With deferred packet sequencing, the PacketSequencer instance is called
directly from the RtpRtcp module while before it was called from within
the RTPSender while holding a lock.
Since sequence number assignment happens on the same thread as actual
packet sending, though thought was that locking was no longer needed.
Unfortunately, SetRtpState()/GetRtpState() also exists - and while they
should only be called on creating/destruction there is a possible race
where a delayed packet from the pacer accesses the sequencer while
GetRtpState() is being called.
For now, this CL just adds a lock to guard sequencer. Follow-ups will
make sure get/set state is never called while module is attached to
the packet router. After that, the lock can be removed again.
Bug: webrtc:11340, webrtc:12470
Change-Id: I123c762fb4afd20b3a6bd03b86234eb9ec34a209
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/228430
Reviewed-by: Danil Chapovalov <danilchap@webrtc.org>
Commit-Queue: Erik Språng <sprang@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#34723}
This is a reland of 320c57b7c6bf68d8612a4f135f44b3d29e802113
Original change's description:
> red: remove special-casing of no-redundancy
>
> removes the special-casing of not sending a RED header when there is no redundant payload.
> This avoids switching back and forth between the primary and the red payload format (primarily at the start of the connection).
>
> BUG=webrtc:11640
>
> Change-Id: I8e0044bef1ed7c4168d9527645522392db2ed068
> Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/220932
> Reviewed-by: Jesus de Vicente Pena <devicentepena@webrtc.org>
> Reviewed-by: Henrik Lundin <henrik.lundin@webrtc.org>
> Commit-Queue: Henrik Lundin <henrik.lundin@webrtc.org>
> Cr-Commit-Position: refs/heads/master@{#34703}
Bug: webrtc:11640
Change-Id: I5e5687be575183ee16d74df4a8170e4fedad739f
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/228422
Commit-Queue: Henrik Lundin <henrik.lundin@webrtc.org>
Reviewed-by: Henrik Lundin <henrik.lundin@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#34713}
This reverts commit 320c57b7c6bf68d8612a4f135f44b3d29e802113.
Reason for revert:
Breaks CI tests: https://ci.chromium.org/ui/p/webrtc/builders/ci/Linux32%20Release/27236/overview
All CI tests: https://ci.chromium.org/p/webrtc/g/ci/console
Error is of the following type:
```
../../modules/audio_coding/codecs/red/audio_encoder_copy_red_unittest.cc:195: Failure
Expected equality of these values:
1u
Which is: 1
encoded_info_.encoded_bytes
Which is: 2
Stack trace:
0x56b298d9: webrtc::AudioEncoderCopyRedTest_CheckPayloadSizesSingle_Test::TestBody()
0x572fe317: testing::internal::HandleExceptionsInMethodIfSupported<>()
0x572fe1d4: testing::Test::Run()
0x572ff2ee: testing::TestInfo::Run()
```
Original change's description:
> red: remove special-casing of no-redundancy
>
> removes the special-casing of not sending a RED header when there is no redundant payload.
> This avoids switching back and forth between the primary and the red payload format (primarily at the start of the connection).
>
> BUG=webrtc:11640
>
> Change-Id: I8e0044bef1ed7c4168d9527645522392db2ed068
> Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/220932
> Reviewed-by: Jesus de Vicente Pena <devicentepena@webrtc.org>
> Reviewed-by: Henrik Lundin <henrik.lundin@webrtc.org>
> Commit-Queue: Henrik Lundin <henrik.lundin@webrtc.org>
> Cr-Commit-Position: refs/heads/master@{#34703}
TBR=henrik.lundin@webrtc.org,devicentepena@webrtc.org,minyue@webrtc.org,philipp.hancke@googlemail.com,webrtc-scoped@luci-project-accounts.iam.gserviceaccount.com
Change-Id: Ide409232720df32b24022f99228f3b6ae81f06fe
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: webrtc:11640
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/228421
Reviewed-by: Andrey Logvin <landrey@webrtc.org>
Commit-Queue: Andrey Logvin <landrey@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#34707}
When pacing is enabled for the low latency rendering path,
frames are sent to the decoder in regular intervals. In case of a
jitter, these frames intervals could add up to create a large latency.
Hence, disable frame pacing if the pre-decode queue grows beyond the
threshold. The threshold for when to disable frame pacing is set
through a field trial. The default value is high enough so that
the behavior is not changed unless the field trial is specified.
Bug: chromium:1237402
Change-Id: I901fd579f68da286eca3d654118f60d3c55e21ce
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/228241
Reviewed-by: Ilya Nikolaevskiy <ilnik@webrtc.org>
Commit-Queue: Johannes Kron <kron@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#34705}
removes the special-casing of not sending a RED header when there is no redundant payload.
This avoids switching back and forth between the primary and the red payload format (primarily at the start of the connection).
BUG=webrtc:11640
Change-Id: I8e0044bef1ed7c4168d9527645522392db2ed068
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/220932
Reviewed-by: Jesus de Vicente Pena <devicentepena@webrtc.org>
Reviewed-by: Henrik Lundin <henrik.lundin@webrtc.org>
Commit-Queue: Henrik Lundin <henrik.lundin@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#34703}
fixes an incorrect redundancy shift and add tests that would have caught this bug.
BUG=webrtc:11640
Change-Id: I6fe2fb21587fffc5fee4d403ac898e12d525a1cd
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/224120
Reviewed-by: Henrik Lundin <henrik.lundin@webrtc.org>
Commit-Queue: Henrik Lundin <henrik.lundin@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#34702}
Denormal numbers (see [1]) may origin in APM when the input is zeroed
after a non-zero signal. In extreme cases, instructions involving
denormal operands may run as much as 100 times slower, which seems to
be the case (to some extent) of crbug.com/1227566.
This CL adds a class that disables denormals only via hardware on x86
and on ARM. The class is used in APM and it is an adaption of [2].
Tested: appr.tc call on Chromium (Win, Mac)
[1] https://en.wikipedia.org/wiki/Denormal_number
[2] https://source.chromium.org/chromium/chromium/src/+/main:third_party/blink/renderer/platform/audio/denormal_disabler.h
Fixed: chromium:1227566
Change-Id: I0ed2eab55dc597529f09f93c26c7a01de051fdbe
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/227768
Reviewed-by: Magnus Flodman <mflodman@webrtc.org>
Reviewed-by: Per Åhgren <peah@webrtc.org>
Commit-Queue: Alessio Bazzica <alessiob@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#34701}
Caches the TCP fairness limit to avoid redundant calculation. Adds option to append the delay based estimate as a candidate. Makes the appending of acknowledged bitrate as a candidate optional. Adds a log-bandwidth bias term.
(submit on behalf of crodbro)
Bug: webrtc:12707
Change-Id: Ic4b0f58e6f0bc3b117fe78a2321a07db65afd9dd
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/228163
Commit-Queue: Ying Wang <yinwa@webrtc.org>
Reviewed-by: Christoffer Rodbro <crodbro@webrtc.org>
Reviewed-by: Danil Chapovalov <danilchap@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#34687}
Remove private members that are no longer used or always have same value
Use less allocations
Bug: None
Change-Id: I5430c2356f0039212baf8b248b92775e8852ce1b
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/227765
Reviewed-by: Erik Språng <sprang@webrtc.org>
Commit-Queue: Danil Chapovalov <danilchap@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#34665}
With this turned on, packets will be sequence number after the pacing
stage rather that during packetization.
This avoids a race where packets may be sent out of order, and paves
the way for the ability to cull packets from the pacer queue without
causing sequence number gaps.
For now, the feature is off by default. Follow-ups will enable it for
video and audio separately.
Bug: webrtc:11340, webrtc:12470
Change-Id: I6d411d8c85b9047e3e9b05ff4c2c3ed97c579aa1
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/208584
Commit-Queue: Erik Språng <sprang@webrtc.org>
Reviewed-by: Danil Chapovalov <danilchap@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#34661}
Schedule the frames to be decoded based on the pacing delay from the
last decode scheduled time. In the current implementation, multiple
threads and different functions in same thread can call
MaxWaitingTime(), thereby increasing the wait time each time the
function is called. Instead of returning the wait time for a future
frame based on the number of times the function is called, return the
wait time only for the next frame to be decoded. Threads can call the
function repeatedly to check the waiting time for next frame and wake
up and then go back to waiting if an encoded frame is not available.
Change-Id: I00886c1619599f94bde5d5eb87405572e435bd73
Bug: chromium:1237402
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/226502
Reviewed-by: Johannes Kron <kron@webrtc.org>
Reviewed-by: Philip Eliasson <philipel@webrtc.org>
Commit-Queue: Johannes Kron <kron@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#34660}
The race can happen when an encoder thread is packetizing a video frame
and is calling RTPSender::AssignSequenceNumber() while the RtpRtcp
module is calling GeneratePadding() and querying
PacketSequencer::CanSendPaddingOnMediaSsrc().
The solution for now is to simply not call
PacketSequencer::CanSendPaddingOnMediaSsrc() from the RtpRtcp module,
as that parameter will be ignored anyway - RTPSender will query that
method internally while holding the send lock.
Once deferred sequencing is implemented, the
can_send_padding_on_media_ssrc parameter can be populated safely since
it is then always called on the pacer thread.
Bug: webrtc:11340, webrtc:12470
Change-Id: I9e90808166453d0e29746df89044e1d3bdffa286
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/227767
Reviewed-by: Tommi <tommi@webrtc.org>
Commit-Queue: Erik Språng <sprang@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#34655}
This prepares for deferred sequence numbering, and is (sort of)
extracted from
https://webrtc-review.googlesource.com/c/src/+/208584
Bug: webrtc:11340, webrtc:12470
Change-Id: I2f3695309e1591b9f7a1ee98556f4f0758de7f69
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/227352
Commit-Queue: Erik Språng <sprang@webrtc.org>
Reviewed-by: Danil Chapovalov <danilchap@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#34643}
This CL is extracted from
https://webrtc-review.googlesource.com/c/src/+/208584
PacketSequencer now has its own unit tests. They are maybe somewhat
redundant with a few RtpSender unit tests, but will defer cleanup to
a later CL.
Bug: webrtc:11340, webrtc:12470
Change-Id: I1c31004b85ae075ddc696bdf1100d2a5044d4ef5
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/227343
Commit-Queue: Erik Språng <sprang@webrtc.org>
Reviewed-by: Danil Chapovalov <danilchap@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#34638}
fill the audio level of the recovery packets from the main packet.
While not exact, this should be close enough. Without this,
the audio level in getStats() will be filled but the audio level
in getSynchronizationSources() will be empty.
In chrome this is easy to test, the audio level graph on
https://webrtc.github.io/samples/src/content/peerconnection/audio/
will be empty all the time prior to this fix.
BUG=webrtc:11640
Change-Id: Ia1e61fd1852445239021a76d08032120a92d83aa
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/226840
Reviewed-by: Henrik Lundin <henrik.lundin@webrtc.org>
Commit-Queue: Henrik Lundin <henrik.lundin@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#34635}
