Re: bug#70411: [bug] install(1) fails to read /dev/stdin on Darwin

Re: bug#70411: [bug] install(1) fails to read /dev/stdin on Darwin

[Paul Eggert]

[-- Attachment #1: Type: text/plain, Size: 1329 bytes --]

On 2024-04-18 14:52, Sergei Trofimovich wrote:
> $ clang simple.c -o simple && echo 42 | ./simple
> 1: ino=3009428657538693161
> 2: ino=3009428657538693161
> 3: ino=1568241705
> 
> Note how stat() and fstat() don't agree on inode.
> 
> Apparently it's documented in
> https://developer.apple.com/library/archive/documentation/System/Conceptual/ManPages_iPhoneOS/man2/fstat.2.html
> as
> 
>    BUGS
>       Applying fstat to a socket (and thus to a pipe) returns a zero'd buffer,
>       except for the blocksize field, and a unique device and inode number.

The BUGS note simply means that a pipe has a unique inode number, which 
is what we want. So that's not indicating any problem.


Oh, I see the problem now.  For a socket or pipe, macOS fstat returns 
the full 64-bit inode number, whereas macOS stat returns only the low 
order 32 bits.  In your example, 3009428657538693161 % (2**32) == 
1568241705.

This is a kernel bug in macOS. Can you report it or otherwise arrange to 
have the kernel bug fixed? I expect that you have better connections 
with Apple than I do. A proposed patch (relative to xnu-10063.101.15) is 
attached; I have not tested it as I don't use macOS. Thanks.

Also, I am documenting this macOS bug in Gnulib by installing the second 
attached patch to Gnulib, and am cc'ing this email to bug-gnulib.

[-- Attachment #2: 0001-Fix-bug-with-stat-truncating-pipe-socket-st_ino.patch --]
[-- Type: text/x-patch, Size: 1330 bytes --]

From 29345117a4cf85aceb88e3901758b19a4867062e Mon Sep 17 00:00:00 2001
From: Paul Eggert <eggert@cs.ucla.edu>
Date: Fri, 19 Apr 2024 00:12:50 -0700
Subject: [PATCH] Fix bug with stat truncating pipe/socket st_ino

Problem reported by Sergei Trofimovich in:
https://bugs.gnu.org/70411
https://github.com/NixOS/nixpkgs/pull/300797
* bsd/miscfs/devfs/devfs_fdesc_support.c (fdesc_attr):
Do not truncate inode numbers to 32 bits.
---
 bsd/miscfs/devfs/devfs_fdesc_support.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/bsd/miscfs/devfs/devfs_fdesc_support.c b/bsd/miscfs/devfs/devfs_fdesc_support.c
index a17c6992..b4a55103 100644
--- a/bsd/miscfs/devfs/devfs_fdesc_support.c
+++ b/bsd/miscfs/devfs/devfs_fdesc_support.c
@@ -437,10 +437,10 @@ fdesc_attr(int fd, struct vnode_attr *vap, vfs_context_t a_context)
 	case DTYPE_PIPE:
 #if SOCKETS
 		if (FILEGLOB_DTYPE(fp->fp_glob) == DTYPE_SOCKET) {
-			error = soo_stat((struct socket *)fp_get_data(fp), (void *)&stb, 0);
+			error = soo_stat((struct socket *)fp_get_data(fp), (void *)&stb, 1);
 		} else
 #endif /* SOCKETS */
-		error = pipe_stat((struct pipe *)fp_get_data(fp), (void *)&stb, 0);
+		error = pipe_stat((struct pipe *)fp_get_data(fp), (void *)&stb, 1);
 
 		if (error == 0) {
 			if (FILEGLOB_DTYPE(fp->fp_glob) == DTYPE_SOCKET) {
-- 
2.44.0


[-- Attachment #3: 0001-fstatat-stat-document-macOS-st_ino-pipe-bug.patch --]
[-- Type: text/x-patch, Size: 2437 bytes --]

From c2174a623d33096b52f4d7fd2963f76acb3e301f Mon Sep 17 00:00:00 2001
From: Paul Eggert <eggert@cs.ucla.edu>
Date: Fri, 19 Apr 2024 00:29:32 -0700
Subject: [PATCH] fstatat, stat: document macOS st_ino pipe bug

* doc/posix-functions/fstatat.texi (fstatat):
* doc/posix-functions/stat.texi (stat):
Document macOS bug (see <https://bugs.gnu.org/70411>).
---
 ChangeLog                        | 7 +++++++
 doc/posix-functions/fstatat.texi | 5 +++++
 doc/posix-functions/stat.texi    | 5 +++++
 3 files changed, 17 insertions(+)

diff --git a/ChangeLog b/ChangeLog
index 7ce75a98a9..1667f90c55 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,10 @@
+2024-04-19  Paul Eggert  <eggert@cs.ucla.edu>
+
+	fstatat, stat: document macOS st_ino pipe bug
+	* doc/posix-functions/fstatat.texi (fstatat):
+	* doc/posix-functions/stat.texi (stat):
+	Document macOS bug (see <https://bugs.gnu.org/70411>).
+
 2024-04-18  Bruno Haible  <bruno@clisp.org>
 
 	totalordermagl: Add tests.
diff --git a/doc/posix-functions/fstatat.texi b/doc/posix-functions/fstatat.texi
index e959a5cc73..90884e2eb1 100644
--- a/doc/posix-functions/fstatat.texi
+++ b/doc/posix-functions/fstatat.texi
@@ -40,5 +40,10 @@ This function does not fail when the second argument is an empty string
 on some platforms, even when @code{AT_EMPTY_PATH} is not used:
 glibc 2.7, Linux 2.6.38.
 @item
+This function sets @code{st_ino} only to the low-order 32 bits of
+the inode number of a socket or pipe, which thus can disagree
+with the @code{st_ino} obtained by @code{fstat}:
+macOS 14.
+@item
 @xref{sys/stat.h}, for general portability problems with @code{struct stat}.
 @end itemize
diff --git a/doc/posix-functions/stat.texi b/doc/posix-functions/stat.texi
index f655451392..8afd3b17bb 100644
--- a/doc/posix-functions/stat.texi
+++ b/doc/posix-functions/stat.texi
@@ -50,6 +50,11 @@ Portability problems not fixed by Gnulib:
 Cygwin's @code{stat} function sometimes sets @code{errno} to @code{EACCES} when
 @code{ENOENT} would be more appropriate.
 @item
+This function sets @code{st_ino} only to the low-order 32 bits of
+the inode number of a socket or pipe, which thus can disagree
+with the @code{st_ino} obtained by @code{fstat}:
+macOS 14.
+@item
 Because of the definition of @code{struct stat}, it is not possible to
 portably replace @code{stat} via an object-like macro.  Therefore,
 expressions such as @code{(islnk ? lstat : stat) (name, buf)} are not
-- 
2.40.1

Re: bug#70411: [bug] install(1) fails to read /dev/stdin on Darwin

[Sergei Trofimovich]

On Fri, 19 Apr 2024 00:33:52 -0700
Paul Eggert <eggert@cs.ucla.edu> wrote:

> On 2024-04-18 14:52, Sergei Trofimovich wrote:
> > $ clang simple.c -o simple && echo 42 | ./simple
> > 1: ino=3009428657538693161
> > 2: ino=3009428657538693161
> > 3: ino=1568241705
> > 
> > Note how stat() and fstat() don't agree on inode.
> > 
> > Apparently it's documented in
> > https://developer.apple.com/library/archive/documentation/System/Conceptual/ManPages_iPhoneOS/man2/fstat.2.html
> > as
> > 
> >    BUGS
> >       Applying fstat to a socket (and thus to a pipe) returns a zero'd buffer,
> >       except for the blocksize field, and a unique device and inode number.  
> 
> The BUGS note simply means that a pipe has a unique inode number, which 
> is what we want. So that's not indicating any problem.
> 
> 
> Oh, I see the problem now.  For a socket or pipe, macOS fstat returns 
> the full 64-bit inode number, whereas macOS stat returns only the low 
> order 32 bits.  In your example, 3009428657538693161 % (2**32) == 
> 1568241705.
> 
> This is a kernel bug in macOS. Can you report it or otherwise arrange to 
> have the kernel bug fixed? I expect that you have better connections 
> with Apple than I do. A proposed patch (relative to xnu-10063.101.15) is 
> attached; I have not tested it as I don't use macOS. Thanks.

I reported it via https://www.apple.com/feedback/macos.html

> Also, I am documenting this macOS bug in Gnulib by installing the second 
> attached patch to Gnulib, and am cc'ing this email to bug-gnulib.

Thank you, Paul!

-- 

  Sergei