PSA: WebRTC M56 Release Notes

[Anatoli Davidson]

M56

WebRTC M56 branch (cut at r15101)

Summary

Chrome M56, currently available in Chrome's beta channel, contains over 20 new features and over 50 bug fixes for WebRTC, including enhancements and stability/performance improvements. As with previous releases, we encourage all developers to run versions of Chrome on the Canary, Dev, and Beta channels frequently and quickly report any issues found. Please take a look at this page, for some pointers on how to file a good bug report. The help we have received has been invaluable!

The Chrome release schedule can be found here.

Important PSAs

Features

RTCPeerConnection is now un-prefixed

You can now invoke an RTCPeerConnection as “new RTCPeerConnection”. Code that uses “new webkitRTCPeerConnection” will continue to work for now, but you don’t have to shim this any more.

Also, as part of our continuous work to improve standards conformance, we’ve made it possible to use the name “iceTransportPolicy” in the RTCPeerConnection initialization parameters; the old, nonstandard name “iceTransports” can still be used.

New RTCPeerConnection.getStats

A promise-based, spec-compliant version of RTCPeerConnection.getStats is under development. It can be enabled in Chrome using command line argument --enable-blink-features=RTCPeerConnectionNewGetStats. In M56, most RTCStats-derived dictionaries are supported but with many dictionary members missing. See master bug and current state of available stats here. Please try out the new getStats API, your feedback is highly appreciated!

Old RED/RTX workarounds removed

This release removes some old workarounds relating to RED over RTX, which were temporarily introduced to mitigate a bug in older Chrome versions. The workaround removal does not affect developers using the default SDP. Developers that munge the SDP with respect to RED and/or RTX, however, now need to ensure that RED over RTX is explicitly enabled, if so desired. The background and more information is available in the public PSA and in the bug tracker.

Hardware encode for H.264 on Windows

On Windows 8.1 and above we will leverage hardware encode support for H.264 video in Chrome when the device hardware supports it. This will reduce significantly CPU usage and enable higher resolution video.

Deprecations

• Build: the last traces of GYP were finally removed.

Platform	Issue	Description	Component
Chrome	653579	Remove support for GpuMemoryBuffers video capture	Blink>MediaStream, Blink>Network>XHR
Chrome	388194	Remove MediaStreamTrack.getSources implementation	Mobile
Chrome	659965	Remove "legacy" stream support from xhr	Video
Native	6687	Remove usage of webrtc::Clock in audio_device/	Audio, Cleanup
Native	6606	Remove deprecated camera enumeration methods from CameraEnumerationAndroid.	Blink>GetUserMedia
Native	6390	Drop support for legacy camera2 devices.	Blink>GetUserMedia>Webcam, Blink>MediaStream
Native	5332	Remove cricket::VideoCodec width/height/framerate	Video
Native	5682	Replace cricket::VideoFrame with webrtc::VideoFrame:	Video
Native	5924	Removal of cricket::VideoRendererFactory and most of the corresponding renderers.	Video
Native	6603	Use accessors for “codec specific” video codec parameters. Accessors will be required in 57.	Video
Native	6027	webrtc/base/rtc_base/macutils.cc uses Gestalt() APIs which are deprecated when building with a 10.8 deployment target	---
Native	6028	base/rtc_base/proxydetect.cc and base/rtc_base/unixfilesystem.cc use FSRef APIs on OS X, which are deprecated with a 10.8 deployment target	---
Native	6029	modules/desktop_capture/desktop_capture/screen_capturer_mac.mm uses deprecated functions with a 10.8 deployment target	---

Features and Bugfixes

Chrome

Type	Issue	Description	Component
Feature	6515	The AudioProcessing class should be a pure interface	Audio
Feature	6525	Implement residual echo detector	Audio
Feature	6462	Add UMA metrics to track ICE regathering reasons	Network
Feature	6531	Create rtc::PacketTransportInterface	Network
Feature	5868	Implement suspend_below_min_bitrate per send stream	Video
Feature	6579	Add stats for frequency offset when converting RTP timestamp to NTP time.	Video
Feature	6601	Implement H264 level-idc negotiation (last two two characters of profile-level-id)	Video
Feature	6650	Remove RED/RTX workarounds.	Video
Feature	6376	Update third_party/libsrtp to version 2.0	---
Feature	658423	Expose unprefixed RTCPeerConnection	Blink>WebRTC
Bugfix	1503	Data Race on access to cftmdl_wk1r	Audio
Bugfix	1504	Data Race on access to cft1st_128	Audio
Bugfix	2521	data racing in audio_processing module	Audio
Bugfix	6518	Audio bwe is just activate on caller side	Audio
Bugfix	6522	centralize the definition of min bwe	Audio
Bugfix	6523	Check for valid observer before call OnTransportFeedback	Audio
Bugfix	5889	UBSan error in NetEQ	Build
Bugfix	6565	BringSelectedWindowToFront has a different behavior on Windows	DesktopCapture
Bugfix	6702	ScreenCapture freezes on external display on Mac	DesktopCapture
Bugfix	6278	Handle 3 bytes AnnexB header in H264 parser	HardwareCodec, Video
Bugfix	6106	Java bindings do not expose DataChannel ID	PeerConnection
Bugfix	6218	data-channel only offer with max-bundle and rtcpMuxPolicy: require fails	PeerConnection
Bugfix	6299	Remove the obsolete enum webrtc::PeerConnectionInterface::IceState.	PeerConnection
Bugfix	6652	PeerConnection::GetStats return true for invalid track	PeerConnection
Bugfix	6400	WebRTC fails to match compatible H264 profile-level-ids	Video
Bugfix	6402	cricket::WebRtcVideoEncoderFactory::VideoCodec does not contain enough information for H264	Video
Bugfix	6501	Potential race condition in VideoReceiveStream shutdown	Video
Bugfix	6510	event_log_visualizer broken	Video
Bugfix	6552	video interop with Edge broken in M56	Video
Bugfix	6563	Fix a regression with quality scaling	Video
Bugfix	6583	WebRTC not compatible with OpenH264 v1.6	Video
Bugfix	6600	Sending CVO rtp header extension compatible with the standard now.	Video
Bugfix	6669	Division by zero in RemoteEstimatorProxy::OnBitrateChanged	Video
Bugfix	6686	Sequential access check hits in RTPSenderVideo	Video
Bugfix	6710	Probe results are ignored midcall	Video
Bugfix	4869	PeerConnection constructor: iceTransports vs iceTransportPolicy	---
Bugfix	6179	Compiler warning possible loss of data in file port.h	---
Bugfix	6432	Rename P2PTransportChannel's worker_thread_ to network_thread_.	---
Bugfix	6442	TWCC produce too much overhead on a low bandwidth situation.	---
Bugfix	6526	BweSimulation uses RemoteBitrateEstimatorAbsSendTime instead of DelayBasedBwe	---
Bugfix	6530	The functionality for emptying the render frame queue when it is full does only work properly in debug mode for AEC and AECM	---
Bugfix	6620	RTC_DCHECK_GE(unsigned, 0u) warns that condition is always true	---
Bugfix	6621	plot_dynamics.py does not plot BWE simulation results	---
Bugfix	6670	AudioSendStream min_bitrate_kbps and max_bitrate_kbps are assigned bit rate with unit of bps.	---
Bugfix	6681	Run into RTC_DCHECK on caller side when ANA is activated.	---
Bugfix	656420	chooseDesktopMedia Window Hides Behind Active Window	Blink>GetUserMedia>Desktop
Bugfix	655198	tabCapture Results in Stretched Mouse Cursor	Blink>GetUserMedia>Tab
Bugfix	657627	Change scaling from MakeHalfFloat to using GPU to do it	Blink>GetUserMedia>Webcam, Internals>Media, OS>Kernel>Video
Bugfix	660204	navigator.mediaDevices.enumerateDevices() not working from non-https origins	Blink>GetUserMedia>Webcam
Bugfix	642663	Cannot clone/forward hardware decoded video tracks	Blink>MediaStream, Blink>WebRTC>Video
Bugfix	567859	MediaRecorder.start() with no duration should buffer until stop() or requestData().	Blink>MediaStream>Recording
Bugfix	657532	Regression: Playback of mediaStream recording shows blank video	Blink>MediaStream>Recording
Bugfix	579430	ConnectDataChannel called when data_channel_ is NULL	Blink>WebRTC
Bugfix	615108	Enable WebRTC H264 on Android with HW enc/dec	Blink>WebRTC
Bugfix	636818	RTCPeerConnection.getStats: RTCPeerConnectionStats (behind flag)	Blink>WebRTC
Bugfix	645599	Chrome Ignores H264 profile-level-id on SDP offer	Blink>WebRTC
Bugfix	654927	RTCPeerConnection.getStats: RTCDataChannelStats (behind flag)	Blink>WebRTC
Bugfix	650533	Volume adjustment does not take effect in Chrome for Android during WebRTC calls	Blink>WebRTC>Audio
Bugfix	586431	Refactor AudioInputDebugWriter	Blink>WebRTC>Tools
Bugfix	661362	[Regression] AppRTC Audio receiving codec is not displayed in stats window and chrome://webrtc-internals	Blink>WebRTC>Tools
Bugfix	660171	Regression: AppRTC H264 loopback/p2p call video is extremely blotchy	Blink>WebRTC>Video
Bugfix	657217	Move video encode accelerator IPC calls to GPU IO thread	Internals>Media>Codecs
Bugfix	658429	Limit VTVideoEncodeAccelerator's keyframe output	Internals>Media>Codecs
Bugfix	645907	googNoiseReduction constraint broken	---

Native Android/iOS

Type	Issue	Description	Component
Feature	6232	Add software fallback for MediaCodecVideoEncoder.	Mobile
Feature	6302	Add UMA stats for camera start time and resolution it is started in.	Mobile
Feature	6334	Android CameraEventsHandler interface: Add onCameraDisconnected()	Mobile
Feature	6355	AVFoundationVideoCapturer: Enumerate formats using AVCaptureDevice.formats instead of SessionPresets	Mobile
Feature	6386	Allow the initial peak level to be specified in the level controller.	Mobile
Feature	6473	Make it possible to set capture video resolution in iOS AppRTCMobile	Mobile
Feature	6496	Add loopback option and improve the UX of the AppRTCMobile for Mac	Mobile
Feature	6617	Remove unnecessary styling of controls in AppRTCMobile/ios/ARDMainView.m	Mobile
Feature	6618	Add UINavigationController to AppRTCMobile/ios/ARDMainViewController	Mobile
Feature	6619	Add H264VideoToolbox as an external codec instead of hardcoded into static H264Encoder::Create()	Mobile
Feature	6674	Android: Add screenshare support to AppRTCMobile.	Mobile
Bugfix	6476	AudioTransport::NeedMorePlayData called from different threads in OpenSL ES	Audio
Bugfix	6524	Include ScreenCapturerAndroid in libjingle_peerconnection_java.jar.	DesktopCapture
Bugfix	6602	Fix a deadlock in EglRenderer.releaseEglSurface.	Mobile
Bugfix	6643	iOS AppRTCMobile hits DCHECK_CALLED_SEQUENTIALLY in RTPSenderVideo	Mobile
Bugfix	6664	Add improved bwe field trial for iOS	Mobile

Regards

Anatoli

[Philipp Hancke]

\o/ I am particularly happy that 4869 is fixed in the unprefixed peerconnection.

WRT to  rtx: "introduced to mitigate a bug in older Chrome version" -- what old Chrome versions?

There are folks out there with chrome versions that can't even parse UDP/TLS/RTP/SAVPF (which IIRC was fixed in ... Chrome 37)

--

---
You received this message because you are subscribed to the Google Groups "discuss-webrtc" group.
To unsubscribe from this group and stop receiving emails from it, send an email to discuss-webrtc+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/discuss-webrtc/CAM1dXWLeh4VKWOymQi7QKWxi3yc7%2BKD6PYPtx5GdPqPwSzmx1w%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

[Stefan Holmer]

The rtx workaround was introduced with Chrome M50, see release notes here.

[Xander Dumaine]

Yay http://crbug.com/650533!

[Szabó Örs]

Hi,

I came across this article about Chrome 56 to remove supprot of SHA-1 signed certificates: https://www.bleepingcomputer.com/news/security/google-to-show-errors-for-sha1-certificates-starting-with-chrome-56/

Could someone clarify if this is true for WebRTC (DTLS-SRTP) calls as well, i.e. if a gateway will not be able to establish a successful DTLS connection using SHA-1 signed ceritifcates?

Thanks.

Ors 

 

[Philipp Hancke]

Still works for me in M57. And while https://tools.ietf.org/html/draft-ietf-rtcweb-security-arch-12 still says "sha-1" everywhere...

Note that the certificate in WebRTC (rather: dtls-srtp) is used in completly different ways, typically the certificate is self-signed and the fingerprint is validated from the SDP.

--

---
You received this message because you are subscribed to the Google Groups "discuss-webrtc" group.
To unsubscribe from this group and stop receiving emails from it, send an email to discuss-webrtc+unsubscribe@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/discuss-webrtc/d7ad51f7-299e-468d-9f96-d5f2c809485a%40googlegroups.com.

[Harald Alvestrand]

The normative requirement on ciphersuites in -security-arch-12 is this:

   All implementations MUST implement DTLS 1.0, with the cipher suite
   TLS_ECDHE_ECDSA_WITH_AES_128_CBC_SHA with the the P-256 curve
   [FIPS186].  The DTLS-SRTP protection profile
   SRTP_AES128_CM_HMAC_SHA1_80 MUST be supported for SRTP.
   Implementations SHOULD implement DTLS 1.2 with the
   TLS_ECDHE_ECDSA_WITH_AES_128_GCM_SHA256 cipher suite.
   Implementations MUST favor cipher suites which support PFS over non-
   PFS cipher suites and SHOULD favor AEAD over non-AEAD cipher suites.

This indicates that we still have to accept SHA-1 certificates.

But webrtc-pc says:

The explicit certificate management functions provided here are optional. If an application does not provide thecertificates configuration option when constructing an RTCPeerConnection a new set of certificates must be generated by the user agent. That set must include an ECDSA certificate with a private key on the P-256 curve and a signature with a SHA-256 hash.

So if we're spec-compliant (and I read the specs correctly), we'll accept SHA-1, but will always send SHA-256.

To view this discussion on the web visit https://groups.google.com/d/msgid/discuss-webrtc/CADxkKiKYe2ztVPwFh%2BebGT4wVYtKtuWN4fyHwY76T1WUSAiMTw%40mail.gmail.com.