internal block size of the AEC differ from the frame
size in the AEC output.
Before this CL, this buffering was done using ringbuffers
as well as secondary internal AEC buffers that were stored
on the state. The internal buffers were redundant, and the
ringbuffers were so short that the benefit of using
ringbuffers were lost.
This CL addresses the above issues by replacing the
ringbuffers by linear buffers. This has the main advantage
of cleaner code but it should significantly less
computational complex.
Furthermore, as the complexity of the function where the
conversion to external and internal AEC frame sizes is done
increased significantly with the changes in this CL, the
CL also include refactoring the near-end buffer handling
to increase readability and reduce code repetition.
After the changes in this CL it is very clear that the
former buffering of the output was incorrectly done for
the first frames. This CL corrects that but in doing that
it breaks the bitexactness with the former code.
The bitexactness is, however, only broken for the first
1000 samples and it has been verified that for a test suite
the CL maintains bitexactness in the AEC output
after the first 1000 samples.
This CL is chained to the CL https://codereview.webrtc.org/2311833002/ and will be
followed by more CLs that refactor the other buffers
inside the AEC.
BUG=webrtc:5298, webrtc:6018
Review-Url: https://codereview.webrtc.org/2321483002
Cr-Commit-Position: refs/heads/master@{#14184}
Reason for revert:
Breaks fuzzer compilation.
Original issue's description:
> Make rtcp parsing implementation private in RtcpReceiver:
> Function just for Parse and for Callbacks moved to private section.
> All handles moved from protected to private section.
>
> BUG=webrtc:5260
> R=sprang@webrtc.org
>
> Committed: https://crrev.com/faf708e238c7b43a732fbebf79ac9298b4b95a95
> Cr-Commit-Position: refs/heads/master@{#14181}
TBR=sprang@webrtc.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=webrtc:5260
Review-Url: https://codereview.webrtc.org/2332673003
Cr-Commit-Position: refs/heads/master@{#14182}
Function just for Parse and for Callbacks moved to private section.
All handles moved from protected to private section.
BUG=webrtc:5260
R=sprang@webrtc.org
Review URL: https://codereview.webrtc.org/2320603002 .
Cr-Commit-Position: refs/heads/master@{#14181}
Reason for revert:
Interface change in the mock breaks downstream code.
Original issue's description:
> The current scheme for setting parameters and specifying the behavior
> of the audio processing module is quite complex and hard to implement
> in a threadsafe and efficient manner. Therefore a new scheme for setting
> the parameters in the audio processing module is introduced in this CL.
>
> The idea is to roll this scheme out gradually and as a first functionality
> in the audio processing module where this is applied the level controller
> was chosen. This CL includes the replacement of the Config-based
> level controller scheme with the new scheme.
>
> BUG=webrtc:5298
>
> Committed: https://crrev.com/c8bbe3fe9aad9e9a1189a42dcaa8f5d6c261ecc8
> Cr-Commit-Position: refs/heads/master@{#14171}
TBR=solenberg@webrtc.org,henrik.lundin@webrtc.org,peah@webrtc.org
BUG=webrtc:5298
NOTRY=True
Review-Url: https://codereview.webrtc.org/2334583002
Cr-Commit-Position: refs/heads/master@{#14177}
functionalities doing sample-rate conversions:
-The implicit resampling done in the AudioBuffer CopyTo,
CopyFrom, InterleaveTo and DeinterleaveFrom methods.
-The multi-band splitting scheme.
The selection of rates in these have been difficult and
complicated, partly due to that the APM API which allows
for activating the APM submodules without notifying
the APM.
This CL adds functionality that for each capture frame
polls all submodules for whether they are active or not
and compares this against a cached result.
Furthermore, new functionality is added that based on the
results of the comparison do a reinitialization of the APM.
This has several advantages
-The code deciding on whether to analysis and synthesis is
needed for the bandsplitting can be much simplified and
centralized.
-The selection of the processing rate can be done such as
to avoid the implicit resampling that was in some cases
unnecessarily done.
-The optimization for whether an output copy is needed
that was done to improve performance due to the implicit
resampling is no longer needed, which simplifies the
code and makes it less error-prone in the sense that
is no longer neccessary to keep track of whether any
module has changed the signal.
Finally, it should be noted that the polling of the state
for all the submodules was done previously as well, but in
a less obvious and distributed manner.
BUG=webrtc:6181, webrtc:6220, webrtc:5298, webrtc:6296, webrtc:6298, webrtc:6297
Review-Url: https://codereview.webrtc.org/2304123002
Cr-Commit-Position: refs/heads/master@{#14175}
of the audio processing module is quite complex and hard to implement
in a threadsafe and efficient manner. Therefore a new scheme for setting
the parameters in the audio processing module is introduced in this CL.
The idea is to roll this scheme out gradually and as a first functionality
in the audio processing module where this is applied the level controller
was chosen. This CL includes the replacement of the Config-based
level controller scheme with the new scheme.
BUG=webrtc:5298
Review-Url: https://codereview.webrtc.org/2292863002
Cr-Commit-Position: refs/heads/master@{#14171}
0 means "pause", so forcing it to the min bitrate means we'll never
allow pausing for internal source encoders.
BUG=
Review-Url: https://codereview.webrtc.org/2304603002
Cr-Commit-Position: refs/heads/master@{#14168}
A left shift by 10 was assumed to never overflow, since "[s]imulation
of the 25 files shows that maximum value in the vector gain_lo_hiQ17[]
is 441344, which means that it is log2((2^31)/441344) = 12.2 shifting
bits from saturation." However, a fuzzer test succeeded in provoking
an overflow, which we ignore in this CL on the theory that only
"abnormal" inputs cause overflow.
Also had to replace a "foo << 1" with "foo * (1 << 1)" in
WEBRTC_SPL_MUL_16_32_RSFT15 because foo could be negative; this
problem showed up as soon as I'd asked UBSan to ignore the overflow
discussed above.
BUG=chromium:615819
Review-Url: https://codereview.webrtc.org/2314413002
Cr-Commit-Position: refs/heads/master@{#14162}
This changes added a simple measurement of levels "close to the audio hardware"
both for playout and for recording. These levels are logged once each 10 seconds.
It also adds WebRTC.Audio.RecordedOnlyZeros UMA stat and it is updated at
destuction. It will report true iff all reported recording leves are zero.
BUG=NONE
R=peah@webrtc.org
Review URL: https://codereview.webrtc.org/2328433003 .
Cr-Commit-Position: refs/heads/master@{#14160}
In order to be able to clear out any potentially stashed old frames from
the RtpFrameReferenceFinder we can now clear frames that contain packets
older than |seq_num|.
BUG=webrtc:5514
Review-Url: https://codereview.webrtc.org/2304723004
Cr-Commit-Position: refs/heads/master@{#14156}
the AEC. This solves the following issues:
-Even though the buffering was previously done using ringbuffers, those
were inefficiently used which caused a lot of hidden memcopys.
-The ringbuffers wasted a lot of space in the AEC state as they were too
long.
-The lowest and two upper bands were decoupled in the buffering, which
required extra code to handle.
-On top of the ringbuffers there was a second linear buffer that was
stored in the state which caused even more data to be stored on the
state.
-The incoming nearend frames were passed to the functions in the form
of buffers on the state, which made the code harder to read as it was
not immediately clear where the nearend signal was used, and when it
was modified.
The CL addresses this by replacing all the buffers by two linear buffers:
-One buffer before the AEC processing for producing nearend
blocks of size 64 that can be processed by the AEC.
-One inside the AEC processing that buffers the current
nearend block until the next block is processed.
The changes have been tested to be bitexact.
This CL will be followed by several other CLs, that refactor the other
buffers in the AEC.
BUG=webrtc:5298, webrtc:6018
Review-Url: https://codereview.webrtc.org/2311833002
Cr-Commit-Position: refs/heads/master@{#14141}
Reason for revert:
Broke all Chromium libFuzzer builds
https://bugs.chromium.org/p/chromium/issues/detail?id=645069
Original issue's description:
> Setting up an RTP input fuzzer for NetEq
>
> This CL introduces a new fuzzer target neteq_rtp_fuzzer that
> manipulates the RTP header fields before inserting the packets into
> NetEq. A few helper classes are also introduced.
>
> BUG=webrtc:5447
> NOTRY=True
>
> Committed: https://crrev.com/2d273f1e97cd5030ed1686f27ce1118291b66395
> Cr-Commit-Position: refs/heads/master@{#14103}
TBR=ivoc@webrtc.org,kjellander@webrtc.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=webrtc:5447
Review-Url: https://codereview.webrtc.org/2328483002
Cr-Commit-Position: refs/heads/master@{#14131}
I could not find a single place it was used, outside of the unittests
for the sync packet support itself.
Review-Url: https://codereview.webrtc.org/2309303002
Cr-Commit-Position: refs/heads/master@{#14130}
Reason for revert:
Downstream build is fixed.
Original issue's description:
> Revert of Ignore Camera and Flip bits in CVO when parsing video rotation (patchset #3 id:80001 of https://codereview.webrtc.org/2280703002/ )
>
> Reason for revert:
> Breaks downstream build.
>
> Original issue's description:
> > Ignore Camera and Flip bits in CVO when parsing video rotation
> >
> > Currently, if WebRTC receives a CVO byte where the Camera or Flip bit is
> > set, then rotation is incorrectly parsed as 0. This CL fixes that issue.
> > The Camera and Flip bit is still unimplemented and will just be ignored
> > though.
> >
> > BUG=webrtc:6120
> > R=danilchap@webrtc.org, pthatcher@webrtc.org, tommi@webrtc.org
> >
> > Committed: f9e1b922ef
>
> TBR=pthatcher@webrtc.org,danilchap@webrtc.org,tommi@webrtc.org
> # Skipping CQ checks because original CL landed less than 1 days ago.
> NOPRESUBMIT=true
> NOTREECHECKS=true
> NOTRY=true
> BUG=webrtc:6120
>
> Committed: https://crrev.com/97667c7746282704acccd896e26175decee349c0
> Cr-Commit-Position: refs/heads/master@{#14035}
TBR=pthatcher@webrtc.org,danilchap@webrtc.org,tommi@webrtc.org
# Not skipping CQ checks because original CL landed more than 1 days ago.
BUG=webrtc:6120
Review-Url: https://codereview.webrtc.org/2320913003
Cr-Commit-Position: refs/heads/master@{#14124}
With this CL, the NetEqReplacementInput class handles reordered and
missing packets in a better way than before, by storing the last
confirmed packet size and using that when the next packet size cannot
be calculated.
NOTRY=True
Review-Url: https://codereview.webrtc.org/2319553003
Cr-Commit-Position: refs/heads/master@{#14122}
1. Use of const in all variable declarations where it is possible
2. Variable names and function arguments changed from CamelCase to match code style
3. A few stale comments removed.
4. Chromium clang plugin check added (now possible thanks to kwiberg@'s work on common.h)
5. Disallow constructor macros added.
NOTRY=true
Review-Url: https://codereview.webrtc.org/2294263002
Cr-Commit-Position: refs/heads/master@{#14120}
Avoids reporting a growing delay (i.e. time b/w current time and oldest packet in the pacer).
BUG=webrtc:6253
Review-Url: https://codereview.webrtc.org/2279283002
Cr-Commit-Position: refs/heads/master@{#14118}
AudioNetworkAdaptor is supposed to facilitate AudioEncoder to adapt to varying network conditions.
This is the first of a sequence of CLs that are to add one implementation of AudioNetworkAdaptor.
This CL illustrates the interfaces of the AudioNetworkAdaptor.
BUG=webrtc:6303
Review-Url: https://codereview.webrtc.org/2308573002
Cr-Commit-Position: refs/heads/master@{#14115}
Reason for revert:
Resubmit capturer tests
Original issue's description:
> Revert of [WebRTC] A real ScreenCapturer test (patchset #8 id:240001 of https://codereview.webrtc.org/2268093002/ )
>
> Reason for revert:
> ScreenCapturerTest.CaptureUpdatedRegion fails on Win DrMemory Full.
>
> Original issue's description:
> > [WebRTC] A real ScreenCapturer test
> >
> > We do not have a real ScreenCapturer test before. And after CL 2210443002, a new
> > ScreenDrawer interface is added to the code base to draw various shapes on the
> > screen. This change is to use ScreenDrawer to test ScreenCapturer. Besides test
> > cases, some other changes are included,
> >
> > 1. A WaitForPendingPaintings() function in ScreenDrawer, to wait for a
> > ScreenDrawer to finish all the pending draws. This function now only sleeps 50
> > milliseconds on X11 and 100 milliseconds on Windows.
> >
> > 2. A Color structure to help handle a big-endian or little-endian safe color and
> > provide functions to compare with DesktopFrame::data(). Both ScreenDrawer and
> > DesktopFrameGenerator (in change 2202443002) can use this class to create colors
> > and compare with or paint to a DesktopFrame.
> >
> > 3. ScreenDrawer now uses Color structure instead of uint32_t.
> >
> > BUG=314516
> >
> > TBR=kjellander@chromium.org
> >
> > Committed: https://crrev.com/9d1c54ace0dc9f68da0152aa1ded2a8dba0a43ae
> > Cr-Commit-Position: refs/heads/master@{#14058}
>
> TBR=sergeyu@chromium.org,jamiewalch@chromium.org,kjellander@chromium.org,zijiehe@chromium.org
> # Not skipping CQ checks because original CL landed more than 1 days ago.
> BUG=314516
>
> Committed: https://crrev.com/4c44202dc348613695a4b529bbd7c9bdab6195ec
> Cr-Commit-Position: refs/heads/master@{#14071}
TBR=sergeyu@chromium.org,jamiewalch@chromium.org,kjellander@chromium.org,asapersson@webrtc.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=644130
Review-Url: https://codereview.webrtc.org/2313653003
Cr-Commit-Position: refs/heads/master@{#14113}
Methods are named more consistently and have a more consistent
signatures. The call structure of mixing is slightly
simplified. Anonymous participants are also ramped up.
NOTRY=True
Review-Url: https://codereview.webrtc.org/2298163002
Cr-Commit-Position: refs/heads/master@{#14110}
This file defines webrtc::Config which was mostly used by modules/audio_processing. The files webrtc/common.h, webrtc/common.cc and webrtc/test/common_unittests.cc are moved to modules/audio_processing and the few remaining uses of webrtc::Config are replaced with simpler code.
- For NetEq and pacing configuration, a VoEBase::ChannelConfig is passed to VoEBase::CreateChannel().
- Removes the need for VoiceEngine::Create(const Config& config). No need to store the webrtc::Config in VoE shared state.
BUG=webrtc:5879
Review-Url: https://codereview.webrtc.org/2307533004
Cr-Commit-Position: refs/heads/master@{#14109}
This CL introduces a new fuzzer target neteq_rtp_fuzzer that
manipulates the RTP header fields before inserting the packets into
NetEq. A few helper classes are also introduced.
BUG=webrtc:5447
NOTRY=True
Review-Url: https://codereview.webrtc.org/2315633002
Cr-Commit-Position: refs/heads/master@{#14103}
We hit a fuzzer bug that caused numDecodedBytesLB + numDecodedBytesUB
> lenEncodedBytes, which is obviously bogus. Check for that, and for
the case whhere the UB decoder itself realized that something was
wrong. (The code already makes the corresponding check for the LB
decoder.)
BUG=chromium:637899
Review-Url: https://codereview.webrtc.org/2315693002
Cr-Commit-Position: refs/heads/master@{#14091}
RembStatus moved to RtcpSender unittest where it fits better
Creating remb in Compound/ReducedSize modes already covered by RtcpSender unittests.
Parsing remb already covered by RtcpReceiverTest.ReceivesRemb
R=stefan@webrtc.org
Review URL: https://codereview.webrtc.org/2312853002 .
Cr-Commit-Position: refs/heads/master@{#14088}
Changed from 10 to 68.
This is to avoid a flake where the limit is exceeded, see
crbug.com/638554. Our performance tests should flag performance
regressions, we shouldn't rely on crashing because the number of
referenced buffers is not tiny to detect this. However, if a really big
number of buffers (>68) are referenced without being dereferenced it is
likely that we have a bug and frames are leaking in which case we can
DCHECK-crash.
BUG=chromium:638554
Review-Url: https://codereview.webrtc.org/2280593002
Cr-Commit-Position: refs/heads/master@{#14084}
If neteq_rtpplay is invoked with the --ssrc option to select packets
matching a specific SSRC, but no matching packets are found, this CL
provides a meaningful error message.
BUG=webrtc:2692
NOTRY=True
TBR=ivoc@webrtc.org
Review-Url: https://codereview.webrtc.org/2318503002
Cr-Commit-Position: refs/heads/master@{#14083}
DirectX capturer won't be able to update the DesktopFrame for the second
ScreenCapturerWinDirectx instance, if AcquireNextFrame() returns timeout or
unchanged error. Current solution uses |last_frame| of the second
ScreenCapturerWinDirectx instance, which also does not contain the updated frame
between this and last AcquireNextFrame() calls. Considering following situation,
(C1: capturer 1, C2: capturer 2, update: screen updated, next AcquireNextFrame()
call will return a new frame, Fx: a frame x)
update -> C2.capture returns F1 -> update -> C1.capture returns F2 ->
C2.capture unchanged
So using F1 to update the last capture frame is not correct, we need to use F2.
Refer to design doc https://goo.gl/hU1ifG for a detail description.
The change also makes DxgiDuplicatorController work with 2+ DesktopFrame queue.
Now TwoDirectxCapturers test can pass.
BUG=314516
Review-Url: https://codereview.webrtc.org/2299663003
Cr-Commit-Position: refs/heads/master@{#14077}
dirty region than the real screen change. A similar behavior may happen on other
platforms with damage notification support. So it's better to have an individual
layer to handle the Differ logic, and remove capturing independent logic out of
each ScreenCapturer* implementation.
So this change does following things,
1. Update differ_block to handle variable height. differ_block_sse2 has been
renamed to differ_vector_sse2.
2. A new ScreenCapturerDifferWrapper implementation to help set
DesktopFrame::updated_region(). It uses an underlying ScreenCapturer to do
the real capture work, and updates the updated region of DesktopFrame returned
from OnCaptureResult function.
3. FakeDesktopCapturer and FakeScreenCapturer to generate controllable
DesktopFrame by using DesktopFrameGenerator and DesktopFramePainter.
4. Test ScreenCapturerDifferWrapper by using FakeScreenCapturer.
After this change, we can eventually remove all Differ logic from
ScreenCapturer* implementations, and fix a potential crash bug in
ScreenCapturerLinux class. It wrongly assumes previous_frame() has a same size
as current_frame(). https://goo.gl/3nSqOC
BUG=633802
TBR=kjellander@webrtc.org
Review-Url: https://codereview.webrtc.org/2202443002
Cr-Commit-Position: refs/heads/master@{#14076}
IncomingPacket(const uint8_t*, size_t) is used as entry point instead
of IncomingRTCPPacket(PacketInformation* out, RtcpParser* in);
Result is validated by checking which callbacks were called instead
of checking intermediate structure PacketInformaion.
This allows to switch parsing implementation.
BUG=webrtc:5260
Review-Url: https://codereview.webrtc.org/2292093002
Cr-Commit-Position: refs/heads/master@{#14074}
this eliminates reparsing of rtp packet on send audio path
BUG=webrtc:5261
Review-Url: https://codereview.webrtc.org/2292883002
Cr-Commit-Position: refs/heads/master@{#14072}
