[Pixman] Make pixman compatible with big image
LE GARREC Vincent
legarrec.vincent at gmail.com
Thu Sep 12 05:44:18 PDT 2013
So, what should I do ? Is it okay to make a cast for every multiplication
of stride by height ? I'm asking because I don't want to work on a patch
that will be denied because my solution is wrong.
2013/9/9 LE GARREC Vincent <legarrec.vincent at gmail.com>
> I didn't wanted doing such a thing because :
> 1) rowstride will never have a value higher than 2^31 so you will
> definitely lose 4 octets. But okay, nobody cases about it.
> 2) I didn't check that bits_image is private so I wanted avoid a
> breakage. So, you're right.
> 3) It still needs lots of change because if I change bits_image.rowstride
> in size_t, I will have to change also lots of variable (like the uint32_t
> byte_stride of the function arm_simd_fill) and parameters of function (like
> mips_dspr2_fill). And this is just one example but I think there's plenty.
> So I think that both ways need lots of work. With my solution, I just have
> to find every variable which have "stride" in its name and cast every
> 2013/9/9 Bill Spitzak <spitzak at gmail.com>
>> It looks like making bits->rowstride into a size_t would be a better
>> solution than doing this casting?
>> LE GARREC Vincent wrote:
>>> I tried to create a very big image (height*stride > 2^31) but it looks
>>> like there's some difficulties. I reported it (
>>> but the solution needs to change more than just few lines.
>>> Does someone could review the patch I made for one file (
>>> ? If it's okay, I'll make a patch for all files.
>>> And after that, I'll make another patch for cairo that have exactly the
>>> same problem.
>>> Vincent LE GARREC
>>> Pixman mailing list
>>> Pixman at lists.freedesktop.org
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Pixman