Bug 214730 – REGRESSION (r264492): Check failed: sink_by_mid_.empty() on 2 webrtc tests, flakily

• Status: NEW
• Component: New Bugs

[Karl Rackler]

REGRESSION (r264492): [ iOS debug wk2 ] Broke 2 webrtc tests that are flaky crash

Description:
imported/w3c/web-platform-tests/webrtc/RTCSctpTransport-constructor.html
webrtc/simulcast-h264.html

This started approximately 07/17/20 and is flaky crash since.

It looks like the changes in https://trac.webkit.org/changeset/264492/webkit caused these tests to crash.

History:
https://results.webkit.org/?platform=ios&limit=50000&flavor=wk2&style=debug&suite=layout-tests&suite=layout-tests&test=imported%2Fw3c%2Fweb-platform-tests%2Fwebrtc%2FRTCSctpTransport-constructor.html&test=webrtc%2Fsimulcast-h264.html

Crash Log:
Test: imported/w3c/web-platform-tests/webrtc/RTCSctpTransport-constructor.html
No crash log found for WebKitTestRunnerApp:10592.

stdout:

stderr:


#
# Fatal error in: /Volumes/Data/slave/ios-simulator-13-debug/build/Source/ThirdParty/libwebrtc/Source/webrtc/call/rtp_demuxer.cc, line 49
# last system error: 316
# Check failed: sink_by_mid_.empty()
# #CRASHED - com.apple.WebKit.WebContent.Development (pid 15054)
LEAK: 1 WebProcessPool
LEAK: 1 WebPageProxy
-------------------------
Test: webrtc/simulcast-h264.html
No crash log found for WebKitTestRunnerApp:15453.

stdout:

stderr:
(used_ids.h:55): Duplicate id found. Reassigning from 102 to 127
(used_ids.h:55): Duplicate id found. Reassigning from 103 to 125
(used_ids.h:55): Duplicate id found. Reassigning from 105 to 124
(used_ids.h:55): Duplicate id found. Reassigning from 1 to 14
(used_ids.h:55): Duplicate id found. Reassigning from 3 to 13
(used_ids.h:55): Duplicate id found. Reassigning from 5 to 12
(used_ids.h:55): Duplicate id found. Reassigning from 6 to 11
(webrtc_video_engine.cc:3160): Absent receive stream; ignoring clearing encoded frame sink for ssrc 0
(webrtc_video_engine.cc:3160): Absent receive stream; ignoring clearing encoded frame sink for ssrc 0
(webrtc_video_engine.cc:3160): Absent receive stream; ignoring clearing encoded frame sink for ssrc 0
(jsep_transport_controller.cc:349): Not adding candidate because the JsepTransport doesn't exist. Ignore it.
(jsep_transport_controller.cc:349): Not adding candidate because the JsepTransport doesn't exist. Ignore it.
(jsep_transport_controller.cc:349): Not adding candidate because the JsepTransport doesn't exist. Ignore it.
(jsep_transport_controller.cc:349): Not adding candidate because the JsepTransport doesn't exist. Ignore it.
(dtls_transport.cc:569): DtlsTransport[0|1|__]: Packet received before we know if we are doing DTLS or not.
(balanced_degradation_settings.cc:96): Unsupported size, value ignored.
(rtp_video_sender.cc:112): Only RED or only ULPFEC enabled, but not both. Disabling both.
(rtp_video_sender.cc:112): Only RED or only ULPFEC enabled, but not both. Disabling both.
(rtp_video_sender.cc:112): Only RED or only ULPFEC enabled, but not both. Disabling both.
(video_send_stream_impl.cc:243): ERROR: Initial encoder max bitrate = -1 which is <= 0!
(simulcast.cc:242): Reducing simulcast layer count from 3 to 2
(helpers.cc:135): VTSessionSetProperty failed to set: DataRateLimits to array value: -12902, vcpSession: 0


#
# Fatal error in: /Volumes/Data/slave/ios-simulator-13-debug/build/Source/ThirdParty/libwebrtc/Source/webrtc/sdk/objc/components/video_codec/RTCVideoEncoderH264.mm, line 428
# last system error: 0
# Check failed: frame.width == _width (640 vs. 320)
# #CRASHED - com.apple.WebKit.WebContent.Development (pid 17127)
LEAK: 1 WebProcessPool
LEAK: 1 WebPageProxy

[Radar WebKit Bug Importer]

<rdar://problem/66047260>

[Karl Rackler]

I have marked these tests as skip while this issue is investigated.
https://trac.webkit.org/changeset/264817/webkit

[Karl Rackler]

This affects both iOS and Mac debug wk2.

Appears to have the same results as Mac. https://bugs.webkit.org/show_bug.cgi?id=214478

It looks like the changes in https://trac.webkit.org/changeset/264492/webkit caused these tests to crash.

Adding iOS debug wk2 test expectation, similar to https://trac.webkit.org/changeset/264520/webkit since skipping one causes the next to crash.

I have marked these tests as skip while this issue is investigated.
https://trac.webkit.org/changeset/264932/webkit

[Karl Rackler]

One more test added.

I have marked these tests as skip while this issue is investigated.
https://trac.webkit.org/changeset/264979/webkit