Bug 9307 - [965g]: Consistent crashes in any GL application with DRI enabled
Summary: [965g]: Consistent crashes in any GL application with DRI enabled
Status: RESOLVED WORKSFORME
Alias: None
Product: Mesa
Classification: Unclassified
Component: Drivers/DRI/i965 (show other bugs)
Version: 6.5
Hardware: x86 (IA32) Linux (All)
: medium normal
Assignee: Eric Anholt
QA Contact:
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2006-12-11 17:42 UTC by Jack Lloyd
Modified: 2007-08-27 18:51 UTC (History)
1 user (show)

See Also:
i915 platform:
i915 features:


Attachments
My X logfile (49.16 KB, text/plain)
2006-12-11 17:43 UTC, Jack Lloyd
Details
My Xorg config (1.51 KB, text/plain)
2006-12-11 17:43 UTC, Jack Lloyd
Details
glxinfo output (4.86 KB, text/plain)
2006-12-11 17:43 UTC, Jack Lloyd
Details
X Log (35.26 KB, text/plain)
2007-07-23 19:45 UTC, Wagner Macedo
Details
X Log with error (37.84 KB, text/plain)
2007-07-23 19:50 UTC, Wagner Macedo
Details
X Log with error (37.84 KB, text/plain)
2007-07-23 19:50 UTC, Wagner Macedo
Details
X Log with debug mode (63.10 KB, text/plain)
2007-07-31 10:32 UTC, Wagner Macedo
Details

Note You need to log in before you can comment on or make changes to this bug.
Description Jack Lloyd 2006-12-11 17:42:33 UTC
With Mesa 6.5.2, running a GL application (such as glxgears or mplayer -vo gl)
results in the application segmentation faulting immediately. This only occurs
when DRI is enabled. I cannot get a backtrace; running within gdb locks up X
entirely and seems to screw up the video hardware state, requiring a reboot.

Currently I'm using Xorg 7.1 w/the latest modesetting branch of the Intel X
driver. It looks like Xorg 7.2 isn't yet official, I will see if this still
occurs with 7.2 as soon as packages hit my distro (probably within a week).

I will note that this is a major improvement over Mesa 6.5.1, where GL
applications would crash with or without DRI enabled. :/ So at least it has gone
from non-working to working but slow.
Comment 1 Jack Lloyd 2006-12-11 17:43:05 UTC
Created attachment 8064 [details]
My X logfile
Comment 2 Jack Lloyd 2006-12-11 17:43:22 UTC
Created attachment 8065 [details]
My Xorg config
Comment 3 Jack Lloyd 2006-12-11 17:43:42 UTC
Created attachment 8066 [details]
glxinfo output
Comment 4 Michel Dänzer 2006-12-30 02:41:00 UTC
(In reply to comment #0)
> With Mesa 6.5.2, running a GL application (such as glxgears or mplayer -vo gl)
> results in the application segmentation faulting immediately. This only occurs
> when DRI is enabled. 

Are you making sure that Mesa builds with -fno-strict-aliasing?

> I cannot get a backtrace; running within gdb locks up X entirely and seems to
> screw up the video hardware state, requiring a reboot.

Yes, you can't debug an X server from within itself, because gdb stops execution
of the X server, which is what provides interaction with gdb... it should work
from a remote login though.
Comment 5 Jack Lloyd 2006-12-30 13:04:42 UTC
Yes, I'm using the Mesa ebuild from Gentoo, which appends -fno-strict-aliasing
to the set of flags (so the full CFLAGS when Mesa is built is '-march=nocona -O2
-momit-leaf-frame-pointer -fno-strict-aliasing').

I'll see if I can get a backtrace using a remote login.
Comment 6 Gordon Jin 2007-03-14 19:38:23 UTC
The bug priority was upgraded (P2->high) with the bugzilla configuration change.
I'm Changing the priority back to the normal one.
Sorry for the spam.
Comment 7 Wagner Macedo 2007-07-23 19:45:19 UTC
Created attachment 10862 [details]
X Log

With me, happens the same, I tested with Ubuntu Feisty and Ubuntu Gutsy (Alpha Version)
Comment 8 Wagner Macedo 2007-07-23 19:50:10 UTC
Created attachment 10863 [details]
X Log with error

Sorry, in my other attachment, I sent the wrong file, now I'm sending the right.
Comment 9 Wagner Macedo 2007-07-23 19:50:23 UTC
Created attachment 10864 [details]
X Log with error

Sorry, in my other attachment, I sent the wrong file, now I'm sending the right.
Comment 10 Wagner Macedo 2007-07-31 10:32:03 UTC
Created attachment 10928 [details]
X Log with debug mode

What more info is needed on this topic? I have an Intel MB DG965SS with Core 2 Duo that I bought less than 3 months and I realy can't use it. I'm having problems that this topic describes well. I decided to buy Intel because this MB is marked as compatible on the site and the open source drivers.

And more, it is not relative to this topic: my sound card don't work microphone. I'm very disappointed with Intel.
Comment 11 Gordon Jin 2007-08-13 19:57:58 UTC
I'm not seeing this problem with upstream driver.

Jack, could you try newer driver?

Wagner, did you report this problem to Ubuntu? I believe Ubuntu has validated on 965. 
Comment 12 Wagner Macedo 2007-08-25 15:43:59 UTC
> Wagner, did you report this problem to Ubuntu? I believe Ubuntu has validated
> on 965. 
>
I tested with Ubuntu Feisty, then Gutsy Tribe 5 arrive in August, 23, so I tested too and I have the same result. So, I report today to Ubuntu.

But if you can test, It's easy to reproduce in Ubuntu (could be Live Tribe 5). In the screensaver preferences, try Endgame, just need less than 5 minutes to crash.
Comment 13 Gordon Jin 2007-08-25 19:03:54 UTC
Wagner, the original reporter says it happens for "any GL application". Are you also in this status? e.g. running "glxgears" will also crash?
Comment 14 Wagner Macedo 2007-08-27 08:07:21 UTC
(In reply to comment #13)
> Wagner, the original reporter says it happens for "any GL application". Are you
> also in this status? e.g. running "glxgears" will also crash?
> 

Thinking about, Gordon, I'm not in this status. Here it happens for some GL applications, not for "any". I run glxgears and don't crash.

Sorry for the inconvenience.

What should I do now? Report a new bug?
Comment 15 Gordon Jin 2007-08-27 18:22:36 UTC
Yes, Wagner, please open a new bug.
Comment 16 Gordon Jin 2007-08-27 18:26:06 UTC
I'm closing as I can't reproduce it and no feedback from the original reporter for long time.

Jack, please reopen if you can reproduce it with the latest driver.
Comment 17 Jack Lloyd 2007-08-27 18:51:04 UTC
I can confirm as fixed with kernel 2.6.22.1, xf86-video-intel 2.1.1, mesa 7.0.1, xorg-server 1.3.0

At first glance, things seem stable and reasonably fast. Thanks for your efforts.


Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct.