Open Bug 1729035 Opened 3 years ago Updated 3 years ago

Condition variable waits fire late on Windows testers

Categories

(Testing :: GTest, defect, P3)

Default
defect

Tracking

(Not tracked)

People

(Reporter: bwc, Unassigned)

References

Details

Noticed this over in bug 1726177, due to new test-cases. Very similar to bug 1726915, but not nearly as severe. Was not happening prior to the creation of the windows 10 2004 jobs.

For reference, grep these logs for "was supposed to wait for":

https://treeherder.mozilla.org/logviewer?job_id=350393068&repo=try

[task 2021-09-03T00:09:11.501Z] 00:09:11 INFO - [Parent 5584: Timer]: D/nsTimerImpl Timer thread woke up after 28.668600ms, was supposed to wait for 1.387800ms
[task 2021-09-03T00:10:23.094Z] 00:10:23 INFO - [Parent 5584: Timer]: D/nsTimerImpl Timer thread woke up after 27.673700ms, was supposed to wait for 0.499100ms
[task 2021-09-03T00:10:23.303Z] 00:10:23 INFO - [Parent 5584: Timer]: D/nsTimerImpl Timer thread woke up after 28.713800ms, was supposed to wait for 1.474900ms
[task 2021-09-03T00:11:59.821Z] 00:11:59 INFO - [Parent 5584: Timer]: D/nsTimerImpl Timer thread woke up after 42.636700ms, was supposed to wait for 19.968000ms
[task 2021-09-03T00:12:11.831Z] 00:12:11 INFO - [Parent 5584: Timer]: D/nsTimerImpl Timer thread woke up after 42.529800ms, was supposed to wait for 19.964800ms
[task 2021-09-03T00:12:15.824Z] 00:12:15 INFO - [Parent 5584: Timer]: D/nsTimerImpl Timer thread woke up after 42.507900ms, was supposed to wait for 19.966400ms
[task 2021-09-03T00:12:26.542Z] 00:12:26 INFO - [Parent 5584: Timer]: D/nsTimerImpl Timer thread woke up after 40.464900ms, was supposed to wait for 19.968500ms
[task 2021-09-03T00:12:35.627Z] 00:12:35 INFO - [Parent 5584: Timer]: D/nsTimerImpl Timer thread woke up after 30.651200ms, was supposed to wait for 7.321700ms
[task 2021-09-03T00:12:38.769Z] 00:12:38 INFO - [Parent 5584: Timer]: D/nsTimerImpl Timer thread woke up after 42.306700ms, was supposed to wait for 19.967600ms
[task 2021-09-03T00:12:43.003Z] 00:12:43 INFO - [Parent 5584: Timer]: D/nsTimerImpl Timer thread woke up after 40.059000ms, was supposed to wait for 19.969500ms
[task 2021-09-03T00:12:57.513Z] 00:12:57 INFO - [Parent 5584: Timer]: D/nsTimerImpl Timer thread woke up after 29.041100ms, was supposed to wait for 4.211500ms
[task 2021-09-03T00:13:28.319Z] 00:13:28 INFO - [Parent 5584: Timer]: D/nsTimerImpl Timer thread woke up after 32.955500ms, was supposed to wait for 9.930700ms

is this something wrong with the test harness, the browser, or the OS? At first glance this looks to be a firefox issue, but maybe this is an OS issue.

Severity: -- → S3
Priority: -- → P3

I think this might be a tester or OS issue, but it is difficult to say for sure. These measurements were taken very close to the OS API, so there's not a whole lot we could be doing to mess this up. It is possible that code elsewhere in Firefox is exacerbating this issue somehow.

You need to log in before you can comment on or make changes to this bug.