On 19 September 2018 at 07:46, BALATON Zoltan <bala...@eik.bme.hu> wrote: > On Wed, 19 Sep 2018, Marcus Comstedt wrote: >> >> The value from twoD_foreground (which is in host endian format) must >> be converted to the endianness of the framebuffer (currently always >> little endian) before it can be used to perform the fill operation. >> >> Signed-off-by: Marcus Comstedt <mar...@mc.pp.se> > > > Reviewed-by: BALATON Zoltan <bala...@eik.bme.hu> > > But also cc-ing Peter Maydell who reviewed endianness fixes before for this > device model in case he has something to add.
There are probably other ways to fix that, but this looks OK to me. We have a pixel value in a host-endianness register field, and we're trying to write it into a little-endian framebuffer with a plain host store of a 32/16/8 bit type. So cpu-to-le is the right thing. (The other approach would be to use the stl_le_p()/stw_le_p()/stb_p() functions to do the store directly in the correct endianness. But I don't think that ends up with significantly cleaner code.) Reviewed-by: Peter Maydell <peter.mayd...@linaro.org> thanks -- PMM
