Pushing some changes from meta-rdk-ext layer #697

emutavchi · 2021-02-09T16:04:18Z

This is a subset of changes from meta-rdk-ext layer from RDK.

Make HTMLMediaElement::ignoreTrackDisplayUpdateRequests() constant time

Web Inspector: Crashes seen under Inspector::ScriptCallFrame::~ScriptCallFrame https://bugs.webkit.org/show_bug.cgi?id=180373 <rdar://problem/33894170> * inspector/AsyncStackTrace.cpp: (Inspector::AsyncStackTrace::truncate): The `lastUnlockedAncestor->remove()` may release the only reference to it's parent which we intend to use later but don't hold a RefPtr to. Keep the parent alive explicitly by protecting it.

…line https://bugs.webkit.org/show_bug.cgi?id=169202

And restart streaming thread after configuring seamless segment

cherry-picked from 8e3ad2d from wpe-20170728

Replacing data during async decoding causes random crashes

eocanha

(deleted comment)

eocanha

Thank you for contributing all these patches. There are many interesting fixes among them. All the multimedia related patches are fine except a few of them, where I have posted comments and questions to ask for clarification about some topics.
About the patches for hole punch or those not related to multimedia, somebody acquainted with those areas (@magomez?) should have a look at them.

Source/WebCore/Modules/mediasource/SourceBuffer.cpp

Source/WebCore/platform/graphics/MediaPlayer.cpp

Source/WebCore/platform/graphics/gstreamer/MediaPlayerPrivateGStreamer.cpp

eocanha · 2021-02-11T18:08:06Z

Source/WebCore/platform/graphics/gstreamer/mse/PlaybackPipeline.cpp

@@ -301,6 +301,25 @@ void PlaybackPipeline::markEndOfStream(MediaSourcePrivate::EndOfStreamStatus)
        gst_app_src_end_of_stream(appsrc);
 }

+GstPadProbeReturn segmentFixerProbe(GstPad*, GstPadProbeInfo* info, gpointer)


The segmentFixerProbe was removed by e81ce7f
Whay kind of issues are you experiencing to need to bring it back? I'd like to discuss them with @ntrrgc

The sample replacement doesn't work after e81ce7f. I tried on different GStreamer version (1.10 to 1.16)

By "sample replacement" I understand a quality change (to higher quality) where the new samples should replace the old ones in the already playing pipeline, right?
In theory, the reset of the running time plus using a zero base time (what @ntrrgc's patch does) should be equivalent of what the old patch did by keeping the running time and tweaking the base time. What kind of problems are you experiencing? Stalls, long waits, audio/video desynchronization?
As the Nexus multimedia system in Broadcom devices uses a common synchronization system between audio and video, maybe having different running times in the audio and the video streams confuses the Broadcom gst elements or the Nexus system. That might explain why you're having problems but everything looked fine on the Raspberry.

Source/WebCore/platform/graphics/gstreamer/mse/PlaybackPipeline.cpp

eocanha

From my side, all the topics except the segmentFixerProbe one have been successfully resolved. I'd like to have more information about the wrong behaviour that happens without the segmentFixerProbe patch.

Source/WebCore/Modules/mediasource/SourceBuffer.cpp

eocanha · 2021-02-18T16:43:28Z

Source/WebCore/platform/graphics/gstreamer/mse/PlaybackPipeline.cpp

@@ -301,6 +301,25 @@ void PlaybackPipeline::markEndOfStream(MediaSourcePrivate::EndOfStreamStatus)
        gst_app_src_end_of_stream(appsrc);
 }

+GstPadProbeReturn segmentFixerProbe(GstPad*, GstPadProbeInfo* info, gpointer)


By "sample replacement" I understand a quality change (to higher quality) where the new samples should replace the old ones in the already playing pipeline, right?
In theory, the reset of the running time plus using a zero base time (what @ntrrgc's patch does) should be equivalent of what the old patch did by keeping the running time and tweaking the base time. What kind of problems are you experiencing? Stalls, long waits, audio/video desynchronization?
As the Nexus multimedia system in Broadcom devices uses a common synchronization system between audio and video, maybe having different running times in the audio and the video streams confuses the Broadcom gst elements or the Nexus system. That might explain why you're having problems but everything looked fine on the Raspberry.

Source/WebCore/platform/graphics/gstreamer/mse/PlaybackPipeline.cpp

magomez

I've given a look to the changes non related to MSE (that's eocanha's field) and they look ok to me. This can be merged as soon as eocanha gives the green light.

…e-2.22

eocanha · 2021-02-25T12:11:18Z

I've been testing all these commits with YouTube TV and other apps and found no regressions. I'm manually merging all the
commits except the ones affecting PlaybackPipeline::flush():

Backport [GStreamer][MSE] Actually implement flush() on playback pipeline
[MSE][GStreamer] Allow independed flushing of video buffer

It's true that they don't cause apparent regressions, but I'd like to understand better how they (or the lack of them) affect you. We can discuss that in a new pull request.

emutavchi added 26 commits February 9, 2021 03:08

[MSE] Don't send caps out of flow

58fa100

Fix crash due to infinite recursion on seek

e0786f0

[MSE] implement maxTimeLoaded

837c587

[MSE] Reset 'm_isEndReached' and 'm_eosMarked' on unmarkEndOfStream

a5f70e4

[EME] Implement MediaKeySession::hasPendingActivity

79e1be5

[GStreamer] Handle gst pad error as format error

6324a4a

[MSE] Propagate parsing errors to MSE SourceBuffer

f57459e

[MSE][SourceBuffer] update m_bufferFull flag when evicted enough

a4d0361

[MSE][SourceBuffer] Fix creating float PTS/DTS when dividing sample

39c2ee1

[MSE][SourceBuffer] Fix removal of paddings from buffered range

4a368b4

Add isEmpty() API to PODRedBlackTree

5751983

Make HTMLMediaElement::ignoreTrackDisplayUpdateRequests() constant time

[MSE][SourceBuffer] Make SourceBuffer evictCodedFrames less aggressive

f1652b4

[MSE] Fix occasional video playback stalls after seek

3a46870

[GStreamer] clarify playback errors with MediaError.message

6b965d8

Avoid auto-selection of hole punching player

08217da

[Westeros-sink] Initial playbin3 support

6d1cb87

Backport [GStreamer][MSE] Actually implement flush() on playback pipe…

91794d0

…line https://bugs.webkit.org/show_bug.cgi?id=169202

Fix detection of EOS in MSE playback

87927b3

[MSE][GStreamer] Allow independed flushing of video buffer

f438d68

And restart streaming thread after configuring seamless segment

Fix ASAN build

b5d9e7c

Fix memory pressure poller thread leak

88cc9ab

Add non-synchronous pressure notifications

58c089e

Backport ScalableImageDecoder: avoid unnecessarily decoding image data.

2cb7548

cherry-picked from 8e3ad2d from wpe-20170728

Disable endcoded data replacement for images with async decoding

ceb3ae8

Replacing data during async decoding causes random crashes

Avoid update of tile content buffers in non composited webgl mode

e12d0c5

emutavchi requested a review from woutermeek February 9, 2021 16:04

woutermeek requested review from magomez and eocanha February 9, 2021 16:10

eocanha reviewed Feb 11, 2021

View reviewed changes

eocanha reviewed Feb 12, 2021

View reviewed changes

eocanha reviewed Feb 18, 2021

View reviewed changes

magomez approved these changes Feb 19, 2021

View reviewed changes

eocanha added a commit that referenced this pull request Feb 25, 2021

Merge subset of pull request #697 from emutavchi/wpe-2.22-wip into wp…

941e508

…e-2.22

eocanha closed this Feb 25, 2021

emutavchi mentioned this pull request Mar 1, 2022

Backport of some of MSE related changes from wpe-2.22 #797

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Pushing some changes from meta-rdk-ext layer #697

Pushing some changes from meta-rdk-ext layer #697

emutavchi commented Feb 9, 2021

eocanha left a comment •

edited

Loading

eocanha left a comment

eocanha Feb 11, 2021

emutavchi Feb 15, 2021 •

edited by eocanha

Loading

eocanha Feb 18, 2021

eocanha left a comment

eocanha Feb 18, 2021

magomez left a comment

eocanha commented Feb 25, 2021

Pushing some changes from meta-rdk-ext layer #697

Pushing some changes from meta-rdk-ext layer #697

Conversation

emutavchi commented Feb 9, 2021

eocanha left a comment • edited Loading

Choose a reason for hiding this comment

eocanha left a comment

Choose a reason for hiding this comment

eocanha Feb 11, 2021

Choose a reason for hiding this comment

emutavchi Feb 15, 2021 • edited by eocanha Loading

Choose a reason for hiding this comment

eocanha Feb 18, 2021

Choose a reason for hiding this comment

eocanha left a comment

Choose a reason for hiding this comment

eocanha Feb 18, 2021

Choose a reason for hiding this comment

magomez left a comment

Choose a reason for hiding this comment

eocanha commented Feb 25, 2021

eocanha left a comment •

edited

Loading

emutavchi Feb 15, 2021 •

edited by eocanha

Loading