git@vger.kernel.org mailing list mirror (one of many)
 help / color / mirror / code / Atom feed
From: Paul Tan <pyokagan@gmail.com>
To: Junio C Hamano <gitster@pobox.com>
Cc: git@vger.kernel.org,
	Johannes Schindelin <johannes.schindelin@gmx.de>,
	Stefan Beller <sbeller@google.com>,
	Remi Lespinet <remi.lespinet@ensimag.grenoble-inp.fr>,
	Jeff King <peff@peff.net>
Subject: Re: [PATCH v2 0/3] am: let command-line options override saved options
Date: Thu, 6 Aug 2015 01:51:17 +0800	[thread overview]
Message-ID: <20150805175117.GA6759@yoshi.chippynet.com> (raw)
In-Reply-To: <xmqq37zx68uf.fsf@gitster.dls.corp.google.com>

On Wed, Aug 05, 2015 at 08:41:44AM -0700, Junio C Hamano wrote:
> Interesting.  This seems to break test under prove.
> 
>     cd t && make T=t4153-am-resume-override-opts.sh prove
> 
> does not seem to return.

The new test-terminal.perl code is the culprit. It seems that if our
wrapped process terminates before our stdin-writing fork does, our
stdin-writing process will stall. I think this occurs with prove because
prove waits until all of its child processes terminate before returning.

So, the solution may be to send a SIGTERM to our stdin-writing fork
should our wrapped process terminate before it does, in order to ensure
that it immediately exits.

The following squash fixes it for me.

Thanks,
Paul

-- >8 --
Subject: [PATCH] squash! test_terminal: redirect child process' stdin to a pty

When the child process terminates before the copy_stdio() finishes
writing all of its data to the child's stdin slave pty, it will stall.

As such, we first move the stdin-pty-writing logic out of copy_stdio()
into its own subroutine copy_stdin() so that we can manage the forked
process ourselves, and then we send SIGTERM to the forked process should
the command we are wrapping terminate before copy_stdin() finishes
writing all of its data to un-stall it.

Signed-off-by: Paul Tan <pyokagan@gmail.com>
---
 t/test-terminal.perl | 27 ++++++++++++++++++---------
 1 file changed, 18 insertions(+), 9 deletions(-)

diff --git a/t/test-terminal.perl b/t/test-terminal.perl
index f6fc9ae..96b6a03 100755
--- a/t/test-terminal.perl
+++ b/t/test-terminal.perl
@@ -51,24 +51,26 @@ sub xsendfile {
 	copy($in, $out, 4096) or $!{EIO} or die "cannot copy from child: $!";
 }
 
-sub copy_stdio {
-	my ($in, $out, $err) = @_;
+sub copy_stdin {
+	my ($in) = @_;
 	my $pid = fork;
 	if (!$pid) {
-		close($out);
-		close($err);
 		xsendfile($in, \*STDIN);
 		exit 0;
 	}
-	$pid = fork;
+	close($in);
+	return $pid;
+}
+
+sub copy_stdio {
+	my ($out, $err) = @_;
+	my $pid = fork;
 	defined $pid or die "fork failed: $!";
 	if (!$pid) {
-		close($in);
 		close($out);
 		xsendfile(\*STDERR, $err);
 		exit 0;
 	}
-	close($in);
 	close($err);
 	xsendfile(\*STDOUT, $out);
 	finish_child($pid) == 0
@@ -91,5 +93,12 @@ my $pid = start_child(\@ARGV, $master_in->slave, $master_out->slave, $master_err
 close $master_in->slave;
 close $master_out->slave;
 close $master_err->slave;
-copy_stdio($master_in, $master_out, $master_err);
-exit(finish_child($pid));
+my $in_pid = copy_stdin($master_in);
+copy_stdio($master_out, $master_err);
+my $ret = finish_child($pid);
+# If the child process terminates before our copy_stdin() process is able to
+# write all of its data to $master_in, the copy_stdin() process could stall.
+# Send SIGTERM to it to ensure it terminates.
+kill 'TERM', $in_pid;
+finish_child($in_pid);
+exit($ret);
-- 
2.5.0.282.gdd6b4b0

      reply	other threads:[~2015-08-05 17:51 UTC|newest]

Thread overview: 26+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-07-24 17:48 "git am" and then "git am -3" regression? Junio C Hamano
2015-07-24 18:09 ` Jeff King
2015-07-26  5:03   ` Paul Tan
2015-07-26  5:21     ` Jeff King
2015-07-27  8:09       ` Matthieu Moy
2015-07-27  8:32         ` Jeff King
2015-07-27 14:21     ` Junio C Hamano
2015-07-28 16:43       ` [PATCH] am: let command-line options override saved options Paul Tan
2015-07-28 16:48         ` Junio C Hamano
2015-07-28 17:09         ` Junio C Hamano
2015-07-31 10:58           ` Paul Tan
2015-07-31 16:04             ` Junio C Hamano
2015-08-01  0:59               ` Paul Tan
2015-08-04 14:05         ` [PATCH v2 0/3] " Paul Tan
2015-08-04 21:12           ` Junio C Hamano
2015-08-04 14:08         ` Paul Tan
2015-08-04 14:08           ` [PATCH v2 1/3] test_terminal: redirect child process' stdin to a pty Paul Tan
2015-08-06 22:15             ` Eric Sunshine
2015-08-12  4:16               ` Paul Tan
2015-08-04 14:08           ` [PATCH v2 2/3] am: let command-line options override saved options Paul Tan
2015-08-04 14:08           ` [PATCH v2 3/3] am: let --signoff override --no-signoff Paul Tan
2015-08-07  9:29             ` Johannes Schindelin
2015-08-12  3:06               ` Paul Tan
2015-08-12  3:07                 ` Paul Tan
2015-08-05 15:41           ` [PATCH v2 0/3] am: let command-line options override saved options Junio C Hamano
2015-08-05 17:51             ` Paul Tan [this message]

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=20150805175117.GA6759@yoshi.chippynet.com \
    --to=pyokagan@gmail.com \
    --cc=git@vger.kernel.org \
    --cc=gitster@pobox.com \
    --cc=johannes.schindelin@gmx.de \
    --cc=peff@peff.net \
    --cc=remi.lespinet@ensimag.grenoble-inp.fr \
    --cc=sbeller@google.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).