Created attachment 130000 [details] Full log of the test The test consistently fails, complaining about assertion failure. Happens every time. To reproduce: ./kms_frontbuffer_tracking --r fbc-suspend The test should SKIP if the hardware doesn't support FBC. IGT-Version: 1.17-geb53664 (x86_64) (Linux: 4.10.0-CI-CI_DRM_2243_15ccc64+ x86_64) Primary screen: HDMI-A 1280x1024, crtc 31 FBC last action not supported Can't test PSR: no usable eDP screen. Sink CRC not supported: primary screen is not eDP FBC status: FBC disabled: no suitable CRTC for FBC (kms_frontbuffer_tracking:2397) CRITICAL: Test assertion failure function enable_prim_screen_and_wait, file kms_frontbuffer_tracking.c:1752: (kms_frontbuffer_tracking:2397) CRITICAL: Failed assertion: false (kms_frontbuffer_tracking:2397) CRITICAL: FBC disabled CPU: HSW-4770R Display: DVI, connected to HDMI port. OS: Fedora 24 kernel: drm-tip: 2017y-02m-27d-07h-11m-41s
Please change back if you disagree. *** This bug has been marked as a duplicate of bug 99169 ***
99169 looks like a different failure mode - timeout: Timed out: CRC reading Subtest fbc-suspend failed. This report is a failed assertion due to fbc not being supported and fails a little earlier. I will reopen so that others can see and confirm.
Same issue on HSW-4770R
I will no longer be able to report on this issue - sorry.
(In reply to Dorota Czaplejewicz from comment #4) > I will no longer be able to report on this issue - sorry. Hi Dorota, I can reproduce this on my setup. I can follow up on this from now on. The output is a little different but it hits the same assertion failure. Can you check if it is the same? This was on SNB.
Created attachment 130737 [details] Attempted reproduction on SNB
I think your report is something else; several asserts are next to each other. The main difference is that my log shows "FBC disabled: no suitable CRTC for FBC" but yours doesn't. According to some feedback I got before, the above "feature" warning means this test should SKIP. Your report, on the other hand, seems to fail on a "correctness" check fbc_wait_for_compression()
(In reply to Dorota Czaplejewicz from comment #7) > I think your report is something else; several asserts are next to each > other. I understand your point and I agree, The SNB issue should be handled separately, I'm opening a new bug for that. Still, I reproduced the bug you reported here on a Haswell this time. Bisecting Linux shows the failure started after f7e9b004b8a3 ("drm/i915/fbc: inline intel_fbc_can_choose()"), which introduced a new error handling for the CRTC FBC selection. The proposed fix is a patch to IGT (attached), which catches the error handling scenario introduced by that patch. Patch also submitted to the IGT list for review.
Created attachment 130821 [details] [review] Proposed fix - Rejected on the list
(In reply to krisman from comment #8) > (In reply to Dorota Czaplejewicz from comment #7) > > I think your report is something else; several asserts are next to each > > other. > > I understand your point and I agree, The SNB issue should be handled > separately, I'm opening a new bug for that. Opened Bug 100677 to track that issue.
(In reply to krisman from comment #9) > Created attachment 130821 [details] [review] [review] > Proposed fix - Rejected on the list Submitted new fix. https://lists.freedesktop.org/archives/intel-gfx/2017-May/128079.html
(In reply to krisman from comment #11) > (In reply to krisman from comment #9) > > Created attachment 130821 [details] [review] [review] [review] > > Proposed fix - Rejected on the list > > Submitted new fix. > > https://lists.freedesktop.org/archives/intel-gfx/2017-May/128079.html this is not solved. Chris requested we make it a patch to igt.
(In reply to krisman from comment #12) > (In reply to krisman from comment #11) > > (In reply to krisman from comment #9) > > > Created attachment 130821 [details] [review] [review] [review] [review] > > > Proposed fix - Rejected on the list > > > > Submitted new fix. > > > > https://lists.freedesktop.org/archives/intel-gfx/2017-May/128079.html > > this is not solved. Chris requested we make it a patch to igt. Changing component since is a IGT related patch.
(In reply to krisman from comment #12) > (In reply to krisman from comment #11) > > (In reply to krisman from comment #9) > > > Created attachment 130821 [details] [review] [review] [review] [review] > > > Proposed fix - Rejected on the list > > > > Submitted new fix. > > > > https://lists.freedesktop.org/archives/intel-gfx/2017-May/128079.html > > this is not solved. Chris requested we make it a patch to igt. We got stuck with people asking to change on igt and igt people asking for kernel fix. I wanna improve the fbc_status reporting interface to actually fix this once and for all.
There is a similar issue open regarding this issue which occurs is BDW 105682. Otherwise issue can be closed. Closing this bug. *** This bug has been marked as a duplicate of bug 105682 ***
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.