Bug 203981

Summary: restart blocks update
Product: [Fedora] Fedora Reporter: jmccann
Component: gdmAssignee: Ray Strode [halfline] <rstrode>
Status: CLOSED NOTABUG QA Contact:
Severity: high Docs Contact:
Priority: medium    
Version: rawhideCC: cschalle
Target Milestone: ---   
Target Release: ---   
Hardware: All   
OS: Linux   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2007-03-06 20:19:20 UTC Type: ---
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:

Description jmccann 2006-08-24 19:39:30 UTC
When I performed a yum update on rawhide today, from VT2, it hung while updating
gdm.  When I checked VT7 for some reason xorg (probably nvidia) didn't restart
properly and was asking me for input.  After I give input then the yum update
continued.  This is a problem when doing unattended updates.

The %post does a /usr/sbin/gdm-safe-restart which does:
exec kill -USR1 `cat $PIDFILE`

The builtin kill (ie. bash's jobs.c:kill_pid()) seems to just wrap kill(2).

In the gdm daemon USR1 seems to be handled by mainloop_sig_callback() which
calls gdm_safe_restart().  And if no one is logged in then does gdm_restart_now():

static void
gdm_restart_now (void)
{
	gdm_info (_("GDM restarting ..."));
	gdm_final_cleanup ();
	gdm_restoreenv ();
	VE_IGNORE_EINTR (execvp (stored_argv[0], stored_argv));
	gdm_error (_("Failed to restart self"));
	_exit (1);
}

Comment 1 jmccann 2006-08-24 20:23:27 UTC
Hmm, I don't see anything that would block there.  Perhaps, there was already an
error prompt on VT7 before the restart was issued.

Comment 2 Ray Strode [halfline] 2007-03-06 20:10:30 UTC
hmm, do you have any idea how to reproduce this? 

Comment 3 jmccann 2007-03-06 20:19:20 UTC
Not really, unfortunately.  GDM is pretty synchronous in places so I guess it
may have been blocking somewhere waiting for input...

You can probably close the bug as worksforme or whatever if you want.  Hopefully
this is just a freak thing.