Created attachment 106352 [details] dmesg ==System Environment== -------------------------- Regression: Yes. They are igt regression issue Non-working platforms: IVB HSW BDW ==kernel== -------------------------- origin/drm-intel-nightly:43df30da20447e2856b2761215ff274886a9f931(fails) drm-intel-nightly: 2014y-09m-12d-15h-35m-20s UTC integration manifest origin/drm-intel-next-queued:648fb42d1e77cf90cddea2b1cf1f4ac4453e65e6(works) drm/i915: Extend BIOS stolen mem handling to all platform origin/drm-intel-fixes: e76c8c5979021483a3a6b480cfe360c644a403c0(works) drm/i915: Fix SRC_COPY width on 830/845g origin/drm-fixes: 83502a5d34386f7c6973bc70e1c423f55f5a2e3a(works) ==Bug detailed description== ----------------------------- igt/pm_rps subcases reset and blocking fail Output: root@x-bdw05:/GFX/Test/Intel_gpu_tools/intel-gpu-tools/tests# ./pm_rps --run-subtest reset IGT-Version: 1.8-g137877f (x86_64) (Linux: 3.17.0-rc4_drm-intel-nightly_43df30_20140916+ x86_64) Test assertion failure function idle_check, file pm_rps.c:408: Failed assertion: freqs[CUR] == freqs[MIN] error: 900 != 100 Subtest reset: FAIL (49.044s) root@x-bdw05:/GFX/Test/Intel_gpu_tools/intel-gpu-tools/tests# ./pm_rps --run-subtest blocking IGT-Version: 1.8-g137877f (x86_64) (Linux: 3.17.0-rc4_drm-intel-nightly_43df30_20140916+ x86_64) Test assertion failure function idle_check, file pm_rps.c:408: Failed assertion: freqs[CUR] == freqs[MIN] error: 900 != 100 Subtest blocking: FAIL (45.020s) ==Reproduce steps== ---------------------------- 1. ./pm_rps --run-subtest reset ==Bisect results== ---------------------------- Bisect shows: 3a1751ef34c32c5d288a328d855bec49ad0eaf9f is the first bad commit commit 3a1751ef34c32c5d288a328d855bec49ad0eaf9f Author: Chris Wilson <chris@chris-wilson.co.uk> AuthorDate: Fri Sep 12 06:46:28 2014 +0100 Commit: Chris Wilson <chris@chris-wilson.co.uk> CommitDate: Fri Sep 12 06:47:20 2014 +0100 igt/pm_rps: Fix the batch count for emitting the flush Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=81858#c2 Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Please confirm the bisect by testing igt commit 3b94d3f8ce21c92d3209e73ab48fda64beb0037e Author: Chris Wilson <chris@chris-wilson.co.uk> Date: Fri Aug 29 13:11:40 2014 +0100 igt: Prettify igt_assert_eq() failure messages
The failure is able to reproduce on latest -nightly(3e5e5913a80f593aa8f67784220eac2658034c6c) on BSW [root@x-bsw01 tests]# ./pm_rps --run-subtest reset IGT-Version: 1.8-g137877f (x86_64) (Linux: 3.17.0-rc5_drm-intel-nightly_3e5e59_20140918+ x86_64) Test assertion failure function idle_check, file pm_rps.c:408: Failed assertion: freqs[CUR] == freqs[MIN] error: 480 != 160 Subtest reset: FAIL (61.798s)
It also fails on BYT. IGT-Version: 1.8-gb7d80d1 (x86_64) (Linux: 3.17.0_drm-intel-nightly_ea4bec_20141011+ x86_64) Test assertion failure function idle_check, file pm_rps.c:407: Failed assertion: freqs[CUR] == freqs[MIN] error: 200 != 178 Subtest blocking: FAIL (45.475s)
I can't reproduce this on my BDW machine. Does it happen on all BDW machines you have? Can you please retest this?
Tests pass on IVB HSW and BDW, but still fail on BYT and BSW. [root@x-ivb9 tests]# ./pm_rps --run-subtest reset IGT-Version: 1.8-g5c7d54e (x86_64) (Linux: 3.17.0_drm-intel-nightly_2ea23c_20141015_debug+ x86_64) Subtest reset: SUCCESS (39.790s) root@x-bdw05:/GFX/Test/Intel_gpu_tools/intel-gpu-tools/tests# ./pm_rps --run-subtest reset IGT-Version: 1.8-ga0b5c6d (x86_64) (Linux: 3.17.0_drm-intel-nightly_2ea23c_20141016+ x86_64) Subtest reset: SUCCESS (50.520s) [root@x-hsw27 tests]# ./pm_rps --run-subtest reset IGT-Version: 1.8-g5c7d54e (x86_64) (Linux: 3.17.0_drm-intel-fixes_bfe01a_20141015+ x86_64) Subtest reset: SUCCESS (44.795s) [root@x-hsw27 tests]# ./pm_rps --run-subtest blocking IGT-Version: 1.8-g5c7d54e (x86_64) (Linux: 3.17.0_drm-intel-fixes_bfe01a_20141015+ x86_64) Subtest blocking: SUCCESS (35.022s) [root@x-bsw01 tests]# ./pm_rps --run-subtest reset IGT-Version: 1.8-ga0b5c6d (x86_64) (Linux: 3.17.0_drm-intel-nightly_2ea23c_20141014+ x86_64) Test assertion failure function loaded_check, file pm_rps.c:430: Failed assertion: freqs[CUR] == freqs[MAX] error: 416 != 624 Subtest reset: FAIL (44.044s) Test assertion failure function load_helper_stop, file pm_rps.c:253: Failed assertion: igt_wait_helper(&lh.igt_proc) == 0 Last errno: 10, No child processes pm_rps: igt_core.c:852: igt_fail: Assertion `!test_with_subtests || in_fixture' failed. Aborted (core dumped) [root@x-bsw01 tests]# ./pm_rps --run-subtest blocking IGT-Version: 1.8-ga0b5c6d (x86_64) (Linux: 3.17.0_drm-intel-nightly_2ea23c_20141014+ x86_64) Subtest blocking: SUCCESS (35.415s) root@x-byt06:/GFX/Test/Intel_gpu_tools/intel-gpu-tools/tests# ./pm_rps --run-subtest reset IGT-Version: 1.8-g5c7d54e (x86_64) (Linux: 3.17.0_drm-intel-nightly_2ea23c_20141016+ x86_64) Test assertion failure function matchit, file pm_rps.c:128: Failed assertion: freqs1[CUR] == freqs2[CUR] error: 200 != 222 Subtest reset: FAIL (49.263s) Test assertion failure function load_helper_stop, file pm_rps.c:253: Failed assertion: igt_wait_helper(&lh.igt_proc) == 0 Last errno: 10, No child processes pm_rps: igt_core.c:852: igt_fail: Assertion `!test_with_subtests || in_fixture' failed. Aborted root@x-byt06:/GFX/Test/Intel_gpu_tools/intel-gpu-tools/tests# ./pm_rps --run-subtest blocking IGT-Version: 1.8-g5c7d54e (x86_64) (Linux: 3.17.0_drm-intel-nightly_2ea23c_20141016+ x86_64) Test assertion failure function idle_check, file pm_rps.c:407: Failed assertion: freqs[CUR] == freqs[MIN] error: 200 != 178 Subtest blocking: FAIL (45.492s)
This is just the usual byt/bsw rpm fail, the original regression has been resolved. Please file any byt/bsw runtime pm issues as separate bugs (but I think we have them already).
Verified on latest -nightly(3b6da75500fb10e59dfb153eba9845c05f8d82d6) on BDW, about the failure on byt and bsw. bug 86363 will tracked.
Closing old verified.
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.