[...] OK: 46 FAIL: 1 SKIP: 0 TIMEOUT: 0 The output from the failed tests: 44/47 mesa:gallium / u_format_test FAIL 0.10 s (exit status 1)
Created attachment 564020 [details] build.log.xz
Created attachment 564022 [details] testlog.txt
Created attachment 564024 [details] emerge --info
Does it fail with 18.3.x? Can you try to bisect the failure if so?
(In reply to Matt Turner from comment #4) > Does it fail with 18.3.x? Can you try to bisect the failure if so? Nope, 18.3.3 passes tests just fine. The "44/47 mesa:gallium / u_format_test" failing in 19.0.0_rc2 is not included in 18.3.3 test suite. [...] >>> Test phase: media-libs/mesa-18.3.3 * abi_ppc_64.ppc64: running multilib-minimal_abi_src_test ninja -v -j34 -l32 -C /var/tmp/portage/media-libs/mesa-18.3.3/work/mesa-18.3.3-abi_ppc_64.ppc64 test ninja: Entering directory `/var/tmp/portage/media-libs/mesa-18.3.3/work/mesa-18.3.3-abi_ppc_64.ppc64' [1/16] /usr/bin/python3.6 ../mesa-18.3.3/bin/git_sha1_gen.py --output src/git_sha1.h [1/2] /usr/lib64/python-exec/python3.6/meson test --no-rebuild --print-errorlogs 1/42 mesa / u_atomic OK 0.04 s 2/42 mesa / roundeven OK 0.04 s 3/42 mesa / mesa-sha1 OK 0.04 s 4/42 mesa / fast_idiv_by_const OK 2.42 s 5/42 mesa / clear OK 0.05 s 6/42 mesa / collision OK 0.04 s 7/42 mesa / delete_and_lookup OK 0.06 s 8/42 mesa / delete_management OK 0.08 s 9/42 mesa / destroy_callback OK 0.08 s 10/42 mesa / insert_and_lookup OK 0.05 s 11/42 mesa / insert_many OK 0.04 s 12/42 mesa / null_destroy OK 0.04 s 13/42 mesa / random_entry OK 0.06 s 14/42 mesa / remove_key OK 0.09 s 15/42 mesa / remove_null OK 0.05 s 16/42 mesa / replacement OK 0.07 s 17/42 mesa / string_buffer OK 0.05 s 18/42 mesa / vma_random OK 0.09 s 19/42 mesa / set OK 0.12 s 20/42 mesa / shared-glapi-test OK 0.11 s 21/42 mesa / es2-ABI-check OK 0.22 s 22/42 mesa / nir_control_flow OK 0.05 s 23/42 mesa / nir_vars OK 0.13 s 24/42 mesa / glcpp test (unix) OK 1.42 s 25/42 mesa / glcpp test (windows) OK 1.43 s 26/42 mesa / glcpp test (oldmac) OK 1.55 s 27/42 mesa / glcpp test (bizarro) OK 1.50 s 28/42 mesa / blob_test OK 0.15 s 29/42 mesa / cache_test OK 4.81 s 30/42 mesa / general_ir_test OK 0.19 s 31/42 mesa / uniform_initializer_test OK 0.24 s 32/42 mesa / sampler_types_test OK 0.23 s 33/42 mesa / glsl compiler warnings OK 0.63 s 34/42 mesa / glsl optimization OK 0.69 s 35/42 mesa:glx / dispatch-index-check OK 0.27 s 36/42 mesa / glx-test OK 0.19 s 37/42 mesa / gbm-symbols-check OK 0.11 s 38/42 mesa / egl-symbols-check OK 0.21 s 39/42 mesa / egl-entrypoint-check OK 0.22 s 40/42 mesa / r300_compiler_test OK 0.18 s 41/42 mesa / st_renumerate_test OK 0.14 s 42/42 mesa / st-array-merge-test OK 0.10 s OK: 42 FAIL: 0 SKIP: 0 TIMEOUT: 0
Took it upstream.
Same for mesa-19.0.8 and mesa-19.1.2.
Created attachment 582898 [details] build.log.xz (19.0.8)
Created attachment 582900 [details] output from u_format_test
Created attachment 589018 [details] build.log.xz (19.1.5, ppc64)
Created attachment 611540 [details] build.log.xz (20.0.0_rc1, ppc64) Re-tested with 20.0.0_rc1 and a more recent toolchain (gcc 9.2.0, glibc 2.29, binutils 2.32, llvm 9.0.1). [...] 28/68 mesa:format / u_format_test FAIL 0.04 s (exit status 1) 40/68 mesa:compiler+nir / load_store_vectorizer FAIL 0.05 s (exit status 1) 57/68 mesa:llvmpipe / lp_test_format FAIL 3.10 s (exit status 1) 58/68 mesa:llvmpipe / lp_test_arit FAIL 0.53 s (exit status 1) Ok: 64 Expected Fail: 0 Fail: 4 Unexpected Pass: 0 Skipped: 0 Timeout: 0
Created attachment 638206 [details] build.log.xz (20.1.0_rc2, ppc64) 20.1.0_rc2 fails one test less. [...] 29/69 mesa:format / u_format_test FAIL 0.29 s (exit status 1) 58/69 mesa:llvmpipe / lp_test_format FAIL 13.56 s (exit status 1) 59/69 mesa:llvmpipe / lp_test_arit FAIL 1.82 s (exit status 1) Ok: 66 Expected Fail: 0 Fail: 3 Unexpected Pass: 0 Skipped: 0 Timeout: 0
Created attachment 638208 [details] emerge --info
*** Bug 758869 has been marked as a duplicate of this bug. ***
Looks more like a general 32bit issue, not?
(In reply to Thomas Deutschmann from comment #15) > Looks more like a general 32bit issue, not? No. The PowerPC failures are due to endianness. The x86 failures are due to x87 FPU precision differences (and are not going to be fixed, short of using -mfpmath=sse -msse2).
Created attachment 677956 [details] build.log.xz (20.2.4, ppc64)
Created attachment 686094 [details] build.log.xz (20.2.6, ppc)
same on sparc
This is now fixed upstream by https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7776 Will be in 21.2.
Created attachment 727341 [details] build.log.xz (21.2.0-rc2, ppc64) Not quite there yet... At least on my Talos II. I'll update the upstream bug accordingly. [...] 72/82 mesa:llvmpipe / lp_test_arit FAIL 0.63s exit status 1 >>> MALLOC_PERTURB_=163 /var/tmp/portage/media-libs/mesa-21.2.0_rc2/work/mesa-21.2.0-rc2-.ppc64/src/gallium/drivers/llvmpipe/lp_test_arit 77/82 mesa:amd / ac_surface_modifier_test FAIL 1.04s killed by signal 6 SIGABRT >>> MALLOC_PERTURB_=138 /var/tmp/portage/media-libs/mesa-21.2.0_rc2/work/mesa-21.2.0-rc2-.ppc64/src/amd/common/ac_surface_modifier_test 80/82 mesa:llvmpipe / lp_test_format FAIL 3.54s exit status 1 >>> MALLOC_PERTURB_=154 /var/tmp/portage/media-libs/mesa-21.2.0_rc2/work/mesa-21.2.0-rc2-.ppc64/src/gallium/drivers/llvmpipe/lp_test_format Ok: 79 Expected Fail: 0 Fail: 3 Unexpected Pass: 0 Skipped: 0 Timeout: 0
Created attachment 727344 [details] emerge --info
Created attachment 727347 [details] testlog.txt (21.2.0-rc2, ppc64)
Oops, sorry I got confused... This original bug report was about "u_format_test" test failing. Which passes now in 21.2.0-rc2 as Matt stated in Comment #20. What is left failing now is "lp_test_format". Also I realize now I linked the wrong upstream bug report here, confusing these two again. Will correct this.
It's still there on sparc: 37/72 mesa:format / u_format_test FAIL 4.82s exit status 1 >>> MALLOC_PERTURB_=69 /var/tmp/portage/media-libs/mesa-21.1.6/work/mesa-21.1.6-.sparc32/src/util/tests/format/u_format_test
(In reply to Rolf Eike Beer from comment #25) > It's still there on sparc: > > 37/72 mesa:format / u_format_test FAIL 4.82s > exit status 1 > >>> MALLOC_PERTURB_=69 /var/tmp/portage/media-libs/mesa-21.1.6/work/mesa-21.1.6-.sparc32/src/util/tests/format/u_format_test Should be gone in 21.2.0. I think upstream explicitly mentioned testing on sparc.
Created attachment 758842 [details] build.log.xz (21.3.1-r1, ppc) Also tested on 32bit ppc, u_format_test passes meanwhile. Also passes on ppc64 and x86 as mentioned in the upstream bugreport. [...] 37/79 mesa:format / u_format_test OK 0.74s 38/79 mesa:compiler+nir / nir_builder OK 0.71s 39/79 mesa:compiler+nir / nir_core OK 0.46s 40/79 mesa:compiler+nir / nir_vars OK 0.38s 41/79 mesa:compiler+nir / nir_control_flow OK 0.78s 42/79 mesa:mapi / shared-glapi symbols check OK 1.36s 43/79 mesa:compiler+nir / negative_equal OK 0.64s 44/79 mesa:compiler+nir / comparison_pre OK 0.57s 45/79 mesa:compiler+nir / load_store_vectorizer OK 0.54s 46/79 mesa:compiler+nir / nir_serialize_test OK 0.58s 47/79 mesa:compiler+nir / nir_opt_if OK 0.58s 48/79 mesa:compiler+nir / nir_lower_returns OK 0.55s 49/79 mesa:compiler+nir / ssa_def_bits_used OK 0.53s 50/79 mesa:compiler+spirv / avail_vis OK 0.50s 51/79 mesa:compiler+spirv / volatile OK 0.53s 52/79 mesa:compiler+nir / algebraic OK 0.98s 53/79 mesa:compiler+glsl / general_ir_test OK 0.38s 54/79 mesa:compiler+glsl / uniform_initializer_test OK 0.27s 55/79 mesa:compiler+glsl / sampler_types_test OK 0.22s 56/79 mesa:compiler+glsl / list_iterators OK 0.23s 57/79 mesa:compiler+nir / nir_algebraic_parser OK 4.88s 58/79 mesa:compiler+glsl / glsl optimization OK 7.27s 59/79 mesa:mesa / main-test OK 0.56s 60/79 mesa:glx / dispatch-index-check OK 1.31s 61/79 mesa:glx / glx-test OK 0.64s 62/79 mesa:gbm / gbm-abi-check OK 0.18s 63/79 mesa:gbm / gbm-symbols-check OK 2.07s 64/79 mesa:r300 / r300_compiler_test OK 0.51s 65/79 mesa:gallium / pipe_barrier_test OK 1.11s 66/79 mesa:gallium / u_half_test OK 0.26s 67/79 mesa:gallium / u_prim_verts_test OK 0.22s 68/79 mesa:st_mesa / st_format_test OK 0.28s 69/79 mesa:compiler+glsl / glsl compiler warnings OK 16.53s 70/79 mesa:st_mesa / st_renumerate_test OK 0.36s 71/79 mesa:st_mesa / st-array-merge-test OK 0.42s 72/79 mesa:egl / egl-symbols-check OK 1.89s 73/79 mesa:egl / egl-entrypoint-check OK 1.76s 74/79 mesa:compiler+glcpp / glcpp test (windows) OK 22.76s 75/79 mesa:compiler+glcpp / glcpp test (unix) OK 23.10s 76/79 mesa:compiler+glcpp / glcpp test (oldmac) OK 23.05s 77/79 mesa:compiler+glcpp / glcpp test (bizarro) OK 22.90s 78/79 mesa:util / sparse_array_multi_threaded OK 26.24s 79/79 mesa:util / fast_idiv_by_const OK 27.84s Ok: 79 Expected Fail: 0 Fail: 0 Unexpected Pass: 0 Skipped: 0 Timeout: 0