Steps to repro: 1. Start mini browser 2. Go to google.com 3. Search for something, for example I entered "autosport". 4. Click a hit. 5. Go back. At this point google.com is indicated in the location bar but nothing is rendered. A bit later mini browser will likely crash.
Assigning to Chris since my bisecting seems to point to https://bugs.webkit.org/show_bug.cgi?id=191781. First of all I would be interested if it is reproducible or just my system/setup though.
Will look, thanks.
Do you process "Process Swap on Cross-Site navigation" enabled in experimental features?
(In reply to Chris Dumez from comment #3) > Do you process "Process Swap on Cross-Site navigation" enabled in > experimental features? Do you *have* "Process Swap on Cross-Site navigation" enabled in experimental features?
> A bit later mini browser will likely crash. Crash log? I wasn't able to reproduce in Safari. Will try Minibrowser soon.
Ok, I was able to reproduce in Minibrowser.
I see a DecidePolicyForResponse sent why the about-to-be-suspended process to the SuspendedPageProxy but no DidFinishLoadForFrame.
(In reply to Chris Dumez from comment #7) > I see a DecidePolicyForResponse sent why the about-to-be-suspended process > to the SuspendedPageProxy but no DidFinishLoadForFrame. In Safari, I get no DecidePolicyForNavigationResponse and I get a DidFinishProgress IPC.
DocumentLoader::maybeLoadEmpty() returns false in Minibrowser but true in Safari. It returns false because SchemeRegistry::shouldLoadURLSchemeAsEmptyDocument(m_request.url().protocol().toStringWithoutCopying())) check fails
Created attachment 355493 [details] Patch
Comment on attachment 355493 [details] Patch Clearing flags on attachment: 355493 Committed r238456: <https://trac.webkit.org/changeset/238456>
All reviewed patches have been landed. Closing bug.
<rdar://problem/46217302>
(In reply to Chris Dumez from comment #3) > Do you process "Process Swap on Cross-Site navigation" enabled in > experimental features? I did have that enabled. Thanks for fixing!