A couple of Intel-GFX-CI hosts started crashing in piglit tests when the following commits were taken in: fae925a49 Specify an explicit window size in builtin uniform tests. 3ecdac456 Fix bogus assertions in builtin uniform test generator. This might be an environment issue on CI, but marking as a behaviour change anyway. All failing tests can be seen in directory level at https://intel-gfx-ci.01.org/tree/drm-tip/IGT_4491/pig-* Approximately 28 tests fail on SKL, 26 on GLK, 6.5k on HSW and 4.5k on SNB. Examples: SNB https://intel-gfx-ci.01.org/tree/drm-tip/IGT_4491/pig-snb-2600/spec%40glsl-1.10%40execution%40built-in-functions%40fs-abs-float.html HSW https://intel-gfx-ci.01.org/tree/drm-tip/IGT_4491/pig-hsw-4770r/spec%40arb_gpu_shader_int64%40execution%40built-in-functions%40cs-abs-i64vec2.html SKL https://intel-gfx-ci.01.org/tree/drm-tip/IGT_4491/pig-skl-6600/spec%40glsl-1.10%40execution%40built-in-functions%40fs-acos-float.html GLK https://intel-gfx-ci.01.org/tree/drm-tip/IGT_4491/pig-glk-j5005/spec%40glsl-1.10%40execution%40built-in-functions%40fs-acos-float.html
This piglit patch was reverted a couple of hours after the regression.
(In reply to Mark Janes from comment #1) > This piglit patch was reverted a couple of hours after the regression. Thanks, we really need to stay on a fixed version of piglit that goes well with the mesa version we run.
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.