Tested with https://hg.mozilla.org/integration/mozilla-inbound/rev/38cd6935fc19

Could you retest this next time you're running an ASAN build? Thanks - I'm hoping it's fixed

Program received signal SIGSEGV, Segmentation fault. [Switching to Thread 7418.7575] 0x00007ff3bb61d91a in mozilla::dom::AudioBufferSourceNodeEngine::CopyFromInputBufferWithResampling (this=0x7ff2f2b7b680, aOutput=0x7ff30508b190, aChannels=1, aOffsetWithinBlock=0x7ff317ffa690, aAvailableInOutput=128, aCurrentPosition=0x7ff317ffa698, aBufferMax=16) at /mnt/ssd1/karl/moz/dev/dom/media/webaudio/AudioBufferSourceNode.cpp:274 (gdb) p /x skipFracNum $1 = 0xdc097ed5 (gdb) p inputLatency $2 = 4 (gdb) p leadTicks $3 = 13681.575461263204 (gdb) p ratioNum $4 = 44100 (gdb) p ratioDen $5 = 2147483647 (gdb) p /x ratioDen $6 = 0x7fffffff skipFracNum was 0xfffffffc before leadTicks * ratioNum was subtracted. ratioDen is the maximum, for the very slow playbackRate. I'll look at changing the assertion.

This assertion no longer fails with this testcase, but there is still overflow in calculating skipFracNum from large ratioDen, which I'll address separately.

This causes even the new assert to fail.

ratioDen can be large when playbackRate is low. Subsample skipping is limited to uint32_t values supported by speex resampler.

Correct comment and s/max/min/.

https://hg.mozilla.org/integration/mozilla-inbound/rev/67c94e508c2e https://hg.mozilla.org/integration/mozilla-inbound/rev/364c352a9ac9 https://hg.mozilla.org/integration/mozilla-inbound/rev/4e57a9fb9a25

https://hg.mozilla.org/mozilla-central/rev/67c94e508c2e https://hg.mozilla.org/mozilla-central/rev/364c352a9ac9 https://hg.mozilla.org/mozilla-central/rev/4e57a9fb9a25