Bug 1748852 Comment 3 Edit History

Note: The actual edited comment in the bug view page will always show the original commenter’s name and original timestamp.

Adding this assert uncovers [some interesting fallout](https://treeherder.mozilla.org/jobs?revision=8f6b385c0b0b24c0894ce0309096a078349ca43c&repo=try&selectedTaskRun=UFMcyIdQQ8yFL6KGuNGebQ.0) and raises questions:

1. Should we use IPC_FAIl to [signal a shutdown](https://searchfox.org/mozilla-central/rev/682e5a82d403974bacb779c1db515962d946be48/dom/messagechannel/MessagePortParent.cpp#53-56) ? My guess would be no, as shutdown is not an unexpected state to be in and killing a child actor during shutdown could result in hangs elsewhere.

2. In [` GMPVideoDecoderParent::RecvDecoded`](https://searchfox.org/mozilla-central/rev/682e5a82d403974bacb779c1db515962d946be48/dom/media/gmp/GMPVideoDecoderParent.cpp#283,291,299) inlining obfuscates which of the two is triggering. For the second reading [this comment](https://searchfox.org/mozilla-central/rev/682e5a82d403974bacb779c1db515962d946be48/dom/media/gmp/GMPVideoi420FrameImpl.cpp#57-60) suggests to ignore the frame, which would mean to do nothing and return IPC_OK. The first seems to be again a shutdown case, as [`mCallback` is nulled out in `Close`](https://searchfox.org/mozilla-central/rev/682e5a82d403974bacb779c1db515962d946be48/dom/media/gmp/GMPVideoDecoderParent.cpp#63).

There might be more future intermittents, but those cases should probably be handled before landing.
Adding this assert uncovers [some interesting fallout](https://treeherder.mozilla.org/jobs?revision=8f6b385c0b0b24c0894ce0309096a078349ca43c&repo=try&selectedTaskRun=UFMcyIdQQ8yFL6KGuNGebQ.0) and raises questions:

1. Should we use IPC_FAIl to [signal a shutdown](https://searchfox.org/mozilla-central/rev/682e5a82d403974bacb779c1db515962d946be48/dom/messagechannel/MessagePortParent.cpp#53-56) ? My guess would be no, as shutdown is not an unexpected state to be in and potentially killing a child actor during shutdown could result in hangs elsewhere.

2. In [` GMPVideoDecoderParent::RecvDecoded`](https://searchfox.org/mozilla-central/rev/682e5a82d403974bacb779c1db515962d946be48/dom/media/gmp/GMPVideoDecoderParent.cpp#283,291,299) inlining obfuscates which of the two is triggering. For the second reading [this comment](https://searchfox.org/mozilla-central/rev/682e5a82d403974bacb779c1db515962d946be48/dom/media/gmp/GMPVideoi420FrameImpl.cpp#57-60) suggests to ignore the frame, which would mean to do nothing and return IPC_OK. The first seems to be again a shutdown case, as [`mCallback` is nulled out in `Close`](https://searchfox.org/mozilla-central/rev/682e5a82d403974bacb779c1db515962d946be48/dom/media/gmp/GMPVideoDecoderParent.cpp#63).

There might be more future intermittents, but those cases should probably be handled before landing.

Back to Bug 1748852 Comment 3