[Webkit-unassigned] [Bug 196062] Certain WebProcesses should opt-out of the freezer

bugzilla-daemon at webkit.org bugzilla-daemon at webkit.org
Thu Mar 21 13:41:08 PDT 2019


https://bugs.webkit.org/show_bug.cgi?id=196062

--- Comment #5 from Chris Dumez <cdumez at apple.com> ---
(In reply to Chris Dumez from comment #4)
> (In reply to Brady Eidson from comment #3)
> > (In reply to Chris Dumez from comment #2)
> > > (In reply to Brady Eidson from comment #1)
> > > > (In reply to Brady Eidson from comment #0)
> > > > > Certain WebProcesses should opt-out of the freezer
> > > > > 
> > > > > These include:
> > > > > -Prewarmed web processes
> > > > > -Processes in the per-domain-process cache
> > > > > -Processes in the back/forward cache list
> > > > 
> > > > Actually that last one is not true. It's "processes that *only host* pages
> > > > in the back/forward cache list" and such processes should also already be in
> > > > the per-domain-process-cache.
> > > > 
> > > > Yay.
> > > 
> > > Processes with suspended pages are NOT in the WebProcess cache, even when
> > > they have no WebPage on the WebProcess side.
> > 
> > This is confusing. You're saying a process that is hosting a suspended
> > webkit.org page is not in the cache and therefore will not be chosen to host
> > other webkit.org pages?
> > 
> > Why not?
> 
> A suspended page processes may get used for navigations if their domain
> matches but they are not in the WebProcessCache. We check both suspended
> pages and the WebProcessCache to find a suitable process.
> In the WebProcessCache, we currently only put processes that were about to
> shutdown (i.e. because they have no pages / suspendedPages).


Note that when we decide to use a suspended page's process for a forward navigation, we often have to destroy the suspended page, this is a destructive optimization.

It is not exactly the same thing as process caching IMO.

-- 
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/20190321/e3488256/attachment-0001.html>


More information about the webkit-unassigned mailing list