon_state;
+
+/*
+ * Initialize platform-specific data for the fsmonitor listener thread.
+ * This will be called from the main thread PRIOR to staring the
+ * fsmonitor_fs_listener thread.
+ *
+ * Returns 0 if successful.
+ * Returns -1 otherwise.
+ */
+int fsm_listen__ctor(struct fsmonitor_daemon_state *state);
+
+/*
+ * Cleanup platform-specific data for the fsmonitor listener thread.
+ * This will be called from the main thread AFTER joining the listener.
+ */
+void fsm_listen__dtor(struct fsmonitor_daemon_state *state);
+
+/*
+ * The main body of the platform-specific event loop to watch for
+ * filesystem events.  This will run in the fsmonitor_fs_listen thread.
+ *
+ * It should call `ipc_server_stop_async()` if the listener thread
+ * prematurely terminates (because of a filesystem error or if it
+ * detects that the .git directory has been deleted).  (It should NOT
+ * do so if the listener thread receives a normal shutdown signal from
+ * the IPC layer.)
+ *
+ * It should set `state->error_code` to -1 if the daemon should exit
+ * with an error.
+ */
+void fsm_listen__loop(struct fsmonitor_daemon_state *state);
+
+/*
+ * Gently request that the fsmonitor listener thread shutdown.
+ * It does not wait for it to stop.  The caller should do a JOIN
+ * to wait for it.
+ */
+void fsm_listen__stop_async(struct fsmonitor_daemon_state *state);
+
+#endif /* HAVE_FSMONITOR_DAEMON_BACKEND */
+#endif /* FSM_LISTEN_H */
diff --git a/config.mak.uname b/config.mak.uname
index 4352ea39e9b..26074f56bed 100644
--- a/config.mak.uname
+++ b/config.mak.uname
@@ -435,6 +435,11 @@ ifeq ($(uname_S),Windows)
 	# so we don't need this:
 	#
 	#   SNPRINTF_RETURNS_BOGUS = YesPlease
+
+	# The builtin FSMonitor requires Named Pipes and Threads on Windows.
+	# These are always available, so we do not have to conditionally
+	# support it.
+	FSMONITOR_DAEMON_BACKEND = win32
 	NO_SVN_TESTS = YesPlease
 	RUNTIME_PREFIX = YesPlease
 	HAVE_WPGMPTR = YesWeDo
@@ -619,6 +624,11 @@ ifeq ($(uname_S),MINGW)
 	NO_STRTOUMAX = YesPlease
 	NO_MKDTEMP = YesPlease
 	NO_SVN_TESTS = YesPlease
+
+	# The builtin FSMonitor requires Named Pipes and Threads on Windows.
+	# These are always available, so we do not have to conditionally
+	# support it.
+	FSMONITOR_DAEMON_BACKEND = win32
 	RUNTIME_PREFIX = YesPlease
 	HAVE_WPGMPTR = YesWeDo
 	NO_ST_BLOCKS_IN_STRUCT_STAT = YesPlease
diff --git a/contrib/buildsystems/CMakeLists.txt b/contrib/buildsystems/CMakeLists.txt
index e44232f85d3..0963629db7f 100644
--- a/contrib/buildsystems/CMakeLists.txt
+++ b/contrib/buildsystems/CMakeLists.txt
@@ -285,6 +285,13 @@ else()
 	endif()
 endif()
 
+if(SUPPORTS_SIMPLE_IPC)
+	if(CMAKE_SYSTEM_NAME STREQUAL "Windows")
+		add_compile_definitions(HAVE_FSMONITOR_DAEMON_BACKEND)
+		list(APPEND compat_SOURCES compat/fsmonitor/fsm-listen-win32.c)
+	endif()
+endif()
+
 set(EXE_EXTENSION ${CMAKE_EXECUTABLE_SUFFIX})
 
 #header checks
diff --git a/repo-settings.c b/repo-settings.c
index b4fbd16cdcc..2dfcb2b6542 100644
--- a/repo-settings.c
+++ b/repo-settings.c
@@ -2,6 +2,7 @@
 #include "config.h"
 #include "repository.h"
 #include "midx.h"
+#include "compat/fsmonitor/fsm-listen.h"
 
 static void repo_cfg_bool(struct repository *r, const char *key, int *dest,
 			  int def)
-- 
gitgitgadget

[PATCH v6 09/30] compat/fsmonitor/fsm-listen-darwin: stub in backend for Darwin

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Stub in empty implementation of fsmonitor--daemon
backend for Darwin (aka MacOS).

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 compat/fsmonitor/fsm-listen-darwin.c | 20 ++++++++++++++++++++
 config.mak.uname                     | 10 ++++++++++
 contrib/buildsystems/CMakeLists.txt  |  3 +++
 3 files changed, 33 insertions(+)
 create mode 100644 compat/fsmonitor/fsm-listen-darwin.c

diff --git a/compat/fsmonitor/fsm-listen-darwin.c b/compat/fsmonitor/fsm-listen-darwin.c
new file mode 100644
index 00000000000..c84e3344ab9
--- /dev/null
+++ b/compat/fsmonitor/fsm-listen-darwin.c
@@ -0,0 +1,20 @@
+#include "cache.h"
+#include "fsmonitor.h"
+#include "fsm-listen.h"
+
+int fsm_listen__ctor(struct fsmonitor_daemon_state *state)
+{
+	return -1;
+}
+
+void fsm_listen__dtor(struct fsmonitor_daemon_state *state)
+{
+}
+
+void fsm_listen__stop_async(struct fsmonitor_daemon_state *state)
+{
+}
+
+void fsm_listen__loop(struct fsmonitor_daemon_state *state)
+{
+}
diff --git a/config.mak.uname b/config.mak.uname
index 26074f56bed..501970902da 100644
--- a/config.mak.uname
+++ b/config.mak.uname
@@ -157,6 +157,16 @@ ifeq ($(uname_S),Darwin)
 			MSGFMT = /usr/local/opt/gettext/bin/msgfmt
 		endif
 	endif
+
+	# The builtin FSMonitor on MacOS builds upon Simple-IPC.  Both require
+	# Unix domain sockets and PThreads.
+	ifndef NO_PTHREADS
+	ifndef NO_UNIX_SOCKETS
+	FSMONITOR_DAEMON_BACKEND = darwin
+	endif
+	endif
+
+	BASIC_LDFLAGS += -framework CoreServices
 endif
 ifeq ($(uname_S),SunOS)
 	NEEDS_SOCKET = YesPlease
diff --git a/contrib/buildsystems/CMakeLists.txt b/contrib/buildsystems/CMakeLists.txt
index 0963629db7f..ee0d7257b77 100644
--- a/contrib/buildsystems/CMakeLists.txt
+++ b/contrib/buildsystems/CMakeLists.txt
@@ -289,6 +289,9 @@ if(SUPPORTS_SIMPLE_IPC)
 	if(CMAKE_SYSTEM_NAME STREQUAL "Windows")
 		add_compile_definitions(HAVE_FSMONITOR_DAEMON_BACKEND)
 		list(APPEND compat_SOURCES compat/fsmonitor/fsm-listen-win32.c)
+	elseif(CMAKE_SYSTEM_NAME STREQUAL "Darwin")
+		add_compile_definitions(HAVE_FSMONITOR_DAEMON_BACKEND)
+		list(APPEND compat_SOURCES compat/fsmonitor/fsm-listen-darwin.c)
 	endif()
 endif()
 
-- 
gitgitgadget

[PATCH v6 10/30] fsmonitor--daemon: implement 'run' command

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Implement `run` command to try to begin listening for file system events.

This version defines the thread structure with a single fsmonitor_fs_listen
thread to watch for file system events and a simple IPC thread pool to
watch for connection from Git clients over a well-known named pipe or
Unix domain socket.

This commit does not actually do anything yet because the platform
backends are still just stubs.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 builtin/fsmonitor--daemon.c | 228 +++++++++++++++++++++++++++++++++++-
 fsmonitor--daemon.h         |  34 ++++++
 2 files changed, 261 insertions(+), 1 deletion(-)
 create mode 100644 fsmonitor--daemon.h

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index 5e3178b8bdd..950614f82f6 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -3,16 +3,52 @@
 #include "parse-options.h"
 #include "fsmonitor.h"
 #include "fsmonitor-ipc.h"
+#include "compat/fsmonitor/fsm-listen.h"
+#include "fsmonitor--daemon.h"
 #include "simple-ipc.h"
 #include "khash.h"
 
 static const char * const builtin_fsmonitor__daemon_usage[] = {
+	N_("git fsmonitor--daemon run [<options>]"),
 	N_("git fsmonitor--daemon stop"),
 	N_("git fsmonitor--daemon status"),
 	NULL
 };
 
 #ifdef HAVE_FSMONITOR_DAEMON_BACKEND
+/*
+ * Global state loaded from config.
+ */
+#define FSMONITOR__IPC_THREADS "fsmonitor.ipcthreads"
+static int fsmonitor__ipc_threads = 8;
+
+#define FSMONITOR__ANNOUNCE_STARTUP "fsmonitor.announcestartup"
+static int fsmonitor__announce_startup = 0;
+
+static int fsmonitor_config(const char *var, const char *value, void *cb)
+{
+	if (!strcmp(var, FSMONITOR__IPC_THREADS)) {
+		int i = git_config_int(var, value);
+		if (i < 1)
+			return error(_("value of '%s' out of range: %d"),
+				     FSMONITOR__IPC_THREADS, i);
+		fsmonitor__ipc_threads = i;
+		return 0;
+	}
+
+	if (!strcmp(var, FSMONITOR__ANNOUNCE_STARTUP)) {
+		int is_bool;
+		int i = git_config_bool_or_int(var, value, &is_bool);
+		if (i < 0)
+			return error(_("value of '%s' not bool or int: %d"),
+				     var, i);
+		fsmonitor__announce_startup = i;
+		return 0;
+	}
+
+	return git_default_config(var, value, cb);
+}
+
 /*
  * Acting as a CLIENT.
  *
@@ -57,15 +93,198 @@ static int do_as_client__status(void)
 	}
 }
 
+static ipc_server_application_cb handle_client;
+
+static int handle_client(void *data,
+			 const char *command, size_t command_len,
+			 ipc_server_reply_cb *reply,
+			 struct ipc_server_reply_data *reply_data)
+{
+	/* struct fsmonitor_daemon_state *state = data; */
+	int result;
+
+	/*
+	 * The Simple IPC API now supports {char*, len} arguments, but
+	 * FSMonitor always uses proper null-terminated strings, so
+	 * we can ignore the command_len argument.  (Trust, but verify.)
+	 */
+	if (command_len != strlen(command))
+		BUG("FSMonitor assumes text messages");
+
+	trace2_region_enter("fsmonitor", "handle_client", the_repository);
+	trace2_data_string("fsmonitor", the_repository, "request", command);
+
+	result = 0; /* TODO Do something here. */
+
+	trace2_region_leave("fsmonitor", "handle_client", the_repository);
+
+	return result;
+}
+
+static void *fsm_listen__thread_proc(void *_state)
+{
+	struct fsmonitor_daemon_state *state = _state;
+
+	trace2_thread_start("fsm-listen");
+
+	trace_printf_key(&trace_fsmonitor, "Watching: worktree '%s'",
+			 state->path_worktree_watch.buf);
+	if (state->nr_paths_watching > 1)
+		trace_printf_key(&trace_fsmonitor, "Watching: gitdir '%s'",
+				 state->path_gitdir_watch.buf);
+
+	fsm_listen__loop(state);
+
+	trace2_thread_exit();
+	return NULL;
+}
+
+static int fsmonitor_run_daemon_1(struct fsmonitor_daemon_state *state)
+{
+	struct ipc_server_opts ipc_opts = {
+		.nr_threads = fsmonitor__ipc_threads,
+
+		/*
+		 * We know that there are no other active threads yet,
+		 * so we can let the IPC layer temporarily chdir() if
+		 * it needs to when creating the server side of the
+		 * Unix domain socket.
+		 */
+		.uds_disallow_chdir = 0
+	};
+
+	/*
+	 * Start the IPC thread pool before the we've started the file
+	 * system event listener thread so that we have the IPC handle
+	 * before we need it.
+	 */
+	if (ipc_server_run_async(&state->ipc_server_data,
+				 fsmonitor_ipc__get_path(), &ipc_opts,
+				 handle_client, state))
+		return error_errno(
+			_("could not start IPC thread pool on '%s'"),
+			fsmonitor_ipc__get_path());
+
+	/*
+	 * Start the fsmonitor listener thread to collect filesystem
+	 * events.
+	 */
+	if (pthread_create(&state->listener_thread, NULL,
+			   fsm_listen__thread_proc, state) < 0) {
+		ipc_server_stop_async(state->ipc_server_data);
+		ipc_server_await(state->ipc_server_data);
+
+		return error(_("could not start fsmonitor listener thread"));
+	}
+
+	/*
+	 * The daemon is now fully functional in background threads.
+	 * Wait for the IPC thread pool to shutdown (whether by client
+	 * request or from filesystem activity).
+	 */
+	ipc_server_await(state->ipc_server_data);
+
+	/*
+	 * The fsmonitor listener thread may have received a shutdown
+	 * event from the IPC thread pool, but it doesn't hurt to tell
+	 * it again.  And wait for it to shutdown.
+	 */
+	fsm_listen__stop_async(state);
+	pthread_join(state->listener_thread, NULL);
+
+	return state->error_code;
+}
+
+static int fsmonitor_run_daemon(void)
+{
+	struct fsmonitor_daemon_state state;
+	int err;
+
+	memset(&state, 0, sizeof(state));
+
+	pthread_mutex_init(&state.main_lock, NULL);
+	state.error_code = 0;
+	state.current_token_data = NULL;
+
+	/* Prepare to (recursively) watch the <worktree-root> directory. */
+	strbuf_init(&state.path_worktree_watch, 0);
+	strbuf_addstr(&state.path_worktree_watch, absolute_path(get_git_work_tree()));
+	state.nr_paths_watching = 1;
+
+	/*
+	 * We create and delete cookie files somewhere inside the .git
+	 * directory to help us keep sync with the file system.  If
+	 * ".git" is not a directory, then <gitdir> is not inside the
+	 * cone of <worktree-root>, so set up a second watch to watch
+	 * the <gitdir> so that we get events for the cookie files.
+	 */
+	strbuf_init(&state.path_gitdir_watch, 0);
+	strbuf_addbuf(&state.path_gitdir_watch, &state.path_worktree_watch);
+	strbuf_addstr(&state.path_gitdir_watch, "/.git");
+	if (!is_directory(state.path_gitdir_watch.buf)) {
+		strbuf_reset(&state.path_gitdir_watch);
+		strbuf_addstr(&state.path_gitdir_watch, absolute_path(get_git_dir()));
+		state.nr_paths_watching = 2;
+	}
+
+	/*
+	 * Confirm that we can create platform-specific resources for the
+	 * filesystem listener before we bother starting all the threads.
+	 */
+	if (fsm_listen__ctor(&state)) {
+		err = error(_("could not initialize listener thread"));
+		goto done;
+	}
+
+	err = fsmonitor_run_daemon_1(&state);
+
+done:
+	pthread_mutex_destroy(&state.main_lock);
+	fsm_listen__dtor(&state);
+
+	ipc_server_free(state.ipc_server_data);
+
+	strbuf_release(&state.path_worktree_watch);
+	strbuf_release(&state.path_gitdir_watch);
+
+	return err;
+}
+
+static int try_to_run_foreground_daemon(void)
+{
+	/*
+	 * Technically, we don't need to probe for an existing daemon
+	 * process, since we could just call `fsmonitor_run_daemon()`
+	 * and let it fail if the pipe/socket is busy.
+	 *
+	 * However, this method gives us a nicer error message for a
+	 * common error case.
+	 */
+	if (fsmonitor_ipc__get_state() == IPC_STATE__LISTENING)
+		die("fsmonitor--daemon is already running '%s'",
+		    the_repository->worktree);
+
+	if (fsmonitor__announce_startup) {
+		fprintf(stderr, _("running fsmonitor-daemon in '%s'\n"),
+			the_repository->worktree);
+		fflush(stderr);
+	}
+
+	return !!fsmonitor_run_daemon();
+}
+
 int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
 {
 	const char *subcmd;
 
 	struct option options[] = {
+		OPT_INTEGER(0, "ipc-threads",
+			    &fsmonitor__ipc_threads,
+			    N_("use <n> ipc worker threads")),
 		OPT_END()
 	};
 
-	git_config(git_default_config, NULL);
+	git_config(fsmonitor_config, NULL);
 
 	argc = parse_options(argc, argv, prefix, options,
 			     builtin_fsmonitor__daemon_usage, 0);
@@ -73,6 +292,13 @@ int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
 		usage_with_options(builtin_fsmonitor__daemon_usage, options);
 	subcmd = argv[0];
 
+	if (fsmonitor__ipc_threads < 1)
+		die(_("invalid 'ipc-threads' value (%d)"),
+		    fsmonitor__ipc_threads);
+
+	if (!strcmp(subcmd, "run"))
+		return !!try_to_run_foreground_daemon();
+
 	if (!strcmp(subcmd, "stop"))
 		return !!do_as_client__send_stop();
 
diff --git a/fsmonitor--daemon.h b/fsmonitor--daemon.h
new file mode 100644
index 00000000000..3009c1a83de
--- /dev/null
+++ b/fsmonitor--daemon.h
@@ -0,0 +1,34 @@
+#ifndef FSMONITOR_DAEMON_H
+#define FSMONITOR_DAEMON_H
+
+#ifdef HAVE_FSMONITOR_DAEMON_BACKEND
+
+#include "cache.h"
+#include "dir.h"
+#include "run-command.h"
+#include "simple-ipc.h"
+#include "thread-utils.h"
+
+struct fsmonitor_batch;
+struct fsmonitor_token_data;
+
+struct fsmonitor_daemon_backend_data; /* opaque platform-specific data */
+
+struct fsmonitor_daemon_state {
+	pthread_t listener_thread;
+	pthread_mutex_t main_lock;
+
+	struct strbuf path_worktree_watch;
+	struct strbuf path_gitdir_watch;
+	int nr_paths_watching;
+
+	struct fsmonitor_token_data *current_token_data;
+
+	int error_code;
+	struct fsmonitor_daemon_backend_data *backend_data;
+
+	struct ipc_server_data *ipc_server_data;
+};
+
+#endif /* HAVE_FSMONITOR_DAEMON_BACKEND */
+#endif /* FSMONITOR_DAEMON_H */
-- 
gitgitgadget

[PATCH v6 11/30] fsmonitor--daemon: implement 'start' command

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Implement 'git fsmonitor--daemon start' command.  This command starts
an instance of 'git fsmonitor--daemon run' in the background using
the new 'start_bg_command()' function.

We avoid the fork-and-call technique on Unix systems in favor of a
fork-and-exec technique.  This gives us more uniform Trace2 child-*
events.  It also makes our usage more consistent with Windows usage.

On Windows, teach 'git fsmonitor--daemon run' to optionally call
'FreeConsole()' to release handles to the inherited Win32 console
(despite being passed invalid handles for stdin/out/err).  Without
this, command prompts and powershell terminal windows could hang
in "exit" until the last background child process exited or released
their Win32 console handle.  (This was not seen with git-bash shells
because they don't have a Win32 console attached to them.)

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 builtin/fsmonitor--daemon.c | 109 +++++++++++++++++++++++++++++++++++-
 1 file changed, 107 insertions(+), 2 deletions(-)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index 950614f82f6..2f721aae016 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -9,6 +9,7 @@
 #include "khash.h"
 
 static const char * const builtin_fsmonitor__daemon_usage[] = {
+	N_("git fsmonitor--daemon start [<options>]"),
 	N_("git fsmonitor--daemon run [<options>]"),
 	N_("git fsmonitor--daemon stop"),
 	N_("git fsmonitor--daemon status"),
@@ -22,6 +23,9 @@ static const char * const builtin_fsmonitor__daemon_usage[] = {
 #define FSMONITOR__IPC_THREADS "fsmonitor.ipcthreads"
 static int fsmonitor__ipc_threads = 8;
 
+#define FSMONITOR__START_TIMEOUT "fsmonitor.starttimeout"
+static int fsmonitor__start_timeout_sec = 60;
+
 #define FSMONITOR__ANNOUNCE_STARTUP "fsmonitor.announcestartup"
 static int fsmonitor__announce_startup = 0;
 
@@ -36,6 +40,15 @@ static int fsmonitor_config(const char *var, const char *value, void *cb)
 		return 0;
 	}
 
+	if (!strcmp(var, FSMONITOR__START_TIMEOUT)) {
+		int i = git_config_int(var, value);
+		if (i < 0)
+			return error(_("value of '%s' out of range: %d"),
+				     FSMONITOR__START_TIMEOUT, i);
+		fsmonitor__start_timeout_sec = i;
+		return 0;
+	}
+
 	if (!strcmp(var, FSMONITOR__ANNOUNCE_STARTUP)) {
 		int is_bool;
 		int i = git_config_bool_or_int(var, value, &is_bool);
@@ -250,7 +263,7 @@ done:
 	return err;
 }
 
-static int try_to_run_foreground_daemon(void)
+static int try_to_run_foreground_daemon(int detach_console)
 {
 	/*
 	 * Technically, we don't need to probe for an existing daemon
@@ -270,17 +283,106 @@ static int try_to_run_foreground_daemon(void)
 		fflush(stderr);
 	}
 
+#ifdef GIT_WINDOWS_NATIVE
+	if (detach_console)
+		FreeConsole();
+#endif
+
 	return !!fsmonitor_run_daemon();
 }
 
+static start_bg_wait_cb bg_wait_cb;
+
+static int bg_wait_cb(const struct child_process *cp, void *cb_data)
+{
+	enum ipc_active_state s = fsmonitor_ipc__get_state();
+
+	switch (s) {
+	case IPC_STATE__LISTENING:
+		/* child is "ready" */
+		return 0;
+
+	case IPC_STATE__NOT_LISTENING:
+	case IPC_STATE__PATH_NOT_FOUND:
+		/* give child more time */
+		return 1;
+
+	default:
+	case IPC_STATE__INVALID_PATH:
+	case IPC_STATE__OTHER_ERROR:
+		/* all the time in world won't help */
+		return -1;
+	}
+}
+
+static int try_to_start_background_daemon(void)
+{
+	struct child_process cp = CHILD_PROCESS_INIT;
+	enum start_bg_result sbgr;
+
+	/*
+	 * Before we try to create a background daemon process, see
+	 * if a daemon process is already listening.  This makes it
+	 * easier for us to report an already-listening error to the
+	 * console, since our spawn/daemon can only report the success
+	 * of creating the background process (and not whether it
+	 * immediately exited).
+	 */
+	if (fsmonitor_ipc__get_state() == IPC_STATE__LISTENING)
+		die("fsmonitor--daemon is already running '%s'",
+		    the_repository->worktree);
+
+	if (fsmonitor__announce_startup) {
+		fprintf(stderr, _("starting fsmonitor-daemon in '%s'\n"),
+			the_repository->worktree);
+		fflush(stderr);
+	}
+
+	cp.git_cmd = 1;
+
+	strvec_push(&cp.args, "fsmonitor--daemon");
+	strvec_push(&cp.args, "run");
+	strvec_push(&cp.args, "--detach");
+	strvec_pushf(&cp.args, "--ipc-threads=%d", fsmonitor__ipc_threads);
+
+	cp.no_stdin = 1;
+	cp.no_stdout = 1;
+	cp.no_stderr = 1;
+
+	sbgr = start_bg_command(&cp, bg_wait_cb, NULL,
+				fsmonitor__start_timeout_sec);
+
+	switch (sbgr) {
+	case SBGR_READY:
+		return 0;
+
+	default:
+	case SBGR_ERROR:
+	case SBGR_CB_ERROR:
+		return error("daemon failed to start");
+
+	case SBGR_TIMEOUT:
+		return error("daemon not online yet");
+
+	case SBGR_DIED:
+		return error("daemon terminated");
+	}
+}
+
 int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
 {
 	const char *subcmd;
+	int detach_console = 0;
 
 	struct option options[] = {
+		OPT_BOOL(0, "detach", &detach_console, N_("detach from console")),
 		OPT_INTEGER(0, "ipc-threads",
 			    &fsmonitor__ipc_threads,
 			    N_("use <n> ipc worker threads")),
+		OPT_INTEGER(0, "start-timeout",
+			    &fsmonitor__start_timeout_sec,
+			    N_("max seconds to wait for background daemon startup")),
+
 		OPT_END()
 	};
 
@@ -296,8 +398,11 @@ int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
 		die(_("invalid 'ipc-threads' value (%d)"),
 		    fsmonitor__ipc_threads);
 
+	if (!strcmp(subcmd, "start"))
+		return !!try_to_start_background_daemon();
+
 	if (!strcmp(subcmd, "run"))
-		return !!try_to_run_foreground_daemon();
+		return !!try_to_run_foreground_daemon(detach_console);
 
 	if (!strcmp(subcmd, "stop"))
 		return !!do_as_client__send_stop();
-- 
gitgitgadget

[PATCH v6 12/30] fsmonitor--daemon: add pathname classification

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach fsmonitor--daemon to classify relative and absolute
pathnames and decide how they should be handled.  This will
be used by the platform-specific backend to respond to each
filesystem event.

When we register for filesystem notifications on a directory,
we get events for everything (recursively) in the directory.
We want to report to clients changes to tracked and untracked
paths within the working directory proper.  We do not want to
report changes within the .git directory, for example.

This classification will be used in a later commit by the
different backends to classify paths as events are received.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 builtin/fsmonitor--daemon.c | 81 ++++++++++++++++++++++++++++++++++
 fsmonitor--daemon.h         | 87 +++++++++++++++++++++++++++++++++++++
 2 files changed, 168 insertions(+)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index 2f721aae016..cb126883832 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -134,6 +134,87 @@ static int handle_client(void *data,
 	return result;
 }
 
+#define FSMONITOR_COOKIE_PREFIX ".fsmonitor-daemon-"
+
+enum fsmonitor_path_type fsmonitor_classify_path_workdir_relative(
+	const char *rel)
+{
+	if (fspathncmp(rel, ".git", 4))
+		return IS_WORKDIR_PATH;
+	rel += 4;
+
+	if (!*rel)
+		return IS_DOT_GIT;
+	if (*rel != '/')
+		return IS_WORKDIR_PATH; /* e.g. .gitignore */
+	rel++;
+
+	if (!fspathncmp(rel, FSMONITOR_COOKIE_PREFIX,
+			strlen(FSMONITOR_COOKIE_PREFIX)))
+		return IS_INSIDE_DOT_GIT_WITH_COOKIE_PREFIX;
+
+	return IS_INSIDE_DOT_GIT;
+}
+
+enum fsmonitor_path_type fsmonitor_classify_path_gitdir_relative(
+	const char *rel)
+{
+	if (!fspathncmp(rel, FSMONITOR_COOKIE_PREFIX,
+			strlen(FSMONITOR_COOKIE_PREFIX)))
+		return IS_INSIDE_GITDIR_WITH_COOKIE_PREFIX;
+
+	return IS_INSIDE_GITDIR;
+}
+
+static enum fsmonitor_path_type try_classify_workdir_abs_path(
+	struct fsmonitor_daemon_state *state,
+	const char *path)
+{
+	const char *rel;
+
+	if (fspathncmp(path, state->path_worktree_watch.buf,
+		       state->path_worktree_watch.len))
+		return IS_OUTSIDE_CONE;
+
+	rel = path + state->path_worktree_watch.len;
+
+	if (!*rel)
+		return IS_WORKDIR_PATH; /* it is the root dir exactly */
+	if (*rel != '/')
+		return IS_OUTSIDE_CONE;
+	rel++;
+
+	return fsmonitor_classify_path_workdir_relative(rel);
+}
+
+enum fsmonitor_path_type fsmonitor_classify_path_absolute(
+	struct fsmonitor_daemon_state *state,
+	const char *path)
+{
+	const char *rel;
+	enum fsmonitor_path_type t;
+
+	t = try_classify_workdir_abs_path(state, path);
+	if (state->nr_paths_watching == 1)
+		return t;
+	if (t != IS_OUTSIDE_CONE)
+		return t;
+
+	if (fspathncmp(path, state->path_gitdir_watch.buf,
+		       state->path_gitdir_watch.len))
+		return IS_OUTSIDE_CONE;
+
+	rel = path + state->path_gitdir_watch.len;
+
+	if (!*rel)
+		return IS_GITDIR; /* it is the <gitdir> exactly */
+	if (*rel != '/')
+		return IS_OUTSIDE_CONE;
+	rel++;
+
+	return fsmonitor_classify_path_gitdir_relative(rel);
+}
+
 static void *fsm_listen__thread_proc(void *_state)
 {
 	struct fsmonitor_daemon_state *state = _state;
diff --git a/fsmonitor--daemon.h b/fsmonitor--daemon.h
index 3009c1a83de..8c3a71a48bd 100644
--- a/fsmonitor--daemon.h
+++ b/fsmonitor--daemon.h
@@ -30,5 +30,92 @@ struct fsmonitor_daemon_state {
 	struct ipc_server_data *ipc_server_data;
 };
 
+/*
+ * Pathname classifications.
+ *
+ * The daemon classifies the pathnames that it receives from file
+ * system notification events into the following categories and uses
+ * that to decide whether clients are told about them.  (And to watch
+ * for file system synchronization events.)
+ *
+ * The daemon only collects and reports on the set of modified paths
+ * within the working directory (proper).
+ *
+ * The client should only care about paths within the working
+ * directory proper (inside the working directory and not ".git" nor
+ * inside of ".git/").  That is, the client has read the index and is
+ * asking for a list of any paths in the working directory that have
+ * been modified since the last token.  The client does not care about
+ * file system changes within the ".git/" directory (such as new loose
+ * objects or packfiles).  So the client will only receive paths that
+ * are classified as IS_WORKDIR_PATH.
+ *
+ * Note that ".git" is usually a directory and is therefore inside
+ * the cone of the FS watch that we have on the working directory root,
+ * so we will also get FS events for disk activity on and within ".git/"
+ * that we need to respond to or filter from the client.
+ *
+ * But Git also allows ".git" to be a *file* that points to a GITDIR
+ * outside of the working directory.  When this happens, we need to
+ * create FS watches on both the working directory root *and* on the
+ * (external) GITDIR root.  (The latter is required because we put
+ * cookie files inside it and use them to sync with the FS event
+ * stream.)
+ *
+ * Note that in the context of this discussion, I'm using "GITDIR"
+ * to only mean an external GITDIR referenced by a ".git" file.
+ *
+ * The platform FS event backends will receive watch-specific
+ * relative paths (except for those OS's that always emit absolute
+ * paths).  We use the following enum and routines to classify each
+ * path so that we know how to handle it.  There is a slight asymmetry
+ * here because ".git/" is inside the working directory and the
+ * (external) GITDIR is not, and therefore how we handle events may
+ * vary slightly, so I have different enums for "IS...DOT_GIT..." and
+ * "IS...GITDIR...".
+ *
+ * The daemon uses the IS_DOT_GIT and IS_GITDIR internally to mean the
+ * exact ".git" file/directory or GITDIR directory.  If the daemon
+ * receives a delete event for either of these paths, it will
+ * automatically shutdown, for example.
+ *
+ * Note that the daemon DOES NOT explicitly watch nor special case the
+ * index.  The daemon does not read the index nor have any internal
+ * index-relative state, so there are no "IS...INDEX..." enum values.
+ */
+enum fsmonitor_path_type {
+	IS_WORKDIR_PATH = 0,
+
+	IS_DOT_GIT,
+	IS_INSIDE_DOT_GIT,
+	IS_INSIDE_DOT_GIT_WITH_COOKIE_PREFIX,
+
+	IS_GITDIR,
+	IS_INSIDE_GITDIR,
+	IS_INSIDE_GITDIR_WITH_COOKIE_PREFIX,
+
+	IS_OUTSIDE_CONE,
+};
+
+/*
+ * Classify a pathname relative to the root of the working directory.
+ */
+enum fsmonitor_path_type fsmonitor_classify_path_workdir_relative(
+	const char *relative_path);
+
+/*
+ * Classify a pathname relative to a <gitdir> that is external to the
+ * worktree directory.
+ */
+enum fsmonitor_path_type fsmonitor_classify_path_gitdir_relative(
+	const char *relative_path);
+
+/*
+ * Classify an absolute pathname received from a filesystem event.
+ */
+enum fsmonitor_path_type fsmonitor_classify_path_absolute(
+	struct fsmonitor_daemon_state *state,
+	const char *path);
+
 #endif /* HAVE_FSMONITOR_DAEMON_BACKEND */
 #endif /* FSMONITOR_DAEMON_H */
-- 
gitgitgadget

[PATCH v6 13/30] fsmonitor--daemon: define token-ids

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach fsmonitor--daemon to create token-ids and define the
overall token naming scheme.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 builtin/fsmonitor--daemon.c | 116 +++++++++++++++++++++++++++++++++++-
 1 file changed, 115 insertions(+), 1 deletion(-)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index cb126883832..c8d1509d87d 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -106,6 +106,120 @@ static int do_as_client__status(void)
 	}
 }
 
+/*
+ * Requests to and from a FSMonitor Protocol V2 provider use an opaque
+ * "token" as a virtual timestamp.  Clients can request a summary of all
+ * created/deleted/modified files relative to a token.  In the response,
+ * clients receive a new token for the next (relative) request.
+ *
+ *
+ * Token Format
+ * ============
+ *
+ * The contents of the token are private and provider-specific.
+ *
+ * For the built-in fsmonitor--daemon, we define a token as follows:
+ *
+ *     "builtin" ":" <token_id> ":" <sequence_nr>
+ *
+ * The "builtin" prefix is used as a namespace to avoid conflicts
+ * with other providers (such as Watchman).
+ *
+ * The <token_id> is an arbitrary OPAQUE string, such as a GUID,
+ * UUID, or {timestamp,pid}.  It is used to group all filesystem
+ * events that happened while the daemon was monitoring (and in-sync
+ * with the filesystem).
+ *
+ *     Unlike FSMonitor Protocol V1, it is not defined as a timestamp
+ *     and does not define less-than/greater-than relationships.
+ *     (There are too many race conditions to rely on file system
+ *     event timestamps.)
+ *
+ * The <sequence_nr> is a simple integer incremented whenever the
+ * daemon needs to make its state public.  For example, if 1000 file
+ * system events come in, but no clients have requested the data,
+ * the daemon can continue to accumulate file changes in the same
+ * bin and does not need to advance the sequence number.  However,
+ * as soon as a client does arrive, the daemon needs to start a new
+ * bin and increment the sequence number.
+ *
+ *     The sequence number serves as the boundary between 2 sets
+ *     of bins -- the older ones that the client has already seen
+ *     and the newer ones that it hasn't.
+ *
+ * When a new <token_id> is created, the <sequence_nr> is reset to
+ * zero.
+ *
+ *
+ * About Token Ids
+ * ===============
+ *
+ * A new token_id is created:
+ *
+ * [1] each time the daemon is started.
+ *
+ * [2] any time that the daemon must re-sync with the filesystem
+ *     (such as when the kernel drops or we miss events on a very
+ *     active volume).
+ *
+ * [3] in response to a client "flush" command (for dropped event
+ *     testing).
+ *
+ * When a new token_id is created, the daemon is free to discard all
+ * cached filesystem events associated with any previous token_ids.
+ * Events associated with a non-current token_id will never be sent
+ * to a client.  A token_id change implicitly means that the daemon
+ * has gap in its event history.
+ *
+ * Therefore, clients that present a token with a stale (non-current)
+ * token_id will always be given a trivial response.
+ */
+struct fsmonitor_token_data {
+	struct strbuf token_id;
+	struct fsmonitor_batch *batch_head;
+	struct fsmonitor_batch *batch_tail;
+	uint64_t client_ref_count;
+};
+
+static struct fsmonitor_token_data *fsmonitor_new_token_data(void)
+{
+	static int test_env_value = -1;
+	static uint64_t flush_count = 0;
+	struct fsmonitor_token_data *token;
+
+	CALLOC_ARRAY(token, 1);
+
+	strbuf_init(&token->token_id, 0);
+	token->batch_head = NULL;
+	token->batch_tail = NULL;
+	token->client_ref_count = 0;
+
+	if (test_env_value < 0)
+		test_env_value = git_env_bool("GIT_TEST_FSMONITOR_TOKEN", 0);
+
+	if (!test_env_value) {
+		struct timeval tv;
+		struct tm tm;
+		time_t secs;
+
+		gettimeofday(&tv, NULL);
+		secs = tv.tv_sec;
+		gmtime_r(&secs, &tm);
+
+		strbuf_addf(&token->token_id,
+			    "%"PRIu64".%d.%4d%02d%02dT%02d%02d%02d.%06ldZ",
+			    flush_count++,
+			    getpid(),
+			    tm.tm_year + 1900, tm.tm_mon + 1, tm.tm_mday,
+			    tm.tm_hour, tm.tm_min, tm.tm_sec,
+			    (long)tv.tv_usec);
+	} else {
+		strbuf_addf(&token->token_id, "test_%08x", test_env_value++);
+	}
+
+	return token;
+}
+
 static ipc_server_application_cb handle_client;
 
 static int handle_client(void *data,
@@ -298,7 +412,7 @@ static int fsmonitor_run_daemon(void)
 
 	pthread_mutex_init(&state.main_lock, NULL);
 	state.error_code = 0;
-	state.current_token_data = NULL;
+	state.current_token_data = fsmonitor_new_token_data();
 
 	/* Prepare to (recursively) watch the <worktree-root> directory. */
 	strbuf_init(&state.path_worktree_watch, 0);
-- 
gitgitgadget

[PATCH v6 14/30] fsmonitor--daemon: create token-based changed path cache

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach fsmonitor--daemon to build a list of changed paths and associate
them with a token-id.  This will be used by the platform-specific
backends to accumulate changed paths in response to filesystem events.

The platform-specific file system listener thread receives file system
events containing one or more changed pathnames (with whatever
bucketing or grouping that is convenient for the file system).  These
paths are accumulated (without locking) by the file system layer into
a `fsmonitor_batch`.

When the file system layer has drained the kernel event queue, it will
"publish" them to our token queue and make them visible to concurrent
client worker threads.  The token layer is free to combine and/or de-dup
paths within these batches for efficient presentation to clients.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 builtin/fsmonitor--daemon.c | 230 +++++++++++++++++++++++++++++++++++-
 fsmonitor--daemon.h         |  40 +++++++
 2 files changed, 268 insertions(+), 2 deletions(-)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index c8d1509d87d..7c44b979035 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -181,17 +181,27 @@ struct fsmonitor_token_data {
 	uint64_t client_ref_count;
 };
 
+struct fsmonitor_batch {
+	struct fsmonitor_batch *next;
+	uint64_t batch_seq_nr;
+	const char **interned_paths;
+	size_t nr, alloc;
+	time_t pinned_time;
+};
+
 static struct fsmonitor_token_data *fsmonitor_new_token_data(void)
 {
 	static int test_env_value = -1;
 	static uint64_t flush_count = 0;
 	struct fsmonitor_token_data *token;
+	struct fsmonitor_batch *batch;
 
 	CALLOC_ARRAY(token, 1);
+	batch = fsmonitor_batch__new();
 
 	strbuf_init(&token->token_id, 0);
-	token->batch_head = NULL;
-	token->batch_tail = NULL;
+	token->batch_head = batch;
+	token->batch_tail = batch;
 	token->client_ref_count = 0;
 
 	if (test_env_value < 0)
@@ -217,9 +227,143 @@ static struct fsmonitor_token_data *fsmonitor_new_token_data(void)
 		strbuf_addf(&token->token_id, "test_%08x", test_env_value++);
 	}
 
+	/*
+	 * We created a new <token_id> and are starting a new series
+	 * of tokens with a zero <seq_nr>.
+	 *
+	 * Since clients cannot guess our new (non test) <token_id>
+	 * they will always receive a trivial response (because of the
+	 * mismatch on the <token_id>).  The trivial response will
+	 * tell them our new <token_id> so that subsequent requests
+	 * will be relative to our new series.  (And when sending that
+	 * response, we pin the current head of the batch list.)
+	 *
+	 * Even if the client correctly guesses the <token_id>, their
+	 * request of "builtin:<token_id>:0" asks for all changes MORE
+	 * RECENT than batch/bin 0.
+	 *
+	 * This implies that it is a waste to accumulate paths in the
+	 * initial batch/bin (because they will never be transmitted).
+	 *
+	 * So the daemon could be running for days and watching the
+	 * file system, but doesn't need to actually accumulate any
+	 * paths UNTIL we need to set a reference point for a later
+	 * relative request.
+	 *
+	 * However, it is very useful for testing to always have a
+	 * reference point set.  Pin batch 0 to force early file system
+	 * events to accumulate.
+	 */
+	if (test_env_value)
+		batch->pinned_time = time(NULL);
+
 	return token;
 }
 
+struct fsmonitor_batch *fsmonitor_batch__new(void)
+{
+	struct fsmonitor_batch *batch;
+
+	CALLOC_ARRAY(batch, 1);
+
+	return batch;
+}
+
+void fsmonitor_batch__free_list(struct fsmonitor_batch *batch)
+{
+	while (batch) {
+		struct fsmonitor_batch *next = batch->next;
+
+		/*
+		 * The actual strings within the array of this batch
+		 * are interned, so we don't own them.  We only own
+		 * the array.
+		 */
+		free(batch->interned_paths);
+		free(batch);
+
+		batch = next;
+	}
+}
+
+void fsmonitor_batch__add_path(struct fsmonitor_batch *batch,
+			       const char *path)
+{
+	const char *interned_path = strintern(path);
+
+	trace_printf_key(&trace_fsmonitor, "event: %s", interned_path);
+
+	ALLOC_GROW(batch->interned_paths, batch->nr + 1, batch->alloc);
+	batch->interned_paths[batch->nr++] = interned_path;
+}
+
+static void fsmonitor_batch__combine(struct fsmonitor_batch *batch_dest,
+				     const struct fsmonitor_batch *batch_src)
+{
+	size_t k;
+
+	ALLOC_GROW(batch_dest->interned_paths,
+		   batch_dest->nr + batch_src->nr + 1,
+		   batch_dest->alloc);
+
+	for (k = 0; k < batch_src->nr; k++)
+		batch_dest->interned_paths[batch_dest->nr++] =
+			batch_src->interned_paths[k];
+}
+
+static void fsmonitor_free_token_data(struct fsmonitor_token_data *token)
+{
+	if (!token)
+		return;
+
+	assert(token->client_ref_count == 0);
+
+	strbuf_release(&token->token_id);
+
+	fsmonitor_batch__free_list(token->batch_head);
+
+	free(token);
+}
+
+/*
+ * Flush all of our cached data about the filesystem.  Call this if we
+ * lose sync with the filesystem and miss some notification events.
+ *
+ * [1] If we are missing events, then we no longer have a complete
+ *     history of the directory (relative to our current start token).
+ *     We should create a new token and start fresh (as if we just
+ *     booted up).
+ *
+ * If there are no concurrent threads reading the current token data
+ * series, we can free it now.  Otherwise, let the last reader free
+ * it.
+ *
+ * Either way, the old token data series is no longer associated with
+ * our state data.
+ */
+static void with_lock__do_force_resync(struct fsmonitor_daemon_state *state)
+{
+	/* assert current thread holding state->main_lock */
+
+	struct fsmonitor_token_data *free_me = NULL;
+	struct fsmonitor_token_data *new_one = NULL;
+
+	new_one = fsmonitor_new_token_data();
+
+	if (state->current_token_data->client_ref_count == 0)
+		free_me = state->current_token_data;
+	state->current_token_data = new_one;
+
+	fsmonitor_free_token_data(free_me);
+}
+
+void fsmonitor_force_resync(struct fsmonitor_daemon_state *state)
+{
+	pthread_mutex_lock(&state->main_lock);
+	with_lock__do_force_resync(state);
+	pthread_mutex_unlock(&state->main_lock);
+}
+
 static ipc_server_application_cb handle_client;
 
 static int handle_client(void *data,
@@ -329,6 +473,81 @@ enum fsmonitor_path_type fsmonitor_classify_path_absolute(
 	return fsmonitor_classify_path_gitdir_relative(rel);
 }
 
+/*
+ * We try to combine small batches at the front of the batch-list to avoid
+ * having a long list.  This hopefully makes it a little easier when we want
+ * to truncate and maintain the list.  However, we don't want the paths array
+ * to just keep growing and growing with realloc, so we insert an arbitrary
+ * limit.
+ */
+#define MY_COMBINE_LIMIT (1024)
+
+void fsmonitor_publish(struct fsmonitor_daemon_state *state,
+		       struct fsmonitor_batch *batch,
+		       const struct string_list *cookie_names)
+{
+	if (!batch && !cookie_names->nr)
+		return;
+
+	pthread_mutex_lock(&state->main_lock);
+
+	if (batch) {
+		struct fsmonitor_batch *head;
+
+		head = state->current_token_data->batch_head;
+		if (!head) {
+			BUG("token does not have batch");
+		} else if (head->pinned_time) {
+			/*
+			 * We cannot alter the current batch list
+			 * because:
+			 *
+			 * [a] it is being transmitted to at least one
+			 * client and the handle_client() thread has a
+			 * ref-count, but not a lock on the batch list
+			 * starting with this item.
+			 *
+			 * [b] it has been transmitted in the past to
+			 * at least one client such that future
+			 * requests are relative to this head batch.
+			 *
+			 * So, we can only prepend a new batch onto
+			 * the front of the list.
+			 */
+			batch->batch_seq_nr = head->batch_seq_nr + 1;
+			batch->next = head;
+			state->current_token_data->batch_head = batch;
+		} else if (!head->batch_seq_nr) {
+			/*
+			 * Batch 0 is unpinned.  See the note in
+			 * `fsmonitor_new_token_data()` about why we
+			 * don't need to accumulate these paths.
+			 */
+			fsmonitor_batch__free_list(batch);
+		} else if (head->nr + batch->nr > MY_COMBINE_LIMIT) {
+			/*
+			 * The head batch in the list has never been
+			 * transmitted to a client, but folding the
+			 * contents of the new batch onto it would
+			 * exceed our arbitrary limit, so just prepend
+			 * the new batch onto the list.
+			 */
+			batch->batch_seq_nr = head->batch_seq_nr + 1;
+			batch->next = head;
+			state->current_token_data->batch_head = batch;
+		} else {
+			/*
+			 * We are free to add the paths in the given
+			 * batch onto the end of the current head batch.
+			 */
+			fsmonitor_batch__combine(head, batch);
+			fsmonitor_batch__free_list(batch);
+		}
+	}
+
+	pthread_mutex_unlock(&state->main_lock);
+}
+
 static void *fsm_listen__thread_proc(void *_state)
 {
 	struct fsmonitor_daemon_state *state = _state;
@@ -343,6 +562,13 @@ static void *fsm_listen__thread_proc(void *_state)
 
 	fsm_listen__loop(state);
 
+	pthread_mutex_lock(&state->main_lock);
+	if (state->current_token_data &&
+	    state->current_token_data->client_ref_count == 0)
+		fsmonitor_free_token_data(state->current_token_data);
+	state->current_token_data = NULL;
+	pthread_mutex_unlock(&state->main_lock);
+
 	trace2_thread_exit();
 	return NULL;
 }
diff --git a/fsmonitor--daemon.h b/fsmonitor--daemon.h
index 8c3a71a48bd..010fbfe60e9 100644
--- a/fsmonitor--daemon.h
+++ b/fsmonitor--daemon.h
@@ -12,6 +12,27 @@
 struct fsmonitor_batch;
 struct fsmonitor_token_data;
 
+/*
+ * Create a new batch of path(s).  The returned batch is considered
+ * private and not linked into the fsmonitor daemon state.  The caller
+ * should fill this batch with one or more paths and then publish it.
+ */
+struct fsmonitor_batch *fsmonitor_batch__new(void);
+
+/*
+ * Free the list of batches starting with this one.
+ */
+void fsmonitor_batch__free_list(struct fsmonitor_batch *batch);
+
+/*
+ * Add this path to this batch of modified files.
+ *
+ * The batch should be private and NOT (yet) linked into the fsmonitor
+ * daemon state and therefore not yet visible to worker threads and so
+ * no locking is required.
+ */
+void fsmonitor_batch__add_path(struct fsmonitor_batch *batch, const char *path);
+
 struct fsmonitor_daemon_backend_data; /* opaque platform-specific data */
 
 struct fsmonitor_daemon_state {
@@ -117,5 +138,24 @@ enum fsmonitor_path_type fsmonitor_classify_path_absolute(
 	struct fsmonitor_daemon_state *state,
 	const char *path);
 
+/*
+ * Prepend the this batch of path(s) onto the list of batches associated
+ * with the current token.  This makes the batch visible to worker threads.
+ *
+ * The caller no longer owns the batch and must not free it.
+ *
+ * Wake up the client threads waiting on these cookies.
+ */
+void fsmonitor_publish(struct fsmonitor_daemon_state *state,
+		       struct fsmonitor_batch *batch,
+		       const struct string_list *cookie_names);
+
+/*
+ * If the platform-specific layer loses sync with the filesystem,
+ * it should call this to invalidate cached data and abort waiting
+ * threads.
+ */
+void fsmonitor_force_resync(struct fsmonitor_daemon_state *state);
+
 #endif /* HAVE_FSMONITOR_DAEMON_BACKEND */
 #endif /* FSMONITOR_DAEMON_H */
-- 
gitgitgadget

[PATCH v6 15/30] compat/fsmonitor/fsm-listen-win32: implement FSMonitor backend on Windows

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach the win32 backend to register a watch on the working tree
root directory (recursively).  Also watch the <gitdir> if it is
not inside the working tree.  And to collect path change notifications
into batches and publish.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 compat/fsmonitor/fsm-listen-win32.c | 565 ++++++++++++++++++++++++++++
 1 file changed, 565 insertions(+)

diff --git a/compat/fsmonitor/fsm-listen-win32.c b/compat/fsmonitor/fsm-listen-win32.c
index 916cbea254f..c2d11acbc1e 100644
--- a/compat/fsmonitor/fsm-listen-win32.c
+++ b/compat/fsmonitor/fsm-listen-win32.c
@@ -2,20 +2,585 @@
 #include "config.h"
 #include "fsmonitor.h"
 #include "fsm-listen.h"
+#include "fsmonitor--daemon.h"
+
+/*
+ * The documentation of ReadDirectoryChangesW() states that the maximum
+ * buffer size is 64K when the monitored directory is remote.
+ *
+ * Larger buffers may be used when the monitored directory is local and
+ * will help us receive events faster from the kernel and avoid dropped
+ * events.
+ *
+ * So we try to use a very large buffer and silently fallback to 64K if
+ * we get an error.
+ */
+#define MAX_RDCW_BUF_FALLBACK (65536)
+#define MAX_RDCW_BUF          (65536 * 8)
+
+struct one_watch
+{
+	char buffer[MAX_RDCW_BUF];
+	DWORD buf_len;
+	DWORD count;
+
+	struct strbuf path;
+	HANDLE hDir;
+	HANDLE hEvent;
+	OVERLAPPED overlapped;
+
+	/*
+	 * Is there an active ReadDirectoryChangesW() call pending.  If so, we
+	 * need to later call GetOverlappedResult() and possibly CancelIoEx().
+	 */
+	BOOL is_active;
+};
+
+struct fsmonitor_daemon_backend_data
+{
+	struct one_watch *watch_worktree;
+	struct one_watch *watch_gitdir;
+
+	HANDLE hEventShutdown;
+
+	HANDLE hListener[3]; /* we don't own these handles */
+#define LISTENER_SHUTDOWN 0
+#define LISTENER_HAVE_DATA_WORKTREE 1
+#define LISTENER_HAVE_DATA_GITDIR 2
+	int nr_listener_handles;
+};
+
+/*
+ * Convert the WCHAR path from the notification into UTF8 and
+ * then normalize it.
+ */
+static int normalize_path_in_utf8(FILE_NOTIFY_INFORMATION *info,
+				  struct strbuf *normalized_path)
+{
+	int reserve;
+	int len = 0;
+
+	strbuf_reset(normalized_path);
+	if (!info->FileNameLength)
+		goto normalize;
+
+	/*
+	 * Pre-reserve enough space in the UTF8 buffer for
+	 * each Unicode WCHAR character to be mapped into a
+	 * sequence of 2 UTF8 characters.  That should let us
+	 * avoid ERROR_INSUFFICIENT_BUFFER 99.9+% of the time.
+	 */
+	reserve = info->FileNameLength + 1;
+	strbuf_grow(normalized_path, reserve);
+
+	for (;;) {
+		len = WideCharToMultiByte(CP_UTF8, 0, info->FileName,
+					  info->FileNameLength / sizeof(WCHAR),
+					  normalized_path->buf,
+					  strbuf_avail(normalized_path) - 1,
+					  NULL, NULL);
+		if (len > 0)
+			goto normalize;
+		if (GetLastError() != ERROR_INSUFFICIENT_BUFFER) {
+			error("[GLE %ld] could not convert path to UTF-8: '%.*ls'",
+			      GetLastError(),
+			      (int)(info->FileNameLength / sizeof(WCHAR)),
+			      info->FileName);
+			return -1;
+		}
+
+		strbuf_grow(normalized_path,
+			    strbuf_avail(normalized_path) + reserve);
+	}
+
+normalize:
+	strbuf_setlen(normalized_path, len);
+	return strbuf_normalize_path(normalized_path);
+}
 
 void fsm_listen__stop_async(struct fsmonitor_daemon_state *state)
 {
+	SetEvent(state->backend_data->hListener[LISTENER_SHUTDOWN]);
+}
+
+static struct one_watch *create_watch(struct fsmonitor_daemon_state *state,
+				      const char *path)
+{
+	struct one_watch *watch = NULL;
+	DWORD desired_access = FILE_LIST_DIRECTORY;
+	DWORD share_mode =
+		FILE_SHARE_WRITE | FILE_SHARE_READ | FILE_SHARE_DELETE;
+	HANDLE hDir;
+	wchar_t wpath[MAX_PATH];
+
+	if (xutftowcs_path(wpath, path) < 0) {
+		error(_("could not convert to wide characters: '%s'"), path);
+		return NULL;
+	}
+
+	hDir = CreateFileW(wpath,
+			   desired_access, share_mode, NULL, OPEN_EXISTING,
+			   FILE_FLAG_BACKUP_SEMANTICS | FILE_FLAG_OVERLAPPED,
+			   NULL);
+	if (hDir == INVALID_HANDLE_VALUE) {
+		error(_("[GLE %ld] could not watch '%s'"),
+		      GetLastError(), path);
+		return NULL;
+	}
+
+	CALLOC_ARRAY(watch, 1);
+
+	watch->buf_len = sizeof(watch->buffer); /* assume full MAX_RDCW_BUF */
+
+	strbuf_init(&watch->path, 0);
+	strbuf_addstr(&watch->path, path);
+
+	watch->hDir = hDir;
+	watch->hEvent = CreateEvent(NULL, TRUE, FALSE, NULL);
+
+	return watch;
+}
+
+static void destroy_watch(struct one_watch *watch)
+{
+	if (!watch)
+		return;
+
+	strbuf_release(&watch->path);
+	if (watch->hDir != INVALID_HANDLE_VALUE)
+		CloseHandle(watch->hDir);
+	if (watch->hEvent != INVALID_HANDLE_VALUE)
+		CloseHandle(watch->hEvent);
+
+	free(watch);
+}
+
+static int start_rdcw_watch(struct fsmonitor_daemon_backend_data *data,
+			    struct one_watch *watch)
+{
+	DWORD dwNotifyFilter =
+		FILE_NOTIFY_CHANGE_FILE_NAME |
+		FILE_NOTIFY_CHANGE_DIR_NAME |
+		FILE_NOTIFY_CHANGE_ATTRIBUTES |
+		FILE_NOTIFY_CHANGE_SIZE |
+		FILE_NOTIFY_CHANGE_LAST_WRITE |
+		FILE_NOTIFY_CHANGE_CREATION;
+
+	ResetEvent(watch->hEvent);
+
+	memset(&watch->overlapped, 0, sizeof(watch->overlapped));
+	watch->overlapped.hEvent = watch->hEvent;
+
+	/*
+	 * Queue an async call using Overlapped IO.  This returns immediately.
+	 * Our event handle will be signalled when the real result is available.
+	 *
+	 * The return value here just means that we successfully queued it.
+	 * We won't know if the Read...() actually produces data until later.
+	 */
+	watch->is_active = ReadDirectoryChangesW(
+		watch->hDir, watch->buffer, watch->buf_len, TRUE,
+		dwNotifyFilter, &watch->count, &watch->overlapped, NULL);
+
+	if (watch->is_active)
+		return 0;
+
+	error("ReadDirectoryChangedW failed on '%s' [GLE %ld]",
+	      watch->path.buf, GetLastError());
+	return -1;
+}
+
+static int recv_rdcw_watch(struct one_watch *watch)
+{
+	DWORD gle;
+
+	watch->is_active = FALSE;
+
+	/*
+	 * The overlapped result is ready.  If the Read...() was successful
+	 * we finally receive the actual result into our buffer.
+	 */
+	if (GetOverlappedResult(watch->hDir, &watch->overlapped, &watch->count,
+				TRUE))
+		return 0;
+
+	gle = GetLastError();
+	if (gle == ERROR_INVALID_PARAMETER &&
+	    /*
+	     * The kernel throws an invalid parameter error when our
+	     * buffer is too big and we are pointed at a remote
+	     * directory (and possibly for other reasons).  Quietly
+	     * set it down and try again.
+	     *
+	     * See note about MAX_RDCW_BUF at the top.
+	     */
+	    watch->buf_len > MAX_RDCW_BUF_FALLBACK) {
+		watch->buf_len = MAX_RDCW_BUF_FALLBACK;
+		return -2;
+	}
+
+	/*
+	 * NEEDSWORK: If an external <gitdir> is deleted, the above
+	 * returns an error.  I'm not sure that there's anything that
+	 * we can do here other than failing -- the <worktree>/.git
+	 * link file would be broken anyway.  We might try to check
+	 * for that and return a better error message, but I'm not
+	 * sure it is worth it.
+	 */
+
+	error("GetOverlappedResult failed on '%s' [GLE %ld]",
+	      watch->path.buf, gle);
+	return -1;
+}
+
+static void cancel_rdcw_watch(struct one_watch *watch)
+{
+	DWORD count;
+
+	if (!watch || !watch->is_active)
+		return;
+
+	/*
+	 * The calls to ReadDirectoryChangesW() and GetOverlappedResult()
+	 * form a "pair" (my term) where we queue an IO and promise to
+	 * hang around and wait for the kernel to give us the result.
+	 *
+	 * If for some reason after we queue the IO, we have to quit
+	 * or otherwise not stick around for the second half, we must
+	 * tell the kernel to abort the IO.  This prevents the kernel
+	 * from writing to our buffer and/or signalling our event
+	 * after we free them.
+	 *
+	 * (Ask me how much fun it was to track that one down).
+	 */
+	CancelIoEx(watch->hDir, &watch->overlapped);
+	GetOverlappedResult(watch->hDir, &watch->overlapped, &count, TRUE);
+	watch->is_active = FALSE;
+}
+
+/*
+ * Process filesystem events that happen anywhere (recursively) under the
+ * <worktree> root directory.  For a normal working directory, this includes
+ * both version controlled files and the contents of the .git/ directory.
+ *
+ * If <worktree>/.git is a file, then we only see events for the file
+ * itself.
+ */
+static int process_worktree_events(struct fsmonitor_daemon_state *state)
+{
+	struct fsmonitor_daemon_backend_data *data = state->backend_data;
+	struct one_watch *watch = data->watch_worktree;
+	struct strbuf path = STRBUF_INIT;
+	struct string_list cookie_list = STRING_LIST_INIT_DUP;
+	struct fsmonitor_batch *batch = NULL;
+	const char *p = watch->buffer;
+
+	/*
+	 * If the kernel gets more events than will fit in the kernel
+	 * buffer associated with our RDCW handle, it drops them and
+	 * returns a count of zero.
+	 *
+	 * Yes, the call returns WITHOUT error and with length zero.
+	 * This is the documented behavior.  (My testing has confirmed
+	 * that it also sets the last error to ERROR_NOTIFY_ENUM_DIR,
+	 * but we do not rely on that since the function did not
+	 * return an error and it is not documented.)
+	 *
+	 * (The "overflow" case is not ambiguous with the "no data" case
+	 * because we did an INFINITE wait.)
+	 *
+	 * This means we have a gap in coverage.  Tell the daemon layer
+	 * to resync.
+	 */
+	if (!watch->count) {
+		trace2_data_string("fsmonitor", NULL, "fsm-listen/kernel",
+				   "overflow");
+		fsmonitor_force_resync(state);
+		return LISTENER_HAVE_DATA_WORKTREE;
+	}
+
+	/*
+	 * On Windows, `info` contains an "array" of paths that are
+	 * relative to the root of whichever directory handle received
+	 * the event.
+	 */
+	for (;;) {
+		FILE_NOTIFY_INFORMATION *info = (void *)p;
+		const char *slash;
+		enum fsmonitor_path_type t;
+
+		strbuf_reset(&path);
+		if (normalize_path_in_utf8(info, &path) == -1)
+			goto skip_this_path;
+
+		t = fsmonitor_classify_path_workdir_relative(path.buf);
+
+		switch (t) {
+		case IS_INSIDE_DOT_GIT_WITH_COOKIE_PREFIX:
+			/* special case cookie files within .git */
+
+			/* Use just the filename of the cookie file. */
+			slash = find_last_dir_sep(path.buf);
+			string_list_append(&cookie_list,
+					   slash ? slash + 1 : path.buf);
+			break;
+
+		case IS_INSIDE_DOT_GIT:
+			/* ignore everything inside of "<worktree>/.git/" */
+			break;
+
+		case IS_DOT_GIT:
+			/* "<worktree>/.git" was deleted (or renamed away) */
+			if ((info->Action == FILE_ACTION_REMOVED) ||
+			    (info->Action == FILE_ACTION_RENAMED_OLD_NAME)) {
+				trace2_data_string("fsmonitor", NULL,
+						   "fsm-listen/dotgit",
+						   "removed");
+				goto force_shutdown;
+			}
+			break;
+
+		case IS_WORKDIR_PATH:
+			/* queue normal pathname */
+			if (!batch)
+				batch = fsmonitor_batch__new();
+			fsmonitor_batch__add_path(batch, path.buf);
+			break;
+
+		case IS_GITDIR:
+		case IS_INSIDE_GITDIR:
+		case IS_INSIDE_GITDIR_WITH_COOKIE_PREFIX:
+		default:
+			BUG("unexpected path classification '%d' for '%s'",
+			    t, path.buf);
+		}
+
+skip_this_path:
+		if (!info->NextEntryOffset)
+			break;
+		p += info->NextEntryOffset;
+	}
+
+	fsmonitor_publish(state, batch, &cookie_list);
+	batch = NULL;
+	string_list_clear(&cookie_list, 0);
+	strbuf_release(&path);
+	return LISTENER_HAVE_DATA_WORKTREE;
+
+force_shutdown:
+	fsmonitor_batch__free_list(batch);
+	string_list_clear(&cookie_list, 0);
+	strbuf_release(&path);
+	return LISTENER_SHUTDOWN;
+}
+
+/*
+ * Process filesystem events that happened anywhere (recursively) under the
+ * external <gitdir> (such as non-primary worktrees or submodules).
+ * We only care about cookie files that our client threads created here.
+ *
+ * Note that we DO NOT get filesystem events on the external <gitdir>
+ * itself (it is not inside something that we are watching).  In particular,
+ * we do not get an event if the external <gitdir> is deleted.
+ */
+static int process_gitdir_events(struct fsmonitor_daemon_state *state)
+{
+	struct fsmonitor_daemon_backend_data *data = state->backend_data;
+	struct one_watch *watch = data->watch_gitdir;
+	struct strbuf path = STRBUF_INIT;
+	struct string_list cookie_list = STRING_LIST_INIT_DUP;
+	const char *p = watch->buffer;
+
+	if (!watch->count) {
+		trace2_data_string("fsmonitor", NULL, "fsm-listen/kernel",
+				   "overflow");
+		fsmonitor_force_resync(state);
+		return LISTENER_HAVE_DATA_GITDIR;
+	}
+
+	for (;;) {
+		FILE_NOTIFY_INFORMATION *info = (void *)p;
+		const char *slash;
+		enum fsmonitor_path_type t;
+
+		strbuf_reset(&path);
+		if (normalize_path_in_utf8(info, &path) == -1)
+			goto skip_this_path;
+
+		t = fsmonitor_classify_path_gitdir_relative(path.buf);
+
+		switch (t) {
+		case IS_INSIDE_GITDIR_WITH_COOKIE_PREFIX:
+			/* special case cookie files within gitdir */
+
+			/* Use just the filename of the cookie file. */
+			slash = find_last_dir_sep(path.buf);
+			string_list_append(&cookie_list,
+					   slash ? slash + 1 : path.buf);
+			break;
+
+		case IS_INSIDE_GITDIR:
+			goto skip_this_path;
+
+		default:
+			BUG("unexpected path classification '%d' for '%s'",
+			    t, path.buf);
+		}
+
+skip_this_path:
+		if (!info->NextEntryOffset)
+			break;
+		p += info->NextEntryOffset;
+	}
+
+	fsmonitor_publish(state, NULL, &cookie_list);
+	string_list_clear(&cookie_list, 0);
+	strbuf_release(&path);
+	return LISTENER_HAVE_DATA_GITDIR;
 }
 
 void fsm_listen__loop(struct fsmonitor_daemon_state *state)
 {
+	struct fsmonitor_daemon_backend_data *data = state->backend_data;
+	DWORD dwWait;
+	int result;
+
+	state->error_code = 0;
+
+	if (start_rdcw_watch(data, data->watch_worktree) == -1)
+		goto force_error_stop;
+
+	if (data->watch_gitdir &&
+	    start_rdcw_watch(data, data->watch_gitdir) == -1)
+		goto force_error_stop;
+
+	for (;;) {
+		dwWait = WaitForMultipleObjects(data->nr_listener_handles,
+						data->hListener,
+						FALSE, INFINITE);
+
+		if (dwWait == WAIT_OBJECT_0 + LISTENER_HAVE_DATA_WORKTREE) {
+			result = recv_rdcw_watch(data->watch_worktree);
+			if (result == -1) {
+				/* hard error */
+				goto force_error_stop;
+			}
+			if (result == -2) {
+				/* retryable error */
+				if (start_rdcw_watch(data, data->watch_worktree) == -1)
+					goto force_error_stop;
+				continue;
+			}
+
+			/* have data */
+			if (process_worktree_events(state) == LISTENER_SHUTDOWN)
+				goto force_shutdown;
+			if (start_rdcw_watch(data, data->watch_worktree) == -1)
+				goto force_error_stop;
+			continue;
+		}
+
+		if (dwWait == WAIT_OBJECT_0 + LISTENER_HAVE_DATA_GITDIR) {
+			result = recv_rdcw_watch(data->watch_gitdir);
+			if (result == -1) {
+				/* hard error */
+				goto force_error_stop;
+			}
+			if (result == -2) {
+				/* retryable error */
+				if (start_rdcw_watch(data, data->watch_gitdir) == -1)
+					goto force_error_stop;
+				continue;
+			}
+
+			/* have data */
+			if (process_gitdir_events(state) == LISTENER_SHUTDOWN)
+				goto force_shutdown;
+			if (start_rdcw_watch(data, data->watch_gitdir) == -1)
+				goto force_error_stop;
+			continue;
+		}
+
+		if (dwWait == WAIT_OBJECT_0 + LISTENER_SHUTDOWN)
+			goto clean_shutdown;
+
+		error(_("could not read directory changes [GLE %ld]"),
+		      GetLastError());
+		goto force_error_stop;
+	}
+
+force_error_stop:
+	state->error_code = -1;
+
+force_shutdown:
+	/*
+	 * Tell the IPC thead pool to stop (which completes the await
+	 * in the main thread (which will also signal this thread (if
+	 * we are still alive))).
+	 */
+	ipc_server_stop_async(state->ipc_server_data);
+
+clean_shutdown:
+	cancel_rdcw_watch(data->watch_worktree);
+	cancel_rdcw_watch(data->watch_gitdir);
 }
 
 int fsm_listen__ctor(struct fsmonitor_daemon_state *state)
 {
+	struct fsmonitor_daemon_backend_data *data;
+
+	CALLOC_ARRAY(data, 1);
+
+	data->hEventShutdown = CreateEvent(NULL, TRUE, FALSE, NULL);
+
+	data->watch_worktree = create_watch(state,
+					    state->path_worktree_watch.buf);
+	if (!data->watch_worktree)
+		goto failed;
+
+	if (state->nr_paths_watching > 1) {
+		data->watch_gitdir = create_watch(state,
+						  state->path_gitdir_watch.buf);
+		if (!data->watch_gitdir)
+			goto failed;
+	}
+
+	data->hListener[LISTENER_SHUTDOWN] = data->hEventShutdown;
+	data->nr_listener_handles++;
+
+	data->hListener[LISTENER_HAVE_DATA_WORKTREE] =
+		data->watch_worktree->hEvent;
+	data->nr_listener_handles++;
+
+	if (data->watch_gitdir) {
+		data->hListener[LISTENER_HAVE_DATA_GITDIR] =
+			data->watch_gitdir->hEvent;
+		data->nr_listener_handles++;
+	}
+
+	state->backend_data = data;
+	return 0;
+
+failed:
+	CloseHandle(data->hEventShutdown);
+	destroy_watch(data->watch_worktree);
+	destroy_watch(data->watch_gitdir);
+
 	return -1;
 }
 
 void fsm_listen__dtor(struct fsmonitor_daemon_state *state)
 {
+	struct fsmonitor_daemon_backend_data *data;
+
+	if (!state || !state->backend_data)
+		return;
+
+	data = state->backend_data;
+
+	CloseHandle(data->hEventShutdown);
+	destroy_watch(data->watch_worktree);
+	destroy_watch(data->watch_gitdir);
+
+	FREE_AND_NULL(state->backend_data);
 }
-- 
gitgitgadget

[PATCH v6 16/30] compat/fsmonitor/fsm-listen-darwin: add MacOS header files for FSEvent

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Include MacOS system declarations to allow us to use FSEvent and
CoreFoundation APIs.  We need different versions of the declarations
for GCC vs. clang because of compiler and header file conflicts.

While it is quite possible to #include Apple's CoreServices.h when
compiling C source code with clang, trying to build it with GCC
currently fails with this error:

In file included
   from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
   ...Library/Frameworks/Security.framework/Headers/AuthSession.h:32,
   from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
   ...Library/Frameworks/Security.framework/Headers/Security.h:42,
   from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
   ...Library/Frameworks/CoreServices.framework/Frameworks/...
   ...OSServices.framework/Headers/CSIdentity.h:43,
   from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
   ...Library/Frameworks/CoreServices.framework/Frameworks/...
   ...OSServices.framework/Headers/OSServices.h:29,
   from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
   ...Library/Frameworks/CoreServices.framework/Frameworks/...
   ...LaunchServices.framework/Headers/IconsCore.h:23,
   from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
   ...Library/Frameworks/CoreServices.framework/Frameworks/...
   ...LaunchServices.framework/Headers/LaunchServices.h:23,
   from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
   ...Library/Frameworks/CoreServices.framework/Headers/CoreServices.h:45,

     /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
     ...Library/Frameworks/Security.framework/Headers/Authorization.h:193:7:
     error: variably modified 'bytes' at file scope
       193 | char bytes[kAuthorizationExternalFormLength];
           |      ^~~~~

The underlying reason is that GCC (rightfully) objects that an `enum`
value such as `kAuthorizationExternalFormLength` is not a constant
(because it is not, the preprocessor has no knowledge of it, only the
actual C compiler does) and can therefore not be used to define the size
of a C array.

This is a known problem and tracked in GCC's bug tracker:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93082

In the meantime, let's not block things and go the slightly ugly route
of declaring/defining the FSEvents constants, data structures and
functions that we need, so that we can avoid above-mentioned issue.

Let's do this _only_ for GCC, though, so that the CI/PR builds (which
build both with clang and with GCC) can guarantee that we _are_ using
the correct data types.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 compat/fsmonitor/fsm-listen-darwin.c | 96 ++++++++++++++++++++++++++++
 1 file changed, 96 insertions(+)

diff --git a/compat/fsmonitor/fsm-listen-darwin.c b/compat/fsmonitor/fsm-listen-darwin.c
index c84e3344ab9..f76341317dd 100644
--- a/compat/fsmonitor/fsm-listen-darwin.c
+++ b/compat/fsmonitor/fsm-listen-darwin.c
@@ -1,3 +1,99 @@
+#ifndef __clang__
+/*
+ * It is possible to #include CoreFoundation/CoreFoundation.h when compiling
+ * with clang, but not with GCC as of time of writing.
+ *
+ * See https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93082 for details.
+ */
+typedef unsigned int FSEventStreamCreateFlags;
+#define kFSEventStreamEventFlagNone               0x00000000
+#define kFSEventStreamEventFlagMustScanSubDirs    0x00000001
+#define kFSEventStreamEventFlagUserDropped        0x00000002
+#define kFSEventStreamEventFlagKernelDropped      0x00000004
+#define kFSEventStreamEventFlagEventIdsWrapped    0x00000008
+#define kFSEventStreamEventFlagHistoryDone        0x00000010
+#define kFSEventStreamEventFlagRootChanged        0x00000020
+#define kFSEventStreamEventFlagMount              0x00000040
+#define kFSEventStreamEventFlagUnmount            0x00000080
+#define kFSEventStreamEventFlagItemCreated        0x00000100
+#define kFSEventStreamEventFlagItemRemoved        0x00000200
+#define kFSEventStreamEventFlagItemInodeMetaMod   0x00000400
+#define kFSEventStreamEventFlagItemRenamed        0x00000800
+#define kFSEventStreamEventFlagItemModified       0x00001000
+#define kFSEventStreamEventFlagItemFinderInfoMod  0x00002000
+#define kFSEventStreamEventFlagItemChangeOwner    0x00004000
+#define kFSEventStreamEventFlagItemXattrMod       0x00008000
+#define kFSEventStreamEventFlagItemIsFile         0x00010000
+#define kFSEventStreamEventFlagItemIsDir          0x00020000
+#define kFSEventStreamEventFlagItemIsSymlink      0x00040000
+#define kFSEventStreamEventFlagOwnEvent           0x00080000
+#define kFSEventStreamEventFlagItemIsHardlink     0x00100000
+#define kFSEventStreamEventFlagItemIsLastHardlink 0x00200000
+#define kFSEventStreamEventFlagItemCloned         0x00400000
+
+typedef struct __FSEventStream *FSEventStreamRef;
+typedef const FSEventStreamRef ConstFSEventStreamRef;
+
+typedef unsigned int CFStringEncoding;
+#define kCFStringEncodingUTF8 0x08000100
+
+typedef const struct __CFString *CFStringRef;
+typedef const struct __CFArray *CFArrayRef;
+typedef const struct __CFRunLoop *CFRunLoopRef;
+
+struct FSEventStreamContext {
+    long long version;
+    void *cb_data, *retain, *release, *copy_description;
+};
+
+typedef struct FSEventStreamContext FSEventStreamContext;
+typedef unsigned int FSEventStreamEventFlags;
+#define kFSEventStreamCreateFlagNoDefer 0x02
+#define kFSEventStreamCreateFlagWatchRoot 0x04
+#define kFSEventStreamCreateFlagFileEvents 0x10
+
+typedef unsigned long long FSEventStreamEventId;
+#define kFSEventStreamEventIdSinceNow 0xFFFFFFFFFFFFFFFFULL
+
+typedef void (*FSEventStreamCallback)(ConstFSEventStreamRef streamRef,
+				      void *context,
+				      __SIZE_TYPE__ num_of_events,
+				      void *event_paths,
+				      const FSEventStreamEventFlags event_flags[],
+				      const FSEventStreamEventId event_ids[]);
+typedef double CFTimeInterval;
+FSEventStreamRef FSEventStreamCreate(void *allocator,
+				     FSEventStreamCallback callback,
+				     FSEventStreamContext *context,
+				     CFArrayRef paths_to_watch,
+				     FSEventStreamEventId since_when,
+				     CFTimeInterval latency,
+				     FSEventStreamCreateFlags flags);
+CFStringRef CFStringCreateWithCString(void *allocator, const char *string,
+				      CFStringEncoding encoding);
+CFArrayRef CFArrayCreate(void *allocator, const void **items, long long count,
+			 void *callbacks);
+void CFRunLoopRun(void);
+void CFRunLoopStop(CFRunLoopRef run_loop);
+CFRunLoopRef CFRunLoopGetCurrent(void);
+extern CFStringRef kCFRunLoopDefaultMode;
+void FSEventStreamScheduleWithRunLoop(FSEventStreamRef stream,
+				      CFRunLoopRef run_loop,
+				      CFStringRef run_loop_mode);
+unsigned char FSEventStreamStart(FSEventStreamRef stream);
+void FSEventStreamStop(FSEventStreamRef stream);
+void FSEventStreamInvalidate(FSEventStreamRef stream);
+void FSEventStreamRelease(FSEventStreamRef stream);
+#else
+/*
+ * Let Apple's headers declare `isalnum()` first, before
+ * Git's headers override it via a constant
+ */
+#include <string.h>
+#include <CoreFoundation/CoreFoundation.h>
+#include <CoreServices/CoreServices.h>
+#endif
+
 #include "cache.h"
 #include "fsmonitor.h"
 #include "fsm-listen.h"
-- 
gitgitgadget

[PATCH v6 17/30] compat/fsmonitor/fsm-listen-darwin: implement FSEvent listener on MacOS

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Implement file system event listener on MacOS using FSEvent,
CoreFoundation, and CoreServices.

Co-authored-by: Kevin Willford <Kevin.Willford@microsoft.com>
Co-authored-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 compat/fsmonitor/fsm-listen-darwin.c | 383 +++++++++++++++++++++++++++
 1 file changed, 383 insertions(+)

diff --git a/compat/fsmonitor/fsm-listen-darwin.c b/compat/fsmonitor/fsm-listen-darwin.c
index f76341317dd..5c5de1ae702 100644
--- a/compat/fsmonitor/fsm-listen-darwin.c
+++ b/compat/fsmonitor/fsm-listen-darwin.c
@@ -97,20 +97,403 @@ void FSEventStreamRelease(FSEventStreamRef stream);
 #include "cache.h"
 #include "fsmonitor.h"
 #include "fsm-listen.h"
+#include "fsmonitor--daemon.h"
+
+struct fsmonitor_daemon_backend_data
+{
+	CFStringRef cfsr_worktree_path;
+	CFStringRef cfsr_gitdir_path;
+
+	CFArrayRef cfar_paths_to_watch;
+	int nr_paths_watching;
+
+	FSEventStreamRef stream;
+
+	CFRunLoopRef rl;
+
+	enum shutdown_style {
+		SHUTDOWN_EVENT = 0,
+		FORCE_SHUTDOWN,
+		FORCE_ERROR_STOP,
+	} shutdown_style;
+
+	unsigned int stream_scheduled:1;
+	unsigned int stream_started:1;
+};
+
+static void log_flags_set(const char *path, const FSEventStreamEventFlags flag)
+{
+	struct strbuf msg = STRBUF_INIT;
+
+	if (flag & kFSEventStreamEventFlagMustScanSubDirs)
+		strbuf_addstr(&msg, "MustScanSubDirs|");
+	if (flag & kFSEventStreamEventFlagUserDropped)
+		strbuf_addstr(&msg, "UserDropped|");
+	if (flag & kFSEventStreamEventFlagKernelDropped)
+		strbuf_addstr(&msg, "KernelDropped|");
+	if (flag & kFSEventStreamEventFlagEventIdsWrapped)
+		strbuf_addstr(&msg, "EventIdsWrapped|");
+	if (flag & kFSEventStreamEventFlagHistoryDone)
+		strbuf_addstr(&msg, "HistoryDone|");
+	if (flag & kFSEventStreamEventFlagRootChanged)
+		strbuf_addstr(&msg, "RootChanged|");
+	if (flag & kFSEventStreamEventFlagMount)
+		strbuf_addstr(&msg, "Mount|");
+	if (flag & kFSEventStreamEventFlagUnmount)
+		strbuf_addstr(&msg, "Unmount|");
+	if (flag & kFSEventStreamEventFlagItemChangeOwner)
+		strbuf_addstr(&msg, "ItemChangeOwner|");
+	if (flag & kFSEventStreamEventFlagItemCreated)
+		strbuf_addstr(&msg, "ItemCreated|");
+	if (flag & kFSEventStreamEventFlagItemFinderInfoMod)
+		strbuf_addstr(&msg, "ItemFinderInfoMod|");
+	if (flag & kFSEventStreamEventFlagItemInodeMetaMod)
+		strbuf_addstr(&msg, "ItemInodeMetaMod|");
+	if (flag & kFSEventStreamEventFlagItemIsDir)
+		strbuf_addstr(&msg, "ItemIsDir|");
+	if (flag & kFSEventStreamEventFlagItemIsFile)
+		strbuf_addstr(&msg, "ItemIsFile|");
+	if (flag & kFSEventStreamEventFlagItemIsHardlink)
+		strbuf_addstr(&msg, "ItemIsHardlink|");
+	if (flag & kFSEventStreamEventFlagItemIsLastHardlink)
+		strbuf_addstr(&msg, "ItemIsLastHardlink|");
+	if (flag & kFSEventStreamEventFlagItemIsSymlink)
+		strbuf_addstr(&msg, "ItemIsSymlink|");
+	if (flag & kFSEventStreamEventFlagItemModified)
+		strbuf_addstr(&msg, "ItemModified|");
+	if (flag & kFSEventStreamEventFlagItemRemoved)
+		strbuf_addstr(&msg, "ItemRemoved|");
+	if (flag & kFSEventStreamEventFlagItemRenamed)
+		strbuf_addstr(&msg, "ItemRenamed|");
+	if (flag & kFSEventStreamEventFlagItemXattrMod)
+		strbuf_addstr(&msg, "ItemXattrMod|");
+	if (flag & kFSEventStreamEventFlagOwnEvent)
+		strbuf_addstr(&msg, "OwnEvent|");
+	if (flag & kFSEventStreamEventFlagItemCloned)
+		strbuf_addstr(&msg, "ItemCloned|");
+
+	trace_printf_key(&trace_fsmonitor, "fsevent: '%s', flags=%u %s",
+			 path, flag, msg.buf);
+
+	strbuf_release(&msg);
+}
+
+static int ef_is_root_delete(const FSEventStreamEventFlags ef)
+{
+	return (ef & kFSEventStreamEventFlagItemIsDir &&
+		ef & kFSEventStreamEventFlagItemRemoved);
+}
+
+static int ef_is_root_renamed(const FSEventStreamEventFlags ef)
+{
+	return (ef & kFSEventStreamEventFlagItemIsDir &&
+		ef & kFSEventStreamEventFlagItemRenamed);
+}
+
+static int ef_is_dropped(const FSEventStreamEventFlags ef)
+{
+	return (ef & kFSEventStreamEventFlagMustScanSubDirs ||
+		ef & kFSEventStreamEventFlagKernelDropped ||
+		ef & kFSEventStreamEventFlagUserDropped);
+}
+
+static void fsevent_callback(ConstFSEventStreamRef streamRef,
+			     void *ctx,
+			     size_t num_of_events,
+			     void *event_paths,
+			     const FSEventStreamEventFlags event_flags[],
+			     const FSEventStreamEventId event_ids[])
+{
+	struct fsmonitor_daemon_state *state = ctx;
+	struct fsmonitor_daemon_backend_data *data = state->backend_data;
+	char **paths = (char **)event_paths;
+	struct fsmonitor_batch *batch = NULL;
+	struct string_list cookie_list = STRING_LIST_INIT_DUP;
+	const char *path_k;
+	const char *slash;
+	int k;
+	struct strbuf tmp = STRBUF_INIT;
+
+	/*
+	 * Build a list of all filesystem changes into a private/local
+	 * list and without holding any locks.
+	 */
+	for (k = 0; k < num_of_events; k++) {
+		/*
+		 * On Mac, we receive an array of absolute paths.
+		 */
+		path_k = paths[k];
+
+		/*
+		 * If you want to debug FSEvents, log them to GIT_TRACE_FSMONITOR.
+		 * Please don't log them to Trace2.
+		 *
+		 * trace_printf_key(&trace_fsmonitor, "Path: '%s'", path_k);
+		 */
+
+		/*
+		 * If event[k] is marked as dropped, we assume that we have
+		 * lost sync with the filesystem and should flush our cached
+		 * data.  We need to:
+		 *
+		 * [1] Abort/wake any client threads waiting for a cookie and
+		 *     flush the cached state data (the current token), and
+		 *     create a new token.
+		 *
+		 * [2] Discard the batch that we were locally building (since
+		 *     they are conceptually relative to the just flushed
+		 *     token).
+		 */
+		if (ef_is_dropped(event_flags[k])) {
+			if (trace_pass_fl(&trace_fsmonitor))
+				log_flags_set(path_k, event_flags[k]);
+
+			fsmonitor_force_resync(state);
+			fsmonitor_batch__free_list(batch);
+			string_list_clear(&cookie_list, 0);
+
+			/*
+			 * We assume that any events that we received
+			 * in this callback after this dropped event
+			 * may still be valid, so we continue rather
+			 * than break.  (And just in case there is a
+			 * delete of ".git" hiding in there.)
+			 */
+			continue;
+		}
+
+		switch (fsmonitor_classify_path_absolute(state, path_k)) {
+
+		case IS_INSIDE_DOT_GIT_WITH_COOKIE_PREFIX:
+		case IS_INSIDE_GITDIR_WITH_COOKIE_PREFIX:
+			/* special case cookie files within .git or gitdir */
+
+			/* Use just the filename of the cookie file. */
+			slash = find_last_dir_sep(path_k);
+			string_list_append(&cookie_list,
+					   slash ? slash + 1 : path_k);
+			break;
+
+		case IS_INSIDE_DOT_GIT:
+		case IS_INSIDE_GITDIR:
+			/* ignore all other paths inside of .git or gitdir */
+			break;
+
+		case IS_DOT_GIT:
+		case IS_GITDIR:
+			/*
+			 * If .git directory is deleted or renamed away,
+			 * we have to quit.
+			 */
+			if (ef_is_root_delete(event_flags[k])) {
+				trace_printf_key(&trace_fsmonitor,
+						 "event: gitdir removed");
+				goto force_shutdown;
+			}
+			if (ef_is_root_renamed(event_flags[k])) {
+				trace_printf_key(&trace_fsmonitor,
+						 "event: gitdir renamed");
+				goto force_shutdown;
+			}
+			break;
+
+		case IS_WORKDIR_PATH:
+			/* try to queue normal pathnames */
+
+			if (trace_pass_fl(&trace_fsmonitor))
+				log_flags_set(path_k, event_flags[k]);
+
+			/*
+			 * Because of the implicit "binning" (the
+			 * kernel calls us at a given frequency) and
+			 * de-duping (the kernel is free to combine
+			 * multiple events for a given pathname), an
+			 * individual fsevent could be marked as both
+			 * a file and directory.  Add it to the queue
+			 * with both spellings so that the client will
+			 * know how much to invalidate/refresh.
+			 */
+
+			if (event_flags[k] & kFSEventStreamEventFlagItemIsFile) {
+				const char *rel = path_k +
+					state->path_worktree_watch.len + 1;
+
+				if (!batch)
+					batch = fsmonitor_batch__new();
+				fsmonitor_batch__add_path(batch, rel);
+			}
+
+			if (event_flags[k] & kFSEventStreamEventFlagItemIsDir) {
+				const char *rel = path_k +
+					state->path_worktree_watch.len + 1;
+
+				strbuf_reset(&tmp);
+				strbuf_addstr(&tmp, rel);
+				strbuf_addch(&tmp, '/');
+
+				if (!batch)
+					batch = fsmonitor_batch__new();
+				fsmonitor_batch__add_path(batch, tmp.buf);
+			}
+
+			break;
+
+		case IS_OUTSIDE_CONE:
+		default:
+			trace_printf_key(&trace_fsmonitor,
+					 "ignoring '%s'", path_k);
+			break;
+		}
+	}
+
+	fsmonitor_publish(state, batch, &cookie_list);
+	string_list_clear(&cookie_list, 0);
+	strbuf_release(&tmp);
+	return;
+
+force_shutdown:
+	fsmonitor_batch__free_list(batch);
+	string_list_clear(&cookie_list, 0);
+
+	data->shutdown_style = FORCE_SHUTDOWN;
+	CFRunLoopStop(data->rl);
+	strbuf_release(&tmp);
+	return;
+}
+
+/*
+ * In the call to `FSEventStreamCreate()` to setup our watch, the
+ * `latency` argument determines the frequency of calls to our callback
+ * with new FS events.  Too slow and events get dropped; too fast and
+ * we burn CPU unnecessarily.  Since it is rather obscure, I don't
+ * think this needs to be a config setting.  I've done extensive
+ * testing on my systems and chosen the value below.  It gives good
+ * results and I've not seen any dropped events.
+ *
+ * With a latency of 0.1, I was seeing lots of dropped events during
+ * the "touch 100000" files test within t/perf/p7519, but with a
+ * latency of 0.001 I did not see any dropped events.  So I'm going
+ * to assume that this is the "correct" value.
+ *
+ * https://developer.apple.com/documentation/coreservices/1443980-fseventstreamcreate
+ */
 
 int fsm_listen__ctor(struct fsmonitor_daemon_state *state)
 {
+	FSEventStreamCreateFlags flags = kFSEventStreamCreateFlagNoDefer |
+		kFSEventStreamCreateFlagWatchRoot |
+		kFSEventStreamCreateFlagFileEvents;
+	FSEventStreamContext ctx = {
+		0,
+		state,
+		NULL,
+		NULL,
+		NULL
+	};
+	struct fsmonitor_daemon_backend_data *data;
+	const void *dir_array[2];
+
+	CALLOC_ARRAY(data, 1);
+	state->backend_data = data;
+
+	data->cfsr_worktree_path = CFStringCreateWithCString(
+		NULL, state->path_worktree_watch.buf, kCFStringEncodingUTF8);
+	dir_array[data->nr_paths_watching++] = data->cfsr_worktree_path;
+
+	if (state->nr_paths_watching > 1) {
+		data->cfsr_gitdir_path = CFStringCreateWithCString(
+			NULL, state->path_gitdir_watch.buf,
+			kCFStringEncodingUTF8);
+		dir_array[data->nr_paths_watching++] = data->cfsr_gitdir_path;
+	}
+
+	data->cfar_paths_to_watch = CFArrayCreate(NULL, dir_array,
+						  data->nr_paths_watching,
+						  NULL);
+	data->stream = FSEventStreamCreate(NULL, fsevent_callback, &ctx,
+					   data->cfar_paths_to_watch,
+					   kFSEventStreamEventIdSinceNow,
+					   0.001, flags);
+	if (data->stream == NULL)
+		goto failed;
+
+	/*
+	 * `data->rl` needs to be set inside the listener thread.
+	 */
+
+	return 0;
+
+failed:
+	error("Unable to create FSEventStream.");
+
+	FREE_AND_NULL(state->backend_data);
 	return -1;
 }
 
 void fsm_listen__dtor(struct fsmonitor_daemon_state *state)
 {
+	struct fsmonitor_daemon_backend_data *data;
+
+	if (!state || !state->backend_data)
+		return;
+
+	data = state->backend_data;
+
+	if (data->stream) {
+		if (data->stream_started)
+			FSEventStreamStop(data->stream);
+		if (data->stream_scheduled)
+			FSEventStreamInvalidate(data->stream);
+		FSEventStreamRelease(data->stream);
+	}
+
+	FREE_AND_NULL(state->backend_data);
 }
 
 void fsm_listen__stop_async(struct fsmonitor_daemon_state *state)
 {
+	struct fsmonitor_daemon_backend_data *data;
+
+	data = state->backend_data;
+	data->shutdown_style = SHUTDOWN_EVENT;
+
+	CFRunLoopStop(data->rl);
 }
 
 void fsm_listen__loop(struct fsmonitor_daemon_state *state)
 {
+	struct fsmonitor_daemon_backend_data *data;
+
+	data = state->backend_data;
+
+	data->rl = CFRunLoopGetCurrent();
+
+	FSEventStreamScheduleWithRunLoop(data->stream, data->rl, kCFRunLoopDefaultMode);
+	data->stream_scheduled = 1;
+
+	if (!FSEventStreamStart(data->stream)) {
+		error("Failed to start the FSEventStream");
+		goto force_error_stop_without_loop;
+	}
+	data->stream_started = 1;
+
+	CFRunLoopRun();
+
+	switch (data->shutdown_style) {
+	case FORCE_ERROR_STOP:
+		state->error_code = -1;
+		/* fall thru */
+	case FORCE_SHUTDOWN:
+		ipc_server_stop_async(state->ipc_server_data);
+		/* fall thru */
+	case SHUTDOWN_EVENT:
+	default:
+		break;
+	}
+	return;
+
+force_error_stop_without_loop:
+	state->error_code = -1;
+	ipc_server_stop_async(state->ipc_server_data);
+	return;
 }
-- 
gitgitgadget

[PATCH v6 18/30] fsmonitor--daemon: implement handle_client callback

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach fsmonitor--daemon to respond to IPC requests from client
Git processes and respond with a list of modified pathnames
relative to the provided token.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 builtin/fsmonitor--daemon.c | 312 +++++++++++++++++++++++++++++++++++-
 1 file changed, 310 insertions(+), 2 deletions(-)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index 7c44b979035..65c1ef7d4bf 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -7,6 +7,7 @@
 #include "fsmonitor--daemon.h"
 #include "simple-ipc.h"
 #include "khash.h"
+#include "pkt-line.h"
 
 static const char * const builtin_fsmonitor__daemon_usage[] = {
 	N_("git fsmonitor--daemon start [<options>]"),
@@ -364,6 +365,311 @@ void fsmonitor_force_resync(struct fsmonitor_daemon_state *state)
 	pthread_mutex_unlock(&state->main_lock);
 }
 
+/*
+ * Format an opaque token string to send to the client.
+ */
+static void with_lock__format_response_token(
+	struct strbuf *response_token,
+	const struct strbuf *response_token_id,
+	const struct fsmonitor_batch *batch)
+{
+	/* assert current thread holding state->main_lock */
+
+	strbuf_reset(response_token);
+	strbuf_addf(response_token, "builtin:%s:%"PRIu64,
+		    response_token_id->buf, batch->batch_seq_nr);
+}
+
+/*
+ * Parse an opaque token from the client.
+ * Returns -1 on error.
+ */
+static int fsmonitor_parse_client_token(const char *buf_token,
+					struct strbuf *requested_token_id,
+					uint64_t *seq_nr)
+{
+	const char *p;
+	char *p_end;
+
+	strbuf_reset(requested_token_id);
+	*seq_nr = 0;
+
+	if (!skip_prefix(buf_token, "builtin:", &p))
+		return -1;
+
+	while (*p && *p != ':')
+		strbuf_addch(requested_token_id, *p++);
+	if (!*p++)
+		return -1;
+
+	*seq_nr = (uint64_t)strtoumax(p, &p_end, 10);
+	if (*p_end)
+		return -1;
+
+	return 0;
+}
+
+KHASH_INIT(str, const char *, int, 0, kh_str_hash_func, kh_str_hash_equal)
+
+static int do_handle_client(struct fsmonitor_daemon_state *state,
+			    const char *command,
+			    ipc_server_reply_cb *reply,
+			    struct ipc_server_reply_data *reply_data)
+{
+	struct fsmonitor_token_data *token_data = NULL;
+	struct strbuf response_token = STRBUF_INIT;
+	struct strbuf requested_token_id = STRBUF_INIT;
+	struct strbuf payload = STRBUF_INIT;
+	uint64_t requested_oldest_seq_nr = 0;
+	uint64_t total_response_len = 0;
+	const char *p;
+	const struct fsmonitor_batch *batch_head;
+	const struct fsmonitor_batch *batch;
+	intmax_t count = 0, duplicates = 0;
+	kh_str_t *shown;
+	int hash_ret;
+	int do_trivial = 0;
+	int do_flush = 0;
+
+	/*
+	 * We expect `command` to be of the form:
+	 *
+	 * <command> := quit NUL
+	 *            | flush NUL
+	 *            | <V1-time-since-epoch-ns> NUL
+	 *            | <V2-opaque-fsmonitor-token> NUL
+	 */
+
+	if (!strcmp(command, "quit")) {
+		/*
+		 * A client has requested over the socket/pipe that the
+		 * daemon shutdown.
+		 *
+		 * Tell the IPC thread pool to shutdown (which completes
+		 * the await in the main thread (which can stop the
+		 * fsmonitor listener thread)).
+		 *
+		 * There is no reply to the client.
+		 */
+		return SIMPLE_IPC_QUIT;
+
+	} else if (!strcmp(command, "flush")) {
+		/*
+		 * Flush all of our cached data and generate a new token
+		 * just like if we lost sync with the filesystem.
+		 *
+		 * Then send a trivial response using the new token.
+		 */
+		do_flush = 1;
+		do_trivial = 1;
+
+	} else if (!skip_prefix(command, "builtin:", &p)) {
+		/* assume V1 timestamp or garbage */
+
+		char *p_end;
+
+		strtoumax(command, &p_end, 10);
+		trace_printf_key(&trace_fsmonitor,
+				 ((*p_end) ?
+				  "fsmonitor: invalid command line '%s'" :
+				  "fsmonitor: unsupported V1 protocol '%s'"),
+				 command);
+		do_trivial = 1;
+
+	} else {
+		/* We have "builtin:*" */
+		if (fsmonitor_parse_client_token(command, &requested_token_id,
+						 &requested_oldest_seq_nr)) {
+			trace_printf_key(&trace_fsmonitor,
+					 "fsmonitor: invalid V2 protocol token '%s'",
+					 command);
+			do_trivial = 1;
+
+		} else {
+			/*
+			 * We have a V2 valid token:
+			 *     "builtin:<token_id>:<seq_nr>"
+			 */
+		}
+	}
+
+	pthread_mutex_lock(&state->main_lock);
+
+	if (!state->current_token_data)
+		BUG("fsmonitor state does not have a current token");
+
+	if (do_flush)
+		with_lock__do_force_resync(state);
+
+	/*
+	 * We mark the current head of the batch list as "pinned" so
+	 * that the listener thread will treat this item as read-only
+	 * (and prevent any more paths from being added to it) from
+	 * now on.
+	 */
+	token_data = state->current_token_data;
+	batch_head = token_data->batch_head;
+	((struct fsmonitor_batch *)batch_head)->pinned_time = time(NULL);
+
+	/*
+	 * FSMonitor Protocol V2 requires that we send a response header
+	 * with a "new current token" and then all of the paths that changed
+	 * since the "requested token".  We send the seq_nr of the just-pinned
+	 * head batch so that future requests from a client will be relative
+	 * to it.
+	 */
+	with_lock__format_response_token(&response_token,
+					 &token_data->token_id, batch_head);
+
+	reply(reply_data, response_token.buf, response_token.len + 1);
+	total_response_len += response_token.len + 1;
+
+	trace2_data_string("fsmonitor", the_repository, "response/token",
+			   response_token.buf);
+	trace_printf_key(&trace_fsmonitor, "response token: %s",
+			 response_token.buf);
+
+	if (!do_trivial) {
+		if (strcmp(requested_token_id.buf, token_data->token_id.buf)) {
+			/*
+			 * The client last spoke to a different daemon
+			 * instance -OR- the daemon had to resync with
+			 * the filesystem (and lost events), so reject.
+			 */
+			trace2_data_string("fsmonitor", the_repository,
+					   "response/token", "different");
+			do_trivial = 1;
+
+		} else if (requested_oldest_seq_nr <
+			   token_data->batch_tail->batch_seq_nr) {
+			/*
+			 * The client wants older events than we have for
+			 * this token_id.  This means that the end of our
+			 * batch list was truncated and we cannot give the
+			 * client a complete snapshot relative to their
+			 * request.
+			 */
+			trace_printf_key(&trace_fsmonitor,
+					 "client requested truncated data");
+			do_trivial = 1;
+		}
+	}
+
+	if (do_trivial) {
+		pthread_mutex_unlock(&state->main_lock);
+
+		reply(reply_data, "/", 2);
+
+		trace2_data_intmax("fsmonitor", the_repository,
+				   "response/trivial", 1);
+
+		strbuf_release(&response_token);
+		strbuf_release(&requested_token_id);
+		return 0;
+	}
+
+	/*
+	 * We're going to hold onto a pointer to the current
+	 * token-data while we walk the list of batches of files.
+	 * During this time, we will NOT be under the lock.
+	 * So we ref-count it.
+	 *
+	 * This allows the listener thread to continue prepending
+	 * new batches of items to the token-data (which we'll ignore).
+	 *
+	 * AND it allows the listener thread to do a token-reset
+	 * (and install a new `current_token_data`).
+	 */
+	token_data->client_ref_count++;
+
+	pthread_mutex_unlock(&state->main_lock);
+
+	/*
+	 * The client request is relative to the token that they sent,
+	 * so walk the batch list backwards from the current head back
+	 * to the batch (sequence number) they named.
+	 *
+	 * We use khash to de-dup the list of pathnames.
+	 *
+	 * NEEDSWORK: each batch contains a list of interned strings,
+	 * so we only need to do pointer comparisons here to build the
+	 * hash table.  Currently, we're still comparing the string
+	 * values.
+	 */
+	shown = kh_init_str();
+	for (batch = batch_head;
+	     batch && batch->batch_seq_nr > requested_oldest_seq_nr;
+	     batch = batch->next) {
+		size_t k;
+
+		for (k = 0; k < batch->nr; k++) {
+			const char *s = batch->interned_paths[k];
+			size_t s_len;
+
+			if (kh_get_str(shown, s) != kh_end(shown))
+				duplicates++;
+			else {
+				kh_put_str(shown, s, &hash_ret);
+
+				trace_printf_key(&trace_fsmonitor,
+						 "send[%"PRIuMAX"]: %s",
+						 count, s);
+
+				/* Each path gets written with a trailing NUL */
+				s_len = strlen(s) + 1;
+
+				if (payload.len + s_len >=
+				    LARGE_PACKET_DATA_MAX) {
+					reply(reply_data, payload.buf,
+					      payload.len);
+					total_response_len += payload.len;
+					strbuf_reset(&payload);
+				}
+
+				strbuf_add(&payload, s, s_len);
+				count++;
+			}
+		}
+	}
+
+	if (payload.len) {
+		reply(reply_data, payload.buf, payload.len);
+		total_response_len += payload.len;
+	}
+
+	kh_release_str(shown);
+
+	pthread_mutex_lock(&state->main_lock);
+
+	if (token_data->client_ref_count > 0)
+		token_data->client_ref_count--;
+
+	if (token_data->client_ref_count == 0) {
+		if (token_data != state->current_token_data) {
+			/*
+			 * The listener thread did a token-reset while we were
+			 * walking the batch list.  Therefore, this token is
+			 * stale and can be discarded completely.  If we are
+			 * the last reader thread using this token, we own
+			 * that work.
+			 */
+			fsmonitor_free_token_data(token_data);
+		}
+	}
+
+	pthread_mutex_unlock(&state->main_lock);
+
+	trace2_data_intmax("fsmonitor", the_repository, "response/length", total_response_len);
+	trace2_data_intmax("fsmonitor", the_repository, "response/count/files", count);
+	trace2_data_intmax("fsmonitor", the_repository, "response/count/duplicates", duplicates);
+
+	strbuf_release(&response_token);
+	strbuf_release(&requested_token_id);
+	strbuf_release(&payload);
+
+	return 0;
+}
+
 static ipc_server_application_cb handle_client;
 
 static int handle_client(void *data,
@@ -371,7 +677,7 @@ static int handle_client(void *data,
 			 ipc_server_reply_cb *reply,
 			 struct ipc_server_reply_data *reply_data)
 {
-	/* struct fsmonitor_daemon_state *state = data; */
+	struct fsmonitor_daemon_state *state = data;
 	int result;
 
 	/*
@@ -382,10 +688,12 @@ static int handle_client(void *data,
 	if (command_len != strlen(command))
 		BUG("FSMonitor assumes text messages");
 
+	trace_printf_key(&trace_fsmonitor, "requested token: %s", command);
+
 	trace2_region_enter("fsmonitor", "handle_client", the_repository);
 	trace2_data_string("fsmonitor", the_repository, "request", command);
 
-	result = 0; /* TODO Do something here. */
+	result = do_handle_client(state, command, reply, reply_data);
 
 	trace2_region_leave("fsmonitor", "handle_client", the_repository);
 
-- 
gitgitgadget

[PATCH v6 19/30] help: include fsmonitor--daemon feature flag in version info

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Add the "feature: fsmonitor--daemon" message to the output of
`git version --build-options`.

The builtin FSMonitor is only available on certain platforms and
even then only when certain Makefile flags are enabled, so print
a message in the verbose version output when it is available.

This can be used by test scripts for prereq testing.  Granted, tests
could just try `git fsmonitor--daemon status` and look for a 128 exit
code or grep for a "not supported" message on stderr, but these
methods are rather obscure.

The main advantage is that the feature message will automatically
appear in bug reports and other support requests.

This concept was also used during the development of Scalar for
similar reasons.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 help.c        | 4 ++++
 t/test-lib.sh | 6 ++++++
 2 files changed, 10 insertions(+)

diff --git a/help.c b/help.c
index 71444906ddf..9112a51e84b 100644
--- a/help.c
+++ b/help.c
@@ -12,6 +12,7 @@
 #include "refs.h"
 #include "parse-options.h"
 #include "prompt.h"
+#include "fsmonitor-ipc.h"
 
 struct category_description {
 	uint32_t category;
@@ -695,6 +696,9 @@ void get_version_info(struct strbuf *buf, int show_build_options)
 		strbuf_addf(buf, "sizeof-size_t: %d\n", (int)sizeof(size_t));
 		strbuf_addf(buf, "shell-path: %s\n", SHELL_PATH);
 		/* NEEDSWORK: also save and output GIT-BUILD_OPTIONS? */
+
+		if (fsmonitor_ipc__is_supported())
+			strbuf_addstr(buf, "feature: fsmonitor--daemon\n");
 	}
 }
 
diff --git a/t/test-lib.sh b/t/test-lib.sh
index e4716b0b867..46cd596e7f5 100644
--- a/t/test-lib.sh
+++ b/t/test-lib.sh
@@ -1799,3 +1799,9 @@ test_lazy_prereq SHA1 '
 # Tests that verify the scheduler integration must set this locally
 # to avoid errors.
 GIT_TEST_MAINT_SCHEDULER="none:exit 1"
+
+# Does this platform support `git fsmonitor--daemon`
+#
+test_lazy_prereq FSMONITOR_DAEMON '
+	git version --build-options | grep "feature:" | grep "fsmonitor--daemon"
+'
-- 
gitgitgadget

[PATCH v6 20/30] t/helper/fsmonitor-client: create IPC client to talk to FSMonitor Daemon

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Create an IPC client to send query and flush commands to the daemon.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 Makefile                         |   1 +
 t/helper/test-fsmonitor-client.c | 121 +++++++++++++++++++++++++++++++
 t/helper/test-tool.c             |   1 +
 t/helper/test-tool.h             |   1 +
 4 files changed, 124 insertions(+)
 create mode 100644 t/helper/test-fsmonitor-client.c

diff --git a/Makefile b/Makefile
index 26567d4f772..daa21bed6c3 100644
--- a/Makefile
+++ b/Makefile
@@ -716,6 +716,7 @@ TEST_BUILTINS_OBJS += test-dump-split-index.o
 TEST_BUILTINS_OBJS += test-dump-untracked-cache.o
 TEST_BUILTINS_OBJS += test-example-decorate.o
 TEST_BUILTINS_OBJS += test-fast-rebase.o
+TEST_BUILTINS_OBJS += test-fsmonitor-client.o
 TEST_BUILTINS_OBJS += test-genrandom.o
 TEST_BUILTINS_OBJS += test-genzeros.o
 TEST_BUILTINS_OBJS += test-getcwd.o
diff --git a/t/helper/test-fsmonitor-client.c b/t/helper/test-fsmonitor-client.c
new file mode 100644
index 00000000000..f7a5b3a32fa
--- /dev/null
+++ b/t/helper/test-fsmonitor-client.c
@@ -0,0 +1,121 @@
+/*
+ * test-fsmonitor-client.c: client code to send commands/requests to
+ * a `git fsmonitor--daemon` daemon.
+ */
+
+#include "test-tool.h"
+#include "cache.h"
+#include "parse-options.h"
+#include "fsmonitor-ipc.h"
+
+#ifndef HAVE_FSMONITOR_DAEMON_BACKEND
+int cmd__fsmonitor_client(int argc, const char **argv)
+{
+	die("fsmonitor--daemon not available on this platform");
+}
+#else
+
+/*
+ * Read the `.git/index` to get the last token written to the
+ * FSMonitor Index Extension.
+ */
+static const char *get_token_from_index(void)
+{
+	struct index_state *istate = the_repository->index;
+
+	if (do_read_index(istate, the_repository->index_file, 0) < 0)
+		die("unable to read index file");
+	if (!istate->fsmonitor_last_update)
+		die("index file does not have fsmonitor extension");
+
+	return istate->fsmonitor_last_update;
+}
+
+/*
+ * Send an IPC query to a `git-fsmonitor--daemon` daemon and
+ * ask for the changes since the given token or from the last
+ * token in the index extension.
+ *
+ * This will implicitly start a daemon process if necessary.  The
+ * daemon process will persist after we exit.
+ */
+static int do_send_query(const char *token)
+{
+	struct strbuf answer = STRBUF_INIT;
+	int ret;
+
+	if (!token || !*token)
+		token = get_token_from_index();
+
+	ret = fsmonitor_ipc__send_query(token, &answer);
+	if (ret < 0)
+		die(_("could not query fsmonitor--daemon"));
+
+	write_in_full(1, answer.buf, answer.len);
+	strbuf_release(&answer);
+
+	return 0;
+}
+
+/*
+ * Send a "flush" command to the `git-fsmonitor--daemon` (if running)
+ * and tell it to flush its cache.
+ *
+ * This feature is primarily used by the test suite to simulate a loss of
+ * sync with the filesystem where we miss kernel events.
+ */
+static int do_send_flush(void)
+{
+	struct strbuf answer = STRBUF_INIT;
+	int ret;
+
+	ret = fsmonitor_ipc__send_command("flush", &answer);
+	if (ret)
+		return ret;
+
+	write_in_full(1, answer.buf, answer.len);
+	strbuf_release(&answer);
+
+	return 0;
+}
+
+int cmd__fsmonitor_client(int argc, const char **argv)
+{
+	const char *subcmd;
+	const char *token = NULL;
+
+	const char * const fsmonitor_client_usage[] = {
+		N_("test-helper fsmonitor-client query [<token>]"),
+		N_("test-helper fsmonitor-client flush"),
+		NULL,
+	};
+
+	struct option options[] = {
+		OPT_STRING(0, "token", &token, N_("token"),
+			   N_("command token to send to the server")),
+		OPT_END()
+	};
+
+	if (argc < 2)
+		usage_with_options(fsmonitor_client_usage, options);
+
+	if (argc == 2 && !strcmp(argv[1], "-h"))
+		usage_with_options(fsmonitor_client_usage, options);
+
+	subcmd = argv[1];
+	argv--;
+	argc++;
+
+	argc = parse_options(argc, argv, NULL, options, fsmonitor_client_usage, 0);
+
+	setup_git_directory();
+
+	if (!strcmp(subcmd, "query"))
+		return !!do_send_query(token);
+
+	if (!strcmp(subcmd, "flush"))
+		return !!do_send_flush();
+
+	die("Unhandled subcommand: '%s'", subcmd);
+}
+#endif
diff --git a/t/helper/test-tool.c b/t/helper/test-tool.c
index e6ec69cf326..0424f7adf5d 100644
--- a/t/helper/test-tool.c
+++ b/t/helper/test-tool.c
@@ -32,6 +32,7 @@ static struct test_cmd cmds[] = {
 	{ "dump-untracked-cache", cmd__dump_untracked_cache },
 	{ "example-decorate", cmd__example_decorate },
 	{ "fast-rebase", cmd__fast_rebase },
+	{ "fsmonitor-client", cmd__fsmonitor_client },
 	{ "genrandom", cmd__genrandom },
 	{ "genzeros", cmd__genzeros },
 	{ "getcwd", cmd__getcwd },
diff --git a/t/helper/test-tool.h b/t/helper/test-tool.h
index 20756eefdda..c876e8246fb 100644
--- a/t/helper/test-tool.h
+++ b/t/helper/test-tool.h
@@ -23,6 +23,7 @@ int cmd__dump_untracked_cache(int argc, const char **argv);
 int cmd__dump_reftable(int argc, const char **argv);
 int cmd__example_decorate(int argc, const char **argv);
 int cmd__fast_rebase(int argc, const char **argv);
+int cmd__fsmonitor_client(int argc, const char **argv);
 int cmd__genrandom(int argc, const char **argv);
 int cmd__genzeros(int argc, const char **argv);
 int cmd__getcwd(int argc, const char **argv);
-- 
gitgitgadget

[PATCH v6 21/30] t7527: create test for fsmonitor--daemon

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 t/t7527-builtin-fsmonitor.sh | 511 +++++++++++++++++++++++++++++++++++
 1 file changed, 511 insertions(+)
 create mode 100755 t/t7527-builtin-fsmonitor.sh

diff --git a/t/t7527-builtin-fsmonitor.sh b/t/t7527-builtin-fsmonitor.sh
new file mode 100755
index 00000000000..5f7b8e54233
--- /dev/null
+++ b/t/t7527-builtin-fsmonitor.sh
@@ -0,0 +1,511 @@
+#!/bin/sh
+
+test_description='built-in file system watcher'
+
+. ./test-lib.sh
+
+if ! test_have_prereq FSMONITOR_DAEMON
+then
+	skip_all="fsmonitor--daemon is not supported on this platform"
+	test_done
+fi
+
+stop_daemon_delete_repo () {
+	r=$1
+	git -C $r fsmonitor--daemon stop >/dev/null 2>/dev/null
+	rm -rf $1
+	return 0
+}
+
+start_daemon () {
+	case "$#" in
+		1) r="-C $1";;
+		*) r="";
+	esac
+
+	git $r fsmonitor--daemon start || return $?
+	git $r fsmonitor--daemon status || return $?
+
+	return 0
+}
+
+# Is a Trace2 data event present with the given catetory and key?
+# We do not care what the value is.
+#
+have_t2_data_event () {
+	c=$1
+	k=$2
+
+	grep -e '"event":"data".*"category":"'"$c"'".*"key":"'"$k"'"'
+}
+
+test_expect_success 'explicit daemon start and stop' '
+	test_when_finished "stop_daemon_delete_repo test_explicit" &&
+
+	git init test_explicit &&
+	start_daemon test_explicit &&
+
+	git -C test_explicit fsmonitor--daemon stop &&
+	test_must_fail git -C test_explicit fsmonitor--daemon status
+'
+
+test_expect_success 'implicit daemon start' '
+	test_when_finished "stop_daemon_delete_repo test_implicit" &&
+
+	git init test_implicit &&
+	test_must_fail git -C test_implicit fsmonitor--daemon status &&
+
+	# query will implicitly start the daemon.
+	#
+	# for test-script simplicity, we send a V1 timestamp rather than
+	# a V2 token.  either way, the daemon response to any query contains
+	# a new V2 token.  (the daemon may complain that we sent a V1 request,
+	# but this test case is only concerned with whether the daemon was
+	# implicitly started.)
+
+	GIT_TRACE2_EVENT="$(pwd)/.git/trace" \
+		test-tool -C test_implicit fsmonitor-client query --token 0 >actual &&
+	nul_to_q <actual >actual.filtered &&
+	grep "builtin:" actual.filtered &&
+
+	# confirm that a daemon was started in the background.
+	#
+	# since the mechanism for starting the background daemon is platform
+	# dependent, just confirm that the foreground command received a
+	# response from the daemon.
+
+	have_t2_data_event fsm_client query/response-length <.git/trace &&
+
+	git -C test_implicit fsmonitor--daemon status &&
+	git -C test_implicit fsmonitor--daemon stop &&
+	test_must_fail git -C test_implicit fsmonitor--daemon status
+'
+
+test_expect_success 'implicit daemon stop (delete .git)' '
+	test_when_finished "stop_daemon_delete_repo test_implicit_1" &&
+
+	git init test_implicit_1 &&
+
+	start_daemon test_implicit_1 &&
+
+	# deleting the .git directory will implicitly stop the daemon.
+	rm -rf test_implicit_1/.git &&
+
+	# [1] Create an empty .git directory so that the following Git
+	#     command will stay relative to the `-C` directory.
+	#
+	#     Without this, the Git command will override the requested
+	#     -C argument and crawl out to the containing Git source tree.
+	#     This would make the test result dependent upon whether we
+	#     were using fsmonitor on our development worktree.
+	#
+	sleep 1 &&
+	mkdir test_implicit_1/.git &&
+
+	test_must_fail git -C test_implicit_1 fsmonitor--daemon status
+'
+
+test_expect_success 'implicit daemon stop (rename .git)' '
+	test_when_finished "stop_daemon_delete_repo test_implicit_2" &&
+
+	git init test_implicit_2 &&
+
+	start_daemon test_implicit_2 &&
+
+	# renaming the .git directory will implicitly stop the daemon.
+	mv test_implicit_2/.git test_implicit_2/.xxx &&
+
+	# See [1] above.
+	#
+	sleep 1 &&
+	mkdir test_implicit_2/.git &&
+
+	test_must_fail git -C test_implicit_2 fsmonitor--daemon status
+'
+
+test_expect_success 'cannot start multiple daemons' '
+	test_when_finished "stop_daemon_delete_repo test_multiple" &&
+
+	git init test_multiple &&
+
+	start_daemon test_multiple &&
+
+	test_must_fail git -C test_multiple fsmonitor--daemon start 2>actual &&
+	grep "fsmonitor--daemon is already running" actual &&
+
+	git -C test_multiple fsmonitor--daemon stop &&
+	test_must_fail git -C test_multiple fsmonitor--daemon status
+'
+
+# These tests use the main repo in the trash directory
+
+test_expect_success 'setup' '
+	>tracked &&
+	>modified &&
+	>delete &&
+	>rename &&
+	mkdir dir1 &&
+	>dir1/tracked &&
+	>dir1/modified &&
+	>dir1/delete &&
+	>dir1/rename &&
+	mkdir dir2 &&
+	>dir2/tracked &&
+	>dir2/modified &&
+	>dir2/delete &&
+	>dir2/rename &&
+	mkdir dirtorename &&
+	>dirtorename/a &&
+	>dirtorename/b &&
+
+	cat >.gitignore <<-\EOF &&
+	.gitignore
+	expect*
+	actual*
+	EOF
+
+	git -c core.fsmonitor=false add . &&
+	test_tick &&
+	git -c core.fsmonitor=false commit -m initial &&
+
+	git config core.fsmonitor true
+'
+
+# The test already explicitly stopped (or tried to stop) the daemon.
+# This is here in case something else fails first.
+#
+redundant_stop_daemon () {
+	git fsmonitor--daemon stop
+	return 0
+}
+
+test_expect_success 'update-index implicitly starts daemon' '
+	test_when_finished redundant_stop_daemon &&
+
+	test_must_fail git fsmonitor--daemon status &&
+
+	GIT_TRACE2_EVENT="$(pwd)/.git/trace_implicit_1" \
+		git update-index --fsmonitor &&
+
+	git fsmonitor--daemon status &&
+	test_might_fail git fsmonitor--daemon stop &&
+
+	# Confirm that the trace2 log contains a record of the
+	# daemon starting.
+	test_subcommand git fsmonitor--daemon start <.git/trace_implicit_1
+'
+
+test_expect_success 'status implicitly starts daemon' '
+	test_when_finished redundant_stop_daemon &&
+
+	test_must_fail git fsmonitor--daemon status &&
+
+	GIT_TRACE2_EVENT="$(pwd)/.git/trace_implicit_2" \
+		git status >actual &&
+
+	git fsmonitor--daemon status &&
+	test_might_fail git fsmonitor--daemon stop &&
+
+	# Confirm that the trace2 log contains a record of the
+	# daemon starting.
+	test_subcommand git fsmonitor--daemon start <.git/trace_implicit_2
+'
+
+edit_files () {
+	echo 1 >modified
+	echo 2 >dir1/modified
+	echo 3 >dir2/modified
+	>dir1/untracked
+}
+
+delete_files () {
+	rm -f delete
+	rm -f dir1/delete
+	rm -f dir2/delete
+}
+
+create_files () {
+	echo 1 >new
+	echo 2 >dir1/new
+	echo 3 >dir2/new
+}
+
+rename_files () {
+	mv rename renamed
+	mv dir1/rename dir1/renamed
+	mv dir2/rename dir2/renamed
+}
+
+file_to_directory () {
+	rm -f delete
+	mkdir delete
+	echo 1 >delete/new
+}
+
+directory_to_file () {
+	rm -rf dir1
+	echo 1 >dir1
+}
+
+verify_status () {
+	git status >actual &&
+	GIT_INDEX_FILE=.git/fresh-index git read-tree master &&
+	GIT_INDEX_FILE=.git/fresh-index git -c core.fsmonitor=false status >expect &&
+	test_cmp expect actual &&
+	echo HELLO AFTER &&
+	cat .git/trace &&
+	echo HELLO AFTER
+}
+
+# The next few test cases confirm that our fsmonitor daemon sees each type
+# of OS filesystem notification that we care about.  At this layer we just
+# ensure we are getting the OS notifications and do not try to confirm what
+# is reported by `git status`.
+#
+# We run a simple query after modifying the filesystem just to introduce
+# a bit of a delay so that the trace logging from the daemon has time to
+# get flushed to disk.
+#
+# We `reset` and `clean` at the bottom of each test (and before stopping the
+# daemon) because these commands might implicitly restart the daemon.
+
+clean_up_repo_and_stop_daemon () {
+	git reset --hard HEAD
+	git clean -fd
+	git fsmonitor--daemon stop
+	rm -f .git/trace
+}
+
+test_expect_success 'edit some files' '
+	test_when_finished clean_up_repo_and_stop_daemon &&
+
+	(
+		GIT_TRACE_FSMONITOR="$(pwd)/.git/trace" &&
+		export GIT_TRACE_FSMONITOR &&
+
+		start_daemon
+	) &&
+
+	edit_files &&
+
+	test-tool fsmonitor-client query --token 0 >/dev/null 2>&1 &&
+
+	grep "^event: dir1/modified$"  .git/trace &&
+	grep "^event: dir2/modified$"  .git/trace &&
+	grep "^event: modified$"       .git/trace &&
+	grep "^event: dir1/untracked$" .git/trace
+'
+
+test_expect_success 'create some files' '
+	test_when_finished clean_up_repo_and_stop_daemon &&
+
+	(
+		GIT_TRACE_FSMONITOR="$(pwd)/.git/trace" &&
+		export GIT_TRACE_FSMONITOR &&
+
+		start_daemon
+	) &&
+
+	create_files &&
+
+	test-tool fsmonitor-client query --token 0 >/dev/null 2>&1 &&
+
+	grep "^event: dir1/new$" .git/trace &&
+	grep "^event: dir2/new$" .git/trace &&
+	grep "^event: new$"      .git/trace
+'
+
+test_expect_success 'delete some files' '
+	test_when_finished clean_up_repo_and_stop_daemon &&
+
+	(
+		GIT_TRACE_FSMONITOR="$(pwd)/.git/trace" &&
+		export GIT_TRACE_FSMONITOR &&
+
+		start_daemon
+	) &&
+
+	delete_files &&
+
+	test-tool fsmonitor-client query --token 0 >/dev/null 2>&1 &&
+
+	grep "^event: dir1/delete$" .git/trace &&
+	grep "^event: dir2/delete$" .git/trace &&
+	grep "^event: delete$"      .git/trace
+'
+
+test_expect_success 'rename some files' '
+	test_when_finished clean_up_repo_and_stop_daemon &&
+
+	(
+		GIT_TRACE_FSMONITOR="$(pwd)/.git/trace" &&
+		export GIT_TRACE_FSMONITOR &&
+
+		start_daemon
+	) &&
+
+	rename_files &&
+
+	test-tool fsmonitor-client query --token 0 >/dev/null 2>&1 &&
+
+	grep "^event: dir1/rename$"  .git/trace &&
+	grep "^event: dir2/rename$"  .git/trace &&
+	grep "^event: rename$"       .git/trace &&
+	grep "^event: dir1/renamed$" .git/trace &&
+	grep "^event: dir2/renamed$" .git/trace &&
+	grep "^event: renamed$"      .git/trace
+'
+
+test_expect_success 'rename directory' '
+	test_when_finished clean_up_repo_and_stop_daemon &&
+
+	(
+		GIT_TRACE_FSMONITOR="$(pwd)/.git/trace" &&
+		export GIT_TRACE_FSMONITOR &&
+
+		start_daemon
+	) &&
+
+	mv dirtorename dirrenamed &&
+
+	test-tool fsmonitor-client query --token 0 >/dev/null 2>&1 &&
+
+	grep "^event: dirtorename/*$" .git/trace &&
+	grep "^event: dirrenamed/*$"  .git/trace
+'
+
+test_expect_success 'file changes to directory' '
+	test_when_finished clean_up_repo_and_stop_daemon &&
+
+	(
+		GIT_TRACE_FSMONITOR="$(pwd)/.git/trace" &&
+		export GIT_TRACE_FSMONITOR &&
+
+		start_daemon
+	) &&
+
+	file_to_directory &&
+
+	test-tool fsmonitor-client query --token 0 >/dev/null 2>&1 &&
+
+	grep "^event: delete$"     .git/trace &&
+	grep "^event: delete/new$" .git/trace
+'
+
+test_expect_success 'directory changes to a file' '
+	test_when_finished clean_up_repo_and_stop_daemon &&
+
+	(
+		GIT_TRACE_FSMONITOR="$(pwd)/.git/trace" &&
+		export GIT_TRACE_FSMONITOR &&
+
+		start_daemon
+	) &&
+
+	directory_to_file &&
+
+	test-tool fsmonitor-client query --token 0 >/dev/null 2>&1 &&
+
+	grep "^event: dir1$" .git/trace
+'
+
+# The next few test cases exercise the token-resync code.  When filesystem
+# drops events (because of filesystem velocity or because the daemon isn't
+# polling fast enough), we need to discard the cached data (relative to the
+# current token) and start collecting events under a new token.
+#
+# the 'test-tool fsmonitor-client flush' command can be used to send a
+# "flush" message to a running daemon and ask it to do a flush/resync.
+
+test_expect_success 'flush cached data' '
+	test_when_finished "stop_daemon_delete_repo test_flush" &&
+
+	git init test_flush &&
+
+	(
+		GIT_TEST_FSMONITOR_TOKEN=true &&
+		export GIT_TEST_FSMONITOR_TOKEN &&
+
+		GIT_TRACE_FSMONITOR="$(pwd)/.git/trace_daemon" &&
+		export GIT_TRACE_FSMONITOR &&
+
+		start_daemon test_flush
+	) &&
+
+	# The daemon should have an initial token with no events in _0 and
+	# then a few (probably platform-specific number of) events in _1.
+	# These should both have the same <token_id>.
+
+	test-tool -C test_flush fsmonitor-client query --token "builtin:test_00000001:0" >actual_0 &&
+	nul_to_q <actual_0 >actual_q0 &&
+
+	touch test_flush/file_1 &&
+	touch test_flush/file_2 &&
+
+	test-tool -C test_flush fsmonitor-client query --token "builtin:test_00000001:0" >actual_1 &&
+	nul_to_q <actual_1 >actual_q1 &&
+
+	grep "file_1" actual_q1 &&
+
+	# Force a flush.  This will change the <token_id>, reset the <seq_nr>, and
+	# flush the file data.  Then create some events and ensure that the file
+	# again appears in the cache.  It should have the new <token_id>.
+
+	test-tool -C test_flush fsmonitor-client flush >flush_0 &&
+	nul_to_q <flush_0 >flush_q0 &&
+	grep "^builtin:test_00000002:0Q/Q$" flush_q0 &&
+
+	test-tool -C test_flush fsmonitor-client query --token "builtin:test_00000002:0" >actual_2 &&
+	nul_to_q <actual_2 >actual_q2 &&
+
+	grep "^builtin:test_00000002:0Q$" actual_q2 &&
+
+	touch test_flush/file_3 &&
+
+	test-tool -C test_flush fsmonitor-client query --token "builtin:test_00000002:0" >actual_3 &&
+	nul_to_q <actual_3 >actual_q3 &&
+
+	grep "file_3" actual_q3
+'
+
+# The next few test cases create repos where the .git directory is NOT
+# inside the one of the working directory.  That is, where .git is a file
+# that points to a directory elsewhere.  This happens for submodules and
+# non-primary worktrees.
+
+test_expect_success 'setup worktree base' '
+	git init wt-base &&
+	echo 1 >wt-base/file1 &&
+	git -C wt-base add file1 &&
+	git -C wt-base commit -m "c1"
+'
+
+test_expect_success 'worktree with .git file' '
+	git -C wt-base worktree add ../wt-secondary &&
+
+	(
+		GIT_TRACE2_PERF="$(pwd)/trace2_wt_secondary" &&
+		export GIT_TRACE2_PERF &&
+
+		GIT_TRACE_FSMONITOR="$(pwd)/trace_wt_secondary" &&
+		export GIT_TRACE_FSMONITOR &&
+
+		start_daemon wt-secondary
+	) &&
+
+	git -C wt-secondary fsmonitor--daemon stop &&
+	test_must_fail git -C wt-secondary fsmonitor--daemon status
+'
+
+# NEEDSWORK: Repeat one of the "edit" tests on wt-secondary and
+# confirm that we get the same events and behavior -- that is, that
+# fsmonitor--daemon correctly watches BOTH the working directory and
+# the external GITDIR directory and behaves the same as when ".git"
+# is a directory inside the working directory.
+
+test_expect_success 'cleanup worktrees' '
+	stop_daemon_delete_repo wt-secondary &&
+	stop_daemon_delete_repo wt-base
+'
+
+test_done
-- 
gitgitgadget

[PATCH v6 22/30] t/perf: avoid copying builtin fsmonitor files into test repo

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Do not copy any of the various fsmonitor--daemon files from the .git
directory of the (GIT_PREF_REPO or GIT_PERF_LARGE_REPO) source repo
into the test's trash directory.

When perf tests start, they copy the contents of the source repo into
the test's trash directory.  If fsmonitor is running in the source repo,
there may be control files, such as the IPC socket and/or fsmonitor
cookie files.  These should not be copied into the test repo.

Unix domain sockets cannot be copied in the manner used by the test
setup, so if present, the test setup fails.

Cookie files are harmless, but we should avoid them.

The builtin fsmonitor keeps all such control files/sockets in
.git/fsmonitor--daemon*, so it is simple to exclude them.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 t/perf/perf-lib.sh | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/t/perf/perf-lib.sh b/t/perf/perf-lib.sh
index 407252bac70..932105cd12c 100644
--- a/t/perf/perf-lib.sh
+++ b/t/perf/perf-lib.sh
@@ -78,7 +78,7 @@ test_perf_copy_repo_contents () {
 	for stuff in "$1"/*
 	do
 		case "$stuff" in
-		*/objects|*/hooks|*/config|*/commondir|*/gitdir|*/worktrees)
+		*/objects|*/hooks|*/config|*/commondir|*/gitdir|*/worktrees|*/fsmonitor--daemon*)
 			;;
 		*)
 			cp -R "$stuff" "$repo/.git/" || exit 1
-- 
gitgitgadget

[PATCH v6 23/30] t/helper/test-chmtime: skip directories on Windows

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach `test-tool.exe chmtime` to ignore errors when setting the mtime
on a directory on Windows.

NEEDSWORK: The Windows version of `utime()` (aka `mingw_utime()`) does
not properly handle directories because it uses `_wopen()`.  It should
be converted to using `CreateFileW()` and backup semantics at a minimum.
Since I'm already in the middle of a large patch series, I did not want
to destabilize other callers of `utime()` right now.  The problem has
only been observed in the t/perf/p7519 test when the test repo contains
an empty directory on disk.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 t/helper/test-chmtime.c | 15 +++++++++++++++
 1 file changed, 15 insertions(+)

diff --git a/t/helper/test-chmtime.c b/t/helper/test-chmtime.c
index 524b55ca496..dc28890a183 100644
--- a/t/helper/test-chmtime.c
+++ b/t/helper/test-chmtime.c
@@ -134,6 +134,21 @@ int cmd__chmtime(int argc, const char **argv)
 		}
 
 		if (utb.modtime != sb.st_mtime && utime(argv[i], &utb) < 0) {
+#ifdef GIT_WINDOWS_NATIVE
+			if (S_ISDIR(sb.st_mode)) {
+				/*
+				 * NEEDSWORK: The Windows version of `utime()`
+				 * (aka `mingw_utime()`) does not correctly
+				 * handle directory arguments, since it uses
+				 * `_wopen()`.  Ignore it for now since this
+				 * is just a test.
+				 */
+				fprintf(stderr,
+					("Failed to modify time on directory %s. "
+					 "Skipping\n"), argv[i]);
+				continue;
+			}
+#endif
 			fprintf(stderr, "Failed to modify time on %s: %s\n",
 			        argv[i], strerror(errno));
 			return 1;
-- 
gitgitgadget

[PATCH v6 24/30] t/perf/p7519: speed up test on Windows

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Change p7519 to use `test_seq` and `xargs` rather than a `for` loop
to touch thousands of files.  This takes minutes off of test runs
on Windows because of process creation overhead.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 t/perf/p7519-fsmonitor.sh | 32 ++++++++++++++++++++------------
 1 file changed, 20 insertions(+), 12 deletions(-)

diff --git a/t/perf/p7519-fsmonitor.sh b/t/perf/p7519-fsmonitor.sh
index c8be58f3c76..aed7b1146b0 100755
--- a/t/perf/p7519-fsmonitor.sh
+++ b/t/perf/p7519-fsmonitor.sh
@@ -72,7 +72,7 @@ then
 	fi
 fi
 
-trace_start() {
+trace_start () {
 	if test -n "$GIT_PERF_7519_TRACE"
 	then
 		name="$1"
@@ -91,13 +91,20 @@ trace_start() {
 	fi
 }
 
-trace_stop() {
+trace_stop () {
 	if test -n "$GIT_PERF_7519_TRACE"
 	then
 		unset GIT_TRACE2_PERF
 	fi
 }
 
+touch_files () {
+	n=$1
+	d="$n"_files
+
+	(cd $d ; test_seq 1 $n | xargs touch )
+}
+
 test_expect_success "one time repo setup" '
 	# set untrackedCache depending on the environment
 	if test -n "$GIT_PERF_7519_UNTRACKED_CACHE"
@@ -119,10 +126,11 @@ test_expect_success "one time repo setup" '
 	fi &&
 
 	mkdir 1_file 10_files 100_files 1000_files 10000_files &&
-	for i in $(test_seq 1 10); do touch 10_files/$i || return 1; done &&
-	for i in $(test_seq 1 100); do touch 100_files/$i || return 1; done &&
-	for i in $(test_seq 1 1000); do touch 1000_files/$i || return 1; done &&
-	for i in $(test_seq 1 10000); do touch 10000_files/$i || return 1; done &&
+	: 1_file directory should be left empty &&
+	touch_files 10 &&
+	touch_files 100 &&
+	touch_files 1000 &&
+	touch_files 10000 &&
 	git add 1_file 10_files 100_files 1000_files 10000_files &&
 	git commit -qm "Add files" &&
 
@@ -133,7 +141,7 @@ test_expect_success "one time repo setup" '
 	fi
 '
 
-setup_for_fsmonitor() {
+setup_for_fsmonitor () {
 	# set INTEGRATION_SCRIPT depending on the environment
 	if test -n "$INTEGRATION_PATH"
 	then
@@ -173,7 +181,7 @@ test_perf_w_drop_caches () {
 	test_perf "$@"
 }
 
-test_fsmonitor_suite() {
+test_fsmonitor_suite () {
 	if test -n "$INTEGRATION_SCRIPT"; then
 		DESC="fsmonitor=$(basename $INTEGRATION_SCRIPT)"
 	else
@@ -199,15 +207,15 @@ test_fsmonitor_suite() {
 
 	# Update the mtimes on upto 100k files to make status think
 	# that they are dirty.  For simplicity, omit any files with
-	# LFs (i.e. anything that ls-files thinks it needs to dquote).
-	# Then fully backslash-quote the paths to capture any
-	# whitespace so that they pass thru xargs properly.
+	# LFs (i.e. anything that ls-files thinks it needs to dquote)
+	# and any files with whitespace so that they pass thru xargs
+	# properly.
 	#
 	test_perf_w_drop_caches "status (dirty) ($DESC)" '
 		git ls-files | \
 			head -100000 | \
 			grep -v \" | \
-			sed '\''s/\(.\)/\\\1/g'\'' | \
+			egrep -v " ." | \
 			xargs test-tool chmtime -300 &&
 		git status
 	'
-- 
gitgitgadget

[PATCH v6 25/30] t/perf/p7519: add fsmonitor--daemon test cases

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Repeat all of the fsmonitor perf tests using `git fsmonitor--daemon` and
the "Simple IPC" interface.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 t/perf/p7519-fsmonitor.sh | 29 ++++++++++++++++++++++++++++-
 1 file changed, 28 insertions(+), 1 deletion(-)

diff --git a/t/perf/p7519-fsmonitor.sh b/t/perf/p7519-fsmonitor.sh
index aed7b1146b0..a1c552129cc 100755
--- a/t/perf/p7519-fsmonitor.sh
+++ b/t/perf/p7519-fsmonitor.sh
@@ -182,7 +182,10 @@ test_perf_w_drop_caches () {
 }
 
 test_fsmonitor_suite () {
-	if test -n "$INTEGRATION_SCRIPT"; then
+	if test -n "$USE_FSMONITOR_DAEMON"
+	then
+		DESC="builtin fsmonitor--daemon"
+	elif test -n "$INTEGRATION_SCRIPT"; then
 		DESC="fsmonitor=$(basename $INTEGRATION_SCRIPT)"
 	else
 		DESC="fsmonitor=disabled"
@@ -293,4 +296,28 @@ test_expect_success "setup without fsmonitor" '
 test_fsmonitor_suite
 trace_stop
 
+#
+# Run a full set of perf tests using the built-in fsmonitor--daemon.
+# It does not use the Hook API, so it has a different setup.
+# Explicitly start the daemon here and before we start client commands
+# so that we can later add custom tracing.
+#
+if test_have_prereq FSMONITOR_DAEMON
+then
+	USE_FSMONITOR_DAEMON=t
+
+	trace_start fsmonitor--daemon--server
+	git fsmonitor--daemon start
+
+	trace_start fsmonitor--daemon--client
+
+	git config core.fsmonitor true
+	git update-index --fsmonitor
+
+	test_fsmonitor_suite
+
+	git fsmonitor--daemon stop
+	trace_stop
+fi
+
 test_done
-- 
gitgitgadget

[PATCH v6 26/30] fsmonitor--daemon: periodically truncate list of modified files

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach fsmonitor--daemon to periodically truncate the list of
modified files to save some memory.

Clients will ask for the set of changes relative to a token that they
found in the FSMN index extension in the index.  (This token is like a
point in time, but different).  Clients will then update the index to
contain the response token (so that subsequent commands will be
relative to this new token).

Therefore, the daemon can gradually truncate the in-memory list of
changed paths as they become obsolete (older than the previous token).
Since we may have multiple clients making concurrent requests with a
skew of tokens and clients may be racing to the talk to the daemon,
we lazily truncate the list.

We introduce a 5 minute delay and truncate batches 5 minutes after
they are considered obsolete.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 builtin/fsmonitor--daemon.c | 88 +++++++++++++++++++++++++++++++++++++
 1 file changed, 88 insertions(+)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index 65c1ef7d4bf..6ca4f5959e6 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -312,6 +312,75 @@ static void fsmonitor_batch__combine(struct fsmonitor_batch *batch_dest,
 			batch_src->interned_paths[k];
 }
 
+/*
+ * To keep the batch list from growing unbounded in response to filesystem
+ * activity, we try to truncate old batches from the end of the list as
+ * they become irrelevant.
+ *
+ * We assume that the .git/index will be updated with the most recent token
+ * any time the index is updated.  And future commands will only ask for
+ * recent changes *since* that new token.  So as tokens advance into the
+ * future, older batch items will never be requested/needed.  So we can
+ * truncate them without loss of functionality.
+ *
+ * However, multiple commands may be talking to the daemon concurrently
+ * or perform a slow command, so a little "token skew" is possible.
+ * Therefore, we want this to be a little bit lazy and have a generous
+ * delay.
+ *
+ * The current reader thread walked backwards in time from `token->batch_head`
+ * back to `batch_marker` somewhere in the middle of the batch list.
+ *
+ * Let's walk backwards in time from that marker an arbitrary delay
+ * and truncate the list there.  Note that these timestamps are completely
+ * artificial (based on when we pinned the batch item) and not on any
+ * filesystem activity.
+ *
+ * Return the obsolete portion of the list after we have removed it from
+ * the official list so that the caller can free it after leaving the lock.
+ */
+#define MY_TIME_DELAY_SECONDS (5 * 60) /* seconds */
+
+static struct fsmonitor_batch *with_lock__truncate_old_batches(
+	struct fsmonitor_daemon_state *state,
+	const struct fsmonitor_batch *batch_marker)
+{
+	/* assert current thread holding state->main_lock */
+
+	const struct fsmonitor_batch *batch;
+	struct fsmonitor_batch *remainder;
+
+	if (!batch_marker)
+		return NULL;
+
+	trace_printf_key(&trace_fsmonitor, "Truncate: mark (%"PRIu64",%"PRIu64")",
+			 batch_marker->batch_seq_nr,
+			 (uint64_t)batch_marker->pinned_time);
+
+	for (batch = batch_marker; batch; batch = batch->next) {
+		time_t t;
+
+		if (!batch->pinned_time) /* an overflow batch */
+			continue;
+
+		t = batch->pinned_time + MY_TIME_DELAY_SECONDS;
+		if (t > batch_marker->pinned_time) /* too close to marker */
+			continue;
+
+		goto truncate_past_here;
+	}
+
+	return NULL;
+
+truncate_past_here:
+	state->current_token_data->batch_tail = (struct fsmonitor_batch *)batch;
+
+	remainder = ((struct fsmonitor_batch *)batch)->next;
+	((struct fsmonitor_batch *)batch)->next = NULL;
+
+	return remainder;
+}
+
 static void fsmonitor_free_token_data(struct fsmonitor_token_data *token)
 {
 	if (!token)
@@ -425,6 +494,7 @@ static int do_handle_client(struct fsmonitor_daemon_state *state,
 	const char *p;
 	const struct fsmonitor_batch *batch_head;
 	const struct fsmonitor_batch *batch;
+	struct fsmonitor_batch *remainder = NULL;
 	intmax_t count = 0, duplicates = 0;
 	kh_str_t *shown;
 	int hash_ret;
@@ -654,11 +724,29 @@ static int do_handle_client(struct fsmonitor_daemon_state *state,
 			 * that work.
 			 */
 			fsmonitor_free_token_data(token_data);
+		} else if (batch) {
+			/*
+			 * We are holding the lock and are the only
+			 * reader of the ref-counted portion of the
+			 * list, so we get the honor of seeing if the
+			 * list can be truncated to save memory.
+			 *
+			 * The main loop did not walk to the end of the
+			 * list, so this batch is the first item in the
+			 * batch-list that is older than the requested
+			 * end-point sequence number.  See if the tail
+			 * end of the list is obsolete.
+			 */
+			remainder = with_lock__truncate_old_batches(state,
+								    batch);
 		}
 	}
 
 	pthread_mutex_unlock(&state->main_lock);
 
+	if (remainder)
+		fsmonitor_batch__free_list(remainder);
+
 	trace2_data_intmax("fsmonitor", the_repository, "response/length", total_response_len);
 	trace2_data_intmax("fsmonitor", the_repository, "response/count/files", count);
 	trace2_data_intmax("fsmonitor", the_repository, "response/count/duplicates", duplicates);
-- 
gitgitgadget

[PATCH v6 27/30] fsmonitor--daemon: use a cookie file to sync with file system

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach fsmonitor--daemon client threads to create a cookie file
inside the .git directory and then wait until FS events for the
cookie are observed by the FS listener thread.

This helps address the racy nature of file system events by
blocking the client response until the kernel has drained any
event backlog.

This is especially important on MacOS where kernel events are
only issued with a limited frequency.  See the `latency` argument
of `FSeventStreamCreate()`.  The kernel only signals every `latency`
seconds, but does not guarantee that the kernel queue is completely
drained, so we may have to wait more than one interval.  If we
increase the latency, the system is more likely to drop events.
We avoid these issues by having each client thread create a unique
cookie file and then wait until it is seen in the event stream.

Co-authored-by: Kevin Willford <Kevin.Willford@microsoft.com>
Co-authored-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 builtin/fsmonitor--daemon.c | 230 +++++++++++++++++++++++++++++++++++-
 fsmonitor--daemon.h         |   5 +
 2 files changed, 234 insertions(+), 1 deletion(-)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index 6ca4f5959e6..97ca2a356e5 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -107,6 +107,155 @@ static int do_as_client__status(void)
 	}
 }
 
+enum fsmonitor_cookie_item_result {
+	FCIR_ERROR = -1, /* could not create cookie file ? */
+	FCIR_INIT = 0,
+	FCIR_SEEN,
+	FCIR_ABORT,
+};
+
+struct fsmonitor_cookie_item {
+	struct hashmap_entry entry;
+	const char *name;
+	enum fsmonitor_cookie_item_result result;
+};
+
+static int cookies_cmp(const void *data, const struct hashmap_entry *he1,
+		     const struct hashmap_entry *he2, const void *keydata)
+{
+	const struct fsmonitor_cookie_item *a =
+		container_of(he1, const struct fsmonitor_cookie_item, entry);
+	const struct fsmonitor_cookie_item *b =
+		container_of(he2, const struct fsmonitor_cookie_item, entry);
+
+	return strcmp(a->name, keydata ? keydata : b->name);
+}
+
+static enum fsmonitor_cookie_item_result with_lock__wait_for_cookie(
+	struct fsmonitor_daemon_state *state)
+{
+	/* assert current thread holding state->main_lock */
+
+	int fd;
+	struct fsmonitor_cookie_item *cookie;
+	struct strbuf cookie_pathname = STRBUF_INIT;
+	struct strbuf cookie_filename = STRBUF_INIT;
+	enum fsmonitor_cookie_item_result result;
+	int my_cookie_seq;
+
+	CALLOC_ARRAY(cookie, 1);
+
+	my_cookie_seq = state->cookie_seq++;
+
+	strbuf_addf(&cookie_filename, "%i-%i", getpid(), my_cookie_seq);
+
+	strbuf_addbuf(&cookie_pathname, &state->path_cookie_prefix);
+	strbuf_addbuf(&cookie_pathname, &cookie_filename);
+
+	cookie->name = strbuf_detach(&cookie_filename, NULL);
+	cookie->result = FCIR_INIT;
+	hashmap_entry_init(&cookie->entry, strhash(cookie->name));
+
+	hashmap_add(&state->cookies, &cookie->entry);
+
+	trace_printf_key(&trace_fsmonitor, "cookie-wait: '%s' '%s'",
+			 cookie->name, cookie_pathname.buf);
+
+	/*
+	 * Create the cookie file on disk and then wait for a notification
+	 * that the listener thread has seen it.
+	 */
+	fd = open(cookie_pathname.buf, O_WRONLY | O_CREAT | O_EXCL, 0600);
+	if (fd >= 0) {
+		close(fd);
+		unlink(cookie_pathname.buf);
+
+		/*
+		 * Technically, this is an infinite wait (well, unless another
+		 * thread sends us an abort).  I'd like to change this to
+		 * use `pthread_cond_timedwait()` and return an error/timeout
+		 * and let the caller do the trivial response thing, but we
+		 * don't have that routine in our thread-utils.
+		 *
+		 * After extensive beta testing I'm not really worried about
+		 * this.  Also note that the above open() and unlink() calls
+		 * will cause at least two FS events on that path, so the odds
+		 * of getting stuck are pretty slim.
+		 */
+		while (cookie->result == FCIR_INIT)
+			pthread_cond_wait(&state->cookies_cond,
+					  &state->main_lock);
+	} else {
+		error_errno(_("could not create fsmonitor cookie '%s'"),
+			    cookie->name);
+
+		cookie->result = FCIR_ERROR;
+	}
+
+	hashmap_remove(&state->cookies, &cookie->entry, NULL);
+
+	result = cookie->result;
+
+	free((char*)cookie->name);
+	free(cookie);
+	strbuf_release(&cookie_pathname);
+
+	return result;
+}
+
+/*
+ * Mark these cookies as _SEEN and wake up the corresponding client threads.
+ */
+static void with_lock__mark_cookies_seen(struct fsmonitor_daemon_state *state,
+					 const struct string_list *cookie_names)
+{
+	/* assert current thread holding state->main_lock */
+
+	int k;
+	int nr_seen = 0;
+
+	for (k = 0; k < cookie_names->nr; k++) {
+		struct fsmonitor_cookie_item key;
+		struct fsmonitor_cookie_item *cookie;
+
+		key.name = cookie_names->items[k].string;
+		hashmap_entry_init(&key.entry, strhash(key.name));
+
+		cookie = hashmap_get_entry(&state->cookies, &key, entry, NULL);
+		if (cookie) {
+			trace_printf_key(&trace_fsmonitor, "cookie-seen: '%s'",
+					 cookie->name);
+			cookie->result = FCIR_SEEN;
+			nr_seen++;
+		}
+	}
+
+	if (nr_seen)
+		pthread_cond_broadcast(&state->cookies_cond);
+}
+
+/*
+ * Set _ABORT on all pending cookies and wake up all client threads.
+ */
+static void with_lock__abort_all_cookies(struct fsmonitor_daemon_state *state)
+{
+	/* assert current thread holding state->main_lock */
+
+	struct hashmap_iter iter;
+	struct fsmonitor_cookie_item *cookie;
+	int nr_aborted = 0;
+
+	hashmap_for_each_entry(&state->cookies, &iter, cookie, entry) {
+		trace_printf_key(&trace_fsmonitor, "cookie-abort: '%s'",
+				 cookie->name);
+		cookie->result = FCIR_ABORT;
+		nr_aborted++;
+	}
+
+	if (nr_aborted)
+		pthread_cond_broadcast(&state->cookies_cond);
+}
+
 /*
  * Requests to and from a FSMonitor Protocol V2 provider use an opaque
  * "token" as a virtual timestamp.  Clients can request a summary of all
@@ -404,6 +553,9 @@ static void fsmonitor_free_token_data(struct fsmonitor_token_data *token)
  *     We should create a new token and start fresh (as if we just
  *     booted up).
  *
+ * [2] Some of those lost events may have been for cookie files.  We
+ *     should assume the worst and abort them rather letting them starve.
+ *
  * If there are no concurrent threads reading the current token data
  * series, we can free it now.  Otherwise, let the last reader free
  * it.
@@ -425,6 +577,8 @@ static void with_lock__do_force_resync(struct fsmonitor_daemon_state *state)
 	state->current_token_data = new_one;
 
 	fsmonitor_free_token_data(free_me);
+
+	with_lock__abort_all_cookies(state);
 }
 
 void fsmonitor_force_resync(struct fsmonitor_daemon_state *state)
@@ -500,6 +654,8 @@ static int do_handle_client(struct fsmonitor_daemon_state *state,
 	int hash_ret;
 	int do_trivial = 0;
 	int do_flush = 0;
+	int do_cookie = 0;
+	enum fsmonitor_cookie_item_result cookie_result;
 
 	/*
 	 * We expect `command` to be of the form:
@@ -560,6 +716,7 @@ static int do_handle_client(struct fsmonitor_daemon_state *state,
 			 * We have a V2 valid token:
 			 *     "builtin:<token_id>:<seq_nr>"
 			 */
+			do_cookie = 1;
 		}
 	}
 
@@ -568,6 +725,30 @@ static int do_handle_client(struct fsmonitor_daemon_state *state,
 	if (!state->current_token_data)
 		BUG("fsmonitor state does not have a current token");
 
+	/*
+	 * Write a cookie file inside the directory being watched in
+	 * an effort to flush out existing filesystem events that we
+	 * actually care about.  Suspend this client thread until we
+	 * see the filesystem events for this cookie file.
+	 *
+	 * Creating the cookie lets us guarantee that our FS listener
+	 * thread has drained the kernel queue and we are caught up
+	 * with the kernel.
+	 *
+	 * If we cannot create the cookie (or otherwise guarantee that
+	 * we are caught up), we send a trivial response.  We have to
+	 * assume that there might be some very, very recent activity
+	 * on the FS still in flight.
+	 */
+	if (do_cookie) {
+		cookie_result = with_lock__wait_for_cookie(state);
+		if (cookie_result != FCIR_SEEN) {
+			error(_("fsmonitor: cookie_result '%d' != SEEN"),
+			      cookie_result);
+			do_trivial = 1;
+		}
+	}
+
 	if (do_flush)
 		with_lock__do_force_resync(state);
 
@@ -788,7 +969,9 @@ static int handle_client(void *data,
 	return result;
 }
 
-#define FSMONITOR_COOKIE_PREFIX ".fsmonitor-daemon-"
+#define FSMONITOR_DIR           "fsmonitor--daemon"
+#define FSMONITOR_COOKIE_DIR    "cookies"
+#define FSMONITOR_COOKIE_PREFIX (FSMONITOR_DIR "/" FSMONITOR_COOKIE_DIR "/")
 
 enum fsmonitor_path_type fsmonitor_classify_path_workdir_relative(
 	const char *rel)
@@ -941,6 +1124,9 @@ void fsmonitor_publish(struct fsmonitor_daemon_state *state,
 		}
 	}
 
+	if (cookie_names->nr)
+		with_lock__mark_cookies_seen(state, cookie_names);
+
 	pthread_mutex_unlock(&state->main_lock);
 }
 
@@ -1032,7 +1218,9 @@ static int fsmonitor_run_daemon(void)
 
 	memset(&state, 0, sizeof(state));
 
+	hashmap_init(&state.cookies, cookies_cmp, NULL, 0);
 	pthread_mutex_init(&state.main_lock, NULL);
+	pthread_cond_init(&state.cookies_cond, NULL);
 	state.error_code = 0;
 	state.current_token_data = fsmonitor_new_token_data();
 
@@ -1057,6 +1245,44 @@ static int fsmonitor_run_daemon(void)
 		state.nr_paths_watching = 2;
 	}
 
+	/*
+	 * We will write filesystem syncing cookie files into
+	 * <gitdir>/<fsmonitor-dir>/<cookie-dir>/<pid>-<seq>.
+	 *
+	 * The extra layers of subdirectories here keep us from
+	 * changing the mtime on ".git/" or ".git/foo/" when we create
+	 * or delete cookie files.
+	 *
+	 * There have been problems with some IDEs that do a
+	 * non-recursive watch of the ".git/" directory and run a
+	 * series of commands any time something happens.
+	 *
+	 * For example, if we place our cookie files directly in
+	 * ".git/" or ".git/foo/" then a `git status` (or similar
+	 * command) from the IDE will cause a cookie file to be
+	 * created in one of those dirs.  This causes the mtime of
+	 * those dirs to change.  This triggers the IDE's watch
+	 * notification.  This triggers the IDE to run those commands
+	 * again.  And the process repeats and the machine never goes
+	 * idle.
+	 *
+	 * Adding the extra layers of subdirectories prevents the
+	 * mtime of ".git/" and ".git/foo" from changing when a
+	 * cookie file is created.
+	 */
+	strbuf_init(&state.path_cookie_prefix, 0);
+	strbuf_addbuf(&state.path_cookie_prefix, &state.path_gitdir_watch);
+
+	strbuf_addch(&state.path_cookie_prefix, '/');
+	strbuf_addstr(&state.path_cookie_prefix, FSMONITOR_DIR);
+	mkdir(state.path_cookie_prefix.buf, 0777);
+
+	strbuf_addch(&state.path_cookie_prefix, '/');
+	strbuf_addstr(&state.path_cookie_prefix, FSMONITOR_COOKIE_DIR);
+	mkdir(state.path_cookie_prefix.buf, 0777);
+
+	strbuf_addch(&state.path_cookie_prefix, '/');
+
 	/*
 	 * Confirm that we can create platform-specific resources for the
 	 * filesystem listener before we bother starting all the threads.
@@ -1069,6 +1295,7 @@ static int fsmonitor_run_daemon(void)
 	err = fsmonitor_run_daemon_1(&state);
 
 done:
+	pthread_cond_destroy(&state.cookies_cond);
 	pthread_mutex_destroy(&state.main_lock);
 	fsm_listen__dtor(&state);
 
@@ -1076,6 +1303,7 @@ done:
 
 	strbuf_release(&state.path_worktree_watch);
 	strbuf_release(&state.path_gitdir_watch);
+	strbuf_release(&state.path_cookie_prefix);
 
 	return err;
 }
diff --git a/fsmonitor--daemon.h b/fsmonitor--daemon.h
index 010fbfe60e9..bd09fffc176 100644
--- a/fsmonitor--daemon.h
+++ b/fsmonitor--daemon.h
@@ -45,6 +45,11 @@ struct fsmonitor_daemon_state {
 
 	struct fsmonitor_token_data *current_token_data;
 
+	struct strbuf path_cookie_prefix;
+	pthread_cond_t cookies_cond;
+	int cookie_seq;
+	struct hashmap cookies;
+
 	int error_code;
 	struct fsmonitor_daemon_backend_data *backend_data;
 
-- 
gitgitgadget

[PATCH v6 28/30] fsmonitor: force update index after large responses

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Measure the time taken to apply the FSMonitor query result
to the index and the untracked-cache.

Set the `FSMONITOR_CHANGED` bit on `istate->cache_changed` when
FSMonitor returns a very large repsonse to ensure that the index is
written to disk.

Normally, when the FSMonitor response includes a tracked file, the
index is always updated.  Similarly, the index might be updated when
the response alters the untracked-cache (when enabled).  However, in
cases where neither of those cause the index to be considered changed,
the FSMonitor response is wasted.  Subsequent Git commands will make
requests with the same token and receive the same response.

If that response is very large, performance may suffer.  It would be
more efficient to force update the index now (and the token in the
index extension) in order to reduce the size of the response received
by future commands.

This was observed on Windows after a large checkout.  On Windows, the
kernel emits events for the files that are changed as they are
changed.  However, it might delay events for the containing
directories until the system is more idle (or someone scans the
directory (so it seems)).  The first status following a checkout would
get the list of files.  The subsequent status commands would get the
list of directories as the events trickled out.  But they would never
catch up because the token was not advanced because the index wasn't
updated.

This list of directories caused `wt_status_collect_untracked()` to
unnecessarily spend time actually scanning them during each command.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 fsmonitor.c | 55 ++++++++++++++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 54 insertions(+), 1 deletion(-)

diff --git a/fsmonitor.c b/fsmonitor.c
index a38b5710eb3..292a6742b4f 100644
--- a/fsmonitor.c
+++ b/fsmonitor.c
@@ -219,6 +219,43 @@ static void fsmonitor_refresh_callback(struct index_state *istate, char *name)
 	untracked_cache_invalidate_path(istate, name, 0);
 }
 
+/*
+ * The number of pathnames that we need to receive from FSMonitor
+ * before we force the index to be updated.
+ *
+ * Note that any pathname within the set of received paths MAY cause
+ * cache-entry or istate flag bits to be updated and thus cause the
+ * index to be updated on disk.
+ *
+ * However, the response may contain many paths (such as ignored
+ * paths) that will not update any flag bits.  And thus not force the
+ * index to be updated.  (This is fine and normal.)  It also means
+ * that the token will not be updated in the FSMonitor index
+ * extension.  So the next Git command will find the same token in the
+ * index, make the same token-relative request, and receive the same
+ * response (plus any newly changed paths).  If this response is large
+ * (and continues to grow), performance could be impacted.
+ *
+ * For example, if the user runs a build and it writes 100K object
+ * files but doesn't modify any source files, the index would not need
+ * to be updated.  The FSMonitor response (after the build and
+ * relative to a pre-build token) might be 5MB.  Each subsequent Git
+ * command will receive that same 100K/5MB response until something
+ * causes the index to be updated.  And `refresh_fsmonitor()` will
+ * have to iterate over those 100K paths each time.
+ *
+ * Performance could be improved if we optionally force update the
+ * index after a very large response and get an updated token into
+ * the FSMonitor index extension.  This should allow subsequent
+ * commands to get smaller and more current responses.
+ *
+ * The value chosen here does not need to be precise.  The index
+ * will be updated automatically the first time the user touches
+ * a tracked file and causes a command like `git status` to
+ * update an mtime to be updated and/or set a flag bit.
+ */
+static int fsmonitor_force_update_threshold = 100;
+
 void refresh_fsmonitor(struct index_state *istate)
 {
 	struct strbuf query_result = STRBUF_INIT;
@@ -362,25 +399,39 @@ apply_results:
 	 *     information and that we should consider everything
 	 *     invalid.  We call this a trivial response.
 	 */
+	trace2_region_enter("fsmonitor", "apply_results", istate->repo);
+
 	if (query_success && !is_trivial) {
 		/*
 		 * Mark all pathnames returned by the monitor as dirty.
 		 *
 		 * This updates both the cache-entries and the untracked-cache.
 		 */
+		int count = 0;
+
 		buf = query_result.buf;
 		for (i = bol; i < query_result.len; i++) {
 			if (buf[i] != '\0')
 				continue;
 			fsmonitor_refresh_callback(istate, buf + bol);
 			bol = i + 1;
+			count++;
 		}
-		if (bol < query_result.len)
+		if (bol < query_result.len) {
 			fsmonitor_refresh_callback(istate, buf + bol);
+			count++;
+		}
 
 		/* Now mark the untracked cache for fsmonitor usage */
 		if (istate->untracked)
 			istate->untracked->use_fsmonitor = 1;
+
+		if (count > fsmonitor_force_update_threshold)
+			istate->cache_changed |= FSMONITOR_CHANGED;
+
+		trace2_data_intmax("fsmonitor", istate->repo, "apply_count",
+				   count);
+
 	} else {
 		/*
 		 * We failed to get a response or received a trivial response,
@@ -409,6 +460,8 @@ apply_results:
 		if (istate->untracked)
 			istate->untracked->use_fsmonitor = 0;
 	}
+	trace2_region_leave("fsmonitor", "apply_results", istate->repo);
+
 	strbuf_release(&query_result);
 
 	/* Now that we've updated istate, save the last_update_token */
-- 
gitgitgadget

[PATCH v6 29/30] t7527: test status with untracked-cache and fsmonitor--daemon

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Create 2x2 test matrix with the untracked-cache and fsmonitor--daemon
features and a series of edits and verify that status output is
identical.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 t/t7527-builtin-fsmonitor.sh | 93 ++++++++++++++++++++++++++++++++++++
 1 file changed, 93 insertions(+)

diff --git a/t/t7527-builtin-fsmonitor.sh b/t/t7527-builtin-fsmonitor.sh
index 5f7b8e54233..0ccbfb9616f 100755
--- a/t/t7527-builtin-fsmonitor.sh
+++ b/t/t7527-builtin-fsmonitor.sh
@@ -162,6 +162,8 @@ test_expect_success 'setup' '
 	.gitignore
 	expect*
 	actual*
+	flush*
+	trace*
 	EOF
 
 	git -c core.fsmonitor=false add . &&
@@ -508,4 +510,95 @@ test_expect_success 'cleanup worktrees' '
 	stop_daemon_delete_repo wt-base
 '
 
+# The next few tests perform arbitrary/contrived file operations and
+# confirm that status is correct.  That is, that the data (or lack of
+# data) from fsmonitor doesn't cause incorrect results.  And doesn't
+# cause incorrect results when the untracked-cache is enabled.
+
+test_lazy_prereq UNTRACKED_CACHE '
+	git update-index --test-untracked-cache
+'
+
+test_expect_success 'Matrix: setup for untracked-cache,fsmonitor matrix' '
+	test_unconfig core.fsmonitor &&
+	git update-index --no-fsmonitor &&
+	test_might_fail git fsmonitor--daemon stop
+'
+
+matrix_clean_up_repo () {
+	git reset --hard HEAD &&
+	git clean -fd
+}
+
+matrix_try () {
+	uc=$1
+	fsm=$2
+	fn=$3
+
+	test_expect_success "Matrix[uc:$uc][fsm:$fsm] $fn" '
+		matrix_clean_up_repo &&
+		$fn &&
+		if test $uc = false && test $fsm = false
+		then
+			git status --porcelain=v1 >.git/expect.$fn
+		else
+			git status --porcelain=v1 >.git/actual.$fn &&
+			test_cmp .git/expect.$fn .git/actual.$fn
+		fi
+	'
+}
+
+uc_values="false"
+test_have_prereq UNTRACKED_CACHE && uc_values="false true"
+for uc_val in $uc_values
+do
+	if test $uc_val = false
+	then
+		test_expect_success "Matrix[uc:$uc_val] disable untracked cache" '
+			git config core.untrackedcache false &&
+			git update-index --no-untracked-cache
+		'
+	else
+		test_expect_success "Matrix[uc:$uc_val] enable untracked cache" '
+			git config core.untrackedcache true &&
+			git update-index --untracked-cache
+		'
+	fi
+
+	fsm_values="false true"
+	for fsm_val in $fsm_values
+	do
+		if test $fsm_val = false
+		then
+			test_expect_success "Matrix[uc:$uc_val][fsm:$fsm_val] disable fsmonitor" '
+				test_unconfig core.fsmonitor &&
+				git update-index --no-fsmonitor &&
+				test_might_fail git fsmonitor--daemon stop
+			'
+		else
+			test_expect_success "Matrix[uc:$uc_val][fsm:$fsm_val] enable fsmonitor" '
+				git config core.fsmonitor true &&
+				git fsmonitor--daemon start &&
+				git update-index --fsmonitor
+			'
+		fi
+
+		matrix_try $uc_val $fsm_val edit_files
+		matrix_try $uc_val $fsm_val delete_files
+		matrix_try $uc_val $fsm_val create_files
+		matrix_try $uc_val $fsm_val rename_files
+		matrix_try $uc_val $fsm_val file_to_directory
+		matrix_try $uc_val $fsm_val directory_to_file
+
+		if test $fsm_val = true
+		then
+			test_expect_success "Matrix[uc:$uc_val][fsm:$fsm_val] disable fsmonitor at end" '
+				test_unconfig core.fsmonitor &&
+				git update-index --no-fsmonitor &&
+				test_might_fail git fsmonitor--daemon stop
+			'
+		fi
+	done
+done
+
 test_done
-- 
gitgitgadget

[PATCH v6 30/30] update-index: convert fsmonitor warnings to advise

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 builtin/update-index.c | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)

diff --git a/builtin/update-index.c b/builtin/update-index.c
index 876112abb21..d335f1ac72a 100644
--- a/builtin/update-index.c
+++ b/builtin/update-index.c
@@ -1238,18 +1238,18 @@ int cmd_update_index(int argc, const char **argv, const char *prefix)
 	if (fsmonitor > 0) {
 		enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
 		if (fsm_mode == FSMONITOR_MODE_DISABLED) {
-			warning(_("core.fsmonitor is unset; "
-				"set it if you really want to "
-				"enable fsmonitor"));
+			advise(_("core.fsmonitor is unset; "
+				 "set it if you really want to "
+				 "enable fsmonitor"));
 		}
 		add_fsmonitor(&the_index);
 		report(_("fsmonitor enabled"));
 	} else if (!fsmonitor) {
 		enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
 		if (fsm_mode > FSMONITOR_MODE_DISABLED)
-			warning(_("core.fsmonitor is set; "
-				"remove it if you really want to "
-				"disable fsmonitor"));
+			advise(_("core.fsmonitor is set; "
+				 "remove it if you really want to "
+				 "disable fsmonitor"));
 		remove_fsmonitor(&the_index);
 		report(_("fsmonitor disabled"));
 	}
-- 
gitgitgadget

Re: [PATCH v6 00/30] Builtin FSMonitor Part 2

[Junio C Hamano]

"Jeff Hostetler via GitGitGadget" <gitgitgadget@gmail.com> writes:

> Here is V6 of Part 2 of my builtin FSmonitor series.
>
> This version contains mostly cleanup based on feedback from V5. Of note:
>
>  * I squashed in the 1_file fix for p7519.
>  * I squashed in a commit from part 3 to optionally print the "running
>    daemon on..." message on stderr.
>  * I added a note to the documentation about incompatible changes around
>    core.fsmonitor.
>  * Removed/rephrased some obsolete NEEDSWORK items.
>
> Tao has an ongoing parallel series to fix test-chmtime on Windows.
> https://lore.kernel.org/all/pull.1166.git.1646041236.gitgitgadget@gmail.com/
>
> If that lands first, we should be able to drop my 't/helper/test-chmtime:
> skip directories on Windows' commit.
>
> A followup Part 3 will contain additional refinements to the daemon and
> additional tests. I drew the line here between Part 2 and 3 to make it
> easier to review.

Hopefully this round will quickly be reviewed solidly to merge it
down to 'next' and lower.

Will queue.

Thanks.

Re: [PATCH v6 00/30] Builtin FSMonitor Part 2

[Johannes Schindelin]

Hi Jeff,

On Tue, 1 Mar 2022, Jeff Hostetler via GitGitGadget wrote:

> Here is V6 of Part 2 of my builtin FSmonitor series.
>
> This version contains mostly cleanup based on feedback from V5. Of note:
>
>  * I squashed in the 1_file fix for p7519.
>  * I squashed in a commit from part 3 to optionally print the "running
>    daemon on..." message on stderr.
>  * I added a note to the documentation about incompatible changes around
>    core.fsmonitor.
>  * Removed/rephrased some obsolete NEEDSWORK items.

Thank you! I also saw a grammar fix ;-)

> Tao has an ongoing parallel series to fix test-chmtime on Windows.
> https://lore.kernel.org/all/pull.1166.git.1646041236.gitgitgadget@gmail.com/
>
> If that lands first, we should be able to drop my 't/helper/test-chmtime:
> skip directories on Windows' commit.

Thank you for this note so that this overwhelmed reviewer knows about
this.

> A followup Part 3 will contain additional refinements to the daemon and
> additional tests. I drew the line here between Part 2 and 3 to make it
> easier to review.

Excellent!

I only found one slight issue with the range-diff: I _assume_ that Junio
likes to add his own Sign-off (but I am being told frequently that my
assumptions are typically incorrect, so take that with a grain of salt
;-)).

Apart from that, I saw that you addressed all of my concerns, and I liked
in particular this elegant change:

> Range-diff vs v5:
>
> [...]
>   3:  384516ce1a1 !  3:  ae622a517cf fsmonitor: config settings are repository-specific
> [...]
>      @@ fsmonitor-settings.c (new)
>       +
>       +enum fsmonitor_mode fsm_settings__get_mode(struct repository *r)
>       +{
>      ++	if (!r)
>      ++		r = the_repository;
>      ++
>       +	lookup_fsmonitor_settings(r);
>       +
>       +	return r->settings.fsmonitor->mode;
> [...]
>      @@ fsmonitor.c: void remove_fsmonitor(struct index_state *istate)
>        {
>        	unsigned int i;
>       -	int fsmonitor_enabled = git_config_get_fsmonitor();
>      -+	struct repository *r = istate->repo ? istate->repo : the_repository;
>      -+	int fsmonitor_enabled = (fsm_settings__get_mode(r) > FSMONITOR_MODE_DISABLED);
>      ++	int fsmonitor_enabled = (fsm_settings__get_mode(istate->repo)
>      ++				 > FSMONITOR_MODE_DISABLED);
>
>        	if (istate->fsmonitor_dirty) {
>        		if (fsmonitor_enabled) {

From my side, this patch series is all clear to advance to `next`.

Thank you,
Dscho

Re: [PATCH v6 16/30] compat/fsmonitor/fsm-listen-darwin: add MacOS header files for FSEvent

[Ævar Arnfjörð Bjarmason]

On Tue, Mar 01 2022, Jeff Hostetler via GitGitGadget wrote:

> From: Jeff Hostetler <jeffhost@microsoft.com>
> [...]
> +#define kFSEventStreamEventFlagNone               0x00000000
> +#define kFSEventStreamEventFlagMustScanSubDirs    0x00000001
> +#define kFSEventStreamEventFlagUserDropped        0x00000002
> +#define kFSEventStreamEventFlagKernelDropped      0x00000004
> +#define kFSEventStreamEventFlagEventIdsWrapped    0x00000008
> +#define kFSEventStreamEventFlagHistoryDone        0x00000010
> +#define kFSEventStreamEventFlagRootChanged        0x00000020
> +#define kFSEventStreamEventFlagMount              0x00000040
> +#define kFSEventStreamEventFlagUnmount            0x00000080
> +#define kFSEventStreamEventFlagItemCreated        0x00000100
> +#define kFSEventStreamEventFlagItemRemoved        0x00000200
> +#define kFSEventStreamEventFlagItemInodeMetaMod   0x00000400
> +#define kFSEventStreamEventFlagItemRenamed        0x00000800
> +#define kFSEventStreamEventFlagItemModified       0x00001000
> +#define kFSEventStreamEventFlagItemFinderInfoMod  0x00002000
> +#define kFSEventStreamEventFlagItemChangeOwner    0x00004000
> +#define kFSEventStreamEventFlagItemXattrMod       0x00008000
> +#define kFSEventStreamEventFlagItemIsFile         0x00010000
> +#define kFSEventStreamEventFlagItemIsDir          0x00020000
> +#define kFSEventStreamEventFlagItemIsSymlink      0x00040000
> +#define kFSEventStreamEventFlagOwnEvent           0x00080000
> +#define kFSEventStreamEventFlagItemIsHardlink     0x00100000
> +#define kFSEventStreamEventFlagItemIsLastHardlink 0x00200000
> +#define kFSEventStreamEventFlagItemCloned         0x00400000

Can we define these as 1<<0, 1<<1, 1<<2 etc.? We do that in most other
places, and it helps to quickly eyeball these and see that they don't
have gaps.

> +#define kCFStringEncodingUTF8 0x08000100

Should this be an OR of some of the above, or is it unrelated?

> +typedef struct FSEventStreamContext FSEventStreamContext;
> +typedef unsigned int FSEventStreamEventFlags;
> +#define kFSEventStreamCreateFlagNoDefer 0x02
> +#define kFSEventStreamCreateFlagWatchRoot 0x04
> +#define kFSEventStreamCreateFlagFileEvents 0x10

Ditto 1<<0 etc.

> +#else
> +/*
> + * Let Apple's headers declare `isalnum()` first, before
> + * Git's headers override it via a constant
> + */




> +#include <string.h>
> +#include <CoreFoundation/CoreFoundation.h>
> +#include <CoreServices/CoreServices.h>
> +#endif

In cache.h which you'rejust about to include we don't include string.h,
but we do in git-compat-util.h, but that one includes string.h before
doing those overrides.

This either isn't needed, or really should be some addition to
git-compat-util.h instead. I.e. if we've missed some edge case with
string.h and ctype.h on OSX we should handle that in git-compat-util.h
rather than <some other file/header> needing a portability workaround.

> +
>  #include "cache.h"
>  #include "fsmonitor.h"
>  #include "fsm-listen.h"

Re: [PATCH v6 19/30] help: include fsmonitor--daemon feature flag in version info

[Ævar Arnfjörð Bjarmason]

On Tue, Mar 01 2022, Jeff Hostetler via GitGitGadget wrote:

> From: Jeff Hostetler <jeffhost@microsoft.com>
>
> Add the "feature: fsmonitor--daemon" message to the output of
> `git version --build-options`.
>
> The builtin FSMonitor is only available on certain platforms and
> even then only when certain Makefile flags are enabled, so print
> a message in the verbose version output when it is available.
>
> This can be used by test scripts for prereq testing.  Granted, tests
> could just try `git fsmonitor--daemon status` and look for a 128 exit
> code or grep for a "not supported" message on stderr, but these
> methods are rather obscure.
>
> The main advantage is that the feature message will automatically
> appear in bug reports and other support requests.
>
> This concept was also used during the development of Scalar for
> similar reasons.
>
> Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
> Signed-off-by: Junio C Hamano <gitster@pobox.com>
> ---
>  help.c        | 4 ++++
>  t/test-lib.sh | 6 ++++++
>  2 files changed, 10 insertions(+)
>
> diff --git a/help.c b/help.c
> index 71444906ddf..9112a51e84b 100644
> --- a/help.c
> +++ b/help.c
> @@ -12,6 +12,7 @@
>  #include "refs.h"
>  #include "parse-options.h"
>  #include "prompt.h"
> +#include "fsmonitor-ipc.h"
>  
>  struct category_description {
>  	uint32_t category;
> @@ -695,6 +696,9 @@ void get_version_info(struct strbuf *buf, int show_build_options)
>  		strbuf_addf(buf, "sizeof-size_t: %d\n", (int)sizeof(size_t));
>  		strbuf_addf(buf, "shell-path: %s\n", SHELL_PATH);
>  		/* NEEDSWORK: also save and output GIT-BUILD_OPTIONS? */
> +
> +		if (fsmonitor_ipc__is_supported())
> +			strbuf_addstr(buf, "feature: fsmonitor--daemon\n");
>  	}
>  }
>  
> diff --git a/t/test-lib.sh b/t/test-lib.sh
> index e4716b0b867..46cd596e7f5 100644
> --- a/t/test-lib.sh
> +++ b/t/test-lib.sh
> @@ -1799,3 +1799,9 @@ test_lazy_prereq SHA1 '
>  # Tests that verify the scheduler integration must set this locally
>  # to avoid errors.
>  GIT_TEST_MAINT_SCHEDULER="none:exit 1"
> +
> +# Does this platform support `git fsmonitor--daemon`
> +#
> +test_lazy_prereq FSMONITOR_DAEMON '
> +	git version --build-options | grep "feature:" | grep "fsmonitor--daemon"
> +'

As I found recently (referenced in another series) the test_lazy_prereq
doesn't currently catch segfaults etc. in git even if test_must_fail and
friends are used.

But it's still better to future-proof things and not add more cases of
git on the LHS of a pipe. So instead:

    git version .. >out &&
    grep ...

The prereqs are run in their own temporary directory, so creating those
files is OK.

Also: You run "grep" here twice, but as the code context shown we could
just run it once.

Re: [PATCH v6 20/30] t/helper/fsmonitor-client: create IPC client to talk to FSMonitor Daemon

[Ævar Arnfjörð Bjarmason]

On Tue, Mar 01 2022, Jeff Hostetler via GitGitGadget wrote:

> From: Jeff Hostetler <jeffhost@microsoft.com>
> [...]
> +/*
> + * test-fsmonitor-client.c: client code to send commands/requests to
> + * a `git fsmonitor--daemon` daemon.
> + */
> +
> +#include "test-tool.h"
> +#include "cache.h"
> +#include "parse-options.h"
> +#include "fsmonitor-ipc.h"
> +
> +#ifndef HAVE_FSMONITOR_DAEMON_BACKEND
> +int cmd__fsmonitor_client(int argc, const char **argv)
> +{
> +	die("fsmonitor--daemon not available on this platform");

UX strings added in the series should be marked for translation with
_(), but in this case this is correct as this is a test helper, so only
git developers will see it. We shouldn't waste translator time on it.

> +		die("unable to read index file");

...ditto

> +		die(_("could not query fsmonitor--daemon"));


...but this one is marked for translation, but shouldn't be.

There are other such cases in this series, found by e.g. diffing
origin/master.. and grepping for '"', not noting all of those in review,
I trust you can eyeball them & see what's missing ... :)

> +	if (argc == 2 && !strcmp(argv[1], "-h"))
> +		usage_with_options(fsmonitor_client_usage, options);

I don't think you need this special handling for -h...

> +
> +	subcmd = argv[1];
> +	argv--;
> +	argc++;
> +
> +	argc = parse_options(argc, argv, NULL, options, fsmonitor_client_usage, 0);

...because this will do it for you.

Re: [PATCH v6 21/30] t7527: create test for fsmonitor--daemon

[Ævar Arnfjörð Bjarmason]

On Tue, Mar 01 2022, Jeff Hostetler via GitGitGadget wrote:

> From: Jeff Hostetler <jeffhost@microsoft.com>
>
> Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
> Signed-off-by: Junio C Hamano <gitster@pobox.com>
> ---
>  t/t7527-builtin-fsmonitor.sh | 511 +++++++++++++++++++++++++++++++++++
>  1 file changed, 511 insertions(+)
>  create mode 100755 t/t7527-builtin-fsmonitor.sh
>
> diff --git a/t/t7527-builtin-fsmonitor.sh b/t/t7527-builtin-fsmonitor.sh
> new file mode 100755
> index 00000000000..5f7b8e54233
> --- /dev/null
> +++ b/t/t7527-builtin-fsmonitor.sh
> @@ -0,0 +1,511 @@
> +#!/bin/sh
> +
> +test_description='built-in file system watcher'
> +
> +. ./test-lib.sh
> +
> +if ! test_have_prereq FSMONITOR_DAEMON
> +then
> +	skip_all="fsmonitor--daemon is not supported on this platform"
> +	test_done
> +fi
> +
> +stop_daemon_delete_repo () {
> +	r=$1
> +	git -C $r fsmonitor--daemon stop >/dev/null 2>/dev/null

Do we really need to quiet all its output? Why not just have the
--verbose option do its thing?

> +	rm -rf $1
> +	return 0

Missing &&-chaining
> +}
> +
> +start_daemon () {
> +	case "$#" in
> +		1) r="-C $1";;
> +		*) r="";

Our usual style is not to indent these.

But maybe just use the same pattern as test_commit et al use? It's a bit
more verbose, but IMO clearer.

> +	esac
> +
> +	git $r fsmonitor--daemon start || return $?
> +	git $r fsmonitor--daemon status || return $?

Maybe don't do all this "return" and just &&-chain these instead (including the case/esac)?

> +
> +	return 0
> +}
> +
> +# Is a Trace2 data event present with the given catetory and key?
> +# We do not care what the value is.
> +#
> +have_t2_data_event () {
> +	c=$1
> +	k=$2
> +
> +	grep -e '"event":"data".*"category":"'"$c"'".*"key":"'"$k"'"'
> +}

Optional & aside: But it would be really nice to just have this amend
the test_region function into something more general, so this could be:

    test_trace2 --event data --category "$c" --key "$k"

And have "test_region" then be a thin wrapper for that, and do the
appropriate "maybe only the start one, not the end one?" logic there.

> +test_expect_success 'explicit daemon start and stop' '
> +	test_when_finished "stop_daemon_delete_repo test_explicit" &&
> +
> +	git init test_explicit &&
> +	start_daemon test_explicit &&
> +
> +	git -C test_explicit fsmonitor--daemon stop &&
> +	test_must_fail git -C test_explicit fsmonitor--daemon status
> +'
> +
> +test_expect_success 'implicit daemon start' '
> +	test_when_finished "stop_daemon_delete_repo test_implicit" &&
> +
> +	git init test_implicit &&
> +	test_must_fail git -C test_implicit fsmonitor--daemon status &&
> +
> +	# query will implicitly start the daemon.
> +	#
> +	# for test-script simplicity, we send a V1 timestamp rather than
> +	# a V2 token.  either way, the daemon response to any query contains
> +	# a new V2 token.  (the daemon may complain that we sent a V1 request,
> +	# but this test case is only concerned with whether the daemon was
> +	# implicitly started.)
> +
> +	GIT_TRACE2_EVENT="$(pwd)/.git/trace" \

Better to use $PWD than $(pwd)

> +delete_files () {
> +	rm -f delete
> +	rm -f dir1/delete
> +	rm -f dir2/delete

More missing &&-chaining.

> +}
> +
> +create_files () {
> +	echo 1 >new
> +	echo 2 >dir1/new
> +	echo 3 >dir2/new
> +}
> +
> +rename_files () {
> +	mv rename renamed
> +	mv dir1/rename dir1/renamed
> +	mv dir2/rename dir2/renamed

ditto.


> +}
> +
> +file_to_directory () {
> +	rm -f delete
> +	mkdir delete
> +	echo 1 >delete/new

ditto.


> +}
> +
> +directory_to_file () {
> +	rm -rf dir1
> +	echo 1 >dir1

ditto.
> +}
> +
> +verify_status () {

This is used by nothing? Maybe it'll be used later, but that commit
could/should add it then?

Hrm, nope, just read ahead and nothing uses it at all.

> +clean_up_repo_and_stop_daemon () {
> +	git reset --hard HEAD
> +	git clean -fd
> +	git fsmonitor--daemon stop
> +	rm -f .git/trace

Missing &&-chaining (will stop noting these now, please look through the rest...)

> +}
> +
> +test_expect_success 'edit some files' '
> +	test_when_finished clean_up_repo_and_stop_daemon &&
> +
> +	(
> +		GIT_TRACE_FSMONITOR="$(pwd)/.git/trace" &&
> +		export GIT_TRACE_FSMONITOR &&
> +
> +		start_daemon

Maybe have this "start_daemon" take an optional --trace argument or
something, allowing us to skip all these subsequent subshells.

> +	) &&
> +
> +	edit_files &&
> +
> +	test-tool fsmonitor-client query --token 0 >/dev/null 2>&1 &&

For these & the rest: just skip the quieting of the output? I.e. let the
test's --verbose do its job?


> +test_expect_success 'flush cached data' '
> +	test_when_finished "stop_daemon_delete_repo test_flush" &&
> +
> +	git init test_flush &&
> +
> +	(
> +		GIT_TEST_FSMONITOR_TOKEN=true &&
> +		export GIT_TEST_FSMONITOR_TOKEN &&
> +
> +		GIT_TRACE_FSMONITOR="$(pwd)/.git/trace_daemon" &&
> +		export GIT_TRACE_FSMONITOR &&
> +
> +		start_daemon test_flush
> +	) &&
> +
> +	# The daemon should have an initial token with no events in _0 and
> +	# then a few (probably platform-specific number of) events in _1.
> +	# These should both have the same <token_id>.
> +
> +	test-tool -C test_flush fsmonitor-client query --token "builtin:test_00000001:0" >actual_0 &&
> +	nul_to_q <actual_0 >actual_q0 &&
> +
> +	touch test_flush/file_1 &&
> +	touch test_flush/file_2 &&

I may be missinga subtlety here, but is "touch" needed v.s. ">", i.e. we
just created "test_flush", so if we create a new file it'll have the
current timestamp.

Or did it get created by the helpers?

Re: [PATCH v6 24/30] t/perf/p7519: speed up test on Windows

[Ævar Arnfjörð Bjarmason]

On Tue, Mar 01 2022, Jeff Hostetler via GitGitGadget wrote:

> From: Jeff Hostetler <jeffhost@microsoft.com>
>
> Change p7519 to use `test_seq` and `xargs` rather than a `for` loop
> to touch thousands of files.  This takes minutes off of test runs
> on Windows because of process creation overhead.
>
> Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
> Signed-off-by: Junio C Hamano <gitster@pobox.com>
> ---
>  t/perf/p7519-fsmonitor.sh | 32 ++++++++++++++++++++------------
>  1 file changed, 20 insertions(+), 12 deletions(-)
>
> diff --git a/t/perf/p7519-fsmonitor.sh b/t/perf/p7519-fsmonitor.sh
> index c8be58f3c76..aed7b1146b0 100755
> --- a/t/perf/p7519-fsmonitor.sh
> +++ b/t/perf/p7519-fsmonitor.sh
> @@ -72,7 +72,7 @@ then
>  	fi
>  fi
>  
> -trace_start() {
> +trace_start () {
>  	if test -n "$GIT_PERF_7519_TRACE"
>  	then
>  		name="$1"
> @@ -91,13 +91,20 @@ trace_start() {
>  	fi
>  }
>  
> -trace_stop() {
> +trace_stop () {
>  	if test -n "$GIT_PERF_7519_TRACE"
>  	then
>  		unset GIT_TRACE2_PERF
>  	fi
>  }

These minor unrelated style fixups could be split up / sent seperately?
Especially as the seem not to conflict hunk-wise with the actual changes
here.

> +touch_files () {
> +	n=$1
> +	d="$n"_files
> +
> +	(cd $d ; test_seq 1 $n | xargs touch )

Missing &&-chaining for "cd"

> -setup_for_fsmonitor() {
> +setup_for_fsmonitor () {
>  	# set INTEGRATION_SCRIPT depending on the environment
>  	if test -n "$INTEGRATION_PATH"
>  	then
> @@ -173,7 +181,7 @@ test_perf_w_drop_caches () {
>  	test_perf "$@"
>  }
>  
> -test_fsmonitor_suite() {
> +test_fsmonitor_suite () {

ditto unrelated style changes.

>  	if test -n "$INTEGRATION_SCRIPT"; then
>  		DESC="fsmonitor=$(basename $INTEGRATION_SCRIPT)"
>  	else
> @@ -199,15 +207,15 @@ test_fsmonitor_suite() {
>  
>  	# Update the mtimes on upto 100k files to make status think
>  	# that they are dirty.  For simplicity, omit any files with
> -	# LFs (i.e. anything that ls-files thinks it needs to dquote).
> -	# Then fully backslash-quote the paths to capture any
> -	# whitespace so that they pass thru xargs properly.
> +	# LFs (i.e. anything that ls-files thinks it needs to dquote)
> +	# and any files with whitespace so that they pass thru xargs
> +	# properly.
>  	#
>  	test_perf_w_drop_caches "status (dirty) ($DESC)" '
>  		git ls-files | \
>  			head -100000 | \
>  			grep -v \" | \
> -			sed '\''s/\(.\)/\\\1/g'\'' | \
> +			egrep -v " ." | \

Per dcf9a748cab (t7700: replace egrep with grep, 2019-12-04) should we
be adding more egrep?

Also, even if we did want that, why does a the " ." regex need ERE over
BRE here?

Re: [PATCH v6 25/30] t/perf/p7519: add fsmonitor--daemon test cases

[Ævar Arnfjörð Bjarmason]

On Tue, Mar 01 2022, Jeff Hostetler via GitGitGadget wrote:

> From: Jeff Hostetler <jeffhost@microsoft.com>
>
> Repeat all of the fsmonitor perf tests using `git fsmonitor--daemon` and
> the "Simple IPC" interface.
>
> Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
> Signed-off-by: Junio C Hamano <gitster@pobox.com>
> ---
>  t/perf/p7519-fsmonitor.sh | 29 ++++++++++++++++++++++++++++-
>  1 file changed, 28 insertions(+), 1 deletion(-)
>
> diff --git a/t/perf/p7519-fsmonitor.sh b/t/perf/p7519-fsmonitor.sh
> index aed7b1146b0..a1c552129cc 100755
> --- a/t/perf/p7519-fsmonitor.sh
> +++ b/t/perf/p7519-fsmonitor.sh
> @@ -182,7 +182,10 @@ test_perf_w_drop_caches () {
>  }
>  
>  test_fsmonitor_suite () {
> -	if test -n "$INTEGRATION_SCRIPT"; then
> +	if test -n "$USE_FSMONITOR_DAEMON"
> +	then

...another candidate for "shell script style change" re my comment on
25/30.

> +		DESC="builtin fsmonitor--daemon"
> +	elif test -n "$INTEGRATION_SCRIPT"; then


...but it's especially odd in this case, becuse here the added code
*doesn't* follow our usual style... :)

Since this hunk-wise would conflict with the meaningful change I don't
mind if we "fix it while we're at it", but it really should use the
preferred style for new code then.

>  		DESC="fsmonitor=$(basename $INTEGRATION_SCRIPT)"
>  	else
>  		DESC="fsmonitor=disabled"
> @@ -293,4 +296,28 @@ test_expect_success "setup without fsmonitor" '
>  test_fsmonitor_suite
>  trace_stop
>  
> +#
> +# Run a full set of perf tests using the built-in fsmonitor--daemon.
> +# It does not use the Hook API, so it has a different setup.
> +# Explicitly start the daemon here and before we start client commands
> +# so that we can later add custom tracing.
> +#
> +if test_have_prereq FSMONITOR_DAEMON
> +then
> +	USE_FSMONITOR_DAEMON=t
> +
> +	trace_start fsmonitor--daemon--server
> +	git fsmonitor--daemon start
> +
> +	trace_start fsmonitor--daemon--client
> +
> +	git config core.fsmonitor true
> +	git update-index --fsmonitor
> +
> +	test_fsmonitor_suite
> +
> +	git fsmonitor--daemon stop
> +	trace_stop
> +fi

Urm, shouldn't this be in a test_perf or test_expect_success?

>  test_done

Re: [PATCH v6 27/30] fsmonitor--daemon: use a cookie file to sync with file system

[Ævar Arnfjörð Bjarmason]

On Tue, Mar 01 2022, Jeff Hostetler via GitGitGadget wrote:

> From: Jeff Hostetler <jeffhost@microsoft.com>
> [...]
> +enum fsmonitor_cookie_item_result {
> +	FCIR_ERROR = -1, /* could not create cookie file ? */
> +	FCIR_INIT = 0,

nit: redundant = 0 assignment, i.e....

> +	FCIR_SEEN,
> +	FCIR_ABORT,

If we're going to make these implicit (which we usually do) let's do the
same for FCIR_INIT.

> +	strbuf_addf(&cookie_filename, "%i-%i", getpid(), my_cookie_seq);

For trace2 we encoded getpid() in the SID using a fixed-width encoding
format, maybe do the same here?

> +	/*
> +	 * Create the cookie file on disk and then wait for a notification
> +	 * that the listener thread has seen it.
> +	 */
> +	fd = open(cookie_pathname.buf, O_WRONLY | O_CREAT | O_EXCL, 0600);
> +	if (fd >= 0) {
> +		close(fd);

Also check the return value of close()...

> +		unlink(cookie_pathname.buf);

...and unlink()? If we explicitly prefer not to, note why?
> +
> +		/*
> +		 * Technically, this is an infinite wait (well, unless another
> +		 * thread sends us an abort).  I'd like to change this to
> +		 * use `pthread_cond_timedwait()` and return an error/timeout
> +		 * and let the caller do the trivial response thing, but we
> +		 * don't have that routine in our thread-utils.
> +		 *
> +		 * After extensive beta testing I'm not really worried about
> +		 * this.  Also note that the above open() and unlink() calls
> +		 * will cause at least two FS events on that path, so the odds
> +		 * of getting stuck are pretty slim.
> +		 */
> +		while (cookie->result == FCIR_INIT)
> +			pthread_cond_wait(&state->cookies_cond,
> +					  &state->main_lock);
> +	} else {
> +		error_errno(_("could not create fsmonitor cookie '%s'"),
> +			    cookie->name);
> +
> +		cookie->result = FCIR_ERROR;

...more readable if we just:

    if (open(...) < 0) {
        ...;
        goto cleanup;

So no need to indent the comment/main logic of the function above?

> +	}
> +
> +	hashmap_remove(&state->cookies, &cookie->entry, NULL);
> +
> +	result = cookie->result;
> +
> +	free((char*)cookie->name);

Since the "name" here is always an allocated string we should just make
it "char *" in the struct and avoid these casts.

In other places we play this casting game because the struct is a public
API, but in this case it's all ours.

Re: [PATCH v6 30/30] update-index: convert fsmonitor warnings to advise

[Ævar Arnfjörð Bjarmason]

On Tue, Mar 01 2022, Jeff Hostetler via GitGitGadget wrote:

> From: Jeff Hostetler <jeffhost@microsoft.com>
>
> Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
> Signed-off-by: Junio C Hamano <gitster@pobox.com>
> ---
>  builtin/update-index.c | 12 ++++++------
>  1 file changed, 6 insertions(+), 6 deletions(-)
>
> diff --git a/builtin/update-index.c b/builtin/update-index.c
> index 876112abb21..d335f1ac72a 100644
> --- a/builtin/update-index.c
> +++ b/builtin/update-index.c
> @@ -1238,18 +1238,18 @@ int cmd_update_index(int argc, const char **argv, const char *prefix)
>  	if (fsmonitor > 0) {
>  		enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
>  		if (fsm_mode == FSMONITOR_MODE_DISABLED) {
> -			warning(_("core.fsmonitor is unset; "
> -				"set it if you really want to "
> -				"enable fsmonitor"));
> +			advise(_("core.fsmonitor is unset; "
> +				 "set it if you really want to "
> +				 "enable fsmonitor"));
>  		}
>  		add_fsmonitor(&the_index);
>  		report(_("fsmonitor enabled"));
>  	} else if (!fsmonitor) {
>  		enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
>  		if (fsm_mode > FSMONITOR_MODE_DISABLED)
> -			warning(_("core.fsmonitor is set; "
> -				"remove it if you really want to "
> -				"disable fsmonitor"));
> +			advise(_("core.fsmonitor is set; "
> +				 "remove it if you really want to "
> +				 "disable fsmonitor"));
>  		remove_fsmonitor(&the_index);
>  		report(_("fsmonitor disabled"));
>  	}

For any new advice you should:

 * Add it to advice.[ch]
 * Add it to Documentation/config/advise.txt
 * This should use advise_if_enabled() with the relevant enum you add
   (which should also be documented in the *.txt)

I'm not oppposed to this change per-se, but I think this narrow change
makes the post-image of builtin/update-index.c worse.

Why are we issuing an advice() only for this config, but not changing
the other two uses of warning()? In the post-image:
    
    6 matches for "\(advise\|warning\)" in buffer: update-index.c                                                                                                                                       1195:                        warning(_("core.splitIndex is set to false; "
       1204:                        warning(_("core.splitIndex is set to true; "
       1216:                        warning(_("core.untrackedCache is set to true; "
       1228:                        warning(_("core.untrackedCache is set to false; "
       1241:                        advise(_("core.fsmonitor is unset; "
       1250:                        advise(_("core.fsmonitor is set; "

We should either leave these alone, or give them all the treatment of
adding a new advise.updateIndexConfig or whatever, and have them all use
advise_if_enabled(ADVICE_UPDATE_INDEX_CONFIG, ...).

Also these existing messages seem really confusing both for the
fsmonitor & others. We're telling the user "do XYZ to disable ABC", but
then right afterwards we say "ABC disabled" via report()?

Re: [PATCH v6 16/30] compat/fsmonitor/fsm-listen-darwin: add MacOS header files for FSEvent

[Jeff Hostetler]

On 3/7/22 5:37 AM, Ævar Arnfjörð Bjarmason wrote:
> 
> On Tue, Mar 01 2022, Jeff Hostetler via GitGitGadget wrote:
> 
>> From: Jeff Hostetler <jeffhost@microsoft.com>
>> [...]
>> +#define kFSEventStreamEventFlagNone               0x00000000
[...]
>> +#define kFSEventStreamEventFlagItemCloned         0x00400000
> 
> Can we define these as 1<<0, 1<<1, 1<<2 etc.? We do that in most other
> places, and it helps to quickly eyeball these and see that they don't
> have gaps.

All of these constants are defined by Apple in their header
files and system documentation.  For example, see:
https://developer.apple.com/documentation/coreservices/1455361-fseventstreameventflags/kfseventstreameventflagitemcloned

The set is relatively fixed by Apple and we won't be adding any
(since they define the bits in a FS event from the kernel).

Changing them to shifts would be wrong.


thanks
Jeff

Re: [PATCH v6 16/30] compat/fsmonitor/fsm-listen-darwin: add MacOS header files for FSEvent

[Jeff Hostetler]

On 3/7/22 5:37 AM, Ævar Arnfjörð Bjarmason wrote:
> 
> On Tue, Mar 01 2022, Jeff Hostetler via GitGitGadget wrote:
> 
>> From: Jeff Hostetler <jeffhost@microsoft.com>
>> [...]
[...]
> 
>> +#else
>> +/*
>> + * Let Apple's headers declare `isalnum()` first, before
>> + * Git's headers override it via a constant
>> + */
> 
> 
> 
> 
>> +#include <string.h>
>> +#include <CoreFoundation/CoreFoundation.h>
>> +#include <CoreServices/CoreServices.h>
>> +#endif
> 
> In cache.h which you'rejust about to include we don't include string.h,
> but we do in git-compat-util.h, but that one includes string.h before
> doing those overrides.
> 
> This either isn't needed, or really should be some addition to
> git-compat-util.h instead. I.e. if we've missed some edge case with
> string.h and ctype.h on OSX we should handle that in git-compat-util.h
> rather than <some other file/header> needing a portability workaround.
> 
>> +
>>   #include "cache.h"
>>   #include "fsmonitor.h"
>>   #include "fsm-listen.h"
> 

You may be right here.  I commented out the <string.h> and
the <...CoreFoundation.h> lines and everything still compiled
and t7527 passed.

I'm not sure why <string.h> was added here (I inherited that
file when I took over the feature).  It may be that recent SDK
updates have eliminated the need for it.  Or it may be that it
was never necessary.  (However, the comment above it suggests
that there was a problem in the past.)

While it may not (now at least) be necessary, it's not doing
any harm, so I'd rather leave it and not interrupt things.
We can always revisit it later if we want.

Thanks,
Jeff

Re: [PATCH v6 19/30] help: include fsmonitor--daemon feature flag in version info

[Jeff Hostetler]

On 3/7/22 5:51 AM, Ævar Arnfjörð Bjarmason wrote:
> 
> On Tue, Mar 01 2022, Jeff Hostetler via GitGitGadget wrote:
> 
>> From: Jeff Hostetler <jeffhost@microsoft.com>
>>
[...]
>> +# Does this platform support `git fsmonitor--daemon`
>> +#
>> +test_lazy_prereq FSMONITOR_DAEMON '
>> +	git version --build-options | grep "feature:" | grep "fsmonitor--daemon"
>> +'
> 
> As I found recently (referenced in another series) the test_lazy_prereq
> doesn't currently catch segfaults etc. in git even if test_must_fail and
> friends are used.
> 
> But it's still better to future-proof things and not add more cases of
> git on the LHS of a pipe. So instead:
> 
>      git version .. >out &&
>      grep ...
> 
> The prereqs are run in their own temporary directory, so creating those
> files is OK.

Yes, I think I saw a series on this topic in my inbox recently.

This command is very safe, so I'd rather not reroll just for this.


> 
> Also: You run "grep" here twice, but as the code context shown we could
> just run it once.
> 

True, but I think this can wait too.

Thanks,
Jeff

Re: [PATCH v6 16/30] compat/fsmonitor/fsm-listen-darwin: add MacOS header files for FSEvent

[Jeff Hostetler]

On 3/7/22 5:37 AM, Ævar Arnfjörð Bjarmason wrote:
> 
> On Tue, Mar 01 2022, Jeff Hostetler via GitGitGadget wrote:
> 
>> From: Jeff Hostetler <jeffhost@microsoft.com>
>> [...]
[...]

Ævar sent feedback (thanks!) on 8 commits in the V6 version
on March 7.  I started responding to each as I got to them
in my inbox yesterday, but I'd like to take a break from
responding individually to each of them inside of Part 2.

Since most of the feedback is for "general cleanup" and since
Part 2 V6 is already in "next", I'd like to revisit these
issues with a few "cleanup" commits on top of Part 3 (which
is still in active review), rather than re-rolling or
appending "fixup" commits onto Part 2.

I think this would be quicker and easier in the long run
and give us the same net result.

Thanks,
Jeff

Re: [PATCH v6 16/30] compat/fsmonitor/fsm-listen-darwin: add MacOS header files for FSEvent

[Ævar Arnfjörð Bjarmason]

On Tue, Mar 08 2022, Jeff Hostetler wrote:

> On 3/7/22 5:37 AM, Ævar Arnfjörð Bjarmason wrote:
>> On Tue, Mar 01 2022, Jeff Hostetler via GitGitGadget wrote:
>> 
>>> From: Jeff Hostetler <jeffhost@microsoft.com>
>>> [...]
>>> +#define kFSEventStreamEventFlagNone               0x00000000
> [...]
>>> +#define kFSEventStreamEventFlagItemCloned         0x00400000
>> Can we define these as 1<<0, 1<<1, 1<<2 etc.? We do that in most
>> other
>> places, and it helps to quickly eyeball these and see that they don't
>> have gaps.
>
> All of these constants are defined by Apple in their header
> files and system documentation.  For example, see:
> https://developer.apple.com/documentation/coreservices/1455361-fseventstreameventflags/kfseventstreameventflagitemcloned
>
> The set is relatively fixed by Apple and we won't be adding any
> (since they define the bits in a FS event from the kernel).
>
> Changing them to shifts would be wrong.

Ah, I missed the "ifndef __clang__" at the start, so most of it is not
needed except with gcc.

FWIW I think having that whole part just split into
compat/fsmonitor/darwin-gcc.h would make it obvious where all the
GCC-specific hackery is.

Re: [PATCH v6 16/30] compat/fsmonitor/fsm-listen-darwin: add MacOS header files for FSEvent

[Ævar Arnfjörð Bjarmason]

On Tue, Mar 08 2022, Jeff Hostetler wrote:

> On 3/7/22 5:37 AM, Ævar Arnfjörð Bjarmason wrote:
>> On Tue, Mar 01 2022, Jeff Hostetler via GitGitGadget wrote:
>> 
>>> From: Jeff Hostetler <jeffhost@microsoft.com>
>>> [...]
> [...]
>> 
>>> +#else
>>> +/*
>>> + * Let Apple's headers declare `isalnum()` first, before
>>> + * Git's headers override it via a constant
>>> + */
>> 
>> 
>>> +#include <string.h>
>>> +#include <CoreFoundation/CoreFoundation.h>
>>> +#include <CoreServices/CoreServices.h>
>>> +#endif
>> In cache.h which you'rejust about to include we don't include
>> string.h,
>> but we do in git-compat-util.h, but that one includes string.h before
>> doing those overrides.
>> This either isn't needed, or really should be some addition to
>> git-compat-util.h instead. I.e. if we've missed some edge case with
>> string.h and ctype.h on OSX we should handle that in git-compat-util.h
>> rather than <some other file/header> needing a portability workaround.
>> 
>>> +
>>>   #include "cache.h"
>>>   #include "fsmonitor.h"
>>>   #include "fsm-listen.h"
>> 
>
> You may be right here.  I commented out the <string.h> and
> the <...CoreFoundation.h> lines and everything still compiled
> and t7527 passed.
>
> I'm not sure why <string.h> was added here (I inherited that
> file when I took over the feature).  It may be that recent SDK
> updates have eliminated the need for it.  Or it may be that it
> was never necessary.  (However, the comment above it suggests
> that there was a problem in the past.)
>
> While it may not (now at least) be necessary, it's not doing
> any harm, so I'd rather leave it and not interrupt things.
> We can always revisit it later if we want.

In terms of figuring out some mysterious portability issue, I think the
right time is now rather than later.

I.e. now this doesn't have anyone relying on it, so we can easily
(re)discover whatever issue this was trying to solve.

Whereas anyone who'd need to figure out why we include string.h on OSX
early in this case later would be left with this otherwise dead-end
thread, and a change at that point would possibly break existing code...

Re: [PATCH v6 16/30] compat/fsmonitor/fsm-listen-darwin: add MacOS header files for FSEvent

[Junio C Hamano]

Jeff Hostetler <git@jeffhostetler.com> writes:

> On 3/7/22 5:37 AM, Ævar Arnfjörð Bjarmason wrote:
>> On Tue, Mar 01 2022, Jeff Hostetler via GitGitGadget wrote:
>> 
>>> From: Jeff Hostetler <jeffhost@microsoft.com>
>>> [...]
> [...]
>
> Ævar sent feedback (thanks!) on 8 commits in the V6 version
> on March 7.  I started responding to each as I got to them
> in my inbox yesterday, but I'd like to take a break from
> responding individually to each of them inside of Part 2.
>
> Since most of the feedback is for "general cleanup" and since
> Part 2 V6 is already in "next", I'd like to revisit these
> issues with a few "cleanup" commits on top of Part 3 (which
> is still in active review), rather than re-rolling or
> appending "fixup" commits onto Part 2.

Sounds good.  Prepending "preliminary clean-up" before part 3 would
be even cleaner, I would suspect.

In any case, let's consider part 2 "more or less done" unless we see
a glaring mistake that requires us to revert and redo it from
scratch.

Thanks.

Re: [PATCH v6 16/30] compat/fsmonitor/fsm-listen-darwin: add MacOS header files for FSEvent

[Ævar Arnfjörð Bjarmason]

On Wed, Mar 09 2022, Junio C Hamano wrote:

> Jeff Hostetler <git@jeffhostetler.com> writes:
>
>> On 3/7/22 5:37 AM, Ævar Arnfjörð Bjarmason wrote:
>>> On Tue, Mar 01 2022, Jeff Hostetler via GitGitGadget wrote:
>>> 
>>>> From: Jeff Hostetler <jeffhost@microsoft.com>
>>>> [...]
>> [...]
>>
>> Ævar sent feedback (thanks!) on 8 commits in the V6 version
>> on March 7.  I started responding to each as I got to them
>> in my inbox yesterday, but I'd like to take a break from
>> responding individually to each of them inside of Part 2.
>>
>> Since most of the feedback is for "general cleanup" and since
>> Part 2 V6 is already in "next", I'd like to revisit these
>> issues with a few "cleanup" commits on top of Part 3 (which
>> is still in active review), rather than re-rolling or
>> appending "fixup" commits onto Part 2.
>
> Sounds good.  Prepending "preliminary clean-up" before part 3 would
> be even cleaner, I would suspect.
>
> In any case, let's consider part 2 "more or less done" unless we see
> a glaring mistake that requires us to revert and redo it from
> scratch.

Sounds good, my comments on v6 today were before I'd noticed that it was
in "next", I think all of those can (well, it would be that way either
way at this point...) be left for some potential follow-up.

Thanks both, especially Jeff for sticking with this fsmonitor topic for
so long & keeping it advancing.

Re: [PATCH v6 16/30] compat/fsmonitor/fsm-listen-darwin: add MacOS header files for FSEvent

[Johannes Schindelin]

[-- Attachment #1: Type: text/plain, Size: 4055 bytes --]

Hi Ævar,

On Wed, 9 Mar 2022, Ævar Arnfjörð Bjarmason wrote:

> On Tue, Mar 08 2022, Jeff Hostetler wrote:
>
> > On 3/7/22 5:37 AM, Ævar Arnfjörð Bjarmason wrote:
> >> On Tue, Mar 01 2022, Jeff Hostetler via GitGitGadget wrote:
> >>
> >>> From: Jeff Hostetler <jeffhost@microsoft.com>
> >>> [...]
> > [...]
> >>
> >>> +#else
> >>> +/*
> >>> + * Let Apple's headers declare `isalnum()` first, before
> >>> + * Git's headers override it via a constant
> >>> + */
> >>
> >>
> >>> +#include <string.h>
> >>> +#include <CoreFoundation/CoreFoundation.h>
> >>> +#include <CoreServices/CoreServices.h>
> >>> +#endif
> >>
> >> In cache.h which you'rejust about to include we don't include
> >> string.h, but we do in git-compat-util.h, but that one includes
> >> string.h before doing those overrides.
> >>
> >> This either isn't needed, or really should be some addition to
> >> git-compat-util.h instead. I.e. if we've missed some edge case with
> >> string.h and ctype.h on OSX we should handle that in git-compat-util.h
> >> rather than <some other file/header> needing a portability workaround.
> >
> > [...]
> >
> > While it may not (now at least) be necessary, it's not doing
> > any harm, so I'd rather leave it and not interrupt things.
> > We can always revisit it later if we want.
>
> In terms of figuring out some mysterious portability issue, I think the
> right time is now rather than later.

I do not see that.

In FSMonitor, this was clearly a problem that needed to be solved (and if
you try to compile on an older macOS, you will run into those problems
again).

If you are talking about the mysterious portability issue with an eye on
`git-compat-util.h`, well... you can successfully compile Git's source
code without this hack in `git-compat-util.h`. That's why the hack is not
needed. Problem solved. Actually, there was not even a problem.

> I.e. now this doesn't have anyone relying on it, so we can easily
> (re)discover whatever issue this was trying to solve.
>
> Whereas anyone who'd need to figure out why we include string.h on OSX
> early in this case later would be left with this otherwise dead-end
> thread, and a change at that point would possibly break existing code...

Anyone who would need to figure out why we `#include` this header early
would read the comment about `isalnum()`, I would hope, and understand
that there are circumstances when Git's `isalnum()` macro interferes with
Apple's, and that this `#include` order addresses that problem.

They might even get to the point where they find
https://github.com/dscho/git/commit/0f89c726a1912dce2bdab14aff8ebfec8550340d,
maybe even the "original original" commit at
https://github.com/kewillford/git/commit/d11ee4698c63347f40a8993ab86ee4e97f695d9b,
which was a still-experimental version of the macOS backend, and where the
`#include` order clearly mattered, else why would Kevin have bothered.

Further, I strongly suspect that it had something to do with
`CoreFoundation.h` or with `CoreServices.h` being `#include`d, and if you
care to check the code above the quoted lines, you will see that you
cannot even `#include` those headers using GCC, it only works with clang:
https://github.com/jeffhostetler/git/commit/cdef9730b3f93a6f0f98d68ffb81bcb89d6e698e#diff-4e865160016fe490b80ad11273a10daca8bc412a75f2da4c6b08fb9e5e3b5e18R3

At this stage, anybody investigating this issue who is a little bit like
me would then be a bit bored with the investigation because there is
actually no breakage here, just a curious `#include` order, and nothing
else. So they might then drop it and move along.

Even you might find it a more satisfying use of your time to implement,
say, a Linux backend for FSMonitor on top of Jeff's work, instead of
worrying about non-issues ;-)

Really, there are so many more interesting issues to discuss than this
`#include` non-issue. And on this note, I will steer my attention to
precisely such more interesting issues.

Ciao,
Johannes

Re: [PATCH v6 16/30] compat/fsmonitor/fsm-listen-darwin: add MacOS header files for FSEvent

[Ævar Arnfjörð Bjarmason]

On Thu, Mar 10 2022, Johannes Schindelin wrote:

> Hi Ævar,
>
> On Wed, 9 Mar 2022, Ævar Arnfjörð Bjarmason wrote:
>
>> On Tue, Mar 08 2022, Jeff Hostetler wrote:
>>
>> > On 3/7/22 5:37 AM, Ævar Arnfjörð Bjarmason wrote:
>> >> On Tue, Mar 01 2022, Jeff Hostetler via GitGitGadget wrote:
>> >>
>> >>> From: Jeff Hostetler <jeffhost@microsoft.com>
>> >>> [...]
>> > [...]
>> >>
>> >>> +#else
>> >>> +/*
>> >>> + * Let Apple's headers declare `isalnum()` first, before
>> >>> + * Git's headers override it via a constant
>> >>> + */
>> >>
>> >>
>> >>> +#include <string.h>
>> >>> +#include <CoreFoundation/CoreFoundation.h>
>> >>> +#include <CoreServices/CoreServices.h>
>> >>> +#endif
>> >>
>> >> In cache.h which you'rejust about to include we don't include
>> >> string.h, but we do in git-compat-util.h, but that one includes
>> >> string.h before doing those overrides.
>> >>
>> >> This either isn't needed, or really should be some addition to
>> >> git-compat-util.h instead. I.e. if we've missed some edge case with
>> >> string.h and ctype.h on OSX we should handle that in git-compat-util.h
>> >> rather than <some other file/header> needing a portability workaround.
>> >
>> > [...]
>> >
>> > While it may not (now at least) be necessary, it's not doing
>> > any harm, so I'd rather leave it and not interrupt things.
>> > We can always revisit it later if we want.
>>
>> In terms of figuring out some mysterious portability issue, I think the
>> right time is now rather than later.
>
> I do not see that.
>
> In FSMonitor, this was clearly a problem that needed to be solved (and if
> you try to compile on an older macOS, you will run into those problems
> again).

So you can reproduce an issue where removing the "#include <string.h>"
from compat/fsmonitor/fsm-listen-darwin.c has an effect? Does swaping it
for "ctype.h" also solve that issue?

I was asking why the non-obvious portability hack was needed, and it
seems Jeff suggested it might not be upthread in
<aa6276f9-8d10-22f9-bfc0-2aa718d002e1@jeffhostetler.com>, but here you
seem to have a reproduction of in being needed, without more of the
relevant details (e.g. what OSX version(s))?

> If you are talking about the mysterious portability issue with an eye on
> `git-compat-util.h`, well... you can successfully compile Git's source
> code without this hack in `git-compat-util.h`. That's why the hack is not
> needed. Problem solved. Actually, there was not even a problem.

Do you mean we don't need the ctype.h overrides in git-compat-util.h at
all? I haven't looked into it, but needing to

>> I.e. now this doesn't have anyone relying on it, so we can easily
>> (re)discover whatever issue this was trying to solve.
>>
>> Whereas anyone who'd need to figure out why we include string.h on OSX
>> early in this case later would be left with this otherwise dead-end
>> thread, and a change at that point would possibly break existing code...
>
> Anyone who would need to figure out why we `#include` this header early
> would read the comment about `isalnum()`, I would hope, and understand
> that there are circumstances when Git's `isalnum()` macro interferes with
> Apple's, and that this `#include` order addresses that problem.
>
> They might even get to the point where they find
> https://github.com/dscho/git/commit/0f89c726a1912dce2bdab14aff8ebfec8550340d,
> maybe even the "original original" commit at
> https://github.com/kewillford/git/commit/d11ee4698c63347f40a8993ab86ee4e97f695d9b,
> which was a still-experimental version of the macOS backend, and where the
> `#include` order clearly mattered, else why would Kevin have bothered.
>
> Further, I strongly suspect that it had something to do with
> `CoreFoundation.h` or with `CoreServices.h` being `#include`d, and if you
> care to check the code above the quoted lines, you will see that you
> cannot even `#include` those headers using GCC, it only works with clang:
> https://github.com/jeffhostetler/git/commit/cdef9730b3f93a6f0f98d68ffb81bcb89d6e698e#diff-4e865160016fe490b80ad11273a10daca8bc412a75f2da4c6b08fb9e5e3b5e18R3
>
> At this stage, anybody investigating this issue who is a little bit like
> me would then be a bit bored with the investigation because there is
> actually no breakage here, just a curious `#include` order, and nothing
> else. So they might then drop it and move along.

My implicit observation upthread is that those sorts of details would
ideally be included in a comment or the commit message. I.e. I didn't
quite see why it was needed, and neither could the person submitting the
patch series when asked.

Sure, it's a minor issue, but patch review is also meant to uncover such
small issues.

Re: [PATCH v6 16/30] compat/fsmonitor/fsm-listen-darwin: add MacOS header files for FSEvent

[Jeff Hostetler]

On 3/10/22 9:42 AM, Ævar Arnfjörð Bjarmason wrote:
> 
> On Thu, Mar 10 2022, Johannes Schindelin wrote:
> 
>> Hi Ævar,
>>
>> On Wed, 9 Mar 2022, Ævar Arnfjörð Bjarmason wrote:
>>
>>> On Tue, Mar 08 2022, Jeff Hostetler wrote:
>>>
>>>> On 3/7/22 5:37 AM, Ævar Arnfjörð Bjarmason wrote:
>>>>> On Tue, Mar 01 2022, Jeff Hostetler via GitGitGadget wrote:
>>>>>
>>>>>> From: Jeff Hostetler <jeffhost@microsoft.com>
>>>>>> [...]
>>>> [...]
>>>>>
>>>>>> +#else
>>>>>> +/*
>>>>>> + * Let Apple's headers declare `isalnum()` first, before
>>>>>> + * Git's headers override it via a constant
>>>>>> + */
>>>>>
>>>>>
>>>>>> +#include <string.h>
>>>>>> +#include <CoreFoundation/CoreFoundation.h>
>>>>>> +#include <CoreServices/CoreServices.h>
>>>>>> +#endif
>>>>>
>>>>> In cache.h which you'rejust about to include we don't include
>>>>> string.h, but we do in git-compat-util.h, but that one includes
>>>>> string.h before doing those overrides.
>>>>>
>>>>> This either isn't needed, or really should be some addition to
>>>>> git-compat-util.h instead. I.e. if we've missed some edge case with
>>>>> string.h and ctype.h on OSX we should handle that in git-compat-util.h
>>>>> rather than <some other file/header> needing a portability workaround.
>>>>
>>>> [...]
>>>>
>>>> While it may not (now at least) be necessary, it's not doing
>>>> any harm, so I'd rather leave it and not interrupt things.
>>>> We can always revisit it later if we want.
>>>
>>> In terms of figuring out some mysterious portability issue, I think the
>>> right time is now rather than later.
>>
>> I do not see that.
>>
>> In FSMonitor, this was clearly a problem that needed to be solved (and if
>> you try to compile on an older macOS, you will run into those problems
>> again).
> 
> So you can reproduce an issue where removing the "#include <string.h>"
> from compat/fsmonitor/fsm-listen-darwin.c has an effect? Does swaping it
> for "ctype.h" also solve that issue?
> 
> I was asking why the non-obvious portability hack was needed, and it
> seems Jeff suggested it might not be upthread in
> <aa6276f9-8d10-22f9-bfc0-2aa718d002e1@jeffhostetler.com>, but here you
> seem to have a reproduction of in being needed, without more of the
> relevant details (e.g. what OSX version(s))?
> 
>> If you are talking about the mysterious portability issue with an eye on
>> `git-compat-util.h`, well... you can successfully compile Git's source
>> code without this hack in `git-compat-util.h`. That's why the hack is not
>> needed. Problem solved. Actually, there was not even a problem.
> 
> Do you mean we don't need the ctype.h overrides in git-compat-util.h at
> all? I haven't looked into it, but needing to
> 
>>> I.e. now this doesn't have anyone relying on it, so we can easily
>>> (re)discover whatever issue this was trying to solve.
>>>
>>> Whereas anyone who'd need to figure out why we include string.h on OSX
>>> early in this case later would be left with this otherwise dead-end
>>> thread, and a change at that point would possibly break existing code...
>>
>> Anyone who would need to figure out why we `#include` this header early
>> would read the comment about `isalnum()`, I would hope, and understand
>> that there are circumstances when Git's `isalnum()` macro interferes with
>> Apple's, and that this `#include` order addresses that problem.
>>
>> They might even get to the point where they find
>> https://github.com/dscho/git/commit/0f89c726a1912dce2bdab14aff8ebfec8550340d,
>> maybe even the "original original" commit at
>> https://github.com/kewillford/git/commit/d11ee4698c63347f40a8993ab86ee4e97f695d9b,
>> which was a still-experimental version of the macOS backend, and where the
>> `#include` order clearly mattered, else why would Kevin have bothered.
>>
>> Further, I strongly suspect that it had something to do with
>> `CoreFoundation.h` or with `CoreServices.h` being `#include`d, and if you
>> care to check the code above the quoted lines, you will see that you
>> cannot even `#include` those headers using GCC, it only works with clang:
>> https://github.com/jeffhostetler/git/commit/cdef9730b3f93a6f0f98d68ffb81bcb89d6e698e#diff-4e865160016fe490b80ad11273a10daca8bc412a75f2da4c6b08fb9e5e3b5e18R3
>>
>> At this stage, anybody investigating this issue who is a little bit like
>> me would then be a bit bored with the investigation because there is
>> actually no breakage here, just a curious `#include` order, and nothing
>> else. So they might then drop it and move along.
> 
> My implicit observation upthread is that those sorts of details would
> ideally be included in a comment or the commit message. I.e. I didn't
> quite see why it was needed, and neither could the person submitting the
> patch series when asked.
> 
> Sure, it's a minor issue, but patch review is also meant to uncover such
> small issues.
> 

There are two independent issues here.
(1) compiling something that includes <CoreServices.h> with GCC.
(2) the need for the #include <string.h> when compiling with clang.

To address (1), we've #ifdef'd the top of the file and insert just
the essential typedefs and prototypes.  (I'll pull them into a separate
local header file as you suggested earlier to make that easier to see.)
But otherwise, GCC is not an issue.

WRT (2) I have tried clang-11 on macOS 10.15 and clang-13 on 11.6 both
with and without the <string.h> and it doesn't matter.  Everything
compiles and t7527 passes in all [2x2] cases.  So I have to assume that
something has changed in the Apple/clang SDK or runtime libraries or
our source code in that single source file in the ~2 years since Kevin
needed to add it.  I do not have access to an older Mac (Apple makes it
hard to test back-compat with older OS's), so I cannot reproduce the
error.  But I don't doubt that there was an error at one point -- I just
don't know what it was.   So that's my context for saying that I don't
think it is needed now, but I was willing to carry it forward in case
it is still helpful for people with older Macs.  FWIW, it really seems
pretty isolated and trivial and would only affect code in this single
source file -- which from a quick scan, doesn't actually reference any
of the functions in <string.h>, so there shouldn't be any need to think
about git-compat-util.h or ctype.h, right?

Jeff

Re: [PATCH v6 03/30] fsmonitor: config settings are repository-specific

[Ævar Arnfjörð Bjarmason]

On Tue, Mar 01 2022, Jeff Hostetler via GitGitGadget wrote:

> From: Jeff Hostetler <jeffhost@microsoft.com>

I know this is in "next", just looking over this code again...

> +static void lookup_fsmonitor_settings(struct repository *r)

Here we'll start loading the settings...

> +{
> +	struct fsmonitor_settings *s;
> +	const char *const_str;
> +	int bool_value;
> +
> +	if (r->settings.fsmonitor)
> +		return;

MARK

> +	CALLOC_ARRAY(s, 1);
> +
> +	r->settings.fsmonitor = s;

And right after we alloc the r->settings.fsmonitor we'll ...

> +	fsm_settings__set_disabled(r);

...call this method...
> +
> +	/*
> +	 * Overload the existing "core.fsmonitor" config setting (which
> +	 * has historically been either unset or a hook pathname) to
> +	 * now allow a boolean value to enable the builtin FSMonitor
> +	 * or to turn everything off.  (This does imply that you can't
> +	 * use a hook script named "true" or "false", but that's OK.)
> +	 */
> +	switch (repo_config_get_maybe_bool(r, "core.fsmonitor", &bool_value)) {
> +
> +	case 0: /* config value was set to <bool> */
> +		if (bool_value)
> +			fsm_settings__set_ipc(r);
> +		return;
> +
> +	case 1: /* config value was unset */
> +		const_str = getenv("GIT_TEST_FSMONITOR");
> +		break;
> +
> +	case -1: /* config value set to an arbitrary string */
> +		if (repo_config_get_pathname(r, "core.fsmonitor", &const_str))
> +			return; /* should not happen */
> +		break;
> +
> +	default: /* should not happen */
> +		return;
> +	}
> +
> +	if (!const_str || !*const_str)
> +		return;
> +
> +	fsm_settings__set_hook(r, const_str);
> +}
> [...]
> +void fsm_settings__set_disabled(struct repository *r)
> +{
> +	if (!r)
> +		r = the_repository;
> +
> +	lookup_fsmonitor_settings(r);

...which here will recurse into lookup_fsmonitor_settings and hit
"MARK".

So isn't that fsm_settings__set_disabled() within that method pointless?

> +	r->settings.fsmonitor->mode = FSMONITOR_MODE_DISABLED;
> +	FREE_AND_NULL(r->settings.fsmonitor->hook_path);

It seems as though the intent was to reach this, but these all happen to
be the same thing you'd get with CALLOC_ARRAY(), so I think this just
happened to work out...

> +enum fsmonitor_mode {
> +	FSMONITOR_MODE_DISABLED = 0,

...I.e. this is luckily zero.

Re: [PATCH v6 03/30] fsmonitor: config settings are repository-specific

[Jeff Hostetler]

On 3/10/22 8:47 PM, Ævar Arnfjörð Bjarmason wrote:
> 
> On Tue, Mar 01 2022, Jeff Hostetler via GitGitGadget wrote:
> 
>> From: Jeff Hostetler <jeffhost@microsoft.com>
> 
> I know this is in "next", just looking over this code again...
> 
>> +static void lookup_fsmonitor_settings(struct repository *r)
> 
> Here we'll start loading the settings...
> 
>> +{
>> +	struct fsmonitor_settings *s;
>> +	const char *const_str;
>> +	int bool_value;
>> +
>> +	if (r->settings.fsmonitor)
>> +		return;
> 
> MARK
> 
>> +	CALLOC_ARRAY(s, 1);
>> +
>> +	r->settings.fsmonitor = s;
> 
> And right after we alloc the r->settings.fsmonitor we'll ...
> 
>> +	fsm_settings__set_disabled(r);
> 
> ...call this method...
>> +
>> +	/*
>> +	 * Overload the existing "core.fsmonitor" config setting (which
>> +	 * has historically been either unset or a hook pathname) to
>> +	 * now allow a boolean value to enable the builtin FSMonitor
>> +	 * or to turn everything off.  (This does imply that you can't
>> +	 * use a hook script named "true" or "false", but that's OK.)
>> +	 */
>> +	switch (repo_config_get_maybe_bool(r, "core.fsmonitor", &bool_value)) {
>> +
>> +	case 0: /* config value was set to <bool> */
>> +		if (bool_value)
>> +			fsm_settings__set_ipc(r);
>> +		return;
>> +
>> +	case 1: /* config value was unset */
>> +		const_str = getenv("GIT_TEST_FSMONITOR");
>> +		break;
>> +
>> +	case -1: /* config value set to an arbitrary string */
>> +		if (repo_config_get_pathname(r, "core.fsmonitor", &const_str))
>> +			return; /* should not happen */
>> +		break;
>> +
>> +	default: /* should not happen */
>> +		return;
>> +	}
>> +
>> +	if (!const_str || !*const_str)
>> +		return;
>> +
>> +	fsm_settings__set_hook(r, const_str);
>> +}
>> [...]
>> +void fsm_settings__set_disabled(struct repository *r)
>> +{
>> +	if (!r)
>> +		r = the_repository;
>> +
>> +	lookup_fsmonitor_settings(r);
> 
> ...which here will recurse into lookup_fsmonitor_settings and hit
> "MARK".
> 
> So isn't that fsm_settings__set_disabled() within that method pointless?

Yeah, that got rolled up in one too many rebase/refactorings...
I'll simplify.

thanks
jeff

> 
>> +	r->settings.fsmonitor->mode = FSMONITOR_MODE_DISABLED;
>> +	FREE_AND_NULL(r->settings.fsmonitor->hook_path);
> 
> It seems as though the intent was to reach this, but these all happen to
> be the same thing you'd get with CALLOC_ARRAY(), so I think this just
> happened to work out...
> 
>> +enum fsmonitor_mode {
>> +	FSMONITOR_MODE_DISABLED = 0,
> 
> ...I.e. this is luckily zero.
> 

Re: [PATCH v6 16/30] compat/fsmonitor/fsm-listen-darwin: add MacOS header files for FSEvent

[Jeff Hostetler]

On 3/9/22 1:57 PM, Junio C Hamano wrote:
> Jeff Hostetler <git@jeffhostetler.com> writes:
> 
>> On 3/7/22 5:37 AM, Ævar Arnfjörð Bjarmason wrote:
>>> On Tue, Mar 01 2022, Jeff Hostetler via GitGitGadget wrote:
>>>
>>>> From: Jeff Hostetler <jeffhost@microsoft.com>
>>>> [...]
>> [...]
>>
>> Ævar sent feedback (thanks!) on 8 commits in the V6 version
>> on March 7.  I started responding to each as I got to them
>> in my inbox yesterday, but I'd like to take a break from
>> responding individually to each of them inside of Part 2.
>>
>> Since most of the feedback is for "general cleanup" and since
>> Part 2 V6 is already in "next", I'd like to revisit these
>> issues with a few "cleanup" commits on top of Part 3 (which
>> is still in active review), rather than re-rolling or
>> appending "fixup" commits onto Part 2.
> 
> Sounds good.  Prepending "preliminary clean-up" before part 3 would
> be even cleaner, I would suspect.
> 
> In any case, let's consider part 2 "more or less done" unless we see
> a glaring mistake that requires us to revert and redo it from
> scratch.
> 
> Thanks.
> 

The cleanup here turned into 16 small commits.  I'll send them as
a "Part 2.5" so that they stand alone and can either be appended
to part 2 or treated as a new branch.

GGG wouldn't let me send fixup! commits, so inside of each
commit message is a "fixup! ..." line which you can use if
you want to squash them into part 2.  But otherwise they
have a normal (non-fixup) commit summary.

After I send that I'll send a new version of part 3 that builds
upon them.

Thanks
Jeff

[PATCH v7 00/29] Builtin FSMonitor Part 2

[Jeff Hostetler via GitGitGadget]

Here is V7 of Part 2 of my builtin FSmonitor series. This version squashes
in the fixups from Part 2.5 and any other last minute suggestions. I'll send
an updated version of Part 3 that builds upon this version.

Here is a range-diff from V6 to V7 relative to 715d08a9e5 (The eighth batch,
2022-02-25). Changes since V6 can be summarized as: (1) Adding/removing
"_()" on die() and error() messages. (2) Splitting GCC-specific declarations
for MacOS FSEvents into a separate header file. (3) Enhancing the
"start_daemon()" function in t7527 to take keyword args and reduce some code
duplication in unit tests. (4) Fixing "&&" chains in the tests. (5) Other
minor style cleanups.

$ git range-diff eight..builtin-fsmonitor-part2-v6 eight..builtin-fsmonitor-part2
 1:  10a34082fc =  1:  e98373f997 fsmonitor: enhance existing comments, clarify trivial response handling
 2:  e1c946af2c !  2:  ab68b94417 fsmonitor-ipc: create client routines for git-fsmonitor--daemon
    @@ fsmonitor-ipc.c (new)
     +    state = ipc_client_try_connect(fsmonitor_ipc__get_path(), &options,
     +                       &connection);
     +    if (state != IPC_STATE__LISTENING) {
    -+        die("fsmonitor--daemon is not running");
    ++        die(_("fsmonitor--daemon is not running"));
     +        return -1;
     +    }
     +
    @@ fsmonitor-ipc.c (new)
     +    ipc_client_close_connection(connection);
     +
     +    if (ret == -1) {
    -+        die("could not send '%s' command to fsmonitor--daemon", c);
    ++        die(_("could not send '%s' command to fsmonitor--daemon"), c);
     +        return -1;
     +    }
     +
 3:  ae622a517c !  3:  e04c7301f2 fsmonitor: config settings are repository-specific
    @@ fsmonitor-settings.c (new)
     +        return;
     +
     +    CALLOC_ARRAY(s, 1);
    ++    s->mode = FSMONITOR_MODE_DISABLED;
     +
     +    r->settings.fsmonitor = s;
     +
    -+    fsm_settings__set_disabled(r);
    -+
     +    /*
     +     * Overload the existing "core.fsmonitor" config setting (which
     +     * has historically been either unset or a hook pathname) to
 4:  55974867da =  4:  ea02ba25d8 fsmonitor: use IPC to query the builtin FSMonitor daemon
 5:  ce42d5bbaf =  5:  6ab7db9cb7 fsmonitor: document builtin fsmonitor
 6:  9ce938c69b =  6:  0ce8ae3f2c fsmonitor--daemon: add a built-in fsmonitor daemon
 7:  3bd7e6a051 =  7:  4624ce2fa4 fsmonitor--daemon: implement 'stop' and 'status' commands
 8:  92d3e54da7 =  8:  a29fe7266a compat/fsmonitor/fsm-listen-win32: stub in backend for Windows
 9:  f121922127 =  9:  2f8a42fdb9 compat/fsmonitor/fsm-listen-darwin: stub in backend for Darwin
10:  a3f94627cf ! 10:  f07800690e fsmonitor--daemon: implement 'run' command
    @@ builtin/fsmonitor--daemon.c: static int do_as_client__status(void)
     +     * common error case.
     +     */
     +    if (fsmonitor_ipc__get_state() == IPC_STATE__LISTENING)
    -+        die("fsmonitor--daemon is already running '%s'",
    ++        die(_("fsmonitor--daemon is already running '%s'"),
     +            the_repository->worktree);
     +
     +    if (fsmonitor__announce_startup) {
11:  8de40b0fe8 ! 11:  a6a39a3306 fsmonitor--daemon: implement 'start' command
    @@ builtin/fsmonitor--daemon.c: static int try_to_run_foreground_daemon(void)
     +     * immediately exited).
     +     */
     +    if (fsmonitor_ipc__get_state() == IPC_STATE__LISTENING)
    -+        die("fsmonitor--daemon is already running '%s'",
    ++        die(_("fsmonitor--daemon is already running '%s'"),
     +            the_repository->worktree);
     +
     +    if (fsmonitor__announce_startup) {
    @@ builtin/fsmonitor--daemon.c: static int try_to_run_foreground_daemon(void)
     +    default:
     +    case SBGR_ERROR:
     +    case SBGR_CB_ERROR:
    -+        return error("daemon failed to start");
    ++        return error(_("daemon failed to start"));
     +
     +    case SBGR_TIMEOUT:
    -+        return error("daemon not online yet");
    ++        return error(_("daemon not online yet"));
     +
     +    case SBGR_DIED:
    -+        return error("daemon terminated");
    ++        return error(_("daemon terminated"));
     +    }
     +}
     +
12:  e5d419b1ea = 12:  d62e338d00 fsmonitor--daemon: add pathname classification
13:  c2f3668a86 = 13:  53e06b4ae5 fsmonitor--daemon: define token-ids
14:  bd492d13fb = 14:  39f43fabe0 fsmonitor--daemon: create token-based changed path cache
15:  6b5b3d0822 ! 15:  239558e34f compat/fsmonitor/fsm-listen-win32: implement FSMonitor backend on Windows
    @@ compat/fsmonitor/fsm-listen-win32.c
     +        if (len > 0)
     +            goto normalize;
     +        if (GetLastError() != ERROR_INSUFFICIENT_BUFFER) {
    -+            error("[GLE %ld] could not convert path to UTF-8: '%.*ls'",
    ++            error(_("[GLE %ld] could not convert path to UTF-8: '%.*ls'"),
     +                  GetLastError(),
     +                  (int)(info->FileNameLength / sizeof(WCHAR)),
     +                  info->FileName);
    @@ compat/fsmonitor/fsm-listen-win32.c
     +    if (watch->is_active)
     +        return 0;
     +
    -+    error("ReadDirectoryChangedW failed on '%s' [GLE %ld]",
    ++    error(_("ReadDirectoryChangedW failed on '%s' [GLE %ld]"),
     +          watch->path.buf, GetLastError());
     +    return -1;
     +}
    @@ compat/fsmonitor/fsm-listen-win32.c
     +     * sure it is worth it.
     +     */
     +
    -+    error("GetOverlappedResult failed on '%s' [GLE %ld]",
    ++    error(_("GetOverlappedResult failed on '%s' [GLE %ld]"),
     +          watch->path.buf, gle);
     +    return -1;
     +}
16:  cdef9730b3 ! 16:  14b775e9d8 compat/fsmonitor/fsm-listen-darwin: add MacOS header files for FSEvent
    @@ Commit message
         Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
         Signed-off-by: Junio C Hamano <gitster@pobox.com>
     
    - ## compat/fsmonitor/fsm-listen-darwin.c ##
    + ## compat/fsmonitor/fsm-darwin-gcc.h (new) ##
     @@
    ++#ifndef FSM_DARWIN_GCC_H
    ++#define FSM_DARWIN_GCC_H
    ++
     +#ifndef __clang__
     +/*
     + * It is possible to #include CoreFoundation/CoreFoundation.h when compiling
    @@ compat/fsmonitor/fsm-listen-darwin.c
     +void FSEventStreamStop(FSEventStreamRef stream);
     +void FSEventStreamInvalidate(FSEventStreamRef stream);
     +void FSEventStreamRelease(FSEventStreamRef stream);
    ++
    ++#endif /* !clang */
    ++#endif /* FSM_DARWIN_GCC_H */
    +
    + ## compat/fsmonitor/fsm-listen-darwin.c ##
    +@@
    ++#ifndef __clang__
    ++#include "fsm-darwin-gcc.h"
     +#else
    -+/*
    -+ * Let Apple's headers declare `isalnum()` first, before
    -+ * Git's headers override it via a constant
    -+ */
    -+#include <string.h>
     +#include <CoreFoundation/CoreFoundation.h>
     +#include <CoreServices/CoreServices.h>
    ++
    ++#ifndef AVAILABLE_MAC_OS_X_VERSION_10_13_AND_LATER
    ++/*
    ++ * This enum value was added in 10.13 to:
    ++ *
    ++ * /Applications/Xcode.app/Contents/Developer/Platforms/ \
    ++ *    MacOSX.platform/Developer/SDKs/MacOSX.sdk/System/ \
    ++ *    Library/Frameworks/CoreServices.framework/Frameworks/ \
    ++ *    FSEvents.framework/Versions/Current/Headers/FSEvents.h
    ++ *
    ++ * If we're compiling against an older SDK, this symbol won't be
    ++ * present.  Silently define it here so that we don't have to ifdef
    ++ * the logging or masking below.  This should be harmless since older
    ++ * versions of macOS won't ever emit this FS event anyway.
    ++ */
    ++#define kFSEventStreamEventFlagItemCloned         0x00400000
    ++#endif
     +#endif
     +
      #include "cache.h"
17:  aabfebd3a4 ! 17:  55bd7aee06 compat/fsmonitor/fsm-listen-darwin: implement FSEvent listener on MacOS
    @@ Commit message
         Signed-off-by: Junio C Hamano <gitster@pobox.com>
     
      ## compat/fsmonitor/fsm-listen-darwin.c ##
    -@@ compat/fsmonitor/fsm-listen-darwin.c: void FSEventStreamRelease(FSEventStreamRef stream);
    +@@
      #include "cache.h"
      #include "fsmonitor.h"
      #include "fsm-listen.h"
    @@ compat/fsmonitor/fsm-listen-darwin.c: void FSEventStreamRelease(FSEventStreamRef
     +    return 0;
     +
     +failed:
    -+    error("Unable to create FSEventStream.");
    ++    error(_("Unable to create FSEventStream."));
     +
     +    FREE_AND_NULL(state->backend_data);
          return -1;
    @@ compat/fsmonitor/fsm-listen-darwin.c: void FSEventStreamRelease(FSEventStreamRef
     +    data->stream_scheduled = 1;
     +
     +    if (!FSEventStreamStart(data->stream)) {
    -+        error("Failed to start the FSEventStream");
    ++        error(_("Failed to start the FSEventStream"));
     +        goto force_error_stop_without_loop;
     +    }
     +    data->stream_started = 1;
18:  0896e72e2b = 18:  1f4b5209bf fsmonitor--daemon: implement handle_client callback
19:  bc240a9e66 ! 19:  8cf62c9fc6 help: include fsmonitor--daemon feature flag in version info
    @@ t/test-lib.sh: test_lazy_prereq SHA1 '
     +# Does this platform support `git fsmonitor--daemon`
     +#
     +test_lazy_prereq FSMONITOR_DAEMON '
    -+    git version --build-options | grep "feature:" | grep "fsmonitor--daemon"
    ++    git version --build-options >output &&
    ++    grep "feature: fsmonitor--daemon" output
     +'
20:  2b563c240e ! 20:  1bd74a8159 t/helper/fsmonitor-client: create IPC client to talk to FSMonitor Daemon
    @@ t/helper/test-fsmonitor-client.c (new)
     +
     +    ret = fsmonitor_ipc__send_query(token, &answer);
     +    if (ret < 0)
    -+        die(_("could not query fsmonitor--daemon"));
    ++        die("could not query fsmonitor--daemon");
     +
     +    write_in_full(1, answer.buf, answer.len);
     +    strbuf_release(&answer);
    @@ t/helper/test-fsmonitor-client.c (new)
     +    const char *token = NULL;
     +
     +    const char * const fsmonitor_client_usage[] = {
    -+        N_("test-helper fsmonitor-client query [<token>]"),
    -+        N_("test-helper fsmonitor-client flush"),
    ++        "test-tool fsmonitor-client query [<token>]",
    ++        "test-tool fsmonitor-client flush",
     +        NULL,
     +    };
     +
     +    struct option options[] = {
    -+        OPT_STRING(0, "token", &token, N_("token"),
    -+               N_("command token to send to the server")),
    ++        OPT_STRING(0, "token", &token, "token",
    ++               "command token to send to the server"),
     +        OPT_END()
     +    };
     +
    -+    if (argc < 2)
    -+        usage_with_options(fsmonitor_client_usage, options);
    ++    argc = parse_options(argc, argv, NULL, options, fsmonitor_client_usage, 0);
     +
    -+    if (argc == 2 && !strcmp(argv[1], "-h"))
    ++    if (argc != 1)
     +        usage_with_options(fsmonitor_client_usage, options);
     +
    -+    subcmd = argv[1];
    -+    argv--;
    -+    argc++;
    -+
    -+    argc = parse_options(argc, argv, NULL, options, fsmonitor_client_usage, 0);
    ++    subcmd = argv[0];
     +
     +    setup_git_directory();
     +
21:  db8809da08 ! 21:  4a920d0b54 t7527: create test for fsmonitor--daemon
    @@ t/t7527-builtin-fsmonitor.sh (new)
     +fi
     +
     +stop_daemon_delete_repo () {
    -+    r=$1
    -+    git -C $r fsmonitor--daemon stop >/dev/null 2>/dev/null
    ++    r=$1 &&
    ++    test_might_fail git -C $r fsmonitor--daemon stop &&
     +    rm -rf $1
    -+    return 0
     +}
     +
     +start_daemon () {
    -+    case "$#" in
    -+        1) r="-C $1";;
    -+        *) r="";
    -+    esac
    ++    r= &&
    ++    tf= &&
    ++    t2= &&
    ++    tk= &&
    ++
    ++    while test "$#" -ne 0
    ++    do
    ++        case "$1" in
    ++        -C)
    ++            shift;
    ++            test "$#" -ne 0 || BUG "error: -C requires arg"
    ++            r="-C $1"
    ++            shift
    ++            ;;
    ++        -tf)
    ++            shift;
    ++            test "$#" -ne 0 || BUG "error: -tf requires arg"
    ++            tf="$1"
    ++            shift
    ++            ;;
    ++        -t2)
    ++            shift;
    ++            test "$#" -ne 0 || BUG "error: -t2 requires arg"
    ++            t2="$1"
    ++            shift
    ++            ;;
    ++        -tk)
    ++            shift;
    ++            test "$#" -ne 0 || BUG "error: -tk requires arg"
    ++            tk="$1"
    ++            shift
    ++            ;;
    ++        *)
    ++            BUG "error: unknown option: '$1'"
    ++            ;;
    ++        esac
    ++    done &&
     +
    -+    git $r fsmonitor--daemon start || return $?
    -+    git $r fsmonitor--daemon status || return $?
    -+
    -+    return 0
    ++    (
    ++        if test -n "$tf"
    ++        then
    ++            GIT_TRACE_FSMONITOR="$tf"
    ++            export GIT_TRACE_FSMONITOR
    ++        fi &&
    ++
    ++        if test -n "$t2"
    ++        then
    ++            GIT_TRACE2_PERF="$t2"
    ++            export GIT_TRACE2_PERF
    ++        fi &&
    ++
    ++        if test -n "$tk"
    ++        then
    ++            GIT_TEST_FSMONITOR_TOKEN="$tk"
    ++            export GIT_TEST_FSMONITOR_TOKEN
    ++        fi &&
    ++
    ++        git $r fsmonitor--daemon start &&
    ++        git $r fsmonitor--daemon status
    ++    )
     +}
     +
     +# Is a Trace2 data event present with the given catetory and key?
     +# We do not care what the value is.
     +#
     +have_t2_data_event () {
    -+    c=$1
    -+    k=$2
    ++    c=$1 &&
    ++    k=$2 &&
     +
     +    grep -e '"event":"data".*"category":"'"$c"'".*"key":"'"$k"'"'
     +}
    @@ t/t7527-builtin-fsmonitor.sh (new)
     +    test_when_finished "stop_daemon_delete_repo test_explicit" &&
     +
     +    git init test_explicit &&
    -+    start_daemon test_explicit &&
    ++    start_daemon -C test_explicit &&
     +
     +    git -C test_explicit fsmonitor--daemon stop &&
     +    test_must_fail git -C test_explicit fsmonitor--daemon status
    @@ t/t7527-builtin-fsmonitor.sh (new)
     +    # but this test case is only concerned with whether the daemon was
     +    # implicitly started.)
     +
    -+    GIT_TRACE2_EVENT="$(pwd)/.git/trace" \
    ++    GIT_TRACE2_EVENT="$PWD/.git/trace" \
     +        test-tool -C test_implicit fsmonitor-client query --token 0 >actual &&
     +    nul_to_q <actual >actual.filtered &&
     +    grep "builtin:" actual.filtered &&
    @@ t/t7527-builtin-fsmonitor.sh (new)
     +
     +    git init test_implicit_1 &&
     +
    -+    start_daemon test_implicit_1 &&
    ++    start_daemon -C test_implicit_1 &&
     +
     +    # deleting the .git directory will implicitly stop the daemon.
     +    rm -rf test_implicit_1/.git &&
    @@ t/t7527-builtin-fsmonitor.sh (new)
     +
     +    git init test_implicit_2 &&
     +
    -+    start_daemon test_implicit_2 &&
    ++    start_daemon -C test_implicit_2 &&
     +
     +    # renaming the .git directory will implicitly stop the daemon.
     +    mv test_implicit_2/.git test_implicit_2/.xxx &&
    @@ t/t7527-builtin-fsmonitor.sh (new)
     +
     +    git init test_multiple &&
     +
    -+    start_daemon test_multiple &&
    ++    start_daemon -C test_multiple &&
     +
     +    test_must_fail git -C test_multiple fsmonitor--daemon start 2>actual &&
     +    grep "fsmonitor--daemon is already running" actual &&
    @@ t/t7527-builtin-fsmonitor.sh (new)
     +# This is here in case something else fails first.
     +#
     +redundant_stop_daemon () {
    -+    git fsmonitor--daemon stop
    -+    return 0
    ++    test_might_fail git fsmonitor--daemon stop
     +}
     +
     +test_expect_success 'update-index implicitly starts daemon' '
    @@ t/t7527-builtin-fsmonitor.sh (new)
     +
     +    test_must_fail git fsmonitor--daemon status &&
     +
    -+    GIT_TRACE2_EVENT="$(pwd)/.git/trace_implicit_1" \
    ++    GIT_TRACE2_EVENT="$PWD/.git/trace_implicit_1" \
     +        git update-index --fsmonitor &&
     +
     +    git fsmonitor--daemon status &&
    @@ t/t7527-builtin-fsmonitor.sh (new)
     +
     +    test_must_fail git fsmonitor--daemon status &&
     +
    -+    GIT_TRACE2_EVENT="$(pwd)/.git/trace_implicit_2" \
    ++    GIT_TRACE2_EVENT="$PWD/.git/trace_implicit_2" \
     +        git status >actual &&
     +
     +    git fsmonitor--daemon status &&
    @@ t/t7527-builtin-fsmonitor.sh (new)
     +'
     +
     +edit_files () {
    -+    echo 1 >modified
    -+    echo 2 >dir1/modified
    -+    echo 3 >dir2/modified
    ++    echo 1 >modified &&
    ++    echo 2 >dir1/modified &&
    ++    echo 3 >dir2/modified &&
     +    >dir1/untracked
     +}
     +
     +delete_files () {
    -+    rm -f delete
    -+    rm -f dir1/delete
    ++    rm -f delete &&
    ++    rm -f dir1/delete &&
     +    rm -f dir2/delete
     +}
     +
     +create_files () {
    -+    echo 1 >new
    -+    echo 2 >dir1/new
    ++    echo 1 >new &&
    ++    echo 2 >dir1/new &&
     +    echo 3 >dir2/new
     +}
     +
     +rename_files () {
    -+    mv rename renamed
    -+    mv dir1/rename dir1/renamed
    ++    mv rename renamed &&
    ++    mv dir1/rename dir1/renamed &&
     +    mv dir2/rename dir2/renamed
     +}
     +
     +file_to_directory () {
    -+    rm -f delete
    -+    mkdir delete
    ++    rm -f delete &&
    ++    mkdir delete &&
     +    echo 1 >delete/new
     +}
     +
     +directory_to_file () {
    -+    rm -rf dir1
    ++    rm -rf dir1 &&
     +    echo 1 >dir1
     +}
     +
    -+verify_status () {
    -+    git status >actual &&
    -+    GIT_INDEX_FILE=.git/fresh-index git read-tree master &&
    -+    GIT_INDEX_FILE=.git/fresh-index git -c core.fsmonitor=false status >expect &&
    -+    test_cmp expect actual &&
    -+    echo HELLO AFTER &&
    -+    cat .git/trace &&
    -+    echo HELLO AFTER
    -+}
    -+
     +# The next few test cases confirm that our fsmonitor daemon sees each type
     +# of OS filesystem notification that we care about.  At this layer we just
     +# ensure we are getting the OS notifications and do not try to confirm what
    @@ t/t7527-builtin-fsmonitor.sh (new)
     +# daemon) because these commands might implicitly restart the daemon.
     +
     +clean_up_repo_and_stop_daemon () {
    -+    git reset --hard HEAD
    -+    git clean -fd
    -+    git fsmonitor--daemon stop
    ++    git reset --hard HEAD &&
    ++    git clean -fd &&
    ++    test_might_fail git fsmonitor--daemon stop &&
     +    rm -f .git/trace
     +}
     +
     +test_expect_success 'edit some files' '
     +    test_when_finished clean_up_repo_and_stop_daemon &&
     +
    -+    (
    -+        GIT_TRACE_FSMONITOR="$(pwd)/.git/trace" &&
    -+        export GIT_TRACE_FSMONITOR &&
    -+
    -+        start_daemon
    -+    ) &&
    ++    start_daemon -tf "$PWD/.git/trace" &&
     +
     +    edit_files &&
     +
    -+    test-tool fsmonitor-client query --token 0 >/dev/null 2>&1 &&
    ++    test-tool fsmonitor-client query --token 0 &&
     +
     +    grep "^event: dir1/modified$"  .git/trace &&
     +    grep "^event: dir2/modified$"  .git/trace &&
    @@ t/t7527-builtin-fsmonitor.sh (new)
     +test_expect_success 'create some files' '
     +    test_when_finished clean_up_repo_and_stop_daemon &&
     +
    -+    (
    -+        GIT_TRACE_FSMONITOR="$(pwd)/.git/trace" &&
    -+        export GIT_TRACE_FSMONITOR &&
    -+
    -+        start_daemon
    -+    ) &&
    ++    start_daemon -tf "$PWD/.git/trace" &&
     +
     +    create_files &&
     +
    -+    test-tool fsmonitor-client query --token 0 >/dev/null 2>&1 &&
    ++    test-tool fsmonitor-client query --token 0 &&
     +
     +    grep "^event: dir1/new$" .git/trace &&
     +    grep "^event: dir2/new$" .git/trace &&
    @@ t/t7527-builtin-fsmonitor.sh (new)
     +test_expect_success 'delete some files' '
     +    test_when_finished clean_up_repo_and_stop_daemon &&
     +
    -+    (
    -+        GIT_TRACE_FSMONITOR="$(pwd)/.git/trace" &&
    -+        export GIT_TRACE_FSMONITOR &&
    -+
    -+        start_daemon
    -+    ) &&
    ++    start_daemon -tf "$PWD/.git/trace" &&
     +
     +    delete_files &&
     +
    -+    test-tool fsmonitor-client query --token 0 >/dev/null 2>&1 &&
    ++    test-tool fsmonitor-client query --token 0 &&
     +
     +    grep "^event: dir1/delete$" .git/trace &&
     +    grep "^event: dir2/delete$" .git/trace &&
    @@ t/t7527-builtin-fsmonitor.sh (new)
     +test_expect_success 'rename some files' '
     +    test_when_finished clean_up_repo_and_stop_daemon &&
     +
    -+    (
    -+        GIT_TRACE_FSMONITOR="$(pwd)/.git/trace" &&
    -+        export GIT_TRACE_FSMONITOR &&
    -+
    -+        start_daemon
    -+    ) &&
    ++    start_daemon -tf "$PWD/.git/trace" &&
     +
     +    rename_files &&
     +
    -+    test-tool fsmonitor-client query --token 0 >/dev/null 2>&1 &&
    ++    test-tool fsmonitor-client query --token 0 &&
     +
     +    grep "^event: dir1/rename$"  .git/trace &&
     +    grep "^event: dir2/rename$"  .git/trace &&
    @@ t/t7527-builtin-fsmonitor.sh (new)
     +test_expect_success 'rename directory' '
     +    test_when_finished clean_up_repo_and_stop_daemon &&
     +
    -+    (
    -+        GIT_TRACE_FSMONITOR="$(pwd)/.git/trace" &&
    -+        export GIT_TRACE_FSMONITOR &&
    -+
    -+        start_daemon
    -+    ) &&
    ++    start_daemon -tf "$PWD/.git/trace" &&
     +
     +    mv dirtorename dirrenamed &&
     +
    -+    test-tool fsmonitor-client query --token 0 >/dev/null 2>&1 &&
    ++    test-tool fsmonitor-client query --token 0 &&
     +
     +    grep "^event: dirtorename/*$" .git/trace &&
     +    grep "^event: dirrenamed/*$"  .git/trace
    @@ t/t7527-builtin-fsmonitor.sh (new)
     +test_expect_success 'file changes to directory' '
     +    test_when_finished clean_up_repo_and_stop_daemon &&
     +
    -+    (
    -+        GIT_TRACE_FSMONITOR="$(pwd)/.git/trace" &&
    -+        export GIT_TRACE_FSMONITOR &&
    -+
    -+        start_daemon
    -+    ) &&
    ++    start_daemon -tf "$PWD/.git/trace" &&
     +
     +    file_to_directory &&
     +
    -+    test-tool fsmonitor-client query --token 0 >/dev/null 2>&1 &&
    ++    test-tool fsmonitor-client query --token 0 &&
     +
     +    grep "^event: delete$"     .git/trace &&
     +    grep "^event: delete/new$" .git/trace
    @@ t/t7527-builtin-fsmonitor.sh (new)
     +test_expect_success 'directory changes to a file' '
     +    test_when_finished clean_up_repo_and_stop_daemon &&
     +
    -+    (
    -+        GIT_TRACE_FSMONITOR="$(pwd)/.git/trace" &&
    -+        export GIT_TRACE_FSMONITOR &&
    -+
    -+        start_daemon
    -+    ) &&
    ++    start_daemon -tf "$PWD/.git/trace" &&
     +
     +    directory_to_file &&
     +
    -+    test-tool fsmonitor-client query --token 0 >/dev/null 2>&1 &&
    ++    test-tool fsmonitor-client query --token 0 &&
     +
     +    grep "^event: dir1$" .git/trace
     +'
    @@ t/t7527-builtin-fsmonitor.sh (new)
     +
     +    git init test_flush &&
     +
    -+    (
    -+        GIT_TEST_FSMONITOR_TOKEN=true &&
    -+        export GIT_TEST_FSMONITOR_TOKEN &&
    -+
    -+        GIT_TRACE_FSMONITOR="$(pwd)/.git/trace_daemon" &&
    -+        export GIT_TRACE_FSMONITOR &&
    -+
    -+        start_daemon test_flush
    -+    ) &&
    ++    start_daemon -C test_flush -tf "$PWD/.git/trace_daemon" -tk true &&
     +
     +    # The daemon should have an initial token with no events in _0 and
     +    # then a few (probably platform-specific number of) events in _1.
    @@ t/t7527-builtin-fsmonitor.sh (new)
     +    test-tool -C test_flush fsmonitor-client query --token "builtin:test_00000001:0" >actual_0 &&
     +    nul_to_q <actual_0 >actual_q0 &&
     +
    -+    touch test_flush/file_1 &&
    -+    touch test_flush/file_2 &&
    ++    >test_flush/file_1 &&
    ++    >test_flush/file_2 &&
     +
     +    test-tool -C test_flush fsmonitor-client query --token "builtin:test_00000001:0" >actual_1 &&
     +    nul_to_q <actual_1 >actual_q1 &&
    @@ t/t7527-builtin-fsmonitor.sh (new)
     +
     +    grep "^builtin:test_00000002:0Q$" actual_q2 &&
     +
    -+    touch test_flush/file_3 &&
    ++    >test_flush/file_3 &&
     +
     +    test-tool -C test_flush fsmonitor-client query --token "builtin:test_00000002:0" >actual_3 &&
     +    nul_to_q <actual_3 >actual_q3 &&
    @@ t/t7527-builtin-fsmonitor.sh (new)
     +test_expect_success 'worktree with .git file' '
     +    git -C wt-base worktree add ../wt-secondary &&
     +
    -+    (
    -+        GIT_TRACE2_PERF="$(pwd)/trace2_wt_secondary" &&
    -+        export GIT_TRACE2_PERF &&
    -+
    -+        GIT_TRACE_FSMONITOR="$(pwd)/trace_wt_secondary" &&
    -+        export GIT_TRACE_FSMONITOR &&
    -+
    -+        start_daemon wt-secondary
    -+    ) &&
    ++    start_daemon -C wt-secondary \
    ++        -tf "$PWD/trace_wt_secondary" \
    ++        -t2 "$PWD/trace2_wt_secondary" &&
     +
     +    git -C wt-secondary fsmonitor--daemon stop &&
     +    test_must_fail git -C wt-secondary fsmonitor--daemon status
22:  15bd5aaff3 = 22:  c925a9a745 t/perf: avoid copying builtin fsmonitor files into test repo
23:  4cb97918d2 = 23:  5b3381c223 t/helper/test-chmtime: skip directories on Windows
24:  a70748b464 ! 24:  803a540cc0 t/perf/p7519: speed up test on Windows
    @@ t/perf/p7519-fsmonitor.sh: test_fsmonitor_suite() {
                  head -100000 | \
                  grep -v \" | \
     -            sed '\''s/\(.\)/\\\1/g'\'' | \
    -+            egrep -v " ." | \
    ++            grep -v " ." | \
                  xargs test-tool chmtime -300 &&
              git status
          '
25:  b0c9c9c7a4 ! 25:  d5ca2df31c t/perf/p7519: add fsmonitor--daemon test cases
    @@ Commit message
         Signed-off-by: Junio C Hamano <gitster@pobox.com>
     
      ## t/perf/p7519-fsmonitor.sh ##
    +@@ t/perf/p7519-fsmonitor.sh: test_expect_success "one time repo setup" '
    +     fi
    + '
    + 
    +-setup_for_fsmonitor () {
    ++setup_for_fsmonitor_hook () {
    +     # set INTEGRATION_SCRIPT depending on the environment
    +     if test -n "$INTEGRATION_PATH"
    +     then
     @@ t/perf/p7519-fsmonitor.sh: test_perf_w_drop_caches () {
      }
      
    @@ t/perf/p7519-fsmonitor.sh: test_perf_w_drop_caches () {
     +    if test -n "$USE_FSMONITOR_DAEMON"
     +    then
     +        DESC="builtin fsmonitor--daemon"
    -+    elif test -n "$INTEGRATION_SCRIPT"; then
    ++    elif test -n "$INTEGRATION_SCRIPT"
    ++    then
              DESC="fsmonitor=$(basename $INTEGRATION_SCRIPT)"
          else
              DESC="fsmonitor=disabled"
    +@@ t/perf/p7519-fsmonitor.sh: test_fsmonitor_suite () {
    + trace_start fsmonitor-watchman
    + if test -n "$GIT_PERF_7519_FSMONITOR"; then
    +     for INTEGRATION_PATH in $GIT_PERF_7519_FSMONITOR; do
    +-        test_expect_success "setup for fsmonitor $INTEGRATION_PATH" 'setup_for_fsmonitor'
    ++        test_expect_success "setup for fsmonitor $INTEGRATION_PATH" 'setup_for_fsmonitor_hook'
    +         test_fsmonitor_suite
    +     done
    + else
    +-    test_expect_success "setup for fsmonitor" 'setup_for_fsmonitor'
    ++    test_expect_success "setup for fsmonitor hook" 'setup_for_fsmonitor_hook'
    +     test_fsmonitor_suite
    + fi
    + 
     @@ t/perf/p7519-fsmonitor.sh: test_expect_success "setup without fsmonitor" '
      test_fsmonitor_suite
      trace_stop
    @@ t/perf/p7519-fsmonitor.sh: test_expect_success "setup without fsmonitor" '
     +then
     +    USE_FSMONITOR_DAEMON=t
     +
    -+    trace_start fsmonitor--daemon--server
    -+    git fsmonitor--daemon start
    ++    test_expect_success "setup for builtin fsmonitor" '
    ++        trace_start fsmonitor--daemon--server &&
    ++        git fsmonitor--daemon start &&
     +
    -+    trace_start fsmonitor--daemon--client
    ++        trace_start fsmonitor--daemon--client &&
     +
    -+    git config core.fsmonitor true
    -+    git update-index --fsmonitor
    ++        git config core.fsmonitor true &&
    ++        git update-index --fsmonitor
    ++    '
     +
     +    test_fsmonitor_suite
     +
26:  64a5b74167 = 26:  42631259e8 fsmonitor--daemon: periodically truncate list of modified files
27:  5b06eb5d0e ! 27:  f256c3cbe8 fsmonitor--daemon: use a cookie file to sync with file system
    @@ builtin/fsmonitor--daemon.c: static int do_as_client__status(void)
      
     +enum fsmonitor_cookie_item_result {
     +    FCIR_ERROR = -1, /* could not create cookie file ? */
    -+    FCIR_INIT = 0,
    ++    FCIR_INIT,
     +    FCIR_SEEN,
     +    FCIR_ABORT,
     +};
     +
     +struct fsmonitor_cookie_item {
     +    struct hashmap_entry entry;
    -+    const char *name;
    ++    char *name;
     +    enum fsmonitor_cookie_item_result result;
     +};
     +
    @@ builtin/fsmonitor--daemon.c: static int do_as_client__status(void)
     +     * that the listener thread has seen it.
     +     */
     +    fd = open(cookie_pathname.buf, O_WRONLY | O_CREAT | O_EXCL, 0600);
    -+    if (fd >= 0) {
    -+        close(fd);
    -+        unlink(cookie_pathname.buf);
    -+
    -+        /*
    -+         * Technically, this is an infinite wait (well, unless another
    -+         * thread sends us an abort).  I'd like to change this to
    -+         * use `pthread_cond_timedwait()` and return an error/timeout
    -+         * and let the caller do the trivial response thing, but we
    -+         * don't have that routine in our thread-utils.
    -+         *
    -+         * After extensive beta testing I'm not really worried about
    -+         * this.  Also note that the above open() and unlink() calls
    -+         * will cause at least two FS events on that path, so the odds
    -+         * of getting stuck are pretty slim.
    -+         */
    -+        while (cookie->result == FCIR_INIT)
    -+            pthread_cond_wait(&state->cookies_cond,
    -+                      &state->main_lock);
    -+    } else {
    ++    if (fd < 0) {
     +        error_errno(_("could not create fsmonitor cookie '%s'"),
     +                cookie->name);
     +
     +        cookie->result = FCIR_ERROR;
    ++        goto done;
     +    }
     +
    ++    /*
    ++     * Technically, close() and unlink() can fail, but we don't
    ++     * care here.  We only created the file to trigger a watch
    ++     * event from the FS to know that when we're up to date.
    ++     */
    ++    close(fd);
    ++    unlink(cookie_pathname.buf);
    ++
    ++    /*
    ++     * Technically, this is an infinite wait (well, unless another
    ++     * thread sends us an abort).  I'd like to change this to
    ++     * use `pthread_cond_timedwait()` and return an error/timeout
    ++     * and let the caller do the trivial response thing, but we
    ++     * don't have that routine in our thread-utils.
    ++     *
    ++     * After extensive beta testing I'm not really worried about
    ++     * this.  Also note that the above open() and unlink() calls
    ++     * will cause at least two FS events on that path, so the odds
    ++     * of getting stuck are pretty slim.
    ++     */
    ++    while (cookie->result == FCIR_INIT)
    ++        pthread_cond_wait(&state->cookies_cond,
    ++                  &state->main_lock);
    ++
    ++done:
     +    hashmap_remove(&state->cookies, &cookie->entry, NULL);
     +
     +    result = cookie->result;
     +
    -+    free((char*)cookie->name);
    ++    free(cookie->name);
     +    free(cookie);
     +    strbuf_release(&cookie_pathname);
     +
28:  1fd5439de0 = 28:  08af8296f9 fsmonitor: force update index after large responses
29:  6fc1430285 ! 29:  e6cf84dc8e t7527: test status with untracked-cache and fsmonitor--daemon
    @@ t/t7527-builtin-fsmonitor.sh: test_expect_success 'cleanup worktrees' '
     +}
     +
     +matrix_try () {
    -+    uc=$1
    -+    fsm=$2
    -+    fn=$3
    ++    uc=$1 &&
    ++    fsm=$2 &&
    ++    fn=$3 &&
     +
     +    test_expect_success "Matrix[uc:$uc][fsm:$fsm] $fn" '
     +        matrix_clean_up_repo &&
30:  b915b95cc2 <  -:  ---------- update-index: convert fsmonitor warnings to advise


Jeff Hostetler (29):
  fsmonitor: enhance existing comments, clarify trivial response
    handling
  fsmonitor-ipc: create client routines for git-fsmonitor--daemon
  fsmonitor: config settings are repository-specific
  fsmonitor: use IPC to query the builtin FSMonitor daemon
  fsmonitor: document builtin fsmonitor
  fsmonitor--daemon: add a built-in fsmonitor daemon
  fsmonitor--daemon: implement 'stop' and 'status' commands
  compat/fsmonitor/fsm-listen-win32: stub in backend for Windows
  compat/fsmonitor/fsm-listen-darwin: stub in backend for Darwin
  fsmonitor--daemon: implement 'run' command
  fsmonitor--daemon: implement 'start' command
  fsmonitor--daemon: add pathname classification
  fsmonitor--daemon: define token-ids
  fsmonitor--daemon: create token-based changed path cache
  compat/fsmonitor/fsm-listen-win32: implement FSMonitor backend on
    Windows
  compat/fsmonitor/fsm-listen-darwin: add MacOS header files for FSEvent
  compat/fsmonitor/fsm-listen-darwin: implement FSEvent listener on
    MacOS
  fsmonitor--daemon: implement handle_client callback
  help: include fsmonitor--daemon feature flag in version info
  t/helper/fsmonitor-client: create IPC client to talk to FSMonitor
    Daemon
  t7527: create test for fsmonitor--daemon
  t/perf: avoid copying builtin fsmonitor files into test repo
  t/helper/test-chmtime: skip directories on Windows
  t/perf/p7519: speed up test on Windows
  t/perf/p7519: add fsmonitor--daemon test cases
  fsmonitor--daemon: periodically truncate list of modified files
  fsmonitor--daemon: use a cookie file to sync with file system
  fsmonitor: force update index after large responses
  t7527: test status with untracked-cache and fsmonitor--daemon

 .gitignore                              |    1 +
 Documentation/config/core.txt           |   60 +-
 Documentation/git-fsmonitor--daemon.txt |   75 ++
 Documentation/git-update-index.txt      |    8 +-
 Makefile                                |   17 +
 builtin.h                               |    1 +
 builtin/fsmonitor--daemon.c             | 1480 +++++++++++++++++++++++
 builtin/update-index.c                  |    7 +-
 cache.h                                 |    1 -
 compat/fsmonitor/fsm-darwin-gcc.h       |   92 ++
 compat/fsmonitor/fsm-listen-darwin.c    |  427 +++++++
 compat/fsmonitor/fsm-listen-win32.c     |  586 +++++++++
 compat/fsmonitor/fsm-listen.h           |   49 +
 config.c                                |   14 -
 config.h                                |    1 -
 config.mak.uname                        |   20 +
 contrib/buildsystems/CMakeLists.txt     |   10 +
 environment.c                           |    1 -
 fsmonitor--daemon.h                     |  166 +++
 fsmonitor-ipc.c                         |  171 +++
 fsmonitor-ipc.h                         |   48 +
 fsmonitor-settings.c                    |  114 ++
 fsmonitor-settings.h                    |   21 +
 fsmonitor.c                             |  216 +++-
 fsmonitor.h                             |   15 +-
 git.c                                   |    1 +
 help.c                                  |    4 +
 repo-settings.c                         |    1 +
 repository.h                            |    3 +
 t/README                                |    4 +-
 t/helper/test-chmtime.c                 |   15 +
 t/helper/test-fsmonitor-client.c        |  116 ++
 t/helper/test-tool.c                    |    1 +
 t/helper/test-tool.h                    |    1 +
 t/perf/p7519-fsmonitor.sh               |   68 +-
 t/perf/perf-lib.sh                      |    2 +-
 t/t7527-builtin-fsmonitor.sh            |  594 +++++++++
 t/test-lib.sh                           |    7 +
 38 files changed, 4312 insertions(+), 106 deletions(-)
 create mode 100644 Documentation/git-fsmonitor--daemon.txt
 create mode 100644 builtin/fsmonitor--daemon.c
 create mode 100644 compat/fsmonitor/fsm-darwin-gcc.h
 create mode 100644 compat/fsmonitor/fsm-listen-darwin.c
 create mode 100644 compat/fsmonitor/fsm-listen-win32.c
 create mode 100644 compat/fsmonitor/fsm-listen.h
 create mode 100644 fsmonitor--daemon.h
 create mode 100644 fsmonitor-ipc.c
 create mode 100644 fsmonitor-ipc.h
 create mode 100644 fsmonitor-settings.c
 create mode 100644 fsmonitor-settings.h
 create mode 100644 t/helper/test-fsmonitor-client.c
 create mode 100755 t/t7527-builtin-fsmonitor.sh


base-commit: 715d08a9e51251ad8290b181b6ac3b9e1f9719d7
Published-As: https://github.com/gitgitgadget/git/releases/tag/pr-1041%2Fjeffhostetler%2Fbuiltin-fsmonitor-part2-v7
Fetch-It-Via: git fetch https://github.com/gitgitgadget/git pr-1041/jeffhostetler/builtin-fsmonitor-part2-v7
Pull-Request: https://github.com/gitgitgadget/git/pull/1041

Range-diff vs v6:

  1:  10a34082fcc =  1:  e98373f997f fsmonitor: enhance existing comments, clarify trivial response handling
  2:  e1c946af2c5 !  2:  ab68b944173 fsmonitor-ipc: create client routines for git-fsmonitor--daemon
     @@ fsmonitor-ipc.c (new)
      +	state = ipc_client_try_connect(fsmonitor_ipc__get_path(), &options,
      +				       &connection);
      +	if (state != IPC_STATE__LISTENING) {
     -+		die("fsmonitor--daemon is not running");
     ++		die(_("fsmonitor--daemon is not running"));
      +		return -1;
      +	}
      +
     @@ fsmonitor-ipc.c (new)
      +	ipc_client_close_connection(connection);
      +
      +	if (ret == -1) {
     -+		die("could not send '%s' command to fsmonitor--daemon", c);
     ++		die(_("could not send '%s' command to fsmonitor--daemon"), c);
      +		return -1;
      +	}
      +
  3:  ae622a517cf !  3:  e04c7301f24 fsmonitor: config settings are repository-specific
     @@ fsmonitor-settings.c (new)
      +		return;
      +
      +	CALLOC_ARRAY(s, 1);
     ++	s->mode = FSMONITOR_MODE_DISABLED;
      +
      +	r->settings.fsmonitor = s;
      +
     -+	fsm_settings__set_disabled(r);
     -+
      +	/*
      +	 * Overload the existing "core.fsmonitor" config setting (which
      +	 * has historically been either unset or a hook pathname) to
  4:  55974867da5 =  4:  ea02ba25d8f fsmonitor: use IPC to query the builtin FSMonitor daemon
  5:  ce42d5bbaf6 =  5:  6ab7db9cb76 fsmonitor: document builtin fsmonitor
  6:  9ce938c69b5 =  6:  0ce8ae3f2cf fsmonitor--daemon: add a built-in fsmonitor daemon
  7:  3bd7e6a051e =  7:  4624ce2fa47 fsmonitor--daemon: implement 'stop' and 'status' commands
  8:  92d3e54da7f =  8:  a29fe7266a4 compat/fsmonitor/fsm-listen-win32: stub in backend for Windows
  9:  f1219221270 =  9:  2f8a42fdb93 compat/fsmonitor/fsm-listen-darwin: stub in backend for Darwin
 10:  a3f94627cf6 ! 10:  f07800690ee fsmonitor--daemon: implement 'run' command
     @@ builtin/fsmonitor--daemon.c: static int do_as_client__status(void)
      +	 * common error case.
      +	 */
      +	if (fsmonitor_ipc__get_state() == IPC_STATE__LISTENING)
     -+		die("fsmonitor--daemon is already running '%s'",
     ++		die(_("fsmonitor--daemon is already running '%s'"),
      +		    the_repository->worktree);
      +
      +	if (fsmonitor__announce_startup) {
 11:  8de40b0fe8b ! 11:  a6a39a3306d fsmonitor--daemon: implement 'start' command
     @@ builtin/fsmonitor--daemon.c: static int try_to_run_foreground_daemon(void)
      +	 * immediately exited).
      +	 */
      +	if (fsmonitor_ipc__get_state() == IPC_STATE__LISTENING)
     -+		die("fsmonitor--daemon is already running '%s'",
     ++		die(_("fsmonitor--daemon is already running '%s'"),
      +		    the_repository->worktree);
      +
      +	if (fsmonitor__announce_startup) {
     @@ builtin/fsmonitor--daemon.c: static int try_to_run_foreground_daemon(void)
      +	default:
      +	case SBGR_ERROR:
      +	case SBGR_CB_ERROR:
     -+		return error("daemon failed to start");
     ++		return error(_("daemon failed to start"));
      +
      +	case SBGR_TIMEOUT:
     -+		return error("daemon not online yet");
     ++		return error(_("daemon not online yet"));
      +
      +	case SBGR_DIED:
     -+		return error("daemon terminated");
     ++		return error(_("daemon terminated"));
      +	}
      +}
      +
 12:  e5d419b1ea0 = 12:  d62e338d008 fsmonitor--daemon: add pathname classification
 13:  c2f3668a866 = 13:  53e06b4ae5d fsmonitor--daemon: define token-ids
 14:  bd492d13fb1 = 14:  39f43fabe02 fsmonitor--daemon: create token-based changed path cache
 15:  6b5b3d08227 ! 15:  239558e34ff compat/fsmonitor/fsm-listen-win32: implement FSMonitor backend on Windows
     @@ compat/fsmonitor/fsm-listen-win32.c
      +		if (len > 0)
      +			goto normalize;
      +		if (GetLastError() != ERROR_INSUFFICIENT_BUFFER) {
     -+			error("[GLE %ld] could not convert path to UTF-8: '%.*ls'",
     ++			error(_("[GLE %ld] could not convert path to UTF-8: '%.*ls'"),
      +			      GetLastError(),
      +			      (int)(info->FileNameLength / sizeof(WCHAR)),
      +			      info->FileName);
     @@ compat/fsmonitor/fsm-listen-win32.c
      +	if (watch->is_active)
      +		return 0;
      +
     -+	error("ReadDirectoryChangedW failed on '%s' [GLE %ld]",
     ++	error(_("ReadDirectoryChangedW failed on '%s' [GLE %ld]"),
      +	      watch->path.buf, GetLastError());
      +	return -1;
      +}
     @@ compat/fsmonitor/fsm-listen-win32.c
      +	 * sure it is worth it.
      +	 */
      +
     -+	error("GetOverlappedResult failed on '%s' [GLE %ld]",
     ++	error(_("GetOverlappedResult failed on '%s' [GLE %ld]"),
      +	      watch->path.buf, gle);
      +	return -1;
      +}
 16:  cdef9730b3f ! 16:  14b775e9d8b compat/fsmonitor/fsm-listen-darwin: add MacOS header files for FSEvent
     @@ Commit message
          Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
          Signed-off-by: Junio C Hamano <gitster@pobox.com>
      
     - ## compat/fsmonitor/fsm-listen-darwin.c ##
     + ## compat/fsmonitor/fsm-darwin-gcc.h (new) ##
      @@
     ++#ifndef FSM_DARWIN_GCC_H
     ++#define FSM_DARWIN_GCC_H
     ++
      +#ifndef __clang__
      +/*
      + * It is possible to #include CoreFoundation/CoreFoundation.h when compiling
     @@ compat/fsmonitor/fsm-listen-darwin.c
      +void FSEventStreamStop(FSEventStreamRef stream);
      +void FSEventStreamInvalidate(FSEventStreamRef stream);
      +void FSEventStreamRelease(FSEventStreamRef stream);
     ++
     ++#endif /* !clang */
     ++#endif /* FSM_DARWIN_GCC_H */
     +
     + ## compat/fsmonitor/fsm-listen-darwin.c ##
     +@@
     ++#ifndef __clang__
     ++#include "fsm-darwin-gcc.h"
      +#else
     -+/*
     -+ * Let Apple's headers declare `isalnum()` first, before
     -+ * Git's headers override it via a constant
     -+ */
     -+#include <string.h>
      +#include <CoreFoundation/CoreFoundation.h>
      +#include <CoreServices/CoreServices.h>
     ++
     ++#ifndef AVAILABLE_MAC_OS_X_VERSION_10_13_AND_LATER
     ++/*
     ++ * This enum value was added in 10.13 to:
     ++ *
     ++ * /Applications/Xcode.app/Contents/Developer/Platforms/ \
     ++ *    MacOSX.platform/Developer/SDKs/MacOSX.sdk/System/ \
     ++ *    Library/Frameworks/CoreServices.framework/Frameworks/ \
     ++ *    FSEvents.framework/Versions/Current/Headers/FSEvents.h
     ++ *
     ++ * If we're compiling against an older SDK, this symbol won't be
     ++ * present.  Silently define it here so that we don't have to ifdef
     ++ * the logging or masking below.  This should be harmless since older
     ++ * versions of macOS won't ever emit this FS event anyway.
     ++ */
     ++#define kFSEventStreamEventFlagItemCloned         0x00400000
     ++#endif
      +#endif
      +
       #include "cache.h"
 17:  aabfebd3a41 ! 17:  55bd7aee06c compat/fsmonitor/fsm-listen-darwin: implement FSEvent listener on MacOS
     @@ Commit message
          Signed-off-by: Junio C Hamano <gitster@pobox.com>
      
       ## compat/fsmonitor/fsm-listen-darwin.c ##
     -@@ compat/fsmonitor/fsm-listen-darwin.c: void FSEventStreamRelease(FSEventStreamRef stream);
     +@@
       #include "cache.h"
       #include "fsmonitor.h"
       #include "fsm-listen.h"
     @@ compat/fsmonitor/fsm-listen-darwin.c: void FSEventStreamRelease(FSEventStreamRef
      +	return 0;
      +
      +failed:
     -+	error("Unable to create FSEventStream.");
     ++	error(_("Unable to create FSEventStream."));
      +
      +	FREE_AND_NULL(state->backend_data);
       	return -1;
     @@ compat/fsmonitor/fsm-listen-darwin.c: void FSEventStreamRelease(FSEventStreamRef
      +	data->stream_scheduled = 1;
      +
      +	if (!FSEventStreamStart(data->stream)) {
     -+		error("Failed to start the FSEventStream");
     ++		error(_("Failed to start the FSEventStream"));
      +		goto force_error_stop_without_loop;
      +	}
      +	data->stream_started = 1;
 18:  0896e72e2b8 = 18:  1f4b5209bf6 fsmonitor--daemon: implement handle_client callback
 19:  bc240a9e665 ! 19:  8cf62c9fc6f help: include fsmonitor--daemon feature flag in version info
     @@ t/test-lib.sh: test_lazy_prereq SHA1 '
      +# Does this platform support `git fsmonitor--daemon`
      +#
      +test_lazy_prereq FSMONITOR_DAEMON '
     -+	git version --build-options | grep "feature:" | grep "fsmonitor--daemon"
     ++	git version --build-options >output &&
     ++	grep "feature: fsmonitor--daemon" output
      +'
 20:  2b563c240eb ! 20:  1bd74a81593 t/helper/fsmonitor-client: create IPC client to talk to FSMonitor Daemon
     @@ t/helper/test-fsmonitor-client.c (new)
      +
      +	ret = fsmonitor_ipc__send_query(token, &answer);
      +	if (ret < 0)
     -+		die(_("could not query fsmonitor--daemon"));
     ++		die("could not query fsmonitor--daemon");
      +
      +	write_in_full(1, answer.buf, answer.len);
      +	strbuf_release(&answer);
     @@ t/helper/test-fsmonitor-client.c (new)
      +	const char *token = NULL;
      +
      +	const char * const fsmonitor_client_usage[] = {
     -+		N_("test-helper fsmonitor-client query [<token>]"),
     -+		N_("test-helper fsmonitor-client flush"),
     ++		"test-tool fsmonitor-client query [<token>]",
     ++		"test-tool fsmonitor-client flush",
      +		NULL,
      +	};
      +
      +	struct option options[] = {
     -+		OPT_STRING(0, "token", &token, N_("token"),
     -+			   N_("command token to send to the server")),
     ++		OPT_STRING(0, "token", &token, "token",
     ++			   "command token to send to the server"),
      +		OPT_END()
      +	};
      +
     -+	if (argc < 2)
     -+		usage_with_options(fsmonitor_client_usage, options);
     ++	argc = parse_options(argc, argv, NULL, options, fsmonitor_client_usage, 0);
      +
     -+	if (argc == 2 && !strcmp(argv[1], "-h"))
     ++	if (argc != 1)
      +		usage_with_options(fsmonitor_client_usage, options);
      +
     -+	subcmd = argv[1];
     -+	argv--;
     -+	argc++;
     -+
     -+	argc = parse_options(argc, argv, NULL, options, fsmonitor_client_usage, 0);
     ++	subcmd = argv[0];
      +
      +	setup_git_directory();
      +
 21:  db8809da089 ! 21:  4a920d0b54a t7527: create test for fsmonitor--daemon
     @@ t/t7527-builtin-fsmonitor.sh (new)
      +fi
      +
      +stop_daemon_delete_repo () {
     -+	r=$1
     -+	git -C $r fsmonitor--daemon stop >/dev/null 2>/dev/null
     ++	r=$1 &&
     ++	test_might_fail git -C $r fsmonitor--daemon stop &&
      +	rm -rf $1
     -+	return 0
      +}
      +
      +start_daemon () {
     -+	case "$#" in
     -+		1) r="-C $1";;
     -+		*) r="";
     -+	esac
     ++	r= &&
     ++	tf= &&
     ++	t2= &&
     ++	tk= &&
     ++
     ++	while test "$#" -ne 0
     ++	do
     ++		case "$1" in
     ++		-C)
     ++			shift;
     ++			test "$#" -ne 0 || BUG "error: -C requires arg"
     ++			r="-C $1"
     ++			shift
     ++			;;
     ++		-tf)
     ++			shift;
     ++			test "$#" -ne 0 || BUG "error: -tf requires arg"
     ++			tf="$1"
     ++			shift
     ++			;;
     ++		-t2)
     ++			shift;
     ++			test "$#" -ne 0 || BUG "error: -t2 requires arg"
     ++			t2="$1"
     ++			shift
     ++			;;
     ++		-tk)
     ++			shift;
     ++			test "$#" -ne 0 || BUG "error: -tk requires arg"
     ++			tk="$1"
     ++			shift
     ++			;;
     ++		*)
     ++			BUG "error: unknown option: '$1'"
     ++			;;
     ++		esac
     ++	done &&
      +
     -+	git $r fsmonitor--daemon start || return $?
     -+	git $r fsmonitor--daemon status || return $?
     -+
     -+	return 0
     ++	(
     ++		if test -n "$tf"
     ++		then
     ++			GIT_TRACE_FSMONITOR="$tf"
     ++			export GIT_TRACE_FSMONITOR
     ++		fi &&
     ++
     ++		if test -n "$t2"
     ++		then
     ++			GIT_TRACE2_PERF="$t2"
     ++			export GIT_TRACE2_PERF
     ++		fi &&
     ++
     ++		if test -n "$tk"
     ++		then
     ++			GIT_TEST_FSMONITOR_TOKEN="$tk"
     ++			export GIT_TEST_FSMONITOR_TOKEN
     ++		fi &&
     ++
     ++		git $r fsmonitor--daemon start &&
     ++		git $r fsmonitor--daemon status
     ++	)
      +}
      +
      +# Is a Trace2 data event present with the given catetory and key?
      +# We do not care what the value is.
      +#
      +have_t2_data_event () {
     -+	c=$1
     -+	k=$2
     ++	c=$1 &&
     ++	k=$2 &&
      +
      +	grep -e '"event":"data".*"category":"'"$c"'".*"key":"'"$k"'"'
      +}
     @@ t/t7527-builtin-fsmonitor.sh (new)
      +	test_when_finished "stop_daemon_delete_repo test_explicit" &&
      +
      +	git init test_explicit &&
     -+	start_daemon test_explicit &&
     ++	start_daemon -C test_explicit &&
      +
      +	git -C test_explicit fsmonitor--daemon stop &&
      +	test_must_fail git -C test_explicit fsmonitor--daemon status
     @@ t/t7527-builtin-fsmonitor.sh (new)
      +	# but this test case is only concerned with whether the daemon was
      +	# implicitly started.)
      +
     -+	GIT_TRACE2_EVENT="$(pwd)/.git/trace" \
     ++	GIT_TRACE2_EVENT="$PWD/.git/trace" \
      +		test-tool -C test_implicit fsmonitor-client query --token 0 >actual &&
      +	nul_to_q <actual >actual.filtered &&
      +	grep "builtin:" actual.filtered &&
     @@ t/t7527-builtin-fsmonitor.sh (new)
      +
      +	git init test_implicit_1 &&
      +
     -+	start_daemon test_implicit_1 &&
     ++	start_daemon -C test_implicit_1 &&
      +
      +	# deleting the .git directory will implicitly stop the daemon.
      +	rm -rf test_implicit_1/.git &&
     @@ t/t7527-builtin-fsmonitor.sh (new)
      +
      +	git init test_implicit_2 &&
      +
     -+	start_daemon test_implicit_2 &&
     ++	start_daemon -C test_implicit_2 &&
      +
      +	# renaming the .git directory will implicitly stop the daemon.
      +	mv test_implicit_2/.git test_implicit_2/.xxx &&
     @@ t/t7527-builtin-fsmonitor.sh (new)
      +
      +	git init test_multiple &&
      +
     -+	start_daemon test_multiple &&
     ++	start_daemon -C test_multiple &&
      +
      +	test_must_fail git -C test_multiple fsmonitor--daemon start 2>actual &&
      +	grep "fsmonitor--daemon is already running" actual &&
     @@ t/t7527-builtin-fsmonitor.sh (new)
      +# This is here in case something else fails first.
      +#
      +redundant_stop_daemon () {
     -+	git fsmonitor--daemon stop
     -+	return 0
     ++	test_might_fail git fsmonitor--daemon stop
      +}
      +
      +test_expect_success 'update-index implicitly starts daemon' '
     @@ t/t7527-builtin-fsmonitor.sh (new)
      +
      +	test_must_fail git fsmonitor--daemon status &&
      +
     -+	GIT_TRACE2_EVENT="$(pwd)/.git/trace_implicit_1" \
     ++	GIT_TRACE2_EVENT="$PWD/.git/trace_implicit_1" \
      +		git update-index --fsmonitor &&
      +
      +	git fsmonitor--daemon status &&
     @@ t/t7527-builtin-fsmonitor.sh (new)
      +
      +	test_must_fail git fsmonitor--daemon status &&
      +
     -+	GIT_TRACE2_EVENT="$(pwd)/.git/trace_implicit_2" \
     ++	GIT_TRACE2_EVENT="$PWD/.git/trace_implicit_2" \
      +		git status >actual &&
      +
      +	git fsmonitor--daemon status &&
     @@ t/t7527-builtin-fsmonitor.sh (new)
      +'
      +
      +edit_files () {
     -+	echo 1 >modified
     -+	echo 2 >dir1/modified
     -+	echo 3 >dir2/modified
     ++	echo 1 >modified &&
     ++	echo 2 >dir1/modified &&
     ++	echo 3 >dir2/modified &&
      +	>dir1/untracked
      +}
      +
      +delete_files () {
     -+	rm -f delete
     -+	rm -f dir1/delete
     ++	rm -f delete &&
     ++	rm -f dir1/delete &&
      +	rm -f dir2/delete
      +}
      +
      +create_files () {
     -+	echo 1 >new
     -+	echo 2 >dir1/new
     ++	echo 1 >new &&
     ++	echo 2 >dir1/new &&
      +	echo 3 >dir2/new
      +}
      +
      +rename_files () {
     -+	mv rename renamed
     -+	mv dir1/rename dir1/renamed
     ++	mv rename renamed &&
     ++	mv dir1/rename dir1/renamed &&
      +	mv dir2/rename dir2/renamed
      +}
      +
      +file_to_directory () {
     -+	rm -f delete
     -+	mkdir delete
     ++	rm -f delete &&
     ++	mkdir delete &&
      +	echo 1 >delete/new
      +}
      +
      +directory_to_file () {
     -+	rm -rf dir1
     ++	rm -rf dir1 &&
      +	echo 1 >dir1
      +}
      +
     -+verify_status () {
     -+	git status >actual &&
     -+	GIT_INDEX_FILE=.git/fresh-index git read-tree master &&
     -+	GIT_INDEX_FILE=.git/fresh-index git -c core.fsmonitor=false status >expect &&
     -+	test_cmp expect actual &&
     -+	echo HELLO AFTER &&
     -+	cat .git/trace &&
     -+	echo HELLO AFTER
     -+}
     -+
      +# The next few test cases confirm that our fsmonitor daemon sees each type
      +# of OS filesystem notification that we care about.  At this layer we just
      +# ensure we are getting the OS notifications and do not try to confirm what
     @@ t/t7527-builtin-fsmonitor.sh (new)
      +# daemon) because these commands might implicitly restart the daemon.
      +
      +clean_up_repo_and_stop_daemon () {
     -+	git reset --hard HEAD
     -+	git clean -fd
     -+	git fsmonitor--daemon stop
     ++	git reset --hard HEAD &&
     ++	git clean -fd &&
     ++	test_might_fail git fsmonitor--daemon stop &&
      +	rm -f .git/trace
      +}
      +
      +test_expect_success 'edit some files' '
      +	test_when_finished clean_up_repo_and_stop_daemon &&
      +
     -+	(
     -+		GIT_TRACE_FSMONITOR="$(pwd)/.git/trace" &&
     -+		export GIT_TRACE_FSMONITOR &&
     -+
     -+		start_daemon
     -+	) &&
     ++	start_daemon -tf "$PWD/.git/trace" &&
      +
      +	edit_files &&
      +
     -+	test-tool fsmonitor-client query --token 0 >/dev/null 2>&1 &&
     ++	test-tool fsmonitor-client query --token 0 &&
      +
      +	grep "^event: dir1/modified$"  .git/trace &&
      +	grep "^event: dir2/modified$"  .git/trace &&
     @@ t/t7527-builtin-fsmonitor.sh (new)
      +test_expect_success 'create some files' '
      +	test_when_finished clean_up_repo_and_stop_daemon &&
      +
     -+	(
     -+		GIT_TRACE_FSMONITOR="$(pwd)/.git/trace" &&
     -+		export GIT_TRACE_FSMONITOR &&
     -+
     -+		start_daemon
     -+	) &&
     ++	start_daemon -tf "$PWD/.git/trace" &&
      +
      +	create_files &&
      +
     -+	test-tool fsmonitor-client query --token 0 >/dev/null 2>&1 &&
     ++	test-tool fsmonitor-client query --token 0 &&
      +
      +	grep "^event: dir1/new$" .git/trace &&
      +	grep "^event: dir2/new$" .git/trace &&
     @@ t/t7527-builtin-fsmonitor.sh (new)
      +test_expect_success 'delete some files' '
      +	test_when_finished clean_up_repo_and_stop_daemon &&
      +
     -+	(
     -+		GIT_TRACE_FSMONITOR="$(pwd)/.git/trace" &&
     -+		export GIT_TRACE_FSMONITOR &&
     -+
     -+		start_daemon
     -+	) &&
     ++	start_daemon -tf "$PWD/.git/trace" &&
      +
      +	delete_files &&
      +
     -+	test-tool fsmonitor-client query --token 0 >/dev/null 2>&1 &&
     ++	test-tool fsmonitor-client query --token 0 &&
      +
      +	grep "^event: dir1/delete$" .git/trace &&
      +	grep "^event: dir2/delete$" .git/trace &&
     @@ t/t7527-builtin-fsmonitor.sh (new)
      +test_expect_success 'rename some files' '
      +	test_when_finished clean_up_repo_and_stop_daemon &&
      +
     -+	(
     -+		GIT_TRACE_FSMONITOR="$(pwd)/.git/trace" &&
     -+		export GIT_TRACE_FSMONITOR &&
     -+
     -+		start_daemon
     -+	) &&
     ++	start_daemon -tf "$PWD/.git/trace" &&
      +
      +	rename_files &&
      +
     -+	test-tool fsmonitor-client query --token 0 >/dev/null 2>&1 &&
     ++	test-tool fsmonitor-client query --token 0 &&
      +
      +	grep "^event: dir1/rename$"  .git/trace &&
      +	grep "^event: dir2/rename$"  .git/trace &&
     @@ t/t7527-builtin-fsmonitor.sh (new)
      +test_expect_success 'rename directory' '
      +	test_when_finished clean_up_repo_and_stop_daemon &&
      +
     -+	(
     -+		GIT_TRACE_FSMONITOR="$(pwd)/.git/trace" &&
     -+		export GIT_TRACE_FSMONITOR &&
     -+
     -+		start_daemon
     -+	) &&
     ++	start_daemon -tf "$PWD/.git/trace" &&
      +
      +	mv dirtorename dirrenamed &&
      +
     -+	test-tool fsmonitor-client query --token 0 >/dev/null 2>&1 &&
     ++	test-tool fsmonitor-client query --token 0 &&
      +
      +	grep "^event: dirtorename/*$" .git/trace &&
      +	grep "^event: dirrenamed/*$"  .git/trace
     @@ t/t7527-builtin-fsmonitor.sh (new)
      +test_expect_success 'file changes to directory' '
      +	test_when_finished clean_up_repo_and_stop_daemon &&
      +
     -+	(
     -+		GIT_TRACE_FSMONITOR="$(pwd)/.git/trace" &&
     -+		export GIT_TRACE_FSMONITOR &&
     -+
     -+		start_daemon
     -+	) &&
     ++	start_daemon -tf "$PWD/.git/trace" &&
      +
      +	file_to_directory &&
      +
     -+	test-tool fsmonitor-client query --token 0 >/dev/null 2>&1 &&
     ++	test-tool fsmonitor-client query --token 0 &&
      +
      +	grep "^event: delete$"     .git/trace &&
      +	grep "^event: delete/new$" .git/trace
     @@ t/t7527-builtin-fsmonitor.sh (new)
      +test_expect_success 'directory changes to a file' '
      +	test_when_finished clean_up_repo_and_stop_daemon &&
      +
     -+	(
     -+		GIT_TRACE_FSMONITOR="$(pwd)/.git/trace" &&
     -+		export GIT_TRACE_FSMONITOR &&
     -+
     -+		start_daemon
     -+	) &&
     ++	start_daemon -tf "$PWD/.git/trace" &&
      +
      +	directory_to_file &&
      +
     -+	test-tool fsmonitor-client query --token 0 >/dev/null 2>&1 &&
     ++	test-tool fsmonitor-client query --token 0 &&
      +
      +	grep "^event: dir1$" .git/trace
      +'
     @@ t/t7527-builtin-fsmonitor.sh (new)
      +
      +	git init test_flush &&
      +
     -+	(
     -+		GIT_TEST_FSMONITOR_TOKEN=true &&
     -+		export GIT_TEST_FSMONITOR_TOKEN &&
     -+
     -+		GIT_TRACE_FSMONITOR="$(pwd)/.git/trace_daemon" &&
     -+		export GIT_TRACE_FSMONITOR &&
     -+
     -+		start_daemon test_flush
     -+	) &&
     ++	start_daemon -C test_flush -tf "$PWD/.git/trace_daemon" -tk true &&
      +
      +	# The daemon should have an initial token with no events in _0 and
      +	# then a few (probably platform-specific number of) events in _1.
     @@ t/t7527-builtin-fsmonitor.sh (new)
      +	test-tool -C test_flush fsmonitor-client query --token "builtin:test_00000001:0" >actual_0 &&
      +	nul_to_q <actual_0 >actual_q0 &&
      +
     -+	touch test_flush/file_1 &&
     -+	touch test_flush/file_2 &&
     ++	>test_flush/file_1 &&
     ++	>test_flush/file_2 &&
      +
      +	test-tool -C test_flush fsmonitor-client query --token "builtin:test_00000001:0" >actual_1 &&
      +	nul_to_q <actual_1 >actual_q1 &&
     @@ t/t7527-builtin-fsmonitor.sh (new)
      +
      +	grep "^builtin:test_00000002:0Q$" actual_q2 &&
      +
     -+	touch test_flush/file_3 &&
     ++	>test_flush/file_3 &&
      +
      +	test-tool -C test_flush fsmonitor-client query --token "builtin:test_00000002:0" >actual_3 &&
      +	nul_to_q <actual_3 >actual_q3 &&
     @@ t/t7527-builtin-fsmonitor.sh (new)
      +test_expect_success 'worktree with .git file' '
      +	git -C wt-base worktree add ../wt-secondary &&
      +
     -+	(
     -+		GIT_TRACE2_PERF="$(pwd)/trace2_wt_secondary" &&
     -+		export GIT_TRACE2_PERF &&
     -+
     -+		GIT_TRACE_FSMONITOR="$(pwd)/trace_wt_secondary" &&
     -+		export GIT_TRACE_FSMONITOR &&
     -+
     -+		start_daemon wt-secondary
     -+	) &&
     ++	start_daemon -C wt-secondary \
     ++		-tf "$PWD/trace_wt_secondary" \
     ++		-t2 "$PWD/trace2_wt_secondary" &&
      +
      +	git -C wt-secondary fsmonitor--daemon stop &&
      +	test_must_fail git -C wt-secondary fsmonitor--daemon status
 22:  15bd5aaff36 = 22:  c925a9a7459 t/perf: avoid copying builtin fsmonitor files into test repo
 23:  4cb97918d20 = 23:  5b3381c223e t/helper/test-chmtime: skip directories on Windows
 24:  a70748b4640 ! 24:  803a540cc00 t/perf/p7519: speed up test on Windows
     @@ t/perf/p7519-fsmonitor.sh: test_fsmonitor_suite() {
       			head -100000 | \
       			grep -v \" | \
      -			sed '\''s/\(.\)/\\\1/g'\'' | \
     -+			egrep -v " ." | \
     ++			grep -v " ." | \
       			xargs test-tool chmtime -300 &&
       		git status
       	'
 25:  b0c9c9c7a40 ! 25:  d5ca2df31c8 t/perf/p7519: add fsmonitor--daemon test cases
     @@ Commit message
          Signed-off-by: Junio C Hamano <gitster@pobox.com>
      
       ## t/perf/p7519-fsmonitor.sh ##
     +@@ t/perf/p7519-fsmonitor.sh: test_expect_success "one time repo setup" '
     + 	fi
     + '
     + 
     +-setup_for_fsmonitor () {
     ++setup_for_fsmonitor_hook () {
     + 	# set INTEGRATION_SCRIPT depending on the environment
     + 	if test -n "$INTEGRATION_PATH"
     + 	then
      @@ t/perf/p7519-fsmonitor.sh: test_perf_w_drop_caches () {
       }
       
     @@ t/perf/p7519-fsmonitor.sh: test_perf_w_drop_caches () {
      +	if test -n "$USE_FSMONITOR_DAEMON"
      +	then
      +		DESC="builtin fsmonitor--daemon"
     -+	elif test -n "$INTEGRATION_SCRIPT"; then
     ++	elif test -n "$INTEGRATION_SCRIPT"
     ++	then
       		DESC="fsmonitor=$(basename $INTEGRATION_SCRIPT)"
       	else
       		DESC="fsmonitor=disabled"
     +@@ t/perf/p7519-fsmonitor.sh: test_fsmonitor_suite () {
     + trace_start fsmonitor-watchman
     + if test -n "$GIT_PERF_7519_FSMONITOR"; then
     + 	for INTEGRATION_PATH in $GIT_PERF_7519_FSMONITOR; do
     +-		test_expect_success "setup for fsmonitor $INTEGRATION_PATH" 'setup_for_fsmonitor'
     ++		test_expect_success "setup for fsmonitor $INTEGRATION_PATH" 'setup_for_fsmonitor_hook'
     + 		test_fsmonitor_suite
     + 	done
     + else
     +-	test_expect_success "setup for fsmonitor" 'setup_for_fsmonitor'
     ++	test_expect_success "setup for fsmonitor hook" 'setup_for_fsmonitor_hook'
     + 	test_fsmonitor_suite
     + fi
     + 
      @@ t/perf/p7519-fsmonitor.sh: test_expect_success "setup without fsmonitor" '
       test_fsmonitor_suite
       trace_stop
     @@ t/perf/p7519-fsmonitor.sh: test_expect_success "setup without fsmonitor" '
      +then
      +	USE_FSMONITOR_DAEMON=t
      +
     -+	trace_start fsmonitor--daemon--server
     -+	git fsmonitor--daemon start
     ++	test_expect_success "setup for builtin fsmonitor" '
     ++		trace_start fsmonitor--daemon--server &&
     ++		git fsmonitor--daemon start &&
      +
     -+	trace_start fsmonitor--daemon--client
     ++		trace_start fsmonitor--daemon--client &&
      +
     -+	git config core.fsmonitor true
     -+	git update-index --fsmonitor
     ++		git config core.fsmonitor true &&
     ++		git update-index --fsmonitor
     ++	'
      +
      +	test_fsmonitor_suite
      +
 26:  64a5b741670 = 26:  42631259e89 fsmonitor--daemon: periodically truncate list of modified files
 27:  5b06eb5d0e6 ! 27:  f256c3cbe8b fsmonitor--daemon: use a cookie file to sync with file system
     @@ builtin/fsmonitor--daemon.c: static int do_as_client__status(void)
       
      +enum fsmonitor_cookie_item_result {
      +	FCIR_ERROR = -1, /* could not create cookie file ? */
     -+	FCIR_INIT = 0,
     ++	FCIR_INIT,
      +	FCIR_SEEN,
      +	FCIR_ABORT,
      +};
      +
      +struct fsmonitor_cookie_item {
      +	struct hashmap_entry entry;
     -+	const char *name;
     ++	char *name;
      +	enum fsmonitor_cookie_item_result result;
      +};
      +
     @@ builtin/fsmonitor--daemon.c: static int do_as_client__status(void)
      +	 * that the listener thread has seen it.
      +	 */
      +	fd = open(cookie_pathname.buf, O_WRONLY | O_CREAT | O_EXCL, 0600);
     -+	if (fd >= 0) {
     -+		close(fd);
     -+		unlink(cookie_pathname.buf);
     -+
     -+		/*
     -+		 * Technically, this is an infinite wait (well, unless another
     -+		 * thread sends us an abort).  I'd like to change this to
     -+		 * use `pthread_cond_timedwait()` and return an error/timeout
     -+		 * and let the caller do the trivial response thing, but we
     -+		 * don't have that routine in our thread-utils.
     -+		 *
     -+		 * After extensive beta testing I'm not really worried about
     -+		 * this.  Also note that the above open() and unlink() calls
     -+		 * will cause at least two FS events on that path, so the odds
     -+		 * of getting stuck are pretty slim.
     -+		 */
     -+		while (cookie->result == FCIR_INIT)
     -+			pthread_cond_wait(&state->cookies_cond,
     -+					  &state->main_lock);
     -+	} else {
     ++	if (fd < 0) {
      +		error_errno(_("could not create fsmonitor cookie '%s'"),
      +			    cookie->name);
      +
      +		cookie->result = FCIR_ERROR;
     ++		goto done;
      +	}
      +
     ++	/*
     ++	 * Technically, close() and unlink() can fail, but we don't
     ++	 * care here.  We only created the file to trigger a watch
     ++	 * event from the FS to know that when we're up to date.
     ++	 */
     ++	close(fd);
     ++	unlink(cookie_pathname.buf);
     ++
     ++	/*
     ++	 * Technically, this is an infinite wait (well, unless another
     ++	 * thread sends us an abort).  I'd like to change this to
     ++	 * use `pthread_cond_timedwait()` and return an error/timeout
     ++	 * and let the caller do the trivial response thing, but we
     ++	 * don't have that routine in our thread-utils.
     ++	 *
     ++	 * After extensive beta testing I'm not really worried about
     ++	 * this.  Also note that the above open() and unlink() calls
     ++	 * will cause at least two FS events on that path, so the odds
     ++	 * of getting stuck are pretty slim.
     ++	 */
     ++	while (cookie->result == FCIR_INIT)
     ++		pthread_cond_wait(&state->cookies_cond,
     ++				  &state->main_lock);
     ++
     ++done:
      +	hashmap_remove(&state->cookies, &cookie->entry, NULL);
      +
      +	result = cookie->result;
      +
     -+	free((char*)cookie->name);
     ++	free(cookie->name);
      +	free(cookie);
      +	strbuf_release(&cookie_pathname);
      +
 28:  1fd5439de03 = 28:  08af8296f96 fsmonitor: force update index after large responses
 29:  6fc1430285f ! 29:  e6cf84dc8eb t7527: test status with untracked-cache and fsmonitor--daemon
     @@ t/t7527-builtin-fsmonitor.sh: test_expect_success 'cleanup worktrees' '
      +}
      +
      +matrix_try () {
     -+	uc=$1
     -+	fsm=$2
     -+	fn=$3
     ++	uc=$1 &&
     ++	fsm=$2 &&
     ++	fn=$3 &&
      +
      +	test_expect_success "Matrix[uc:$uc][fsm:$fsm] $fn" '
      +		matrix_clean_up_repo &&
 30:  b915b95cc2f <  -:  ----------- update-index: convert fsmonitor warnings to advise

-- 
gitgitgadget

[PATCH v7 01/29] fsmonitor: enhance existing comments, clarify trivial response handling

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 fsmonitor.c | 64 ++++++++++++++++++++++++++++++++++-------------------
 1 file changed, 41 insertions(+), 23 deletions(-)

diff --git a/fsmonitor.c b/fsmonitor.c
index ab9bfc60b34..448d0ee33f5 100644
--- a/fsmonitor.c
+++ b/fsmonitor.c
@@ -168,29 +168,15 @@ static int query_fsmonitor(int version, const char *last_update, struct strbuf *
 
 	if (result)
 		trace2_data_intmax("fsm_hook", NULL, "query/failed", result);
-	else {
+	else
 		trace2_data_intmax("fsm_hook", NULL, "query/response-length",
 				   query_result->len);
 
-		if (fsmonitor_is_trivial_response(query_result))
-			trace2_data_intmax("fsm_hook", NULL,
-					   "query/trivial-response", 1);
-	}
-
 	trace2_region_leave("fsm_hook", "query", NULL);
 
 	return result;
 }
 
-int fsmonitor_is_trivial_response(const struct strbuf *query_result)
-{
-	static char trivial_response[3] = { '\0', '/', '\0' };
-
-	return query_result->len >= 3 &&
-		!memcmp(trivial_response,
-			&query_result->buf[query_result->len - 3], 3);
-}
-
 static void fsmonitor_refresh_callback(struct index_state *istate, char *name)
 {
 	int i, len = strlen(name);
@@ -238,6 +224,7 @@ void refresh_fsmonitor(struct index_state *istate)
 	struct strbuf last_update_token = STRBUF_INIT;
 	char *buf;
 	unsigned int i;
+	int is_trivial = 0;
 
 	if (!core_fsmonitor || istate->fsmonitor_has_run_once)
 		return;
@@ -283,6 +270,7 @@ void refresh_fsmonitor(struct index_state *istate)
 					query_success = 0;
 				} else {
 					bol = last_update_token.len + 1;
+					is_trivial = query_result.buf[bol] == '/';
 				}
 			} else if (hook_version < 0) {
 				hook_version = HOOK_INTERFACE_VERSION1;
@@ -294,16 +282,38 @@ void refresh_fsmonitor(struct index_state *istate)
 		if (hook_version == HOOK_INTERFACE_VERSION1) {
 			query_success = !query_fsmonitor(HOOK_INTERFACE_VERSION1,
 				istate->fsmonitor_last_update, &query_result);
+			if (query_success)
+				is_trivial = query_result.buf[0] == '/';
 		}
 
+		if (is_trivial)
+			trace2_data_intmax("fsm_hook", NULL,
+					   "query/trivial-response", 1);
+
 		trace_performance_since(last_update, "fsmonitor process '%s'", core_fsmonitor);
 		trace_printf_key(&trace_fsmonitor, "fsmonitor process '%s' returned %s",
 			core_fsmonitor, query_success ? "success" : "failure");
 	}
 
-	/* a fsmonitor process can return '/' to indicate all entries are invalid */
-	if (query_success && query_result.buf[bol] != '/') {
-		/* Mark all entries returned by the monitor as dirty */
+	/*
+	 * The response from FSMonitor (excluding the header token) is
+	 * either:
+	 *
+	 * [a] a (possibly empty) list of NUL delimited relative
+	 *     pathnames of changed paths.  This list can contain
+	 *     files and directories.  Directories have a trailing
+	 *     slash.
+	 *
+	 * [b] a single '/' to indicate the provider had no
+	 *     information and that we should consider everything
+	 *     invalid.  We call this a trivial response.
+	 */
+	if (query_success && !is_trivial) {
+		/*
+		 * Mark all pathnames returned by the monitor as dirty.
+		 *
+		 * This updates both the cache-entries and the untracked-cache.
+		 */
 		buf = query_result.buf;
 		for (i = bol; i < query_result.len; i++) {
 			if (buf[i] != '\0')
@@ -318,11 +328,16 @@ void refresh_fsmonitor(struct index_state *istate)
 		if (istate->untracked)
 			istate->untracked->use_fsmonitor = 1;
 	} else {
-
-		/* We only want to run the post index changed hook if we've actually changed entries, so keep track
-		 * if we actually changed entries or not */
+		/*
+		 * We failed to get a response or received a trivial response,
+		 * so invalidate everything.
+		 *
+		 * We only want to run the post index changed hook if
+		 * we've actually changed entries, so keep track if we
+		 * actually changed entries or not.
+		 */
 		int is_cache_changed = 0;
-		/* Mark all entries invalid */
+
 		for (i = 0; i < istate->cache_nr; i++) {
 			if (istate->cache[i]->ce_flags & CE_FSMONITOR_VALID) {
 				is_cache_changed = 1;
@@ -330,7 +345,10 @@ void refresh_fsmonitor(struct index_state *istate)
 			}
 		}
 
-		/* If we're going to check every file, ensure we save the results */
+		/*
+		 * If we're going to check every file, ensure we save
+		 * the results.
+		 */
 		if (is_cache_changed)
 			istate->cache_changed |= FSMONITOR_CHANGED;
 
-- 
gitgitgadget

[PATCH v7 02/29] fsmonitor-ipc: create client routines for git-fsmonitor--daemon

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Create fsmonitor_ipc__*() client routines to spawn the built-in file
system monitor daemon and send it an IPC request using the `Simple
IPC` API.

Stub in empty fsmonitor_ipc__*() functions for unsupported platforms.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 Makefile        |   1 +
 fsmonitor-ipc.c | 171 ++++++++++++++++++++++++++++++++++++++++++++++++
 fsmonitor-ipc.h |  48 ++++++++++++++
 3 files changed, 220 insertions(+)
 create mode 100644 fsmonitor-ipc.c
 create mode 100644 fsmonitor-ipc.h

diff --git a/Makefile b/Makefile
index 6f0b4b775fe..a19d850e716 100644
--- a/Makefile
+++ b/Makefile
@@ -907,6 +907,7 @@ LIB_OBJS += fetch-pack.o
 LIB_OBJS += fmt-merge-msg.o
 LIB_OBJS += fsck.o
 LIB_OBJS += fsmonitor.o
+LIB_OBJS += fsmonitor-ipc.o
 LIB_OBJS += gettext.o
 LIB_OBJS += gpg-interface.o
 LIB_OBJS += graph.o
diff --git a/fsmonitor-ipc.c b/fsmonitor-ipc.c
new file mode 100644
index 00000000000..789e7397baa
--- /dev/null
+++ b/fsmonitor-ipc.c
@@ -0,0 +1,171 @@
+#include "cache.h"
+#include "fsmonitor.h"
+#include "simple-ipc.h"
+#include "fsmonitor-ipc.h"
+#include "run-command.h"
+#include "strbuf.h"
+#include "trace2.h"
+
+#ifndef HAVE_FSMONITOR_DAEMON_BACKEND
+
+/*
+ * A trivial implementation of the fsmonitor_ipc__ API for unsupported
+ * platforms.
+ */
+
+int fsmonitor_ipc__is_supported(void)
+{
+	return 0;
+}
+
+const char *fsmonitor_ipc__get_path(void)
+{
+	return NULL;
+}
+
+enum ipc_active_state fsmonitor_ipc__get_state(void)
+{
+	return IPC_STATE__OTHER_ERROR;
+}
+
+int fsmonitor_ipc__send_query(const char *since_token,
+			      struct strbuf *answer)
+{
+	return -1;
+}
+
+int fsmonitor_ipc__send_command(const char *command,
+				struct strbuf *answer)
+{
+	return -1;
+}
+
+#else
+
+int fsmonitor_ipc__is_supported(void)
+{
+	return 1;
+}
+
+GIT_PATH_FUNC(fsmonitor_ipc__get_path, "fsmonitor--daemon.ipc")
+
+enum ipc_active_state fsmonitor_ipc__get_state(void)
+{
+	return ipc_get_active_state(fsmonitor_ipc__get_path());
+}
+
+static int spawn_daemon(void)
+{
+	const char *args[] = { "fsmonitor--daemon", "start", NULL };
+
+	return run_command_v_opt_tr2(args, RUN_COMMAND_NO_STDIN | RUN_GIT_CMD,
+				    "fsmonitor");
+}
+
+int fsmonitor_ipc__send_query(const char *since_token,
+			      struct strbuf *answer)
+{
+	int ret = -1;
+	int tried_to_spawn = 0;
+	enum ipc_active_state state = IPC_STATE__OTHER_ERROR;
+	struct ipc_client_connection *connection = NULL;
+	struct ipc_client_connect_options options
+		= IPC_CLIENT_CONNECT_OPTIONS_INIT;
+	const char *tok = since_token ? since_token : "";
+	size_t tok_len = since_token ? strlen(since_token) : 0;
+
+	options.wait_if_busy = 1;
+	options.wait_if_not_found = 0;
+
+	trace2_region_enter("fsm_client", "query", NULL);
+	trace2_data_string("fsm_client", NULL, "query/command", tok);
+
+try_again:
+	state = ipc_client_try_connect(fsmonitor_ipc__get_path(), &options,
+				       &connection);
+
+	switch (state) {
+	case IPC_STATE__LISTENING:
+		ret = ipc_client_send_command_to_connection(
+			connection, tok, tok_len, answer);
+		ipc_client_close_connection(connection);
+
+		trace2_data_intmax("fsm_client", NULL,
+				   "query/response-length", answer->len);
+		goto done;
+
+	case IPC_STATE__NOT_LISTENING:
+	case IPC_STATE__PATH_NOT_FOUND:
+		if (tried_to_spawn)
+			goto done;
+
+		tried_to_spawn++;
+		if (spawn_daemon())
+			goto done;
+
+		/*
+		 * Try again, but this time give the daemon a chance to
+		 * actually create the pipe/socket.
+		 *
+		 * Granted, the daemon just started so it can't possibly have
+		 * any FS cached yet, so we'll always get a trivial answer.
+		 * BUT the answer should include a new token that can serve
+		 * as the basis for subsequent requests.
+		 */
+		options.wait_if_not_found = 1;
+		goto try_again;
+
+	case IPC_STATE__INVALID_PATH:
+		ret = error(_("fsmonitor_ipc__send_query: invalid path '%s'"),
+			    fsmonitor_ipc__get_path());
+		goto done;
+
+	case IPC_STATE__OTHER_ERROR:
+	default:
+		ret = error(_("fsmonitor_ipc__send_query: unspecified error on '%s'"),
+			    fsmonitor_ipc__get_path());
+		goto done;
+	}
+
+done:
+	trace2_region_leave("fsm_client", "query", NULL);
+
+	return ret;
+}
+
+int fsmonitor_ipc__send_command(const char *command,
+				struct strbuf *answer)
+{
+	struct ipc_client_connection *connection = NULL;
+	struct ipc_client_connect_options options
+		= IPC_CLIENT_CONNECT_OPTIONS_INIT;
+	int ret;
+	enum ipc_active_state state;
+	const char *c = command ? command : "";
+	size_t c_len = command ? strlen(command) : 0;
+
+	strbuf_reset(answer);
+
+	options.wait_if_busy = 1;
+	options.wait_if_not_found = 0;
+
+	state = ipc_client_try_connect(fsmonitor_ipc__get_path(), &options,
+				       &connection);
+	if (state != IPC_STATE__LISTENING) {
+		die(_("fsmonitor--daemon is not running"));
+		return -1;
+	}
+
+	ret = ipc_client_send_command_to_connection(connection, c, c_len,
+						    answer);
+	ipc_client_close_connection(connection);
+
+	if (ret == -1) {
+		die(_("could not send '%s' command to fsmonitor--daemon"), c);
+		return -1;
+	}
+
+	return 0;
+}
+
+#endif
diff --git a/fsmonitor-ipc.h b/fsmonitor-ipc.h
new file mode 100644
index 00000000000..b6a7067c3af
--- /dev/null
+++ b/fsmonitor-ipc.h
@@ -0,0 +1,48 @@
+#ifndef FSMONITOR_IPC_H
+#define FSMONITOR_IPC_H
+
+#include "simple-ipc.h"
+
+/*
+ * Returns true if built-in file system monitor daemon is defined
+ * for this platform.
+ */
+int fsmonitor_ipc__is_supported(void);
+
+/*
+ * Returns the pathname to the IPC named pipe or Unix domain socket
+ * where a `git-fsmonitor--daemon` process will listen.  This is a
+ * per-worktree value.
+ *
+ * Returns NULL if the daemon is not supported on this platform.
+ */
+const char *fsmonitor_ipc__get_path(void);
+
+/*
+ * Try to determine whether there is a `git-fsmonitor--daemon` process
+ * listening on the IPC pipe/socket.
+ */
+enum ipc_active_state fsmonitor_ipc__get_state(void);
+
+/*
+ * Connect to a `git-fsmonitor--daemon` process via simple-ipc
+ * and ask for the set of changed files since the given token.
+ *
+ * Spawn a daemon process in the background if necessary.
+ *
+ * Returns -1 on error; 0 on success.
+ */
+int fsmonitor_ipc__send_query(const char *since_token,
+			      struct strbuf *answer);
+
+/*
+ * Connect to a `git-fsmonitor--daemon` process via simple-ipc and
+ * send a command verb.  If no daemon is available, we DO NOT try to
+ * start one.
+ *
+ * Returns -1 on error; 0 on success.
+ */
+int fsmonitor_ipc__send_command(const char *command,
+				struct strbuf *answer);
+
+#endif /* FSMONITOR_IPC_H */
-- 
gitgitgadget

[PATCH v7 03/29] fsmonitor: config settings are repository-specific

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Move fsmonitor config settings to a new and opaque
`struct fsmonitor_settings` structure.  Add a lazily-loaded pointer
to this into `struct repo_settings`

Create an `enum fsmonitor_mode` type in `struct fsmonitor_settings` to
represent the state of fsmonitor.  This lets us represent which, if
any, fsmonitor provider (hook or IPC) is enabled.

Create `fsm_settings__get_*()` getters to lazily look up fsmonitor-
related config settings.

Get rid of the `core_fsmonitor` global variable.  Move the code to
lookup the existing `core.fsmonitor` config value into the fsmonitor
settings.

Create a hook pathname variable in `struct fsmonitor-settings` and
only set it when in hook mode.

Extend the definition of `core.fsmonitor` to be either a boolean
or a hook pathname.  When true, the builtin FSMonitor is used.
When false or unset, no FSMonitor (neither builtin nor hook) is
used.

The existing `core_fsmonitor` global variable was used to store the
pathname to the fsmonitor hook *and* it was used as a boolean to see
if fsmonitor was enabled.  This dual usage and global visibility leads
to confusion when we add the IPC-based provider.  So lets hide the
details in fsmonitor-settings.c and let it decide which provider to
use in the case of multiple settings.  This avoids cluttering up
repo-settings.c with these private details.

A future commit in builtin-fsmonitor series will add the ability to
disqualify worktrees for various reasons, such as being mounted from a
remote volume, where fsmonitor should not be started.  Having the
config settings hidden in fsmonitor-settings.c allows such worktree
restrictions to override the config values used.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 Makefile               |   1 +
 builtin/update-index.c |   7 ++-
 cache.h                |   1 -
 config.c               |  14 -----
 config.h               |   1 -
 environment.c          |   1 -
 fsmonitor-settings.c   | 114 +++++++++++++++++++++++++++++++++++++++++
 fsmonitor-settings.h   |  21 ++++++++
 fsmonitor.c            |  63 ++++++++++++++---------
 fsmonitor.h            |  15 ++++--
 repository.h           |   3 ++
 t/README               |   4 +-
 12 files changed, 196 insertions(+), 49 deletions(-)
 create mode 100644 fsmonitor-settings.c
 create mode 100644 fsmonitor-settings.h

diff --git a/Makefile b/Makefile
index a19d850e716..707a56d4c11 100644
--- a/Makefile
+++ b/Makefile
@@ -908,6 +908,7 @@ LIB_OBJS += fmt-merge-msg.o
 LIB_OBJS += fsck.o
 LIB_OBJS += fsmonitor.o
 LIB_OBJS += fsmonitor-ipc.o
+LIB_OBJS += fsmonitor-settings.o
 LIB_OBJS += gettext.o
 LIB_OBJS += gpg-interface.o
 LIB_OBJS += graph.o
diff --git a/builtin/update-index.c b/builtin/update-index.c
index aafe7eeac2a..876112abb21 100644
--- a/builtin/update-index.c
+++ b/builtin/update-index.c
@@ -1236,14 +1236,17 @@ int cmd_update_index(int argc, const char **argv, const char *prefix)
 	}
 
 	if (fsmonitor > 0) {
-		if (git_config_get_fsmonitor() == 0)
+		enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
+		if (fsm_mode == FSMONITOR_MODE_DISABLED) {
 			warning(_("core.fsmonitor is unset; "
 				"set it if you really want to "
 				"enable fsmonitor"));
+		}
 		add_fsmonitor(&the_index);
 		report(_("fsmonitor enabled"));
 	} else if (!fsmonitor) {
-		if (git_config_get_fsmonitor() == 1)
+		enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
+		if (fsm_mode > FSMONITOR_MODE_DISABLED)
 			warning(_("core.fsmonitor is set; "
 				"remove it if you really want to "
 				"disable fsmonitor"));
diff --git a/cache.h b/cache.h
index 04d4d2db25c..aaf334e2aa4 100644
--- a/cache.h
+++ b/cache.h
@@ -999,7 +999,6 @@ extern int core_preload_index;
 extern int precomposed_unicode;
 extern int protect_hfs;
 extern int protect_ntfs;
-extern const char *core_fsmonitor;
 
 extern int core_apply_sparse_checkout;
 extern int core_sparse_checkout_cone;
diff --git a/config.c b/config.c
index 383b1a4885b..3f9b0739a78 100644
--- a/config.c
+++ b/config.c
@@ -2626,20 +2626,6 @@ int git_config_get_max_percent_split_change(void)
 	return -1; /* default value */
 }
 
-int git_config_get_fsmonitor(void)
-{
-	if (git_config_get_pathname("core.fsmonitor", &core_fsmonitor))
-		core_fsmonitor = getenv("GIT_TEST_FSMONITOR");
-
-	if (core_fsmonitor && !*core_fsmonitor)
-		core_fsmonitor = NULL;
-
-	if (core_fsmonitor)
-		return 1;
-
-	return 0;
-}
-
 int git_config_get_index_threads(int *dest)
 {
 	int is_bool, val;
diff --git a/config.h b/config.h
index bb49baf1ee0..7654f61c634 100644
--- a/config.h
+++ b/config.h
@@ -597,7 +597,6 @@ int git_config_get_pathname(const char *key, const char **dest);
 int git_config_get_index_threads(int *dest);
 int git_config_get_split_index(void);
 int git_config_get_max_percent_split_change(void);
-int git_config_get_fsmonitor(void);
 
 /* This dies if the configured or default date is in the future */
 int git_config_get_expiry(const char *key, const char **output);
diff --git a/environment.c b/environment.c
index fd0501e77a5..00682e638d7 100644
--- a/environment.c
+++ b/environment.c
@@ -84,7 +84,6 @@ int protect_hfs = PROTECT_HFS_DEFAULT;
 #define PROTECT_NTFS_DEFAULT 1
 #endif
 int protect_ntfs = PROTECT_NTFS_DEFAULT;
-const char *core_fsmonitor;
 
 /*
  * The character that begins a commented line in user-editable file
diff --git a/fsmonitor-settings.c b/fsmonitor-settings.c
new file mode 100644
index 00000000000..757d230d538
--- /dev/null
+++ b/fsmonitor-settings.c
@@ -0,0 +1,114 @@
+#include "cache.h"
+#include "config.h"
+#include "repository.h"
+#include "fsmonitor-settings.h"
+
+/*
+ * We keep this structure defintion private and have getters
+ * for all fields so that we can lazy load it as needed.
+ */
+struct fsmonitor_settings {
+	enum fsmonitor_mode mode;
+	char *hook_path;
+};
+
+static void lookup_fsmonitor_settings(struct repository *r)
+{
+	struct fsmonitor_settings *s;
+	const char *const_str;
+	int bool_value;
+
+	if (r->settings.fsmonitor)
+		return;
+
+	CALLOC_ARRAY(s, 1);
+	s->mode = FSMONITOR_MODE_DISABLED;
+
+	r->settings.fsmonitor = s;
+
+	/*
+	 * Overload the existing "core.fsmonitor" config setting (which
+	 * has historically been either unset or a hook pathname) to
+	 * now allow a boolean value to enable the builtin FSMonitor
+	 * or to turn everything off.  (This does imply that you can't
+	 * use a hook script named "true" or "false", but that's OK.)
+	 */
+	switch (repo_config_get_maybe_bool(r, "core.fsmonitor", &bool_value)) {
+
+	case 0: /* config value was set to <bool> */
+		if (bool_value)
+			fsm_settings__set_ipc(r);
+		return;
+
+	case 1: /* config value was unset */
+		const_str = getenv("GIT_TEST_FSMONITOR");
+		break;
+
+	case -1: /* config value set to an arbitrary string */
+		if (repo_config_get_pathname(r, "core.fsmonitor", &const_str))
+			return; /* should not happen */
+		break;
+
+	default: /* should not happen */
+		return;
+	}
+
+	if (!const_str || !*const_str)
+		return;
+
+	fsm_settings__set_hook(r, const_str);
+}
+
+enum fsmonitor_mode fsm_settings__get_mode(struct repository *r)
+{
+	if (!r)
+		r = the_repository;
+
+	lookup_fsmonitor_settings(r);
+
+	return r->settings.fsmonitor->mode;
+}
+
+const char *fsm_settings__get_hook_path(struct repository *r)
+{
+	if (!r)
+		r = the_repository;
+
+	lookup_fsmonitor_settings(r);
+
+	return r->settings.fsmonitor->hook_path;
+}
+
+void fsm_settings__set_ipc(struct repository *r)
+{
+	if (!r)
+		r = the_repository;
+
+	lookup_fsmonitor_settings(r);
+
+	r->settings.fsmonitor->mode = FSMONITOR_MODE_IPC;
+	FREE_AND_NULL(r->settings.fsmonitor->hook_path);
+}
+
+void fsm_settings__set_hook(struct repository *r, const char *path)
+{
+	if (!r)
+		r = the_repository;
+
+	lookup_fsmonitor_settings(r);
+
+	r->settings.fsmonitor->mode = FSMONITOR_MODE_HOOK;
+	FREE_AND_NULL(r->settings.fsmonitor->hook_path);
+	r->settings.fsmonitor->hook_path = strdup(path);
+}
+
+void fsm_settings__set_disabled(struct repository *r)
+{
+	if (!r)
+		r = the_repository;
+
+	lookup_fsmonitor_settings(r);
+
+	r->settings.fsmonitor->mode = FSMONITOR_MODE_DISABLED;
+	FREE_AND_NULL(r->settings.fsmonitor->hook_path);
+}
diff --git a/fsmonitor-settings.h b/fsmonitor-settings.h
new file mode 100644
index 00000000000..a4c5d7b4889
--- /dev/null
+++ b/fsmonitor-settings.h
@@ -0,0 +1,21 @@
+#ifndef FSMONITOR_SETTINGS_H
+#define FSMONITOR_SETTINGS_H
+
+struct repository;
+
+enum fsmonitor_mode {
+	FSMONITOR_MODE_DISABLED = 0,
+	FSMONITOR_MODE_HOOK = 1, /* core.fsmonitor=<hook_path> */
+	FSMONITOR_MODE_IPC = 2,  /* core.fsmonitor=<true> */
+};
+
+void fsm_settings__set_ipc(struct repository *r);
+void fsm_settings__set_hook(struct repository *r, const char *path);
+void fsm_settings__set_disabled(struct repository *r);
+
+enum fsmonitor_mode fsm_settings__get_mode(struct repository *r);
+const char *fsm_settings__get_hook_path(struct repository *r);
+
+struct fsmonitor_settings;
+
+#endif /* FSMONITOR_SETTINGS_H */
diff --git a/fsmonitor.c b/fsmonitor.c
index 448d0ee33f5..0e961b74d82 100644
--- a/fsmonitor.c
+++ b/fsmonitor.c
@@ -3,6 +3,7 @@
 #include "dir.h"
 #include "ewah/ewok.h"
 #include "fsmonitor.h"
+#include "fsmonitor-ipc.h"
 #include "run-command.h"
 #include "strbuf.h"
 
@@ -148,15 +149,18 @@ void write_fsmonitor_extension(struct strbuf *sb, struct index_state *istate)
 /*
  * Call the query-fsmonitor hook passing the last update token of the saved results.
  */
-static int query_fsmonitor(int version, const char *last_update, struct strbuf *query_result)
+static int query_fsmonitor_hook(struct repository *r,
+				int version,
+				const char *last_update,
+				struct strbuf *query_result)
 {
 	struct child_process cp = CHILD_PROCESS_INIT;
 	int result;
 
-	if (!core_fsmonitor)
+	if (fsm_settings__get_mode(r) != FSMONITOR_MODE_HOOK)
 		return -1;
 
-	strvec_push(&cp.args, core_fsmonitor);
+	strvec_push(&cp.args, fsm_settings__get_hook_path(r));
 	strvec_pushf(&cp.args, "%d", version);
 	strvec_pushf(&cp.args, "%s", last_update);
 	cp.use_shell = 1;
@@ -225,17 +229,28 @@ void refresh_fsmonitor(struct index_state *istate)
 	char *buf;
 	unsigned int i;
 	int is_trivial = 0;
+	struct repository *r = istate->repo ? istate->repo : the_repository;
+	enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
 
-	if (!core_fsmonitor || istate->fsmonitor_has_run_once)
+	if (fsm_mode <= FSMONITOR_MODE_DISABLED ||
+	    istate->fsmonitor_has_run_once)
 		return;
 
-	hook_version = fsmonitor_hook_version();
-
 	istate->fsmonitor_has_run_once = 1;
 
 	trace_printf_key(&trace_fsmonitor, "refresh fsmonitor");
+
+	if (fsm_mode == FSMONITOR_MODE_IPC) {
+		/* TODO */
+		return;
+	}
+
+	assert(fsm_mode == FSMONITOR_MODE_HOOK);
+
+	hook_version = fsmonitor_hook_version();
+
 	/*
-	 * This could be racy so save the date/time now and query_fsmonitor
+	 * This could be racy so save the date/time now and query_fsmonitor_hook
 	 * should be inclusive to ensure we don't miss potential changes.
 	 */
 	last_update = getnanotime();
@@ -243,13 +258,14 @@ void refresh_fsmonitor(struct index_state *istate)
 		strbuf_addf(&last_update_token, "%"PRIu64"", last_update);
 
 	/*
-	 * If we have a last update token, call query_fsmonitor for the set of
+	 * If we have a last update token, call query_fsmonitor_hook for the set of
 	 * changes since that token, else assume everything is possibly dirty
 	 * and check it all.
 	 */
 	if (istate->fsmonitor_last_update) {
 		if (hook_version == -1 || hook_version == HOOK_INTERFACE_VERSION2) {
-			query_success = !query_fsmonitor(HOOK_INTERFACE_VERSION2,
+			query_success = !query_fsmonitor_hook(
+				r, HOOK_INTERFACE_VERSION2,
 				istate->fsmonitor_last_update, &query_result);
 
 			if (query_success) {
@@ -280,7 +296,8 @@ void refresh_fsmonitor(struct index_state *istate)
 		}
 
 		if (hook_version == HOOK_INTERFACE_VERSION1) {
-			query_success = !query_fsmonitor(HOOK_INTERFACE_VERSION1,
+			query_success = !query_fsmonitor_hook(
+				r, HOOK_INTERFACE_VERSION1,
 				istate->fsmonitor_last_update, &query_result);
 			if (query_success)
 				is_trivial = query_result.buf[0] == '/';
@@ -290,9 +307,12 @@ void refresh_fsmonitor(struct index_state *istate)
 			trace2_data_intmax("fsm_hook", NULL,
 					   "query/trivial-response", 1);
 
-		trace_performance_since(last_update, "fsmonitor process '%s'", core_fsmonitor);
-		trace_printf_key(&trace_fsmonitor, "fsmonitor process '%s' returned %s",
-			core_fsmonitor, query_success ? "success" : "failure");
+		trace_performance_since(last_update, "fsmonitor process '%s'",
+					fsm_settings__get_hook_path(r));
+		trace_printf_key(&trace_fsmonitor,
+				 "fsmonitor process '%s' returned %s",
+				 fsm_settings__get_hook_path(r),
+				 query_success ? "success" : "failure");
 	}
 
 	/*
@@ -429,7 +449,8 @@ void remove_fsmonitor(struct index_state *istate)
 void tweak_fsmonitor(struct index_state *istate)
 {
 	unsigned int i;
-	int fsmonitor_enabled = git_config_get_fsmonitor();
+	int fsmonitor_enabled = (fsm_settings__get_mode(istate->repo)
+				 > FSMONITOR_MODE_DISABLED);
 
 	if (istate->fsmonitor_dirty) {
 		if (fsmonitor_enabled) {
@@ -449,16 +470,8 @@ void tweak_fsmonitor(struct index_state *istate)
 		istate->fsmonitor_dirty = NULL;
 	}
 
-	switch (fsmonitor_enabled) {
-	case -1: /* keep: do nothing */
-		break;
-	case 0: /* false */
-		remove_fsmonitor(istate);
-		break;
-	case 1: /* true */
+	if (fsmonitor_enabled)
 		add_fsmonitor(istate);
-		break;
-	default: /* unknown value: do nothing */
-		break;
-	}
+	else
+		remove_fsmonitor(istate);
 }
diff --git a/fsmonitor.h b/fsmonitor.h
index f20d72631d7..3f41f653691 100644
--- a/fsmonitor.h
+++ b/fsmonitor.h
@@ -3,6 +3,7 @@
 
 #include "cache.h"
 #include "dir.h"
+#include "fsmonitor-settings.h"
 
 extern struct trace_key trace_fsmonitor;
 
@@ -57,7 +58,10 @@ int fsmonitor_is_trivial_response(const struct strbuf *query_result);
  */
 static inline int is_fsmonitor_refreshed(const struct index_state *istate)
 {
-	return !core_fsmonitor || istate->fsmonitor_has_run_once;
+	enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(istate->repo);
+
+	return fsm_mode <= FSMONITOR_MODE_DISABLED ||
+		istate->fsmonitor_has_run_once;
 }
 
 /*
@@ -67,7 +71,10 @@ static inline int is_fsmonitor_refreshed(const struct index_state *istate)
  */
 static inline void mark_fsmonitor_valid(struct index_state *istate, struct cache_entry *ce)
 {
-	if (core_fsmonitor && !(ce->ce_flags & CE_FSMONITOR_VALID)) {
+	enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(istate->repo);
+
+	if (fsm_mode > FSMONITOR_MODE_DISABLED &&
+	    !(ce->ce_flags & CE_FSMONITOR_VALID)) {
 		istate->cache_changed = 1;
 		ce->ce_flags |= CE_FSMONITOR_VALID;
 		trace_printf_key(&trace_fsmonitor, "mark_fsmonitor_clean '%s'", ce->name);
@@ -83,7 +90,9 @@ static inline void mark_fsmonitor_valid(struct index_state *istate, struct cache
  */
 static inline void mark_fsmonitor_invalid(struct index_state *istate, struct cache_entry *ce)
 {
-	if (core_fsmonitor) {
+	enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(istate->repo);
+
+	if (fsm_mode > FSMONITOR_MODE_DISABLED) {
 		ce->ce_flags &= ~CE_FSMONITOR_VALID;
 		untracked_cache_invalidate_path(istate, ce->name, 1);
 		trace_printf_key(&trace_fsmonitor, "mark_fsmonitor_invalid '%s'", ce->name);
diff --git a/repository.h b/repository.h
index ca837cb9e91..9bbb4659cc8 100644
--- a/repository.h
+++ b/repository.h
@@ -4,6 +4,7 @@
 #include "path.h"
 
 struct config_set;
+struct fsmonitor_settings;
 struct git_hash_algo;
 struct index_state;
 struct lock_file;
@@ -35,6 +36,8 @@ struct repo_settings {
 	int command_requires_full_index;
 	int sparse_index;
 
+	struct fsmonitor_settings *fsmonitor; /* lazily loaded */
+
 	int index_version;
 	enum untracked_cache_setting core_untracked_cache;
 
diff --git a/t/README b/t/README
index f48e0542cdc..9ffea1d3147 100644
--- a/t/README
+++ b/t/README
@@ -405,8 +405,8 @@ every 'git commit-graph write', as if the `--changed-paths` option was
 passed in.
 
 GIT_TEST_FSMONITOR=$PWD/t7519/fsmonitor-all exercises the fsmonitor
-code path for utilizing a file system monitor to speed up detecting
-new or changed files.
+code paths for utilizing a (hook based) file system monitor to speed up
+detecting new or changed files.
 
 GIT_TEST_INDEX_VERSION=<n> exercises the index read/write code path
 for the index version specified.  Can be set to any valid version
-- 
gitgitgadget

[PATCH v7 04/29] fsmonitor: use IPC to query the builtin FSMonitor daemon

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Use simple IPC to directly communicate with the new builtin file
system monitor daemon when `core.fsmonitor` is set to true.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 fsmonitor.c | 38 ++++++++++++++++++++++++++++++++++++--
 1 file changed, 36 insertions(+), 2 deletions(-)

diff --git a/fsmonitor.c b/fsmonitor.c
index 0e961b74d82..a38b5710eb3 100644
--- a/fsmonitor.c
+++ b/fsmonitor.c
@@ -241,8 +241,41 @@ void refresh_fsmonitor(struct index_state *istate)
 	trace_printf_key(&trace_fsmonitor, "refresh fsmonitor");
 
 	if (fsm_mode == FSMONITOR_MODE_IPC) {
-		/* TODO */
-		return;
+		query_success = !fsmonitor_ipc__send_query(
+			istate->fsmonitor_last_update ?
+			istate->fsmonitor_last_update : "builtin:fake",
+			&query_result);
+		if (query_success) {
+			/*
+			 * The response contains a series of nul terminated
+			 * strings.  The first is the new token.
+			 *
+			 * Use `char *buf` as an interlude to trick the CI
+			 * static analysis to let us use `strbuf_addstr()`
+			 * here (and only copy the token) rather than
+			 * `strbuf_addbuf()`.
+			 */
+			buf = query_result.buf;
+			strbuf_addstr(&last_update_token, buf);
+			bol = last_update_token.len + 1;
+			is_trivial = query_result.buf[bol] == '/';
+			if (is_trivial)
+				trace2_data_intmax("fsm_client", NULL,
+						   "query/trivial-response", 1);
+		} else {
+			/*
+			 * The builtin daemon is not available on this
+			 * platform -OR- we failed to get a response.
+			 *
+			 * Generate a fake token (rather than a V1
+			 * timestamp) for the index extension.  (If
+			 * they switch back to the hook API, we don't
+			 * want ambiguous state.)
+			 */
+			strbuf_addstr(&last_update_token, "builtin:fake");
+		}
+
+		goto apply_results;
 	}
 
 	assert(fsm_mode == FSMONITOR_MODE_HOOK);
@@ -315,6 +348,7 @@ void refresh_fsmonitor(struct index_state *istate)
 				 query_success ? "success" : "failure");
 	}
 
+apply_results:
 	/*
 	 * The response from FSMonitor (excluding the header token) is
 	 * either:
-- 
gitgitgadget

[PATCH v7 05/29] fsmonitor: document builtin fsmonitor

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Document how `core.fsmonitor` can be set to a boolean to enable
or disable the builtin FSMonitor.

Update references to `core.fsmonitor` and `core.fsmonitorHookVersion` and
pointers to `Watchman` to refer to it.

Create `git-fsmonitor--daemon` manual page and describe its features.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 Documentation/config/core.txt           | 60 +++++++++++++++-----
 Documentation/git-fsmonitor--daemon.txt | 75 +++++++++++++++++++++++++
 Documentation/git-update-index.txt      |  8 ++-
 3 files changed, 126 insertions(+), 17 deletions(-)
 create mode 100644 Documentation/git-fsmonitor--daemon.txt

diff --git a/Documentation/config/core.txt b/Documentation/config/core.txt
index c04f62a54a1..6303c36c7ed 100644
--- a/Documentation/config/core.txt
+++ b/Documentation/config/core.txt
@@ -62,22 +62,54 @@ core.protectNTFS::
 	Defaults to `true` on Windows, and `false` elsewhere.
 
 core.fsmonitor::
-	If set, the value of this variable is used as a command which
-	will identify all files that may have changed since the
-	requested date/time. This information is used to speed up git by
-	avoiding unnecessary processing of files that have not changed.
-	See the "fsmonitor-watchman" section of linkgit:githooks[5].
+	If set to true, enable the built-in file system monitor
+	daemon for this working directory (linkgit:git-fsmonitor--daemon[1]).
++
+Like hook-based file system monitors, the built-in file system monitor
+can speed up Git commands that need to refresh the Git index
+(e.g. `git status`) in a working directory with many files.  The
+built-in monitor eliminates the need to install and maintain an
+external third-party tool.
++
+The built-in file system monitor is currently available only on a
+limited set of supported platforms.  Currently, this includes Windows
+and MacOS.
++
+	Otherwise, this variable contains the pathname of the "fsmonitor"
+	hook command.
++
+This hook command is used to identify all files that may have changed
+since the requested date/time. This information is used to speed up
+git by avoiding unnecessary scanning of files that have not changed.
++
+See the "fsmonitor-watchman" section of linkgit:githooks[5].
++
+Note that if you concurrently use multiple versions of Git, such
+as one version on the command line and another version in an IDE
+tool, that the definition of `core.fsmonitor` was extended to
+allow boolean values in addition to hook pathnames.  Git versions
+2.35.1 and prior will not understand the boolean values and will
+consider the "true" or "false" values as hook pathnames to be
+invoked.  Git versions 2.26 thru 2.35.1 default to hook protocol
+V2 and will fall back to no fsmonitor (full scan).  Git versions
+prior to 2.26 default to hook protocol V1 and will silently
+assume there were no changes to report (no scan), so status
+commands may report incomplete results.  For this reason, it is
+best to upgrade all of your Git versions before using the built-in
+file system monitor.
 
 core.fsmonitorHookVersion::
-	Sets the version of hook that is to be used when calling fsmonitor.
-	There are currently versions 1 and 2. When this is not set,
-	version 2 will be tried first and if it fails then version 1
-	will be tried. Version 1 uses a timestamp as input to determine
-	which files have changes since that time but some monitors
-	like watchman have race conditions when used with a timestamp.
-	Version 2 uses an opaque string so that the monitor can return
-	something that can be used to determine what files have changed
-	without race conditions.
+	Sets the protocol version to be used when invoking the
+	"fsmonitor" hook.
++
+There are currently versions 1 and 2. When this is not set,
+version 2 will be tried first and if it fails then version 1
+will be tried. Version 1 uses a timestamp as input to determine
+which files have changes since that time but some monitors
+like Watchman have race conditions when used with a timestamp.
+Version 2 uses an opaque string so that the monitor can return
+something that can be used to determine what files have changed
+without race conditions.
 
 core.trustctime::
 	If false, the ctime differences between the index and the
diff --git a/Documentation/git-fsmonitor--daemon.txt b/Documentation/git-fsmonitor--daemon.txt
new file mode 100644
index 00000000000..0fedf5a4565
--- /dev/null
+++ b/Documentation/git-fsmonitor--daemon.txt
@@ -0,0 +1,75 @@
+git-fsmonitor--daemon(1)
+========================
+
+NAME
+----
+git-fsmonitor--daemon - A Built-in File System Monitor
+
+SYNOPSIS
+--------
+[verse]
+'git fsmonitor--daemon' start
+'git fsmonitor--daemon' run
+'git fsmonitor--daemon' stop
+'git fsmonitor--daemon' status
+
+DESCRIPTION
+-----------
+
+A daemon to watch the working directory for file and directory
+changes using platform-specific file system notification facilities.
+
+This daemon communicates directly with commands like `git status`
+using the link:technical/api-simple-ipc.html[simple IPC] interface
+instead of the slower linkgit:githooks[5] interface.
+
+This daemon is built into Git so that no third-party tools are
+required.
+
+OPTIONS
+-------
+
+start::
+	Starts a daemon in the background.
+
+run::
+	Runs a daemon in the foreground.
+
+stop::
+	Stops the daemon running in the current working
+	directory, if present.
+
+status::
+	Exits with zero status if a daemon is watching the
+	current working directory.
+
+REMARKS
+-------
+
+This daemon is a long running process used to watch a single working
+directory and maintain a list of the recently changed files and
+directories.  Performance of commands such as `git status` can be
+increased if they just ask for a summary of changes to the working
+directory and can avoid scanning the disk.
+
+When `core.fsmonitor` is set to `true` (see linkgit:git-config[1])
+commands, such as `git status`, will ask the daemon for changes and
+automatically start it (if necessary).
+
+For more information see the "File System Monitor" section in
+linkgit:git-update-index[1].
+
+CAVEATS
+-------
+
+The fsmonitor daemon does not currently know about submodules and does
+not know to filter out file system events that happen within a
+submodule.  If fsmonitor daemon is watching a super repo and a file is
+modified within the working directory of a submodule, it will report
+the change (as happening against the super repo).  However, the client
+will properly ignore these extra events, so performance may be affected
+but it will not cause an incorrect result.
+
+GIT
+---
+Part of the linkgit:git[1] suite
diff --git a/Documentation/git-update-index.txt b/Documentation/git-update-index.txt
index 2853f168d97..53ea48a04e2 100644
--- a/Documentation/git-update-index.txt
+++ b/Documentation/git-update-index.txt
@@ -498,7 +498,9 @@ FILE SYSTEM MONITOR
 This feature is intended to speed up git operations for repos that have
 large working directories.
 
-It enables git to work together with a file system monitor (see the
+It enables git to work together with a file system monitor (see
+linkgit:git-fsmonitor--daemon[1]
+and the
 "fsmonitor-watchman" section of linkgit:githooks[5]) that can
 inform it as to what files have been modified. This enables git to avoid
 having to lstat() every file to find modified files.
@@ -509,8 +511,8 @@ looking for new files.
 
 If you want to enable (or disable) this feature, it is easier to use
 the `core.fsmonitor` configuration variable (see
-linkgit:git-config[1]) than using the `--fsmonitor` option to
-`git update-index` in each repository, especially if you want to do so
+linkgit:git-config[1]) than using the `--fsmonitor` option to `git
+update-index` in each repository, especially if you want to do so
 across all repositories you use, because you can set the configuration
 variable in your `$HOME/.gitconfig` just once and have it affect all
 repositories you touch.
-- 
gitgitgadget

[PATCH v7 06/29] fsmonitor--daemon: add a built-in fsmonitor daemon

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Create a built-in file system monitoring daemon that can be used by
the existing `fsmonitor` feature (protocol API and index extension)
to improve the performance of various Git commands, such as `status`.

The `fsmonitor--daemon` feature builds upon the `Simple IPC` API and
provides an alternative to hook access to existing fsmonitors such
as `watchman`.

This commit merely adds the new command without any functionality.

Co-authored-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 .gitignore                  |  1 +
 Makefile                    |  1 +
 builtin.h                   |  1 +
 builtin/fsmonitor--daemon.c | 46 +++++++++++++++++++++++++++++++++++++
 git.c                       |  1 +
 5 files changed, 50 insertions(+)
 create mode 100644 builtin/fsmonitor--daemon.c

diff --git a/.gitignore b/.gitignore
index f817c509ec0..e81de1063a4 100644
--- a/.gitignore
+++ b/.gitignore
@@ -72,6 +72,7 @@
 /git-format-patch
 /git-fsck
 /git-fsck-objects
+/git-fsmonitor--daemon
 /git-gc
 /git-get-tar-commit-id
 /git-grep
diff --git a/Makefile b/Makefile
index 707a56d4c11..5af1d5b112e 100644
--- a/Makefile
+++ b/Makefile
@@ -1114,6 +1114,7 @@ BUILTIN_OBJS += builtin/fmt-merge-msg.o
 BUILTIN_OBJS += builtin/for-each-ref.o
 BUILTIN_OBJS += builtin/for-each-repo.o
 BUILTIN_OBJS += builtin/fsck.o
+BUILTIN_OBJS += builtin/fsmonitor--daemon.o
 BUILTIN_OBJS += builtin/gc.o
 BUILTIN_OBJS += builtin/get-tar-commit-id.o
 BUILTIN_OBJS += builtin/grep.o
diff --git a/builtin.h b/builtin.h
index 83379f3832c..40e9ecc8485 100644
--- a/builtin.h
+++ b/builtin.h
@@ -159,6 +159,7 @@ int cmd_for_each_ref(int argc, const char **argv, const char *prefix);
 int cmd_for_each_repo(int argc, const char **argv, const char *prefix);
 int cmd_format_patch(int argc, const char **argv, const char *prefix);
 int cmd_fsck(int argc, const char **argv, const char *prefix);
+int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix);
 int cmd_gc(int argc, const char **argv, const char *prefix);
 int cmd_get_tar_commit_id(int argc, const char **argv, const char *prefix);
 int cmd_grep(int argc, const char **argv, const char *prefix);
diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
new file mode 100644
index 00000000000..f0498793379
--- /dev/null
+++ b/builtin/fsmonitor--daemon.c
@@ -0,0 +1,46 @@
+#include "builtin.h"
+#include "config.h"
+#include "parse-options.h"
+#include "fsmonitor.h"
+#include "fsmonitor-ipc.h"
+#include "simple-ipc.h"
+#include "khash.h"
+
+static const char * const builtin_fsmonitor__daemon_usage[] = {
+	NULL
+};
+
+#ifdef HAVE_FSMONITOR_DAEMON_BACKEND
+
+int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
+{
+	const char *subcmd;
+
+	struct option options[] = {
+		OPT_END()
+	};
+
+	git_config(git_default_config, NULL);
+
+	argc = parse_options(argc, argv, prefix, options,
+			     builtin_fsmonitor__daemon_usage, 0);
+	if (argc != 1)
+		usage_with_options(builtin_fsmonitor__daemon_usage, options);
+	subcmd = argv[0];
+
+	die(_("Unhandled subcommand '%s'"), subcmd);
+}
+
+#else
+int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
+{
+	struct option options[] = {
+		OPT_END()
+	};
+
+	if (argc == 2 && !strcmp(argv[1], "-h"))
+		usage_with_options(builtin_fsmonitor__daemon_usage, options);
+
+	die(_("fsmonitor--daemon not supported on this platform"));
+}
+#endif
diff --git a/git.c b/git.c
index a25940d72e8..3d8e48cf555 100644
--- a/git.c
+++ b/git.c
@@ -537,6 +537,7 @@ static struct cmd_struct commands[] = {
 	{ "format-patch", cmd_format_patch, RUN_SETUP },
 	{ "fsck", cmd_fsck, RUN_SETUP },
 	{ "fsck-objects", cmd_fsck, RUN_SETUP },
+	{ "fsmonitor--daemon", cmd_fsmonitor__daemon, RUN_SETUP },
 	{ "gc", cmd_gc, RUN_SETUP },
 	{ "get-tar-commit-id", cmd_get_tar_commit_id, NO_PARSEOPT },
 	{ "grep", cmd_grep, RUN_SETUP_GENTLY },
-- 
gitgitgadget

[PATCH v7 07/29] fsmonitor--daemon: implement 'stop' and 'status' commands

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Implement `stop` and `status` client commands to control and query the
status of a `fsmonitor--daemon` server process (and implicitly start a
server process if necessary).

Later commits will implement the actual server and monitor the file
system.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 builtin/fsmonitor--daemon.c | 51 +++++++++++++++++++++++++++++++++++++
 1 file changed, 51 insertions(+)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index f0498793379..5e3178b8bdd 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -7,10 +7,55 @@
 #include "khash.h"
 
 static const char * const builtin_fsmonitor__daemon_usage[] = {
+	N_("git fsmonitor--daemon stop"),
+	N_("git fsmonitor--daemon status"),
 	NULL
 };
 
 #ifdef HAVE_FSMONITOR_DAEMON_BACKEND
+/*
+ * Acting as a CLIENT.
+ *
+ * Send a "quit" command to the `git-fsmonitor--daemon` (if running)
+ * and wait for it to shutdown.
+ */
+static int do_as_client__send_stop(void)
+{
+	struct strbuf answer = STRBUF_INIT;
+	int ret;
+
+	ret = fsmonitor_ipc__send_command("quit", &answer);
+
+	/* The quit command does not return any response data. */
+	strbuf_release(&answer);
+
+	if (ret)
+		return ret;
+
+	trace2_region_enter("fsm_client", "polling-for-daemon-exit", NULL);
+	while (fsmonitor_ipc__get_state() == IPC_STATE__LISTENING)
+		sleep_millisec(50);
+	trace2_region_leave("fsm_client", "polling-for-daemon-exit", NULL);
+
+	return 0;
+}
+
+static int do_as_client__status(void)
+{
+	enum ipc_active_state state = fsmonitor_ipc__get_state();
+
+	switch (state) {
+	case IPC_STATE__LISTENING:
+		printf(_("fsmonitor-daemon is watching '%s'\n"),
+		       the_repository->worktree);
+		return 0;
+
+	default:
+		printf(_("fsmonitor-daemon is not watching '%s'\n"),
+		       the_repository->worktree);
+		return 1;
+	}
+}
 
 int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
 {
@@ -28,6 +73,12 @@ int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
 		usage_with_options(builtin_fsmonitor__daemon_usage, options);
 	subcmd = argv[0];
 
+	if (!strcmp(subcmd, "stop"))
+		return !!do_as_client__send_stop();
+
+	if (!strcmp(subcmd, "status"))
+		return !!do_as_client__status();
+
 	die(_("Unhandled subcommand '%s'"), subcmd);
 }
 
-- 
gitgitgadget

[PATCH v7 08/29] compat/fsmonitor/fsm-listen-win32: stub in backend for Windows

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Stub in empty filesystem listener backend for fsmonitor--daemon on Windows.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 Makefile                            | 13 ++++++++
 compat/fsmonitor/fsm-listen-win32.c | 21 +++++++++++++
 compat/fsmonitor/fsm-listen.h       | 49 +++++++++++++++++++++++++++++
 config.mak.uname                    | 10 ++++++
 contrib/buildsystems/CMakeLists.txt |  7 +++++
 repo-settings.c                     |  1 +
 6 files changed, 101 insertions(+)
 create mode 100644 compat/fsmonitor/fsm-listen-win32.c
 create mode 100644 compat/fsmonitor/fsm-listen.h

diff --git a/Makefile b/Makefile
index 5af1d5b112e..26567d4f772 100644
--- a/Makefile
+++ b/Makefile
@@ -470,6 +470,11 @@ all::
 # directory, and the JSON compilation database 'compile_commands.json' will be
 # created at the root of the repository.
 #
+# If your platform supports a built-in fsmonitor backend, set
+# FSMONITOR_DAEMON_BACKEND to the "<name>" of the corresponding
+# `compat/fsmonitor/fsm-listen-<name>.c` that implements the
+# `fsm_listen__*()` routines.
+#
 # Define DEVELOPER to enable more compiler warnings. Compiler version
 # and family are auto detected, but could be overridden by defining
 # COMPILER_FEATURES (see config.mak.dev). You can still set
@@ -1968,6 +1973,11 @@ ifdef NEED_ACCESS_ROOT_HANDLER
 	COMPAT_OBJS += compat/access.o
 endif
 
+ifdef FSMONITOR_DAEMON_BACKEND
+	COMPAT_CFLAGS += -DHAVE_FSMONITOR_DAEMON_BACKEND
+	COMPAT_OBJS += compat/fsmonitor/fsm-listen-$(FSMONITOR_DAEMON_BACKEND).o
+endif
+
 ifeq ($(TCLTK_PATH),)
 NO_TCLTK = NoThanks
 endif
@@ -2887,6 +2897,9 @@ GIT-BUILD-OPTIONS: FORCE
 	@echo DC_SHA1=\''$(subst ','\'',$(subst ','\'',$(DC_SHA1)))'\' >>$@+
 	@echo SANITIZE_LEAK=\''$(subst ','\'',$(subst ','\'',$(SANITIZE_LEAK)))'\' >>$@+
 	@echo X=\'$(X)\' >>$@+
+ifdef FSMONITOR_DAEMON_BACKEND
+	@echo FSMONITOR_DAEMON_BACKEND=\''$(subst ','\'',$(subst ','\'',$(FSMONITOR_DAEMON_BACKEND)))'\' >>$@+
+endif
 ifdef TEST_OUTPUT_DIRECTORY
 	@echo TEST_OUTPUT_DIRECTORY=\''$(subst ','\'',$(subst ','\'',$(TEST_OUTPUT_DIRECTORY)))'\' >>$@+
 endif
diff --git a/compat/fsmonitor/fsm-listen-win32.c b/compat/fsmonitor/fsm-listen-win32.c
new file mode 100644
index 00000000000..916cbea254f
--- /dev/null
+++ b/compat/fsmonitor/fsm-listen-win32.c
@@ -0,0 +1,21 @@
+#include "cache.h"
+#include "config.h"
+#include "fsmonitor.h"
+#include "fsm-listen.h"
+
+void fsm_listen__stop_async(struct fsmonitor_daemon_state *state)
+{
+}
+
+void fsm_listen__loop(struct fsmonitor_daemon_state *state)
+{
+}
+
+int fsm_listen__ctor(struct fsmonitor_daemon_state *state)
+{
+	return -1;
+}
+
+void fsm_listen__dtor(struct fsmonitor_daemon_state *state)
+{
+}
diff --git a/compat/fsmonitor/fsm-listen.h b/compat/fsmonitor/fsm-listen.h
new file mode 100644
index 00000000000..f0539349baf
--- /dev/null
+++ b/compat/fsmonitor/fsm-listen.h
@@ -0,0 +1,49 @@
+#ifndef FSM_LISTEN_H
+#define FSM_LISTEN_H
+
+/* This needs to be implemented by each backend */
+
+#ifdef HAVE_FSMONITOR_DAEMON_BACKEND
+
+struct fsmonitor_daemon_state;
+
+/*
+ * Initialize platform-specific data for the fsmonitor listener thread.
+ * This will be called from the main thread PRIOR to staring the
+ * fsmonitor_fs_listener thread.
+ *
+ * Returns 0 if successful.
+ * Returns -1 otherwise.
+ */
+int fsm_listen__ctor(struct fsmonitor_daemon_state *state);
+
+/*
+ * Cleanup platform-specific data for the fsmonitor listener thread.
+ * This will be called from the main thread AFTER joining the listener.
+ */
+void fsm_listen__dtor(struct fsmonitor_daemon_state *state);
+
+/*
+ * The main body of the platform-specific event loop to watch for
+ * filesystem events.  This will run in the fsmonitor_fs_listen thread.
+ *
+ * It should call `ipc_server_stop_async()` if the listener thread
+ * prematurely terminates (because of a filesystem error or if it
+ * detects that the .git directory has been deleted).  (It should NOT
+ * do so if the listener thread receives a normal shutdown signal from
+ * the IPC layer.)
+ *
+ * It should set `state->error_code` to -1 if the daemon should exit
+ * with an error.
+ */
+void fsm_listen__loop(struct fsmonitor_daemon_state *state);
+
+/*
+ * Gently request that the fsmonitor listener thread shutdown.
+ * It does not wait for it to stop.  The caller should do a JOIN
+ * to wait for it.
+ */
+void fsm_listen__stop_async(struct fsmonitor_daemon_state *state);
+
+#endif /* HAVE_FSMONITOR_DAEMON_BACKEND */
+#endif /* FSM_LISTEN_H */
diff --git a/config.mak.uname b/config.mak.uname
index 4352ea39e9b..26074f56bed 100644
--- a/config.mak.uname
+++ b/config.mak.uname
@@ -435,6 +435,11 @@ ifeq ($(uname_S),Windows)
 	# so we don't need this:
 	#
 	#   SNPRINTF_RETURNS_BOGUS = YesPlease
+
+	# The builtin FSMonitor requires Named Pipes and Threads on Windows.
+	# These are always available, so we do not have to conditionally
+	# support it.
+	FSMONITOR_DAEMON_BACKEND = win32
 	NO_SVN_TESTS = YesPlease
 	RUNTIME_PREFIX = YesPlease
 	HAVE_WPGMPTR = YesWeDo
@@ -619,6 +624,11 @@ ifeq ($(uname_S),MINGW)
 	NO_STRTOUMAX = YesPlease
 	NO_MKDTEMP = YesPlease
 	NO_SVN_TESTS = YesPlease
+
+	# The builtin FSMonitor requires Named Pipes and Threads on Windows.
+	# These are always available, so we do not have to conditionally
+	# support it.
+	FSMONITOR_DAEMON_BACKEND = win32
 	RUNTIME_PREFIX = YesPlease
 	HAVE_WPGMPTR = YesWeDo
 	NO_ST_BLOCKS_IN_STRUCT_STAT = YesPlease
diff --git a/contrib/buildsystems/CMakeLists.txt b/contrib/buildsystems/CMakeLists.txt
index e44232f85d3..0963629db7f 100644
--- a/contrib/buildsystems/CMakeLists.txt
+++ b/contrib/buildsystems/CMakeLists.txt
@@ -285,6 +285,13 @@ else()
 	endif()
 endif()
 
+if(SUPPORTS_SIMPLE_IPC)
+	if(CMAKE_SYSTEM_NAME STREQUAL "Windows")
+		add_compile_definitions(HAVE_FSMONITOR_DAEMON_BACKEND)
+		list(APPEND compat_SOURCES compat/fsmonitor/fsm-listen-win32.c)
+	endif()
+endif()
+
 set(EXE_EXTENSION ${CMAKE_EXECUTABLE_SUFFIX})
 
 #header checks
diff --git a/repo-settings.c b/repo-settings.c
index b4fbd16cdcc..2dfcb2b6542 100644
--- a/repo-settings.c
+++ b/repo-settings.c
@@ -2,6 +2,7 @@
 #include "config.h"
 #include "repository.h"
 #include "midx.h"
+#include "compat/fsmonitor/fsm-listen.h"
 
 static void repo_cfg_bool(struct repository *r, const char *key, int *dest,
 			  int def)
-- 
gitgitgadget

[PATCH v7 09/29] compat/fsmonitor/fsm-listen-darwin: stub in backend for Darwin

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Stub in empty implementation of fsmonitor--daemon
backend for Darwin (aka MacOS).

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 compat/fsmonitor/fsm-listen-darwin.c | 20 ++++++++++++++++++++
 config.mak.uname                     | 10 ++++++++++
 contrib/buildsystems/CMakeLists.txt  |  3 +++
 3 files changed, 33 insertions(+)
 create mode 100644 compat/fsmonitor/fsm-listen-darwin.c

diff --git a/compat/fsmonitor/fsm-listen-darwin.c b/compat/fsmonitor/fsm-listen-darwin.c
new file mode 100644
index 00000000000..c84e3344ab9
--- /dev/null
+++ b/compat/fsmonitor/fsm-listen-darwin.c
@@ -0,0 +1,20 @@
+#include "cache.h"
+#include "fsmonitor.h"
+#include "fsm-listen.h"
+
+int fsm_listen__ctor(struct fsmonitor_daemon_state *state)
+{
+	return -1;
+}
+
+void fsm_listen__dtor(struct fsmonitor_daemon_state *state)
+{
+}
+
+void fsm_listen__stop_async(struct fsmonitor_daemon_state *state)
+{
+}
+
+void fsm_listen__loop(struct fsmonitor_daemon_state *state)
+{
+}
diff --git a/config.mak.uname b/config.mak.uname
index 26074f56bed..501970902da 100644
--- a/config.mak.uname
+++ b/config.mak.uname
@@ -157,6 +157,16 @@ ifeq ($(uname_S),Darwin)
 			MSGFMT = /usr/local/opt/gettext/bin/msgfmt
 		endif
 	endif
+
+	# The builtin FSMonitor on MacOS builds upon Simple-IPC.  Both require
+	# Unix domain sockets and PThreads.
+	ifndef NO_PTHREADS
+	ifndef NO_UNIX_SOCKETS
+	FSMONITOR_DAEMON_BACKEND = darwin
+	endif
+	endif
+
+	BASIC_LDFLAGS += -framework CoreServices
 endif
 ifeq ($(uname_S),SunOS)
 	NEEDS_SOCKET = YesPlease
diff --git a/contrib/buildsystems/CMakeLists.txt b/contrib/buildsystems/CMakeLists.txt
index 0963629db7f..ee0d7257b77 100644
--- a/contrib/buildsystems/CMakeLists.txt
+++ b/contrib/buildsystems/CMakeLists.txt
@@ -289,6 +289,9 @@ if(SUPPORTS_SIMPLE_IPC)
 	if(CMAKE_SYSTEM_NAME STREQUAL "Windows")
 		add_compile_definitions(HAVE_FSMONITOR_DAEMON_BACKEND)
 		list(APPEND compat_SOURCES compat/fsmonitor/fsm-listen-win32.c)
+	elseif(CMAKE_SYSTEM_NAME STREQUAL "Darwin")
+		add_compile_definitions(HAVE_FSMONITOR_DAEMON_BACKEND)
+		list(APPEND compat_SOURCES compat/fsmonitor/fsm-listen-darwin.c)
 	endif()
 endif()
 
-- 
gitgitgadget

[PATCH v7 10/29] fsmonitor--daemon: implement 'run' command

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Implement `run` command to try to begin listening for file system events.

This version defines the thread structure with a single fsmonitor_fs_listen
thread to watch for file system events and a simple IPC thread pool to
watch for connection from Git clients over a well-known named pipe or
Unix domain socket.

This commit does not actually do anything yet because the platform
backends are still just stubs.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 builtin/fsmonitor--daemon.c | 228 +++++++++++++++++++++++++++++++++++-
 fsmonitor--daemon.h         |  34 ++++++
 2 files changed, 261 insertions(+), 1 deletion(-)
 create mode 100644 fsmonitor--daemon.h

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index 5e3178b8bdd..5591339399a 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -3,16 +3,52 @@
 #include "parse-options.h"
 #include "fsmonitor.h"
 #include "fsmonitor-ipc.h"
+#include "compat/fsmonitor/fsm-listen.h"
+#include "fsmonitor--daemon.h"
 #include "simple-ipc.h"
 #include "khash.h"
 
 static const char * const builtin_fsmonitor__daemon_usage[] = {
+	N_("git fsmonitor--daemon run [<options>]"),
 	N_("git fsmonitor--daemon stop"),
 	N_("git fsmonitor--daemon status"),
 	NULL
 };
 
 #ifdef HAVE_FSMONITOR_DAEMON_BACKEND
+/*
+ * Global state loaded from config.
+ */
+#define FSMONITOR__IPC_THREADS "fsmonitor.ipcthreads"
+static int fsmonitor__ipc_threads = 8;
+
+#define FSMONITOR__ANNOUNCE_STARTUP "fsmonitor.announcestartup"
+static int fsmonitor__announce_startup = 0;
+
+static int fsmonitor_config(const char *var, const char *value, void *cb)
+{
+	if (!strcmp(var, FSMONITOR__IPC_THREADS)) {
+		int i = git_config_int(var, value);
+		if (i < 1)
+			return error(_("value of '%s' out of range: %d"),
+				     FSMONITOR__IPC_THREADS, i);
+		fsmonitor__ipc_threads = i;
+		return 0;
+	}
+
+	if (!strcmp(var, FSMONITOR__ANNOUNCE_STARTUP)) {
+		int is_bool;
+		int i = git_config_bool_or_int(var, value, &is_bool);
+		if (i < 0)
+			return error(_("value of '%s' not bool or int: %d"),
+				     var, i);
+		fsmonitor__announce_startup = i;
+		return 0;
+	}
+
+	return git_default_config(var, value, cb);
+}
+
 /*
  * Acting as a CLIENT.
  *
@@ -57,15 +93,198 @@ static int do_as_client__status(void)
 	}
 }
 
+static ipc_server_application_cb handle_client;
+
+static int handle_client(void *data,
+			 const char *command, size_t command_len,
+			 ipc_server_reply_cb *reply,
+			 struct ipc_server_reply_data *reply_data)
+{
+	/* struct fsmonitor_daemon_state *state = data; */
+	int result;
+
+	/*
+	 * The Simple IPC API now supports {char*, len} arguments, but
+	 * FSMonitor always uses proper null-terminated strings, so
+	 * we can ignore the command_len argument.  (Trust, but verify.)
+	 */
+	if (command_len != strlen(command))
+		BUG("FSMonitor assumes text messages");
+
+	trace2_region_enter("fsmonitor", "handle_client", the_repository);
+	trace2_data_string("fsmonitor", the_repository, "request", command);
+
+	result = 0; /* TODO Do something here. */
+
+	trace2_region_leave("fsmonitor", "handle_client", the_repository);
+
+	return result;
+}
+
+static void *fsm_listen__thread_proc(void *_state)
+{
+	struct fsmonitor_daemon_state *state = _state;
+
+	trace2_thread_start("fsm-listen");
+
+	trace_printf_key(&trace_fsmonitor, "Watching: worktree '%s'",
+			 state->path_worktree_watch.buf);
+	if (state->nr_paths_watching > 1)
+		trace_printf_key(&trace_fsmonitor, "Watching: gitdir '%s'",
+				 state->path_gitdir_watch.buf);
+
+	fsm_listen__loop(state);
+
+	trace2_thread_exit();
+	return NULL;
+}
+
+static int fsmonitor_run_daemon_1(struct fsmonitor_daemon_state *state)
+{
+	struct ipc_server_opts ipc_opts = {
+		.nr_threads = fsmonitor__ipc_threads,
+
+		/*
+		 * We know that there are no other active threads yet,
+		 * so we can let the IPC layer temporarily chdir() if
+		 * it needs to when creating the server side of the
+		 * Unix domain socket.
+		 */
+		.uds_disallow_chdir = 0
+	};
+
+	/*
+	 * Start the IPC thread pool before the we've started the file
+	 * system event listener thread so that we have the IPC handle
+	 * before we need it.
+	 */
+	if (ipc_server_run_async(&state->ipc_server_data,
+				 fsmonitor_ipc__get_path(), &ipc_opts,
+				 handle_client, state))
+		return error_errno(
+			_("could not start IPC thread pool on '%s'"),
+			fsmonitor_ipc__get_path());
+
+	/*
+	 * Start the fsmonitor listener thread to collect filesystem
+	 * events.
+	 */
+	if (pthread_create(&state->listener_thread, NULL,
+			   fsm_listen__thread_proc, state) < 0) {
+		ipc_server_stop_async(state->ipc_server_data);
+		ipc_server_await(state->ipc_server_data);
+
+		return error(_("could not start fsmonitor listener thread"));
+	}
+
+	/*
+	 * The daemon is now fully functional in background threads.
+	 * Wait for the IPC thread pool to shutdown (whether by client
+	 * request or from filesystem activity).
+	 */
+	ipc_server_await(state->ipc_server_data);
+
+	/*
+	 * The fsmonitor listener thread may have received a shutdown
+	 * event from the IPC thread pool, but it doesn't hurt to tell
+	 * it again.  And wait for it to shutdown.
+	 */
+	fsm_listen__stop_async(state);
+	pthread_join(state->listener_thread, NULL);
+
+	return state->error_code;
+}
+
+static int fsmonitor_run_daemon(void)
+{
+	struct fsmonitor_daemon_state state;
+	int err;
+
+	memset(&state, 0, sizeof(state));
+
+	pthread_mutex_init(&state.main_lock, NULL);
+	state.error_code = 0;
+	state.current_token_data = NULL;
+
+	/* Prepare to (recursively) watch the <worktree-root> directory. */
+	strbuf_init(&state.path_worktree_watch, 0);
+	strbuf_addstr(&state.path_worktree_watch, absolute_path(get_git_work_tree()));
+	state.nr_paths_watching = 1;
+
+	/*
+	 * We create and delete cookie files somewhere inside the .git
+	 * directory to help us keep sync with the file system.  If
+	 * ".git" is not a directory, then <gitdir> is not inside the
+	 * cone of <worktree-root>, so set up a second watch to watch
+	 * the <gitdir> so that we get events for the cookie files.
+	 */
+	strbuf_init(&state.path_gitdir_watch, 0);
+	strbuf_addbuf(&state.path_gitdir_watch, &state.path_worktree_watch);
+	strbuf_addstr(&state.path_gitdir_watch, "/.git");
+	if (!is_directory(state.path_gitdir_watch.buf)) {
+		strbuf_reset(&state.path_gitdir_watch);
+		strbuf_addstr(&state.path_gitdir_watch, absolute_path(get_git_dir()));
+		state.nr_paths_watching = 2;
+	}
+
+	/*
+	 * Confirm that we can create platform-specific resources for the
+	 * filesystem listener before we bother starting all the threads.
+	 */
+	if (fsm_listen__ctor(&state)) {
+		err = error(_("could not initialize listener thread"));
+		goto done;
+	}
+
+	err = fsmonitor_run_daemon_1(&state);
+
+done:
+	pthread_mutex_destroy(&state.main_lock);
+	fsm_listen__dtor(&state);
+
+	ipc_server_free(state.ipc_server_data);
+
+	strbuf_release(&state.path_worktree_watch);
+	strbuf_release(&state.path_gitdir_watch);
+
+	return err;
+}
+
+static int try_to_run_foreground_daemon(void)
+{
+	/*
+	 * Technically, we don't need to probe for an existing daemon
+	 * process, since we could just call `fsmonitor_run_daemon()`
+	 * and let it fail if the pipe/socket is busy.
+	 *
+	 * However, this method gives us a nicer error message for a
+	 * common error case.
+	 */
+	if (fsmonitor_ipc__get_state() == IPC_STATE__LISTENING)
+		die(_("fsmonitor--daemon is already running '%s'"),
+		    the_repository->worktree);
+
+	if (fsmonitor__announce_startup) {
+		fprintf(stderr, _("running fsmonitor-daemon in '%s'\n"),
+			the_repository->worktree);
+		fflush(stderr);
+	}
+
+	return !!fsmonitor_run_daemon();
+}
+
 int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
 {
 	const char *subcmd;
 
 	struct option options[] = {
+		OPT_INTEGER(0, "ipc-threads",
+			    &fsmonitor__ipc_threads,
+			    N_("use <n> ipc worker threads")),
 		OPT_END()
 	};
 
-	git_config(git_default_config, NULL);
+	git_config(fsmonitor_config, NULL);
 
 	argc = parse_options(argc, argv, prefix, options,
 			     builtin_fsmonitor__daemon_usage, 0);
@@ -73,6 +292,13 @@ int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
 		usage_with_options(builtin_fsmonitor__daemon_usage, options);
 	subcmd = argv[0];
 
+	if (fsmonitor__ipc_threads < 1)
+		die(_("invalid 'ipc-threads' value (%d)"),
+		    fsmonitor__ipc_threads);
+
+	if (!strcmp(subcmd, "run"))
+		return !!try_to_run_foreground_daemon();
+
 	if (!strcmp(subcmd, "stop"))
 		return !!do_as_client__send_stop();
 
diff --git a/fsmonitor--daemon.h b/fsmonitor--daemon.h
new file mode 100644
index 00000000000..3009c1a83de
--- /dev/null
+++ b/fsmonitor--daemon.h
@@ -0,0 +1,34 @@
+#ifndef FSMONITOR_DAEMON_H
+#define FSMONITOR_DAEMON_H
+
+#ifdef HAVE_FSMONITOR_DAEMON_BACKEND
+
+#include "cache.h"
+#include "dir.h"
+#include "run-command.h"
+#include "simple-ipc.h"
+#include "thread-utils.h"
+
+struct fsmonitor_batch;
+struct fsmonitor_token_data;
+
+struct fsmonitor_daemon_backend_data; /* opaque platform-specific data */
+
+struct fsmonitor_daemon_state {
+	pthread_t listener_thread;
+	pthread_mutex_t main_lock;
+
+	struct strbuf path_worktree_watch;
+	struct strbuf path_gitdir_watch;
+	int nr_paths_watching;
+
+	struct fsmonitor_token_data *current_token_data;
+
+	int error_code;
+	struct fsmonitor_daemon_backend_data *backend_data;
+
+	struct ipc_server_data *ipc_server_data;
+};
+
+#endif /* HAVE_FSMONITOR_DAEMON_BACKEND */
+#endif /* FSMONITOR_DAEMON_H */
-- 
gitgitgadget

[PATCH v7 11/29] fsmonitor--daemon: implement 'start' command

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Implement 'git fsmonitor--daemon start' command.  This command starts
an instance of 'git fsmonitor--daemon run' in the background using
the new 'start_bg_command()' function.

We avoid the fork-and-call technique on Unix systems in favor of a
fork-and-exec technique.  This gives us more uniform Trace2 child-*
events.  It also makes our usage more consistent with Windows usage.

On Windows, teach 'git fsmonitor--daemon run' to optionally call
'FreeConsole()' to release handles to the inherited Win32 console
(despite being passed invalid handles for stdin/out/err).  Without
this, command prompts and powershell terminal windows could hang
in "exit" until the last background child process exited or released
their Win32 console handle.  (This was not seen with git-bash shells
because they don't have a Win32 console attached to them.)

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 builtin/fsmonitor--daemon.c | 109 +++++++++++++++++++++++++++++++++++-
 1 file changed, 107 insertions(+), 2 deletions(-)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index 5591339399a..69dd39121a3 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -9,6 +9,7 @@
 #include "khash.h"
 
 static const char * const builtin_fsmonitor__daemon_usage[] = {
+	N_("git fsmonitor--daemon start [<options>]"),
 	N_("git fsmonitor--daemon run [<options>]"),
 	N_("git fsmonitor--daemon stop"),
 	N_("git fsmonitor--daemon status"),
@@ -22,6 +23,9 @@ static const char * const builtin_fsmonitor__daemon_usage[] = {
 #define FSMONITOR__IPC_THREADS "fsmonitor.ipcthreads"
 static int fsmonitor__ipc_threads = 8;
 
+#define FSMONITOR__START_TIMEOUT "fsmonitor.starttimeout"
+static int fsmonitor__start_timeout_sec = 60;
+
 #define FSMONITOR__ANNOUNCE_STARTUP "fsmonitor.announcestartup"
 static int fsmonitor__announce_startup = 0;
 
@@ -36,6 +40,15 @@ static int fsmonitor_config(const char *var, const char *value, void *cb)
 		return 0;
 	}
 
+	if (!strcmp(var, FSMONITOR__START_TIMEOUT)) {
+		int i = git_config_int(var, value);
+		if (i < 0)
+			return error(_("value of '%s' out of range: %d"),
+				     FSMONITOR__START_TIMEOUT, i);
+		fsmonitor__start_timeout_sec = i;
+		return 0;
+	}
+
 	if (!strcmp(var, FSMONITOR__ANNOUNCE_STARTUP)) {
 		int is_bool;
 		int i = git_config_bool_or_int(var, value, &is_bool);
@@ -250,7 +263,7 @@ done:
 	return err;
 }
 
-static int try_to_run_foreground_daemon(void)
+static int try_to_run_foreground_daemon(int detach_console)
 {
 	/*
 	 * Technically, we don't need to probe for an existing daemon
@@ -270,17 +283,106 @@ static int try_to_run_foreground_daemon(void)
 		fflush(stderr);
 	}
 
+#ifdef GIT_WINDOWS_NATIVE
+	if (detach_console)
+		FreeConsole();
+#endif
+
 	return !!fsmonitor_run_daemon();
 }
 
+static start_bg_wait_cb bg_wait_cb;
+
+static int bg_wait_cb(const struct child_process *cp, void *cb_data)
+{
+	enum ipc_active_state s = fsmonitor_ipc__get_state();
+
+	switch (s) {
+	case IPC_STATE__LISTENING:
+		/* child is "ready" */
+		return 0;
+
+	case IPC_STATE__NOT_LISTENING:
+	case IPC_STATE__PATH_NOT_FOUND:
+		/* give child more time */
+		return 1;
+
+	default:
+	case IPC_STATE__INVALID_PATH:
+	case IPC_STATE__OTHER_ERROR:
+		/* all the time in world won't help */
+		return -1;
+	}
+}
+
+static int try_to_start_background_daemon(void)
+{
+	struct child_process cp = CHILD_PROCESS_INIT;
+	enum start_bg_result sbgr;
+
+	/*
+	 * Before we try to create a background daemon process, see
+	 * if a daemon process is already listening.  This makes it
+	 * easier for us to report an already-listening error to the
+	 * console, since our spawn/daemon can only report the success
+	 * of creating the background process (and not whether it
+	 * immediately exited).
+	 */
+	if (fsmonitor_ipc__get_state() == IPC_STATE__LISTENING)
+		die(_("fsmonitor--daemon is already running '%s'"),
+		    the_repository->worktree);
+
+	if (fsmonitor__announce_startup) {
+		fprintf(stderr, _("starting fsmonitor-daemon in '%s'\n"),
+			the_repository->worktree);
+		fflush(stderr);
+	}
+
+	cp.git_cmd = 1;
+
+	strvec_push(&cp.args, "fsmonitor--daemon");
+	strvec_push(&cp.args, "run");
+	strvec_push(&cp.args, "--detach");
+	strvec_pushf(&cp.args, "--ipc-threads=%d", fsmonitor__ipc_threads);
+
+	cp.no_stdin = 1;
+	cp.no_stdout = 1;
+	cp.no_stderr = 1;
+
+	sbgr = start_bg_command(&cp, bg_wait_cb, NULL,
+				fsmonitor__start_timeout_sec);
+
+	switch (sbgr) {
+	case SBGR_READY:
+		return 0;
+
+	default:
+	case SBGR_ERROR:
+	case SBGR_CB_ERROR:
+		return error(_("daemon failed to start"));
+
+	case SBGR_TIMEOUT:
+		return error(_("daemon not online yet"));
+
+	case SBGR_DIED:
+		return error(_("daemon terminated"));
+	}
+}
+
 int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
 {
 	const char *subcmd;
+	int detach_console = 0;
 
 	struct option options[] = {
+		OPT_BOOL(0, "detach", &detach_console, N_("detach from console")),
 		OPT_INTEGER(0, "ipc-threads",
 			    &fsmonitor__ipc_threads,
 			    N_("use <n> ipc worker threads")),
+		OPT_INTEGER(0, "start-timeout",
+			    &fsmonitor__start_timeout_sec,
+			    N_("max seconds to wait for background daemon startup")),
+
 		OPT_END()
 	};
 
@@ -296,8 +398,11 @@ int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
 		die(_("invalid 'ipc-threads' value (%d)"),
 		    fsmonitor__ipc_threads);
 
+	if (!strcmp(subcmd, "start"))
+		return !!try_to_start_background_daemon();
+
 	if (!strcmp(subcmd, "run"))
-		return !!try_to_run_foreground_daemon();
+		return !!try_to_run_foreground_daemon(detach_console);
 
 	if (!strcmp(subcmd, "stop"))
 		return !!do_as_client__send_stop();
-- 
gitgitgadget

[PATCH v7 12/29] fsmonitor--daemon: add pathname classification

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach fsmonitor--daemon to classify relative and absolute
pathnames and decide how they should be handled.  This will
be used by the platform-specific backend to respond to each
filesystem event.

When we register for filesystem notifications on a directory,
we get events for everything (recursively) in the directory.
We want to report to clients changes to tracked and untracked
paths within the working directory proper.  We do not want to
report changes within the .git directory, for example.

This classification will be used in a later commit by the
different backends to classify paths as events are received.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 builtin/fsmonitor--daemon.c | 81 ++++++++++++++++++++++++++++++++++
 fsmonitor--daemon.h         | 87 +++++++++++++++++++++++++++++++++++++
 2 files changed, 168 insertions(+)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index 69dd39121a3..1ce00b7c150 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -134,6 +134,87 @@ static int handle_client(void *data,
 	return result;
 }
 
+#define FSMONITOR_COOKIE_PREFIX ".fsmonitor-daemon-"
+
+enum fsmonitor_path_type fsmonitor_classify_path_workdir_relative(
+	const char *rel)
+{
+	if (fspathncmp(rel, ".git", 4))
+		return IS_WORKDIR_PATH;
+	rel += 4;
+
+	if (!*rel)
+		return IS_DOT_GIT;
+	if (*rel != '/')
+		return IS_WORKDIR_PATH; /* e.g. .gitignore */
+	rel++;
+
+	if (!fspathncmp(rel, FSMONITOR_COOKIE_PREFIX,
+			strlen(FSMONITOR_COOKIE_PREFIX)))
+		return IS_INSIDE_DOT_GIT_WITH_COOKIE_PREFIX;
+
+	return IS_INSIDE_DOT_GIT;
+}
+
+enum fsmonitor_path_type fsmonitor_classify_path_gitdir_relative(
+	const char *rel)
+{
+	if (!fspathncmp(rel, FSMONITOR_COOKIE_PREFIX,
+			strlen(FSMONITOR_COOKIE_PREFIX)))
+		return IS_INSIDE_GITDIR_WITH_COOKIE_PREFIX;
+
+	return IS_INSIDE_GITDIR;
+}
+
+static enum fsmonitor_path_type try_classify_workdir_abs_path(
+	struct fsmonitor_daemon_state *state,
+	const char *path)
+{
+	const char *rel;
+
+	if (fspathncmp(path, state->path_worktree_watch.buf,
+		       state->path_worktree_watch.len))
+		return IS_OUTSIDE_CONE;
+
+	rel = path + state->path_worktree_watch.len;
+
+	if (!*rel)
+		return IS_WORKDIR_PATH; /* it is the root dir exactly */
+	if (*rel != '/')
+		return IS_OUTSIDE_CONE;
+	rel++;
+
+	return fsmonitor_classify_path_workdir_relative(rel);
+}
+
+enum fsmonitor_path_type fsmonitor_classify_path_absolute(
+	struct fsmonitor_daemon_state *state,
+	const char *path)
+{
+	const char *rel;
+	enum fsmonitor_path_type t;
+
+	t = try_classify_workdir_abs_path(state, path);
+	if (state->nr_paths_watching == 1)
+		return t;
+	if (t != IS_OUTSIDE_CONE)
+		return t;
+
+	if (fspathncmp(path, state->path_gitdir_watch.buf,
+		       state->path_gitdir_watch.len))
+		return IS_OUTSIDE_CONE;
+
+	rel = path + state->path_gitdir_watch.len;
+
+	if (!*rel)
+		return IS_GITDIR; /* it is the <gitdir> exactly */
+	if (*rel != '/')
+		return IS_OUTSIDE_CONE;
+	rel++;
+
+	return fsmonitor_classify_path_gitdir_relative(rel);
+}
+
 static void *fsm_listen__thread_proc(void *_state)
 {
 	struct fsmonitor_daemon_state *state = _state;
diff --git a/fsmonitor--daemon.h b/fsmonitor--daemon.h
index 3009c1a83de..8c3a71a48bd 100644
--- a/fsmonitor--daemon.h
+++ b/fsmonitor--daemon.h
@@ -30,5 +30,92 @@ struct fsmonitor_daemon_state {
 	struct ipc_server_data *ipc_server_data;
 };
 
+/*
+ * Pathname classifications.
+ *
+ * The daemon classifies the pathnames that it receives from file
+ * system notification events into the following categories and uses
+ * that to decide whether clients are told about them.  (And to watch
+ * for file system synchronization events.)
+ *
+ * The daemon only collects and reports on the set of modified paths
+ * within the working directory (proper).
+ *
+ * The client should only care about paths within the working
+ * directory proper (inside the working directory and not ".git" nor
+ * inside of ".git/").  That is, the client has read the index and is
+ * asking for a list of any paths in the working directory that have
+ * been modified since the last token.  The client does not care about
+ * file system changes within the ".git/" directory (such as new loose
+ * objects or packfiles).  So the client will only receive paths that
+ * are classified as IS_WORKDIR_PATH.
+ *
+ * Note that ".git" is usually a directory and is therefore inside
+ * the cone of the FS watch that we have on the working directory root,
+ * so we will also get FS events for disk activity on and within ".git/"
+ * that we need to respond to or filter from the client.
+ *
+ * But Git also allows ".git" to be a *file* that points to a GITDIR
+ * outside of the working directory.  When this happens, we need to
+ * create FS watches on both the working directory root *and* on the
+ * (external) GITDIR root.  (The latter is required because we put
+ * cookie files inside it and use them to sync with the FS event
+ * stream.)
+ *
+ * Note that in the context of this discussion, I'm using "GITDIR"
+ * to only mean an external GITDIR referenced by a ".git" file.
+ *
+ * The platform FS event backends will receive watch-specific
+ * relative paths (except for those OS's that always emit absolute
+ * paths).  We use the following enum and routines to classify each
+ * path so that we know how to handle it.  There is a slight asymmetry
+ * here because ".git/" is inside the working directory and the
+ * (external) GITDIR is not, and therefore how we handle events may
+ * vary slightly, so I have different enums for "IS...DOT_GIT..." and
+ * "IS...GITDIR...".
+ *
+ * The daemon uses the IS_DOT_GIT and IS_GITDIR internally to mean the
+ * exact ".git" file/directory or GITDIR directory.  If the daemon
+ * receives a delete event for either of these paths, it will
+ * automatically shutdown, for example.
+ *
+ * Note that the daemon DOES NOT explicitly watch nor special case the
+ * index.  The daemon does not read the index nor have any internal
+ * index-relative state, so there are no "IS...INDEX..." enum values.
+ */
+enum fsmonitor_path_type {
+	IS_WORKDIR_PATH = 0,
+
+	IS_DOT_GIT,
+	IS_INSIDE_DOT_GIT,
+	IS_INSIDE_DOT_GIT_WITH_COOKIE_PREFIX,
+
+	IS_GITDIR,
+	IS_INSIDE_GITDIR,
+	IS_INSIDE_GITDIR_WITH_COOKIE_PREFIX,
+
+	IS_OUTSIDE_CONE,
+};
+
+/*
+ * Classify a pathname relative to the root of the working directory.
+ */
+enum fsmonitor_path_type fsmonitor_classify_path_workdir_relative(
+	const char *relative_path);
+
+/*
+ * Classify a pathname relative to a <gitdir> that is external to the
+ * worktree directory.
+ */
+enum fsmonitor_path_type fsmonitor_classify_path_gitdir_relative(
+	const char *relative_path);
+
+/*
+ * Classify an absolute pathname received from a filesystem event.
+ */
+enum fsmonitor_path_type fsmonitor_classify_path_absolute(
+	struct fsmonitor_daemon_state *state,
+	const char *path);
+
 #endif /* HAVE_FSMONITOR_DAEMON_BACKEND */
 #endif /* FSMONITOR_DAEMON_H */
-- 
gitgitgadget

[PATCH v7 13/29] fsmonitor--daemon: define token-ids

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach fsmonitor--daemon to create token-ids and define the
overall token naming scheme.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 builtin/fsmonitor--daemon.c | 116 +++++++++++++++++++++++++++++++++++-
 1 file changed, 115 insertions(+), 1 deletion(-)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index 1ce00b7c150..1c7c156288d 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -106,6 +106,120 @@ static int do_as_client__status(void)
 	}
 }
 
+/*
+ * Requests to and from a FSMonitor Protocol V2 provider use an opaque
+ * "token" as a virtual timestamp.  Clients can request a summary of all
+ * created/deleted/modified files relative to a token.  In the response,
+ * clients receive a new token for the next (relative) request.
+ *
+ *
+ * Token Format
+ * ============
+ *
+ * The contents of the token are private and provider-specific.
+ *
+ * For the built-in fsmonitor--daemon, we define a token as follows:
+ *
+ *     "builtin" ":" <token_id> ":" <sequence_nr>
+ *
+ * The "builtin" prefix is used as a namespace to avoid conflicts
+ * with other providers (such as Watchman).
+ *
+ * The <token_id> is an arbitrary OPAQUE string, such as a GUID,
+ * UUID, or {timestamp,pid}.  It is used to group all filesystem
+ * events that happened while the daemon was monitoring (and in-sync
+ * with the filesystem).
+ *
+ *     Unlike FSMonitor Protocol V1, it is not defined as a timestamp
+ *     and does not define less-than/greater-than relationships.
+ *     (There are too many race conditions to rely on file system
+ *     event timestamps.)
+ *
+ * The <sequence_nr> is a simple integer incremented whenever the
+ * daemon needs to make its state public.  For example, if 1000 file
+ * system events come in, but no clients have requested the data,
+ * the daemon can continue to accumulate file changes in the same
+ * bin and does not need to advance the sequence number.  However,
+ * as soon as a client does arrive, the daemon needs to start a new
+ * bin and increment the sequence number.
+ *
+ *     The sequence number serves as the boundary between 2 sets
+ *     of bins -- the older ones that the client has already seen
+ *     and the newer ones that it hasn't.
+ *
+ * When a new <token_id> is created, the <sequence_nr> is reset to
+ * zero.
+ *
+ *
+ * About Token Ids
+ * ===============
+ *
+ * A new token_id is created:
+ *
+ * [1] each time the daemon is started.
+ *
+ * [2] any time that the daemon must re-sync with the filesystem
+ *     (such as when the kernel drops or we miss events on a very
+ *     active volume).
+ *
+ * [3] in response to a client "flush" command (for dropped event
+ *     testing).
+ *
+ * When a new token_id is created, the daemon is free to discard all
+ * cached filesystem events associated with any previous token_ids.
+ * Events associated with a non-current token_id will never be sent
+ * to a client.  A token_id change implicitly means that the daemon
+ * has gap in its event history.
+ *
+ * Therefore, clients that present a token with a stale (non-current)
+ * token_id will always be given a trivial response.
+ */
+struct fsmonitor_token_data {
+	struct strbuf token_id;
+	struct fsmonitor_batch *batch_head;
+	struct fsmonitor_batch *batch_tail;
+	uint64_t client_ref_count;
+};
+
+static struct fsmonitor_token_data *fsmonitor_new_token_data(void)
+{
+	static int test_env_value = -1;
+	static uint64_t flush_count = 0;
+	struct fsmonitor_token_data *token;
+
+	CALLOC_ARRAY(token, 1);
+
+	strbuf_init(&token->token_id, 0);
+	token->batch_head = NULL;
+	token->batch_tail = NULL;
+	token->client_ref_count = 0;
+
+	if (test_env_value < 0)
+		test_env_value = git_env_bool("GIT_TEST_FSMONITOR_TOKEN", 0);
+
+	if (!test_env_value) {
+		struct timeval tv;
+		struct tm tm;
+		time_t secs;
+
+		gettimeofday(&tv, NULL);
+		secs = tv.tv_sec;
+		gmtime_r(&secs, &tm);
+
+		strbuf_addf(&token->token_id,
+			    "%"PRIu64".%d.%4d%02d%02dT%02d%02d%02d.%06ldZ",
+			    flush_count++,
+			    getpid(),
+			    tm.tm_year + 1900, tm.tm_mon + 1, tm.tm_mday,
+			    tm.tm_hour, tm.tm_min, tm.tm_sec,
+			    (long)tv.tv_usec);
+	} else {
+		strbuf_addf(&token->token_id, "test_%08x", test_env_value++);
+	}
+
+	return token;
+}
+
 static ipc_server_application_cb handle_client;
 
 static int handle_client(void *data,
@@ -298,7 +412,7 @@ static int fsmonitor_run_daemon(void)
 
 	pthread_mutex_init(&state.main_lock, NULL);
 	state.error_code = 0;
-	state.current_token_data = NULL;
+	state.current_token_data = fsmonitor_new_token_data();
 
 	/* Prepare to (recursively) watch the <worktree-root> directory. */
 	strbuf_init(&state.path_worktree_watch, 0);
-- 
gitgitgadget

[PATCH v7 14/29] fsmonitor--daemon: create token-based changed path cache

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach fsmonitor--daemon to build a list of changed paths and associate
them with a token-id.  This will be used by the platform-specific
backends to accumulate changed paths in response to filesystem events.

The platform-specific file system listener thread receives file system
events containing one or more changed pathnames (with whatever
bucketing or grouping that is convenient for the file system).  These
paths are accumulated (without locking) by the file system layer into
a `fsmonitor_batch`.

When the file system layer has drained the kernel event queue, it will
"publish" them to our token queue and make them visible to concurrent
client worker threads.  The token layer is free to combine and/or de-dup
paths within these batches for efficient presentation to clients.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 builtin/fsmonitor--daemon.c | 230 +++++++++++++++++++++++++++++++++++-
 fsmonitor--daemon.h         |  40 +++++++
 2 files changed, 268 insertions(+), 2 deletions(-)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index 1c7c156288d..69312119b07 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -181,17 +181,27 @@ struct fsmonitor_token_data {
 	uint64_t client_ref_count;
 };
 
+struct fsmonitor_batch {
+	struct fsmonitor_batch *next;
+	uint64_t batch_seq_nr;
+	const char **interned_paths;
+	size_t nr, alloc;
+	time_t pinned_time;
+};
+
 static struct fsmonitor_token_data *fsmonitor_new_token_data(void)
 {
 	static int test_env_value = -1;
 	static uint64_t flush_count = 0;
 	struct fsmonitor_token_data *token;
+	struct fsmonitor_batch *batch;
 
 	CALLOC_ARRAY(token, 1);
+	batch = fsmonitor_batch__new();
 
 	strbuf_init(&token->token_id, 0);
-	token->batch_head = NULL;
-	token->batch_tail = NULL;
+	token->batch_head = batch;
+	token->batch_tail = batch;
 	token->client_ref_count = 0;
 
 	if (test_env_value < 0)
@@ -217,9 +227,143 @@ static struct fsmonitor_token_data *fsmonitor_new_token_data(void)
 		strbuf_addf(&token->token_id, "test_%08x", test_env_value++);
 	}
 
+	/*
+	 * We created a new <token_id> and are starting a new series
+	 * of tokens with a zero <seq_nr>.
+	 *
+	 * Since clients cannot guess our new (non test) <token_id>
+	 * they will always receive a trivial response (because of the
+	 * mismatch on the <token_id>).  The trivial response will
+	 * tell them our new <token_id> so that subsequent requests
+	 * will be relative to our new series.  (And when sending that
+	 * response, we pin the current head of the batch list.)
+	 *
+	 * Even if the client correctly guesses the <token_id>, their
+	 * request of "builtin:<token_id>:0" asks for all changes MORE
+	 * RECENT than batch/bin 0.
+	 *
+	 * This implies that it is a waste to accumulate paths in the
+	 * initial batch/bin (because they will never be transmitted).
+	 *
+	 * So the daemon could be running for days and watching the
+	 * file system, but doesn't need to actually accumulate any
+	 * paths UNTIL we need to set a reference point for a later
+	 * relative request.
+	 *
+	 * However, it is very useful for testing to always have a
+	 * reference point set.  Pin batch 0 to force early file system
+	 * events to accumulate.
+	 */
+	if (test_env_value)
+		batch->pinned_time = time(NULL);
+
 	return token;
 }
 
+struct fsmonitor_batch *fsmonitor_batch__new(void)
+{
+	struct fsmonitor_batch *batch;
+
+	CALLOC_ARRAY(batch, 1);
+
+	return batch;
+}
+
+void fsmonitor_batch__free_list(struct fsmonitor_batch *batch)
+{
+	while (batch) {
+		struct fsmonitor_batch *next = batch->next;
+
+		/*
+		 * The actual strings within the array of this batch
+		 * are interned, so we don't own them.  We only own
+		 * the array.
+		 */
+		free(batch->interned_paths);
+		free(batch);
+
+		batch = next;
+	}
+}
+
+void fsmonitor_batch__add_path(struct fsmonitor_batch *batch,
+			       const char *path)
+{
+	const char *interned_path = strintern(path);
+
+	trace_printf_key(&trace_fsmonitor, "event: %s", interned_path);
+
+	ALLOC_GROW(batch->interned_paths, batch->nr + 1, batch->alloc);
+	batch->interned_paths[batch->nr++] = interned_path;
+}
+
+static void fsmonitor_batch__combine(struct fsmonitor_batch *batch_dest,
+				     const struct fsmonitor_batch *batch_src)
+{
+	size_t k;
+
+	ALLOC_GROW(batch_dest->interned_paths,
+		   batch_dest->nr + batch_src->nr + 1,
+		   batch_dest->alloc);
+
+	for (k = 0; k < batch_src->nr; k++)
+		batch_dest->interned_paths[batch_dest->nr++] =
+			batch_src->interned_paths[k];
+}
+
+static void fsmonitor_free_token_data(struct fsmonitor_token_data *token)
+{
+	if (!token)
+		return;
+
+	assert(token->client_ref_count == 0);
+
+	strbuf_release(&token->token_id);
+
+	fsmonitor_batch__free_list(token->batch_head);
+
+	free(token);
+}
+
+/*
+ * Flush all of our cached data about the filesystem.  Call this if we
+ * lose sync with the filesystem and miss some notification events.
+ *
+ * [1] If we are missing events, then we no longer have a complete
+ *     history of the directory (relative to our current start token).
+ *     We should create a new token and start fresh (as if we just
+ *     booted up).
+ *
+ * If there are no concurrent threads reading the current token data
+ * series, we can free it now.  Otherwise, let the last reader free
+ * it.
+ *
+ * Either way, the old token data series is no longer associated with
+ * our state data.
+ */
+static void with_lock__do_force_resync(struct fsmonitor_daemon_state *state)
+{
+	/* assert current thread holding state->main_lock */
+
+	struct fsmonitor_token_data *free_me = NULL;
+	struct fsmonitor_token_data *new_one = NULL;
+
+	new_one = fsmonitor_new_token_data();
+
+	if (state->current_token_data->client_ref_count == 0)
+		free_me = state->current_token_data;
+	state->current_token_data = new_one;
+
+	fsmonitor_free_token_data(free_me);
+}
+
+void fsmonitor_force_resync(struct fsmonitor_daemon_state *state)
+{
+	pthread_mutex_lock(&state->main_lock);
+	with_lock__do_force_resync(state);
+	pthread_mutex_unlock(&state->main_lock);
+}
+
 static ipc_server_application_cb handle_client;
 
 static int handle_client(void *data,
@@ -329,6 +473,81 @@ enum fsmonitor_path_type fsmonitor_classify_path_absolute(
 	return fsmonitor_classify_path_gitdir_relative(rel);
 }
 
+/*
+ * We try to combine small batches at the front of the batch-list to avoid
+ * having a long list.  This hopefully makes it a little easier when we want
+ * to truncate and maintain the list.  However, we don't want the paths array
+ * to just keep growing and growing with realloc, so we insert an arbitrary
+ * limit.
+ */
+#define MY_COMBINE_LIMIT (1024)
+
+void fsmonitor_publish(struct fsmonitor_daemon_state *state,
+		       struct fsmonitor_batch *batch,
+		       const struct string_list *cookie_names)
+{
+	if (!batch && !cookie_names->nr)
+		return;
+
+	pthread_mutex_lock(&state->main_lock);
+
+	if (batch) {
+		struct fsmonitor_batch *head;
+
+		head = state->current_token_data->batch_head;
+		if (!head) {
+			BUG("token does not have batch");
+		} else if (head->pinned_time) {
+			/*
+			 * We cannot alter the current batch list
+			 * because:
+			 *
+			 * [a] it is being transmitted to at least one
+			 * client and the handle_client() thread has a
+			 * ref-count, but not a lock on the batch list
+			 * starting with this item.
+			 *
+			 * [b] it has been transmitted in the past to
+			 * at least one client such that future
+			 * requests are relative to this head batch.
+			 *
+			 * So, we can only prepend a new batch onto
+			 * the front of the list.
+			 */
+			batch->batch_seq_nr = head->batch_seq_nr + 1;
+			batch->next = head;
+			state->current_token_data->batch_head = batch;
+		} else if (!head->batch_seq_nr) {
+			/*
+			 * Batch 0 is unpinned.  See the note in
+			 * `fsmonitor_new_token_data()` about why we
+			 * don't need to accumulate these paths.
+			 */
+			fsmonitor_batch__free_list(batch);
+		} else if (head->nr + batch->nr > MY_COMBINE_LIMIT) {
+			/*
+			 * The head batch in the list has never been
+			 * transmitted to a client, but folding the
+			 * contents of the new batch onto it would
+			 * exceed our arbitrary limit, so just prepend
+			 * the new batch onto the list.
+			 */
+			batch->batch_seq_nr = head->batch_seq_nr + 1;
+			batch->next = head;
+			state->current_token_data->batch_head = batch;
+		} else {
+			/*
+			 * We are free to add the paths in the given
+			 * batch onto the end of the current head batch.
+			 */
+			fsmonitor_batch__combine(head, batch);
+			fsmonitor_batch__free_list(batch);
+		}
+	}
+
+	pthread_mutex_unlock(&state->main_lock);
+}
+
 static void *fsm_listen__thread_proc(void *_state)
 {
 	struct fsmonitor_daemon_state *state = _state;
@@ -343,6 +562,13 @@ static void *fsm_listen__thread_proc(void *_state)
 
 	fsm_listen__loop(state);
 
+	pthread_mutex_lock(&state->main_lock);
+	if (state->current_token_data &&
+	    state->current_token_data->client_ref_count == 0)
+		fsmonitor_free_token_data(state->current_token_data);
+	state->current_token_data = NULL;
+	pthread_mutex_unlock(&state->main_lock);
+
 	trace2_thread_exit();
 	return NULL;
 }
diff --git a/fsmonitor--daemon.h b/fsmonitor--daemon.h
index 8c3a71a48bd..010fbfe60e9 100644
--- a/fsmonitor--daemon.h
+++ b/fsmonitor--daemon.h
@@ -12,6 +12,27 @@
 struct fsmonitor_batch;
 struct fsmonitor_token_data;
 
+/*
+ * Create a new batch of path(s).  The returned batch is considered
+ * private and not linked into the fsmonitor daemon state.  The caller
+ * should fill this batch with one or more paths and then publish it.
+ */
+struct fsmonitor_batch *fsmonitor_batch__new(void);
+
+/*
+ * Free the list of batches starting with this one.
+ */
+void fsmonitor_batch__free_list(struct fsmonitor_batch *batch);
+
+/*
+ * Add this path to this batch of modified files.
+ *
+ * The batch should be private and NOT (yet) linked into the fsmonitor
+ * daemon state and therefore not yet visible to worker threads and so
+ * no locking is required.
+ */
+void fsmonitor_batch__add_path(struct fsmonitor_batch *batch, const char *path);
+
 struct fsmonitor_daemon_backend_data; /* opaque platform-specific data */
 
 struct fsmonitor_daemon_state {
@@ -117,5 +138,24 @@ enum fsmonitor_path_type fsmonitor_classify_path_absolute(
 	struct fsmonitor_daemon_state *state,
 	const char *path);
 
+/*
+ * Prepend the this batch of path(s) onto the list of batches associated
+ * with the current token.  This makes the batch visible to worker threads.
+ *
+ * The caller no longer owns the batch and must not free it.
+ *
+ * Wake up the client threads waiting on these cookies.
+ */
+void fsmonitor_publish(struct fsmonitor_daemon_state *state,
+		       struct fsmonitor_batch *batch,
+		       const struct string_list *cookie_names);
+
+/*
+ * If the platform-specific layer loses sync with the filesystem,
+ * it should call this to invalidate cached data and abort waiting
+ * threads.
+ */
+void fsmonitor_force_resync(struct fsmonitor_daemon_state *state);
+
 #endif /* HAVE_FSMONITOR_DAEMON_BACKEND */
 #endif /* FSMONITOR_DAEMON_H */
-- 
gitgitgadget

[PATCH v7 15/29] compat/fsmonitor/fsm-listen-win32: implement FSMonitor backend on Windows

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach the win32 backend to register a watch on the working tree
root directory (recursively).  Also watch the <gitdir> if it is
not inside the working tree.  And to collect path change notifications
into batches and publish.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 compat/fsmonitor/fsm-listen-win32.c | 565 ++++++++++++++++++++++++++++
 1 file changed, 565 insertions(+)

diff --git a/compat/fsmonitor/fsm-listen-win32.c b/compat/fsmonitor/fsm-listen-win32.c
index 916cbea254f..5b928ab66e5 100644
--- a/compat/fsmonitor/fsm-listen-win32.c
+++ b/compat/fsmonitor/fsm-listen-win32.c
@@ -2,20 +2,585 @@
 #include "config.h"
 #include "fsmonitor.h"
 #include "fsm-listen.h"
+#include "fsmonitor--daemon.h"
+
+/*
+ * The documentation of ReadDirectoryChangesW() states that the maximum
+ * buffer size is 64K when the monitored directory is remote.
+ *
+ * Larger buffers may be used when the monitored directory is local and
+ * will help us receive events faster from the kernel and avoid dropped
+ * events.
+ *
+ * So we try to use a very large buffer and silently fallback to 64K if
+ * we get an error.
+ */
+#define MAX_RDCW_BUF_FALLBACK (65536)
+#define MAX_RDCW_BUF          (65536 * 8)
+
+struct one_watch
+{
+	char buffer[MAX_RDCW_BUF];
+	DWORD buf_len;
+	DWORD count;
+
+	struct strbuf path;
+	HANDLE hDir;
+	HANDLE hEvent;
+	OVERLAPPED overlapped;
+
+	/*
+	 * Is there an active ReadDirectoryChangesW() call pending.  If so, we
+	 * need to later call GetOverlappedResult() and possibly CancelIoEx().
+	 */
+	BOOL is_active;
+};
+
+struct fsmonitor_daemon_backend_data
+{
+	struct one_watch *watch_worktree;
+	struct one_watch *watch_gitdir;
+
+	HANDLE hEventShutdown;
+
+	HANDLE hListener[3]; /* we don't own these handles */
+#define LISTENER_SHUTDOWN 0
+#define LISTENER_HAVE_DATA_WORKTREE 1
+#define LISTENER_HAVE_DATA_GITDIR 2
+	int nr_listener_handles;
+};
+
+/*
+ * Convert the WCHAR path from the notification into UTF8 and
+ * then normalize it.
+ */
+static int normalize_path_in_utf8(FILE_NOTIFY_INFORMATION *info,
+				  struct strbuf *normalized_path)
+{
+	int reserve;
+	int len = 0;
+
+	strbuf_reset(normalized_path);
+	if (!info->FileNameLength)
+		goto normalize;
+
+	/*
+	 * Pre-reserve enough space in the UTF8 buffer for
+	 * each Unicode WCHAR character to be mapped into a
+	 * sequence of 2 UTF8 characters.  That should let us
+	 * avoid ERROR_INSUFFICIENT_BUFFER 99.9+% of the time.
+	 */
+	reserve = info->FileNameLength + 1;
+	strbuf_grow(normalized_path, reserve);
+
+	for (;;) {
+		len = WideCharToMultiByte(CP_UTF8, 0, info->FileName,
+					  info->FileNameLength / sizeof(WCHAR),
+					  normalized_path->buf,
+					  strbuf_avail(normalized_path) - 1,
+					  NULL, NULL);
+		if (len > 0)
+			goto normalize;
+		if (GetLastError() != ERROR_INSUFFICIENT_BUFFER) {
+			error(_("[GLE %ld] could not convert path to UTF-8: '%.*ls'"),
+			      GetLastError(),
+			      (int)(info->FileNameLength / sizeof(WCHAR)),
+			      info->FileName);
+			return -1;
+		}
+
+		strbuf_grow(normalized_path,
+			    strbuf_avail(normalized_path) + reserve);
+	}
+
+normalize:
+	strbuf_setlen(normalized_path, len);
+	return strbuf_normalize_path(normalized_path);
+}
 
 void fsm_listen__stop_async(struct fsmonitor_daemon_state *state)
 {
+	SetEvent(state->backend_data->hListener[LISTENER_SHUTDOWN]);
+}
+
+static struct one_watch *create_watch(struct fsmonitor_daemon_state *state,
+				      const char *path)
+{
+	struct one_watch *watch = NULL;
+	DWORD desired_access = FILE_LIST_DIRECTORY;
+	DWORD share_mode =
+		FILE_SHARE_WRITE | FILE_SHARE_READ | FILE_SHARE_DELETE;
+	HANDLE hDir;
+	wchar_t wpath[MAX_PATH];
+
+	if (xutftowcs_path(wpath, path) < 0) {
+		error(_("could not convert to wide characters: '%s'"), path);
+		return NULL;
+	}
+
+	hDir = CreateFileW(wpath,
+			   desired_access, share_mode, NULL, OPEN_EXISTING,
+			   FILE_FLAG_BACKUP_SEMANTICS | FILE_FLAG_OVERLAPPED,
+			   NULL);
+	if (hDir == INVALID_HANDLE_VALUE) {
+		error(_("[GLE %ld] could not watch '%s'"),
+		      GetLastError(), path);
+		return NULL;
+	}
+
+	CALLOC_ARRAY(watch, 1);
+
+	watch->buf_len = sizeof(watch->buffer); /* assume full MAX_RDCW_BUF */
+
+	strbuf_init(&watch->path, 0);
+	strbuf_addstr(&watch->path, path);
+
+	watch->hDir = hDir;
+	watch->hEvent = CreateEvent(NULL, TRUE, FALSE, NULL);
+
+	return watch;
+}
+
+static void destroy_watch(struct one_watch *watch)
+{
+	if (!watch)
+		return;
+
+	strbuf_release(&watch->path);
+	if (watch->hDir != INVALID_HANDLE_VALUE)
+		CloseHandle(watch->hDir);
+	if (watch->hEvent != INVALID_HANDLE_VALUE)
+		CloseHandle(watch->hEvent);
+
+	free(watch);
+}
+
+static int start_rdcw_watch(struct fsmonitor_daemon_backend_data *data,
+			    struct one_watch *watch)
+{
+	DWORD dwNotifyFilter =
+		FILE_NOTIFY_CHANGE_FILE_NAME |
+		FILE_NOTIFY_CHANGE_DIR_NAME |
+		FILE_NOTIFY_CHANGE_ATTRIBUTES |
+		FILE_NOTIFY_CHANGE_SIZE |
+		FILE_NOTIFY_CHANGE_LAST_WRITE |
+		FILE_NOTIFY_CHANGE_CREATION;
+
+	ResetEvent(watch->hEvent);
+
+	memset(&watch->overlapped, 0, sizeof(watch->overlapped));
+	watch->overlapped.hEvent = watch->hEvent;
+
+	/*
+	 * Queue an async call using Overlapped IO.  This returns immediately.
+	 * Our event handle will be signalled when the real result is available.
+	 *
+	 * The return value here just means that we successfully queued it.
+	 * We won't know if the Read...() actually produces data until later.
+	 */
+	watch->is_active = ReadDirectoryChangesW(
+		watch->hDir, watch->buffer, watch->buf_len, TRUE,
+		dwNotifyFilter, &watch->count, &watch->overlapped, NULL);
+
+	if (watch->is_active)
+		return 0;
+
+	error(_("ReadDirectoryChangedW failed on '%s' [GLE %ld]"),
+	      watch->path.buf, GetLastError());
+	return -1;
+}
+
+static int recv_rdcw_watch(struct one_watch *watch)
+{
+	DWORD gle;
+
+	watch->is_active = FALSE;
+
+	/*
+	 * The overlapped result is ready.  If the Read...() was successful
+	 * we finally receive the actual result into our buffer.
+	 */
+	if (GetOverlappedResult(watch->hDir, &watch->overlapped, &watch->count,
+				TRUE))
+		return 0;
+
+	gle = GetLastError();
+	if (gle == ERROR_INVALID_PARAMETER &&
+	    /*
+	     * The kernel throws an invalid parameter error when our
+	     * buffer is too big and we are pointed at a remote
+	     * directory (and possibly for other reasons).  Quietly
+	     * set it down and try again.
+	     *
+	     * See note about MAX_RDCW_BUF at the top.
+	     */
+	    watch->buf_len > MAX_RDCW_BUF_FALLBACK) {
+		watch->buf_len = MAX_RDCW_BUF_FALLBACK;
+		return -2;
+	}
+
+	/*
+	 * NEEDSWORK: If an external <gitdir> is deleted, the above
+	 * returns an error.  I'm not sure that there's anything that
+	 * we can do here other than failing -- the <worktree>/.git
+	 * link file would be broken anyway.  We might try to check
+	 * for that and return a better error message, but I'm not
+	 * sure it is worth it.
+	 */
+
+	error(_("GetOverlappedResult failed on '%s' [GLE %ld]"),
+	      watch->path.buf, gle);
+	return -1;
+}
+
+static void cancel_rdcw_watch(struct one_watch *watch)
+{
+	DWORD count;
+
+	if (!watch || !watch->is_active)
+		return;
+
+	/*
+	 * The calls to ReadDirectoryChangesW() and GetOverlappedResult()
+	 * form a "pair" (my term) where we queue an IO and promise to
+	 * hang around and wait for the kernel to give us the result.
+	 *
+	 * If for some reason after we queue the IO, we have to quit
+	 * or otherwise not stick around for the second half, we must
+	 * tell the kernel to abort the IO.  This prevents the kernel
+	 * from writing to our buffer and/or signalling our event
+	 * after we free them.
+	 *
+	 * (Ask me how much fun it was to track that one down).
+	 */
+	CancelIoEx(watch->hDir, &watch->overlapped);
+	GetOverlappedResult(watch->hDir, &watch->overlapped, &count, TRUE);
+	watch->is_active = FALSE;
+}
+
+/*
+ * Process filesystem events that happen anywhere (recursively) under the
+ * <worktree> root directory.  For a normal working directory, this includes
+ * both version controlled files and the contents of the .git/ directory.
+ *
+ * If <worktree>/.git is a file, then we only see events for the file
+ * itself.
+ */
+static int process_worktree_events(struct fsmonitor_daemon_state *state)
+{
+	struct fsmonitor_daemon_backend_data *data = state->backend_data;
+	struct one_watch *watch = data->watch_worktree;
+	struct strbuf path = STRBUF_INIT;
+	struct string_list cookie_list = STRING_LIST_INIT_DUP;
+	struct fsmonitor_batch *batch = NULL;
+	const char *p = watch->buffer;
+
+	/*
+	 * If the kernel gets more events than will fit in the kernel
+	 * buffer associated with our RDCW handle, it drops them and
+	 * returns a count of zero.
+	 *
+	 * Yes, the call returns WITHOUT error and with length zero.
+	 * This is the documented behavior.  (My testing has confirmed
+	 * that it also sets the last error to ERROR_NOTIFY_ENUM_DIR,
+	 * but we do not rely on that since the function did not
+	 * return an error and it is not documented.)
+	 *
+	 * (The "overflow" case is not ambiguous with the "no data" case
+	 * because we did an INFINITE wait.)
+	 *
+	 * This means we have a gap in coverage.  Tell the daemon layer
+	 * to resync.
+	 */
+	if (!watch->count) {
+		trace2_data_string("fsmonitor", NULL, "fsm-listen/kernel",
+				   "overflow");
+		fsmonitor_force_resync(state);
+		return LISTENER_HAVE_DATA_WORKTREE;
+	}
+
+	/*
+	 * On Windows, `info` contains an "array" of paths that are
+	 * relative to the root of whichever directory handle received
+	 * the event.
+	 */
+	for (;;) {
+		FILE_NOTIFY_INFORMATION *info = (void *)p;
+		const char *slash;
+		enum fsmonitor_path_type t;
+
+		strbuf_reset(&path);
+		if (normalize_path_in_utf8(info, &path) == -1)
+			goto skip_this_path;
+
+		t = fsmonitor_classify_path_workdir_relative(path.buf);
+
+		switch (t) {
+		case IS_INSIDE_DOT_GIT_WITH_COOKIE_PREFIX:
+			/* special case cookie files within .git */
+
+			/* Use just the filename of the cookie file. */
+			slash = find_last_dir_sep(path.buf);
+			string_list_append(&cookie_list,
+					   slash ? slash + 1 : path.buf);
+			break;
+
+		case IS_INSIDE_DOT_GIT:
+			/* ignore everything inside of "<worktree>/.git/" */
+			break;
+
+		case IS_DOT_GIT:
+			/* "<worktree>/.git" was deleted (or renamed away) */
+			if ((info->Action == FILE_ACTION_REMOVED) ||
+			    (info->Action == FILE_ACTION_RENAMED_OLD_NAME)) {
+				trace2_data_string("fsmonitor", NULL,
+						   "fsm-listen/dotgit",
+						   "removed");
+				goto force_shutdown;
+			}
+			break;
+
+		case IS_WORKDIR_PATH:
+			/* queue normal pathname */
+			if (!batch)
+				batch = fsmonitor_batch__new();
+			fsmonitor_batch__add_path(batch, path.buf);
+			break;
+
+		case IS_GITDIR:
+		case IS_INSIDE_GITDIR:
+		case IS_INSIDE_GITDIR_WITH_COOKIE_PREFIX:
+		default:
+			BUG("unexpected path classification '%d' for '%s'",
+			    t, path.buf);
+		}
+
+skip_this_path:
+		if (!info->NextEntryOffset)
+			break;
+		p += info->NextEntryOffset;
+	}
+
+	fsmonitor_publish(state, batch, &cookie_list);
+	batch = NULL;
+	string_list_clear(&cookie_list, 0);
+	strbuf_release(&path);
+	return LISTENER_HAVE_DATA_WORKTREE;
+
+force_shutdown:
+	fsmonitor_batch__free_list(batch);
+	string_list_clear(&cookie_list, 0);
+	strbuf_release(&path);
+	return LISTENER_SHUTDOWN;
+}
+
+/*
+ * Process filesystem events that happened anywhere (recursively) under the
+ * external <gitdir> (such as non-primary worktrees or submodules).
+ * We only care about cookie files that our client threads created here.
+ *
+ * Note that we DO NOT get filesystem events on the external <gitdir>
+ * itself (it is not inside something that we are watching).  In particular,
+ * we do not get an event if the external <gitdir> is deleted.
+ */
+static int process_gitdir_events(struct fsmonitor_daemon_state *state)
+{
+	struct fsmonitor_daemon_backend_data *data = state->backend_data;
+	struct one_watch *watch = data->watch_gitdir;
+	struct strbuf path = STRBUF_INIT;
+	struct string_list cookie_list = STRING_LIST_INIT_DUP;
+	const char *p = watch->buffer;
+
+	if (!watch->count) {
+		trace2_data_string("fsmonitor", NULL, "fsm-listen/kernel",
+				   "overflow");
+		fsmonitor_force_resync(state);
+		return LISTENER_HAVE_DATA_GITDIR;
+	}
+
+	for (;;) {
+		FILE_NOTIFY_INFORMATION *info = (void *)p;
+		const char *slash;
+		enum fsmonitor_path_type t;
+
+		strbuf_reset(&path);
+		if (normalize_path_in_utf8(info, &path) == -1)
+			goto skip_this_path;
+
+		t = fsmonitor_classify_path_gitdir_relative(path.buf);
+
+		switch (t) {
+		case IS_INSIDE_GITDIR_WITH_COOKIE_PREFIX:
+			/* special case cookie files within gitdir */
+
+			/* Use just the filename of the cookie file. */
+			slash = find_last_dir_sep(path.buf);
+			string_list_append(&cookie_list,
+					   slash ? slash + 1 : path.buf);
+			break;
+
+		case IS_INSIDE_GITDIR:
+			goto skip_this_path;
+
+		default:
+			BUG("unexpected path classification '%d' for '%s'",
+			    t, path.buf);
+		}
+
+skip_this_path:
+		if (!info->NextEntryOffset)
+			break;
+		p += info->NextEntryOffset;
+	}
+
+	fsmonitor_publish(state, NULL, &cookie_list);
+	string_list_clear(&cookie_list, 0);
+	strbuf_release(&path);
+	return LISTENER_HAVE_DATA_GITDIR;
 }
 
 void fsm_listen__loop(struct fsmonitor_daemon_state *state)
 {
+	struct fsmonitor_daemon_backend_data *data = state->backend_data;
+	DWORD dwWait;
+	int result;
+
+	state->error_code = 0;
+
+	if (start_rdcw_watch(data, data->watch_worktree) == -1)
+		goto force_error_stop;
+
+	if (data->watch_gitdir &&
+	    start_rdcw_watch(data, data->watch_gitdir) == -1)
+		goto force_error_stop;
+
+	for (;;) {
+		dwWait = WaitForMultipleObjects(data->nr_listener_handles,
+						data->hListener,
+						FALSE, INFINITE);
+
+		if (dwWait == WAIT_OBJECT_0 + LISTENER_HAVE_DATA_WORKTREE) {
+			result = recv_rdcw_watch(data->watch_worktree);
+			if (result == -1) {
+				/* hard error */
+				goto force_error_stop;
+			}
+			if (result == -2) {
+				/* retryable error */
+				if (start_rdcw_watch(data, data->watch_worktree) == -1)
+					goto force_error_stop;
+				continue;
+			}
+
+			/* have data */
+			if (process_worktree_events(state) == LISTENER_SHUTDOWN)
+				goto force_shutdown;
+			if (start_rdcw_watch(data, data->watch_worktree) == -1)
+				goto force_error_stop;
+			continue;
+		}
+
+		if (dwWait == WAIT_OBJECT_0 + LISTENER_HAVE_DATA_GITDIR) {
+			result = recv_rdcw_watch(data->watch_gitdir);
+			if (result == -1) {
+				/* hard error */
+				goto force_error_stop;
+			}
+			if (result == -2) {
+				/* retryable error */
+				if (start_rdcw_watch(data, data->watch_gitdir) == -1)
+					goto force_error_stop;
+				continue;
+			}
+
+			/* have data */
+			if (process_gitdir_events(state) == LISTENER_SHUTDOWN)
+				goto force_shutdown;
+			if (start_rdcw_watch(data, data->watch_gitdir) == -1)
+				goto force_error_stop;
+			continue;
+		}
+
+		if (dwWait == WAIT_OBJECT_0 + LISTENER_SHUTDOWN)
+			goto clean_shutdown;
+
+		error(_("could not read directory changes [GLE %ld]"),
+		      GetLastError());
+		goto force_error_stop;
+	}
+
+force_error_stop:
+	state->error_code = -1;
+
+force_shutdown:
+	/*
+	 * Tell the IPC thead pool to stop (which completes the await
+	 * in the main thread (which will also signal this thread (if
+	 * we are still alive))).
+	 */
+	ipc_server_stop_async(state->ipc_server_data);
+
+clean_shutdown:
+	cancel_rdcw_watch(data->watch_worktree);
+	cancel_rdcw_watch(data->watch_gitdir);
 }
 
 int fsm_listen__ctor(struct fsmonitor_daemon_state *state)
 {
+	struct fsmonitor_daemon_backend_data *data;
+
+	CALLOC_ARRAY(data, 1);
+
+	data->hEventShutdown = CreateEvent(NULL, TRUE, FALSE, NULL);
+
+	data->watch_worktree = create_watch(state,
+					    state->path_worktree_watch.buf);
+	if (!data->watch_worktree)
+		goto failed;
+
+	if (state->nr_paths_watching > 1) {
+		data->watch_gitdir = create_watch(state,
+						  state->path_gitdir_watch.buf);
+		if (!data->watch_gitdir)
+			goto failed;
+	}
+
+	data->hListener[LISTENER_SHUTDOWN] = data->hEventShutdown;
+	data->nr_listener_handles++;
+
+	data->hListener[LISTENER_HAVE_DATA_WORKTREE] =
+		data->watch_worktree->hEvent;
+	data->nr_listener_handles++;
+
+	if (data->watch_gitdir) {
+		data->hListener[LISTENER_HAVE_DATA_GITDIR] =
+			data->watch_gitdir->hEvent;
+		data->nr_listener_handles++;
+	}
+
+	state->backend_data = data;
+	return 0;
+
+failed:
+	CloseHandle(data->hEventShutdown);
+	destroy_watch(data->watch_worktree);
+	destroy_watch(data->watch_gitdir);
+
 	return -1;
 }
 
 void fsm_listen__dtor(struct fsmonitor_daemon_state *state)
 {
+	struct fsmonitor_daemon_backend_data *data;
+
+	if (!state || !state->backend_data)
+		return;
+
+	data = state->backend_data;
+
+	CloseHandle(data->hEventShutdown);
+	destroy_watch(data->watch_worktree);
+	destroy_watch(data->watch_gitdir);
+
+	FREE_AND_NULL(state->backend_data);
 }
-- 
gitgitgadget

[PATCH v7 16/29] compat/fsmonitor/fsm-listen-darwin: add MacOS header files for FSEvent

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Include MacOS system declarations to allow us to use FSEvent and
CoreFoundation APIs.  We need different versions of the declarations
for GCC vs. clang because of compiler and header file conflicts.

While it is quite possible to #include Apple's CoreServices.h when
compiling C source code with clang, trying to build it with GCC
currently fails with this error:

In file included
   from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
   ...Library/Frameworks/Security.framework/Headers/AuthSession.h:32,
   from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
   ...Library/Frameworks/Security.framework/Headers/Security.h:42,
   from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
   ...Library/Frameworks/CoreServices.framework/Frameworks/...
   ...OSServices.framework/Headers/CSIdentity.h:43,
   from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
   ...Library/Frameworks/CoreServices.framework/Frameworks/...
   ...OSServices.framework/Headers/OSServices.h:29,
   from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
   ...Library/Frameworks/CoreServices.framework/Frameworks/...
   ...LaunchServices.framework/Headers/IconsCore.h:23,
   from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
   ...Library/Frameworks/CoreServices.framework/Frameworks/...
   ...LaunchServices.framework/Headers/LaunchServices.h:23,
   from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
   ...Library/Frameworks/CoreServices.framework/Headers/CoreServices.h:45,

     /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
     ...Library/Frameworks/Security.framework/Headers/Authorization.h:193:7:
     error: variably modified 'bytes' at file scope
       193 | char bytes[kAuthorizationExternalFormLength];
           |      ^~~~~

The underlying reason is that GCC (rightfully) objects that an `enum`
value such as `kAuthorizationExternalFormLength` is not a constant
(because it is not, the preprocessor has no knowledge of it, only the
actual C compiler does) and can therefore not be used to define the size
of a C array.

This is a known problem and tracked in GCC's bug tracker:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93082

In the meantime, let's not block things and go the slightly ugly route
of declaring/defining the FSEvents constants, data structures and
functions that we need, so that we can avoid above-mentioned issue.

Let's do this _only_ for GCC, though, so that the CI/PR builds (which
build both with clang and with GCC) can guarantee that we _are_ using
the correct data types.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 compat/fsmonitor/fsm-darwin-gcc.h    | 92 ++++++++++++++++++++++++++++
 compat/fsmonitor/fsm-listen-darwin.c | 24 ++++++++
 2 files changed, 116 insertions(+)
 create mode 100644 compat/fsmonitor/fsm-darwin-gcc.h

diff --git a/compat/fsmonitor/fsm-darwin-gcc.h b/compat/fsmonitor/fsm-darwin-gcc.h
new file mode 100644
index 00000000000..1c75c3d48e7
--- /dev/null
+++ b/compat/fsmonitor/fsm-darwin-gcc.h
@@ -0,0 +1,92 @@
+#ifndef FSM_DARWIN_GCC_H
+#define FSM_DARWIN_GCC_H
+
+#ifndef __clang__
+/*
+ * It is possible to #include CoreFoundation/CoreFoundation.h when compiling
+ * with clang, but not with GCC as of time of writing.
+ *
+ * See https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93082 for details.
+ */
+typedef unsigned int FSEventStreamCreateFlags;
+#define kFSEventStreamEventFlagNone               0x00000000
+#define kFSEventStreamEventFlagMustScanSubDirs    0x00000001
+#define kFSEventStreamEventFlagUserDropped        0x00000002
+#define kFSEventStreamEventFlagKernelDropped      0x00000004
+#define kFSEventStreamEventFlagEventIdsWrapped    0x00000008
+#define kFSEventStreamEventFlagHistoryDone        0x00000010
+#define kFSEventStreamEventFlagRootChanged        0x00000020
+#define kFSEventStreamEventFlagMount              0x00000040
+#define kFSEventStreamEventFlagUnmount            0x00000080
+#define kFSEventStreamEventFlagItemCreated        0x00000100
+#define kFSEventStreamEventFlagItemRemoved        0x00000200
+#define kFSEventStreamEventFlagItemInodeMetaMod   0x00000400
+#define kFSEventStreamEventFlagItemRenamed        0x00000800
+#define kFSEventStreamEventFlagItemModified       0x00001000
+#define kFSEventStreamEventFlagItemFinderInfoMod  0x00002000
+#define kFSEventStreamEventFlagItemChangeOwner    0x00004000
+#define kFSEventStreamEventFlagItemXattrMod       0x00008000
+#define kFSEventStreamEventFlagItemIsFile         0x00010000
+#define kFSEventStreamEventFlagItemIsDir          0x00020000
+#define kFSEventStreamEventFlagItemIsSymlink      0x00040000
+#define kFSEventStreamEventFlagOwnEvent           0x00080000
+#define kFSEventStreamEventFlagItemIsHardlink     0x00100000
+#define kFSEventStreamEventFlagItemIsLastHardlink 0x00200000
+#define kFSEventStreamEventFlagItemCloned         0x00400000
+
+typedef struct __FSEventStream *FSEventStreamRef;
+typedef const FSEventStreamRef ConstFSEventStreamRef;
+
+typedef unsigned int CFStringEncoding;
+#define kCFStringEncodingUTF8 0x08000100
+
+typedef const struct __CFString *CFStringRef;
+typedef const struct __CFArray *CFArrayRef;
+typedef const struct __CFRunLoop *CFRunLoopRef;
+
+struct FSEventStreamContext {
+    long long version;
+    void *cb_data, *retain, *release, *copy_description;
+};
+
+typedef struct FSEventStreamContext FSEventStreamContext;
+typedef unsigned int FSEventStreamEventFlags;
+#define kFSEventStreamCreateFlagNoDefer 0x02
+#define kFSEventStreamCreateFlagWatchRoot 0x04
+#define kFSEventStreamCreateFlagFileEvents 0x10
+
+typedef unsigned long long FSEventStreamEventId;
+#define kFSEventStreamEventIdSinceNow 0xFFFFFFFFFFFFFFFFULL
+
+typedef void (*FSEventStreamCallback)(ConstFSEventStreamRef streamRef,
+				      void *context,
+				      __SIZE_TYPE__ num_of_events,
+				      void *event_paths,
+				      const FSEventStreamEventFlags event_flags[],
+				      const FSEventStreamEventId event_ids[]);
+typedef double CFTimeInterval;
+FSEventStreamRef FSEventStreamCreate(void *allocator,
+				     FSEventStreamCallback callback,
+				     FSEventStreamContext *context,
+				     CFArrayRef paths_to_watch,
+				     FSEventStreamEventId since_when,
+				     CFTimeInterval latency,
+				     FSEventStreamCreateFlags flags);
+CFStringRef CFStringCreateWithCString(void *allocator, const char *string,
+				      CFStringEncoding encoding);
+CFArrayRef CFArrayCreate(void *allocator, const void **items, long long count,
+			 void *callbacks);
+void CFRunLoopRun(void);
+void CFRunLoopStop(CFRunLoopRef run_loop);
+CFRunLoopRef CFRunLoopGetCurrent(void);
+extern CFStringRef kCFRunLoopDefaultMode;
+void FSEventStreamScheduleWithRunLoop(FSEventStreamRef stream,
+				      CFRunLoopRef run_loop,
+				      CFStringRef run_loop_mode);
+unsigned char FSEventStreamStart(FSEventStreamRef stream);
+void FSEventStreamStop(FSEventStreamRef stream);
+void FSEventStreamInvalidate(FSEventStreamRef stream);
+void FSEventStreamRelease(FSEventStreamRef stream);
+
+#endif /* !clang */
+#endif /* FSM_DARWIN_GCC_H */
diff --git a/compat/fsmonitor/fsm-listen-darwin.c b/compat/fsmonitor/fsm-listen-darwin.c
index c84e3344ab9..d2ce942cade 100644
--- a/compat/fsmonitor/fsm-listen-darwin.c
+++ b/compat/fsmonitor/fsm-listen-darwin.c
@@ -1,3 +1,27 @@
+#ifndef __clang__
+#include "fsm-darwin-gcc.h"
+#else
+#include <CoreFoundation/CoreFoundation.h>
+#include <CoreServices/CoreServices.h>
+
+#ifndef AVAILABLE_MAC_OS_X_VERSION_10_13_AND_LATER
+/*
+ * This enum value was added in 10.13 to:
+ *
+ * /Applications/Xcode.app/Contents/Developer/Platforms/ \
+ *    MacOSX.platform/Developer/SDKs/MacOSX.sdk/System/ \
+ *    Library/Frameworks/CoreServices.framework/Frameworks/ \
+ *    FSEvents.framework/Versions/Current/Headers/FSEvents.h
+ *
+ * If we're compiling against an older SDK, this symbol won't be
+ * present.  Silently define it here so that we don't have to ifdef
+ * the logging or masking below.  This should be harmless since older
+ * versions of macOS won't ever emit this FS event anyway.
+ */
+#define kFSEventStreamEventFlagItemCloned         0x00400000
+#endif
+#endif
+
 #include "cache.h"
 #include "fsmonitor.h"
 #include "fsm-listen.h"
-- 
gitgitgadget

[PATCH v7 17/29] compat/fsmonitor/fsm-listen-darwin: implement FSEvent listener on MacOS

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Implement file system event listener on MacOS using FSEvent,
CoreFoundation, and CoreServices.

Co-authored-by: Kevin Willford <Kevin.Willford@microsoft.com>
Co-authored-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 compat/fsmonitor/fsm-listen-darwin.c | 383 +++++++++++++++++++++++++++
 1 file changed, 383 insertions(+)

diff --git a/compat/fsmonitor/fsm-listen-darwin.c b/compat/fsmonitor/fsm-listen-darwin.c
index d2ce942cade..0741fe834c3 100644
--- a/compat/fsmonitor/fsm-listen-darwin.c
+++ b/compat/fsmonitor/fsm-listen-darwin.c
@@ -25,20 +25,403 @@
 #include "cache.h"
 #include "fsmonitor.h"
 #include "fsm-listen.h"
+#include "fsmonitor--daemon.h"
+
+struct fsmonitor_daemon_backend_data
+{
+	CFStringRef cfsr_worktree_path;
+	CFStringRef cfsr_gitdir_path;
+
+	CFArrayRef cfar_paths_to_watch;
+	int nr_paths_watching;
+
+	FSEventStreamRef stream;
+
+	CFRunLoopRef rl;
+
+	enum shutdown_style {
+		SHUTDOWN_EVENT = 0,
+		FORCE_SHUTDOWN,
+		FORCE_ERROR_STOP,
+	} shutdown_style;
+
+	unsigned int stream_scheduled:1;
+	unsigned int stream_started:1;
+};
+
+static void log_flags_set(const char *path, const FSEventStreamEventFlags flag)
+{
+	struct strbuf msg = STRBUF_INIT;
+
+	if (flag & kFSEventStreamEventFlagMustScanSubDirs)
+		strbuf_addstr(&msg, "MustScanSubDirs|");
+	if (flag & kFSEventStreamEventFlagUserDropped)
+		strbuf_addstr(&msg, "UserDropped|");
+	if (flag & kFSEventStreamEventFlagKernelDropped)
+		strbuf_addstr(&msg, "KernelDropped|");
+	if (flag & kFSEventStreamEventFlagEventIdsWrapped)
+		strbuf_addstr(&msg, "EventIdsWrapped|");
+	if (flag & kFSEventStreamEventFlagHistoryDone)
+		strbuf_addstr(&msg, "HistoryDone|");
+	if (flag & kFSEventStreamEventFlagRootChanged)
+		strbuf_addstr(&msg, "RootChanged|");
+	if (flag & kFSEventStreamEventFlagMount)
+		strbuf_addstr(&msg, "Mount|");
+	if (flag & kFSEventStreamEventFlagUnmount)
+		strbuf_addstr(&msg, "Unmount|");
+	if (flag & kFSEventStreamEventFlagItemChangeOwner)
+		strbuf_addstr(&msg, "ItemChangeOwner|");
+	if (flag & kFSEventStreamEventFlagItemCreated)
+		strbuf_addstr(&msg, "ItemCreated|");
+	if (flag & kFSEventStreamEventFlagItemFinderInfoMod)
+		strbuf_addstr(&msg, "ItemFinderInfoMod|");
+	if (flag & kFSEventStreamEventFlagItemInodeMetaMod)
+		strbuf_addstr(&msg, "ItemInodeMetaMod|");
+	if (flag & kFSEventStreamEventFlagItemIsDir)
+		strbuf_addstr(&msg, "ItemIsDir|");
+	if (flag & kFSEventStreamEventFlagItemIsFile)
+		strbuf_addstr(&msg, "ItemIsFile|");
+	if (flag & kFSEventStreamEventFlagItemIsHardlink)
+		strbuf_addstr(&msg, "ItemIsHardlink|");
+	if (flag & kFSEventStreamEventFlagItemIsLastHardlink)
+		strbuf_addstr(&msg, "ItemIsLastHardlink|");
+	if (flag & kFSEventStreamEventFlagItemIsSymlink)
+		strbuf_addstr(&msg, "ItemIsSymlink|");
+	if (flag & kFSEventStreamEventFlagItemModified)
+		strbuf_addstr(&msg, "ItemModified|");
+	if (flag & kFSEventStreamEventFlagItemRemoved)
+		strbuf_addstr(&msg, "ItemRemoved|");
+	if (flag & kFSEventStreamEventFlagItemRenamed)
+		strbuf_addstr(&msg, "ItemRenamed|");
+	if (flag & kFSEventStreamEventFlagItemXattrMod)
+		strbuf_addstr(&msg, "ItemXattrMod|");
+	if (flag & kFSEventStreamEventFlagOwnEvent)
+		strbuf_addstr(&msg, "OwnEvent|");
+	if (flag & kFSEventStreamEventFlagItemCloned)
+		strbuf_addstr(&msg, "ItemCloned|");
+
+	trace_printf_key(&trace_fsmonitor, "fsevent: '%s', flags=%u %s",
+			 path, flag, msg.buf);
+
+	strbuf_release(&msg);
+}
+
+static int ef_is_root_delete(const FSEventStreamEventFlags ef)
+{
+	return (ef & kFSEventStreamEventFlagItemIsDir &&
+		ef & kFSEventStreamEventFlagItemRemoved);
+}
+
+static int ef_is_root_renamed(const FSEventStreamEventFlags ef)
+{
+	return (ef & kFSEventStreamEventFlagItemIsDir &&
+		ef & kFSEventStreamEventFlagItemRenamed);
+}
+
+static int ef_is_dropped(const FSEventStreamEventFlags ef)
+{
+	return (ef & kFSEventStreamEventFlagMustScanSubDirs ||
+		ef & kFSEventStreamEventFlagKernelDropped ||
+		ef & kFSEventStreamEventFlagUserDropped);
+}
+
+static void fsevent_callback(ConstFSEventStreamRef streamRef,
+			     void *ctx,
+			     size_t num_of_events,
+			     void *event_paths,
+			     const FSEventStreamEventFlags event_flags[],
+			     const FSEventStreamEventId event_ids[])
+{
+	struct fsmonitor_daemon_state *state = ctx;
+	struct fsmonitor_daemon_backend_data *data = state->backend_data;
+	char **paths = (char **)event_paths;
+	struct fsmonitor_batch *batch = NULL;
+	struct string_list cookie_list = STRING_LIST_INIT_DUP;
+	const char *path_k;
+	const char *slash;
+	int k;
+	struct strbuf tmp = STRBUF_INIT;
+
+	/*
+	 * Build a list of all filesystem changes into a private/local
+	 * list and without holding any locks.
+	 */
+	for (k = 0; k < num_of_events; k++) {
+		/*
+		 * On Mac, we receive an array of absolute paths.
+		 */
+		path_k = paths[k];
+
+		/*
+		 * If you want to debug FSEvents, log them to GIT_TRACE_FSMONITOR.
+		 * Please don't log them to Trace2.
+		 *
+		 * trace_printf_key(&trace_fsmonitor, "Path: '%s'", path_k);
+		 */
+
+		/*
+		 * If event[k] is marked as dropped, we assume that we have
+		 * lost sync with the filesystem and should flush our cached
+		 * data.  We need to:
+		 *
+		 * [1] Abort/wake any client threads waiting for a cookie and
+		 *     flush the cached state data (the current token), and
+		 *     create a new token.
+		 *
+		 * [2] Discard the batch that we were locally building (since
+		 *     they are conceptually relative to the just flushed
+		 *     token).
+		 */
+		if (ef_is_dropped(event_flags[k])) {
+			if (trace_pass_fl(&trace_fsmonitor))
+				log_flags_set(path_k, event_flags[k]);
+
+			fsmonitor_force_resync(state);
+			fsmonitor_batch__free_list(batch);
+			string_list_clear(&cookie_list, 0);
+
+			/*
+			 * We assume that any events that we received
+			 * in this callback after this dropped event
+			 * may still be valid, so we continue rather
+			 * than break.  (And just in case there is a
+			 * delete of ".git" hiding in there.)
+			 */
+			continue;
+		}
+
+		switch (fsmonitor_classify_path_absolute(state, path_k)) {
+
+		case IS_INSIDE_DOT_GIT_WITH_COOKIE_PREFIX:
+		case IS_INSIDE_GITDIR_WITH_COOKIE_PREFIX:
+			/* special case cookie files within .git or gitdir */
+
+			/* Use just the filename of the cookie file. */
+			slash = find_last_dir_sep(path_k);
+			string_list_append(&cookie_list,
+					   slash ? slash + 1 : path_k);
+			break;
+
+		case IS_INSIDE_DOT_GIT:
+		case IS_INSIDE_GITDIR:
+			/* ignore all other paths inside of .git or gitdir */
+			break;
+
+		case IS_DOT_GIT:
+		case IS_GITDIR:
+			/*
+			 * If .git directory is deleted or renamed away,
+			 * we have to quit.
+			 */
+			if (ef_is_root_delete(event_flags[k])) {
+				trace_printf_key(&trace_fsmonitor,
+						 "event: gitdir removed");
+				goto force_shutdown;
+			}
+			if (ef_is_root_renamed(event_flags[k])) {
+				trace_printf_key(&trace_fsmonitor,
+						 "event: gitdir renamed");
+				goto force_shutdown;
+			}
+			break;
+
+		case IS_WORKDIR_PATH:
+			/* try to queue normal pathnames */
+
+			if (trace_pass_fl(&trace_fsmonitor))
+				log_flags_set(path_k, event_flags[k]);
+
+			/*
+			 * Because of the implicit "binning" (the
+			 * kernel calls us at a given frequency) and
+			 * de-duping (the kernel is free to combine
+			 * multiple events for a given pathname), an
+			 * individual fsevent could be marked as both
+			 * a file and directory.  Add it to the queue
+			 * with both spellings so that the client will
+			 * know how much to invalidate/refresh.
+			 */
+
+			if (event_flags[k] & kFSEventStreamEventFlagItemIsFile) {
+				const char *rel = path_k +
+					state->path_worktree_watch.len + 1;
+
+				if (!batch)
+					batch = fsmonitor_batch__new();
+				fsmonitor_batch__add_path(batch, rel);
+			}
+
+			if (event_flags[k] & kFSEventStreamEventFlagItemIsDir) {
+				const char *rel = path_k +
+					state->path_worktree_watch.len + 1;
+
+				strbuf_reset(&tmp);
+				strbuf_addstr(&tmp, rel);
+				strbuf_addch(&tmp, '/');
+
+				if (!batch)
+					batch = fsmonitor_batch__new();
+				fsmonitor_batch__add_path(batch, tmp.buf);
+			}
+
+			break;
+
+		case IS_OUTSIDE_CONE:
+		default:
+			trace_printf_key(&trace_fsmonitor,
+					 "ignoring '%s'", path_k);
+			break;
+		}
+	}
+
+	fsmonitor_publish(state, batch, &cookie_list);
+	string_list_clear(&cookie_list, 0);
+	strbuf_release(&tmp);
+	return;
+
+force_shutdown:
+	fsmonitor_batch__free_list(batch);
+	string_list_clear(&cookie_list, 0);
+
+	data->shutdown_style = FORCE_SHUTDOWN;
+	CFRunLoopStop(data->rl);
+	strbuf_release(&tmp);
+	return;
+}
+
+/*
+ * In the call to `FSEventStreamCreate()` to setup our watch, the
+ * `latency` argument determines the frequency of calls to our callback
+ * with new FS events.  Too slow and events get dropped; too fast and
+ * we burn CPU unnecessarily.  Since it is rather obscure, I don't
+ * think this needs to be a config setting.  I've done extensive
+ * testing on my systems and chosen the value below.  It gives good
+ * results and I've not seen any dropped events.
+ *
+ * With a latency of 0.1, I was seeing lots of dropped events during
+ * the "touch 100000" files test within t/perf/p7519, but with a
+ * latency of 0.001 I did not see any dropped events.  So I'm going
+ * to assume that this is the "correct" value.
+ *
+ * https://developer.apple.com/documentation/coreservices/1443980-fseventstreamcreate
+ */
 
 int fsm_listen__ctor(struct fsmonitor_daemon_state *state)
 {
+	FSEventStreamCreateFlags flags = kFSEventStreamCreateFlagNoDefer |
+		kFSEventStreamCreateFlagWatchRoot |
+		kFSEventStreamCreateFlagFileEvents;
+	FSEventStreamContext ctx = {
+		0,
+		state,
+		NULL,
+		NULL,
+		NULL
+	};
+	struct fsmonitor_daemon_backend_data *data;
+	const void *dir_array[2];
+
+	CALLOC_ARRAY(data, 1);
+	state->backend_data = data;
+
+	data->cfsr_worktree_path = CFStringCreateWithCString(
+		NULL, state->path_worktree_watch.buf, kCFStringEncodingUTF8);
+	dir_array[data->nr_paths_watching++] = data->cfsr_worktree_path;
+
+	if (state->nr_paths_watching > 1) {
+		data->cfsr_gitdir_path = CFStringCreateWithCString(
+			NULL, state->path_gitdir_watch.buf,
+			kCFStringEncodingUTF8);
+		dir_array[data->nr_paths_watching++] = data->cfsr_gitdir_path;
+	}
+
+	data->cfar_paths_to_watch = CFArrayCreate(NULL, dir_array,
+						  data->nr_paths_watching,
+						  NULL);
+	data->stream = FSEventStreamCreate(NULL, fsevent_callback, &ctx,
+					   data->cfar_paths_to_watch,
+					   kFSEventStreamEventIdSinceNow,
+					   0.001, flags);
+	if (data->stream == NULL)
+		goto failed;
+
+	/*
+	 * `data->rl` needs to be set inside the listener thread.
+	 */
+
+	return 0;
+
+failed:
+	error(_("Unable to create FSEventStream."));
+
+	FREE_AND_NULL(state->backend_data);
 	return -1;
 }
 
 void fsm_listen__dtor(struct fsmonitor_daemon_state *state)
 {
+	struct fsmonitor_daemon_backend_data *data;
+
+	if (!state || !state->backend_data)
+		return;
+
+	data = state->backend_data;
+
+	if (data->stream) {
+		if (data->stream_started)
+			FSEventStreamStop(data->stream);
+		if (data->stream_scheduled)
+			FSEventStreamInvalidate(data->stream);
+		FSEventStreamRelease(data->stream);
+	}
+
+	FREE_AND_NULL(state->backend_data);
 }
 
 void fsm_listen__stop_async(struct fsmonitor_daemon_state *state)
 {
+	struct fsmonitor_daemon_backend_data *data;
+
+	data = state->backend_data;
+	data->shutdown_style = SHUTDOWN_EVENT;
+
+	CFRunLoopStop(data->rl);
 }
 
 void fsm_listen__loop(struct fsmonitor_daemon_state *state)
 {
+	struct fsmonitor_daemon_backend_data *data;
+
+	data = state->backend_data;
+
+	data->rl = CFRunLoopGetCurrent();
+
+	FSEventStreamScheduleWithRunLoop(data->stream, data->rl, kCFRunLoopDefaultMode);
+	data->stream_scheduled = 1;
+
+	if (!FSEventStreamStart(data->stream)) {
+		error(_("Failed to start the FSEventStream"));
+		goto force_error_stop_without_loop;
+	}
+	data->stream_started = 1;
+
+	CFRunLoopRun();
+
+	switch (data->shutdown_style) {
+	case FORCE_ERROR_STOP:
+		state->error_code = -1;
+		/* fall thru */
+	case FORCE_SHUTDOWN:
+		ipc_server_stop_async(state->ipc_server_data);
+		/* fall thru */
+	case SHUTDOWN_EVENT:
+	default:
+		break;
+	}
+	return;
+
+force_error_stop_without_loop:
+	state->error_code = -1;
+	ipc_server_stop_async(state->ipc_server_data);
+	return;
 }
-- 
gitgitgadget

[PATCH v7 18/29] fsmonitor--daemon: implement handle_client callback

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach fsmonitor--daemon to respond to IPC requests from client
Git processes and respond with a list of modified pathnames
relative to the provided token.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 builtin/fsmonitor--daemon.c | 312 +++++++++++++++++++++++++++++++++++-
 1 file changed, 310 insertions(+), 2 deletions(-)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index 69312119b07..fc3aee0ada0 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -7,6 +7,7 @@
 #include "fsmonitor--daemon.h"
 #include "simple-ipc.h"
 #include "khash.h"
+#include "pkt-line.h"
 
 static const char * const builtin_fsmonitor__daemon_usage[] = {
 	N_("git fsmonitor--daemon start [<options>]"),
@@ -364,6 +365,311 @@ void fsmonitor_force_resync(struct fsmonitor_daemon_state *state)
 	pthread_mutex_unlock(&state->main_lock);
 }
 
+/*
+ * Format an opaque token string to send to the client.
+ */
+static void with_lock__format_response_token(
+	struct strbuf *response_token,
+	const struct strbuf *response_token_id,
+	const struct fsmonitor_batch *batch)
+{
+	/* assert current thread holding state->main_lock */
+
+	strbuf_reset(response_token);
+	strbuf_addf(response_token, "builtin:%s:%"PRIu64,
+		    response_token_id->buf, batch->batch_seq_nr);
+}
+
+/*
+ * Parse an opaque token from the client.
+ * Returns -1 on error.
+ */
+static int fsmonitor_parse_client_token(const char *buf_token,
+					struct strbuf *requested_token_id,
+					uint64_t *seq_nr)
+{
+	const char *p;
+	char *p_end;
+
+	strbuf_reset(requested_token_id);
+	*seq_nr = 0;
+
+	if (!skip_prefix(buf_token, "builtin:", &p))
+		return -1;
+
+	while (*p && *p != ':')
+		strbuf_addch(requested_token_id, *p++);
+	if (!*p++)
+		return -1;
+
+	*seq_nr = (uint64_t)strtoumax(p, &p_end, 10);
+	if (*p_end)
+		return -1;
+
+	return 0;
+}
+
+KHASH_INIT(str, const char *, int, 0, kh_str_hash_func, kh_str_hash_equal)
+
+static int do_handle_client(struct fsmonitor_daemon_state *state,
+			    const char *command,
+			    ipc_server_reply_cb *reply,
+			    struct ipc_server_reply_data *reply_data)
+{
+	struct fsmonitor_token_data *token_data = NULL;
+	struct strbuf response_token = STRBUF_INIT;
+	struct strbuf requested_token_id = STRBUF_INIT;
+	struct strbuf payload = STRBUF_INIT;
+	uint64_t requested_oldest_seq_nr = 0;
+	uint64_t total_response_len = 0;
+	const char *p;
+	const struct fsmonitor_batch *batch_head;
+	const struct fsmonitor_batch *batch;
+	intmax_t count = 0, duplicates = 0;
+	kh_str_t *shown;
+	int hash_ret;
+	int do_trivial = 0;
+	int do_flush = 0;
+
+	/*
+	 * We expect `command` to be of the form:
+	 *
+	 * <command> := quit NUL
+	 *            | flush NUL
+	 *            | <V1-time-since-epoch-ns> NUL
+	 *            | <V2-opaque-fsmonitor-token> NUL
+	 */
+
+	if (!strcmp(command, "quit")) {
+		/*
+		 * A client has requested over the socket/pipe that the
+		 * daemon shutdown.
+		 *
+		 * Tell the IPC thread pool to shutdown (which completes
+		 * the await in the main thread (which can stop the
+		 * fsmonitor listener thread)).
+		 *
+		 * There is no reply to the client.
+		 */
+		return SIMPLE_IPC_QUIT;
+
+	} else if (!strcmp(command, "flush")) {
+		/*
+		 * Flush all of our cached data and generate a new token
+		 * just like if we lost sync with the filesystem.
+		 *
+		 * Then send a trivial response using the new token.
+		 */
+		do_flush = 1;
+		do_trivial = 1;
+
+	} else if (!skip_prefix(command, "builtin:", &p)) {
+		/* assume V1 timestamp or garbage */
+
+		char *p_end;
+
+		strtoumax(command, &p_end, 10);
+		trace_printf_key(&trace_fsmonitor,
+				 ((*p_end) ?
+				  "fsmonitor: invalid command line '%s'" :
+				  "fsmonitor: unsupported V1 protocol '%s'"),
+				 command);
+		do_trivial = 1;
+
+	} else {
+		/* We have "builtin:*" */
+		if (fsmonitor_parse_client_token(command, &requested_token_id,
+						 &requested_oldest_seq_nr)) {
+			trace_printf_key(&trace_fsmonitor,
+					 "fsmonitor: invalid V2 protocol token '%s'",
+					 command);
+			do_trivial = 1;
+
+		} else {
+			/*
+			 * We have a V2 valid token:
+			 *     "builtin:<token_id>:<seq_nr>"
+			 */
+		}
+	}
+
+	pthread_mutex_lock(&state->main_lock);
+
+	if (!state->current_token_data)
+		BUG("fsmonitor state does not have a current token");
+
+	if (do_flush)
+		with_lock__do_force_resync(state);
+
+	/*
+	 * We mark the current head of the batch list as "pinned" so
+	 * that the listener thread will treat this item as read-only
+	 * (and prevent any more paths from being added to it) from
+	 * now on.
+	 */
+	token_data = state->current_token_data;
+	batch_head = token_data->batch_head;
+	((struct fsmonitor_batch *)batch_head)->pinned_time = time(NULL);
+
+	/*
+	 * FSMonitor Protocol V2 requires that we send a response header
+	 * with a "new current token" and then all of the paths that changed
+	 * since the "requested token".  We send the seq_nr of the just-pinned
+	 * head batch so that future requests from a client will be relative
+	 * to it.
+	 */
+	with_lock__format_response_token(&response_token,
+					 &token_data->token_id, batch_head);
+
+	reply(reply_data, response_token.buf, response_token.len + 1);
+	total_response_len += response_token.len + 1;
+
+	trace2_data_string("fsmonitor", the_repository, "response/token",
+			   response_token.buf);
+	trace_printf_key(&trace_fsmonitor, "response token: %s",
+			 response_token.buf);
+
+	if (!do_trivial) {
+		if (strcmp(requested_token_id.buf, token_data->token_id.buf)) {
+			/*
+			 * The client last spoke to a different daemon
+			 * instance -OR- the daemon had to resync with
+			 * the filesystem (and lost events), so reject.
+			 */
+			trace2_data_string("fsmonitor", the_repository,
+					   "response/token", "different");
+			do_trivial = 1;
+
+		} else if (requested_oldest_seq_nr <
+			   token_data->batch_tail->batch_seq_nr) {
+			/*
+			 * The client wants older events than we have for
+			 * this token_id.  This means that the end of our
+			 * batch list was truncated and we cannot give the
+			 * client a complete snapshot relative to their
+			 * request.
+			 */
+			trace_printf_key(&trace_fsmonitor,
+					 "client requested truncated data");
+			do_trivial = 1;
+		}
+	}
+
+	if (do_trivial) {
+		pthread_mutex_unlock(&state->main_lock);
+
+		reply(reply_data, "/", 2);
+
+		trace2_data_intmax("fsmonitor", the_repository,
+				   "response/trivial", 1);
+
+		strbuf_release(&response_token);
+		strbuf_release(&requested_token_id);
+		return 0;
+	}
+
+	/*
+	 * We're going to hold onto a pointer to the current
+	 * token-data while we walk the list of batches of files.
+	 * During this time, we will NOT be under the lock.
+	 * So we ref-count it.
+	 *
+	 * This allows the listener thread to continue prepending
+	 * new batches of items to the token-data (which we'll ignore).
+	 *
+	 * AND it allows the listener thread to do a token-reset
+	 * (and install a new `current_token_data`).
+	 */
+	token_data->client_ref_count++;
+
+	pthread_mutex_unlock(&state->main_lock);
+
+	/*
+	 * The client request is relative to the token that they sent,
+	 * so walk the batch list backwards from the current head back
+	 * to the batch (sequence number) they named.
+	 *
+	 * We use khash to de-dup the list of pathnames.
+	 *
+	 * NEEDSWORK: each batch contains a list of interned strings,
+	 * so we only need to do pointer comparisons here to build the
+	 * hash table.  Currently, we're still comparing the string
+	 * values.
+	 */
+	shown = kh_init_str();
+	for (batch = batch_head;
+	     batch && batch->batch_seq_nr > requested_oldest_seq_nr;
+	     batch = batch->next) {
+		size_t k;
+
+		for (k = 0; k < batch->nr; k++) {
+			const char *s = batch->interned_paths[k];
+			size_t s_len;
+
+			if (kh_get_str(shown, s) != kh_end(shown))
+				duplicates++;
+			else {
+				kh_put_str(shown, s, &hash_ret);
+
+				trace_printf_key(&trace_fsmonitor,
+						 "send[%"PRIuMAX"]: %s",
+						 count, s);
+
+				/* Each path gets written with a trailing NUL */
+				s_len = strlen(s) + 1;
+
+				if (payload.len + s_len >=
+				    LARGE_PACKET_DATA_MAX) {
+					reply(reply_data, payload.buf,
+					      payload.len);
+					total_response_len += payload.len;
+					strbuf_reset(&payload);
+				}
+
+				strbuf_add(&payload, s, s_len);
+				count++;
+			}
+		}
+	}
+
+	if (payload.len) {
+		reply(reply_data, payload.buf, payload.len);
+		total_response_len += payload.len;
+	}
+
+	kh_release_str(shown);
+
+	pthread_mutex_lock(&state->main_lock);
+
+	if (token_data->client_ref_count > 0)
+		token_data->client_ref_count--;
+
+	if (token_data->client_ref_count == 0) {
+		if (token_data != state->current_token_data) {
+			/*
+			 * The listener thread did a token-reset while we were
+			 * walking the batch list.  Therefore, this token is
+			 * stale and can be discarded completely.  If we are
+			 * the last reader thread using this token, we own
+			 * that work.
+			 */
+			fsmonitor_free_token_data(token_data);
+		}
+	}
+
+	pthread_mutex_unlock(&state->main_lock);
+
+	trace2_data_intmax("fsmonitor", the_repository, "response/length", total_response_len);
+	trace2_data_intmax("fsmonitor", the_repository, "response/count/files", count);
+	trace2_data_intmax("fsmonitor", the_repository, "response/count/duplicates", duplicates);
+
+	strbuf_release(&response_token);
+	strbuf_release(&requested_token_id);
+	strbuf_release(&payload);
+
+	return 0;
+}
+
 static ipc_server_application_cb handle_client;
 
 static int handle_client(void *data,
@@ -371,7 +677,7 @@ static int handle_client(void *data,
 			 ipc_server_reply_cb *reply,
 			 struct ipc_server_reply_data *reply_data)
 {
-	/* struct fsmonitor_daemon_state *state = data; */
+	struct fsmonitor_daemon_state *state = data;
 	int result;
 
 	/*
@@ -382,10 +688,12 @@ static int handle_client(void *data,
 	if (command_len != strlen(command))
 		BUG("FSMonitor assumes text messages");
 
+	trace_printf_key(&trace_fsmonitor, "requested token: %s", command);
+
 	trace2_region_enter("fsmonitor", "handle_client", the_repository);
 	trace2_data_string("fsmonitor", the_repository, "request", command);
 
-	result = 0; /* TODO Do something here. */
+	result = do_handle_client(state, command, reply, reply_data);
 
 	trace2_region_leave("fsmonitor", "handle_client", the_repository);
 
-- 
gitgitgadget

[PATCH v7 19/29] help: include fsmonitor--daemon feature flag in version info

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Add the "feature: fsmonitor--daemon" message to the output of
`git version --build-options`.

The builtin FSMonitor is only available on certain platforms and
even then only when certain Makefile flags are enabled, so print
a message in the verbose version output when it is available.

This can be used by test scripts for prereq testing.  Granted, tests
could just try `git fsmonitor--daemon status` and look for a 128 exit
code or grep for a "not supported" message on stderr, but these
methods are rather obscure.

The main advantage is that the feature message will automatically
appear in bug reports and other support requests.

This concept was also used during the development of Scalar for
similar reasons.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 help.c        | 4 ++++
 t/test-lib.sh | 7 +++++++
 2 files changed, 11 insertions(+)

diff --git a/help.c b/help.c
index 71444906ddf..9112a51e84b 100644
--- a/help.c
+++ b/help.c
@@ -12,6 +12,7 @@
 #include "refs.h"
 #include "parse-options.h"
 #include "prompt.h"
+#include "fsmonitor-ipc.h"
 
 struct category_description {
 	uint32_t category;
@@ -695,6 +696,9 @@ void get_version_info(struct strbuf *buf, int show_build_options)
 		strbuf_addf(buf, "sizeof-size_t: %d\n", (int)sizeof(size_t));
 		strbuf_addf(buf, "shell-path: %s\n", SHELL_PATH);
 		/* NEEDSWORK: also save and output GIT-BUILD_OPTIONS? */
+
+		if (fsmonitor_ipc__is_supported())
+			strbuf_addstr(buf, "feature: fsmonitor--daemon\n");
 	}
 }
 
diff --git a/t/test-lib.sh b/t/test-lib.sh
index e4716b0b867..5d819c1bc11 100644
--- a/t/test-lib.sh
+++ b/t/test-lib.sh
@@ -1799,3 +1799,10 @@ test_lazy_prereq SHA1 '
 # Tests that verify the scheduler integration must set this locally
 # to avoid errors.
 GIT_TEST_MAINT_SCHEDULER="none:exit 1"
+
+# Does this platform support `git fsmonitor--daemon`
+#
+test_lazy_prereq FSMONITOR_DAEMON '
+	git version --build-options >output &&
+	grep "feature: fsmonitor--daemon" output
+'
-- 
gitgitgadget

[PATCH v7 20/29] t/helper/fsmonitor-client: create IPC client to talk to FSMonitor Daemon

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Create an IPC client to send query and flush commands to the daemon.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 Makefile                         |   1 +
 t/helper/test-fsmonitor-client.c | 116 +++++++++++++++++++++++++++++++
 t/helper/test-tool.c             |   1 +
 t/helper/test-tool.h             |   1 +
 4 files changed, 119 insertions(+)
 create mode 100644 t/helper/test-fsmonitor-client.c

diff --git a/Makefile b/Makefile
index 26567d4f772..daa21bed6c3 100644
--- a/Makefile
+++ b/Makefile
@@ -716,6 +716,7 @@ TEST_BUILTINS_OBJS += test-dump-split-index.o
 TEST_BUILTINS_OBJS += test-dump-untracked-cache.o
 TEST_BUILTINS_OBJS += test-example-decorate.o
 TEST_BUILTINS_OBJS += test-fast-rebase.o
+TEST_BUILTINS_OBJS += test-fsmonitor-client.o
 TEST_BUILTINS_OBJS += test-genrandom.o
 TEST_BUILTINS_OBJS += test-genzeros.o
 TEST_BUILTINS_OBJS += test-getcwd.o
diff --git a/t/helper/test-fsmonitor-client.c b/t/helper/test-fsmonitor-client.c
new file mode 100644
index 00000000000..3062c8a3c2b
--- /dev/null
+++ b/t/helper/test-fsmonitor-client.c
@@ -0,0 +1,116 @@
+/*
+ * test-fsmonitor-client.c: client code to send commands/requests to
+ * a `git fsmonitor--daemon` daemon.
+ */
+
+#include "test-tool.h"
+#include "cache.h"
+#include "parse-options.h"
+#include "fsmonitor-ipc.h"
+
+#ifndef HAVE_FSMONITOR_DAEMON_BACKEND
+int cmd__fsmonitor_client(int argc, const char **argv)
+{
+	die("fsmonitor--daemon not available on this platform");
+}
+#else
+
+/*
+ * Read the `.git/index` to get the last token written to the
+ * FSMonitor Index Extension.
+ */
+static const char *get_token_from_index(void)
+{
+	struct index_state *istate = the_repository->index;
+
+	if (do_read_index(istate, the_repository->index_file, 0) < 0)
+		die("unable to read index file");
+	if (!istate->fsmonitor_last_update)
+		die("index file does not have fsmonitor extension");
+
+	return istate->fsmonitor_last_update;
+}
+
+/*
+ * Send an IPC query to a `git-fsmonitor--daemon` daemon and
+ * ask for the changes since the given token or from the last
+ * token in the index extension.
+ *
+ * This will implicitly start a daemon process if necessary.  The
+ * daemon process will persist after we exit.
+ */
+static int do_send_query(const char *token)
+{
+	struct strbuf answer = STRBUF_INIT;
+	int ret;
+
+	if (!token || !*token)
+		token = get_token_from_index();
+
+	ret = fsmonitor_ipc__send_query(token, &answer);
+	if (ret < 0)
+		die("could not query fsmonitor--daemon");
+
+	write_in_full(1, answer.buf, answer.len);
+	strbuf_release(&answer);
+
+	return 0;
+}
+
+/*
+ * Send a "flush" command to the `git-fsmonitor--daemon` (if running)
+ * and tell it to flush its cache.
+ *
+ * This feature is primarily used by the test suite to simulate a loss of
+ * sync with the filesystem where we miss kernel events.
+ */
+static int do_send_flush(void)
+{
+	struct strbuf answer = STRBUF_INIT;
+	int ret;
+
+	ret = fsmonitor_ipc__send_command("flush", &answer);
+	if (ret)
+		return ret;
+
+	write_in_full(1, answer.buf, answer.len);
+	strbuf_release(&answer);
+
+	return 0;
+}
+
+int cmd__fsmonitor_client(int argc, const char **argv)
+{
+	const char *subcmd;
+	const char *token = NULL;
+
+	const char * const fsmonitor_client_usage[] = {
+		"test-tool fsmonitor-client query [<token>]",
+		"test-tool fsmonitor-client flush",
+		NULL,
+	};
+
+	struct option options[] = {
+		OPT_STRING(0, "token", &token, "token",
+			   "command token to send to the server"),
+		OPT_END()
+	};
+
+	argc = parse_options(argc, argv, NULL, options, fsmonitor_client_usage, 0);
+
+	if (argc != 1)
+		usage_with_options(fsmonitor_client_usage, options);
+
+	subcmd = argv[0];
+
+	setup_git_directory();
+
+	if (!strcmp(subcmd, "query"))
+		return !!do_send_query(token);
+
+	if (!strcmp(subcmd, "flush"))
+		return !!do_send_flush();
+
+	die("Unhandled subcommand: '%s'", subcmd);
+}
+#endif
diff --git a/t/helper/test-tool.c b/t/helper/test-tool.c
index e6ec69cf326..0424f7adf5d 100644
--- a/t/helper/test-tool.c
+++ b/t/helper/test-tool.c
@@ -32,6 +32,7 @@ static struct test_cmd cmds[] = {
 	{ "dump-untracked-cache", cmd__dump_untracked_cache },
 	{ "example-decorate", cmd__example_decorate },
 	{ "fast-rebase", cmd__fast_rebase },
+	{ "fsmonitor-client", cmd__fsmonitor_client },
 	{ "genrandom", cmd__genrandom },
 	{ "genzeros", cmd__genzeros },
 	{ "getcwd", cmd__getcwd },
diff --git a/t/helper/test-tool.h b/t/helper/test-tool.h
index 20756eefdda..c876e8246fb 100644
--- a/t/helper/test-tool.h
+++ b/t/helper/test-tool.h
@@ -23,6 +23,7 @@ int cmd__dump_untracked_cache(int argc, const char **argv);
 int cmd__dump_reftable(int argc, const char **argv);
 int cmd__example_decorate(int argc, const char **argv);
 int cmd__fast_rebase(int argc, const char **argv);
+int cmd__fsmonitor_client(int argc, const char **argv);
 int cmd__genrandom(int argc, const char **argv);
 int cmd__genzeros(int argc, const char **argv);
 int cmd__getcwd(int argc, const char **argv);
-- 
gitgitgadget

[PATCH v7 21/29] t7527: create test for fsmonitor--daemon

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 t/t7527-builtin-fsmonitor.sh | 501 +++++++++++++++++++++++++++++++++++
 1 file changed, 501 insertions(+)
 create mode 100755 t/t7527-builtin-fsmonitor.sh

diff --git a/t/t7527-builtin-fsmonitor.sh b/t/t7527-builtin-fsmonitor.sh
new file mode 100755
index 00000000000..d79635e7596
--- /dev/null
+++ b/t/t7527-builtin-fsmonitor.sh
@@ -0,0 +1,501 @@
+#!/bin/sh
+
+test_description='built-in file system watcher'
+
+. ./test-lib.sh
+
+if ! test_have_prereq FSMONITOR_DAEMON
+then
+	skip_all="fsmonitor--daemon is not supported on this platform"
+	test_done
+fi
+
+stop_daemon_delete_repo () {
+	r=$1 &&
+	test_might_fail git -C $r fsmonitor--daemon stop &&
+	rm -rf $1
+}
+
+start_daemon () {
+	r= &&
+	tf= &&
+	t2= &&
+	tk= &&
+
+	while test "$#" -ne 0
+	do
+		case "$1" in
+		-C)
+			shift;
+			test "$#" -ne 0 || BUG "error: -C requires arg"
+			r="-C $1"
+			shift
+			;;
+		-tf)
+			shift;
+			test "$#" -ne 0 || BUG "error: -tf requires arg"
+			tf="$1"
+			shift
+			;;
+		-t2)
+			shift;
+			test "$#" -ne 0 || BUG "error: -t2 requires arg"
+			t2="$1"
+			shift
+			;;
+		-tk)
+			shift;
+			test "$#" -ne 0 || BUG "error: -tk requires arg"
+			tk="$1"
+			shift
+			;;
+		*)
+			BUG "error: unknown option: '$1'"
+			;;
+		esac
+	done &&
+
+	(
+		if test -n "$tf"
+		then
+			GIT_TRACE_FSMONITOR="$tf"
+			export GIT_TRACE_FSMONITOR
+		fi &&
+
+		if test -n "$t2"
+		then
+			GIT_TRACE2_PERF="$t2"
+			export GIT_TRACE2_PERF
+		fi &&
+
+		if test -n "$tk"
+		then
+			GIT_TEST_FSMONITOR_TOKEN="$tk"
+			export GIT_TEST_FSMONITOR_TOKEN
+		fi &&
+
+		git $r fsmonitor--daemon start &&
+		git $r fsmonitor--daemon status
+	)
+}
+
+# Is a Trace2 data event present with the given catetory and key?
+# We do not care what the value is.
+#
+have_t2_data_event () {
+	c=$1 &&
+	k=$2 &&
+
+	grep -e '"event":"data".*"category":"'"$c"'".*"key":"'"$k"'"'
+}
+
+test_expect_success 'explicit daemon start and stop' '
+	test_when_finished "stop_daemon_delete_repo test_explicit" &&
+
+	git init test_explicit &&
+	start_daemon -C test_explicit &&
+
+	git -C test_explicit fsmonitor--daemon stop &&
+	test_must_fail git -C test_explicit fsmonitor--daemon status
+'
+
+test_expect_success 'implicit daemon start' '
+	test_when_finished "stop_daemon_delete_repo test_implicit" &&
+
+	git init test_implicit &&
+	test_must_fail git -C test_implicit fsmonitor--daemon status &&
+
+	# query will implicitly start the daemon.
+	#
+	# for test-script simplicity, we send a V1 timestamp rather than
+	# a V2 token.  either way, the daemon response to any query contains
+	# a new V2 token.  (the daemon may complain that we sent a V1 request,
+	# but this test case is only concerned with whether the daemon was
+	# implicitly started.)
+
+	GIT_TRACE2_EVENT="$PWD/.git/trace" \
+		test-tool -C test_implicit fsmonitor-client query --token 0 >actual &&
+	nul_to_q <actual >actual.filtered &&
+	grep "builtin:" actual.filtered &&
+
+	# confirm that a daemon was started in the background.
+	#
+	# since the mechanism for starting the background daemon is platform
+	# dependent, just confirm that the foreground command received a
+	# response from the daemon.
+
+	have_t2_data_event fsm_client query/response-length <.git/trace &&
+
+	git -C test_implicit fsmonitor--daemon status &&
+	git -C test_implicit fsmonitor--daemon stop &&
+	test_must_fail git -C test_implicit fsmonitor--daemon status
+'
+
+test_expect_success 'implicit daemon stop (delete .git)' '
+	test_when_finished "stop_daemon_delete_repo test_implicit_1" &&
+
+	git init test_implicit_1 &&
+
+	start_daemon -C test_implicit_1 &&
+
+	# deleting the .git directory will implicitly stop the daemon.
+	rm -rf test_implicit_1/.git &&
+
+	# [1] Create an empty .git directory so that the following Git
+	#     command will stay relative to the `-C` directory.
+	#
+	#     Without this, the Git command will override the requested
+	#     -C argument and crawl out to the containing Git source tree.
+	#     This would make the test result dependent upon whether we
+	#     were using fsmonitor on our development worktree.
+	#
+	sleep 1 &&
+	mkdir test_implicit_1/.git &&
+
+	test_must_fail git -C test_implicit_1 fsmonitor--daemon status
+'
+
+test_expect_success 'implicit daemon stop (rename .git)' '
+	test_when_finished "stop_daemon_delete_repo test_implicit_2" &&
+
+	git init test_implicit_2 &&
+
+	start_daemon -C test_implicit_2 &&
+
+	# renaming the .git directory will implicitly stop the daemon.
+	mv test_implicit_2/.git test_implicit_2/.xxx &&
+
+	# See [1] above.
+	#
+	sleep 1 &&
+	mkdir test_implicit_2/.git &&
+
+	test_must_fail git -C test_implicit_2 fsmonitor--daemon status
+'
+
+test_expect_success 'cannot start multiple daemons' '
+	test_when_finished "stop_daemon_delete_repo test_multiple" &&
+
+	git init test_multiple &&
+
+	start_daemon -C test_multiple &&
+
+	test_must_fail git -C test_multiple fsmonitor--daemon start 2>actual &&
+	grep "fsmonitor--daemon is already running" actual &&
+
+	git -C test_multiple fsmonitor--daemon stop &&
+	test_must_fail git -C test_multiple fsmonitor--daemon status
+'
+
+# These tests use the main repo in the trash directory
+
+test_expect_success 'setup' '
+	>tracked &&
+	>modified &&
+	>delete &&
+	>rename &&
+	mkdir dir1 &&
+	>dir1/tracked &&
+	>dir1/modified &&
+	>dir1/delete &&
+	>dir1/rename &&
+	mkdir dir2 &&
+	>dir2/tracked &&
+	>dir2/modified &&
+	>dir2/delete &&
+	>dir2/rename &&
+	mkdir dirtorename &&
+	>dirtorename/a &&
+	>dirtorename/b &&
+
+	cat >.gitignore <<-\EOF &&
+	.gitignore
+	expect*
+	actual*
+	EOF
+
+	git -c core.fsmonitor=false add . &&
+	test_tick &&
+	git -c core.fsmonitor=false commit -m initial &&
+
+	git config core.fsmonitor true
+'
+
+# The test already explicitly stopped (or tried to stop) the daemon.
+# This is here in case something else fails first.
+#
+redundant_stop_daemon () {
+	test_might_fail git fsmonitor--daemon stop
+}
+
+test_expect_success 'update-index implicitly starts daemon' '
+	test_when_finished redundant_stop_daemon &&
+
+	test_must_fail git fsmonitor--daemon status &&
+
+	GIT_TRACE2_EVENT="$PWD/.git/trace_implicit_1" \
+		git update-index --fsmonitor &&
+
+	git fsmonitor--daemon status &&
+	test_might_fail git fsmonitor--daemon stop &&
+
+	# Confirm that the trace2 log contains a record of the
+	# daemon starting.
+	test_subcommand git fsmonitor--daemon start <.git/trace_implicit_1
+'
+
+test_expect_success 'status implicitly starts daemon' '
+	test_when_finished redundant_stop_daemon &&
+
+	test_must_fail git fsmonitor--daemon status &&
+
+	GIT_TRACE2_EVENT="$PWD/.git/trace_implicit_2" \
+		git status >actual &&
+
+	git fsmonitor--daemon status &&
+	test_might_fail git fsmonitor--daemon stop &&
+
+	# Confirm that the trace2 log contains a record of the
+	# daemon starting.
+	test_subcommand git fsmonitor--daemon start <.git/trace_implicit_2
+'
+
+edit_files () {
+	echo 1 >modified &&
+	echo 2 >dir1/modified &&
+	echo 3 >dir2/modified &&
+	>dir1/untracked
+}
+
+delete_files () {
+	rm -f delete &&
+	rm -f dir1/delete &&
+	rm -f dir2/delete
+}
+
+create_files () {
+	echo 1 >new &&
+	echo 2 >dir1/new &&
+	echo 3 >dir2/new
+}
+
+rename_files () {
+	mv rename renamed &&
+	mv dir1/rename dir1/renamed &&
+	mv dir2/rename dir2/renamed
+}
+
+file_to_directory () {
+	rm -f delete &&
+	mkdir delete &&
+	echo 1 >delete/new
+}
+
+directory_to_file () {
+	rm -rf dir1 &&
+	echo 1 >dir1
+}
+
+# The next few test cases confirm that our fsmonitor daemon sees each type
+# of OS filesystem notification that we care about.  At this layer we just
+# ensure we are getting the OS notifications and do not try to confirm what
+# is reported by `git status`.
+#
+# We run a simple query after modifying the filesystem just to introduce
+# a bit of a delay so that the trace logging from the daemon has time to
+# get flushed to disk.
+#
+# We `reset` and `clean` at the bottom of each test (and before stopping the
+# daemon) because these commands might implicitly restart the daemon.
+
+clean_up_repo_and_stop_daemon () {
+	git reset --hard HEAD &&
+	git clean -fd &&
+	test_might_fail git fsmonitor--daemon stop &&
+	rm -f .git/trace
+}
+
+test_expect_success 'edit some files' '
+	test_when_finished clean_up_repo_and_stop_daemon &&
+
+	start_daemon -tf "$PWD/.git/trace" &&
+
+	edit_files &&
+
+	test-tool fsmonitor-client query --token 0 &&
+
+	grep "^event: dir1/modified$"  .git/trace &&
+	grep "^event: dir2/modified$"  .git/trace &&
+	grep "^event: modified$"       .git/trace &&
+	grep "^event: dir1/untracked$" .git/trace
+'
+
+test_expect_success 'create some files' '
+	test_when_finished clean_up_repo_and_stop_daemon &&
+
+	start_daemon -tf "$PWD/.git/trace" &&
+
+	create_files &&
+
+	test-tool fsmonitor-client query --token 0 &&
+
+	grep "^event: dir1/new$" .git/trace &&
+	grep "^event: dir2/new$" .git/trace &&
+	grep "^event: new$"      .git/trace
+'
+
+test_expect_success 'delete some files' '
+	test_when_finished clean_up_repo_and_stop_daemon &&
+
+	start_daemon -tf "$PWD/.git/trace" &&
+
+	delete_files &&
+
+	test-tool fsmonitor-client query --token 0 &&
+
+	grep "^event: dir1/delete$" .git/trace &&
+	grep "^event: dir2/delete$" .git/trace &&
+	grep "^event: delete$"      .git/trace
+'
+
+test_expect_success 'rename some files' '
+	test_when_finished clean_up_repo_and_stop_daemon &&
+
+	start_daemon -tf "$PWD/.git/trace" &&
+
+	rename_files &&
+
+	test-tool fsmonitor-client query --token 0 &&
+
+	grep "^event: dir1/rename$"  .git/trace &&
+	grep "^event: dir2/rename$"  .git/trace &&
+	grep "^event: rename$"       .git/trace &&
+	grep "^event: dir1/renamed$" .git/trace &&
+	grep "^event: dir2/renamed$" .git/trace &&
+	grep "^event: renamed$"      .git/trace
+'
+
+test_expect_success 'rename directory' '
+	test_when_finished clean_up_repo_and_stop_daemon &&
+
+	start_daemon -tf "$PWD/.git/trace" &&
+
+	mv dirtorename dirrenamed &&
+
+	test-tool fsmonitor-client query --token 0 &&
+
+	grep "^event: dirtorename/*$" .git/trace &&
+	grep "^event: dirrenamed/*$"  .git/trace
+'
+
+test_expect_success 'file changes to directory' '
+	test_when_finished clean_up_repo_and_stop_daemon &&
+
+	start_daemon -tf "$PWD/.git/trace" &&
+
+	file_to_directory &&
+
+	test-tool fsmonitor-client query --token 0 &&
+
+	grep "^event: delete$"     .git/trace &&
+	grep "^event: delete/new$" .git/trace
+'
+
+test_expect_success 'directory changes to a file' '
+	test_when_finished clean_up_repo_and_stop_daemon &&
+
+	start_daemon -tf "$PWD/.git/trace" &&
+
+	directory_to_file &&
+
+	test-tool fsmonitor-client query --token 0 &&
+
+	grep "^event: dir1$" .git/trace
+'
+
+# The next few test cases exercise the token-resync code.  When filesystem
+# drops events (because of filesystem velocity or because the daemon isn't
+# polling fast enough), we need to discard the cached data (relative to the
+# current token) and start collecting events under a new token.
+#
+# the 'test-tool fsmonitor-client flush' command can be used to send a
+# "flush" message to a running daemon and ask it to do a flush/resync.
+
+test_expect_success 'flush cached data' '
+	test_when_finished "stop_daemon_delete_repo test_flush" &&
+
+	git init test_flush &&
+
+	start_daemon -C test_flush -tf "$PWD/.git/trace_daemon" -tk true &&
+
+	# The daemon should have an initial token with no events in _0 and
+	# then a few (probably platform-specific number of) events in _1.
+	# These should both have the same <token_id>.
+
+	test-tool -C test_flush fsmonitor-client query --token "builtin:test_00000001:0" >actual_0 &&
+	nul_to_q <actual_0 >actual_q0 &&
+
+	>test_flush/file_1 &&
+	>test_flush/file_2 &&
+
+	test-tool -C test_flush fsmonitor-client query --token "builtin:test_00000001:0" >actual_1 &&
+	nul_to_q <actual_1 >actual_q1 &&
+
+	grep "file_1" actual_q1 &&
+
+	# Force a flush.  This will change the <token_id>, reset the <seq_nr>, and
+	# flush the file data.  Then create some events and ensure that the file
+	# again appears in the cache.  It should have the new <token_id>.
+
+	test-tool -C test_flush fsmonitor-client flush >flush_0 &&
+	nul_to_q <flush_0 >flush_q0 &&
+	grep "^builtin:test_00000002:0Q/Q$" flush_q0 &&
+
+	test-tool -C test_flush fsmonitor-client query --token "builtin:test_00000002:0" >actual_2 &&
+	nul_to_q <actual_2 >actual_q2 &&
+
+	grep "^builtin:test_00000002:0Q$" actual_q2 &&
+
+	>test_flush/file_3 &&
+
+	test-tool -C test_flush fsmonitor-client query --token "builtin:test_00000002:0" >actual_3 &&
+	nul_to_q <actual_3 >actual_q3 &&
+
+	grep "file_3" actual_q3
+'
+
+# The next few test cases create repos where the .git directory is NOT
+# inside the one of the working directory.  That is, where .git is a file
+# that points to a directory elsewhere.  This happens for submodules and
+# non-primary worktrees.
+
+test_expect_success 'setup worktree base' '
+	git init wt-base &&
+	echo 1 >wt-base/file1 &&
+	git -C wt-base add file1 &&
+	git -C wt-base commit -m "c1"
+'
+
+test_expect_success 'worktree with .git file' '
+	git -C wt-base worktree add ../wt-secondary &&
+
+	start_daemon -C wt-secondary \
+		-tf "$PWD/trace_wt_secondary" \
+		-t2 "$PWD/trace2_wt_secondary" &&
+
+	git -C wt-secondary fsmonitor--daemon stop &&
+	test_must_fail git -C wt-secondary fsmonitor--daemon status
+'
+
+# NEEDSWORK: Repeat one of the "edit" tests on wt-secondary and
+# confirm that we get the same events and behavior -- that is, that
+# fsmonitor--daemon correctly watches BOTH the working directory and
+# the external GITDIR directory and behaves the same as when ".git"
+# is a directory inside the working directory.
+
+test_expect_success 'cleanup worktrees' '
+	stop_daemon_delete_repo wt-secondary &&
+	stop_daemon_delete_repo wt-base
+'
+
+test_done
-- 
gitgitgadget

[PATCH v7 22/29] t/perf: avoid copying builtin fsmonitor files into test repo

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Do not copy any of the various fsmonitor--daemon files from the .git
directory of the (GIT_PREF_REPO or GIT_PERF_LARGE_REPO) source repo
into the test's trash directory.

When perf tests start, they copy the contents of the source repo into
the test's trash directory.  If fsmonitor is running in the source repo,
there may be control files, such as the IPC socket and/or fsmonitor
cookie files.  These should not be copied into the test repo.

Unix domain sockets cannot be copied in the manner used by the test
setup, so if present, the test setup fails.

Cookie files are harmless, but we should avoid them.

The builtin fsmonitor keeps all such control files/sockets in
.git/fsmonitor--daemon*, so it is simple to exclude them.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 t/perf/perf-lib.sh | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/t/perf/perf-lib.sh b/t/perf/perf-lib.sh
index 407252bac70..932105cd12c 100644
--- a/t/perf/perf-lib.sh
+++ b/t/perf/perf-lib.sh
@@ -78,7 +78,7 @@ test_perf_copy_repo_contents () {
 	for stuff in "$1"/*
 	do
 		case "$stuff" in
-		*/objects|*/hooks|*/config|*/commondir|*/gitdir|*/worktrees)
+		*/objects|*/hooks|*/config|*/commondir|*/gitdir|*/worktrees|*/fsmonitor--daemon*)
 			;;
 		*)
 			cp -R "$stuff" "$repo/.git/" || exit 1
-- 
gitgitgadget

[PATCH v7 23/29] t/helper/test-chmtime: skip directories on Windows

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach `test-tool.exe chmtime` to ignore errors when setting the mtime
on a directory on Windows.

NEEDSWORK: The Windows version of `utime()` (aka `mingw_utime()`) does
not properly handle directories because it uses `_wopen()`.  It should
be converted to using `CreateFileW()` and backup semantics at a minimum.
Since I'm already in the middle of a large patch series, I did not want
to destabilize other callers of `utime()` right now.  The problem has
only been observed in the t/perf/p7519 test when the test repo contains
an empty directory on disk.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 t/helper/test-chmtime.c | 15 +++++++++++++++
 1 file changed, 15 insertions(+)

diff --git a/t/helper/test-chmtime.c b/t/helper/test-chmtime.c
index 524b55ca496..dc28890a183 100644
--- a/t/helper/test-chmtime.c
+++ b/t/helper/test-chmtime.c
@@ -134,6 +134,21 @@ int cmd__chmtime(int argc, const char **argv)
 		}
 
 		if (utb.modtime != sb.st_mtime && utime(argv[i], &utb) < 0) {
+#ifdef GIT_WINDOWS_NATIVE
+			if (S_ISDIR(sb.st_mode)) {
+				/*
+				 * NEEDSWORK: The Windows version of `utime()`
+				 * (aka `mingw_utime()`) does not correctly
+				 * handle directory arguments, since it uses
+				 * `_wopen()`.  Ignore it for now since this
+				 * is just a test.
+				 */
+				fprintf(stderr,
+					("Failed to modify time on directory %s. "
+					 "Skipping\n"), argv[i]);
+				continue;
+			}
+#endif
 			fprintf(stderr, "Failed to modify time on %s: %s\n",
 			        argv[i], strerror(errno));
 			return 1;
-- 
gitgitgadget

[PATCH v7 24/29] t/perf/p7519: speed up test on Windows

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Change p7519 to use `test_seq` and `xargs` rather than a `for` loop
to touch thousands of files.  This takes minutes off of test runs
on Windows because of process creation overhead.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 t/perf/p7519-fsmonitor.sh | 32 ++++++++++++++++++++------------
 1 file changed, 20 insertions(+), 12 deletions(-)

diff --git a/t/perf/p7519-fsmonitor.sh b/t/perf/p7519-fsmonitor.sh
index c8be58f3c76..0611e533951 100755
--- a/t/perf/p7519-fsmonitor.sh
+++ b/t/perf/p7519-fsmonitor.sh
@@ -72,7 +72,7 @@ then
 	fi
 fi
 
-trace_start() {
+trace_start () {
 	if test -n "$GIT_PERF_7519_TRACE"
 	then
 		name="$1"
@@ -91,13 +91,20 @@ trace_start() {
 	fi
 }
 
-trace_stop() {
+trace_stop () {
 	if test -n "$GIT_PERF_7519_TRACE"
 	then
 		unset GIT_TRACE2_PERF
 	fi
 }
 
+touch_files () {
+	n=$1
+	d="$n"_files
+
+	(cd $d ; test_seq 1 $n | xargs touch )
+}
+
 test_expect_success "one time repo setup" '
 	# set untrackedCache depending on the environment
 	if test -n "$GIT_PERF_7519_UNTRACKED_CACHE"
@@ -119,10 +126,11 @@ test_expect_success "one time repo setup" '
 	fi &&
 
 	mkdir 1_file 10_files 100_files 1000_files 10000_files &&
-	for i in $(test_seq 1 10); do touch 10_files/$i || return 1; done &&
-	for i in $(test_seq 1 100); do touch 100_files/$i || return 1; done &&
-	for i in $(test_seq 1 1000); do touch 1000_files/$i || return 1; done &&
-	for i in $(test_seq 1 10000); do touch 10000_files/$i || return 1; done &&
+	: 1_file directory should be left empty &&
+	touch_files 10 &&
+	touch_files 100 &&
+	touch_files 1000 &&
+	touch_files 10000 &&
 	git add 1_file 10_files 100_files 1000_files 10000_files &&
 	git commit -qm "Add files" &&
 
@@ -133,7 +141,7 @@ test_expect_success "one time repo setup" '
 	fi
 '
 
-setup_for_fsmonitor() {
+setup_for_fsmonitor () {
 	# set INTEGRATION_SCRIPT depending on the environment
 	if test -n "$INTEGRATION_PATH"
 	then
@@ -173,7 +181,7 @@ test_perf_w_drop_caches () {
 	test_perf "$@"
 }
 
-test_fsmonitor_suite() {
+test_fsmonitor_suite () {
 	if test -n "$INTEGRATION_SCRIPT"; then
 		DESC="fsmonitor=$(basename $INTEGRATION_SCRIPT)"
 	else
@@ -199,15 +207,15 @@ test_fsmonitor_suite() {
 
 	# Update the mtimes on upto 100k files to make status think
 	# that they are dirty.  For simplicity, omit any files with
-	# LFs (i.e. anything that ls-files thinks it needs to dquote).
-	# Then fully backslash-quote the paths to capture any
-	# whitespace so that they pass thru xargs properly.
+	# LFs (i.e. anything that ls-files thinks it needs to dquote)
+	# and any files with whitespace so that they pass thru xargs
+	# properly.
 	#
 	test_perf_w_drop_caches "status (dirty) ($DESC)" '
 		git ls-files | \
 			head -100000 | \
 			grep -v \" | \
-			sed '\''s/\(.\)/\\\1/g'\'' | \
+			grep -v " ." | \
 			xargs test-tool chmtime -300 &&
 		git status
 	'
-- 
gitgitgadget

[PATCH v7 25/29] t/perf/p7519: add fsmonitor--daemon test cases

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Repeat all of the fsmonitor perf tests using `git fsmonitor--daemon` and
the "Simple IPC" interface.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 t/perf/p7519-fsmonitor.sh | 38 ++++++++++++++++++++++++++++++++++----
 1 file changed, 34 insertions(+), 4 deletions(-)

diff --git a/t/perf/p7519-fsmonitor.sh b/t/perf/p7519-fsmonitor.sh
index 0611e533951..3b232e386b9 100755
--- a/t/perf/p7519-fsmonitor.sh
+++ b/t/perf/p7519-fsmonitor.sh
@@ -141,7 +141,7 @@ test_expect_success "one time repo setup" '
 	fi
 '
 
-setup_for_fsmonitor () {
+setup_for_fsmonitor_hook () {
 	# set INTEGRATION_SCRIPT depending on the environment
 	if test -n "$INTEGRATION_PATH"
 	then
@@ -182,7 +182,11 @@ test_perf_w_drop_caches () {
 }
 
 test_fsmonitor_suite () {
-	if test -n "$INTEGRATION_SCRIPT"; then
+	if test -n "$USE_FSMONITOR_DAEMON"
+	then
+		DESC="builtin fsmonitor--daemon"
+	elif test -n "$INTEGRATION_SCRIPT"
+	then
 		DESC="fsmonitor=$(basename $INTEGRATION_SCRIPT)"
 	else
 		DESC="fsmonitor=disabled"
@@ -261,11 +265,11 @@ test_fsmonitor_suite () {
 trace_start fsmonitor-watchman
 if test -n "$GIT_PERF_7519_FSMONITOR"; then
 	for INTEGRATION_PATH in $GIT_PERF_7519_FSMONITOR; do
-		test_expect_success "setup for fsmonitor $INTEGRATION_PATH" 'setup_for_fsmonitor'
+		test_expect_success "setup for fsmonitor $INTEGRATION_PATH" 'setup_for_fsmonitor_hook'
 		test_fsmonitor_suite
 	done
 else
-	test_expect_success "setup for fsmonitor" 'setup_for_fsmonitor'
+	test_expect_success "setup for fsmonitor hook" 'setup_for_fsmonitor_hook'
 	test_fsmonitor_suite
 fi
 
@@ -293,4 +297,30 @@ test_expect_success "setup without fsmonitor" '
 test_fsmonitor_suite
 trace_stop
 
+#
+# Run a full set of perf tests using the built-in fsmonitor--daemon.
+# It does not use the Hook API, so it has a different setup.
+# Explicitly start the daemon here and before we start client commands
+# so that we can later add custom tracing.
+#
+if test_have_prereq FSMONITOR_DAEMON
+then
+	USE_FSMONITOR_DAEMON=t
+
+	test_expect_success "setup for builtin fsmonitor" '
+		trace_start fsmonitor--daemon--server &&
+		git fsmonitor--daemon start &&
+
+		trace_start fsmonitor--daemon--client &&
+
+		git config core.fsmonitor true &&
+		git update-index --fsmonitor
+	'
+
+	test_fsmonitor_suite
+
+	git fsmonitor--daemon stop
+	trace_stop
+fi
+
 test_done
-- 
gitgitgadget

[PATCH v7 26/29] fsmonitor--daemon: periodically truncate list of modified files

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach fsmonitor--daemon to periodically truncate the list of
modified files to save some memory.

Clients will ask for the set of changes relative to a token that they
found in the FSMN index extension in the index.  (This token is like a
point in time, but different).  Clients will then update the index to
contain the response token (so that subsequent commands will be
relative to this new token).

Therefore, the daemon can gradually truncate the in-memory list of
changed paths as they become obsolete (older than the previous token).
Since we may have multiple clients making concurrent requests with a
skew of tokens and clients may be racing to the talk to the daemon,
we lazily truncate the list.

We introduce a 5 minute delay and truncate batches 5 minutes after
they are considered obsolete.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 builtin/fsmonitor--daemon.c | 88 +++++++++++++++++++++++++++++++++++++
 1 file changed, 88 insertions(+)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index fc3aee0ada0..edd00379cca 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -312,6 +312,75 @@ static void fsmonitor_batch__combine(struct fsmonitor_batch *batch_dest,
 			batch_src->interned_paths[k];
 }
 
+/*
+ * To keep the batch list from growing unbounded in response to filesystem
+ * activity, we try to truncate old batches from the end of the list as
+ * they become irrelevant.
+ *
+ * We assume that the .git/index will be updated with the most recent token
+ * any time the index is updated.  And future commands will only ask for
+ * recent changes *since* that new token.  So as tokens advance into the
+ * future, older batch items will never be requested/needed.  So we can
+ * truncate them without loss of functionality.
+ *
+ * However, multiple commands may be talking to the daemon concurrently
+ * or perform a slow command, so a little "token skew" is possible.
+ * Therefore, we want this to be a little bit lazy and have a generous
+ * delay.
+ *
+ * The current reader thread walked backwards in time from `token->batch_head`
+ * back to `batch_marker` somewhere in the middle of the batch list.
+ *
+ * Let's walk backwards in time from that marker an arbitrary delay
+ * and truncate the list there.  Note that these timestamps are completely
+ * artificial (based on when we pinned the batch item) and not on any
+ * filesystem activity.
+ *
+ * Return the obsolete portion of the list after we have removed it from
+ * the official list so that the caller can free it after leaving the lock.
+ */
+#define MY_TIME_DELAY_SECONDS (5 * 60) /* seconds */
+
+static struct fsmonitor_batch *with_lock__truncate_old_batches(
+	struct fsmonitor_daemon_state *state,
+	const struct fsmonitor_batch *batch_marker)
+{
+	/* assert current thread holding state->main_lock */
+
+	const struct fsmonitor_batch *batch;
+	struct fsmonitor_batch *remainder;
+
+	if (!batch_marker)
+		return NULL;
+
+	trace_printf_key(&trace_fsmonitor, "Truncate: mark (%"PRIu64",%"PRIu64")",
+			 batch_marker->batch_seq_nr,
+			 (uint64_t)batch_marker->pinned_time);
+
+	for (batch = batch_marker; batch; batch = batch->next) {
+		time_t t;
+
+		if (!batch->pinned_time) /* an overflow batch */
+			continue;
+
+		t = batch->pinned_time + MY_TIME_DELAY_SECONDS;
+		if (t > batch_marker->pinned_time) /* too close to marker */
+			continue;
+
+		goto truncate_past_here;
+	}
+
+	return NULL;
+
+truncate_past_here:
+	state->current_token_data->batch_tail = (struct fsmonitor_batch *)batch;
+
+	remainder = ((struct fsmonitor_batch *)batch)->next;
+	((struct fsmonitor_batch *)batch)->next = NULL;
+
+	return remainder;
+}
+
 static void fsmonitor_free_token_data(struct fsmonitor_token_data *token)
 {
 	if (!token)
@@ -425,6 +494,7 @@ static int do_handle_client(struct fsmonitor_daemon_state *state,
 	const char *p;
 	const struct fsmonitor_batch *batch_head;
 	const struct fsmonitor_batch *batch;
+	struct fsmonitor_batch *remainder = NULL;
 	intmax_t count = 0, duplicates = 0;
 	kh_str_t *shown;
 	int hash_ret;
@@ -654,11 +724,29 @@ static int do_handle_client(struct fsmonitor_daemon_state *state,
 			 * that work.
 			 */
 			fsmonitor_free_token_data(token_data);
+		} else if (batch) {
+			/*
+			 * We are holding the lock and are the only
+			 * reader of the ref-counted portion of the
+			 * list, so we get the honor of seeing if the
+			 * list can be truncated to save memory.
+			 *
+			 * The main loop did not walk to the end of the
+			 * list, so this batch is the first item in the
+			 * batch-list that is older than the requested
+			 * end-point sequence number.  See if the tail
+			 * end of the list is obsolete.
+			 */
+			remainder = with_lock__truncate_old_batches(state,
+								    batch);
 		}
 	}
 
 	pthread_mutex_unlock(&state->main_lock);
 
+	if (remainder)
+		fsmonitor_batch__free_list(remainder);
+
 	trace2_data_intmax("fsmonitor", the_repository, "response/length", total_response_len);
 	trace2_data_intmax("fsmonitor", the_repository, "response/count/files", count);
 	trace2_data_intmax("fsmonitor", the_repository, "response/count/duplicates", duplicates);
-- 
gitgitgadget

[PATCH v7 27/29] fsmonitor--daemon: use a cookie file to sync with file system

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach fsmonitor--daemon client threads to create a cookie file
inside the .git directory and then wait until FS events for the
cookie are observed by the FS listener thread.

This helps address the racy nature of file system events by
blocking the client response until the kernel has drained any
event backlog.

This is especially important on MacOS where kernel events are
only issued with a limited frequency.  See the `latency` argument
of `FSeventStreamCreate()`.  The kernel only signals every `latency`
seconds, but does not guarantee that the kernel queue is completely
drained, so we may have to wait more than one interval.  If we
increase the latency, the system is more likely to drop events.
We avoid these issues by having each client thread create a unique
cookie file and then wait until it is seen in the event stream.

Co-authored-by: Kevin Willford <Kevin.Willford@microsoft.com>
Co-authored-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 builtin/fsmonitor--daemon.c | 237 +++++++++++++++++++++++++++++++++++-
 fsmonitor--daemon.h         |   5 +
 2 files changed, 241 insertions(+), 1 deletion(-)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index edd00379cca..a30ecf6cfac 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -107,6 +107,162 @@ static int do_as_client__status(void)
 	}
 }
 
+enum fsmonitor_cookie_item_result {
+	FCIR_ERROR = -1, /* could not create cookie file ? */
+	FCIR_INIT,
+	FCIR_SEEN,
+	FCIR_ABORT,
+};
+
+struct fsmonitor_cookie_item {
+	struct hashmap_entry entry;
+	char *name;
+	enum fsmonitor_cookie_item_result result;
+};
+
+static int cookies_cmp(const void *data, const struct hashmap_entry *he1,
+		     const struct hashmap_entry *he2, const void *keydata)
+{
+	const struct fsmonitor_cookie_item *a =
+		container_of(he1, const struct fsmonitor_cookie_item, entry);
+	const struct fsmonitor_cookie_item *b =
+		container_of(he2, const struct fsmonitor_cookie_item, entry);
+
+	return strcmp(a->name, keydata ? keydata : b->name);
+}
+
+static enum fsmonitor_cookie_item_result with_lock__wait_for_cookie(
+	struct fsmonitor_daemon_state *state)
+{
+	/* assert current thread holding state->main_lock */
+
+	int fd;
+	struct fsmonitor_cookie_item *cookie;
+	struct strbuf cookie_pathname = STRBUF_INIT;
+	struct strbuf cookie_filename = STRBUF_INIT;
+	enum fsmonitor_cookie_item_result result;
+	int my_cookie_seq;
+
+	CALLOC_ARRAY(cookie, 1);
+
+	my_cookie_seq = state->cookie_seq++;
+
+	strbuf_addf(&cookie_filename, "%i-%i", getpid(), my_cookie_seq);
+
+	strbuf_addbuf(&cookie_pathname, &state->path_cookie_prefix);
+	strbuf_addbuf(&cookie_pathname, &cookie_filename);
+
+	cookie->name = strbuf_detach(&cookie_filename, NULL);
+	cookie->result = FCIR_INIT;
+	hashmap_entry_init(&cookie->entry, strhash(cookie->name));
+
+	hashmap_add(&state->cookies, &cookie->entry);
+
+	trace_printf_key(&trace_fsmonitor, "cookie-wait: '%s' '%s'",
+			 cookie->name, cookie_pathname.buf);
+
+	/*
+	 * Create the cookie file on disk and then wait for a notification
+	 * that the listener thread has seen it.
+	 */
+	fd = open(cookie_pathname.buf, O_WRONLY | O_CREAT | O_EXCL, 0600);
+	if (fd < 0) {
+		error_errno(_("could not create fsmonitor cookie '%s'"),
+			    cookie->name);
+
+		cookie->result = FCIR_ERROR;
+		goto done;
+	}
+
+	/*
+	 * Technically, close() and unlink() can fail, but we don't
+	 * care here.  We only created the file to trigger a watch
+	 * event from the FS to know that when we're up to date.
+	 */
+	close(fd);
+	unlink(cookie_pathname.buf);
+
+	/*
+	 * Technically, this is an infinite wait (well, unless another
+	 * thread sends us an abort).  I'd like to change this to
+	 * use `pthread_cond_timedwait()` and return an error/timeout
+	 * and let the caller do the trivial response thing, but we
+	 * don't have that routine in our thread-utils.
+	 *
+	 * After extensive beta testing I'm not really worried about
+	 * this.  Also note that the above open() and unlink() calls
+	 * will cause at least two FS events on that path, so the odds
+	 * of getting stuck are pretty slim.
+	 */
+	while (cookie->result == FCIR_INIT)
+		pthread_cond_wait(&state->cookies_cond,
+				  &state->main_lock);
+
+done:
+	hashmap_remove(&state->cookies, &cookie->entry, NULL);
+
+	result = cookie->result;
+
+	free(cookie->name);
+	free(cookie);
+	strbuf_release(&cookie_pathname);
+
+	return result;
+}
+
+/*
+ * Mark these cookies as _SEEN and wake up the corresponding client threads.
+ */
+static void with_lock__mark_cookies_seen(struct fsmonitor_daemon_state *state,
+					 const struct string_list *cookie_names)
+{
+	/* assert current thread holding state->main_lock */
+
+	int k;
+	int nr_seen = 0;
+
+	for (k = 0; k < cookie_names->nr; k++) {
+		struct fsmonitor_cookie_item key;
+		struct fsmonitor_cookie_item *cookie;
+
+		key.name = cookie_names->items[k].string;
+		hashmap_entry_init(&key.entry, strhash(key.name));
+
+		cookie = hashmap_get_entry(&state->cookies, &key, entry, NULL);
+		if (cookie) {
+			trace_printf_key(&trace_fsmonitor, "cookie-seen: '%s'",
+					 cookie->name);
+			cookie->result = FCIR_SEEN;
+			nr_seen++;
+		}
+	}
+
+	if (nr_seen)
+		pthread_cond_broadcast(&state->cookies_cond);
+}
+
+/*
+ * Set _ABORT on all pending cookies and wake up all client threads.
+ */
+static void with_lock__abort_all_cookies(struct fsmonitor_daemon_state *state)
+{
+	/* assert current thread holding state->main_lock */
+
+	struct hashmap_iter iter;
+	struct fsmonitor_cookie_item *cookie;
+	int nr_aborted = 0;
+
+	hashmap_for_each_entry(&state->cookies, &iter, cookie, entry) {
+		trace_printf_key(&trace_fsmonitor, "cookie-abort: '%s'",
+				 cookie->name);
+		cookie->result = FCIR_ABORT;
+		nr_aborted++;
+	}
+
+	if (nr_aborted)
+		pthread_cond_broadcast(&state->cookies_cond);
+}
+
 /*
  * Requests to and from a FSMonitor Protocol V2 provider use an opaque
  * "token" as a virtual timestamp.  Clients can request a summary of all
@@ -404,6 +560,9 @@ static void fsmonitor_free_token_data(struct fsmonitor_token_data *token)
  *     We should create a new token and start fresh (as if we just
  *     booted up).
  *
+ * [2] Some of those lost events may have been for cookie files.  We
+ *     should assume the worst and abort them rather letting them starve.
+ *
  * If there are no concurrent threads reading the current token data
  * series, we can free it now.  Otherwise, let the last reader free
  * it.
@@ -425,6 +584,8 @@ static void with_lock__do_force_resync(struct fsmonitor_daemon_state *state)
 	state->current_token_data = new_one;
 
 	fsmonitor_free_token_data(free_me);
+
+	with_lock__abort_all_cookies(state);
 }
 
 void fsmonitor_force_resync(struct fsmonitor_daemon_state *state)
@@ -500,6 +661,8 @@ static int do_handle_client(struct fsmonitor_daemon_state *state,
 	int hash_ret;
 	int do_trivial = 0;
 	int do_flush = 0;
+	int do_cookie = 0;
+	enum fsmonitor_cookie_item_result cookie_result;
 
 	/*
 	 * We expect `command` to be of the form:
@@ -560,6 +723,7 @@ static int do_handle_client(struct fsmonitor_daemon_state *state,
 			 * We have a V2 valid token:
 			 *     "builtin:<token_id>:<seq_nr>"
 			 */
+			do_cookie = 1;
 		}
 	}
 
@@ -568,6 +732,30 @@ static int do_handle_client(struct fsmonitor_daemon_state *state,
 	if (!state->current_token_data)
 		BUG("fsmonitor state does not have a current token");
 
+	/*
+	 * Write a cookie file inside the directory being watched in
+	 * an effort to flush out existing filesystem events that we
+	 * actually care about.  Suspend this client thread until we
+	 * see the filesystem events for this cookie file.
+	 *
+	 * Creating the cookie lets us guarantee that our FS listener
+	 * thread has drained the kernel queue and we are caught up
+	 * with the kernel.
+	 *
+	 * If we cannot create the cookie (or otherwise guarantee that
+	 * we are caught up), we send a trivial response.  We have to
+	 * assume that there might be some very, very recent activity
+	 * on the FS still in flight.
+	 */
+	if (do_cookie) {
+		cookie_result = with_lock__wait_for_cookie(state);
+		if (cookie_result != FCIR_SEEN) {
+			error(_("fsmonitor: cookie_result '%d' != SEEN"),
+			      cookie_result);
+			do_trivial = 1;
+		}
+	}
+
 	if (do_flush)
 		with_lock__do_force_resync(state);
 
@@ -788,7 +976,9 @@ static int handle_client(void *data,
 	return result;
 }
 
-#define FSMONITOR_COOKIE_PREFIX ".fsmonitor-daemon-"
+#define FSMONITOR_DIR           "fsmonitor--daemon"
+#define FSMONITOR_COOKIE_DIR    "cookies"
+#define FSMONITOR_COOKIE_PREFIX (FSMONITOR_DIR "/" FSMONITOR_COOKIE_DIR "/")
 
 enum fsmonitor_path_type fsmonitor_classify_path_workdir_relative(
 	const char *rel)
@@ -941,6 +1131,9 @@ void fsmonitor_publish(struct fsmonitor_daemon_state *state,
 		}
 	}
 
+	if (cookie_names->nr)
+		with_lock__mark_cookies_seen(state, cookie_names);
+
 	pthread_mutex_unlock(&state->main_lock);
 }
 
@@ -1032,7 +1225,9 @@ static int fsmonitor_run_daemon(void)
 
 	memset(&state, 0, sizeof(state));
 
+	hashmap_init(&state.cookies, cookies_cmp, NULL, 0);
 	pthread_mutex_init(&state.main_lock, NULL);
+	pthread_cond_init(&state.cookies_cond, NULL);
 	state.error_code = 0;
 	state.current_token_data = fsmonitor_new_token_data();
 
@@ -1057,6 +1252,44 @@ static int fsmonitor_run_daemon(void)
 		state.nr_paths_watching = 2;
 	}
 
+	/*
+	 * We will write filesystem syncing cookie files into
+	 * <gitdir>/<fsmonitor-dir>/<cookie-dir>/<pid>-<seq>.
+	 *
+	 * The extra layers of subdirectories here keep us from
+	 * changing the mtime on ".git/" or ".git/foo/" when we create
+	 * or delete cookie files.
+	 *
+	 * There have been problems with some IDEs that do a
+	 * non-recursive watch of the ".git/" directory and run a
+	 * series of commands any time something happens.
+	 *
+	 * For example, if we place our cookie files directly in
+	 * ".git/" or ".git/foo/" then a `git status` (or similar
+	 * command) from the IDE will cause a cookie file to be
+	 * created in one of those dirs.  This causes the mtime of
+	 * those dirs to change.  This triggers the IDE's watch
+	 * notification.  This triggers the IDE to run those commands
+	 * again.  And the process repeats and the machine never goes
+	 * idle.
+	 *
+	 * Adding the extra layers of subdirectories prevents the
+	 * mtime of ".git/" and ".git/foo" from changing when a
+	 * cookie file is created.
+	 */
+	strbuf_init(&state.path_cookie_prefix, 0);
+	strbuf_addbuf(&state.path_cookie_prefix, &state.path_gitdir_watch);
+
+	strbuf_addch(&state.path_cookie_prefix, '/');
+	strbuf_addstr(&state.path_cookie_prefix, FSMONITOR_DIR);
+	mkdir(state.path_cookie_prefix.buf, 0777);
+
+	strbuf_addch(&state.path_cookie_prefix, '/');
+	strbuf_addstr(&state.path_cookie_prefix, FSMONITOR_COOKIE_DIR);
+	mkdir(state.path_cookie_prefix.buf, 0777);
+
+	strbuf_addch(&state.path_cookie_prefix, '/');
+
 	/*
 	 * Confirm that we can create platform-specific resources for the
 	 * filesystem listener before we bother starting all the threads.
@@ -1069,6 +1302,7 @@ static int fsmonitor_run_daemon(void)
 	err = fsmonitor_run_daemon_1(&state);
 
 done:
+	pthread_cond_destroy(&state.cookies_cond);
 	pthread_mutex_destroy(&state.main_lock);
 	fsm_listen__dtor(&state);
 
@@ -1076,6 +1310,7 @@ done:
 
 	strbuf_release(&state.path_worktree_watch);
 	strbuf_release(&state.path_gitdir_watch);
+	strbuf_release(&state.path_cookie_prefix);
 
 	return err;
 }
diff --git a/fsmonitor--daemon.h b/fsmonitor--daemon.h
index 010fbfe60e9..bd09fffc176 100644
--- a/fsmonitor--daemon.h
+++ b/fsmonitor--daemon.h
@@ -45,6 +45,11 @@ struct fsmonitor_daemon_state {
 
 	struct fsmonitor_token_data *current_token_data;
 
+	struct strbuf path_cookie_prefix;
+	pthread_cond_t cookies_cond;
+	int cookie_seq;
+	struct hashmap cookies;
+
 	int error_code;
 	struct fsmonitor_daemon_backend_data *backend_data;
 
-- 
gitgitgadget

[PATCH v7 28/29] fsmonitor: force update index after large responses

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Measure the time taken to apply the FSMonitor query result
to the index and the untracked-cache.

Set the `FSMONITOR_CHANGED` bit on `istate->cache_changed` when
FSMonitor returns a very large repsonse to ensure that the index is
written to disk.

Normally, when the FSMonitor response includes a tracked file, the
index is always updated.  Similarly, the index might be updated when
the response alters the untracked-cache (when enabled).  However, in
cases where neither of those cause the index to be considered changed,
the FSMonitor response is wasted.  Subsequent Git commands will make
requests with the same token and receive the same response.

If that response is very large, performance may suffer.  It would be
more efficient to force update the index now (and the token in the
index extension) in order to reduce the size of the response received
by future commands.

This was observed on Windows after a large checkout.  On Windows, the
kernel emits events for the files that are changed as they are
changed.  However, it might delay events for the containing
directories until the system is more idle (or someone scans the
directory (so it seems)).  The first status following a checkout would
get the list of files.  The subsequent status commands would get the
list of directories as the events trickled out.  But they would never
catch up because the token was not advanced because the index wasn't
updated.

This list of directories caused `wt_status_collect_untracked()` to
unnecessarily spend time actually scanning them during each command.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 fsmonitor.c | 55 ++++++++++++++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 54 insertions(+), 1 deletion(-)

diff --git a/fsmonitor.c b/fsmonitor.c
index a38b5710eb3..292a6742b4f 100644
--- a/fsmonitor.c
+++ b/fsmonitor.c
@@ -219,6 +219,43 @@ static void fsmonitor_refresh_callback(struct index_state *istate, char *name)
 	untracked_cache_invalidate_path(istate, name, 0);
 }
 
+/*
+ * The number of pathnames that we need to receive from FSMonitor
+ * before we force the index to be updated.
+ *
+ * Note that any pathname within the set of received paths MAY cause
+ * cache-entry or istate flag bits to be updated and thus cause the
+ * index to be updated on disk.
+ *
+ * However, the response may contain many paths (such as ignored
+ * paths) that will not update any flag bits.  And thus not force the
+ * index to be updated.  (This is fine and normal.)  It also means
+ * that the token will not be updated in the FSMonitor index
+ * extension.  So the next Git command will find the same token in the
+ * index, make the same token-relative request, and receive the same
+ * response (plus any newly changed paths).  If this response is large
+ * (and continues to grow), performance could be impacted.
+ *
+ * For example, if the user runs a build and it writes 100K object
+ * files but doesn't modify any source files, the index would not need
+ * to be updated.  The FSMonitor response (after the build and
+ * relative to a pre-build token) might be 5MB.  Each subsequent Git
+ * command will receive that same 100K/5MB response until something
+ * causes the index to be updated.  And `refresh_fsmonitor()` will
+ * have to iterate over those 100K paths each time.
+ *
+ * Performance could be improved if we optionally force update the
+ * index after a very large response and get an updated token into
+ * the FSMonitor index extension.  This should allow subsequent
+ * commands to get smaller and more current responses.
+ *
+ * The value chosen here does not need to be precise.  The index
+ * will be updated automatically the first time the user touches
+ * a tracked file and causes a command like `git status` to
+ * update an mtime to be updated and/or set a flag bit.
+ */
+static int fsmonitor_force_update_threshold = 100;
+
 void refresh_fsmonitor(struct index_state *istate)
 {
 	struct strbuf query_result = STRBUF_INIT;
@@ -362,25 +399,39 @@ apply_results:
 	 *     information and that we should consider everything
 	 *     invalid.  We call this a trivial response.
 	 */
+	trace2_region_enter("fsmonitor", "apply_results", istate->repo);
+
 	if (query_success && !is_trivial) {
 		/*
 		 * Mark all pathnames returned by the monitor as dirty.
 		 *
 		 * This updates both the cache-entries and the untracked-cache.
 		 */
+		int count = 0;
+
 		buf = query_result.buf;
 		for (i = bol; i < query_result.len; i++) {
 			if (buf[i] != '\0')
 				continue;
 			fsmonitor_refresh_callback(istate, buf + bol);
 			bol = i + 1;
+			count++;
 		}
-		if (bol < query_result.len)
+		if (bol < query_result.len) {
 			fsmonitor_refresh_callback(istate, buf + bol);
+			count++;
+		}
 
 		/* Now mark the untracked cache for fsmonitor usage */
 		if (istate->untracked)
 			istate->untracked->use_fsmonitor = 1;
+
+		if (count > fsmonitor_force_update_threshold)
+			istate->cache_changed |= FSMONITOR_CHANGED;
+
+		trace2_data_intmax("fsmonitor", istate->repo, "apply_count",
+				   count);
+
 	} else {
 		/*
 		 * We failed to get a response or received a trivial response,
@@ -409,6 +460,8 @@ apply_results:
 		if (istate->untracked)
 			istate->untracked->use_fsmonitor = 0;
 	}
+	trace2_region_leave("fsmonitor", "apply_results", istate->repo);
+
 	strbuf_release(&query_result);
 
 	/* Now that we've updated istate, save the last_update_token */
-- 
gitgitgadget

[PATCH v7 29/29] t7527: test status with untracked-cache and fsmonitor--daemon

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Create 2x2 test matrix with the untracked-cache and fsmonitor--daemon
features and a series of edits and verify that status output is
identical.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 t/t7527-builtin-fsmonitor.sh | 93 ++++++++++++++++++++++++++++++++++++
 1 file changed, 93 insertions(+)

diff --git a/t/t7527-builtin-fsmonitor.sh b/t/t7527-builtin-fsmonitor.sh
index d79635e7596..389ebf431c6 100755
--- a/t/t7527-builtin-fsmonitor.sh
+++ b/t/t7527-builtin-fsmonitor.sh
@@ -212,6 +212,8 @@ test_expect_success 'setup' '
 	.gitignore
 	expect*
 	actual*
+	flush*
+	trace*
 	EOF
 
 	git -c core.fsmonitor=false add . &&
@@ -498,4 +500,95 @@ test_expect_success 'cleanup worktrees' '
 	stop_daemon_delete_repo wt-base
 '
 
+# The next few tests perform arbitrary/contrived file operations and
+# confirm that status is correct.  That is, that the data (or lack of
+# data) from fsmonitor doesn't cause incorrect results.  And doesn't
+# cause incorrect results when the untracked-cache is enabled.
+
+test_lazy_prereq UNTRACKED_CACHE '
+	git update-index --test-untracked-cache
+'
+
+test_expect_success 'Matrix: setup for untracked-cache,fsmonitor matrix' '
+	test_unconfig core.fsmonitor &&
+	git update-index --no-fsmonitor &&
+	test_might_fail git fsmonitor--daemon stop
+'
+
+matrix_clean_up_repo () {
+	git reset --hard HEAD &&
+	git clean -fd
+}
+
+matrix_try () {
+	uc=$1 &&
+	fsm=$2 &&
+	fn=$3 &&
+
+	test_expect_success "Matrix[uc:$uc][fsm:$fsm] $fn" '
+		matrix_clean_up_repo &&
+		$fn &&
+		if test $uc = false && test $fsm = false
+		then
+			git status --porcelain=v1 >.git/expect.$fn
+		else
+			git status --porcelain=v1 >.git/actual.$fn &&
+			test_cmp .git/expect.$fn .git/actual.$fn
+		fi
+	'
+}
+
+uc_values="false"
+test_have_prereq UNTRACKED_CACHE && uc_values="false true"
+for uc_val in $uc_values
+do
+	if test $uc_val = false
+	then
+		test_expect_success "Matrix[uc:$uc_val] disable untracked cache" '
+			git config core.untrackedcache false &&
+			git update-index --no-untracked-cache
+		'
+	else
+		test_expect_success "Matrix[uc:$uc_val] enable untracked cache" '
+			git config core.untrackedcache true &&
+			git update-index --untracked-cache
+		'
+	fi
+
+	fsm_values="false true"
+	for fsm_val in $fsm_values
+	do
+		if test $fsm_val = false
+		then
+			test_expect_success "Matrix[uc:$uc_val][fsm:$fsm_val] disable fsmonitor" '
+				test_unconfig core.fsmonitor &&
+				git update-index --no-fsmonitor &&
+				test_might_fail git fsmonitor--daemon stop
+			'
+		else
+			test_expect_success "Matrix[uc:$uc_val][fsm:$fsm_val] enable fsmonitor" '
+				git config core.fsmonitor true &&
+				git fsmonitor--daemon start &&
+				git update-index --fsmonitor
+			'
+		fi
+
+		matrix_try $uc_val $fsm_val edit_files
+		matrix_try $uc_val $fsm_val delete_files
+		matrix_try $uc_val $fsm_val create_files
+		matrix_try $uc_val $fsm_val rename_files
+		matrix_try $uc_val $fsm_val file_to_directory
+		matrix_try $uc_val $fsm_val directory_to_file
+
+		if test $fsm_val = true
+		then
+			test_expect_success "Matrix[uc:$uc_val][fsm:$fsm_val] disable fsmonitor at end" '
+				test_unconfig core.fsmonitor &&
+				git update-index --no-fsmonitor &&
+				test_might_fail git fsmonitor--daemon stop
+			'
+		fi
+	done
+done
+
 test_done
-- 
gitgitgadget

Re: [PATCH v7 16/29] compat/fsmonitor/fsm-listen-darwin: add MacOS header files for FSEvent

[Ævar Arnfjörð Bjarmason]

On Tue, Mar 22 2022, Jeff Hostetler via GitGitGadget wrote:

> From: Jeff Hostetler <jeffhost@microsoft.com>
> 
>  compat/fsmonitor/fsm-darwin-gcc.h    | 92 ++++++++++++++++++++++++++++

It's much nicer to have this compat-for-the-compat in its own
header. Thanks.

>  compat/fsmonitor/fsm-listen-darwin.c | 24 ++++++++
>  2 files changed, 116 insertions(+)
>  create mode 100644 compat/fsmonitor/fsm-darwin-gcc.h
>
> diff --git a/compat/fsmonitor/fsm-darwin-gcc.h b/compat/fsmonitor/fsm-darwin-gcc.h
> new file mode 100644
> index 00000000000..1c75c3d48e7
> --- /dev/null
> +++ b/compat/fsmonitor/fsm-darwin-gcc.h
> @@ -0,0 +1,92 @@
> +#ifndef FSM_DARWIN_GCC_H
> +#define FSM_DARWIN_GCC_H
> +
> +#ifndef __clang__

This was surprising, until I remembered that clang tries really hard to
pretend to be other compilers. I wonder if we should steal the macro
check from compat/compiler.h into something more generic & use it here,
probably best as a follow-up...

> [...]
> +#endif /* !clang */
> +#endif /* FSM_DARWIN_GCC_H */
> diff --git a/compat/fsmonitor/fsm-listen-darwin.c b/compat/fsmonitor/fsm-listen-darwin.c
> index c84e3344ab9..d2ce942cade 100644
> --- a/compat/fsmonitor/fsm-listen-darwin.c
> +++ b/compat/fsmonitor/fsm-listen-darwin.c
> @@ -1,3 +1,27 @@
> +#ifndef __clang__
> +#include "fsm-darwin-gcc.h"
> +#else
> +#include <CoreFoundation/CoreFoundation.h>
> +#include <CoreServices/CoreServices.h>
> +
> [...]

Re: [PATCH v7 18/29] fsmonitor--daemon: implement handle_client callback

[Ævar Arnfjörð Bjarmason]

On Tue, Mar 22 2022, Jeff Hostetler via GitGitGadget wrote:

> From: Jeff Hostetler <jeffhost@microsoft.com>
> [...]
> +static int do_handle_client(struct fsmonitor_daemon_state *state,
> +			    const char *command,
> +			    ipc_server_reply_cb *reply,
> +			    struct ipc_server_reply_data *reply_data)
> +{
> +	struct fsmonitor_token_data *token_data = NULL;
> +	struct strbuf response_token = STRBUF_INIT;
> +	struct strbuf requested_token_id = STRBUF_INIT;
> +	struct strbuf payload = STRBUF_INIT;
> +	uint64_t requested_oldest_seq_nr = 0;
> +	uint64_t total_response_len = 0;
> +	const char *p;
> +	const struct fsmonitor_batch *batch_head;
> +	const struct fsmonitor_batch *batch;
> +	intmax_t count = 0, duplicates = 0;
> +	kh_str_t *shown;
> +	int hash_ret;
> +	int do_trivial = 0;
> +	int do_flush = 0;

Just noticed while skimming, isn't initializing do_trivial here (didn't
exhaustively check the others) getting in the way of uninitialized
analysis the compiler will do...

> +
> +	/*
> +	 * We expect `command` to be of the form:
> +	 *
> +	 * <command> := quit NUL
> +	 *            | flush NUL
> +	 *            | <V1-time-since-epoch-ns> NUL
> +	 *            | <V2-opaque-fsmonitor-token> NUL
> +	 */
> +
> +	if (!strcmp(command, "quit")) {
> +		/*
> +		 * A client has requested over the socket/pipe that the
> +		 * daemon shutdown.
> +		 *
> +		 * Tell the IPC thread pool to shutdown (which completes
> +		 * the await in the main thread (which can stop the
> +		 * fsmonitor listener thread)).
> +		 *
> +		 * There is no reply to the client.
> +		 */
> +		return SIMPLE_IPC_QUIT;
> +
> +	} else if (!strcmp(command, "flush")) {
> +		/*
> +		 * Flush all of our cached data and generate a new token
> +		 * just like if we lost sync with the filesystem.
> +		 *
> +		 * Then send a trivial response using the new token.
> +		 */
> +		do_flush = 1;
> +		do_trivial = 1;
> +
> +	} else if (!skip_prefix(command, "builtin:", &p)) {
> +		/* assume V1 timestamp or garbage */
> +
> +		char *p_end;
> +
> +		strtoumax(command, &p_end, 10);
> +		trace_printf_key(&trace_fsmonitor,
> +				 ((*p_end) ?
> +				  "fsmonitor: invalid command line '%s'" :
> +				  "fsmonitor: unsupported V1 protocol '%s'"),
> +				 command);
> +		do_trivial = 1;
> +
> +	} else {
> +		/* We have "builtin:*" */
> +		if (fsmonitor_parse_client_token(command, &requested_token_id,
> +						 &requested_oldest_seq_nr)) {
> +			trace_printf_key(&trace_fsmonitor,
> +					 "fsmonitor: invalid V2 protocol token '%s'",
> +					 command);
> +			do_trivial = 1;
> +
> +		} else {
> +			/*
> +			 * We have a V2 valid token:
> +			 *     "builtin:<token_id>:<seq_nr>"
> +			 */
> +		}
> +	}

Since we'll set it here in all branches except the "else" branch,
i.e. if you move this to "we have a v2 valid" we'll catch future bugs if
we ever have it uninitialized still, 

> +
> +	pthread_mutex_lock(&state->main_lock);
> +
> +	if (!state->current_token_data)
> +		BUG("fsmonitor state does not have a current token");
> +
> +	if (do_flush)
> +		with_lock__do_force_resync(state);
> +
> +	/*
> +	 * We mark the current head of the batch list as "pinned" so
> +	 * that the listener thread will treat this item as read-only
> +	 * (and prevent any more paths from being added to it) from
> +	 * now on.
> +	 */
> +	token_data = state->current_token_data;
> +	batch_head = token_data->batch_head;
> +	((struct fsmonitor_batch *)batch_head)->pinned_time = time(NULL);
> +
> +	/*
> +	 * FSMonitor Protocol V2 requires that we send a response header
> +	 * with a "new current token" and then all of the paths that changed
> +	 * since the "requested token".  We send the seq_nr of the just-pinned
> +	 * head batch so that future requests from a client will be relative
> +	 * to it.
> +	 */
> +	with_lock__format_response_token(&response_token,
> +					 &token_data->token_id, batch_head);
> +
> +	reply(reply_data, response_token.buf, response_token.len + 1);
> +	total_response_len += response_token.len + 1;
> +
> +	trace2_data_string("fsmonitor", the_repository, "response/token",
> +			   response_token.buf);
> +	trace_printf_key(&trace_fsmonitor, "response token: %s",
> +			 response_token.buf);
> +
> +	if (!do_trivial) {

I.e. this would start warning.

> +		if (strcmp(requested_token_id.buf, token_data->token_id.buf)) {
> +			/*
> +			 * The client last spoke to a different daemon
> +			 * instance -OR- the daemon had to resync with
> +			 * the filesystem (and lost events), so reject.
> +			 */
> +			trace2_data_string("fsmonitor", the_repository,
> +					   "response/token", "different");
> +			do_trivial = 1;
> +
> +		} else if (requested_oldest_seq_nr <
> +			   token_data->batch_tail->batch_seq_nr) {
> +			/*
> +			 * The client wants older events than we have for
> +			 * this token_id.  This means that the end of our
> +			 * batch list was truncated and we cannot give the
> +			 * client a complete snapshot relative to their
> +			 * request.
> +			 */
> +			trace_printf_key(&trace_fsmonitor,
> +					 "client requested truncated data");
> +			do_trivial = 1;
> +		}
> +	}
> +
> +	if (do_trivial) {
> +		pthread_mutex_unlock(&state->main_lock);
> +
> +		reply(reply_data, "/", 2);
> +
> +		trace2_data_intmax("fsmonitor", the_repository,
> +				   "response/trivial", 1);
> +
> +		strbuf_release(&response_token);
> +		strbuf_release(&requested_token_id);
> +		return 0;

Nit: instead of strbuf_release() here, just a 'goto cleanup' and ...

> +	}
> +
> +	/*
> +	 * We're going to hold onto a pointer to the current
> +	 * token-data while we walk the list of batches of files.
> +	 * During this time, we will NOT be under the lock.
> +	 * So we ref-count it.
> +	 *
> +	 * This allows the listener thread to continue prepending
> +	 * new batches of items to the token-data (which we'll ignore).
> +	 *
> +	 * AND it allows the listener thread to do a token-reset
> +	 * (and install a new `current_token_data`).
> +	 */
> +	token_data->client_ref_count++;
> +
> +	pthread_mutex_unlock(&state->main_lock);
> +
> +	/*
> +	 * The client request is relative to the token that they sent,
> +	 * so walk the batch list backwards from the current head back
> +	 * to the batch (sequence number) they named.
> +	 *
> +	 * We use khash to de-dup the list of pathnames.
> +	 *
> +	 * NEEDSWORK: each batch contains a list of interned strings,
> +	 * so we only need to do pointer comparisons here to build the
> +	 * hash table.  Currently, we're still comparing the string
> +	 * values.
> +	 */
> +	shown = kh_init_str();
> +	for (batch = batch_head;
> +	     batch && batch->batch_seq_nr > requested_oldest_seq_nr;
> +	     batch = batch->next) {
> +		size_t k;
> +
> +		for (k = 0; k < batch->nr; k++) {
> +			const char *s = batch->interned_paths[k];
> +			size_t s_len;
> +
> +			if (kh_get_str(shown, s) != kh_end(shown))
> +				duplicates++;
> +			else {
> +				kh_put_str(shown, s, &hash_ret);
> +
> +				trace_printf_key(&trace_fsmonitor,
> +						 "send[%"PRIuMAX"]: %s",
> +						 count, s);
> +
> +				/* Each path gets written with a trailing NUL */
> +				s_len = strlen(s) + 1;
> +
> +				if (payload.len + s_len >=
> +				    LARGE_PACKET_DATA_MAX) {
> +					reply(reply_data, payload.buf,
> +					      payload.len);
> +					total_response_len += payload.len;
> +					strbuf_reset(&payload);
> +				}
> +
> +				strbuf_add(&payload, s, s_len);
> +				count++;
> +			}
> +		}
> +	}
> +
> +	if (payload.len) {
> +		reply(reply_data, payload.buf, payload.len);
> +		total_response_len += payload.len;
> +	}
> +
> +	kh_release_str(shown);
> +
> +	pthread_mutex_lock(&state->main_lock);
> +
> +	if (token_data->client_ref_count > 0)
> +		token_data->client_ref_count--;
> +
> +	if (token_data->client_ref_count == 0) {
> +		if (token_data != state->current_token_data) {
> +			/*
> +			 * The listener thread did a token-reset while we were
> +			 * walking the batch list.  Therefore, this token is
> +			 * stale and can be discarded completely.  If we are
> +			 * the last reader thread using this token, we own
> +			 * that work.
> +			 */
> +			fsmonitor_free_token_data(token_data);
> +		}
> +	}
> +
> +	pthread_mutex_unlock(&state->main_lock);
> +
> +	trace2_data_intmax("fsmonitor", the_repository, "response/length", total_response_len);
> +	trace2_data_intmax("fsmonitor", the_repository, "response/count/files", count);
> +	trace2_data_intmax("fsmonitor", the_repository, "response/count/duplicates", duplicates);
> +

add a "cleanup" label here?

> +	strbuf_release(&response_token);
> +	strbuf_release(&requested_token_id);
> +	strbuf_release(&payload);
> +
> +	return 0;
> +}
> +
>  static ipc_server_application_cb handle_client;

Re: [PATCH v7 21/29] t7527: create test for fsmonitor--daemon

[Ævar Arnfjörð Bjarmason]

On Tue, Mar 22 2022, Jeff Hostetler via GitGitGadget wrote:

> From: Jeff Hostetler <jeffhost@microsoft.com>
>
> Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
> Signed-off-by: Junio C Hamano <gitster@pobox.com>
> ---
>  t/t7527-builtin-fsmonitor.sh | 501 +++++++++++++++++++++++++++++++++++
>  1 file changed, 501 insertions(+)
>  create mode 100755 t/t7527-builtin-fsmonitor.sh
>
> diff --git a/t/t7527-builtin-fsmonitor.sh b/t/t7527-builtin-fsmonitor.sh
> new file mode 100755
> index 00000000000..d79635e7596
> --- /dev/null
> +++ b/t/t7527-builtin-fsmonitor.sh
> @@ -0,0 +1,501 @@
> +#!/bin/sh
> +
> +test_description='built-in file system watcher'
> +
> +. ./test-lib.sh
> +
> +if ! test_have_prereq FSMONITOR_DAEMON
> +then
> +	skip_all="fsmonitor--daemon is not supported on this platform"
> +	test_done
> +fi
> +
> +stop_daemon_delete_repo () {
> +	r=$1 &&
> +	test_might_fail git -C $r fsmonitor--daemon stop &&
> +	rm -rf $1
> +}
> +
> +start_daemon () {
> +	r= &&
> +	tf= &&
> +	t2= &&
> +	tk= &&

I think these all have &&-chains added since previous versions, good...

> +	while test "$#" -ne 0
> +	do
> +		case "$1" in
> +		-C)
> +			shift;
> +			test "$#" -ne 0 || BUG "error: -C requires arg"
> +			r="-C $1"
> +			shift
> +			;;
> +		-tf)
> +			shift;
> +			test "$#" -ne 0 || BUG "error: -tf requires arg"
> +			tf="$1"
> +			shift
> +			;;
> +		-t2)
> +			shift;
> +			test "$#" -ne 0 || BUG "error: -t2 requires arg"
> +			t2="$1"
> +			shift
> +			;;
> +		-tk)
> +			shift;
> +			test "$#" -ne 0 || BUG "error: -tk requires arg"
> +			tk="$1"
> +			shift
> +			;;

But (and IIRC I noted this in a previous iteration) if you &&-chain the
"shift" here you can lose the more verbose BUG

> +	start_daemon -tf "$PWD/.git/trace" &&

FWIW having an option parser take -tf to mean --tf is quite unlike our
common conventions, usually it means both -t and -f.

In this case every single caller added here does provide -tf
argument. Perhaps better as as unconditional $1 then?

Re: [PATCH v7 24/29] t/perf/p7519: speed up test on Windows

[Ævar Arnfjörð Bjarmason]

On Tue, Mar 22 2022, Jeff Hostetler via GitGitGadget wrote:

> From: Jeff Hostetler <jeffhost@microsoft.com>
>
> Change p7519 to use `test_seq` and `xargs` rather than a `for` loop
> to touch thousands of files.  This takes minutes off of test runs
> on Windows because of process creation overhead.
>
> Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
> Signed-off-by: Junio C Hamano <gitster@pobox.com>
> ---
>  t/perf/p7519-fsmonitor.sh | 32 ++++++++++++++++++++------------
>  1 file changed, 20 insertions(+), 12 deletions(-)
>
> diff --git a/t/perf/p7519-fsmonitor.sh b/t/perf/p7519-fsmonitor.sh
> index c8be58f3c76..0611e533951 100755
> --- a/t/perf/p7519-fsmonitor.sh
> +++ b/t/perf/p7519-fsmonitor.sh
> @@ -72,7 +72,7 @@ then
>  	fi
>  fi
>  
> -trace_start() {
> +trace_start () {
>  	if test -n "$GIT_PERF_7519_TRACE"
>  	then
>  		name="$1"
> @@ -91,13 +91,20 @@ trace_start() {
>  	fi
>  }
>  
> -trace_stop() {
> +trace_stop () {
>  	if test -n "$GIT_PERF_7519_TRACE"
>  	then
>  		unset GIT_TRACE2_PERF
>  	fi
>  }

(I think I noted in a previous version): Would be nice to have an
optimization change not do unrelated refactoring....

> +touch_files () {
> +	n=$1
> +	d="$n"_files
> +
> +	(cd $d ; test_seq 1 $n | xargs touch )


...and here we don't &&-chain.

> +}
> +
>  test_expect_success "one time repo setup" '
>  	# set untrackedCache depending on the environment
>  	if test -n "$GIT_PERF_7519_UNTRACKED_CACHE"
> @@ -119,10 +126,11 @@ test_expect_success "one time repo setup" '
>  	fi &&
>  
>  	mkdir 1_file 10_files 100_files 1000_files 10000_files &&
> -	for i in $(test_seq 1 10); do touch 10_files/$i || return 1; done &&
> -	for i in $(test_seq 1 100); do touch 100_files/$i || return 1; done &&
> -	for i in $(test_seq 1 1000); do touch 1000_files/$i || return 1; done &&
> -	for i in $(test_seq 1 10000); do touch 10000_files/$i || return 1; done &&
> +	: 1_file directory should be left empty &&
> +	touch_files 10 &&
> +	touch_files 100 &&
> +	touch_files 1000 &&
> +	touch_files 10000 &&
>  	git add 1_file 10_files 100_files 1000_files 10000_files &&
>  	git commit -qm "Add files" &&
>  
> @@ -133,7 +141,7 @@ test_expect_success "one time repo setup" '
>  	fi
>  '
>  
> -setup_for_fsmonitor() {
> +setup_for_fsmonitor () {
>  	# set INTEGRATION_SCRIPT depending on the environment
>  	if test -n "$INTEGRATION_PATH"
>  	then
> @@ -173,7 +181,7 @@ test_perf_w_drop_caches () {
>  	test_perf "$@"
>  }
>  
> -test_fsmonitor_suite() {
> +test_fsmonitor_suite () {

ditto refactoring..

Re: [PATCH v7 16/29] compat/fsmonitor/fsm-listen-darwin: add MacOS header files for FSEvent

[Jeff Hostetler]

On 3/22/22 2:19 PM, Ævar Arnfjörð Bjarmason wrote:
> 
> On Tue, Mar 22 2022, Jeff Hostetler via GitGitGadget wrote:
> 
>> From: Jeff Hostetler <jeffhost@microsoft.com>
[...]
>> +++ b/compat/fsmonitor/fsm-darwin-gcc.h
>> @@ -0,0 +1,92 @@
>> +#ifndef FSM_DARWIN_GCC_H
>> +#define FSM_DARWIN_GCC_H
>> +
>> +#ifndef __clang__
> 
> This was surprising, until I remembered that clang tries really hard to
> pretend to be other compilers. I wonder if we should steal the macro
> check from compat/compiler.h into something more generic & use it here,
> probably best as a follow-up...
> 
>> [...]

yeah, V5 had this as an ifdef __GNUC__, I changed it in V6 because
we were always using the hack local declarations rather than the
official header files in clang builds.  (Technically, it doesn't
really matter, since the net result is the same, but it felt
cleaner (er, less mysterious).)

As for extracting a better macro from compat/compiler.h for more
general use, maybe, but that's not something I want to think about
right now.  Doing a quick "git grep __GNUC__" turns up a lot of dark
magic.  There are only a couple of references to __clang__, so it
might be that a good comment somewhere (say in compat/compiler.h)
would be sufficient.

Jeff

Re: [PATCH v7 18/29] fsmonitor--daemon: implement handle_client callback

[Jeff Hostetler]

On 3/22/22 2:30 PM, Ævar Arnfjörð Bjarmason wrote:
> 
> On Tue, Mar 22 2022, Jeff Hostetler via GitGitGadget wrote:
> 
>> From: Jeff Hostetler <jeffhost@microsoft.com>
>> [...]
>> +static int do_handle_client(struct fsmonitor_daemon_state *state,
>> +			    const char *command,
>> +			    ipc_server_reply_cb *reply,
>> +			    struct ipc_server_reply_data *reply_data)
>> +{
>> +	struct fsmonitor_token_data *token_data = NULL;
>> +	struct strbuf response_token = STRBUF_INIT;
>> +	struct strbuf requested_token_id = STRBUF_INIT;
>> +	struct strbuf payload = STRBUF_INIT;
>> +	uint64_t requested_oldest_seq_nr = 0;
>> +	uint64_t total_response_len = 0;
>> +	const char *p;
>> +	const struct fsmonitor_batch *batch_head;
>> +	const struct fsmonitor_batch *batch;
>> +	intmax_t count = 0, duplicates = 0;
>> +	kh_str_t *shown;
>> +	int hash_ret;
>> +	int do_trivial = 0;
>> +	int do_flush = 0;
> 
> Just noticed while skimming, isn't initializing do_trivial here (didn't
> exhaustively check the others) getting in the way of uninitialized
> analysis the compiler will do...
> 
>> +
>> +	/*
>> +	 * We expect `command` to be of the form:
>> +	 *
>> +	 * <command> := quit NUL
>> +	 *            | flush NUL
>> +	 *            | <V1-time-since-epoch-ns> NUL
>> +	 *            | <V2-opaque-fsmonitor-token> NUL
>> +	 */
>> +
>> +	if (!strcmp(command, "quit")) {
>> +		/*
>> +		 * A client has requested over the socket/pipe that the
>> +		 * daemon shutdown.
>> +		 *
>> +		 * Tell the IPC thread pool to shutdown (which completes
>> +		 * the await in the main thread (which can stop the
>> +		 * fsmonitor listener thread)).
>> +		 *
>> +		 * There is no reply to the client.
>> +		 */
>> +		return SIMPLE_IPC_QUIT;
>> +
>> +	} else if (!strcmp(command, "flush")) {
>> +		/*
>> +		 * Flush all of our cached data and generate a new token
>> +		 * just like if we lost sync with the filesystem.
>> +		 *
>> +		 * Then send a trivial response using the new token.
>> +		 */
>> +		do_flush = 1;
>> +		do_trivial = 1;
>> +
>> +	} else if (!skip_prefix(command, "builtin:", &p)) {
>> +		/* assume V1 timestamp or garbage */
>> +
>> +		char *p_end;
>> +
>> +		strtoumax(command, &p_end, 10);
>> +		trace_printf_key(&trace_fsmonitor,
>> +				 ((*p_end) ?
>> +				  "fsmonitor: invalid command line '%s'" :
>> +				  "fsmonitor: unsupported V1 protocol '%s'"),
>> +				 command);
>> +		do_trivial = 1;
>> +
>> +	} else {
>> +		/* We have "builtin:*" */
>> +		if (fsmonitor_parse_client_token(command, &requested_token_id,
>> +						 &requested_oldest_seq_nr)) {
>> +			trace_printf_key(&trace_fsmonitor,
>> +					 "fsmonitor: invalid V2 protocol token '%s'",
>> +					 command);
>> +			do_trivial = 1;
>> +
>> +		} else {
>> +			/*
>> +			 * We have a V2 valid token:
>> +			 *     "builtin:<token_id>:<seq_nr>"
>> +			 */
>> +		}
>> +	}
> 
> Since we'll set it here in all branches except the "else" branch,
> i.e. if you move this to "we have a v2 valid" we'll catch future bugs if
> we ever have it uninitialized still,

I suppose we could do it that way.  We'd want to do the same
for "do_flush" in that case.  I could go either way on this
change.


> 
>> +
>> +	pthread_mutex_lock(&state->main_lock);
>> +
>> +	if (!state->current_token_data)
>> +		BUG("fsmonitor state does not have a current token");
>> +
>> +	if (do_flush)
>> +		with_lock__do_force_resync(state);
>> +
>> +	/*
>> +	 * We mark the current head of the batch list as "pinned" so
>> +	 * that the listener thread will treat this item as read-only
>> +	 * (and prevent any more paths from being added to it) from
>> +	 * now on.
>> +	 */
>> +	token_data = state->current_token_data;
>> +	batch_head = token_data->batch_head;
>> +	((struct fsmonitor_batch *)batch_head)->pinned_time = time(NULL);
>> +
>> +	/*
>> +	 * FSMonitor Protocol V2 requires that we send a response header
>> +	 * with a "new current token" and then all of the paths that changed
>> +	 * since the "requested token".  We send the seq_nr of the just-pinned
>> +	 * head batch so that future requests from a client will be relative
>> +	 * to it.
>> +	 */
>> +	with_lock__format_response_token(&response_token,
>> +					 &token_data->token_id, batch_head);
>> +
>> +	reply(reply_data, response_token.buf, response_token.len + 1);
>> +	total_response_len += response_token.len + 1;
>> +
>> +	trace2_data_string("fsmonitor", the_repository, "response/token",
>> +			   response_token.buf);
>> +	trace_printf_key(&trace_fsmonitor, "response token: %s",
>> +			 response_token.buf);
>> +
>> +	if (!do_trivial) {
> 
> I.e. this would start warning.
> 
>> +		if (strcmp(requested_token_id.buf, token_data->token_id.buf)) {
>> +			/*
>> +			 * The client last spoke to a different daemon
>> +			 * instance -OR- the daemon had to resync with
>> +			 * the filesystem (and lost events), so reject.
>> +			 */
>> +			trace2_data_string("fsmonitor", the_repository,
>> +					   "response/token", "different");
>> +			do_trivial = 1;
>> +
>> +		} else if (requested_oldest_seq_nr <
>> +			   token_data->batch_tail->batch_seq_nr) {
>> +			/*
>> +			 * The client wants older events than we have for
>> +			 * this token_id.  This means that the end of our
>> +			 * batch list was truncated and we cannot give the
>> +			 * client a complete snapshot relative to their
>> +			 * request.
>> +			 */
>> +			trace_printf_key(&trace_fsmonitor,
>> +					 "client requested truncated data");
>> +			do_trivial = 1;
>> +		}
>> +	}
>> +
>> +	if (do_trivial) {
>> +		pthread_mutex_unlock(&state->main_lock);
>> +
>> +		reply(reply_data, "/", 2);
>> +
>> +		trace2_data_intmax("fsmonitor", the_repository,
>> +				   "response/trivial", 1);
>> +
>> +		strbuf_release(&response_token);
>> +		strbuf_release(&requested_token_id);
>> +		return 0;
> 
> Nit: instead of strbuf_release() here, just a 'goto cleanup' and ...

Yeah, we could do that.  That would also do an unnecesssary release
of "payload", but that's not that bad (or I could order them at the
bottom).

I'm not sure either of these changes are enough for a re-roll,
but I'll include them if I do.

Thanks
Jeff

> 
>> +	}
>> +
>> +	/*
>> +	 * We're going to hold onto a pointer to the current
>> +	 * token-data while we walk the list of batches of files.
>> +	 * During this time, we will NOT be under the lock.
>> +	 * So we ref-count it.
>> +	 *
>> +	 * This allows the listener thread to continue prepending
>> +	 * new batches of items to the token-data (which we'll ignore).
>> +	 *
>> +	 * AND it allows the listener thread to do a token-reset
>> +	 * (and install a new `current_token_data`).
>> +	 */
>> +	token_data->client_ref_count++;
>> +
>> +	pthread_mutex_unlock(&state->main_lock);
>> +
>> +	/*
>> +	 * The client request is relative to the token that they sent,
>> +	 * so walk the batch list backwards from the current head back
>> +	 * to the batch (sequence number) they named.
>> +	 *
>> +	 * We use khash to de-dup the list of pathnames.
>> +	 *
>> +	 * NEEDSWORK: each batch contains a list of interned strings,
>> +	 * so we only need to do pointer comparisons here to build the
>> +	 * hash table.  Currently, we're still comparing the string
>> +	 * values.
>> +	 */
>> +	shown = kh_init_str();
>> +	for (batch = batch_head;
>> +	     batch && batch->batch_seq_nr > requested_oldest_seq_nr;
>> +	     batch = batch->next) {
>> +		size_t k;
>> +
>> +		for (k = 0; k < batch->nr; k++) {
>> +			const char *s = batch->interned_paths[k];
>> +			size_t s_len;
>> +
>> +			if (kh_get_str(shown, s) != kh_end(shown))
>> +				duplicates++;
>> +			else {
>> +				kh_put_str(shown, s, &hash_ret);
>> +
>> +				trace_printf_key(&trace_fsmonitor,
>> +						 "send[%"PRIuMAX"]: %s",
>> +						 count, s);
>> +
>> +				/* Each path gets written with a trailing NUL */
>> +				s_len = strlen(s) + 1;
>> +
>> +				if (payload.len + s_len >=
>> +				    LARGE_PACKET_DATA_MAX) {
>> +					reply(reply_data, payload.buf,
>> +					      payload.len);
>> +					total_response_len += payload.len;
>> +					strbuf_reset(&payload);
>> +				}
>> +
>> +				strbuf_add(&payload, s, s_len);
>> +				count++;
>> +			}
>> +		}
>> +	}
>> +
>> +	if (payload.len) {
>> +		reply(reply_data, payload.buf, payload.len);
>> +		total_response_len += payload.len;
>> +	}
>> +
>> +	kh_release_str(shown);
>> +
>> +	pthread_mutex_lock(&state->main_lock);
>> +
>> +	if (token_data->client_ref_count > 0)
>> +		token_data->client_ref_count--;
>> +
>> +	if (token_data->client_ref_count == 0) {
>> +		if (token_data != state->current_token_data) {
>> +			/*
>> +			 * The listener thread did a token-reset while we were
>> +			 * walking the batch list.  Therefore, this token is
>> +			 * stale and can be discarded completely.  If we are
>> +			 * the last reader thread using this token, we own
>> +			 * that work.
>> +			 */
>> +			fsmonitor_free_token_data(token_data);
>> +		}
>> +	}
>> +
>> +	pthread_mutex_unlock(&state->main_lock);
>> +
>> +	trace2_data_intmax("fsmonitor", the_repository, "response/length", total_response_len);
>> +	trace2_data_intmax("fsmonitor", the_repository, "response/count/files", count);
>> +	trace2_data_intmax("fsmonitor", the_repository, "response/count/duplicates", duplicates);
>> +
> 
> add a "cleanup" label here?
> 
>> +	strbuf_release(&response_token);
>> +	strbuf_release(&requested_token_id);
>> +	strbuf_release(&payload);
>> +
>> +	return 0;
>> +}
>> +
>>   static ipc_server_application_cb handle_client;

Re: [PATCH v7 21/29] t7527: create test for fsmonitor--daemon

[Jeff Hostetler]

On 3/22/22 2:35 PM, Ævar Arnfjörð Bjarmason wrote:
> 
> On Tue, Mar 22 2022, Jeff Hostetler via GitGitGadget wrote:
> 
>> From: Jeff Hostetler <jeffhost@microsoft.com>
>>
>> Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
>> Signed-off-by: Junio C Hamano <gitster@pobox.com>
>> ---
>>   t/t7527-builtin-fsmonitor.sh | 501 +++++++++++++++++++++++++++++++++++
>>   1 file changed, 501 insertions(+)
>>   create mode 100755 t/t7527-builtin-fsmonitor.sh
>>
>> diff --git a/t/t7527-builtin-fsmonitor.sh b/t/t7527-builtin-fsmonitor.sh
>> new file mode 100755
>> index 00000000000..d79635e7596
[...]
> 
>> +	while test "$#" -ne 0
>> +	do
>> +		case "$1" in
>> +		-C)
>> +			shift;
>> +			test "$#" -ne 0 || BUG "error: -C requires arg"
>> +			r="-C $1"
>> +			shift
>> +			;;
>> +		-tf)
>> +			shift;
>> +			test "$#" -ne 0 || BUG "error: -tf requires arg"
>> +			tf="$1"
>> +			shift
>> +			;;
>> +		-t2)
>> +			shift;
>> +			test "$#" -ne 0 || BUG "error: -t2 requires arg"
>> +			t2="$1"
>> +			shift
>> +			;;
>> +		-tk)
>> +			shift;
>> +			test "$#" -ne 0 || BUG "error: -tk requires arg"
>> +			tk="$1"
>> +			shift
>> +			;;
> 
> But (and IIRC I noted this in a previous iteration) if you &&-chain the
> "shift" here you can lose the more verbose BUG

Yeah, I looked at the test_commit() example that you referenced.
I thought it was too subtle and misleading.

I mean, the "shift" after the "case...esac" will either clobber
the "--key" or the "value" depending on whether the particular
case-arm shifted.  The shift error would only be thrown on a
missing value, since the while loop already tested $# for non-zero,
but at the point of the error, we'll just have a generic error message
and not know which key should have had a value -- without reading
the script in detail.

Also, in the k/v case-arms, it references $2 without confirming
that it exists.  In test_commit(), it just loads up local variables
(in advance of the soon-to-be-thrown shift error, so no big deal)
but if other people copy this as a model, they may do more in their
case-arms that may be more serious.


My version on the other hand, shifts away the key immediately,
tests whether the required value is present and errors with a
detailed message, and then references the value and shifts away
the value.

My way (IMHO) feels more straight-forward and easier for casual
readers to follow.  Yes, it is a bit more wordy, but I think it
is worth it.


FWIW, as I was writing this note I noticed that both test_commit()
and my start_daemon() examples have a bug where they won't detect
a missing value.  For example, if someone changes
    "test_commit -C repo"
to "test_commit -C --no-tag repo"
then $indir will be "--no-tag" and "repo" will be unclaimed and
an error will follow at some point later (when $indir is used
in a Git command).


I think I'll fix my function to handle that error case, but keep
the basic design that I have.


> 
>> +	start_daemon -tf "$PWD/.git/trace" &&
> 
> FWIW having an option parser take -tf to mean --tf is quite unlike our
> common conventions, usually it means both -t and -f.
> 
> In this case every single caller added here does provide -tf
> argument. Perhaps better as as unconditional $1 then?
> 

yes, very old-school of me to use single dashes here.  I'll change
it/them to use double-dashes (or relabel the keys to be single chars).
There are callers that do not pass the -tf key, so I'd rather keep it
as a key/value than assume a fixed $1.

Thanks
Jeff

Re: [PATCH v7 24/29] t/perf/p7519: speed up test on Windows

[Jeff Hostetler]

On 3/22/22 2:43 PM, Ævar Arnfjörð Bjarmason wrote:
> 
> On Tue, Mar 22 2022, Jeff Hostetler via GitGitGadget wrote:
> 
>> From: Jeff Hostetler <jeffhost@microsoft.com>
>>
>> Change p7519 to use `test_seq` and `xargs` rather than a `for` loop
>> to touch thousands of files.  This takes minutes off of test runs
>> on Windows because of process creation overhead.
>>
>> Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
>> Signed-off-by: Junio C Hamano <gitster@pobox.com>
>> ---
>>   t/perf/p7519-fsmonitor.sh | 32 ++++++++++++++++++++------------
>>   1 file changed, 20 insertions(+), 12 deletions(-)
>>
>> diff --git a/t/perf/p7519-fsmonitor.sh b/t/perf/p7519-fsmonitor.sh
>> index c8be58f3c76..0611e533951 100755
>> --- a/t/perf/p7519-fsmonitor.sh
>> +++ b/t/perf/p7519-fsmonitor.sh
>> @@ -72,7 +72,7 @@ then
[...]
>>   
>> -trace_stop() {
>> +trace_stop () {
>>   	if test -n "$GIT_PERF_7519_TRACE"
>>   	then
>>   		unset GIT_TRACE2_PERF
>>   	fi
>>   }
> 
> (I think I noted in a previous version): Would be nice to have an
> optimization change not do unrelated refactoring....

In V4 24/29, Junio pushed a fixup! commit onto my branch to
add the space between the function name and the parens of the
function that I added.  And fixed the other existing instances
while he was at it.  I squashed that down into my branch during
cleanups between V4 and V5.  It seemed less trouble/noise than
splitting it into two commits.  (I was also at the 30 commit
limit of GGG in V5 and V6, so that was another reason not to
split it.)

Now that I'm at 29 in V7, I suppose that I could split it.



> 
>> +touch_files () {
>> +	n=$1
>> +	d="$n"_files
>> +
>> +	(cd $d ; test_seq 1 $n | xargs touch )
> 
> 
> ...and here we don't &&-chain.
> 

Right, thanks.

Jeff

[PATCH v8 00/30] Builtin FSMonitor Part 2

[Jeff Hostetler via GitGitGadget]

Here is V8 of Part 2 of my builtin FSMonitor series. This contains last
minute comments on V7. I'll also send an updated version of Part 3 that
builds upon this version.

Here is a range-diff from V7 to V8 relative to 715d08a9e5 (The eighth batch,
2022-02-25). Changes since V7 can be summarized as:

 * [] Improved arg checking in start_daemon() in t7527.
 * [] Switch start_daemon() args to use double-dash keywords.
 * [] Split out coding style changes in p7519 into separate commit from
   speed up changes.
 * [] Disabled some of the matrix tests in t7527 concerning the untracked
   cache.
 * [] Other minor style cleanups.

 1:  e98373f997 =  1:  e98373f997 fsmonitor: enhance existing comments, clarify trivial response handling
 2:  ab68b94417 =  2:  ab68b94417 fsmonitor-ipc: create client routines for git-fsmonitor--daemon
 3:  e04c7301f2 =  3:  e04c7301f2 fsmonitor: config settings are repository-specific
 4:  ea02ba25d8 =  4:  ea02ba25d8 fsmonitor: use IPC to query the builtin FSMonitor daemon
 5:  6ab7db9cb7 =  5:  6ab7db9cb7 fsmonitor: document builtin fsmonitor
 6:  0ce8ae3f2c =  6:  0ce8ae3f2c fsmonitor--daemon: add a built-in fsmonitor daemon
 7:  4624ce2fa4 =  7:  4624ce2fa4 fsmonitor--daemon: implement 'stop' and 'status' commands
 8:  a29fe7266a =  8:  a29fe7266a compat/fsmonitor/fsm-listen-win32: stub in backend for Windows
 9:  2f8a42fdb9 =  9:  2f8a42fdb9 compat/fsmonitor/fsm-listen-darwin: stub in backend for Darwin
10:  f07800690e = 10:  f07800690e fsmonitor--daemon: implement 'run' command
11:  a6a39a3306 = 11:  a6a39a3306 fsmonitor--daemon: implement 'start' command
12:  d62e338d00 = 12:  d62e338d00 fsmonitor--daemon: add pathname classification
13:  53e06b4ae5 = 13:  53e06b4ae5 fsmonitor--daemon: define token-ids
14:  39f43fabe0 = 14:  39f43fabe0 fsmonitor--daemon: create token-based changed path cache
15:  239558e34f = 15:  239558e34f compat/fsmonitor/fsm-listen-win32: implement FSMonitor backend on Windows
16:  14b775e9d8 = 16:  14b775e9d8 compat/fsmonitor/fsm-listen-darwin: add MacOS header files for FSEvent
17:  55bd7aee06 = 17:  55bd7aee06 compat/fsmonitor/fsm-listen-darwin: implement FSEvent listener on MacOS
18:  1f4b5209bf ! 18:  c43009124f fsmonitor--daemon: implement handle_client callback
    @@ builtin/fsmonitor--daemon.c: void fsmonitor_force_resync(struct fsmonitor_daemon
     +        trace2_data_intmax("fsmonitor", the_repository,
     +                   "response/trivial", 1);
     +
    -+        strbuf_release(&response_token);
    -+        strbuf_release(&requested_token_id);
    -+        return 0;
    ++        goto cleanup;
     +    }
     +
     +    /*
    @@ builtin/fsmonitor--daemon.c: void fsmonitor_force_resync(struct fsmonitor_daemon
     +    trace2_data_intmax("fsmonitor", the_repository, "response/count/files", count);
     +    trace2_data_intmax("fsmonitor", the_repository, "response/count/duplicates", duplicates);
     +
    ++cleanup:
     +    strbuf_release(&response_token);
     +    strbuf_release(&requested_token_id);
     +    strbuf_release(&payload);
19:  8cf62c9fc6 = 19:  ed338777b5 help: include fsmonitor--daemon feature flag in version info
20:  1bd74a8159 = 20:  c99bac29d4 t/helper/fsmonitor-client: create IPC client to talk to FSMonitor Daemon
21:  4a920d0b54 ! 21:  c8709da945 t7527: create test for fsmonitor--daemon
    @@ t/t7527-builtin-fsmonitor.sh (new)
     +    rm -rf $1
     +}
     +
    ++is_value () {
    ++    test -n "$1" && test "${1::1}" != "-"
    ++}
    ++
     +start_daemon () {
     +    r= &&
     +    tf= &&
    @@ t/t7527-builtin-fsmonitor.sh (new)
     +        case "$1" in
     +        -C)
     +            shift;
    -+            test "$#" -ne 0 || BUG "error: -C requires arg"
    ++            is_value $1 || BUG "error: -C requires value"
     +            r="-C $1"
     +            shift
     +            ;;
    -+        -tf)
    ++        --tf)
     +            shift;
    -+            test "$#" -ne 0 || BUG "error: -tf requires arg"
    ++            is_value $1 || BUG "error: --tf requires value"
     +            tf="$1"
     +            shift
     +            ;;
    -+        -t2)
    ++        --t2)
     +            shift;
    -+            test "$#" -ne 0 || BUG "error: -t2 requires arg"
    ++            is_value $1 || BUG "error: --t2 requires value"
     +            t2="$1"
     +            shift
     +            ;;
    -+        -tk)
    ++        --tk)
     +            shift;
    -+            test "$#" -ne 0 || BUG "error: -tk requires arg"
    ++            is_value $1 || BUG "error: --tk requires value"
     +            tk="$1"
     +            shift
     +            ;;
    @@ t/t7527-builtin-fsmonitor.sh (new)
     +test_expect_success 'edit some files' '
     +    test_when_finished clean_up_repo_and_stop_daemon &&
     +
    -+    start_daemon -tf "$PWD/.git/trace" &&
    ++    start_daemon --tf "$PWD/.git/trace" &&
     +
     +    edit_files &&
     +
    @@ t/t7527-builtin-fsmonitor.sh (new)
     +test_expect_success 'create some files' '
     +    test_when_finished clean_up_repo_and_stop_daemon &&
     +
    -+    start_daemon -tf "$PWD/.git/trace" &&
    ++    start_daemon --tf "$PWD/.git/trace" &&
     +
     +    create_files &&
     +
    @@ t/t7527-builtin-fsmonitor.sh (new)
     +test_expect_success 'delete some files' '
     +    test_when_finished clean_up_repo_and_stop_daemon &&
     +
    -+    start_daemon -tf "$PWD/.git/trace" &&
    ++    start_daemon --tf "$PWD/.git/trace" &&
     +
     +    delete_files &&
     +
    @@ t/t7527-builtin-fsmonitor.sh (new)
     +test_expect_success 'rename some files' '
     +    test_when_finished clean_up_repo_and_stop_daemon &&
     +
    -+    start_daemon -tf "$PWD/.git/trace" &&
    ++    start_daemon --tf "$PWD/.git/trace" &&
     +
     +    rename_files &&
     +
    @@ t/t7527-builtin-fsmonitor.sh (new)
     +test_expect_success 'rename directory' '
     +    test_when_finished clean_up_repo_and_stop_daemon &&
     +
    -+    start_daemon -tf "$PWD/.git/trace" &&
    ++    start_daemon --tf "$PWD/.git/trace" &&
     +
     +    mv dirtorename dirrenamed &&
     +
    @@ t/t7527-builtin-fsmonitor.sh (new)
     +test_expect_success 'file changes to directory' '
     +    test_when_finished clean_up_repo_and_stop_daemon &&
     +
    -+    start_daemon -tf "$PWD/.git/trace" &&
    ++    start_daemon --tf "$PWD/.git/trace" &&
     +
     +    file_to_directory &&
     +
    @@ t/t7527-builtin-fsmonitor.sh (new)
     +test_expect_success 'directory changes to a file' '
     +    test_when_finished clean_up_repo_and_stop_daemon &&
     +
    -+    start_daemon -tf "$PWD/.git/trace" &&
    ++    start_daemon --tf "$PWD/.git/trace" &&
     +
     +    directory_to_file &&
     +
    @@ t/t7527-builtin-fsmonitor.sh (new)
     +
     +    git init test_flush &&
     +
    -+    start_daemon -C test_flush -tf "$PWD/.git/trace_daemon" -tk true &&
    ++    start_daemon -C test_flush --tf "$PWD/.git/trace_daemon" --tk true &&
     +
     +    # The daemon should have an initial token with no events in _0 and
     +    # then a few (probably platform-specific number of) events in _1.
    @@ t/t7527-builtin-fsmonitor.sh (new)
     +    git -C wt-base worktree add ../wt-secondary &&
     +
     +    start_daemon -C wt-secondary \
    -+        -tf "$PWD/trace_wt_secondary" \
    -+        -t2 "$PWD/trace2_wt_secondary" &&
    ++        --tf "$PWD/trace_wt_secondary" \
    ++        --t2 "$PWD/trace2_wt_secondary" &&
     +
     +    git -C wt-secondary fsmonitor--daemon stop &&
     +    test_must_fail git -C wt-secondary fsmonitor--daemon status
22:  c925a9a745 = 22:  cc39ecf10a t/perf: avoid copying builtin fsmonitor files into test repo
23:  5b3381c223 = 23:  2bb3eb8476 t/helper/test-chmtime: skip directories on Windows
 -:  ---------- > 24:  bab9a9b080 t/perf/p7519: fix coding style
24:  803a540cc0 ! 25:  2dd06ad2f7 t/perf/p7519: speed up test on Windows
    @@ Commit message
         Signed-off-by: Junio C Hamano <gitster@pobox.com>
     
      ## t/perf/p7519-fsmonitor.sh ##
    -@@ t/perf/p7519-fsmonitor.sh: then
    -     fi
    - fi
    - 
    --trace_start() {
    -+trace_start () {
    -     if test -n "$GIT_PERF_7519_TRACE"
    -     then
    -         name="$1"
    -@@ t/perf/p7519-fsmonitor.sh: trace_start() {
    -     fi
    - }
    - 
    --trace_stop() {
    -+trace_stop () {
    -     if test -n "$GIT_PERF_7519_TRACE"
    -     then
    -         unset GIT_TRACE2_PERF
    +@@ t/perf/p7519-fsmonitor.sh: trace_stop () {
          fi
      }
      
     +touch_files () {
    -+    n=$1
    -+    d="$n"_files
    ++    n=$1 &&
    ++    d="$n"_files &&
     +
    -+    (cd $d ; test_seq 1 $n | xargs touch )
    ++    (cd $d && test_seq 1 $n | xargs touch )
     +}
     +
      test_expect_success "one time repo setup" '
    @@ t/perf/p7519-fsmonitor.sh: test_expect_success "one time repo setup" '
          git add 1_file 10_files 100_files 1000_files 10000_files &&
          git commit -qm "Add files" &&
      
    -@@ t/perf/p7519-fsmonitor.sh: test_expect_success "one time repo setup" '
    -     fi
    - '
    - 
    --setup_for_fsmonitor() {
    -+setup_for_fsmonitor () {
    -     # set INTEGRATION_SCRIPT depending on the environment
    -     if test -n "$INTEGRATION_PATH"
    -     then
    -@@ t/perf/p7519-fsmonitor.sh: test_perf_w_drop_caches () {
    -     test_perf "$@"
    - }
    - 
    --test_fsmonitor_suite() {
    -+test_fsmonitor_suite () {
    -     if test -n "$INTEGRATION_SCRIPT"; then
    -         DESC="fsmonitor=$(basename $INTEGRATION_SCRIPT)"
    -     else
    -@@ t/perf/p7519-fsmonitor.sh: test_fsmonitor_suite() {
    +@@ t/perf/p7519-fsmonitor.sh: test_fsmonitor_suite () {
      
          # Update the mtimes on upto 100k files to make status think
          # that they are dirty.  For simplicity, omit any files with
25:  d5ca2df31c = 26:  6eaa5765ae t/perf/p7519: add fsmonitor--daemon test cases
26:  42631259e8 = 27:  30957f3930 fsmonitor--daemon: periodically truncate list of modified files
27:  f256c3cbe8 = 28:  c8ca2a1727 fsmonitor--daemon: use a cookie file to sync with file system
28:  08af8296f9 = 29:  4caf1d89b8 fsmonitor: force update index after large responses
29:  e6cf84dc8e ! 30:  f87a1eba69 t7527: test status with untracked-cache and fsmonitor--daemon
    @@ t/t7527-builtin-fsmonitor.sh: test_expect_success 'cleanup worktrees' '
     +    fsm=$2 &&
     +    fn=$3 &&
     +
    ++    if test $uc = true && test $fsm = false
    ++    then
    ++        # The untracked-cache is buggy when FSMonitor is
    ++        # DISABLED, so skip the tests for this matrix
    ++        # combination.
    ++        #
    ++        # We've observed random, occasional test failures on
    ++        # Windows and MacOS when the UC is turned on and FSM
    ++        # is turned off.  These are rare, but they do happen
    ++        # indicating that it is probably a race condition within
    ++        # the untracked cache itself.
    ++        #
    ++        # It usually happens when a test does F/D trickery and
    ++        # then the NEXT test fails because of extra status
    ++        # output from stale UC data from the previous test.
    ++        #
    ++        # Since FSMonitor is not involved in the error, skip
    ++        # the tests for this matrix combination.
    ++        #
    ++        return 0
    ++    fi &&
    ++
     +    test_expect_success "Matrix[uc:$uc][fsm:$fsm] $fn" '
     +        matrix_clean_up_repo &&
     +        $fn &&


Jeff Hostetler (30):
  fsmonitor: enhance existing comments, clarify trivial response
    handling
  fsmonitor-ipc: create client routines for git-fsmonitor--daemon
  fsmonitor: config settings are repository-specific
  fsmonitor: use IPC to query the builtin FSMonitor daemon
  fsmonitor: document builtin fsmonitor
  fsmonitor--daemon: add a built-in fsmonitor daemon
  fsmonitor--daemon: implement 'stop' and 'status' commands
  compat/fsmonitor/fsm-listen-win32: stub in backend for Windows
  compat/fsmonitor/fsm-listen-darwin: stub in backend for Darwin
  fsmonitor--daemon: implement 'run' command
  fsmonitor--daemon: implement 'start' command
  fsmonitor--daemon: add pathname classification
  fsmonitor--daemon: define token-ids
  fsmonitor--daemon: create token-based changed path cache
  compat/fsmonitor/fsm-listen-win32: implement FSMonitor backend on
    Windows
  compat/fsmonitor/fsm-listen-darwin: add MacOS header files for FSEvent
  compat/fsmonitor/fsm-listen-darwin: implement FSEvent listener on
    MacOS
  fsmonitor--daemon: implement handle_client callback
  help: include fsmonitor--daemon feature flag in version info
  t/helper/fsmonitor-client: create IPC client to talk to FSMonitor
    Daemon
  t7527: create test for fsmonitor--daemon
  t/perf: avoid copying builtin fsmonitor files into test repo
  t/helper/test-chmtime: skip directories on Windows
  t/perf/p7519: fix coding style
  t/perf/p7519: speed up test on Windows
  t/perf/p7519: add fsmonitor--daemon test cases
  fsmonitor--daemon: periodically truncate list of modified files
  fsmonitor--daemon: use a cookie file to sync with file system
  fsmonitor: force update index after large responses
  t7527: test status with untracked-cache and fsmonitor--daemon

 .gitignore                              |    1 +
 Documentation/config/core.txt           |   60 +-
 Documentation/git-fsmonitor--daemon.txt |   75 ++
 Documentation/git-update-index.txt      |    8 +-
 Makefile                                |   17 +
 builtin.h                               |    1 +
 builtin/fsmonitor--daemon.c             | 1479 +++++++++++++++++++++++
 builtin/update-index.c                  |    7 +-
 cache.h                                 |    1 -
 compat/fsmonitor/fsm-darwin-gcc.h       |   92 ++
 compat/fsmonitor/fsm-listen-darwin.c    |  427 +++++++
 compat/fsmonitor/fsm-listen-win32.c     |  586 +++++++++
 compat/fsmonitor/fsm-listen.h           |   49 +
 config.c                                |   14 -
 config.h                                |    1 -
 config.mak.uname                        |   20 +
 contrib/buildsystems/CMakeLists.txt     |   10 +
 environment.c                           |    1 -
 fsmonitor--daemon.h                     |  166 +++
 fsmonitor-ipc.c                         |  171 +++
 fsmonitor-ipc.h                         |   48 +
 fsmonitor-settings.c                    |  114 ++
 fsmonitor-settings.h                    |   21 +
 fsmonitor.c                             |  216 +++-
 fsmonitor.h                             |   15 +-
 git.c                                   |    1 +
 help.c                                  |    4 +
 repo-settings.c                         |    1 +
 repository.h                            |    3 +
 t/README                                |    4 +-
 t/helper/test-chmtime.c                 |   15 +
 t/helper/test-fsmonitor-client.c        |  116 ++
 t/helper/test-tool.c                    |    1 +
 t/helper/test-tool.h                    |    1 +
 t/perf/p7519-fsmonitor.sh               |   68 +-
 t/perf/perf-lib.sh                      |    2 +-
 t/t7527-builtin-fsmonitor.sh            |  620 ++++++++++
 t/test-lib.sh                           |    7 +
 38 files changed, 4337 insertions(+), 106 deletions(-)
 create mode 100644 Documentation/git-fsmonitor--daemon.txt
 create mode 100644 builtin/fsmonitor--daemon.c
 create mode 100644 compat/fsmonitor/fsm-darwin-gcc.h
 create mode 100644 compat/fsmonitor/fsm-listen-darwin.c
 create mode 100644 compat/fsmonitor/fsm-listen-win32.c
 create mode 100644 compat/fsmonitor/fsm-listen.h
 create mode 100644 fsmonitor--daemon.h
 create mode 100644 fsmonitor-ipc.c
 create mode 100644 fsmonitor-ipc.h
 create mode 100644 fsmonitor-settings.c
 create mode 100644 fsmonitor-settings.h
 create mode 100644 t/helper/test-fsmonitor-client.c
 create mode 100755 t/t7527-builtin-fsmonitor.sh


base-commit: 715d08a9e51251ad8290b181b6ac3b9e1f9719d7
Published-As: https://github.com/gitgitgadget/git/releases/tag/pr-1041%2Fjeffhostetler%2Fbuiltin-fsmonitor-part2-v8
Fetch-It-Via: git fetch https://github.com/gitgitgadget/git pr-1041/jeffhostetler/builtin-fsmonitor-part2-v8
Pull-Request: https://github.com/gitgitgadget/git/pull/1041

Range-diff vs v7:

  1:  e98373f997f =  1:  e98373f997f fsmonitor: enhance existing comments, clarify trivial response handling
  2:  ab68b944173 =  2:  ab68b944173 fsmonitor-ipc: create client routines for git-fsmonitor--daemon
  3:  e04c7301f24 =  3:  e04c7301f24 fsmonitor: config settings are repository-specific
  4:  ea02ba25d8f =  4:  ea02ba25d8f fsmonitor: use IPC to query the builtin FSMonitor daemon
  5:  6ab7db9cb76 =  5:  6ab7db9cb76 fsmonitor: document builtin fsmonitor
  6:  0ce8ae3f2cf =  6:  0ce8ae3f2cf fsmonitor--daemon: add a built-in fsmonitor daemon
  7:  4624ce2fa47 =  7:  4624ce2fa47 fsmonitor--daemon: implement 'stop' and 'status' commands
  8:  a29fe7266a4 =  8:  a29fe7266a4 compat/fsmonitor/fsm-listen-win32: stub in backend for Windows
  9:  2f8a42fdb93 =  9:  2f8a42fdb93 compat/fsmonitor/fsm-listen-darwin: stub in backend for Darwin
 10:  f07800690ee = 10:  f07800690ee fsmonitor--daemon: implement 'run' command
 11:  a6a39a3306d = 11:  a6a39a3306d fsmonitor--daemon: implement 'start' command
 12:  d62e338d008 = 12:  d62e338d008 fsmonitor--daemon: add pathname classification
 13:  53e06b4ae5d = 13:  53e06b4ae5d fsmonitor--daemon: define token-ids
 14:  39f43fabe02 = 14:  39f43fabe02 fsmonitor--daemon: create token-based changed path cache
 15:  239558e34ff = 15:  239558e34ff compat/fsmonitor/fsm-listen-win32: implement FSMonitor backend on Windows
 16:  14b775e9d8b = 16:  14b775e9d8b compat/fsmonitor/fsm-listen-darwin: add MacOS header files for FSEvent
 17:  55bd7aee06c = 17:  55bd7aee06c compat/fsmonitor/fsm-listen-darwin: implement FSEvent listener on MacOS
 18:  1f4b5209bf6 ! 18:  c43009124fb fsmonitor--daemon: implement handle_client callback
     @@ builtin/fsmonitor--daemon.c: void fsmonitor_force_resync(struct fsmonitor_daemon
      +		trace2_data_intmax("fsmonitor", the_repository,
      +				   "response/trivial", 1);
      +
     -+		strbuf_release(&response_token);
     -+		strbuf_release(&requested_token_id);
     -+		return 0;
     ++		goto cleanup;
      +	}
      +
      +	/*
     @@ builtin/fsmonitor--daemon.c: void fsmonitor_force_resync(struct fsmonitor_daemon
      +	trace2_data_intmax("fsmonitor", the_repository, "response/count/files", count);
      +	trace2_data_intmax("fsmonitor", the_repository, "response/count/duplicates", duplicates);
      +
     ++cleanup:
      +	strbuf_release(&response_token);
      +	strbuf_release(&requested_token_id);
      +	strbuf_release(&payload);
 19:  8cf62c9fc6f = 19:  ed338777b56 help: include fsmonitor--daemon feature flag in version info
 20:  1bd74a81593 = 20:  c99bac29d42 t/helper/fsmonitor-client: create IPC client to talk to FSMonitor Daemon
 21:  4a920d0b54a ! 21:  c8709da9457 t7527: create test for fsmonitor--daemon
     @@ t/t7527-builtin-fsmonitor.sh (new)
      +	rm -rf $1
      +}
      +
     ++is_value () {
     ++	test -n "$1" && test "${1::1}" != "-"
     ++}
     ++
      +start_daemon () {
      +	r= &&
      +	tf= &&
     @@ t/t7527-builtin-fsmonitor.sh (new)
      +		case "$1" in
      +		-C)
      +			shift;
     -+			test "$#" -ne 0 || BUG "error: -C requires arg"
     ++			is_value $1 || BUG "error: -C requires value"
      +			r="-C $1"
      +			shift
      +			;;
     -+		-tf)
     ++		--tf)
      +			shift;
     -+			test "$#" -ne 0 || BUG "error: -tf requires arg"
     ++			is_value $1 || BUG "error: --tf requires value"
      +			tf="$1"
      +			shift
      +			;;
     -+		-t2)
     ++		--t2)
      +			shift;
     -+			test "$#" -ne 0 || BUG "error: -t2 requires arg"
     ++			is_value $1 || BUG "error: --t2 requires value"
      +			t2="$1"
      +			shift
      +			;;
     -+		-tk)
     ++		--tk)
      +			shift;
     -+			test "$#" -ne 0 || BUG "error: -tk requires arg"
     ++			is_value $1 || BUG "error: --tk requires value"
      +			tk="$1"
      +			shift
      +			;;
     @@ t/t7527-builtin-fsmonitor.sh (new)
      +test_expect_success 'edit some files' '
      +	test_when_finished clean_up_repo_and_stop_daemon &&
      +
     -+	start_daemon -tf "$PWD/.git/trace" &&
     ++	start_daemon --tf "$PWD/.git/trace" &&
      +
      +	edit_files &&
      +
     @@ t/t7527-builtin-fsmonitor.sh (new)
      +test_expect_success 'create some files' '
      +	test_when_finished clean_up_repo_and_stop_daemon &&
      +
     -+	start_daemon -tf "$PWD/.git/trace" &&
     ++	start_daemon --tf "$PWD/.git/trace" &&
      +
      +	create_files &&
      +
     @@ t/t7527-builtin-fsmonitor.sh (new)
      +test_expect_success 'delete some files' '
      +	test_when_finished clean_up_repo_and_stop_daemon &&
      +
     -+	start_daemon -tf "$PWD/.git/trace" &&
     ++	start_daemon --tf "$PWD/.git/trace" &&
      +
      +	delete_files &&
      +
     @@ t/t7527-builtin-fsmonitor.sh (new)
      +test_expect_success 'rename some files' '
      +	test_when_finished clean_up_repo_and_stop_daemon &&
      +
     -+	start_daemon -tf "$PWD/.git/trace" &&
     ++	start_daemon --tf "$PWD/.git/trace" &&
      +
      +	rename_files &&
      +
     @@ t/t7527-builtin-fsmonitor.sh (new)
      +test_expect_success 'rename directory' '
      +	test_when_finished clean_up_repo_and_stop_daemon &&
      +
     -+	start_daemon -tf "$PWD/.git/trace" &&
     ++	start_daemon --tf "$PWD/.git/trace" &&
      +
      +	mv dirtorename dirrenamed &&
      +
     @@ t/t7527-builtin-fsmonitor.sh (new)
      +test_expect_success 'file changes to directory' '
      +	test_when_finished clean_up_repo_and_stop_daemon &&
      +
     -+	start_daemon -tf "$PWD/.git/trace" &&
     ++	start_daemon --tf "$PWD/.git/trace" &&
      +
      +	file_to_directory &&
      +
     @@ t/t7527-builtin-fsmonitor.sh (new)
      +test_expect_success 'directory changes to a file' '
      +	test_when_finished clean_up_repo_and_stop_daemon &&
      +
     -+	start_daemon -tf "$PWD/.git/trace" &&
     ++	start_daemon --tf "$PWD/.git/trace" &&
      +
      +	directory_to_file &&
      +
     @@ t/t7527-builtin-fsmonitor.sh (new)
      +
      +	git init test_flush &&
      +
     -+	start_daemon -C test_flush -tf "$PWD/.git/trace_daemon" -tk true &&
     ++	start_daemon -C test_flush --tf "$PWD/.git/trace_daemon" --tk true &&
      +
      +	# The daemon should have an initial token with no events in _0 and
      +	# then a few (probably platform-specific number of) events in _1.
     @@ t/t7527-builtin-fsmonitor.sh (new)
      +	git -C wt-base worktree add ../wt-secondary &&
      +
      +	start_daemon -C wt-secondary \
     -+		-tf "$PWD/trace_wt_secondary" \
     -+		-t2 "$PWD/trace2_wt_secondary" &&
     ++		--tf "$PWD/trace_wt_secondary" \
     ++		--t2 "$PWD/trace2_wt_secondary" &&
      +
      +	git -C wt-secondary fsmonitor--daemon stop &&
      +	test_must_fail git -C wt-secondary fsmonitor--daemon status
 22:  c925a9a7459 = 22:  cc39ecf10ae t/perf: avoid copying builtin fsmonitor files into test repo
 23:  5b3381c223e = 23:  2bb3eb84767 t/helper/test-chmtime: skip directories on Windows
  -:  ----------- > 24:  bab9a9b0802 t/perf/p7519: fix coding style
 24:  803a540cc00 ! 25:  2dd06ad2f71 t/perf/p7519: speed up test on Windows
     @@ Commit message
          Signed-off-by: Junio C Hamano <gitster@pobox.com>
      
       ## t/perf/p7519-fsmonitor.sh ##
     -@@ t/perf/p7519-fsmonitor.sh: then
     - 	fi
     - fi
     - 
     --trace_start() {
     -+trace_start () {
     - 	if test -n "$GIT_PERF_7519_TRACE"
     - 	then
     - 		name="$1"
     -@@ t/perf/p7519-fsmonitor.sh: trace_start() {
     - 	fi
     - }
     - 
     --trace_stop() {
     -+trace_stop () {
     - 	if test -n "$GIT_PERF_7519_TRACE"
     - 	then
     - 		unset GIT_TRACE2_PERF
     +@@ t/perf/p7519-fsmonitor.sh: trace_stop () {
       	fi
       }
       
      +touch_files () {
     -+	n=$1
     -+	d="$n"_files
     ++	n=$1 &&
     ++	d="$n"_files &&
      +
     -+	(cd $d ; test_seq 1 $n | xargs touch )
     ++	(cd $d && test_seq 1 $n | xargs touch )
      +}
      +
       test_expect_success "one time repo setup" '
     @@ t/perf/p7519-fsmonitor.sh: test_expect_success "one time repo setup" '
       	git add 1_file 10_files 100_files 1000_files 10000_files &&
       	git commit -qm "Add files" &&
       
     -@@ t/perf/p7519-fsmonitor.sh: test_expect_success "one time repo setup" '
     - 	fi
     - '
     - 
     --setup_for_fsmonitor() {
     -+setup_for_fsmonitor () {
     - 	# set INTEGRATION_SCRIPT depending on the environment
     - 	if test -n "$INTEGRATION_PATH"
     - 	then
     -@@ t/perf/p7519-fsmonitor.sh: test_perf_w_drop_caches () {
     - 	test_perf "$@"
     - }
     - 
     --test_fsmonitor_suite() {
     -+test_fsmonitor_suite () {
     - 	if test -n "$INTEGRATION_SCRIPT"; then
     - 		DESC="fsmonitor=$(basename $INTEGRATION_SCRIPT)"
     - 	else
     -@@ t/perf/p7519-fsmonitor.sh: test_fsmonitor_suite() {
     +@@ t/perf/p7519-fsmonitor.sh: test_fsmonitor_suite () {
       
       	# Update the mtimes on upto 100k files to make status think
       	# that they are dirty.  For simplicity, omit any files with
 25:  d5ca2df31c8 = 26:  6eaa5765ae1 t/perf/p7519: add fsmonitor--daemon test cases
 26:  42631259e89 = 27:  30957f3930e fsmonitor--daemon: periodically truncate list of modified files
 27:  f256c3cbe8b = 28:  c8ca2a17277 fsmonitor--daemon: use a cookie file to sync with file system
 28:  08af8296f96 = 29:  4caf1d89b84 fsmonitor: force update index after large responses
 29:  e6cf84dc8eb ! 30:  f87a1eba693 t7527: test status with untracked-cache and fsmonitor--daemon
     @@ t/t7527-builtin-fsmonitor.sh: test_expect_success 'cleanup worktrees' '
      +	fsm=$2 &&
      +	fn=$3 &&
      +
     ++	if test $uc = true && test $fsm = false
     ++	then
     ++		# The untracked-cache is buggy when FSMonitor is
     ++		# DISABLED, so skip the tests for this matrix
     ++		# combination.
     ++		#
     ++		# We've observed random, occasional test failures on
     ++		# Windows and MacOS when the UC is turned on and FSM
     ++		# is turned off.  These are rare, but they do happen
     ++		# indicating that it is probably a race condition within
     ++		# the untracked cache itself.
     ++		#
     ++		# It usually happens when a test does F/D trickery and
     ++		# then the NEXT test fails because of extra status
     ++		# output from stale UC data from the previous test.
     ++		#
     ++		# Since FSMonitor is not involved in the error, skip
     ++		# the tests for this matrix combination.
     ++		#
     ++		return 0
     ++	fi &&
     ++
      +	test_expect_success "Matrix[uc:$uc][fsm:$fsm] $fn" '
      +		matrix_clean_up_repo &&
      +		$fn &&

-- 
gitgitgadget

[PATCH v8 01/30] fsmonitor: enhance existing comments, clarify trivial response handling

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 fsmonitor.c | 64 ++++++++++++++++++++++++++++++++++-------------------
 1 file changed, 41 insertions(+), 23 deletions(-)

diff --git a/fsmonitor.c b/fsmonitor.c
index ab9bfc60b34..448d0ee33f5 100644
--- a/fsmonitor.c
+++ b/fsmonitor.c
@@ -168,29 +168,15 @@ static int query_fsmonitor(int version, const char *last_update, struct strbuf *
 
 	if (result)
 		trace2_data_intmax("fsm_hook", NULL, "query/failed", result);
-	else {
+	else
 		trace2_data_intmax("fsm_hook", NULL, "query/response-length",
 				   query_result->len);
 
-		if (fsmonitor_is_trivial_response(query_result))
-			trace2_data_intmax("fsm_hook", NULL,
-					   "query/trivial-response", 1);
-	}
-
 	trace2_region_leave("fsm_hook", "query", NULL);
 
 	return result;
 }
 
-int fsmonitor_is_trivial_response(const struct strbuf *query_result)
-{
-	static char trivial_response[3] = { '\0', '/', '\0' };
-
-	return query_result->len >= 3 &&
-		!memcmp(trivial_response,
-			&query_result->buf[query_result->len - 3], 3);
-}
-
 static void fsmonitor_refresh_callback(struct index_state *istate, char *name)
 {
 	int i, len = strlen(name);
@@ -238,6 +224,7 @@ void refresh_fsmonitor(struct index_state *istate)
 	struct strbuf last_update_token = STRBUF_INIT;
 	char *buf;
 	unsigned int i;
+	int is_trivial = 0;
 
 	if (!core_fsmonitor || istate->fsmonitor_has_run_once)
 		return;
@@ -283,6 +270,7 @@ void refresh_fsmonitor(struct index_state *istate)
 					query_success = 0;
 				} else {
 					bol = last_update_token.len + 1;
+					is_trivial = query_result.buf[bol] == '/';
 				}
 			} else if (hook_version < 0) {
 				hook_version = HOOK_INTERFACE_VERSION1;
@@ -294,16 +282,38 @@ void refresh_fsmonitor(struct index_state *istate)
 		if (hook_version == HOOK_INTERFACE_VERSION1) {
 			query_success = !query_fsmonitor(HOOK_INTERFACE_VERSION1,
 				istate->fsmonitor_last_update, &query_result);
+			if (query_success)
+				is_trivial = query_result.buf[0] == '/';
 		}
 
+		if (is_trivial)
+			trace2_data_intmax("fsm_hook", NULL,
+					   "query/trivial-response", 1);
+
 		trace_performance_since(last_update, "fsmonitor process '%s'", core_fsmonitor);
 		trace_printf_key(&trace_fsmonitor, "fsmonitor process '%s' returned %s",
 			core_fsmonitor, query_success ? "success" : "failure");
 	}
 
-	/* a fsmonitor process can return '/' to indicate all entries are invalid */
-	if (query_success && query_result.buf[bol] != '/') {
-		/* Mark all entries returned by the monitor as dirty */
+	/*
+	 * The response from FSMonitor (excluding the header token) is
+	 * either:
+	 *
+	 * [a] a (possibly empty) list of NUL delimited relative
+	 *     pathnames of changed paths.  This list can contain
+	 *     files and directories.  Directories have a trailing
+	 *     slash.
+	 *
+	 * [b] a single '/' to indicate the provider had no
+	 *     information and that we should consider everything
+	 *     invalid.  We call this a trivial response.
+	 */
+	if (query_success && !is_trivial) {
+		/*
+		 * Mark all pathnames returned by the monitor as dirty.
+		 *
+		 * This updates both the cache-entries and the untracked-cache.
+		 */
 		buf = query_result.buf;
 		for (i = bol; i < query_result.len; i++) {
 			if (buf[i] != '\0')
@@ -318,11 +328,16 @@ void refresh_fsmonitor(struct index_state *istate)
 		if (istate->untracked)
 			istate->untracked->use_fsmonitor = 1;
 	} else {
-
-		/* We only want to run the post index changed hook if we've actually changed entries, so keep track
-		 * if we actually changed entries or not */
+		/*
+		 * We failed to get a response or received a trivial response,
+		 * so invalidate everything.
+		 *
+		 * We only want to run the post index changed hook if
+		 * we've actually changed entries, so keep track if we
+		 * actually changed entries or not.
+		 */
 		int is_cache_changed = 0;
-		/* Mark all entries invalid */
+
 		for (i = 0; i < istate->cache_nr; i++) {
 			if (istate->cache[i]->ce_flags & CE_FSMONITOR_VALID) {
 				is_cache_changed = 1;
@@ -330,7 +345,10 @@ void refresh_fsmonitor(struct index_state *istate)
 			}
 		}
 
-		/* If we're going to check every file, ensure we save the results */
+		/*
+		 * If we're going to check every file, ensure we save
+		 * the results.
+		 */
 		if (is_cache_changed)
 			istate->cache_changed |= FSMONITOR_CHANGED;
 
-- 
gitgitgadget

[PATCH v8 02/30] fsmonitor-ipc: create client routines for git-fsmonitor--daemon

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Create fsmonitor_ipc__*() client routines to spawn the built-in file
system monitor daemon and send it an IPC request using the `Simple
IPC` API.

Stub in empty fsmonitor_ipc__*() functions for unsupported platforms.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 Makefile        |   1 +
 fsmonitor-ipc.c | 171 ++++++++++++++++++++++++++++++++++++++++++++++++
 fsmonitor-ipc.h |  48 ++++++++++++++
 3 files changed, 220 insertions(+)
 create mode 100644 fsmonitor-ipc.c
 create mode 100644 fsmonitor-ipc.h

diff --git a/Makefile b/Makefile
index 6f0b4b775fe..a19d850e716 100644
--- a/Makefile
+++ b/Makefile
@@ -907,6 +907,7 @@ LIB_OBJS += fetch-pack.o
 LIB_OBJS += fmt-merge-msg.o
 LIB_OBJS += fsck.o
 LIB_OBJS += fsmonitor.o
+LIB_OBJS += fsmonitor-ipc.o
 LIB_OBJS += gettext.o
 LIB_OBJS += gpg-interface.o
 LIB_OBJS += graph.o
diff --git a/fsmonitor-ipc.c b/fsmonitor-ipc.c
new file mode 100644
index 00000000000..789e7397baa
--- /dev/null
+++ b/fsmonitor-ipc.c
@@ -0,0 +1,171 @@
+#include "cache.h"
+#include "fsmonitor.h"
+#include "simple-ipc.h"
+#include "fsmonitor-ipc.h"
+#include "run-command.h"
+#include "strbuf.h"
+#include "trace2.h"
+
+#ifndef HAVE_FSMONITOR_DAEMON_BACKEND
+
+/*
+ * A trivial implementation of the fsmonitor_ipc__ API for unsupported
+ * platforms.
+ */
+
+int fsmonitor_ipc__is_supported(void)
+{
+	return 0;
+}
+
+const char *fsmonitor_ipc__get_path(void)
+{
+	return NULL;
+}
+
+enum ipc_active_state fsmonitor_ipc__get_state(void)
+{
+	return IPC_STATE__OTHER_ERROR;
+}
+
+int fsmonitor_ipc__send_query(const char *since_token,
+			      struct strbuf *answer)
+{
+	return -1;
+}
+
+int fsmonitor_ipc__send_command(const char *command,
+				struct strbuf *answer)
+{
+	return -1;
+}
+
+#else
+
+int fsmonitor_ipc__is_supported(void)
+{
+	return 1;
+}
+
+GIT_PATH_FUNC(fsmonitor_ipc__get_path, "fsmonitor--daemon.ipc")
+
+enum ipc_active_state fsmonitor_ipc__get_state(void)
+{
+	return ipc_get_active_state(fsmonitor_ipc__get_path());
+}
+
+static int spawn_daemon(void)
+{
+	const char *args[] = { "fsmonitor--daemon", "start", NULL };
+
+	return run_command_v_opt_tr2(args, RUN_COMMAND_NO_STDIN | RUN_GIT_CMD,
+				    "fsmonitor");
+}
+
+int fsmonitor_ipc__send_query(const char *since_token,
+			      struct strbuf *answer)
+{
+	int ret = -1;
+	int tried_to_spawn = 0;
+	enum ipc_active_state state = IPC_STATE__OTHER_ERROR;
+	struct ipc_client_connection *connection = NULL;
+	struct ipc_client_connect_options options
+		= IPC_CLIENT_CONNECT_OPTIONS_INIT;
+	const char *tok = since_token ? since_token : "";
+	size_t tok_len = since_token ? strlen(since_token) : 0;
+
+	options.wait_if_busy = 1;
+	options.wait_if_not_found = 0;
+
+	trace2_region_enter("fsm_client", "query", NULL);
+	trace2_data_string("fsm_client", NULL, "query/command", tok);
+
+try_again:
+	state = ipc_client_try_connect(fsmonitor_ipc__get_path(), &options,
+				       &connection);
+
+	switch (state) {
+	case IPC_STATE__LISTENING:
+		ret = ipc_client_send_command_to_connection(
+			connection, tok, tok_len, answer);
+		ipc_client_close_connection(connection);
+
+		trace2_data_intmax("fsm_client", NULL,
+				   "query/response-length", answer->len);
+		goto done;
+
+	case IPC_STATE__NOT_LISTENING:
+	case IPC_STATE__PATH_NOT_FOUND:
+		if (tried_to_spawn)
+			goto done;
+
+		tried_to_spawn++;
+		if (spawn_daemon())
+			goto done;
+
+		/*
+		 * Try again, but this time give the daemon a chance to
+		 * actually create the pipe/socket.
+		 *
+		 * Granted, the daemon just started so it can't possibly have
+		 * any FS cached yet, so we'll always get a trivial answer.
+		 * BUT the answer should include a new token that can serve
+		 * as the basis for subsequent requests.
+		 */
+		options.wait_if_not_found = 1;
+		goto try_again;
+
+	case IPC_STATE__INVALID_PATH:
+		ret = error(_("fsmonitor_ipc__send_query: invalid path '%s'"),
+			    fsmonitor_ipc__get_path());
+		goto done;
+
+	case IPC_STATE__OTHER_ERROR:
+	default:
+		ret = error(_("fsmonitor_ipc__send_query: unspecified error on '%s'"),
+			    fsmonitor_ipc__get_path());
+		goto done;
+	}
+
+done:
+	trace2_region_leave("fsm_client", "query", NULL);
+
+	return ret;
+}
+
+int fsmonitor_ipc__send_command(const char *command,
+				struct strbuf *answer)
+{
+	struct ipc_client_connection *connection = NULL;
+	struct ipc_client_connect_options options
+		= IPC_CLIENT_CONNECT_OPTIONS_INIT;
+	int ret;
+	enum ipc_active_state state;
+	const char *c = command ? command : "";
+	size_t c_len = command ? strlen(command) : 0;
+
+	strbuf_reset(answer);
+
+	options.wait_if_busy = 1;
+	options.wait_if_not_found = 0;
+
+	state = ipc_client_try_connect(fsmonitor_ipc__get_path(), &options,
+				       &connection);
+	if (state != IPC_STATE__LISTENING) {
+		die(_("fsmonitor--daemon is not running"));
+		return -1;
+	}
+
+	ret = ipc_client_send_command_to_connection(connection, c, c_len,
+						    answer);
+	ipc_client_close_connection(connection);
+
+	if (ret == -1) {
+		die(_("could not send '%s' command to fsmonitor--daemon"), c);
+		return -1;
+	}
+
+	return 0;
+}
+
+#endif
diff --git a/fsmonitor-ipc.h b/fsmonitor-ipc.h
new file mode 100644
index 00000000000..b6a7067c3af
--- /dev/null
+++ b/fsmonitor-ipc.h
@@ -0,0 +1,48 @@
+#ifndef FSMONITOR_IPC_H
+#define FSMONITOR_IPC_H
+
+#include "simple-ipc.h"
+
+/*
+ * Returns true if built-in file system monitor daemon is defined
+ * for this platform.
+ */
+int fsmonitor_ipc__is_supported(void);
+
+/*
+ * Returns the pathname to the IPC named pipe or Unix domain socket
+ * where a `git-fsmonitor--daemon` process will listen.  This is a
+ * per-worktree value.
+ *
+ * Returns NULL if the daemon is not supported on this platform.
+ */
+const char *fsmonitor_ipc__get_path(void);
+
+/*
+ * Try to determine whether there is a `git-fsmonitor--daemon` process
+ * listening on the IPC pipe/socket.
+ */
+enum ipc_active_state fsmonitor_ipc__get_state(void);
+
+/*
+ * Connect to a `git-fsmonitor--daemon` process via simple-ipc
+ * and ask for the set of changed files since the given token.
+ *
+ * Spawn a daemon process in the background if necessary.
+ *
+ * Returns -1 on error; 0 on success.
+ */
+int fsmonitor_ipc__send_query(const char *since_token,
+			      struct strbuf *answer);
+
+/*
+ * Connect to a `git-fsmonitor--daemon` process via simple-ipc and
+ * send a command verb.  If no daemon is available, we DO NOT try to
+ * start one.
+ *
+ * Returns -1 on error; 0 on success.
+ */
+int fsmonitor_ipc__send_command(const char *command,
+				struct strbuf *answer);
+
+#endif /* FSMONITOR_IPC_H */
-- 
gitgitgadget

[PATCH v8 03/30] fsmonitor: config settings are repository-specific

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Move fsmonitor config settings to a new and opaque
`struct fsmonitor_settings` structure.  Add a lazily-loaded pointer
to this into `struct repo_settings`

Create an `enum fsmonitor_mode` type in `struct fsmonitor_settings` to
represent the state of fsmonitor.  This lets us represent which, if
any, fsmonitor provider (hook or IPC) is enabled.

Create `fsm_settings__get_*()` getters to lazily look up fsmonitor-
related config settings.

Get rid of the `core_fsmonitor` global variable.  Move the code to
lookup the existing `core.fsmonitor` config value into the fsmonitor
settings.

Create a hook pathname variable in `struct fsmonitor-settings` and
only set it when in hook mode.

Extend the definition of `core.fsmonitor` to be either a boolean
or a hook pathname.  When true, the builtin FSMonitor is used.
When false or unset, no FSMonitor (neither builtin nor hook) is
used.

The existing `core_fsmonitor` global variable was used to store the
pathname to the fsmonitor hook *and* it was used as a boolean to see
if fsmonitor was enabled.  This dual usage and global visibility leads
to confusion when we add the IPC-based provider.  So lets hide the
details in fsmonitor-settings.c and let it decide which provider to
use in the case of multiple settings.  This avoids cluttering up
repo-settings.c with these private details.

A future commit in builtin-fsmonitor series will add the ability to
disqualify worktrees for various reasons, such as being mounted from a
remote volume, where fsmonitor should not be started.  Having the
config settings hidden in fsmonitor-settings.c allows such worktree
restrictions to override the config values used.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 Makefile               |   1 +
 builtin/update-index.c |   7 ++-
 cache.h                |   1 -
 config.c               |  14 -----
 config.h               |   1 -
 environment.c          |   1 -
 fsmonitor-settings.c   | 114 +++++++++++++++++++++++++++++++++++++++++
 fsmonitor-settings.h   |  21 ++++++++
 fsmonitor.c            |  63 ++++++++++++++---------
 fsmonitor.h            |  15 ++++--
 repository.h           |   3 ++
 t/README               |   4 +-
 12 files changed, 196 insertions(+), 49 deletions(-)
 create mode 100644 fsmonitor-settings.c
 create mode 100644 fsmonitor-settings.h

diff --git a/Makefile b/Makefile
index a19d850e716..707a56d4c11 100644
--- a/Makefile
+++ b/Makefile
@@ -908,6 +908,7 @@ LIB_OBJS += fmt-merge-msg.o
 LIB_OBJS += fsck.o
 LIB_OBJS += fsmonitor.o
 LIB_OBJS += fsmonitor-ipc.o
+LIB_OBJS += fsmonitor-settings.o
 LIB_OBJS += gettext.o
 LIB_OBJS += gpg-interface.o
 LIB_OBJS += graph.o
diff --git a/builtin/update-index.c b/builtin/update-index.c
index aafe7eeac2a..876112abb21 100644
--- a/builtin/update-index.c
+++ b/builtin/update-index.c
@@ -1236,14 +1236,17 @@ int cmd_update_index(int argc, const char **argv, const char *prefix)
 	}
 
 	if (fsmonitor > 0) {
-		if (git_config_get_fsmonitor() == 0)
+		enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
+		if (fsm_mode == FSMONITOR_MODE_DISABLED) {
 			warning(_("core.fsmonitor is unset; "
 				"set it if you really want to "
 				"enable fsmonitor"));
+		}
 		add_fsmonitor(&the_index);
 		report(_("fsmonitor enabled"));
 	} else if (!fsmonitor) {
-		if (git_config_get_fsmonitor() == 1)
+		enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
+		if (fsm_mode > FSMONITOR_MODE_DISABLED)
 			warning(_("core.fsmonitor is set; "
 				"remove it if you really want to "
 				"disable fsmonitor"));
diff --git a/cache.h b/cache.h
index 04d4d2db25c..aaf334e2aa4 100644
--- a/cache.h
+++ b/cache.h
@@ -999,7 +999,6 @@ extern int core_preload_index;
 extern int precomposed_unicode;
 extern int protect_hfs;
 extern int protect_ntfs;
-extern const char *core_fsmonitor;
 
 extern int core_apply_sparse_checkout;
 extern int core_sparse_checkout_cone;
diff --git a/config.c b/config.c
index 383b1a4885b..3f9b0739a78 100644
--- a/config.c
+++ b/config.c
@@ -2626,20 +2626,6 @@ int git_config_get_max_percent_split_change(void)
 	return -1; /* default value */
 }
 
-int git_config_get_fsmonitor(void)
-{
-	if (git_config_get_pathname("core.fsmonitor", &core_fsmonitor))
-		core_fsmonitor = getenv("GIT_TEST_FSMONITOR");
-
-	if (core_fsmonitor && !*core_fsmonitor)
-		core_fsmonitor = NULL;
-
-	if (core_fsmonitor)
-		return 1;
-
-	return 0;
-}
-
 int git_config_get_index_threads(int *dest)
 {
 	int is_bool, val;
diff --git a/config.h b/config.h
index bb49baf1ee0..7654f61c634 100644
--- a/config.h
+++ b/config.h
@@ -597,7 +597,6 @@ int git_config_get_pathname(const char *key, const char **dest);
 int git_config_get_index_threads(int *dest);
 int git_config_get_split_index(void);
 int git_config_get_max_percent_split_change(void);
-int git_config_get_fsmonitor(void);
 
 /* This dies if the configured or default date is in the future */
 int git_config_get_expiry(const char *key, const char **output);
diff --git a/environment.c b/environment.c
index fd0501e77a5..00682e638d7 100644
--- a/environment.c
+++ b/environment.c
@@ -84,7 +84,6 @@ int protect_hfs = PROTECT_HFS_DEFAULT;
 #define PROTECT_NTFS_DEFAULT 1
 #endif
 int protect_ntfs = PROTECT_NTFS_DEFAULT;
-const char *core_fsmonitor;
 
 /*
  * The character that begins a commented line in user-editable file
diff --git a/fsmonitor-settings.c b/fsmonitor-settings.c
new file mode 100644
index 00000000000..757d230d538
--- /dev/null
+++ b/fsmonitor-settings.c
@@ -0,0 +1,114 @@
+#include "cache.h"
+#include "config.h"
+#include "repository.h"
+#include "fsmonitor-settings.h"
+
+/*
+ * We keep this structure defintion private and have getters
+ * for all fields so that we can lazy load it as needed.
+ */
+struct fsmonitor_settings {
+	enum fsmonitor_mode mode;
+	char *hook_path;
+};
+
+static void lookup_fsmonitor_settings(struct repository *r)
+{
+	struct fsmonitor_settings *s;
+	const char *const_str;
+	int bool_value;
+
+	if (r->settings.fsmonitor)
+		return;
+
+	CALLOC_ARRAY(s, 1);
+	s->mode = FSMONITOR_MODE_DISABLED;
+
+	r->settings.fsmonitor = s;
+
+	/*
+	 * Overload the existing "core.fsmonitor" config setting (which
+	 * has historically been either unset or a hook pathname) to
+	 * now allow a boolean value to enable the builtin FSMonitor
+	 * or to turn everything off.  (This does imply that you can't
+	 * use a hook script named "true" or "false", but that's OK.)
+	 */
+	switch (repo_config_get_maybe_bool(r, "core.fsmonitor", &bool_value)) {
+
+	case 0: /* config value was set to <bool> */
+		if (bool_value)
+			fsm_settings__set_ipc(r);
+		return;
+
+	case 1: /* config value was unset */
+		const_str = getenv("GIT_TEST_FSMONITOR");
+		break;
+
+	case -1: /* config value set to an arbitrary string */
+		if (repo_config_get_pathname(r, "core.fsmonitor", &const_str))
+			return; /* should not happen */
+		break;
+
+	default: /* should not happen */
+		return;
+	}
+
+	if (!const_str || !*const_str)
+		return;
+
+	fsm_settings__set_hook(r, const_str);
+}
+
+enum fsmonitor_mode fsm_settings__get_mode(struct repository *r)
+{
+	if (!r)
+		r = the_repository;
+
+	lookup_fsmonitor_settings(r);
+
+	return r->settings.fsmonitor->mode;
+}
+
+const char *fsm_settings__get_hook_path(struct repository *r)
+{
+	if (!r)
+		r = the_repository;
+
+	lookup_fsmonitor_settings(r);
+
+	return r->settings.fsmonitor->hook_path;
+}
+
+void fsm_settings__set_ipc(struct repository *r)
+{
+	if (!r)
+		r = the_repository;
+
+	lookup_fsmonitor_settings(r);
+
+	r->settings.fsmonitor->mode = FSMONITOR_MODE_IPC;
+	FREE_AND_NULL(r->settings.fsmonitor->hook_path);
+}
+
+void fsm_settings__set_hook(struct repository *r, const char *path)
+{
+	if (!r)
+		r = the_repository;
+
+	lookup_fsmonitor_settings(r);
+
+	r->settings.fsmonitor->mode = FSMONITOR_MODE_HOOK;
+	FREE_AND_NULL(r->settings.fsmonitor->hook_path);
+	r->settings.fsmonitor->hook_path = strdup(path);
+}
+
+void fsm_settings__set_disabled(struct repository *r)
+{
+	if (!r)
+		r = the_repository;
+
+	lookup_fsmonitor_settings(r);
+
+	r->settings.fsmonitor->mode = FSMONITOR_MODE_DISABLED;
+	FREE_AND_NULL(r->settings.fsmonitor->hook_path);
+}
diff --git a/fsmonitor-settings.h b/fsmonitor-settings.h
new file mode 100644
index 00000000000..a4c5d7b4889
--- /dev/null
+++ b/fsmonitor-settings.h
@@ -0,0 +1,21 @@
+#ifndef FSMONITOR_SETTINGS_H
+#define FSMONITOR_SETTINGS_H
+
+struct repository;
+
+enum fsmonitor_mode {
+	FSMONITOR_MODE_DISABLED = 0,
+	FSMONITOR_MODE_HOOK = 1, /* core.fsmonitor=<hook_path> */
+	FSMONITOR_MODE_IPC = 2,  /* core.fsmonitor=<true> */
+};
+
+void fsm_settings__set_ipc(struct repository *r);
+void fsm_settings__set_hook(struct repository *r, const char *path);
+void fsm_settings__set_disabled(struct repository *r);
+
+enum fsmonitor_mode fsm_settings__get_mode(struct repository *r);
+const char *fsm_settings__get_hook_path(struct repository *r);
+
+struct fsmonitor_settings;
+
+#endif /* FSMONITOR_SETTINGS_H */
diff --git a/fsmonitor.c b/fsmonitor.c
index 448d0ee33f5..0e961b74d82 100644
--- a/fsmonitor.c
+++ b/fsmonitor.c
@@ -3,6 +3,7 @@
 #include "dir.h"
 #include "ewah/ewok.h"
 #include "fsmonitor.h"
+#include "fsmonitor-ipc.h"
 #include "run-command.h"
 #include "strbuf.h"
 
@@ -148,15 +149,18 @@ void write_fsmonitor_extension(struct strbuf *sb, struct index_state *istate)
 /*
  * Call the query-fsmonitor hook passing the last update token of the saved results.
  */
-static int query_fsmonitor(int version, const char *last_update, struct strbuf *query_result)
+static int query_fsmonitor_hook(struct repository *r,
+				int version,
+				const char *last_update,
+				struct strbuf *query_result)
 {
 	struct child_process cp = CHILD_PROCESS_INIT;
 	int result;
 
-	if (!core_fsmonitor)
+	if (fsm_settings__get_mode(r) != FSMONITOR_MODE_HOOK)
 		return -1;
 
-	strvec_push(&cp.args, core_fsmonitor);
+	strvec_push(&cp.args, fsm_settings__get_hook_path(r));
 	strvec_pushf(&cp.args, "%d", version);
 	strvec_pushf(&cp.args, "%s", last_update);
 	cp.use_shell = 1;
@@ -225,17 +229,28 @@ void refresh_fsmonitor(struct index_state *istate)
 	char *buf;
 	unsigned int i;
 	int is_trivial = 0;
+	struct repository *r = istate->repo ? istate->repo : the_repository;
+	enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
 
-	if (!core_fsmonitor || istate->fsmonitor_has_run_once)
+	if (fsm_mode <= FSMONITOR_MODE_DISABLED ||
+	    istate->fsmonitor_has_run_once)
 		return;
 
-	hook_version = fsmonitor_hook_version();
-
 	istate->fsmonitor_has_run_once = 1;
 
 	trace_printf_key(&trace_fsmonitor, "refresh fsmonitor");
+
+	if (fsm_mode == FSMONITOR_MODE_IPC) {
+		/* TODO */
+		return;
+	}
+
+	assert(fsm_mode == FSMONITOR_MODE_HOOK);
+
+	hook_version = fsmonitor_hook_version();
+
 	/*
-	 * This could be racy so save the date/time now and query_fsmonitor
+	 * This could be racy so save the date/time now and query_fsmonitor_hook
 	 * should be inclusive to ensure we don't miss potential changes.
 	 */
 	last_update = getnanotime();
@@ -243,13 +258,14 @@ void refresh_fsmonitor(struct index_state *istate)
 		strbuf_addf(&last_update_token, "%"PRIu64"", last_update);
 
 	/*
-	 * If we have a last update token, call query_fsmonitor for the set of
+	 * If we have a last update token, call query_fsmonitor_hook for the set of
 	 * changes since that token, else assume everything is possibly dirty
 	 * and check it all.
 	 */
 	if (istate->fsmonitor_last_update) {
 		if (hook_version == -1 || hook_version == HOOK_INTERFACE_VERSION2) {
-			query_success = !query_fsmonitor(HOOK_INTERFACE_VERSION2,
+			query_success = !query_fsmonitor_hook(
+				r, HOOK_INTERFACE_VERSION2,
 				istate->fsmonitor_last_update, &query_result);
 
 			if (query_success) {
@@ -280,7 +296,8 @@ void refresh_fsmonitor(struct index_state *istate)
 		}
 
 		if (hook_version == HOOK_INTERFACE_VERSION1) {
-			query_success = !query_fsmonitor(HOOK_INTERFACE_VERSION1,
+			query_success = !query_fsmonitor_hook(
+				r, HOOK_INTERFACE_VERSION1,
 				istate->fsmonitor_last_update, &query_result);
 			if (query_success)
 				is_trivial = query_result.buf[0] == '/';
@@ -290,9 +307,12 @@ void refresh_fsmonitor(struct index_state *istate)
 			trace2_data_intmax("fsm_hook", NULL,
 					   "query/trivial-response", 1);
 
-		trace_performance_since(last_update, "fsmonitor process '%s'", core_fsmonitor);
-		trace_printf_key(&trace_fsmonitor, "fsmonitor process '%s' returned %s",
-			core_fsmonitor, query_success ? "success" : "failure");
+		trace_performance_since(last_update, "fsmonitor process '%s'",
+					fsm_settings__get_hook_path(r));
+		trace_printf_key(&trace_fsmonitor,
+				 "fsmonitor process '%s' returned %s",
+				 fsm_settings__get_hook_path(r),
+				 query_success ? "success" : "failure");
 	}
 
 	/*
@@ -429,7 +449,8 @@ void remove_fsmonitor(struct index_state *istate)
 void tweak_fsmonitor(struct index_state *istate)
 {
 	unsigned int i;
-	int fsmonitor_enabled = git_config_get_fsmonitor();
+	int fsmonitor_enabled = (fsm_settings__get_mode(istate->repo)
+				 > FSMONITOR_MODE_DISABLED);
 
 	if (istate->fsmonitor_dirty) {
 		if (fsmonitor_enabled) {
@@ -449,16 +470,8 @@ void tweak_fsmonitor(struct index_state *istate)
 		istate->fsmonitor_dirty = NULL;
 	}
 
-	switch (fsmonitor_enabled) {
-	case -1: /* keep: do nothing */
-		break;
-	case 0: /* false */
-		remove_fsmonitor(istate);
-		break;
-	case 1: /* true */
+	if (fsmonitor_enabled)
 		add_fsmonitor(istate);
-		break;
-	default: /* unknown value: do nothing */
-		break;
-	}
+	else
+		remove_fsmonitor(istate);
 }
diff --git a/fsmonitor.h b/fsmonitor.h
index f20d72631d7..3f41f653691 100644
--- a/fsmonitor.h
+++ b/fsmonitor.h
@@ -3,6 +3,7 @@
 
 #include "cache.h"
 #include "dir.h"
+#include "fsmonitor-settings.h"
 
 extern struct trace_key trace_fsmonitor;
 
@@ -57,7 +58,10 @@ int fsmonitor_is_trivial_response(const struct strbuf *query_result);
  */
 static inline int is_fsmonitor_refreshed(const struct index_state *istate)
 {
-	return !core_fsmonitor || istate->fsmonitor_has_run_once;
+	enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(istate->repo);
+
+	return fsm_mode <= FSMONITOR_MODE_DISABLED ||
+		istate->fsmonitor_has_run_once;
 }
 
 /*
@@ -67,7 +71,10 @@ static inline int is_fsmonitor_refreshed(const struct index_state *istate)
  */
 static inline void mark_fsmonitor_valid(struct index_state *istate, struct cache_entry *ce)
 {
-	if (core_fsmonitor && !(ce->ce_flags & CE_FSMONITOR_VALID)) {
+	enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(istate->repo);
+
+	if (fsm_mode > FSMONITOR_MODE_DISABLED &&
+	    !(ce->ce_flags & CE_FSMONITOR_VALID)) {
 		istate->cache_changed = 1;
 		ce->ce_flags |= CE_FSMONITOR_VALID;
 		trace_printf_key(&trace_fsmonitor, "mark_fsmonitor_clean '%s'", ce->name);
@@ -83,7 +90,9 @@ static inline void mark_fsmonitor_valid(struct index_state *istate, struct cache
  */
 static inline void mark_fsmonitor_invalid(struct index_state *istate, struct cache_entry *ce)
 {
-	if (core_fsmonitor) {
+	enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(istate->repo);
+
+	if (fsm_mode > FSMONITOR_MODE_DISABLED) {
 		ce->ce_flags &= ~CE_FSMONITOR_VALID;
 		untracked_cache_invalidate_path(istate, ce->name, 1);
 		trace_printf_key(&trace_fsmonitor, "mark_fsmonitor_invalid '%s'", ce->name);
diff --git a/repository.h b/repository.h
index ca837cb9e91..9bbb4659cc8 100644
--- a/repository.h
+++ b/repository.h
@@ -4,6 +4,7 @@
 #include "path.h"
 
 struct config_set;
+struct fsmonitor_settings;
 struct git_hash_algo;
 struct index_state;
 struct lock_file;
@@ -35,6 +36,8 @@ struct repo_settings {
 	int command_requires_full_index;
 	int sparse_index;
 
+	struct fsmonitor_settings *fsmonitor; /* lazily loaded */
+
 	int index_version;
 	enum untracked_cache_setting core_untracked_cache;
 
diff --git a/t/README b/t/README
index f48e0542cdc..9ffea1d3147 100644
--- a/t/README
+++ b/t/README
@@ -405,8 +405,8 @@ every 'git commit-graph write', as if the `--changed-paths` option was
 passed in.
 
 GIT_TEST_FSMONITOR=$PWD/t7519/fsmonitor-all exercises the fsmonitor
-code path for utilizing a file system monitor to speed up detecting
-new or changed files.
+code paths for utilizing a (hook based) file system monitor to speed up
+detecting new or changed files.
 
 GIT_TEST_INDEX_VERSION=<n> exercises the index read/write code path
 for the index version specified.  Can be set to any valid version
-- 
gitgitgadget

[PATCH v8 04/30] fsmonitor: use IPC to query the builtin FSMonitor daemon

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Use simple IPC to directly communicate with the new builtin file
system monitor daemon when `core.fsmonitor` is set to true.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 fsmonitor.c | 38 ++++++++++++++++++++++++++++++++++++--
 1 file changed, 36 insertions(+), 2 deletions(-)

diff --git a/fsmonitor.c b/fsmonitor.c
index 0e961b74d82..a38b5710eb3 100644
--- a/fsmonitor.c
+++ b/fsmonitor.c
@@ -241,8 +241,41 @@ void refresh_fsmonitor(struct index_state *istate)
 	trace_printf_key(&trace_fsmonitor, "refresh fsmonitor");
 
 	if (fsm_mode == FSMONITOR_MODE_IPC) {
-		/* TODO */
-		return;
+		query_success = !fsmonitor_ipc__send_query(
+			istate->fsmonitor_last_update ?
+			istate->fsmonitor_last_update : "builtin:fake",
+			&query_result);
+		if (query_success) {
+			/*
+			 * The response contains a series of nul terminated
+			 * strings.  The first is the new token.
+			 *
+			 * Use `char *buf` as an interlude to trick the CI
+			 * static analysis to let us use `strbuf_addstr()`
+			 * here (and only copy the token) rather than
+			 * `strbuf_addbuf()`.
+			 */
+			buf = query_result.buf;
+			strbuf_addstr(&last_update_token, buf);
+			bol = last_update_token.len + 1;
+			is_trivial = query_result.buf[bol] == '/';
+			if (is_trivial)
+				trace2_data_intmax("fsm_client", NULL,
+						   "query/trivial-response", 1);
+		} else {
+			/*
+			 * The builtin daemon is not available on this
+			 * platform -OR- we failed to get a response.
+			 *
+			 * Generate a fake token (rather than a V1
+			 * timestamp) for the index extension.  (If
+			 * they switch back to the hook API, we don't
+			 * want ambiguous state.)
+			 */
+			strbuf_addstr(&last_update_token, "builtin:fake");
+		}
+
+		goto apply_results;
 	}
 
 	assert(fsm_mode == FSMONITOR_MODE_HOOK);
@@ -315,6 +348,7 @@ void refresh_fsmonitor(struct index_state *istate)
 				 query_success ? "success" : "failure");
 	}
 
+apply_results:
 	/*
 	 * The response from FSMonitor (excluding the header token) is
 	 * either:
-- 
gitgitgadget

[PATCH v8 05/30] fsmonitor: document builtin fsmonitor

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Document how `core.fsmonitor` can be set to a boolean to enable
or disable the builtin FSMonitor.

Update references to `core.fsmonitor` and `core.fsmonitorHookVersion` and
pointers to `Watchman` to refer to it.

Create `git-fsmonitor--daemon` manual page and describe its features.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 Documentation/config/core.txt           | 60 +++++++++++++++-----
 Documentation/git-fsmonitor--daemon.txt | 75 +++++++++++++++++++++++++
 Documentation/git-update-index.txt      |  8 ++-
 3 files changed, 126 insertions(+), 17 deletions(-)
 create mode 100644 Documentation/git-fsmonitor--daemon.txt

diff --git a/Documentation/config/core.txt b/Documentation/config/core.txt
index c04f62a54a1..6303c36c7ed 100644
--- a/Documentation/config/core.txt
+++ b/Documentation/config/core.txt
@@ -62,22 +62,54 @@ core.protectNTFS::
 	Defaults to `true` on Windows, and `false` elsewhere.
 
 core.fsmonitor::
-	If set, the value of this variable is used as a command which
-	will identify all files that may have changed since the
-	requested date/time. This information is used to speed up git by
-	avoiding unnecessary processing of files that have not changed.
-	See the "fsmonitor-watchman" section of linkgit:githooks[5].
+	If set to true, enable the built-in file system monitor
+	daemon for this working directory (linkgit:git-fsmonitor--daemon[1]).
++
+Like hook-based file system monitors, the built-in file system monitor
+can speed up Git commands that need to refresh the Git index
+(e.g. `git status`) in a working directory with many files.  The
+built-in monitor eliminates the need to install and maintain an
+external third-party tool.
++
+The built-in file system monitor is currently available only on a
+limited set of supported platforms.  Currently, this includes Windows
+and MacOS.
++
+	Otherwise, this variable contains the pathname of the "fsmonitor"
+	hook command.
++
+This hook command is used to identify all files that may have changed
+since the requested date/time. This information is used to speed up
+git by avoiding unnecessary scanning of files that have not changed.
++
+See the "fsmonitor-watchman" section of linkgit:githooks[5].
++
+Note that if you concurrently use multiple versions of Git, such
+as one version on the command line and another version in an IDE
+tool, that the definition of `core.fsmonitor` was extended to
+allow boolean values in addition to hook pathnames.  Git versions
+2.35.1 and prior will not understand the boolean values and will
+consider the "true" or "false" values as hook pathnames to be
+invoked.  Git versions 2.26 thru 2.35.1 default to hook protocol
+V2 and will fall back to no fsmonitor (full scan).  Git versions
+prior to 2.26 default to hook protocol V1 and will silently
+assume there were no changes to report (no scan), so status
+commands may report incomplete results.  For this reason, it is
+best to upgrade all of your Git versions before using the built-in
+file system monitor.
 
 core.fsmonitorHookVersion::
-	Sets the version of hook that is to be used when calling fsmonitor.
-	There are currently versions 1 and 2. When this is not set,
-	version 2 will be tried first and if it fails then version 1
-	will be tried. Version 1 uses a timestamp as input to determine
-	which files have changes since that time but some monitors
-	like watchman have race conditions when used with a timestamp.
-	Version 2 uses an opaque string so that the monitor can return
-	something that can be used to determine what files have changed
-	without race conditions.
+	Sets the protocol version to be used when invoking the
+	"fsmonitor" hook.
++
+There are currently versions 1 and 2. When this is not set,
+version 2 will be tried first and if it fails then version 1
+will be tried. Version 1 uses a timestamp as input to determine
+which files have changes since that time but some monitors
+like Watchman have race conditions when used with a timestamp.
+Version 2 uses an opaque string so that the monitor can return
+something that can be used to determine what files have changed
+without race conditions.
 
 core.trustctime::
 	If false, the ctime differences between the index and the
diff --git a/Documentation/git-fsmonitor--daemon.txt b/Documentation/git-fsmonitor--daemon.txt
new file mode 100644
index 00000000000..0fedf5a4565
--- /dev/null
+++ b/Documentation/git-fsmonitor--daemon.txt
@@ -0,0 +1,75 @@
+git-fsmonitor--daemon(1)
+========================
+
+NAME
+----
+git-fsmonitor--daemon - A Built-in File System Monitor
+
+SYNOPSIS
+--------
+[verse]
+'git fsmonitor--daemon' start
+'git fsmonitor--daemon' run
+'git fsmonitor--daemon' stop
+'git fsmonitor--daemon' status
+
+DESCRIPTION
+-----------
+
+A daemon to watch the working directory for file and directory
+changes using platform-specific file system notification facilities.
+
+This daemon communicates directly with commands like `git status`
+using the link:technical/api-simple-ipc.html[simple IPC] interface
+instead of the slower linkgit:githooks[5] interface.
+
+This daemon is built into Git so that no third-party tools are
+required.
+
+OPTIONS
+-------
+
+start::
+	Starts a daemon in the background.
+
+run::
+	Runs a daemon in the foreground.
+
+stop::
+	Stops the daemon running in the current working
+	directory, if present.
+
+status::
+	Exits with zero status if a daemon is watching the
+	current working directory.
+
+REMARKS
+-------
+
+This daemon is a long running process used to watch a single working
+directory and maintain a list of the recently changed files and
+directories.  Performance of commands such as `git status` can be
+increased if they just ask for a summary of changes to the working
+directory and can avoid scanning the disk.
+
+When `core.fsmonitor` is set to `true` (see linkgit:git-config[1])
+commands, such as `git status`, will ask the daemon for changes and
+automatically start it (if necessary).
+
+For more information see the "File System Monitor" section in
+linkgit:git-update-index[1].
+
+CAVEATS
+-------
+
+The fsmonitor daemon does not currently know about submodules and does
+not know to filter out file system events that happen within a
+submodule.  If fsmonitor daemon is watching a super repo and a file is
+modified within the working directory of a submodule, it will report
+the change (as happening against the super repo).  However, the client
+will properly ignore these extra events, so performance may be affected
+but it will not cause an incorrect result.
+
+GIT
+---
+Part of the linkgit:git[1] suite
diff --git a/Documentation/git-update-index.txt b/Documentation/git-update-index.txt
index 2853f168d97..53ea48a04e2 100644
--- a/Documentation/git-update-index.txt
+++ b/Documentation/git-update-index.txt
@@ -498,7 +498,9 @@ FILE SYSTEM MONITOR
 This feature is intended to speed up git operations for repos that have
 large working directories.
 
-It enables git to work together with a file system monitor (see the
+It enables git to work together with a file system monitor (see
+linkgit:git-fsmonitor--daemon[1]
+and the
 "fsmonitor-watchman" section of linkgit:githooks[5]) that can
 inform it as to what files have been modified. This enables git to avoid
 having to lstat() every file to find modified files.
@@ -509,8 +511,8 @@ looking for new files.
 
 If you want to enable (or disable) this feature, it is easier to use
 the `core.fsmonitor` configuration variable (see
-linkgit:git-config[1]) than using the `--fsmonitor` option to
-`git update-index` in each repository, especially if you want to do so
+linkgit:git-config[1]) than using the `--fsmonitor` option to `git
+update-index` in each repository, especially if you want to do so
 across all repositories you use, because you can set the configuration
 variable in your `$HOME/.gitconfig` just once and have it affect all
 repositories you touch.
-- 
gitgitgadget

[PATCH v8 06/30] fsmonitor--daemon: add a built-in fsmonitor daemon

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Create a built-in file system monitoring daemon that can be used by
the existing `fsmonitor` feature (protocol API and index extension)
to improve the performance of various Git commands, such as `status`.

The `fsmonitor--daemon` feature builds upon the `Simple IPC` API and
provides an alternative to hook access to existing fsmonitors such
as `watchman`.

This commit merely adds the new command without any functionality.

Co-authored-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 .gitignore                  |  1 +
 Makefile                    |  1 +
 builtin.h                   |  1 +
 builtin/fsmonitor--daemon.c | 46 +++++++++++++++++++++++++++++++++++++
 git.c                       |  1 +
 5 files changed, 50 insertions(+)
 create mode 100644 builtin/fsmonitor--daemon.c

diff --git a/.gitignore b/.gitignore
index f817c509ec0..e81de1063a4 100644
--- a/.gitignore
+++ b/.gitignore
@@ -72,6 +72,7 @@
 /git-format-patch
 /git-fsck
 /git-fsck-objects
+/git-fsmonitor--daemon
 /git-gc
 /git-get-tar-commit-id
 /git-grep
diff --git a/Makefile b/Makefile
index 707a56d4c11..5af1d5b112e 100644
--- a/Makefile
+++ b/Makefile
@@ -1114,6 +1114,7 @@ BUILTIN_OBJS += builtin/fmt-merge-msg.o
 BUILTIN_OBJS += builtin/for-each-ref.o
 BUILTIN_OBJS += builtin/for-each-repo.o
 BUILTIN_OBJS += builtin/fsck.o
+BUILTIN_OBJS += builtin/fsmonitor--daemon.o
 BUILTIN_OBJS += builtin/gc.o
 BUILTIN_OBJS += builtin/get-tar-commit-id.o
 BUILTIN_OBJS += builtin/grep.o
diff --git a/builtin.h b/builtin.h
index 83379f3832c..40e9ecc8485 100644
--- a/builtin.h
+++ b/builtin.h
@@ -159,6 +159,7 @@ int cmd_for_each_ref(int argc, const char **argv, const char *prefix);
 int cmd_for_each_repo(int argc, const char **argv, const char *prefix);
 int cmd_format_patch(int argc, const char **argv, const char *prefix);
 int cmd_fsck(int argc, const char **argv, const char *prefix);
+int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix);
 int cmd_gc(int argc, const char **argv, const char *prefix);
 int cmd_get_tar_commit_id(int argc, const char **argv, const char *prefix);
 int cmd_grep(int argc, const char **argv, const char *prefix);
diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
new file mode 100644
index 00000000000..f0498793379
--- /dev/null
+++ b/builtin/fsmonitor--daemon.c
@@ -0,0 +1,46 @@
+#include "builtin.h"
+#include "config.h"
+#include "parse-options.h"
+#include "fsmonitor.h"
+#include "fsmonitor-ipc.h"
+#include "simple-ipc.h"
+#include "khash.h"
+
+static const char * const builtin_fsmonitor__daemon_usage[] = {
+	NULL
+};
+
+#ifdef HAVE_FSMONITOR_DAEMON_BACKEND
+
+int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
+{
+	const char *subcmd;
+
+	struct option options[] = {
+		OPT_END()
+	};
+
+	git_config(git_default_config, NULL);
+
+	argc = parse_options(argc, argv, prefix, options,
+			     builtin_fsmonitor__daemon_usage, 0);
+	if (argc != 1)
+		usage_with_options(builtin_fsmonitor__daemon_usage, options);
+	subcmd = argv[0];
+
+	die(_("Unhandled subcommand '%s'"), subcmd);
+}
+
+#else
+int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
+{
+	struct option options[] = {
+		OPT_END()
+	};
+
+	if (argc == 2 && !strcmp(argv[1], "-h"))
+		usage_with_options(builtin_fsmonitor__daemon_usage, options);
+
+	die(_("fsmonitor--daemon not supported on this platform"));
+}
+#endif
diff --git a/git.c b/git.c
index a25940d72e8..3d8e48cf555 100644
--- a/git.c
+++ b/git.c
@@ -537,6 +537,7 @@ static struct cmd_struct commands[] = {
 	{ "format-patch", cmd_format_patch, RUN_SETUP },
 	{ "fsck", cmd_fsck, RUN_SETUP },
 	{ "fsck-objects", cmd_fsck, RUN_SETUP },
+	{ "fsmonitor--daemon", cmd_fsmonitor__daemon, RUN_SETUP },
 	{ "gc", cmd_gc, RUN_SETUP },
 	{ "get-tar-commit-id", cmd_get_tar_commit_id, NO_PARSEOPT },
 	{ "grep", cmd_grep, RUN_SETUP_GENTLY },
-- 
gitgitgadget

[PATCH v8 07/30] fsmonitor--daemon: implement 'stop' and 'status' commands

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Implement `stop` and `status` client commands to control and query the
status of a `fsmonitor--daemon` server process (and implicitly start a
server process if necessary).

Later commits will implement the actual server and monitor the file
system.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 builtin/fsmonitor--daemon.c | 51 +++++++++++++++++++++++++++++++++++++
 1 file changed, 51 insertions(+)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index f0498793379..5e3178b8bdd 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -7,10 +7,55 @@
 #include "khash.h"
 
 static const char * const builtin_fsmonitor__daemon_usage[] = {
+	N_("git fsmonitor--daemon stop"),
+	N_("git fsmonitor--daemon status"),
 	NULL
 };
 
 #ifdef HAVE_FSMONITOR_DAEMON_BACKEND
+/*
+ * Acting as a CLIENT.
+ *
+ * Send a "quit" command to the `git-fsmonitor--daemon` (if running)
+ * and wait for it to shutdown.
+ */
+static int do_as_client__send_stop(void)
+{
+	struct strbuf answer = STRBUF_INIT;
+	int ret;
+
+	ret = fsmonitor_ipc__send_command("quit", &answer);
+
+	/* The quit command does not return any response data. */
+	strbuf_release(&answer);
+
+	if (ret)
+		return ret;
+
+	trace2_region_enter("fsm_client", "polling-for-daemon-exit", NULL);
+	while (fsmonitor_ipc__get_state() == IPC_STATE__LISTENING)
+		sleep_millisec(50);
+	trace2_region_leave("fsm_client", "polling-for-daemon-exit", NULL);
+
+	return 0;
+}
+
+static int do_as_client__status(void)
+{
+	enum ipc_active_state state = fsmonitor_ipc__get_state();
+
+	switch (state) {
+	case IPC_STATE__LISTENING:
+		printf(_("fsmonitor-daemon is watching '%s'\n"),
+		       the_repository->worktree);
+		return 0;
+
+	default:
+		printf(_("fsmonitor-daemon is not watching '%s'\n"),
+		       the_repository->worktree);
+		return 1;
+	}
+}
 
 int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
 {
@@ -28,6 +73,12 @@ int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
 		usage_with_options(builtin_fsmonitor__daemon_usage, options);
 	subcmd = argv[0];
 
+	if (!strcmp(subcmd, "stop"))
+		return !!do_as_client__send_stop();
+
+	if (!strcmp(subcmd, "status"))
+		return !!do_as_client__status();
+
 	die(_("Unhandled subcommand '%s'"), subcmd);
 }
 
-- 
gitgitgadget

[PATCH v8 08/30] compat/fsmonitor/fsm-listen-win32: stub in backend for Windows

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Stub in empty filesystem listener backend for fsmonitor--daemon on Windows.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 Makefile                            | 13 ++++++++
 compat/fsmonitor/fsm-listen-win32.c | 21 +++++++++++++
 compat/fsmonitor/fsm-listen.h       | 49 +++++++++++++++++++++++++++++
 config.mak.uname                    | 10 ++++++
 contrib/buildsystems/CMakeLists.txt |  7 +++++
 repo-settings.c                     |  1 +
 6 files changed, 101 insertions(+)
 create mode 100644 compat/fsmonitor/fsm-listen-win32.c
 create mode 100644 compat/fsmonitor/fsm-listen.h

diff --git a/Makefile b/Makefile
index 5af1d5b112e..26567d4f772 100644
--- a/Makefile
+++ b/Makefile
@@ -470,6 +470,11 @@ all::
 # directory, and the JSON compilation database 'compile_commands.json' will be
 # created at the root of the repository.
 #
+# If your platform supports a built-in fsmonitor backend, set
+# FSMONITOR_DAEMON_BACKEND to the "<name>" of the corresponding
+# `compat/fsmonitor/fsm-listen-<name>.c` that implements the
+# `fsm_listen__*()` routines.
+#
 # Define DEVELOPER to enable more compiler warnings. Compiler version
 # and family are auto detected, but could be overridden by defining
 # COMPILER_FEATURES (see config.mak.dev). You can still set
@@ -1968,6 +1973,11 @@ ifdef NEED_ACCESS_ROOT_HANDLER
 	COMPAT_OBJS += compat/access.o
 endif
 
+ifdef FSMONITOR_DAEMON_BACKEND
+	COMPAT_CFLAGS += -DHAVE_FSMONITOR_DAEMON_BACKEND
+	COMPAT_OBJS += compat/fsmonitor/fsm-listen-$(FSMONITOR_DAEMON_BACKEND).o
+endif
+
 ifeq ($(TCLTK_PATH),)
 NO_TCLTK = NoThanks
 endif
@@ -2887,6 +2897,9 @@ GIT-BUILD-OPTIONS: FORCE
 	@echo DC_SHA1=\''$(subst ','\'',$(subst ','\'',$(DC_SHA1)))'\' >>$@+
 	@echo SANITIZE_LEAK=\''$(subst ','\'',$(subst ','\'',$(SANITIZE_LEAK)))'\' >>$@+
 	@echo X=\'$(X)\' >>$@+
+ifdef FSMONITOR_DAEMON_BACKEND
+	@echo FSMONITOR_DAEMON_BACKEND=\''$(subst ','\'',$(subst ','\'',$(FSMONITOR_DAEMON_BACKEND)))'\' >>$@+
+endif
 ifdef TEST_OUTPUT_DIRECTORY
 	@echo TEST_OUTPUT_DIRECTORY=\''$(subst ','\'',$(subst ','\'',$(TEST_OUTPUT_DIRECTORY)))'\' >>$@+
 endif
diff --git a/compat/fsmonitor/fsm-listen-win32.c b/compat/fsmonitor/fsm-listen-win32.c
new file mode 100644
index 00000000000..916cbea254f
--- /dev/null
+++ b/compat/fsmonitor/fsm-listen-win32.c
@@ -0,0 +1,21 @@
+#include "cache.h"
+#include "config.h"
+#include "fsmonitor.h"
+#include "fsm-listen.h"
+
+void fsm_listen__stop_async(struct fsmonitor_daemon_state *state)
+{
+}
+
+void fsm_listen__loop(struct fsmonitor_daemon_state *state)
+{
+}
+
+int fsm_listen__ctor(struct fsmonitor_daemon_state *state)
+{
+	return -1;
+}
+
+void fsm_listen__dtor(struct fsmonitor_daemon_state *state)
+{
+}
diff --git a/compat/fsmonitor/fsm-listen.h b/compat/fsmonitor/fsm-listen.h
new file mode 100644
index 00000000000..f0539349baf
--- /dev/null
+++ b/compat/fsmonitor/fsm-listen.h
@@ -0,0 +1,49 @@
+#ifndef FSM_LISTEN_H
+#define FSM_LISTEN_H
+
+/* This needs to be implemented by each backend */
+
+#ifdef HAVE_FSMONITOR_DAEMON_BACKEND
+
+struct fsmonitor_daemon_state;
+
+/*
+ * Initialize platform-specific data for the fsmonitor listener thread.
+ * This will be called from the main thread PRIOR to staring the
+ * fsmonitor_fs_listener thread.
+ *
+ * Returns 0 if successful.
+ * Returns -1 otherwise.
+ */
+int fsm_listen__ctor(struct fsmonitor_daemon_state *state);
+
+/*
+ * Cleanup platform-specific data for the fsmonitor listener thread.
+ * This will be called from the main thread AFTER joining the listener.
+ */
+void fsm_listen__dtor(struct fsmonitor_daemon_state *state);
+
+/*
+ * The main body of the platform-specific event loop to watch for
+ * filesystem events.  This will run in the fsmonitor_fs_listen thread.
+ *
+ * It should call `ipc_server_stop_async()` if the listener thread
+ * prematurely terminates (because of a filesystem error or if it
+ * detects that the .git directory has been deleted).  (It should NOT
+ * do so if the listener thread receives a normal shutdown signal from
+ * the IPC layer.)
+ *
+ * It should set `state->error_code` to -1 if the daemon should exit
+ * with an error.
+ */
+void fsm_listen__loop(struct fsmonitor_daemon_state *state);
+
+/*
+ * Gently request that the fsmonitor listener thread shutdown.
+ * It does not wait for it to stop.  The caller should do a JOIN
+ * to wait for it.
+ */
+void fsm_listen__stop_async(struct fsmonitor_daemon_state *state);
+
+#endif /* HAVE_FSMONITOR_DAEMON_BACKEND */
+#endif /* FSM_LISTEN_H */
diff --git a/config.mak.uname b/config.mak.uname
index 4352ea39e9b..26074f56bed 100644
--- a/config.mak.uname
+++ b/config.mak.uname
@@ -435,6 +435,11 @@ ifeq ($(uname_S),Windows)
 	# so we don't need this:
 	#
 	#   SNPRINTF_RETURNS_BOGUS = YesPlease
+
+	# The builtin FSMonitor requires Named Pipes and Threads on Windows.
+	# These are always available, so we do not have to conditionally
+	# support it.
+	FSMONITOR_DAEMON_BACKEND = win32
 	NO_SVN_TESTS = YesPlease
 	RUNTIME_PREFIX = YesPlease
 	HAVE_WPGMPTR = YesWeDo
@@ -619,6 +624,11 @@ ifeq ($(uname_S),MINGW)
 	NO_STRTOUMAX = YesPlease
 	NO_MKDTEMP = YesPlease
 	NO_SVN_TESTS = YesPlease
+
+	# The builtin FSMonitor requires Named Pipes and Threads on Windows.
+	# These are always available, so we do not have to conditionally
+	# support it.
+	FSMONITOR_DAEMON_BACKEND = win32
 	RUNTIME_PREFIX = YesPlease
 	HAVE_WPGMPTR = YesWeDo
 	NO_ST_BLOCKS_IN_STRUCT_STAT = YesPlease
diff --git a/contrib/buildsystems/CMakeLists.txt b/contrib/buildsystems/CMakeLists.txt
index e44232f85d3..0963629db7f 100644
--- a/contrib/buildsystems/CMakeLists.txt
+++ b/contrib/buildsystems/CMakeLists.txt
@@ -285,6 +285,13 @@ else()
 	endif()
 endif()
 
+if(SUPPORTS_SIMPLE_IPC)
+	if(CMAKE_SYSTEM_NAME STREQUAL "Windows")
+		add_compile_definitions(HAVE_FSMONITOR_DAEMON_BACKEND)
+		list(APPEND compat_SOURCES compat/fsmonitor/fsm-listen-win32.c)
+	endif()
+endif()
+
 set(EXE_EXTENSION ${CMAKE_EXECUTABLE_SUFFIX})
 
 #header checks
diff --git a/repo-settings.c b/repo-settings.c
index b4fbd16cdcc..2dfcb2b6542 100644
--- a/repo-settings.c
+++ b/repo-settings.c
@@ -2,6 +2,7 @@
 #include "config.h"
 #include "repository.h"
 #include "midx.h"
+#include "compat/fsmonitor/fsm-listen.h"
 
 static void repo_cfg_bool(struct repository *r, const char *key, int *dest,
 			  int def)
-- 
gitgitgadget

[PATCH v8 09/30] compat/fsmonitor/fsm-listen-darwin: stub in backend for Darwin

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Stub in empty implementation of fsmonitor--daemon
backend for Darwin (aka MacOS).

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 compat/fsmonitor/fsm-listen-darwin.c | 20 ++++++++++++++++++++
 config.mak.uname                     | 10 ++++++++++
 contrib/buildsystems/CMakeLists.txt  |  3 +++
 3 files changed, 33 insertions(+)
 create mode 100644 compat/fsmonitor/fsm-listen-darwin.c

diff --git a/compat/fsmonitor/fsm-listen-darwin.c b/compat/fsmonitor/fsm-listen-darwin.c
new file mode 100644
index 00000000000..c84e3344ab9
--- /dev/null
+++ b/compat/fsmonitor/fsm-listen-darwin.c
@@ -0,0 +1,20 @@
+#include "cache.h"
+#include "fsmonitor.h"
+#include "fsm-listen.h"
+
+int fsm_listen__ctor(struct fsmonitor_daemon_state *state)
+{
+	return -1;
+}
+
+void fsm_listen__dtor(struct fsmonitor_daemon_state *state)
+{
+}
+
+void fsm_listen__stop_async(struct fsmonitor_daemon_state *state)
+{
+}
+
+void fsm_listen__loop(struct fsmonitor_daemon_state *state)
+{
+}
diff --git a/config.mak.uname b/config.mak.uname
index 26074f56bed..501970902da 100644
--- a/config.mak.uname
+++ b/config.mak.uname
@@ -157,6 +157,16 @@ ifeq ($(uname_S),Darwin)
 			MSGFMT = /usr/local/opt/gettext/bin/msgfmt
 		endif
 	endif
+
+	# The builtin FSMonitor on MacOS builds upon Simple-IPC.  Both require
+	# Unix domain sockets and PThreads.
+	ifndef NO_PTHREADS
+	ifndef NO_UNIX_SOCKETS
+	FSMONITOR_DAEMON_BACKEND = darwin
+	endif
+	endif
+
+	BASIC_LDFLAGS += -framework CoreServices
 endif
 ifeq ($(uname_S),SunOS)
 	NEEDS_SOCKET = YesPlease
diff --git a/contrib/buildsystems/CMakeLists.txt b/contrib/buildsystems/CMakeLists.txt
index 0963629db7f..ee0d7257b77 100644
--- a/contrib/buildsystems/CMakeLists.txt
+++ b/contrib/buildsystems/CMakeLists.txt
@@ -289,6 +289,9 @@ if(SUPPORTS_SIMPLE_IPC)
 	if(CMAKE_SYSTEM_NAME STREQUAL "Windows")
 		add_compile_definitions(HAVE_FSMONITOR_DAEMON_BACKEND)
 		list(APPEND compat_SOURCES compat/fsmonitor/fsm-listen-win32.c)
+	elseif(CMAKE_SYSTEM_NAME STREQUAL "Darwin")
+		add_compile_definitions(HAVE_FSMONITOR_DAEMON_BACKEND)
+		list(APPEND compat_SOURCES compat/fsmonitor/fsm-listen-darwin.c)
 	endif()
 endif()
 
-- 
gitgitgadget

[PATCH v8 10/30] fsmonitor--daemon: implement 'run' command

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Implement `run` command to try to begin listening for file system events.

This version defines the thread structure with a single fsmonitor_fs_listen
thread to watch for file system events and a simple IPC thread pool to
watch for connection from Git clients over a well-known named pipe or
Unix domain socket.

This commit does not actually do anything yet because the platform
backends are still just stubs.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 builtin/fsmonitor--daemon.c | 228 +++++++++++++++++++++++++++++++++++-
 fsmonitor--daemon.h         |  34 ++++++
 2 files changed, 261 insertions(+), 1 deletion(-)
 create mode 100644 fsmonitor--daemon.h

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index 5e3178b8bdd..5591339399a 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -3,16 +3,52 @@
 #include "parse-options.h"
 #include "fsmonitor.h"
 #include "fsmonitor-ipc.h"
+#include "compat/fsmonitor/fsm-listen.h"
+#include "fsmonitor--daemon.h"
 #include "simple-ipc.h"
 #include "khash.h"
 
 static const char * const builtin_fsmonitor__daemon_usage[] = {
+	N_("git fsmonitor--daemon run [<options>]"),
 	N_("git fsmonitor--daemon stop"),
 	N_("git fsmonitor--daemon status"),
 	NULL
 };
 
 #ifdef HAVE_FSMONITOR_DAEMON_BACKEND
+/*
+ * Global state loaded from config.
+ */
+#define FSMONITOR__IPC_THREADS "fsmonitor.ipcthreads"
+static int fsmonitor__ipc_threads = 8;
+
+#define FSMONITOR__ANNOUNCE_STARTUP "fsmonitor.announcestartup"
+static int fsmonitor__announce_startup = 0;
+
+static int fsmonitor_config(const char *var, const char *value, void *cb)
+{
+	if (!strcmp(var, FSMONITOR__IPC_THREADS)) {
+		int i = git_config_int(var, value);
+		if (i < 1)
+			return error(_("value of '%s' out of range: %d"),
+				     FSMONITOR__IPC_THREADS, i);
+		fsmonitor__ipc_threads = i;
+		return 0;
+	}
+
+	if (!strcmp(var, FSMONITOR__ANNOUNCE_STARTUP)) {
+		int is_bool;
+		int i = git_config_bool_or_int(var, value, &is_bool);
+		if (i < 0)
+			return error(_("value of '%s' not bool or int: %d"),
+				     var, i);
+		fsmonitor__announce_startup = i;
+		return 0;
+	}
+
+	return git_default_config(var, value, cb);
+}
+
 /*
  * Acting as a CLIENT.
  *
@@ -57,15 +93,198 @@ static int do_as_client__status(void)
 	}
 }
 
+static ipc_server_application_cb handle_client;
+
+static int handle_client(void *data,
+			 const char *command, size_t command_len,
+			 ipc_server_reply_cb *reply,
+			 struct ipc_server_reply_data *reply_data)
+{
+	/* struct fsmonitor_daemon_state *state = data; */
+	int result;
+
+	/*
+	 * The Simple IPC API now supports {char*, len} arguments, but
+	 * FSMonitor always uses proper null-terminated strings, so
+	 * we can ignore the command_len argument.  (Trust, but verify.)
+	 */
+	if (command_len != strlen(command))
+		BUG("FSMonitor assumes text messages");
+
+	trace2_region_enter("fsmonitor", "handle_client", the_repository);
+	trace2_data_string("fsmonitor", the_repository, "request", command);
+
+	result = 0; /* TODO Do something here. */
+
+	trace2_region_leave("fsmonitor", "handle_client", the_repository);
+
+	return result;
+}
+
+static void *fsm_listen__thread_proc(void *_state)
+{
+	struct fsmonitor_daemon_state *state = _state;
+
+	trace2_thread_start("fsm-listen");
+
+	trace_printf_key(&trace_fsmonitor, "Watching: worktree '%s'",
+			 state->path_worktree_watch.buf);
+	if (state->nr_paths_watching > 1)
+		trace_printf_key(&trace_fsmonitor, "Watching: gitdir '%s'",
+				 state->path_gitdir_watch.buf);
+
+	fsm_listen__loop(state);
+
+	trace2_thread_exit();
+	return NULL;
+}
+
+static int fsmonitor_run_daemon_1(struct fsmonitor_daemon_state *state)
+{
+	struct ipc_server_opts ipc_opts = {
+		.nr_threads = fsmonitor__ipc_threads,
+
+		/*
+		 * We know that there are no other active threads yet,
+		 * so we can let the IPC layer temporarily chdir() if
+		 * it needs to when creating the server side of the
+		 * Unix domain socket.
+		 */
+		.uds_disallow_chdir = 0
+	};
+
+	/*
+	 * Start the IPC thread pool before the we've started the file
+	 * system event listener thread so that we have the IPC handle
+	 * before we need it.
+	 */
+	if (ipc_server_run_async(&state->ipc_server_data,
+				 fsmonitor_ipc__get_path(), &ipc_opts,
+				 handle_client, state))
+		return error_errno(
+			_("could not start IPC thread pool on '%s'"),
+			fsmonitor_ipc__get_path());
+
+	/*
+	 * Start the fsmonitor listener thread to collect filesystem
+	 * events.
+	 */
+	if (pthread_create(&state->listener_thread, NULL,
+			   fsm_listen__thread_proc, state) < 0) {
+		ipc_server_stop_async(state->ipc_server_data);
+		ipc_server_await(state->ipc_server_data);
+
+		return error(_("could not start fsmonitor listener thread"));
+	}
+
+	/*
+	 * The daemon is now fully functional in background threads.
+	 * Wait for the IPC thread pool to shutdown (whether by client
+	 * request or from filesystem activity).
+	 */
+	ipc_server_await(state->ipc_server_data);
+
+	/*
+	 * The fsmonitor listener thread may have received a shutdown
+	 * event from the IPC thread pool, but it doesn't hurt to tell
+	 * it again.  And wait for it to shutdown.
+	 */
+	fsm_listen__stop_async(state);
+	pthread_join(state->listener_thread, NULL);
+
+	return state->error_code;
+}
+
+static int fsmonitor_run_daemon(void)
+{
+	struct fsmonitor_daemon_state state;
+	int err;
+
+	memset(&state, 0, sizeof(state));
+
+	pthread_mutex_init(&state.main_lock, NULL);
+	state.error_code = 0;
+	state.current_token_data = NULL;
+
+	/* Prepare to (recursively) watch the <worktree-root> directory. */
+	strbuf_init(&state.path_worktree_watch, 0);
+	strbuf_addstr(&state.path_worktree_watch, absolute_path(get_git_work_tree()));
+	state.nr_paths_watching = 1;
+
+	/*
+	 * We create and delete cookie files somewhere inside the .git
+	 * directory to help us keep sync with the file system.  If
+	 * ".git" is not a directory, then <gitdir> is not inside the
+	 * cone of <worktree-root>, so set up a second watch to watch
+	 * the <gitdir> so that we get events for the cookie files.
+	 */
+	strbuf_init(&state.path_gitdir_watch, 0);
+	strbuf_addbuf(&state.path_gitdir_watch, &state.path_worktree_watch);
+	strbuf_addstr(&state.path_gitdir_watch, "/.git");
+	if (!is_directory(state.path_gitdir_watch.buf)) {
+		strbuf_reset(&state.path_gitdir_watch);
+		strbuf_addstr(&state.path_gitdir_watch, absolute_path(get_git_dir()));
+		state.nr_paths_watching = 2;
+	}
+
+	/*
+	 * Confirm that we can create platform-specific resources for the
+	 * filesystem listener before we bother starting all the threads.
+	 */
+	if (fsm_listen__ctor(&state)) {
+		err = error(_("could not initialize listener thread"));
+		goto done;
+	}
+
+	err = fsmonitor_run_daemon_1(&state);
+
+done:
+	pthread_mutex_destroy(&state.main_lock);
+	fsm_listen__dtor(&state);
+
+	ipc_server_free(state.ipc_server_data);
+
+	strbuf_release(&state.path_worktree_watch);
+	strbuf_release(&state.path_gitdir_watch);
+
+	return err;
+}
+
+static int try_to_run_foreground_daemon(void)
+{
+	/*
+	 * Technically, we don't need to probe for an existing daemon
+	 * process, since we could just call `fsmonitor_run_daemon()`
+	 * and let it fail if the pipe/socket is busy.
+	 *
+	 * However, this method gives us a nicer error message for a
+	 * common error case.
+	 */
+	if (fsmonitor_ipc__get_state() == IPC_STATE__LISTENING)
+		die(_("fsmonitor--daemon is already running '%s'"),
+		    the_repository->worktree);
+
+	if (fsmonitor__announce_startup) {
+		fprintf(stderr, _("running fsmonitor-daemon in '%s'\n"),
+			the_repository->worktree);
+		fflush(stderr);
+	}
+
+	return !!fsmonitor_run_daemon();
+}
+
 int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
 {
 	const char *subcmd;
 
 	struct option options[] = {
+		OPT_INTEGER(0, "ipc-threads",
+			    &fsmonitor__ipc_threads,
+			    N_("use <n> ipc worker threads")),
 		OPT_END()
 	};
 
-	git_config(git_default_config, NULL);
+	git_config(fsmonitor_config, NULL);
 
 	argc = parse_options(argc, argv, prefix, options,
 			     builtin_fsmonitor__daemon_usage, 0);
@@ -73,6 +292,13 @@ int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
 		usage_with_options(builtin_fsmonitor__daemon_usage, options);
 	subcmd = argv[0];
 
+	if (fsmonitor__ipc_threads < 1)
+		die(_("invalid 'ipc-threads' value (%d)"),
+		    fsmonitor__ipc_threads);
+
+	if (!strcmp(subcmd, "run"))
+		return !!try_to_run_foreground_daemon();
+
 	if (!strcmp(subcmd, "stop"))
 		return !!do_as_client__send_stop();
 
diff --git a/fsmonitor--daemon.h b/fsmonitor--daemon.h
new file mode 100644
index 00000000000..3009c1a83de
--- /dev/null
+++ b/fsmonitor--daemon.h
@@ -0,0 +1,34 @@
+#ifndef FSMONITOR_DAEMON_H
+#define FSMONITOR_DAEMON_H
+
+#ifdef HAVE_FSMONITOR_DAEMON_BACKEND
+
+#include "cache.h"
+#include "dir.h"
+#include "run-command.h"
+#include "simple-ipc.h"
+#include "thread-utils.h"
+
+struct fsmonitor_batch;
+struct fsmonitor_token_data;
+
+struct fsmonitor_daemon_backend_data; /* opaque platform-specific data */
+
+struct fsmonitor_daemon_state {
+	pthread_t listener_thread;
+	pthread_mutex_t main_lock;
+
+	struct strbuf path_worktree_watch;
+	struct strbuf path_gitdir_watch;
+	int nr_paths_watching;
+
+	struct fsmonitor_token_data *current_token_data;
+
+	int error_code;
+	struct fsmonitor_daemon_backend_data *backend_data;
+
+	struct ipc_server_data *ipc_server_data;
+};
+
+#endif /* HAVE_FSMONITOR_DAEMON_BACKEND */
+#endif /* FSMONITOR_DAEMON_H */
-- 
gitgitgadget

[PATCH v8 11/30] fsmonitor--daemon: implement 'start' command

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Implement 'git fsmonitor--daemon start' command.  This command starts
an instance of 'git fsmonitor--daemon run' in the background using
the new 'start_bg_command()' function.

We avoid the fork-and-call technique on Unix systems in favor of a
fork-and-exec technique.  This gives us more uniform Trace2 child-*
events.  It also makes our usage more consistent with Windows usage.

On Windows, teach 'git fsmonitor--daemon run' to optionally call
'FreeConsole()' to release handles to the inherited Win32 console
(despite being passed invalid handles for stdin/out/err).  Without
this, command prompts and powershell terminal windows could hang
in "exit" until the last background child process exited or released
their Win32 console handle.  (This was not seen with git-bash shells
because they don't have a Win32 console attached to them.)

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 builtin/fsmonitor--daemon.c | 109 +++++++++++++++++++++++++++++++++++-
 1 file changed, 107 insertions(+), 2 deletions(-)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index 5591339399a..69dd39121a3 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -9,6 +9,7 @@
 #include "khash.h"
 
 static const char * const builtin_fsmonitor__daemon_usage[] = {
+	N_("git fsmonitor--daemon start [<options>]"),
 	N_("git fsmonitor--daemon run [<options>]"),
 	N_("git fsmonitor--daemon stop"),
 	N_("git fsmonitor--daemon status"),
@@ -22,6 +23,9 @@ static const char * const builtin_fsmonitor__daemon_usage[] = {
 #define FSMONITOR__IPC_THREADS "fsmonitor.ipcthreads"
 static int fsmonitor__ipc_threads = 8;
 
+#define FSMONITOR__START_TIMEOUT "fsmonitor.starttimeout"
+static int fsmonitor__start_timeout_sec = 60;
+
 #define FSMONITOR__ANNOUNCE_STARTUP "fsmonitor.announcestartup"
 static int fsmonitor__announce_startup = 0;
 
@@ -36,6 +40,15 @@ static int fsmonitor_config(const char *var, const char *value, void *cb)
 		return 0;
 	}
 
+	if (!strcmp(var, FSMONITOR__START_TIMEOUT)) {
+		int i = git_config_int(var, value);
+		if (i < 0)
+			return error(_("value of '%s' out of range: %d"),
+				     FSMONITOR__START_TIMEOUT, i);
+		fsmonitor__start_timeout_sec = i;
+		return 0;
+	}
+
 	if (!strcmp(var, FSMONITOR__ANNOUNCE_STARTUP)) {
 		int is_bool;
 		int i = git_config_bool_or_int(var, value, &is_bool);
@@ -250,7 +263,7 @@ done:
 	return err;
 }
 
-static int try_to_run_foreground_daemon(void)
+static int try_to_run_foreground_daemon(int detach_console)
 {
 	/*
 	 * Technically, we don't need to probe for an existing daemon
@@ -270,17 +283,106 @@ static int try_to_run_foreground_daemon(void)
 		fflush(stderr);
 	}
 
+#ifdef GIT_WINDOWS_NATIVE
+	if (detach_console)
+		FreeConsole();
+#endif
+
 	return !!fsmonitor_run_daemon();
 }
 
+static start_bg_wait_cb bg_wait_cb;
+
+static int bg_wait_cb(const struct child_process *cp, void *cb_data)
+{
+	enum ipc_active_state s = fsmonitor_ipc__get_state();
+
+	switch (s) {
+	case IPC_STATE__LISTENING:
+		/* child is "ready" */
+		return 0;
+
+	case IPC_STATE__NOT_LISTENING:
+	case IPC_STATE__PATH_NOT_FOUND:
+		/* give child more time */
+		return 1;
+
+	default:
+	case IPC_STATE__INVALID_PATH:
+	case IPC_STATE__OTHER_ERROR:
+		/* all the time in world won't help */
+		return -1;
+	}
+}
+
+static int try_to_start_background_daemon(void)
+{
+	struct child_process cp = CHILD_PROCESS_INIT;
+	enum start_bg_result sbgr;
+
+	/*
+	 * Before we try to create a background daemon process, see
+	 * if a daemon process is already listening.  This makes it
+	 * easier for us to report an already-listening error to the
+	 * console, since our spawn/daemon can only report the success
+	 * of creating the background process (and not whether it
+	 * immediately exited).
+	 */
+	if (fsmonitor_ipc__get_state() == IPC_STATE__LISTENING)
+		die(_("fsmonitor--daemon is already running '%s'"),
+		    the_repository->worktree);
+
+	if (fsmonitor__announce_startup) {
+		fprintf(stderr, _("starting fsmonitor-daemon in '%s'\n"),
+			the_repository->worktree);
+		fflush(stderr);
+	}
+
+	cp.git_cmd = 1;
+
+	strvec_push(&cp.args, "fsmonitor--daemon");
+	strvec_push(&cp.args, "run");
+	strvec_push(&cp.args, "--detach");
+	strvec_pushf(&cp.args, "--ipc-threads=%d", fsmonitor__ipc_threads);
+
+	cp.no_stdin = 1;
+	cp.no_stdout = 1;
+	cp.no_stderr = 1;
+
+	sbgr = start_bg_command(&cp, bg_wait_cb, NULL,
+				fsmonitor__start_timeout_sec);
+
+	switch (sbgr) {
+	case SBGR_READY:
+		return 0;
+
+	default:
+	case SBGR_ERROR:
+	case SBGR_CB_ERROR:
+		return error(_("daemon failed to start"));
+
+	case SBGR_TIMEOUT:
+		return error(_("daemon not online yet"));
+
+	case SBGR_DIED:
+		return error(_("daemon terminated"));
+	}
+}
+
 int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
 {
 	const char *subcmd;
+	int detach_console = 0;
 
 	struct option options[] = {
+		OPT_BOOL(0, "detach", &detach_console, N_("detach from console")),
 		OPT_INTEGER(0, "ipc-threads",
 			    &fsmonitor__ipc_threads,
 			    N_("use <n> ipc worker threads")),
+		OPT_INTEGER(0, "start-timeout",
+			    &fsmonitor__start_timeout_sec,
+			    N_("max seconds to wait for background daemon startup")),
+
 		OPT_END()
 	};
 
@@ -296,8 +398,11 @@ int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
 		die(_("invalid 'ipc-threads' value (%d)"),
 		    fsmonitor__ipc_threads);
 
+	if (!strcmp(subcmd, "start"))
+		return !!try_to_start_background_daemon();
+
 	if (!strcmp(subcmd, "run"))
-		return !!try_to_run_foreground_daemon();
+		return !!try_to_run_foreground_daemon(detach_console);
 
 	if (!strcmp(subcmd, "stop"))
 		return !!do_as_client__send_stop();
-- 
gitgitgadget

[PATCH v8 12/30] fsmonitor--daemon: add pathname classification

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach fsmonitor--daemon to classify relative and absolute
pathnames and decide how they should be handled.  This will
be used by the platform-specific backend to respond to each
filesystem event.

When we register for filesystem notifications on a directory,
we get events for everything (recursively) in the directory.
We want to report to clients changes to tracked and untracked
paths within the working directory proper.  We do not want to
report changes within the .git directory, for example.

This classification will be used in a later commit by the
different backends to classify paths as events are received.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 builtin/fsmonitor--daemon.c | 81 ++++++++++++++++++++++++++++++++++
 fsmonitor--daemon.h         | 87 +++++++++++++++++++++++++++++++++++++
 2 files changed, 168 insertions(+)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index 69dd39121a3..1ce00b7c150 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -134,6 +134,87 @@ static int handle_client(void *data,
 	return result;
 }
 
+#define FSMONITOR_COOKIE_PREFIX ".fsmonitor-daemon-"
+
+enum fsmonitor_path_type fsmonitor_classify_path_workdir_relative(
+	const char *rel)
+{
+	if (fspathncmp(rel, ".git", 4))
+		return IS_WORKDIR_PATH;
+	rel += 4;
+
+	if (!*rel)
+		return IS_DOT_GIT;
+	if (*rel != '/')
+		return IS_WORKDIR_PATH; /* e.g. .gitignore */
+	rel++;
+
+	if (!fspathncmp(rel, FSMONITOR_COOKIE_PREFIX,
+			strlen(FSMONITOR_COOKIE_PREFIX)))
+		return IS_INSIDE_DOT_GIT_WITH_COOKIE_PREFIX;
+
+	return IS_INSIDE_DOT_GIT;
+}
+
+enum fsmonitor_path_type fsmonitor_classify_path_gitdir_relative(
+	const char *rel)
+{
+	if (!fspathncmp(rel, FSMONITOR_COOKIE_PREFIX,
+			strlen(FSMONITOR_COOKIE_PREFIX)))
+		return IS_INSIDE_GITDIR_WITH_COOKIE_PREFIX;
+
+	return IS_INSIDE_GITDIR;
+}
+
+static enum fsmonitor_path_type try_classify_workdir_abs_path(
+	struct fsmonitor_daemon_state *state,
+	const char *path)
+{
+	const char *rel;
+
+	if (fspathncmp(path, state->path_worktree_watch.buf,
+		       state->path_worktree_watch.len))
+		return IS_OUTSIDE_CONE;
+
+	rel = path + state->path_worktree_watch.len;
+
+	if (!*rel)
+		return IS_WORKDIR_PATH; /* it is the root dir exactly */
+	if (*rel != '/')
+		return IS_OUTSIDE_CONE;
+	rel++;
+
+	return fsmonitor_classify_path_workdir_relative(rel);
+}
+
+enum fsmonitor_path_type fsmonitor_classify_path_absolute(
+	struct fsmonitor_daemon_state *state,
+	const char *path)
+{
+	const char *rel;
+	enum fsmonitor_path_type t;
+
+	t = try_classify_workdir_abs_path(state, path);
+	if (state->nr_paths_watching == 1)
+		return t;
+	if (t != IS_OUTSIDE_CONE)
+		return t;
+
+	if (fspathncmp(path, state->path_gitdir_watch.buf,
+		       state->path_gitdir_watch.len))
+		return IS_OUTSIDE_CONE;
+
+	rel = path + state->path_gitdir_watch.len;
+
+	if (!*rel)
+		return IS_GITDIR; /* it is the <gitdir> exactly */
+	if (*rel != '/')
+		return IS_OUTSIDE_CONE;
+	rel++;
+
+	return fsmonitor_classify_path_gitdir_relative(rel);
+}
+
 static void *fsm_listen__thread_proc(void *_state)
 {
 	struct fsmonitor_daemon_state *state = _state;
diff --git a/fsmonitor--daemon.h b/fsmonitor--daemon.h
index 3009c1a83de..8c3a71a48bd 100644
--- a/fsmonitor--daemon.h
+++ b/fsmonitor--daemon.h
@@ -30,5 +30,92 @@ struct fsmonitor_daemon_state {
 	struct ipc_server_data *ipc_server_data;
 };
 
+/*
+ * Pathname classifications.
+ *
+ * The daemon classifies the pathnames that it receives from file
+ * system notification events into the following categories and uses
+ * that to decide whether clients are told about them.  (And to watch
+ * for file system synchronization events.)
+ *
+ * The daemon only collects and reports on the set of modified paths
+ * within the working directory (proper).
+ *
+ * The client should only care about paths within the working
+ * directory proper (inside the working directory and not ".git" nor
+ * inside of ".git/").  That is, the client has read the index and is
+ * asking for a list of any paths in the working directory that have
+ * been modified since the last token.  The client does not care about
+ * file system changes within the ".git/" directory (such as new loose
+ * objects or packfiles).  So the client will only receive paths that
+ * are classified as IS_WORKDIR_PATH.
+ *
+ * Note that ".git" is usually a directory and is therefore inside
+ * the cone of the FS watch that we have on the working directory root,
+ * so we will also get FS events for disk activity on and within ".git/"
+ * that we need to respond to or filter from the client.
+ *
+ * But Git also allows ".git" to be a *file* that points to a GITDIR
+ * outside of the working directory.  When this happens, we need to
+ * create FS watches on both the working directory root *and* on the
+ * (external) GITDIR root.  (The latter is required because we put
+ * cookie files inside it and use them to sync with the FS event
+ * stream.)
+ *
+ * Note that in the context of this discussion, I'm using "GITDIR"
+ * to only mean an external GITDIR referenced by a ".git" file.
+ *
+ * The platform FS event backends will receive watch-specific
+ * relative paths (except for those OS's that always emit absolute
+ * paths).  We use the following enum and routines to classify each
+ * path so that we know how to handle it.  There is a slight asymmetry
+ * here because ".git/" is inside the working directory and the
+ * (external) GITDIR is not, and therefore how we handle events may
+ * vary slightly, so I have different enums for "IS...DOT_GIT..." and
+ * "IS...GITDIR...".
+ *
+ * The daemon uses the IS_DOT_GIT and IS_GITDIR internally to mean the
+ * exact ".git" file/directory or GITDIR directory.  If the daemon
+ * receives a delete event for either of these paths, it will
+ * automatically shutdown, for example.
+ *
+ * Note that the daemon DOES NOT explicitly watch nor special case the
+ * index.  The daemon does not read the index nor have any internal
+ * index-relative state, so there are no "IS...INDEX..." enum values.
+ */
+enum fsmonitor_path_type {
+	IS_WORKDIR_PATH = 0,
+
+	IS_DOT_GIT,
+	IS_INSIDE_DOT_GIT,
+	IS_INSIDE_DOT_GIT_WITH_COOKIE_PREFIX,
+
+	IS_GITDIR,
+	IS_INSIDE_GITDIR,
+	IS_INSIDE_GITDIR_WITH_COOKIE_PREFIX,
+
+	IS_OUTSIDE_CONE,
+};
+
+/*
+ * Classify a pathname relative to the root of the working directory.
+ */
+enum fsmonitor_path_type fsmonitor_classify_path_workdir_relative(
+	const char *relative_path);
+
+/*
+ * Classify a pathname relative to a <gitdir> that is external to the
+ * worktree directory.
+ */
+enum fsmonitor_path_type fsmonitor_classify_path_gitdir_relative(
+	const char *relative_path);
+
+/*
+ * Classify an absolute pathname received from a filesystem event.
+ */
+enum fsmonitor_path_type fsmonitor_classify_path_absolute(
+	struct fsmonitor_daemon_state *state,
+	const char *path);
+
 #endif /* HAVE_FSMONITOR_DAEMON_BACKEND */
 #endif /* FSMONITOR_DAEMON_H */
-- 
gitgitgadget

[PATCH v8 13/30] fsmonitor--daemon: define token-ids

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach fsmonitor--daemon to create token-ids and define the
overall token naming scheme.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 builtin/fsmonitor--daemon.c | 116 +++++++++++++++++++++++++++++++++++-
 1 file changed, 115 insertions(+), 1 deletion(-)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index 1ce00b7c150..1c7c156288d 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -106,6 +106,120 @@ static int do_as_client__status(void)
 	}
 }
 
+/*
+ * Requests to and from a FSMonitor Protocol V2 provider use an opaque
+ * "token" as a virtual timestamp.  Clients can request a summary of all
+ * created/deleted/modified files relative to a token.  In the response,
+ * clients receive a new token for the next (relative) request.
+ *
+ *
+ * Token Format
+ * ============
+ *
+ * The contents of the token are private and provider-specific.
+ *
+ * For the built-in fsmonitor--daemon, we define a token as follows:
+ *
+ *     "builtin" ":" <token_id> ":" <sequence_nr>
+ *
+ * The "builtin" prefix is used as a namespace to avoid conflicts
+ * with other providers (such as Watchman).
+ *
+ * The <token_id> is an arbitrary OPAQUE string, such as a GUID,
+ * UUID, or {timestamp,pid}.  It is used to group all filesystem
+ * events that happened while the daemon was monitoring (and in-sync
+ * with the filesystem).
+ *
+ *     Unlike FSMonitor Protocol V1, it is not defined as a timestamp
+ *     and does not define less-than/greater-than relationships.
+ *     (There are too many race conditions to rely on file system
+ *     event timestamps.)
+ *
+ * The <sequence_nr> is a simple integer incremented whenever the
+ * daemon needs to make its state public.  For example, if 1000 file
+ * system events come in, but no clients have requested the data,
+ * the daemon can continue to accumulate file changes in the same
+ * bin and does not need to advance the sequence number.  However,
+ * as soon as a client does arrive, the daemon needs to start a new
+ * bin and increment the sequence number.
+ *
+ *     The sequence number serves as the boundary between 2 sets
+ *     of bins -- the older ones that the client has already seen
+ *     and the newer ones that it hasn't.
+ *
+ * When a new <token_id> is created, the <sequence_nr> is reset to
+ * zero.
+ *
+ *
+ * About Token Ids
+ * ===============
+ *
+ * A new token_id is created:
+ *
+ * [1] each time the daemon is started.
+ *
+ * [2] any time that the daemon must re-sync with the filesystem
+ *     (such as when the kernel drops or we miss events on a very
+ *     active volume).
+ *
+ * [3] in response to a client "flush" command (for dropped event
+ *     testing).
+ *
+ * When a new token_id is created, the daemon is free to discard all
+ * cached filesystem events associated with any previous token_ids.
+ * Events associated with a non-current token_id will never be sent
+ * to a client.  A token_id change implicitly means that the daemon
+ * has gap in its event history.
+ *
+ * Therefore, clients that present a token with a stale (non-current)
+ * token_id will always be given a trivial response.
+ */
+struct fsmonitor_token_data {
+	struct strbuf token_id;
+	struct fsmonitor_batch *batch_head;
+	struct fsmonitor_batch *batch_tail;
+	uint64_t client_ref_count;
+};
+
+static struct fsmonitor_token_data *fsmonitor_new_token_data(void)
+{
+	static int test_env_value = -1;
+	static uint64_t flush_count = 0;
+	struct fsmonitor_token_data *token;
+
+	CALLOC_ARRAY(token, 1);
+
+	strbuf_init(&token->token_id, 0);
+	token->batch_head = NULL;
+	token->batch_tail = NULL;
+	token->client_ref_count = 0;
+
+	if (test_env_value < 0)
+		test_env_value = git_env_bool("GIT_TEST_FSMONITOR_TOKEN", 0);
+
+	if (!test_env_value) {
+		struct timeval tv;
+		struct tm tm;
+		time_t secs;
+
+		gettimeofday(&tv, NULL);
+		secs = tv.tv_sec;
+		gmtime_r(&secs, &tm);
+
+		strbuf_addf(&token->token_id,
+			    "%"PRIu64".%d.%4d%02d%02dT%02d%02d%02d.%06ldZ",
+			    flush_count++,
+			    getpid(),
+			    tm.tm_year + 1900, tm.tm_mon + 1, tm.tm_mday,
+			    tm.tm_hour, tm.tm_min, tm.tm_sec,
+			    (long)tv.tv_usec);
+	} else {
+		strbuf_addf(&token->token_id, "test_%08x", test_env_value++);
+	}
+
+	return token;
+}
+
 static ipc_server_application_cb handle_client;
 
 static int handle_client(void *data,
@@ -298,7 +412,7 @@ static int fsmonitor_run_daemon(void)
 
 	pthread_mutex_init(&state.main_lock, NULL);
 	state.error_code = 0;
-	state.current_token_data = NULL;
+	state.current_token_data = fsmonitor_new_token_data();
 
 	/* Prepare to (recursively) watch the <worktree-root> directory. */
 	strbuf_init(&state.path_worktree_watch, 0);
-- 
gitgitgadget

[PATCH v8 14/30] fsmonitor--daemon: create token-based changed path cache

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach fsmonitor--daemon to build a list of changed paths and associate
them with a token-id.  This will be used by the platform-specific
backends to accumulate changed paths in response to filesystem events.

The platform-specific file system listener thread receives file system
events containing one or more changed pathnames (with whatever
bucketing or grouping that is convenient for the file system).  These
paths are accumulated (without locking) by the file system layer into
a `fsmonitor_batch`.

When the file system layer has drained the kernel event queue, it will
"publish" them to our token queue and make them visible to concurrent
client worker threads.  The token layer is free to combine and/or de-dup
paths within these batches for efficient presentation to clients.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 builtin/fsmonitor--daemon.c | 230 +++++++++++++++++++++++++++++++++++-
 fsmonitor--daemon.h         |  40 +++++++
 2 files changed, 268 insertions(+), 2 deletions(-)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index 1c7c156288d..69312119b07 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -181,17 +181,27 @@ struct fsmonitor_token_data {
 	uint64_t client_ref_count;
 };
 
+struct fsmonitor_batch {
+	struct fsmonitor_batch *next;
+	uint64_t batch_seq_nr;
+	const char **interned_paths;
+	size_t nr, alloc;
+	time_t pinned_time;
+};
+
 static struct fsmonitor_token_data *fsmonitor_new_token_data(void)
 {
 	static int test_env_value = -1;
 	static uint64_t flush_count = 0;
 	struct fsmonitor_token_data *token;
+	struct fsmonitor_batch *batch;
 
 	CALLOC_ARRAY(token, 1);
+	batch = fsmonitor_batch__new();
 
 	strbuf_init(&token->token_id, 0);
-	token->batch_head = NULL;
-	token->batch_tail = NULL;
+	token->batch_head = batch;
+	token->batch_tail = batch;
 	token->client_ref_count = 0;
 
 	if (test_env_value < 0)
@@ -217,9 +227,143 @@ static struct fsmonitor_token_data *fsmonitor_new_token_data(void)
 		strbuf_addf(&token->token_id, "test_%08x", test_env_value++);
 	}
 
+	/*
+	 * We created a new <token_id> and are starting a new series
+	 * of tokens with a zero <seq_nr>.
+	 *
+	 * Since clients cannot guess our new (non test) <token_id>
+	 * they will always receive a trivial response (because of the
+	 * mismatch on the <token_id>).  The trivial response will
+	 * tell them our new <token_id> so that subsequent requests
+	 * will be relative to our new series.  (And when sending that
+	 * response, we pin the current head of the batch list.)
+	 *
+	 * Even if the client correctly guesses the <token_id>, their
+	 * request of "builtin:<token_id>:0" asks for all changes MORE
+	 * RECENT than batch/bin 0.
+	 *
+	 * This implies that it is a waste to accumulate paths in the
+	 * initial batch/bin (because they will never be transmitted).
+	 *
+	 * So the daemon could be running for days and watching the
+	 * file system, but doesn't need to actually accumulate any
+	 * paths UNTIL we need to set a reference point for a later
+	 * relative request.
+	 *
+	 * However, it is very useful for testing to always have a
+	 * reference point set.  Pin batch 0 to force early file system
+	 * events to accumulate.
+	 */
+	if (test_env_value)
+		batch->pinned_time = time(NULL);
+
 	return token;
 }
 
+struct fsmonitor_batch *fsmonitor_batch__new(void)
+{
+	struct fsmonitor_batch *batch;
+
+	CALLOC_ARRAY(batch, 1);
+
+	return batch;
+}
+
+void fsmonitor_batch__free_list(struct fsmonitor_batch *batch)
+{
+	while (batch) {
+		struct fsmonitor_batch *next = batch->next;
+
+		/*
+		 * The actual strings within the array of this batch
+		 * are interned, so we don't own them.  We only own
+		 * the array.
+		 */
+		free(batch->interned_paths);
+		free(batch);
+
+		batch = next;
+	}
+}
+
+void fsmonitor_batch__add_path(struct fsmonitor_batch *batch,
+			       const char *path)
+{
+	const char *interned_path = strintern(path);
+
+	trace_printf_key(&trace_fsmonitor, "event: %s", interned_path);
+
+	ALLOC_GROW(batch->interned_paths, batch->nr + 1, batch->alloc);
+	batch->interned_paths[batch->nr++] = interned_path;
+}
+
+static void fsmonitor_batch__combine(struct fsmonitor_batch *batch_dest,
+				     const struct fsmonitor_batch *batch_src)
+{
+	size_t k;
+
+	ALLOC_GROW(batch_dest->interned_paths,
+		   batch_dest->nr + batch_src->nr + 1,
+		   batch_dest->alloc);
+
+	for (k = 0; k < batch_src->nr; k++)
+		batch_dest->interned_paths[batch_dest->nr++] =
+			batch_src->interned_paths[k];
+}
+
+static void fsmonitor_free_token_data(struct fsmonitor_token_data *token)
+{
+	if (!token)
+		return;
+
+	assert(token->client_ref_count == 0);
+
+	strbuf_release(&token->token_id);
+
+	fsmonitor_batch__free_list(token->batch_head);
+
+	free(token);
+}
+
+/*
+ * Flush all of our cached data about the filesystem.  Call this if we
+ * lose sync with the filesystem and miss some notification events.
+ *
+ * [1] If we are missing events, then we no longer have a complete
+ *     history of the directory (relative to our current start token).
+ *     We should create a new token and start fresh (as if we just
+ *     booted up).
+ *
+ * If there are no concurrent threads reading the current token data
+ * series, we can free it now.  Otherwise, let the last reader free
+ * it.
+ *
+ * Either way, the old token data series is no longer associated with
+ * our state data.
+ */
+static void with_lock__do_force_resync(struct fsmonitor_daemon_state *state)
+{
+	/* assert current thread holding state->main_lock */
+
+	struct fsmonitor_token_data *free_me = NULL;
+	struct fsmonitor_token_data *new_one = NULL;
+
+	new_one = fsmonitor_new_token_data();
+
+	if (state->current_token_data->client_ref_count == 0)
+		free_me = state->current_token_data;
+	state->current_token_data = new_one;
+
+	fsmonitor_free_token_data(free_me);
+}
+
+void fsmonitor_force_resync(struct fsmonitor_daemon_state *state)
+{
+	pthread_mutex_lock(&state->main_lock);
+	with_lock__do_force_resync(state);
+	pthread_mutex_unlock(&state->main_lock);
+}
+
 static ipc_server_application_cb handle_client;
 
 static int handle_client(void *data,
@@ -329,6 +473,81 @@ enum fsmonitor_path_type fsmonitor_classify_path_absolute(
 	return fsmonitor_classify_path_gitdir_relative(rel);
 }
 
+/*
+ * We try to combine small batches at the front of the batch-list to avoid
+ * having a long list.  This hopefully makes it a little easier when we want
+ * to truncate and maintain the list.  However, we don't want the paths array
+ * to just keep growing and growing with realloc, so we insert an arbitrary
+ * limit.
+ */
+#define MY_COMBINE_LIMIT (1024)
+
+void fsmonitor_publish(struct fsmonitor_daemon_state *state,
+		       struct fsmonitor_batch *batch,
+		       const struct string_list *cookie_names)
+{
+	if (!batch && !cookie_names->nr)
+		return;
+
+	pthread_mutex_lock(&state->main_lock);
+
+	if (batch) {
+		struct fsmonitor_batch *head;
+
+		head = state->current_token_data->batch_head;
+		if (!head) {
+			BUG("token does not have batch");
+		} else if (head->pinned_time) {
+			/*
+			 * We cannot alter the current batch list
+			 * because:
+			 *
+			 * [a] it is being transmitted to at least one
+			 * client and the handle_client() thread has a
+			 * ref-count, but not a lock on the batch list
+			 * starting with this item.
+			 *
+			 * [b] it has been transmitted in the past to
+			 * at least one client such that future
+			 * requests are relative to this head batch.
+			 *
+			 * So, we can only prepend a new batch onto
+			 * the front of the list.
+			 */
+			batch->batch_seq_nr = head->batch_seq_nr + 1;
+			batch->next = head;
+			state->current_token_data->batch_head = batch;
+		} else if (!head->batch_seq_nr) {
+			/*
+			 * Batch 0 is unpinned.  See the note in
+			 * `fsmonitor_new_token_data()` about why we
+			 * don't need to accumulate these paths.
+			 */
+			fsmonitor_batch__free_list(batch);
+		} else if (head->nr + batch->nr > MY_COMBINE_LIMIT) {
+			/*
+			 * The head batch in the list has never been
+			 * transmitted to a client, but folding the
+			 * contents of the new batch onto it would
+			 * exceed our arbitrary limit, so just prepend
+			 * the new batch onto the list.
+			 */
+			batch->batch_seq_nr = head->batch_seq_nr + 1;
+			batch->next = head;
+			state->current_token_data->batch_head = batch;
+		} else {
+			/*
+			 * We are free to add the paths in the given
+			 * batch onto the end of the current head batch.
+			 */
+			fsmonitor_batch__combine(head, batch);
+			fsmonitor_batch__free_list(batch);
+		}
+	}
+
+	pthread_mutex_unlock(&state->main_lock);
+}
+
 static void *fsm_listen__thread_proc(void *_state)
 {
 	struct fsmonitor_daemon_state *state = _state;
@@ -343,6 +562,13 @@ static void *fsm_listen__thread_proc(void *_state)
 
 	fsm_listen__loop(state);
 
+	pthread_mutex_lock(&state->main_lock);
+	if (state->current_token_data &&
+	    state->current_token_data->client_ref_count == 0)
+		fsmonitor_free_token_data(state->current_token_data);
+	state->current_token_data = NULL;
+	pthread_mutex_unlock(&state->main_lock);
+
 	trace2_thread_exit();
 	return NULL;
 }
diff --git a/fsmonitor--daemon.h b/fsmonitor--daemon.h
index 8c3a71a48bd..010fbfe60e9 100644
--- a/fsmonitor--daemon.h
+++ b/fsmonitor--daemon.h
@@ -12,6 +12,27 @@
 struct fsmonitor_batch;
 struct fsmonitor_token_data;
 
+/*
+ * Create a new batch of path(s).  The returned batch is considered
+ * private and not linked into the fsmonitor daemon state.  The caller
+ * should fill this batch with one or more paths and then publish it.
+ */
+struct fsmonitor_batch *fsmonitor_batch__new(void);
+
+/*
+ * Free the list of batches starting with this one.
+ */
+void fsmonitor_batch__free_list(struct fsmonitor_batch *batch);
+
+/*
+ * Add this path to this batch of modified files.
+ *
+ * The batch should be private and NOT (yet) linked into the fsmonitor
+ * daemon state and therefore not yet visible to worker threads and so
+ * no locking is required.
+ */
+void fsmonitor_batch__add_path(struct fsmonitor_batch *batch, const char *path);
+
 struct fsmonitor_daemon_backend_data; /* opaque platform-specific data */
 
 struct fsmonitor_daemon_state {
@@ -117,5 +138,24 @@ enum fsmonitor_path_type fsmonitor_classify_path_absolute(
 	struct fsmonitor_daemon_state *state,
 	const char *path);
 
+/*
+ * Prepend the this batch of path(s) onto the list of batches associated
+ * with the current token.  This makes the batch visible to worker threads.
+ *
+ * The caller no longer owns the batch and must not free it.
+ *
+ * Wake up the client threads waiting on these cookies.
+ */
+void fsmonitor_publish(struct fsmonitor_daemon_state *state,
+		       struct fsmonitor_batch *batch,
+		       const struct string_list *cookie_names);
+
+/*
+ * If the platform-specific layer loses sync with the filesystem,
+ * it should call this to invalidate cached data and abort waiting
+ * threads.
+ */
+void fsmonitor_force_resync(struct fsmonitor_daemon_state *state);
+
 #endif /* HAVE_FSMONITOR_DAEMON_BACKEND */
 #endif /* FSMONITOR_DAEMON_H */
-- 
gitgitgadget

[PATCH v8 15/30] compat/fsmonitor/fsm-listen-win32: implement FSMonitor backend on Windows

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach the win32 backend to register a watch on the working tree
root directory (recursively).  Also watch the <gitdir> if it is
not inside the working tree.  And to collect path change notifications
into batches and publish.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 compat/fsmonitor/fsm-listen-win32.c | 565 ++++++++++++++++++++++++++++
 1 file changed, 565 insertions(+)

diff --git a/compat/fsmonitor/fsm-listen-win32.c b/compat/fsmonitor/fsm-listen-win32.c
index 916cbea254f..5b928ab66e5 100644
--- a/compat/fsmonitor/fsm-listen-win32.c
+++ b/compat/fsmonitor/fsm-listen-win32.c
@@ -2,20 +2,585 @@
 #include "config.h"
 #include "fsmonitor.h"
 #include "fsm-listen.h"
+#include "fsmonitor--daemon.h"
+
+/*
+ * The documentation of ReadDirectoryChangesW() states that the maximum
+ * buffer size is 64K when the monitored directory is remote.
+ *
+ * Larger buffers may be used when the monitored directory is local and
+ * will help us receive events faster from the kernel and avoid dropped
+ * events.
+ *
+ * So we try to use a very large buffer and silently fallback to 64K if
+ * we get an error.
+ */
+#define MAX_RDCW_BUF_FALLBACK (65536)
+#define MAX_RDCW_BUF          (65536 * 8)
+
+struct one_watch
+{
+	char buffer[MAX_RDCW_BUF];
+	DWORD buf_len;
+	DWORD count;
+
+	struct strbuf path;
+	HANDLE hDir;
+	HANDLE hEvent;
+	OVERLAPPED overlapped;
+
+	/*
+	 * Is there an active ReadDirectoryChangesW() call pending.  If so, we
+	 * need to later call GetOverlappedResult() and possibly CancelIoEx().
+	 */
+	BOOL is_active;
+};
+
+struct fsmonitor_daemon_backend_data
+{
+	struct one_watch *watch_worktree;
+	struct one_watch *watch_gitdir;
+
+	HANDLE hEventShutdown;
+
+	HANDLE hListener[3]; /* we don't own these handles */
+#define LISTENER_SHUTDOWN 0
+#define LISTENER_HAVE_DATA_WORKTREE 1
+#define LISTENER_HAVE_DATA_GITDIR 2
+	int nr_listener_handles;
+};
+
+/*
+ * Convert the WCHAR path from the notification into UTF8 and
+ * then normalize it.
+ */
+static int normalize_path_in_utf8(FILE_NOTIFY_INFORMATION *info,
+				  struct strbuf *normalized_path)
+{
+	int reserve;
+	int len = 0;
+
+	strbuf_reset(normalized_path);
+	if (!info->FileNameLength)
+		goto normalize;
+
+	/*
+	 * Pre-reserve enough space in the UTF8 buffer for
+	 * each Unicode WCHAR character to be mapped into a
+	 * sequence of 2 UTF8 characters.  That should let us
+	 * avoid ERROR_INSUFFICIENT_BUFFER 99.9+% of the time.
+	 */
+	reserve = info->FileNameLength + 1;
+	strbuf_grow(normalized_path, reserve);
+
+	for (;;) {
+		len = WideCharToMultiByte(CP_UTF8, 0, info->FileName,
+					  info->FileNameLength / sizeof(WCHAR),
+					  normalized_path->buf,
+					  strbuf_avail(normalized_path) - 1,
+					  NULL, NULL);
+		if (len > 0)
+			goto normalize;
+		if (GetLastError() != ERROR_INSUFFICIENT_BUFFER) {
+			error(_("[GLE %ld] could not convert path to UTF-8: '%.*ls'"),
+			      GetLastError(),
+			      (int)(info->FileNameLength / sizeof(WCHAR)),
+			      info->FileName);
+			return -1;
+		}
+
+		strbuf_grow(normalized_path,
+			    strbuf_avail(normalized_path) + reserve);
+	}
+
+normalize:
+	strbuf_setlen(normalized_path, len);
+	return strbuf_normalize_path(normalized_path);
+}
 
 void fsm_listen__stop_async(struct fsmonitor_daemon_state *state)
 {
+	SetEvent(state->backend_data->hListener[LISTENER_SHUTDOWN]);
+}
+
+static struct one_watch *create_watch(struct fsmonitor_daemon_state *state,
+				      const char *path)
+{
+	struct one_watch *watch = NULL;
+	DWORD desired_access = FILE_LIST_DIRECTORY;
+	DWORD share_mode =
+		FILE_SHARE_WRITE | FILE_SHARE_READ | FILE_SHARE_DELETE;
+	HANDLE hDir;
+	wchar_t wpath[MAX_PATH];
+
+	if (xutftowcs_path(wpath, path) < 0) {
+		error(_("could not convert to wide characters: '%s'"), path);
+		return NULL;
+	}
+
+	hDir = CreateFileW(wpath,
+			   desired_access, share_mode, NULL, OPEN_EXISTING,
+			   FILE_FLAG_BACKUP_SEMANTICS | FILE_FLAG_OVERLAPPED,
+			   NULL);
+	if (hDir == INVALID_HANDLE_VALUE) {
+		error(_("[GLE %ld] could not watch '%s'"),
+		      GetLastError(), path);
+		return NULL;
+	}
+
+	CALLOC_ARRAY(watch, 1);
+
+	watch->buf_len = sizeof(watch->buffer); /* assume full MAX_RDCW_BUF */
+
+	strbuf_init(&watch->path, 0);
+	strbuf_addstr(&watch->path, path);
+
+	watch->hDir = hDir;
+	watch->hEvent = CreateEvent(NULL, TRUE, FALSE, NULL);
+
+	return watch;
+}
+
+static void destroy_watch(struct one_watch *watch)
+{
+	if (!watch)
+		return;
+
+	strbuf_release(&watch->path);
+	if (watch->hDir != INVALID_HANDLE_VALUE)
+		CloseHandle(watch->hDir);
+	if (watch->hEvent != INVALID_HANDLE_VALUE)
+		CloseHandle(watch->hEvent);
+
+	free(watch);
+}
+
+static int start_rdcw_watch(struct fsmonitor_daemon_backend_data *data,
+			    struct one_watch *watch)
+{
+	DWORD dwNotifyFilter =
+		FILE_NOTIFY_CHANGE_FILE_NAME |
+		FILE_NOTIFY_CHANGE_DIR_NAME |
+		FILE_NOTIFY_CHANGE_ATTRIBUTES |
+		FILE_NOTIFY_CHANGE_SIZE |
+		FILE_NOTIFY_CHANGE_LAST_WRITE |
+		FILE_NOTIFY_CHANGE_CREATION;
+
+	ResetEvent(watch->hEvent);
+
+	memset(&watch->overlapped, 0, sizeof(watch->overlapped));
+	watch->overlapped.hEvent = watch->hEvent;
+
+	/*
+	 * Queue an async call using Overlapped IO.  This returns immediately.
+	 * Our event handle will be signalled when the real result is available.
+	 *
+	 * The return value here just means that we successfully queued it.
+	 * We won't know if the Read...() actually produces data until later.
+	 */
+	watch->is_active = ReadDirectoryChangesW(
+		watch->hDir, watch->buffer, watch->buf_len, TRUE,
+		dwNotifyFilter, &watch->count, &watch->overlapped, NULL);
+
+	if (watch->is_active)
+		return 0;
+
+	error(_("ReadDirectoryChangedW failed on '%s' [GLE %ld]"),
+	      watch->path.buf, GetLastError());
+	return -1;
+}
+
+static int recv_rdcw_watch(struct one_watch *watch)
+{
+	DWORD gle;
+
+	watch->is_active = FALSE;
+
+	/*
+	 * The overlapped result is ready.  If the Read...() was successful
+	 * we finally receive the actual result into our buffer.
+	 */
+	if (GetOverlappedResult(watch->hDir, &watch->overlapped, &watch->count,
+				TRUE))
+		return 0;
+
+	gle = GetLastError();
+	if (gle == ERROR_INVALID_PARAMETER &&
+	    /*
+	     * The kernel throws an invalid parameter error when our
+	     * buffer is too big and we are pointed at a remote
+	     * directory (and possibly for other reasons).  Quietly
+	     * set it down and try again.
+	     *
+	     * See note about MAX_RDCW_BUF at the top.
+	     */
+	    watch->buf_len > MAX_RDCW_BUF_FALLBACK) {
+		watch->buf_len = MAX_RDCW_BUF_FALLBACK;
+		return -2;
+	}
+
+	/*
+	 * NEEDSWORK: If an external <gitdir> is deleted, the above
+	 * returns an error.  I'm not sure that there's anything that
+	 * we can do here other than failing -- the <worktree>/.git
+	 * link file would be broken anyway.  We might try to check
+	 * for that and return a better error message, but I'm not
+	 * sure it is worth it.
+	 */
+
+	error(_("GetOverlappedResult failed on '%s' [GLE %ld]"),
+	      watch->path.buf, gle);
+	return -1;
+}
+
+static void cancel_rdcw_watch(struct one_watch *watch)
+{
+	DWORD count;
+
+	if (!watch || !watch->is_active)
+		return;
+
+	/*
+	 * The calls to ReadDirectoryChangesW() and GetOverlappedResult()
+	 * form a "pair" (my term) where we queue an IO and promise to
+	 * hang around and wait for the kernel to give us the result.
+	 *
+	 * If for some reason after we queue the IO, we have to quit
+	 * or otherwise not stick around for the second half, we must
+	 * tell the kernel to abort the IO.  This prevents the kernel
+	 * from writing to our buffer and/or signalling our event
+	 * after we free them.
+	 *
+	 * (Ask me how much fun it was to track that one down).
+	 */
+	CancelIoEx(watch->hDir, &watch->overlapped);
+	GetOverlappedResult(watch->hDir, &watch->overlapped, &count, TRUE);
+	watch->is_active = FALSE;
+}
+
+/*
+ * Process filesystem events that happen anywhere (recursively) under the
+ * <worktree> root directory.  For a normal working directory, this includes
+ * both version controlled files and the contents of the .git/ directory.
+ *
+ * If <worktree>/.git is a file, then we only see events for the file
+ * itself.
+ */
+static int process_worktree_events(struct fsmonitor_daemon_state *state)
+{
+	struct fsmonitor_daemon_backend_data *data = state->backend_data;
+	struct one_watch *watch = data->watch_worktree;
+	struct strbuf path = STRBUF_INIT;
+	struct string_list cookie_list = STRING_LIST_INIT_DUP;
+	struct fsmonitor_batch *batch = NULL;
+	const char *p = watch->buffer;
+
+	/*
+	 * If the kernel gets more events than will fit in the kernel
+	 * buffer associated with our RDCW handle, it drops them and
+	 * returns a count of zero.
+	 *
+	 * Yes, the call returns WITHOUT error and with length zero.
+	 * This is the documented behavior.  (My testing has confirmed
+	 * that it also sets the last error to ERROR_NOTIFY_ENUM_DIR,
+	 * but we do not rely on that since the function did not
+	 * return an error and it is not documented.)
+	 *
+	 * (The "overflow" case is not ambiguous with the "no data" case
+	 * because we did an INFINITE wait.)
+	 *
+	 * This means we have a gap in coverage.  Tell the daemon layer
+	 * to resync.
+	 */
+	if (!watch->count) {
+		trace2_data_string("fsmonitor", NULL, "fsm-listen/kernel",
+				   "overflow");
+		fsmonitor_force_resync(state);
+		return LISTENER_HAVE_DATA_WORKTREE;
+	}
+
+	/*
+	 * On Windows, `info` contains an "array" of paths that are
+	 * relative to the root of whichever directory handle received
+	 * the event.
+	 */
+	for (;;) {
+		FILE_NOTIFY_INFORMATION *info = (void *)p;
+		const char *slash;
+		enum fsmonitor_path_type t;
+
+		strbuf_reset(&path);
+		if (normalize_path_in_utf8(info, &path) == -1)
+			goto skip_this_path;
+
+		t = fsmonitor_classify_path_workdir_relative(path.buf);
+
+		switch (t) {
+		case IS_INSIDE_DOT_GIT_WITH_COOKIE_PREFIX:
+			/* special case cookie files within .git */
+
+			/* Use just the filename of the cookie file. */
+			slash = find_last_dir_sep(path.buf);
+			string_list_append(&cookie_list,
+					   slash ? slash + 1 : path.buf);
+			break;
+
+		case IS_INSIDE_DOT_GIT:
+			/* ignore everything inside of "<worktree>/.git/" */
+			break;
+
+		case IS_DOT_GIT:
+			/* "<worktree>/.git" was deleted (or renamed away) */
+			if ((info->Action == FILE_ACTION_REMOVED) ||
+			    (info->Action == FILE_ACTION_RENAMED_OLD_NAME)) {
+				trace2_data_string("fsmonitor", NULL,
+						   "fsm-listen/dotgit",
+						   "removed");
+				goto force_shutdown;
+			}
+			break;
+
+		case IS_WORKDIR_PATH:
+			/* queue normal pathname */
+			if (!batch)
+				batch = fsmonitor_batch__new();
+			fsmonitor_batch__add_path(batch, path.buf);
+			break;
+
+		case IS_GITDIR:
+		case IS_INSIDE_GITDIR:
+		case IS_INSIDE_GITDIR_WITH_COOKIE_PREFIX:
+		default:
+			BUG("unexpected path classification '%d' for '%s'",
+			    t, path.buf);
+		}
+
+skip_this_path:
+		if (!info->NextEntryOffset)
+			break;
+		p += info->NextEntryOffset;
+	}
+
+	fsmonitor_publish(state, batch, &cookie_list);
+	batch = NULL;
+	string_list_clear(&cookie_list, 0);
+	strbuf_release(&path);
+	return LISTENER_HAVE_DATA_WORKTREE;
+
+force_shutdown:
+	fsmonitor_batch__free_list(batch);
+	string_list_clear(&cookie_list, 0);
+	strbuf_release(&path);
+	return LISTENER_SHUTDOWN;
+}
+
+/*
+ * Process filesystem events that happened anywhere (recursively) under the
+ * external <gitdir> (such as non-primary worktrees or submodules).
+ * We only care about cookie files that our client threads created here.
+ *
+ * Note that we DO NOT get filesystem events on the external <gitdir>
+ * itself (it is not inside something that we are watching).  In particular,
+ * we do not get an event if the external <gitdir> is deleted.
+ */
+static int process_gitdir_events(struct fsmonitor_daemon_state *state)
+{
+	struct fsmonitor_daemon_backend_data *data = state->backend_data;
+	struct one_watch *watch = data->watch_gitdir;
+	struct strbuf path = STRBUF_INIT;
+	struct string_list cookie_list = STRING_LIST_INIT_DUP;
+	const char *p = watch->buffer;
+
+	if (!watch->count) {
+		trace2_data_string("fsmonitor", NULL, "fsm-listen/kernel",
+				   "overflow");
+		fsmonitor_force_resync(state);
+		return LISTENER_HAVE_DATA_GITDIR;
+	}
+
+	for (;;) {
+		FILE_NOTIFY_INFORMATION *info = (void *)p;
+		const char *slash;
+		enum fsmonitor_path_type t;
+
+		strbuf_reset(&path);
+		if (normalize_path_in_utf8(info, &path) == -1)
+			goto skip_this_path;
+
+		t = fsmonitor_classify_path_gitdir_relative(path.buf);
+
+		switch (t) {
+		case IS_INSIDE_GITDIR_WITH_COOKIE_PREFIX:
+			/* special case cookie files within gitdir */
+
+			/* Use just the filename of the cookie file. */
+			slash = find_last_dir_sep(path.buf);
+			string_list_append(&cookie_list,
+					   slash ? slash + 1 : path.buf);
+			break;
+
+		case IS_INSIDE_GITDIR:
+			goto skip_this_path;
+
+		default:
+			BUG("unexpected path classification '%d' for '%s'",
+			    t, path.buf);
+		}
+
+skip_this_path:
+		if (!info->NextEntryOffset)
+			break;
+		p += info->NextEntryOffset;
+	}
+
+	fsmonitor_publish(state, NULL, &cookie_list);
+	string_list_clear(&cookie_list, 0);
+	strbuf_release(&path);
+	return LISTENER_HAVE_DATA_GITDIR;
 }
 
 void fsm_listen__loop(struct fsmonitor_daemon_state *state)
 {
+	struct fsmonitor_daemon_backend_data *data = state->backend_data;
+	DWORD dwWait;
+	int result;
+
+	state->error_code = 0;
+
+	if (start_rdcw_watch(data, data->watch_worktree) == -1)
+		goto force_error_stop;
+
+	if (data->watch_gitdir &&
+	    start_rdcw_watch(data, data->watch_gitdir) == -1)
+		goto force_error_stop;
+
+	for (;;) {
+		dwWait = WaitForMultipleObjects(data->nr_listener_handles,
+						data->hListener,
+						FALSE, INFINITE);
+
+		if (dwWait == WAIT_OBJECT_0 + LISTENER_HAVE_DATA_WORKTREE) {
+			result = recv_rdcw_watch(data->watch_worktree);
+			if (result == -1) {
+				/* hard error */
+				goto force_error_stop;
+			}
+			if (result == -2) {
+				/* retryable error */
+				if (start_rdcw_watch(data, data->watch_worktree) == -1)
+					goto force_error_stop;
+				continue;
+			}
+
+			/* have data */
+			if (process_worktree_events(state) == LISTENER_SHUTDOWN)
+				goto force_shutdown;
+			if (start_rdcw_watch(data, data->watch_worktree) == -1)
+				goto force_error_stop;
+			continue;
+		}
+
+		if (dwWait == WAIT_OBJECT_0 + LISTENER_HAVE_DATA_GITDIR) {
+			result = recv_rdcw_watch(data->watch_gitdir);
+			if (result == -1) {
+				/* hard error */
+				goto force_error_stop;
+			}
+			if (result == -2) {
+				/* retryable error */
+				if (start_rdcw_watch(data, data->watch_gitdir) == -1)
+					goto force_error_stop;
+				continue;
+			}
+
+			/* have data */
+			if (process_gitdir_events(state) == LISTENER_SHUTDOWN)
+				goto force_shutdown;
+			if (start_rdcw_watch(data, data->watch_gitdir) == -1)
+				goto force_error_stop;
+			continue;
+		}
+
+		if (dwWait == WAIT_OBJECT_0 + LISTENER_SHUTDOWN)
+			goto clean_shutdown;
+
+		error(_("could not read directory changes [GLE %ld]"),
+		      GetLastError());
+		goto force_error_stop;
+	}
+
+force_error_stop:
+	state->error_code = -1;
+
+force_shutdown:
+	/*
+	 * Tell the IPC thead pool to stop (which completes the await
+	 * in the main thread (which will also signal this thread (if
+	 * we are still alive))).
+	 */
+	ipc_server_stop_async(state->ipc_server_data);
+
+clean_shutdown:
+	cancel_rdcw_watch(data->watch_worktree);
+	cancel_rdcw_watch(data->watch_gitdir);
 }
 
 int fsm_listen__ctor(struct fsmonitor_daemon_state *state)
 {
+	struct fsmonitor_daemon_backend_data *data;
+
+	CALLOC_ARRAY(data, 1);
+
+	data->hEventShutdown = CreateEvent(NULL, TRUE, FALSE, NULL);
+
+	data->watch_worktree = create_watch(state,
+					    state->path_worktree_watch.buf);
+	if (!data->watch_worktree)
+		goto failed;
+
+	if (state->nr_paths_watching > 1) {
+		data->watch_gitdir = create_watch(state,
+						  state->path_gitdir_watch.buf);
+		if (!data->watch_gitdir)
+			goto failed;
+	}
+
+	data->hListener[LISTENER_SHUTDOWN] = data->hEventShutdown;
+	data->nr_listener_handles++;
+
+	data->hListener[LISTENER_HAVE_DATA_WORKTREE] =
+		data->watch_worktree->hEvent;
+	data->nr_listener_handles++;
+
+	if (data->watch_gitdir) {
+		data->hListener[LISTENER_HAVE_DATA_GITDIR] =
+			data->watch_gitdir->hEvent;
+		data->nr_listener_handles++;
+	}
+
+	state->backend_data = data;
+	return 0;
+
+failed:
+	CloseHandle(data->hEventShutdown);
+	destroy_watch(data->watch_worktree);
+	destroy_watch(data->watch_gitdir);
+
 	return -1;
 }
 
 void fsm_listen__dtor(struct fsmonitor_daemon_state *state)
 {
+	struct fsmonitor_daemon_backend_data *data;
+
+	if (!state || !state->backend_data)
+		return;
+
+	data = state->backend_data;
+
+	CloseHandle(data->hEventShutdown);
+	destroy_watch(data->watch_worktree);
+	destroy_watch(data->watch_gitdir);
+
+	FREE_AND_NULL(state->backend_data);
 }
-- 
gitgitgadget

[PATCH v8 16/30] compat/fsmonitor/fsm-listen-darwin: add MacOS header files for FSEvent

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Include MacOS system declarations to allow us to use FSEvent and
CoreFoundation APIs.  We need different versions of the declarations
for GCC vs. clang because of compiler and header file conflicts.

While it is quite possible to #include Apple's CoreServices.h when
compiling C source code with clang, trying to build it with GCC
currently fails with this error:

In file included
   from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
   ...Library/Frameworks/Security.framework/Headers/AuthSession.h:32,
   from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
   ...Library/Frameworks/Security.framework/Headers/Security.h:42,
   from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
   ...Library/Frameworks/CoreServices.framework/Frameworks/...
   ...OSServices.framework/Headers/CSIdentity.h:43,
   from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
   ...Library/Frameworks/CoreServices.framework/Frameworks/...
   ...OSServices.framework/Headers/OSServices.h:29,
   from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
   ...Library/Frameworks/CoreServices.framework/Frameworks/...
   ...LaunchServices.framework/Headers/IconsCore.h:23,
   from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
   ...Library/Frameworks/CoreServices.framework/Frameworks/...
   ...LaunchServices.framework/Headers/LaunchServices.h:23,
   from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
   ...Library/Frameworks/CoreServices.framework/Headers/CoreServices.h:45,

     /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
     ...Library/Frameworks/Security.framework/Headers/Authorization.h:193:7:
     error: variably modified 'bytes' at file scope
       193 | char bytes[kAuthorizationExternalFormLength];
           |      ^~~~~

The underlying reason is that GCC (rightfully) objects that an `enum`
value such as `kAuthorizationExternalFormLength` is not a constant
(because it is not, the preprocessor has no knowledge of it, only the
actual C compiler does) and can therefore not be used to define the size
of a C array.

This is a known problem and tracked in GCC's bug tracker:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93082

In the meantime, let's not block things and go the slightly ugly route
of declaring/defining the FSEvents constants, data structures and
functions that we need, so that we can avoid above-mentioned issue.

Let's do this _only_ for GCC, though, so that the CI/PR builds (which
build both with clang and with GCC) can guarantee that we _are_ using
the correct data types.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 compat/fsmonitor/fsm-darwin-gcc.h    | 92 ++++++++++++++++++++++++++++
 compat/fsmonitor/fsm-listen-darwin.c | 24 ++++++++
 2 files changed, 116 insertions(+)
 create mode 100644 compat/fsmonitor/fsm-darwin-gcc.h

diff --git a/compat/fsmonitor/fsm-darwin-gcc.h b/compat/fsmonitor/fsm-darwin-gcc.h
new file mode 100644
index 00000000000..1c75c3d48e7
--- /dev/null
+++ b/compat/fsmonitor/fsm-darwin-gcc.h
@@ -0,0 +1,92 @@
+#ifndef FSM_DARWIN_GCC_H
+#define FSM_DARWIN_GCC_H
+
+#ifndef __clang__
+/*
+ * It is possible to #include CoreFoundation/CoreFoundation.h when compiling
+ * with clang, but not with GCC as of time of writing.
+ *
+ * See https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93082 for details.
+ */
+typedef unsigned int FSEventStreamCreateFlags;
+#define kFSEventStreamEventFlagNone               0x00000000
+#define kFSEventStreamEventFlagMustScanSubDirs    0x00000001
+#define kFSEventStreamEventFlagUserDropped        0x00000002
+#define kFSEventStreamEventFlagKernelDropped      0x00000004
+#define kFSEventStreamEventFlagEventIdsWrapped    0x00000008
+#define kFSEventStreamEventFlagHistoryDone        0x00000010
+#define kFSEventStreamEventFlagRootChanged        0x00000020
+#define kFSEventStreamEventFlagMount              0x00000040
+#define kFSEventStreamEventFlagUnmount            0x00000080
+#define kFSEventStreamEventFlagItemCreated        0x00000100
+#define kFSEventStreamEventFlagItemRemoved        0x00000200
+#define kFSEventStreamEventFlagItemInodeMetaMod   0x00000400
+#define kFSEventStreamEventFlagItemRenamed        0x00000800
+#define kFSEventStreamEventFlagItemModified       0x00001000
+#define kFSEventStreamEventFlagItemFinderInfoMod  0x00002000
+#define kFSEventStreamEventFlagItemChangeOwner    0x00004000
+#define kFSEventStreamEventFlagItemXattrMod       0x00008000
+#define kFSEventStreamEventFlagItemIsFile         0x00010000
+#define kFSEventStreamEventFlagItemIsDir          0x00020000
+#define kFSEventStreamEventFlagItemIsSymlink      0x00040000
+#define kFSEventStreamEventFlagOwnEvent           0x00080000
+#define kFSEventStreamEventFlagItemIsHardlink     0x00100000
+#define kFSEventStreamEventFlagItemIsLastHardlink 0x00200000
+#define kFSEventStreamEventFlagItemCloned         0x00400000
+
+typedef struct __FSEventStream *FSEventStreamRef;
+typedef const FSEventStreamRef ConstFSEventStreamRef;
+
+typedef unsigned int CFStringEncoding;
+#define kCFStringEncodingUTF8 0x08000100
+
+typedef const struct __CFString *CFStringRef;
+typedef const struct __CFArray *CFArrayRef;
+typedef const struct __CFRunLoop *CFRunLoopRef;
+
+struct FSEventStreamContext {
+    long long version;
+    void *cb_data, *retain, *release, *copy_description;
+};
+
+typedef struct FSEventStreamContext FSEventStreamContext;
+typedef unsigned int FSEventStreamEventFlags;
+#define kFSEventStreamCreateFlagNoDefer 0x02
+#define kFSEventStreamCreateFlagWatchRoot 0x04
+#define kFSEventStreamCreateFlagFileEvents 0x10
+
+typedef unsigned long long FSEventStreamEventId;
+#define kFSEventStreamEventIdSinceNow 0xFFFFFFFFFFFFFFFFULL
+
+typedef void (*FSEventStreamCallback)(ConstFSEventStreamRef streamRef,
+				      void *context,
+				      __SIZE_TYPE__ num_of_events,
+				      void *event_paths,
+				      const FSEventStreamEventFlags event_flags[],
+				      const FSEventStreamEventId event_ids[]);
+typedef double CFTimeInterval;
+FSEventStreamRef FSEventStreamCreate(void *allocator,
+				     FSEventStreamCallback callback,
+				     FSEventStreamContext *context,
+				     CFArrayRef paths_to_watch,
+				     FSEventStreamEventId since_when,
+				     CFTimeInterval latency,
+				     FSEventStreamCreateFlags flags);
+CFStringRef CFStringCreateWithCString(void *allocator, const char *string,
+				      CFStringEncoding encoding);
+CFArrayRef CFArrayCreate(void *allocator, const void **items, long long count,
+			 void *callbacks);
+void CFRunLoopRun(void);
+void CFRunLoopStop(CFRunLoopRef run_loop);
+CFRunLoopRef CFRunLoopGetCurrent(void);
+extern CFStringRef kCFRunLoopDefaultMode;
+void FSEventStreamScheduleWithRunLoop(FSEventStreamRef stream,
+				      CFRunLoopRef run_loop,
+				      CFStringRef run_loop_mode);
+unsigned char FSEventStreamStart(FSEventStreamRef stream);
+void FSEventStreamStop(FSEventStreamRef stream);
+void FSEventStreamInvalidate(FSEventStreamRef stream);
+void FSEventStreamRelease(FSEventStreamRef stream);
+
+#endif /* !clang */
+#endif /* FSM_DARWIN_GCC_H */
diff --git a/compat/fsmonitor/fsm-listen-darwin.c b/compat/fsmonitor/fsm-listen-darwin.c
index c84e3344ab9..d2ce942cade 100644
--- a/compat/fsmonitor/fsm-listen-darwin.c
+++ b/compat/fsmonitor/fsm-listen-darwin.c
@@ -1,3 +1,27 @@
+#ifndef __clang__
+#include "fsm-darwin-gcc.h"
+#else
+#include <CoreFoundation/CoreFoundation.h>
+#include <CoreServices/CoreServices.h>
+
+#ifndef AVAILABLE_MAC_OS_X_VERSION_10_13_AND_LATER
+/*
+ * This enum value was added in 10.13 to:
+ *
+ * /Applications/Xcode.app/Contents/Developer/Platforms/ \
+ *    MacOSX.platform/Developer/SDKs/MacOSX.sdk/System/ \
+ *    Library/Frameworks/CoreServices.framework/Frameworks/ \
+ *    FSEvents.framework/Versions/Current/Headers/FSEvents.h
+ *
+ * If we're compiling against an older SDK, this symbol won't be
+ * present.  Silently define it here so that we don't have to ifdef
+ * the logging or masking below.  This should be harmless since older
+ * versions of macOS won't ever emit this FS event anyway.
+ */
+#define kFSEventStreamEventFlagItemCloned         0x00400000
+#endif
+#endif
+
 #include "cache.h"
 #include "fsmonitor.h"
 #include "fsm-listen.h"
-- 
gitgitgadget

[PATCH v8 17/30] compat/fsmonitor/fsm-listen-darwin: implement FSEvent listener on MacOS

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Implement file system event listener on MacOS using FSEvent,
CoreFoundation, and CoreServices.

Co-authored-by: Kevin Willford <Kevin.Willford@microsoft.com>
Co-authored-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 compat/fsmonitor/fsm-listen-darwin.c | 383 +++++++++++++++++++++++++++
 1 file changed, 383 insertions(+)

diff --git a/compat/fsmonitor/fsm-listen-darwin.c b/compat/fsmonitor/fsm-listen-darwin.c
index d2ce942cade..0741fe834c3 100644
--- a/compat/fsmonitor/fsm-listen-darwin.c
+++ b/compat/fsmonitor/fsm-listen-darwin.c
@@ -25,20 +25,403 @@
 #include "cache.h"
 #include "fsmonitor.h"
 #include "fsm-listen.h"
+#include "fsmonitor--daemon.h"
+
+struct fsmonitor_daemon_backend_data
+{
+	CFStringRef cfsr_worktree_path;
+	CFStringRef cfsr_gitdir_path;
+
+	CFArrayRef cfar_paths_to_watch;
+	int nr_paths_watching;
+
+	FSEventStreamRef stream;
+
+	CFRunLoopRef rl;
+
+	enum shutdown_style {
+		SHUTDOWN_EVENT = 0,
+		FORCE_SHUTDOWN,
+		FORCE_ERROR_STOP,
+	} shutdown_style;
+
+	unsigned int stream_scheduled:1;
+	unsigned int stream_started:1;
+};
+
+static void log_flags_set(const char *path, const FSEventStreamEventFlags flag)
+{
+	struct strbuf msg = STRBUF_INIT;
+
+	if (flag & kFSEventStreamEventFlagMustScanSubDirs)
+		strbuf_addstr(&msg, "MustScanSubDirs|");
+	if (flag & kFSEventStreamEventFlagUserDropped)
+		strbuf_addstr(&msg, "UserDropped|");
+	if (flag & kFSEventStreamEventFlagKernelDropped)
+		strbuf_addstr(&msg, "KernelDropped|");
+	if (flag & kFSEventStreamEventFlagEventIdsWrapped)
+		strbuf_addstr(&msg, "EventIdsWrapped|");
+	if (flag & kFSEventStreamEventFlagHistoryDone)
+		strbuf_addstr(&msg, "HistoryDone|");
+	if (flag & kFSEventStreamEventFlagRootChanged)
+		strbuf_addstr(&msg, "RootChanged|");
+	if (flag & kFSEventStreamEventFlagMount)
+		strbuf_addstr(&msg, "Mount|");
+	if (flag & kFSEventStreamEventFlagUnmount)
+		strbuf_addstr(&msg, "Unmount|");
+	if (flag & kFSEventStreamEventFlagItemChangeOwner)
+		strbuf_addstr(&msg, "ItemChangeOwner|");
+	if (flag & kFSEventStreamEventFlagItemCreated)
+		strbuf_addstr(&msg, "ItemCreated|");
+	if (flag & kFSEventStreamEventFlagItemFinderInfoMod)
+		strbuf_addstr(&msg, "ItemFinderInfoMod|");
+	if (flag & kFSEventStreamEventFlagItemInodeMetaMod)
+		strbuf_addstr(&msg, "ItemInodeMetaMod|");
+	if (flag & kFSEventStreamEventFlagItemIsDir)
+		strbuf_addstr(&msg, "ItemIsDir|");
+	if (flag & kFSEventStreamEventFlagItemIsFile)
+		strbuf_addstr(&msg, "ItemIsFile|");
+	if (flag & kFSEventStreamEventFlagItemIsHardlink)
+		strbuf_addstr(&msg, "ItemIsHardlink|");
+	if (flag & kFSEventStreamEventFlagItemIsLastHardlink)
+		strbuf_addstr(&msg, "ItemIsLastHardlink|");
+	if (flag & kFSEventStreamEventFlagItemIsSymlink)
+		strbuf_addstr(&msg, "ItemIsSymlink|");
+	if (flag & kFSEventStreamEventFlagItemModified)
+		strbuf_addstr(&msg, "ItemModified|");
+	if (flag & kFSEventStreamEventFlagItemRemoved)
+		strbuf_addstr(&msg, "ItemRemoved|");
+	if (flag & kFSEventStreamEventFlagItemRenamed)
+		strbuf_addstr(&msg, "ItemRenamed|");
+	if (flag & kFSEventStreamEventFlagItemXattrMod)
+		strbuf_addstr(&msg, "ItemXattrMod|");
+	if (flag & kFSEventStreamEventFlagOwnEvent)
+		strbuf_addstr(&msg, "OwnEvent|");
+	if (flag & kFSEventStreamEventFlagItemCloned)
+		strbuf_addstr(&msg, "ItemCloned|");
+
+	trace_printf_key(&trace_fsmonitor, "fsevent: '%s', flags=%u %s",
+			 path, flag, msg.buf);
+
+	strbuf_release(&msg);
+}
+
+static int ef_is_root_delete(const FSEventStreamEventFlags ef)
+{
+	return (ef & kFSEventStreamEventFlagItemIsDir &&
+		ef & kFSEventStreamEventFlagItemRemoved);
+}
+
+static int ef_is_root_renamed(const FSEventStreamEventFlags ef)
+{
+	return (ef & kFSEventStreamEventFlagItemIsDir &&
+		ef & kFSEventStreamEventFlagItemRenamed);
+}
+
+static int ef_is_dropped(const FSEventStreamEventFlags ef)
+{
+	return (ef & kFSEventStreamEventFlagMustScanSubDirs ||
+		ef & kFSEventStreamEventFlagKernelDropped ||
+		ef & kFSEventStreamEventFlagUserDropped);
+}
+
+static void fsevent_callback(ConstFSEventStreamRef streamRef,
+			     void *ctx,
+			     size_t num_of_events,
+			     void *event_paths,
+			     const FSEventStreamEventFlags event_flags[],
+			     const FSEventStreamEventId event_ids[])
+{
+	struct fsmonitor_daemon_state *state = ctx;
+	struct fsmonitor_daemon_backend_data *data = state->backend_data;
+	char **paths = (char **)event_paths;
+	struct fsmonitor_batch *batch = NULL;
+	struct string_list cookie_list = STRING_LIST_INIT_DUP;
+	const char *path_k;
+	const char *slash;
+	int k;
+	struct strbuf tmp = STRBUF_INIT;
+
+	/*
+	 * Build a list of all filesystem changes into a private/local
+	 * list and without holding any locks.
+	 */
+	for (k = 0; k < num_of_events; k++) {
+		/*
+		 * On Mac, we receive an array of absolute paths.
+		 */
+		path_k = paths[k];
+
+		/*
+		 * If you want to debug FSEvents, log them to GIT_TRACE_FSMONITOR.
+		 * Please don't log them to Trace2.
+		 *
+		 * trace_printf_key(&trace_fsmonitor, "Path: '%s'", path_k);
+		 */
+
+		/*
+		 * If event[k] is marked as dropped, we assume that we have
+		 * lost sync with the filesystem and should flush our cached
+		 * data.  We need to:
+		 *
+		 * [1] Abort/wake any client threads waiting for a cookie and
+		 *     flush the cached state data (the current token), and
+		 *     create a new token.
+		 *
+		 * [2] Discard the batch that we were locally building (since
+		 *     they are conceptually relative to the just flushed
+		 *     token).
+		 */
+		if (ef_is_dropped(event_flags[k])) {
+			if (trace_pass_fl(&trace_fsmonitor))
+				log_flags_set(path_k, event_flags[k]);
+
+			fsmonitor_force_resync(state);
+			fsmonitor_batch__free_list(batch);
+			string_list_clear(&cookie_list, 0);
+
+			/*
+			 * We assume that any events that we received
+			 * in this callback after this dropped event
+			 * may still be valid, so we continue rather
+			 * than break.  (And just in case there is a
+			 * delete of ".git" hiding in there.)
+			 */
+			continue;
+		}
+
+		switch (fsmonitor_classify_path_absolute(state, path_k)) {
+
+		case IS_INSIDE_DOT_GIT_WITH_COOKIE_PREFIX:
+		case IS_INSIDE_GITDIR_WITH_COOKIE_PREFIX:
+			/* special case cookie files within .git or gitdir */
+
+			/* Use just the filename of the cookie file. */
+			slash = find_last_dir_sep(path_k);
+			string_list_append(&cookie_list,
+					   slash ? slash + 1 : path_k);
+			break;
+
+		case IS_INSIDE_DOT_GIT:
+		case IS_INSIDE_GITDIR:
+			/* ignore all other paths inside of .git or gitdir */
+			break;
+
+		case IS_DOT_GIT:
+		case IS_GITDIR:
+			/*
+			 * If .git directory is deleted or renamed away,
+			 * we have to quit.
+			 */
+			if (ef_is_root_delete(event_flags[k])) {
+				trace_printf_key(&trace_fsmonitor,
+						 "event: gitdir removed");
+				goto force_shutdown;
+			}
+			if (ef_is_root_renamed(event_flags[k])) {
+				trace_printf_key(&trace_fsmonitor,
+						 "event: gitdir renamed");
+				goto force_shutdown;
+			}
+			break;
+
+		case IS_WORKDIR_PATH:
+			/* try to queue normal pathnames */
+
+			if (trace_pass_fl(&trace_fsmonitor))
+				log_flags_set(path_k, event_flags[k]);
+
+			/*
+			 * Because of the implicit "binning" (the
+			 * kernel calls us at a given frequency) and
+			 * de-duping (the kernel is free to combine
+			 * multiple events for a given pathname), an
+			 * individual fsevent could be marked as both
+			 * a file and directory.  Add it to the queue
+			 * with both spellings so that the client will
+			 * know how much to invalidate/refresh.
+			 */
+
+			if (event_flags[k] & kFSEventStreamEventFlagItemIsFile) {
+				const char *rel = path_k +
+					state->path_worktree_watch.len + 1;
+
+				if (!batch)
+					batch = fsmonitor_batch__new();
+				fsmonitor_batch__add_path(batch, rel);
+			}
+
+			if (event_flags[k] & kFSEventStreamEventFlagItemIsDir) {
+				const char *rel = path_k +
+					state->path_worktree_watch.len + 1;
+
+				strbuf_reset(&tmp);
+				strbuf_addstr(&tmp, rel);
+				strbuf_addch(&tmp, '/');
+
+				if (!batch)
+					batch = fsmonitor_batch__new();
+				fsmonitor_batch__add_path(batch, tmp.buf);
+			}
+
+			break;
+
+		case IS_OUTSIDE_CONE:
+		default:
+			trace_printf_key(&trace_fsmonitor,
+					 "ignoring '%s'", path_k);
+			break;
+		}
+	}
+
+	fsmonitor_publish(state, batch, &cookie_list);
+	string_list_clear(&cookie_list, 0);
+	strbuf_release(&tmp);
+	return;
+
+force_shutdown:
+	fsmonitor_batch__free_list(batch);
+	string_list_clear(&cookie_list, 0);
+
+	data->shutdown_style = FORCE_SHUTDOWN;
+	CFRunLoopStop(data->rl);
+	strbuf_release(&tmp);
+	return;
+}
+
+/*
+ * In the call to `FSEventStreamCreate()` to setup our watch, the
+ * `latency` argument determines the frequency of calls to our callback
+ * with new FS events.  Too slow and events get dropped; too fast and
+ * we burn CPU unnecessarily.  Since it is rather obscure, I don't
+ * think this needs to be a config setting.  I've done extensive
+ * testing on my systems and chosen the value below.  It gives good
+ * results and I've not seen any dropped events.
+ *
+ * With a latency of 0.1, I was seeing lots of dropped events during
+ * the "touch 100000" files test within t/perf/p7519, but with a
+ * latency of 0.001 I did not see any dropped events.  So I'm going
+ * to assume that this is the "correct" value.
+ *
+ * https://developer.apple.com/documentation/coreservices/1443980-fseventstreamcreate
+ */
 
 int fsm_listen__ctor(struct fsmonitor_daemon_state *state)
 {
+	FSEventStreamCreateFlags flags = kFSEventStreamCreateFlagNoDefer |
+		kFSEventStreamCreateFlagWatchRoot |
+		kFSEventStreamCreateFlagFileEvents;
+	FSEventStreamContext ctx = {
+		0,
+		state,
+		NULL,
+		NULL,
+		NULL
+	};
+	struct fsmonitor_daemon_backend_data *data;
+	const void *dir_array[2];
+
+	CALLOC_ARRAY(data, 1);
+	state->backend_data = data;
+
+	data->cfsr_worktree_path = CFStringCreateWithCString(
+		NULL, state->path_worktree_watch.buf, kCFStringEncodingUTF8);
+	dir_array[data->nr_paths_watching++] = data->cfsr_worktree_path;
+
+	if (state->nr_paths_watching > 1) {
+		data->cfsr_gitdir_path = CFStringCreateWithCString(
+			NULL, state->path_gitdir_watch.buf,
+			kCFStringEncodingUTF8);
+		dir_array[data->nr_paths_watching++] = data->cfsr_gitdir_path;
+	}
+
+	data->cfar_paths_to_watch = CFArrayCreate(NULL, dir_array,
+						  data->nr_paths_watching,
+						  NULL);
+	data->stream = FSEventStreamCreate(NULL, fsevent_callback, &ctx,
+					   data->cfar_paths_to_watch,
+					   kFSEventStreamEventIdSinceNow,
+					   0.001, flags);
+	if (data->stream == NULL)
+		goto failed;
+
+	/*
+	 * `data->rl` needs to be set inside the listener thread.
+	 */
+
+	return 0;
+
+failed:
+	error(_("Unable to create FSEventStream."));
+
+	FREE_AND_NULL(state->backend_data);
 	return -1;
 }
 
 void fsm_listen__dtor(struct fsmonitor_daemon_state *state)
 {
+	struct fsmonitor_daemon_backend_data *data;
+
+	if (!state || !state->backend_data)
+		return;
+
+	data = state->backend_data;
+
+	if (data->stream) {
+		if (data->stream_started)
+			FSEventStreamStop(data->stream);
+		if (data->stream_scheduled)
+			FSEventStreamInvalidate(data->stream);
+		FSEventStreamRelease(data->stream);
+	}
+
+	FREE_AND_NULL(state->backend_data);
 }
 
 void fsm_listen__stop_async(struct fsmonitor_daemon_state *state)
 {
+	struct fsmonitor_daemon_backend_data *data;
+
+	data = state->backend_data;
+	data->shutdown_style = SHUTDOWN_EVENT;
+
+	CFRunLoopStop(data->rl);
 }
 
 void fsm_listen__loop(struct fsmonitor_daemon_state *state)
 {
+	struct fsmonitor_daemon_backend_data *data;
+
+	data = state->backend_data;
+
+	data->rl = CFRunLoopGetCurrent();
+
+	FSEventStreamScheduleWithRunLoop(data->stream, data->rl, kCFRunLoopDefaultMode);
+	data->stream_scheduled = 1;
+
+	if (!FSEventStreamStart(data->stream)) {
+		error(_("Failed to start the FSEventStream"));
+		goto force_error_stop_without_loop;
+	}
+	data->stream_started = 1;
+
+	CFRunLoopRun();
+
+	switch (data->shutdown_style) {
+	case FORCE_ERROR_STOP:
+		state->error_code = -1;
+		/* fall thru */
+	case FORCE_SHUTDOWN:
+		ipc_server_stop_async(state->ipc_server_data);
+		/* fall thru */
+	case SHUTDOWN_EVENT:
+	default:
+		break;
+	}
+	return;
+
+force_error_stop_without_loop:
+	state->error_code = -1;
+	ipc_server_stop_async(state->ipc_server_data);
+	return;
 }
-- 
gitgitgadget

[PATCH v8 18/30] fsmonitor--daemon: implement handle_client callback

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach fsmonitor--daemon to respond to IPC requests from client
Git processes and respond with a list of modified pathnames
relative to the provided token.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 builtin/fsmonitor--daemon.c | 311 +++++++++++++++++++++++++++++++++++-
 1 file changed, 309 insertions(+), 2 deletions(-)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index 69312119b07..eafaafb45b1 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -7,6 +7,7 @@
 #include "fsmonitor--daemon.h"
 #include "simple-ipc.h"
 #include "khash.h"
+#include "pkt-line.h"
 
 static const char * const builtin_fsmonitor__daemon_usage[] = {
 	N_("git fsmonitor--daemon start [<options>]"),
@@ -364,6 +365,310 @@ void fsmonitor_force_resync(struct fsmonitor_daemon_state *state)
 	pthread_mutex_unlock(&state->main_lock);
 }
 
+/*
+ * Format an opaque token string to send to the client.
+ */
+static void with_lock__format_response_token(
+	struct strbuf *response_token,
+	const struct strbuf *response_token_id,
+	const struct fsmonitor_batch *batch)
+{
+	/* assert current thread holding state->main_lock */
+
+	strbuf_reset(response_token);
+	strbuf_addf(response_token, "builtin:%s:%"PRIu64,
+		    response_token_id->buf, batch->batch_seq_nr);
+}
+
+/*
+ * Parse an opaque token from the client.
+ * Returns -1 on error.
+ */
+static int fsmonitor_parse_client_token(const char *buf_token,
+					struct strbuf *requested_token_id,
+					uint64_t *seq_nr)
+{
+	const char *p;
+	char *p_end;
+
+	strbuf_reset(requested_token_id);
+	*seq_nr = 0;
+
+	if (!skip_prefix(buf_token, "builtin:", &p))
+		return -1;
+
+	while (*p && *p != ':')
+		strbuf_addch(requested_token_id, *p++);
+	if (!*p++)
+		return -1;
+
+	*seq_nr = (uint64_t)strtoumax(p, &p_end, 10);
+	if (*p_end)
+		return -1;
+
+	return 0;
+}
+
+KHASH_INIT(str, const char *, int, 0, kh_str_hash_func, kh_str_hash_equal)
+
+static int do_handle_client(struct fsmonitor_daemon_state *state,
+			    const char *command,
+			    ipc_server_reply_cb *reply,
+			    struct ipc_server_reply_data *reply_data)
+{
+	struct fsmonitor_token_data *token_data = NULL;
+	struct strbuf response_token = STRBUF_INIT;
+	struct strbuf requested_token_id = STRBUF_INIT;
+	struct strbuf payload = STRBUF_INIT;
+	uint64_t requested_oldest_seq_nr = 0;
+	uint64_t total_response_len = 0;
+	const char *p;
+	const struct fsmonitor_batch *batch_head;
+	const struct fsmonitor_batch *batch;
+	intmax_t count = 0, duplicates = 0;
+	kh_str_t *shown;
+	int hash_ret;
+	int do_trivial = 0;
+	int do_flush = 0;
+
+	/*
+	 * We expect `command` to be of the form:
+	 *
+	 * <command> := quit NUL
+	 *            | flush NUL
+	 *            | <V1-time-since-epoch-ns> NUL
+	 *            | <V2-opaque-fsmonitor-token> NUL
+	 */
+
+	if (!strcmp(command, "quit")) {
+		/*
+		 * A client has requested over the socket/pipe that the
+		 * daemon shutdown.
+		 *
+		 * Tell the IPC thread pool to shutdown (which completes
+		 * the await in the main thread (which can stop the
+		 * fsmonitor listener thread)).
+		 *
+		 * There is no reply to the client.
+		 */
+		return SIMPLE_IPC_QUIT;
+
+	} else if (!strcmp(command, "flush")) {
+		/*
+		 * Flush all of our cached data and generate a new token
+		 * just like if we lost sync with the filesystem.
+		 *
+		 * Then send a trivial response using the new token.
+		 */
+		do_flush = 1;
+		do_trivial = 1;
+
+	} else if (!skip_prefix(command, "builtin:", &p)) {
+		/* assume V1 timestamp or garbage */
+
+		char *p_end;
+
+		strtoumax(command, &p_end, 10);
+		trace_printf_key(&trace_fsmonitor,
+				 ((*p_end) ?
+				  "fsmonitor: invalid command line '%s'" :
+				  "fsmonitor: unsupported V1 protocol '%s'"),
+				 command);
+		do_trivial = 1;
+
+	} else {
+		/* We have "builtin:*" */
+		if (fsmonitor_parse_client_token(command, &requested_token_id,
+						 &requested_oldest_seq_nr)) {
+			trace_printf_key(&trace_fsmonitor,
+					 "fsmonitor: invalid V2 protocol token '%s'",
+					 command);
+			do_trivial = 1;
+
+		} else {
+			/*
+			 * We have a V2 valid token:
+			 *     "builtin:<token_id>:<seq_nr>"
+			 */
+		}
+	}
+
+	pthread_mutex_lock(&state->main_lock);
+
+	if (!state->current_token_data)
+		BUG("fsmonitor state does not have a current token");
+
+	if (do_flush)
+		with_lock__do_force_resync(state);
+
+	/*
+	 * We mark the current head of the batch list as "pinned" so
+	 * that the listener thread will treat this item as read-only
+	 * (and prevent any more paths from being added to it) from
+	 * now on.
+	 */
+	token_data = state->current_token_data;
+	batch_head = token_data->batch_head;
+	((struct fsmonitor_batch *)batch_head)->pinned_time = time(NULL);
+
+	/*
+	 * FSMonitor Protocol V2 requires that we send a response header
+	 * with a "new current token" and then all of the paths that changed
+	 * since the "requested token".  We send the seq_nr of the just-pinned
+	 * head batch so that future requests from a client will be relative
+	 * to it.
+	 */
+	with_lock__format_response_token(&response_token,
+					 &token_data->token_id, batch_head);
+
+	reply(reply_data, response_token.buf, response_token.len + 1);
+	total_response_len += response_token.len + 1;
+
+	trace2_data_string("fsmonitor", the_repository, "response/token",
+			   response_token.buf);
+	trace_printf_key(&trace_fsmonitor, "response token: %s",
+			 response_token.buf);
+
+	if (!do_trivial) {
+		if (strcmp(requested_token_id.buf, token_data->token_id.buf)) {
+			/*
+			 * The client last spoke to a different daemon
+			 * instance -OR- the daemon had to resync with
+			 * the filesystem (and lost events), so reject.
+			 */
+			trace2_data_string("fsmonitor", the_repository,
+					   "response/token", "different");
+			do_trivial = 1;
+
+		} else if (requested_oldest_seq_nr <
+			   token_data->batch_tail->batch_seq_nr) {
+			/*
+			 * The client wants older events than we have for
+			 * this token_id.  This means that the end of our
+			 * batch list was truncated and we cannot give the
+			 * client a complete snapshot relative to their
+			 * request.
+			 */
+			trace_printf_key(&trace_fsmonitor,
+					 "client requested truncated data");
+			do_trivial = 1;
+		}
+	}
+
+	if (do_trivial) {
+		pthread_mutex_unlock(&state->main_lock);
+
+		reply(reply_data, "/", 2);
+
+		trace2_data_intmax("fsmonitor", the_repository,
+				   "response/trivial", 1);
+
+		goto cleanup;
+	}
+
+	/*
+	 * We're going to hold onto a pointer to the current
+	 * token-data while we walk the list of batches of files.
+	 * During this time, we will NOT be under the lock.
+	 * So we ref-count it.
+	 *
+	 * This allows the listener thread to continue prepending
+	 * new batches of items to the token-data (which we'll ignore).
+	 *
+	 * AND it allows the listener thread to do a token-reset
+	 * (and install a new `current_token_data`).
+	 */
+	token_data->client_ref_count++;
+
+	pthread_mutex_unlock(&state->main_lock);
+
+	/*
+	 * The client request is relative to the token that they sent,
+	 * so walk the batch list backwards from the current head back
+	 * to the batch (sequence number) they named.
+	 *
+	 * We use khash to de-dup the list of pathnames.
+	 *
+	 * NEEDSWORK: each batch contains a list of interned strings,
+	 * so we only need to do pointer comparisons here to build the
+	 * hash table.  Currently, we're still comparing the string
+	 * values.
+	 */
+	shown = kh_init_str();
+	for (batch = batch_head;
+	     batch && batch->batch_seq_nr > requested_oldest_seq_nr;
+	     batch = batch->next) {
+		size_t k;
+
+		for (k = 0; k < batch->nr; k++) {
+			const char *s = batch->interned_paths[k];
+			size_t s_len;
+
+			if (kh_get_str(shown, s) != kh_end(shown))
+				duplicates++;
+			else {
+				kh_put_str(shown, s, &hash_ret);
+
+				trace_printf_key(&trace_fsmonitor,
+						 "send[%"PRIuMAX"]: %s",
+						 count, s);
+
+				/* Each path gets written with a trailing NUL */
+				s_len = strlen(s) + 1;
+
+				if (payload.len + s_len >=
+				    LARGE_PACKET_DATA_MAX) {
+					reply(reply_data, payload.buf,
+					      payload.len);
+					total_response_len += payload.len;
+					strbuf_reset(&payload);
+				}
+
+				strbuf_add(&payload, s, s_len);
+				count++;
+			}
+		}
+	}
+
+	if (payload.len) {
+		reply(reply_data, payload.buf, payload.len);
+		total_response_len += payload.len;
+	}
+
+	kh_release_str(shown);
+
+	pthread_mutex_lock(&state->main_lock);
+
+	if (token_data->client_ref_count > 0)
+		token_data->client_ref_count--;
+
+	if (token_data->client_ref_count == 0) {
+		if (token_data != state->current_token_data) {
+			/*
+			 * The listener thread did a token-reset while we were
+			 * walking the batch list.  Therefore, this token is
+			 * stale and can be discarded completely.  If we are
+			 * the last reader thread using this token, we own
+			 * that work.
+			 */
+			fsmonitor_free_token_data(token_data);
+		}
+	}
+
+	pthread_mutex_unlock(&state->main_lock);
+
+	trace2_data_intmax("fsmonitor", the_repository, "response/length", total_response_len);
+	trace2_data_intmax("fsmonitor", the_repository, "response/count/files", count);
+	trace2_data_intmax("fsmonitor", the_repository, "response/count/duplicates", duplicates);
+
+cleanup:
+	strbuf_release(&response_token);
+	strbuf_release(&requested_token_id);
+	strbuf_release(&payload);
+
+	return 0;
+}
+
 static ipc_server_application_cb handle_client;
 
 static int handle_client(void *data,
@@ -371,7 +676,7 @@ static int handle_client(void *data,
 			 ipc_server_reply_cb *reply,
 			 struct ipc_server_reply_data *reply_data)
 {
-	/* struct fsmonitor_daemon_state *state = data; */
+	struct fsmonitor_daemon_state *state = data;
 	int result;
 
 	/*
@@ -382,10 +687,12 @@ static int handle_client(void *data,
 	if (command_len != strlen(command))
 		BUG("FSMonitor assumes text messages");
 
+	trace_printf_key(&trace_fsmonitor, "requested token: %s", command);
+
 	trace2_region_enter("fsmonitor", "handle_client", the_repository);
 	trace2_data_string("fsmonitor", the_repository, "request", command);
 
-	result = 0; /* TODO Do something here. */
+	result = do_handle_client(state, command, reply, reply_data);
 
 	trace2_region_leave("fsmonitor", "handle_client", the_repository);
 
-- 
gitgitgadget

[PATCH v8 19/30] help: include fsmonitor--daemon feature flag in version info

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Add the "feature: fsmonitor--daemon" message to the output of
`git version --build-options`.

The builtin FSMonitor is only available on certain platforms and
even then only when certain Makefile flags are enabled, so print
a message in the verbose version output when it is available.

This can be used by test scripts for prereq testing.  Granted, tests
could just try `git fsmonitor--daemon status` and look for a 128 exit
code or grep for a "not supported" message on stderr, but these
methods are rather obscure.

The main advantage is that the feature message will automatically
appear in bug reports and other support requests.

This concept was also used during the development of Scalar for
similar reasons.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 help.c        | 4 ++++
 t/test-lib.sh | 7 +++++++
 2 files changed, 11 insertions(+)

diff --git a/help.c b/help.c
index 71444906ddf..9112a51e84b 100644
--- a/help.c
+++ b/help.c
@@ -12,6 +12,7 @@
 #include "refs.h"
 #include "parse-options.h"
 #include "prompt.h"
+#include "fsmonitor-ipc.h"
 
 struct category_description {
 	uint32_t category;
@@ -695,6 +696,9 @@ void get_version_info(struct strbuf *buf, int show_build_options)
 		strbuf_addf(buf, "sizeof-size_t: %d\n", (int)sizeof(size_t));
 		strbuf_addf(buf, "shell-path: %s\n", SHELL_PATH);
 		/* NEEDSWORK: also save and output GIT-BUILD_OPTIONS? */
+
+		if (fsmonitor_ipc__is_supported())
+			strbuf_addstr(buf, "feature: fsmonitor--daemon\n");
 	}
 }
 
diff --git a/t/test-lib.sh b/t/test-lib.sh
index e4716b0b867..5d819c1bc11 100644
--- a/t/test-lib.sh
+++ b/t/test-lib.sh
@@ -1799,3 +1799,10 @@ test_lazy_prereq SHA1 '
 # Tests that verify the scheduler integration must set this locally
 # to avoid errors.
 GIT_TEST_MAINT_SCHEDULER="none:exit 1"
+
+# Does this platform support `git fsmonitor--daemon`
+#
+test_lazy_prereq FSMONITOR_DAEMON '
+	git version --build-options >output &&
+	grep "feature: fsmonitor--daemon" output
+'
-- 
gitgitgadget

[PATCH v8 20/30] t/helper/fsmonitor-client: create IPC client to talk to FSMonitor Daemon

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Create an IPC client to send query and flush commands to the daemon.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 Makefile                         |   1 +
 t/helper/test-fsmonitor-client.c | 116 +++++++++++++++++++++++++++++++
 t/helper/test-tool.c             |   1 +
 t/helper/test-tool.h             |   1 +
 4 files changed, 119 insertions(+)
 create mode 100644 t/helper/test-fsmonitor-client.c

diff --git a/Makefile b/Makefile
index 26567d4f772..daa21bed6c3 100644
--- a/Makefile
+++ b/Makefile
@@ -716,6 +716,7 @@ TEST_BUILTINS_OBJS += test-dump-split-index.o
 TEST_BUILTINS_OBJS += test-dump-untracked-cache.o
 TEST_BUILTINS_OBJS += test-example-decorate.o
 TEST_BUILTINS_OBJS += test-fast-rebase.o
+TEST_BUILTINS_OBJS += test-fsmonitor-client.o
 TEST_BUILTINS_OBJS += test-genrandom.o
 TEST_BUILTINS_OBJS += test-genzeros.o
 TEST_BUILTINS_OBJS += test-getcwd.o
diff --git a/t/helper/test-fsmonitor-client.c b/t/helper/test-fsmonitor-client.c
new file mode 100644
index 00000000000..3062c8a3c2b
--- /dev/null
+++ b/t/helper/test-fsmonitor-client.c
@@ -0,0 +1,116 @@
+/*
+ * test-fsmonitor-client.c: client code to send commands/requests to
+ * a `git fsmonitor--daemon` daemon.
+ */
+
+#include "test-tool.h"
+#include "cache.h"
+#include "parse-options.h"
+#include "fsmonitor-ipc.h"
+
+#ifndef HAVE_FSMONITOR_DAEMON_BACKEND
+int cmd__fsmonitor_client(int argc, const char **argv)
+{
+	die("fsmonitor--daemon not available on this platform");
+}
+#else
+
+/*
+ * Read the `.git/index` to get the last token written to the
+ * FSMonitor Index Extension.
+ */
+static const char *get_token_from_index(void)
+{
+	struct index_state *istate = the_repository->index;
+
+	if (do_read_index(istate, the_repository->index_file, 0) < 0)
+		die("unable to read index file");
+	if (!istate->fsmonitor_last_update)
+		die("index file does not have fsmonitor extension");
+
+	return istate->fsmonitor_last_update;
+}
+
+/*
+ * Send an IPC query to a `git-fsmonitor--daemon` daemon and
+ * ask for the changes since the given token or from the last
+ * token in the index extension.
+ *
+ * This will implicitly start a daemon process if necessary.  The
+ * daemon process will persist after we exit.
+ */
+static int do_send_query(const char *token)
+{
+	struct strbuf answer = STRBUF_INIT;
+	int ret;
+
+	if (!token || !*token)
+		token = get_token_from_index();
+
+	ret = fsmonitor_ipc__send_query(token, &answer);
+	if (ret < 0)
+		die("could not query fsmonitor--daemon");
+
+	write_in_full(1, answer.buf, answer.len);
+	strbuf_release(&answer);
+
+	return 0;
+}
+
+/*
+ * Send a "flush" command to the `git-fsmonitor--daemon` (if running)
+ * and tell it to flush its cache.
+ *
+ * This feature is primarily used by the test suite to simulate a loss of
+ * sync with the filesystem where we miss kernel events.
+ */
+static int do_send_flush(void)
+{
+	struct strbuf answer = STRBUF_INIT;
+	int ret;
+
+	ret = fsmonitor_ipc__send_command("flush", &answer);
+	if (ret)
+		return ret;
+
+	write_in_full(1, answer.buf, answer.len);
+	strbuf_release(&answer);
+
+	return 0;
+}
+
+int cmd__fsmonitor_client(int argc, const char **argv)
+{
+	const char *subcmd;
+	const char *token = NULL;
+
+	const char * const fsmonitor_client_usage[] = {
+		"test-tool fsmonitor-client query [<token>]",
+		"test-tool fsmonitor-client flush",
+		NULL,
+	};
+
+	struct option options[] = {
+		OPT_STRING(0, "token", &token, "token",
+			   "command token to send to the server"),
+		OPT_END()
+	};
+
+	argc = parse_options(argc, argv, NULL, options, fsmonitor_client_usage, 0);
+
+	if (argc != 1)
+		usage_with_options(fsmonitor_client_usage, options);
+
+	subcmd = argv[0];
+
+	setup_git_directory();
+
+	if (!strcmp(subcmd, "query"))
+		return !!do_send_query(token);
+
+	if (!strcmp(subcmd, "flush"))
+		return !!do_send_flush();
+
+	die("Unhandled subcommand: '%s'", subcmd);
+}
+#endif
diff --git a/t/helper/test-tool.c b/t/helper/test-tool.c
index e6ec69cf326..0424f7adf5d 100644
--- a/t/helper/test-tool.c
+++ b/t/helper/test-tool.c
@@ -32,6 +32,7 @@ static struct test_cmd cmds[] = {
 	{ "dump-untracked-cache", cmd__dump_untracked_cache },
 	{ "example-decorate", cmd__example_decorate },
 	{ "fast-rebase", cmd__fast_rebase },
+	{ "fsmonitor-client", cmd__fsmonitor_client },
 	{ "genrandom", cmd__genrandom },
 	{ "genzeros", cmd__genzeros },
 	{ "getcwd", cmd__getcwd },
diff --git a/t/helper/test-tool.h b/t/helper/test-tool.h
index 20756eefdda..c876e8246fb 100644
--- a/t/helper/test-tool.h
+++ b/t/helper/test-tool.h
@@ -23,6 +23,7 @@ int cmd__dump_untracked_cache(int argc, const char **argv);
 int cmd__dump_reftable(int argc, const char **argv);
 int cmd__example_decorate(int argc, const char **argv);
 int cmd__fast_rebase(int argc, const char **argv);
+int cmd__fsmonitor_client(int argc, const char **argv);
 int cmd__genrandom(int argc, const char **argv);
 int cmd__genzeros(int argc, const char **argv);
 int cmd__getcwd(int argc, const char **argv);
-- 
gitgitgadget

[PATCH v8 21/30] t7527: create test for fsmonitor--daemon

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 t/t7527-builtin-fsmonitor.sh | 505 +++++++++++++++++++++++++++++++++++
 1 file changed, 505 insertions(+)
 create mode 100755 t/t7527-builtin-fsmonitor.sh

diff --git a/t/t7527-builtin-fsmonitor.sh b/t/t7527-builtin-fsmonitor.sh
new file mode 100755
index 00000000000..3570bf764aa
--- /dev/null
+++ b/t/t7527-builtin-fsmonitor.sh
@@ -0,0 +1,505 @@
+#!/bin/sh
+
+test_description='built-in file system watcher'
+
+. ./test-lib.sh
+
+if ! test_have_prereq FSMONITOR_DAEMON
+then
+	skip_all="fsmonitor--daemon is not supported on this platform"
+	test_done
+fi
+
+stop_daemon_delete_repo () {
+	r=$1 &&
+	test_might_fail git -C $r fsmonitor--daemon stop &&
+	rm -rf $1
+}
+
+is_value () {
+	test -n "$1" && test "${1::1}" != "-"
+}
+
+start_daemon () {
+	r= &&
+	tf= &&
+	t2= &&
+	tk= &&
+
+	while test "$#" -ne 0
+	do
+		case "$1" in
+		-C)
+			shift;
+			is_value $1 || BUG "error: -C requires value"
+			r="-C $1"
+			shift
+			;;
+		--tf)
+			shift;
+			is_value $1 || BUG "error: --tf requires value"
+			tf="$1"
+			shift
+			;;
+		--t2)
+			shift;
+			is_value $1 || BUG "error: --t2 requires value"
+			t2="$1"
+			shift
+			;;
+		--tk)
+			shift;
+			is_value $1 || BUG "error: --tk requires value"
+			tk="$1"
+			shift
+			;;
+		*)
+			BUG "error: unknown option: '$1'"
+			;;
+		esac
+	done &&
+
+	(
+		if test -n "$tf"
+		then
+			GIT_TRACE_FSMONITOR="$tf"
+			export GIT_TRACE_FSMONITOR
+		fi &&
+
+		if test -n "$t2"
+		then
+			GIT_TRACE2_PERF="$t2"
+			export GIT_TRACE2_PERF
+		fi &&
+
+		if test -n "$tk"
+		then
+			GIT_TEST_FSMONITOR_TOKEN="$tk"
+			export GIT_TEST_FSMONITOR_TOKEN
+		fi &&
+
+		git $r fsmonitor--daemon start &&
+		git $r fsmonitor--daemon status
+	)
+}
+
+# Is a Trace2 data event present with the given catetory and key?
+# We do not care what the value is.
+#
+have_t2_data_event () {
+	c=$1 &&
+	k=$2 &&
+
+	grep -e '"event":"data".*"category":"'"$c"'".*"key":"'"$k"'"'
+}
+
+test_expect_success 'explicit daemon start and stop' '
+	test_when_finished "stop_daemon_delete_repo test_explicit" &&
+
+	git init test_explicit &&
+	start_daemon -C test_explicit &&
+
+	git -C test_explicit fsmonitor--daemon stop &&
+	test_must_fail git -C test_explicit fsmonitor--daemon status
+'
+
+test_expect_success 'implicit daemon start' '
+	test_when_finished "stop_daemon_delete_repo test_implicit" &&
+
+	git init test_implicit &&
+	test_must_fail git -C test_implicit fsmonitor--daemon status &&
+
+	# query will implicitly start the daemon.
+	#
+	# for test-script simplicity, we send a V1 timestamp rather than
+	# a V2 token.  either way, the daemon response to any query contains
+	# a new V2 token.  (the daemon may complain that we sent a V1 request,
+	# but this test case is only concerned with whether the daemon was
+	# implicitly started.)
+
+	GIT_TRACE2_EVENT="$PWD/.git/trace" \
+		test-tool -C test_implicit fsmonitor-client query --token 0 >actual &&
+	nul_to_q <actual >actual.filtered &&
+	grep "builtin:" actual.filtered &&
+
+	# confirm that a daemon was started in the background.
+	#
+	# since the mechanism for starting the background daemon is platform
+	# dependent, just confirm that the foreground command received a
+	# response from the daemon.
+
+	have_t2_data_event fsm_client query/response-length <.git/trace &&
+
+	git -C test_implicit fsmonitor--daemon status &&
+	git -C test_implicit fsmonitor--daemon stop &&
+	test_must_fail git -C test_implicit fsmonitor--daemon status
+'
+
+test_expect_success 'implicit daemon stop (delete .git)' '
+	test_when_finished "stop_daemon_delete_repo test_implicit_1" &&
+
+	git init test_implicit_1 &&
+
+	start_daemon -C test_implicit_1 &&
+
+	# deleting the .git directory will implicitly stop the daemon.
+	rm -rf test_implicit_1/.git &&
+
+	# [1] Create an empty .git directory so that the following Git
+	#     command will stay relative to the `-C` directory.
+	#
+	#     Without this, the Git command will override the requested
+	#     -C argument and crawl out to the containing Git source tree.
+	#     This would make the test result dependent upon whether we
+	#     were using fsmonitor on our development worktree.
+	#
+	sleep 1 &&
+	mkdir test_implicit_1/.git &&
+
+	test_must_fail git -C test_implicit_1 fsmonitor--daemon status
+'
+
+test_expect_success 'implicit daemon stop (rename .git)' '
+	test_when_finished "stop_daemon_delete_repo test_implicit_2" &&
+
+	git init test_implicit_2 &&
+
+	start_daemon -C test_implicit_2 &&
+
+	# renaming the .git directory will implicitly stop the daemon.
+	mv test_implicit_2/.git test_implicit_2/.xxx &&
+
+	# See [1] above.
+	#
+	sleep 1 &&
+	mkdir test_implicit_2/.git &&
+
+	test_must_fail git -C test_implicit_2 fsmonitor--daemon status
+'
+
+test_expect_success 'cannot start multiple daemons' '
+	test_when_finished "stop_daemon_delete_repo test_multiple" &&
+
+	git init test_multiple &&
+
+	start_daemon -C test_multiple &&
+
+	test_must_fail git -C test_multiple fsmonitor--daemon start 2>actual &&
+	grep "fsmonitor--daemon is already running" actual &&
+
+	git -C test_multiple fsmonitor--daemon stop &&
+	test_must_fail git -C test_multiple fsmonitor--daemon status
+'
+
+# These tests use the main repo in the trash directory
+
+test_expect_success 'setup' '
+	>tracked &&
+	>modified &&
+	>delete &&
+	>rename &&
+	mkdir dir1 &&
+	>dir1/tracked &&
+	>dir1/modified &&
+	>dir1/delete &&
+	>dir1/rename &&
+	mkdir dir2 &&
+	>dir2/tracked &&
+	>dir2/modified &&
+	>dir2/delete &&
+	>dir2/rename &&
+	mkdir dirtorename &&
+	>dirtorename/a &&
+	>dirtorename/b &&
+
+	cat >.gitignore <<-\EOF &&
+	.gitignore
+	expect*
+	actual*
+	EOF
+
+	git -c core.fsmonitor=false add . &&
+	test_tick &&
+	git -c core.fsmonitor=false commit -m initial &&
+
+	git config core.fsmonitor true
+'
+
+# The test already explicitly stopped (or tried to stop) the daemon.
+# This is here in case something else fails first.
+#
+redundant_stop_daemon () {
+	test_might_fail git fsmonitor--daemon stop
+}
+
+test_expect_success 'update-index implicitly starts daemon' '
+	test_when_finished redundant_stop_daemon &&
+
+	test_must_fail git fsmonitor--daemon status &&
+
+	GIT_TRACE2_EVENT="$PWD/.git/trace_implicit_1" \
+		git update-index --fsmonitor &&
+
+	git fsmonitor--daemon status &&
+	test_might_fail git fsmonitor--daemon stop &&
+
+	# Confirm that the trace2 log contains a record of the
+	# daemon starting.
+	test_subcommand git fsmonitor--daemon start <.git/trace_implicit_1
+'
+
+test_expect_success 'status implicitly starts daemon' '
+	test_when_finished redundant_stop_daemon &&
+
+	test_must_fail git fsmonitor--daemon status &&
+
+	GIT_TRACE2_EVENT="$PWD/.git/trace_implicit_2" \
+		git status >actual &&
+
+	git fsmonitor--daemon status &&
+	test_might_fail git fsmonitor--daemon stop &&
+
+	# Confirm that the trace2 log contains a record of the
+	# daemon starting.
+	test_subcommand git fsmonitor--daemon start <.git/trace_implicit_2
+'
+
+edit_files () {
+	echo 1 >modified &&
+	echo 2 >dir1/modified &&
+	echo 3 >dir2/modified &&
+	>dir1/untracked
+}
+
+delete_files () {
+	rm -f delete &&
+	rm -f dir1/delete &&
+	rm -f dir2/delete
+}
+
+create_files () {
+	echo 1 >new &&
+	echo 2 >dir1/new &&
+	echo 3 >dir2/new
+}
+
+rename_files () {
+	mv rename renamed &&
+	mv dir1/rename dir1/renamed &&
+	mv dir2/rename dir2/renamed
+}
+
+file_to_directory () {
+	rm -f delete &&
+	mkdir delete &&
+	echo 1 >delete/new
+}
+
+directory_to_file () {
+	rm -rf dir1 &&
+	echo 1 >dir1
+}
+
+# The next few test cases confirm that our fsmonitor daemon sees each type
+# of OS filesystem notification that we care about.  At this layer we just
+# ensure we are getting the OS notifications and do not try to confirm what
+# is reported by `git status`.
+#
+# We run a simple query after modifying the filesystem just to introduce
+# a bit of a delay so that the trace logging from the daemon has time to
+# get flushed to disk.
+#
+# We `reset` and `clean` at the bottom of each test (and before stopping the
+# daemon) because these commands might implicitly restart the daemon.
+
+clean_up_repo_and_stop_daemon () {
+	git reset --hard HEAD &&
+	git clean -fd &&
+	test_might_fail git fsmonitor--daemon stop &&
+	rm -f .git/trace
+}
+
+test_expect_success 'edit some files' '
+	test_when_finished clean_up_repo_and_stop_daemon &&
+
+	start_daemon --tf "$PWD/.git/trace" &&
+
+	edit_files &&
+
+	test-tool fsmonitor-client query --token 0 &&
+
+	grep "^event: dir1/modified$"  .git/trace &&
+	grep "^event: dir2/modified$"  .git/trace &&
+	grep "^event: modified$"       .git/trace &&
+	grep "^event: dir1/untracked$" .git/trace
+'
+
+test_expect_success 'create some files' '
+	test_when_finished clean_up_repo_and_stop_daemon &&
+
+	start_daemon --tf "$PWD/.git/trace" &&
+
+	create_files &&
+
+	test-tool fsmonitor-client query --token 0 &&
+
+	grep "^event: dir1/new$" .git/trace &&
+	grep "^event: dir2/new$" .git/trace &&
+	grep "^event: new$"      .git/trace
+'
+
+test_expect_success 'delete some files' '
+	test_when_finished clean_up_repo_and_stop_daemon &&
+
+	start_daemon --tf "$PWD/.git/trace" &&
+
+	delete_files &&
+
+	test-tool fsmonitor-client query --token 0 &&
+
+	grep "^event: dir1/delete$" .git/trace &&
+	grep "^event: dir2/delete$" .git/trace &&
+	grep "^event: delete$"      .git/trace
+'
+
+test_expect_success 'rename some files' '
+	test_when_finished clean_up_repo_and_stop_daemon &&
+
+	start_daemon --tf "$PWD/.git/trace" &&
+
+	rename_files &&
+
+	test-tool fsmonitor-client query --token 0 &&
+
+	grep "^event: dir1/rename$"  .git/trace &&
+	grep "^event: dir2/rename$"  .git/trace &&
+	grep "^event: rename$"       .git/trace &&
+	grep "^event: dir1/renamed$" .git/trace &&
+	grep "^event: dir2/renamed$" .git/trace &&
+	grep "^event: renamed$"      .git/trace
+'
+
+test_expect_success 'rename directory' '
+	test_when_finished clean_up_repo_and_stop_daemon &&
+
+	start_daemon --tf "$PWD/.git/trace" &&
+
+	mv dirtorename dirrenamed &&
+
+	test-tool fsmonitor-client query --token 0 &&
+
+	grep "^event: dirtorename/*$" .git/trace &&
+	grep "^event: dirrenamed/*$"  .git/trace
+'
+
+test_expect_success 'file changes to directory' '
+	test_when_finished clean_up_repo_and_stop_daemon &&
+
+	start_daemon --tf "$PWD/.git/trace" &&
+
+	file_to_directory &&
+
+	test-tool fsmonitor-client query --token 0 &&
+
+	grep "^event: delete$"     .git/trace &&
+	grep "^event: delete/new$" .git/trace
+'
+
+test_expect_success 'directory changes to a file' '
+	test_when_finished clean_up_repo_and_stop_daemon &&
+
+	start_daemon --tf "$PWD/.git/trace" &&
+
+	directory_to_file &&
+
+	test-tool fsmonitor-client query --token 0 &&
+
+	grep "^event: dir1$" .git/trace
+'
+
+# The next few test cases exercise the token-resync code.  When filesystem
+# drops events (because of filesystem velocity or because the daemon isn't
+# polling fast enough), we need to discard the cached data (relative to the
+# current token) and start collecting events under a new token.
+#
+# the 'test-tool fsmonitor-client flush' command can be used to send a
+# "flush" message to a running daemon and ask it to do a flush/resync.
+
+test_expect_success 'flush cached data' '
+	test_when_finished "stop_daemon_delete_repo test_flush" &&
+
+	git init test_flush &&
+
+	start_daemon -C test_flush --tf "$PWD/.git/trace_daemon" --tk true &&
+
+	# The daemon should have an initial token with no events in _0 and
+	# then a few (probably platform-specific number of) events in _1.
+	# These should both have the same <token_id>.
+
+	test-tool -C test_flush fsmonitor-client query --token "builtin:test_00000001:0" >actual_0 &&
+	nul_to_q <actual_0 >actual_q0 &&
+
+	>test_flush/file_1 &&
+	>test_flush/file_2 &&
+
+	test-tool -C test_flush fsmonitor-client query --token "builtin:test_00000001:0" >actual_1 &&
+	nul_to_q <actual_1 >actual_q1 &&
+
+	grep "file_1" actual_q1 &&
+
+	# Force a flush.  This will change the <token_id>, reset the <seq_nr>, and
+	# flush the file data.  Then create some events and ensure that the file
+	# again appears in the cache.  It should have the new <token_id>.
+
+	test-tool -C test_flush fsmonitor-client flush >flush_0 &&
+	nul_to_q <flush_0 >flush_q0 &&
+	grep "^builtin:test_00000002:0Q/Q$" flush_q0 &&
+
+	test-tool -C test_flush fsmonitor-client query --token "builtin:test_00000002:0" >actual_2 &&
+	nul_to_q <actual_2 >actual_q2 &&
+
+	grep "^builtin:test_00000002:0Q$" actual_q2 &&
+
+	>test_flush/file_3 &&
+
+	test-tool -C test_flush fsmonitor-client query --token "builtin:test_00000002:0" >actual_3 &&
+	nul_to_q <actual_3 >actual_q3 &&
+
+	grep "file_3" actual_q3
+'
+
+# The next few test cases create repos where the .git directory is NOT
+# inside the one of the working directory.  That is, where .git is a file
+# that points to a directory elsewhere.  This happens for submodules and
+# non-primary worktrees.
+
+test_expect_success 'setup worktree base' '
+	git init wt-base &&
+	echo 1 >wt-base/file1 &&
+	git -C wt-base add file1 &&
+	git -C wt-base commit -m "c1"
+'
+
+test_expect_success 'worktree with .git file' '
+	git -C wt-base worktree add ../wt-secondary &&
+
+	start_daemon -C wt-secondary \
+		--tf "$PWD/trace_wt_secondary" \
+		--t2 "$PWD/trace2_wt_secondary" &&
+
+	git -C wt-secondary fsmonitor--daemon stop &&
+	test_must_fail git -C wt-secondary fsmonitor--daemon status
+'
+
+# NEEDSWORK: Repeat one of the "edit" tests on wt-secondary and
+# confirm that we get the same events and behavior -- that is, that
+# fsmonitor--daemon correctly watches BOTH the working directory and
+# the external GITDIR directory and behaves the same as when ".git"
+# is a directory inside the working directory.
+
+test_expect_success 'cleanup worktrees' '
+	stop_daemon_delete_repo wt-secondary &&
+	stop_daemon_delete_repo wt-base
+'
+
+test_done
-- 
gitgitgadget

[PATCH v8 22/30] t/perf: avoid copying builtin fsmonitor files into test repo

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Do not copy any of the various fsmonitor--daemon files from the .git
directory of the (GIT_PREF_REPO or GIT_PERF_LARGE_REPO) source repo
into the test's trash directory.

When perf tests start, they copy the contents of the source repo into
the test's trash directory.  If fsmonitor is running in the source repo,
there may be control files, such as the IPC socket and/or fsmonitor
cookie files.  These should not be copied into the test repo.

Unix domain sockets cannot be copied in the manner used by the test
setup, so if present, the test setup fails.

Cookie files are harmless, but we should avoid them.

The builtin fsmonitor keeps all such control files/sockets in
.git/fsmonitor--daemon*, so it is simple to exclude them.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 t/perf/perf-lib.sh | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/t/perf/perf-lib.sh b/t/perf/perf-lib.sh
index 407252bac70..932105cd12c 100644
--- a/t/perf/perf-lib.sh
+++ b/t/perf/perf-lib.sh
@@ -78,7 +78,7 @@ test_perf_copy_repo_contents () {
 	for stuff in "$1"/*
 	do
 		case "$stuff" in
-		*/objects|*/hooks|*/config|*/commondir|*/gitdir|*/worktrees)
+		*/objects|*/hooks|*/config|*/commondir|*/gitdir|*/worktrees|*/fsmonitor--daemon*)
 			;;
 		*)
 			cp -R "$stuff" "$repo/.git/" || exit 1
-- 
gitgitgadget

[PATCH v8 23/30] t/helper/test-chmtime: skip directories on Windows

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach `test-tool.exe chmtime` to ignore errors when setting the mtime
on a directory on Windows.

NEEDSWORK: The Windows version of `utime()` (aka `mingw_utime()`) does
not properly handle directories because it uses `_wopen()`.  It should
be converted to using `CreateFileW()` and backup semantics at a minimum.
Since I'm already in the middle of a large patch series, I did not want
to destabilize other callers of `utime()` right now.  The problem has
only been observed in the t/perf/p7519 test when the test repo contains
an empty directory on disk.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 t/helper/test-chmtime.c | 15 +++++++++++++++
 1 file changed, 15 insertions(+)

diff --git a/t/helper/test-chmtime.c b/t/helper/test-chmtime.c
index 524b55ca496..dc28890a183 100644
--- a/t/helper/test-chmtime.c
+++ b/t/helper/test-chmtime.c
@@ -134,6 +134,21 @@ int cmd__chmtime(int argc, const char **argv)
 		}
 
 		if (utb.modtime != sb.st_mtime && utime(argv[i], &utb) < 0) {
+#ifdef GIT_WINDOWS_NATIVE
+			if (S_ISDIR(sb.st_mode)) {
+				/*
+				 * NEEDSWORK: The Windows version of `utime()`
+				 * (aka `mingw_utime()`) does not correctly
+				 * handle directory arguments, since it uses
+				 * `_wopen()`.  Ignore it for now since this
+				 * is just a test.
+				 */
+				fprintf(stderr,
+					("Failed to modify time on directory %s. "
+					 "Skipping\n"), argv[i]);
+				continue;
+			}
+#endif
 			fprintf(stderr, "Failed to modify time on %s: %s\n",
 			        argv[i], strerror(errno));
 			return 1;
-- 
gitgitgadget

[PATCH v8 24/30] t/perf/p7519: fix coding style

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 t/perf/p7519-fsmonitor.sh | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/t/perf/p7519-fsmonitor.sh b/t/perf/p7519-fsmonitor.sh
index c8be58f3c76..5241eb6c4e5 100755
--- a/t/perf/p7519-fsmonitor.sh
+++ b/t/perf/p7519-fsmonitor.sh
@@ -72,7 +72,7 @@ then
 	fi
 fi
 
-trace_start() {
+trace_start () {
 	if test -n "$GIT_PERF_7519_TRACE"
 	then
 		name="$1"
@@ -91,7 +91,7 @@ trace_start() {
 	fi
 }
 
-trace_stop() {
+trace_stop () {
 	if test -n "$GIT_PERF_7519_TRACE"
 	then
 		unset GIT_TRACE2_PERF
@@ -133,7 +133,7 @@ test_expect_success "one time repo setup" '
 	fi
 '
 
-setup_for_fsmonitor() {
+setup_for_fsmonitor () {
 	# set INTEGRATION_SCRIPT depending on the environment
 	if test -n "$INTEGRATION_PATH"
 	then
@@ -173,7 +173,7 @@ test_perf_w_drop_caches () {
 	test_perf "$@"
 }
 
-test_fsmonitor_suite() {
+test_fsmonitor_suite () {
 	if test -n "$INTEGRATION_SCRIPT"; then
 		DESC="fsmonitor=$(basename $INTEGRATION_SCRIPT)"
 	else
-- 
gitgitgadget

[PATCH v8 25/30] t/perf/p7519: speed up test on Windows

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Change p7519 to use `test_seq` and `xargs` rather than a `for` loop
to touch thousands of files.  This takes minutes off of test runs
on Windows because of process creation overhead.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 t/perf/p7519-fsmonitor.sh | 24 ++++++++++++++++--------
 1 file changed, 16 insertions(+), 8 deletions(-)

diff --git a/t/perf/p7519-fsmonitor.sh b/t/perf/p7519-fsmonitor.sh
index 5241eb6c4e5..a6c2a910e70 100755
--- a/t/perf/p7519-fsmonitor.sh
+++ b/t/perf/p7519-fsmonitor.sh
@@ -98,6 +98,13 @@ trace_stop () {
 	fi
 }
 
+touch_files () {
+	n=$1 &&
+	d="$n"_files &&
+
+	(cd $d && test_seq 1 $n | xargs touch )
+}
+
 test_expect_success "one time repo setup" '
 	# set untrackedCache depending on the environment
 	if test -n "$GIT_PERF_7519_UNTRACKED_CACHE"
@@ -119,10 +126,11 @@ test_expect_success "one time repo setup" '
 	fi &&
 
 	mkdir 1_file 10_files 100_files 1000_files 10000_files &&
-	for i in $(test_seq 1 10); do touch 10_files/$i || return 1; done &&
-	for i in $(test_seq 1 100); do touch 100_files/$i || return 1; done &&
-	for i in $(test_seq 1 1000); do touch 1000_files/$i || return 1; done &&
-	for i in $(test_seq 1 10000); do touch 10000_files/$i || return 1; done &&
+	: 1_file directory should be left empty &&
+	touch_files 10 &&
+	touch_files 100 &&
+	touch_files 1000 &&
+	touch_files 10000 &&
 	git add 1_file 10_files 100_files 1000_files 10000_files &&
 	git commit -qm "Add files" &&
 
@@ -199,15 +207,15 @@ test_fsmonitor_suite () {
 
 	# Update the mtimes on upto 100k files to make status think
 	# that they are dirty.  For simplicity, omit any files with
-	# LFs (i.e. anything that ls-files thinks it needs to dquote).
-	# Then fully backslash-quote the paths to capture any
-	# whitespace so that they pass thru xargs properly.
+	# LFs (i.e. anything that ls-files thinks it needs to dquote)
+	# and any files with whitespace so that they pass thru xargs
+	# properly.
 	#
 	test_perf_w_drop_caches "status (dirty) ($DESC)" '
 		git ls-files | \
 			head -100000 | \
 			grep -v \" | \
-			sed '\''s/\(.\)/\\\1/g'\'' | \
+			grep -v " ." | \
 			xargs test-tool chmtime -300 &&
 		git status
 	'
-- 
gitgitgadget

[PATCH v8 26/30] t/perf/p7519: add fsmonitor--daemon test cases

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Repeat all of the fsmonitor perf tests using `git fsmonitor--daemon` and
the "Simple IPC" interface.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 t/perf/p7519-fsmonitor.sh | 38 ++++++++++++++++++++++++++++++++++----
 1 file changed, 34 insertions(+), 4 deletions(-)

diff --git a/t/perf/p7519-fsmonitor.sh b/t/perf/p7519-fsmonitor.sh
index a6c2a910e70..0b9129ca7bc 100755
--- a/t/perf/p7519-fsmonitor.sh
+++ b/t/perf/p7519-fsmonitor.sh
@@ -141,7 +141,7 @@ test_expect_success "one time repo setup" '
 	fi
 '
 
-setup_for_fsmonitor () {
+setup_for_fsmonitor_hook () {
 	# set INTEGRATION_SCRIPT depending on the environment
 	if test -n "$INTEGRATION_PATH"
 	then
@@ -182,7 +182,11 @@ test_perf_w_drop_caches () {
 }
 
 test_fsmonitor_suite () {
-	if test -n "$INTEGRATION_SCRIPT"; then
+	if test -n "$USE_FSMONITOR_DAEMON"
+	then
+		DESC="builtin fsmonitor--daemon"
+	elif test -n "$INTEGRATION_SCRIPT"
+	then
 		DESC="fsmonitor=$(basename $INTEGRATION_SCRIPT)"
 	else
 		DESC="fsmonitor=disabled"
@@ -261,11 +265,11 @@ test_fsmonitor_suite () {
 trace_start fsmonitor-watchman
 if test -n "$GIT_PERF_7519_FSMONITOR"; then
 	for INTEGRATION_PATH in $GIT_PERF_7519_FSMONITOR; do
-		test_expect_success "setup for fsmonitor $INTEGRATION_PATH" 'setup_for_fsmonitor'
+		test_expect_success "setup for fsmonitor $INTEGRATION_PATH" 'setup_for_fsmonitor_hook'
 		test_fsmonitor_suite
 	done
 else
-	test_expect_success "setup for fsmonitor" 'setup_for_fsmonitor'
+	test_expect_success "setup for fsmonitor hook" 'setup_for_fsmonitor_hook'
 	test_fsmonitor_suite
 fi
 
@@ -293,4 +297,30 @@ test_expect_success "setup without fsmonitor" '
 test_fsmonitor_suite
 trace_stop
 
+#
+# Run a full set of perf tests using the built-in fsmonitor--daemon.
+# It does not use the Hook API, so it has a different setup.
+# Explicitly start the daemon here and before we start client commands
+# so that we can later add custom tracing.
+#
+if test_have_prereq FSMONITOR_DAEMON
+then
+	USE_FSMONITOR_DAEMON=t
+
+	test_expect_success "setup for builtin fsmonitor" '
+		trace_start fsmonitor--daemon--server &&
+		git fsmonitor--daemon start &&
+
+		trace_start fsmonitor--daemon--client &&
+
+		git config core.fsmonitor true &&
+		git update-index --fsmonitor
+	'
+
+	test_fsmonitor_suite
+
+	git fsmonitor--daemon stop
+	trace_stop
+fi
+
 test_done
-- 
gitgitgadget

[PATCH v8 27/30] fsmonitor--daemon: periodically truncate list of modified files

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach fsmonitor--daemon to periodically truncate the list of
modified files to save some memory.

Clients will ask for the set of changes relative to a token that they
found in the FSMN index extension in the index.  (This token is like a
point in time, but different).  Clients will then update the index to
contain the response token (so that subsequent commands will be
relative to this new token).

Therefore, the daemon can gradually truncate the in-memory list of
changed paths as they become obsolete (older than the previous token).
Since we may have multiple clients making concurrent requests with a
skew of tokens and clients may be racing to the talk to the daemon,
we lazily truncate the list.

We introduce a 5 minute delay and truncate batches 5 minutes after
they are considered obsolete.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 builtin/fsmonitor--daemon.c | 88 +++++++++++++++++++++++++++++++++++++
 1 file changed, 88 insertions(+)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index eafaafb45b1..ab9cc09f7ce 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -312,6 +312,75 @@ static void fsmonitor_batch__combine(struct fsmonitor_batch *batch_dest,
 			batch_src->interned_paths[k];
 }
 
+/*
+ * To keep the batch list from growing unbounded in response to filesystem
+ * activity, we try to truncate old batches from the end of the list as
+ * they become irrelevant.
+ *
+ * We assume that the .git/index will be updated with the most recent token
+ * any time the index is updated.  And future commands will only ask for
+ * recent changes *since* that new token.  So as tokens advance into the
+ * future, older batch items will never be requested/needed.  So we can
+ * truncate them without loss of functionality.
+ *
+ * However, multiple commands may be talking to the daemon concurrently
+ * or perform a slow command, so a little "token skew" is possible.
+ * Therefore, we want this to be a little bit lazy and have a generous
+ * delay.
+ *
+ * The current reader thread walked backwards in time from `token->batch_head`
+ * back to `batch_marker` somewhere in the middle of the batch list.
+ *
+ * Let's walk backwards in time from that marker an arbitrary delay
+ * and truncate the list there.  Note that these timestamps are completely
+ * artificial (based on when we pinned the batch item) and not on any
+ * filesystem activity.
+ *
+ * Return the obsolete portion of the list after we have removed it from
+ * the official list so that the caller can free it after leaving the lock.
+ */
+#define MY_TIME_DELAY_SECONDS (5 * 60) /* seconds */
+
+static struct fsmonitor_batch *with_lock__truncate_old_batches(
+	struct fsmonitor_daemon_state *state,
+	const struct fsmonitor_batch *batch_marker)
+{
+	/* assert current thread holding state->main_lock */
+
+	const struct fsmonitor_batch *batch;
+	struct fsmonitor_batch *remainder;
+
+	if (!batch_marker)
+		return NULL;
+
+	trace_printf_key(&trace_fsmonitor, "Truncate: mark (%"PRIu64",%"PRIu64")",
+			 batch_marker->batch_seq_nr,
+			 (uint64_t)batch_marker->pinned_time);
+
+	for (batch = batch_marker; batch; batch = batch->next) {
+		time_t t;
+
+		if (!batch->pinned_time) /* an overflow batch */
+			continue;
+
+		t = batch->pinned_time + MY_TIME_DELAY_SECONDS;
+		if (t > batch_marker->pinned_time) /* too close to marker */
+			continue;
+
+		goto truncate_past_here;
+	}
+
+	return NULL;
+
+truncate_past_here:
+	state->current_token_data->batch_tail = (struct fsmonitor_batch *)batch;
+
+	remainder = ((struct fsmonitor_batch *)batch)->next;
+	((struct fsmonitor_batch *)batch)->next = NULL;
+
+	return remainder;
+}
+
 static void fsmonitor_free_token_data(struct fsmonitor_token_data *token)
 {
 	if (!token)
@@ -425,6 +494,7 @@ static int do_handle_client(struct fsmonitor_daemon_state *state,
 	const char *p;
 	const struct fsmonitor_batch *batch_head;
 	const struct fsmonitor_batch *batch;
+	struct fsmonitor_batch *remainder = NULL;
 	intmax_t count = 0, duplicates = 0;
 	kh_str_t *shown;
 	int hash_ret;
@@ -652,11 +722,29 @@ static int do_handle_client(struct fsmonitor_daemon_state *state,
 			 * that work.
 			 */
 			fsmonitor_free_token_data(token_data);
+		} else if (batch) {
+			/*
+			 * We are holding the lock and are the only
+			 * reader of the ref-counted portion of the
+			 * list, so we get the honor of seeing if the
+			 * list can be truncated to save memory.
+			 *
+			 * The main loop did not walk to the end of the
+			 * list, so this batch is the first item in the
+			 * batch-list that is older than the requested
+			 * end-point sequence number.  See if the tail
+			 * end of the list is obsolete.
+			 */
+			remainder = with_lock__truncate_old_batches(state,
+								    batch);
 		}
 	}
 
 	pthread_mutex_unlock(&state->main_lock);
 
+	if (remainder)
+		fsmonitor_batch__free_list(remainder);
+
 	trace2_data_intmax("fsmonitor", the_repository, "response/length", total_response_len);
 	trace2_data_intmax("fsmonitor", the_repository, "response/count/files", count);
 	trace2_data_intmax("fsmonitor", the_repository, "response/count/duplicates", duplicates);
-- 
gitgitgadget

[PATCH v8 28/30] fsmonitor--daemon: use a cookie file to sync with file system

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach fsmonitor--daemon client threads to create a cookie file
inside the .git directory and then wait until FS events for the
cookie are observed by the FS listener thread.

This helps address the racy nature of file system events by
blocking the client response until the kernel has drained any
event backlog.

This is especially important on MacOS where kernel events are
only issued with a limited frequency.  See the `latency` argument
of `FSeventStreamCreate()`.  The kernel only signals every `latency`
seconds, but does not guarantee that the kernel queue is completely
drained, so we may have to wait more than one interval.  If we
increase the latency, the system is more likely to drop events.
We avoid these issues by having each client thread create a unique
cookie file and then wait until it is seen in the event stream.

Co-authored-by: Kevin Willford <Kevin.Willford@microsoft.com>
Co-authored-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 builtin/fsmonitor--daemon.c | 237 +++++++++++++++++++++++++++++++++++-
 fsmonitor--daemon.h         |   5 +
 2 files changed, 241 insertions(+), 1 deletion(-)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index ab9cc09f7ce..46be55a4618 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -107,6 +107,162 @@ static int do_as_client__status(void)
 	}
 }
 
+enum fsmonitor_cookie_item_result {
+	FCIR_ERROR = -1, /* could not create cookie file ? */
+	FCIR_INIT,
+	FCIR_SEEN,
+	FCIR_ABORT,
+};
+
+struct fsmonitor_cookie_item {
+	struct hashmap_entry entry;
+	char *name;
+	enum fsmonitor_cookie_item_result result;
+};
+
+static int cookies_cmp(const void *data, const struct hashmap_entry *he1,
+		     const struct hashmap_entry *he2, const void *keydata)
+{
+	const struct fsmonitor_cookie_item *a =
+		container_of(he1, const struct fsmonitor_cookie_item, entry);
+	const struct fsmonitor_cookie_item *b =
+		container_of(he2, const struct fsmonitor_cookie_item, entry);
+
+	return strcmp(a->name, keydata ? keydata : b->name);
+}
+
+static enum fsmonitor_cookie_item_result with_lock__wait_for_cookie(
+	struct fsmonitor_daemon_state *state)
+{
+	/* assert current thread holding state->main_lock */
+
+	int fd;
+	struct fsmonitor_cookie_item *cookie;
+	struct strbuf cookie_pathname = STRBUF_INIT;
+	struct strbuf cookie_filename = STRBUF_INIT;
+	enum fsmonitor_cookie_item_result result;
+	int my_cookie_seq;
+
+	CALLOC_ARRAY(cookie, 1);
+
+	my_cookie_seq = state->cookie_seq++;
+
+	strbuf_addf(&cookie_filename, "%i-%i", getpid(), my_cookie_seq);
+
+	strbuf_addbuf(&cookie_pathname, &state->path_cookie_prefix);
+	strbuf_addbuf(&cookie_pathname, &cookie_filename);
+
+	cookie->name = strbuf_detach(&cookie_filename, NULL);
+	cookie->result = FCIR_INIT;
+	hashmap_entry_init(&cookie->entry, strhash(cookie->name));
+
+	hashmap_add(&state->cookies, &cookie->entry);
+
+	trace_printf_key(&trace_fsmonitor, "cookie-wait: '%s' '%s'",
+			 cookie->name, cookie_pathname.buf);
+
+	/*
+	 * Create the cookie file on disk and then wait for a notification
+	 * that the listener thread has seen it.
+	 */
+	fd = open(cookie_pathname.buf, O_WRONLY | O_CREAT | O_EXCL, 0600);
+	if (fd < 0) {
+		error_errno(_("could not create fsmonitor cookie '%s'"),
+			    cookie->name);
+
+		cookie->result = FCIR_ERROR;
+		goto done;
+	}
+
+	/*
+	 * Technically, close() and unlink() can fail, but we don't
+	 * care here.  We only created the file to trigger a watch
+	 * event from the FS to know that when we're up to date.
+	 */
+	close(fd);
+	unlink(cookie_pathname.buf);
+
+	/*
+	 * Technically, this is an infinite wait (well, unless another
+	 * thread sends us an abort).  I'd like to change this to
+	 * use `pthread_cond_timedwait()` and return an error/timeout
+	 * and let the caller do the trivial response thing, but we
+	 * don't have that routine in our thread-utils.
+	 *
+	 * After extensive beta testing I'm not really worried about
+	 * this.  Also note that the above open() and unlink() calls
+	 * will cause at least two FS events on that path, so the odds
+	 * of getting stuck are pretty slim.
+	 */
+	while (cookie->result == FCIR_INIT)
+		pthread_cond_wait(&state->cookies_cond,
+				  &state->main_lock);
+
+done:
+	hashmap_remove(&state->cookies, &cookie->entry, NULL);
+
+	result = cookie->result;
+
+	free(cookie->name);
+	free(cookie);
+	strbuf_release(&cookie_pathname);
+
+	return result;
+}
+
+/*
+ * Mark these cookies as _SEEN and wake up the corresponding client threads.
+ */
+static void with_lock__mark_cookies_seen(struct fsmonitor_daemon_state *state,
+					 const struct string_list *cookie_names)
+{
+	/* assert current thread holding state->main_lock */
+
+	int k;
+	int nr_seen = 0;
+
+	for (k = 0; k < cookie_names->nr; k++) {
+		struct fsmonitor_cookie_item key;
+		struct fsmonitor_cookie_item *cookie;
+
+		key.name = cookie_names->items[k].string;
+		hashmap_entry_init(&key.entry, strhash(key.name));
+
+		cookie = hashmap_get_entry(&state->cookies, &key, entry, NULL);
+		if (cookie) {
+			trace_printf_key(&trace_fsmonitor, "cookie-seen: '%s'",
+					 cookie->name);
+			cookie->result = FCIR_SEEN;
+			nr_seen++;
+		}
+	}
+
+	if (nr_seen)
+		pthread_cond_broadcast(&state->cookies_cond);
+}
+
+/*
+ * Set _ABORT on all pending cookies and wake up all client threads.
+ */
+static void with_lock__abort_all_cookies(struct fsmonitor_daemon_state *state)
+{
+	/* assert current thread holding state->main_lock */
+
+	struct hashmap_iter iter;
+	struct fsmonitor_cookie_item *cookie;
+	int nr_aborted = 0;
+
+	hashmap_for_each_entry(&state->cookies, &iter, cookie, entry) {
+		trace_printf_key(&trace_fsmonitor, "cookie-abort: '%s'",
+				 cookie->name);
+		cookie->result = FCIR_ABORT;
+		nr_aborted++;
+	}
+
+	if (nr_aborted)
+		pthread_cond_broadcast(&state->cookies_cond);
+}
+
 /*
  * Requests to and from a FSMonitor Protocol V2 provider use an opaque
  * "token" as a virtual timestamp.  Clients can request a summary of all
@@ -404,6 +560,9 @@ static void fsmonitor_free_token_data(struct fsmonitor_token_data *token)
  *     We should create a new token and start fresh (as if we just
  *     booted up).
  *
+ * [2] Some of those lost events may have been for cookie files.  We
+ *     should assume the worst and abort them rather letting them starve.
+ *
  * If there are no concurrent threads reading the current token data
  * series, we can free it now.  Otherwise, let the last reader free
  * it.
@@ -425,6 +584,8 @@ static void with_lock__do_force_resync(struct fsmonitor_daemon_state *state)
 	state->current_token_data = new_one;
 
 	fsmonitor_free_token_data(free_me);
+
+	with_lock__abort_all_cookies(state);
 }
 
 void fsmonitor_force_resync(struct fsmonitor_daemon_state *state)
@@ -500,6 +661,8 @@ static int do_handle_client(struct fsmonitor_daemon_state *state,
 	int hash_ret;
 	int do_trivial = 0;
 	int do_flush = 0;
+	int do_cookie = 0;
+	enum fsmonitor_cookie_item_result cookie_result;
 
 	/*
 	 * We expect `command` to be of the form:
@@ -560,6 +723,7 @@ static int do_handle_client(struct fsmonitor_daemon_state *state,
 			 * We have a V2 valid token:
 			 *     "builtin:<token_id>:<seq_nr>"
 			 */
+			do_cookie = 1;
 		}
 	}
 
@@ -568,6 +732,30 @@ static int do_handle_client(struct fsmonitor_daemon_state *state,
 	if (!state->current_token_data)
 		BUG("fsmonitor state does not have a current token");
 
+	/*
+	 * Write a cookie file inside the directory being watched in
+	 * an effort to flush out existing filesystem events that we
+	 * actually care about.  Suspend this client thread until we
+	 * see the filesystem events for this cookie file.
+	 *
+	 * Creating the cookie lets us guarantee that our FS listener
+	 * thread has drained the kernel queue and we are caught up
+	 * with the kernel.
+	 *
+	 * If we cannot create the cookie (or otherwise guarantee that
+	 * we are caught up), we send a trivial response.  We have to
+	 * assume that there might be some very, very recent activity
+	 * on the FS still in flight.
+	 */
+	if (do_cookie) {
+		cookie_result = with_lock__wait_for_cookie(state);
+		if (cookie_result != FCIR_SEEN) {
+			error(_("fsmonitor: cookie_result '%d' != SEEN"),
+			      cookie_result);
+			do_trivial = 1;
+		}
+	}
+
 	if (do_flush)
 		with_lock__do_force_resync(state);
 
@@ -787,7 +975,9 @@ static int handle_client(void *data,
 	return result;
 }
 
-#define FSMONITOR_COOKIE_PREFIX ".fsmonitor-daemon-"
+#define FSMONITOR_DIR           "fsmonitor--daemon"
+#define FSMONITOR_COOKIE_DIR    "cookies"
+#define FSMONITOR_COOKIE_PREFIX (FSMONITOR_DIR "/" FSMONITOR_COOKIE_DIR "/")
 
 enum fsmonitor_path_type fsmonitor_classify_path_workdir_relative(
 	const char *rel)
@@ -940,6 +1130,9 @@ void fsmonitor_publish(struct fsmonitor_daemon_state *state,
 		}
 	}
 
+	if (cookie_names->nr)
+		with_lock__mark_cookies_seen(state, cookie_names);
+
 	pthread_mutex_unlock(&state->main_lock);
 }
 
@@ -1031,7 +1224,9 @@ static int fsmonitor_run_daemon(void)
 
 	memset(&state, 0, sizeof(state));
 
+	hashmap_init(&state.cookies, cookies_cmp, NULL, 0);
 	pthread_mutex_init(&state.main_lock, NULL);
+	pthread_cond_init(&state.cookies_cond, NULL);
 	state.error_code = 0;
 	state.current_token_data = fsmonitor_new_token_data();
 
@@ -1056,6 +1251,44 @@ static int fsmonitor_run_daemon(void)
 		state.nr_paths_watching = 2;
 	}
 
+	/*
+	 * We will write filesystem syncing cookie files into
+	 * <gitdir>/<fsmonitor-dir>/<cookie-dir>/<pid>-<seq>.
+	 *
+	 * The extra layers of subdirectories here keep us from
+	 * changing the mtime on ".git/" or ".git/foo/" when we create
+	 * or delete cookie files.
+	 *
+	 * There have been problems with some IDEs that do a
+	 * non-recursive watch of the ".git/" directory and run a
+	 * series of commands any time something happens.
+	 *
+	 * For example, if we place our cookie files directly in
+	 * ".git/" or ".git/foo/" then a `git status` (or similar
+	 * command) from the IDE will cause a cookie file to be
+	 * created in one of those dirs.  This causes the mtime of
+	 * those dirs to change.  This triggers the IDE's watch
+	 * notification.  This triggers the IDE to run those commands
+	 * again.  And the process repeats and the machine never goes
+	 * idle.
+	 *
+	 * Adding the extra layers of subdirectories prevents the
+	 * mtime of ".git/" and ".git/foo" from changing when a
+	 * cookie file is created.
+	 */
+	strbuf_init(&state.path_cookie_prefix, 0);
+	strbuf_addbuf(&state.path_cookie_prefix, &state.path_gitdir_watch);
+
+	strbuf_addch(&state.path_cookie_prefix, '/');
+	strbuf_addstr(&state.path_cookie_prefix, FSMONITOR_DIR);
+	mkdir(state.path_cookie_prefix.buf, 0777);
+
+	strbuf_addch(&state.path_cookie_prefix, '/');
+	strbuf_addstr(&state.path_cookie_prefix, FSMONITOR_COOKIE_DIR);
+	mkdir(state.path_cookie_prefix.buf, 0777);
+
+	strbuf_addch(&state.path_cookie_prefix, '/');
+
 	/*
 	 * Confirm that we can create platform-specific resources for the
 	 * filesystem listener before we bother starting all the threads.
@@ -1068,6 +1301,7 @@ static int fsmonitor_run_daemon(void)
 	err = fsmonitor_run_daemon_1(&state);
 
 done:
+	pthread_cond_destroy(&state.cookies_cond);
 	pthread_mutex_destroy(&state.main_lock);
 	fsm_listen__dtor(&state);
 
@@ -1075,6 +1309,7 @@ done:
 
 	strbuf_release(&state.path_worktree_watch);
 	strbuf_release(&state.path_gitdir_watch);
+	strbuf_release(&state.path_cookie_prefix);
 
 	return err;
 }
diff --git a/fsmonitor--daemon.h b/fsmonitor--daemon.h
index 010fbfe60e9..bd09fffc176 100644
--- a/fsmonitor--daemon.h
+++ b/fsmonitor--daemon.h
@@ -45,6 +45,11 @@ struct fsmonitor_daemon_state {
 
 	struct fsmonitor_token_data *current_token_data;
 
+	struct strbuf path_cookie_prefix;
+	pthread_cond_t cookies_cond;
+	int cookie_seq;
+	struct hashmap cookies;
+
 	int error_code;
 	struct fsmonitor_daemon_backend_data *backend_data;
 
-- 
gitgitgadget

[PATCH v8 29/30] fsmonitor: force update index after large responses

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Measure the time taken to apply the FSMonitor query result
to the index and the untracked-cache.

Set the `FSMONITOR_CHANGED` bit on `istate->cache_changed` when
FSMonitor returns a very large repsonse to ensure that the index is
written to disk.

Normally, when the FSMonitor response includes a tracked file, the
index is always updated.  Similarly, the index might be updated when
the response alters the untracked-cache (when enabled).  However, in
cases where neither of those cause the index to be considered changed,
the FSMonitor response is wasted.  Subsequent Git commands will make
requests with the same token and receive the same response.

If that response is very large, performance may suffer.  It would be
more efficient to force update the index now (and the token in the
index extension) in order to reduce the size of the response received
by future commands.

This was observed on Windows after a large checkout.  On Windows, the
kernel emits events for the files that are changed as they are
changed.  However, it might delay events for the containing
directories until the system is more idle (or someone scans the
directory (so it seems)).  The first status following a checkout would
get the list of files.  The subsequent status commands would get the
list of directories as the events trickled out.  But they would never
catch up because the token was not advanced because the index wasn't
updated.

This list of directories caused `wt_status_collect_untracked()` to
unnecessarily spend time actually scanning them during each command.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 fsmonitor.c | 55 ++++++++++++++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 54 insertions(+), 1 deletion(-)

diff --git a/fsmonitor.c b/fsmonitor.c
index a38b5710eb3..292a6742b4f 100644
--- a/fsmonitor.c
+++ b/fsmonitor.c
@@ -219,6 +219,43 @@ static void fsmonitor_refresh_callback(struct index_state *istate, char *name)
 	untracked_cache_invalidate_path(istate, name, 0);
 }
 
+/*
+ * The number of pathnames that we need to receive from FSMonitor
+ * before we force the index to be updated.
+ *
+ * Note that any pathname within the set of received paths MAY cause
+ * cache-entry or istate flag bits to be updated and thus cause the
+ * index to be updated on disk.
+ *
+ * However, the response may contain many paths (such as ignored
+ * paths) that will not update any flag bits.  And thus not force the
+ * index to be updated.  (This is fine and normal.)  It also means
+ * that the token will not be updated in the FSMonitor index
+ * extension.  So the next Git command will find the same token in the
+ * index, make the same token-relative request, and receive the same
+ * response (plus any newly changed paths).  If this response is large
+ * (and continues to grow), performance could be impacted.
+ *
+ * For example, if the user runs a build and it writes 100K object
+ * files but doesn't modify any source files, the index would not need
+ * to be updated.  The FSMonitor response (after the build and
+ * relative to a pre-build token) might be 5MB.  Each subsequent Git
+ * command will receive that same 100K/5MB response until something
+ * causes the index to be updated.  And `refresh_fsmonitor()` will
+ * have to iterate over those 100K paths each time.
+ *
+ * Performance could be improved if we optionally force update the
+ * index after a very large response and get an updated token into
+ * the FSMonitor index extension.  This should allow subsequent
+ * commands to get smaller and more current responses.
+ *
+ * The value chosen here does not need to be precise.  The index
+ * will be updated automatically the first time the user touches
+ * a tracked file and causes a command like `git status` to
+ * update an mtime to be updated and/or set a flag bit.
+ */
+static int fsmonitor_force_update_threshold = 100;
+
 void refresh_fsmonitor(struct index_state *istate)
 {
 	struct strbuf query_result = STRBUF_INIT;
@@ -362,25 +399,39 @@ apply_results:
 	 *     information and that we should consider everything
 	 *     invalid.  We call this a trivial response.
 	 */
+	trace2_region_enter("fsmonitor", "apply_results", istate->repo);
+
 	if (query_success && !is_trivial) {
 		/*
 		 * Mark all pathnames returned by the monitor as dirty.
 		 *
 		 * This updates both the cache-entries and the untracked-cache.
 		 */
+		int count = 0;
+
 		buf = query_result.buf;
 		for (i = bol; i < query_result.len; i++) {
 			if (buf[i] != '\0')
 				continue;
 			fsmonitor_refresh_callback(istate, buf + bol);
 			bol = i + 1;
+			count++;
 		}
-		if (bol < query_result.len)
+		if (bol < query_result.len) {
 			fsmonitor_refresh_callback(istate, buf + bol);
+			count++;
+		}
 
 		/* Now mark the untracked cache for fsmonitor usage */
 		if (istate->untracked)
 			istate->untracked->use_fsmonitor = 1;
+
+		if (count > fsmonitor_force_update_threshold)
+			istate->cache_changed |= FSMONITOR_CHANGED;
+
+		trace2_data_intmax("fsmonitor", istate->repo, "apply_count",
+				   count);
+
 	} else {
 		/*
 		 * We failed to get a response or received a trivial response,
@@ -409,6 +460,8 @@ apply_results:
 		if (istate->untracked)
 			istate->untracked->use_fsmonitor = 0;
 	}
+	trace2_region_leave("fsmonitor", "apply_results", istate->repo);
+
 	strbuf_release(&query_result);
 
 	/* Now that we've updated istate, save the last_update_token */
-- 
gitgitgadget

[PATCH v8 30/30] t7527: test status with untracked-cache and fsmonitor--daemon

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Create 2x2 test matrix with the untracked-cache and fsmonitor--daemon
features and a series of edits and verify that status output is
identical.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 t/t7527-builtin-fsmonitor.sh | 115 +++++++++++++++++++++++++++++++++++
 1 file changed, 115 insertions(+)

diff --git a/t/t7527-builtin-fsmonitor.sh b/t/t7527-builtin-fsmonitor.sh
index 3570bf764aa..1c40d8d2035 100755
--- a/t/t7527-builtin-fsmonitor.sh
+++ b/t/t7527-builtin-fsmonitor.sh
@@ -216,6 +216,8 @@ test_expect_success 'setup' '
 	.gitignore
 	expect*
 	actual*
+	flush*
+	trace*
 	EOF
 
 	git -c core.fsmonitor=false add . &&
@@ -502,4 +504,117 @@ test_expect_success 'cleanup worktrees' '
 	stop_daemon_delete_repo wt-base
 '
 
+# The next few tests perform arbitrary/contrived file operations and
+# confirm that status is correct.  That is, that the data (or lack of
+# data) from fsmonitor doesn't cause incorrect results.  And doesn't
+# cause incorrect results when the untracked-cache is enabled.
+
+test_lazy_prereq UNTRACKED_CACHE '
+	git update-index --test-untracked-cache
+'
+
+test_expect_success 'Matrix: setup for untracked-cache,fsmonitor matrix' '
+	test_unconfig core.fsmonitor &&
+	git update-index --no-fsmonitor &&
+	test_might_fail git fsmonitor--daemon stop
+'
+
+matrix_clean_up_repo () {
+	git reset --hard HEAD &&
+	git clean -fd
+}
+
+matrix_try () {
+	uc=$1 &&
+	fsm=$2 &&
+	fn=$3 &&
+
+	if test $uc = true && test $fsm = false
+	then
+		# The untracked-cache is buggy when FSMonitor is
+		# DISABLED, so skip the tests for this matrix
+		# combination.
+		#
+		# We've observed random, occasional test failures on
+		# Windows and MacOS when the UC is turned on and FSM
+		# is turned off.  These are rare, but they do happen
+		# indicating that it is probably a race condition within
+		# the untracked cache itself.
+		#
+		# It usually happens when a test does F/D trickery and
+		# then the NEXT test fails because of extra status
+		# output from stale UC data from the previous test.
+		#
+		# Since FSMonitor is not involved in the error, skip
+		# the tests for this matrix combination.
+		#
+		return 0
+	fi &&
+
+	test_expect_success "Matrix[uc:$uc][fsm:$fsm] $fn" '
+		matrix_clean_up_repo &&
+		$fn &&
+		if test $uc = false && test $fsm = false
+		then
+			git status --porcelain=v1 >.git/expect.$fn
+		else
+			git status --porcelain=v1 >.git/actual.$fn &&
+			test_cmp .git/expect.$fn .git/actual.$fn
+		fi
+	'
+}
+
+uc_values="false"
+test_have_prereq UNTRACKED_CACHE && uc_values="false true"
+for uc_val in $uc_values
+do
+	if test $uc_val = false
+	then
+		test_expect_success "Matrix[uc:$uc_val] disable untracked cache" '
+			git config core.untrackedcache false &&
+			git update-index --no-untracked-cache
+		'
+	else
+		test_expect_success "Matrix[uc:$uc_val] enable untracked cache" '
+			git config core.untrackedcache true &&
+			git update-index --untracked-cache
+		'
+	fi
+
+	fsm_values="false true"
+	for fsm_val in $fsm_values
+	do
+		if test $fsm_val = false
+		then
+			test_expect_success "Matrix[uc:$uc_val][fsm:$fsm_val] disable fsmonitor" '
+				test_unconfig core.fsmonitor &&
+				git update-index --no-fsmonitor &&
+				test_might_fail git fsmonitor--daemon stop
+			'
+		else
+			test_expect_success "Matrix[uc:$uc_val][fsm:$fsm_val] enable fsmonitor" '
+				git config core.fsmonitor true &&
+				git fsmonitor--daemon start &&
+				git update-index --fsmonitor
+			'
+		fi
+
+		matrix_try $uc_val $fsm_val edit_files
+		matrix_try $uc_val $fsm_val delete_files
+		matrix_try $uc_val $fsm_val create_files
+		matrix_try $uc_val $fsm_val rename_files
+		matrix_try $uc_val $fsm_val file_to_directory
+		matrix_try $uc_val $fsm_val directory_to_file
+
+		if test $fsm_val = true
+		then
+			test_expect_success "Matrix[uc:$uc_val][fsm:$fsm_val] disable fsmonitor at end" '
+				test_unconfig core.fsmonitor &&
+				git update-index --no-fsmonitor &&
+				test_might_fail git fsmonitor--daemon stop
+			'
+		fi
+	done
+done
+
 test_done
-- 
gitgitgadget

Re: [PATCH v8 21/30] t7527: create test for fsmonitor--daemon

[Junio C Hamano]

"Jeff Hostetler via GitGitGadget" <gitgitgadget@gmail.com> writes:

> From: Jeff Hostetler <jeffhost@microsoft.com>
>
> Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
> Signed-off-by: Junio C Hamano <gitster@pobox.com>

I hadn't signed off on this one yet ;-)

> +is_value () {
> +	test -n "$1" && test "${1::1}" != "-"
> +}

${var:ofs:len} is a bash-ism.  If you run this test under /bin/dash
instead of /bin/dash, you'll likely see it fail.

If it were a good idea to see if $1 begins with a dash, a more
natural (to shell programmers) way to do so is

	case "$1" in -*) false ;; ?*) true ;; *) false ;; esac

but given how this is used below, we do not want to special case
dash.  

There isn't anything wrong in "mkdir ./-foo && start_daemon -C -foo"
in other words.

> +start_daemon () {
> +	r= &&
> +	tf= &&
> +	t2= &&
> +	tk= &&

FYI, you can write these on a single line, i.e.

	r= tf= t2= tk= &&

Spending lines and spaces for the meat of the script would enhance
readability but for things like a boilerplate "we clear variables
before using them", being concise may be less distracting.

> +	while test "$#" -ne 0
> +	do
> +		case "$1" in
> +		-C)
> +			shift;
> +			is_value $1 || BUG "error: -C requires value"
> +			r="-C $1"
> +			shift
> +			;;
> +	...
> +		esac
> +	done &&

A more natural way to write these loops is

	while ...
	do
		case "$1" in
		-C)
			r="-C ${2?}"
			shift
			;;
		... all other options you handle ...
		-*)
			echo >&2 "unknown option $1"
			exit 1
			;;
		*)
			break
			;;
		esac
		shift
	done

i.e. shifting out what we just saw is the default and happens
immediately after the case/esac, and extra shift after consuming
an option parameter happens in each case arm.

An acceptable slight variation is

		-C)
			shift
			r="-C ${1?}"
			;;

but the first form is more logical and clear, i.e. "when we see '-C',
we want two on the command line, -C itself and the parameter it takes"
is conveyed more strongly with "${2?}" there.

For an additional bonus, we could also accept the stuck form, i.e.

		case "$1" in
		-C)
			r="-C ${2?}"
			shift
			;;
		-C*)
			r="-C {$1#-C}"
			;;
		...

RE: [PATCH v8 21/30] t7527: create test for fsmonitor--daemon

[rsbecker]

On March 24, 2022 3:00 PM, Junio C Hamano wrote:
>Subject: Re: [PATCH v8 21/30] t7527: create test for fsmonitor--daemon
>
>"Jeff Hostetler via GitGitGadget" <gitgitgadget@gmail.com> writes:
>
>> From: Jeff Hostetler <jeffhost@microsoft.com>
>>
>> Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
>> Signed-off-by: Junio C Hamano <gitster@pobox.com>
>
>I hadn't signed off on this one yet ;-)
>
>> +is_value () {
>> +	test -n "$1" && test "${1::1}" != "-"
>> +}
>
>${var:ofs:len} is a bash-ism.  If you run this test under /bin/dash instead of
>/bin/dash, you'll likely see it fail.
>
>If it were a good idea to see if $1 begins with a dash, a more natural (to shell
>programmers) way to do so is
>
>	case "$1" in -*) false ;; ?*) true ;; *) false ;; esac
>
>but given how this is used below, we do not want to special case dash.
>
>There isn't anything wrong in "mkdir ./-foo && start_daemon -C -foo"
>in other words.
>
>> +start_daemon () {
>> +	r= &&
>> +	tf= &&
>> +	t2= &&
>> +	tk= &&
>
>FYI, you can write these on a single line, i.e.
>
>	r= tf= t2= tk= &&
>
>Spending lines and spaces for the meat of the script would enhance readability but
>for things like a boilerplate "we clear variables before using them", being concise
>may be less distracting.
>
>> +	while test "$#" -ne 0
>> +	do
>> +		case "$1" in
>> +		-C)
>> +			shift;
>> +			is_value $1 || BUG "error: -C requires value"
>> +			r="-C $1"
>> +			shift
>> +			;;
>> +	...
>> +		esac
>> +	done &&
>
>A more natural way to write these loops is
>
>	while ...
>	do
>		case "$1" in
>		-C)
>			r="-C ${2?}"
>			shift
>			;;
>		... all other options you handle ...
>		-*)
>			echo >&2 "unknown option $1"
>			exit 1
>			;;
>		*)
>			break
>			;;
>		esac
>		shift
>	done
>
>i.e. shifting out what we just saw is the default and happens immediately after the
>case/esac, and extra shift after consuming an option parameter happens in each
>case arm.
>
>An acceptable slight variation is
>
>		-C)
>			shift
>			r="-C ${1?}"
>			;;
>
>but the first form is more logical and clear, i.e. "when we see '-C', we want two on
>the command line, -C itself and the parameter it takes"
>is conveyed more strongly with "${2?}" there.
>
>For an additional bonus, we could also accept the stuck form, i.e.
>
>		case "$1" in
>		-C)
>			r="-C ${2?}"
>			shift
>			;;
>		-C*)
>			r="-C {$1#-C}"
>			;;
>		...

May I request a bit of extra time on the -rc0 to -rc1 cycle for this? I have a feeling that while testing this is probably going to go well, I would like to have a bit of extra time for anything that might come up. There are a lot of moving parts to this series. Not being critical, but debugging scripts on my platforms can be a bit rough at times.

Thanks,
Randall

Re: [PATCH v8 21/30] t7527: create test for fsmonitor--daemon

[Jeff Hostetler]

On 3/24/22 3:05 PM, rsbecker@nexbridge.com wrote:
> On March 24, 2022 3:00 PM, Junio C Hamano wrote:
>> Subject: Re: [PATCH v8 21/30] t7527: create test for fsmonitor--daemon
>>
>> "Jeff Hostetler via GitGitGadget" <gitgitgadget@gmail.com> writes:
>>
>>> From: Jeff Hostetler <jeffhost@microsoft.com>
[...]
> 
> May I request a bit of extra time on the -rc0 to -rc1 cycle for this? I have a feeling that while testing this is probably going to go well, I would like to have a bit of extra time for anything that might come up. There are a lot of moving parts to this series. Not being critical, but debugging scripts on my platforms can be a bit rough at times.
> 
> Thanks,
> Randall
> 

I'll simplify the `start_daemon()` function as Junio suggests,
so hopefully that'll reduce the amount of debugging that you need.

BTW, which platforms are you worried about?

Jeff

RE: [PATCH v8 21/30] t7527: create test for fsmonitor--daemon

[rsbecker]

On March 24, 2022 4:28 PM, Jeff Hostetler wrote:
>On 3/24/22 3:05 PM, rsbecker@nexbridge.com wrote:
>> On March 24, 2022 3:00 PM, Junio C Hamano wrote:
>>> Subject: Re: [PATCH v8 21/30] t7527: create test for
>>> fsmonitor--daemon
>>>
>>> "Jeff Hostetler via GitGitGadget" <gitgitgadget@gmail.com> writes:
>>>
>>>> From: Jeff Hostetler <jeffhost@microsoft.com>
>[...]
>>
>> May I request a bit of extra time on the -rc0 to -rc1 cycle for this? I have a feeling
>that while testing this is probably going to go well, I would like to have a bit of
>extra time for anything that might come up. There are a lot of moving parts to this
>series. Not being critical, but debugging scripts on my platforms can be a bit rough
>at times.
>>
>> Thanks,
>> Randall
>>
>
>I'll simplify the `start_daemon()` function as Junio suggests, so hopefully that'll
>reduce the amount of debugging that you need.
Thanks.

>BTW, which platforms are you worried about?
I'm worried about NonStop ia64 and x86. It's not just this series but also the fsync series. I think it's going to be a bit of a rid this time, but hoping not.
--Randall

Re: [PATCH v8 21/30] t7527: create test for fsmonitor--daemon

[Jeff Hostetler]

On 3/24/22 4:36 PM, rsbecker@nexbridge.com wrote:
> On March 24, 2022 4:28 PM, Jeff Hostetler wrote:
>> On 3/24/22 3:05 PM, rsbecker@nexbridge.com wrote:
>>> On March 24, 2022 3:00 PM, Junio C Hamano wrote:
>>>> Subject: Re: [PATCH v8 21/30] t7527: create test for
>>>> fsmonitor--daemon
>>>>
>>>> "Jeff Hostetler via GitGitGadget" <gitgitgadget@gmail.com> writes:
>>>>
>>>>> From: Jeff Hostetler <jeffhost@microsoft.com>
>> [...]
>>>
>>> May I request a bit of extra time on the -rc0 to -rc1 cycle for this? I have a feeling
>> that while testing this is probably going to go well, I would like to have a bit of
>> extra time for anything that might come up. There are a lot of moving parts to this
>> series. Not being critical, but debugging scripts on my platforms can be a bit rough
>> at times.
>>>
>>> Thanks,
>>> Randall
>>>
>>
>> I'll simplify the `start_daemon()` function as Junio suggests, so hopefully that'll
>> reduce the amount of debugging that you need.
> Thanks.
> 
>> BTW, which platforms are you worried about?
> I'm worried about NonStop ia64 and x86. It's not just this series but also the fsync series. I think it's going to be a bit of a rid this time, but hoping not.
> --Randall
> 

I currently only have drivers/backends for Windows and MacOS for the
builtin FSMonitor daemon feature.  Since it relies on platform-specific
code to read whatever filesystem journal or event stream that the
the platform provides.

So t7527 should do a skip_all on NonStop IIUC.

Jeff

Re: [PATCH v8 21/30] t7527: create test for fsmonitor--daemon

[Jeff Hostetler]

On 3/24/22 2:59 PM, Junio C Hamano wrote:
> "Jeff Hostetler via GitGitGadget" <gitgitgadget@gmail.com> writes:
> 
>> From: Jeff Hostetler <jeffhost@microsoft.com>
>>
>> Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
>> Signed-off-by: Junio C Hamano <gitster@pobox.com>
> 
> I hadn't signed off on this one yet ;-)

Sorry, I started a recent rebase based upon your version
of the commits and it had your signoff in it.

> 
>> +is_value () {
>> +	test -n "$1" && test "${1::1}" != "-"
>> +}
> 
> ${var:ofs:len} is a bash-ism.  If you run this test under /bin/dash
> instead of /bin/dash, you'll likely see it fail.
[...]

I'll simplify the function as you suggested and send a V9 in
the morning after GGG's CI finishes.

Thanks
Jeff

RE: [PATCH v8 21/30] t7527: create test for fsmonitor--daemon

[rsbecker]

On: March 24, 2022 4:43 PM, Jeff Hostetler wrote:
>On 3/24/22 4:36 PM, rsbecker@nexbridge.com wrote:
>> On March 24, 2022 4:28 PM, Jeff Hostetler wrote:
>>> On 3/24/22 3:05 PM, rsbecker@nexbridge.com wrote:
>>>> On March 24, 2022 3:00 PM, Junio C Hamano wrote:
>>>>> Subject: Re: [PATCH v8 21/30] t7527: create test for
>>>>> fsmonitor--daemon
>>>>>
>>>>> "Jeff Hostetler via GitGitGadget" <gitgitgadget@gmail.com> writes:
>>>>>
>>>>>> From: Jeff Hostetler <jeffhost@microsoft.com>
>>> [...]
>>>>
>>>> May I request a bit of extra time on the -rc0 to -rc1 cycle for
>>>> this? I have a feeling
>>> that while testing this is probably going to go well, I would like to
>>> have a bit of extra time for anything that might come up. There are a
>>> lot of moving parts to this series. Not being critical, but debugging
>>> scripts on my platforms can be a bit rough at times.
>>>>
>>>> Thanks,
>>>> Randall
>>>>
>>>
>>> I'll simplify the `start_daemon()` function as Junio suggests, so
>>> hopefully that'll reduce the amount of debugging that you need.
>> Thanks.
>>
>>> BTW, which platforms are you worried about?
>> I'm worried about NonStop ia64 and x86. It's not just this series but also the
>fsync series. I think it's going to be a bit of a rid this time, but hoping not.
>> --Randall
>>
>
>I currently only have drivers/backends for Windows and MacOS for the builtin
>FSMonitor daemon feature.  Since it relies on platform-specific code to read
>whatever filesystem journal or event stream that the the platform provides.
>
>So t7527 should do a skip_all on NonStop IIUC.

I would have thought this was applicable to Linux also. So use, it should skip. Be in touch when if you do down that path.
--Randall

Re: [PATCH v8 21/30] t7527: create test for fsmonitor--daemon

[Jeff Hostetler]

On 3/24/22 4:46 PM, rsbecker@nexbridge.com wrote:
> On: March 24, 2022 4:43 PM, Jeff Hostetler wrote:
>> On 3/24/22 4:36 PM, rsbecker@nexbridge.com wrote:
>>> On March 24, 2022 4:28 PM, Jeff Hostetler wrote:
>>>> On 3/24/22 3:05 PM, rsbecker@nexbridge.com wrote:
>>>>> On March 24, 2022 3:00 PM, Junio C Hamano wrote:
>>>>>> Subject: Re: [PATCH v8 21/30] t7527: create test for
>>>>>> fsmonitor--daemon
>>>>>>
>>>>>> "Jeff Hostetler via GitGitGadget" <gitgitgadget@gmail.com> writes:
>>>>>>
>>>>>>> From: Jeff Hostetler <jeffhost@microsoft.com>
>>>> [...]
>>>>>
>>>>> May I request a bit of extra time on the -rc0 to -rc1 cycle for
>>>>> this? I have a feeling
>>>> that while testing this is probably going to go well, I would like to
>>>> have a bit of extra time for anything that might come up. There are a
>>>> lot of moving parts to this series. Not being critical, but debugging
>>>> scripts on my platforms can be a bit rough at times.
>>>>>
>>>>> Thanks,
>>>>> Randall
>>>>>
>>>>
>>>> I'll simplify the `start_daemon()` function as Junio suggests, so
>>>> hopefully that'll reduce the amount of debugging that you need.
>>> Thanks.
>>>
>>>> BTW, which platforms are you worried about?
>>> I'm worried about NonStop ia64 and x86. It's not just this series but also the
>> fsync series. I think it's going to be a bit of a rid this time, but hoping not.
>>> --Randall
>>>
>>
>> I currently only have drivers/backends for Windows and MacOS for the builtin
>> FSMonitor daemon feature.  Since it relies on platform-specific code to read
>> whatever filesystem journal or event stream that the the platform provides.
>>
>> So t7527 should do a skip_all on NonStop IIUC.
> 
> I would have thought this was applicable to Linux also. So use, it should skip. Be in touch when if you do down that path.
> --Randall
> 

Sorry, yeah, I wanted to do a Linux version too, but there was just too
much platform-specific stuff to juggle three different backends at
the same time.  I have the APIs in place between platform-specific and
-independent code so that it shouldn't be hard to add a Linux version.
I just didn't have the capacity to do all three (and the rest of
$dayjob....)

Jeff

[PATCH v9 00/30] Builtin FSMonitor Part 2

[Jeff Hostetler via GitGitGadget]

Here is V9 of Part 2 of my builtin FSMonitor series. This version addresses
bash style issues in t7527 raised on V8. These changes do not require a new
version of Part 3.

Here is a range-diff from V8 to V9 relative to 715d08a9e5 (The eighth batch,
2022-02-25).

 1:  e98373f997 =  1:  e98373f997 fsmonitor: enhance existing comments, clarify trivial response handling
 2:  ab68b94417 =  2:  ab68b94417 fsmonitor-ipc: create client routines for git-fsmonitor--daemon
 3:  e04c7301f2 =  3:  e04c7301f2 fsmonitor: config settings are repository-specific
 4:  ea02ba25d8 =  4:  ea02ba25d8 fsmonitor: use IPC to query the builtin FSMonitor daemon
 5:  6ab7db9cb7 =  5:  6ab7db9cb7 fsmonitor: document builtin fsmonitor
 6:  0ce8ae3f2c =  6:  0ce8ae3f2c fsmonitor--daemon: add a built-in fsmonitor daemon
 7:  4624ce2fa4 =  7:  4624ce2fa4 fsmonitor--daemon: implement 'stop' and 'status' commands
 8:  a29fe7266a =  8:  a29fe7266a compat/fsmonitor/fsm-listen-win32: stub in backend for Windows
 9:  2f8a42fdb9 =  9:  2f8a42fdb9 compat/fsmonitor/fsm-listen-darwin: stub in backend for Darwin
10:  f07800690e = 10:  f07800690e fsmonitor--daemon: implement 'run' command
11:  a6a39a3306 = 11:  a6a39a3306 fsmonitor--daemon: implement 'start' command
12:  d62e338d00 = 12:  d62e338d00 fsmonitor--daemon: add pathname classification
13:  53e06b4ae5 = 13:  53e06b4ae5 fsmonitor--daemon: define token-ids
14:  39f43fabe0 = 14:  39f43fabe0 fsmonitor--daemon: create token-based changed path cache
15:  239558e34f = 15:  239558e34f compat/fsmonitor/fsm-listen-win32: implement FSMonitor backend on Windows
16:  14b775e9d8 = 16:  14b775e9d8 compat/fsmonitor/fsm-listen-darwin: add MacOS header files for FSEvent
17:  55bd7aee06 = 17:  55bd7aee06 compat/fsmonitor/fsm-listen-darwin: implement FSEvent listener on MacOS
18:  c43009124f = 18:  c43009124f fsmonitor--daemon: implement handle_client callback
19:  ed338777b5 = 19:  ed338777b5 help: include fsmonitor--daemon feature flag in version info
20:  c99bac29d4 = 20:  c99bac29d4 t/helper/fsmonitor-client: create IPC client to talk to FSMonitor Daemon
21:  c8709da945 ! 21:  bc94e379b0 t7527: create test for fsmonitor--daemon
    @@ Commit message
         t7527: create test for fsmonitor--daemon
     
         Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
    -    Signed-off-by: Junio C Hamano <gitster@pobox.com>
     
      ## t/t7527-builtin-fsmonitor.sh (new) ##
     @@
    @@ t/t7527-builtin-fsmonitor.sh (new)
     +    rm -rf $1
     +}
     +
    -+is_value () {
    -+    test -n "$1" && test "${1::1}" != "-"
    -+}
    -+
     +start_daemon () {
    -+    r= &&
    -+    tf= &&
    -+    t2= &&
    -+    tk= &&
    ++    r= tf= t2= tk= &&
     +
     +    while test "$#" -ne 0
     +    do
     +        case "$1" in
     +        -C)
    -+            shift;
    -+            is_value $1 || BUG "error: -C requires value"
    -+            r="-C $1"
    ++            r="-C ${2?}"
     +            shift
     +            ;;
     +        --tf)
    -+            shift;
    -+            is_value $1 || BUG "error: --tf requires value"
    -+            tf="$1"
    ++            tf="${2?}"
     +            shift
     +            ;;
     +        --t2)
    -+            shift;
    -+            is_value $1 || BUG "error: --t2 requires value"
    -+            t2="$1"
    ++            t2="${2?}"
     +            shift
     +            ;;
     +        --tk)
    -+            shift;
    -+            is_value $1 || BUG "error: --tk requires value"
    -+            tk="$1"
    ++            tk="${2?}"
     +            shift
     +            ;;
    -+        *)
    ++        -*)
     +            BUG "error: unknown option: '$1'"
     +            ;;
    ++        *)
    ++            BUG "error: unbound argument: '$1'"
    ++            ;;
     +        esac
    ++        shift
     +    done &&
     +
     +    (
22:  cc39ecf10a = 22:  06d56d3a73 t/perf: avoid copying builtin fsmonitor files into test repo
23:  2bb3eb8476 = 23:  2dd0215127 t/helper/test-chmtime: skip directories on Windows
24:  bab9a9b080 = 24:  bb88cddc13 t/perf/p7519: fix coding style
25:  2dd06ad2f7 = 25:  50c2afaa49 t/perf/p7519: speed up test on Windows
26:  6eaa5765ae = 26:  5b18e3b692 t/perf/p7519: add fsmonitor--daemon test cases
27:  30957f3930 = 27:  899c23f63c fsmonitor--daemon: periodically truncate list of modified files
28:  c8ca2a1727 = 28:  597a7192f9 fsmonitor--daemon: use a cookie file to sync with file system
29:  4caf1d89b8 = 29:  68a05fd289 fsmonitor: force update index after large responses
30:  f87a1eba69 = 30:  5eb696daba t7527: test status with untracked-cache and fsmonitor--daemon


Jeff Hostetler (30):
  fsmonitor: enhance existing comments, clarify trivial response
    handling
  fsmonitor-ipc: create client routines for git-fsmonitor--daemon
  fsmonitor: config settings are repository-specific
  fsmonitor: use IPC to query the builtin FSMonitor daemon
  fsmonitor: document builtin fsmonitor
  fsmonitor--daemon: add a built-in fsmonitor daemon
  fsmonitor--daemon: implement 'stop' and 'status' commands
  compat/fsmonitor/fsm-listen-win32: stub in backend for Windows
  compat/fsmonitor/fsm-listen-darwin: stub in backend for Darwin
  fsmonitor--daemon: implement 'run' command
  fsmonitor--daemon: implement 'start' command
  fsmonitor--daemon: add pathname classification
  fsmonitor--daemon: define token-ids
  fsmonitor--daemon: create token-based changed path cache
  compat/fsmonitor/fsm-listen-win32: implement FSMonitor backend on
    Windows
  compat/fsmonitor/fsm-listen-darwin: add MacOS header files for FSEvent
  compat/fsmonitor/fsm-listen-darwin: implement FSEvent listener on
    MacOS
  fsmonitor--daemon: implement handle_client callback
  help: include fsmonitor--daemon feature flag in version info
  t/helper/fsmonitor-client: create IPC client to talk to FSMonitor
    Daemon
  t7527: create test for fsmonitor--daemon
  t/perf: avoid copying builtin fsmonitor files into test repo
  t/helper/test-chmtime: skip directories on Windows
  t/perf/p7519: fix coding style
  t/perf/p7519: speed up test on Windows
  t/perf/p7519: add fsmonitor--daemon test cases
  fsmonitor--daemon: periodically truncate list of modified files
  fsmonitor--daemon: use a cookie file to sync with file system
  fsmonitor: force update index after large responses
  t7527: test status with untracked-cache and fsmonitor--daemon

 .gitignore                              |    1 +
 Documentation/config/core.txt           |   60 +-
 Documentation/git-fsmonitor--daemon.txt |   75 ++
 Documentation/git-update-index.txt      |    8 +-
 Makefile                                |   17 +
 builtin.h                               |    1 +
 builtin/fsmonitor--daemon.c             | 1479 +++++++++++++++++++++++
 builtin/update-index.c                  |    7 +-
 cache.h                                 |    1 -
 compat/fsmonitor/fsm-darwin-gcc.h       |   92 ++
 compat/fsmonitor/fsm-listen-darwin.c    |  427 +++++++
 compat/fsmonitor/fsm-listen-win32.c     |  586 +++++++++
 compat/fsmonitor/fsm-listen.h           |   49 +
 config.c                                |   14 -
 config.h                                |    1 -
 config.mak.uname                        |   20 +
 contrib/buildsystems/CMakeLists.txt     |   10 +
 environment.c                           |    1 -
 fsmonitor--daemon.h                     |  166 +++
 fsmonitor-ipc.c                         |  171 +++
 fsmonitor-ipc.h                         |   48 +
 fsmonitor-settings.c                    |  114 ++
 fsmonitor-settings.h                    |   21 +
 fsmonitor.c                             |  216 +++-
 fsmonitor.h                             |   15 +-
 git.c                                   |    1 +
 help.c                                  |    4 +
 repo-settings.c                         |    1 +
 repository.h                            |    3 +
 t/README                                |    4 +-
 t/helper/test-chmtime.c                 |   15 +
 t/helper/test-fsmonitor-client.c        |  116 ++
 t/helper/test-tool.c                    |    1 +
 t/helper/test-tool.h                    |    1 +
 t/perf/p7519-fsmonitor.sh               |   68 +-
 t/perf/perf-lib.sh                      |    2 +-
 t/t7527-builtin-fsmonitor.sh            |  609 ++++++++++
 t/test-lib.sh                           |    7 +
 38 files changed, 4326 insertions(+), 106 deletions(-)
 create mode 100644 Documentation/git-fsmonitor--daemon.txt
 create mode 100644 builtin/fsmonitor--daemon.c
 create mode 100644 compat/fsmonitor/fsm-darwin-gcc.h
 create mode 100644 compat/fsmonitor/fsm-listen-darwin.c
 create mode 100644 compat/fsmonitor/fsm-listen-win32.c
 create mode 100644 compat/fsmonitor/fsm-listen.h
 create mode 100644 fsmonitor--daemon.h
 create mode 100644 fsmonitor-ipc.c
 create mode 100644 fsmonitor-ipc.h
 create mode 100644 fsmonitor-settings.c
 create mode 100644 fsmonitor-settings.h
 create mode 100644 t/helper/test-fsmonitor-client.c
 create mode 100755 t/t7527-builtin-fsmonitor.sh


base-commit: 715d08a9e51251ad8290b181b6ac3b9e1f9719d7
Published-As: https://github.com/gitgitgadget/git/releases/tag/pr-1041%2Fjeffhostetler%2Fbuiltin-fsmonitor-part2-v9
Fetch-It-Via: git fetch https://github.com/gitgitgadget/git pr-1041/jeffhostetler/builtin-fsmonitor-part2-v9
Pull-Request: https://github.com/gitgitgadget/git/pull/1041

Range-diff vs v8:

  1:  e98373f997f =  1:  e98373f997f fsmonitor: enhance existing comments, clarify trivial response handling
  2:  ab68b944173 =  2:  ab68b944173 fsmonitor-ipc: create client routines for git-fsmonitor--daemon
  3:  e04c7301f24 =  3:  e04c7301f24 fsmonitor: config settings are repository-specific
  4:  ea02ba25d8f =  4:  ea02ba25d8f fsmonitor: use IPC to query the builtin FSMonitor daemon
  5:  6ab7db9cb76 =  5:  6ab7db9cb76 fsmonitor: document builtin fsmonitor
  6:  0ce8ae3f2cf =  6:  0ce8ae3f2cf fsmonitor--daemon: add a built-in fsmonitor daemon
  7:  4624ce2fa47 =  7:  4624ce2fa47 fsmonitor--daemon: implement 'stop' and 'status' commands
  8:  a29fe7266a4 =  8:  a29fe7266a4 compat/fsmonitor/fsm-listen-win32: stub in backend for Windows
  9:  2f8a42fdb93 =  9:  2f8a42fdb93 compat/fsmonitor/fsm-listen-darwin: stub in backend for Darwin
 10:  f07800690ee = 10:  f07800690ee fsmonitor--daemon: implement 'run' command
 11:  a6a39a3306d = 11:  a6a39a3306d fsmonitor--daemon: implement 'start' command
 12:  d62e338d008 = 12:  d62e338d008 fsmonitor--daemon: add pathname classification
 13:  53e06b4ae5d = 13:  53e06b4ae5d fsmonitor--daemon: define token-ids
 14:  39f43fabe02 = 14:  39f43fabe02 fsmonitor--daemon: create token-based changed path cache
 15:  239558e34ff = 15:  239558e34ff compat/fsmonitor/fsm-listen-win32: implement FSMonitor backend on Windows
 16:  14b775e9d8b = 16:  14b775e9d8b compat/fsmonitor/fsm-listen-darwin: add MacOS header files for FSEvent
 17:  55bd7aee06c = 17:  55bd7aee06c compat/fsmonitor/fsm-listen-darwin: implement FSEvent listener on MacOS
 18:  c43009124fb = 18:  c43009124fb fsmonitor--daemon: implement handle_client callback
 19:  ed338777b56 = 19:  ed338777b56 help: include fsmonitor--daemon feature flag in version info
 20:  c99bac29d42 = 20:  c99bac29d42 t/helper/fsmonitor-client: create IPC client to talk to FSMonitor Daemon
 21:  c8709da9457 ! 21:  bc94e379b03 t7527: create test for fsmonitor--daemon
     @@ Commit message
          t7527: create test for fsmonitor--daemon
      
          Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
     -    Signed-off-by: Junio C Hamano <gitster@pobox.com>
      
       ## t/t7527-builtin-fsmonitor.sh (new) ##
      @@
     @@ t/t7527-builtin-fsmonitor.sh (new)
      +	rm -rf $1
      +}
      +
     -+is_value () {
     -+	test -n "$1" && test "${1::1}" != "-"
     -+}
     -+
      +start_daemon () {
     -+	r= &&
     -+	tf= &&
     -+	t2= &&
     -+	tk= &&
     ++	r= tf= t2= tk= &&
      +
      +	while test "$#" -ne 0
      +	do
      +		case "$1" in
      +		-C)
     -+			shift;
     -+			is_value $1 || BUG "error: -C requires value"
     -+			r="-C $1"
     ++			r="-C ${2?}"
      +			shift
      +			;;
      +		--tf)
     -+			shift;
     -+			is_value $1 || BUG "error: --tf requires value"
     -+			tf="$1"
     ++			tf="${2?}"
      +			shift
      +			;;
      +		--t2)
     -+			shift;
     -+			is_value $1 || BUG "error: --t2 requires value"
     -+			t2="$1"
     ++			t2="${2?}"
      +			shift
      +			;;
      +		--tk)
     -+			shift;
     -+			is_value $1 || BUG "error: --tk requires value"
     -+			tk="$1"
     ++			tk="${2?}"
      +			shift
      +			;;
     -+		*)
     ++		-*)
      +			BUG "error: unknown option: '$1'"
      +			;;
     ++		*)
     ++			BUG "error: unbound argument: '$1'"
     ++			;;
      +		esac
     ++		shift
      +	done &&
      +
      +	(
 22:  cc39ecf10ae = 22:  06d56d3a733 t/perf: avoid copying builtin fsmonitor files into test repo
 23:  2bb3eb84767 = 23:  2dd02151278 t/helper/test-chmtime: skip directories on Windows
 24:  bab9a9b0802 = 24:  bb88cddc137 t/perf/p7519: fix coding style
 25:  2dd06ad2f71 = 25:  50c2afaa49e t/perf/p7519: speed up test on Windows
 26:  6eaa5765ae1 = 26:  5b18e3b6926 t/perf/p7519: add fsmonitor--daemon test cases
 27:  30957f3930e = 27:  899c23f63c3 fsmonitor--daemon: periodically truncate list of modified files
 28:  c8ca2a17277 = 28:  597a7192f94 fsmonitor--daemon: use a cookie file to sync with file system
 29:  4caf1d89b84 = 29:  68a05fd2892 fsmonitor: force update index after large responses
 30:  f87a1eba693 = 30:  5eb696daba2 t7527: test status with untracked-cache and fsmonitor--daemon

-- 
gitgitgadget

[PATCH v9 01/30] fsmonitor: enhance existing comments, clarify trivial response handling

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 fsmonitor.c | 64 ++++++++++++++++++++++++++++++++++-------------------
 1 file changed, 41 insertions(+), 23 deletions(-)

diff --git a/fsmonitor.c b/fsmonitor.c
index ab9bfc60b34..448d0ee33f5 100644
--- a/fsmonitor.c
+++ b/fsmonitor.c
@@ -168,29 +168,15 @@ static int query_fsmonitor(int version, const char *last_update, struct strbuf *
 
 	if (result)
 		trace2_data_intmax("fsm_hook", NULL, "query/failed", result);
-	else {
+	else
 		trace2_data_intmax("fsm_hook", NULL, "query/response-length",
 				   query_result->len);
 
-		if (fsmonitor_is_trivial_response(query_result))
-			trace2_data_intmax("fsm_hook", NULL,
-					   "query/trivial-response", 1);
-	}
-
 	trace2_region_leave("fsm_hook", "query", NULL);
 
 	return result;
 }
 
-int fsmonitor_is_trivial_response(const struct strbuf *query_result)
-{
-	static char trivial_response[3] = { '\0', '/', '\0' };
-
-	return query_result->len >= 3 &&
-		!memcmp(trivial_response,
-			&query_result->buf[query_result->len - 3], 3);
-}
-
 static void fsmonitor_refresh_callback(struct index_state *istate, char *name)
 {
 	int i, len = strlen(name);
@@ -238,6 +224,7 @@ void refresh_fsmonitor(struct index_state *istate)
 	struct strbuf last_update_token = STRBUF_INIT;
 	char *buf;
 	unsigned int i;
+	int is_trivial = 0;
 
 	if (!core_fsmonitor || istate->fsmonitor_has_run_once)
 		return;
@@ -283,6 +270,7 @@ void refresh_fsmonitor(struct index_state *istate)
 					query_success = 0;
 				} else {
 					bol = last_update_token.len + 1;
+					is_trivial = query_result.buf[bol] == '/';
 				}
 			} else if (hook_version < 0) {
 				hook_version = HOOK_INTERFACE_VERSION1;
@@ -294,16 +282,38 @@ void refresh_fsmonitor(struct index_state *istate)
 		if (hook_version == HOOK_INTERFACE_VERSION1) {
 			query_success = !query_fsmonitor(HOOK_INTERFACE_VERSION1,
 				istate->fsmonitor_last_update, &query_result);
+			if (query_success)
+				is_trivial = query_result.buf[0] == '/';
 		}
 
+		if (is_trivial)
+			trace2_data_intmax("fsm_hook", NULL,
+					   "query/trivial-response", 1);
+
 		trace_performance_since(last_update, "fsmonitor process '%s'", core_fsmonitor);
 		trace_printf_key(&trace_fsmonitor, "fsmonitor process '%s' returned %s",
 			core_fsmonitor, query_success ? "success" : "failure");
 	}
 
-	/* a fsmonitor process can return '/' to indicate all entries are invalid */
-	if (query_success && query_result.buf[bol] != '/') {
-		/* Mark all entries returned by the monitor as dirty */
+	/*
+	 * The response from FSMonitor (excluding the header token) is
+	 * either:
+	 *
+	 * [a] a (possibly empty) list of NUL delimited relative
+	 *     pathnames of changed paths.  This list can contain
+	 *     files and directories.  Directories have a trailing
+	 *     slash.
+	 *
+	 * [b] a single '/' to indicate the provider had no
+	 *     information and that we should consider everything
+	 *     invalid.  We call this a trivial response.
+	 */
+	if (query_success && !is_trivial) {
+		/*
+		 * Mark all pathnames returned by the monitor as dirty.
+		 *
+		 * This updates both the cache-entries and the untracked-cache.
+		 */
 		buf = query_result.buf;
 		for (i = bol; i < query_result.len; i++) {
 			if (buf[i] != '\0')
@@ -318,11 +328,16 @@ void refresh_fsmonitor(struct index_state *istate)
 		if (istate->untracked)
 			istate->untracked->use_fsmonitor = 1;
 	} else {
-
-		/* We only want to run the post index changed hook if we've actually changed entries, so keep track
-		 * if we actually changed entries or not */
+		/*
+		 * We failed to get a response or received a trivial response,
+		 * so invalidate everything.
+		 *
+		 * We only want to run the post index changed hook if
+		 * we've actually changed entries, so keep track if we
+		 * actually changed entries or not.
+		 */
 		int is_cache_changed = 0;
-		/* Mark all entries invalid */
+
 		for (i = 0; i < istate->cache_nr; i++) {
 			if (istate->cache[i]->ce_flags & CE_FSMONITOR_VALID) {
 				is_cache_changed = 1;
@@ -330,7 +345,10 @@ void refresh_fsmonitor(struct index_state *istate)
 			}
 		}
 
-		/* If we're going to check every file, ensure we save the results */
+		/*
+		 * If we're going to check every file, ensure we save
+		 * the results.
+		 */
 		if (is_cache_changed)
 			istate->cache_changed |= FSMONITOR_CHANGED;
 
-- 
gitgitgadget

[PATCH v9 02/30] fsmonitor-ipc: create client routines for git-fsmonitor--daemon

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Create fsmonitor_ipc__*() client routines to spawn the built-in file
system monitor daemon and send it an IPC request using the `Simple
IPC` API.

Stub in empty fsmonitor_ipc__*() functions for unsupported platforms.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 Makefile        |   1 +
 fsmonitor-ipc.c | 171 ++++++++++++++++++++++++++++++++++++++++++++++++
 fsmonitor-ipc.h |  48 ++++++++++++++
 3 files changed, 220 insertions(+)
 create mode 100644 fsmonitor-ipc.c
 create mode 100644 fsmonitor-ipc.h

diff --git a/Makefile b/Makefile
index 6f0b4b775fe..a19d850e716 100644
--- a/Makefile
+++ b/Makefile
@@ -907,6 +907,7 @@ LIB_OBJS += fetch-pack.o
 LIB_OBJS += fmt-merge-msg.o
 LIB_OBJS += fsck.o
 LIB_OBJS += fsmonitor.o
+LIB_OBJS += fsmonitor-ipc.o
 LIB_OBJS += gettext.o
 LIB_OBJS += gpg-interface.o
 LIB_OBJS += graph.o
diff --git a/fsmonitor-ipc.c b/fsmonitor-ipc.c
new file mode 100644
index 00000000000..789e7397baa
--- /dev/null
+++ b/fsmonitor-ipc.c
@@ -0,0 +1,171 @@
+#include "cache.h"
+#include "fsmonitor.h"
+#include "simple-ipc.h"
+#include "fsmonitor-ipc.h"
+#include "run-command.h"
+#include "strbuf.h"
+#include "trace2.h"
+
+#ifndef HAVE_FSMONITOR_DAEMON_BACKEND
+
+/*
+ * A trivial implementation of the fsmonitor_ipc__ API for unsupported
+ * platforms.
+ */
+
+int fsmonitor_ipc__is_supported(void)
+{
+	return 0;
+}
+
+const char *fsmonitor_ipc__get_path(void)
+{
+	return NULL;
+}
+
+enum ipc_active_state fsmonitor_ipc__get_state(void)
+{
+	return IPC_STATE__OTHER_ERROR;
+}
+
+int fsmonitor_ipc__send_query(const char *since_token,
+			      struct strbuf *answer)
+{
+	return -1;
+}
+
+int fsmonitor_ipc__send_command(const char *command,
+				struct strbuf *answer)
+{
+	return -1;
+}
+
+#else
+
+int fsmonitor_ipc__is_supported(void)
+{
+	return 1;
+}
+
+GIT_PATH_FUNC(fsmonitor_ipc__get_path, "fsmonitor--daemon.ipc")
+
+enum ipc_active_state fsmonitor_ipc__get_state(void)
+{
+	return ipc_get_active_state(fsmonitor_ipc__get_path());
+}
+
+static int spawn_daemon(void)
+{
+	const char *args[] = { "fsmonitor--daemon", "start", NULL };
+
+	return run_command_v_opt_tr2(args, RUN_COMMAND_NO_STDIN | RUN_GIT_CMD,
+				    "fsmonitor");
+}
+
+int fsmonitor_ipc__send_query(const char *since_token,
+			      struct strbuf *answer)
+{
+	int ret = -1;
+	int tried_to_spawn = 0;
+	enum ipc_active_state state = IPC_STATE__OTHER_ERROR;
+	struct ipc_client_connection *connection = NULL;
+	struct ipc_client_connect_options options
+		= IPC_CLIENT_CONNECT_OPTIONS_INIT;
+	const char *tok = since_token ? since_token : "";
+	size_t tok_len = since_token ? strlen(since_token) : 0;
+
+	options.wait_if_busy = 1;
+	options.wait_if_not_found = 0;
+
+	trace2_region_enter("fsm_client", "query", NULL);
+	trace2_data_string("fsm_client", NULL, "query/command", tok);
+
+try_again:
+	state = ipc_client_try_connect(fsmonitor_ipc__get_path(), &options,
+				       &connection);
+
+	switch (state) {
+	case IPC_STATE__LISTENING:
+		ret = ipc_client_send_command_to_connection(
+			connection, tok, tok_len, answer);
+		ipc_client_close_connection(connection);
+
+		trace2_data_intmax("fsm_client", NULL,
+				   "query/response-length", answer->len);
+		goto done;
+
+	case IPC_STATE__NOT_LISTENING:
+	case IPC_STATE__PATH_NOT_FOUND:
+		if (tried_to_spawn)
+			goto done;
+
+		tried_to_spawn++;
+		if (spawn_daemon())
+			goto done;
+
+		/*
+		 * Try again, but this time give the daemon a chance to
+		 * actually create the pipe/socket.
+		 *
+		 * Granted, the daemon just started so it can't possibly have
+		 * any FS cached yet, so we'll always get a trivial answer.
+		 * BUT the answer should include a new token that can serve
+		 * as the basis for subsequent requests.
+		 */
+		options.wait_if_not_found = 1;
+		goto try_again;
+
+	case IPC_STATE__INVALID_PATH:
+		ret = error(_("fsmonitor_ipc__send_query: invalid path '%s'"),
+			    fsmonitor_ipc__get_path());
+		goto done;
+
+	case IPC_STATE__OTHER_ERROR:
+	default:
+		ret = error(_("fsmonitor_ipc__send_query: unspecified error on '%s'"),
+			    fsmonitor_ipc__get_path());
+		goto done;
+	}
+
+done:
+	trace2_region_leave("fsm_client", "query", NULL);
+
+	return ret;
+}
+
+int fsmonitor_ipc__send_command(const char *command,
+				struct strbuf *answer)
+{
+	struct ipc_client_connection *connection = NULL;
+	struct ipc_client_connect_options options
+		= IPC_CLIENT_CONNECT_OPTIONS_INIT;
+	int ret;
+	enum ipc_active_state state;
+	const char *c = command ? command : "";
+	size_t c_len = command ? strlen(command) : 0;
+
+	strbuf_reset(answer);
+
+	options.wait_if_busy = 1;
+	options.wait_if_not_found = 0;
+
+	state = ipc_client_try_connect(fsmonitor_ipc__get_path(), &options,
+				       &connection);
+	if (state != IPC_STATE__LISTENING) {
+		die(_("fsmonitor--daemon is not running"));
+		return -1;
+	}
+
+	ret = ipc_client_send_command_to_connection(connection, c, c_len,
+						    answer);
+	ipc_client_close_connection(connection);
+
+	if (ret == -1) {
+		die(_("could not send '%s' command to fsmonitor--daemon"), c);
+		return -1;
+	}
+
+	return 0;
+}
+
+#endif
diff --git a/fsmonitor-ipc.h b/fsmonitor-ipc.h
new file mode 100644
index 00000000000..b6a7067c3af
--- /dev/null
+++ b/fsmonitor-ipc.h
@@ -0,0 +1,48 @@
+#ifndef FSMONITOR_IPC_H
+#define FSMONITOR_IPC_H
+
+#include "simple-ipc.h"
+
+/*
+ * Returns true if built-in file system monitor daemon is defined
+ * for this platform.
+ */
+int fsmonitor_ipc__is_supported(void);
+
+/*
+ * Returns the pathname to the IPC named pipe or Unix domain socket
+ * where a `git-fsmonitor--daemon` process will listen.  This is a
+ * per-worktree value.
+ *
+ * Returns NULL if the daemon is not supported on this platform.
+ */
+const char *fsmonitor_ipc__get_path(void);
+
+/*
+ * Try to determine whether there is a `git-fsmonitor--daemon` process
+ * listening on the IPC pipe/socket.
+ */
+enum ipc_active_state fsmonitor_ipc__get_state(void);
+
+/*
+ * Connect to a `git-fsmonitor--daemon` process via simple-ipc
+ * and ask for the set of changed files since the given token.
+ *
+ * Spawn a daemon process in the background if necessary.
+ *
+ * Returns -1 on error; 0 on success.
+ */
+int fsmonitor_ipc__send_query(const char *since_token,
+			      struct strbuf *answer);
+
+/*
+ * Connect to a `git-fsmonitor--daemon` process via simple-ipc and
+ * send a command verb.  If no daemon is available, we DO NOT try to
+ * start one.
+ *
+ * Returns -1 on error; 0 on success.
+ */
+int fsmonitor_ipc__send_command(const char *command,
+				struct strbuf *answer);
+
+#endif /* FSMONITOR_IPC_H */
-- 
gitgitgadget

[PATCH v9 03/30] fsmonitor: config settings are repository-specific

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Move fsmonitor config settings to a new and opaque
`struct fsmonitor_settings` structure.  Add a lazily-loaded pointer
to this into `struct repo_settings`

Create an `enum fsmonitor_mode` type in `struct fsmonitor_settings` to
represent the state of fsmonitor.  This lets us represent which, if
any, fsmonitor provider (hook or IPC) is enabled.

Create `fsm_settings__get_*()` getters to lazily look up fsmonitor-
related config settings.

Get rid of the `core_fsmonitor` global variable.  Move the code to
lookup the existing `core.fsmonitor` config value into the fsmonitor
settings.

Create a hook pathname variable in `struct fsmonitor-settings` and
only set it when in hook mode.

Extend the definition of `core.fsmonitor` to be either a boolean
or a hook pathname.  When true, the builtin FSMonitor is used.
When false or unset, no FSMonitor (neither builtin nor hook) is
used.

The existing `core_fsmonitor` global variable was used to store the
pathname to the fsmonitor hook *and* it was used as a boolean to see
if fsmonitor was enabled.  This dual usage and global visibility leads
to confusion when we add the IPC-based provider.  So lets hide the
details in fsmonitor-settings.c and let it decide which provider to
use in the case of multiple settings.  This avoids cluttering up
repo-settings.c with these private details.

A future commit in builtin-fsmonitor series will add the ability to
disqualify worktrees for various reasons, such as being mounted from a
remote volume, where fsmonitor should not be started.  Having the
config settings hidden in fsmonitor-settings.c allows such worktree
restrictions to override the config values used.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 Makefile               |   1 +
 builtin/update-index.c |   7 ++-
 cache.h                |   1 -
 config.c               |  14 -----
 config.h               |   1 -
 environment.c          |   1 -
 fsmonitor-settings.c   | 114 +++++++++++++++++++++++++++++++++++++++++
 fsmonitor-settings.h   |  21 ++++++++
 fsmonitor.c            |  63 ++++++++++++++---------
 fsmonitor.h            |  15 ++++--
 repository.h           |   3 ++
 t/README               |   4 +-
 12 files changed, 196 insertions(+), 49 deletions(-)
 create mode 100644 fsmonitor-settings.c
 create mode 100644 fsmonitor-settings.h

diff --git a/Makefile b/Makefile
index a19d850e716..707a56d4c11 100644
--- a/Makefile
+++ b/Makefile
@@ -908,6 +908,7 @@ LIB_OBJS += fmt-merge-msg.o
 LIB_OBJS += fsck.o
 LIB_OBJS += fsmonitor.o
 LIB_OBJS += fsmonitor-ipc.o
+LIB_OBJS += fsmonitor-settings.o
 LIB_OBJS += gettext.o
 LIB_OBJS += gpg-interface.o
 LIB_OBJS += graph.o
diff --git a/builtin/update-index.c b/builtin/update-index.c
index aafe7eeac2a..876112abb21 100644
--- a/builtin/update-index.c
+++ b/builtin/update-index.c
@@ -1236,14 +1236,17 @@ int cmd_update_index(int argc, const char **argv, const char *prefix)
 	}
 
 	if (fsmonitor > 0) {
-		if (git_config_get_fsmonitor() == 0)
+		enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
+		if (fsm_mode == FSMONITOR_MODE_DISABLED) {
 			warning(_("core.fsmonitor is unset; "
 				"set it if you really want to "
 				"enable fsmonitor"));
+		}
 		add_fsmonitor(&the_index);
 		report(_("fsmonitor enabled"));
 	} else if (!fsmonitor) {
-		if (git_config_get_fsmonitor() == 1)
+		enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
+		if (fsm_mode > FSMONITOR_MODE_DISABLED)
 			warning(_("core.fsmonitor is set; "
 				"remove it if you really want to "
 				"disable fsmonitor"));
diff --git a/cache.h b/cache.h
index 04d4d2db25c..aaf334e2aa4 100644
--- a/cache.h
+++ b/cache.h
@@ -999,7 +999,6 @@ extern int core_preload_index;
 extern int precomposed_unicode;
 extern int protect_hfs;
 extern int protect_ntfs;
-extern const char *core_fsmonitor;
 
 extern int core_apply_sparse_checkout;
 extern int core_sparse_checkout_cone;
diff --git a/config.c b/config.c
index 383b1a4885b..3f9b0739a78 100644
--- a/config.c
+++ b/config.c
@@ -2626,20 +2626,6 @@ int git_config_get_max_percent_split_change(void)
 	return -1; /* default value */
 }
 
-int git_config_get_fsmonitor(void)
-{
-	if (git_config_get_pathname("core.fsmonitor", &core_fsmonitor))
-		core_fsmonitor = getenv("GIT_TEST_FSMONITOR");
-
-	if (core_fsmonitor && !*core_fsmonitor)
-		core_fsmonitor = NULL;
-
-	if (core_fsmonitor)
-		return 1;
-
-	return 0;
-}
-
 int git_config_get_index_threads(int *dest)
 {
 	int is_bool, val;
diff --git a/config.h b/config.h
index bb49baf1ee0..7654f61c634 100644
--- a/config.h
+++ b/config.h
@@ -597,7 +597,6 @@ int git_config_get_pathname(const char *key, const char **dest);
 int git_config_get_index_threads(int *dest);
 int git_config_get_split_index(void);
 int git_config_get_max_percent_split_change(void);
-int git_config_get_fsmonitor(void);
 
 /* This dies if the configured or default date is in the future */
 int git_config_get_expiry(const char *key, const char **output);
diff --git a/environment.c b/environment.c
index fd0501e77a5..00682e638d7 100644
--- a/environment.c
+++ b/environment.c
@@ -84,7 +84,6 @@ int protect_hfs = PROTECT_HFS_DEFAULT;
 #define PROTECT_NTFS_DEFAULT 1
 #endif
 int protect_ntfs = PROTECT_NTFS_DEFAULT;
-const char *core_fsmonitor;
 
 /*
  * The character that begins a commented line in user-editable file
diff --git a/fsmonitor-settings.c b/fsmonitor-settings.c
new file mode 100644
index 00000000000..757d230d538
--- /dev/null
+++ b/fsmonitor-settings.c
@@ -0,0 +1,114 @@
+#include "cache.h"
+#include "config.h"
+#include "repository.h"
+#include "fsmonitor-settings.h"
+
+/*
+ * We keep this structure defintion private and have getters
+ * for all fields so that we can lazy load it as needed.
+ */
+struct fsmonitor_settings {
+	enum fsmonitor_mode mode;
+	char *hook_path;
+};
+
+static void lookup_fsmonitor_settings(struct repository *r)
+{
+	struct fsmonitor_settings *s;
+	const char *const_str;
+	int bool_value;
+
+	if (r->settings.fsmonitor)
+		return;
+
+	CALLOC_ARRAY(s, 1);
+	s->mode = FSMONITOR_MODE_DISABLED;
+
+	r->settings.fsmonitor = s;
+
+	/*
+	 * Overload the existing "core.fsmonitor" config setting (which
+	 * has historically been either unset or a hook pathname) to
+	 * now allow a boolean value to enable the builtin FSMonitor
+	 * or to turn everything off.  (This does imply that you can't
+	 * use a hook script named "true" or "false", but that's OK.)
+	 */
+	switch (repo_config_get_maybe_bool(r, "core.fsmonitor", &bool_value)) {
+
+	case 0: /* config value was set to <bool> */
+		if (bool_value)
+			fsm_settings__set_ipc(r);
+		return;
+
+	case 1: /* config value was unset */
+		const_str = getenv("GIT_TEST_FSMONITOR");
+		break;
+
+	case -1: /* config value set to an arbitrary string */
+		if (repo_config_get_pathname(r, "core.fsmonitor", &const_str))
+			return; /* should not happen */
+		break;
+
+	default: /* should not happen */
+		return;
+	}
+
+	if (!const_str || !*const_str)
+		return;
+
+	fsm_settings__set_hook(r, const_str);
+}
+
+enum fsmonitor_mode fsm_settings__get_mode(struct repository *r)
+{
+	if (!r)
+		r = the_repository;
+
+	lookup_fsmonitor_settings(r);
+
+	return r->settings.fsmonitor->mode;
+}
+
+const char *fsm_settings__get_hook_path(struct repository *r)
+{
+	if (!r)
+		r = the_repository;
+
+	lookup_fsmonitor_settings(r);
+
+	return r->settings.fsmonitor->hook_path;
+}
+
+void fsm_settings__set_ipc(struct repository *r)
+{
+	if (!r)
+		r = the_repository;
+
+	lookup_fsmonitor_settings(r);
+
+	r->settings.fsmonitor->mode = FSMONITOR_MODE_IPC;
+	FREE_AND_NULL(r->settings.fsmonitor->hook_path);
+}
+
+void fsm_settings__set_hook(struct repository *r, const char *path)
+{
+	if (!r)
+		r = the_repository;
+
+	lookup_fsmonitor_settings(r);
+
+	r->settings.fsmonitor->mode = FSMONITOR_MODE_HOOK;
+	FREE_AND_NULL(r->settings.fsmonitor->hook_path);
+	r->settings.fsmonitor->hook_path = strdup(path);
+}
+
+void fsm_settings__set_disabled(struct repository *r)
+{
+	if (!r)
+		r = the_repository;
+
+	lookup_fsmonitor_settings(r);
+
+	r->settings.fsmonitor->mode = FSMONITOR_MODE_DISABLED;
+	FREE_AND_NULL(r->settings.fsmonitor->hook_path);
+}
diff --git a/fsmonitor-settings.h b/fsmonitor-settings.h
new file mode 100644
index 00000000000..a4c5d7b4889
--- /dev/null
+++ b/fsmonitor-settings.h
@@ -0,0 +1,21 @@
+#ifndef FSMONITOR_SETTINGS_H
+#define FSMONITOR_SETTINGS_H
+
+struct repository;
+
+enum fsmonitor_mode {
+	FSMONITOR_MODE_DISABLED = 0,
+	FSMONITOR_MODE_HOOK = 1, /* core.fsmonitor=<hook_path> */
+	FSMONITOR_MODE_IPC = 2,  /* core.fsmonitor=<true> */
+};
+
+void fsm_settings__set_ipc(struct repository *r);
+void fsm_settings__set_hook(struct repository *r, const char *path);
+void fsm_settings__set_disabled(struct repository *r);
+
+enum fsmonitor_mode fsm_settings__get_mode(struct repository *r);
+const char *fsm_settings__get_hook_path(struct repository *r);
+
+struct fsmonitor_settings;
+
+#endif /* FSMONITOR_SETTINGS_H */
diff --git a/fsmonitor.c b/fsmonitor.c
index 448d0ee33f5..0e961b74d82 100644
--- a/fsmonitor.c
+++ b/fsmonitor.c
@@ -3,6 +3,7 @@
 #include "dir.h"
 #include "ewah/ewok.h"
 #include "fsmonitor.h"
+#include "fsmonitor-ipc.h"
 #include "run-command.h"
 #include "strbuf.h"
 
@@ -148,15 +149,18 @@ void write_fsmonitor_extension(struct strbuf *sb, struct index_state *istate)
 /*
  * Call the query-fsmonitor hook passing the last update token of the saved results.
  */
-static int query_fsmonitor(int version, const char *last_update, struct strbuf *query_result)
+static int query_fsmonitor_hook(struct repository *r,
+				int version,
+				const char *last_update,
+				struct strbuf *query_result)
 {
 	struct child_process cp = CHILD_PROCESS_INIT;
 	int result;
 
-	if (!core_fsmonitor)
+	if (fsm_settings__get_mode(r) != FSMONITOR_MODE_HOOK)
 		return -1;
 
-	strvec_push(&cp.args, core_fsmonitor);
+	strvec_push(&cp.args, fsm_settings__get_hook_path(r));
 	strvec_pushf(&cp.args, "%d", version);
 	strvec_pushf(&cp.args, "%s", last_update);
 	cp.use_shell = 1;
@@ -225,17 +229,28 @@ void refresh_fsmonitor(struct index_state *istate)
 	char *buf;
 	unsigned int i;
 	int is_trivial = 0;
+	struct repository *r = istate->repo ? istate->repo : the_repository;
+	enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(r);
 
-	if (!core_fsmonitor || istate->fsmonitor_has_run_once)
+	if (fsm_mode <= FSMONITOR_MODE_DISABLED ||
+	    istate->fsmonitor_has_run_once)
 		return;
 
-	hook_version = fsmonitor_hook_version();
-
 	istate->fsmonitor_has_run_once = 1;
 
 	trace_printf_key(&trace_fsmonitor, "refresh fsmonitor");
+
+	if (fsm_mode == FSMONITOR_MODE_IPC) {
+		/* TODO */
+		return;
+	}
+
+	assert(fsm_mode == FSMONITOR_MODE_HOOK);
+
+	hook_version = fsmonitor_hook_version();
+
 	/*
-	 * This could be racy so save the date/time now and query_fsmonitor
+	 * This could be racy so save the date/time now and query_fsmonitor_hook
 	 * should be inclusive to ensure we don't miss potential changes.
 	 */
 	last_update = getnanotime();
@@ -243,13 +258,14 @@ void refresh_fsmonitor(struct index_state *istate)
 		strbuf_addf(&last_update_token, "%"PRIu64"", last_update);
 
 	/*
-	 * If we have a last update token, call query_fsmonitor for the set of
+	 * If we have a last update token, call query_fsmonitor_hook for the set of
 	 * changes since that token, else assume everything is possibly dirty
 	 * and check it all.
 	 */
 	if (istate->fsmonitor_last_update) {
 		if (hook_version == -1 || hook_version == HOOK_INTERFACE_VERSION2) {
-			query_success = !query_fsmonitor(HOOK_INTERFACE_VERSION2,
+			query_success = !query_fsmonitor_hook(
+				r, HOOK_INTERFACE_VERSION2,
 				istate->fsmonitor_last_update, &query_result);
 
 			if (query_success) {
@@ -280,7 +296,8 @@ void refresh_fsmonitor(struct index_state *istate)
 		}
 
 		if (hook_version == HOOK_INTERFACE_VERSION1) {
-			query_success = !query_fsmonitor(HOOK_INTERFACE_VERSION1,
+			query_success = !query_fsmonitor_hook(
+				r, HOOK_INTERFACE_VERSION1,
 				istate->fsmonitor_last_update, &query_result);
 			if (query_success)
 				is_trivial = query_result.buf[0] == '/';
@@ -290,9 +307,12 @@ void refresh_fsmonitor(struct index_state *istate)
 			trace2_data_intmax("fsm_hook", NULL,
 					   "query/trivial-response", 1);
 
-		trace_performance_since(last_update, "fsmonitor process '%s'", core_fsmonitor);
-		trace_printf_key(&trace_fsmonitor, "fsmonitor process '%s' returned %s",
-			core_fsmonitor, query_success ? "success" : "failure");
+		trace_performance_since(last_update, "fsmonitor process '%s'",
+					fsm_settings__get_hook_path(r));
+		trace_printf_key(&trace_fsmonitor,
+				 "fsmonitor process '%s' returned %s",
+				 fsm_settings__get_hook_path(r),
+				 query_success ? "success" : "failure");
 	}
 
 	/*
@@ -429,7 +449,8 @@ void remove_fsmonitor(struct index_state *istate)
 void tweak_fsmonitor(struct index_state *istate)
 {
 	unsigned int i;
-	int fsmonitor_enabled = git_config_get_fsmonitor();
+	int fsmonitor_enabled = (fsm_settings__get_mode(istate->repo)
+				 > FSMONITOR_MODE_DISABLED);
 
 	if (istate->fsmonitor_dirty) {
 		if (fsmonitor_enabled) {
@@ -449,16 +470,8 @@ void tweak_fsmonitor(struct index_state *istate)
 		istate->fsmonitor_dirty = NULL;
 	}
 
-	switch (fsmonitor_enabled) {
-	case -1: /* keep: do nothing */
-		break;
-	case 0: /* false */
-		remove_fsmonitor(istate);
-		break;
-	case 1: /* true */
+	if (fsmonitor_enabled)
 		add_fsmonitor(istate);
-		break;
-	default: /* unknown value: do nothing */
-		break;
-	}
+	else
+		remove_fsmonitor(istate);
 }
diff --git a/fsmonitor.h b/fsmonitor.h
index f20d72631d7..3f41f653691 100644
--- a/fsmonitor.h
+++ b/fsmonitor.h
@@ -3,6 +3,7 @@
 
 #include "cache.h"
 #include "dir.h"
+#include "fsmonitor-settings.h"
 
 extern struct trace_key trace_fsmonitor;
 
@@ -57,7 +58,10 @@ int fsmonitor_is_trivial_response(const struct strbuf *query_result);
  */
 static inline int is_fsmonitor_refreshed(const struct index_state *istate)
 {
-	return !core_fsmonitor || istate->fsmonitor_has_run_once;
+	enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(istate->repo);
+
+	return fsm_mode <= FSMONITOR_MODE_DISABLED ||
+		istate->fsmonitor_has_run_once;
 }
 
 /*
@@ -67,7 +71,10 @@ static inline int is_fsmonitor_refreshed(const struct index_state *istate)
  */
 static inline void mark_fsmonitor_valid(struct index_state *istate, struct cache_entry *ce)
 {
-	if (core_fsmonitor && !(ce->ce_flags & CE_FSMONITOR_VALID)) {
+	enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(istate->repo);
+
+	if (fsm_mode > FSMONITOR_MODE_DISABLED &&
+	    !(ce->ce_flags & CE_FSMONITOR_VALID)) {
 		istate->cache_changed = 1;
 		ce->ce_flags |= CE_FSMONITOR_VALID;
 		trace_printf_key(&trace_fsmonitor, "mark_fsmonitor_clean '%s'", ce->name);
@@ -83,7 +90,9 @@ static inline void mark_fsmonitor_valid(struct index_state *istate, struct cache
  */
 static inline void mark_fsmonitor_invalid(struct index_state *istate, struct cache_entry *ce)
 {
-	if (core_fsmonitor) {
+	enum fsmonitor_mode fsm_mode = fsm_settings__get_mode(istate->repo);
+
+	if (fsm_mode > FSMONITOR_MODE_DISABLED) {
 		ce->ce_flags &= ~CE_FSMONITOR_VALID;
 		untracked_cache_invalidate_path(istate, ce->name, 1);
 		trace_printf_key(&trace_fsmonitor, "mark_fsmonitor_invalid '%s'", ce->name);
diff --git a/repository.h b/repository.h
index ca837cb9e91..9bbb4659cc8 100644
--- a/repository.h
+++ b/repository.h
@@ -4,6 +4,7 @@
 #include "path.h"
 
 struct config_set;
+struct fsmonitor_settings;
 struct git_hash_algo;
 struct index_state;
 struct lock_file;
@@ -35,6 +36,8 @@ struct repo_settings {
 	int command_requires_full_index;
 	int sparse_index;
 
+	struct fsmonitor_settings *fsmonitor; /* lazily loaded */
+
 	int index_version;
 	enum untracked_cache_setting core_untracked_cache;
 
diff --git a/t/README b/t/README
index f48e0542cdc..9ffea1d3147 100644
--- a/t/README
+++ b/t/README
@@ -405,8 +405,8 @@ every 'git commit-graph write', as if the `--changed-paths` option was
 passed in.
 
 GIT_TEST_FSMONITOR=$PWD/t7519/fsmonitor-all exercises the fsmonitor
-code path for utilizing a file system monitor to speed up detecting
-new or changed files.
+code paths for utilizing a (hook based) file system monitor to speed up
+detecting new or changed files.
 
 GIT_TEST_INDEX_VERSION=<n> exercises the index read/write code path
 for the index version specified.  Can be set to any valid version
-- 
gitgitgadget

[PATCH v9 04/30] fsmonitor: use IPC to query the builtin FSMonitor daemon

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Use simple IPC to directly communicate with the new builtin file
system monitor daemon when `core.fsmonitor` is set to true.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 fsmonitor.c | 38 ++++++++++++++++++++++++++++++++++++--
 1 file changed, 36 insertions(+), 2 deletions(-)

diff --git a/fsmonitor.c b/fsmonitor.c
index 0e961b74d82..a38b5710eb3 100644
--- a/fsmonitor.c
+++ b/fsmonitor.c
@@ -241,8 +241,41 @@ void refresh_fsmonitor(struct index_state *istate)
 	trace_printf_key(&trace_fsmonitor, "refresh fsmonitor");
 
 	if (fsm_mode == FSMONITOR_MODE_IPC) {
-		/* TODO */
-		return;
+		query_success = !fsmonitor_ipc__send_query(
+			istate->fsmonitor_last_update ?
+			istate->fsmonitor_last_update : "builtin:fake",
+			&query_result);
+		if (query_success) {
+			/*
+			 * The response contains a series of nul terminated
+			 * strings.  The first is the new token.
+			 *
+			 * Use `char *buf` as an interlude to trick the CI
+			 * static analysis to let us use `strbuf_addstr()`
+			 * here (and only copy the token) rather than
+			 * `strbuf_addbuf()`.
+			 */
+			buf = query_result.buf;
+			strbuf_addstr(&last_update_token, buf);
+			bol = last_update_token.len + 1;
+			is_trivial = query_result.buf[bol] == '/';
+			if (is_trivial)
+				trace2_data_intmax("fsm_client", NULL,
+						   "query/trivial-response", 1);
+		} else {
+			/*
+			 * The builtin daemon is not available on this
+			 * platform -OR- we failed to get a response.
+			 *
+			 * Generate a fake token (rather than a V1
+			 * timestamp) for the index extension.  (If
+			 * they switch back to the hook API, we don't
+			 * want ambiguous state.)
+			 */
+			strbuf_addstr(&last_update_token, "builtin:fake");
+		}
+
+		goto apply_results;
 	}
 
 	assert(fsm_mode == FSMONITOR_MODE_HOOK);
@@ -315,6 +348,7 @@ void refresh_fsmonitor(struct index_state *istate)
 				 query_success ? "success" : "failure");
 	}
 
+apply_results:
 	/*
 	 * The response from FSMonitor (excluding the header token) is
 	 * either:
-- 
gitgitgadget

[PATCH v9 05/30] fsmonitor: document builtin fsmonitor

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Document how `core.fsmonitor` can be set to a boolean to enable
or disable the builtin FSMonitor.

Update references to `core.fsmonitor` and `core.fsmonitorHookVersion` and
pointers to `Watchman` to refer to it.

Create `git-fsmonitor--daemon` manual page and describe its features.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 Documentation/config/core.txt           | 60 +++++++++++++++-----
 Documentation/git-fsmonitor--daemon.txt | 75 +++++++++++++++++++++++++
 Documentation/git-update-index.txt      |  8 ++-
 3 files changed, 126 insertions(+), 17 deletions(-)
 create mode 100644 Documentation/git-fsmonitor--daemon.txt

diff --git a/Documentation/config/core.txt b/Documentation/config/core.txt
index c04f62a54a1..6303c36c7ed 100644
--- a/Documentation/config/core.txt
+++ b/Documentation/config/core.txt
@@ -62,22 +62,54 @@ core.protectNTFS::
 	Defaults to `true` on Windows, and `false` elsewhere.
 
 core.fsmonitor::
-	If set, the value of this variable is used as a command which
-	will identify all files that may have changed since the
-	requested date/time. This information is used to speed up git by
-	avoiding unnecessary processing of files that have not changed.
-	See the "fsmonitor-watchman" section of linkgit:githooks[5].
+	If set to true, enable the built-in file system monitor
+	daemon for this working directory (linkgit:git-fsmonitor--daemon[1]).
++
+Like hook-based file system monitors, the built-in file system monitor
+can speed up Git commands that need to refresh the Git index
+(e.g. `git status`) in a working directory with many files.  The
+built-in monitor eliminates the need to install and maintain an
+external third-party tool.
++
+The built-in file system monitor is currently available only on a
+limited set of supported platforms.  Currently, this includes Windows
+and MacOS.
++
+	Otherwise, this variable contains the pathname of the "fsmonitor"
+	hook command.
++
+This hook command is used to identify all files that may have changed
+since the requested date/time. This information is used to speed up
+git by avoiding unnecessary scanning of files that have not changed.
++
+See the "fsmonitor-watchman" section of linkgit:githooks[5].
++
+Note that if you concurrently use multiple versions of Git, such
+as one version on the command line and another version in an IDE
+tool, that the definition of `core.fsmonitor` was extended to
+allow boolean values in addition to hook pathnames.  Git versions
+2.35.1 and prior will not understand the boolean values and will
+consider the "true" or "false" values as hook pathnames to be
+invoked.  Git versions 2.26 thru 2.35.1 default to hook protocol
+V2 and will fall back to no fsmonitor (full scan).  Git versions
+prior to 2.26 default to hook protocol V1 and will silently
+assume there were no changes to report (no scan), so status
+commands may report incomplete results.  For this reason, it is
+best to upgrade all of your Git versions before using the built-in
+file system monitor.
 
 core.fsmonitorHookVersion::
-	Sets the version of hook that is to be used when calling fsmonitor.
-	There are currently versions 1 and 2. When this is not set,
-	version 2 will be tried first and if it fails then version 1
-	will be tried. Version 1 uses a timestamp as input to determine
-	which files have changes since that time but some monitors
-	like watchman have race conditions when used with a timestamp.
-	Version 2 uses an opaque string so that the monitor can return
-	something that can be used to determine what files have changed
-	without race conditions.
+	Sets the protocol version to be used when invoking the
+	"fsmonitor" hook.
++
+There are currently versions 1 and 2. When this is not set,
+version 2 will be tried first and if it fails then version 1
+will be tried. Version 1 uses a timestamp as input to determine
+which files have changes since that time but some monitors
+like Watchman have race conditions when used with a timestamp.
+Version 2 uses an opaque string so that the monitor can return
+something that can be used to determine what files have changed
+without race conditions.
 
 core.trustctime::
 	If false, the ctime differences between the index and the
diff --git a/Documentation/git-fsmonitor--daemon.txt b/Documentation/git-fsmonitor--daemon.txt
new file mode 100644
index 00000000000..0fedf5a4565
--- /dev/null
+++ b/Documentation/git-fsmonitor--daemon.txt
@@ -0,0 +1,75 @@
+git-fsmonitor--daemon(1)
+========================
+
+NAME
+----
+git-fsmonitor--daemon - A Built-in File System Monitor
+
+SYNOPSIS
+--------
+[verse]
+'git fsmonitor--daemon' start
+'git fsmonitor--daemon' run
+'git fsmonitor--daemon' stop
+'git fsmonitor--daemon' status
+
+DESCRIPTION
+-----------
+
+A daemon to watch the working directory for file and directory
+changes using platform-specific file system notification facilities.
+
+This daemon communicates directly with commands like `git status`
+using the link:technical/api-simple-ipc.html[simple IPC] interface
+instead of the slower linkgit:githooks[5] interface.
+
+This daemon is built into Git so that no third-party tools are
+required.
+
+OPTIONS
+-------
+
+start::
+	Starts a daemon in the background.
+
+run::
+	Runs a daemon in the foreground.
+
+stop::
+	Stops the daemon running in the current working
+	directory, if present.
+
+status::
+	Exits with zero status if a daemon is watching the
+	current working directory.
+
+REMARKS
+-------
+
+This daemon is a long running process used to watch a single working
+directory and maintain a list of the recently changed files and
+directories.  Performance of commands such as `git status` can be
+increased if they just ask for a summary of changes to the working
+directory and can avoid scanning the disk.
+
+When `core.fsmonitor` is set to `true` (see linkgit:git-config[1])
+commands, such as `git status`, will ask the daemon for changes and
+automatically start it (if necessary).
+
+For more information see the "File System Monitor" section in
+linkgit:git-update-index[1].
+
+CAVEATS
+-------
+
+The fsmonitor daemon does not currently know about submodules and does
+not know to filter out file system events that happen within a
+submodule.  If fsmonitor daemon is watching a super repo and a file is
+modified within the working directory of a submodule, it will report
+the change (as happening against the super repo).  However, the client
+will properly ignore these extra events, so performance may be affected
+but it will not cause an incorrect result.
+
+GIT
+---
+Part of the linkgit:git[1] suite
diff --git a/Documentation/git-update-index.txt b/Documentation/git-update-index.txt
index 2853f168d97..53ea48a04e2 100644
--- a/Documentation/git-update-index.txt
+++ b/Documentation/git-update-index.txt
@@ -498,7 +498,9 @@ FILE SYSTEM MONITOR
 This feature is intended to speed up git operations for repos that have
 large working directories.
 
-It enables git to work together with a file system monitor (see the
+It enables git to work together with a file system monitor (see
+linkgit:git-fsmonitor--daemon[1]
+and the
 "fsmonitor-watchman" section of linkgit:githooks[5]) that can
 inform it as to what files have been modified. This enables git to avoid
 having to lstat() every file to find modified files.
@@ -509,8 +511,8 @@ looking for new files.
 
 If you want to enable (or disable) this feature, it is easier to use
 the `core.fsmonitor` configuration variable (see
-linkgit:git-config[1]) than using the `--fsmonitor` option to
-`git update-index` in each repository, especially if you want to do so
+linkgit:git-config[1]) than using the `--fsmonitor` option to `git
+update-index` in each repository, especially if you want to do so
 across all repositories you use, because you can set the configuration
 variable in your `$HOME/.gitconfig` just once and have it affect all
 repositories you touch.
-- 
gitgitgadget

[PATCH v9 06/30] fsmonitor--daemon: add a built-in fsmonitor daemon

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Create a built-in file system monitoring daemon that can be used by
the existing `fsmonitor` feature (protocol API and index extension)
to improve the performance of various Git commands, such as `status`.

The `fsmonitor--daemon` feature builds upon the `Simple IPC` API and
provides an alternative to hook access to existing fsmonitors such
as `watchman`.

This commit merely adds the new command without any functionality.

Co-authored-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 .gitignore                  |  1 +
 Makefile                    |  1 +
 builtin.h                   |  1 +
 builtin/fsmonitor--daemon.c | 46 +++++++++++++++++++++++++++++++++++++
 git.c                       |  1 +
 5 files changed, 50 insertions(+)
 create mode 100644 builtin/fsmonitor--daemon.c

diff --git a/.gitignore b/.gitignore
index f817c509ec0..e81de1063a4 100644
--- a/.gitignore
+++ b/.gitignore
@@ -72,6 +72,7 @@
 /git-format-patch
 /git-fsck
 /git-fsck-objects
+/git-fsmonitor--daemon
 /git-gc
 /git-get-tar-commit-id
 /git-grep
diff --git a/Makefile b/Makefile
index 707a56d4c11..5af1d5b112e 100644
--- a/Makefile
+++ b/Makefile
@@ -1114,6 +1114,7 @@ BUILTIN_OBJS += builtin/fmt-merge-msg.o
 BUILTIN_OBJS += builtin/for-each-ref.o
 BUILTIN_OBJS += builtin/for-each-repo.o
 BUILTIN_OBJS += builtin/fsck.o
+BUILTIN_OBJS += builtin/fsmonitor--daemon.o
 BUILTIN_OBJS += builtin/gc.o
 BUILTIN_OBJS += builtin/get-tar-commit-id.o
 BUILTIN_OBJS += builtin/grep.o
diff --git a/builtin.h b/builtin.h
index 83379f3832c..40e9ecc8485 100644
--- a/builtin.h
+++ b/builtin.h
@@ -159,6 +159,7 @@ int cmd_for_each_ref(int argc, const char **argv, const char *prefix);
 int cmd_for_each_repo(int argc, const char **argv, const char *prefix);
 int cmd_format_patch(int argc, const char **argv, const char *prefix);
 int cmd_fsck(int argc, const char **argv, const char *prefix);
+int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix);
 int cmd_gc(int argc, const char **argv, const char *prefix);
 int cmd_get_tar_commit_id(int argc, const char **argv, const char *prefix);
 int cmd_grep(int argc, const char **argv, const char *prefix);
diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
new file mode 100644
index 00000000000..f0498793379
--- /dev/null
+++ b/builtin/fsmonitor--daemon.c
@@ -0,0 +1,46 @@
+#include "builtin.h"
+#include "config.h"
+#include "parse-options.h"
+#include "fsmonitor.h"
+#include "fsmonitor-ipc.h"
+#include "simple-ipc.h"
+#include "khash.h"
+
+static const char * const builtin_fsmonitor__daemon_usage[] = {
+	NULL
+};
+
+#ifdef HAVE_FSMONITOR_DAEMON_BACKEND
+
+int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
+{
+	const char *subcmd;
+
+	struct option options[] = {
+		OPT_END()
+	};
+
+	git_config(git_default_config, NULL);
+
+	argc = parse_options(argc, argv, prefix, options,
+			     builtin_fsmonitor__daemon_usage, 0);
+	if (argc != 1)
+		usage_with_options(builtin_fsmonitor__daemon_usage, options);
+	subcmd = argv[0];
+
+	die(_("Unhandled subcommand '%s'"), subcmd);
+}
+
+#else
+int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
+{
+	struct option options[] = {
+		OPT_END()
+	};
+
+	if (argc == 2 && !strcmp(argv[1], "-h"))
+		usage_with_options(builtin_fsmonitor__daemon_usage, options);
+
+	die(_("fsmonitor--daemon not supported on this platform"));
+}
+#endif
diff --git a/git.c b/git.c
index a25940d72e8..3d8e48cf555 100644
--- a/git.c
+++ b/git.c
@@ -537,6 +537,7 @@ static struct cmd_struct commands[] = {
 	{ "format-patch", cmd_format_patch, RUN_SETUP },
 	{ "fsck", cmd_fsck, RUN_SETUP },
 	{ "fsck-objects", cmd_fsck, RUN_SETUP },
+	{ "fsmonitor--daemon", cmd_fsmonitor__daemon, RUN_SETUP },
 	{ "gc", cmd_gc, RUN_SETUP },
 	{ "get-tar-commit-id", cmd_get_tar_commit_id, NO_PARSEOPT },
 	{ "grep", cmd_grep, RUN_SETUP_GENTLY },
-- 
gitgitgadget

[PATCH v9 07/30] fsmonitor--daemon: implement 'stop' and 'status' commands

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Implement `stop` and `status` client commands to control and query the
status of a `fsmonitor--daemon` server process (and implicitly start a
server process if necessary).

Later commits will implement the actual server and monitor the file
system.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 builtin/fsmonitor--daemon.c | 51 +++++++++++++++++++++++++++++++++++++
 1 file changed, 51 insertions(+)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index f0498793379..5e3178b8bdd 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -7,10 +7,55 @@
 #include "khash.h"
 
 static const char * const builtin_fsmonitor__daemon_usage[] = {
+	N_("git fsmonitor--daemon stop"),
+	N_("git fsmonitor--daemon status"),
 	NULL
 };
 
 #ifdef HAVE_FSMONITOR_DAEMON_BACKEND
+/*
+ * Acting as a CLIENT.
+ *
+ * Send a "quit" command to the `git-fsmonitor--daemon` (if running)
+ * and wait for it to shutdown.
+ */
+static int do_as_client__send_stop(void)
+{
+	struct strbuf answer = STRBUF_INIT;
+	int ret;
+
+	ret = fsmonitor_ipc__send_command("quit", &answer);
+
+	/* The quit command does not return any response data. */
+	strbuf_release(&answer);
+
+	if (ret)
+		return ret;
+
+	trace2_region_enter("fsm_client", "polling-for-daemon-exit", NULL);
+	while (fsmonitor_ipc__get_state() == IPC_STATE__LISTENING)
+		sleep_millisec(50);
+	trace2_region_leave("fsm_client", "polling-for-daemon-exit", NULL);
+
+	return 0;
+}
+
+static int do_as_client__status(void)
+{
+	enum ipc_active_state state = fsmonitor_ipc__get_state();
+
+	switch (state) {
+	case IPC_STATE__LISTENING:
+		printf(_("fsmonitor-daemon is watching '%s'\n"),
+		       the_repository->worktree);
+		return 0;
+
+	default:
+		printf(_("fsmonitor-daemon is not watching '%s'\n"),
+		       the_repository->worktree);
+		return 1;
+	}
+}
 
 int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
 {
@@ -28,6 +73,12 @@ int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
 		usage_with_options(builtin_fsmonitor__daemon_usage, options);
 	subcmd = argv[0];
 
+	if (!strcmp(subcmd, "stop"))
+		return !!do_as_client__send_stop();
+
+	if (!strcmp(subcmd, "status"))
+		return !!do_as_client__status();
+
 	die(_("Unhandled subcommand '%s'"), subcmd);
 }
 
-- 
gitgitgadget

[PATCH v9 08/30] compat/fsmonitor/fsm-listen-win32: stub in backend for Windows

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Stub in empty filesystem listener backend for fsmonitor--daemon on Windows.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 Makefile                            | 13 ++++++++
 compat/fsmonitor/fsm-listen-win32.c | 21 +++++++++++++
 compat/fsmonitor/fsm-listen.h       | 49 +++++++++++++++++++++++++++++
 config.mak.uname                    | 10 ++++++
 contrib/buildsystems/CMakeLists.txt |  7 +++++
 repo-settings.c                     |  1 +
 6 files changed, 101 insertions(+)
 create mode 100644 compat/fsmonitor/fsm-listen-win32.c
 create mode 100644 compat/fsmonitor/fsm-listen.h

diff --git a/Makefile b/Makefile
index 5af1d5b112e..26567d4f772 100644
--- a/Makefile
+++ b/Makefile
@@ -470,6 +470,11 @@ all::
 # directory, and the JSON compilation database 'compile_commands.json' will be
 # created at the root of the repository.
 #
+# If your platform supports a built-in fsmonitor backend, set
+# FSMONITOR_DAEMON_BACKEND to the "<name>" of the corresponding
+# `compat/fsmonitor/fsm-listen-<name>.c` that implements the
+# `fsm_listen__*()` routines.
+#
 # Define DEVELOPER to enable more compiler warnings. Compiler version
 # and family are auto detected, but could be overridden by defining
 # COMPILER_FEATURES (see config.mak.dev). You can still set
@@ -1968,6 +1973,11 @@ ifdef NEED_ACCESS_ROOT_HANDLER
 	COMPAT_OBJS += compat/access.o
 endif
 
+ifdef FSMONITOR_DAEMON_BACKEND
+	COMPAT_CFLAGS += -DHAVE_FSMONITOR_DAEMON_BACKEND
+	COMPAT_OBJS += compat/fsmonitor/fsm-listen-$(FSMONITOR_DAEMON_BACKEND).o
+endif
+
 ifeq ($(TCLTK_PATH),)
 NO_TCLTK = NoThanks
 endif
@@ -2887,6 +2897,9 @@ GIT-BUILD-OPTIONS: FORCE
 	@echo DC_SHA1=\''$(subst ','\'',$(subst ','\'',$(DC_SHA1)))'\' >>$@+
 	@echo SANITIZE_LEAK=\''$(subst ','\'',$(subst ','\'',$(SANITIZE_LEAK)))'\' >>$@+
 	@echo X=\'$(X)\' >>$@+
+ifdef FSMONITOR_DAEMON_BACKEND
+	@echo FSMONITOR_DAEMON_BACKEND=\''$(subst ','\'',$(subst ','\'',$(FSMONITOR_DAEMON_BACKEND)))'\' >>$@+
+endif
 ifdef TEST_OUTPUT_DIRECTORY
 	@echo TEST_OUTPUT_DIRECTORY=\''$(subst ','\'',$(subst ','\'',$(TEST_OUTPUT_DIRECTORY)))'\' >>$@+
 endif
diff --git a/compat/fsmonitor/fsm-listen-win32.c b/compat/fsmonitor/fsm-listen-win32.c
new file mode 100644
index 00000000000..916cbea254f
--- /dev/null
+++ b/compat/fsmonitor/fsm-listen-win32.c
@@ -0,0 +1,21 @@
+#include "cache.h"
+#include "config.h"
+#include "fsmonitor.h"
+#include "fsm-listen.h"
+
+void fsm_listen__stop_async(struct fsmonitor_daemon_state *state)
+{
+}
+
+void fsm_listen__loop(struct fsmonitor_daemon_state *state)
+{
+}
+
+int fsm_listen__ctor(struct fsmonitor_daemon_state *state)
+{
+	return -1;
+}
+
+void fsm_listen__dtor(struct fsmonitor_daemon_state *state)
+{
+}
diff --git a/compat/fsmonitor/fsm-listen.h b/compat/fsmonitor/fsm-listen.h
new file mode 100644
index 00000000000..f0539349baf
--- /dev/null
+++ b/compat/fsmonitor/fsm-listen.h
@@ -0,0 +1,49 @@
+#ifndef FSM_LISTEN_H
+#define FSM_LISTEN_H
+
+/* This needs to be implemented by each backend */
+
+#ifdef HAVE_FSMONITOR_DAEMON_BACKEND
+
+struct fsmonitor_daemon_state;
+
+/*
+ * Initialize platform-specific data for the fsmonitor listener thread.
+ * This will be called from the main thread PRIOR to staring the
+ * fsmonitor_fs_listener thread.
+ *
+ * Returns 0 if successful.
+ * Returns -1 otherwise.
+ */
+int fsm_listen__ctor(struct fsmonitor_daemon_state *state);
+
+/*
+ * Cleanup platform-specific data for the fsmonitor listener thread.
+ * This will be called from the main thread AFTER joining the listener.
+ */
+void fsm_listen__dtor(struct fsmonitor_daemon_state *state);
+
+/*
+ * The main body of the platform-specific event loop to watch for
+ * filesystem events.  This will run in the fsmonitor_fs_listen thread.
+ *
+ * It should call `ipc_server_stop_async()` if the listener thread
+ * prematurely terminates (because of a filesystem error or if it
+ * detects that the .git directory has been deleted).  (It should NOT
+ * do so if the listener thread receives a normal shutdown signal from
+ * the IPC layer.)
+ *
+ * It should set `state->error_code` to -1 if the daemon should exit
+ * with an error.
+ */
+void fsm_listen__loop(struct fsmonitor_daemon_state *state);
+
+/*
+ * Gently request that the fsmonitor listener thread shutdown.
+ * It does not wait for it to stop.  The caller should do a JOIN
+ * to wait for it.
+ */
+void fsm_listen__stop_async(struct fsmonitor_daemon_state *state);
+
+#endif /* HAVE_FSMONITOR_DAEMON_BACKEND */
+#endif /* FSM_LISTEN_H */
diff --git a/config.mak.uname b/config.mak.uname
index 4352ea39e9b..26074f56bed 100644
--- a/config.mak.uname
+++ b/config.mak.uname
@@ -435,6 +435,11 @@ ifeq ($(uname_S),Windows)
 	# so we don't need this:
 	#
 	#   SNPRINTF_RETURNS_BOGUS = YesPlease
+
+	# The builtin FSMonitor requires Named Pipes and Threads on Windows.
+	# These are always available, so we do not have to conditionally
+	# support it.
+	FSMONITOR_DAEMON_BACKEND = win32
 	NO_SVN_TESTS = YesPlease
 	RUNTIME_PREFIX = YesPlease
 	HAVE_WPGMPTR = YesWeDo
@@ -619,6 +624,11 @@ ifeq ($(uname_S),MINGW)
 	NO_STRTOUMAX = YesPlease
 	NO_MKDTEMP = YesPlease
 	NO_SVN_TESTS = YesPlease
+
+	# The builtin FSMonitor requires Named Pipes and Threads on Windows.
+	# These are always available, so we do not have to conditionally
+	# support it.
+	FSMONITOR_DAEMON_BACKEND = win32
 	RUNTIME_PREFIX = YesPlease
 	HAVE_WPGMPTR = YesWeDo
 	NO_ST_BLOCKS_IN_STRUCT_STAT = YesPlease
diff --git a/contrib/buildsystems/CMakeLists.txt b/contrib/buildsystems/CMakeLists.txt
index e44232f85d3..0963629db7f 100644
--- a/contrib/buildsystems/CMakeLists.txt
+++ b/contrib/buildsystems/CMakeLists.txt
@@ -285,6 +285,13 @@ else()
 	endif()
 endif()
 
+if(SUPPORTS_SIMPLE_IPC)
+	if(CMAKE_SYSTEM_NAME STREQUAL "Windows")
+		add_compile_definitions(HAVE_FSMONITOR_DAEMON_BACKEND)
+		list(APPEND compat_SOURCES compat/fsmonitor/fsm-listen-win32.c)
+	endif()
+endif()
+
 set(EXE_EXTENSION ${CMAKE_EXECUTABLE_SUFFIX})
 
 #header checks
diff --git a/repo-settings.c b/repo-settings.c
index b4fbd16cdcc..2dfcb2b6542 100644
--- a/repo-settings.c
+++ b/repo-settings.c
@@ -2,6 +2,7 @@
 #include "config.h"
 #include "repository.h"
 #include "midx.h"
+#include "compat/fsmonitor/fsm-listen.h"
 
 static void repo_cfg_bool(struct repository *r, const char *key, int *dest,
 			  int def)
-- 
gitgitgadget

[PATCH v9 09/30] compat/fsmonitor/fsm-listen-darwin: stub in backend for Darwin

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Stub in empty implementation of fsmonitor--daemon
backend for Darwin (aka MacOS).

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 compat/fsmonitor/fsm-listen-darwin.c | 20 ++++++++++++++++++++
 config.mak.uname                     | 10 ++++++++++
 contrib/buildsystems/CMakeLists.txt  |  3 +++
 3 files changed, 33 insertions(+)
 create mode 100644 compat/fsmonitor/fsm-listen-darwin.c

diff --git a/compat/fsmonitor/fsm-listen-darwin.c b/compat/fsmonitor/fsm-listen-darwin.c
new file mode 100644
index 00000000000..c84e3344ab9
--- /dev/null
+++ b/compat/fsmonitor/fsm-listen-darwin.c
@@ -0,0 +1,20 @@
+#include "cache.h"
+#include "fsmonitor.h"
+#include "fsm-listen.h"
+
+int fsm_listen__ctor(struct fsmonitor_daemon_state *state)
+{
+	return -1;
+}
+
+void fsm_listen__dtor(struct fsmonitor_daemon_state *state)
+{
+}
+
+void fsm_listen__stop_async(struct fsmonitor_daemon_state *state)
+{
+}
+
+void fsm_listen__loop(struct fsmonitor_daemon_state *state)
+{
+}
diff --git a/config.mak.uname b/config.mak.uname
index 26074f56bed..501970902da 100644
--- a/config.mak.uname
+++ b/config.mak.uname
@@ -157,6 +157,16 @@ ifeq ($(uname_S),Darwin)
 			MSGFMT = /usr/local/opt/gettext/bin/msgfmt
 		endif
 	endif
+
+	# The builtin FSMonitor on MacOS builds upon Simple-IPC.  Both require
+	# Unix domain sockets and PThreads.
+	ifndef NO_PTHREADS
+	ifndef NO_UNIX_SOCKETS
+	FSMONITOR_DAEMON_BACKEND = darwin
+	endif
+	endif
+
+	BASIC_LDFLAGS += -framework CoreServices
 endif
 ifeq ($(uname_S),SunOS)
 	NEEDS_SOCKET = YesPlease
diff --git a/contrib/buildsystems/CMakeLists.txt b/contrib/buildsystems/CMakeLists.txt
index 0963629db7f..ee0d7257b77 100644
--- a/contrib/buildsystems/CMakeLists.txt
+++ b/contrib/buildsystems/CMakeLists.txt
@@ -289,6 +289,9 @@ if(SUPPORTS_SIMPLE_IPC)
 	if(CMAKE_SYSTEM_NAME STREQUAL "Windows")
 		add_compile_definitions(HAVE_FSMONITOR_DAEMON_BACKEND)
 		list(APPEND compat_SOURCES compat/fsmonitor/fsm-listen-win32.c)
+	elseif(CMAKE_SYSTEM_NAME STREQUAL "Darwin")
+		add_compile_definitions(HAVE_FSMONITOR_DAEMON_BACKEND)
+		list(APPEND compat_SOURCES compat/fsmonitor/fsm-listen-darwin.c)
 	endif()
 endif()
 
-- 
gitgitgadget

[PATCH v9 10/30] fsmonitor--daemon: implement 'run' command

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Implement `run` command to try to begin listening for file system events.

This version defines the thread structure with a single fsmonitor_fs_listen
thread to watch for file system events and a simple IPC thread pool to
watch for connection from Git clients over a well-known named pipe or
Unix domain socket.

This commit does not actually do anything yet because the platform
backends are still just stubs.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 builtin/fsmonitor--daemon.c | 228 +++++++++++++++++++++++++++++++++++-
 fsmonitor--daemon.h         |  34 ++++++
 2 files changed, 261 insertions(+), 1 deletion(-)
 create mode 100644 fsmonitor--daemon.h

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index 5e3178b8bdd..5591339399a 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -3,16 +3,52 @@
 #include "parse-options.h"
 #include "fsmonitor.h"
 #include "fsmonitor-ipc.h"
+#include "compat/fsmonitor/fsm-listen.h"
+#include "fsmonitor--daemon.h"
 #include "simple-ipc.h"
 #include "khash.h"
 
 static const char * const builtin_fsmonitor__daemon_usage[] = {
+	N_("git fsmonitor--daemon run [<options>]"),
 	N_("git fsmonitor--daemon stop"),
 	N_("git fsmonitor--daemon status"),
 	NULL
 };
 
 #ifdef HAVE_FSMONITOR_DAEMON_BACKEND
+/*
+ * Global state loaded from config.
+ */
+#define FSMONITOR__IPC_THREADS "fsmonitor.ipcthreads"
+static int fsmonitor__ipc_threads = 8;
+
+#define FSMONITOR__ANNOUNCE_STARTUP "fsmonitor.announcestartup"
+static int fsmonitor__announce_startup = 0;
+
+static int fsmonitor_config(const char *var, const char *value, void *cb)
+{
+	if (!strcmp(var, FSMONITOR__IPC_THREADS)) {
+		int i = git_config_int(var, value);
+		if (i < 1)
+			return error(_("value of '%s' out of range: %d"),
+				     FSMONITOR__IPC_THREADS, i);
+		fsmonitor__ipc_threads = i;
+		return 0;
+	}
+
+	if (!strcmp(var, FSMONITOR__ANNOUNCE_STARTUP)) {
+		int is_bool;
+		int i = git_config_bool_or_int(var, value, &is_bool);
+		if (i < 0)
+			return error(_("value of '%s' not bool or int: %d"),
+				     var, i);
+		fsmonitor__announce_startup = i;
+		return 0;
+	}
+
+	return git_default_config(var, value, cb);
+}
+
 /*
  * Acting as a CLIENT.
  *
@@ -57,15 +93,198 @@ static int do_as_client__status(void)
 	}
 }
 
+static ipc_server_application_cb handle_client;
+
+static int handle_client(void *data,
+			 const char *command, size_t command_len,
+			 ipc_server_reply_cb *reply,
+			 struct ipc_server_reply_data *reply_data)
+{
+	/* struct fsmonitor_daemon_state *state = data; */
+	int result;
+
+	/*
+	 * The Simple IPC API now supports {char*, len} arguments, but
+	 * FSMonitor always uses proper null-terminated strings, so
+	 * we can ignore the command_len argument.  (Trust, but verify.)
+	 */
+	if (command_len != strlen(command))
+		BUG("FSMonitor assumes text messages");
+
+	trace2_region_enter("fsmonitor", "handle_client", the_repository);
+	trace2_data_string("fsmonitor", the_repository, "request", command);
+
+	result = 0; /* TODO Do something here. */
+
+	trace2_region_leave("fsmonitor", "handle_client", the_repository);
+
+	return result;
+}
+
+static void *fsm_listen__thread_proc(void *_state)
+{
+	struct fsmonitor_daemon_state *state = _state;
+
+	trace2_thread_start("fsm-listen");
+
+	trace_printf_key(&trace_fsmonitor, "Watching: worktree '%s'",
+			 state->path_worktree_watch.buf);
+	if (state->nr_paths_watching > 1)
+		trace_printf_key(&trace_fsmonitor, "Watching: gitdir '%s'",
+				 state->path_gitdir_watch.buf);
+
+	fsm_listen__loop(state);
+
+	trace2_thread_exit();
+	return NULL;
+}
+
+static int fsmonitor_run_daemon_1(struct fsmonitor_daemon_state *state)
+{
+	struct ipc_server_opts ipc_opts = {
+		.nr_threads = fsmonitor__ipc_threads,
+
+		/*
+		 * We know that there are no other active threads yet,
+		 * so we can let the IPC layer temporarily chdir() if
+		 * it needs to when creating the server side of the
+		 * Unix domain socket.
+		 */
+		.uds_disallow_chdir = 0
+	};
+
+	/*
+	 * Start the IPC thread pool before the we've started the file
+	 * system event listener thread so that we have the IPC handle
+	 * before we need it.
+	 */
+	if (ipc_server_run_async(&state->ipc_server_data,
+				 fsmonitor_ipc__get_path(), &ipc_opts,
+				 handle_client, state))
+		return error_errno(
+			_("could not start IPC thread pool on '%s'"),
+			fsmonitor_ipc__get_path());
+
+	/*
+	 * Start the fsmonitor listener thread to collect filesystem
+	 * events.
+	 */
+	if (pthread_create(&state->listener_thread, NULL,
+			   fsm_listen__thread_proc, state) < 0) {
+		ipc_server_stop_async(state->ipc_server_data);
+		ipc_server_await(state->ipc_server_data);
+
+		return error(_("could not start fsmonitor listener thread"));
+	}
+
+	/*
+	 * The daemon is now fully functional in background threads.
+	 * Wait for the IPC thread pool to shutdown (whether by client
+	 * request or from filesystem activity).
+	 */
+	ipc_server_await(state->ipc_server_data);
+
+	/*
+	 * The fsmonitor listener thread may have received a shutdown
+	 * event from the IPC thread pool, but it doesn't hurt to tell
+	 * it again.  And wait for it to shutdown.
+	 */
+	fsm_listen__stop_async(state);
+	pthread_join(state->listener_thread, NULL);
+
+	return state->error_code;
+}
+
+static int fsmonitor_run_daemon(void)
+{
+	struct fsmonitor_daemon_state state;
+	int err;
+
+	memset(&state, 0, sizeof(state));
+
+	pthread_mutex_init(&state.main_lock, NULL);
+	state.error_code = 0;
+	state.current_token_data = NULL;
+
+	/* Prepare to (recursively) watch the <worktree-root> directory. */
+	strbuf_init(&state.path_worktree_watch, 0);
+	strbuf_addstr(&state.path_worktree_watch, absolute_path(get_git_work_tree()));
+	state.nr_paths_watching = 1;
+
+	/*
+	 * We create and delete cookie files somewhere inside the .git
+	 * directory to help us keep sync with the file system.  If
+	 * ".git" is not a directory, then <gitdir> is not inside the
+	 * cone of <worktree-root>, so set up a second watch to watch
+	 * the <gitdir> so that we get events for the cookie files.
+	 */
+	strbuf_init(&state.path_gitdir_watch, 0);
+	strbuf_addbuf(&state.path_gitdir_watch, &state.path_worktree_watch);
+	strbuf_addstr(&state.path_gitdir_watch, "/.git");
+	if (!is_directory(state.path_gitdir_watch.buf)) {
+		strbuf_reset(&state.path_gitdir_watch);
+		strbuf_addstr(&state.path_gitdir_watch, absolute_path(get_git_dir()));
+		state.nr_paths_watching = 2;
+	}
+
+	/*
+	 * Confirm that we can create platform-specific resources for the
+	 * filesystem listener before we bother starting all the threads.
+	 */
+	if (fsm_listen__ctor(&state)) {
+		err = error(_("could not initialize listener thread"));
+		goto done;
+	}
+
+	err = fsmonitor_run_daemon_1(&state);
+
+done:
+	pthread_mutex_destroy(&state.main_lock);
+	fsm_listen__dtor(&state);
+
+	ipc_server_free(state.ipc_server_data);
+
+	strbuf_release(&state.path_worktree_watch);
+	strbuf_release(&state.path_gitdir_watch);
+
+	return err;
+}
+
+static int try_to_run_foreground_daemon(void)
+{
+	/*
+	 * Technically, we don't need to probe for an existing daemon
+	 * process, since we could just call `fsmonitor_run_daemon()`
+	 * and let it fail if the pipe/socket is busy.
+	 *
+	 * However, this method gives us a nicer error message for a
+	 * common error case.
+	 */
+	if (fsmonitor_ipc__get_state() == IPC_STATE__LISTENING)
+		die(_("fsmonitor--daemon is already running '%s'"),
+		    the_repository->worktree);
+
+	if (fsmonitor__announce_startup) {
+		fprintf(stderr, _("running fsmonitor-daemon in '%s'\n"),
+			the_repository->worktree);
+		fflush(stderr);
+	}
+
+	return !!fsmonitor_run_daemon();
+}
+
 int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
 {
 	const char *subcmd;
 
 	struct option options[] = {
+		OPT_INTEGER(0, "ipc-threads",
+			    &fsmonitor__ipc_threads,
+			    N_("use <n> ipc worker threads")),
 		OPT_END()
 	};
 
-	git_config(git_default_config, NULL);
+	git_config(fsmonitor_config, NULL);
 
 	argc = parse_options(argc, argv, prefix, options,
 			     builtin_fsmonitor__daemon_usage, 0);
@@ -73,6 +292,13 @@ int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
 		usage_with_options(builtin_fsmonitor__daemon_usage, options);
 	subcmd = argv[0];
 
+	if (fsmonitor__ipc_threads < 1)
+		die(_("invalid 'ipc-threads' value (%d)"),
+		    fsmonitor__ipc_threads);
+
+	if (!strcmp(subcmd, "run"))
+		return !!try_to_run_foreground_daemon();
+
 	if (!strcmp(subcmd, "stop"))
 		return !!do_as_client__send_stop();
 
diff --git a/fsmonitor--daemon.h b/fsmonitor--daemon.h
new file mode 100644
index 00000000000..3009c1a83de
--- /dev/null
+++ b/fsmonitor--daemon.h
@@ -0,0 +1,34 @@
+#ifndef FSMONITOR_DAEMON_H
+#define FSMONITOR_DAEMON_H
+
+#ifdef HAVE_FSMONITOR_DAEMON_BACKEND
+
+#include "cache.h"
+#include "dir.h"
+#include "run-command.h"
+#include "simple-ipc.h"
+#include "thread-utils.h"
+
+struct fsmonitor_batch;
+struct fsmonitor_token_data;
+
+struct fsmonitor_daemon_backend_data; /* opaque platform-specific data */
+
+struct fsmonitor_daemon_state {
+	pthread_t listener_thread;
+	pthread_mutex_t main_lock;
+
+	struct strbuf path_worktree_watch;
+	struct strbuf path_gitdir_watch;
+	int nr_paths_watching;
+
+	struct fsmonitor_token_data *current_token_data;
+
+	int error_code;
+	struct fsmonitor_daemon_backend_data *backend_data;
+
+	struct ipc_server_data *ipc_server_data;
+};
+
+#endif /* HAVE_FSMONITOR_DAEMON_BACKEND */
+#endif /* FSMONITOR_DAEMON_H */
-- 
gitgitgadget

[PATCH v9 11/30] fsmonitor--daemon: implement 'start' command

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Implement 'git fsmonitor--daemon start' command.  This command starts
an instance of 'git fsmonitor--daemon run' in the background using
the new 'start_bg_command()' function.

We avoid the fork-and-call technique on Unix systems in favor of a
fork-and-exec technique.  This gives us more uniform Trace2 child-*
events.  It also makes our usage more consistent with Windows usage.

On Windows, teach 'git fsmonitor--daemon run' to optionally call
'FreeConsole()' to release handles to the inherited Win32 console
(despite being passed invalid handles for stdin/out/err).  Without
this, command prompts and powershell terminal windows could hang
in "exit" until the last background child process exited or released
their Win32 console handle.  (This was not seen with git-bash shells
because they don't have a Win32 console attached to them.)

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 builtin/fsmonitor--daemon.c | 109 +++++++++++++++++++++++++++++++++++-
 1 file changed, 107 insertions(+), 2 deletions(-)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index 5591339399a..69dd39121a3 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -9,6 +9,7 @@
 #include "khash.h"
 
 static const char * const builtin_fsmonitor__daemon_usage[] = {
+	N_("git fsmonitor--daemon start [<options>]"),
 	N_("git fsmonitor--daemon run [<options>]"),
 	N_("git fsmonitor--daemon stop"),
 	N_("git fsmonitor--daemon status"),
@@ -22,6 +23,9 @@ static const char * const builtin_fsmonitor__daemon_usage[] = {
 #define FSMONITOR__IPC_THREADS "fsmonitor.ipcthreads"
 static int fsmonitor__ipc_threads = 8;
 
+#define FSMONITOR__START_TIMEOUT "fsmonitor.starttimeout"
+static int fsmonitor__start_timeout_sec = 60;
+
 #define FSMONITOR__ANNOUNCE_STARTUP "fsmonitor.announcestartup"
 static int fsmonitor__announce_startup = 0;
 
@@ -36,6 +40,15 @@ static int fsmonitor_config(const char *var, const char *value, void *cb)
 		return 0;
 	}
 
+	if (!strcmp(var, FSMONITOR__START_TIMEOUT)) {
+		int i = git_config_int(var, value);
+		if (i < 0)
+			return error(_("value of '%s' out of range: %d"),
+				     FSMONITOR__START_TIMEOUT, i);
+		fsmonitor__start_timeout_sec = i;
+		return 0;
+	}
+
 	if (!strcmp(var, FSMONITOR__ANNOUNCE_STARTUP)) {
 		int is_bool;
 		int i = git_config_bool_or_int(var, value, &is_bool);
@@ -250,7 +263,7 @@ done:
 	return err;
 }
 
-static int try_to_run_foreground_daemon(void)
+static int try_to_run_foreground_daemon(int detach_console)
 {
 	/*
 	 * Technically, we don't need to probe for an existing daemon
@@ -270,17 +283,106 @@ static int try_to_run_foreground_daemon(void)
 		fflush(stderr);
 	}
 
+#ifdef GIT_WINDOWS_NATIVE
+	if (detach_console)
+		FreeConsole();
+#endif
+
 	return !!fsmonitor_run_daemon();
 }
 
+static start_bg_wait_cb bg_wait_cb;
+
+static int bg_wait_cb(const struct child_process *cp, void *cb_data)
+{
+	enum ipc_active_state s = fsmonitor_ipc__get_state();
+
+	switch (s) {
+	case IPC_STATE__LISTENING:
+		/* child is "ready" */
+		return 0;
+
+	case IPC_STATE__NOT_LISTENING:
+	case IPC_STATE__PATH_NOT_FOUND:
+		/* give child more time */
+		return 1;
+
+	default:
+	case IPC_STATE__INVALID_PATH:
+	case IPC_STATE__OTHER_ERROR:
+		/* all the time in world won't help */
+		return -1;
+	}
+}
+
+static int try_to_start_background_daemon(void)
+{
+	struct child_process cp = CHILD_PROCESS_INIT;
+	enum start_bg_result sbgr;
+
+	/*
+	 * Before we try to create a background daemon process, see
+	 * if a daemon process is already listening.  This makes it
+	 * easier for us to report an already-listening error to the
+	 * console, since our spawn/daemon can only report the success
+	 * of creating the background process (and not whether it
+	 * immediately exited).
+	 */
+	if (fsmonitor_ipc__get_state() == IPC_STATE__LISTENING)
+		die(_("fsmonitor--daemon is already running '%s'"),
+		    the_repository->worktree);
+
+	if (fsmonitor__announce_startup) {
+		fprintf(stderr, _("starting fsmonitor-daemon in '%s'\n"),
+			the_repository->worktree);
+		fflush(stderr);
+	}
+
+	cp.git_cmd = 1;
+
+	strvec_push(&cp.args, "fsmonitor--daemon");
+	strvec_push(&cp.args, "run");
+	strvec_push(&cp.args, "--detach");
+	strvec_pushf(&cp.args, "--ipc-threads=%d", fsmonitor__ipc_threads);
+
+	cp.no_stdin = 1;
+	cp.no_stdout = 1;
+	cp.no_stderr = 1;
+
+	sbgr = start_bg_command(&cp, bg_wait_cb, NULL,
+				fsmonitor__start_timeout_sec);
+
+	switch (sbgr) {
+	case SBGR_READY:
+		return 0;
+
+	default:
+	case SBGR_ERROR:
+	case SBGR_CB_ERROR:
+		return error(_("daemon failed to start"));
+
+	case SBGR_TIMEOUT:
+		return error(_("daemon not online yet"));
+
+	case SBGR_DIED:
+		return error(_("daemon terminated"));
+	}
+}
+
 int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
 {
 	const char *subcmd;
+	int detach_console = 0;
 
 	struct option options[] = {
+		OPT_BOOL(0, "detach", &detach_console, N_("detach from console")),
 		OPT_INTEGER(0, "ipc-threads",
 			    &fsmonitor__ipc_threads,
 			    N_("use <n> ipc worker threads")),
+		OPT_INTEGER(0, "start-timeout",
+			    &fsmonitor__start_timeout_sec,
+			    N_("max seconds to wait for background daemon startup")),
+
 		OPT_END()
 	};
 
@@ -296,8 +398,11 @@ int cmd_fsmonitor__daemon(int argc, const char **argv, const char *prefix)
 		die(_("invalid 'ipc-threads' value (%d)"),
 		    fsmonitor__ipc_threads);
 
+	if (!strcmp(subcmd, "start"))
+		return !!try_to_start_background_daemon();
+
 	if (!strcmp(subcmd, "run"))
-		return !!try_to_run_foreground_daemon();
+		return !!try_to_run_foreground_daemon(detach_console);
 
 	if (!strcmp(subcmd, "stop"))
 		return !!do_as_client__send_stop();
-- 
gitgitgadget

[PATCH v9 12/30] fsmonitor--daemon: add pathname classification

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach fsmonitor--daemon to classify relative and absolute
pathnames and decide how they should be handled.  This will
be used by the platform-specific backend to respond to each
filesystem event.

When we register for filesystem notifications on a directory,
we get events for everything (recursively) in the directory.
We want to report to clients changes to tracked and untracked
paths within the working directory proper.  We do not want to
report changes within the .git directory, for example.

This classification will be used in a later commit by the
different backends to classify paths as events are received.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 builtin/fsmonitor--daemon.c | 81 ++++++++++++++++++++++++++++++++++
 fsmonitor--daemon.h         | 87 +++++++++++++++++++++++++++++++++++++
 2 files changed, 168 insertions(+)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index 69dd39121a3..1ce00b7c150 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -134,6 +134,87 @@ static int handle_client(void *data,
 	return result;
 }
 
+#define FSMONITOR_COOKIE_PREFIX ".fsmonitor-daemon-"
+
+enum fsmonitor_path_type fsmonitor_classify_path_workdir_relative(
+	const char *rel)
+{
+	if (fspathncmp(rel, ".git", 4))
+		return IS_WORKDIR_PATH;
+	rel += 4;
+
+	if (!*rel)
+		return IS_DOT_GIT;
+	if (*rel != '/')
+		return IS_WORKDIR_PATH; /* e.g. .gitignore */
+	rel++;
+
+	if (!fspathncmp(rel, FSMONITOR_COOKIE_PREFIX,
+			strlen(FSMONITOR_COOKIE_PREFIX)))
+		return IS_INSIDE_DOT_GIT_WITH_COOKIE_PREFIX;
+
+	return IS_INSIDE_DOT_GIT;
+}
+
+enum fsmonitor_path_type fsmonitor_classify_path_gitdir_relative(
+	const char *rel)
+{
+	if (!fspathncmp(rel, FSMONITOR_COOKIE_PREFIX,
+			strlen(FSMONITOR_COOKIE_PREFIX)))
+		return IS_INSIDE_GITDIR_WITH_COOKIE_PREFIX;
+
+	return IS_INSIDE_GITDIR;
+}
+
+static enum fsmonitor_path_type try_classify_workdir_abs_path(
+	struct fsmonitor_daemon_state *state,
+	const char *path)
+{
+	const char *rel;
+
+	if (fspathncmp(path, state->path_worktree_watch.buf,
+		       state->path_worktree_watch.len))
+		return IS_OUTSIDE_CONE;
+
+	rel = path + state->path_worktree_watch.len;
+
+	if (!*rel)
+		return IS_WORKDIR_PATH; /* it is the root dir exactly */
+	if (*rel != '/')
+		return IS_OUTSIDE_CONE;
+	rel++;
+
+	return fsmonitor_classify_path_workdir_relative(rel);
+}
+
+enum fsmonitor_path_type fsmonitor_classify_path_absolute(
+	struct fsmonitor_daemon_state *state,
+	const char *path)
+{
+	const char *rel;
+	enum fsmonitor_path_type t;
+
+	t = try_classify_workdir_abs_path(state, path);
+	if (state->nr_paths_watching == 1)
+		return t;
+	if (t != IS_OUTSIDE_CONE)
+		return t;
+
+	if (fspathncmp(path, state->path_gitdir_watch.buf,
+		       state->path_gitdir_watch.len))
+		return IS_OUTSIDE_CONE;
+
+	rel = path + state->path_gitdir_watch.len;
+
+	if (!*rel)
+		return IS_GITDIR; /* it is the <gitdir> exactly */
+	if (*rel != '/')
+		return IS_OUTSIDE_CONE;
+	rel++;
+
+	return fsmonitor_classify_path_gitdir_relative(rel);
+}
+
 static void *fsm_listen__thread_proc(void *_state)
 {
 	struct fsmonitor_daemon_state *state = _state;
diff --git a/fsmonitor--daemon.h b/fsmonitor--daemon.h
index 3009c1a83de..8c3a71a48bd 100644
--- a/fsmonitor--daemon.h
+++ b/fsmonitor--daemon.h
@@ -30,5 +30,92 @@ struct fsmonitor_daemon_state {
 	struct ipc_server_data *ipc_server_data;
 };
 
+/*
+ * Pathname classifications.
+ *
+ * The daemon classifies the pathnames that it receives from file
+ * system notification events into the following categories and uses
+ * that to decide whether clients are told about them.  (And to watch
+ * for file system synchronization events.)
+ *
+ * The daemon only collects and reports on the set of modified paths
+ * within the working directory (proper).
+ *
+ * The client should only care about paths within the working
+ * directory proper (inside the working directory and not ".git" nor
+ * inside of ".git/").  That is, the client has read the index and is
+ * asking for a list of any paths in the working directory that have
+ * been modified since the last token.  The client does not care about
+ * file system changes within the ".git/" directory (such as new loose
+ * objects or packfiles).  So the client will only receive paths that
+ * are classified as IS_WORKDIR_PATH.
+ *
+ * Note that ".git" is usually a directory and is therefore inside
+ * the cone of the FS watch that we have on the working directory root,
+ * so we will also get FS events for disk activity on and within ".git/"
+ * that we need to respond to or filter from the client.
+ *
+ * But Git also allows ".git" to be a *file* that points to a GITDIR
+ * outside of the working directory.  When this happens, we need to
+ * create FS watches on both the working directory root *and* on the
+ * (external) GITDIR root.  (The latter is required because we put
+ * cookie files inside it and use them to sync with the FS event
+ * stream.)
+ *
+ * Note that in the context of this discussion, I'm using "GITDIR"
+ * to only mean an external GITDIR referenced by a ".git" file.
+ *
+ * The platform FS event backends will receive watch-specific
+ * relative paths (except for those OS's that always emit absolute
+ * paths).  We use the following enum and routines to classify each
+ * path so that we know how to handle it.  There is a slight asymmetry
+ * here because ".git/" is inside the working directory and the
+ * (external) GITDIR is not, and therefore how we handle events may
+ * vary slightly, so I have different enums for "IS...DOT_GIT..." and
+ * "IS...GITDIR...".
+ *
+ * The daemon uses the IS_DOT_GIT and IS_GITDIR internally to mean the
+ * exact ".git" file/directory or GITDIR directory.  If the daemon
+ * receives a delete event for either of these paths, it will
+ * automatically shutdown, for example.
+ *
+ * Note that the daemon DOES NOT explicitly watch nor special case the
+ * index.  The daemon does not read the index nor have any internal
+ * index-relative state, so there are no "IS...INDEX..." enum values.
+ */
+enum fsmonitor_path_type {
+	IS_WORKDIR_PATH = 0,
+
+	IS_DOT_GIT,
+	IS_INSIDE_DOT_GIT,
+	IS_INSIDE_DOT_GIT_WITH_COOKIE_PREFIX,
+
+	IS_GITDIR,
+	IS_INSIDE_GITDIR,
+	IS_INSIDE_GITDIR_WITH_COOKIE_PREFIX,
+
+	IS_OUTSIDE_CONE,
+};
+
+/*
+ * Classify a pathname relative to the root of the working directory.
+ */
+enum fsmonitor_path_type fsmonitor_classify_path_workdir_relative(
+	const char *relative_path);
+
+/*
+ * Classify a pathname relative to a <gitdir> that is external to the
+ * worktree directory.
+ */
+enum fsmonitor_path_type fsmonitor_classify_path_gitdir_relative(
+	const char *relative_path);
+
+/*
+ * Classify an absolute pathname received from a filesystem event.
+ */
+enum fsmonitor_path_type fsmonitor_classify_path_absolute(
+	struct fsmonitor_daemon_state *state,
+	const char *path);
+
 #endif /* HAVE_FSMONITOR_DAEMON_BACKEND */
 #endif /* FSMONITOR_DAEMON_H */
-- 
gitgitgadget

[PATCH v9 13/30] fsmonitor--daemon: define token-ids

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach fsmonitor--daemon to create token-ids and define the
overall token naming scheme.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 builtin/fsmonitor--daemon.c | 116 +++++++++++++++++++++++++++++++++++-
 1 file changed, 115 insertions(+), 1 deletion(-)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index 1ce00b7c150..1c7c156288d 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -106,6 +106,120 @@ static int do_as_client__status(void)
 	}
 }
 
+/*
+ * Requests to and from a FSMonitor Protocol V2 provider use an opaque
+ * "token" as a virtual timestamp.  Clients can request a summary of all
+ * created/deleted/modified files relative to a token.  In the response,
+ * clients receive a new token for the next (relative) request.
+ *
+ *
+ * Token Format
+ * ============
+ *
+ * The contents of the token are private and provider-specific.
+ *
+ * For the built-in fsmonitor--daemon, we define a token as follows:
+ *
+ *     "builtin" ":" <token_id> ":" <sequence_nr>
+ *
+ * The "builtin" prefix is used as a namespace to avoid conflicts
+ * with other providers (such as Watchman).
+ *
+ * The <token_id> is an arbitrary OPAQUE string, such as a GUID,
+ * UUID, or {timestamp,pid}.  It is used to group all filesystem
+ * events that happened while the daemon was monitoring (and in-sync
+ * with the filesystem).
+ *
+ *     Unlike FSMonitor Protocol V1, it is not defined as a timestamp
+ *     and does not define less-than/greater-than relationships.
+ *     (There are too many race conditions to rely on file system
+ *     event timestamps.)
+ *
+ * The <sequence_nr> is a simple integer incremented whenever the
+ * daemon needs to make its state public.  For example, if 1000 file
+ * system events come in, but no clients have requested the data,
+ * the daemon can continue to accumulate file changes in the same
+ * bin and does not need to advance the sequence number.  However,
+ * as soon as a client does arrive, the daemon needs to start a new
+ * bin and increment the sequence number.
+ *
+ *     The sequence number serves as the boundary between 2 sets
+ *     of bins -- the older ones that the client has already seen
+ *     and the newer ones that it hasn't.
+ *
+ * When a new <token_id> is created, the <sequence_nr> is reset to
+ * zero.
+ *
+ *
+ * About Token Ids
+ * ===============
+ *
+ * A new token_id is created:
+ *
+ * [1] each time the daemon is started.
+ *
+ * [2] any time that the daemon must re-sync with the filesystem
+ *     (such as when the kernel drops or we miss events on a very
+ *     active volume).
+ *
+ * [3] in response to a client "flush" command (for dropped event
+ *     testing).
+ *
+ * When a new token_id is created, the daemon is free to discard all
+ * cached filesystem events associated with any previous token_ids.
+ * Events associated with a non-current token_id will never be sent
+ * to a client.  A token_id change implicitly means that the daemon
+ * has gap in its event history.
+ *
+ * Therefore, clients that present a token with a stale (non-current)
+ * token_id will always be given a trivial response.
+ */
+struct fsmonitor_token_data {
+	struct strbuf token_id;
+	struct fsmonitor_batch *batch_head;
+	struct fsmonitor_batch *batch_tail;
+	uint64_t client_ref_count;
+};
+
+static struct fsmonitor_token_data *fsmonitor_new_token_data(void)
+{
+	static int test_env_value = -1;
+	static uint64_t flush_count = 0;
+	struct fsmonitor_token_data *token;
+
+	CALLOC_ARRAY(token, 1);
+
+	strbuf_init(&token->token_id, 0);
+	token->batch_head = NULL;
+	token->batch_tail = NULL;
+	token->client_ref_count = 0;
+
+	if (test_env_value < 0)
+		test_env_value = git_env_bool("GIT_TEST_FSMONITOR_TOKEN", 0);
+
+	if (!test_env_value) {
+		struct timeval tv;
+		struct tm tm;
+		time_t secs;
+
+		gettimeofday(&tv, NULL);
+		secs = tv.tv_sec;
+		gmtime_r(&secs, &tm);
+
+		strbuf_addf(&token->token_id,
+			    "%"PRIu64".%d.%4d%02d%02dT%02d%02d%02d.%06ldZ",
+			    flush_count++,
+			    getpid(),
+			    tm.tm_year + 1900, tm.tm_mon + 1, tm.tm_mday,
+			    tm.tm_hour, tm.tm_min, tm.tm_sec,
+			    (long)tv.tv_usec);
+	} else {
+		strbuf_addf(&token->token_id, "test_%08x", test_env_value++);
+	}
+
+	return token;
+}
+
 static ipc_server_application_cb handle_client;
 
 static int handle_client(void *data,
@@ -298,7 +412,7 @@ static int fsmonitor_run_daemon(void)
 
 	pthread_mutex_init(&state.main_lock, NULL);
 	state.error_code = 0;
-	state.current_token_data = NULL;
+	state.current_token_data = fsmonitor_new_token_data();
 
 	/* Prepare to (recursively) watch the <worktree-root> directory. */
 	strbuf_init(&state.path_worktree_watch, 0);
-- 
gitgitgadget

[PATCH v9 14/30] fsmonitor--daemon: create token-based changed path cache

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach fsmonitor--daemon to build a list of changed paths and associate
them with a token-id.  This will be used by the platform-specific
backends to accumulate changed paths in response to filesystem events.

The platform-specific file system listener thread receives file system
events containing one or more changed pathnames (with whatever
bucketing or grouping that is convenient for the file system).  These
paths are accumulated (without locking) by the file system layer into
a `fsmonitor_batch`.

When the file system layer has drained the kernel event queue, it will
"publish" them to our token queue and make them visible to concurrent
client worker threads.  The token layer is free to combine and/or de-dup
paths within these batches for efficient presentation to clients.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 builtin/fsmonitor--daemon.c | 230 +++++++++++++++++++++++++++++++++++-
 fsmonitor--daemon.h         |  40 +++++++
 2 files changed, 268 insertions(+), 2 deletions(-)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index 1c7c156288d..69312119b07 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -181,17 +181,27 @@ struct fsmonitor_token_data {
 	uint64_t client_ref_count;
 };
 
+struct fsmonitor_batch {
+	struct fsmonitor_batch *next;
+	uint64_t batch_seq_nr;
+	const char **interned_paths;
+	size_t nr, alloc;
+	time_t pinned_time;
+};
+
 static struct fsmonitor_token_data *fsmonitor_new_token_data(void)
 {
 	static int test_env_value = -1;
 	static uint64_t flush_count = 0;
 	struct fsmonitor_token_data *token;
+	struct fsmonitor_batch *batch;
 
 	CALLOC_ARRAY(token, 1);
+	batch = fsmonitor_batch__new();
 
 	strbuf_init(&token->token_id, 0);
-	token->batch_head = NULL;
-	token->batch_tail = NULL;
+	token->batch_head = batch;
+	token->batch_tail = batch;
 	token->client_ref_count = 0;
 
 	if (test_env_value < 0)
@@ -217,9 +227,143 @@ static struct fsmonitor_token_data *fsmonitor_new_token_data(void)
 		strbuf_addf(&token->token_id, "test_%08x", test_env_value++);
 	}
 
+	/*
+	 * We created a new <token_id> and are starting a new series
+	 * of tokens with a zero <seq_nr>.
+	 *
+	 * Since clients cannot guess our new (non test) <token_id>
+	 * they will always receive a trivial response (because of the
+	 * mismatch on the <token_id>).  The trivial response will
+	 * tell them our new <token_id> so that subsequent requests
+	 * will be relative to our new series.  (And when sending that
+	 * response, we pin the current head of the batch list.)
+	 *
+	 * Even if the client correctly guesses the <token_id>, their
+	 * request of "builtin:<token_id>:0" asks for all changes MORE
+	 * RECENT than batch/bin 0.
+	 *
+	 * This implies that it is a waste to accumulate paths in the
+	 * initial batch/bin (because they will never be transmitted).
+	 *
+	 * So the daemon could be running for days and watching the
+	 * file system, but doesn't need to actually accumulate any
+	 * paths UNTIL we need to set a reference point for a later
+	 * relative request.
+	 *
+	 * However, it is very useful for testing to always have a
+	 * reference point set.  Pin batch 0 to force early file system
+	 * events to accumulate.
+	 */
+	if (test_env_value)
+		batch->pinned_time = time(NULL);
+
 	return token;
 }
 
+struct fsmonitor_batch *fsmonitor_batch__new(void)
+{
+	struct fsmonitor_batch *batch;
+
+	CALLOC_ARRAY(batch, 1);
+
+	return batch;
+}
+
+void fsmonitor_batch__free_list(struct fsmonitor_batch *batch)
+{
+	while (batch) {
+		struct fsmonitor_batch *next = batch->next;
+
+		/*
+		 * The actual strings within the array of this batch
+		 * are interned, so we don't own them.  We only own
+		 * the array.
+		 */
+		free(batch->interned_paths);
+		free(batch);
+
+		batch = next;
+	}
+}
+
+void fsmonitor_batch__add_path(struct fsmonitor_batch *batch,
+			       const char *path)
+{
+	const char *interned_path = strintern(path);
+
+	trace_printf_key(&trace_fsmonitor, "event: %s", interned_path);
+
+	ALLOC_GROW(batch->interned_paths, batch->nr + 1, batch->alloc);
+	batch->interned_paths[batch->nr++] = interned_path;
+}
+
+static void fsmonitor_batch__combine(struct fsmonitor_batch *batch_dest,
+				     const struct fsmonitor_batch *batch_src)
+{
+	size_t k;
+
+	ALLOC_GROW(batch_dest->interned_paths,
+		   batch_dest->nr + batch_src->nr + 1,
+		   batch_dest->alloc);
+
+	for (k = 0; k < batch_src->nr; k++)
+		batch_dest->interned_paths[batch_dest->nr++] =
+			batch_src->interned_paths[k];
+}
+
+static void fsmonitor_free_token_data(struct fsmonitor_token_data *token)
+{
+	if (!token)
+		return;
+
+	assert(token->client_ref_count == 0);
+
+	strbuf_release(&token->token_id);
+
+	fsmonitor_batch__free_list(token->batch_head);
+
+	free(token);
+}
+
+/*
+ * Flush all of our cached data about the filesystem.  Call this if we
+ * lose sync with the filesystem and miss some notification events.
+ *
+ * [1] If we are missing events, then we no longer have a complete
+ *     history of the directory (relative to our current start token).
+ *     We should create a new token and start fresh (as if we just
+ *     booted up).
+ *
+ * If there are no concurrent threads reading the current token data
+ * series, we can free it now.  Otherwise, let the last reader free
+ * it.
+ *
+ * Either way, the old token data series is no longer associated with
+ * our state data.
+ */
+static void with_lock__do_force_resync(struct fsmonitor_daemon_state *state)
+{
+	/* assert current thread holding state->main_lock */
+
+	struct fsmonitor_token_data *free_me = NULL;
+	struct fsmonitor_token_data *new_one = NULL;
+
+	new_one = fsmonitor_new_token_data();
+
+	if (state->current_token_data->client_ref_count == 0)
+		free_me = state->current_token_data;
+	state->current_token_data = new_one;
+
+	fsmonitor_free_token_data(free_me);
+}
+
+void fsmonitor_force_resync(struct fsmonitor_daemon_state *state)
+{
+	pthread_mutex_lock(&state->main_lock);
+	with_lock__do_force_resync(state);
+	pthread_mutex_unlock(&state->main_lock);
+}
+
 static ipc_server_application_cb handle_client;
 
 static int handle_client(void *data,
@@ -329,6 +473,81 @@ enum fsmonitor_path_type fsmonitor_classify_path_absolute(
 	return fsmonitor_classify_path_gitdir_relative(rel);
 }
 
+/*
+ * We try to combine small batches at the front of the batch-list to avoid
+ * having a long list.  This hopefully makes it a little easier when we want
+ * to truncate and maintain the list.  However, we don't want the paths array
+ * to just keep growing and growing with realloc, so we insert an arbitrary
+ * limit.
+ */
+#define MY_COMBINE_LIMIT (1024)
+
+void fsmonitor_publish(struct fsmonitor_daemon_state *state,
+		       struct fsmonitor_batch *batch,
+		       const struct string_list *cookie_names)
+{
+	if (!batch && !cookie_names->nr)
+		return;
+
+	pthread_mutex_lock(&state->main_lock);
+
+	if (batch) {
+		struct fsmonitor_batch *head;
+
+		head = state->current_token_data->batch_head;
+		if (!head) {
+			BUG("token does not have batch");
+		} else if (head->pinned_time) {
+			/*
+			 * We cannot alter the current batch list
+			 * because:
+			 *
+			 * [a] it is being transmitted to at least one
+			 * client and the handle_client() thread has a
+			 * ref-count, but not a lock on the batch list
+			 * starting with this item.
+			 *
+			 * [b] it has been transmitted in the past to
+			 * at least one client such that future
+			 * requests are relative to this head batch.
+			 *
+			 * So, we can only prepend a new batch onto
+			 * the front of the list.
+			 */
+			batch->batch_seq_nr = head->batch_seq_nr + 1;
+			batch->next = head;
+			state->current_token_data->batch_head = batch;
+		} else if (!head->batch_seq_nr) {
+			/*
+			 * Batch 0 is unpinned.  See the note in
+			 * `fsmonitor_new_token_data()` about why we
+			 * don't need to accumulate these paths.
+			 */
+			fsmonitor_batch__free_list(batch);
+		} else if (head->nr + batch->nr > MY_COMBINE_LIMIT) {
+			/*
+			 * The head batch in the list has never been
+			 * transmitted to a client, but folding the
+			 * contents of the new batch onto it would
+			 * exceed our arbitrary limit, so just prepend
+			 * the new batch onto the list.
+			 */
+			batch->batch_seq_nr = head->batch_seq_nr + 1;
+			batch->next = head;
+			state->current_token_data->batch_head = batch;
+		} else {
+			/*
+			 * We are free to add the paths in the given
+			 * batch onto the end of the current head batch.
+			 */
+			fsmonitor_batch__combine(head, batch);
+			fsmonitor_batch__free_list(batch);
+		}
+	}
+
+	pthread_mutex_unlock(&state->main_lock);
+}
+
 static void *fsm_listen__thread_proc(void *_state)
 {
 	struct fsmonitor_daemon_state *state = _state;
@@ -343,6 +562,13 @@ static void *fsm_listen__thread_proc(void *_state)
 
 	fsm_listen__loop(state);
 
+	pthread_mutex_lock(&state->main_lock);
+	if (state->current_token_data &&
+	    state->current_token_data->client_ref_count == 0)
+		fsmonitor_free_token_data(state->current_token_data);
+	state->current_token_data = NULL;
+	pthread_mutex_unlock(&state->main_lock);
+
 	trace2_thread_exit();
 	return NULL;
 }
diff --git a/fsmonitor--daemon.h b/fsmonitor--daemon.h
index 8c3a71a48bd..010fbfe60e9 100644
--- a/fsmonitor--daemon.h
+++ b/fsmonitor--daemon.h
@@ -12,6 +12,27 @@
 struct fsmonitor_batch;
 struct fsmonitor_token_data;
 
+/*
+ * Create a new batch of path(s).  The returned batch is considered
+ * private and not linked into the fsmonitor daemon state.  The caller
+ * should fill this batch with one or more paths and then publish it.
+ */
+struct fsmonitor_batch *fsmonitor_batch__new(void);
+
+/*
+ * Free the list of batches starting with this one.
+ */
+void fsmonitor_batch__free_list(struct fsmonitor_batch *batch);
+
+/*
+ * Add this path to this batch of modified files.
+ *
+ * The batch should be private and NOT (yet) linked into the fsmonitor
+ * daemon state and therefore not yet visible to worker threads and so
+ * no locking is required.
+ */
+void fsmonitor_batch__add_path(struct fsmonitor_batch *batch, const char *path);
+
 struct fsmonitor_daemon_backend_data; /* opaque platform-specific data */
 
 struct fsmonitor_daemon_state {
@@ -117,5 +138,24 @@ enum fsmonitor_path_type fsmonitor_classify_path_absolute(
 	struct fsmonitor_daemon_state *state,
 	const char *path);
 
+/*
+ * Prepend the this batch of path(s) onto the list of batches associated
+ * with the current token.  This makes the batch visible to worker threads.
+ *
+ * The caller no longer owns the batch and must not free it.
+ *
+ * Wake up the client threads waiting on these cookies.
+ */
+void fsmonitor_publish(struct fsmonitor_daemon_state *state,
+		       struct fsmonitor_batch *batch,
+		       const struct string_list *cookie_names);
+
+/*
+ * If the platform-specific layer loses sync with the filesystem,
+ * it should call this to invalidate cached data and abort waiting
+ * threads.
+ */
+void fsmonitor_force_resync(struct fsmonitor_daemon_state *state);
+
 #endif /* HAVE_FSMONITOR_DAEMON_BACKEND */
 #endif /* FSMONITOR_DAEMON_H */
-- 
gitgitgadget

[PATCH v9 15/30] compat/fsmonitor/fsm-listen-win32: implement FSMonitor backend on Windows

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach the win32 backend to register a watch on the working tree
root directory (recursively).  Also watch the <gitdir> if it is
not inside the working tree.  And to collect path change notifications
into batches and publish.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 compat/fsmonitor/fsm-listen-win32.c | 565 ++++++++++++++++++++++++++++
 1 file changed, 565 insertions(+)

diff --git a/compat/fsmonitor/fsm-listen-win32.c b/compat/fsmonitor/fsm-listen-win32.c
index 916cbea254f..5b928ab66e5 100644
--- a/compat/fsmonitor/fsm-listen-win32.c
+++ b/compat/fsmonitor/fsm-listen-win32.c
@@ -2,20 +2,585 @@
 #include "config.h"
 #include "fsmonitor.h"
 #include "fsm-listen.h"
+#include "fsmonitor--daemon.h"
+
+/*
+ * The documentation of ReadDirectoryChangesW() states that the maximum
+ * buffer size is 64K when the monitored directory is remote.
+ *
+ * Larger buffers may be used when the monitored directory is local and
+ * will help us receive events faster from the kernel and avoid dropped
+ * events.
+ *
+ * So we try to use a very large buffer and silently fallback to 64K if
+ * we get an error.
+ */
+#define MAX_RDCW_BUF_FALLBACK (65536)
+#define MAX_RDCW_BUF          (65536 * 8)
+
+struct one_watch
+{
+	char buffer[MAX_RDCW_BUF];
+	DWORD buf_len;
+	DWORD count;
+
+	struct strbuf path;
+	HANDLE hDir;
+	HANDLE hEvent;
+	OVERLAPPED overlapped;
+
+	/*
+	 * Is there an active ReadDirectoryChangesW() call pending.  If so, we
+	 * need to later call GetOverlappedResult() and possibly CancelIoEx().
+	 */
+	BOOL is_active;
+};
+
+struct fsmonitor_daemon_backend_data
+{
+	struct one_watch *watch_worktree;
+	struct one_watch *watch_gitdir;
+
+	HANDLE hEventShutdown;
+
+	HANDLE hListener[3]; /* we don't own these handles */
+#define LISTENER_SHUTDOWN 0
+#define LISTENER_HAVE_DATA_WORKTREE 1
+#define LISTENER_HAVE_DATA_GITDIR 2
+	int nr_listener_handles;
+};
+
+/*
+ * Convert the WCHAR path from the notification into UTF8 and
+ * then normalize it.
+ */
+static int normalize_path_in_utf8(FILE_NOTIFY_INFORMATION *info,
+				  struct strbuf *normalized_path)
+{
+	int reserve;
+	int len = 0;
+
+	strbuf_reset(normalized_path);
+	if (!info->FileNameLength)
+		goto normalize;
+
+	/*
+	 * Pre-reserve enough space in the UTF8 buffer for
+	 * each Unicode WCHAR character to be mapped into a
+	 * sequence of 2 UTF8 characters.  That should let us
+	 * avoid ERROR_INSUFFICIENT_BUFFER 99.9+% of the time.
+	 */
+	reserve = info->FileNameLength + 1;
+	strbuf_grow(normalized_path, reserve);
+
+	for (;;) {
+		len = WideCharToMultiByte(CP_UTF8, 0, info->FileName,
+					  info->FileNameLength / sizeof(WCHAR),
+					  normalized_path->buf,
+					  strbuf_avail(normalized_path) - 1,
+					  NULL, NULL);
+		if (len > 0)
+			goto normalize;
+		if (GetLastError() != ERROR_INSUFFICIENT_BUFFER) {
+			error(_("[GLE %ld] could not convert path to UTF-8: '%.*ls'"),
+			      GetLastError(),
+			      (int)(info->FileNameLength / sizeof(WCHAR)),
+			      info->FileName);
+			return -1;
+		}
+
+		strbuf_grow(normalized_path,
+			    strbuf_avail(normalized_path) + reserve);
+	}
+
+normalize:
+	strbuf_setlen(normalized_path, len);
+	return strbuf_normalize_path(normalized_path);
+}
 
 void fsm_listen__stop_async(struct fsmonitor_daemon_state *state)
 {
+	SetEvent(state->backend_data->hListener[LISTENER_SHUTDOWN]);
+}
+
+static struct one_watch *create_watch(struct fsmonitor_daemon_state *state,
+				      const char *path)
+{
+	struct one_watch *watch = NULL;
+	DWORD desired_access = FILE_LIST_DIRECTORY;
+	DWORD share_mode =
+		FILE_SHARE_WRITE | FILE_SHARE_READ | FILE_SHARE_DELETE;
+	HANDLE hDir;
+	wchar_t wpath[MAX_PATH];
+
+	if (xutftowcs_path(wpath, path) < 0) {
+		error(_("could not convert to wide characters: '%s'"), path);
+		return NULL;
+	}
+
+	hDir = CreateFileW(wpath,
+			   desired_access, share_mode, NULL, OPEN_EXISTING,
+			   FILE_FLAG_BACKUP_SEMANTICS | FILE_FLAG_OVERLAPPED,
+			   NULL);
+	if (hDir == INVALID_HANDLE_VALUE) {
+		error(_("[GLE %ld] could not watch '%s'"),
+		      GetLastError(), path);
+		return NULL;
+	}
+
+	CALLOC_ARRAY(watch, 1);
+
+	watch->buf_len = sizeof(watch->buffer); /* assume full MAX_RDCW_BUF */
+
+	strbuf_init(&watch->path, 0);
+	strbuf_addstr(&watch->path, path);
+
+	watch->hDir = hDir;
+	watch->hEvent = CreateEvent(NULL, TRUE, FALSE, NULL);
+
+	return watch;
+}
+
+static void destroy_watch(struct one_watch *watch)
+{
+	if (!watch)
+		return;
+
+	strbuf_release(&watch->path);
+	if (watch->hDir != INVALID_HANDLE_VALUE)
+		CloseHandle(watch->hDir);
+	if (watch->hEvent != INVALID_HANDLE_VALUE)
+		CloseHandle(watch->hEvent);
+
+	free(watch);
+}
+
+static int start_rdcw_watch(struct fsmonitor_daemon_backend_data *data,
+			    struct one_watch *watch)
+{
+	DWORD dwNotifyFilter =
+		FILE_NOTIFY_CHANGE_FILE_NAME |
+		FILE_NOTIFY_CHANGE_DIR_NAME |
+		FILE_NOTIFY_CHANGE_ATTRIBUTES |
+		FILE_NOTIFY_CHANGE_SIZE |
+		FILE_NOTIFY_CHANGE_LAST_WRITE |
+		FILE_NOTIFY_CHANGE_CREATION;
+
+	ResetEvent(watch->hEvent);
+
+	memset(&watch->overlapped, 0, sizeof(watch->overlapped));
+	watch->overlapped.hEvent = watch->hEvent;
+
+	/*
+	 * Queue an async call using Overlapped IO.  This returns immediately.
+	 * Our event handle will be signalled when the real result is available.
+	 *
+	 * The return value here just means that we successfully queued it.
+	 * We won't know if the Read...() actually produces data until later.
+	 */
+	watch->is_active = ReadDirectoryChangesW(
+		watch->hDir, watch->buffer, watch->buf_len, TRUE,
+		dwNotifyFilter, &watch->count, &watch->overlapped, NULL);
+
+	if (watch->is_active)
+		return 0;
+
+	error(_("ReadDirectoryChangedW failed on '%s' [GLE %ld]"),
+	      watch->path.buf, GetLastError());
+	return -1;
+}
+
+static int recv_rdcw_watch(struct one_watch *watch)
+{
+	DWORD gle;
+
+	watch->is_active = FALSE;
+
+	/*
+	 * The overlapped result is ready.  If the Read...() was successful
+	 * we finally receive the actual result into our buffer.
+	 */
+	if (GetOverlappedResult(watch->hDir, &watch->overlapped, &watch->count,
+				TRUE))
+		return 0;
+
+	gle = GetLastError();
+	if (gle == ERROR_INVALID_PARAMETER &&
+	    /*
+	     * The kernel throws an invalid parameter error when our
+	     * buffer is too big and we are pointed at a remote
+	     * directory (and possibly for other reasons).  Quietly
+	     * set it down and try again.
+	     *
+	     * See note about MAX_RDCW_BUF at the top.
+	     */
+	    watch->buf_len > MAX_RDCW_BUF_FALLBACK) {
+		watch->buf_len = MAX_RDCW_BUF_FALLBACK;
+		return -2;
+	}
+
+	/*
+	 * NEEDSWORK: If an external <gitdir> is deleted, the above
+	 * returns an error.  I'm not sure that there's anything that
+	 * we can do here other than failing -- the <worktree>/.git
+	 * link file would be broken anyway.  We might try to check
+	 * for that and return a better error message, but I'm not
+	 * sure it is worth it.
+	 */
+
+	error(_("GetOverlappedResult failed on '%s' [GLE %ld]"),
+	      watch->path.buf, gle);
+	return -1;
+}
+
+static void cancel_rdcw_watch(struct one_watch *watch)
+{
+	DWORD count;
+
+	if (!watch || !watch->is_active)
+		return;
+
+	/*
+	 * The calls to ReadDirectoryChangesW() and GetOverlappedResult()
+	 * form a "pair" (my term) where we queue an IO and promise to
+	 * hang around and wait for the kernel to give us the result.
+	 *
+	 * If for some reason after we queue the IO, we have to quit
+	 * or otherwise not stick around for the second half, we must
+	 * tell the kernel to abort the IO.  This prevents the kernel
+	 * from writing to our buffer and/or signalling our event
+	 * after we free them.
+	 *
+	 * (Ask me how much fun it was to track that one down).
+	 */
+	CancelIoEx(watch->hDir, &watch->overlapped);
+	GetOverlappedResult(watch->hDir, &watch->overlapped, &count, TRUE);
+	watch->is_active = FALSE;
+}
+
+/*
+ * Process filesystem events that happen anywhere (recursively) under the
+ * <worktree> root directory.  For a normal working directory, this includes
+ * both version controlled files and the contents of the .git/ directory.
+ *
+ * If <worktree>/.git is a file, then we only see events for the file
+ * itself.
+ */
+static int process_worktree_events(struct fsmonitor_daemon_state *state)
+{
+	struct fsmonitor_daemon_backend_data *data = state->backend_data;
+	struct one_watch *watch = data->watch_worktree;
+	struct strbuf path = STRBUF_INIT;
+	struct string_list cookie_list = STRING_LIST_INIT_DUP;
+	struct fsmonitor_batch *batch = NULL;
+	const char *p = watch->buffer;
+
+	/*
+	 * If the kernel gets more events than will fit in the kernel
+	 * buffer associated with our RDCW handle, it drops them and
+	 * returns a count of zero.
+	 *
+	 * Yes, the call returns WITHOUT error and with length zero.
+	 * This is the documented behavior.  (My testing has confirmed
+	 * that it also sets the last error to ERROR_NOTIFY_ENUM_DIR,
+	 * but we do not rely on that since the function did not
+	 * return an error and it is not documented.)
+	 *
+	 * (The "overflow" case is not ambiguous with the "no data" case
+	 * because we did an INFINITE wait.)
+	 *
+	 * This means we have a gap in coverage.  Tell the daemon layer
+	 * to resync.
+	 */
+	if (!watch->count) {
+		trace2_data_string("fsmonitor", NULL, "fsm-listen/kernel",
+				   "overflow");
+		fsmonitor_force_resync(state);
+		return LISTENER_HAVE_DATA_WORKTREE;
+	}
+
+	/*
+	 * On Windows, `info` contains an "array" of paths that are
+	 * relative to the root of whichever directory handle received
+	 * the event.
+	 */
+	for (;;) {
+		FILE_NOTIFY_INFORMATION *info = (void *)p;
+		const char *slash;
+		enum fsmonitor_path_type t;
+
+		strbuf_reset(&path);
+		if (normalize_path_in_utf8(info, &path) == -1)
+			goto skip_this_path;
+
+		t = fsmonitor_classify_path_workdir_relative(path.buf);
+
+		switch (t) {
+		case IS_INSIDE_DOT_GIT_WITH_COOKIE_PREFIX:
+			/* special case cookie files within .git */
+
+			/* Use just the filename of the cookie file. */
+			slash = find_last_dir_sep(path.buf);
+			string_list_append(&cookie_list,
+					   slash ? slash + 1 : path.buf);
+			break;
+
+		case IS_INSIDE_DOT_GIT:
+			/* ignore everything inside of "<worktree>/.git/" */
+			break;
+
+		case IS_DOT_GIT:
+			/* "<worktree>/.git" was deleted (or renamed away) */
+			if ((info->Action == FILE_ACTION_REMOVED) ||
+			    (info->Action == FILE_ACTION_RENAMED_OLD_NAME)) {
+				trace2_data_string("fsmonitor", NULL,
+						   "fsm-listen/dotgit",
+						   "removed");
+				goto force_shutdown;
+			}
+			break;
+
+		case IS_WORKDIR_PATH:
+			/* queue normal pathname */
+			if (!batch)
+				batch = fsmonitor_batch__new();
+			fsmonitor_batch__add_path(batch, path.buf);
+			break;
+
+		case IS_GITDIR:
+		case IS_INSIDE_GITDIR:
+		case IS_INSIDE_GITDIR_WITH_COOKIE_PREFIX:
+		default:
+			BUG("unexpected path classification '%d' for '%s'",
+			    t, path.buf);
+		}
+
+skip_this_path:
+		if (!info->NextEntryOffset)
+			break;
+		p += info->NextEntryOffset;
+	}
+
+	fsmonitor_publish(state, batch, &cookie_list);
+	batch = NULL;
+	string_list_clear(&cookie_list, 0);
+	strbuf_release(&path);
+	return LISTENER_HAVE_DATA_WORKTREE;
+
+force_shutdown:
+	fsmonitor_batch__free_list(batch);
+	string_list_clear(&cookie_list, 0);
+	strbuf_release(&path);
+	return LISTENER_SHUTDOWN;
+}
+
+/*
+ * Process filesystem events that happened anywhere (recursively) under the
+ * external <gitdir> (such as non-primary worktrees or submodules).
+ * We only care about cookie files that our client threads created here.
+ *
+ * Note that we DO NOT get filesystem events on the external <gitdir>
+ * itself (it is not inside something that we are watching).  In particular,
+ * we do not get an event if the external <gitdir> is deleted.
+ */
+static int process_gitdir_events(struct fsmonitor_daemon_state *state)
+{
+	struct fsmonitor_daemon_backend_data *data = state->backend_data;
+	struct one_watch *watch = data->watch_gitdir;
+	struct strbuf path = STRBUF_INIT;
+	struct string_list cookie_list = STRING_LIST_INIT_DUP;
+	const char *p = watch->buffer;
+
+	if (!watch->count) {
+		trace2_data_string("fsmonitor", NULL, "fsm-listen/kernel",
+				   "overflow");
+		fsmonitor_force_resync(state);
+		return LISTENER_HAVE_DATA_GITDIR;
+	}
+
+	for (;;) {
+		FILE_NOTIFY_INFORMATION *info = (void *)p;
+		const char *slash;
+		enum fsmonitor_path_type t;
+
+		strbuf_reset(&path);
+		if (normalize_path_in_utf8(info, &path) == -1)
+			goto skip_this_path;
+
+		t = fsmonitor_classify_path_gitdir_relative(path.buf);
+
+		switch (t) {
+		case IS_INSIDE_GITDIR_WITH_COOKIE_PREFIX:
+			/* special case cookie files within gitdir */
+
+			/* Use just the filename of the cookie file. */
+			slash = find_last_dir_sep(path.buf);
+			string_list_append(&cookie_list,
+					   slash ? slash + 1 : path.buf);
+			break;
+
+		case IS_INSIDE_GITDIR:
+			goto skip_this_path;
+
+		default:
+			BUG("unexpected path classification '%d' for '%s'",
+			    t, path.buf);
+		}
+
+skip_this_path:
+		if (!info->NextEntryOffset)
+			break;
+		p += info->NextEntryOffset;
+	}
+
+	fsmonitor_publish(state, NULL, &cookie_list);
+	string_list_clear(&cookie_list, 0);
+	strbuf_release(&path);
+	return LISTENER_HAVE_DATA_GITDIR;
 }
 
 void fsm_listen__loop(struct fsmonitor_daemon_state *state)
 {
+	struct fsmonitor_daemon_backend_data *data = state->backend_data;
+	DWORD dwWait;
+	int result;
+
+	state->error_code = 0;
+
+	if (start_rdcw_watch(data, data->watch_worktree) == -1)
+		goto force_error_stop;
+
+	if (data->watch_gitdir &&
+	    start_rdcw_watch(data, data->watch_gitdir) == -1)
+		goto force_error_stop;
+
+	for (;;) {
+		dwWait = WaitForMultipleObjects(data->nr_listener_handles,
+						data->hListener,
+						FALSE, INFINITE);
+
+		if (dwWait == WAIT_OBJECT_0 + LISTENER_HAVE_DATA_WORKTREE) {
+			result = recv_rdcw_watch(data->watch_worktree);
+			if (result == -1) {
+				/* hard error */
+				goto force_error_stop;
+			}
+			if (result == -2) {
+				/* retryable error */
+				if (start_rdcw_watch(data, data->watch_worktree) == -1)
+					goto force_error_stop;
+				continue;
+			}
+
+			/* have data */
+			if (process_worktree_events(state) == LISTENER_SHUTDOWN)
+				goto force_shutdown;
+			if (start_rdcw_watch(data, data->watch_worktree) == -1)
+				goto force_error_stop;
+			continue;
+		}
+
+		if (dwWait == WAIT_OBJECT_0 + LISTENER_HAVE_DATA_GITDIR) {
+			result = recv_rdcw_watch(data->watch_gitdir);
+			if (result == -1) {
+				/* hard error */
+				goto force_error_stop;
+			}
+			if (result == -2) {
+				/* retryable error */
+				if (start_rdcw_watch(data, data->watch_gitdir) == -1)
+					goto force_error_stop;
+				continue;
+			}
+
+			/* have data */
+			if (process_gitdir_events(state) == LISTENER_SHUTDOWN)
+				goto force_shutdown;
+			if (start_rdcw_watch(data, data->watch_gitdir) == -1)
+				goto force_error_stop;
+			continue;
+		}
+
+		if (dwWait == WAIT_OBJECT_0 + LISTENER_SHUTDOWN)
+			goto clean_shutdown;
+
+		error(_("could not read directory changes [GLE %ld]"),
+		      GetLastError());
+		goto force_error_stop;
+	}
+
+force_error_stop:
+	state->error_code = -1;
+
+force_shutdown:
+	/*
+	 * Tell the IPC thead pool to stop (which completes the await
+	 * in the main thread (which will also signal this thread (if
+	 * we are still alive))).
+	 */
+	ipc_server_stop_async(state->ipc_server_data);
+
+clean_shutdown:
+	cancel_rdcw_watch(data->watch_worktree);
+	cancel_rdcw_watch(data->watch_gitdir);
 }
 
 int fsm_listen__ctor(struct fsmonitor_daemon_state *state)
 {
+	struct fsmonitor_daemon_backend_data *data;
+
+	CALLOC_ARRAY(data, 1);
+
+	data->hEventShutdown = CreateEvent(NULL, TRUE, FALSE, NULL);
+
+	data->watch_worktree = create_watch(state,
+					    state->path_worktree_watch.buf);
+	if (!data->watch_worktree)
+		goto failed;
+
+	if (state->nr_paths_watching > 1) {
+		data->watch_gitdir = create_watch(state,
+						  state->path_gitdir_watch.buf);
+		if (!data->watch_gitdir)
+			goto failed;
+	}
+
+	data->hListener[LISTENER_SHUTDOWN] = data->hEventShutdown;
+	data->nr_listener_handles++;
+
+	data->hListener[LISTENER_HAVE_DATA_WORKTREE] =
+		data->watch_worktree->hEvent;
+	data->nr_listener_handles++;
+
+	if (data->watch_gitdir) {
+		data->hListener[LISTENER_HAVE_DATA_GITDIR] =
+			data->watch_gitdir->hEvent;
+		data->nr_listener_handles++;
+	}
+
+	state->backend_data = data;
+	return 0;
+
+failed:
+	CloseHandle(data->hEventShutdown);
+	destroy_watch(data->watch_worktree);
+	destroy_watch(data->watch_gitdir);
+
 	return -1;
 }
 
 void fsm_listen__dtor(struct fsmonitor_daemon_state *state)
 {
+	struct fsmonitor_daemon_backend_data *data;
+
+	if (!state || !state->backend_data)
+		return;
+
+	data = state->backend_data;
+
+	CloseHandle(data->hEventShutdown);
+	destroy_watch(data->watch_worktree);
+	destroy_watch(data->watch_gitdir);
+
+	FREE_AND_NULL(state->backend_data);
 }
-- 
gitgitgadget

[PATCH v9 16/30] compat/fsmonitor/fsm-listen-darwin: add MacOS header files for FSEvent

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Include MacOS system declarations to allow us to use FSEvent and
CoreFoundation APIs.  We need different versions of the declarations
for GCC vs. clang because of compiler and header file conflicts.

While it is quite possible to #include Apple's CoreServices.h when
compiling C source code with clang, trying to build it with GCC
currently fails with this error:

In file included
   from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
   ...Library/Frameworks/Security.framework/Headers/AuthSession.h:32,
   from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
   ...Library/Frameworks/Security.framework/Headers/Security.h:42,
   from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
   ...Library/Frameworks/CoreServices.framework/Frameworks/...
   ...OSServices.framework/Headers/CSIdentity.h:43,
   from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
   ...Library/Frameworks/CoreServices.framework/Frameworks/...
   ...OSServices.framework/Headers/OSServices.h:29,
   from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
   ...Library/Frameworks/CoreServices.framework/Frameworks/...
   ...LaunchServices.framework/Headers/IconsCore.h:23,
   from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
   ...Library/Frameworks/CoreServices.framework/Frameworks/...
   ...LaunchServices.framework/Headers/LaunchServices.h:23,
   from /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
   ...Library/Frameworks/CoreServices.framework/Headers/CoreServices.h:45,

     /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/...
     ...Library/Frameworks/Security.framework/Headers/Authorization.h:193:7:
     error: variably modified 'bytes' at file scope
       193 | char bytes[kAuthorizationExternalFormLength];
           |      ^~~~~

The underlying reason is that GCC (rightfully) objects that an `enum`
value such as `kAuthorizationExternalFormLength` is not a constant
(because it is not, the preprocessor has no knowledge of it, only the
actual C compiler does) and can therefore not be used to define the size
of a C array.

This is a known problem and tracked in GCC's bug tracker:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93082

In the meantime, let's not block things and go the slightly ugly route
of declaring/defining the FSEvents constants, data structures and
functions that we need, so that we can avoid above-mentioned issue.

Let's do this _only_ for GCC, though, so that the CI/PR builds (which
build both with clang and with GCC) can guarantee that we _are_ using
the correct data types.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 compat/fsmonitor/fsm-darwin-gcc.h    | 92 ++++++++++++++++++++++++++++
 compat/fsmonitor/fsm-listen-darwin.c | 24 ++++++++
 2 files changed, 116 insertions(+)
 create mode 100644 compat/fsmonitor/fsm-darwin-gcc.h

diff --git a/compat/fsmonitor/fsm-darwin-gcc.h b/compat/fsmonitor/fsm-darwin-gcc.h
new file mode 100644
index 00000000000..1c75c3d48e7
--- /dev/null
+++ b/compat/fsmonitor/fsm-darwin-gcc.h
@@ -0,0 +1,92 @@
+#ifndef FSM_DARWIN_GCC_H
+#define FSM_DARWIN_GCC_H
+
+#ifndef __clang__
+/*
+ * It is possible to #include CoreFoundation/CoreFoundation.h when compiling
+ * with clang, but not with GCC as of time of writing.
+ *
+ * See https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93082 for details.
+ */
+typedef unsigned int FSEventStreamCreateFlags;
+#define kFSEventStreamEventFlagNone               0x00000000
+#define kFSEventStreamEventFlagMustScanSubDirs    0x00000001
+#define kFSEventStreamEventFlagUserDropped        0x00000002
+#define kFSEventStreamEventFlagKernelDropped      0x00000004
+#define kFSEventStreamEventFlagEventIdsWrapped    0x00000008
+#define kFSEventStreamEventFlagHistoryDone        0x00000010
+#define kFSEventStreamEventFlagRootChanged        0x00000020
+#define kFSEventStreamEventFlagMount              0x00000040
+#define kFSEventStreamEventFlagUnmount            0x00000080
+#define kFSEventStreamEventFlagItemCreated        0x00000100
+#define kFSEventStreamEventFlagItemRemoved        0x00000200
+#define kFSEventStreamEventFlagItemInodeMetaMod   0x00000400
+#define kFSEventStreamEventFlagItemRenamed        0x00000800
+#define kFSEventStreamEventFlagItemModified       0x00001000
+#define kFSEventStreamEventFlagItemFinderInfoMod  0x00002000
+#define kFSEventStreamEventFlagItemChangeOwner    0x00004000
+#define kFSEventStreamEventFlagItemXattrMod       0x00008000
+#define kFSEventStreamEventFlagItemIsFile         0x00010000
+#define kFSEventStreamEventFlagItemIsDir          0x00020000
+#define kFSEventStreamEventFlagItemIsSymlink      0x00040000
+#define kFSEventStreamEventFlagOwnEvent           0x00080000
+#define kFSEventStreamEventFlagItemIsHardlink     0x00100000
+#define kFSEventStreamEventFlagItemIsLastHardlink 0x00200000
+#define kFSEventStreamEventFlagItemCloned         0x00400000
+
+typedef struct __FSEventStream *FSEventStreamRef;
+typedef const FSEventStreamRef ConstFSEventStreamRef;
+
+typedef unsigned int CFStringEncoding;
+#define kCFStringEncodingUTF8 0x08000100
+
+typedef const struct __CFString *CFStringRef;
+typedef const struct __CFArray *CFArrayRef;
+typedef const struct __CFRunLoop *CFRunLoopRef;
+
+struct FSEventStreamContext {
+    long long version;
+    void *cb_data, *retain, *release, *copy_description;
+};
+
+typedef struct FSEventStreamContext FSEventStreamContext;
+typedef unsigned int FSEventStreamEventFlags;
+#define kFSEventStreamCreateFlagNoDefer 0x02
+#define kFSEventStreamCreateFlagWatchRoot 0x04
+#define kFSEventStreamCreateFlagFileEvents 0x10
+
+typedef unsigned long long FSEventStreamEventId;
+#define kFSEventStreamEventIdSinceNow 0xFFFFFFFFFFFFFFFFULL
+
+typedef void (*FSEventStreamCallback)(ConstFSEventStreamRef streamRef,
+				      void *context,
+				      __SIZE_TYPE__ num_of_events,
+				      void *event_paths,
+				      const FSEventStreamEventFlags event_flags[],
+				      const FSEventStreamEventId event_ids[]);
+typedef double CFTimeInterval;
+FSEventStreamRef FSEventStreamCreate(void *allocator,
+				     FSEventStreamCallback callback,
+				     FSEventStreamContext *context,
+				     CFArrayRef paths_to_watch,
+				     FSEventStreamEventId since_when,
+				     CFTimeInterval latency,
+				     FSEventStreamCreateFlags flags);
+CFStringRef CFStringCreateWithCString(void *allocator, const char *string,
+				      CFStringEncoding encoding);
+CFArrayRef CFArrayCreate(void *allocator, const void **items, long long count,
+			 void *callbacks);
+void CFRunLoopRun(void);
+void CFRunLoopStop(CFRunLoopRef run_loop);
+CFRunLoopRef CFRunLoopGetCurrent(void);
+extern CFStringRef kCFRunLoopDefaultMode;
+void FSEventStreamScheduleWithRunLoop(FSEventStreamRef stream,
+				      CFRunLoopRef run_loop,
+				      CFStringRef run_loop_mode);
+unsigned char FSEventStreamStart(FSEventStreamRef stream);
+void FSEventStreamStop(FSEventStreamRef stream);
+void FSEventStreamInvalidate(FSEventStreamRef stream);
+void FSEventStreamRelease(FSEventStreamRef stream);
+
+#endif /* !clang */
+#endif /* FSM_DARWIN_GCC_H */
diff --git a/compat/fsmonitor/fsm-listen-darwin.c b/compat/fsmonitor/fsm-listen-darwin.c
index c84e3344ab9..d2ce942cade 100644
--- a/compat/fsmonitor/fsm-listen-darwin.c
+++ b/compat/fsmonitor/fsm-listen-darwin.c
@@ -1,3 +1,27 @@
+#ifndef __clang__
+#include "fsm-darwin-gcc.h"
+#else
+#include <CoreFoundation/CoreFoundation.h>
+#include <CoreServices/CoreServices.h>
+
+#ifndef AVAILABLE_MAC_OS_X_VERSION_10_13_AND_LATER
+/*
+ * This enum value was added in 10.13 to:
+ *
+ * /Applications/Xcode.app/Contents/Developer/Platforms/ \
+ *    MacOSX.platform/Developer/SDKs/MacOSX.sdk/System/ \
+ *    Library/Frameworks/CoreServices.framework/Frameworks/ \
+ *    FSEvents.framework/Versions/Current/Headers/FSEvents.h
+ *
+ * If we're compiling against an older SDK, this symbol won't be
+ * present.  Silently define it here so that we don't have to ifdef
+ * the logging or masking below.  This should be harmless since older
+ * versions of macOS won't ever emit this FS event anyway.
+ */
+#define kFSEventStreamEventFlagItemCloned         0x00400000
+#endif
+#endif
+
 #include "cache.h"
 #include "fsmonitor.h"
 #include "fsm-listen.h"
-- 
gitgitgadget

[PATCH v9 17/30] compat/fsmonitor/fsm-listen-darwin: implement FSEvent listener on MacOS

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Implement file system event listener on MacOS using FSEvent,
CoreFoundation, and CoreServices.

Co-authored-by: Kevin Willford <Kevin.Willford@microsoft.com>
Co-authored-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 compat/fsmonitor/fsm-listen-darwin.c | 383 +++++++++++++++++++++++++++
 1 file changed, 383 insertions(+)

diff --git a/compat/fsmonitor/fsm-listen-darwin.c b/compat/fsmonitor/fsm-listen-darwin.c
index d2ce942cade..0741fe834c3 100644
--- a/compat/fsmonitor/fsm-listen-darwin.c
+++ b/compat/fsmonitor/fsm-listen-darwin.c
@@ -25,20 +25,403 @@
 #include "cache.h"
 #include "fsmonitor.h"
 #include "fsm-listen.h"
+#include "fsmonitor--daemon.h"
+
+struct fsmonitor_daemon_backend_data
+{
+	CFStringRef cfsr_worktree_path;
+	CFStringRef cfsr_gitdir_path;
+
+	CFArrayRef cfar_paths_to_watch;
+	int nr_paths_watching;
+
+	FSEventStreamRef stream;
+
+	CFRunLoopRef rl;
+
+	enum shutdown_style {
+		SHUTDOWN_EVENT = 0,
+		FORCE_SHUTDOWN,
+		FORCE_ERROR_STOP,
+	} shutdown_style;
+
+	unsigned int stream_scheduled:1;
+	unsigned int stream_started:1;
+};
+
+static void log_flags_set(const char *path, const FSEventStreamEventFlags flag)
+{
+	struct strbuf msg = STRBUF_INIT;
+
+	if (flag & kFSEventStreamEventFlagMustScanSubDirs)
+		strbuf_addstr(&msg, "MustScanSubDirs|");
+	if (flag & kFSEventStreamEventFlagUserDropped)
+		strbuf_addstr(&msg, "UserDropped|");
+	if (flag & kFSEventStreamEventFlagKernelDropped)
+		strbuf_addstr(&msg, "KernelDropped|");
+	if (flag & kFSEventStreamEventFlagEventIdsWrapped)
+		strbuf_addstr(&msg, "EventIdsWrapped|");
+	if (flag & kFSEventStreamEventFlagHistoryDone)
+		strbuf_addstr(&msg, "HistoryDone|");
+	if (flag & kFSEventStreamEventFlagRootChanged)
+		strbuf_addstr(&msg, "RootChanged|");
+	if (flag & kFSEventStreamEventFlagMount)
+		strbuf_addstr(&msg, "Mount|");
+	if (flag & kFSEventStreamEventFlagUnmount)
+		strbuf_addstr(&msg, "Unmount|");
+	if (flag & kFSEventStreamEventFlagItemChangeOwner)
+		strbuf_addstr(&msg, "ItemChangeOwner|");
+	if (flag & kFSEventStreamEventFlagItemCreated)
+		strbuf_addstr(&msg, "ItemCreated|");
+	if (flag & kFSEventStreamEventFlagItemFinderInfoMod)
+		strbuf_addstr(&msg, "ItemFinderInfoMod|");
+	if (flag & kFSEventStreamEventFlagItemInodeMetaMod)
+		strbuf_addstr(&msg, "ItemInodeMetaMod|");
+	if (flag & kFSEventStreamEventFlagItemIsDir)
+		strbuf_addstr(&msg, "ItemIsDir|");
+	if (flag & kFSEventStreamEventFlagItemIsFile)
+		strbuf_addstr(&msg, "ItemIsFile|");
+	if (flag & kFSEventStreamEventFlagItemIsHardlink)
+		strbuf_addstr(&msg, "ItemIsHardlink|");
+	if (flag & kFSEventStreamEventFlagItemIsLastHardlink)
+		strbuf_addstr(&msg, "ItemIsLastHardlink|");
+	if (flag & kFSEventStreamEventFlagItemIsSymlink)
+		strbuf_addstr(&msg, "ItemIsSymlink|");
+	if (flag & kFSEventStreamEventFlagItemModified)
+		strbuf_addstr(&msg, "ItemModified|");
+	if (flag & kFSEventStreamEventFlagItemRemoved)
+		strbuf_addstr(&msg, "ItemRemoved|");
+	if (flag & kFSEventStreamEventFlagItemRenamed)
+		strbuf_addstr(&msg, "ItemRenamed|");
+	if (flag & kFSEventStreamEventFlagItemXattrMod)
+		strbuf_addstr(&msg, "ItemXattrMod|");
+	if (flag & kFSEventStreamEventFlagOwnEvent)
+		strbuf_addstr(&msg, "OwnEvent|");
+	if (flag & kFSEventStreamEventFlagItemCloned)
+		strbuf_addstr(&msg, "ItemCloned|");
+
+	trace_printf_key(&trace_fsmonitor, "fsevent: '%s', flags=%u %s",
+			 path, flag, msg.buf);
+
+	strbuf_release(&msg);
+}
+
+static int ef_is_root_delete(const FSEventStreamEventFlags ef)
+{
+	return (ef & kFSEventStreamEventFlagItemIsDir &&
+		ef & kFSEventStreamEventFlagItemRemoved);
+}
+
+static int ef_is_root_renamed(const FSEventStreamEventFlags ef)
+{
+	return (ef & kFSEventStreamEventFlagItemIsDir &&
+		ef & kFSEventStreamEventFlagItemRenamed);
+}
+
+static int ef_is_dropped(const FSEventStreamEventFlags ef)
+{
+	return (ef & kFSEventStreamEventFlagMustScanSubDirs ||
+		ef & kFSEventStreamEventFlagKernelDropped ||
+		ef & kFSEventStreamEventFlagUserDropped);
+}
+
+static void fsevent_callback(ConstFSEventStreamRef streamRef,
+			     void *ctx,
+			     size_t num_of_events,
+			     void *event_paths,
+			     const FSEventStreamEventFlags event_flags[],
+			     const FSEventStreamEventId event_ids[])
+{
+	struct fsmonitor_daemon_state *state = ctx;
+	struct fsmonitor_daemon_backend_data *data = state->backend_data;
+	char **paths = (char **)event_paths;
+	struct fsmonitor_batch *batch = NULL;
+	struct string_list cookie_list = STRING_LIST_INIT_DUP;
+	const char *path_k;
+	const char *slash;
+	int k;
+	struct strbuf tmp = STRBUF_INIT;
+
+	/*
+	 * Build a list of all filesystem changes into a private/local
+	 * list and without holding any locks.
+	 */
+	for (k = 0; k < num_of_events; k++) {
+		/*
+		 * On Mac, we receive an array of absolute paths.
+		 */
+		path_k = paths[k];
+
+		/*
+		 * If you want to debug FSEvents, log them to GIT_TRACE_FSMONITOR.
+		 * Please don't log them to Trace2.
+		 *
+		 * trace_printf_key(&trace_fsmonitor, "Path: '%s'", path_k);
+		 */
+
+		/*
+		 * If event[k] is marked as dropped, we assume that we have
+		 * lost sync with the filesystem and should flush our cached
+		 * data.  We need to:
+		 *
+		 * [1] Abort/wake any client threads waiting for a cookie and
+		 *     flush the cached state data (the current token), and
+		 *     create a new token.
+		 *
+		 * [2] Discard the batch that we were locally building (since
+		 *     they are conceptually relative to the just flushed
+		 *     token).
+		 */
+		if (ef_is_dropped(event_flags[k])) {
+			if (trace_pass_fl(&trace_fsmonitor))
+				log_flags_set(path_k, event_flags[k]);
+
+			fsmonitor_force_resync(state);
+			fsmonitor_batch__free_list(batch);
+			string_list_clear(&cookie_list, 0);
+
+			/*
+			 * We assume that any events that we received
+			 * in this callback after this dropped event
+			 * may still be valid, so we continue rather
+			 * than break.  (And just in case there is a
+			 * delete of ".git" hiding in there.)
+			 */
+			continue;
+		}
+
+		switch (fsmonitor_classify_path_absolute(state, path_k)) {
+
+		case IS_INSIDE_DOT_GIT_WITH_COOKIE_PREFIX:
+		case IS_INSIDE_GITDIR_WITH_COOKIE_PREFIX:
+			/* special case cookie files within .git or gitdir */
+
+			/* Use just the filename of the cookie file. */
+			slash = find_last_dir_sep(path_k);
+			string_list_append(&cookie_list,
+					   slash ? slash + 1 : path_k);
+			break;
+
+		case IS_INSIDE_DOT_GIT:
+		case IS_INSIDE_GITDIR:
+			/* ignore all other paths inside of .git or gitdir */
+			break;
+
+		case IS_DOT_GIT:
+		case IS_GITDIR:
+			/*
+			 * If .git directory is deleted or renamed away,
+			 * we have to quit.
+			 */
+			if (ef_is_root_delete(event_flags[k])) {
+				trace_printf_key(&trace_fsmonitor,
+						 "event: gitdir removed");
+				goto force_shutdown;
+			}
+			if (ef_is_root_renamed(event_flags[k])) {
+				trace_printf_key(&trace_fsmonitor,
+						 "event: gitdir renamed");
+				goto force_shutdown;
+			}
+			break;
+
+		case IS_WORKDIR_PATH:
+			/* try to queue normal pathnames */
+
+			if (trace_pass_fl(&trace_fsmonitor))
+				log_flags_set(path_k, event_flags[k]);
+
+			/*
+			 * Because of the implicit "binning" (the
+			 * kernel calls us at a given frequency) and
+			 * de-duping (the kernel is free to combine
+			 * multiple events for a given pathname), an
+			 * individual fsevent could be marked as both
+			 * a file and directory.  Add it to the queue
+			 * with both spellings so that the client will
+			 * know how much to invalidate/refresh.
+			 */
+
+			if (event_flags[k] & kFSEventStreamEventFlagItemIsFile) {
+				const char *rel = path_k +
+					state->path_worktree_watch.len + 1;
+
+				if (!batch)
+					batch = fsmonitor_batch__new();
+				fsmonitor_batch__add_path(batch, rel);
+			}
+
+			if (event_flags[k] & kFSEventStreamEventFlagItemIsDir) {
+				const char *rel = path_k +
+					state->path_worktree_watch.len + 1;
+
+				strbuf_reset(&tmp);
+				strbuf_addstr(&tmp, rel);
+				strbuf_addch(&tmp, '/');
+
+				if (!batch)
+					batch = fsmonitor_batch__new();
+				fsmonitor_batch__add_path(batch, tmp.buf);
+			}
+
+			break;
+
+		case IS_OUTSIDE_CONE:
+		default:
+			trace_printf_key(&trace_fsmonitor,
+					 "ignoring '%s'", path_k);
+			break;
+		}
+	}
+
+	fsmonitor_publish(state, batch, &cookie_list);
+	string_list_clear(&cookie_list, 0);
+	strbuf_release(&tmp);
+	return;
+
+force_shutdown:
+	fsmonitor_batch__free_list(batch);
+	string_list_clear(&cookie_list, 0);
+
+	data->shutdown_style = FORCE_SHUTDOWN;
+	CFRunLoopStop(data->rl);
+	strbuf_release(&tmp);
+	return;
+}
+
+/*
+ * In the call to `FSEventStreamCreate()` to setup our watch, the
+ * `latency` argument determines the frequency of calls to our callback
+ * with new FS events.  Too slow and events get dropped; too fast and
+ * we burn CPU unnecessarily.  Since it is rather obscure, I don't
+ * think this needs to be a config setting.  I've done extensive
+ * testing on my systems and chosen the value below.  It gives good
+ * results and I've not seen any dropped events.
+ *
+ * With a latency of 0.1, I was seeing lots of dropped events during
+ * the "touch 100000" files test within t/perf/p7519, but with a
+ * latency of 0.001 I did not see any dropped events.  So I'm going
+ * to assume that this is the "correct" value.
+ *
+ * https://developer.apple.com/documentation/coreservices/1443980-fseventstreamcreate
+ */
 
 int fsm_listen__ctor(struct fsmonitor_daemon_state *state)
 {
+	FSEventStreamCreateFlags flags = kFSEventStreamCreateFlagNoDefer |
+		kFSEventStreamCreateFlagWatchRoot |
+		kFSEventStreamCreateFlagFileEvents;
+	FSEventStreamContext ctx = {
+		0,
+		state,
+		NULL,
+		NULL,
+		NULL
+	};
+	struct fsmonitor_daemon_backend_data *data;
+	const void *dir_array[2];
+
+	CALLOC_ARRAY(data, 1);
+	state->backend_data = data;
+
+	data->cfsr_worktree_path = CFStringCreateWithCString(
+		NULL, state->path_worktree_watch.buf, kCFStringEncodingUTF8);
+	dir_array[data->nr_paths_watching++] = data->cfsr_worktree_path;
+
+	if (state->nr_paths_watching > 1) {
+		data->cfsr_gitdir_path = CFStringCreateWithCString(
+			NULL, state->path_gitdir_watch.buf,
+			kCFStringEncodingUTF8);
+		dir_array[data->nr_paths_watching++] = data->cfsr_gitdir_path;
+	}
+
+	data->cfar_paths_to_watch = CFArrayCreate(NULL, dir_array,
+						  data->nr_paths_watching,
+						  NULL);
+	data->stream = FSEventStreamCreate(NULL, fsevent_callback, &ctx,
+					   data->cfar_paths_to_watch,
+					   kFSEventStreamEventIdSinceNow,
+					   0.001, flags);
+	if (data->stream == NULL)
+		goto failed;
+
+	/*
+	 * `data->rl` needs to be set inside the listener thread.
+	 */
+
+	return 0;
+
+failed:
+	error(_("Unable to create FSEventStream."));
+
+	FREE_AND_NULL(state->backend_data);
 	return -1;
 }
 
 void fsm_listen__dtor(struct fsmonitor_daemon_state *state)
 {
+	struct fsmonitor_daemon_backend_data *data;
+
+	if (!state || !state->backend_data)
+		return;
+
+	data = state->backend_data;
+
+	if (data->stream) {
+		if (data->stream_started)
+			FSEventStreamStop(data->stream);
+		if (data->stream_scheduled)
+			FSEventStreamInvalidate(data->stream);
+		FSEventStreamRelease(data->stream);
+	}
+
+	FREE_AND_NULL(state->backend_data);
 }
 
 void fsm_listen__stop_async(struct fsmonitor_daemon_state *state)
 {
+	struct fsmonitor_daemon_backend_data *data;
+
+	data = state->backend_data;
+	data->shutdown_style = SHUTDOWN_EVENT;
+
+	CFRunLoopStop(data->rl);
 }
 
 void fsm_listen__loop(struct fsmonitor_daemon_state *state)
 {
+	struct fsmonitor_daemon_backend_data *data;
+
+	data = state->backend_data;
+
+	data->rl = CFRunLoopGetCurrent();
+
+	FSEventStreamScheduleWithRunLoop(data->stream, data->rl, kCFRunLoopDefaultMode);
+	data->stream_scheduled = 1;
+
+	if (!FSEventStreamStart(data->stream)) {
+		error(_("Failed to start the FSEventStream"));
+		goto force_error_stop_without_loop;
+	}
+	data->stream_started = 1;
+
+	CFRunLoopRun();
+
+	switch (data->shutdown_style) {
+	case FORCE_ERROR_STOP:
+		state->error_code = -1;
+		/* fall thru */
+	case FORCE_SHUTDOWN:
+		ipc_server_stop_async(state->ipc_server_data);
+		/* fall thru */
+	case SHUTDOWN_EVENT:
+	default:
+		break;
+	}
+	return;
+
+force_error_stop_without_loop:
+	state->error_code = -1;
+	ipc_server_stop_async(state->ipc_server_data);
+	return;
 }
-- 
gitgitgadget

[PATCH v9 18/30] fsmonitor--daemon: implement handle_client callback

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach fsmonitor--daemon to respond to IPC requests from client
Git processes and respond with a list of modified pathnames
relative to the provided token.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 builtin/fsmonitor--daemon.c | 311 +++++++++++++++++++++++++++++++++++-
 1 file changed, 309 insertions(+), 2 deletions(-)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index 69312119b07..eafaafb45b1 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -7,6 +7,7 @@
 #include "fsmonitor--daemon.h"
 #include "simple-ipc.h"
 #include "khash.h"
+#include "pkt-line.h"
 
 static const char * const builtin_fsmonitor__daemon_usage[] = {
 	N_("git fsmonitor--daemon start [<options>]"),
@@ -364,6 +365,310 @@ void fsmonitor_force_resync(struct fsmonitor_daemon_state *state)
 	pthread_mutex_unlock(&state->main_lock);
 }
 
+/*
+ * Format an opaque token string to send to the client.
+ */
+static void with_lock__format_response_token(
+	struct strbuf *response_token,
+	const struct strbuf *response_token_id,
+	const struct fsmonitor_batch *batch)
+{
+	/* assert current thread holding state->main_lock */
+
+	strbuf_reset(response_token);
+	strbuf_addf(response_token, "builtin:%s:%"PRIu64,
+		    response_token_id->buf, batch->batch_seq_nr);
+}
+
+/*
+ * Parse an opaque token from the client.
+ * Returns -1 on error.
+ */
+static int fsmonitor_parse_client_token(const char *buf_token,
+					struct strbuf *requested_token_id,
+					uint64_t *seq_nr)
+{
+	const char *p;
+	char *p_end;
+
+	strbuf_reset(requested_token_id);
+	*seq_nr = 0;
+
+	if (!skip_prefix(buf_token, "builtin:", &p))
+		return -1;
+
+	while (*p && *p != ':')
+		strbuf_addch(requested_token_id, *p++);
+	if (!*p++)
+		return -1;
+
+	*seq_nr = (uint64_t)strtoumax(p, &p_end, 10);
+	if (*p_end)
+		return -1;
+
+	return 0;
+}
+
+KHASH_INIT(str, const char *, int, 0, kh_str_hash_func, kh_str_hash_equal)
+
+static int do_handle_client(struct fsmonitor_daemon_state *state,
+			    const char *command,
+			    ipc_server_reply_cb *reply,
+			    struct ipc_server_reply_data *reply_data)
+{
+	struct fsmonitor_token_data *token_data = NULL;
+	struct strbuf response_token = STRBUF_INIT;
+	struct strbuf requested_token_id = STRBUF_INIT;
+	struct strbuf payload = STRBUF_INIT;
+	uint64_t requested_oldest_seq_nr = 0;
+	uint64_t total_response_len = 0;
+	const char *p;
+	const struct fsmonitor_batch *batch_head;
+	const struct fsmonitor_batch *batch;
+	intmax_t count = 0, duplicates = 0;
+	kh_str_t *shown;
+	int hash_ret;
+	int do_trivial = 0;
+	int do_flush = 0;
+
+	/*
+	 * We expect `command` to be of the form:
+	 *
+	 * <command> := quit NUL
+	 *            | flush NUL
+	 *            | <V1-time-since-epoch-ns> NUL
+	 *            | <V2-opaque-fsmonitor-token> NUL
+	 */
+
+	if (!strcmp(command, "quit")) {
+		/*
+		 * A client has requested over the socket/pipe that the
+		 * daemon shutdown.
+		 *
+		 * Tell the IPC thread pool to shutdown (which completes
+		 * the await in the main thread (which can stop the
+		 * fsmonitor listener thread)).
+		 *
+		 * There is no reply to the client.
+		 */
+		return SIMPLE_IPC_QUIT;
+
+	} else if (!strcmp(command, "flush")) {
+		/*
+		 * Flush all of our cached data and generate a new token
+		 * just like if we lost sync with the filesystem.
+		 *
+		 * Then send a trivial response using the new token.
+		 */
+		do_flush = 1;
+		do_trivial = 1;
+
+	} else if (!skip_prefix(command, "builtin:", &p)) {
+		/* assume V1 timestamp or garbage */
+
+		char *p_end;
+
+		strtoumax(command, &p_end, 10);
+		trace_printf_key(&trace_fsmonitor,
+				 ((*p_end) ?
+				  "fsmonitor: invalid command line '%s'" :
+				  "fsmonitor: unsupported V1 protocol '%s'"),
+				 command);
+		do_trivial = 1;
+
+	} else {
+		/* We have "builtin:*" */
+		if (fsmonitor_parse_client_token(command, &requested_token_id,
+						 &requested_oldest_seq_nr)) {
+			trace_printf_key(&trace_fsmonitor,
+					 "fsmonitor: invalid V2 protocol token '%s'",
+					 command);
+			do_trivial = 1;
+
+		} else {
+			/*
+			 * We have a V2 valid token:
+			 *     "builtin:<token_id>:<seq_nr>"
+			 */
+		}
+	}
+
+	pthread_mutex_lock(&state->main_lock);
+
+	if (!state->current_token_data)
+		BUG("fsmonitor state does not have a current token");
+
+	if (do_flush)
+		with_lock__do_force_resync(state);
+
+	/*
+	 * We mark the current head of the batch list as "pinned" so
+	 * that the listener thread will treat this item as read-only
+	 * (and prevent any more paths from being added to it) from
+	 * now on.
+	 */
+	token_data = state->current_token_data;
+	batch_head = token_data->batch_head;
+	((struct fsmonitor_batch *)batch_head)->pinned_time = time(NULL);
+
+	/*
+	 * FSMonitor Protocol V2 requires that we send a response header
+	 * with a "new current token" and then all of the paths that changed
+	 * since the "requested token".  We send the seq_nr of the just-pinned
+	 * head batch so that future requests from a client will be relative
+	 * to it.
+	 */
+	with_lock__format_response_token(&response_token,
+					 &token_data->token_id, batch_head);
+
+	reply(reply_data, response_token.buf, response_token.len + 1);
+	total_response_len += response_token.len + 1;
+
+	trace2_data_string("fsmonitor", the_repository, "response/token",
+			   response_token.buf);
+	trace_printf_key(&trace_fsmonitor, "response token: %s",
+			 response_token.buf);
+
+	if (!do_trivial) {
+		if (strcmp(requested_token_id.buf, token_data->token_id.buf)) {
+			/*
+			 * The client last spoke to a different daemon
+			 * instance -OR- the daemon had to resync with
+			 * the filesystem (and lost events), so reject.
+			 */
+			trace2_data_string("fsmonitor", the_repository,
+					   "response/token", "different");
+			do_trivial = 1;
+
+		} else if (requested_oldest_seq_nr <
+			   token_data->batch_tail->batch_seq_nr) {
+			/*
+			 * The client wants older events than we have for
+			 * this token_id.  This means that the end of our
+			 * batch list was truncated and we cannot give the
+			 * client a complete snapshot relative to their
+			 * request.
+			 */
+			trace_printf_key(&trace_fsmonitor,
+					 "client requested truncated data");
+			do_trivial = 1;
+		}
+	}
+
+	if (do_trivial) {
+		pthread_mutex_unlock(&state->main_lock);
+
+		reply(reply_data, "/", 2);
+
+		trace2_data_intmax("fsmonitor", the_repository,
+				   "response/trivial", 1);
+
+		goto cleanup;
+	}
+
+	/*
+	 * We're going to hold onto a pointer to the current
+	 * token-data while we walk the list of batches of files.
+	 * During this time, we will NOT be under the lock.
+	 * So we ref-count it.
+	 *
+	 * This allows the listener thread to continue prepending
+	 * new batches of items to the token-data (which we'll ignore).
+	 *
+	 * AND it allows the listener thread to do a token-reset
+	 * (and install a new `current_token_data`).
+	 */
+	token_data->client_ref_count++;
+
+	pthread_mutex_unlock(&state->main_lock);
+
+	/*
+	 * The client request is relative to the token that they sent,
+	 * so walk the batch list backwards from the current head back
+	 * to the batch (sequence number) they named.
+	 *
+	 * We use khash to de-dup the list of pathnames.
+	 *
+	 * NEEDSWORK: each batch contains a list of interned strings,
+	 * so we only need to do pointer comparisons here to build the
+	 * hash table.  Currently, we're still comparing the string
+	 * values.
+	 */
+	shown = kh_init_str();
+	for (batch = batch_head;
+	     batch && batch->batch_seq_nr > requested_oldest_seq_nr;
+	     batch = batch->next) {
+		size_t k;
+
+		for (k = 0; k < batch->nr; k++) {
+			const char *s = batch->interned_paths[k];
+			size_t s_len;
+
+			if (kh_get_str(shown, s) != kh_end(shown))
+				duplicates++;
+			else {
+				kh_put_str(shown, s, &hash_ret);
+
+				trace_printf_key(&trace_fsmonitor,
+						 "send[%"PRIuMAX"]: %s",
+						 count, s);
+
+				/* Each path gets written with a trailing NUL */
+				s_len = strlen(s) + 1;
+
+				if (payload.len + s_len >=
+				    LARGE_PACKET_DATA_MAX) {
+					reply(reply_data, payload.buf,
+					      payload.len);
+					total_response_len += payload.len;
+					strbuf_reset(&payload);
+				}
+
+				strbuf_add(&payload, s, s_len);
+				count++;
+			}
+		}
+	}
+
+	if (payload.len) {
+		reply(reply_data, payload.buf, payload.len);
+		total_response_len += payload.len;
+	}
+
+	kh_release_str(shown);
+
+	pthread_mutex_lock(&state->main_lock);
+
+	if (token_data->client_ref_count > 0)
+		token_data->client_ref_count--;
+
+	if (token_data->client_ref_count == 0) {
+		if (token_data != state->current_token_data) {
+			/*
+			 * The listener thread did a token-reset while we were
+			 * walking the batch list.  Therefore, this token is
+			 * stale and can be discarded completely.  If we are
+			 * the last reader thread using this token, we own
+			 * that work.
+			 */
+			fsmonitor_free_token_data(token_data);
+		}
+	}
+
+	pthread_mutex_unlock(&state->main_lock);
+
+	trace2_data_intmax("fsmonitor", the_repository, "response/length", total_response_len);
+	trace2_data_intmax("fsmonitor", the_repository, "response/count/files", count);
+	trace2_data_intmax("fsmonitor", the_repository, "response/count/duplicates", duplicates);
+
+cleanup:
+	strbuf_release(&response_token);
+	strbuf_release(&requested_token_id);
+	strbuf_release(&payload);
+
+	return 0;
+}
+
 static ipc_server_application_cb handle_client;
 
 static int handle_client(void *data,
@@ -371,7 +676,7 @@ static int handle_client(void *data,
 			 ipc_server_reply_cb *reply,
 			 struct ipc_server_reply_data *reply_data)
 {
-	/* struct fsmonitor_daemon_state *state = data; */
+	struct fsmonitor_daemon_state *state = data;
 	int result;
 
 	/*
@@ -382,10 +687,12 @@ static int handle_client(void *data,
 	if (command_len != strlen(command))
 		BUG("FSMonitor assumes text messages");
 
+	trace_printf_key(&trace_fsmonitor, "requested token: %s", command);
+
 	trace2_region_enter("fsmonitor", "handle_client", the_repository);
 	trace2_data_string("fsmonitor", the_repository, "request", command);
 
-	result = 0; /* TODO Do something here. */
+	result = do_handle_client(state, command, reply, reply_data);
 
 	trace2_region_leave("fsmonitor", "handle_client", the_repository);
 
-- 
gitgitgadget

[PATCH v9 19/30] help: include fsmonitor--daemon feature flag in version info

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Add the "feature: fsmonitor--daemon" message to the output of
`git version --build-options`.

The builtin FSMonitor is only available on certain platforms and
even then only when certain Makefile flags are enabled, so print
a message in the verbose version output when it is available.

This can be used by test scripts for prereq testing.  Granted, tests
could just try `git fsmonitor--daemon status` and look for a 128 exit
code or grep for a "not supported" message on stderr, but these
methods are rather obscure.

The main advantage is that the feature message will automatically
appear in bug reports and other support requests.

This concept was also used during the development of Scalar for
similar reasons.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 help.c        | 4 ++++
 t/test-lib.sh | 7 +++++++
 2 files changed, 11 insertions(+)

diff --git a/help.c b/help.c
index 71444906ddf..9112a51e84b 100644
--- a/help.c
+++ b/help.c
@@ -12,6 +12,7 @@
 #include "refs.h"
 #include "parse-options.h"
 #include "prompt.h"
+#include "fsmonitor-ipc.h"
 
 struct category_description {
 	uint32_t category;
@@ -695,6 +696,9 @@ void get_version_info(struct strbuf *buf, int show_build_options)
 		strbuf_addf(buf, "sizeof-size_t: %d\n", (int)sizeof(size_t));
 		strbuf_addf(buf, "shell-path: %s\n", SHELL_PATH);
 		/* NEEDSWORK: also save and output GIT-BUILD_OPTIONS? */
+
+		if (fsmonitor_ipc__is_supported())
+			strbuf_addstr(buf, "feature: fsmonitor--daemon\n");
 	}
 }
 
diff --git a/t/test-lib.sh b/t/test-lib.sh
index e4716b0b867..5d819c1bc11 100644
--- a/t/test-lib.sh
+++ b/t/test-lib.sh
@@ -1799,3 +1799,10 @@ test_lazy_prereq SHA1 '
 # Tests that verify the scheduler integration must set this locally
 # to avoid errors.
 GIT_TEST_MAINT_SCHEDULER="none:exit 1"
+
+# Does this platform support `git fsmonitor--daemon`
+#
+test_lazy_prereq FSMONITOR_DAEMON '
+	git version --build-options >output &&
+	grep "feature: fsmonitor--daemon" output
+'
-- 
gitgitgadget

[PATCH v9 20/30] t/helper/fsmonitor-client: create IPC client to talk to FSMonitor Daemon

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Create an IPC client to send query and flush commands to the daemon.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 Makefile                         |   1 +
 t/helper/test-fsmonitor-client.c | 116 +++++++++++++++++++++++++++++++
 t/helper/test-tool.c             |   1 +
 t/helper/test-tool.h             |   1 +
 4 files changed, 119 insertions(+)
 create mode 100644 t/helper/test-fsmonitor-client.c

diff --git a/Makefile b/Makefile
index 26567d4f772..daa21bed6c3 100644
--- a/Makefile
+++ b/Makefile
@@ -716,6 +716,7 @@ TEST_BUILTINS_OBJS += test-dump-split-index.o
 TEST_BUILTINS_OBJS += test-dump-untracked-cache.o
 TEST_BUILTINS_OBJS += test-example-decorate.o
 TEST_BUILTINS_OBJS += test-fast-rebase.o
+TEST_BUILTINS_OBJS += test-fsmonitor-client.o
 TEST_BUILTINS_OBJS += test-genrandom.o
 TEST_BUILTINS_OBJS += test-genzeros.o
 TEST_BUILTINS_OBJS += test-getcwd.o
diff --git a/t/helper/test-fsmonitor-client.c b/t/helper/test-fsmonitor-client.c
new file mode 100644
index 00000000000..3062c8a3c2b
--- /dev/null
+++ b/t/helper/test-fsmonitor-client.c
@@ -0,0 +1,116 @@
+/*
+ * test-fsmonitor-client.c: client code to send commands/requests to
+ * a `git fsmonitor--daemon` daemon.
+ */
+
+#include "test-tool.h"
+#include "cache.h"
+#include "parse-options.h"
+#include "fsmonitor-ipc.h"
+
+#ifndef HAVE_FSMONITOR_DAEMON_BACKEND
+int cmd__fsmonitor_client(int argc, const char **argv)
+{
+	die("fsmonitor--daemon not available on this platform");
+}
+#else
+
+/*
+ * Read the `.git/index` to get the last token written to the
+ * FSMonitor Index Extension.
+ */
+static const char *get_token_from_index(void)
+{
+	struct index_state *istate = the_repository->index;
+
+	if (do_read_index(istate, the_repository->index_file, 0) < 0)
+		die("unable to read index file");
+	if (!istate->fsmonitor_last_update)
+		die("index file does not have fsmonitor extension");
+
+	return istate->fsmonitor_last_update;
+}
+
+/*
+ * Send an IPC query to a `git-fsmonitor--daemon` daemon and
+ * ask for the changes since the given token or from the last
+ * token in the index extension.
+ *
+ * This will implicitly start a daemon process if necessary.  The
+ * daemon process will persist after we exit.
+ */
+static int do_send_query(const char *token)
+{
+	struct strbuf answer = STRBUF_INIT;
+	int ret;
+
+	if (!token || !*token)
+		token = get_token_from_index();
+
+	ret = fsmonitor_ipc__send_query(token, &answer);
+	if (ret < 0)
+		die("could not query fsmonitor--daemon");
+
+	write_in_full(1, answer.buf, answer.len);
+	strbuf_release(&answer);
+
+	return 0;
+}
+
+/*
+ * Send a "flush" command to the `git-fsmonitor--daemon` (if running)
+ * and tell it to flush its cache.
+ *
+ * This feature is primarily used by the test suite to simulate a loss of
+ * sync with the filesystem where we miss kernel events.
+ */
+static int do_send_flush(void)
+{
+	struct strbuf answer = STRBUF_INIT;
+	int ret;
+
+	ret = fsmonitor_ipc__send_command("flush", &answer);
+	if (ret)
+		return ret;
+
+	write_in_full(1, answer.buf, answer.len);
+	strbuf_release(&answer);
+
+	return 0;
+}
+
+int cmd__fsmonitor_client(int argc, const char **argv)
+{
+	const char *subcmd;
+	const char *token = NULL;
+
+	const char * const fsmonitor_client_usage[] = {
+		"test-tool fsmonitor-client query [<token>]",
+		"test-tool fsmonitor-client flush",
+		NULL,
+	};
+
+	struct option options[] = {
+		OPT_STRING(0, "token", &token, "token",
+			   "command token to send to the server"),
+		OPT_END()
+	};
+
+	argc = parse_options(argc, argv, NULL, options, fsmonitor_client_usage, 0);
+
+	if (argc != 1)
+		usage_with_options(fsmonitor_client_usage, options);
+
+	subcmd = argv[0];
+
+	setup_git_directory();
+
+	if (!strcmp(subcmd, "query"))
+		return !!do_send_query(token);
+
+	if (!strcmp(subcmd, "flush"))
+		return !!do_send_flush();
+
+	die("Unhandled subcommand: '%s'", subcmd);
+}
+#endif
diff --git a/t/helper/test-tool.c b/t/helper/test-tool.c
index e6ec69cf326..0424f7adf5d 100644
--- a/t/helper/test-tool.c
+++ b/t/helper/test-tool.c
@@ -32,6 +32,7 @@ static struct test_cmd cmds[] = {
 	{ "dump-untracked-cache", cmd__dump_untracked_cache },
 	{ "example-decorate", cmd__example_decorate },
 	{ "fast-rebase", cmd__fast_rebase },
+	{ "fsmonitor-client", cmd__fsmonitor_client },
 	{ "genrandom", cmd__genrandom },
 	{ "genzeros", cmd__genzeros },
 	{ "getcwd", cmd__getcwd },
diff --git a/t/helper/test-tool.h b/t/helper/test-tool.h
index 20756eefdda..c876e8246fb 100644
--- a/t/helper/test-tool.h
+++ b/t/helper/test-tool.h
@@ -23,6 +23,7 @@ int cmd__dump_untracked_cache(int argc, const char **argv);
 int cmd__dump_reftable(int argc, const char **argv);
 int cmd__example_decorate(int argc, const char **argv);
 int cmd__fast_rebase(int argc, const char **argv);
+int cmd__fsmonitor_client(int argc, const char **argv);
 int cmd__genrandom(int argc, const char **argv);
 int cmd__genzeros(int argc, const char **argv);
 int cmd__getcwd(int argc, const char **argv);
-- 
gitgitgadget

[PATCH v9 21/30] t7527: create test for fsmonitor--daemon

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 t/t7527-builtin-fsmonitor.sh | 494 +++++++++++++++++++++++++++++++++++
 1 file changed, 494 insertions(+)
 create mode 100755 t/t7527-builtin-fsmonitor.sh

diff --git a/t/t7527-builtin-fsmonitor.sh b/t/t7527-builtin-fsmonitor.sh
new file mode 100755
index 00000000000..062e01c0dfc
--- /dev/null
+++ b/t/t7527-builtin-fsmonitor.sh
@@ -0,0 +1,494 @@
+#!/bin/sh
+
+test_description='built-in file system watcher'
+
+. ./test-lib.sh
+
+if ! test_have_prereq FSMONITOR_DAEMON
+then
+	skip_all="fsmonitor--daemon is not supported on this platform"
+	test_done
+fi
+
+stop_daemon_delete_repo () {
+	r=$1 &&
+	test_might_fail git -C $r fsmonitor--daemon stop &&
+	rm -rf $1
+}
+
+start_daemon () {
+	r= tf= t2= tk= &&
+
+	while test "$#" -ne 0
+	do
+		case "$1" in
+		-C)
+			r="-C ${2?}"
+			shift
+			;;
+		--tf)
+			tf="${2?}"
+			shift
+			;;
+		--t2)
+			t2="${2?}"
+			shift
+			;;
+		--tk)
+			tk="${2?}"
+			shift
+			;;
+		-*)
+			BUG "error: unknown option: '$1'"
+			;;
+		*)
+			BUG "error: unbound argument: '$1'"
+			;;
+		esac
+		shift
+	done &&
+
+	(
+		if test -n "$tf"
+		then
+			GIT_TRACE_FSMONITOR="$tf"
+			export GIT_TRACE_FSMONITOR
+		fi &&
+
+		if test -n "$t2"
+		then
+			GIT_TRACE2_PERF="$t2"
+			export GIT_TRACE2_PERF
+		fi &&
+
+		if test -n "$tk"
+		then
+			GIT_TEST_FSMONITOR_TOKEN="$tk"
+			export GIT_TEST_FSMONITOR_TOKEN
+		fi &&
+
+		git $r fsmonitor--daemon start &&
+		git $r fsmonitor--daemon status
+	)
+}
+
+# Is a Trace2 data event present with the given catetory and key?
+# We do not care what the value is.
+#
+have_t2_data_event () {
+	c=$1 &&
+	k=$2 &&
+
+	grep -e '"event":"data".*"category":"'"$c"'".*"key":"'"$k"'"'
+}
+
+test_expect_success 'explicit daemon start and stop' '
+	test_when_finished "stop_daemon_delete_repo test_explicit" &&
+
+	git init test_explicit &&
+	start_daemon -C test_explicit &&
+
+	git -C test_explicit fsmonitor--daemon stop &&
+	test_must_fail git -C test_explicit fsmonitor--daemon status
+'
+
+test_expect_success 'implicit daemon start' '
+	test_when_finished "stop_daemon_delete_repo test_implicit" &&
+
+	git init test_implicit &&
+	test_must_fail git -C test_implicit fsmonitor--daemon status &&
+
+	# query will implicitly start the daemon.
+	#
+	# for test-script simplicity, we send a V1 timestamp rather than
+	# a V2 token.  either way, the daemon response to any query contains
+	# a new V2 token.  (the daemon may complain that we sent a V1 request,
+	# but this test case is only concerned with whether the daemon was
+	# implicitly started.)
+
+	GIT_TRACE2_EVENT="$PWD/.git/trace" \
+		test-tool -C test_implicit fsmonitor-client query --token 0 >actual &&
+	nul_to_q <actual >actual.filtered &&
+	grep "builtin:" actual.filtered &&
+
+	# confirm that a daemon was started in the background.
+	#
+	# since the mechanism for starting the background daemon is platform
+	# dependent, just confirm that the foreground command received a
+	# response from the daemon.
+
+	have_t2_data_event fsm_client query/response-length <.git/trace &&
+
+	git -C test_implicit fsmonitor--daemon status &&
+	git -C test_implicit fsmonitor--daemon stop &&
+	test_must_fail git -C test_implicit fsmonitor--daemon status
+'
+
+test_expect_success 'implicit daemon stop (delete .git)' '
+	test_when_finished "stop_daemon_delete_repo test_implicit_1" &&
+
+	git init test_implicit_1 &&
+
+	start_daemon -C test_implicit_1 &&
+
+	# deleting the .git directory will implicitly stop the daemon.
+	rm -rf test_implicit_1/.git &&
+
+	# [1] Create an empty .git directory so that the following Git
+	#     command will stay relative to the `-C` directory.
+	#
+	#     Without this, the Git command will override the requested
+	#     -C argument and crawl out to the containing Git source tree.
+	#     This would make the test result dependent upon whether we
+	#     were using fsmonitor on our development worktree.
+	#
+	sleep 1 &&
+	mkdir test_implicit_1/.git &&
+
+	test_must_fail git -C test_implicit_1 fsmonitor--daemon status
+'
+
+test_expect_success 'implicit daemon stop (rename .git)' '
+	test_when_finished "stop_daemon_delete_repo test_implicit_2" &&
+
+	git init test_implicit_2 &&
+
+	start_daemon -C test_implicit_2 &&
+
+	# renaming the .git directory will implicitly stop the daemon.
+	mv test_implicit_2/.git test_implicit_2/.xxx &&
+
+	# See [1] above.
+	#
+	sleep 1 &&
+	mkdir test_implicit_2/.git &&
+
+	test_must_fail git -C test_implicit_2 fsmonitor--daemon status
+'
+
+test_expect_success 'cannot start multiple daemons' '
+	test_when_finished "stop_daemon_delete_repo test_multiple" &&
+
+	git init test_multiple &&
+
+	start_daemon -C test_multiple &&
+
+	test_must_fail git -C test_multiple fsmonitor--daemon start 2>actual &&
+	grep "fsmonitor--daemon is already running" actual &&
+
+	git -C test_multiple fsmonitor--daemon stop &&
+	test_must_fail git -C test_multiple fsmonitor--daemon status
+'
+
+# These tests use the main repo in the trash directory
+
+test_expect_success 'setup' '
+	>tracked &&
+	>modified &&
+	>delete &&
+	>rename &&
+	mkdir dir1 &&
+	>dir1/tracked &&
+	>dir1/modified &&
+	>dir1/delete &&
+	>dir1/rename &&
+	mkdir dir2 &&
+	>dir2/tracked &&
+	>dir2/modified &&
+	>dir2/delete &&
+	>dir2/rename &&
+	mkdir dirtorename &&
+	>dirtorename/a &&
+	>dirtorename/b &&
+
+	cat >.gitignore <<-\EOF &&
+	.gitignore
+	expect*
+	actual*
+	EOF
+
+	git -c core.fsmonitor=false add . &&
+	test_tick &&
+	git -c core.fsmonitor=false commit -m initial &&
+
+	git config core.fsmonitor true
+'
+
+# The test already explicitly stopped (or tried to stop) the daemon.
+# This is here in case something else fails first.
+#
+redundant_stop_daemon () {
+	test_might_fail git fsmonitor--daemon stop
+}
+
+test_expect_success 'update-index implicitly starts daemon' '
+	test_when_finished redundant_stop_daemon &&
+
+	test_must_fail git fsmonitor--daemon status &&
+
+	GIT_TRACE2_EVENT="$PWD/.git/trace_implicit_1" \
+		git update-index --fsmonitor &&
+
+	git fsmonitor--daemon status &&
+	test_might_fail git fsmonitor--daemon stop &&
+
+	# Confirm that the trace2 log contains a record of the
+	# daemon starting.
+	test_subcommand git fsmonitor--daemon start <.git/trace_implicit_1
+'
+
+test_expect_success 'status implicitly starts daemon' '
+	test_when_finished redundant_stop_daemon &&
+
+	test_must_fail git fsmonitor--daemon status &&
+
+	GIT_TRACE2_EVENT="$PWD/.git/trace_implicit_2" \
+		git status >actual &&
+
+	git fsmonitor--daemon status &&
+	test_might_fail git fsmonitor--daemon stop &&
+
+	# Confirm that the trace2 log contains a record of the
+	# daemon starting.
+	test_subcommand git fsmonitor--daemon start <.git/trace_implicit_2
+'
+
+edit_files () {
+	echo 1 >modified &&
+	echo 2 >dir1/modified &&
+	echo 3 >dir2/modified &&
+	>dir1/untracked
+}
+
+delete_files () {
+	rm -f delete &&
+	rm -f dir1/delete &&
+	rm -f dir2/delete
+}
+
+create_files () {
+	echo 1 >new &&
+	echo 2 >dir1/new &&
+	echo 3 >dir2/new
+}
+
+rename_files () {
+	mv rename renamed &&
+	mv dir1/rename dir1/renamed &&
+	mv dir2/rename dir2/renamed
+}
+
+file_to_directory () {
+	rm -f delete &&
+	mkdir delete &&
+	echo 1 >delete/new
+}
+
+directory_to_file () {
+	rm -rf dir1 &&
+	echo 1 >dir1
+}
+
+# The next few test cases confirm that our fsmonitor daemon sees each type
+# of OS filesystem notification that we care about.  At this layer we just
+# ensure we are getting the OS notifications and do not try to confirm what
+# is reported by `git status`.
+#
+# We run a simple query after modifying the filesystem just to introduce
+# a bit of a delay so that the trace logging from the daemon has time to
+# get flushed to disk.
+#
+# We `reset` and `clean` at the bottom of each test (and before stopping the
+# daemon) because these commands might implicitly restart the daemon.
+
+clean_up_repo_and_stop_daemon () {
+	git reset --hard HEAD &&
+	git clean -fd &&
+	test_might_fail git fsmonitor--daemon stop &&
+	rm -f .git/trace
+}
+
+test_expect_success 'edit some files' '
+	test_when_finished clean_up_repo_and_stop_daemon &&
+
+	start_daemon --tf "$PWD/.git/trace" &&
+
+	edit_files &&
+
+	test-tool fsmonitor-client query --token 0 &&
+
+	grep "^event: dir1/modified$"  .git/trace &&
+	grep "^event: dir2/modified$"  .git/trace &&
+	grep "^event: modified$"       .git/trace &&
+	grep "^event: dir1/untracked$" .git/trace
+'
+
+test_expect_success 'create some files' '
+	test_when_finished clean_up_repo_and_stop_daemon &&
+
+	start_daemon --tf "$PWD/.git/trace" &&
+
+	create_files &&
+
+	test-tool fsmonitor-client query --token 0 &&
+
+	grep "^event: dir1/new$" .git/trace &&
+	grep "^event: dir2/new$" .git/trace &&
+	grep "^event: new$"      .git/trace
+'
+
+test_expect_success 'delete some files' '
+	test_when_finished clean_up_repo_and_stop_daemon &&
+
+	start_daemon --tf "$PWD/.git/trace" &&
+
+	delete_files &&
+
+	test-tool fsmonitor-client query --token 0 &&
+
+	grep "^event: dir1/delete$" .git/trace &&
+	grep "^event: dir2/delete$" .git/trace &&
+	grep "^event: delete$"      .git/trace
+'
+
+test_expect_success 'rename some files' '
+	test_when_finished clean_up_repo_and_stop_daemon &&
+
+	start_daemon --tf "$PWD/.git/trace" &&
+
+	rename_files &&
+
+	test-tool fsmonitor-client query --token 0 &&
+
+	grep "^event: dir1/rename$"  .git/trace &&
+	grep "^event: dir2/rename$"  .git/trace &&
+	grep "^event: rename$"       .git/trace &&
+	grep "^event: dir1/renamed$" .git/trace &&
+	grep "^event: dir2/renamed$" .git/trace &&
+	grep "^event: renamed$"      .git/trace
+'
+
+test_expect_success 'rename directory' '
+	test_when_finished clean_up_repo_and_stop_daemon &&
+
+	start_daemon --tf "$PWD/.git/trace" &&
+
+	mv dirtorename dirrenamed &&
+
+	test-tool fsmonitor-client query --token 0 &&
+
+	grep "^event: dirtorename/*$" .git/trace &&
+	grep "^event: dirrenamed/*$"  .git/trace
+'
+
+test_expect_success 'file changes to directory' '
+	test_when_finished clean_up_repo_and_stop_daemon &&
+
+	start_daemon --tf "$PWD/.git/trace" &&
+
+	file_to_directory &&
+
+	test-tool fsmonitor-client query --token 0 &&
+
+	grep "^event: delete$"     .git/trace &&
+	grep "^event: delete/new$" .git/trace
+'
+
+test_expect_success 'directory changes to a file' '
+	test_when_finished clean_up_repo_and_stop_daemon &&
+
+	start_daemon --tf "$PWD/.git/trace" &&
+
+	directory_to_file &&
+
+	test-tool fsmonitor-client query --token 0 &&
+
+	grep "^event: dir1$" .git/trace
+'
+
+# The next few test cases exercise the token-resync code.  When filesystem
+# drops events (because of filesystem velocity or because the daemon isn't
+# polling fast enough), we need to discard the cached data (relative to the
+# current token) and start collecting events under a new token.
+#
+# the 'test-tool fsmonitor-client flush' command can be used to send a
+# "flush" message to a running daemon and ask it to do a flush/resync.
+
+test_expect_success 'flush cached data' '
+	test_when_finished "stop_daemon_delete_repo test_flush" &&
+
+	git init test_flush &&
+
+	start_daemon -C test_flush --tf "$PWD/.git/trace_daemon" --tk true &&
+
+	# The daemon should have an initial token with no events in _0 and
+	# then a few (probably platform-specific number of) events in _1.
+	# These should both have the same <token_id>.
+
+	test-tool -C test_flush fsmonitor-client query --token "builtin:test_00000001:0" >actual_0 &&
+	nul_to_q <actual_0 >actual_q0 &&
+
+	>test_flush/file_1 &&
+	>test_flush/file_2 &&
+
+	test-tool -C test_flush fsmonitor-client query --token "builtin:test_00000001:0" >actual_1 &&
+	nul_to_q <actual_1 >actual_q1 &&
+
+	grep "file_1" actual_q1 &&
+
+	# Force a flush.  This will change the <token_id>, reset the <seq_nr>, and
+	# flush the file data.  Then create some events and ensure that the file
+	# again appears in the cache.  It should have the new <token_id>.
+
+	test-tool -C test_flush fsmonitor-client flush >flush_0 &&
+	nul_to_q <flush_0 >flush_q0 &&
+	grep "^builtin:test_00000002:0Q/Q$" flush_q0 &&
+
+	test-tool -C test_flush fsmonitor-client query --token "builtin:test_00000002:0" >actual_2 &&
+	nul_to_q <actual_2 >actual_q2 &&
+
+	grep "^builtin:test_00000002:0Q$" actual_q2 &&
+
+	>test_flush/file_3 &&
+
+	test-tool -C test_flush fsmonitor-client query --token "builtin:test_00000002:0" >actual_3 &&
+	nul_to_q <actual_3 >actual_q3 &&
+
+	grep "file_3" actual_q3
+'
+
+# The next few test cases create repos where the .git directory is NOT
+# inside the one of the working directory.  That is, where .git is a file
+# that points to a directory elsewhere.  This happens for submodules and
+# non-primary worktrees.
+
+test_expect_success 'setup worktree base' '
+	git init wt-base &&
+	echo 1 >wt-base/file1 &&
+	git -C wt-base add file1 &&
+	git -C wt-base commit -m "c1"
+'
+
+test_expect_success 'worktree with .git file' '
+	git -C wt-base worktree add ../wt-secondary &&
+
+	start_daemon -C wt-secondary \
+		--tf "$PWD/trace_wt_secondary" \
+		--t2 "$PWD/trace2_wt_secondary" &&
+
+	git -C wt-secondary fsmonitor--daemon stop &&
+	test_must_fail git -C wt-secondary fsmonitor--daemon status
+'
+
+# NEEDSWORK: Repeat one of the "edit" tests on wt-secondary and
+# confirm that we get the same events and behavior -- that is, that
+# fsmonitor--daemon correctly watches BOTH the working directory and
+# the external GITDIR directory and behaves the same as when ".git"
+# is a directory inside the working directory.
+
+test_expect_success 'cleanup worktrees' '
+	stop_daemon_delete_repo wt-secondary &&
+	stop_daemon_delete_repo wt-base
+'
+
+test_done
-- 
gitgitgadget

[PATCH v9 22/30] t/perf: avoid copying builtin fsmonitor files into test repo

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Do not copy any of the various fsmonitor--daemon files from the .git
directory of the (GIT_PREF_REPO or GIT_PERF_LARGE_REPO) source repo
into the test's trash directory.

When perf tests start, they copy the contents of the source repo into
the test's trash directory.  If fsmonitor is running in the source repo,
there may be control files, such as the IPC socket and/or fsmonitor
cookie files.  These should not be copied into the test repo.

Unix domain sockets cannot be copied in the manner used by the test
setup, so if present, the test setup fails.

Cookie files are harmless, but we should avoid them.

The builtin fsmonitor keeps all such control files/sockets in
.git/fsmonitor--daemon*, so it is simple to exclude them.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 t/perf/perf-lib.sh | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/t/perf/perf-lib.sh b/t/perf/perf-lib.sh
index 407252bac70..932105cd12c 100644
--- a/t/perf/perf-lib.sh
+++ b/t/perf/perf-lib.sh
@@ -78,7 +78,7 @@ test_perf_copy_repo_contents () {
 	for stuff in "$1"/*
 	do
 		case "$stuff" in
-		*/objects|*/hooks|*/config|*/commondir|*/gitdir|*/worktrees)
+		*/objects|*/hooks|*/config|*/commondir|*/gitdir|*/worktrees|*/fsmonitor--daemon*)
 			;;
 		*)
 			cp -R "$stuff" "$repo/.git/" || exit 1
-- 
gitgitgadget

[PATCH v9 23/30] t/helper/test-chmtime: skip directories on Windows

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach `test-tool.exe chmtime` to ignore errors when setting the mtime
on a directory on Windows.

NEEDSWORK: The Windows version of `utime()` (aka `mingw_utime()`) does
not properly handle directories because it uses `_wopen()`.  It should
be converted to using `CreateFileW()` and backup semantics at a minimum.
Since I'm already in the middle of a large patch series, I did not want
to destabilize other callers of `utime()` right now.  The problem has
only been observed in the t/perf/p7519 test when the test repo contains
an empty directory on disk.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 t/helper/test-chmtime.c | 15 +++++++++++++++
 1 file changed, 15 insertions(+)

diff --git a/t/helper/test-chmtime.c b/t/helper/test-chmtime.c
index 524b55ca496..dc28890a183 100644
--- a/t/helper/test-chmtime.c
+++ b/t/helper/test-chmtime.c
@@ -134,6 +134,21 @@ int cmd__chmtime(int argc, const char **argv)
 		}
 
 		if (utb.modtime != sb.st_mtime && utime(argv[i], &utb) < 0) {
+#ifdef GIT_WINDOWS_NATIVE
+			if (S_ISDIR(sb.st_mode)) {
+				/*
+				 * NEEDSWORK: The Windows version of `utime()`
+				 * (aka `mingw_utime()`) does not correctly
+				 * handle directory arguments, since it uses
+				 * `_wopen()`.  Ignore it for now since this
+				 * is just a test.
+				 */
+				fprintf(stderr,
+					("Failed to modify time on directory %s. "
+					 "Skipping\n"), argv[i]);
+				continue;
+			}
+#endif
 			fprintf(stderr, "Failed to modify time on %s: %s\n",
 			        argv[i], strerror(errno));
 			return 1;
-- 
gitgitgadget

[PATCH v9 24/30] t/perf/p7519: fix coding style

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
---
 t/perf/p7519-fsmonitor.sh | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/t/perf/p7519-fsmonitor.sh b/t/perf/p7519-fsmonitor.sh
index c8be58f3c76..5241eb6c4e5 100755
--- a/t/perf/p7519-fsmonitor.sh
+++ b/t/perf/p7519-fsmonitor.sh
@@ -72,7 +72,7 @@ then
 	fi
 fi
 
-trace_start() {
+trace_start () {
 	if test -n "$GIT_PERF_7519_TRACE"
 	then
 		name="$1"
@@ -91,7 +91,7 @@ trace_start() {
 	fi
 }
 
-trace_stop() {
+trace_stop () {
 	if test -n "$GIT_PERF_7519_TRACE"
 	then
 		unset GIT_TRACE2_PERF
@@ -133,7 +133,7 @@ test_expect_success "one time repo setup" '
 	fi
 '
 
-setup_for_fsmonitor() {
+setup_for_fsmonitor () {
 	# set INTEGRATION_SCRIPT depending on the environment
 	if test -n "$INTEGRATION_PATH"
 	then
@@ -173,7 +173,7 @@ test_perf_w_drop_caches () {
 	test_perf "$@"
 }
 
-test_fsmonitor_suite() {
+test_fsmonitor_suite () {
 	if test -n "$INTEGRATION_SCRIPT"; then
 		DESC="fsmonitor=$(basename $INTEGRATION_SCRIPT)"
 	else
-- 
gitgitgadget

[PATCH v9 25/30] t/perf/p7519: speed up test on Windows

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Change p7519 to use `test_seq` and `xargs` rather than a `for` loop
to touch thousands of files.  This takes minutes off of test runs
on Windows because of process creation overhead.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 t/perf/p7519-fsmonitor.sh | 24 ++++++++++++++++--------
 1 file changed, 16 insertions(+), 8 deletions(-)

diff --git a/t/perf/p7519-fsmonitor.sh b/t/perf/p7519-fsmonitor.sh
index 5241eb6c4e5..a6c2a910e70 100755
--- a/t/perf/p7519-fsmonitor.sh
+++ b/t/perf/p7519-fsmonitor.sh
@@ -98,6 +98,13 @@ trace_stop () {
 	fi
 }
 
+touch_files () {
+	n=$1 &&
+	d="$n"_files &&
+
+	(cd $d && test_seq 1 $n | xargs touch )
+}
+
 test_expect_success "one time repo setup" '
 	# set untrackedCache depending on the environment
 	if test -n "$GIT_PERF_7519_UNTRACKED_CACHE"
@@ -119,10 +126,11 @@ test_expect_success "one time repo setup" '
 	fi &&
 
 	mkdir 1_file 10_files 100_files 1000_files 10000_files &&
-	for i in $(test_seq 1 10); do touch 10_files/$i || return 1; done &&
-	for i in $(test_seq 1 100); do touch 100_files/$i || return 1; done &&
-	for i in $(test_seq 1 1000); do touch 1000_files/$i || return 1; done &&
-	for i in $(test_seq 1 10000); do touch 10000_files/$i || return 1; done &&
+	: 1_file directory should be left empty &&
+	touch_files 10 &&
+	touch_files 100 &&
+	touch_files 1000 &&
+	touch_files 10000 &&
 	git add 1_file 10_files 100_files 1000_files 10000_files &&
 	git commit -qm "Add files" &&
 
@@ -199,15 +207,15 @@ test_fsmonitor_suite () {
 
 	# Update the mtimes on upto 100k files to make status think
 	# that they are dirty.  For simplicity, omit any files with
-	# LFs (i.e. anything that ls-files thinks it needs to dquote).
-	# Then fully backslash-quote the paths to capture any
-	# whitespace so that they pass thru xargs properly.
+	# LFs (i.e. anything that ls-files thinks it needs to dquote)
+	# and any files with whitespace so that they pass thru xargs
+	# properly.
 	#
 	test_perf_w_drop_caches "status (dirty) ($DESC)" '
 		git ls-files | \
 			head -100000 | \
 			grep -v \" | \
-			sed '\''s/\(.\)/\\\1/g'\'' | \
+			grep -v " ." | \
 			xargs test-tool chmtime -300 &&
 		git status
 	'
-- 
gitgitgadget

[PATCH v9 26/30] t/perf/p7519: add fsmonitor--daemon test cases

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Repeat all of the fsmonitor perf tests using `git fsmonitor--daemon` and
the "Simple IPC" interface.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 t/perf/p7519-fsmonitor.sh | 38 ++++++++++++++++++++++++++++++++++----
 1 file changed, 34 insertions(+), 4 deletions(-)

diff --git a/t/perf/p7519-fsmonitor.sh b/t/perf/p7519-fsmonitor.sh
index a6c2a910e70..0b9129ca7bc 100755
--- a/t/perf/p7519-fsmonitor.sh
+++ b/t/perf/p7519-fsmonitor.sh
@@ -141,7 +141,7 @@ test_expect_success "one time repo setup" '
 	fi
 '
 
-setup_for_fsmonitor () {
+setup_for_fsmonitor_hook () {
 	# set INTEGRATION_SCRIPT depending on the environment
 	if test -n "$INTEGRATION_PATH"
 	then
@@ -182,7 +182,11 @@ test_perf_w_drop_caches () {
 }
 
 test_fsmonitor_suite () {
-	if test -n "$INTEGRATION_SCRIPT"; then
+	if test -n "$USE_FSMONITOR_DAEMON"
+	then
+		DESC="builtin fsmonitor--daemon"
+	elif test -n "$INTEGRATION_SCRIPT"
+	then
 		DESC="fsmonitor=$(basename $INTEGRATION_SCRIPT)"
 	else
 		DESC="fsmonitor=disabled"
@@ -261,11 +265,11 @@ test_fsmonitor_suite () {
 trace_start fsmonitor-watchman
 if test -n "$GIT_PERF_7519_FSMONITOR"; then
 	for INTEGRATION_PATH in $GIT_PERF_7519_FSMONITOR; do
-		test_expect_success "setup for fsmonitor $INTEGRATION_PATH" 'setup_for_fsmonitor'
+		test_expect_success "setup for fsmonitor $INTEGRATION_PATH" 'setup_for_fsmonitor_hook'
 		test_fsmonitor_suite
 	done
 else
-	test_expect_success "setup for fsmonitor" 'setup_for_fsmonitor'
+	test_expect_success "setup for fsmonitor hook" 'setup_for_fsmonitor_hook'
 	test_fsmonitor_suite
 fi
 
@@ -293,4 +297,30 @@ test_expect_success "setup without fsmonitor" '
 test_fsmonitor_suite
 trace_stop
 
+#
+# Run a full set of perf tests using the built-in fsmonitor--daemon.
+# It does not use the Hook API, so it has a different setup.
+# Explicitly start the daemon here and before we start client commands
+# so that we can later add custom tracing.
+#
+if test_have_prereq FSMONITOR_DAEMON
+then
+	USE_FSMONITOR_DAEMON=t
+
+	test_expect_success "setup for builtin fsmonitor" '
+		trace_start fsmonitor--daemon--server &&
+		git fsmonitor--daemon start &&
+
+		trace_start fsmonitor--daemon--client &&
+
+		git config core.fsmonitor true &&
+		git update-index --fsmonitor
+	'
+
+	test_fsmonitor_suite
+
+	git fsmonitor--daemon stop
+	trace_stop
+fi
+
 test_done
-- 
gitgitgadget

[PATCH v9 27/30] fsmonitor--daemon: periodically truncate list of modified files

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach fsmonitor--daemon to periodically truncate the list of
modified files to save some memory.

Clients will ask for the set of changes relative to a token that they
found in the FSMN index extension in the index.  (This token is like a
point in time, but different).  Clients will then update the index to
contain the response token (so that subsequent commands will be
relative to this new token).

Therefore, the daemon can gradually truncate the in-memory list of
changed paths as they become obsolete (older than the previous token).
Since we may have multiple clients making concurrent requests with a
skew of tokens and clients may be racing to the talk to the daemon,
we lazily truncate the list.

We introduce a 5 minute delay and truncate batches 5 minutes after
they are considered obsolete.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 builtin/fsmonitor--daemon.c | 88 +++++++++++++++++++++++++++++++++++++
 1 file changed, 88 insertions(+)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index eafaafb45b1..ab9cc09f7ce 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -312,6 +312,75 @@ static void fsmonitor_batch__combine(struct fsmonitor_batch *batch_dest,
 			batch_src->interned_paths[k];
 }
 
+/*
+ * To keep the batch list from growing unbounded in response to filesystem
+ * activity, we try to truncate old batches from the end of the list as
+ * they become irrelevant.
+ *
+ * We assume that the .git/index will be updated with the most recent token
+ * any time the index is updated.  And future commands will only ask for
+ * recent changes *since* that new token.  So as tokens advance into the
+ * future, older batch items will never be requested/needed.  So we can
+ * truncate them without loss of functionality.
+ *
+ * However, multiple commands may be talking to the daemon concurrently
+ * or perform a slow command, so a little "token skew" is possible.
+ * Therefore, we want this to be a little bit lazy and have a generous
+ * delay.
+ *
+ * The current reader thread walked backwards in time from `token->batch_head`
+ * back to `batch_marker` somewhere in the middle of the batch list.
+ *
+ * Let's walk backwards in time from that marker an arbitrary delay
+ * and truncate the list there.  Note that these timestamps are completely
+ * artificial (based on when we pinned the batch item) and not on any
+ * filesystem activity.
+ *
+ * Return the obsolete portion of the list after we have removed it from
+ * the official list so that the caller can free it after leaving the lock.
+ */
+#define MY_TIME_DELAY_SECONDS (5 * 60) /* seconds */
+
+static struct fsmonitor_batch *with_lock__truncate_old_batches(
+	struct fsmonitor_daemon_state *state,
+	const struct fsmonitor_batch *batch_marker)
+{
+	/* assert current thread holding state->main_lock */
+
+	const struct fsmonitor_batch *batch;
+	struct fsmonitor_batch *remainder;
+
+	if (!batch_marker)
+		return NULL;
+
+	trace_printf_key(&trace_fsmonitor, "Truncate: mark (%"PRIu64",%"PRIu64")",
+			 batch_marker->batch_seq_nr,
+			 (uint64_t)batch_marker->pinned_time);
+
+	for (batch = batch_marker; batch; batch = batch->next) {
+		time_t t;
+
+		if (!batch->pinned_time) /* an overflow batch */
+			continue;
+
+		t = batch->pinned_time + MY_TIME_DELAY_SECONDS;
+		if (t > batch_marker->pinned_time) /* too close to marker */
+			continue;
+
+		goto truncate_past_here;
+	}
+
+	return NULL;
+
+truncate_past_here:
+	state->current_token_data->batch_tail = (struct fsmonitor_batch *)batch;
+
+	remainder = ((struct fsmonitor_batch *)batch)->next;
+	((struct fsmonitor_batch *)batch)->next = NULL;
+
+	return remainder;
+}
+
 static void fsmonitor_free_token_data(struct fsmonitor_token_data *token)
 {
 	if (!token)
@@ -425,6 +494,7 @@ static int do_handle_client(struct fsmonitor_daemon_state *state,
 	const char *p;
 	const struct fsmonitor_batch *batch_head;
 	const struct fsmonitor_batch *batch;
+	struct fsmonitor_batch *remainder = NULL;
 	intmax_t count = 0, duplicates = 0;
 	kh_str_t *shown;
 	int hash_ret;
@@ -652,11 +722,29 @@ static int do_handle_client(struct fsmonitor_daemon_state *state,
 			 * that work.
 			 */
 			fsmonitor_free_token_data(token_data);
+		} else if (batch) {
+			/*
+			 * We are holding the lock and are the only
+			 * reader of the ref-counted portion of the
+			 * list, so we get the honor of seeing if the
+			 * list can be truncated to save memory.
+			 *
+			 * The main loop did not walk to the end of the
+			 * list, so this batch is the first item in the
+			 * batch-list that is older than the requested
+			 * end-point sequence number.  See if the tail
+			 * end of the list is obsolete.
+			 */
+			remainder = with_lock__truncate_old_batches(state,
+								    batch);
 		}
 	}
 
 	pthread_mutex_unlock(&state->main_lock);
 
+	if (remainder)
+		fsmonitor_batch__free_list(remainder);
+
 	trace2_data_intmax("fsmonitor", the_repository, "response/length", total_response_len);
 	trace2_data_intmax("fsmonitor", the_repository, "response/count/files", count);
 	trace2_data_intmax("fsmonitor", the_repository, "response/count/duplicates", duplicates);
-- 
gitgitgadget

[PATCH v9 28/30] fsmonitor--daemon: use a cookie file to sync with file system

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Teach fsmonitor--daemon client threads to create a cookie file
inside the .git directory and then wait until FS events for the
cookie are observed by the FS listener thread.

This helps address the racy nature of file system events by
blocking the client response until the kernel has drained any
event backlog.

This is especially important on MacOS where kernel events are
only issued with a limited frequency.  See the `latency` argument
of `FSeventStreamCreate()`.  The kernel only signals every `latency`
seconds, but does not guarantee that the kernel queue is completely
drained, so we may have to wait more than one interval.  If we
increase the latency, the system is more likely to drop events.
We avoid these issues by having each client thread create a unique
cookie file and then wait until it is seen in the event stream.

Co-authored-by: Kevin Willford <Kevin.Willford@microsoft.com>
Co-authored-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 builtin/fsmonitor--daemon.c | 237 +++++++++++++++++++++++++++++++++++-
 fsmonitor--daemon.h         |   5 +
 2 files changed, 241 insertions(+), 1 deletion(-)

diff --git a/builtin/fsmonitor--daemon.c b/builtin/fsmonitor--daemon.c
index ab9cc09f7ce..46be55a4618 100644
--- a/builtin/fsmonitor--daemon.c
+++ b/builtin/fsmonitor--daemon.c
@@ -107,6 +107,162 @@ static int do_as_client__status(void)
 	}
 }
 
+enum fsmonitor_cookie_item_result {
+	FCIR_ERROR = -1, /* could not create cookie file ? */
+	FCIR_INIT,
+	FCIR_SEEN,
+	FCIR_ABORT,
+};
+
+struct fsmonitor_cookie_item {
+	struct hashmap_entry entry;
+	char *name;
+	enum fsmonitor_cookie_item_result result;
+};
+
+static int cookies_cmp(const void *data, const struct hashmap_entry *he1,
+		     const struct hashmap_entry *he2, const void *keydata)
+{
+	const struct fsmonitor_cookie_item *a =
+		container_of(he1, const struct fsmonitor_cookie_item, entry);
+	const struct fsmonitor_cookie_item *b =
+		container_of(he2, const struct fsmonitor_cookie_item, entry);
+
+	return strcmp(a->name, keydata ? keydata : b->name);
+}
+
+static enum fsmonitor_cookie_item_result with_lock__wait_for_cookie(
+	struct fsmonitor_daemon_state *state)
+{
+	/* assert current thread holding state->main_lock */
+
+	int fd;
+	struct fsmonitor_cookie_item *cookie;
+	struct strbuf cookie_pathname = STRBUF_INIT;
+	struct strbuf cookie_filename = STRBUF_INIT;
+	enum fsmonitor_cookie_item_result result;
+	int my_cookie_seq;
+
+	CALLOC_ARRAY(cookie, 1);
+
+	my_cookie_seq = state->cookie_seq++;
+
+	strbuf_addf(&cookie_filename, "%i-%i", getpid(), my_cookie_seq);
+
+	strbuf_addbuf(&cookie_pathname, &state->path_cookie_prefix);
+	strbuf_addbuf(&cookie_pathname, &cookie_filename);
+
+	cookie->name = strbuf_detach(&cookie_filename, NULL);
+	cookie->result = FCIR_INIT;
+	hashmap_entry_init(&cookie->entry, strhash(cookie->name));
+
+	hashmap_add(&state->cookies, &cookie->entry);
+
+	trace_printf_key(&trace_fsmonitor, "cookie-wait: '%s' '%s'",
+			 cookie->name, cookie_pathname.buf);
+
+	/*
+	 * Create the cookie file on disk and then wait for a notification
+	 * that the listener thread has seen it.
+	 */
+	fd = open(cookie_pathname.buf, O_WRONLY | O_CREAT | O_EXCL, 0600);
+	if (fd < 0) {
+		error_errno(_("could not create fsmonitor cookie '%s'"),
+			    cookie->name);
+
+		cookie->result = FCIR_ERROR;
+		goto done;
+	}
+
+	/*
+	 * Technically, close() and unlink() can fail, but we don't
+	 * care here.  We only created the file to trigger a watch
+	 * event from the FS to know that when we're up to date.
+	 */
+	close(fd);
+	unlink(cookie_pathname.buf);
+
+	/*
+	 * Technically, this is an infinite wait (well, unless another
+	 * thread sends us an abort).  I'd like to change this to
+	 * use `pthread_cond_timedwait()` and return an error/timeout
+	 * and let the caller do the trivial response thing, but we
+	 * don't have that routine in our thread-utils.
+	 *
+	 * After extensive beta testing I'm not really worried about
+	 * this.  Also note that the above open() and unlink() calls
+	 * will cause at least two FS events on that path, so the odds
+	 * of getting stuck are pretty slim.
+	 */
+	while (cookie->result == FCIR_INIT)
+		pthread_cond_wait(&state->cookies_cond,
+				  &state->main_lock);
+
+done:
+	hashmap_remove(&state->cookies, &cookie->entry, NULL);
+
+	result = cookie->result;
+
+	free(cookie->name);
+	free(cookie);
+	strbuf_release(&cookie_pathname);
+
+	return result;
+}
+
+/*
+ * Mark these cookies as _SEEN and wake up the corresponding client threads.
+ */
+static void with_lock__mark_cookies_seen(struct fsmonitor_daemon_state *state,
+					 const struct string_list *cookie_names)
+{
+	/* assert current thread holding state->main_lock */
+
+	int k;
+	int nr_seen = 0;
+
+	for (k = 0; k < cookie_names->nr; k++) {
+		struct fsmonitor_cookie_item key;
+		struct fsmonitor_cookie_item *cookie;
+
+		key.name = cookie_names->items[k].string;
+		hashmap_entry_init(&key.entry, strhash(key.name));
+
+		cookie = hashmap_get_entry(&state->cookies, &key, entry, NULL);
+		if (cookie) {
+			trace_printf_key(&trace_fsmonitor, "cookie-seen: '%s'",
+					 cookie->name);
+			cookie->result = FCIR_SEEN;
+			nr_seen++;
+		}
+	}
+
+	if (nr_seen)
+		pthread_cond_broadcast(&state->cookies_cond);
+}
+
+/*
+ * Set _ABORT on all pending cookies and wake up all client threads.
+ */
+static void with_lock__abort_all_cookies(struct fsmonitor_daemon_state *state)
+{
+	/* assert current thread holding state->main_lock */
+
+	struct hashmap_iter iter;
+	struct fsmonitor_cookie_item *cookie;
+	int nr_aborted = 0;
+
+	hashmap_for_each_entry(&state->cookies, &iter, cookie, entry) {
+		trace_printf_key(&trace_fsmonitor, "cookie-abort: '%s'",
+				 cookie->name);
+		cookie->result = FCIR_ABORT;
+		nr_aborted++;
+	}
+
+	if (nr_aborted)
+		pthread_cond_broadcast(&state->cookies_cond);
+}
+
 /*
  * Requests to and from a FSMonitor Protocol V2 provider use an opaque
  * "token" as a virtual timestamp.  Clients can request a summary of all
@@ -404,6 +560,9 @@ static void fsmonitor_free_token_data(struct fsmonitor_token_data *token)
  *     We should create a new token and start fresh (as if we just
  *     booted up).
  *
+ * [2] Some of those lost events may have been for cookie files.  We
+ *     should assume the worst and abort them rather letting them starve.
+ *
  * If there are no concurrent threads reading the current token data
  * series, we can free it now.  Otherwise, let the last reader free
  * it.
@@ -425,6 +584,8 @@ static void with_lock__do_force_resync(struct fsmonitor_daemon_state *state)
 	state->current_token_data = new_one;
 
 	fsmonitor_free_token_data(free_me);
+
+	with_lock__abort_all_cookies(state);
 }
 
 void fsmonitor_force_resync(struct fsmonitor_daemon_state *state)
@@ -500,6 +661,8 @@ static int do_handle_client(struct fsmonitor_daemon_state *state,
 	int hash_ret;
 	int do_trivial = 0;
 	int do_flush = 0;
+	int do_cookie = 0;
+	enum fsmonitor_cookie_item_result cookie_result;
 
 	/*
 	 * We expect `command` to be of the form:
@@ -560,6 +723,7 @@ static int do_handle_client(struct fsmonitor_daemon_state *state,
 			 * We have a V2 valid token:
 			 *     "builtin:<token_id>:<seq_nr>"
 			 */
+			do_cookie = 1;
 		}
 	}
 
@@ -568,6 +732,30 @@ static int do_handle_client(struct fsmonitor_daemon_state *state,
 	if (!state->current_token_data)
 		BUG("fsmonitor state does not have a current token");
 
+	/*
+	 * Write a cookie file inside the directory being watched in
+	 * an effort to flush out existing filesystem events that we
+	 * actually care about.  Suspend this client thread until we
+	 * see the filesystem events for this cookie file.
+	 *
+	 * Creating the cookie lets us guarantee that our FS listener
+	 * thread has drained the kernel queue and we are caught up
+	 * with the kernel.
+	 *
+	 * If we cannot create the cookie (or otherwise guarantee that
+	 * we are caught up), we send a trivial response.  We have to
+	 * assume that there might be some very, very recent activity
+	 * on the FS still in flight.
+	 */
+	if (do_cookie) {
+		cookie_result = with_lock__wait_for_cookie(state);
+		if (cookie_result != FCIR_SEEN) {
+			error(_("fsmonitor: cookie_result '%d' != SEEN"),
+			      cookie_result);
+			do_trivial = 1;
+		}
+	}
+
 	if (do_flush)
 		with_lock__do_force_resync(state);
 
@@ -787,7 +975,9 @@ static int handle_client(void *data,
 	return result;
 }
 
-#define FSMONITOR_COOKIE_PREFIX ".fsmonitor-daemon-"
+#define FSMONITOR_DIR           "fsmonitor--daemon"
+#define FSMONITOR_COOKIE_DIR    "cookies"
+#define FSMONITOR_COOKIE_PREFIX (FSMONITOR_DIR "/" FSMONITOR_COOKIE_DIR "/")
 
 enum fsmonitor_path_type fsmonitor_classify_path_workdir_relative(
 	const char *rel)
@@ -940,6 +1130,9 @@ void fsmonitor_publish(struct fsmonitor_daemon_state *state,
 		}
 	}
 
+	if (cookie_names->nr)
+		with_lock__mark_cookies_seen(state, cookie_names);
+
 	pthread_mutex_unlock(&state->main_lock);
 }
 
@@ -1031,7 +1224,9 @@ static int fsmonitor_run_daemon(void)
 
 	memset(&state, 0, sizeof(state));
 
+	hashmap_init(&state.cookies, cookies_cmp, NULL, 0);
 	pthread_mutex_init(&state.main_lock, NULL);
+	pthread_cond_init(&state.cookies_cond, NULL);
 	state.error_code = 0;
 	state.current_token_data = fsmonitor_new_token_data();
 
@@ -1056,6 +1251,44 @@ static int fsmonitor_run_daemon(void)
 		state.nr_paths_watching = 2;
 	}
 
+	/*
+	 * We will write filesystem syncing cookie files into
+	 * <gitdir>/<fsmonitor-dir>/<cookie-dir>/<pid>-<seq>.
+	 *
+	 * The extra layers of subdirectories here keep us from
+	 * changing the mtime on ".git/" or ".git/foo/" when we create
+	 * or delete cookie files.
+	 *
+	 * There have been problems with some IDEs that do a
+	 * non-recursive watch of the ".git/" directory and run a
+	 * series of commands any time something happens.
+	 *
+	 * For example, if we place our cookie files directly in
+	 * ".git/" or ".git/foo/" then a `git status` (or similar
+	 * command) from the IDE will cause a cookie file to be
+	 * created in one of those dirs.  This causes the mtime of
+	 * those dirs to change.  This triggers the IDE's watch
+	 * notification.  This triggers the IDE to run those commands
+	 * again.  And the process repeats and the machine never goes
+	 * idle.
+	 *
+	 * Adding the extra layers of subdirectories prevents the
+	 * mtime of ".git/" and ".git/foo" from changing when a
+	 * cookie file is created.
+	 */
+	strbuf_init(&state.path_cookie_prefix, 0);
+	strbuf_addbuf(&state.path_cookie_prefix, &state.path_gitdir_watch);
+
+	strbuf_addch(&state.path_cookie_prefix, '/');
+	strbuf_addstr(&state.path_cookie_prefix, FSMONITOR_DIR);
+	mkdir(state.path_cookie_prefix.buf, 0777);
+
+	strbuf_addch(&state.path_cookie_prefix, '/');
+	strbuf_addstr(&state.path_cookie_prefix, FSMONITOR_COOKIE_DIR);
+	mkdir(state.path_cookie_prefix.buf, 0777);
+
+	strbuf_addch(&state.path_cookie_prefix, '/');
+
 	/*
 	 * Confirm that we can create platform-specific resources for the
 	 * filesystem listener before we bother starting all the threads.
@@ -1068,6 +1301,7 @@ static int fsmonitor_run_daemon(void)
 	err = fsmonitor_run_daemon_1(&state);
 
 done:
+	pthread_cond_destroy(&state.cookies_cond);
 	pthread_mutex_destroy(&state.main_lock);
 	fsm_listen__dtor(&state);
 
@@ -1075,6 +1309,7 @@ done:
 
 	strbuf_release(&state.path_worktree_watch);
 	strbuf_release(&state.path_gitdir_watch);
+	strbuf_release(&state.path_cookie_prefix);
 
 	return err;
 }
diff --git a/fsmonitor--daemon.h b/fsmonitor--daemon.h
index 010fbfe60e9..bd09fffc176 100644
--- a/fsmonitor--daemon.h
+++ b/fsmonitor--daemon.h
@@ -45,6 +45,11 @@ struct fsmonitor_daemon_state {
 
 	struct fsmonitor_token_data *current_token_data;
 
+	struct strbuf path_cookie_prefix;
+	pthread_cond_t cookies_cond;
+	int cookie_seq;
+	struct hashmap cookies;
+
 	int error_code;
 	struct fsmonitor_daemon_backend_data *backend_data;
 
-- 
gitgitgadget

[PATCH v9 29/30] fsmonitor: force update index after large responses

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Measure the time taken to apply the FSMonitor query result
to the index and the untracked-cache.

Set the `FSMONITOR_CHANGED` bit on `istate->cache_changed` when
FSMonitor returns a very large repsonse to ensure that the index is
written to disk.

Normally, when the FSMonitor response includes a tracked file, the
index is always updated.  Similarly, the index might be updated when
the response alters the untracked-cache (when enabled).  However, in
cases where neither of those cause the index to be considered changed,
the FSMonitor response is wasted.  Subsequent Git commands will make
requests with the same token and receive the same response.

If that response is very large, performance may suffer.  It would be
more efficient to force update the index now (and the token in the
index extension) in order to reduce the size of the response received
by future commands.

This was observed on Windows after a large checkout.  On Windows, the
kernel emits events for the files that are changed as they are
changed.  However, it might delay events for the containing
directories until the system is more idle (or someone scans the
directory (so it seems)).  The first status following a checkout would
get the list of files.  The subsequent status commands would get the
list of directories as the events trickled out.  But they would never
catch up because the token was not advanced because the index wasn't
updated.

This list of directories caused `wt_status_collect_untracked()` to
unnecessarily spend time actually scanning them during each command.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 fsmonitor.c | 55 ++++++++++++++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 54 insertions(+), 1 deletion(-)

diff --git a/fsmonitor.c b/fsmonitor.c
index a38b5710eb3..292a6742b4f 100644
--- a/fsmonitor.c
+++ b/fsmonitor.c
@@ -219,6 +219,43 @@ static void fsmonitor_refresh_callback(struct index_state *istate, char *name)
 	untracked_cache_invalidate_path(istate, name, 0);
 }
 
+/*
+ * The number of pathnames that we need to receive from FSMonitor
+ * before we force the index to be updated.
+ *
+ * Note that any pathname within the set of received paths MAY cause
+ * cache-entry or istate flag bits to be updated and thus cause the
+ * index to be updated on disk.
+ *
+ * However, the response may contain many paths (such as ignored
+ * paths) that will not update any flag bits.  And thus not force the
+ * index to be updated.  (This is fine and normal.)  It also means
+ * that the token will not be updated in the FSMonitor index
+ * extension.  So the next Git command will find the same token in the
+ * index, make the same token-relative request, and receive the same
+ * response (plus any newly changed paths).  If this response is large
+ * (and continues to grow), performance could be impacted.
+ *
+ * For example, if the user runs a build and it writes 100K object
+ * files but doesn't modify any source files, the index would not need
+ * to be updated.  The FSMonitor response (after the build and
+ * relative to a pre-build token) might be 5MB.  Each subsequent Git
+ * command will receive that same 100K/5MB response until something
+ * causes the index to be updated.  And `refresh_fsmonitor()` will
+ * have to iterate over those 100K paths each time.
+ *
+ * Performance could be improved if we optionally force update the
+ * index after a very large response and get an updated token into
+ * the FSMonitor index extension.  This should allow subsequent
+ * commands to get smaller and more current responses.
+ *
+ * The value chosen here does not need to be precise.  The index
+ * will be updated automatically the first time the user touches
+ * a tracked file and causes a command like `git status` to
+ * update an mtime to be updated and/or set a flag bit.
+ */
+static int fsmonitor_force_update_threshold = 100;
+
 void refresh_fsmonitor(struct index_state *istate)
 {
 	struct strbuf query_result = STRBUF_INIT;
@@ -362,25 +399,39 @@ apply_results:
 	 *     information and that we should consider everything
 	 *     invalid.  We call this a trivial response.
 	 */
+	trace2_region_enter("fsmonitor", "apply_results", istate->repo);
+
 	if (query_success && !is_trivial) {
 		/*
 		 * Mark all pathnames returned by the monitor as dirty.
 		 *
 		 * This updates both the cache-entries and the untracked-cache.
 		 */
+		int count = 0;
+
 		buf = query_result.buf;
 		for (i = bol; i < query_result.len; i++) {
 			if (buf[i] != '\0')
 				continue;
 			fsmonitor_refresh_callback(istate, buf + bol);
 			bol = i + 1;
+			count++;
 		}
-		if (bol < query_result.len)
+		if (bol < query_result.len) {
 			fsmonitor_refresh_callback(istate, buf + bol);
+			count++;
+		}
 
 		/* Now mark the untracked cache for fsmonitor usage */
 		if (istate->untracked)
 			istate->untracked->use_fsmonitor = 1;
+
+		if (count > fsmonitor_force_update_threshold)
+			istate->cache_changed |= FSMONITOR_CHANGED;
+
+		trace2_data_intmax("fsmonitor", istate->repo, "apply_count",
+				   count);
+
 	} else {
 		/*
 		 * We failed to get a response or received a trivial response,
@@ -409,6 +460,8 @@ apply_results:
 		if (istate->untracked)
 			istate->untracked->use_fsmonitor = 0;
 	}
+	trace2_region_leave("fsmonitor", "apply_results", istate->repo);
+
 	strbuf_release(&query_result);
 
 	/* Now that we've updated istate, save the last_update_token */
-- 
gitgitgadget

[PATCH v9 30/30] t7527: test status with untracked-cache and fsmonitor--daemon

[Jeff Hostetler via GitGitGadget]

From: Jeff Hostetler <jeffhost@microsoft.com>

Create 2x2 test matrix with the untracked-cache and fsmonitor--daemon
features and a series of edits and verify that status output is
identical.

Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 t/t7527-builtin-fsmonitor.sh | 115 +++++++++++++++++++++++++++++++++++
 1 file changed, 115 insertions(+)

diff --git a/t/t7527-builtin-fsmonitor.sh b/t/t7527-builtin-fsmonitor.sh
index 062e01c0dfc..bd0c952a116 100755
--- a/t/t7527-builtin-fsmonitor.sh
+++ b/t/t7527-builtin-fsmonitor.sh
@@ -205,6 +205,8 @@ test_expect_success 'setup' '
 	.gitignore
 	expect*
 	actual*
+	flush*
+	trace*
 	EOF
 
 	git -c core.fsmonitor=false add . &&
@@ -491,4 +493,117 @@ test_expect_success 'cleanup worktrees' '
 	stop_daemon_delete_repo wt-base
 '
 
+# The next few tests perform arbitrary/contrived file operations and
+# confirm that status is correct.  That is, that the data (or lack of
+# data) from fsmonitor doesn't cause incorrect results.  And doesn't
+# cause incorrect results when the untracked-cache is enabled.
+
+test_lazy_prereq UNTRACKED_CACHE '
+	git update-index --test-untracked-cache
+'
+
+test_expect_success 'Matrix: setup for untracked-cache,fsmonitor matrix' '
+	test_unconfig core.fsmonitor &&
+	git update-index --no-fsmonitor &&
+	test_might_fail git fsmonitor--daemon stop
+'
+
+matrix_clean_up_repo () {
+	git reset --hard HEAD &&
+	git clean -fd
+}
+
+matrix_try () {
+	uc=$1 &&
+	fsm=$2 &&
+	fn=$3 &&
+
+	if test $uc = true && test $fsm = false
+	then
+		# The untracked-cache is buggy when FSMonitor is
+		# DISABLED, so skip the tests for this matrix
+		# combination.
+		#
+		# We've observed random, occasional test failures on
+		# Windows and MacOS when the UC is turned on and FSM
+		# is turned off.  These are rare, but they do happen
+		# indicating that it is probably a race condition within
+		# the untracked cache itself.
+		#
+		# It usually happens when a test does F/D trickery and
+		# then the NEXT test fails because of extra status
+		# output from stale UC data from the previous test.
+		#
+		# Since FSMonitor is not involved in the error, skip
+		# the tests for this matrix combination.
+		#
+		return 0
+	fi &&
+
+	test_expect_success "Matrix[uc:$uc][fsm:$fsm] $fn" '
+		matrix_clean_up_repo &&
+		$fn &&
+		if test $uc = false && test $fsm = false
+		then
+			git status --porcelain=v1 >.git/expect.$fn
+		else
+			git status --porcelain=v1 >.git/actual.$fn &&
+			test_cmp .git/expect.$fn .git/actual.$fn
+		fi
+	'
+}
+
+uc_values="false"
+test_have_prereq UNTRACKED_CACHE && uc_values="false true"
+for uc_val in $uc_values
+do
+	if test $uc_val = false
+	then
+		test_expect_success "Matrix[uc:$uc_val] disable untracked cache" '
+			git config core.untrackedcache false &&
+			git update-index --no-untracked-cache
+		'
+	else
+		test_expect_success "Matrix[uc:$uc_val] enable untracked cache" '
+			git config core.untrackedcache true &&
+			git update-index --untracked-cache
+		'
+	fi
+
+	fsm_values="false true"
+	for fsm_val in $fsm_values
+	do
+		if test $fsm_val = false
+		then
+			test_expect_success "Matrix[uc:$uc_val][fsm:$fsm_val] disable fsmonitor" '
+				test_unconfig core.fsmonitor &&
+				git update-index --no-fsmonitor &&
+				test_might_fail git fsmonitor--daemon stop
+			'
+		else
+			test_expect_success "Matrix[uc:$uc_val][fsm:$fsm_val] enable fsmonitor" '
+				git config core.fsmonitor true &&
+				git fsmonitor--daemon start &&
+				git update-index --fsmonitor
+			'
+		fi
+
+		matrix_try $uc_val $fsm_val edit_files
+		matrix_try $uc_val $fsm_val delete_files
+		matrix_try $uc_val $fsm_val create_files
+		matrix_try $uc_val $fsm_val rename_files
+		matrix_try $uc_val $fsm_val file_to_directory
+		matrix_try $uc_val $fsm_val directory_to_file
+
+		if test $fsm_val = true
+		then
+			test_expect_success "Matrix[uc:$uc_val][fsm:$fsm_val] disable fsmonitor at end" '
+				test_unconfig core.fsmonitor &&
+				git update-index --no-fsmonitor &&
+				test_might_fail git fsmonitor--daemon stop
+			'
+		fi
+	done
+done
+
 test_done
-- 
gitgitgadget

RE: [PATCH v9 00/30] Builtin FSMonitor Part 2

[rsbecker]

On March 25, 2022 2:03 PM, Jeff Hostetler wrote:
>Here is V9 of Part 2 of my builtin FSMonitor series. This version addresses bash
>style issues in t7527 raised on V8. These changes do not require a new version of
>Part 3.
>
>Here is a range-diff from V8 to V9 relative to 715d08a9e5 (The eighth batch, 2022-
>02-25).
>
> 1:  e98373f997 =  1:  e98373f997 fsmonitor: enhance existing comments, clarify
>trivial response handling
> 2:  ab68b94417 =  2:  ab68b94417 fsmonitor-ipc: create client routines for git-
>fsmonitor--daemon
> 3:  e04c7301f2 =  3:  e04c7301f2 fsmonitor: config settings are repository-specific
> 4:  ea02ba25d8 =  4:  ea02ba25d8 fsmonitor: use IPC to query the builtin FSMonitor
>daemon
> 5:  6ab7db9cb7 =  5:  6ab7db9cb7 fsmonitor: document builtin fsmonitor
> 6:  0ce8ae3f2c =  6:  0ce8ae3f2c fsmonitor--daemon: add a built-in fsmonitor
>daemon
> 7:  4624ce2fa4 =  7:  4624ce2fa4 fsmonitor--daemon: implement 'stop' and 'status'
>commands
> 8:  a29fe7266a =  8:  a29fe7266a compat/fsmonitor/fsm-listen-win32: stub in
>backend for Windows
> 9:  2f8a42fdb9 =  9:  2f8a42fdb9 compat/fsmonitor/fsm-listen-darwin: stub in
>backend for Darwin
>10:  f07800690e = 10:  f07800690e fsmonitor--daemon: implement 'run' command
>11:  a6a39a3306 = 11:  a6a39a3306 fsmonitor--daemon: implement 'start' command
>12:  d62e338d00 = 12:  d62e338d00 fsmonitor--daemon: add pathname
>classification
>13:  53e06b4ae5 = 13:  53e06b4ae5 fsmonitor--daemon: define token-ids
>14:  39f43fabe0 = 14:  39f43fabe0 fsmonitor--daemon: create token-based
>changed path cache
>15:  239558e34f = 15:  239558e34f compat/fsmonitor/fsm-listen-win32: implement
>FSMonitor backend on Windows
>16:  14b775e9d8 = 16:  14b775e9d8 compat/fsmonitor/fsm-listen-darwin: add
>MacOS header files for FSEvent
>17:  55bd7aee06 = 17:  55bd7aee06 compat/fsmonitor/fsm-listen-darwin:
>implement FSEvent listener on MacOS
>18:  c43009124f = 18:  c43009124f fsmonitor--daemon: implement handle_client
>callback
>19:  ed338777b5 = 19:  ed338777b5 help: include fsmonitor--daemon feature flag in
>version info
>20:  c99bac29d4 = 20:  c99bac29d4 t/helper/fsmonitor-client: create IPC client to
>talk to FSMonitor Daemon
>21:  c8709da945 ! 21:  bc94e379b0 t7527: create test for fsmonitor--daemon
>    @@ Commit message
>         t7527: create test for fsmonitor--daemon
>
>         Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
>    -    Signed-off-by: Junio C Hamano <gitster@pobox.com>
>
>      ## t/t7527-builtin-fsmonitor.sh (new) ##
>     @@
>    @@ t/t7527-builtin-fsmonitor.sh (new)
>     +    rm -rf $1
>     +}
>     +
>    -+is_value () {
>    -+    test -n "$1" && test "${1::1}" != "-"
>    -+}
>    -+
>     +start_daemon () {
>    -+    r= &&
>    -+    tf= &&
>    -+    t2= &&
>    -+    tk= &&
>    ++    r= tf= t2= tk= &&
>     +
>     +    while test "$#" -ne 0
>     +    do
>     +        case "$1" in
>     +        -C)
>    -+            shift;
>    -+            is_value $1 || BUG "error: -C requires value"
>    -+            r="-C $1"
>    ++            r="-C ${2?}"
>     +            shift
>     +            ;;
>     +        --tf)
>    -+            shift;
>    -+            is_value $1 || BUG "error: --tf requires value"
>    -+            tf="$1"
>    ++            tf="${2?}"
>     +            shift
>     +            ;;
>     +        --t2)
>    -+            shift;
>    -+            is_value $1 || BUG "error: --t2 requires value"
>    -+            t2="$1"
>    ++            t2="${2?}"
>     +            shift
>     +            ;;
>     +        --tk)
>    -+            shift;
>    -+            is_value $1 || BUG "error: --tk requires value"
>    -+            tk="$1"
>    ++            tk="${2?}"
>     +            shift
>     +            ;;
>    -+        *)
>    ++        -*)
>     +            BUG "error: unknown option: '$1'"
>     +            ;;
>    ++        *)
>    ++            BUG "error: unbound argument: '$1'"
>    ++            ;;
>     +        esac
>    ++        shift
>     +    done &&
>     +
>     +    (
>22:  cc39ecf10a = 22:  06d56d3a73 t/perf: avoid copying builtin fsmonitor files into
>test repo
>23:  2bb3eb8476 = 23:  2dd0215127 t/helper/test-chmtime: skip directories on
>Windows
>24:  bab9a9b080 = 24:  bb88cddc13 t/perf/p7519: fix coding style
>25:  2dd06ad2f7 = 25:  50c2afaa49 t/perf/p7519: speed up test on Windows
>26:  6eaa5765ae = 26:  5b18e3b692 t/perf/p7519: add fsmonitor--daemon test
>cases
>27:  30957f3930 = 27:  899c23f63c fsmonitor--daemon: periodically truncate list of
>modified files
>28:  c8ca2a1727 = 28:  597a7192f9 fsmonitor--daemon: use a cookie file to sync with
>file system
>29:  4caf1d89b8 = 29:  68a05fd289 fsmonitor: force update index after large
>responses
>30:  f87a1eba69 = 30:  5eb696daba t7527: test status with untracked-cache and
>fsmonitor--daemon
>
>
>Jeff Hostetler (30):
>  fsmonitor: enhance existing comments, clarify trivial response
>    handling
>  fsmonitor-ipc: create client routines for git-fsmonitor--daemon
>  fsmonitor: config settings are repository-specific
>  fsmonitor: use IPC to query the builtin FSMonitor daemon
>  fsmonitor: document builtin fsmonitor
>  fsmonitor--daemon: add a built-in fsmonitor daemon
>  fsmonitor--daemon: implement 'stop' and 'status' commands
>  compat/fsmonitor/fsm-listen-win32: stub in backend for Windows
>  compat/fsmonitor/fsm-listen-darwin: stub in backend for Darwin
>  fsmonitor--daemon: implement 'run' command
>  fsmonitor--daemon: implement 'start' command
>  fsmonitor--daemon: add pathname classification
>  fsmonitor--daemon: define token-ids
>  fsmonitor--daemon: create token-based changed path cache
>  compat/fsmonitor/fsm-listen-win32: implement FSMonitor backend on
>    Windows
>  compat/fsmonitor/fsm-listen-darwin: add MacOS header files for FSEvent
>  compat/fsmonitor/fsm-listen-darwin: implement FSEvent listener on
>    MacOS
>  fsmonitor--daemon: implement handle_client callback
>  help: include fsmonitor--daemon feature flag in version info
>  t/helper/fsmonitor-client: create IPC client to talk to FSMonitor
>    Daemon
>  t7527: create test for fsmonitor--daemon
>  t/perf: avoid copying builtin fsmonitor files into test repo
>  t/helper/test-chmtime: skip directories on Windows
>  t/perf/p7519: fix coding style
>  t/perf/p7519: speed up test on Windows
>  t/perf/p7519: add fsmonitor--daemon test cases
>  fsmonitor--daemon: periodically truncate list of modified files
>  fsmonitor--daemon: use a cookie file to sync with file system
>  fsmonitor: force update index after large responses
>  t7527: test status with untracked-cache and fsmonitor--daemon
>
> .gitignore                              |    1 +
> Documentation/config/core.txt           |   60 +-
> Documentation/git-fsmonitor--daemon.txt |   75 ++
> Documentation/git-update-index.txt      |    8 +-
> Makefile                                |   17 +
> builtin.h                               |    1 +
> builtin/fsmonitor--daemon.c             | 1479 +++++++++++++++++++++++
> builtin/update-index.c                  |    7 +-
> cache.h                                 |    1 -
> compat/fsmonitor/fsm-darwin-gcc.h       |   92 ++
> compat/fsmonitor/fsm-listen-darwin.c    |  427 +++++++
> compat/fsmonitor/fsm-listen-win32.c     |  586 +++++++++
> compat/fsmonitor/fsm-listen.h           |   49 +
> config.c                                |   14 -
> config.h                                |    1 -
> config.mak.uname                        |   20 +
> contrib/buildsystems/CMakeLists.txt     |   10 +
> environment.c                           |    1 -
> fsmonitor--daemon.h                     |  166 +++
> fsmonitor-ipc.c                         |  171 +++
> fsmonitor-ipc.h                         |   48 +
> fsmonitor-settings.c                    |  114 ++
> fsmonitor-settings.h                    |   21 +
> fsmonitor.c                             |  216 +++-
> fsmonitor.h                             |   15 +-
> git.c                                   |    1 +
> help.c                                  |    4 +
> repo-settings.c                         |    1 +
> repository.h                            |    3 +
> t/README                                |    4 +-
> t/helper/test-chmtime.c                 |   15 +
> t/helper/test-fsmonitor-client.c        |  116 ++
> t/helper/test-tool.c                    |    1 +
> t/helper/test-tool.h                    |    1 +
> t/perf/p7519-fsmonitor.sh               |   68 +-
> t/perf/perf-lib.sh                      |    2 +-
> t/t7527-builtin-fsmonitor.sh            |  609 ++++++++++
> t/test-lib.sh                           |    7 +
> 38 files changed, 4326 insertions(+), 106 deletions(-)  create mode 100644
>Documentation/git-fsmonitor--daemon.txt
> create mode 100644 builtin/fsmonitor--daemon.c  create mode 100644
>compat/fsmonitor/fsm-darwin-gcc.h  create mode 100644 compat/fsmonitor/fsm-
>listen-darwin.c
> create mode 100644 compat/fsmonitor/fsm-listen-win32.c
> create mode 100644 compat/fsmonitor/fsm-listen.h  create mode 100644
>fsmonitor--daemon.h  create mode 100644 fsmonitor-ipc.c  create mode 100644
>fsmonitor-ipc.h  create mode 100644 fsmonitor-settings.c  create mode 100644
>fsmonitor-settings.h  create mode 100644 t/helper/test-fsmonitor-client.c  create
>mode 100755 t/t7527-builtin-fsmonitor.sh
>
>
>base-commit: 715d08a9e51251ad8290b181b6ac3b9e1f9719d7
>Published-As: https://github.com/gitgitgadget/git/releases/tag/pr-
>1041%2Fjeffhostetler%2Fbuiltin-fsmonitor-part2-v9
>Fetch-It-Via: git fetch https://github.com/gitgitgadget/git pr-
>1041/jeffhostetler/builtin-fsmonitor-part2-v9
>Pull-Request: https://github.com/gitgitgadget/git/pull/1041
>
>Range-diff vs v8:
>
>  1:  e98373f997f =  1:  e98373f997f fsmonitor: enhance existing comments, clarify
>trivial response handling
>  2:  ab68b944173 =  2:  ab68b944173 fsmonitor-ipc: create client routines for git-
>fsmonitor--daemon
>  3:  e04c7301f24 =  3:  e04c7301f24 fsmonitor: config settings are repository-
>specific
>  4:  ea02ba25d8f =  4:  ea02ba25d8f fsmonitor: use IPC to query the builtin
>FSMonitor daemon
>  5:  6ab7db9cb76 =  5:  6ab7db9cb76 fsmonitor: document builtin fsmonitor
>  6:  0ce8ae3f2cf =  6:  0ce8ae3f2cf fsmonitor--daemon: add a built-in fsmonitor
>daemon
>  7:  4624ce2fa47 =  7:  4624ce2fa47 fsmonitor--daemon: implement 'stop' and
>'status' commands
>  8:  a29fe7266a4 =  8:  a29fe7266a4 compat/fsmonitor/fsm-listen-win32: stub in
>backend for Windows
>  9:  2f8a42fdb93 =  9:  2f8a42fdb93 compat/fsmonitor/fsm-listen-darwin: stub in
>backend for Darwin
> 10:  f07800690ee = 10:  f07800690ee fsmonitor--daemon: implement 'run'
>command
> 11:  a6a39a3306d = 11:  a6a39a3306d fsmonitor--daemon: implement 'start'
>command
> 12:  d62e338d008 = 12:  d62e338d008 fsmonitor--daemon: add pathname
>classification
> 13:  53e06b4ae5d = 13:  53e06b4ae5d fsmonitor--daemon: define token-ids
> 14:  39f43fabe02 = 14:  39f43fabe02 fsmonitor--daemon: create token-based
>changed path cache
> 15:  239558e34ff = 15:  239558e34ff compat/fsmonitor/fsm-listen-win32:
>implement FSMonitor backend on Windows
> 16:  14b775e9d8b = 16:  14b775e9d8b compat/fsmonitor/fsm-listen-darwin: add
>MacOS header files for FSEvent
> 17:  55bd7aee06c = 17:  55bd7aee06c compat/fsmonitor/fsm-listen-darwin:
>implement FSEvent listener on MacOS
> 18:  c43009124fb = 18:  c43009124fb fsmonitor--daemon: implement handle_client
>callback
> 19:  ed338777b56 = 19:  ed338777b56 help: include fsmonitor--daemon feature
>flag in version info
> 20:  c99bac29d42 = 20:  c99bac29d42 t/helper/fsmonitor-client: create IPC client to
>talk to FSMonitor Daemon
> 21:  c8709da9457 ! 21:  bc94e379b03 t7527: create test for fsmonitor--daemon
>     @@ Commit message
>          t7527: create test for fsmonitor--daemon
>
>          Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com>
>     -    Signed-off-by: Junio C Hamano <gitster@pobox.com>
>
>       ## t/t7527-builtin-fsmonitor.sh (new) ##
>      @@
>     @@ t/t7527-builtin-fsmonitor.sh (new)
>      +	rm -rf $1
>      +}
>      +
>     -+is_value () {
>     -+	test -n "$1" && test "${1::1}" != "-"
>     -+}
>     -+
>      +start_daemon () {
>     -+	r= &&
>     -+	tf= &&
>     -+	t2= &&
>     -+	tk= &&
>     ++	r= tf= t2= tk= &&
>      +
>      +	while test "$#" -ne 0
>      +	do
>      +		case "$1" in
>      +		-C)
>     -+			shift;
>     -+			is_value $1 || BUG "error: -C requires value"
>     -+			r="-C $1"
>     ++			r="-C ${2?}"
>      +			shift
>      +			;;
>      +		--tf)
>     -+			shift;
>     -+			is_value $1 || BUG "error: --tf requires value"
>     -+			tf="$1"
>     ++			tf="${2?}"
>      +			shift
>      +			;;
>      +		--t2)
>     -+			shift;
>     -+			is_value $1 || BUG "error: --t2 requires value"
>     -+			t2="$1"
>     ++			t2="${2?}"
>      +			shift
>      +			;;
>      +		--tk)
>     -+			shift;
>     -+			is_value $1 || BUG "error: --tk requires value"
>     -+			tk="$1"
>     ++			tk="${2?}"
>      +			shift
>      +			;;
>     -+		*)
>     ++		-*)
>      +			BUG "error: unknown option: '$1'"
>      +			;;
>     ++		*)
>     ++			BUG "error: unbound argument: '$1'"
>     ++			;;
>      +		esac
>     ++		shift
>      +	done &&
>      +
>      +	(
> 22:  cc39ecf10ae = 22:  06d56d3a733 t/perf: avoid copying builtin fsmonitor files
>into test repo
> 23:  2bb3eb84767 = 23:  2dd02151278 t/helper/test-chmtime: skip directories on
>Windows
> 24:  bab9a9b0802 = 24:  bb88cddc137 t/perf/p7519: fix coding style
> 25:  2dd06ad2f71 = 25:  50c2afaa49e t/perf/p7519: speed up test on Windows
> 26:  6eaa5765ae1 = 26:  5b18e3b6926 t/perf/p7519: add fsmonitor--daemon test
>cases
> 27:  30957f3930e = 27:  899c23f63c3 fsmonitor--daemon: periodically truncate list
>of modified files
> 28:  c8ca2a17277 = 28:  597a7192f94 fsmonitor--daemon: use a cookie file to sync
>with file system
> 29:  4caf1d89b84 = 29:  68a05fd2892 fsmonitor: force update index after large
>responses
> 30:  f87a1eba693 = 30:  5eb696daba2 t7527: test status with untracked-cache and
>fsmonitor--daemon

So that we can plan for next time, would you mind giving some thought to what would be required for Linux and whether it makes sense to extend this. No urgency, but I would not mind participating - assuming my $DAYJOB lets me. I have a backlog for git that I need to get done first anyway.

Kind Regards,
Randall

Re: [PATCH v9 00/30] Builtin FSMonitor Part 2

[Jeff Hostetler]

On 3/25/22 3:02 PM, rsbecker@nexbridge.com wrote:
> On March 25, 2022 2:03 PM, Jeff Hostetler wrote:
[...]
> 
> So that we can plan for next time, would you mind giving some thought to what would be required for Linux and whether it makes sense to extend this. No urgency, but I would not mind participating - assuming my $DAYJOB lets me. I have a backlog for git that I need to get done first anyway.
> 
> Kind Regards,
> Randall
> 

A Linux backend would need to:

(1) stub in compat/fsmonitor/fsm-listen-linux.c (see commits
     8/30 and 9/30 in this series).  That gives you enough for
     the builtin/fsmonitor--daemon.c to link with your new
     backend.

(2) populate those 4 routines.

(2a) __ctor() and __dtor() will be called from the main thread
      before and after the listener thread is created.  You can
      do anything you need there to register/deregister a watch
      on the FS.

      See inotify() and/or fanotify().  I haven't looked at those
      routines for a while, so all I can say is google it.

(2b) __stop_async() will be called by "some" thread to request
      that the listener thread stop listening.  This is an async
      request, so just notify the listener thread and return.
      (There is an example of this for Unix in
       compat/simple-ipc/ipc-unix-sockets.c)

(2c) __loop() will be called (once) by the "listener" thread-proc
      to process/service events from the FS until a shutdown event
      is received.  This runs in the body of the "listener" thread.
      It should probably use poll()/select()/whatever on fd's from
      inotify()/fsnotify() *and* whatever you set up in (2b) to
      wait for a shutdown event.

      (There are examples of this wait-loop in the Windows and Mac
      backends, but they have *very* different FS event and wait
      models, so they might not be very helpful here.)

      When you get a "batch" of one or more paths from the FS,
      use the fsmonitor_classify_*() routines to classify or
      discard them and then use fsmonitor_publish() to publish
      newly changed paths to the other threads.

      The core code will handle path de-dup and all locking so
      you don't have to.

It is important that you figure out how to get recursive data
from the FS.  We want to watch the complete worktree.  Windows
and Mac let you register the root directory of the watch and
automatically give me events for anything under it.  IIRC,
inotify() only gave you a single directory and you had to
readdir() and recurse to get fd's to the subdirs.  I haven't
looked to see if fanotify() solves that or not.  So there may
be some fd juggling and tree walking required.  That could/should
all be hidden inside the __ctor() and/or __loop() routines.

(3) stub in compat/fsmonitor/fsm-settings-linux.c (see the
     peers.  This is needed to link.

(4) fill in any platform-specific reasons why you might want
     to reject a worktree.  for example, when they are remote
     (NFS/SMB might support it, but do you trust it....)

(5) stub in compat/fsmonitor/fsm-health-linux.c (see
     fsm-health-*.c in part 3).  That will give you enough
     to link the health thread.

(6) this part is probably optional (at least for now).  we
     can use this if we want to add platform-specific things
     like auto-shutdown after idle.  The mac version is currently
     empty, but the Windows version needs to watch the worktree
     root separately (because the FS watch is limited to what
     is *within* watched directory root).

That should help get you get started.
Let me know if you have questions.

Jeff

Re: [PATCH v9 00/30] Builtin FSMonitor Part 2

[Junio C Hamano]

"Jeff Hostetler via GitGitGadget" <gitgitgadget@gmail.com> writes:

> Here is V9 of Part 2 of my builtin FSMonitor series. This version addresses
> bash style issues in t7527 raised on V8. These changes do not require a new
> version of Part 3.

This addresses all the "oops" fix-ups from part 2.5, and good base
to build part 3 on top, right?

Queued.

Thanks.

Re: [PATCH v9 00/30] Builtin FSMonitor Part 2

[Ævar Arnfjörð Bjarmason]

On Fri, Mar 25 2022, Jeff Hostetler wrote:

> On 3/25/22 3:02 PM, rsbecker@nexbridge.com wrote:
>> On March 25, 2022 2:03 PM, Jeff Hostetler wrote:
> [...]
>> So that we can plan for next time, would you mind giving some
>> thought to what would be required for Linux and whether it makes
>> sense to extend this. No urgency, but I would not mind participating
>> - assuming my $DAYJOB lets me. I have a backlog for git that I need
>> to get done first anyway.
>> Kind Regards,
>> Randall
>> 
>
> A Linux backend would need to:
>
> (1) stub in compat/fsmonitor/fsm-listen-linux.c (see commits
>     8/30 and 9/30 in this series).  That gives you enough for
>     the builtin/fsmonitor--daemon.c to link with your new
>     backend.
>
> (2) populate those 4 routines.
>
> (2a) __ctor() and __dtor() will be called from the main thread
>      before and after the listener thread is created.  You can
>      do anything you need there to register/deregister a watch
>      on the FS.
>
>      See inotify() and/or fanotify().  I haven't looked at those
>      routines for a while, so all I can say is google it.
>
> (2b) __stop_async() will be called by "some" thread to request
>      that the listener thread stop listening.  This is an async
>      request, so just notify the listener thread and return.
>      (There is an example of this for Unix in
>       compat/simple-ipc/ipc-unix-sockets.c)
>
> (2c) __loop() will be called (once) by the "listener" thread-proc
>      to process/service events from the FS until a shutdown event
>      is received.  This runs in the body of the "listener" thread.
>      It should probably use poll()/select()/whatever on fd's from
>      inotify()/fsnotify() *and* whatever you set up in (2b) to
>      wait for a shutdown event.
>
>      (There are examples of this wait-loop in the Windows and Mac
>      backends, but they have *very* different FS event and wait
>      models, so they might not be very helpful here.)
>
>      When you get a "batch" of one or more paths from the FS,
>      use the fsmonitor_classify_*() routines to classify or
>      discard them and then use fsmonitor_publish() to publish
>      newly changed paths to the other threads.
>
>      The core code will handle path de-dup and all locking so
>      you don't have to.
>
> It is important that you figure out how to get recursive data
> from the FS.  We want to watch the complete worktree.  Windows
> and Mac let you register the root directory of the watch and
> automatically give me events for anything under it.  IIRC,
> inotify() only gave you a single directory and you had to
> readdir() and recurse to get fd's to the subdirs.  I haven't
> looked to see if fanotify() solves that or not.  So there may
> be some fd juggling and tree walking required.  That could/should
> all be hidden inside the __ctor() and/or __loop() routines.
>
> (3) stub in compat/fsmonitor/fsm-settings-linux.c (see the
>     peers.  This is needed to link.
>
> (4) fill in any platform-specific reasons why you might want
>     to reject a worktree.  for example, when they are remote
>     (NFS/SMB might support it, but do you trust it....)
>
> (5) stub in compat/fsmonitor/fsm-health-linux.c (see
>     fsm-health-*.c in part 3).  That will give you enough
>     to link the health thread.
>
> (6) this part is probably optional (at least for now).  we
>     can use this if we want to add platform-specific things
>     like auto-shutdown after idle.  The mac version is currently
>     empty, but the Windows version needs to watch the worktree
>     root separately (because the FS watch is limited to what
>     is *within* watched directory root).
>
> That should help get you get started.
> Let me know if you have questions.

Wouldn't it be much simpler POC in this case to write "watchman
backend"?  Then we'd both get a Linux backend, and an alternate backend
for the other platforms to validate their implementation.

Some past references to that:
https://lore.kernel.org/git/871r8c73ej.fsf@evledraar.gmail.com/ &
https://lore.kernel.org/git/87h7lgfchm.fsf@evledraar.gmail.com/

Re: [PATCH v9 00/30] Builtin FSMonitor Part 2

[Jeff Hostetler]

On 3/25/22 7:09 PM, Junio C Hamano wrote:
> "Jeff Hostetler via GitGitGadget" <gitgitgadget@gmail.com> writes:
> 
>> Here is V9 of Part 2 of my builtin FSMonitor series. This version addresses
>> bash style issues in t7527 raised on V8. These changes do not require a new
>> version of Part 3.
> 
> This addresses all the "oops" fix-ups from part 2.5, and good base
> to build part 3 on top, right?
> 
> Queued.
> 
> Thanks.
> 

Yes, this version of part 2 includes all of the fixups from 2.5.
And 2.5 should be dropped.
And yes, this part 2 should be the basis for part 3.

Thanks again!
Jeff

Re: [PATCH v9 00/30] Builtin FSMonitor Part 2

[Jeff Hostetler]

On 3/25/22 8:48 PM, Ævar Arnfjörð Bjarmason wrote:
> 
> On Fri, Mar 25 2022, Jeff Hostetler wrote:
> 
>> On 3/25/22 3:02 PM, rsbecker@nexbridge.com wrote:
>>> On March 25, 2022 2:03 PM, Jeff Hostetler wrote:
>> [...]
[...]
> 
> Wouldn't it be much simpler POC in this case to write "watchman
> backend"?  Then we'd both get a Linux backend, and an alternate backend
> for the other platforms to validate their implementation.
> 
> Some past references to that:
> https://lore.kernel.org/git/871r8c73ej.fsf@evledraar.gmail.com/ &
> https://lore.kernel.org/git/87h7lgfchm.fsf@evledraar.gmail.com/
> 

Yes, there are several ways for a client command, such as anyone
who calls read_index/refresh_index, to get FS change data from a
monitoring service.

Let's go thru the options here for the sake of conversation:

(option 1): Use the hook-like mechanism that Ben built in 2017
             to talk to an interlude program, shell script, perl
             script, etc.  That "script" itself then talks to a
             long-running service/daemon, such as Watchman, to get
             the list of changes and relays them back to the client.

     * This "proxy" has to handle protocol format conversions.
     * It may also have to start the service on new repos.
     * And depends upon a third-party service being installed.
     * We are limited to supporting platforms where the third-party
       tool is supported.

(option 2): Replace the hook with builtin client code to talk
             directly to the service and bypass the need for
             the proxy script/executable.

     * Git client code would need client-side IPC to talk to
       an established and running service.  (Similar to the client
       side of Simple-IPC but probably not pkt-line based.)
     * Git client code would now need to handle any protocol
       format conversions.
     * Git client code might also have to start the service.
     * And we'd still be dependent on a third-party service being
       installed.
     * And we are still limited to supporting platforms where
       the third-party tool is supported.
     * So far we've been assuming that that third-party tool is
       "Watchman", but technically, you could have other such
       services available.
       * So you may need multiple implementations of option 2,
         one for each third-party tool.
       * I'm not saying that this is hard, but just yet another
         detail that would have to be encoded in the Git source
         to get this "free" feature.

(option 3): Git implements a daemon to monitor the file system
             directly.

     * Git owns the protocol between client and service.
     * Git owns the backend, so no third-party tools required.
     * Git owns service startup.
     * Unfortunately, we are also responsible for building the
       backends on each platform we want to support.

     * In the future, we could augment the service to be more
       "Git-aware", such as discarding data for ignored files,
       but that is just speculation at this point.


Now, with that context in place:

[1] Nothing prohibits us from having all three options be available
     on a platform.  They should all be able to coexist.

[2] One of my stated goals was to reduce the dependency on
     third-party tools -- especially on platforms that don't have
     a simple package management system.  The point here was to
     make it easier for enterprises to deploy Git to 1k's or 10k's
     of users (and possibly unattended build machines) and make use
     of the feature without *also* having to deploy and track updates
     to yet-another third-party tool or otherwise complicate their ES
     deployment setups.  Only option 3 gets rid of the third-party
     tool requirement.

[3] Option 2 is a valuable suggestion, don't get me wrong.  It can/
     will/should improve performance over option 1 by eliminating an
     extra process creation and the overhead of pumping all of that
     data thru another socket-pair/process and all of the context
     switches that that requires.

[4] Option 2 and option 3 could/should perform relatively equally.
     And if we wanted to deprecate the hook-like interface, doing
     an option 2 implementation would allow us to transition the
     platforms for which I don't currently have a backend.

[5] However, option 2 does not eliminate the need for a third-party
     tool, so it is of limited interest to me at this time.  Yes, it
     would be nice to have it for testing and perf testing purposes
     and comparisons with option 3, but if I have to budget my time,
     I would rather spend my efforts on additional backends.

     I consider the question of doing option 2 and a Linux backend
     as two completely independent topics -- topics that we can
     discuss and/or pursue in parallel if there is interest.

[6] Randall's question was about doing option 3 and I hope that I
     provided helpful information should he or anyone else want to
     pick up that effort before I can.

[7] If you want to start a parallel conversation on option 2, let's
     do that in a new top-level email thread.

Cheers,
Jeff