git.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [PATCH v2] Clear fd after closing to avoid double-close error
@ 2013-10-22 13:36 Jens Lindström
  2013-10-22 15:29 ` Jeff King
  2013-10-23 13:00 ` Duy Nguyen
  0 siblings, 2 replies; 3+ messages in thread
From: Jens Lindström @ 2013-10-22 13:36 UTC (permalink / raw)
  To: git; +Cc: Duy Nguyen

From: Jens Lindstrom <jl@opera.com>

In send_pack(), clear the fd passed to pack_objects() by setting
it to -1, since pack_objects() closes the fd (via a call to
run_command()).  Likewise, in get_pack(), clear the fd passed to
run_command().

Not doing so risks having git_transport_push(), caller of
send_pack(), closing the fd again, possibly incorrectly closing
some other open file; or similarly with fetch_refs_from_pack(),
indirect caller of get_pack().

Signed-off-by: Jens Lindström <jl@opera.com>
---
 fetch-pack.c | 4 ++++
 send-pack.c  | 4 ++++
 2 files changed, 8 insertions(+)

diff --git a/fetch-pack.c b/fetch-pack.c
index f5d99c1..29b711a 100644
--- a/fetch-pack.c
+++ b/fetch-pack.c
@@ -776,6 +776,10 @@ static int get_pack(struct fetch_pack_args *args,
 		close(cmd.out);
 	}
 
+	if (!use_sideband)
+		/* Closed by start_command() */
+		xd[0] = -1;
+
 	ret = finish_command(&cmd);
 	if (!ret || (args->check_self_contained_and_connected && ret == 1))
 		args->self_contained_and_connected =
diff --git a/send-pack.c b/send-pack.c
index 7d172ef..edbfd07 100644
--- a/send-pack.c
+++ b/send-pack.c
@@ -300,8 +300,12 @@ int send_pack(struct send_pack_args *args,
 				shutdown(fd[0], SHUT_WR);
 			if (use_sideband)
 				finish_async(&demux);
+			fd[1] = -1;
 			return -1;
 		}
+		if (!args->stateless_rpc)
+			/* Closed by pack_objects() via start_command() */
+			fd[1] = -1;
 	}
 	if (args->stateless_rpc && cmds_sent)
 		packet_flush(out);
-- 
1.8.1.2

^ permalink raw reply related	[flat|nested] 3+ messages in thread

* Re: [PATCH v2] Clear fd after closing to avoid double-close error
  2013-10-22 13:36 [PATCH v2] Clear fd after closing to avoid double-close error Jens Lindström
@ 2013-10-22 15:29 ` Jeff King
  2013-10-23 13:00 ` Duy Nguyen
  1 sibling, 0 replies; 3+ messages in thread
From: Jeff King @ 2013-10-22 15:29 UTC (permalink / raw)
  To: Jens Lindström; +Cc: git, Duy Nguyen

On Tue, Oct 22, 2013 at 03:36:02PM +0200, Jens Lindström wrote:

> In send_pack(), clear the fd passed to pack_objects() by setting
> it to -1, since pack_objects() closes the fd (via a call to
> run_command()).  Likewise, in get_pack(), clear the fd passed to
> run_command().
> 
> Not doing so risks having git_transport_push(), caller of
> send_pack(), closing the fd again, possibly incorrectly closing
> some other open file; or similarly with fetch_refs_from_pack(),
> indirect caller of get_pack().

Thanks, this looks like the right thing to do.

You had mentioned earlier in the thread the possibility of just dropping
the close() in git_transport_push. I agree that it is not worth going
that route, as it serves as a safety check to avoid leaking in early
returns (and indeed, in both the fetch and push cases, we can return
early without spawning a subprocess when the receiver already has all of
the necessary objects).

> Signed-off-by: Jens Lindström <jl@opera.com>

Acked-by: Jeff King <peff@peff.net>

-Peff

^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: [PATCH v2] Clear fd after closing to avoid double-close error
  2013-10-22 13:36 [PATCH v2] Clear fd after closing to avoid double-close error Jens Lindström
  2013-10-22 15:29 ` Jeff King
@ 2013-10-23 13:00 ` Duy Nguyen
  1 sibling, 0 replies; 3+ messages in thread
From: Duy Nguyen @ 2013-10-23 13:00 UTC (permalink / raw)
  To: Jens Lindström; +Cc: Git Mailing List

On Tue, Oct 22, 2013 at 8:36 PM, Jens Lindström <jl@opera.com> wrote:
> From: Jens Lindstrom <jl@opera.com>
>
> In send_pack(), clear the fd passed to pack_objects() by setting
> it to -1, since pack_objects() closes the fd (via a call to
> run_command()).  Likewise, in get_pack(), clear the fd passed to
> run_command().
>
> Not doing so risks having git_transport_push(), caller of
> send_pack(), closing the fd again, possibly incorrectly closing
> some other open file; or similarly with fetch_refs_from_pack(),
> indirect caller of get_pack().
>
> Signed-off-by: Jens Lindström <jl@opera.com>
> ---
>  fetch-pack.c | 4 ++++
>  send-pack.c  | 4 ++++
>  2 files changed, 8 insertions(+)
>
> diff --git a/fetch-pack.c b/fetch-pack.c
> index f5d99c1..29b711a 100644
> --- a/fetch-pack.c
> +++ b/fetch-pack.c
> @@ -776,6 +776,10 @@ static int get_pack(struct fetch_pack_args *args,
>                 close(cmd.out);
>         }
>
> +       if (!use_sideband)
> +               /* Closed by start_command() */
> +               xd[0] = -1;
> +

Further up demux.out still holds this handle. But we don't actually
start_async when use_sideband is false so demux.out can't be used
anywhere else. So it's good.

Acked-by: me

>         ret = finish_command(&cmd);
>         if (!ret || (args->check_self_contained_and_connected && ret == 1))
>                 args->self_contained_and_connected =
> diff --git a/send-pack.c b/send-pack.c
> index 7d172ef..edbfd07 100644
> --- a/send-pack.c
> +++ b/send-pack.c
> @@ -300,8 +300,12 @@ int send_pack(struct send_pack_args *args,
>                                 shutdown(fd[0], SHUT_WR);
>                         if (use_sideband)
>                                 finish_async(&demux);
> +                       fd[1] = -1;
>                         return -1;
>                 }
> +               if (!args->stateless_rpc)
> +                       /* Closed by pack_objects() via start_command() */
> +                       fd[1] = -1;
>         }
>         if (args->stateless_rpc && cmds_sent)
>                 packet_flush(out);
> --
> 1.8.1.2
>



-- 
Duy

^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2013-10-23 13:00 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2013-10-22 13:36 [PATCH v2] Clear fd after closing to avoid double-close error Jens Lindström
2013-10-22 15:29 ` Jeff King
2013-10-23 13:00 ` Duy Nguyen

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).