git@vger.kernel.org mailing list mirror (one of many)
 help / color / mirror / code / Atom feed
From: Jonathan Nieder <jrnieder@gmail.com>
To: David Barr <davidbarr@google.com>
Cc: Junio C Hamano <gitster@pobox.com>,
	Git Mailing List <git@vger.kernel.org>,
	Andrew Sayers <andrew-git@pileofstuff.org>,
	Dmitry Ivankov <divanorama@gmail.com>,
	Sverre Rabbelier <srabbelier@gmail.com>
Subject: Re: [PATCH 2/2] fast-import: teach ls command to accept empty path
Date: Fri, 9 Mar 2012 03:29:40 -0600	[thread overview]
Message-ID: <20120309092940.GB2229@burratino> (raw)
In-Reply-To: <CAFfmPPNnQ21qwKb_w1FCRL7Vx7CSQKYurM2zqziTw01kkRoMog@mail.gmail.com>

David Barr wrote:
> On Fri, Mar 9, 2012 at 7:33 AM, Jonathan Nieder <jrnieder@gmail.com> wrote:

>> +               /*
>> +                * store_tree scribbles over version[0] in leaf.tree's
>> +                * entries, so we need a deep copy.
>> +                */
>> +               if (root->tree && is_null_sha1(root->versions[1].sha1))
>> +                       leaf.tree = dup_tree_content(root->tree);
>
> Is it ok to call store_tree(root)?

Yes.

>                                     If so, could we not introduce a
> pointer rather than a deep copy?

If using 'ls' with an empty path after dirtying the root tree is
common, then that would work as an optimization.  The fussy bit is
making sure the call to

	release_tree_content_recursive(leaf.tree);

is skipped in this case and not skipped when tree_content_get() made a
copy.  That is, something like this (patch against fast-import-pu on
repo.or.cz/git/jrn.git):

-- >8 --
From: David Barr <davidbarr@google.com>
Subject: fast-import: optimize 'ls' command with empty path to avoid a copy

fast-import's "ls" command normally copies a tree (implicitly, by
calling tree_content_get) before passing it to store_tree.  Otherwise:

 - after versions[0] is overwritten by versions[1] in child
   directories, it would be impossible to rebuild the tree object for
   version 0 of the current tree, so parse_ls would need to

	hashcpy(leaf.versions[0].sha1, leaf.versions[1].sha1)

   so version 0 points to a tree that can be rebuilt.

 - in turn, that would make it impossible to rebuild the tree object
   for version 0 of the parent tree.  And so on.

The above considerations do not apply when the tree we are examining
with 'ls' has no parent.  Avoid a copy in that case.

Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
---
 fast-import.c |   22 +++++++++++++++-------
 1 file changed, 15 insertions(+), 7 deletions(-)

diff --git a/fast-import.c b/fast-import.c
index 1e5d59b4..28fe4c35 100644
--- a/fast-import.c
+++ b/fast-import.c
@@ -3002,7 +3002,8 @@ static void parse_ls(struct branch *b)
 {
 	const char *p;
 	struct tree_entry *root = NULL;
-	struct tree_entry leaf = {NULL};
+	struct tree_entry leaf_storage = {NULL};
+	struct tree_entry *leaf = &leaf_storage;
 
 	/* ls SP (<treeish> SP)? <path> */
 	p = command_buf.buf + strlen("ls ");
@@ -3029,17 +3030,24 @@ static void parse_ls(struct branch *b)
 			die("Garbage after path in: %s", command_buf.buf);
 		p = uq.buf;
 	}
-	tree_content_get(root, p, &leaf);
+	if (*p)
+		tree_content_get(root, p, leaf);
+	else
+		leaf = root;
+
 	/*
 	 * A directory in preparation would have a sha1 of zero
 	 * until it is saved.  Save, for simplicity.
 	 */
-	if (S_ISDIR(leaf.versions[1].mode))
-		store_tree(&leaf);
+	if (S_ISDIR(leaf->versions[1].mode)
+	    && is_null_sha1(leaf->versions[1].sha1)) {
+		store_tree(leaf);
+		hashcpy(leaf->versions[0].sha1, leaf->versions[1].sha1);
+	}
 
-	print_ls(leaf.versions[1].mode, leaf.versions[1].sha1, p);
-	if (leaf.tree)
-		release_tree_content_recursive(leaf.tree);
+	print_ls(leaf->versions[1].mode, leaf->versions[1].sha1, p);
+	if (*p && leaf->tree)
+		release_tree_content_recursive(leaf->tree);
 	if (!b || root != &b->branch_tree)
 		release_tree_entry(root);
 }
-- 
1.7.9.2

  reply	other threads:[~2012-03-09  9:30 UTC|newest]

Thread overview: 23+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2012-03-08  3:13 [BUG] fast-import: ls command on commit root returns missing (was: Bug in svn-fe: copying the root directory acts as if it's an empty directory) David Barr
2012-03-08  5:30 ` [PATCH] fast-import: fix ls command with empty path David Barr
2012-03-08  7:09   ` Jonathan Nieder
2012-03-08 15:29     ` Sverre Rabbelier
2012-03-08 16:39     ` Junio C Hamano
2012-03-08 17:33       ` Dmitry Ivankov
2012-03-08 19:32         ` Jonathan Nieder
2012-03-10  9:48         ` Jonathan Nieder
2012-03-08 18:15       ` Jonathan Nieder
2012-03-10  3:12       ` Jonathan Nieder
2012-03-10  3:20         ` [PATCH maint-1.7.6] fast-import: leakfix for 'ls' of dirty trees Jonathan Nieder
2012-03-10  4:00         ` [PATCH maint-1.7.6] fast-import: don't allow 'ls' of path with empty components Jonathan Nieder
2012-03-10  4:30         ` [PULL maint] two fast-import "ls" fixes Jonathan Nieder
2012-03-10  7:00         ` [NON-PATCH] vcs-svn: avoid 'ls' and filedelete with empty path Jonathan Nieder
2013-06-21 16:33         ` [PATCH] fast-import: fix ls command " Dave Abrahams
2012-03-08 20:27   ` [PATCH v2 0/2] " Jonathan Nieder
2012-03-08 20:31     ` [PATCH 1/2] fast-import: plug leak of dirty trees in 'ls' command Jonathan Nieder
2012-03-08 20:33     ` [PATCH 2/2] fast-import: teach ls command to accept empty path Jonathan Nieder
2012-03-09  4:28       ` David Barr
2012-03-09  9:29         ` Jonathan Nieder [this message]
2012-03-10  8:53   ` [PATCH v3] fast-import: allow 'ls' and filecopy to read the root Jonathan Nieder
2012-03-10  9:01     ` Jonathan Nieder
2012-03-10  9:18     ` [PATCH 2/1] fixup! " Jonathan Nieder

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

  List information: http://vger.kernel.org/majordomo-info.html

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20120309092940.GB2229@burratino \
    --to=jrnieder@gmail.com \
    --cc=andrew-git@pileofstuff.org \
    --cc=davidbarr@google.com \
    --cc=divanorama@gmail.com \
    --cc=git@vger.kernel.org \
    --cc=gitster@pobox.com \
    --cc=srabbelier@gmail.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
Code repositories for project(s) associated with this public inbox

	https://80x24.org/mirrors/git.git

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).