[Webkit-unassigned] [Bug 174805] [GTK][WPE] Eats 100% CPU when starting a new WebProcess fails

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=174805

--- Comment #3 from Bastien Nocera <bugzilla at hadess.net> ---
(In reply to Bastien Nocera from comment #2)
> > We need some general limit on how many times the web process can be launched in a row. If it's failed to launch, say, five times in under one second, it's time to just give up. This is really unfortunate, as there's no way to report an error if we can't display an error page in the web process. In this case, I'd say it's sufficiently-fatal that the UI process should probably crash with SIGABRT or something.
> 
> I'd really rather you didn't, as the error is recoverable (I closed a few
> apps and could open a new tab to file this bug), and you could definitely
> swap out the WebKit error page for something native client side.

The problem happened again, and I was seeing this white page. I closed a web app to free up some RAM, and epiphany immediately showed the "there's a problem" page. It tried :)

> I don't know how much RAM a new WebKitProcess needs but I seem to remember
> it pre-allocating at least 1GB for the Javascript stack. That should
> definitely be enough to show a GTK+ native error message.

-- 
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/20170725/58ee5152/attachment-0001.html>