| Summary: | GLX Visual Strangeness | ||
|---|---|---|---|
| Product: | xorg | Reporter: | Ben Gamari <bgamari> |
| Component: | Server/General | Assignee: | Xorg Project Team <xorg-team> |
| Status: | RESOLVED FIXED | QA Contact: | Xorg Project Team <xorg-team> |
| Severity: | normal | ||
| Priority: | medium | CC: | krh |
| Version: | git | ||
| Hardware: | Other | ||
| OS: | All | ||
| Whiteboard: | |||
| i915 platform: | i915 features: | ||
| Attachments: | |||
Created attachment 16093 [details]
glxinfo -t when run with a Typical set of Glx Visuals
It looks like someone else has experienced this bug as well: http://people.freedesktop.org/~cbrill/dri-log/index.php?date=2008-03-03&channel=dri-devel&show_html=true&highlight_names=krh%3Bghepeu&update=Update&date=2008-03-31 . Kristian , do you know anything about this? It looks like this might be an intel-specific issue actually, judging by http://bugs.freedesktop.org/show_bug.cgi?id=14340#c17. Any input anyone? Please attach the X log file. In particular, is AIGLX enabled? I think GLX visuals are known to be busted currently when it isn't. Created attachment 16362 [details]
Xorg log
I just attached a log which shows that AIGLX should be enabled. There are also some messages in that log which might not look familiar. These I added to the xserver to figure out what is happening. If they look like they might help, I attached the log, otherwise ignore them. Created attachment 16363 [details] [review] Patch adding debug output to xserver I'm not seeing this with radeon or intel, I suspect it could be 64 bit related. (In reply to comment #8) > I'm not seeing this with radeon or intel, I suspect it could be 64 bit related. > Ahh, touche. I completely neglected to consider this as an option. Created attachment 16744 [details]
Minimal test case for issue
On my machine, this fails with,
ben@mercury ~/c $ ./visual-test
Using visual 0x21
Root visual is not a GL visual
This is quite strange. It appears that my default root config is getting filtered out in filterOutNativeConfigs(). Surely, this can't be the intended behavior? Created attachment 16766 [details] [review] Fix to use X Visual IDs instead of GLX visual ids Thanks krh. Created attachment 16767 [details] [review] New patch that actually applies against master Comment on attachment 16767 [details] [review] New patch that actually applies against master this patch appears to completely mix up the indices into arrays, and would result in bad results if the glx visuals weren't a 1:1 mapping to X visuals The "full" option appears to be broken -- it doesn't attempt to match the visuals to existing X visuals, so the root visual never gets a corresponding GLX visual even though one exists. Lame. Eric fixed this for xserver 1.6. |
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.
Created attachment 16092 [details] glxinfo -t when run with a Full set of GlxVisuals xserver from git master (along with mesa and xf86-video-intel from git) produces extremely strange sets of GL visuals, so much so that compiz refuses to start. When the GlxVisuals ServerFlags option is set to "typical" (which appears to be the same as minimal from GL/glx/glxscreens.c?), the server exposes three GL visuals, none of which are double-buffered (which compiz requires). When GlxVisuals is set to "full", the server exposes 36 visuals, none of which are the root visual (0x21), again prohibiting compiz from starting. Surely there is something wrong with this picture.