linux-mips
[Top] [All Lists]

Re: X server

To: "William J. Earl" <wje@cthulhu.engr.sgi.com>
Subject: Re: X server
From: Ralf Baechle <ralf@oss.sgi.com>
Date: Wed, 12 Jan 2000 01:28:58 +0100
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>, Bill Halchin <bhalchin@hotmail.com>, ralf@oss.sgi.com, linux@cthulhu.engr.sgi.com
In-reply-to: <14459.34255.60876.700493@liveoak.engr.sgi.com>
References: <20000111030346.42694.qmail@hotmail.com> <E1281JK-0004hu-00@the-village.bc.nu> <14459.34255.60876.700493@liveoak.engr.sgi.com>
Sender: owner-linuxmips@oss.sgi.com
On Tue, Jan 11, 2000 at 11:34:39AM -0800, William J. Earl wrote:

>  > > Ralf,
>  > >   What I am saying is why don't we try this approach.
>  > 
>  > "we". I have this cool idea - "You" 
>  > 
>  > Actually  for an Indy X server your best model is still the 8514 driver in 
> XFree
> 
>       That sounds right to me as well.  The problem with fbcon or GGI
> is that the Indy graphics hardware does not have a CPU-addressable
> frame buffer.  You can always fake one in main memory, and DMA any
> modified portions to the real frame buffer, but a naive implementation
> would use more memory bandwidth than is available and a clever
> implementation would incur a lot of VM overhead (and still use a lot
> of bandwidth).  Starting from an X server designed for hardware without
> an addressable frame buffer is more appropriate.

As I remember GGI has been improved such that it also can handle things
like a 8514 or Newport card.  That still doesn't invalidate other
reasons why GGI isn't such a good idea.  That said, Alan's suggestion
is the right one.

  Ralf

<Prev in Thread] Current Thread [Next in Thread>