Re: [PATCH] Hooks for new fbdev api

["Petr Vandrovec" <VANDROVE@vc.cvut.cz>]

On 29 Nov 01 at 9:35, James Simmons wrote:
> +struct fbcursor {
> +   __u16 set;      /* what to set */
> +   __u16 enable;/* cursor on/off */
> +   struct fbcurpos pos;/* cursor position */
> +   struct fbcurpos hot;/* cursor hot spot */
> +   struct fb_cmap cmap;/* color map info */

>From what you wrote I assume that cmap.start must be 0 and cmap.len
some length, and it must be always set, as otherwise it is impossible
to guess image/mask depth from it.

> +   struct fbcurpos size;/* cursor bit map size */
> +   char *image;/* cursor image bits */
> +   char *mask;/* cursor mask bits */

And maybe it is better to go with Geert idea? Remove mask, and
make image just really use cmap - if cmap entry is 100% transparent,
it is like that bit(mask) == 1, and add one more field for inverted
cmap entry. As no driver can use image/mask immediately anyway, there
is no big problem.

I hope that mask/image format is going to be specified somewhere
more exactly - like whether each image/mask line consist of 
non-fractional number of bytes, what happens if cmap has 8 entries 
and other legal, but hard to implement, features...
                                                    Petr Vandrovec
                                                    vandrove@vc.cvut.cz