webrtc_m130/audio/channel_send.cc
Jiawei Ou 55718120e6 Adding rtcp report interval into RTCConfiguration.
This is a follow up of https://webrtc-review.googlesource.com/c/src/+/43201.

Issue 43201 didn't do the job properly.
1. The audio rtcp report interval is not properly hooked up.
2. We don't need to propagate audio rtcp interval into video send stream or vice versa.
3. We don't need to propagate rtcp report interval to any receiving streams.

Bug: webrtc:8789
Change-Id: I1f637d6e5173608564ef0702d7eda6fc93b3200f
Reviewed-on: https://webrtc-review.googlesource.com/c/110105
Reviewed-by: Erik Språng <sprang@webrtc.org>
Reviewed-by: Magnus Flodman <mflodman@webrtc.org>
Reviewed-by: Niels Moller <nisse@webrtc.org>
Commit-Queue: Jiawei Ou <ouj@fb.com>
Cr-Commit-Position: refs/heads/master@{#25610}
2018-11-12 20:00:00 +00:00

1125 lines
39 KiB
C++

/*
* Copyright (c) 2012 The WebRTC project authors. All Rights Reserved.
*
* Use of this source code is governed by a BSD-style license
* that can be found in the LICENSE file in the root of the source
* tree. An additional intellectual property rights grant can be found
* in the file PATENTS. All contributing project authors may
* be found in the AUTHORS file in the root of the source tree.
*/
#include "audio/channel_send.h"
#include <algorithm>
#include <map>
#include <memory>
#include <string>
#include <utility>
#include <vector>
#include "absl/memory/memory.h"
#include "api/array_view.h"
#include "api/crypto/frameencryptorinterface.h"
#include "audio/utility/audio_frame_operations.h"
#include "call/rtp_transport_controller_send_interface.h"
#include "logging/rtc_event_log/events/rtc_event_audio_playout.h"
#include "logging/rtc_event_log/rtc_event_log.h"
#include "modules/audio_coding/audio_network_adaptor/include/audio_network_adaptor_config.h"
#include "modules/pacing/packet_router.h"
#include "modules/utility/include/process_thread.h"
#include "rtc_base/checks.h"
#include "rtc_base/criticalsection.h"
#include "rtc_base/event.h"
#include "rtc_base/format_macros.h"
#include "rtc_base/location.h"
#include "rtc_base/logging.h"
#include "rtc_base/rate_limiter.h"
#include "rtc_base/task_queue.h"
#include "rtc_base/thread_checker.h"
#include "rtc_base/timeutils.h"
#include "system_wrappers/include/field_trial.h"
#include "system_wrappers/include/metrics.h"
namespace webrtc {
namespace voe {
namespace {
constexpr int64_t kMaxRetransmissionWindowMs = 1000;
constexpr int64_t kMinRetransmissionWindowMs = 30;
MediaTransportEncodedAudioFrame::FrameType
MediaTransportFrameTypeForWebrtcFrameType(webrtc::FrameType frame_type) {
switch (frame_type) {
case kAudioFrameSpeech:
return MediaTransportEncodedAudioFrame::FrameType::kSpeech;
break;
case kAudioFrameCN:
return MediaTransportEncodedAudioFrame::FrameType::
kDiscontinuousTransmission;
break;
default:
RTC_CHECK(false) << "Unexpected frame type=" << frame_type;
break;
}
}
} // namespace
const int kTelephoneEventAttenuationdB = 10;
class TransportFeedbackProxy : public TransportFeedbackObserver {
public:
TransportFeedbackProxy() : feedback_observer_(nullptr) {
pacer_thread_.DetachFromThread();
network_thread_.DetachFromThread();
}
void SetTransportFeedbackObserver(
TransportFeedbackObserver* feedback_observer) {
RTC_DCHECK(thread_checker_.CalledOnValidThread());
rtc::CritScope lock(&crit_);
feedback_observer_ = feedback_observer;
}
// Implements TransportFeedbackObserver.
void AddPacket(uint32_t ssrc,
uint16_t sequence_number,
size_t length,
const PacedPacketInfo& pacing_info) override {
RTC_DCHECK(pacer_thread_.CalledOnValidThread());
rtc::CritScope lock(&crit_);
if (feedback_observer_)
feedback_observer_->AddPacket(ssrc, sequence_number, length, pacing_info);
}
void OnTransportFeedback(const rtcp::TransportFeedback& feedback) override {
RTC_DCHECK(network_thread_.CalledOnValidThread());
rtc::CritScope lock(&crit_);
if (feedback_observer_)
feedback_observer_->OnTransportFeedback(feedback);
}
private:
rtc::CriticalSection crit_;
rtc::ThreadChecker thread_checker_;
rtc::ThreadChecker pacer_thread_;
rtc::ThreadChecker network_thread_;
TransportFeedbackObserver* feedback_observer_ RTC_GUARDED_BY(&crit_);
};
class TransportSequenceNumberProxy : public TransportSequenceNumberAllocator {
public:
TransportSequenceNumberProxy() : seq_num_allocator_(nullptr) {
pacer_thread_.DetachFromThread();
}
void SetSequenceNumberAllocator(
TransportSequenceNumberAllocator* seq_num_allocator) {
RTC_DCHECK(thread_checker_.CalledOnValidThread());
rtc::CritScope lock(&crit_);
seq_num_allocator_ = seq_num_allocator;
}
// Implements TransportSequenceNumberAllocator.
uint16_t AllocateSequenceNumber() override {
RTC_DCHECK(pacer_thread_.CalledOnValidThread());
rtc::CritScope lock(&crit_);
if (!seq_num_allocator_)
return 0;
return seq_num_allocator_->AllocateSequenceNumber();
}
private:
rtc::CriticalSection crit_;
rtc::ThreadChecker thread_checker_;
rtc::ThreadChecker pacer_thread_;
TransportSequenceNumberAllocator* seq_num_allocator_ RTC_GUARDED_BY(&crit_);
};
class RtpPacketSenderProxy : public RtpPacketSender {
public:
RtpPacketSenderProxy() : rtp_packet_sender_(nullptr) {}
void SetPacketSender(RtpPacketSender* rtp_packet_sender) {
RTC_DCHECK(thread_checker_.CalledOnValidThread());
rtc::CritScope lock(&crit_);
rtp_packet_sender_ = rtp_packet_sender;
}
// Implements RtpPacketSender.
void InsertPacket(Priority priority,
uint32_t ssrc,
uint16_t sequence_number,
int64_t capture_time_ms,
size_t bytes,
bool retransmission) override {
rtc::CritScope lock(&crit_);
if (rtp_packet_sender_) {
rtp_packet_sender_->InsertPacket(priority, ssrc, sequence_number,
capture_time_ms, bytes, retransmission);
}
}
void SetAccountForAudioPackets(bool account_for_audio) override {
RTC_NOTREACHED();
}
private:
rtc::ThreadChecker thread_checker_;
rtc::CriticalSection crit_;
RtpPacketSender* rtp_packet_sender_ RTC_GUARDED_BY(&crit_);
};
class VoERtcpObserver : public RtcpBandwidthObserver {
public:
explicit VoERtcpObserver(ChannelSend* owner)
: owner_(owner), bandwidth_observer_(nullptr) {}
virtual ~VoERtcpObserver() {}
void SetBandwidthObserver(RtcpBandwidthObserver* bandwidth_observer) {
rtc::CritScope lock(&crit_);
bandwidth_observer_ = bandwidth_observer;
}
void OnReceivedEstimatedBitrate(uint32_t bitrate) override {
rtc::CritScope lock(&crit_);
if (bandwidth_observer_) {
bandwidth_observer_->OnReceivedEstimatedBitrate(bitrate);
}
}
void OnReceivedRtcpReceiverReport(const ReportBlockList& report_blocks,
int64_t rtt,
int64_t now_ms) override {
{
rtc::CritScope lock(&crit_);
if (bandwidth_observer_) {
bandwidth_observer_->OnReceivedRtcpReceiverReport(report_blocks, rtt,
now_ms);
}
}
// TODO(mflodman): Do we need to aggregate reports here or can we jut send
// what we get? I.e. do we ever get multiple reports bundled into one RTCP
// report for VoiceEngine?
if (report_blocks.empty())
return;
int fraction_lost_aggregate = 0;
int total_number_of_packets = 0;
// If receiving multiple report blocks, calculate the weighted average based
// on the number of packets a report refers to.
for (ReportBlockList::const_iterator block_it = report_blocks.begin();
block_it != report_blocks.end(); ++block_it) {
// Find the previous extended high sequence number for this remote SSRC,
// to calculate the number of RTP packets this report refers to. Ignore if
// we haven't seen this SSRC before.
std::map<uint32_t, uint32_t>::iterator seq_num_it =
extended_max_sequence_number_.find(block_it->source_ssrc);
int number_of_packets = 0;
if (seq_num_it != extended_max_sequence_number_.end()) {
number_of_packets =
block_it->extended_highest_sequence_number - seq_num_it->second;
}
fraction_lost_aggregate += number_of_packets * block_it->fraction_lost;
total_number_of_packets += number_of_packets;
extended_max_sequence_number_[block_it->source_ssrc] =
block_it->extended_highest_sequence_number;
}
int weighted_fraction_lost = 0;
if (total_number_of_packets > 0) {
weighted_fraction_lost =
(fraction_lost_aggregate + total_number_of_packets / 2) /
total_number_of_packets;
}
owner_->OnUplinkPacketLossRate(weighted_fraction_lost / 255.0f);
}
private:
ChannelSend* owner_;
// Maps remote side ssrc to extended highest sequence number received.
std::map<uint32_t, uint32_t> extended_max_sequence_number_;
rtc::CriticalSection crit_;
RtcpBandwidthObserver* bandwidth_observer_ RTC_GUARDED_BY(crit_);
};
class ChannelSend::ProcessAndEncodeAudioTask : public rtc::QueuedTask {
public:
ProcessAndEncodeAudioTask(std::unique_ptr<AudioFrame> audio_frame,
ChannelSend* channel)
: audio_frame_(std::move(audio_frame)), channel_(channel) {
RTC_DCHECK(channel_);
}
private:
bool Run() override {
RTC_DCHECK_RUN_ON(channel_->encoder_queue_);
channel_->ProcessAndEncodeAudioOnTaskQueue(audio_frame_.get());
return true;
}
std::unique_ptr<AudioFrame> audio_frame_;
ChannelSend* const channel_;
};
int32_t ChannelSend::SendData(FrameType frameType,
uint8_t payloadType,
uint32_t timeStamp,
const uint8_t* payloadData,
size_t payloadSize,
const RTPFragmentationHeader* fragmentation) {
RTC_DCHECK_RUN_ON(encoder_queue_);
rtc::ArrayView<const uint8_t> payload(payloadData, payloadSize);
if (media_transport() != nullptr) {
return SendMediaTransportAudio(frameType, payloadType, timeStamp, payload,
fragmentation);
} else {
return SendRtpAudio(frameType, payloadType, timeStamp, payload,
fragmentation);
}
}
int32_t ChannelSend::SendRtpAudio(FrameType frameType,
uint8_t payloadType,
uint32_t timeStamp,
rtc::ArrayView<const uint8_t> payload,
const RTPFragmentationHeader* fragmentation) {
RTC_DCHECK_RUN_ON(encoder_queue_);
if (_includeAudioLevelIndication) {
// Store current audio level in the RTP/RTCP module.
// The level will be used in combination with voice-activity state
// (frameType) to add an RTP header extension
_rtpRtcpModule->SetAudioLevel(rms_level_.Average());
}
// E2EE Custom Audio Frame Encryption (This is optional).
// Keep this buffer around for the lifetime of the send call.
rtc::Buffer encrypted_audio_payload;
if (frame_encryptor_ != nullptr) {
// TODO(benwright@webrtc.org) - Allocate enough to always encrypt inline.
// Allocate a buffer to hold the maximum possible encrypted payload.
size_t max_ciphertext_size = frame_encryptor_->GetMaxCiphertextByteSize(
cricket::MEDIA_TYPE_AUDIO, payload.size());
encrypted_audio_payload.SetSize(max_ciphertext_size);
// Encrypt the audio payload into the buffer.
size_t bytes_written = 0;
int encrypt_status = frame_encryptor_->Encrypt(
cricket::MEDIA_TYPE_AUDIO, _rtpRtcpModule->SSRC(),
/*additional_data=*/nullptr, payload, encrypted_audio_payload,
&bytes_written);
if (encrypt_status != 0) {
RTC_DLOG(LS_ERROR) << "Channel::SendData() failed encrypt audio payload: "
<< encrypt_status;
return -1;
}
// Resize the buffer to the exact number of bytes actually used.
encrypted_audio_payload.SetSize(bytes_written);
// Rewrite the payloadData and size to the new encrypted payload.
payload = encrypted_audio_payload;
} else if (crypto_options_.sframe.require_frame_encryption) {
RTC_DLOG(LS_ERROR) << "Channel::SendData() failed sending audio payload: "
<< "A frame encryptor is required but one is not set.";
return -1;
}
// Push data from ACM to RTP/RTCP-module to deliver audio frame for
// packetization.
// This call will trigger Transport::SendPacket() from the RTP/RTCP module.
if (!_rtpRtcpModule->SendOutgoingData((FrameType&)frameType, payloadType,
timeStamp,
// Leaving the time when this frame was
// received from the capture device as
// undefined for voice for now.
-1, payload.data(), payload.size(),
fragmentation, nullptr, nullptr)) {
RTC_DLOG(LS_ERROR)
<< "ChannelSend::SendData() failed to send data to RTP/RTCP module";
return -1;
}
return 0;
}
int32_t ChannelSend::SendMediaTransportAudio(
FrameType frameType,
uint8_t payloadType,
uint32_t timeStamp,
rtc::ArrayView<const uint8_t> payload,
const RTPFragmentationHeader* fragmentation) {
RTC_DCHECK_RUN_ON(encoder_queue_);
// TODO(nisse): Use null _transportPtr for MediaTransport.
// RTC_DCHECK(_transportPtr == nullptr);
uint64_t channel_id;
int sampling_rate_hz;
{
rtc::CritScope cs(&media_transport_lock_);
if (media_transport_payload_type_ != payloadType) {
// Payload type is being changed, media_transport_sampling_frequency_,
// no longer current.
return -1;
}
sampling_rate_hz = media_transport_sampling_frequency_;
channel_id = media_transport_channel_id_;
}
const MediaTransportEncodedAudioFrame frame(
/*sampling_rate_hz=*/sampling_rate_hz,
// TODO(nisse): Timestamp and sample index are the same for all supported
// audio codecs except G722. Refactor audio coding module to only use
// sample index, and leave translation to RTP time, when needed, for
// RTP-specific code.
/*starting_sample_index=*/timeStamp,
// Sample count isn't conveniently available from the AudioCodingModule,
// and needs some refactoring to wire up in a good way. For now, left as
// zero.
/*sample_count=*/0,
/*sequence_number=*/media_transport_sequence_number_,
MediaTransportFrameTypeForWebrtcFrameType(frameType), payloadType,
std::vector<uint8_t>(payload.begin(), payload.end()));
// TODO(nisse): Introduce a MediaTransportSender object bound to a specific
// channel id.
RTCError rtc_error =
media_transport()->SendAudioFrame(channel_id, std::move(frame));
if (!rtc_error.ok()) {
RTC_LOG(LS_ERROR) << "Failed to send frame, rtc_error="
<< ToString(rtc_error.type()) << ", "
<< rtc_error.message();
return -1;
}
++media_transport_sequence_number_;
return 0;
}
bool ChannelSend::SendRtp(const uint8_t* data,
size_t len,
const PacketOptions& options) {
// We should not be sending RTP packets if media transport is available.
RTC_CHECK(!media_transport());
rtc::CritScope cs(&_callbackCritSect);
if (_transportPtr == NULL) {
RTC_DLOG(LS_ERROR)
<< "ChannelSend::SendPacket() failed to send RTP packet due to"
<< " invalid transport object";
return false;
}
if (!_transportPtr->SendRtp(data, len, options)) {
RTC_DLOG(LS_ERROR) << "ChannelSend::SendPacket() RTP transmission failed";
return false;
}
return true;
}
bool ChannelSend::SendRtcp(const uint8_t* data, size_t len) {
rtc::CritScope cs(&_callbackCritSect);
if (_transportPtr == NULL) {
RTC_DLOG(LS_ERROR)
<< "ChannelSend::SendRtcp() failed to send RTCP packet due to"
<< " invalid transport object";
return false;
}
int n = _transportPtr->SendRtcp(data, len);
if (n < 0) {
RTC_DLOG(LS_ERROR) << "ChannelSend::SendRtcp() transmission failed";
return false;
}
return true;
}
int ChannelSend::PreferredSampleRate() const {
// Return the bigger of playout and receive frequency in the ACM.
return std::max(audio_coding_->ReceiveFrequency(),
audio_coding_->PlayoutFrequency());
}
ChannelSend::ChannelSend(rtc::TaskQueue* encoder_queue,
ProcessThread* module_process_thread,
MediaTransportInterface* media_transport,
RtcpRttStats* rtcp_rtt_stats,
RtcEventLog* rtc_event_log,
FrameEncryptorInterface* frame_encryptor,
const webrtc::CryptoOptions& crypto_options,
bool extmap_allow_mixed,
int rtcp_report_interval_ms)
: event_log_(rtc_event_log),
_timeStamp(0), // This is just an offset, RTP module will add it's own
// random offset
send_sequence_number_(0),
_moduleProcessThreadPtr(module_process_thread),
_transportPtr(NULL),
input_mute_(false),
previous_frame_muted_(false),
_includeAudioLevelIndication(false),
transport_overhead_per_packet_(0),
rtp_overhead_per_packet_(0),
rtcp_observer_(new VoERtcpObserver(this)),
feedback_observer_proxy_(new TransportFeedbackProxy()),
seq_num_allocator_proxy_(new TransportSequenceNumberProxy()),
rtp_packet_sender_proxy_(new RtpPacketSenderProxy()),
retransmission_rate_limiter_(new RateLimiter(Clock::GetRealTimeClock(),
kMaxRetransmissionWindowMs)),
use_twcc_plr_for_ana_(
webrtc::field_trial::FindFullName("UseTwccPlrForAna") == "Enabled"),
encoder_queue_(encoder_queue),
media_transport_(media_transport),
frame_encryptor_(frame_encryptor),
crypto_options_(crypto_options) {
RTC_DCHECK(module_process_thread);
RTC_DCHECK(encoder_queue);
audio_coding_.reset(AudioCodingModule::Create(AudioCodingModule::Config()));
RtpRtcp::Configuration configuration;
configuration.audio = true;
configuration.outgoing_transport = this;
configuration.overhead_observer = this;
configuration.bandwidth_callback = rtcp_observer_.get();
configuration.paced_sender = rtp_packet_sender_proxy_.get();
configuration.transport_sequence_number_allocator =
seq_num_allocator_proxy_.get();
configuration.transport_feedback_callback = feedback_observer_proxy_.get();
configuration.event_log = event_log_;
configuration.rtt_stats = rtcp_rtt_stats;
configuration.retransmission_rate_limiter =
retransmission_rate_limiter_.get();
configuration.extmap_allow_mixed = extmap_allow_mixed;
configuration.rtcp_interval_config.audio_interval_ms =
rtcp_report_interval_ms;
_rtpRtcpModule.reset(RtpRtcp::CreateRtpRtcp(configuration));
_rtpRtcpModule->SetSendingMediaStatus(false);
Init();
}
ChannelSend::~ChannelSend() {
Terminate();
RTC_DCHECK(!channel_state_.Get().sending);
}
void ChannelSend::Init() {
channel_state_.Reset();
// --- Add modules to process thread (for periodic schedulation)
_moduleProcessThreadPtr->RegisterModule(_rtpRtcpModule.get(), RTC_FROM_HERE);
// --- ACM initialization
int error = audio_coding_->InitializeReceiver();
RTC_DCHECK_EQ(0, error);
// --- RTP/RTCP module initialization
// Ensure that RTCP is enabled by default for the created channel.
// Note that, the module will keep generating RTCP until it is explicitly
// disabled by the user.
// After StopListen (when no sockets exists), RTCP packets will no longer
// be transmitted since the Transport object will then be invalid.
// RTCP is enabled by default.
_rtpRtcpModule->SetRTCPStatus(RtcpMode::kCompound);
// --- Register all permanent callbacks
error = audio_coding_->RegisterTransportCallback(this);
RTC_DCHECK_EQ(0, error);
}
void ChannelSend::Terminate() {
RTC_DCHECK(construction_thread_.CalledOnValidThread());
// Must be called on the same thread as Init().
StopSend();
// The order to safely shutdown modules in a channel is:
// 1. De-register callbacks in modules
// 2. De-register modules in process thread
// 3. Destroy modules
int error = audio_coding_->RegisterTransportCallback(NULL);
RTC_DCHECK_EQ(0, error);
// De-register modules in process thread
if (_moduleProcessThreadPtr)
_moduleProcessThreadPtr->DeRegisterModule(_rtpRtcpModule.get());
// End of modules shutdown
}
int32_t ChannelSend::StartSend() {
if (channel_state_.Get().sending) {
return 0;
}
channel_state_.SetSending(true);
// Resume the previous sequence number which was reset by StopSend(). This
// needs to be done before |sending| is set to true on the RTP/RTCP module.
if (send_sequence_number_) {
_rtpRtcpModule->SetSequenceNumber(send_sequence_number_);
}
_rtpRtcpModule->SetSendingMediaStatus(true);
if (_rtpRtcpModule->SetSendingStatus(true) != 0) {
RTC_DLOG(LS_ERROR) << "StartSend() RTP/RTCP failed to start sending";
_rtpRtcpModule->SetSendingMediaStatus(false);
rtc::CritScope cs(&_callbackCritSect);
channel_state_.SetSending(false);
return -1;
}
{
// It is now OK to start posting tasks to the encoder task queue.
rtc::CritScope cs(&encoder_queue_lock_);
encoder_queue_is_active_ = true;
}
return 0;
}
void ChannelSend::StopSend() {
if (!channel_state_.Get().sending) {
return;
}
channel_state_.SetSending(false);
// Post a task to the encoder thread which sets an event when the task is
// executed. We know that no more encoding tasks will be added to the task
// queue for this channel since sending is now deactivated. It means that,
// if we wait for the event to bet set, we know that no more pending tasks
// exists and it is therfore guaranteed that the task queue will never try
// to acccess and invalid channel object.
RTC_DCHECK(encoder_queue_);
rtc::Event flush;
{
// Clear |encoder_queue_is_active_| under lock to prevent any other tasks
// than this final "flush task" to be posted on the queue.
rtc::CritScope cs(&encoder_queue_lock_);
encoder_queue_is_active_ = false;
encoder_queue_->PostTask([&flush]() { flush.Set(); });
}
flush.Wait(rtc::Event::kForever);
// Store the sequence number to be able to pick up the same sequence for
// the next StartSend(). This is needed for restarting device, otherwise
// it might cause libSRTP to complain about packets being replayed.
// TODO(xians): Remove this workaround after RtpRtcpModule's refactoring
// CL is landed. See issue
// https://code.google.com/p/webrtc/issues/detail?id=2111 .
send_sequence_number_ = _rtpRtcpModule->SequenceNumber();
// Reset sending SSRC and sequence number and triggers direct transmission
// of RTCP BYE
if (_rtpRtcpModule->SetSendingStatus(false) == -1) {
RTC_DLOG(LS_ERROR) << "StartSend() RTP/RTCP failed to stop sending";
}
_rtpRtcpModule->SetSendingMediaStatus(false);
}
bool ChannelSend::SetEncoder(int payload_type,
std::unique_ptr<AudioEncoder> encoder) {
RTC_DCHECK_GE(payload_type, 0);
RTC_DCHECK_LE(payload_type, 127);
// TODO(ossu): Make CodecInsts up, for now: one for the RTP/RTCP module and
// one for for us to keep track of sample rate and number of channels, etc.
// The RTP/RTCP module needs to know the RTP timestamp rate (i.e. clockrate)
// as well as some other things, so we collect this info and send it along.
CodecInst rtp_codec;
rtp_codec.pltype = payload_type;
strncpy(rtp_codec.plname, "audio", sizeof(rtp_codec.plname));
rtp_codec.plname[sizeof(rtp_codec.plname) - 1] = 0;
// Seems unclear if it should be clock rate or sample rate. CodecInst
// supposedly carries the sample rate, but only clock rate seems sensible to
// send to the RTP/RTCP module.
rtp_codec.plfreq = encoder->RtpTimestampRateHz();
rtp_codec.pacsize = rtc::CheckedDivExact(
static_cast<int>(encoder->Max10MsFramesInAPacket() * rtp_codec.plfreq),
100);
rtp_codec.channels = encoder->NumChannels();
rtp_codec.rate = 0;
if (_rtpRtcpModule->RegisterSendPayload(rtp_codec) != 0) {
_rtpRtcpModule->DeRegisterSendPayload(payload_type);
if (_rtpRtcpModule->RegisterSendPayload(rtp_codec) != 0) {
RTC_DLOG(LS_ERROR)
<< "SetEncoder() failed to register codec to RTP/RTCP module";
return false;
}
}
if (media_transport_) {
rtc::CritScope cs(&media_transport_lock_);
media_transport_payload_type_ = payload_type;
// TODO(nisse): Currently broken for G722, since timestamps passed through
// encoder use RTP clock rather than sample count, and they differ for G722.
media_transport_sampling_frequency_ = encoder->RtpTimestampRateHz();
}
audio_coding_->SetEncoder(std::move(encoder));
return true;
}
void ChannelSend::ModifyEncoder(
rtc::FunctionView<void(std::unique_ptr<AudioEncoder>*)> modifier) {
audio_coding_->ModifyEncoder(modifier);
}
void ChannelSend::SetBitRate(int bitrate_bps, int64_t probing_interval_ms) {
audio_coding_->ModifyEncoder([&](std::unique_ptr<AudioEncoder>* encoder) {
if (*encoder) {
(*encoder)->OnReceivedUplinkBandwidth(bitrate_bps, probing_interval_ms);
}
});
retransmission_rate_limiter_->SetMaxRate(bitrate_bps);
configured_bitrate_bps_ = bitrate_bps;
}
int ChannelSend::GetBitRate() const {
return configured_bitrate_bps_;
}
void ChannelSend::OnTwccBasedUplinkPacketLossRate(float packet_loss_rate) {
if (!use_twcc_plr_for_ana_)
return;
audio_coding_->ModifyEncoder([&](std::unique_ptr<AudioEncoder>* encoder) {
if (*encoder) {
(*encoder)->OnReceivedUplinkPacketLossFraction(packet_loss_rate);
}
});
}
void ChannelSend::OnRecoverableUplinkPacketLossRate(
float recoverable_packet_loss_rate) {
audio_coding_->ModifyEncoder([&](std::unique_ptr<AudioEncoder>* encoder) {
if (*encoder) {
(*encoder)->OnReceivedUplinkRecoverablePacketLossFraction(
recoverable_packet_loss_rate);
}
});
}
void ChannelSend::OnUplinkPacketLossRate(float packet_loss_rate) {
if (use_twcc_plr_for_ana_)
return;
audio_coding_->ModifyEncoder([&](std::unique_ptr<AudioEncoder>* encoder) {
if (*encoder) {
(*encoder)->OnReceivedUplinkPacketLossFraction(packet_loss_rate);
}
});
}
bool ChannelSend::EnableAudioNetworkAdaptor(const std::string& config_string) {
bool success = false;
audio_coding_->ModifyEncoder([&](std::unique_ptr<AudioEncoder>* encoder) {
if (*encoder) {
success =
(*encoder)->EnableAudioNetworkAdaptor(config_string, event_log_);
}
});
return success;
}
void ChannelSend::DisableAudioNetworkAdaptor() {
audio_coding_->ModifyEncoder([&](std::unique_ptr<AudioEncoder>* encoder) {
if (*encoder)
(*encoder)->DisableAudioNetworkAdaptor();
});
}
void ChannelSend::SetReceiverFrameLengthRange(int min_frame_length_ms,
int max_frame_length_ms) {
audio_coding_->ModifyEncoder([&](std::unique_ptr<AudioEncoder>* encoder) {
if (*encoder) {
(*encoder)->SetReceiverFrameLengthRange(min_frame_length_ms,
max_frame_length_ms);
}
});
}
void ChannelSend::RegisterTransport(Transport* transport) {
rtc::CritScope cs(&_callbackCritSect);
_transportPtr = transport;
}
int32_t ChannelSend::ReceivedRTCPPacket(const uint8_t* data, size_t length) {
// Deliver RTCP packet to RTP/RTCP module for parsing
_rtpRtcpModule->IncomingRtcpPacket(data, length);
int64_t rtt = GetRTT();
if (rtt == 0) {
// Waiting for valid RTT.
return 0;
}
int64_t nack_window_ms = rtt;
if (nack_window_ms < kMinRetransmissionWindowMs) {
nack_window_ms = kMinRetransmissionWindowMs;
} else if (nack_window_ms > kMaxRetransmissionWindowMs) {
nack_window_ms = kMaxRetransmissionWindowMs;
}
retransmission_rate_limiter_->SetWindowSize(nack_window_ms);
// Invoke audio encoders OnReceivedRtt().
audio_coding_->ModifyEncoder([&](std::unique_ptr<AudioEncoder>* encoder) {
if (*encoder)
(*encoder)->OnReceivedRtt(rtt);
});
return 0;
}
void ChannelSend::SetInputMute(bool enable) {
rtc::CritScope cs(&volume_settings_critsect_);
input_mute_ = enable;
}
bool ChannelSend::InputMute() const {
rtc::CritScope cs(&volume_settings_critsect_);
return input_mute_;
}
int ChannelSend::SendTelephoneEventOutband(int event, int duration_ms) {
RTC_DCHECK_LE(0, event);
RTC_DCHECK_GE(255, event);
RTC_DCHECK_LE(0, duration_ms);
RTC_DCHECK_GE(65535, duration_ms);
if (!Sending()) {
return -1;
}
if (_rtpRtcpModule->SendTelephoneEventOutband(
event, duration_ms, kTelephoneEventAttenuationdB) != 0) {
RTC_DLOG(LS_ERROR) << "SendTelephoneEventOutband() failed to send event";
return -1;
}
return 0;
}
int ChannelSend::SetSendTelephoneEventPayloadType(int payload_type,
int payload_frequency) {
RTC_DCHECK_LE(0, payload_type);
RTC_DCHECK_GE(127, payload_type);
CodecInst codec = {0};
codec.pltype = payload_type;
codec.plfreq = payload_frequency;
memcpy(codec.plname, "telephone-event", 16);
if (_rtpRtcpModule->RegisterSendPayload(codec) != 0) {
_rtpRtcpModule->DeRegisterSendPayload(codec.pltype);
if (_rtpRtcpModule->RegisterSendPayload(codec) != 0) {
RTC_DLOG(LS_ERROR)
<< "SetSendTelephoneEventPayloadType() failed to register "
"send payload type";
return -1;
}
}
return 0;
}
int ChannelSend::SetLocalSSRC(unsigned int ssrc) {
if (channel_state_.Get().sending) {
RTC_DLOG(LS_ERROR) << "SetLocalSSRC() already sending";
return -1;
}
if (media_transport_) {
rtc::CritScope cs(&media_transport_lock_);
media_transport_channel_id_ = ssrc;
}
_rtpRtcpModule->SetSSRC(ssrc);
return 0;
}
void ChannelSend::SetMid(const std::string& mid, int extension_id) {
int ret = SetSendRtpHeaderExtension(true, kRtpExtensionMid, extension_id);
RTC_DCHECK_EQ(0, ret);
_rtpRtcpModule->SetMid(mid);
}
void ChannelSend::SetExtmapAllowMixed(bool extmap_allow_mixed) {
_rtpRtcpModule->SetExtmapAllowMixed(extmap_allow_mixed);
}
int ChannelSend::SetSendAudioLevelIndicationStatus(bool enable,
unsigned char id) {
_includeAudioLevelIndication = enable;
return SetSendRtpHeaderExtension(enable, kRtpExtensionAudioLevel, id);
}
void ChannelSend::EnableSendTransportSequenceNumber(int id) {
int ret =
SetSendRtpHeaderExtension(true, kRtpExtensionTransportSequenceNumber, id);
RTC_DCHECK_EQ(0, ret);
}
void ChannelSend::RegisterSenderCongestionControlObjects(
RtpTransportControllerSendInterface* transport,
RtcpBandwidthObserver* bandwidth_observer) {
RtpPacketSender* rtp_packet_sender = transport->packet_sender();
TransportFeedbackObserver* transport_feedback_observer =
transport->transport_feedback_observer();
PacketRouter* packet_router = transport->packet_router();
RTC_DCHECK(rtp_packet_sender);
RTC_DCHECK(transport_feedback_observer);
RTC_DCHECK(packet_router);
RTC_DCHECK(!packet_router_);
rtcp_observer_->SetBandwidthObserver(bandwidth_observer);
feedback_observer_proxy_->SetTransportFeedbackObserver(
transport_feedback_observer);
seq_num_allocator_proxy_->SetSequenceNumberAllocator(packet_router);
rtp_packet_sender_proxy_->SetPacketSender(rtp_packet_sender);
_rtpRtcpModule->SetStorePacketsStatus(true, 600);
constexpr bool remb_candidate = false;
packet_router->AddSendRtpModule(_rtpRtcpModule.get(), remb_candidate);
packet_router_ = packet_router;
}
void ChannelSend::ResetSenderCongestionControlObjects() {
RTC_DCHECK(packet_router_);
_rtpRtcpModule->SetStorePacketsStatus(false, 600);
rtcp_observer_->SetBandwidthObserver(nullptr);
feedback_observer_proxy_->SetTransportFeedbackObserver(nullptr);
seq_num_allocator_proxy_->SetSequenceNumberAllocator(nullptr);
packet_router_->RemoveSendRtpModule(_rtpRtcpModule.get());
packet_router_ = nullptr;
rtp_packet_sender_proxy_->SetPacketSender(nullptr);
}
void ChannelSend::SetRTCPStatus(bool enable) {
_rtpRtcpModule->SetRTCPStatus(enable ? RtcpMode::kCompound : RtcpMode::kOff);
}
int ChannelSend::SetRTCP_CNAME(const char cName[256]) {
if (_rtpRtcpModule->SetCNAME(cName) != 0) {
RTC_DLOG(LS_ERROR) << "SetRTCP_CNAME() failed to set RTCP CNAME";
return -1;
}
return 0;
}
int ChannelSend::GetRemoteRTCPReportBlocks(
std::vector<ReportBlock>* report_blocks) {
if (report_blocks == NULL) {
RTC_DLOG(LS_ERROR) << "GetRemoteRTCPReportBlock()s invalid report_blocks.";
return -1;
}
// Get the report blocks from the latest received RTCP Sender or Receiver
// Report. Each element in the vector contains the sender's SSRC and a
// report block according to RFC 3550.
std::vector<RTCPReportBlock> rtcp_report_blocks;
if (_rtpRtcpModule->RemoteRTCPStat(&rtcp_report_blocks) != 0) {
return -1;
}
if (rtcp_report_blocks.empty())
return 0;
std::vector<RTCPReportBlock>::const_iterator it = rtcp_report_blocks.begin();
for (; it != rtcp_report_blocks.end(); ++it) {
ReportBlock report_block;
report_block.sender_SSRC = it->sender_ssrc;
report_block.source_SSRC = it->source_ssrc;
report_block.fraction_lost = it->fraction_lost;
report_block.cumulative_num_packets_lost = it->packets_lost;
report_block.extended_highest_sequence_number =
it->extended_highest_sequence_number;
report_block.interarrival_jitter = it->jitter;
report_block.last_SR_timestamp = it->last_sender_report_timestamp;
report_block.delay_since_last_SR = it->delay_since_last_sender_report;
report_blocks->push_back(report_block);
}
return 0;
}
int ChannelSend::GetRTPStatistics(CallSendStatistics& stats) {
// --- RtcpStatistics
// --- RTT
stats.rttMs = GetRTT();
// --- Data counters
size_t bytesSent(0);
uint32_t packetsSent(0);
if (_rtpRtcpModule->DataCountersRTP(&bytesSent, &packetsSent) != 0) {
RTC_DLOG(LS_WARNING)
<< "GetRTPStatistics() failed to retrieve RTP datacounters"
<< " => output will not be complete";
}
stats.bytesSent = bytesSent;
stats.packetsSent = packetsSent;
return 0;
}
void ChannelSend::SetNACKStatus(bool enable, int maxNumberOfPackets) {
// None of these functions can fail.
if (enable)
audio_coding_->EnableNack(maxNumberOfPackets);
else
audio_coding_->DisableNack();
}
// Called when we are missing one or more packets.
int ChannelSend::ResendPackets(const uint16_t* sequence_numbers, int length) {
return _rtpRtcpModule->SendNACK(sequence_numbers, length);
}
void ChannelSend::ProcessAndEncodeAudio(
std::unique_ptr<AudioFrame> audio_frame) {
// Avoid posting any new tasks if sending was already stopped in StopSend().
rtc::CritScope cs(&encoder_queue_lock_);
if (!encoder_queue_is_active_) {
return;
}
// Profile time between when the audio frame is added to the task queue and
// when the task is actually executed.
audio_frame->UpdateProfileTimeStamp();
encoder_queue_->PostTask(std::unique_ptr<rtc::QueuedTask>(
new ProcessAndEncodeAudioTask(std::move(audio_frame), this)));
}
void ChannelSend::ProcessAndEncodeAudioOnTaskQueue(AudioFrame* audio_input) {
RTC_DCHECK_RUN_ON(encoder_queue_);
RTC_DCHECK_GT(audio_input->samples_per_channel_, 0);
RTC_DCHECK_LE(audio_input->num_channels_, 2);
// Measure time between when the audio frame is added to the task queue and
// when the task is actually executed. Goal is to keep track of unwanted
// extra latency added by the task queue.
RTC_HISTOGRAM_COUNTS_10000("WebRTC.Audio.EncodingTaskQueueLatencyMs",
audio_input->ElapsedProfileTimeMs());
bool is_muted = InputMute();
AudioFrameOperations::Mute(audio_input, previous_frame_muted_, is_muted);
if (_includeAudioLevelIndication) {
size_t length =
audio_input->samples_per_channel_ * audio_input->num_channels_;
RTC_CHECK_LE(length, AudioFrame::kMaxDataSizeBytes);
if (is_muted && previous_frame_muted_) {
rms_level_.AnalyzeMuted(length);
} else {
rms_level_.Analyze(
rtc::ArrayView<const int16_t>(audio_input->data(), length));
}
}
previous_frame_muted_ = is_muted;
// Add 10ms of raw (PCM) audio data to the encoder @ 32kHz.
// The ACM resamples internally.
audio_input->timestamp_ = _timeStamp;
// This call will trigger AudioPacketizationCallback::SendData if encoding
// is done and payload is ready for packetization and transmission.
// Otherwise, it will return without invoking the callback.
if (audio_coding_->Add10MsData(*audio_input) < 0) {
RTC_DLOG(LS_ERROR) << "ACM::Add10MsData() failed.";
return;
}
_timeStamp += static_cast<uint32_t>(audio_input->samples_per_channel_);
}
void ChannelSend::UpdateOverheadForEncoder() {
size_t overhead_per_packet =
transport_overhead_per_packet_ + rtp_overhead_per_packet_;
audio_coding_->ModifyEncoder([&](std::unique_ptr<AudioEncoder>* encoder) {
if (*encoder) {
(*encoder)->OnReceivedOverhead(overhead_per_packet);
}
});
}
void ChannelSend::SetTransportOverhead(size_t transport_overhead_per_packet) {
rtc::CritScope cs(&overhead_per_packet_lock_);
transport_overhead_per_packet_ = transport_overhead_per_packet;
UpdateOverheadForEncoder();
}
// TODO(solenberg): Make AudioSendStream an OverheadObserver instead.
void ChannelSend::OnOverheadChanged(size_t overhead_bytes_per_packet) {
rtc::CritScope cs(&overhead_per_packet_lock_);
rtp_overhead_per_packet_ = overhead_bytes_per_packet;
UpdateOverheadForEncoder();
}
ANAStats ChannelSend::GetANAStatistics() const {
return audio_coding_->GetANAStats();
}
RtpRtcp* ChannelSend::GetRtpRtcp() const {
return _rtpRtcpModule.get();
}
int ChannelSend::SetSendRtpHeaderExtension(bool enable,
RTPExtensionType type,
unsigned char id) {
int error = 0;
_rtpRtcpModule->DeregisterSendRtpHeaderExtension(type);
if (enable) {
error = _rtpRtcpModule->RegisterSendRtpHeaderExtension(type, id);
}
return error;
}
int ChannelSend::GetRtpTimestampRateHz() const {
const auto format = audio_coding_->ReceiveFormat();
// Default to the playout frequency if we've not gotten any packets yet.
// TODO(ossu): Zero clockrate can only happen if we've added an external
// decoder for a format we don't support internally. Remove once that way of
// adding decoders is gone!
return (format && format->clockrate_hz != 0)
? format->clockrate_hz
: audio_coding_->PlayoutFrequency();
}
int64_t ChannelSend::GetRTT() const {
RtcpMode method = _rtpRtcpModule->RTCP();
if (method == RtcpMode::kOff) {
return 0;
}
std::vector<RTCPReportBlock> report_blocks;
_rtpRtcpModule->RemoteRTCPStat(&report_blocks);
if (report_blocks.empty()) {
return 0;
}
int64_t rtt = 0;
int64_t avg_rtt = 0;
int64_t max_rtt = 0;
int64_t min_rtt = 0;
// We don't know in advance the remote ssrc used by the other end's receiver
// reports, so use the SSRC of the first report block for calculating the RTT.
if (_rtpRtcpModule->RTT(report_blocks[0].sender_ssrc, &rtt, &avg_rtt,
&min_rtt, &max_rtt) != 0) {
return 0;
}
return rtt;
}
void ChannelSend::SetFrameEncryptor(
rtc::scoped_refptr<FrameEncryptorInterface> frame_encryptor) {
rtc::CritScope cs(&encoder_queue_lock_);
if (encoder_queue_is_active_) {
encoder_queue_->PostTask([this, frame_encryptor]() {
this->frame_encryptor_ = std::move(frame_encryptor);
});
} else {
frame_encryptor_ = std::move(frame_encryptor);
}
}
} // namespace voe
} // namespace webrtc