FS#31359 - [linux, firefox] hangs the system on a particular website, kernel 3.5.3

Attached to Project: Arch Linux
Opened by grzech (grzech) - Friday, 31 August 2012, 12:15 GMT
Last edited by Andrea Scarpino (BaSh) - Monday, 24 September 2012, 06:50 GMT
Task Type Bug Report
Category Kernel
Status Closed
Assigned To Tobias Powalowski (tpowa)
Jan de Groot (JGC)
Thomas Bächler (brain0)
Ionut Biru (wonder)
Jan Alexander Steffens (heftig)
Architecture x86_64
Severity Critical
Priority Normal
Reported Version
Due in Version Undecided
Due Date Undecided
Percent Complete 100%
Votes 0
Private No

Details

Description:
fresh firefox 15 hangs all system on latest kernel 3.5.3. X crash to console with oops which starts with:
...
kernel BUG at drivers/gpu/drm/ttm/ttm_bo.c:273!
invalid opcode: 0000 [#1] PREEMPT SMP
...

only hard restart helps. probably problem is in graphic system with ATI/AMD DRI driver.
hardware: core i5 2500 with mobile amd/ati 6450 (HP Elitebook 8460p).
same bug observed on own compiled kernel 3.4.10

Additional info:
* package version(s)
* config and/or log files etc.


Steps to reproduce:
start firefox and enter the website: http://schumann.elis.ugent.be/
This task depends upon

Closed by  Andrea Scarpino (BaSh)
Monday, 24 September 2012, 06:50 GMT
Reason for closing:  Fixed
Additional comments about closing:  linux 3.5.4
Comment by Gaetan Bisson (vesath) - Friday, 31 August 2012, 12:50 GMT
No crash for me with an Intel IGP, but the website says "An error occured during shader loading. The following GL error occured: INVALID_OPERATION".
Comment by ogaby (ogaby) - Wednesday, 12 September 2012, 09:37 GMT
With Windows 7 and Firefox 15.0.1, Firefox crashes and restart automatically. I am not sure that this bug has his place here.
Comment by grzech (grzech) - Wednesday, 12 September 2012, 10:21 GMT
Maby firefox is guilty also, but IMHO it's linux kernel problem. Any program from userland shouldn't "touch" kernel space driver in this way.
Comment by Kevin (anonymous_user) - Wednesday, 12 September 2012, 13:42 GMT
No crashes on Linux Firefox with NVIDIA drivers and the demo works fine. On Windows 7, it doesn't crash but it also doesn't work either.

@grzech - Have you tried with the Catalyst drivers?
Comment by grzech (grzech) - Wednesday, 12 September 2012, 14:49 GMT
No. I don't tried Catalyst drivers. I have bad experiences with proprietary ATI drivers. I don't have to use site http://schumann.elis.ugent.be/ in everyday work. I only reported a problem found accidentaly. Thank You for advice.
Comment by Martin Raivio (MurdRav) - Wednesday, 12 September 2012, 15:32 GMT
This happen to me also. I have ATI radeon 4650 card with open source ATI/AMD driver. Problem is I think
with video driver and the fact that firefox enables webgl when driver can not handle it. That site works for me
with both Ubuntu and Slackware but they have the ATI proprietary Catalyst drivers.
But for me the repair was

start up firefox
type about.config
click OK on warranty thingy (What warranty ???)
scroll to almost very bottom
find webgl.disabled
double click till it turns to true.
Go to : http://schumann.elis.ugent.be/
Get message
It doesn't appear your computer can support WebGL.
Click here for more information.

NO CRASH
Comment by grzech (grzech) - Thursday, 13 September 2012, 07:35 GMT
OK that's the way. I found this bug testing WebGL in firefox. But the point is we got the bug in kernel code.
Comment by grzech (grzech) - Saturday, 15 September 2012, 08:19 GMT
Problem resolved 10 days before bug report;) Kernel developers are really fast:) For kernel 3.5 bug is fixed in patch 3.5.4:

commit 929b707162f994145a5bd4f8490aab17c0205153
Author: Alex Deucher <alexander.deucher@amd.com>
Date: Tue Aug 21 09:55:01 2012 -0400

Revert "drm/radeon: fix bo creation retry path"

commit 676bc2e1e4f9072f7a640d5b7c99ffdf9709a6e7 upstream.

This reverts commit d1c7871ddb1f588b8eb35affd9ee1a3d5e11cd0c.

ttm_bo_init() destroys the BO on failure. So this patch makes
the retry path work with freed memory. This ends up causing
kernel panics when this path is hit.

Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Patches 3.2.29 and 3.4.11 also fixed this bug. I tested it on own compiled kernel 3.4.11. Thanks for all - for investigating and ways to resolve.

Loading...