See [1] and bug 1284102 for evidence of these races. I believe the race comes from [2]. The test waits for the same event that triggers this code, but this code needs another async event to update the state we are about to check. Florian might know the details better. [1] http://searchfox.org/mozilla-central/rev/8910ca900f826a9b714607fd23bfa1b37a191eca/browser/base/content/test/webrtc/head.js#91 [2] http://searchfox.org/mozilla-central/rev/8910ca900f826a9b714607fd23bfa1b37a191eca/browser/modules/ContentWebRTC.jsm#310

Comment from Florian on IRC on what should be done: > [10:21:38] <flo-retina> in the future I think it would be nice to investigate if both of these > timer-based hacks could be removed and replaced with something listening for the > webrtc:UpdateBrowserIndicators message in the parent process.