Reason for revert:
Makes perf and Chromium FYI bots unhappy.
Original issue's description:
> WebRtcVoiceMediaChannel::AddRecvStream: Don't call SetRecPayloadType
>
> This removes one more place where we were unable to handle codecs not
> in the built-in set.
>
> BUG=webrtc:5805
>
> Review-Url: https://codereview.webrtc.org/2686043006
> Cr-Commit-Position: refs/heads/master@{#17370}
> Committed: 1724cfbdbaTBR=ossu@webrtc.org,solenberg@webrtc.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=webrtc:5805
Review-Url: https://codereview.webrtc.org/2772043002
Cr-Commit-Position: refs/heads/master@{#17374}
This removes one more place where we were unable to handle codecs not
in the built-in set.
BUG=webrtc:5805
Review-Url: https://codereview.webrtc.org/2686043006
Cr-Commit-Position: refs/heads/master@{#17370}
different sample rate frequency.
BUG=webrtc:7327
Problems before the fix:
1. NetEqImpl::timestamp_ is inconsistent. Initially it is set to
the original RTP timestamp, but later gets updated with the
scaled timestamp.
2. NetEqImpl::InsertPacketInternal::main_timestamp is set with
the original RTP timestamp, but later gets compared with the
NetEqImpl::timestamp_ which may or may not be with the same
sample rate frequency and this results in major problems.
3. IncreaseEndTimestamp(main_timestamp - timestamp_) will be
incorrect when SSRC is changed and not the first packet.
4. delay_manager_->Update() may not be always invoked, since
the (main_timestamp - timestamp_) >= 0 will not be true when
the previous scaled timestamp_ is bigger than the main_timestamp
(current RTP timestamp) even if the current RTP timestamp is
bigger than the previous RTP timestamp.
5. delay_manager_->Update() parameters are main_timestamp
which increments with the RTP sample rate frequency and the
fs_hz_ which is the decoder sample rate frequency. When these
two frequencies are different as is the case with g.722, the
DelayManager::Update() will misfire and calculate incorrect
packet_len_ms and inter-arrival time (IAT) as a result. This
in effect will cause neteq to enter kPreemptiveExpand operation
and will keep expanding the jitter buffer even if the RTP packets
arrive with no jitter at all.
The fix corrects all these problems by making sure the
main_timestamp and the timestamp_ are always set with the scaled
timestamp and increment with the decoder sample rate frequency.
Review-Url: https://codereview.webrtc.org/2743063005
Cr-Commit-Position: refs/heads/master@{#17232}
In short, what I did was to
* Remove acm_common_defs.h (the stuff in it was used only by
acm_codec_database.cc).
* Move audio_coding_module_typedefs.h to a new build target.
* Move the NetEqDecoder enum (and the associated
NetEqDecoderToSdpAudioFormat function) to a new file in a new
build target.
BUG=webrtc:7243, webrtc:7244
Review-Url: https://codereview.webrtc.org/2723253005
Cr-Commit-Position: refs/heads/master@{#17005}
It's the faster, less strict cousin of checked_cast.
BUG=none
Review-Url: https://codereview.webrtc.org/2714063002
Cr-Commit-Position: refs/heads/master@{#16958}
The paths of the protobuf output files needs to match the actual tree path
in order for Bazel builds to work.
BUG=webrtc:6412
NOTRY=True
Review-Url: https://codereview.webrtc.org/2716413004
Cr-Commit-Position: refs/heads/master@{#16912}
This is a follow-up to https://codereview.webrtc.org/2670643007/. That
CL provided significant improvement to Mac, Linux and ARM-based
platforms, but failed to improve the performance for Windows. The
problem is that the MSVC compiler did not produce branch-free code for
that fix. This new change produces the same result for non-Windows
platforms, as well as introduces branch-free code for Windows.
H/t to kwiberg@ for providing the solution.
BUG=webrtc:7159
Review-Url: https://codereview.webrtc.org/2700633003
Cr-Commit-Position: refs/heads/master@{#16649}
After https://codereview.webrtc.org/2340773002,
the path from webrtc::test::ResourcePath in
webrtc/modules/audio_coding/neteq/tools/neteq_performance_test.cc is wrong.
It is
/path/to/repos/resources/audio_coding/testfile32kHz.pcm
It should be
/path/to/repos/webrtc-temp/src/resources/audio_coding/testfile32kHz.pcm.
The middle part is missing.
The reason this target is affected is because
webrtc::test::SetExecutablePath(argv[0]);
was not called.
That call is necessary for us to know that the test is being run from src/
and not from out/Default (as is assumed, when that function is not called.)
BUG=chromium:497757
R=kjellander@webrtc.org, henrik.lundin@webrtc.org
Review-Url: https://codereview.webrtc.org/2698743002
Cr-Commit-Position: refs/heads/master@{#16641}
As it is, the test fails to compile on some downstream compilers with the following error:
webrtc/modules/audio_coding/neteq/neteq_impl_unittest.cc:316:25: error: lambda capture 'kPayloadLength' is not required to be captured for this use [-Werror,-Wunused-lambda-capture]
.WillOnce(Invoke([kPayloadLength, kFirstSequenceNumber, kFirstTimestamp,
^
webrtc/modules/audio_coding/neteq/neteq_impl_unittest.cc:316:41: error: lambda capture 'kFirstSequenceNumber' is not required to be captured for this use [-Werror,-Wunused-lambda-capture]
.WillOnce(Invoke([kPayloadLength, kFirstSequenceNumber, kFirstTimestamp,
^
webrtc/modules/audio_coding/neteq/neteq_impl_unittest.cc:316:63: error: lambda capture 'kFirstTimestamp' is not required to be captured for this use [-Werror,-Wunused-lambda-capture]
.WillOnce(Invoke([kPayloadLength, kFirstSequenceNumber, kFirstTimestamp,
^
webrtc/modules/audio_coding/neteq/neteq_impl_unittest.cc:317:25: error: lambda capture 'kFirstReceiveTime' is not required to be captured for this use [-Werror,-Wunused-lambda-capture]
kFirstReceiveTime](const SdpAudioFormat& format,
BUG=webrtc:7107
Review-Url: https://codereview.webrtc.org/2672823002
Cr-Commit-Position: refs/heads/master@{#16422}
Left shifting negative integers is undefined behavior, and should be prevented. This CL fixes one such instance in the NetEq Expand function.
BUG=chromium:677106
Review-Url: https://codereview.webrtc.org/2616363003
Cr-Commit-Position: refs/heads/master@{#15966}
The file was aldready pruned down to the point where it only included
webrtc/typedefs.h. Therefore, all includes of
voice_engine_configurations.h are replaced with typedefs.h, except on
two occasions where it was obvously not needed.
BUG=webrtc:6506
Review-Url: https://codereview.webrtc.org/2553583002
Cr-Commit-Position: refs/heads/master@{#15547}
There's no longer any need to make the two arguments have the same
signedness, so we can remove a bunch of superfluous (and sometimes
dangerous) casts.
It turned out I also had to fix the safe_cmp functions to properly handle
enums that are implicitly convertible to integers.
NOPRESUBMIT=true
BUG=webrtc:6645
Review-Url: https://codereview.webrtc.org/2534683002
Cr-Commit-Position: refs/heads/master@{#15281}
There's no longer any need to make the two arguments have the same
signedness, so we can drop the "u" suffix on literal integer
arguments.
NOPRESUBMIT=true
BUG=webrtc:6645
Review-Url: https://codereview.webrtc.org/2535593002
Cr-Commit-Position: refs/heads/master@{#15280}
This is undefined behavior, according to specification.
BUG=chromium:661133
Review-Url: https://codereview.webrtc.org/2500953003
Cr-Commit-Position: refs/heads/master@{#15078}
This can happen in rare and strange cases.
Also taking the opportunity to replace all asserts with DCHECKs in
that file.
BUG=chromium:659225
Review-Url: https://codereview.webrtc.org/2499013002
Cr-Commit-Position: refs/heads/master@{#15070}
With this change, instead of
RTC_DCHECK_GE(unsigned_var, 17u);
we can simply write
RTC_DCHECK_GE(unsigned_var, 17);
or even
RTC_DCHECK_GE(unsigned_var, -17); // Always true.
and the mathematically sensible thing will happen.
Perhaps more importantly, we can replace checks like
// index is size_t, num_channels is int.
RTC_DCHECK(num_channels >= 0
&& index < static_cast<size_t>(num_channels));
or, even worse, just
// Surely num_channels isn't negative. That would be absurd!
RTC_DCHECK_LT(index, static_cast<size_t>(num_channels));
with simply
RTC_DCHECK_LT(index, num_channels);
In short, you no longer have to keep track of the signedness of the arguments, because the sensible thing will happen.
BUG=webrtc:6645
Review-Url: https://codereview.webrtc.org/2459793002
Cr-Commit-Position: refs/heads/master@{#14878}
The timestamps are 32 bit and can (conceivably) be spaced far enough
apart for the calculation, which is done in Q4, to overflow.
BUG=chromium:653268
Review-Url: https://codereview.webrtc.org/2460393002
Cr-Commit-Position: refs/heads/master@{#14856}
PacketList is now list<Packet> instead of list<Packet*>.
Splicing the lists in NetEqImpl::InsertPacketInternal instead of
moving packets. Avoid moving the packet when doing Rfc3389Cng.
Removed PacketBuffer::DeleteFirstPacket and DeleteAllPackets.
BUG=chromium:657300
Review-Url: https://codereview.webrtc.org/2425223002
Cr-Commit-Position: refs/heads/master@{#14747}
Only three items in the (rather large) header were actually used after
InsertPacket: payloadType, timestamp and sequenceNumber. They are now
put directly into Packet. This saves 129 bytes per Packet that no
longer need to be allocated and deallocated.
This also works towards decoupling NetEq from RTP. As part of that,
I've moved the NACK code earlier in InsertPacketInternal, together
with other things that directly reference the RTPHeader.
BUG=webrtc:6549
Review-Url: https://codereview.webrtc.org/2411183003
Cr-Commit-Position: refs/heads/master@{#14658}
With this change, the calculations inside AverageIAT are changed to be
in double-precision floating point instead of in fixed point. Also,
the method's name is changed to EstimatedClockDriftPpm to better
reflect what it returns.
A few unit tests had to be updated because of minor numerical
differences.
Also removing the UBSan suppression related to this issue.
BUG=webrtc:5889
Review-Url: https://codereview.webrtc.org/2408653002
Cr-Commit-Position: refs/heads/master@{#14628}
This was an ill tested special case which turned out to be more problem
than benefit. The special case was only triggered when the decoder frame
size was smaller than 10 ms, which is more or less unsupported by NetEq.
Also fixed a bug in a test, a bug which was exposed by the code change.
BUG=chromium:654983
Review-Url: https://codereview.webrtc.org/2412883002
Cr-Commit-Position: refs/heads/master@{#14627}
This CL deletes the old and not used video defines in
engine_configurations.h and pre-pends voice_ to indicate there are only
voice/audio defines left in the file.
BUG=none
R=solenberg@webrtc.org
Review URL: https://codereview.webrtc.org/2401673002 .
Cr-Commit-Position: refs/heads/master@{#14558}
The original CL (https://codereview.webrtc.org/2315633002) was
reverted since the fuzzer depended on gflags and files in the
resources folder; neither of this is allowed for a fuzzer test in
Chromium. This new version streamlines the dependencies, and changes
the test to generate a sinusoid input audio signal instead of reading
from a file.
Original commit message:
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
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.win:win_chromium_rel_ng;master.tryserver.chromium.android:android_compile_dbg,linux_android_rel_ng;master.tryserver.chromium.linux:linux_chromium_rel_ng;master.tryserver.chromium.mac:mac_chromium_rel_ng,ios-device
Review-Url: https://codereview.webrtc.org/2384423002
Cr-Commit-Position: refs/heads/master@{#14523}
