Composite and damage status

[Keith Packard]

Around 16 o'clock on May 12, Egbert Eich wrote:

>  > Correct.  "fixing" XAA would still be a good idea, but it's not required.
>  > Using the 'shim' layer causes duplicate allocation of GCs, Pictures and a 
>  > lot of extra region manipulation.  Functional, but sub-optimal.
>  > 
> 
> :-((( 

Deron's code includes a function which the driver calls to enable the new 
pixmap fields, so when XAA is fixed, the drivers can just call that and 
get the more efficient mechanism.  The best part is that it's all 
compatible; if anyone doesn't deal with the new offsets, you just use the 
'shim' and all of the other code will work correctly.

My reading of the XAA interface says that the "better" method is ABI/API 
compatible with existing video drivers.

> OK. Then you have to tell the DDX in some way what is 'on screen' and
> what is 'off screen'  I suppose. I don't see how this will work
> with the XAA implementation as this uses the drawable type to decide
> where the data goes.

Not really; as long as XAA sticks pixmaps in video memory, things will be 
accelerated.  Fixing XAA to migrate pixmaps as appropriate might make some 
sense for cards with limited memory (less than 1/2GB).

-keith


-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 228 bytes
Desc: not available
Url : http://freedesktop.org/pipermail/xserver/attachments/20040512/abaf3321/attachment.pgp