From owner-freebsd-current  Mon Aug 26 13:53:40 1996
Return-Path: owner-current
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.5/8.7.3) id NAA15120
          for current-outgoing; Mon, 26 Aug 1996 13:53:40 -0700 (PDT)
Received: from FileServ1.MI.Uni-Koeln.DE (FileServ1.MI.Uni-Koeln.DE [134.95.212.1])
          by freefall.freebsd.org (8.7.5/8.7.3) with SMTP id NAA15112;
          Mon, 26 Aug 1996 13:53:36 -0700 (PDT)
Received: from x14.mi.uni-koeln.de (annexr3-13.slip.Uni-Koeln.DE) by FileServ1.MI.Uni-Koeln.DE with SMTP id AA15792
  (5.67b/IDA-1.5); Mon, 26 Aug 1996 22:53:24 +0200
Received: (from se@localhost) by x14.mi.uni-koeln.de (8.7.5/8.6.9) id WAA03904; Mon, 26 Aug 1996 22:47:02 +0200 (MET DST)
Date: Mon, 26 Aug 1996 22:47:02 +0200 (MET DST)
Message-Id: <199608262047.WAA03904@x14.mi.uni-koeln.de>
From: Stefan Esser <se@zpr.uni-koeln.de>
To: Guido Muesch <g.muesch@stochastik.rwth-aachen.de>
Cc: hardware@FreeBSD.ORG, current@FreeBSD.ORG
Subject: Re: NCR  & AMD DX/4-133 problem
In-Reply-To: <199608261904.VAA06960@jupiter.stochastik.rwth-aachen.de>
References: <199608261904.VAA06960@jupiter.stochastik.rwth-aachen.de>
Sender: owner-current@FreeBSD.ORG
X-Loop: FreeBSD.org
Precedence: bulk

Guido Muesch writes:
 > I get the following when booting FreeBSD-current from my SCSI disk:
 > 
 > Using an AMD-133:
 > 
 > ncr0 <ncr 53c810 scsi> rev 17 int a irq 9 on pci0:29
 > ncr waiting fo scsi devices to settle
 > ncr0:0: ERROR (81:0) (8-0-0) (0/13) @ (ffdac008:f000f773)

Hmm, the 81 means "Illegal Instruction", and the pair of values
at the end of the line does fully support the NCR chips complains :)

Seems that the NCR reads corrupt data from memory when fetching 
its instructions.

 >         reg:         ca 00 00 13 47 00 00 1f 31 08 04 00 80 00 08 02.
 > ncr0: restart (fatal error).
 > (ncr0:0:0): COMMAND FAILED  (9 ff) @ f09fa8dc.
 > 
 > When I have my old AMD-100 plugged in, everything works fine.

Yes, the error message above indicates some kind of hardware 
problem, though not everything is lost ...

 > I get this with two different boards. (One of them did not support the
 > AMD-133 right). But the DOS driver seems to work with the second.

Actually, that doesn't prove a lot :)

The DOS drivers tend to ignore most of the features that make
the NCR controllers interesting under a multi-tasking OS.

 > Is it a hardware problem? Should I get just another mainboard?
 > (This mainboard is a Pine PT-432B)

Don't know that motherboard. As always: Please boot with "-v" and 
send me the message log (as written to /var/log/messages). I'll 
know which chip set there is, and possibly which cache configuration
and PCI performance options are in effect.

My current suspicion is that you are using the Write-Back mode of 
the AMD5x86s primary cache, and that your motherboard's chip set 
does not actually support that feature. Please try again with the 
primary cache set to Write-Through, or with the primary cache 
disabled. Send me your results ...


Gruss, STefan