Created attachment 136010 [details] kern_log_from_cycle In CNL we're getting this message in our FF tests. [ 309.667455] [drm:intel_cpu_fifo_underrun_irq_handler [i915]] *ERROR* CPU pipe A FIFO underrun It have been appearing at least a couple of time in the dmesg by cycle, and though the system capture the warns while executing a specific test, it don't seems to be triggered by that test, since repeating the test various times won't give it. Some of the test that have seen the messages at least once over the week: igt@gem_exec_flush@basic-wb-prw-default - commit-1a0d67e igt@gem_sync@basic-each - commit-1a0d67e igt@gem_exec_create@basic -commit-807db75 igt@gem_exec_flush@basic-uc-ro-default -commit-0d0fe91 igt@gem_exec_flush@basic-batch-kernel-default-uc -commit-0645c6d igt@gem_exec_flush@basic-uc-set-default -commit-6d6c48b igt@gem_sync@basic-all - -commit-807db75
This test have the same dmesg-warn on CNL QA using IGT-IGT-Version: 1.20-g1db1246 (x86_64) (Linux: 4.15.0-rc2-drm-tip-ww49-commit-66be577+ x86_64) test list: igt@gem_exec_await@wide-all igt@gem_exec_flush@batch-cpu-bsd-uc igt@gem_exec_flush@batch-kernel-render-uc igt@gem_exec_flush@batch-user-vebox-uc igt@gem_exec_flush@batch-wc-blt-wb igt@gem_exec_flush@batch-wc-bsd-wb igt@gem_exec_flush@batch-wc-vebox-uc igt@gem_exec_flush@stream-pro-blt igt@gem_exec_flush@stream-prw-blt igt@gem_exec_flush@stream-prw-bsd igt@gem_exec_flush@stream-prw-render-interruptible igt@gem_exec_flush@stream-rw-before-blt igt@gem_exec_flush@stream-rw-before-blt-interruptible igt@gem_exec_flush@stream-set-blt-interruptible igt@gem_exec_flush@stream-set-vebox-interruptible igt@gem_exec_flush@uc-pro-bsd-interruptible igt@gem_exec_flush@uc-pro-vebox igt@gem_exec_flush@uc-prw-vebox igt@gem_exec_flush@uc-ro-before-bsd igt@gem_exec_flush@uc-ro-before-vebox igt@gem_exec_flush@uc-ro-bsd igt@gem_exec_flush@uc-rw-before-blt-interruptible igt@gem_exec_flush@uc-rw-blt igt@gem_exec_flush@uc-set-blt-interruptible igt@gem_exec_flush@wb-ro-before-vebox igt@gem_exec_flush@wb-ro-bsd igt@gem_exec_flush@wb-rw-before-blt-interruptible igt@gem_exec_flush@wb-rw-blt igt@gem_exec_flush@wb-set-bsd-interruptible igt@gem_exec_flush@wb-set-vebox-interruptible igt@gem_exec_flush@wc-pro-vebox igt@gem_exec_flush@wc-prw-blt igt@gem_exec_flush@wc-prw-blt-interruptible igt@gem_exec_flush@wc-prw-vebox-interruptible igt@gem_exec_flush@wc-ro-before-bsd igt@gem_exec_flush@wc-ro-before-vebox igt@gem_exec_flush@wc-ro-blt igt@gem_exec_flush@wc-ro-vebox igt@gem_exec_flush@wc-rw-blt igt@gem_exec_flush@wc-set-blt igt@gem_exec_flush@wc-set-bsd-interruptible igt@gem_exec_flush@wc-set-vebox-interruptible igt@gem_exec_gttfill@default igt@gem_exec_whisper@blt-contexts-forked igt@gem_exec_whisper@blt-forked igt@gem_exec_whisper@bsd-contexts-forked igt@gem_exec_whisper@contexts-priority igt@gem_exec_whisper@vebox-fds-priority igt@gem_exec_whisper@vebox-forked igt@gem_ring_sync_loop igt@gem_sync@forked-all igt@gem_sync@forked-each dmesg-warn: [ 70.917898] Setting dangerous option reset - tainting kernel [ 85.769493] [drm:intel_cpu_fifo_underrun_irq_handler [i915]] *ERROR* CPU pipe A FIFO underrun
With today's commit: igt@gem_sync@basic-all - b5f297e0 igt@gem_close_race@basic-threads - b5f297e0
igt@gvt_basic@invalid-placeholder-test - commit-8fa442b igt@gem_exec_flush@basic-batch-kernel-default-wb - commit-8fa442b
This test are dmesg-warning on CNL QA igt@kms_cursor_crc@cursor-128x128-rapid-movement igt@kms_cursor_crc@cursor-128x42-onscreen igt@kms_rotation_crc@sprite-rotation-270 using IGT-Version: 1.20-g7440741 (x86_64) (Linux: 4.15.0-rc3-drm-tip-ww50-commit-62d9383+ x86_64) intel-gpu-tools commit: 74407418720ff7a9de7caabec05d4c3afe9a5c51 [ 633.804498] [drm:intel_cpu_fifo_underrun_irq_handler [i915]] *ERROR* CPU pipe A FIFO underrun [ 634.916498] [drm:intel_cpu_fifo_underrun_irq_handler [i915]] *ERROR* CPU pipe A FIFO underrun
This test has admesg-warn on CNL QA igt@gem_sync@basic-each IGT-Version: 1.20-gc0be331 (x86_64) (Linux: 4.15.0-rc3-drm-intel-qa-ww50-commit-ad43db1+ x86_64) Using Execlists submission Has kernel scheduler - With priority sorting - With preemption enabled bsd completed 138240 cycles: 36.263 us vebox completed 141312 cycles: 35.510 us blt completed 151552 cycles: 33.172 us render completed 50176 cycles: 100.522 us Subtest basic-each: SUCCESS (5.144s) Stderr Environment PIGLIT_SOURCE_DIR="/home/gfx/intel-graphics/intel-gpu-tools/piglit" PIGLIT_PLATFORM="mixed_glx_egl" Command /home/gfx/intel-graphics/intel-gpu-tools/tests/gem_sync --run-subtest basic-each dmesg [ 296.441985] Setting dangerous option reset - tainting kernel [ 300.223391] [drm:intel_cpu_fifo_underrun_irq_handler [i915]] *ERROR* CPU pipe A FIFO underrun
With commit-e044c9e, test igt@gem_sync@basic-each, got the warn today.
igt@gem_sync@basic-all IGT-Version: 1.20-gd86d53b (x86_64) (Linux: 4.15.0-rc6-drm-intel-qa-ww1-commit-cb4a985+ x86_64) Using Execlists submission Has kernel scheduler - With priority sorting - With preemption enabled Completed 76800 cycles: 65.539 us Subtest basic-all: SUCCESS (5.119s) dmesg [ 299.506280] Setting dangerous option reset - tainting kernel [ 301.732360] [drm:intel_cpu_fifo_underrun_irq_handler [i915]] *ERROR* CPU pipe A FIFO underrun
In FF seems to be repeating with the same test this week: igt@gem_sync@basic-each, commit-11030d7 and commit-d373aa7 [ 297.165895] Setting dangerous option reset - tainting kernel [ 298.124755] [drm:intel_cpu_fifo_underrun_irq_handler [i915]] *ERROR* CPU pipe A FIFO underrun
The following test has the same dmesg-warn on CNL with latest configuration. igt@gem_streaming_writes@cpu-sync [ 304.496178] [drm:intel_cpu_fifo_underrun_irq_handler [i915]] *ERROR* CPU pipe A FIFO underrun
The following test case has the same dmesg-warn on CNL igt@gem_exec_whisper@normal [ 156.427644] Setting dangerous option reset - tainting kernel [ 177.729420] [drm:intel_cpu_fifo_underrun_irq_handler [i915]] *ERROR* CPU pipe A FIFO underrun
Is this reproducible when booting with nvme_core.default_ps_max_latency_us=0 module parameter?
(In reply to Jani Nikula from comment #11) > Is this reproducible when booting with nvme_core.default_ps_max_latency_us=0 > module parameter? Haven't tested the parameter, but we have a whole week that the message doesn't appear. I'll be monitoring it.
Is this still valid?
(In reply to Rodrigo Vivi from comment #13) > Is this still valid? Haven't seen it for almost three weeks, I'll close by the end of the week if same status.
(In reply to Elizabeth from comment #14) > (In reply to Rodrigo Vivi from comment #13) > > Is this still valid? > Haven't seen it for almost three weeks, I'll close by the end of the week if > same status. Closing. Thanks for your time.
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.