Created attachment 31347 [details]
screenshot of etqw
In the games ETQW and Quake 4, some objects have a pink tint when viewed from a distance. In the attached screenshot it can be seen on the wall, almost in the middle of the image.
-- chipset: G45 / ICH10R
-- system architecture: 32-bit
-- Linux distribution: Debian unstable
-- Machine or mobo model: Asus P5Q-EM
-- Display connector: DVI
-- KMS: enabled
-- xf86-video-intel: 14109abf285866ad4cd99d0cd16b0954a0a73a62
-- xserver: 1.7.0
-- mesa: e8eec9385497053a4172deeff9c0ff2726a10f3b
-- drm: 83a35b68f45cebc70152e55ed3f99db485c9a7cd
-- kernel: 2.6.32-rc7 (debian)
Yeah, I've seen this too.
Trimmed trace for this failure on ivb at http://people.freedesktop.org/~anholt/etqw-trim2.trace -- there are 2 different draw commands where if either is removed, the failure looks like it goes away. One is a 6-vert TRIANGLES, and the other is a big draw operation that presumably is drawing the building I'm looking at.
Author: Eric Anholt <email@example.com>
Date: Wed Feb 13 21:21:45 2013 -0800
i965/gen7: Set up all samplers even if samplers are sparsely used.
In GLSL, sampler indices are allocated contiguously from 0. But in the
case of ARB_fragment_program (and possibly fixed function), an app that
uses texture 0 and 2 will use sampler indices 0 and 2, so we were only
allocating space for samplers 0 and 1 and setting up sampler 0. We
would read garbage for sampler 2, resulting in flickering textures and
an angry simulator.
Fixes bad rendering in 0 A.D. and ETQW. This was fixed for pre-gen7 by
Reviewed-by: Kenneth Graunke <firstname.lastname@example.org>
NOTE: This is a candidate for stable branches.
Cherry-picked to the 9.0 branch: be7048477a2989bd11ab590c65ea1515f764a86c
It will be available in Mesa 9.0.3.