New acceleration architecture

[Thomas Winischhofer]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Thomas Winischhofer wrote:
> 
> As I already told Zack, I have a problem with EXA in terms of 24/32bpp
> support. I have to support some old hardware which operates at 24bpp for
> depth 24.
> 
> My driver sets the SupportConvert32To24 and PreferConvert32to24 flags
> for xf86SetDepthBpp(). I never looked into the fb code in detail, but I
> was told many times (IIRC by Mark Vojkovich and Keith Packard) that this
> is ok, and even required for some applications which expect 32bit
> visuals. The fb layer is said to handle the conversion to 24bpp.

Not even QT/KDE can handle pure 24bpp visuals. Heavy drawing errors with
acceleration entirely disabled. So these flags must be set in order to
use depth 24 and a 24bpp framebuffer nowadays, with everybody and his
brother obviously expecting 32bpp visuals...

> 
> However, as long as these two flags are set, the server, with EXA, uses
> *exclusively* 32bpp pixmaps - which leads to drawing errors and a sig 11
> pretty soon after start. Also, my acceleration primitives are passed
> 32bpp pixmaps which this old hardware can't handle.
> 
> Anybody any idea?
> 

- --
Thomas Winischhofer
Vienna/Austria
thomas AT winischhofer DOT net          http://www.winischhofer.net/
twini AT xfree86 DOT org
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFCxngLzydIRAktyUcRAuMIAKDDIbKYvbJOv21jCSPaVpNEMMhczgCglzXb
PRYZO292KlngFasTYW6UNQM=
=ut4E
-----END PGP SIGNATURE-----