Index: [thread] [date] [subject] [author]
  From: Brian Julin <bri@forcade.calyx.net>
  To  : ggi-develop@eskimo.com
  Date: Tue, 11 Aug 1998 09:08:04 -0400 (EDT)

Re: kgicon problems...

On Tue, 11 Aug 1998, Andrew Apted wrote:

> Brian writes:
> 
> >  The Degas drivers fix this,
> >  but use a totally different structure.
> 
> "Degas drivers" ?  You've mentioned that a few times, but there isn't
> any other drivers in the CVS.  Are they in Steffen's tree ?

>From EvStack tree.  Actually better yet, myself and a few others actually had
EvStack drivers working with libGGI and the dpp/bpp/text member changes.
There are handy differences like this one:

< == evstack 
> == Dali

351,362c389,395
<                       DEBUG2("chipset_setup_mode: Setting up for GT_TEXT16");
<                       mode->textop    =&text16_operations;
<                       mode->fb.addr   =bus_to_virt(fb_text);
<                       mode->fb.stride =sizeof(uint16)*
<                                        (mode->request.virt.x/mode->request.text.x);
<                       mode->fb.size   =mode->fb.stride*
<                                        (mode->request.virt.y/mode->request.text.y);
<                       mode->fb.frames =1;
<                       mode->setup.chipset=text16_data_alloc();
<                       if (mode->setup.chipset==NULL)
<                               return -ENOMEM;
<                       return EOK;
---
>               mode->fb        = (mem_vaddr)fb_text;
>               mode->fb_size   = fb_text_len;
>               mode->fsize     = (mode->request.virt.x / mode->request.text.x)
*
>                       (mode->request.virt.y / mode->request.text.y) * 2;
>               mode->mem       = &text16_color_mem;
>               mode->io        = &text16_color_io;
>               return;         /* Done setting up */


> >  Shouldn't be too hard to check all uses of this member as long
> >  as you chop out the text16 stuff which isn't needed
> 
> It is needed, for when kgicon does text modes, which it will do sometime
> soon (RSN :).

Maintaining the code that is already there would _not_
help us prepare for GGI-console as it would all have to be extensively
re-patched to make up for the differences in driver structure.  So
if we are not using it we should remove it, or we should at
least upgrade to the EvStack code.

--
Brian S. Julin

Index: [thread] [date] [subject] [author]