Full list of tests that crash can be found here: https://mesa-ci.01.org/mesa_master/builds/17304/group/63a9f0ea7bb98050796b649e85481845#fails Test output: Test case 'dEQP-GLES3.functional.transform_feedback.array.separate.points.lowp_mat4x2'.. deqp-gles3: ../src/compiler/nir/nir_opt_find_array_copies.c:122: node_for_deref: Assertion `idx < parent->num_children' failed. Bisected to the following commit: commit 156306e5e62fb4fda0e829c5798fffdf7eb1907c Author: Connor Abbott <cwabbott0@gmail.com> Date: Tue Jun 18 12:12:49 2019 +0200 nir/find_array_copies: Handle wildcards and overlapping copies
This should be fixed by https://gitlab.freedesktop.org/mesa/mesa/merge_requests/1501.
Looks good to me, tests are now passing.
Upon closer inspection, it seems the following tests are still failing on all i965 platforms even after MR 1501 was merged: deqp: dEQP-GLES31.functional.shaders.linkage.es31.geometry.varying.types.float_struct dEQP-GLES31.functional.shaders.linkage.es31.geometry.varying.types.mat3x2 dEQP-GLES31.functional.shaders.linkage.es31.tessellation_geometry.varying.types.float_struct dEQP-GLES31.functional.shaders.linkage.es31.tessellation_geometry.varying.types.mat3x2 dEQP-GLES31.functional.shaders.linkage.es32.geometry.varying.types.float_struct dEQP-GLES31.functional.shaders.linkage.es32.geometry.varying.types.mat3x2 dEQP-GLES31.functional.shaders.linkage.es32.tessellation_geometry.varying.types.float_struct dEQP-GLES31.functional.shaders.linkage.es32.tessellation_geometry.varying.types.mat3x2 piglit: spec.glsl-1_50.execution.variable-indexing.gs-input-array-vec2-index-rd spec.glsl-1_50.execution.variable-indexing.vs-output-array-vec2-index-wr-before-gs spec.glsl-1_50.execution.varying-struct-basic-vs-gs stderr from tests can be viewed by putting in the test name in the search box at the top of this page: https://mesa-ci.01.org/mesa_master_daily/builds/5141/group/63a9f0ea7bb98050796b649e85481845 For example the output from dEQP-GLES31.functional.shaders.linkage.es31.geometry.varying.types.float_struct: https://mesa-ci.01.org/mesa_master_daily/builds/5141/results/741381931
I created https://gitlab.freedesktop.org/mesa/mesa/merge_requests/1521 which I think fixes the remaining errors.
I'm running that patch through CI now. FYI, CI is still set up to trigger builds when you (Conner) push to the 'intel-jenkins' branch in your mesa repo, and you can use that in the future to test any patches/fixed before merging.
Looks like MR 1521 fixes the remaining tests in CI.
Pushed to master.
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.