From owner-freebsd-hackers  Fri Apr 11 11:01:33 1997
Return-Path: <owner-hackers>
Received: (from root@localhost)
          by freefall.freebsd.org (8.8.5/8.8.5) id LAA07147
          for hackers-outgoing; Fri, 11 Apr 1997 11:01:33 -0700 (PDT)
Received: from phaeton.artisoft.com (phaeton.Artisoft.COM [198.17.250.50])
          by freefall.freebsd.org (8.8.5/8.8.5) with SMTP id LAA07131
          for <hackers@FreeBSD.ORG>; Fri, 11 Apr 1997 11:01:24 -0700 (PDT)
Received: (from terry@localhost) by phaeton.artisoft.com (8.6.11/8.6.9) id KAA11690; Fri, 11 Apr 1997 10:41:22 -0700
From: Terry Lambert <terry@lambert.org>
Message-Id: <199704111741.KAA11690@phaeton.artisoft.com>
Subject: Re: Solved problem with com ports & Mach64.
To: kta@alien.ru (Tatiana Kichkaylo)
Date: Fri, 11 Apr 1997 10:41:22 -0700 (MST)
Cc: hackers@FreeBSD.ORG
In-Reply-To: <199704100450.LAA23340@robin.alien.ru> from "Tatiana Kichkaylo" at Apr 10, 97 10:23:08 am
X-Mailer: ELM [version 2.4 PL24]
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Sender: owner-hackers@FreeBSD.ORG
X-Loop: FreeBSD.org
Precedence: bulk

> It's 1997 now, and I think I'm not the first with this problem, but...
> 
> >Environment:
> 
> Release:        FreeBSD 2.1-STABLE i386 from Walnut Creec CDROM, Jan 1996
> Intel motherboard with chipset Triton rev 2 and on-board video Mach64.
> Only com1 & com2 present. Address space of com3 & com4 is used by video
> card.

Actually, the standard for the interface implemented by the Mach32 and
Mach64 chips require that they ignore the ports unless they are selected
into a graphic mode that uses them.  The implementation is buggy.


> >Description: 
> 	
> In boot time, when a kernel probes sio, monitor falls into green mode.
> Reason: in /usr/src/sys/i386/isa/sio.c in function sioprobe kernel writes
> into all likely_com_ports, while some of them are video ports now.

Yes, this is a destructive probe.


> >How-To-Repeat: 
> 
> Boot from any kernel with any sio turned on. With all of com ports turned
> off booting is Ok.
> 
> >Fix: 
> 		
> With /usr/src/sys/i386/isa/sio.c
> 
> 341a342
> > #if NSIO>2
> 342a344,346
> > #else
> > static Port_t likely_com_ports[] = { 0x3f8, 0x2f8, };
> > #endif
> 
> I've just decreased the length of this array for sioprobe
> (and other functions) never writes to my video ports.

This was discussed on -current under the name "Mach32".

The typical "fix" is to turn off the ports until you can rebuild a
kernel that options out COM4: (you can leave the COM3: probe in; it
won't upset the chips).

The actual fix is to go to the ATI WWW site and download the Mach32
assembly language detection code, and incorporate the detection as
a negative probe for COM4:.

I actually posted the ATI code in the previous discussion of the
problem.


					Regards,
					Terry Lambert
					terry@lambert.org
---
Any opinions in this posting are my own and not those of my present
or previous employers.