[Webkit-unassigned] [Bug 180643] New: [MSE] SourceBuffer::provideMediaData() may remove sample from decode queue unexpectedly

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=180643

            Bug ID: 180643
           Summary: [MSE] SourceBuffer::provideMediaData() may remove
                    sample from decode queue unexpectedly
           Product: WebKit
           Version: WebKit Nightly Build
          Hardware: Unspecified
                OS: Unspecified
            Status: NEW
          Severity: Normal
          Priority: P2
         Component: Media Elements
          Assignee: webkit-unassigned at lists.webkit.org
          Reporter: zhangququ at gmail.com

I found in some cases, a sample already in trackBuffer.decodeQueue could be removed unexpectedly when there is gap ahead of it. For example, if samples of the following ranges are already in the trackBuffer.decodeQueue,

    [0 .. 5.0) [10.0, ... 30.0)

when SourceBuffer::provideMediaData() is called on such a decodeQueue, it may remove sample 10.0 (decode time) due to one second gap. Then, if the gap is filled (segment re-appended) later, it will enqueue samples from this decodeQueue but without sample 10.0.

The problem usually happens after playback for a while, e.g. buffered [10.0, ... 30.0), then seeking to the beginning. During re-appending the first segments, provideMediaData() may remove sample 10.0. As the removed sample is usually a sync sample, it results in playback issue if the enqueued samples are sent to decoder (not flushed).

I suggest to move the following removal code to be after the one-second gap checking:

    trackBuffer.decodeQueue.erase(trackBuffer.decodeQueue.begin());

How do you think? Thanks.

Best Regards,
Daniel ZHANG

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.webkit.org/pipermail/webkit-unassigned/attachments/20171211/f3f45b04/attachment.html>