Bug 103857 - [CI] igt@perf_pmu@* - fail - Failed assertion: (double)(val[i]) <= (1.0 + tolerance) * (double)0.0f && (double)(val[i]) >= (1.0 - tolerance) * (double)0.0f
Summary: [CI] igt@perf_pmu@* - fail - Failed assertion: (double)(val[i]) <= (1.0 + tol...
Status: CLOSED FIXED
Alias: None
Product: DRI
Classification: Unclassified
Component: DRM/Intel (show other bugs)
Version: DRI git
Hardware: Other All
: medium normal
Assignee: Intel GFX Bugs mailing list
QA Contact: Intel GFX Bugs mailing list
URL:
Whiteboard: ReadyForDev
Keywords:
Depends on:
Blocks:
 
Reported: 2017-11-23 07:50 UTC by Marta Löfstedt
Modified: 2018-01-12 08:25 UTC (History)
1 user (show)

See Also:
i915 platform: BXT, GLK, KBL
i915 features: power/Other


Attachments

Description Marta Löfstedt 2017-11-23 07:50:16 UTC
https://intel-gfx-ci.01.org/tree/drm-tip/IGT_4004/shard-hsw6/igt@perf_pmu@most-busy-check-all-rcs0.html

(perf_pmu:6061) CRITICAL: Test assertion failure function most_busy_check_all, file perf_pmu.c:279:
(perf_pmu:6061) CRITICAL: Failed assertion: (double)(val[i]) <= (1.0 + tolerance) * (double)0.0f && (double)(val[i]) >= (1.0 - tolerance) * (double)0.0f
(perf_pmu:6061) CRITICAL: 'val[i]' != '0.0f' (2500000.000000 not within 5.000000% tolerance of 0.000000)
Subtest most-busy-check-all-rcs0 failed.
---
https://intel-gfx-ci.01.org/tree/drm-tip/IGT_4004/shard-snb5/igt@perf_pmu@most-busy-check-all-rcs0.html

(perf_pmu:11862) CRITICAL: Test assertion failure function most_busy_check_all, file perf_pmu.c:279:
(perf_pmu:11862) CRITICAL: Failed assertion: (double)(val[i]) <= (1.0 + tolerance) * (double)0.0f && (double)(val[i]) >= (1.0 - tolerance) * (double)0.0f
(perf_pmu:11862) CRITICAL: 'val[i]' != '0.0f' (2500000.000000 not within 5.000000% tolerance of 0.000000)
Subtest most-busy-check-all-rcs0 failed.
---
https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_3373/shard-hsw2/igt@perf_pmu@busy-check-all-bcs0.html
	
(perf_pmu:7876) CRITICAL: Test assertion failure function busy_check_all, file perf_pmu.c:225:
(perf_pmu:7876) CRITICAL: Failed assertion: (double)(val[i]) <= (1.0 + tolerance) * (double)0.0f && (double)(val[i]) >= (1.0 - tolerance) * (double)0.0f
(perf_pmu:7876) CRITICAL: 'val[i]' != '0.0f' (2500000.000000 not within 5.000000% tolerance of 0.000000)
Subtest busy-check-all-bcs0 failed.
---
https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_3373/shard-snb2/igt@perf_pmu@busy-check-all-bcs0.html
	
(perf_pmu:6626) CRITICAL: Test assertion failure function busy_check_all, file perf_pmu.c:225:
(perf_pmu:6626) CRITICAL: Failed assertion: (double)(val[i]) <= (1.0 + tolerance) * (double)0.0f && (double)(val[i]) >= (1.0 - tolerance) * (double)0.0f
(perf_pmu:6626) CRITICAL: 'val[i]' != '0.0f' (2500000.000000 not within 5.000000% tolerance of 0.000000)
Subtest busy-check-all-bcs0 failed.
---
https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_3373/shard-snb1/igt@perf_pmu@rc6p.html

(perf_pmu:1795) CRITICAL: Test assertion failure function test_rc6p, file perf_pmu.c:1058:
(perf_pmu:1795) CRITICAL: Failed assertion: (double)(idle[i] - prev[i]) <= (1.0 + tolerance) * (double)slept && (double)(idle[i] - prev[i]) >= (1.0 - tolerance) * (double)slept
(perf_pmu:1795) CRITICAL: 'idle[i] - prev[i]' != 'slept' (0.000000 not within 5.000000% tolerance of 2001073709.000000)
Subtest rc6p failed.
Comment 1 Chris Wilson 2017-11-23 07:52:19 UTC
See https://patchwork.freedesktop.org/series/34262/
Comment 2 Marta Löfstedt 2017-11-23 09:46:38 UTC
https://intel-gfx-ci.01.org/tree/drm-tip/IGT_4006/shard-snb2/igt@perf_pmu@most-busy-check-all-vcs0.html

(perf_pmu:4244) CRITICAL: Test assertion failure function most_busy_check_all, file perf_pmu.c:277:
(perf_pmu:4244) CRITICAL: Failed assertion: (double)(val[i]) <= (1.0 + (tolerance)) * (double)(batch_duration_ns) && (double)(val[i]) >= (1.0 - (tolerance)) * (double)(batch_duration_ns)
(perf_pmu:4244) CRITICAL: 'val[i]' != 'batch_duration_ns' (92500000.000000 not within 5.000000% tolerance of 100000000.000000)
Subtest most-busy-check-all-vcs0 failed.
Comment 3 Marta Löfstedt 2017-11-23 09:47:18 UTC
https://intel-gfx-ci.01.org/tree/drm-tip/IGT_4006/shard-hsw8/igt@perf_pmu@multi-client-vecs0.html
(perf_pmu:8872) CRITICAL: Test assertion failure function multi_client, file perf_pmu.c:650:
(perf_pmu:8872) CRITICAL: Failed assertion: (double)(val[1]) <= (1.0 + (tolerance)) * (double)(slept) && (double)(val[1]) >= (1.0 - (tolerance)) * (double)(slept)
(perf_pmu:8872) CRITICAL: 'val[1]' != 'slept' (27500000.000000 not within 5.000000% tolerance of 33410702.000000)
Subtest multi-client-vecs0 failed.
Comment 4 Marta Löfstedt 2017-11-23 09:47:39 UTC
https://intel-gfx-ci.01.org/tree/drm-tip/IGT_4006/shard-hsw1/igt@perf_pmu@most-busy-check-all-bcs0.html
(perf_pmu:7384) CRITICAL: Test assertion failure function most_busy_check_all, file perf_pmu.c:274:
(perf_pmu:7384) CRITICAL: Failed assertion: (double)(val[i]) <= (1.0 + (tolerance)) * (double)(0.0f) && (double)(val[i]) >= (1.0 - (tolerance)) * (double)(0.0f)
(perf_pmu:7384) CRITICAL: 'val[i]' != '0.0f' (2500000.000000 not within 5.000000% tolerance of 0.000000)
Subtest most-busy-check-all-bcs0 failed.
Comment 5 Marta Löfstedt 2017-11-23 09:48:23 UTC
https://intel-gfx-ci.01.org/tree/drm-tip/IGT_4006/shard-hsw4/igt@perf_pmu@multi-client-vcs0.html
(perf_pmu:1635) CRITICAL: Test assertion failure function multi_client, file perf_pmu.c:650:
(perf_pmu:1635) CRITICAL: Failed assertion: (double)(val[1]) <= (1.0 + (tolerance)) * (double)(slept) && (double)(val[1]) >= (1.0 - (tolerance)) * (double)(slept)
(perf_pmu:1635) CRITICAL: 'val[1]' != 'slept' (27500000.000000 not within 5.000000% tolerance of 33432390.000000)
Subtest multi-client-vcs0 failed.
---
https://intel-gfx-ci.01.org/tree/drm-tip/IGT_4006/shard-snb4/igt@perf_pmu@multi-client-vcs0.html
(perf_pmu:1635) CRITICAL: Test assertion failure function multi_client, file perf_pmu.c:650:
(perf_pmu:1635) CRITICAL: Failed assertion: (double)(val[1]) <= (1.0 + (tolerance)) * (double)(slept) && (double)(val[1]) >= (1.0 - (tolerance)) * (double)(slept)
(perf_pmu:1635) CRITICAL: 'val[1]' != 'slept' (27500000.000000 not within 5.000000% tolerance of 33425707.000000)
Subtest multi-client-vcs0 failed.
Comment 6 Marta Löfstedt 2017-11-23 09:49:01 UTC
https://intel-gfx-ci.01.org/tree/drm-tip/IGT_4006/shard-hsw5/igt@perf_pmu@most-busy-check-all-vecs0.html
(perf_pmu:8233) CRITICAL: Test assertion failure function most_busy_check_all, file perf_pmu.c:274:
(perf_pmu:8233) CRITICAL: Failed assertion: (double)(val[i]) <= (1.0 + (tolerance)) * (double)(0.0f) && (double)(val[i]) >= (1.0 - (tolerance)) * (double)(0.0f)
(perf_pmu:8233) CRITICAL: 'val[i]' != '0.0f' (2500000.000000 not within 5.000000% tolerance of 0.000000)
Subtest most-busy-check-all-vecs0 failed.
Comment 7 Marta Löfstedt 2017-11-23 09:49:36 UTC
https://intel-gfx-ci.01.org/tree/drm-tip/IGT_4006/shard-apl3/igt@perf_pmu@render-node-busy-vcs0.html
(perf_pmu:7994) CRITICAL: Test assertion failure function single, file perf_pmu.c:158:
(perf_pmu:7994) CRITICAL: Failed assertion: (double)(val) <= (1.0 + (tolerance)) * (double)(ref) && (double)(val) >= (1.0 - (tolerance)) * (double)(ref)
(perf_pmu:7994) CRITICAL: 'val' != 'ref' (112573715.000000 not within 5.000000% tolerance of 100000000.000000)
Subtest render-node-busy-vcs0 failed.
Comment 8 Marta Löfstedt 2017-11-23 09:50:03 UTC
https://intel-gfx-ci.01.org/tree/drm-tip/IGT_4006/shard-hsw3/igt@perf_pmu@busy-check-all-vecs0.html
(perf_pmu:4062) CRITICAL: Test assertion failure function busy_check_all, file perf_pmu.c:220:
(perf_pmu:4062) CRITICAL: Failed assertion: (double)(val[i]) <= (1.0 + (tolerance)) * (double)(0.0f) && (double)(val[i]) >= (1.0 - (tolerance)) * (double)(0.0f)
(perf_pmu:4062) CRITICAL: 'val[i]' != '0.0f' (2500000.000000 not within 5.000000% tolerance of 0.000000)
Subtest busy-check-all-vecs0 failed.
Comment 9 Marta Löfstedt 2017-11-23 09:50:59 UTC
https://intel-gfx-ci.01.org/tree/drm-tip/IGT_4006/shard-hsw6/igt@perf_pmu@multi-client-bcs0.html
(perf_pmu:6648) CRITICAL: Test assertion failure function multi_client, file perf_pmu.c:650:
(perf_pmu:6648) CRITICAL: Failed assertion: (double)(val[1]) <= (1.0 + (tolerance)) * (double)(slept) && (double)(val[1]) >= (1.0 - (tolerance)) * (double)(slept)
(perf_pmu:6648) CRITICAL: 'val[1]' != 'slept' (27500000.000000 not within 5.000000% tolerance of 33428966.000000)
Subtest multi-client-bcs0 failed.
---
https://intel-gfx-ci.01.org/tree/drm-tip/IGT_4006/shard-snb5/igt@perf_pmu@multi-client-bcs0.html
(perf_pmu:9277) CRITICAL: Test assertion failure function multi_client, file perf_pmu.c:650:
(perf_pmu:9277) CRITICAL: Failed assertion: (double)(val[1]) <= (1.0 + (tolerance)) * (double)(slept) && (double)(val[1]) >= (1.0 - (tolerance)) * (double)(slept)
(perf_pmu:9277) CRITICAL: 'val[1]' != 'slept' (27500000.000000 not within 5.000000% tolerance of 33409932.000000)
Subtest multi-client-bcs0 failed.
Comment 10 Marta Löfstedt 2017-11-23 09:52:15 UTC
https://intel-gfx-ci.01.org/tree/drm-tip/IGT_4006/shard-snb2/igt@perf_pmu@most-busy-check-all-vcs0.html

(perf_pmu:4244) CRITICAL: Test assertion failure function most_busy_check_all, file perf_pmu.c:277:
(perf_pmu:4244) CRITICAL: Failed assertion: (double)(val[i]) <= (1.0 + (tolerance)) * (double)(batch_duration_ns) && (double)(val[i]) >= (1.0 - (tolerance)) * (double)(batch_duration_ns)
(perf_pmu:4244) CRITICAL: 'val[i]' != 'batch_duration_ns' (92500000.000000 not within 5.000000% tolerance of 100000000.000000)
Subtest most-busy-check-all-vcs0 failed.
Comment 11 Marta Löfstedt 2017-11-23 09:53:11 UTC
https://intel-gfx-ci.01.org/tree/drm-tip/IGT_4006/shard-hsw4/igt@perf_pmu@multi-client-rcs0.html
(perf_pmu:6243) CRITICAL: Test assertion failure function multi_client, file perf_pmu.c:650:
(perf_pmu:6243) CRITICAL: Failed assertion: (double)(val[1]) <= (1.0 + (tolerance)) * (double)(slept) && (double)(val[1]) >= (1.0 - (tolerance)) * (double)(slept)
(perf_pmu:6243) CRITICAL: 'val[1]' != 'slept' (30000000.000000 not within 5.000000% tolerance of 33421990.000000)
Subtest multi-client-rcs0 failed.
---
https://intel-gfx-ci.01.org/tree/drm-tip/IGT_4006/shard-snb5/igt@perf_pmu@multi-client-rcs0.html
(perf_pmu:8766) CRITICAL: Test assertion failure function multi_client, file perf_pmu.c:650:
(perf_pmu:8766) CRITICAL: Failed assertion: (double)(val[1]) <= (1.0 + (tolerance)) * (double)(slept) && (double)(val[1]) >= (1.0 - (tolerance)) * (double)(slept)
(perf_pmu:8766) CRITICAL: 'val[1]' != 'slept' (30000000.000000 not within 5.000000% tolerance of 33417046.000000)
Subtest multi-client-rcs0 failed.
Comment 12 Marta Löfstedt 2017-11-23 09:54:19 UTC
https://intel-gfx-ci.01.org/tree/drm-tip/IGT_4006/shard-hsw2/igt@perf_pmu@busy-check-all-rcs0.html
(perf_pmu:9398) CRITICAL: Test assertion failure function busy_check_all, file perf_pmu.c:220:
(perf_pmu:9398) CRITICAL: Failed assertion: (double)(val[i]) <= (1.0 + (tolerance)) * (double)(0.0f) && (double)(val[i]) >= (1.0 - (tolerance)) * (double)(0.0f)
(perf_pmu:9398) CRITICAL: 'val[i]' != '0.0f' (2500000.000000 not within 5.000000% tolerance of 0.000000)
Subtest busy-check-all-rcs0 failed.
---
https://intel-gfx-ci.01.org/tree/drm-tip/IGT_4006/shard-snb4/igt@perf_pmu@busy-check-all-rcs0.html
(perf_pmu:7984) CRITICAL: Test assertion failure function busy_check_all, file perf_pmu.c:220:
(perf_pmu:7984) CRITICAL: Failed assertion: (double)(val[i]) <= (1.0 + (tolerance)) * (double)(0.0f) && (double)(val[i]) >= (1.0 - (tolerance)) * (double)(0.0f)
(perf_pmu:7984) CRITICAL: 'val[i]' != '0.0f' (2500000.000000 not within 5.000000% tolerance of 0.000000)
Subtest busy-check-all-rcs0 failed.
Comment 13 Chris Wilson 2017-11-23 12:55:24 UTC
Should be improved with

commit fbba5559d91e812db0fd446fb383a37fde48b1cb
Author: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Date:   Thu Nov 23 10:26:54 2017 +0000

    drm/i915/pmu: Clear the previous sample value when parking
    
    When turning off the engines, and the pmu sampling, clear the previous
    value as the current measurement should be 0.
    
    v2: Use a for-loop
    v3:
     * Move clearing to timer self-dis-arm to avoid race with parking.
     * Clear frequency samples as well.
    v4:
     * Init frequency to idle_freq. (Chris Wilson)
    
    Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
    Suggested-by: Chris Wilson <chris@chris-wilson.co.uk>
    Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> (v3)
    Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
    Link: https://patchwork.freedesktop.org/patch/msgid/20171123102654.29296-1-tvrtko.ursulin@linux.intel.com

Taking a chance...
Comment 14 Chris Wilson 2017-11-23 12:55:40 UTC
Oh rc6p maybe something different, still checking that out.
Comment 15 Marta Löfstedt 2017-11-23 14:11:32 UTC
(In reply to Chris Wilson from comment #14)
> Oh rc6p maybe something different, still checking that out.

Tvrtkos fix is included in CI_DRM_3377. Unfortunately there is no data from that run yet, so I will have to keep filing the "old" stuff on this bug. If it turns out that there are still issues with some of the subtests after Ci_DRM_3377 I will either file a new bug or remove green subtests from this.
Comment 17 Chris Wilson 2017-11-23 21:38:53 UTC
(In reply to Chris Wilson from comment #14)
> Oh rc6p maybe something different, still checking that out.

rc6p is failing on that machine as it has rc6p disabled. Hmm. Still fix to the test on the list.
Comment 18 Marta Löfstedt 2017-11-24 09:26:44 UTC
There are only two remaining issues with the perf_mmu tests after the patch was integrated are:
https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_3377/shard-glkb6/igt@perf_pmu@multi-client-vecs0.html

(perf_pmu:4132) CRITICAL: Test assertion failure function multi_client, file perf_pmu.c:650:
(perf_pmu:4132) CRITICAL: Failed assertion: (double)(val[1]) <= (1.0 + (tolerance)) * (double)(slept) && (double)(val[1]) >= (1.0 - (tolerance)) * (double)(slept)
(perf_pmu:4132) CRITICAL: 'val[1]' != 'slept' (190889525.000000 not within 5.000000% tolerance of 166780280.000000)
Subtest multi-client-vecs0 failed.

and:
https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_3378/shard-snb1/igt@perf_pmu@rc6p.html

(perf_pmu:4323) CRITICAL: Test assertion failure function test_rc6p, file perf_pmu.c:1051:
(perf_pmu:4323) CRITICAL: Failed assertion: (double)(idle[i] - prev[i]) <= (1.0 + (tolerance)) * (double)(slept) && (double)(idle[i] - prev[i]) >= (1.0 - (tolerance)) * (double)(slept)
(perf_pmu:4323) CRITICAL: 'idle[i] - prev[i]' != 'slept' (0.000000 not within 5.000000% tolerance of 2000125801.000000)
Subtest rc6p failed.

the GLK issue is now handled in bug 103886
the SNB issue is now handled in bug 103885
Comment 20 Marta Löfstedt 2018-01-02 07:19:07 UTC
This looks like a regression that started flip-flopping from CI_DRM_3570 on APL, KBL, GLK-shards.
Comment 21 Marta Löfstedt 2018-01-02 11:11:18 UTC
https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_3587/shard-kbl3/igt@perf_pmu@most-busy-check-all-vcs1.html

(perf_pmu:4720) CRITICAL: Test assertion failure function most_busy_check_all, file perf_pmu.c:280:
(perf_pmu:4720) CRITICAL: Failed assertion: (double)(val[i]) <= (1.0 + (tolerance)) * (double)(slept) && (double)(val[i]) >= (1.0 - (tolerance)) * (double)(slept)
(perf_pmu:4720) CRITICAL: 'val[i]' != 'slept' (0.000000 not within 5.000000% tolerance of 500087307.000000)
Subtest most-busy-check-all-vcs1 failed.
Comment 22 Chris Wilson 2018-01-11 17:39:14 UTC
commit 4900727d35bb20028f9bd83146ec4bf78afffe30
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date:   Thu Jan 11 07:30:31 2018 +0000

    drm/i915/pmu: Reconstruct active state on starting busy-stats
    
    We have a hole in our busy-stat accounting if the pmu is enabled during
    a long running batch, the pmu will not start accumulating busy-time
    until the next context switch. This then fails tests that are only
    sampling a single batch.
    
    v2: Count each active port just once (context in/out events are only on
    the first and last assignment to a port).
    v3: Avoid hardcoding knowledge of 2 submission ports
    
    Fixes: 30e17b7847f5 ("drm/i915: Engine busy time tracking")
    Testcase: igt/perf_pmu/busy-start
    Testcase: igt/perf_pmu/busy-double-start
    Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
    Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
    Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
    Link: https://patchwork.freedesktop.org/patch/msgid/20180111073031.14614-1-chris@chris-wilson.co.uk
Comment 23 Marta Löfstedt 2018-01-12 08:25:17 UTC
Fix integrated in CI_DRM_3622 tests are green. Thanks Chris


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.