The testcase requires approx. 10-15 seconds to take affect; the used media file in the testcase is not required. content/media/webaudio/MediaStreamAudioSourceNode.cpp:19 * NS_INTERFACE_MAP_BEGIN_CYCLE_COLLECTION_INHERITED(MediaStreamAudioSourceNode) The stack shows from frame 0 to frame 37367 the same call: #37367 0x0000000106aefcce in mozilla::dom::MediaStreamAudioSourceNode::QueryInterface (this=0x60e00016f840, aIID=@0x10a266580, aInstancePtr=0x7fff5fbfb880) at /Users/cdiehl/dev/repos/mozilla/mozilla-inbound/content/media/webaudio/MediaStreamAudioSourceNode.cpp:20 The rest of the stack is attached. Tested with http://hg.mozilla.org/integration/mozilla-inbound/rev/3fbe86d67b6a

Roc: assigning this to you because it appears to be a regression from your landing for bug 856361 on inbound. That was backed out so I guess this is in some sense "fixed", but presumably you intend to re-land it and we need to make sure that this is handled. It may or may not be fixed as a side-effect of fixing the testsuite failures that caused the backout. Leaving open for now until you decide how you want to track this.

https://hg.mozilla.org/integration/mozilla-inbound/rev/f7496fddb076 fixed this (before it was backed out). You guys are fast :-).

We'd like to get an ESR24 patch for this since it is a sec-critical and people running ESR may change the pref and turn on webaudio.

Ah, Dveditz says this is a 25 and higher only regression so we won't need it for ESR24.