Files
sousa-gecko/testing/web-platform/meta/webrtc/RTCPeerConnection-videoDetectorTest.html.ini
T
Daniel Baker 18f6f8603f Bug 2050179 - Fix perma-TIMEOUT of RTCPeerConnection-videoDetectorTest on Android r=webrtc-reviewers,bwc
Add per-test prefs on Android to disable hardware video decoders
(mediadatadecoder_h264_enabled, mediadatadecoder_vpx_enabled). Hardware
decoders produce SurfaceTextureImage frames; once the compositor GL context
attaches to the SurfaceTexture, GetAsSourceSurface() from another context
fails silently, causing canvas drawImage() to produce black pixels and the
signal detector to never fire. Software decoders (libvpx/GMP) produce
PlanarYCbCrImage frames that GetAsSourceSurface() can always read.

Also remove the [TIMEOUT, OK] expected results since the test now passes
with software decoders.

Differential Revision: https://phabricator.services.mozilla.com/D309971
2026-07-07 18:37:29 +00:00

17 lines
909 B
INI

[RTCPeerConnection-videoDetectorTest.html]
bug: https://bugzilla.mozilla.org/show_bug.cgi?id=1526207
prefs:
# Hardware video decoders on Android produce SurfaceTextureImage frames
# that cannot be read back via canvas drawImage once the compositor has
# attached to the SurfaceTexture GL context. Use software decoders so
# GetAsSourceSurface() works. Bug 2050179.
if os == "android": [media.navigator.mediadatadecoder_h264_enabled:false, media.navigator.mediadatadecoder_vpx_enabled:false]
expected:
if (os == "linux") and debug and fission: [OK, TIMEOUT]
if (os == "linux") and debug and not fission: [OK, CRASH]
if (os == "mac") and (processor == "aarch64"): TIMEOUT
[Signal detector detects track change within reasonable time]
bug: https://bugzilla.mozilla.org/show_bug.cgi?id=1526207
expected:
if (os == "mac") and (processor == "aarch64"): TIMEOUT