1271517 - Remove use of FlushableTaskQueue::Flush() from FFmpegDataDecoder::Flush()

Status: RESOLVED FIXED

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

Description

[JW Wang [:jwwang] [:jw_wang]]

As step forward to get rid of FlushableTaskQueue.

We will borrow the code pattern from WMFMediaDataDecoder::mIsFlushing so FFmpegDataDecoder::Decode can bail out early when seeing mIsFlushing is true. Then we won't need FlushableTaskQueue::Flush() to clean decoding jobs.

Comment 1

[JW Wang [:jwwang] [:jw_wang]]

try: https://treeherder.mozilla.org/#/jobs?repo=try&revision=7da2558a67dd

Comment 2

[Jean-Yves Avenard [:jya]]

(In reply to JW Wang [:jwwang] from comment #0)
> As step forward to get rid of FlushableTaskQueue.
> 
> We will borrow the code pattern from WMFMediaDataDecoder::mIsFlushing so
> FFmpegDataDecoder::Decode can bail out early when seeing mIsFlushing is
> true. Then we won't need FlushableTaskQueue::Flush() to clean decoding jobs.

I don't think this will be sufficient.

After a Flush() we don't expect any callback at all from the MediaDataDecoder, that includes calls to InputExhausted or DrainComplete()..

The MediaFormatReader will have to be adjusted accordingly.

Comment 3

[JW Wang [:jwwang] [:jw_wang]]

(In reply to Jean-Yves Avenard [:jya] from comment #2)
> After a Flush() we don't expect any callback at all from the
> MediaDataDecoder, that includes calls to InputExhausted or DrainComplete()..

This behavior is preserved without this bug. And early bail out from FFmpegDataDecoder::Decode() will not change this behavior. So this bug will not break the behavior, right?

Comment 4

[Jean-Yves Avenard [:jya]]

(In reply to JW Wang [:jwwang] from comment #3)
> (In reply to Jean-Yves Avenard [:jya] from comment #2)
> > After a Flush() we don't expect any callback at all from the
> > MediaDataDecoder, that includes calls to InputExhausted or DrainComplete()..
> 
> This behavior is preserved without this bug. And early bail out from
> FFmpegDataDecoder::Decode() will not change this behavior. So this bug will
> not break the behavior, right?

What about: mDecoder->Drain(); immediately followed by mDecoder->Flush();

Comment 5

[Jean-Yves Avenard [:jya]]

Actually never mind, the MFR already checks that flush() wasn't called and ignore any following calls to DrainComplete().

Comment 6

[JW Wang [:jwwang] [:jw_wang]]

try: https://treeherder.mozilla.org/#/jobs?repo=try&revision=0b9680efd654&selectedJob=20620946 looks fine.

Comment 7

[JW Wang [:jwwang] [:jw_wang]]

Attachment: MozReview Request: Bug 1271517. Part 1 - remove use of FlushableTaskQueue::Flush() from FFmpegDataDecoder::Flush(). r=jya.

Review commit: https://reviewboard.mozilla.org/r/51787/diff/#index_header
See other reviews: https://reviewboard.mozilla.org/r/51787/

Comment 8

[JW Wang [:jwwang] [:jw_wang]]

Attachment: MozReview Request: Bug 1271517. Part 2 - remove use of FlushableTaskQueue. r=jya.

Review commit: https://reviewboard.mozilla.org/r/51789/diff/#index_header
See other reviews: https://reviewboard.mozilla.org/r/51789/

Comment 9

[Jean-Yves Avenard [:jya]]

Comment on attachment 8751057 [details]
MozReview Request: Bug 1271517. Part 1 - remove use of FlushableTaskQueue::Flush() from FFmpegDataDecoder::Flush(). r=jya.

https://reviewboard.mozilla.org/r/51787/#review48697

Comment 10

[Jean-Yves Avenard [:jya]]

Comment on attachment 8751058 [details]
MozReview Request: Bug 1271517. Part 2 - remove use of FlushableTaskQueue. r=jya.

https://reviewboard.mozilla.org/r/51789/#review48699

Comment 11

[JW Wang [:jwwang] [:jw_wang]]

Thanks!

Comment 12

[Pulsebot]

https://hg.mozilla.org/integration/mozilla-inbound/rev/fef82e636b41
https://hg.mozilla.org/integration/mozilla-inbound/rev/1db715403fbe

Comment 13

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/fef82e636b41
https://hg.mozilla.org/mozilla-central/rev/1db715403fbe