Subject: Re: Why don't we have 'options GATEWAY' in GENERIC?
To: Gordon W. Ross <gwr@mc.com>
From: Virtually Here <sellis@rohan.sdsu.edu>
List: port-sun3
Date: 05/22/1996 13:48:30
Gordon W. Ross just said:
> 
> > Date: Tue, 21 May 1996 23:46:04 -0700 (PDT)
> > From: Curt Sampson <curt@portal.ca>
> 
> > 
> > Unfortunately, the kernels since 1.1B crash when they try to
> > autoconfig the cgfour. So if I compile a new kernel, I don't get
> > a cgfour, and thus don't get X. (I can config the bwtwo plane on
> > that card to bwtwo0, but unfortunately XSunMono doesn't seem to
> > want to find it.)
> 
> Has anybody found out what happened with the cg4 driver?
> I haven't had much time to work on it...

I looked into it a small bit (actually, I played with the debugger
while somebody who knows 68k asm real well told me where I was in the C
code!), and it looks like subt_autoconf.c is the problem, specifically
config_search() where it's looping on 'p' (which is pv[] in ioconf.c).

It looked like p went to zero too soon which caused one of the
other functions (can't remember the name..whatever is inside the loop
with cf_parent) to choke when it tried to use 'p' as input.

cfdata struct seemed to be okay..the cgfour entry looked the same as the
bwtwo entry, and bwtwo works fine.  pv[] (also in ioconf) looked normal..
no strange numbers in there, and again bwtwo was using the same values.

Diffs of bw and cg .c and .h files from -current to 1.1A showed no 
changes other than the new cfdriver struct.  

I've no idea where to go from here.  Ideas?
   //////////////////////////////////////////////////////////////////////
  //  Scott Ellis   //   sellis@gi.com   //   sellis@rohan.sdsu.edu   //
 //////////////////////////////////////////////////////////////////////
// WARNING: This signature warps  time and space in its vicinity    //