Bug 679404

Summary: mutter crashes, flickering on graphics test day iso image
Product: [Fedora] Fedora Reporter: cam <camilo>
Component: mutterAssignee: Peter Robinson <pbrobinson>
Status: CLOSED DUPLICATE QA Contact: Fedora Extras Quality Assurance <extras-qa>
Severity: unspecified Docs Contact:
Priority: unspecified    
Version: 15CC: airlied, ajax, awilliam, bskeggs, maxamillion, mcepl, mcepl, otaylor, pbrobinson, walters
Target Milestone: ---   
Target Release: ---   
Hardware: Unspecified   
OS: Unspecified   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2011-02-28 06:09:11 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:
Attachments:
Description Flags
Xorg log
none
xsession errors
none
dmesg output
none
log of gdb showing backtrace
none
gdb log showing detailed backtrace none

Description cam 2011-02-22 13:53:43 UTC
Created attachment 480135 [details]
Xorg log

Description of problem:
Mutter crashes repeatedly after login

Version-Release number of selected component (if applicable):
graphics test day iso version

How reproducible:
100% (two out of two boots)

Steps to Reproduce:
1.boot image from usb
2.click on live user login
3.observe flickering
  
Actual results:
flickering and no usable desktop

Expected results:
usable desktop

Additional info:
logs attached as requested on irc

Comment 1 cam 2011-02-22 13:55:15 UTC
Created attachment 480136 [details]
xsession errors

Comment 2 cam 2011-02-22 13:55:51 UTC
Created attachment 480137 [details]
dmesg output

Comment 3 Matěj Cepl 2011-02-22 15:22:06 UTC
I don't see anything else in the logs than

repeated crash in mutter-2.91.6-4.fc15, not showing the notification

reassigned to mutter for further investigation.

Comment 4 Owen Taylor 2011-02-22 16:15:18 UTC
Since it isn't happening for everybody with the live CD, we can strongly guess this is a problem with the Nouveau drivers rather than Mutter, maybe the same as bug 671294

Comment 5 Ben Skeggs 2011-02-22 22:31:00 UTC
I'm not seeing any good reason for this either.  The chipset is NVAC, thus uses the nv50 driver paths, which have also shown to be working.

Comment 6 Ben Skeggs 2011-02-22 22:31:29 UTC
Without a backtrace or something to see where this is messing up, it'll be hard to track down.

Comment 7 cam 2011-02-22 23:44:26 UTC
For what it's worth, I tested a Gnome3 live CD and it worked OK, no crashing mutter.

Comment 8 cam 2011-02-23 22:21:22 UTC
Created attachment 480590 [details]
log of gdb showing backtrace

Comment 9 cam 2011-02-23 22:21:56 UTC
I tried the latest Fedora 15 Desktop Alpha iso and saw much the same problem.
However the desktop seemed to eventually go into a fallback mode of some kind where mutter wasn't running at all.

I manually downloaded some debuginfos by hook or by crook and opened a terminal, gave ulimit -c unlimited and ran mutter --replace, which crashed fairly promptly.

I opened mutter and the core file with gdb.

I did this twice to see if there was variation in the traces (they look v similar to me).

I have attached the backtraces, hope they help.

Comment 10 cam 2011-02-23 22:22:31 UTC
Created attachment 480591 [details]
gdb log showing detailed backtrace

Comment 11 Adam Williamson 2011-02-25 20:16:03 UTC
are you using the 32-bit or 64-bit image? if 32-bit, can you test with 64, or is the system incapable of it?

Comment 12 cam 2011-02-25 20:51:15 UTC
I was using the 32 bit image, I have this:

cpu family	: 6
model		: 28
model name	: Intel(R) Atom(TM) CPU N270   @ 1.60GHz

I believe it isn't able to run 64-bit code

Comment 13 Adam Williamson 2011-02-25 21:01:20 UTC
yeah, it's not :/

ben, do you think this is a case of the 32-bit mesa crash?

Comment 14 Ben Skeggs 2011-02-27 22:45:12 UTC
It seems likely, yes.

Comment 15 cam 2011-02-27 22:54:35 UTC
This problem has been introduced recently it seems - is the same true of the '32-bit mesa crash'?

* not present in F14
* not present in Gnome 3 gnome3_test_20110203_i686.iso

* present in:
Fedora-15-Alpha-i686-Live-Desktop.iso
gfx_test_week_20110221_i386.iso

Is there anything else I can do to help test / diagnose the problem?

Comment 16 Adam Williamson 2011-02-28 01:11:00 UTC
"This problem has been introduced recently it seems - is the same true of the
'32-bit mesa crash'?"

yes.

"Is there anything else I can do to help test / diagnose the problem?"

try alpha rc2 - http://serverbeach1.fedoraproject.org/pub/alt/stage/15-Alpha.RC2/ .

Comment 17 cam 2011-02-28 05:55:55 UTC
I tried once and it behaved differently to RC1.

Logging in from GDM seemed OK, but the desktop seemed to be running in a fallback mode (menus and tray). Ps showed mutter wasn't running. Starting it with mutter --replace didn't crash.
Closer inspection of /var/log/messages showed some desktop related errors.  I was able to get the networking started but the system locked up before I could copy the log off. I will try again later.

I also noticed a messsage at early boot time about a PCI mapping conflict.

Any other advice?

Comment 18 Adam Williamson 2011-02-28 06:09:11 UTC
"but the desktop seemed to be running in a fallback mode (menus and tray)"

that's https://bugzilla.redhat.com/show_bug.cgi?id=679326 . it sounds like this is the 32-bit issue.

*** This bug has been marked as a duplicate of bug 679924 ***

Comment 19 cam 2011-02-28 10:26:41 UTC
Having read through 679326 it does sound like what I was seeing on the desktop and in the logs. 

Thanks for the attention and prompt replies!