Closed Bug 1775978 Opened 4 years ago Closed 1 year ago

Intermittent dom/media/test/browser/browser_tab_visibility_and_play_time.js | single tracking bug

Categories

(Core :: Audio/Video: Playback, defect, P5)

defect

Tracking

()

RESOLVED FIXED
141 Branch
Tracking Status
firefox141 --- fixed

People

(Reporter: jmaher, Assigned: alwu)

References

Details

(Keywords: intermittent-failure, intermittent-testcase)

Attachments

(3 external files, 1 obsolete file)

No description provided.

Additional information about this bug failures and frequency patterns can be found by running: ./mach test-info failure-report --bug 1775978

Severity: -- → S4
Priority: P3 → P5
Whiteboard: [stockwell disable-recommended]

FWIW, I just got a failure on this run here.

I spent a few minutes poking around and wanted to note a few things that I observed:

  1. In all cases (failing and not failing/not failing the test start is immediately followed by an error:
    [Child XXX, MediaDecoderStateMachine #1] WARNING: Decoder=XXX Decode error: NS_ERROR_DOM_MEDIA_DECODE_ERR (0x806e0004) - RefPtr<MediaDataDecoder::DecodePromise> __thiscall mozilla::WMFMediaDataDecoder::ProcessError(HRESULT, const char *): MFTManager::Output(2):80004005: file /builds/worker/checkouts/gecko/dom/media/MediaDecoderStateMachineBase.cpp:168
    This seems to me like something that shouldn't be happening right at the start of a test like this.
  2. In the fail case, there is output about a whole bunch more tests running and passing up until the failure, whereas in the passing runs the next output is the "MEMORY STATS" line and then the "TEST-OK" line. (How is the test OK when one of the sub-tests failed?) Interestingly both the failing and passing runs all report similar amounts of time taken for the tests.
  3. In the failing run, there is a couple error message pertaining to not being able to decode https://example.com/browser/dom/media/test/browser/TestPatternHDR.mp4 (here and here).

Beyond the intermittent failure itself (but perhaps causing it), it seems to me that this test has probably rotted a bit.

I am NI'ing padenot in case any of this is interesting, seems like it might be indicating a problem worth looking into, or would just be an easy fix. Feel free to ignore or clear.

Flags: needinfo?(padenot)

Alastor is better person for Windows stuff, but thanks a lot for your analysis!

Flags: needinfo?(padenot) → needinfo?(alwu)

That error is caused by playing a HDR video, which is not supported on Windows. I will have a patch for this to see if that helps.

Assignee: nobody → alwu
Flags: needinfo?(alwu)

totalVideoHDRPlayTime is measured when the video's bit depth exceeds 8 bits, so
it isn’t dependent on platform color support. This is because we're not testing
whether the color is displayed correctly—only whether the probe is recorded
properly during playback of a 10-bit video.

Therefore, in this patch, we use an HDR file that is supported across all
platforms and enable the HDR play time check universally.

Attachment #9494454 - Attachment description: Bug 1775978 - part3 : enable HDR playtime test for all platforms. → Bug 1775978 - part2 : enable HDR playtime test for all platforms.
Attachment #9494455 - Attachment description: Bug 1775978 - part4 : wait longer to avoid intermittent test failures. → Bug 1775978 - part3 : wait longer to avoid intermittent test failures.
Attachment #9494453 - Attachment is obsolete: true
Pushed by alwu@mozilla.com: https://github.com/mozilla-firefox/firefox/commit/3894840cff81 https://hg.mozilla.org/integration/autoland/rev/5423a3c7638c part1 : apply changes from ESLint. r=media-playback-reviewers,padenot https://github.com/mozilla-firefox/firefox/commit/77de40d40f2e https://hg.mozilla.org/integration/autoland/rev/cf888b808249 part2 : enable HDR playtime test for all platforms. r=media-playback-reviewers,padenot https://github.com/mozilla-firefox/firefox/commit/bcd71dbec352 https://hg.mozilla.org/integration/autoland/rev/5d10a9a3ca58 part3 : wait longer to avoid intermittent test failures. r=media-playback-reviewers,padenot
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: