This bug was filed from the Socorro interface and is report bp-42bec3a2-35ea-45c7-99aa-fde182130415 . ============================================================= I can reproduce this crash very reliably by running the emscripten browser tests, python tests/runner.py browser The crash happens in a different test each time, I guess one of the web audio tests causes the crash some time after its tab is closed and it is cleaned up.

Ah yeah, Context() is probably wrong at that point, or something.

kripken: thanks, I could repro without problems.

Comment on attachment 737415 [details] [diff] [review] Nullcheck when unregistering the pannernode. r= Review of attachment 737415 [details] [diff] [review]: ----------------------------------------------------------------- r=me I guess, but this fix is already on inbound. ::: content/media/AudioStream.cpp @@ +65,5 @@ > // value, the longer stream volume changes will take to become > // audible. > uint32_t value = Preferences::GetUint(aPref, 100); > MutexAutoLock lock(*gAudioPrefsLock); > + gCubebLatency = std::min<uint32_t>(std::max<uint32_t>(value, 0), 1000); Trying to sneak in this change here? ;-) ::: content/media/webaudio/PannerNode.cpp @@ +181,5 @@ > PannerNode::~PannerNode() > { > + if (Context()) { > + Context()->UnregisterPannerNode(this); > + } I've already fixed this in my patch to bug 853298, currently on inbound.