Bug 26480 - Segfault / hang, and texture problems on Mobility X300
Summary: Segfault / hang, and texture problems on Mobility X300
Status: RESOLVED INVALID
Alias: None
Product: xorg
Classification: Unclassified
Component: Driver/Radeon (show other bugs)
Version: unspecified
Hardware: Other All
: medium normal
Assignee: xf86-video-ati maintainers
QA Contact: Xorg Project Team
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2010-02-08 13:01 UTC by Alexey Spiridonov
Modified: 2016-02-25 02:07 UTC (History)
1 user (show)

See Also:
i915 platform:
i915 features:


Attachments
dmesg (41.67 KB, application/octet-stream)
2010-02-08 13:01 UTC, Alexey Spiridonov
no flags Details
xdpyinfo (3.94 KB, application/octet-stream)
2010-02-08 13:01 UTC, Alexey Spiridonov
no flags Details
xorg.conf (I later added EXA) (1.77 KB, application/octet-stream)
2010-02-08 13:02 UTC, Alexey Spiridonov
no flags Details
X session hung with XAA when I ran Q3A (just when the level was supposed to appear) (60.42 KB, application/x-trash)
2010-02-08 13:03 UTC, Alexey Spiridonov
no flags Details
XAA log on the boot after the Q3A hang (60.09 KB, text/x-log)
2010-02-08 13:03 UTC, Alexey Spiridonov
no flags Details
EXA segfault after trying to enable Compiz (88.97 KB, text/plain)
2010-02-08 13:04 UTC, Alexey Spiridonov
no flags Details

Description Alexey Spiridonov 2010-02-08 13:01:03 UTC
Created attachment 33174 [details]
dmesg

I have an Ubuntu/Karmic 9.10 system (xorg server 1.6.4), radeon module version 6.12.99 with an ATI Mobility Radeon X300 M22 PCI Express card. I used to use the proprietary driver, and had neither stability nor performance problems. With the open-source driver, I have both. I hope that you will be able to help me.

$ lspci | grep VGA
01:00.0 VGA compatible controller: ATI Technologies Inc M22 [Mobility Radeon X300]

I carefully read through these bugs:
  https://bugs.freedesktop.org/show_bug.cgi?id=15096
  https://bugs.freedesktop.org/show_bug.cgi?id=24506
They didn't completely resolve my issues.

1) EXA should be enabled by default for my card. 

At first, all my 3D applications triggered X server lockups almost immediately. The backtrace was:
 (gdb) bt
 #0  0x00c2d422 in __kernel_vsyscall ()
 #1  0x0023d719 in ioctl () from /lib/tls/i686/cmov/libc.so.6
 #2  0x002d7b2d in drmIoctl () from /usr/lib/libdrm.so.2
 #3  0x002d7f22 in drmCommandNone () from /usr/lib/libdrm.so.2
 #4  0x00598709 in ?? () from /usr/lib/xorg/modules/drivers//radeon_drv.so
 #5  0x00d89a48 in ?? () from /usr/lib/xorg/modules//libxaa.so
 #6  0x00dcdeac in ?? () from /usr/lib/xorg/modules//libxaa.so
 #7  0x0817fd65 in ?? ()
 #8  0x0816feed in ?? ()
 #9  0x0817000e in miCompositeRects ()
 #10 0x00dcf47b in ?? () from /usr/lib/xorg/modules//libxaa.so
 #11 0x081714f2 in CompositeRects ()
 #12 0x081780a5 in ?? ()
 #13 0x081742e5 in ?? ()
 #14 0x0808d17f in Dispatch ()
 #15 0x08072515 in main ()
I tried a bunch of things, including "nomodeset", "radeon.modeset=0", etc. I finally found a suggestion to try EXA. That got rid of a number of crashes. Then I discovered this line hidden deep inside the log file:
 (II) RADEON(0): XAA Render acceleration unsupported on Radeon 9500/9700 and newer. Please use EXA instead.

I'm attaching X logs for this XAA setup, (1) from the time the crash happened -- I rebooted the system with REISUB, (2) from the boot immediately after the crash.

Can we please make this the default? If the user has not overridden the setting by explicitly writing XAA, make the right choice for them. Putting an II line saying "do this or your system will crash" in a log is hardly a good user experience.


2) Performance and texture issues with EXA. With this setting, my 3D test apps both started "working". They are Neverball and Quake 3 Arena. Performance is abysmal. Q3 at 800x600 refreshes at about 15fps, and stutters badly every second or so. Neverball is just barely playable in an 800x600 window (15-20fps).

Worse yet, the textures in Q3 are completely broken if there scene is complex enough: they cycle through some progression of colors and splotches in a psychedelic progression. A similar bug also shows up in Neverball: some of the objects are supposed to glisten -- but with this driver, they flash entirely white in a fairly rapid strobe.


3) Compiz causes a segfault and X hang with EXA. The log is attached.



I'm very happy to help debug these, just let me know what to do. 

Thanks!
Comment 1 Alexey Spiridonov 2010-02-08 13:01:30 UTC
Created attachment 33175 [details]
xdpyinfo
Comment 2 Alexey Spiridonov 2010-02-08 13:02:12 UTC
Created attachment 33176 [details]
xorg.conf (I later added EXA)
Comment 3 Alexey Spiridonov 2010-02-08 13:03:11 UTC
Created attachment 33177 [details]
X session hung with XAA when I ran Q3A (just when the level was supposed to appear)
Comment 4 Alexey Spiridonov 2010-02-08 13:03:42 UTC
Created attachment 33178 [details]
XAA log on the boot after the Q3A hang
Comment 5 Alexey Spiridonov 2010-02-08 13:04:17 UTC
Created attachment 33179 [details]
EXA segfault after trying to enable Compiz
Comment 6 Michel Dänzer 2010-02-16 02:31:58 UTC
1) Already the case in current upstream and AFAIK in the Ubuntu packages you're using, but only with more than 32 MB of VRAM, due to other bug reports about EXA performance issues with little VRAM.

2) Two issues that need to be tracked separately. Though I suspect it's mostly due to there being almost no VRAM available for textures. If possible, reducing the display depth or resolution might alleviate this.

3) Yet another issue that needs to be tracked separately. I think there's a good chance this is already fixed in a newer version of Mesa, I suppose you're using libgl1-mesa-dri 7.6.0-1ubuntu4 from Ubuntu karmic?
Comment 7 Pauli 2010-02-16 02:35:17 UTC
(In reply to comment #5)
> Created an attachment (id=33179) [details]
> EXA segfault after trying to enable Compiz 
> 

Can you still find the kernel log matching this log? There should be more info
about this crash in kernel log. Please open separate bug report like Michel asked.
Comment 8 Christopher M. Penalver 2016-02-25 02:07:14 UTC
Alexey Spiridonov, Ubuntu 9.10 reached EOL on April 30, 2011. For more on this, please see https://wiki.ubuntu.com/Releases .

If this is reproducible in a supported release, it will help immensely if you filed a new report with Ubuntu by ensuring you have the package xdiagnose installed, and that you click the Yes button for attaching additional debugging information running the following from a terminal:
ubuntu-bug xorg

Also, please feel free to subscribe me to it.

For more on why this is helpful, please see https://wiki.ubuntu.com/ReportingBugs.


Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.