[Bug 110728] [CI][SHARDS] igt at perf@(blocking|polling) - fail - Failed assertion: kernel_ns <= (test_duration_ns / 100ull)

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=110728

--- Comment #3 from emersion <contact at emersion.fr> ---
In [1], 170ms (±10ms) are spent blocking in the kernel. This is more than 1% of
the test duration (10s = 10000ms, time spent in kernel is 1.7%). In other logs,
similar issues can be observed.

This means that we're burning too many CPU cycles in the kernel when reading
from Observation Architecture.

User impact: some performance issues when measuring performance.

I suspect these performance issues only happen when actively using OA, so most
users shouldn't be affected, but I'm not familiar with OA, so don't quote me on
this.

[1]:
https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6176/shard-skl6/igt@perf@blocking.html

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
You are the assignee for the bug.
You are on the CC list for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/intel-gfx-bugs/attachments/20190610/cb000b57/attachment.html>