It first showed up in 25.0a1/20130711. The regression range might be (low volume): http://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=04d8c309fe72&tochange=dde4dcd6fa46 It's likely a regression from bug 866675. Signature _VEC_memcpy | mozilla::BufferedAudioStream::GetUnprocessed(void*, long) More Reports Search UUID e53ccc4c-2567-43d1-88f2-e3de12130731 Date Processed 2013-07-31 19:53:40.928182 Uptime 516 Last Crash 520 seconds before submission Install Age 757 since version was first installed. Install Time 2013-07-31 19:40:56 Product Firefox Version 25.0a1 Build ID 20130731030203 Release Channel nightly OS Windows NT OS Version 6.1.7601 Service Pack 1 Build Architecture x86 Build Architecture Info GenuineIntel family 6 model 37 stepping 5 | 4 Crash Reason EXCEPTION_ACCESS_VIOLATION_WRITE Crash Address 0x209fa000 App Notes Cisco VPN AdapterVendorID: 0x10de, AdapterDeviceID: 0x0a6c, AdapterSubsysID: 172b103c, AdapterDriverVersion: 8.17.12.6128 D2D? D2D+ DWrite? DWrite+ D3D10 Layers? D3D10 Layers+ Frame Module Signature Source 0 msvcr100.dll _VEC_memcpy 1 xul.dll mozilla::BufferedAudioStream::GetUnprocessed(void *,long) content/media/AudioStream.cpp 2 xul.dll mozilla::BufferedAudioStream::DataCallback(void *,long) content/media/AudioStream.cpp 3 xul.dll mozilla::BufferedAudioStream::DataCallback_S(cubeb_stream *,void *,void *,long) content/media/AudioStream.cpp 4 gkmedias.dll `anonymous namespace'::refill_with_resampling(cubeb_stream *,float *,long) media/libcubeb/src/cubeb_wasapi.cpp 5 gkmedias.dll `anonymous namespace'::wasapi_stream_render_loop media/libcubeb/src/cubeb_wasapi.cpp 6 msvcr100.dll _callthreadstartex f:\\dd\\vctools\\crt_bld\\self_x86\\crt\\src\\threadex.c 7 msvcr100.dll _threadstartex f:\\dd\\vctools\\crt_bld\\self_x86\\crt\\src\\threadex.c 8 kernel32.dll BaseThreadInitThunk 9 ntdll.dll __RtlUserThreadStart 10 ntdll.dll _RtlUserThreadStart More reports at: https://crash-stats.mozilla.com/report/list?product=Firefox&signature=_VEC_memcpy+|+mozilla%3A%3ABufferedAudioStream%3A%3AGetUnprocessed%28void*%2C+long%29 https://crash-stats.mozilla.com/report/list?product=Firefox&signature=memcpy+|+mozilla%3A%3ABufferedAudioStream%3A%3AGetUnprocessed%28void*%2C+long%29

So, this is because we call the refill callback with an invalid address for the buffer: we segfault on the first memcpy when writing to the buffer. My theory so far is that I screwed up the shutdown sequence somehow.

I can't find any crashes with this stack in the last 4 weeks, is it possible we fixed this in another bug? Or is it just *really* low volume?

We don't have BufferedAudioStream anymore.