Implement RTP keepalive in native stack.

webrtc/modules/rtp_rtcp/source/rtp_sender.cc

 /*
  *  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.
  */
 
@@ skipping 1271 matching lines @@
     rtc::CritScope lock(&send_critsect_);
     if (rtp_overhead_bytes_per_packet_ == packet.headers_size()) {
       return;
     }
     rtp_overhead_bytes_per_packet_ = packet.headers_size();
     overhead_bytes_per_packet = rtp_overhead_bytes_per_packet_;
   }
   overhead_observer_->OnOverheadChanged(overhead_bytes_per_packet);
 }
 
+int64_t RTPSender::LastTimestampTimeMs() const {
+  rtc::CritScope lock(&send_critsect_);
+  return last_timestamp_time_ms_;
+}
+
+void RTPSender::SendKeepAlive(uint8_t payload_type) {
+  rtc::CritScope lock(&send_critsect_);
+  if (!ssrc_)
+    return;
+
+  std::unique_ptr<RtpPacketToSend> packet = AllocatePacket();
+  packet->SetMarker(true);
+  packet->SetPayloadType(payload_type);
+  packet->SetTimestamp(last_rtp_timestamp_);

[åsapersson, 2017/06/30 15:09:13]

Could last_rtp_timestamp_ be 0 here and if so ok?

[sprang_webrtc, 2017/06/30 16:21:00]

Hm, if the stream is started with media being sent, then yes it can be 0.
This can potentially lead to a problem since it seems that the bandwidth
estimator
is triggered for the packet, even though it get's discarded.

However the alternative is to not keep the connection open, which is a
problem...
Maybe we should use the timestamp offset logic from SendPadData() though, for
temporary video mute?

Should we also try to ignore these packets in the bwe?

+  packet->set_capture_time_ms(capture_time_ms_);

[åsapersson, 2017/06/30 15:09:13]

is locked needed for SendToNetwork?

And AllocatePacket?

[sprang_webrtc, 2017/06/30 16:21:00]

No, in fact it probably shouldn't. Missed cleaning this up.
Only needed for reading timestamps.

+  SendToNetwork(std::move(packet), StorageType::kDontRetransmit,
+                RtpPacketSender::Priority::kLowPriority);
+}
+
 }  // namespace webrtc