I noticed two bugs during testing. One of them is that the vsync labeling has always been broken. There's a missing QI which means that we were not actually labeling vsyncs. The other problem is that, since docshells start out in the active state, we need to treat new TabChilds as active.
Attachment #8922637 - Flags: review?(btseng)
Attachment #8922637 - Flags: review?(btseng) → review+
Pushing this to try exposed a bug where a caller was using nsDocShell::SetIsActive directly in the child process. Consequently, we were missing the change in the active state, so sActiveTabs was wrong. The caller is this one: http://searchfox.org/mozilla-central/rev/21363323fd4aa21db074c808fb5358a46df6d698/toolkit/components/thumbnails/content/backgroundPageThumbsContent.js#172
Attachment #8923027 - Flags: review?(btseng)
Attachment #8923027 - Flags: review?(btseng) → review+
Pushed by email@example.com: https://hg.mozilla.org/integration/mozilla-inbound/rev/8d8519a2b6dc Track all changes to TabChild active state (r=bevis)
You need to log in before you can comment on or make changes to this bug.