This cl is in preparation for https://codereview.webrtc.org/2060403002/ Add task queue to Call.
In the coming cl the video_sender, and i420_buffer_pool will be used on a task queue and therefore SequencedTaskChecker is needed instead of a ThreadChecker.
BUG=webrtc:5687
Review-Url: https://codereview.webrtc.org/2149553002
Cr-Commit-Position: refs/heads/master@{#13474}
The decode thread should be stopped before triggering shutdown of the
video receiver, so that the decoder doesn't try to insert a new frame
while the jitter buffer is being shut down.
BUG=webrtc:6102
Review-Url: https://codereview.webrtc.org/2146883002
Cr-Commit-Position: refs/heads/master@{#13467}
Reason for revert:
Upstream fixes in place, should be OK now.
Original issue's description:
> Revert of Refactor NACK bitrate allocation (patchset #16 id:300001 of https://codereview.webrtc.org/2061423003/ )
>
> Reason for revert:
> Breaks upstream code.
>
> Original issue's description:
> > Refactor NACK bitrate allocation
> >
> > Nack bitrate allocation should not be done on a per-rtp-module basis,
> > but rather shared bitrate pool per call. This CL moves allocation to the
> > pacer and cleans up a bunch if bitrate stats handling.
> >
> > BUG=
> > R=danilchap@webrtc.org, stefan@webrtc.org, tommi@webrtc.org
> >
> > Committed: 5fc59e810b
>
> TBR=tommi@webrtc.org,danilchap@webrtc.org,stefan@webrtc.org
> # Skipping CQ checks because original CL landed less than 1 days ago.
> NOPRESUBMIT=true
> NOTREECHECKS=true
> NOTRY=true
> BUG=
>
> Committed: https://crrev.com/e5dd44101eca485f5ad12e5f7ce6f6b0d204116b
> Cr-Commit-Position: refs/heads/master@{#13417}
TBR=tommi@webrtc.org,danilchap@webrtc.org,stefan@webrtc.org
# Not skipping CQ checks because original CL landed more than 1 days ago.
BUG=
Review-Url: https://codereview.webrtc.org/2146013002
Cr-Commit-Position: refs/heads/master@{#13465}
These arguments are not really known when calling SetEncodingData. They are still provided as argument to ProtectionBitrateCalculator::SetTargetRates though.
This cl is broken out from https://codereview.webrtc.org/2060403002/
BUG=webrtc:5687
Review-Url: https://codereview.webrtc.org/2121983002
Cr-Commit-Position: refs/heads/master@{#13429}
Reason for revert:
It keeps breaking upstream.
Original issue's description:
> Reland Issue 2061423003: Refactor NACK bitrate allocation
>
> This is a reland of https://codereview.webrtc.org/2061423003/
> Which was reverted in https://codereview.webrtc.org/2131913003/
>
> The reason for the revert was that some upstream code used
> RtpSender::SetTargetBitrate(). I've added that back as a no-op until we
> it's been brought up to date.
>
> TBR=tommi@webrtc.org
>
> Committed: 05ce4ae31fTBR=tommi@webrtc.org,sprang@webrtc.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
Review-Url: https://codereview.webrtc.org/2130423002
Cr-Commit-Position: refs/heads/master@{#13419}
Reason for revert:
Breaks upstream code.
Original issue's description:
> Refactor NACK bitrate allocation
>
> Nack bitrate allocation should not be done on a per-rtp-module basis,
> but rather shared bitrate pool per call. This CL moves allocation to the
> pacer and cleans up a bunch if bitrate stats handling.
>
> BUG=
> R=danilchap@webrtc.org, stefan@webrtc.org, tommi@webrtc.org
>
> Committed: 5fc59e810bTBR=tommi@webrtc.org,danilchap@webrtc.org,stefan@webrtc.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=
Review-Url: https://codereview.webrtc.org/2131913003
Cr-Commit-Position: refs/heads/master@{#13417}
A recent refactoring (r13192) introduced a bug where the min transmit
config wasn't being respected. Specifically, if a VideoSendStream was
created without it and the reconfigured, the min transmit bitrate would
not take effect. Probably the other way around as well.
BUG=webrtc::5687
Review-Url: https://codereview.webrtc.org/2106183002
Cr-Commit-Position: refs/heads/master@{#13390}
When the target bitrate is zero, currently VideoSendStream.Stats.target_media_bitrate_bps show the last set rate before the target was set to zero.
BUG=webrtc::5687 b/29574845
Review-Url: https://codereview.webrtc.org/2122743003
Cr-Commit-Position: refs/heads/master@{#13386}
This CL changes the auto-pause logic to suspend a stream based on the
encoder target bitrate instead of the allocated bitrate for a stream,
to account for possible protection, e.g. FEC and NACK.
This CL also adds periodic logging of the current BWE and possibility
to run with suspension in video loopback test.
BUG=webrtc:5868
R=stefan@webrtc.org
Review URL: https://codereview.webrtc.org/2117493002 .
Cr-Commit-Position: refs/heads/master@{#13360}
Was thought to be only flaky on Mac, but just failed on Win SyzyASan.
So, disabling until flakiness is fixed.
BUG=webrtc:4332
TBR=pbos@webrtc.org
NOTRY=True
Review-Url: https://codereview.webrtc.org/2104583002
Cr-Commit-Position: refs/heads/master@{#13303}
Reason for revert:
Reverting the revert. This change is not related to the failure on the Windows FYI bots. The cause of the failure has been reverted in Chromium:
https://codereview.chromium.org/2081653004/
Original issue's description:
> Revert of Split IncomingVideoStream into two implementations, with smoothing and without. (patchset #5 id:340001 of https://codereview.webrtc.org/2078873002/ )
>
> Reason for revert:
> Breaks chromium.webrtc.fyi
>
> https://uberchromegw.corp.google.com/i/chromium.webrtc.fyi/builders/Win7%20Tester/builds/4719
> https://uberchromegw.corp.google.com/i/chromium.webrtc.fyi/builders/Win10%20Tester/builds/3120
>
> Original issue's description:
> > Reland of IncomingVideoStream refactoring.
> > This reland does not contain the non-smoothing part of the original implementation. Instead, when smoothing is turned off, frame callbacks run on the decoder thread, as they did before. This code path is used in Chrome. As far as Chrome goes, the difference now is that there won't be an instance of IncomingVideoStream in between the decoder and the callback (i.e. fewer locks). Other than that, no change for Chrome.
> >
> > Original issue's description (with non-smoothing references removed):
> >
> > Split IncomingVideoStream into two implementations, with smoothing and without.
> >
> > * Added TODOs and documentation for VideoReceiveStream::OnFrame, where we today grab 6 locks.
> >
> > * Removed the Start/Stop methods from the IncomingVideoStream implementations. Now, when an instance is created, it should be considered to be "running" and when it is deleted, it's "not running". This saves on resources and also reduces the amount of locking required and I could remove one critical section altogether.
> >
> > * Changed the VideoStreamDecoder class to not depend on IncomingVideoStream but rather use the generic rtc::VideoSinkInterface<VideoFrame> interface. This means that any implementation of that interface can be used and the decoder can be made to just use the 'renderer' from the config. Once we do that, we can decouple the IncomingVideoStream implementations from the decoder and VideoReceiveStream implementations and leave it up to the application for how to do smoothing. The app can choose to use the Incoming* classes or roll its own (which may be preferable since applications often have their own scheduling mechanisms).
> >
> > * The lifetime of the VideoStreamDecoder instance is now bound to Start/Stop in VideoReceiveStream and not all of the lifetime of VideoReceiveStream.
> >
> > * Fixed VideoStreamDecoder to unregister callbacks in the dtor that were registered in the ctor. (this was open to a use-after-free regression)
> >
> > * Delay and callback pointers are now passed via the ctors to the IncomingVideoStream classes. The thread primitives in the IncomingVideoStream classes are also constructed/destructed at the same time as the owning object, which allowed me to remove one more lock.
> >
> > * Removed code in the VideoStreamDecoder that could overwrite the VideoReceiveStream render delay with a fixed value of 10ms on construction. This wasn't a problem with the previous implementation (it would be now though) but seemed to me like the wrong place to be setting that value.
> >
> > * Made the render delay value in VideoRenderFrames, const.
> >
> > BUG=chromium:620232
> > R=mflodman@webrtc.org, nisse@webrtc.org
> >
> > Committed: https://crrev.com/884c336c345d988974c2a69cea402b0fb8b07a63
> > Cr-Commit-Position: refs/heads/master@{#13219}
>
> TBR=nisse@webrtc.org,philipel@webrtc.org,mflodman@webrtc.org,tommi@webrtc.org
> # Skipping CQ checks because original CL landed less than 1 days ago.
> NOPRESUBMIT=true
> NOTREECHECKS=true
> NOTRY=true
> BUG=chromium:620232
>
> Committed: https://crrev.com/a536bfe70de38fe877245317a7f0b00bcf69cbd0
> Cr-Commit-Position: refs/heads/master@{#13229}
TBR=nisse@webrtc.org,philipel@webrtc.org,mflodman@webrtc.org,sakal@webrtc.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=chromium:620232
Review-Url: https://codereview.webrtc.org/2089613002
Cr-Commit-Position: refs/heads/master@{#13230}
Reason for revert:
Breaks chromium.webrtc.fyi
https://uberchromegw.corp.google.com/i/chromium.webrtc.fyi/builders/Win7%20Tester/builds/4719https://uberchromegw.corp.google.com/i/chromium.webrtc.fyi/builders/Win10%20Tester/builds/3120
Original issue's description:
> Reland of IncomingVideoStream refactoring.
> This reland does not contain the non-smoothing part of the original implementation. Instead, when smoothing is turned off, frame callbacks run on the decoder thread, as they did before. This code path is used in Chrome. As far as Chrome goes, the difference now is that there won't be an instance of IncomingVideoStream in between the decoder and the callback (i.e. fewer locks). Other than that, no change for Chrome.
>
> Original issue's description (with non-smoothing references removed):
>
> Split IncomingVideoStream into two implementations, with smoothing and without.
>
> * Added TODOs and documentation for VideoReceiveStream::OnFrame, where we today grab 6 locks.
>
> * Removed the Start/Stop methods from the IncomingVideoStream implementations. Now, when an instance is created, it should be considered to be "running" and when it is deleted, it's "not running". This saves on resources and also reduces the amount of locking required and I could remove one critical section altogether.
>
> * Changed the VideoStreamDecoder class to not depend on IncomingVideoStream but rather use the generic rtc::VideoSinkInterface<VideoFrame> interface. This means that any implementation of that interface can be used and the decoder can be made to just use the 'renderer' from the config. Once we do that, we can decouple the IncomingVideoStream implementations from the decoder and VideoReceiveStream implementations and leave it up to the application for how to do smoothing. The app can choose to use the Incoming* classes or roll its own (which may be preferable since applications often have their own scheduling mechanisms).
>
> * The lifetime of the VideoStreamDecoder instance is now bound to Start/Stop in VideoReceiveStream and not all of the lifetime of VideoReceiveStream.
>
> * Fixed VideoStreamDecoder to unregister callbacks in the dtor that were registered in the ctor. (this was open to a use-after-free regression)
>
> * Delay and callback pointers are now passed via the ctors to the IncomingVideoStream classes. The thread primitives in the IncomingVideoStream classes are also constructed/destructed at the same time as the owning object, which allowed me to remove one more lock.
>
> * Removed code in the VideoStreamDecoder that could overwrite the VideoReceiveStream render delay with a fixed value of 10ms on construction. This wasn't a problem with the previous implementation (it would be now though) but seemed to me like the wrong place to be setting that value.
>
> * Made the render delay value in VideoRenderFrames, const.
>
> BUG=chromium:620232
> R=mflodman@webrtc.org, nisse@webrtc.org
>
> Committed: https://crrev.com/884c336c345d988974c2a69cea402b0fb8b07a63
> Cr-Commit-Position: refs/heads/master@{#13219}
TBR=nisse@webrtc.org,philipel@webrtc.org,mflodman@webrtc.org,tommi@webrtc.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=chromium:620232
Review-Url: https://codereview.webrtc.org/2084873002
Cr-Commit-Position: refs/heads/master@{#13229}
This reland does not contain the non-smoothing part of the original implementation. Instead, when smoothing is turned off, frame callbacks run on the decoder thread, as they did before. This code path is used in Chrome. As far as Chrome goes, the difference now is that there won't be an instance of IncomingVideoStream in between the decoder and the callback (i.e. fewer locks). Other than that, no change for Chrome.
Original issue's description (with non-smoothing references removed):
Split IncomingVideoStream into two implementations, with smoothing and without.
* Added TODOs and documentation for VideoReceiveStream::OnFrame, where we today grab 6 locks.
* Removed the Start/Stop methods from the IncomingVideoStream implementations. Now, when an instance is created, it should be considered to be "running" and when it is deleted, it's "not running". This saves on resources and also reduces the amount of locking required and I could remove one critical section altogether.
* Changed the VideoStreamDecoder class to not depend on IncomingVideoStream but rather use the generic rtc::VideoSinkInterface<VideoFrame> interface. This means that any implementation of that interface can be used and the decoder can be made to just use the 'renderer' from the config. Once we do that, we can decouple the IncomingVideoStream implementations from the decoder and VideoReceiveStream implementations and leave it up to the application for how to do smoothing. The app can choose to use the Incoming* classes or roll its own (which may be preferable since applications often have their own scheduling mechanisms).
* The lifetime of the VideoStreamDecoder instance is now bound to Start/Stop in VideoReceiveStream and not all of the lifetime of VideoReceiveStream.
* Fixed VideoStreamDecoder to unregister callbacks in the dtor that were registered in the ctor. (this was open to a use-after-free regression)
* Delay and callback pointers are now passed via the ctors to the IncomingVideoStream classes. The thread primitives in the IncomingVideoStream classes are also constructed/destructed at the same time as the owning object, which allowed me to remove one more lock.
* Removed code in the VideoStreamDecoder that could overwrite the VideoReceiveStream render delay with a fixed value of 10ms on construction. This wasn't a problem with the previous implementation (it would be now though) but seemed to me like the wrong place to be setting that value.
* Made the render delay value in VideoRenderFrames, const.
BUG=chromium:620232
R=mflodman@webrtc.org, nisse@webrtc.org
Review URL: https://codereview.webrtc.org/2078873002 .
Cr-Commit-Position: refs/heads/master@{#13219}
Allows detecting large-enough audio packets as part of a probe,
speculative fix for a rampup-time regression in M50. These packets are
accounted on the send side when probing.
BUG=webrtc:5985
R=mflodman@webrtc.org, philipel@webrtc.org
Review URL: https://codereview.webrtc.org/2061193002 .
Cr-Commit-Position: refs/heads/master@{#13210}
This cl change so that VideoSendStream::Start adds the stream as a BitrateObserver and VideoSendStream::Stop removes the stream as observer.
That also means that start will trigger a VideoEncoder::SetRate call with the most recent bitrate estimate.
VideoSendStream::Stop will trigger a VideoEncoder::SetRate with bitrate = 0.
BUG=webrtc:5687 b/28636240
Review-Url: https://codereview.webrtc.org/2070343002
Cr-Commit-Position: refs/heads/master@{#13192}
Reason for revert:
Reverting again. The perf regression does not seem to be related to dropping frames.
Original issue's description:
> Reland of Split IncomingVideoStream into two implementations, with smoothing and without.
>
> Original issue's description:
>
> Split IncomingVideoStream into two implementations, with smoothing and without.
>
> This CL fixes an issue with the non-smoothing implementation where frames were delivered on the decoder thread. No-smoothing is now done in a separate class that uses a TaskQueue. The implementation may drop frames if the renderer doesn't keep up and it doesn't block the decoder thread.
>
> Further work done:
>
> * I added TODOs and documentation for VideoReceiveStream::OnFrame, where we today grab 5 locks.
>
> * I removed the Start/Stop methods from the IncomingVideoStream implementations. Now, when an instance is created, it should be considered to be "running" and when it is deleted, it's "not running". This saves on resources and also reduces the amount of locking required and I could remove one critical section altogether.
>
> * I changed the VideoStreamDecoder class to not depend on IncomingVideoStream but rather use the generic rtc::VideoSinkInterface<VideoFrame> interface. This means that any implementation of that interface can be used and the decoder can be made to just use the 'renderer' from the config. Once we do that, we can decouple the IncomingVideoStream implementations from the decoder and VideoReceiveStream implementations and leave it up to the application for how to do smoothing. The app can choose to use the Incoming* classes or roll its own (which may be preferable since applications often have their own scheduling mechanisms).
>
> * The non-smoothing IncomingVideoStream implementation currently allows only 1 outstanding pending frame. If we exceed that, the current frame won't be delivered to the renderer and instead we deliver the next one (since when this happens, the renderer is falling behind).
>
> * The lifetime of the VideoStreamDecoder instance is now bound to Start/Stop in VideoReceiveStream and not all of the lifetime of VideoReceiveStream.
>
> * Fixed VideoStreamDecoder to unregister callbacks in the dtor that were registered in the ctor. (this was open to a use-after-free regression)
>
> * Delay and callback pointers are now passed via the ctors to the IncomingVideoStream classes. The thread primitives in the IncomingVideoStream classes are also constructed/destructed at the same time as the owning object, which allowed me to remove one more lock.
>
> * Removed code in the VideoStreamDecoder that could overwrite the VideoReceiveStream render delay with a fixed value of 10ms on construction. This wasn't a problem with the previous implementation (it would be now though) but seemed to me like the wrong place to be setting that value.
>
> * Made the render delay value in VideoRenderFrames, const.
>
> BUG=chromium:620232
> TBR=mflodman
>
> Committed: https://crrev.com/e03f8787377bbc03a4e00184bb14b7561b108cbb
> Cr-Commit-Position: refs/heads/master@{#13175}
TBR=mflodman@webrtc.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=chromium:620232
Review-Url: https://codereview.webrtc.org/2071093002
Cr-Commit-Position: refs/heads/master@{#13176}
Original issue's description:
Split IncomingVideoStream into two implementations, with smoothing and without.
This CL fixes an issue with the non-smoothing implementation where frames were delivered on the decoder thread. No-smoothing is now done in a separate class that uses a TaskQueue. The implementation may drop frames if the renderer doesn't keep up and it doesn't block the decoder thread.
Further work done:
* I added TODOs and documentation for VideoReceiveStream::OnFrame, where we today grab 5 locks.
* I removed the Start/Stop methods from the IncomingVideoStream implementations. Now, when an instance is created, it should be considered to be "running" and when it is deleted, it's "not running". This saves on resources and also reduces the amount of locking required and I could remove one critical section altogether.
* I changed the VideoStreamDecoder class to not depend on IncomingVideoStream but rather use the generic rtc::VideoSinkInterface<VideoFrame> interface. This means that any implementation of that interface can be used and the decoder can be made to just use the 'renderer' from the config. Once we do that, we can decouple the IncomingVideoStream implementations from the decoder and VideoReceiveStream implementations and leave it up to the application for how to do smoothing. The app can choose to use the Incoming* classes or roll its own (which may be preferable since applications often have their own scheduling mechanisms).
* The non-smoothing IncomingVideoStream implementation currently allows only 1 outstanding pending frame. If we exceed that, the current frame won't be delivered to the renderer and instead we deliver the next one (since when this happens, the renderer is falling behind).
* The lifetime of the VideoStreamDecoder instance is now bound to Start/Stop in VideoReceiveStream and not all of the lifetime of VideoReceiveStream.
* Fixed VideoStreamDecoder to unregister callbacks in the dtor that were registered in the ctor. (this was open to a use-after-free regression)
* Delay and callback pointers are now passed via the ctors to the IncomingVideoStream classes. The thread primitives in the IncomingVideoStream classes are also constructed/destructed at the same time as the owning object, which allowed me to remove one more lock.
* Removed code in the VideoStreamDecoder that could overwrite the VideoReceiveStream render delay with a fixed value of 10ms on construction. This wasn't a problem with the previous implementation (it would be now though) but seemed to me like the wrong place to be setting that value.
* Made the render delay value in VideoRenderFrames, const.
BUG=chromium:620232
TBR=mflodman
Review-Url: https://codereview.webrtc.org/2071473002
Cr-Commit-Position: refs/heads/master@{#13175}
This change reduces the number of times the Android hardware video
encoder is reconfigured when making an outgoing call. With this change,
the encoder should only be initialized once as opposed to the ~3 times
it happens currently.
Before the fix, the following sequence of events caused the extra
reconfigurations:
1. After the SetLocalDescription call, the WebRtcVideoSendStream is created.
All frames from the camera are dropped until the corresponding
VideoSendStream is created.
2. SetRemoteDescription() triggers the VideoSendStream creation. At
this point, the encoder is configured for the first time, with the
frame dimensions set to a low resolution default (176x144).
3. When the first video frame is received from the camera after the
VideoSendStreamIsCreated, the encoder is reconfigured to the correct
dimensions. If we are using the Android hardware encoder, the default
configuration is set to encode from a memory buffer (use_surface=false).
4. When the frame is passed down to the encoder in
androidmediaencoder_jni.cc EncodeOnCodecThread(), it may be stored in
a texture instead of a memory buffer. In this case, yet another
reconfiguration takes place to enable encoding from a texture.
5. Even if the resolution and texture flag were known at the start of
the call, there would be a reconfiguration involved if the camera is
rotated (such as when making a call from a phone in portrait orientation).
The reason for that is that at construction time, WebRtcVideoEngine2
sets the VideoSinkWants structure parameter to request frames rotated
by the source; the early frames will then arrive in portrait resolution.
When the remote description is finally set, if the rotation RTP extension
is supported by the remote receiver, the source is asked to provide
non-rotated frames. The very next frame will then arrive in landscape
resolution with a non-zero rotation value to be applied by the receiver.
Since the encoder was configured with the last (portrait) frame size,
it's going to need to be reconfigured again.
The fix makes the following changes:
1. WebRtcVideoSendStream::OnFrame() now caches the last seen frame
dimensions, and whether the frame was stored in a texture.
2. When the encoder is configured the first time
(WebRtcVideoSendStream::SetCodec()) - the last seen frame dimensions
are used instead of the default dimensions.
3. A flag that indicates if encoding is to be done from a texture has
been added to the webrtc::VideoStream and webrtc::VideoCodec structs,
and it's been wired up to be passed down all the way to the JNI code in
androidmediaencoder_jni.cc.
4. MediaCodecVideoEncoder::InitEncode is now reading the is_surface
flag from the VideoCodec structure instead of guessing the default as
false. This way we end up with the correct encoder configuration the
first time around.
5. WebRtcVideoSendStream now takes an optimistic guess and requests non-
rotated frames when the supported RtpExtensions list is not available.
This makes the "early" frames arrive non-rotated, and the cached dimensions
will be correct for the common case when the rotation extension is supported.
If the other side is an older endpoint which does not support rotation,
the encoder will have to be reconfigured - but it's better to penalize the
uncommon case rather than the common one.
Review-Url: https://codereview.webrtc.org/2067103002
Cr-Commit-Position: refs/heads/master@{#13173}
1. It moves calculation of the needed padding to VideoSendStream instead of ViEEncoder and only does it once per send Stream instead of every time the network estimate changes.
2. The maximum amount of padding sent was prior to this cl calculated and updated based on network estimate changes. However, it can only change based on encoder configuration changes and if send streams are added or removed. This cl change the VideoSendStream/VieEncoder to notify the BitrateAllocator of changes to the needed padding bitrate and for BitrateAllocator to notify Call of these changes.
3. Fixed an issue in the SendPacer where it could send a padding packet before sending a real packet. This caused the test EndToEndTest.RestartingSendStreamPreservesRtpStatesWithRtx to fail with these refactorings since the pacer suddenly could send a padding packet before the encoder had produced its first frame.
BUG=webrtc:5687
Review-Url: https://codereview.webrtc.org/1993113003
Cr-Commit-Position: refs/heads/master@{#13149}
Reason for revert:
Reverting while we track down the issue on the Win10 bot.
Original issue's description:
> Split IncomingVideoStream into two implementations, with smoothing and without.
>
> This CL fixes an issue with the non-smoothing implementation where frames were delivered on the decoder thread. No-smoothing is now done in a separate class that uses a TaskQueue. The implementation may drop frames if the renderer doesn't keep up and it doesn't block the decoder thread.
>
> Further work done:
>
> * I added TODOs and documentation for VideoReceiveStream::OnFrame, where we today grab 5 locks.
>
> * I removed the Start/Stop methods from the IncomingVideoStream implementations. Now, when an instance is created, it should be considered to be "running" and when it is deleted, it's "not running". This saves on resources and also reduces the amount of locking required and I could remove one critical section altogether.
>
> * I changed the VideoStreamDecoder class to not depend on IncomingVideoStream but rather use the generic rtc::VideoSinkInterface<VideoFrame> interface. This means that any implementation of that interface can be used and the decoder can be made to just use the 'renderer' from the config. Once we do that, we can decouple the IncomingVideoStream implementations from the decoder and VideoReceiveStream implementations and leave it up to the application for how to do smoothing. The app can choose to use the Incoming* classes or roll its own (which may be preferable since applications often have their own scheduling mechanisms).
>
> * The non-smoothing IncomingVideoStream implementation currently allows only 1 outstanding pending frame. If we exceed that, the current frame won't be delivered to the renderer and instead we deliver the next one (since when this happens, the renderer is falling behind).
>
> * The lifetime of the VideoStreamDecoder instance is now bound to Start/Stop in VideoReceiveStream and not all of the lifetime of VideoReceiveStream.
>
> * Fixed VideoStreamDecoder to unregister callbacks in the dtor that were registered in the ctor. (this was open to a use-after-free regression)
>
> * Delay and callback pointers are now passed via the ctors to the IncomingVideoStream classes. The thread primitives in the IncomingVideoStream classes are also constructed/destructed at the same time as the owning object, which allowed me to remove one more lock.
>
> * Removed code in the VideoStreamDecoder that could overwrite the VideoReceiveStream render delay with a fixed value of 10ms on construction. This wasn't a problem with the previous implementation (it would be now though) but seemed to me like the wrong place to be setting that value.
>
> * Made the render delay value in VideoRenderFrames, const.
>
> BUG=
>
> Committed: https://crrev.com/1c7eef652b0aa22d8ebb0bfe2b547094a794be22
> Cr-Commit-Position: refs/heads/master@{#13129}
TBR=mflodman@webrtc.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=
Review-Url: https://codereview.webrtc.org/2061363002
Cr-Commit-Position: refs/heads/master@{#13146}
EncodedFrameCallbackAdapter was used VideoSendStream and
VideoReceiveStream, but there is no reason to have it as these classes
can call EncodedFrameObserver directly.
Review-Url: https://codereview.webrtc.org/2068463004
Cr-Commit-Position: refs/heads/master@{#13145}
This CL fixes an issue with the non-smoothing implementation where frames were delivered on the decoder thread. No-smoothing is now done in a separate class that uses a TaskQueue. The implementation may drop frames if the renderer doesn't keep up and it doesn't block the decoder thread.
Further work done:
* I added TODOs and documentation for VideoReceiveStream::OnFrame, where we today grab 5 locks.
* I removed the Start/Stop methods from the IncomingVideoStream implementations. Now, when an instance is created, it should be considered to be "running" and when it is deleted, it's "not running". This saves on resources and also reduces the amount of locking required and I could remove one critical section altogether.
* I changed the VideoStreamDecoder class to not depend on IncomingVideoStream but rather use the generic rtc::VideoSinkInterface<VideoFrame> interface. This means that any implementation of that interface can be used and the decoder can be made to just use the 'renderer' from the config. Once we do that, we can decouple the IncomingVideoStream implementations from the decoder and VideoReceiveStream implementations and leave it up to the application for how to do smoothing. The app can choose to use the Incoming* classes or roll its own (which may be preferable since applications often have their own scheduling mechanisms).
* The non-smoothing IncomingVideoStream implementation currently allows only 1 outstanding pending frame. If we exceed that, the current frame won't be delivered to the renderer and instead we deliver the next one (since when this happens, the renderer is falling behind).
* The lifetime of the VideoStreamDecoder instance is now bound to Start/Stop in VideoReceiveStream and not all of the lifetime of VideoReceiveStream.
* Fixed VideoStreamDecoder to unregister callbacks in the dtor that were registered in the ctor. (this was open to a use-after-free regression)
* Delay and callback pointers are now passed via the ctors to the IncomingVideoStream classes. The thread primitives in the IncomingVideoStream classes are also constructed/destructed at the same time as the owning object, which allowed me to remove one more lock.
* Removed code in the VideoStreamDecoder that could overwrite the VideoReceiveStream render delay with a fixed value of 10ms on construction. This wasn't a problem with the previous implementation (it would be now though) but seemed to me like the wrong place to be setting that value.
* Made the render delay value in VideoRenderFrames, const.
BUG=
Review-Url: https://codereview.webrtc.org/2035173002
Cr-Commit-Position: refs/heads/master@{#13129}
To avoid the case where a single data point or too short window is used,
causing bad behavior due to bad stats, update RateStatistics to return
an Optional rather than a plain rate.
There was also a strange off by one bug where the rate was slightly
overestimated (N + 1 buckets, N ms time window).
These changes requires updates to a number of places, and may very well
cause seeming perf regressions (but the stats were probablty more wrong
previously).
BUG=
R=mflodman@webrtc.org, stefan@webrtc.org
Review URL: https://codereview.webrtc.org/2029593002 .
Cr-Commit-Position: refs/heads/master@{#13103}
Instead of the default copy constructor, the Copy() method has to be used. In this CL, the number of copies has been reduced significantly in production code. One case in the video engine remains, where we need to restart a video stream. Even in that case, I'm sure we could avoid it, but for this particular CL, I decided against it to keep things simple (and it's also an edge case). Most importantly, creating copies is made harder and the interface encourages ownership transfers.
R=mflodman@webrtc.org, pbos@webrtc.org
Review URL: https://codereview.webrtc.org/2042603002 .
Cr-Commit-Position: refs/heads/master@{#13102}
This CL implements auto pausing video streams per stream with logic to
avoid toggling state too often.
Also re-enabling tests disabled for Mac, with the assumption the new
logic removes flakiness.
BUG=webrtc:5868,webrtc:5407
R=pbos@webrtc.org, stefan@webrtc.org
Review URL: https://codereview.webrtc.org/2035383002 .
Cr-Commit-Position: refs/heads/master@{#13092}
This CL adds support for an extension on RTP frames to allow the sender
to specify the minimum and maximum playout delay limits.
The receiver makes a best-effort attempt to keep the capture-to-render delay
within this range. This allows different types of application to specify
different end-to-end delay goals. For example gaming can support rendering
of frames as soon as received on receiver to minimize delay. A movie playback
application can specify a minimum playout delay to allow fixed buffering
in presence of network jitter.
There are no tests at this time and most of testing is done with chromium
webrtc prototype.
On chromoting performance tests, this extension helps bring down end-to-end
delay by about 150 ms on small frames.
BUG=webrtc:5895
Review-Url: https://codereview.webrtc.org/2007743003
Cr-Commit-Position: refs/heads/master@{#13059}
This shouldn't be needed, but because the receiver assumes RTX packets
contain RED if configured to receive them (due to an incompatibility
issue), we also have to make sure we send them for now.
BUG=webrtc:5675
Review-Url: https://codereview.webrtc.org/2033763002
Cr-Commit-Position: refs/heads/master@{#13024}
This cl split the class MediaOptimization into two parts. One that deals with frame dropping and stats and one new class called ProtectionBitrateCalculator that deals with calculating the needed FEC parameters and how much of the estimated network bitrate that can be used by an encoder
Note that the logic of how FEC and the needed bitrates is not changed.
BUG=webrtc:5687
R=asapersson@webrtc.org, stefan@webrtc.org
Review URL: https://codereview.webrtc.org/1972083002 .
Cr-Commit-Position: refs/heads/master@{#13018}
The fake network pipe will still only drop packets at an average rate of
|loss_percent| but in bursts at an average length specified by
|avg_burst_loss_length|.
Also added the flag -avg_burst_loss_length to video loopback.
BUG=
Review-Url: https://codereview.webrtc.org/1995683003
Cr-Commit-Position: refs/heads/master@{#12969}
Currently there are two structs that are identical and track extension details:
webrtc::RtpExtension
cricket::RtpHeaderExtension
The use of the structs is mixed in the code to track the extensions being
supported. This results in duplicate definition of
the URI constants and there is code to convert between the two structs.
Clean up to use a single RtpHeader throughout the codebase. The actual location
of RtpHeader may change in future (perhaps to be located in api/). Additionally,
this CL renames some of the constants to clarify Uri and Id use.
BUG= webrtc:5895
Review-Url: https://codereview.webrtc.org/1984983002
Cr-Commit-Position: refs/heads/master@{#12924}
Also set the Configuration parameters in CreateRtpRtcpModules in the same order as the members are declared.
BUG=webrtc:5917
Review-Url: https://codereview.webrtc.org/2011433002
Cr-Commit-Position: refs/heads/master@{#12905}
During a period (e.g. 2 seconds), different metrics can be computed, e.g.:
- average of samples
- percentage/permille of samples
- units per second
Each periodic metric can be either:
- reported to an observer each period
- aggregated during the call (e.g. min, max, average)
BUG=webrtc:5283
Review-Url: https://codereview.webrtc.org/1640053003
Cr-Commit-Position: refs/heads/master@{#12847}
