[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Message-ID: <20121015171437.32e9a4ac@mj>
Date: Mon, 15 Oct 2012 17:14:37 -0400
From: Pavel Roskin <proski@....org>
To: Al Viro <viro@...iv.linux.org.uk>, linux-kernel@...r.kernel.org
Subject: Bisected regression in Linux 3.7.0-rc1, hang on login caused by
replace_fd()
Hello!
I tried the current mainline Linux on Fedora 16 x64_64 and found that I
cannot login in gdm. I'm using gdm with LXDE. After a few minutes,
kernel messages appear indicating a hung process gnome-settings-daemon.
Bisection found that the commit to blame is
commit 8280d16172243702ed43432f826ca6130edb4086
Author: Al Viro <viro@...iv.linux.org.uk>
Date: Tue Aug 21 12:11:46 2012 -0400
new helper: replace_fd()
analog of dup2(), except that it takes struct file * as source.
Signed-off-by: Al Viro <viro@...iv.linux.org.uk>
Reverting the part of the patch that changes fs/exec.c is sufficient to
fix the login behavior.
I was able to forward port the change to the current mainline Linux.
The call to replace_fd() was moved to fs/coredump.c, so the patch
changes it. Also, it copies __clear_close_on_exec() and
__set_open_fd() from fs/file.c
The .config file, the kernel log and the forward ported patch are
attached.
I'll try to figure out why replace_fd() is not good in
umh_pipe_setup(), but I'd like to report the problem as soon as
possible, as it's very visible to the users and I don't want others to
spend hours bisecting it.
--
Regards,
Pavel Roskin
Download attachment "dotconfig" of type "application/octet-stream" (78210 bytes)
View attachment "kernel.log" of type "text/x-log" (194528 bytes)
View attachment "revert-replace_fd.patch" of type "text/x-patch" (2659 bytes)
Powered by blists - more mailing lists