https://treeherder.mozilla.org/logviewer.html#?job_id=4078181&repo=fx-team 05:59:38 WARNING - gtest TEST-UNEXPECTED-FAIL | gtest | test failed with return code 11 05:54:38 INFO - TEST-START | ImageDecodeToSurface.PNG 5:54:38 INFO - Program /builds/slave/fx-team-lx-d-00000000000000000/build/src/obj-firefox/dist/bin/firefox (pid = 3003) received signal 11. 05:54:38 INFO - Stack: 05:54:38 INFO - #01: ???[/builds/slave/fx-team-lx-d-00000000000000000/build/src/obj-firefox/dist/bin/gtest/libxul.so +0x4930de] 05:54:38 INFO - #02: ???[/builds/slave/fx-team-lx-d-00000000000000000/build/src/obj-firefox/dist/bin/gtest/libxul.so +0x4931a6] 05:54:38 INFO - #03: ???[/builds/slave/fx-team-lx-d-00000000000000000/build/src/obj-firefox/dist/bin/gtest/libxul.so +0x5d8ff2] 05:54:38 INFO - #04: ???[/builds/slave/fx-team-lx-d-00000000000000000/build/src/obj-firefox/dist/bin/gtest/libxul.so +0x5d8e1d] 05:54:38 INFO - #05: ???[/builds/slave/fx-team-lx-d-00000000000000000/build/src/obj-firefox/dist/bin/gtest/libxul.so +0x600f7b] 05:54:38 INFO - #06: ???[/builds/slave/fx-team-lx-d-00000000000000000/build/src/obj-firefox/dist/bin/gtest/libxul.so +0x88daa9] 05:54:38 INFO - #07: ???[/builds/slave/fx-team-lx-d-00000000000000000/build/src/obj-firefox/dist/bin/gtest/libxul.so +0x86d98e] 05:54:38 INFO - #08: ???[/builds/slave/fx-team-lx-d-00000000000000000/build/src/obj-firefox/dist/bin/gtest/libxul.so +0x86d9b4] 05:54:38 INFO - #09: ???[/builds/slave/fx-team-lx-d-00000000000000000/build/src/obj-firefox/dist/bin/gtest/libxul.so +0x5d9143] 05:54:38 INFO - #10: ???[/builds/slave/fx-team-lx-d-00000000000000000/build/src/obj-firefox/dist/bin/libnspr4.so +0x2f365] 05:54:38 INFO - #11: ???[/lib/libpthread.so.0 +0x6a09] 05:54:38 INFO - Sleeping for 300 seconds.

seth: could you take a look, thanks!

Loving the fact that the linux failures don't have symbols and the OS X failures don't even print stacks.

Ryan provided this ASAN failure which should help track down the issue: https://treeherder.mozilla.org/logviewer.html#?job_id=12635283&repo=mozilla-inbound

(In reply to Seth Fowler [:seth] [:s2h] from comment #13) > Ryan provided this ASAN failure which should help track down the issue: Any updates here?

A new variant of this recently popped up: https://treeherder.mozilla.org/logviewer.html#?job_id=14389424&repo=mozilla-inbound I am suspicious that the cause is that we are releasing a SourceSurface off-main-thread, which is done with a runnable (i.e., the release *actually* happens on the main thread, later). Gtests are a fairly weird environment and I'm not convinced that we're handling that correctly. I just pushed a patch that may fix the problem by ensuring that we explicitly release SourceSurfaces on the main thread in this test.

Comment 47 actually happened before the patch in comment 45 landed. (I just realized I didn't upload the patch to this bug; I'll do that.) So right now it looks like the issue hasn't reoccurred since the patch landed. Since it's intermittent, this bug may need to be reopened, but I'm hopeful that it's fixed.

Here's the patch that I landed previously.

Oh, hey, I bet I'd have an easier time finding this to star the ongoing failures (which are, to be sure, much less frequent) if I reopened it, wouldn't I?

It looks an awful lot like this got fixed at some point.