Description of problem:
X completely freezes at GDM after booting up 2.6.32-53.el6.x86_64 on a machine running the nouveau driver (GeForce 9600M GT.) System boots fine, no messages that I've been able to locate to this point, but once GDM starts up, the display is completely frozen.
Version-Release number of selected component (if applicable):
Always. Dropping back to -52.el6 and everything works fine again
Steps to Reproduce:
1. Boot up -53.el6.x86_64 on a nouveau-enabled system
Seeing this on another nouveau machine as well (GeForce GTX 285)
Additional note. Although access via ssh is fine, the directly attached keyboard and mouse are completely frozen as well (unable to cycle the capslock/numlock, etc.)
-53 has the following patches pull in..
- [drm] nouveau: cleanup connector/encoder creation (Ben Skeggs) 
- [drm] nouveau: move LVDS detection back to connector detect() time (Ben Skeggs) 
- [drm] nouveau: fix race condition when under memory pressure (Ben Skeggs) 
The 3rd patch listed above (602663) was independently verified in a scratch-build kernel. Just as a data point.
All 3 patches were actually present in the scratch build mentioned.
Can you provide dmesg and Xorg.0.log, hopefully they'll reveal something.
this isn't a nouveau bug,
something has broken hal and input devices.
lshal | grep x11
from a working kernel and non-working kernel are different,
restarting hal fixes the problem, I suspect
cc'd Harald and Richard Hughes.
This may be related to the fix for bug 617526
airlied just confirmed this problem only happens if you rebuild the initrd after having dracut-004-27 installed. Rebuilding the initrd with any other version of dracut works. Moving to dracut...
if I remove the umask in the 0196 patch in dracut it all seems to work okay.
I would be very interested, which files were affected by the umask. Can you find that out for me?
I'll try and reproduce, at a guess some files hal was going to read.
*** Bug 619374 has been marked as a duplicate of this bug. ***
Confirmed. Installing new dracut package and rebuilding the initramfs fixed the issue for me.
+1 for comment 15.
regression from bugfix of bug 617526
*** Bug 619708 has been marked as a duplicate of this bug. ***
*** Bug 620812 has been marked as a duplicate of this bug. ***