Bug 100415

Summary: [BYT][BAT] few tests on kms_cursor_legacy fails always on CI with BYT (n2820)
Product: DRI Reporter: Jani Saarinen <jani.saarinen>
Component: DRM/IntelAssignee: Ville Syrjala <ville.syrjala>
Status: CLOSED FIXED QA Contact: Intel GFX Bugs mailing list <intel-gfx-bugs>
Severity: normal    
Priority: medium CC: intel-gfx-bugs
Version: DRI git   
Hardware: Other   
OS: All   
Whiteboard:
i915 platform: BYT i915 features: display/atomic

Description Jani Saarinen 2017-03-27 12:30:41 UTC
On CI

basic-busy-flip-before-cursor-atomic
basic-busy-flip-before-cursor-legacy
basic-flip-after-cursor-atomic
basic-flip-after-cursor-legacy
basic-flip-after-cursor-varying-size
basic-flip-before-cursor-atomic
basic-flip-before-cursor-legacy
basic-flip-before-cursor-varying-size

Documenting. 

This is known issues and cannot make system passing so document bug to get atomic enabled on BYT/VLV/CHV/BSW

Last seen on CI on https://patchwork.freedesktop.org/series/20634/

Has been tried number of methods but system is just slow on these tests.
Comment 1 Chris Wilson 2017-03-27 12:37:57 UTC
Is this that known? Afaik we only see it when enabing atomic modesetting for byt/bsw.
Comment 2 Jani Saarinen 2017-03-27 12:50:02 UTC
Hi, Yes, known when that will be enables. Was discussing with Ville that we could make bug and then enable so that no flickering on results. 

Always those tests fail on our CI BYT
Comment 3 Jani Saarinen 2017-03-27 12:50:55 UTC
So this is making but that we know to fail and we have not been able to changes that help on that spesific BYT device under CI.
Comment 4 Jani Saarinen 2017-04-05 06:09:33 UTC
Passes now on CI:
https://patchwork.freedesktop.org/series/20634/ but still 2 tests fails:
Test kms_cursor_legacy:
        Subgroup basic-flip-after-cursor-varying-size:
                pass       -> FAIL       (fi-byt-n2820) fdo#100415
        Subgroup basic-flip-before-cursor-varying-size:
                pass       -> FAIL       (fi-byt-n2820) fdo#100415
Comment 5 Jani Saarinen 2017-04-05 06:10:38 UTC
Assigning to Ville to make suggestions here.
Comment 6 Jani Saarinen 2017-04-05 06:12:15 UTC
See error:
Out	
IGT-Version: 1.18-gc774395 (x86_64) (Linux: 4.11.0-rc5-CI-Patchwork_4398+ x86_64)
Test requirement not met in function __real_main1388, file kms_cursor_legacy.c:1409:
Test requirement: !(n >= display.n_pipes)
Stack trace:
  #0 [__igt_fail_assert+0x101]
  #1 [basic_flip_cursor+0x608]
  #2 [<unknown>+0x608]
Subtest basic-flip-after-cursor-varying-size: FAIL (0.223s)
Err	
(kms_cursor_legacy:9923) CRITICAL: Test assertion failure function basic_flip_cursor, file kms_cursor_legacy.c:611:
(kms_cursor_legacy:9923) CRITICAL: Failed assertion: get_vblank(display->drm_fd, pipe, 0) == vblank_start
(kms_cursor_legacy:9923) CRITICAL: error: 24534 != 24533
Subtest basic-flip-after-cursor-varying-size failed.
**** DEBUG ****
(kms_cursor_legacy:9923) DEBUG: Test requirement passed: found
(kms_cursor_legacy:9923) igt-kms-DEBUG: display: HDMI-A-1: set_pipe(A)
(kms_cursor_legacy:9923) igt-kms-DEBUG: display: HDMI-A-1: Selecting pipe A
(kms_cursor_legacy:9923) igt-fb-DEBUG: igt_create_fb_with_bo_size(width=1600, height=900, format=0x34325258, tiling=0x0, size=0)
(kms_cursor_legacy:9923) igt-fb-DEBUG: igt_create_fb_with_bo_size(handle=1, pitch=6400)
(kms_cursor_legacy:9923) igt-kms-DEBUG: display: A.0: plane_set_fb(70)
(kms_cursor_legacy:9923) DEBUG: Test requirement passed: set_fb_on_crtc(display, pipe, &fb_info)
(kms_cursor_legacy:9923) igt-fb-DEBUG: igt_create_fb_with_bo_size(width=64, height=64, format=0x34325241, tiling=0x0, size=0)
(kms_cursor_legacy:9923) igt-fb-DEBUG: igt_create_fb_with_bo_size(handle=2, pitch=256)
(kms_cursor_legacy:9923) DEBUG: Test requirement passed: cursor
(kms_cursor_legacy:9923) igt-kms-DEBUG: display: A.3: plane_set_fb(71)
(kms_cursor_legacy:9923) DEBUG: Test requirement passed: !(width <= 64 && height <= 64)
(kms_cursor_legacy:9923) igt-fb-DEBUG: igt_create_fb_with_bo_size(width=256, height=256, format=0x34325241, tiling=0x0, size=0)
(kms_cursor_legacy:9923) igt-fb-DEBUG: igt_create_fb_with_bo_size(handle=3, pitch=1024)
(kms_cursor_legacy:9923) igt-kms-DEBUG: display: commit {
(kms_cursor_legacy:9923) igt-kms-DEBUG: display:     A: Setting mode 1600x900 from HDMI-A-1
(kms_cursor_legacy:9923) igt-kms-DEBUG: display:     populating plane data: A.0, fb 70
(kms_cursor_legacy:9923) igt-kms-DEBUG: display:     src = (0, 0) 1600 x 900 dst = (0, 0) 1600 x 900
(kms_cursor_legacy:9923) igt-kms-DEBUG: display:     populating plane data: A.1, fb 0
(kms_cursor_legacy:9923) igt-kms-DEBUG: display:     populating plane data: A.2, fb 0
(kms_cursor_legacy:9923) igt-kms-DEBUG: display:     populating plane data: A.3, fb 71
(kms_cursor_legacy:9923) igt-kms-DEBUG: display:     src = (0, 0) 64 x 64 dst = (0, 0) 64 x 64
(kms_cursor_legacy:9923) igt-kms-DEBUG: display:     B: Setting NULL mode
(kms_cursor_legacy:9923) igt-kms-DEBUG: display:     populating plane data: B.0, fb 0
(kms_cursor_legacy:9923) igt-kms-DEBUG: display:     populating plane data: B.1, fb 0
(kms_cursor_legacy:9923) igt-kms-DEBUG: display:     populating plane data: B.2, fb 0
(kms_cursor_legacy:9923) igt-kms-DEBUG: display:     populating plane data: B.3, fb 0
(kms_cursor_legacy:9923) igt-kms-DEBUG: display:     DP-1: preparing atomic, pipe: None
(kms_cursor_legacy:9923) igt-kms-DEBUG: display:     HDMI-A-1: preparing atomic, pipe: A
(kms_cursor_legacy:9923) igt-kms-DEBUG: display: }
(kms_cursor_legacy:9923) CRITICAL: Test assertion failure function basic_flip_cursor, file kms_cursor_legacy.c:611:
(kms_cursor_legacy:9923) CRITICAL: Failed assertion: get_vblank(display->drm_fd, pipe, 0) == vblank_start
(kms_cursor_legacy:9923) CRITICAL: error: 24534 != 24533
****  END  ****
Comment 7 Jani Saarinen 2017-04-12 12:23:19 UTC
With decisions made that cpumax=2 helps fixing 2 last issues.
And plane optimizations done by Ville.
Comment 8 Chris Wilson 2017-04-12 12:29:37 UTC
Just to be clear, you are not suggesting that maxcpu=2 is a fix for this?
Comment 9 Jani Saarinen 2017-04-12 13:25:46 UTC
Well at least those two last cases 
est kms_cursor_legacy:
        Subgroup basic-flip-after-cursor-varying-size:
        Subgroup basic-flip-before-cursor-varying-size:

that was left after Ville's fix was fixed by that change. Ofcourse eventually we need to fix somehow on driver side I guess.

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.