| Summary: | mesa_test_os_sse_exception_support fails inside gdb | ||
|---|---|---|---|
| Product: | Mesa | Reporter: | Moritz Muehlenhoff <jmm> |
| Component: | Mesa core | Assignee: | Default DRI bug account <dri-devel> |
| Status: | RESOLVED FIXED | QA Contact: | |
| Severity: | normal | ||
| Priority: | high | ||
| Version: | unspecified | ||
| Hardware: | Other | ||
| OS: | Linux (All) | ||
| Whiteboard: | |||
| i915 platform: | i915 features: | ||
| Attachments: |
Backtrace of a debug run inside gdb
add IN_DRI_DRIVER Define into xc/lib/GL/mesa subdir where it is missing |
||
|
Description
Moritz Muehlenhoff
2004-10-25 15:36:36 UTC
Created attachment 1164 [details]
Backtrace of a debug run inside gdb
The backtrace. BTW, I'm using gdb 6.1.
hi Just a guess, I had one similar bug and recompile glut with Mesa - sources and the bug is gone i strongly doubt this is a bug. the whole point of mesa_test_os_sse_exception is to generate an FPE to see how the OS handles it. try saying "handle SIGFPE nostop" in gdb before running your application. my guess is it'll work just fine. It's not a bug in anything. Just type 'cont', hit return, and be happy. The exception is just a side-effect of the way Mesa determines if the CPU and the OS fully support SSE. I see the point, but is there no gentler way to determine if the CPU has SSE support? I guess the current solution leads almost every programmer to believe something went wrong with his own code (and I guess there are very few setups that have built their X11 libs with debug symbols). I've added a comment on the _mesa_test_os_sse_support() function to tell developers to just type 'continue' in gdb when the exception is caught. Hopefully this will help people in the future. Created attachment 3848 [details] [review] add IN_DRI_DRIVER Define into xc/lib/GL/mesa subdir where it is missing Ian Romanick has point that Mesa code should have IN_DRI_DRIVER Define, to avoid mesa_test_os_sse_exception_support. After analyze how build Mesa from CVS with "make linux-x86-dri" and how xorg built same Mesa source with "make World" So, I found on xc/lib/GL/mesa all subdirs: shader swrast tnl x86 main math and tnl_dd have missing at least -DIN_DRI_DRIVER Xorg build also misses -DPTHREADS -DUSE_EXTERNAL_DXTN_LIB=1 -DHAVE_ALIAS Ian, and those ones are needed ? damnation, this should have been on the blocker list. withdrawn, far too late in the process now. Patch in attachment #3848 [details] [review] has been applied to Xorg monolith for inclusion in 6.9. (In reply to comment #10) > Patch in attachment #3848 [details] [review] [edit] has been applied to Xorg monolith for inclusion in 6.9. I would prefer that has been applied attachment #4069 [details] [review] from bug #5057 because is more safety on files xc/lib/GL/mesa/main/Imakefile and xc/lib/GL/mesa/x86/Imakefile use #if BuildXF86DRI DRI_DEFINES = -DIN_DRI_DRIVER #endif in other hand, add other Mesa Missing Define, this one much less critical used and problematic. Patch in attachment #3848 [details] [review] has been applied to Xorg 6.9 monolith. Closing ... |
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.