Red Hat Bugzilla – Bug 215155
glBlendFunc() segfaults on R300
Last modified: 2008-05-01 11:38:07 EDT
Description of problem:
I have a Radeon Mobility X600, and when I try to run software which calls
glBlendFunc() it crashes with a segfault. I don't have any other hardware
accelerated 3D hardware that works well enough to test this, so it might be
specific to the R300 driver but I can't prove it.
In my own software I can work around this bug by removing calls to glBlendFunc()
although of course this means I no longer have transparency, which makes things
look very ugly.
Version-Release number of selected component (if applicable):
Completely reproducible, happens every time. May need specific hardware to
reproduce if it's only this particular driver.
Steps to Reproduce:
1. Build or install software which uses glBlendFunc() such as the Mesa Teapot demo
2. Run it on FC6
Working GL transparency support as per OpenGL specification
Here's an example backtrace from the GL Teapot example which I compiled from the
Mesa source used for FC6. Other applications, whether my own or third party
crash with a similar trace
[ ie _mesa_BlendFunc() calls _mesa_loopback_init_api_table() and then we crash ]
Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread -1208289584 (LWP 3710)]
0x002739a1 in _mesa_loopback_init_api_table () from /usr/lib/dri/r300_dri.so
#0 0x002739a1 in _mesa_loopback_init_api_table () from /usr/lib/dri/r300_dri.so
#1 0x0027eb44 in _mesa_BlendFunc () from /usr/lib/dri/r300_dri.so
#2 0x08049bb5 in main (ac=1, av=0xbfa0d9c4) at teapot.c:655
Sorry, bug was caused by a stale build environment from some tests I ran before
upgrading to FC6. Not a Fedore Core bug, thus closing NOTABUG.