*** This bug has been marked as a duplicate of bug 103886 ***

reproduced:
https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_3413/shard-kbl2/igt@perf_pmu@rc6.html


(perf_pmu:1511) CRITICAL: Test assertion failure function test_rc6, file perf_pmu.c:1006:
(perf_pmu:1511) CRITICAL: Failed assertion: (double)(idle - prev) <= (1.0 + (tolerance)) * (double)(slept) && (double)(idle - prev) >= (1.0 - (tolerance)) * (double)(slept)
(perf_pmu:1511) CRITICAL: 'idle - prev' != 'slept' (1808775680.000000 not within 5.000000% tolerance of 2001331815.000000)
Subtest rc6 failed.

It was also failing on:
https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_3407/shard-kbl5/igt@perf_pmu@rc6.html
and 
https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_3412/shard-kbl4/igt@perf_pmu@rc6.html

but that was shadowed by the 4.15.0.rc1 fire.

Once more into the breach, (sleepily by accident pushed):

commit ba6c4e6e94f43857c6fa13e5c4cfddad780dc42b (upstream/master, origin/master, origin/HEAD)
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date:   Thu Nov 30 07:48:46 2017 +0000

    igt/perf_pmu: Increase delay for rc6 to start
    
    I was thinking of the RC6 threshold parameter, but needed to consider
    the RC6 evaluation interval instead. RC6 doesn't enable until activity
    is below the threshold inside an evaluation interval, therefore we need
    to wait at least 2 EI after idling before we can expect RC6 to be
    enabled.
    
    Fixes: 55a17bc2d040 ("igt/perf_pmu: Reduce arbitrary delays before rc6")
    Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
    Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>

There is no IGT run for this yet, so I will have to wait to close it again...

Fix integrated in CI_DRM_3415

This is looking green from CI_DRM_3415, I will close

Meh:
https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_3452/shard-kbl5/igt@perf_pmu@rc6.html
	

(perf_pmu:1553) CRITICAL: Test assertion failure function test_rc6, file perf_pmu.c:1029:
(perf_pmu:1553) CRITICAL: Failed assertion: (double)(idle - prev) <= (1.0 + (tolerance)) * (double)(slept) && (double)(idle - prev) >= (1.0 - (tolerance)) * (double)(slept)
(perf_pmu:1553) CRITICAL: 'idle - prev' != 'slept' (1317283840.000000 not within 5.000000% tolerance of 2000169218.000000)
Subtest rc6 failed.

(In reply to Marta Löfstedt from comment #7)
> Meh:
> https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_3452/shard-kbl5/
> igt@perf_pmu@rc6.html
> 	
> 
> (perf_pmu:1553) CRITICAL: Test assertion failure function test_rc6, file
> perf_pmu.c:1029:
> (perf_pmu:1553) CRITICAL: Failed assertion: (double)(idle - prev) <= (1.0 +
> (tolerance)) * (double)(slept) && (double)(idle - prev) >= (1.0 -
> (tolerance)) * (double)(slept)
> (perf_pmu:1553) CRITICAL: 'idle - prev' != 'slept' (1317283840.000000 not
> within 5.000000% tolerance of 2000169218.000000)
> Subtest rc6 failed.

The previous ones where just outside the tolerance; this is 65% of the target, definitely something a bit more major than the previous systematic sampling errors.

Next attempt,

commit 55da075197d1d7810043655d952e60a20b9e8fa2 (HEAD, upstream/master)
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date:   Tue Dec 5 10:37:13 2017 +0000

    igt/perf_pmu: Replace hard-coded sleep before rc6 with a probe
    
    Instead of trying to sleep for 2 evaluations intervals and then assuming
    that rc6 is working, poll the rc6 residency instead.
    
    v2: dce
    
    References: https://bugs.freedesktop.org/show_bug.cgi?id=103929
    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>

Fix included in CI_DRM_3459, has been green since, closing