ProcessSwap.ReuseSuspendedProcessLoadHTMLString API test is flaky:
Expected equality of these values:
Which is: 70197
Which is: 70198
I have trouble reproducing on my first machine, will try on another machine.
Flakiness was indeed introduced by r244144.
I see SuspendedPageProxy::pageEnteredAcceleratedCompositingMode() getting called and it destroys the SuspendedPageProxy because m_suspensionState == SuspensionState::FailedToSuspend.
Suspension likely fails because there is no HistoryItem in WebContent process side. I believe failure to suspend is expected here.
Depending on how fast SuspendedPageProxy::pageEnteredAcceleratedCompositingMode() gets called, the SuspendedPageProxy may or may not still be there when we need it for the navigation.
Created attachment 367176 [details]
Created attachment 367178 [details]
Comment on attachment 367178 [details]
Clearing flags on attachment: 367178
Committed r244186: <https://trac.webkit.org/changeset/244186>
All reviewed patches have been landed. Closing bug.