Closed Bug 895720 Opened 7 years ago Closed 7 years ago
_audio Buffer Source Node Offset .html | Correct number of samples received (expected: 24000, actual: 37120).
https://tbpl.mozilla.org/php/getParsedLog.php?id=25462320&tree=Mozilla-Inbound Rev4 MacOSX Snow Leopard 10.6 mozilla-inbound debug test mochitest-1 on 2013-07-18 16:28:46 PDT for push 5955cd778745 slave: talos-r4-snow-030 17:21:40 INFO - 152896 INFO TEST-START | /tests/content/media/webaudio/test/test_audioBufferSourceNodeOffset.html 17:21:40 INFO - System JS : ERROR (null):0 17:21:40 INFO - uncaught exception: 2147746065 17:21:40 INFO - ++DOMWINDOW == 24 (0x14ee7e758) [serial = 4732] [outer = 0x118330528] 17:21:41 INFO - System JS : ERROR (null):0 17:21:41 INFO - uncaught exception: 2147746065 17:21:41 INFO - System JS : ERROR (null):0 17:21:41 INFO - uncaught exception: 2147746065 17:21:41 INFO - 152897 ERROR TEST-UNEXPECTED-FAIL | /tests/content/media/webaudio/test/test_audioBufferSourceNodeOffset.html | Correct number of samples received (expected: 24000, actual: 37120). 17:21:41 INFO - 152898 INFO TEST-END | /tests/content/media/webaudio/test/test_audioBufferSourceNodeOffset.html | finished in 1428ms
Well, I a gremlin must have triggered a super long gc in between the oncomplete and the onprocessaudio.
I'll let those run in a loop locally to see if I can repro, and check if this is solely due to the oncomplete handler being called too late because of gc or some other stuff, or if we have a bug.
Assignee: nobody → paul
Note - Assuming that the failure seen in https://bugzilla.mozilla.org/show_bug.cgi?id=896582#c24 is alignment with this bug, this test failure in this bug resulted in this web audio test being turned off on Firefox OS emulators.
I can repro very very rarely locally on a Macbook pro, maybe we should just increase the fuzz on MacOS. The test would still be valid, because "actual" would have been 0 before the fix.
(In reply to comment #10) > I can repro very very rarely locally on a Macbook pro, maybe we should just > increase the fuzz on MacOS. The test would still be valid, because "actual" > would have been 0 before the fix. Sure.
Status: NEW → RESOLVED
Closed: 7 years ago
Resolution: --- → FIXED
Target Milestone: --- → mozilla25
Why increase the fuzz factor only on mac?
Because other platform work just fine? The fuzzing is scheduler and load related, and it seems to work fine on other platforms.
Ok, it just seems so fragile to me.
You need to log in before you can comment on or make changes to this bug.