[Bug 105312] [CI][SHARDS] igt at kms_flip@*(flip|wf_vblank_ts_check)* - fail - Failed assertion: elapsed >= min && elapsed <= max

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=105312

Arek Hiler <arkadiusz.hiler at intel.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|high                        |medium

--- Comment #16 from Arek Hiler <arkadiusz.hiler at intel.com> ---
This bug has extremely low rate of reproduction - we see it about once every
quarter. From the historical diggings it seems that in the last year or so we
missed a single frame each failure.

This family of failing tests makes sure that we managed to squeeze given number
of frames in a given interval of time.

The test does vblank interval measurement/calibration in calibrate_ts(). Some
statistical checks on the timing samples are performed to make sure that 99.7%
samples withing 0.5% of the means and that 94% of samples are within 0.5% of
the
expected value.

Every so often we will an event where we hit more intervals on the bad side and
the rounding done on the timeouts (we don't have any leeway here) won't be in
our favour.

Statistical character of this test and non-deterministic timings means we will
fail from time to time.

Impact on the users: low (a frame a quater on the whole farm in delays), as
long
as the reproduction rate and number of missed frames stays the same.

-- 
You are receiving this mail because:
You are the assignee for the bug.
You are the QA Contact 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/20190617/c777d65a/attachment.html>