Closed Bug 1081842 Opened 10 years ago Closed 10 years ago

Intermittent test_streams_autoplay.html | Test timed out. - expected PASS

Categories

(Core :: WebRTC, defect)

Product:
Core
Component:
WebRTC
Platform:
x86
Gonk (Firefox OS)
Type:
defect
Priority:
Not set
Severity:
normal

Tracking

()

Status:
RESOLVED FIXED
Milestone:
mozilla36
Tracking Flags:
Tracking Status
firefox34 --- unaffected
firefox35 --- fixed
firefox36 --- fixed
firefox-esr31 --- unaffected

People

(Reporter: cbook, Assigned: pehrsons)

References

(
URL
)

Details

(Keywords: intermittent-failure) 

b2g_emulator_vm mozilla-inbound opt test mochitest-3

https://treeherder.mozilla.org/ui/logviewer.html#?job_id=2951193&repo=mozilla-inbound

23:34:06 INFO - 428 INFO TEST-UNEXPECTED-FAIL | /tests/content/media/test/test_streams_autoplay.html | Test timed out. - expected PASS
jesup: could you take a look at this one, thanks
Flags: needinfo?(rjesup)
From bug 879717, looks like we have another race condition on our hands.

I believe we get these intermittents because the MediaStream ends before the DOMMediaStream has reported track added. With bug 879717 we need at least one track to proceed from HAVE_METADATA, and we need to get to HAVE_FUTURE_DATA to get a playing event (not getting one is causing the test timeout).

The reason it only happens on Android 2.3 and B2G ICS would be that they are too slow..?

The file that fails in this case is 320x240.ogv. It is only 0.27 seconds long. The simplest fix to this test case would probably be to use a longer one.


jesup, what do you think?
My analysis appears to be correct at least: https://treeherder.mozilla.org/ui/#/jobs?repo=try&revision=0fbc00699bb2

The patch I used to test it: https://hg.mozilla.org/try/rev/5bfb87914f6e
So this was in the merge down to Aurora. Whatever fix comes needs to land there as well.
status-firefox35: --- → affected
Is VideoFrameContainer::InvalidateWithFlags not called at all? If it's called at all (even before the main thread detects there is a video track) then we should get past the "if (container && mMediaSize == nsIntSize(-1,-1))" check.
backing out bug 897717 for b2g emulator and android 2.3 failures in test_streams_autoplay and test_streams_element_capture.

Adding "loop" to v1 in test_streams_autoplay got it past the timing race where the video could finish before the second video element was ready, but then it still fails in test_streams_element_capture.

What that merges we should close this bug.
Flags: needinfo?(rjesup)
We backed out the cause - bug 879717
Status: NEW → RESOLVED
Closed: 10 years ago
Resolution: --- → FIXED
Assignee: nobody → pehrsons
status-firefox34: --- → unaffected
status-firefox35: affected → fixed
status-firefox36: --- → fixed
status-firefox-esr31: --- → unaffected
Target Milestone: --- → mozilla36
You need to log in before you can comment on or make changes to this bug.