[Webkit-unassigned] [Bug 186545] PSON: http/tests/resourceLoadStatistics/classify-as-prevalent-based-on-top-frame-redirect-collusion.html ASSERTS with process swap enabled

bugzilla-daemon at webkit.org bugzilla-daemon at webkit.org
Mon Jun 11 16:39:07 PDT 2018


https://bugs.webkit.org/show_bug.cgi?id=186545

--- Comment #6 from Chris Dumez <cdumez at apple.com> ---
(In reply to Brady Eidson from comment #5)
> (In reply to Chris Dumez from comment #4)
> > (In reply to Brady Eidson from comment #3)
> > > (In reply to Brady Eidson from comment #2)
> > > > ResourceLoadStatistics is fundamentally broken in PSON
> > > > 
> > > > We either need to move a statistics map from the old process to the new one,
> > > > or just host statistics in the UIProcess.
> > > 
> > > Chris tells me that RLS are hosted in the UI process, but driven by data
> > > pushed out from WebProcesses. This test failure a case of a new WebProcess
> > > not having what is expected of it!
> > 
> > The test is relying on a test runner API to force the WebProcess to send its
> > RLS to the UIProcess:
> > 6   0x10675f049 WKBundleResourceLoadStatisticsNotifyObserver
> > 7   0x645f189e8 WTR::InjectedBundle::statisticsNotifyObserver()
> > 8   0x645f4f399 WTR::TestRunner::statisticsNotifyObserver()
> > 
> > Because of process swap, it is asking to new WebProcess instead of the old
> > one, we hit an assertion because we're not supposed to notify the client of
> > new statistics with no statistics.
> 
> Yup, and when I simply make the ASSERT a no-op instead the test passes.
> 
> /me shrugs

The test is marked as flaky, which is why it may have passed from you locally. However, I do think we would break the test here. We need the previous WebProcess do send its RLS to the UIProcess.

We may want to update the testRunner API to ask all WebProcesses to send their data to the UIProcess, instead of merely the current one.

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.webkit.org/pipermail/webkit-unassigned/attachments/20180611/b845a936/attachment.html>


More information about the webkit-unassigned mailing list