Problem seems to be that once the estimate drops, "sample_uncertainty"
becomes very large, and it therefore takes a long time to recover.
Fix is under config for further downstream verification.
Bug: webrtc:10462
Change-Id: I5c2035f06e8a5088db0f0cb6ca511ef900e07645
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/128902
Commit-Queue: Christoffer Rodbro <crodbro@webrtc.org>
Reviewed-by: Sebastian Jansson <srte@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27320}
This change prevents FilterAnalyzer from accessing memory out-of-bounds
when the filter is resized.
Bug: chromium:946439
Change-Id: I7e2392c8b1ff0ff55566c663bf6d7a40d7754501
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/129928
Reviewed-by: Per Åhgren <peah@webrtc.org>
Commit-Queue: Gustaf Ullberg <gustaf@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27318}
This CL allows us to control how many probes we send when the bandwidth
allocation is updated, and how big they are.
Bug: webrtc:10394
Change-Id: I19e40740a528f83384b65d7509295034cc9a3031
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/129904
Commit-Queue: Jonas Olsson <jonasolsson@webrtc.org>
Reviewed-by: Sebastian Jansson <srte@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27317}
The pointer-to-submodule interfaces are being removed.
This CL:
1) introduces AudioProcessing::Config::GainController1 with most config,
2) adds functions to APM for setting and getting analog gain,
3) creates a temporary GainControlConfigProxy to support the transition
to the new config.
4) Moves the lock references in GainControlForExperimentalAgc and
GainControlImpl into the GainControlConfigProxy, as it becomes the
sole AGC object with functionality exposed to the client.
Bug: webrtc:9947, webrtc:9878
Change-Id: Ic31e15e9bb26d6497a92b77874e0b6cab21ff2b2
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/126485
Commit-Queue: Sam Zackrisson <saza@webrtc.org>
Reviewed-by: Fredrik Solenberg <solenberg@webrtc.org>
Reviewed-by: Alessio Bazzica <alessiob@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27316}
This CL changes the API for webrtc::VideoEncoder.
There is a legacy method called SetRates(). This is indicated as being
deprecated, but there seem to be a number of usages still left.
Then there is the new SetRateAllocation() method which takes a
VideoBitrateAllocation instance instead of a single target bitrate.
This CL adds a new version of SetRates() which moves all the existing
parameters in a RateControlParameters struct, and adds a bandwidth
allocation signal. The intent of this signal is to allow the encoder
to know how close to the target it needs to stay. If the encoder rate
is network restricted, it will need to be more aggressive in keep the
rate down and possibly drop frames. If there is some margin for
overshoot, it might do different trade-offs.
Furthermore, the frame rate signal is changes from an integer to a
floating point type in order to get more precise rates at low frame
rates and the return type has been changed to void since the only use
of the previous values to log it and that is better done inside encoder
where the error condition originates.
The intent is to properly deprecate the "old" SetRates() /
SetRateAllocation() methods, send out a PSA and then remove them in two
weeks. Changes required by users should be trivial, as using the new
headroom signal is optional.
Bug: webrtc:10155, webrtc:10481
Change-Id: I4f797b0b0c73086111869ef4ee5f42bf530f5fde
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/129724
Commit-Queue: Erik Språng <sprang@webrtc.org>
Reviewed-by: Niels Moller <nisse@webrtc.org>
Reviewed-by: Rasmus Brandt <brandtr@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27314}
As a library, WebRTC should not assume UNICODE and _UNICODE to be
defined globally.
This CL explicitly selects wide character functions and types in
order to build WebRTC with /UUNICODE and /U_UNICODE.
Bug: None
Change-Id: Ie4e2bcb4c5c34aee6f68dc7b5b54b76f088ee3e4
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/128904
Commit-Queue: Tommi <tommi@webrtc.org>
Reviewed-by: Tommi <tommi@webrtc.org>
Reviewed-by: Noah Richards <noahric@chromium.org>
Cr-Commit-Position: refs/heads/master@{#27313}
This lets us change how many bytes and packets goes into the probes, as
well as some other things.
Bug: webrtc:10394
Change-Id: I26bb26a644e6f00366e9275228760c8744d63735
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/128424
Commit-Queue: Jonas Olsson <jonasolsson@webrtc.org>
Reviewed-by: Sebastian Jansson <srte@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27312}
Even if neither frame height nor frame width is <=0 we can end up
with <=0 dimensions in renderHeight or renderWidth. With this change,
we perform the check on the latter.
Bug: webrtc:10367
Change-Id: I9672672659ad7d12cf1e7ccab5b5cde583ae7e8c
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/129760
Reviewed-by: Magnus Jedvert <magjed@webrtc.org>
Commit-Queue: Paulina Hensman <phensman@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27307}
- Add GetFrameStatistics API:
This is useful for downstream test users that want to read frame-level stats.
- Remove other APIs that are not used by downstream tests:
* AddFrame
* GetFrame
* GetFrameWithTimestamp
* SliceAndCalcAggregatedVideoStatistic
* PrintFrameStatistics
* Size
* Clear
The implementations, which are used by the fixture implementation, are kept.
Bug: webrtc:10349
Change-Id: Id2f6fa5a36b8341a5ccb365725f71ebe0c0f1570
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/128779
Reviewed-by: Karl Wiberg <kwiberg@webrtc.org>
Reviewed-by: Sergey Silkin <ssilkin@webrtc.org>
Commit-Queue: Rasmus Brandt <brandtr@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27306}
This reverts commit a39254da593bbdb0b1e072a44827229680afe3ee.
Reason for revert: Tests are failing due to ThreadChecker's called on valid thread.
Original change's description:
> in WebrtcVoiceEngine allow to set TaskQueueFactory
>
> in production code keep using GlobalTaskQueueFactory()
> in tests switch to use DefaultTaskQueueFactory directly.
>
> Bug: webrtc:10284
> Change-Id: I170274a98324796623089a965a39f0cbb7e281d9
> Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/128878
> Reviewed-by: Steve Anton <steveanton@webrtc.org>
> Commit-Queue: Danil Chapovalov <danilchap@webrtc.org>
> Cr-Commit-Position: refs/heads/master@{#27296}
TBR=danilchap@webrtc.org,steveanton@webrtc.org
Change-Id: I9742e5d0171a94f3840e197c40fdb44523e4963b
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: webrtc:10284
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/129780
Reviewed-by: Amit Hilbuch <amithi@webrtc.org>
Commit-Queue: Amit Hilbuch <amithi@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27297}
in production code keep using GlobalTaskQueueFactory()
in tests switch to use DefaultTaskQueueFactory directly.
Bug: webrtc:10284
Change-Id: I170274a98324796623089a965a39f0cbb7e281d9
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/128878
Reviewed-by: Steve Anton <steveanton@webrtc.org>
Commit-Queue: Danil Chapovalov <danilchap@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27296}
It isn't used and can easily be replaced with posting a task, as shown
in the changes to the unit test.
Also removing the sequenced task checker that no longer adds any value.
We now ensure that the task can only be stopped with a reference to the
task queue it runs on.
Bug: webrtc:10365
Change-Id: Ie8aef6f742c55db1fb686f20c2a28c606c721fa0
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/129725
Commit-Queue: Sebastian Jansson <srte@webrtc.org>
Reviewed-by: Karl Wiberg <kwiberg@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27292}
Earlier CLs assumed that the object pointed to by call_ had to be
accessed on the worker thread. While this is generally the case,
Call::MediaTransportChange is explicitly thread safe, so
PeerConnection::OnTransportChanged doesn't have to run on the worker
thread for that reason.
Which is fortunate, because it actually runs on the network thread.
The RTC_RUN_ON(worker_thread()) annotation on the method declaration
was ineffective because this method is being called via a base class
pointer; replacing it with a call to
RTC_DCHECK_RUN_ON(worker_thread()) in the function body immediately
triggered assertions in the unit tests.
Bug: webrtc:9987
Change-Id: I08cf558a74f4ca2b2eff8ef4810ebbd1287a9726
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/129442
Commit-Queue: Karl Wiberg <kwiberg@webrtc.org>
Reviewed-by: Steve Anton <steveanton@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27287}
In case protection bitrate (fec/nack) is used, don't allow encoder
(rate adjuster) to utilize those bits as that may result in pacer delay.
Bug: webrtc:10155
Change-Id: I5bceb100396f0ae2131db51039e8524ca068d13f
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/128873
Reviewed-by: Ilya Nikolaevskiy <ilnik@webrtc.org>
Commit-Queue: Erik Språng <sprang@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27283}
Injection is made possible through VP8Encoder::Create.
According to native-api.md, it is a defacto public API despite
not being in the api/ folder.
Bug: webrtc:10259, webrtc:10382
Change-Id: Ifc5d55aa99613cfee0fcb4f0c6690121c85b2e3e
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/128883
Reviewed-by: Karl Wiberg <kwiberg@webrtc.org>
Reviewed-by: Erik Språng <sprang@webrtc.org>
Commit-Queue: Elad Alon <eladalon@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27281}
Replaced multiple helpers for running code on a different thread with single helper.
Move logic under test and expectation from helpers to tests
Remove unique_ptr indirection where not used.
Replace rtc::TaskQueue with TaskQueueForTest to keep use convenient constructor
and to use convenient synchronous calls
Bug: webrtc:10284
Change-Id: I3f1ec2adc4dc16b44d35e769ef8b456407ee66c2
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/128903
Reviewed-by: Karl Wiberg <kwiberg@webrtc.org>
Commit-Queue: Danil Chapovalov <danilchap@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27280}
Plus all the annotations that were necessary to make things compile
again.
We needed a special twist for call_. The value it points to is owned
by the worker thread, but the signal thread needs to read the pointer.
We could have made the pointer const, except that we explicitly reset
it in the destructor (in an invoke to the worker thread).
Bug: webrtc:9987
Change-Id: I31f024547f4be0e50967133b0d452c80ae38d7ed
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/128863
Commit-Queue: Karl Wiberg <kwiberg@webrtc.org>
Reviewed-by: Steve Anton <steveanton@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27278}
VCMReceiveStatisticsCallback originates in the old jitter buffer, and
is no longer used.
VCMFrameTypeCallback originates in VideoReceiver::RequestKeyFrame,
which is called from OncomingPacket, Process, Decode(uint16_t
maxWaitTimeMs), all of which are unused by VideoReceiveStream.
So delete the code to wire them up via VideoStreamDecoder.
Bug: webrtc:7408
Change-Id: I173bc94eb32f2641f943c125083db038c3bcaeb1
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/128870
Reviewed-by: Philip Eliasson <philipel@webrtc.org>
Reviewed-by: Erik Språng <sprang@webrtc.org>
Commit-Queue: Niels Moller <nisse@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27277}
