https://blogs.gentoo.org/ago/2020/07/04/gentoo-tinderbox/ Issue: media-libs/opus-1.5.1-r1 fails tests. Discovered on: arm (internal ref: tinderbox_arm) Info about the issue: https://wiki.gentoo.org/wiki/Project:Tinderbox/Common_Issues_Helper#CF0015
Created attachment 886778 [details] build.log build log and emerge --info
Created attachment 886779 [details] 1-testlog.txt 1-testlog.txt
i can't reproduce it here: >>> Test phase: media-libs/opus-1.5.1-r1 * abi_x86_32.x86: running multilib-minimal_abi_src_test meson test --print-errorlogs --num-processes 33 --timeout-multiplier=2 ninja: Entering directory `/var/tmp/portage/media-libs/opus-1.5.1-r1/work/opus-1.5.1-abi_x86_32.x86' ninja: no work to do. 1/15 test_unit_types OK 0.03s 2/15 test_unit_mathops OK 0.03s 3/15 test_unit_laplace OK 0.02s 4/15 test_unit_rotation OK 0.02s 5/15 test_opus_padding OK 0.01s 6/15 test_unit_dft OK 0.03s 7/15 test_opus_projection OK 0.03s 8/15 test_unit_mdct OK 0.11s 9/15 test_unit_cwrs32 OK 0.99s 10/15 test_unit_entropy OK 1.16s 11/15 test_unit_cwrs32 OK 2.99s 12/15 test_opus_api OK 7.99s 13/15 test_opus_extensions OK 13.67s 14/15 test_opus_decode OK 25.94s 15/15 test_opus_encode OK 32.03s Ok: 15 Expected Fail: 0 Fail: 0 Unexpected Pass: 0 Skipped: 0 Timeout: 0 Full log written to /var/tmp/portage/media-libs/opus-1.5.1-r1/work/opus-1.5.1-abi_x86_32.x86/meson-logs/testlog.txt * abi_x86_64.amd64: running multilib-minimal_abi_src_test meson test --print-errorlogs --num-processes 33 --timeout-multiplier=2 ninja: Entering directory `/var/tmp/portage/media-libs/opus-1.5.1-r1/work/opus-1.5.1-abi_x86_64.amd64' ninja: no work to do. 1/15 test_unit_types OK 0.03s 2/15 test_unit_mathops OK 0.03s 3/15 test_unit_laplace OK 0.03s 4/15 test_unit_dft OK 0.03s 5/15 test_unit_rotation OK 0.03s 6/15 test_opus_padding OK 0.01s 7/15 test_opus_projection OK 0.02s 8/15 test_unit_mdct OK 0.09s 9/15 test_unit_cwrs32 OK 0.80s 10/15 test_unit_entropy OK 1.03s 11/15 test_unit_cwrs32 OK 2.47s 12/15 test_opus_api OK 7.24s 13/15 test_opus_extensions OK 12.76s 14/15 test_opus_decode OK 28.08s 15/15 test_opus_encode OK 31.51s Ok: 15 Expected Fail: 0 Fail: 0 Unexpected Pass: 0 Skipped: 0 Timeout: 0
(In reply to Miroslav Šulc from comment #3) > i can't reproduce it here: did you try on the arch where it is supposed to fail (arm) ?
(In reply to Agostino Sarubbo from comment #4) > (In reply to Miroslav Šulc from comment #3) > > i can't reproduce it here: > > did you try on the arch where it is supposed to fail (arm) ? oh, sorry, i didn't notice that. i tested on amd64. no arm here.
I see this as well on other arches, seems codepath is much less-optimized on non-amd64. E.x. ppc64 BE: 14/15 test_opus_encode OK 639.72s 15/15 test_opus_decode OK 991.52s ppc64 LE: 14/15 test_opus_encode OK 221.68s 15/15 test_opus_decode OK 384.82s I would say timeout-multiplier needs to go to at least 20 or so instead of 2.
loong: 14/15 test_opus_encode OK 422.98s 15/15 test_opus_decode OK 479.72s These are some of my faster machines too, probably 2x or more faster single-threaded than the Cortex-A72 I test on for arm.
