[webkit-gtk] extremely slow reflow and/or painting on certain pages (due to xorg xserver 1.16?)

Adrian Perez de Castro aperez at igalia.com
Wed Sep 17 05:45:52 PDT 2014

Yosef Or Boczko <yoseforb at gnome.org> writes:

> Here the bug report about this bug (DRI3 broken on intel driver):
> https://bugs.freedesktop.org/show_bug.cgi?id=81623
> For me, in ArchLinux, it look like it fixes.
> For now, you can to run with LIBGL_DRI3_DISABLE=1 .

Right, I am using ArchLinux, too.

This post explains that the DRI3 support has been removed at build time
From the Intel video driver:


A very quick, dirty hack would be to set the “LIBGL_DRI1_DISABLE=1”
environment variable before WebKitGTK spawns a “WebKitWebProcess”,
then remove the hack later when drivers that supports DRI3 are known
to work properly — or just let users define the environment variable
themselves, I can imagine more cases in which DRI3 could make things
break for users.



> (I'm not sure what fixed this).
> Regards,
> Yosef Or Boczko
> בתאריך ד', ספט 17, 2014 בשעה 2:06 PM, Adrian Perez de 
> Castro <aperez at igalia.com> כתב:
>> Michael Gratton <mike at vee.net> writes:
>>>  I'm running Ephy 3.12.1 w/ WebKitGtk 2.4.5 on the current Ubuntu 
>>>  beta. Recently, certain pages started updating *really* slowly, such
>>>  that it takes a second or two before the effects of scrolling or key
>>>  presses are seen. Duck Duck Go's search interface
>>>  <https://duckduckgo.com/> and results lists are two examples. [...]
>> Here I also remember DuckDuckGo being sluggish for a while, but the
>> thing seems to have solved itself.
>>>  It's hard to pin down exactly when the issue started, but I think 
>>> it is
>>>  related to upgrading to xorg's xserver 1.16.
>> I am running Xorg server version 1.16.0, so I suspect the Intel video
>> driver can be a better guess, because the Intel driver and mesa got
>> several updates for me in the last weeks. I am using version 2.99.916,
>> so the issue must be somewhere in the versions of the Intel driver in
>> between versions 2.99.910 and 2.99.916
>> As a recap, we have the following versions tested:
>>   Xorg     Intel driver  Slowness?
>>   ======== ============= =========
>>   1.15     2.99.910      No
>>   1.16     2.99.914      Yes
>>   1.16     2.99.916      No
>> If this is a problem caused by the driver, probably it won't be worth
>> it to provide a workaround in WebKitGTK (and JFTR, it may not be even
>> possible to workaround).
>> Cheers,
>> --
>>  ☺ Adrian
>> _______________________________________________
>> webkit-gtk mailing list
>> webkit-gtk at lists.webkit.org
>> https://lists.webkit.org/mailman/listinfo/webkit-gtk

 ☺ Adrian
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 180 bytes
Desc: not available
URL: <https://lists.webkit.org/pipermail/webkit-gtk/attachments/20140917/90bce37a/attachment.sig>

More information about the webkit-gtk mailing list