[directfb-dev] Re: how to assure a video-only surface

[Antonino Daplas <adaplas@pol.net>]

On Wed, 2001-10-31 at 02:54, Denis Oliver Kropp wrote:

> 
> Cool, could you point me to the location of the framebuffer driver?

I don't have a website/ftp site.  I could send it to you by e-mail, just
let me know.

> Graphics drivers can assume that the surface is currently in video memory;
> before the driver gets called the surface management assures a valid
> video memory instance. For the destination "surface->back_buffer->video.*"
> is valid and to be programmed. For the source "surface->front_buffer->video.*"
> is valid and to be programmed.

I found that out minutes after posting to this list :) sorry for the
bother.
> > I've just got the source for a few days so I haven't really studied the
> > source code.
> 
> Have a look at the NeoMagic and the Matrox driver.
> 

Thanks, I'm starting to implement the acceleration code already, and it
already has a Blit function, no effects yet.  It's a bit difficult to
program the i810 since it uses instruction packets via a DMA buffer for
the acceleration engine.  Most/all of the drivers use mmio, but I'm
already looking at the matrox and tdfx codes for the framework.

Does anyone have any experience with these kinds of cards?  I tried to
mmap the DMA buffer via a custom fb_mmap routine but I get inconsistent
results (approximately 70% of instructions processed, the rest are
skipped).  Presently and, hopefully temporarily, I'm using the ugly
(ugh!) ioctl interface to send instruction packets.

Thanks

Tony




-- 
Info:  To unsubscribe send a mail to listar@directfb.org with 
"unsubscribe directfb-dev" as subject.