[Webkit-unassigned] [Bug 161605] [GTK] Crash of WebProcess on the last WebView disconnect

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=161605

--- Comment #13 from Michael Catanzaro <mcatanzaro at igalia.com> ---
(In reply to comment #6)
> This is not accurate. PlatformDisplay is not deleted in an exit handler,
> it's a static std::unique_ptr that is deleted after all exit handlers.

Where do you think static objects get deleted. :)

> Never liked that, TBH, I was tempted to remove it when reworked the
> GLContext. It's probably a workaround for a bug in nvidia drivers that might
> have already been fixed. In any case, all contexts created now except the
> sharing context that is owned by the PlatformDisplay should be freed already
> on exit.

Agreed.

> This is not possible because the GLContext exit handler is a workaround used
> on for X11 contexts, and other classes shouldn't rely on that.

Good point.

> > create a throwaway GLContext the
> > first time PlatformDisplay::initializeEGLDisplay is called to ensure the
> > GLContext exit handler gets registered first (and therefore executed last at
> > exit time)... let's see if Carlos Garcia has an opinion on what's best.
> 
> This has nothing to do with EGLDisplay, the crash is happening on
> ~PlatformDisplay and the exit handler of the EGLDisplay doesn't delete the
> PlatformDisplay.

Yes, I was completely wrong; for some reason I thought we were deleting the PlatformDisplay in the exit handler registered in that function, but clearly that's not the case.

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.webkit.org/pipermail/webkit-unassigned/attachments/20160907/ee42a36a/attachment.html>