[Bug 111821] [CI][SHARDS] igt at perf@gen8-unprivileged-single-ctx-counters - dmesg-fail - Failed assertion: ret >= 0, Cannot allocate memory

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=111821

--- Comment #4 from umesh <umesh.nerlige.ramappa at intel.com> ---
The test captures counter snapshots using MI RPC before and after a few render
workloads. The batch consisting of the MI-RPC also does a PIPE_FLUSH and
reads/stores the TIMESTAMP value from the render timestamp register. Since
there may be a context switch during the execution of these workloads, the test
also collects reports from the OA buffer between the OA timestamps collected
from the MI RPCs. Is uses all these reports to calculate the actual delta
between the counters A0, A21 and A26. The delta of the counters is expected to
be a certain fixed value for the render copy performed.

The failure is likely from a large deviance between "timestamp delta from MI
RPC OA timestamps" and "timestamp delta between render timestamps collected
along with MI-RPC". The large deviance causes the test to try again. Every time
the test tries again, it allocates some resources. Eventually results in OOM.

Machines: skl, kbl
Repro rate: 5%

Impact: Accuracy of perf snapshots from MI-RPC
Severity: Medium
Pririty: High

-- 
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/20191101/3c436234/attachment-0001.html>