The issue specified in 762282 comment 1 is fixed by bug 1001317. <=== - mDidNotifyDataEnded is flipped to true, which prevent subsequent call of this method to call ChannelMediaResource::CacheClientNotifyDataEnded, which prevent calling nsHTMLMediaElement::ResourceLoaded, and that does not change the networkState and does not fire the last progress, so the buffered bar does not update. ===> We should revert the change in bug 762282. Otherwise it is possible to download and cache the whole file during the execution of OggReader::RangeEndTime() before MediaResource::EndSeekingForMetadata() is called. Then download end event will not be notified and the media element will stay in NETWORK_LOADING. Since the whole file is already cached, subsequent access to the media resource will not trigger HTTP channel activities and thus no download end event at all. This bug results in timeout of test_playback.html which expect both "ended" and "suspend" events to finish the test.  http://hg.mozilla.org/mozilla-central/file/af6c928893c0/content/media/test/test_playback.html#l48  http://hg.mozilla.org/mozilla-central/file/af6c928893c0/content/media/test/test_playback.html#l62
Hi roc, Does comment 0 make sense to you?
(In reply to JW Wang [:jwwang] from comment #1) > Does comment 0 make sense to you? Yes.
Created attachment 8500208 [details] [diff] [review] 1076691_revert_bug_762282.patch Per comment 0, remove the changes in bug 762282. Try: https://tbpl.mozilla.org/?tree=Try&rev=10721aa24fdd https://tbpl.mozilla.org/?tree=Try&rev=b037132b3777 I guess those oranges have been there for a while.