The loadeddata event is fired with our HTMLMediaElement as soons a readyState is >= HAVE_CURRENT_DATA Deciding the value of readyState is based on how the file content being cached or available. This operation is disabled for small files https://hg.mozilla.org/mozilla-central/file/f029ccdee154/dom/html/HTMLMediaElement.cpp#l3799 checks that NextFrameStatus() != MediaDecoderOwner::NEXT_FRAME_AVAILABLE and set readyState to HAVE_CURRENT_DATA. This is the wrong behaviour, as even we if had a decode error and were unable to decode the first frame we will fire the loadeddata event.

We must have at least a decoded frame available to transition to HAVE_ENOUGH_DATA, as otherwise canplay/canplaythrough will always be fired even if the data was invalid so long that it was small.

Comment on attachment 8677890 [details] [diff] [review] P2. Do not transition to HAVE_ENOUGH_DATA readyState until we do have data. got a better idea...

We must have at least a decoded frame available to transition to HAVE_ENOUGH_DATA, as otherwise canplay/canplaythrough will always be fired even if the data was invalid so long that it was small.

Comment on attachment 8677896 [details] [diff] [review] P2. Do not transition to HAVE_ENOUGH_DATA readyState until we do have data. Review of attachment 8677896 [details] [diff] [review]: ----------------------------------------------------------------- ::: dom/html/HTMLMediaElement.cpp @@ +3817,5 @@ > } > return; > } > > + if (!mFirstFrameLoaded) { Might worth adding an assertion somewhere to ensure we only transition to HAVE_CURRENT_DATA or above when mFirstFrameLoaded is true. It will help check if we regress the readyState transition algorithm.

https://hg.mozilla.org/integration/mozilla-inbound/rev/590ee345a6ed https://hg.mozilla.org/integration/mozilla-inbound/rev/d673112518b5

https://hg.mozilla.org/mozilla-central/rev/590ee345a6ed https://hg.mozilla.org/mozilla-central/rev/d673112518b5