Saw intermittent segfault running GLBenchmark 3.0's Manhattan test. Narrowed issue down to writing outside the memory bounds of the texture when decompressing ETC2 data. Probably the root cause of bug 71002 (which has a nice repro) as well.
See patch mesa-add-bounds-checking-to-eliminate-buffer-overrun on mesadev for more details.
manhattan has been working well, this was fixed:
--- 8< ---
Author: Courtney Goeltzenleuchter <firstname.lastname@example.org>
Date: Tue Apr 8 09:10:09 2014 -0600
mesa: add bounds checking to eliminate buffer overrun
Decompressing ETC2 textures was causing intermitent segfault
by copying resulting 4x4 texel block to the destination texture
regardless of the size of the destination texture. Issue found
via application crash in GLBenchmark 3.0's Manhattan test.
v2: add more detail comment. Compute limit outside inner loops.
v3: add bugzilla reference
v4: Correct cc syntax in commit log
v5: really grab the right patch
Cc: "9.2 10.0 10.1" <email@example.com>
Reviewed-by: Kenneth Graunke <firstname.lastname@example.org>
Reviewed-by: Ian Romanick <email@example.com> [v1, suggested v2-3]