git.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Linus Torvalds <torvalds@linux-foundation.org>
To: Brandon Casey <casey@nrlssc.navy.mil>
Cc: "Git Mailing List" <git@vger.kernel.org>,
	drafnel@gmail.com, "Junio C Hamano" <gitster@pobox.com>,
	"Alex Riesen" <raa.lkml@gmail.com>,
	"Kristian Høgsberg" <krh@redhat.com>
Subject: Re: git-commit fatal: Out of memory? mmap failed: Bad file descriptor
Date: Tue, 15 Jan 2008 10:50:19 -0800 (PST)	[thread overview]
Message-ID: <alpine.LFD.1.00.0801151036110.2806@woody.linux-foundation.org> (raw)
In-Reply-To: <478CFAFF.6010006@nrlssc.navy.mil>



On Tue, 15 Jan 2008, Brandon Casey wrote:
>
> Linus Torvalds wrote:
> > It would obviously be interesting to see the base repository and the 
> > commit you are trying to do - is that possibly publicly available?
> 
> I wish it was.

It's ok, I found the bug in your full strace.

The bug really is pretty stupid:

 - prepare_index() does a 

	fd = hold_lock_file_for_update(&false_lock, ...
	...
	if (write_cache(fd, active_cache, active_nr) || close(fd))
		die("unable to write temporary index file");

and the magic here is that *it*closes*the*fd*.

But that's not how "hold_lock_file_for_update()" works. It still has that 
fd squirrelled away in it's "false_lock.fd", and later on, when we do 

	rollback_lock_file(&false_lock);

(in the COMMIT_PARTIAL case of either "commit_index_files()" or 
"rollback_index_files()"), that rollback_lock_file() will do:

	void rollback_lock_file(struct lock_file *lk)
	{
	        if (lk->filename[0]) {
	                close(lk->fd);
	                unlink(lk->filename);
	        }
	        lk->filename[0] = 0;
	}

and now it's trying to close that fd *again* and would normally get a 
EBADF there. But in the meantime, somebody already re-used it for 
something else, and what rollback_lock_file() ends up doing is to just 
close some random file descriptor.

In other words, I'm pretty sure that the bug goes away with this really 
ugly hack. The real problem is that that "false_lockfile" thing simply 
mis-uses the whole lockfile interface. So this is not a pretty fix, but it 
at least should hide the effects of the mis-use of the interface.

In other words: the file descriptor that is returned by the lock_file 
interface functions *MUST*NOT* be closed. But if you violate that rule, 
you'd better make sure that you also fix the effects.

		Linus

---
 builtin-commit.c |    3 +++
 1 files changed, 3 insertions(+), 0 deletions(-)

diff --git a/builtin-commit.c b/builtin-commit.c
index 265ba6b..7a52224 100644
--- a/builtin-commit.c
+++ b/builtin-commit.c
@@ -308,6 +308,9 @@ static char *prepare_index(int argc, const char **argv, const char *prefix)
 
 	if (write_cache(fd, active_cache, active_nr) || close(fd))
 		die("unable to write temporary index file");
+
+	/* We closed the false lock-file fd, make sure we don't do anything else to it */
+	false_lock.fd = -1;
 	return false_lock.filename;
 }
 

  reply	other threads:[~2008-01-15 18:51 UTC|newest]

Thread overview: 39+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2008-01-11 22:11 git-commit fatal: Out of memory? mmap failed: Bad file descriptor Brandon Casey
2008-01-11 22:18 ` Charles Bailey
2008-01-12  4:56   ` Jeff King
2008-01-11 22:19 ` Marco Costalba
2008-01-11 22:47 ` Brandon Casey
2008-01-11 23:48   ` Junio C Hamano
2008-01-12  0:43     ` Brandon Casey
2008-01-12  1:08       ` Junio C Hamano
2008-01-12 20:16   ` Alex Riesen
2008-01-14 23:22     ` Brandon Casey
2008-01-15  2:42 ` Brandon Casey
2008-01-15  5:42   ` Linus Torvalds
2008-01-15 17:26     ` Brandon Casey
2008-01-15 17:36       ` Linus Torvalds
2008-01-15 18:27         ` Brandon Casey
2008-01-15 18:50           ` Linus Torvalds [this message]
2008-01-15 19:43             ` Brandon Casey
2008-01-15 20:00               ` Kristian Høgsberg
2008-01-15 20:27                 ` Brandon Casey
2008-01-15 20:39                 ` Brandon Casey
2008-01-15 20:09               ` Linus Torvalds
2008-01-15 20:20                 ` Linus Torvalds
2008-01-15 23:10             ` Junio C Hamano
2008-01-16  1:27               ` Junio C Hamano
2008-01-16  2:11                 ` Brandon Casey
2008-01-16 19:00                   ` [PATCH 1/2] Document lockfile API Junio C Hamano
2008-01-16 19:05                   ` [PATCH 2/2] close_lock_file(): new function in the " Junio C Hamano
2008-01-16 20:08                     ` Linus Torvalds
2008-01-16 20:36                       ` Junio C Hamano
2008-01-16 20:46                         ` Brandon Casey
2008-01-16 21:57                           ` Junio C Hamano
2008-01-16 22:46                             ` Brandon Casey
2008-01-16 22:55                               ` Junio C Hamano
2008-01-16 23:08                                 ` Brandon Casey
2008-01-16 23:16                                   ` Brandon Casey
2008-01-16 23:19                           ` Junio C Hamano
2008-01-16 23:28                             ` Brandon Casey
2008-01-16  7:53               ` git-commit fatal: Out of memory? mmap failed: Bad file descriptor Johannes Sixt
2008-01-15 12:21   ` Junio C Hamano

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

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

  git send-email \
    --in-reply-to=alpine.LFD.1.00.0801151036110.2806@woody.linux-foundation.org \
    --to=torvalds@linux-foundation.org \
    --cc=casey@nrlssc.navy.mil \
    --cc=drafnel@gmail.com \
    --cc=git@vger.kernel.org \
    --cc=gitster@pobox.com \
    --cc=krh@redhat.com \
    --cc=raa.lkml@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.
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 NNTP newsgroup(s).