From owner-freebsd-hackers  Mon Mar 11 16:52:57 1996
Return-Path: owner-hackers
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.3/8.7.3) id QAA05281
          for hackers-outgoing; Mon, 11 Mar 1996 16:52:57 -0800 (PST)
Received: from doleh.com (doleh.nai.net [205.139.1.114])
          by freefall.freebsd.org (8.7.3/8.7.3) with SMTP id QAA05276
          for <freebsd-hackers@FreeBSD.org>; Mon, 11 Mar 1996 16:52:54 -0800 (PST)
Received: (from yaser@localhost) by doleh.com (8.6.12/96.03.01.0) id TAA05485; Mon, 11 Mar 1996 19:54:07 -0500
From: "Yaser K. Doleh" <yaser@doleh.com>
Message-Id: <199603120054.TAA05485@doleh.com>
Subject: Re: First attempt to write a device driver
To: msmith@atrad.adelaide.edu.au (Michael Smith)
Date: Mon, 11 Mar 1996 19:54:06 -0500 (EST)
Cc: freebsd-hackers@FreeBSD.org
In-Reply-To: <199603120016.KAA24351@genesis.atrad.adelaide.edu.au> from "Michael Smith" at Mar 12, 96 10:46:43 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


I have one more questions which I could not figure out by looking at the 
source.

The card have 64K shared memory which can be mapped into the machine
address space. I can map 16K, 32K or the whole 64K into the address
space. My question is how can I find an empty slot in the area 640 - 1MB
that I can map this memory into. The driver must choose the location
and set a register in the Token Ring card to tell it
where to map the memory.

My second question is "Do I use pmap_mapdev function ?"


Michael Smith said in a letter
> 
> Make sure the above is in your config, make sure you have interrupts
> enabled on the card.
> 

It must be the card not sending any interrupts. I guess I have to RTFM 
that IBM manual again.

> You mean the 'visual' mode doesn't know your driver?  That's because there's
> no way for it to find all the names of drivers installed itself, so that
> information is in the table at the top of /sys/i386/i386/userconfig.c.
> Look for the string "EDIT THIS LIST", and create an entry like the other
> network devices.
> 

Yes, that's what I was looking for. Thank you.

> Looks like you've got the rest of the skeleton working fine.
> 

Yes, most of it is there.

> > I am also thinking of really hooking the driver to the ethernet code
> > and just convert tokenring<->ethernet in my driver. So to the machine
> > it will look like this is an ethernet driver but to the network
> > it will look like its a tokenring. 
> 
> Um, that's an interesting approach.  I'd be inclined to seperate the two so
> that the token-specific code could be reused for other hardware though.
> 

That was a suggestion of somebody else which made me think "This is a good
idea, I can get the Hardware driver working first and test it this way
then I can worry about the higher layer later". I am not at that point
yet, I just wanted to see what people will think about this approach.
If I decide to take this approach, I will definitly seperate the 
driver from the conversion routines.

I also have been looking at if_ethersubr.c and if_fddisubr.c files
to see how easy will it be to write similar functions to handle Token Ring.

> Have a look at /sys/net/if_ether.c.  There's already code in there to handle
> LLC (look for "LLC" 8) which I believe is the major obstacle.  Your driver
> shouldn't try to interpret the contents of packets itself.

Thanks, I'll check it out.

Very truly yours
Yaser Doleh
Yaser@Doleh.Com