Created attachment 48195 [details] backtraces from bioshock I have been trying to get the game Bioshock (running in Wine) to run with Mesa drivers. The game will start, but when it's loading a map it just sits there, pegging the CPU. A backtrace seems to indicate that it's busy running the shader compiler. Backtraces usually differs, but only up to a point, eventually they all point back to do_common_optimization. This might be a regression. The game was running at one point on r300g, but I never made a note of the version of Mesa and Wine used, and so far haven't been able to find a good starting point for a bisect. (Earlier versions of r600g causes a GPU hang and AFAIK i965 have never worked with this game). Setting MESA_GLSL="nopt" will cause a crash in the driver, "nopfrag" or "nopvert" makes no difference on the bug, and using "dump" causes a crash in hash_table_string_hash (might be separate bug?) This is using Mesa 2fe39b46e73aea37152777fe11d489e0b1bc3f92 and Wine 1.3.22.
Created attachment 48196 [details] backtrace from MESA_GLSL=dump crash
Same problem here, try reverting: commit 6a35cbb656e0f8a2479a63eadefb1ab85f42d490 Author: Eric Anholt <eric@anholt.net> Date: Sat Apr 2 20:19:20 2011 -0700 glsl/opt_cpe: Reenable opt_copy_propagation_elements.cpp pass. Note: You may first have to check out mesa from commit 8ab1c5328b12e8b075f62599a84672024aaf2982 , as later commits introduce other problems here. Works for me.
Thanks for the tip! In that case it's probably the same as bug 34006 (Which I really should have figured out myself, having filed the bug in first case :/ )
*** This bug has been marked as a duplicate of bug 34006 ***
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.