[Webkit-unassigned] [Bug 201507] [GTK] Accelerated compositing is disabled due to failure in EGL display creation

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=201507

--- Comment #18 from Carlos Garcia Campos <cgarcia at igalia.com> ---
(In reply to Michael Catanzaro from comment #16)
> (In reply to Michael Catanzaro from comment #3)
> > OK I have a better reproducer! Visit https://q13fox.com/ and the page will
> > crash immediately.
> 
> What I didn't realize when I added that comment is the bug is only
> reproducible once Epiphany gets into the "bad state" such that bug #202362
> also occurs. Then it will crash 100% until Epiphany is restarted. But if
> you're not yet in this bad state, it works fine.

The fact that doesn't always happen, and it starts happening after a while, makes me think it's not a problem with the EGL config, because it's always the same. It could be something like OOM or that you run out of file descriptors or something like that.

> (In reply to Carlos Garcia Campos from comment #8)
> > about:gpu info is still useful in any case
> 
> I can add a patch to our runtime to add about:gpu, if you want to provide a
> patch that builds against 2.26.1.
> 
> I won't update the runtime to 2.27.1 due to (a) the GitLab CI regression, it
> will become impossible to use Epiphany to develop Epiphany, and (b) the MSE
> regressions, I like to watch YouTube. So I want to keep Tech Preview on 2.26
> for now.
> 
> (In reply to Carlos Garcia Campos from comment #14)
> > Michael, it would help if you could add printfs to
> > PlatformDisplayLibWPE::initialize() to show the hostFd and the native
> > display returned by wpe_renderer_backend_egl_get_native_display() as a
> > pointer.
> 
> This is more work than just adding a patch locally though. A local build
> isn't good enough because this bug is not reproducible; we need the patch in
> the real Tech Preview build for the debug to be available next time I notice
> the issue.
> 
> So I will add a debug patch to the runtime tomorrow. Please confirm that you
> still need it to print (a) hostFd, and (b) native display as a pointer, and
> (c) nothing else. If more would be helpful, now is the best time to add it
> because it would be nice to update the patch as few times as possible.

No, I don't really need that, because it also happens when initializing the main display, so it's not related to wpe renderer.

-- 
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/20191009/0ab77123/attachment.html>