glmark2 crashed with SIGSEGV in _swrast_render_texture() [with MIR]

Bug #1203742 reported by Mark Shuttleworth
28
This bug affects 3 people
Affects Status Importance Assigned to Milestone
glmark2 (Ubuntu)
Expired
Undecided
Unassigned

Bug Description

Crashes after a few minutes of benchmarking.

ProblemType: Crash
DistroRelease: Ubuntu 13.10
Package: glmark2 2012.08-0ubuntu2
ProcVersionSignature: Ubuntu 3.10.0-4.13-generic 3.10.1
Uname: Linux 3.10.0-4-generic x86_64
ApportVersion: 2.11-0ubuntu1
Architecture: amd64
CrashCounter: 1
Date: Mon Jul 22 14:00:11 2013
DistributionChannelDescriptor:
 # This is a distribution channel descriptor
 # For more information see http://wiki.ubuntu.com/DistributionChannelDescriptor
 canonical-oem-somerville-precise-amd64-20120703-2
EcryptfsInUse: Yes
ExecutablePath: /usr/bin/glmark2
InstallationDate: Installed on 2013-05-13 (69 days ago)
InstallationMedia: Ubuntu 12.04 "Precise" - Build amd64 LIVE Binary 20120703-15:08
MarkForUpload: True
ProcCmdline: glmark2
SegvAnalysis:
 Segfault happened at: 0x7fefaf0c6e34 <_swrast_render_texture+68>: mov (%rax,%r12,1),%rax
 PC (0x7fefaf0c6e34) ok
 source "(%rax,%r12,1)" (0x00000000) not located in a known VMA region (needed readable region)!
 destination "%rax" ok
SegvReason: reading NULL VMA
Signal: 11
SourcePackage: glmark2
StacktraceTop:
 _swrast_render_texture () from /usr/lib/x86_64-linux-gnu/libdricore9.2.0-devel.so.1
 ?? () from /usr/lib/x86_64-linux-gnu/libdricore9.2.0-devel.so.1
 _mesa_FramebufferTexture2D () from /usr/lib/x86_64-linux-gnu/libdricore9.2.0-devel.so.1
 ?? ()
 ?? ()
Title: glmark2 crashed with SIGSEGV in _swrast_render_texture()
UpgradeStatus: No upgrade log present (probably fresh install)
UserGroups: adm cdrom dip libvirtd lpadmin plugdev sambashare sudo

Revision history for this message
Mark Shuttleworth (sabdfl) wrote :
Revision history for this message
Apport retracing service (apport) wrote :

StacktraceTop:
 ?? ()
 ?? ()
 ?? ()
 ?? ()
 ?? ()

Revision history for this message
Apport retracing service (apport) wrote : Stacktrace.txt
Revision history for this message
Apport retracing service (apport) wrote : ThreadStacktrace.txt
Changed in glmark2 (Ubuntu):
status: New → Invalid
Revision history for this message
Apport retracing service (apport) wrote : Crash report cannot be processed

Thank you for your report!

However, processing it in order to get sufficient information for the
developers failed (it does not generate a useful symbolic stack trace). This
might be caused by some outdated packages which were installed on your system
at the time of the report:

outdated debug symbol package for libexpat1: package version 2.1.0-4 dbgsym version 2.1.0-2

Please upgrade your system to the latest package versions. If you still
encounter the crash, please file a new report.

Thank you for your understanding, and sorry for the inconvenience!

tags: removed: need-amd64-retrace
Revision history for this message
Mark Shuttleworth (sabdfl) wrote : Re: glmark2 crashed with SIGSEGV in _swrast_render_texture()

Hmm... don't think apport's debug tracer got that right. Libexpat is up to date here relative to the saucy archive.

Revision history for this message
Martin Pitt (pitti) wrote :

Mark,

right, your installed version of libexpat1 is up to date, but http://ddebs.ubuntu.com/pool/main/e/expat/ is missing the amd64 debug symbols for 2.1.0-4. (There's a long story behind why ddebs.u.c. is unreliable..)

But the libexpat warning is mostly moot. Apport invalidates bugs if it encounters any obsolete libraries and the stack trace is unusable, such as in this case. Here it seems that gdb is unable to determine a stack trace; memory could have gotten thoroughly corrupted, or there is something specific wrt. these libraries that gdb doesn't get along with. Hard to say without a core dump, unfortunately the retracer already removed the core attachment (secret data leakage paranoia..)

So in summary, I'm afraid we need a core dump here. If this problem is easily reproducible (trying now..), I can get one and check further if it's a problem in gdb or in our retracer setup.

Revision history for this message
Martin Pitt (pitti) wrote :

glmark2 finished here ("Mesa DRI Intel(R) Ironlake Mobile" renderer, Intel Arrandale chipset, saucy amd64 du jour). If you can reproduce this, perhaps you can attach the raw apport .crash file from /var/crash/ here?

Thanks!

information type: Private → Public
Changed in glmark2 (Ubuntu):
status: Invalid → Incomplete
Revision history for this message
Mark Shuttleworth (sabdfl) wrote :

Am attaching the raw apport crash file for this crash, which still reliably happens on Intel drivers under Mir. Thanks!

Revision history for this message
Martin Pitt (pitti) wrote :

Ah, https://launchpadlibrarian.net/145584178/Dependencies.txt shows quite a lot of packages from a PPA, and thus our retracer wouldn't have the debug symbols for the important libraries (libGL etc.). Now the PPA is empty as the packages all went into saucy, so reporting the crash again actually ought to give us something more sensible from the auto-retracers. Unfortunately the attached .crash file does not contain any package/dependency information, I forgot to tell that you have to tell apport to collect these details first. It's easiest by clicking on "show details", wait until that's done, then closing apport, and attaching the .crash file.

I tried to process the attached .crash file on current saucy (without package/dependency info), but I'm afraid it still gives an useless backtrace. Do you have the current packages from saucy now, without the PPA? I could also try to reproduce the crash by configuring my system like your's, with using Mir as a system compositor.

Revision history for this message
Martin Pitt (pitti) wrote :

For the record, I cannot reproduce the glmark2 crash on my intel laptop (ThinkPad X201, Arrandale) with today's saucy Mir system compositor, I ran it three times.

Revision history for this message
Mark Shuttleworth (sabdfl) wrote : Re: [Bug 1203742] Re: glmark2 crashed with SIGSEGV in _swrast_render_texture()

On 20/08/13 14:30, Martin Pitt wrote:
> For the record, I cannot reproduce the glmark2 crash on my intel laptop
> (ThinkPad X201, Arrandale) with today's saucy Mir system compositor, I
> ran it three times.

Ah, interesting. I will do a ppa-purge and try again, thanks Martin!

Revision history for this message
Robert Ancell (robert-ancell) wrote : Re: glmark2 crashed with SIGSEGV in _swrast_render_texture()

I can reproduce this running under XMir from the saucy archive (Intel Corporation Core Processor Integrated Graphics Controller (rev 18))

Specifically it fails after the pulsar test:

bob@alchemy:~$ glmark2
=======================================================
    glmark2 2012.08
=======================================================
    OpenGL Information
    GL_VENDOR: Intel Open Source Technology Center
    GL_RENDERER: Mesa DRI Intel(R) Ironlake Mobile
    GL_VERSION: 2.1 Mesa 9.2.0-devel
=======================================================
[build] use-vbo=false: FPS: 113 FrameTime: 8.850 ms
[build] use-vbo=true: FPS: 215 FrameTime: 4.651 ms
[texture] texture-filter=nearest: FPS: 280 FrameTime: 3.571 ms
[texture] texture-filter=linear: FPS: 302 FrameTime: 3.311 ms
[texture] texture-filter=mipmap: FPS: 276 FrameTime: 3.623 ms
[shading] shading=gouraud: FPS: 248 FrameTime: 4.032 ms
[shading] shading=blinn-phong-inf: FPS: 200 FrameTime: 5.000 ms
[shading] shading=phong: FPS: 169 FrameTime: 5.917 ms
[bump] bump-render=high-poly: FPS: 105 FrameTime: 9.524 ms
[bump] bump-render=normals: FPS: 244 FrameTime: 4.098 ms
[bump] bump-render=height: FPS: 251 FrameTime: 3.984 ms
[effect2d] kernel=0,1,0;1,-4,1;0,1,0;: FPS: 134 FrameTime: 7.463 ms
[effect2d] kernel=1,1,1,1,1;1,1,1,1,1;1,1,1,1,1;: FPS: 66 FrameTime: 15.152 ms
[pulsar] light=false:quads=5:texture=false: FPS: 222 FrameTime: 4.505 ms
Segmentation fault (core dumped)

tags: added: xmir
Martin Pitt (pitti)
summary: - glmark2 crashed with SIGSEGV in _swrast_render_texture()
+ glmark2 crashed with SIGSEGV in _swrast_render_texture() [with MIR]
Revision history for this message
Launchpad Janitor (janitor) wrote :

[Expired for glmark2 (Ubuntu) because there has been no activity for 60 days.]

Changed in glmark2 (Ubuntu):
status: Incomplete → Expired
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Duplicates of this bug

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.