[Intel-gfx] panning crash

[Felix Miata]

Chris Wilson composed on 2015-06-13 09:19 (UTC+0100):

Thank you for responding!

> On Sat, Jun 13, 2015 at 02:17:34AM -0400, Felix Miata wrote:

>> Dell Optiplex GX280
>> P4 32 bit Hyperthreading enabled 0F34 2.8GHz CPU
>> 00:02.0 VGA compatible controller: Intel Corporation 82915G/GV/910GL Integrated Graphics Controller (rev 04)
 
>> /etc/X11/xinit/xinitrc.d/setup contains:
>> xrandr --fbmm 387x218 --fb 2560x1440 --output VGA1 --mode 1920x1440 --panning 2560x1440
 
>> Panning mouse out of initial CRTC desktop space into virtual desktop space and back in
>> some installations causes crash, in others not. Can someone here tell if this is a kernel,
>> server or other issue and not a driver issue from a look at Xorg.0.log? My suspicion is
>> kernel as primary if not sole.

> I think the crash is in the rejection handling code. If the kernel
> rejects the modeset, we try again with via a shadow. The crash is very
> unexpected. You can run the stacktraces though addr2line (e.g.
> addr2line -i -e /usr/lib/xorg/modules/drivers/intel_drv.so 0x638f7
> 0x68fbc 0x6a02b)

Brief googling this didn't find anything I can grok. It's probably doable with
minimal hand-holding, e.g. a URL explaining what tools need to be installed
(gdb + ???) and how to use them. Debugging is something I do so infrequently I
forget everything I managed to slog through from one session to the next.

> but also compiling your own ddx with
> --enable-debug=full would be very useful.

Not doable. I tried programming in college over 4 decades ago. English remains
as much language as I can cope with, and not even that much WRT OBS. OTOH, RPMs
I can deal with.

> A drm.debug=6 dmesg would tell you what is going on kernel side.

Tell you maybe. :-)

http://fm.no-ip.com/Tmp/Linux/Xorg/Igfx/panfault/

Fedora 21, kernel 4.0.4:
dmsg-i915G-f21-k404crash.txt

Tumbleweed, kernel 4.0.4:
dmsg-i915G-os133-k404crash.txt

Looks like maybe the crash occurred in the vicinity of [433.26] in F21,
[111.57] in TW, but there's minimal sense I can make of dmesg generally.

All this said, crash didn't happen in Rawhide, so maybe it's a kernel
problem that has resolved itself post-4.0.4?

HTH
-- 
"The wise are known for their understanding, and pleasant
words are persuasive." Proverbs 16:21 (New Living Translation)

 Team OS/2 ** Reg. Linux User #211409 ** a11y rocks!

Felix Miata  ***  http://fm.no-ip.com/