Bug 24972

(In reply to comment #0)
> hardware:
> intel gma945
> 
> software:
> intel 2.9.1 drivers
> xorg 1.7.1
> Mesa 7.6
> kernel 2.6.31 (tested 2.6.32-rc6)
> the issue is only apparent after a suspend. when booting with display attached,
> everything works as it should.
> 
> if i have the dual screen setup, and i suspend, after resume 100% it will die
> back to gdm login screen
> 
> dmesg comes out clean. so does Xorg.0.log
> 
> my xorg.conf has a modeline set since this monitor lost its hability to provide
> an EDID, but removing it didnt fix the issue.
> 
> any tests i can try out to further debug this?
> 
> thanks
> 

Xorg 1.7.1.901 appears to have fixed the issue

(In reply to comment #1)
> (In reply to comment #0)
> > hardware:
> > intel gma945
> > 
> > software:
> > intel 2.9.1 drivers
> > xorg 1.7.1
> > Mesa 7.6
> > kernel 2.6.31 (tested 2.6.32-rc6)
> > the issue is only apparent after a suspend. when booting with display attached,
> > everything works as it should.
> > 
> > if i have the dual screen setup, and i suspend, after resume 100% it will die
> > back to gdm login screen
> > 
> > dmesg comes out clean. so does Xorg.0.log
> > 
> > my xorg.conf has a modeline set since this monitor lost its hability to provide
> > an EDID, but removing it didnt fix the issue.
> > 
> > any tests i can try out to further debug this?
> > 
> > thanks
> > 
> 
> Xorg 1.7.1.901 appears to have fixed the issue
> 

i guess i jumped too early into a conclusion...the issue is still there.

May be a dupe of http://bugzilla.kernel.org/show_bug.cgi?id=14484...

(In reply to comment #3)
> May be a dupe of http://bugzilla.kernel.org/show_bug.cgi?id=14484...
> 

hmm, that report is for kernel 2.6.32-rc5

havent tested the 32 series yet here...

ive used archlinux's 2.6.31 default package, and my own 2.6.31-brain fuck scheduler.

both kernels have the same issue..

this worked correctly with the kernel im using now, and xserver 1.6

i cannot reproduce this issue with the VGA port disconnected (LVDS only).

under an active session, plugging an exteran monitor will most likely freeze the entire thing (sysrq-REISUB) to reboot.

some times, sysreq-k or ctrl-alt-space will suffice.

everything smells of the intel driver or mesa and how they handle the VGA port.
(these were updated with xorg) at the time. i guess i could narrow it down building an older mesa, all this is comming to mind as i type, sorry for adding lag to something i should have done before... will report back when i do ;)

one extra note.

if i log out of X (to GDM or a tty) and plug the VGA port, and log in again, it works ok,

downgraded:
mesa 7.5.2
libgl 7.5.2
intel-dri 7.5.2

kept:
xerver 1.7
xf86-video-intel 2.9.1


the issue is gone with this configuration.

Any updates on this?  We've had several hotplug fixes land in 2.6.33-rc, so if this isn't already fixed you could try that.  X and your desktop also handle hotplug events, so it's worth updating them too.

Please reopen if things are still broken.  If it's a kernel issue a bisect would help track it down (git help bisect will tell you how).