Bug 56542

Created attachment 69252 [details]
backtrace

Reproducible with llvmpipe or r300g here.

/piglit/bin/select alpha
PIGLIT: {'result': 'pass' }
pure virtual method called
terminate called without an active exception
Neúspěšně ukončen (SIGABRT)

Full backtrace attached.
This works fine in 8.0.5 so I did a regression test, but there were some build failures:

There are only 'skip'ped commits left to test. The first bad commit could be any of: 23f1e551cc6553ce7f99b9264efc8776f5f35b00 1f2c87cc8f50d0a0ec3c9be1ddf6cf035002e5c7 
f42e601ce0263808ba36c88763196856ecdb79c1 
2023bf996ed5c3797233b8d70670c28e15bdff75 
3469715a8a171512cf9b528702e70393f01c6041 
bf484024b944a452e9022a1098313663e0028b29 
We cannot bisect more!

However almost all of them are build fixes, or some gbm stuff which leaves:
commit 3469715a8a171512cf9b528702e70393f01c6041
Author: José Fonseca <jfonseca@vmware.com>
Date:   Fri Jul 13 18:09:30 2012 +0100

    gallivm,draw,llvmpipe: Support wider native registers.

as the most likely candidate.

LLVM version: 3.0-14.fc17
Mesa: e8f2bec25ea9bbaedac7abfa7eabb62e1aaf3163

This is with an old intel prescott cpu, 32 bit Fedora 17:
from /proc/cpuinfo
vendor_id	: GenuineIntel
cpu family	: 6
model		: 14
model name	: Genuine Intel(R) CPU T2300  @ 1.66GHz

printed with GALLIUM_DUMP_CPU=1
util_cpu_caps.nr_cpus = 2
util_cpu_caps.x86_cpu_type = 6
util_cpu_caps.cacheline = 64
util_cpu_caps.has_tsc = 1
util_cpu_caps.has_mmx = 1
util_cpu_caps.has_mmx2 = 1
util_cpu_caps.has_sse = 1
util_cpu_caps.has_sse2 = 1
util_cpu_caps.has_sse3 = 1
util_cpu_caps.has_ssse3 = 0
util_cpu_caps.has_sse4_1 = 0
util_cpu_caps.has_sse4_2 = 0
util_cpu_caps.has_avx = 0
util_cpu_caps.has_3dnow = 0
util_cpu_caps.has_3dnow_ext = 0
util_cpu_caps.has_altivec = 0

It seems this is highly cpu dependent, as this test doesn't crash on my other machine (core2) with 64bit fedora ( there is just hang forever in __lll_lock_wait() from libpthread.so.0 on exit )