Turnbull China Bikeride/Turnbull China Bikeride - Disc 2.iso/BARNET/ARMLINUX/MAIL/9805/000136_owner-linux-arm…r.rutgers.edu _Sat May 23 01:51:46 1998.msg

Internet Message Format | 1998-06-02 | 4KB

Return-Path: <[e-mail]>
Received: from virtual.bbc.co.uk (virtual.bbc.co.uk [132.185.132.199])
    by odie.barnet.ac.uk (8.8.6/8.8.6) with ESMTP id BAA16319
    for <[e-mail]>; Sat, 23 May 1998 01:51:45 +0100
Received: from nic.funet.fi (nic.funet.fi [128.214.248.6])
    by virtual.bbc.co.uk (8.8.5/8.8.5) with ESMTP id BAA12371
    for <[e-mail]>; Sat, 23 May 1998 01:52:16 +0100 (BST)
Received: from vger.rutgers.edu ([128.6.190.2]:51768 "EHLO vger.rutgers.edu" ident: "root") by nic.funet.fi with ESMTP id <2740-502>; Sat, 23 May 1998 02:57:01 +0300
Received: by vger.rutgers.edu id <971722-31161>; Fri, 22 May 1998 17:07:49 -0400
Received: from post-11.mail.demon.net ([194.217.242.40]:62913 "HELO post.mail.demon.net" ident: "NO-IDENT-SERVICE[2]") by vger.rutgers.edu with SMTP id <977936-31161>; Fri, 22 May 1998 15:15:48 -0400
Received: from (marketto.demon.co.uk) [194.222.32.5]
    by post.mail.demon.net with smtp (Exim 1.82 #2)
    id 0ycxd5-0001zb-00; Fri, 22 May 1998 19:37:24 +0000
To: [e-mail]
Cc: [e-mail]
From: [e-mail] (Theo Markettos)
Subject: Re: PowerROMmed cards
Reply-To: [e-mail]
In-Reply-To: <[e-mail]>
Date:     Fri, 22 May 1998 20:20:14 +0100
Message-ID: <[e-mail]>
X-Mailer: TTFN version 0.41 (Acorn RISC OS)
X-Posting-Agent: RISC OS Newsbase 0.60g
X-Orcpt: rfc822;[e-mail]
Sender: [e-mail]
Precedence: bulk
X-Loop: [e-mail]
Status: RO

On Mon, 18 May 1998 19:56:28 +0100 (BST), [e-mail] said:

> Theo Markettos writes:
> It runs in 32-bit modes, and I have to maintain a blacklist of the loader
> code that doesn't work.  It seems that most code does work, but there are
> exceptions (eg, the PowerROMs of course).  Their code crashes if you call
> it with 32-bit addresses etc.
>
> I think that it may be possible to detect the card by the location of the
> FAS216 (or NCR53C9x) chip.  I personally think that the drivers should
> not load/initialise if they can't detect the SCSI chip where they believe
> it should be.  This would then get around these problems.

That sounds _very_ dodgy.  Firstly, IIRC some cards used a 5380 (A3000-style
8 bit ones spring to mind).  Secondly, if I had to design a podule with a
ROM, page latch and SCSI chip in 4K of address space, there are a few obvious
ways to do it, which might be used on more than one card, which would make
telling the difference difficult.  Also, hardware probing may well be
dangerous - you may be accessing read sensitive hardware, which could do
nasty things (like generate an interrupt or FIQ) that you're not expecting.

A better (IMHO) solution I would propose would be to find out where the page
latch for each card is (should be hackable out of the PowerROM or original
loader), and then drive the ROM paging directly.  If the loader doesn't do
funny things, like map even pages to the first half of ROM and odd pages to
the second, it should be a case of bits 0 - x-1 of the ROM space address,
where x=LOG2(ROM size), are part of the podule address, and bits x - x+7 are
the page register value.  If you hack the Risc OS loader, you can also find
out the purpose of any other latched bits (eg reset, interrupt control), and
how they should be set for normal operation.

--
Theo Markettos          Home:   [e-mail]
Liphook                 Work:   [e-mail]
Hampshire
UK                      Web site, including Acorn backup software
                        http://www.marketto.demon.co.uk/

unsubscribe: body of `unsubscribe linux-arm' to [e-mail]