In order to reduce contention, this CL avoids taking locks per packet
and prepares for forwarding all packets for a frame in one call, rather
than one at a time. This will especially reduce contention in the paced
sender during very high packet rates.
Bug: webrtc:10809
Change-Id: Ifc5fe3759b76a2a45f418b69d29c329e876f96d0
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/154358
Reviewed-by: Ilya Nikolaevskiy <ilnik@webrtc.org>
Commit-Queue: Erik Språng <sprang@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#29323}
Passing an empty arg is working at the moment but it is not
guaranteed to continue to work in the future.
This CL has been generated with:
git grep -l "INSTANTIATE_TEST_SUITE_P(," | xargs sed -i \
"s/INSTANTIATE_TEST_SUITE_P(,/INSTANTIATE_TEST_SUITE_P(All,/g"
Bug: None
Change-Id: Icd2fb9d9d29aed5d692a234124bd990d0f097db4
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/153890
Reviewed-by: Karl Wiberg <kwiberg@webrtc.org>
Commit-Queue: Mirko Bonadei <mbonadei@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#29282}
This is a reland of 9e380fd484db09c37323b90a19c5ce7965927975
Patchset 1 is the original CL. The follow-ups adds fix for a test failure
and test for that change.
Original change's description:
> Improve performance of RtpPacketHistory
>
> The data structures in RtpPacketHistory were chosen based on assumption
> of few packets with possible sparse segments due to missing acking.
> In practice high bitrate usages with full histories seem to be more of
> a problem.
> Due to that, change storage from an std::map to an std::deque and live
> with potential segments of nullptr. Also limit size of padding prio
> set so that doesn't become a bottleneck.
>
> Bug: webrtc:8975
> Change-Id: I3b6314fb3255937d25362ff2cd906efb7b1397f7
> Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/145901
> Commit-Queue: Erik Språng <sprang@webrtc.org>
> Reviewed-by: Danil Chapovalov <danilchap@webrtc.org>
> Cr-Commit-Position: refs/heads/master@{#29117}
Bug: webrtc:8975
Change-Id: I5038e5ad2eb79ce75710d2d8b0b3ac01dd41c013
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/152282
Commit-Queue: Erik Språng <sprang@webrtc.org>
Reviewed-by: Danil Chapovalov <danilchap@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#29152}
This reverts commit 9e380fd484db09c37323b90a19c5ce7965927975.
Reason for revert: breaking downstream projects
Original change's description:
> Improve performance of RtpPacketHistory
>
> The data structures in RtpPacketHistory were chosen based on assumption
> of few packets with possible sparse segments due to missing acking.
> In practice high bitrate usages with full histories seem to be more of
> a problem.
> Due to that, change storage from an std::map to an std::deque and live
> with potential segments of nullptr. Also limit size of padding prio
> set so that doesn't become a bottleneck.
>
> Bug: webrtc:8975
> Change-Id: I3b6314fb3255937d25362ff2cd906efb7b1397f7
> Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/145901
> Commit-Queue: Erik Språng <sprang@webrtc.org>
> Reviewed-by: Danil Chapovalov <danilchap@webrtc.org>
> Cr-Commit-Position: refs/heads/master@{#29117}
TBR=danilchap@webrtc.org,sprang@webrtc.org
Change-Id: I5d5b74a6f4d60588e01a52dafe33e26deb9bdf77
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: webrtc:8975
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/152220
Reviewed-by: Qingsi Wang <qingsi@webrtc.org>
Commit-Queue: Qingsi Wang <qingsi@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#29121}
The data structures in RtpPacketHistory were chosen based on assumption
of few packets with possible sparse segments due to missing acking.
In practice high bitrate usages with full histories seem to be more of
a problem.
Due to that, change storage from an std::map to an std::deque and live
with potential segments of nullptr. Also limit size of padding prio
set so that doesn't become a bottleneck.
Bug: webrtc:8975
Change-Id: I3b6314fb3255937d25362ff2cd906efb7b1397f7
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/145901
Commit-Queue: Erik Språng <sprang@webrtc.org>
Reviewed-by: Danil Chapovalov <danilchap@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#29117}
Reland with fixes for fuzzer found crashes.
This refactoring helps to reduce unnecessary memcpy calls on the receive side.
This CL replaces |uint8 data[IP_PACKET_SIZE]| with |rtc::CopyOnWriteBuffer data| in Packet class, removes |length| field there, and does necessary changes.
Original Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/145332
Bug: webrtc:10750
Change-Id: I6775a701bcb2ae25ec1666e1db90041cd49013b7
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/151131
Reviewed-by: Rasmus Brandt <brandtr@webrtc.org>
Reviewed-by: Stefan Holmer <stefan@webrtc.org>
Reviewed-by: Ilya Nikolaevskiy <ilnik@webrtc.org>
Commit-Queue: Ilya Nikolaevskiy <ilnik@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#29116}
Update the |cumulative_lost_| counter per received packet. The rules
follow from RFC 3550 and are fairly simple: Decrement the counter by
one for every received packet. For every in-order packet, i.e., increasing
|received_seq_max_|, add that change to |cumulative_lost_|.
Net change is zero as long as packets are received in proper sequence.
This way, GetStats() always returns an up-to-date value, independent
of the timing of RTCP report blocks.
For RTCP reports, keep a workaround to never report negative cumulative loss.
Bug: webrtc:10679
Change-Id: I47ff3bf266ff2382f405ec9828d34f7fad7068b4
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/150641
Commit-Queue: Niels Moller <nisse@webrtc.org>
Reviewed-by: Danil Chapovalov <danilchap@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#29058}
These fixes are automatically created by various analysis tools, but have been manually triggered to be applied.
* the 'empty' method should be used to check for emptiness instead of 'size' (3 times)
* using decl 'Return' is unused (4 times)
* using decl '_' is unused (3 times)
* using decl 'DoAll' is unused (2 times)
* using decl 'SetArgPointee' is unused
* using decl 'Dlrr' is unused
* using decl 'IsEmpty' is unused
* redundant get() call on smart pointer
* using decl 'Invoke' is unused (2 times)
* using decl 'SizeIs' is unused (3 times)
* using decl 'make_tuple' is unused
* using decl 'NiceMock' is unused
* using decl 'SaveArg' is unused (2 times)
* using decl 'AtLeast' is unused
* using decl 'ElementsAre' is unused
* using decl 'Gt' is unused
Bug: None
Change-Id: I97658fb0e94620b8319d7c3da29b15e27ec23188
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/151133
Reviewed-by: Niels Moller <nisse@webrtc.org>
Commit-Queue: Danil Chapovalov <danilchap@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#29056}
This reverts commit eec5fff4df92b2330e5fec67ff08c7cbb4c4ab8d.
Reason for revert: Some crashes found by the fuzzer
Original change's description:
> Refactor FEC code to use COW buffers
>
> This refactoring helps to reduce unnecessary memcpy calls on the receive
> side.
>
> This CL replaces
> |uint8 data[IP_PACKET_SIZE]| with |rtc::CopyOnWriteBuffer data| in Packet class,
> removes |length| field there, and does necessary changes.
>
> This is a reland of these two CLs with fixes:
> https://webrtc-review.googlesource.com/c/src/+/144942
> https://webrtc-review.googlesource.com/c/src/+/144881
>
> Bug: webrtc:10750
> Change-Id: I76f6dee5a57ade59942ea2822ca4737edfe6438b
> Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/145332
> Commit-Queue: Ilya Nikolaevskiy <ilnik@webrtc.org>
> Reviewed-by: Rasmus Brandt <brandtr@webrtc.org>
> Reviewed-by: Stefan Holmer <stefan@webrtc.org>
> Cr-Commit-Position: refs/heads/master@{#29035}
TBR=brandtr@webrtc.org,ilnik@webrtc.org,stefan@webrtc.org
Change-Id: Id3d65fb1324b9f1b0446fe217012115ecacf2b40
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: webrtc:10750
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/151130
Reviewed-by: Ilya Nikolaevskiy <ilnik@webrtc.org>
Commit-Queue: Ilya Nikolaevskiy <ilnik@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#29043}
And moved declaration into a new api directory, as
api/transport/rtp/rtp_source.h.
Bug: webrtc:8733
Change-Id: Ia73b7b0630e6065de4707a37633adddfa00a2b8a
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/150880
Commit-Queue: Niels Moller <nisse@webrtc.org>
Reviewed-by: Mirko Bonadei <mbonadei@webrtc.org>
Reviewed-by: Karl Wiberg <kwiberg@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#29039}
This refactoring helps to reduce unnecessary memcpy calls on the receive
side.
This CL replaces
|uint8 data[IP_PACKET_SIZE]| with |rtc::CopyOnWriteBuffer data| in Packet class,
removes |length| field there, and does necessary changes.
This is a reland of these two CLs with fixes:
https://webrtc-review.googlesource.com/c/src/+/144942https://webrtc-review.googlesource.com/c/src/+/144881
Bug: webrtc:10750
Change-Id: I76f6dee5a57ade59942ea2822ca4737edfe6438b
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/145332
Commit-Queue: Ilya Nikolaevskiy <ilnik@webrtc.org>
Reviewed-by: Rasmus Brandt <brandtr@webrtc.org>
Reviewed-by: Stefan Holmer <stefan@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#29035}
This reverts commit 7db900e2e78d1644a173a0bc505ad52c61c43f9b.
Reason for revert: Speculative revert
Original change's description:
> Simplify pacer queue
>
> This CL simplifies the pacer queue by removing the now unnecessary
> beginpop/cancelpop/finalizepop methods. Instead there's a const top()
> and a pop() much like an stl queue.
> Old methods using the deprecated pacing code path are cleaned away.
>
> Bug: webrtc:10633
> Change-Id: Ib6da4d46a571bf56415172b790cc9e3f63206a38
> Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/150522
> Commit-Queue: Erik Språng <sprang@webrtc.org>
> Reviewed-by: Philip Eliasson <philipel@webrtc.org>
> Cr-Commit-Position: refs/heads/master@{#28997}
TBR=sprang@webrtc.org,philipel@webrtc.org
# Not skipping CQ checks because original CL landed > 1 day ago.
Bug: webrtc:10633
Change-Id: I38f61afed4f4d542e236bcce3152a3aab52c6e6b
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/151120
Reviewed-by: Erik Språng <sprang@webrtc.org>
Commit-Queue: Erik Språng <sprang@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#29030}
There are a few reasons for making this test only:
* The code is only used by tests and utilities.
* The pure interface has only a single implementation so an interface isn't really needed.
(a followup change could remove it altogether)
* The implementation always incorporates locking regardless of how the class gets used.
See e.g. previous use in the Packet class.
* The implementation is a layer on top of RtpUtility::RtpHeaderParser which is
sufficient for most production cases.
Change-Id: Ide6d50567cf8ae5127a2eb04cceeb10cf317ec36
Bug: none
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/150658
Commit-Queue: Tommi <tommi@webrtc.org>
Reviewed-by: Karl Wiberg <kwiberg@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#29010}
This CL simplifies the pacer queue by removing the now unnecessary
beginpop/cancelpop/finalizepop methods. Instead there's a const top()
and a pop() much like an stl queue.
Old methods using the deprecated pacing code path are cleaned away.
Bug: webrtc:10633
Change-Id: Ib6da4d46a571bf56415172b790cc9e3f63206a38
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/150522
Commit-Queue: Erik Språng <sprang@webrtc.org>
Reviewed-by: Philip Eliasson <philipel@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#28997}
This interface/config field is now unused, let's remove it.
Bug: webrtc:10633
Change-Id: I56ff3d47ba784d973de411ada52ec9485bad9864
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/150531
Reviewed-by: Stefan Holmer <stefan@webrtc.org>
Commit-Queue: Erik Språng <sprang@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#28978}
Number of received FEC bytes is used for the
WebRTC.Video.FecBitrateReceivedInKbps UMA histogram. Before this cl,
that value is based on a FEC packet counter updated by
ReceiveStatistics::FecPacketReceived. This cl deletes that method, and
instead adds a byte count to the FecPacketCounter struct, which is
maintained by the UlpFecReceiver and used for other FEC-related stats.
Bug: webrtc:10917
Change-Id: I24bd494b6909a2fe109d28e2b71ca8f413d05911
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/150533
Reviewed-by: Åsa Persson <asapersson@webrtc.org>
Commit-Queue: Niels Moller <nisse@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#28976}
Previously the kDontRetransmit value was used to indicate packets that
should not be retransmitted but were put in the RtpPacketHistory anyway
as a temporary storage while waiting for a callback from PacedSender.
Since PacedSender now always owns the delayed packets directly, we can
remove all usage of StorageTye in RtpPacketHistory, and only put
packets there after pacing if RtpPacketToSend::allow_retransmission()
returns true.
Bug: webrtc:10633
Change-Id: I003b76ba43bd87658cc2a39e908fd28ebcd403f7
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/150521
Commit-Queue: Erik Språng <sprang@webrtc.org>
Reviewed-by: Åsa Persson <asapersson@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#28974}
This CL removes the old non-paced code path and instead uses a helper
class to just immediately pass the packet through the same code path as
when an actual pacer is used.
Bug: webrtc:10633
Change-Id: Id9a3ee4719829ad07710f5468e5452aa4bc8570b
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/150530
Commit-Queue: Erik Språng <sprang@webrtc.org>
Reviewed-by: Sebastian Jansson <srte@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#28963}
The methods are no longer in use, this CL cleans away references and
updates any tests using them.
Bug: webrtc:10633
Change-Id: I2db301e0a021a2f85a8b9a74e409303baba407da
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/150520
Reviewed-by: Stefan Holmer <stefan@webrtc.org>
Reviewed-by: Sebastian Jansson <srte@webrtc.org>
Commit-Queue: Erik Språng <sprang@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#28956}
And a corresponding struct RtpReceiveStats. This is intended
to hold the information exposed via GetStats, which is quite
different from the stats reported to the peer via RTCP.
This is a preparation for moving ReceiveStatistics out of the
individual receive stream objects, and instead have a shared instance
owned by RtpStreamReceiverController or maybe Call.
Bug: webrtc:10679,chromium:677543
Change-Id: Ibb52ee769516ddc51da109b7f2319405693be5d5
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/148982
Reviewed-by: Karl Wiberg <kwiberg@webrtc.org>
Reviewed-by: Åsa Persson <asapersson@webrtc.org>
Reviewed-by: Henrik Boström <hbos@webrtc.org>
Commit-Queue: Niels Moller <nisse@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#28943}
This change adds helper classes to manipulate Absolute Capture Time header extensions. Both classes support the "timestamp interpolation" optimization.
Bug: webrtc:10739
Change-Id: I08eff46eb8910842a6dbaa3288b976004fabe1c7
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/149801
Commit-Queue: Chen Xing <chxg@google.com>
Reviewed-by: Erik Språng <sprang@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#28936}
This flag has been default-off since Jul 24th (m77 branch) and apart
from a bug fixed on Aug 5th, there have been no reports of issues, so
let's remove it and start cleaning away the old code path.
Most of the usage within RtpSender/PacingController and their
respective unit tests are removed with this CL, but there will be
several more to follow.
Bug: webrtc:10633
Change-Id: I1986ccf093434ac8fbd8d6db82a0bb44f50b514e
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/149838
Reviewed-by: Sebastian Jansson <srte@webrtc.org>
Commit-Queue: Erik Språng <sprang@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#28930}
The name media_send_ssrc makes less sense when used mostly for the
RtcpReceiver functionality.
The old member is still there and used as a fallback. That will be
cleaned away after downstream code is fixed.
Bug: webrtc:10774
Change-Id: I4ec18db76910f31dfe76bc9b137ffe89220d3fa8
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/149836
Reviewed-by: Sebastian Jansson <srte@webrtc.org>
Reviewed-by: Fredrik Solenberg <solenberg@webrtc.org>
Commit-Queue: Erik Språng <sprang@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#28923}
There are currently three overloads with different number of arguments,
and one of those return a raw pointer. This cl changes that to unique_ptr.
The transition plan is to update those downstream call sites that
currently require a raw pointer to use one of the other overloads.
Bug: webrtc:10679
Change-Id: I234605e99c04a59fbe6f478581ed8edd96a9b05a
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/148447
Reviewed-by: Danil Chapovalov <danilchap@webrtc.org>
Reviewed-by: Karl Wiberg <kwiberg@webrtc.org>
Commit-Queue: Niels Moller <nisse@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#28804}
