[Pixman] [PATCH 1/4] bits: Implement PAD support in the simple fetcher

Ben Avison bavison at riscosopen.org
Thu Jan 31 06:49:42 PST 2013


On Thu, 31 Jan 2013 05:21:52 -0000, Jeff Muizelaar <jmuizelaar at mozilla.com> wrote:
> On 2013-01-30, at 9:14 PM, Søren Sandmann wrote:
>> This might be a good idea, though it's still worth finding out what is
>> causing an untransformed image with PAD repeat to be drawn without a
>> mask onto a destination bigger than source image.
>
> I've I had to guess, I'd guess that this is unintentional.

If it helps, I just tried dumping out the source coordinates for this
composite type in the cairo firefox-chalkboard trace. There's a fair bit
of a 1x4 texture being accessed well outside its bounds; using a single
pixel-wide image is a trick I've seen used by a few websites to achieve
a graduated fill effect, so I'd suggest it's probably worth pixman
special-casing this.

The real speed killer, though, is a 128x128 image that's being accessed
 from 0,0 to 2560x1400 (approx). I'm not sure why firefox would be doing
that?

Ben


More information about the Pixman mailing list