RE: [AD] [TODO BUGFIX] Wrong framebuffer 24 bit scrolling

[Shawn Hargreaves <shargreaves@xxxxxxxxxx>]

Title: RE: [AD] [TODO BUGFIX] Wrong framebuffer 24 bit scrolling

Grzegorz Adam Hankiewicz writes:
> I have just noticed that scrolling in 24 bit modes with framebuffer on my
> matrox card makes funny effects: looks like allegro is presuming that 24
> bit is really 32 with a padding byte, while the framebufer seems to be
> working as real continuos 24 bit pixels, meaning that depending on the x
> position of the scroll, colors are messed, possibly because the order of
> red, green, bytes are taken wrongly from nearby pixels.

I've only very rarely seen scrolling work in 24 bit modes. Most cards have
4, 8, or 16 bit alignment restrictions on the possible addreses they can
scroll to, which can limit you to jerks of 2 or 4 pixels in 16 or 32 bit
modes, but leaves it reading from a totally wrongly aligned part of the
framebuffer if the card is in a true 24 bit mode. There's no nice solution
for that: maybe the drivers could force the address to the nearest common
multiple of 3 and 16 or whatever, but most don't seem to do that, and in
any case this would make such horrible jerks as to be basically unusable.

24 bit packed modes suck :-)

        Shawn.