ANV: regressions in unit tests
Since 356a60bd from !17897 (merged) was merged, ninja test
when configured with -Dbuild-tests=true
has started failing:
[128/129] Running all tests.
1/58 mesa:util / process OK 0.02s
2/58 mesa:util / clear OK 0.02s
3/58 mesa:util / collision OK 0.02s
4/58 mesa:util / delete_and_lookup OK 0.01s
5/58 mesa:util / delete_management OK 0.01s
6/58 mesa:util / destroy_callback OK 0.01s
7/58 mesa:util / xmlconfig OK 0.06s
8/58 mesa:util / insert_and_lookup OK 0.03s
9/58 mesa:util / insert_many OK 0.03s
10/58 mesa:util / null_destroy OK 0.03s
11/58 mesa:util / random_entry OK 0.02s
12/58 mesa:util / remove_key OK 0.02s
13/58 mesa:util / remove_null OK 0.02s
14/58 mesa:util / replacement OK 0.04s
15/58 mesa:util / vma_random OK 0.04s
16/58 mesa:format / srgb OK 0.04s
17/58 mesa:format / u_format_compatible_test OK 0.03s
18/58 mesa:mapi / shared-glapi-test OK 0.03s
19/58 mesa:format / u_format_test OK 0.08s
20/58 mesa:compiler+nir / load_store_vectorizer OK 0.03s
21/58 mesa:compiler+nir / nir_opt_dce OK 0.03s
22/58 mesa:compiler+spirv / spirv_tests OK 0.02s
23/58 mesa:compiler+nir / nir_tests OK 0.11s
24/58 mesa:mapi / shared-glapi symbols check OK 0.25s
25/58 mesa:mapi / es1-ABI-check OK 0.25s
26/58 mesa:mapi / es2-ABI-check OK 0.25s
27/58 mesa:compiler+glsl / general_ir_test OK 0.03s
28/58 mesa:compiler+glsl / sampler_types_test OK 0.03s
29/58 mesa:compiler+glsl / list_iterators OK 0.05s
30/58 mesa:compiler+nir / nir_algebraic_parser OK 0.33s
31/58 mesa:intel / intel_device_info_test OK 0.01s
32/58 mesa:intel / isl_surf_get_image_offset OK 0.02s
33/58 mesa:intel / isl_aux_info OK 0.02s
34/58 mesa:intel / genxml_test OK 0.01s
35/58 mesa:compiler+glcpp / glcpp test (unix) OK 0.37s
36/58 mesa:compiler+glcpp / glcpp test (windows) OK 0.40s
37/58 mesa:compiler+glcpp / glcpp test (oldmac) OK 0.41s
38/58 mesa:compiler+glsl / glsl optimization OK 0.22s
39/58 mesa:intel / anv symbols check OK 0.19s
40/58 mesa:compiler+glcpp / glcpp test (bizarro) OK 0.44s
41/58 mesa:compiler+glsl / glsl compiler warnings OK 0.75s
42/58 mesa:intel / anv_state_pool_free_list_only FAIL 2.88s killed by signal 11 SIGSEGV
>>> MALLOC_PERTURB_=47 /home/kusma/build/mesa/zink-anv/src/intel/vulkan/state_pool_free_list_only
43/58 mesa:intel / anv_block_pool_no_free FAIL 3.02s killed by signal 11 SIGSEGV
>>> MALLOC_PERTURB_=211 /home/kusma/build/mesa/zink-anv/src/intel/vulkan/block_pool_no_free
44/58 mesa:intel / anv_state_pool_padding FAIL 2.47s killed by signal 11 SIGSEGV
>>> MALLOC_PERTURB_=230 /home/kusma/build/mesa/zink-anv/src/intel/vulkan/state_pool_padding
45/58 mesa:intel / anv_block_pool_grow_first FAIL 2.99s killed by signal 11 SIGSEGV
>>> MALLOC_PERTURB_=29 /home/kusma/build/mesa/zink-anv/src/intel/vulkan/block_pool_grow_first
46/58 mesa:intel / anv_state_pool_no_free FAIL 2.98s killed by signal 11 SIGSEGV
>>> MALLOC_PERTURB_=198 /home/kusma/build/mesa/zink-anv/src/intel/vulkan/state_pool_no_free
47/58 mesa:intel / anv_state_pool FAIL 2.85s killed by signal 11 SIGSEGV
>>> MALLOC_PERTURB_=189 /home/kusma/build/mesa/zink-anv/src/intel/vulkan/state_pool
48/58 mesa:gbm / gbm-abi-check OK 0.04s
49/58 mesa:zink / zink_format_test OK 0.04s
50/58 mesa:glx / dispatch-index-check OK 0.05s
51/58 mesa:util / util_tests OK 3.70s
52/58 mesa:mesa / main-test OK 0.06s
53/58 mesa:st_mesa / st_format_test OK 0.05s
54/58 mesa:glx / glx-test OK 0.11s
55/58 mesa:egl / egl-entrypoint-check OK 0.07s
56/58 mesa:gbm / gbm-symbols-check OK 0.14s
57/58 mesa:egl / egl-symbols-check OK 0.13s
58/58 mesa:intel / intel_compiler_tests OK 4.50s
Summary of Failures:
42/58 mesa:intel / anv_state_pool_free_list_only FAIL 2.88s killed by signal 11 SIGSEGV
43/58 mesa:intel / anv_block_pool_no_free FAIL 3.02s killed by signal 11 SIGSEGV
44/58 mesa:intel / anv_state_pool_padding FAIL 2.47s killed by signal 11 SIGSEGV
45/58 mesa:intel / anv_block_pool_grow_first FAIL 2.99s killed by signal 11 SIGSEGV
46/58 mesa:intel / anv_state_pool_no_free FAIL 2.98s killed by signal 11 SIGSEGV
47/58 mesa:intel / anv_state_pool FAIL 2.85s killed by signal 11 SIGSEGV
It seems we no longer run these tests on CI, which seems to be why this wasn't caught earlier.