[webkit-qt] Inconsistent acceleration of linear gradients and atlas maps

[Jonathan Morton]

On Mon, 2010-07-12 at 21:53 +0200, benjamin.poulain at nokia.com wrote:
> You should start by adding a reduction to
> https://bugs.webkit.org/show_bug.cgi?id=41682 in order to get people
> interested.

We're now trying to make a test page which reproduces the problem
independently of the AOL page.  This should turn out a lot simpler than
the test case presently attached, and will probably not depend on the
atlas map being infeasibly large.

> You also mentioned you modified Qt, but you do not contribute the code
> back to Qt. So people have now way reproduce the problem you mention,
> and some contributors probably feel you take code but don't contribute
> back.

Attached are the patches we're using.  They're against Qt 4.6.2.  As
previously mentioned, there are minor problems with these patches,
mostly in ConstantOpacity support which can be bypassed by setting
QT_X11_NO_CONSTANT_OPACITY in the environment.

The key thing we're looking for is the software fallbacks, which have
relatively little performance impact on desktop platforms but which are
catastrophically slow on our device.  The best way to spot them is to
watch for GetImage calls in xtrace output.  (The xtrace that is an X11
protocol logger, not the low-level debugging tool.)

-- 
------
From: Jonathan Morton
      jonathan.morton at movial.com

-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0002-XRender-add-gradient-and-transform.patch
Type: text/x-patch
Size: 13546 bytes
Desc: 
URL: <http://lists.webkit.org/pipermail/webkit-qt/attachments/20100713/470915ec/attachment-0002.bin>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0003-Add-remainder-of-ConstantOpacity-support.patch
Type: text/x-patch
Size: 5911 bytes
Desc: not available
URL: <http://lists.webkit.org/pipermail/webkit-qt/attachments/20100713/470915ec/attachment-0003.bin>