Summary: | glxgears segfault only when resolution > 1024x768 | ||
---|---|---|---|
Product: | Mesa | Reporter: | Huw Jenkins <huwtjenkins> |
Component: | Drivers/DRI/r128 | Assignee: | Default DRI bug account <dri-devel> |
Status: | RESOLVED DUPLICATE | QA Contact: | |
Severity: | normal | ||
Priority: | high | ||
Version: | unspecified | ||
Hardware: | x86 (IA32) | ||
OS: | Linux (All) | ||
Whiteboard: | |||
i915 platform: | i915 features: | ||
Attachments: |
Xorg log
glxinfo when running at 1024x768 gdb log of glxinfo segfault drirc file |
Description
Huw Jenkins
2004-07-27 08:51:47 UTC
Created attachment 528 [details]
Xorg log
Created attachment 529 [details]
glxinfo when running at 1024x768
please run glxinfo under gdb and obtain a backtrace. type 'gdb /usr/X11R6/bin/glxinfo' at a shell prompt. at the (gdb) prompt, type 'run', then 'bt' after the segfault. this may be caused by, or fixed by, fxkuehl's xmlconfig.c changes, depending on timing. could you also try the most recent DRI snapshot? Beware that debugging a DRI client from the same display will deadlock the X server if execution is stopped while the client holds the hardware lock. Run gdb from another machine, e.g. via ssh, to be safe. Created attachment 535 [details]
gdb log of glxinfo segfault
This doesn't look very useful - not sure why debugging symbols missing...
Occurs with latest snapshot (rage128-20040727) Could you attach your configuration file /home/huw/.drirc. This would help me reproduce the problem here if it's really related to the configuration parser. Created attachment 542 [details]
drirc file
(I should note that the fix described in 1730 isn't going to make the card work at higher resolutions, it should just make glxgears/glxinfo not segfault when failing to initialize DRI) I don't know how I never noticed this bug before. Hmm...anyway, the r128 driver seems to have to odd quirk that it will try to enable direct-rendering even when there is no on-card memory available for textures. I don't think any of the other drivers do this, but I'd have to check. The issues that the DRI driver assumes that there will be on-card memory but that there may or may not be AGP memory available. It should be fairly easy to modify the init code in the DRI driver (r128_context.c, line 159) to not crash if r128scrn->texSize[0] == 0. The only tricky part is that there may be places that assume rmesa->texture_heaps[0] is on-card memory. This bug is most likely a dupe of 5115 https://bugs.freedesktop.org/show_bug.cgi?id=5115 *** This bug has been marked as a duplicate of 5115 *** |
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.