Nubus video cards

Hi,

Just before I got a headache with MacsBug,

On 2006/12/13, at 1:18, Robert Swindells wrote:
> Kazuyuki Inanaga wrote:
>
>> Main chip is "smsc FEAST(TM) FD LAN91C100FDQFP
B9904-B467AIC
>> 6H111735-6".
>>
>
> Ok, this is supported by the sm(4) driver. The NuBus
attachment stub
> for it is in sys/arch/mac68k/nubus/if_sm_nubus.c.
>
>
>> How can I look at the driver?  With
"MacsBug" like Video ?
>> "Delving for Video Interrupt Information"
>> http://
www.macbsd.com/macbsd/howto/video.html
>>
>
> You only need to follow the "Identifying your
card" section on that  
> page.
>
"Identifying your card"
 >we only need to know how to identify your card to BSD.

Oh, I see. Thanks! Robert-san.    I
remember a FAQ;

"NetBSD/mac68k Frequently Asked Questions with Answers/
7. Porting and Programming"
htt
p://www.netbsd.org/Ports/mac68k/faq/faq-7.html

> 7.9 Are there any Mac OS utilities for investigating
hardware
> so I can begin implementing support for it?

> Yes. Apple has quite a few useful utilities located at:
>
ftp://ftp.apple.com/developer/Tool_Chest/Devices_-_Hardware/
> There are at least three useful subfolders in this
directory:
> •Apple_Desktop_Bus - utilities to work with ADB
> •NuBus_Slot_Manager - contains the Slots utility
> •Display_&_Video_Hardware - utilities for
investigating video hardware

I got "Slots.sit.hqx" from
"NuBus_Slot_Manager" directory.

It is unbelievably easy, with just double click
"Slots", I get;
sResources
Directory:
81  --------->  01  --------> 0004 0001 00FF FFD4

What we need is this?

------------------------------------------------------------
-

I posted above before I received kind suggestions from
Michael-san,
but my post was not delivered.

On 2006/12/13, at 22:25, Michael R. Zucca wrote:
>
> There's a program called "Slots" around
somewhere that will
> give you some nice information about the card and there
may be
> a hint in some of that information about where the
ethernet chip
> lives in the card's address space.

Thanks! As I wrote above, I got "Slots".

> However, look at the driver. There may be some nice
some easy
> to find value like the MAC address for your card (or at
least
> the vendor part of the MAC address encoded in hex (try
both
> big and little endian) or some other easily
identifyable info.

0004 0001 00FF FFD4    Need more?
(MAC address is printed as 0000C54B11F4 on this card.
Honestly, I almost don't understand what I'm doing..)

> Then use MacsBug to search the card's address space.

Ouch!

> If you get a match, use the driver's offsets and look
at
> the values around the matched address to see if it
matches
> the way the chip should look. If so, you should be able
to
> determine the chip's base address.
>
> Good luck!
>

I think I should get a "MacsBug guidebook" written
in Japanese.

Thanks a lot,
Kazu Inanaga

On Dec 14, 2006, at 11:18 AM, Kazuyuki Inanaga wrote:

> 0004 0001 00FF FFD4    Need more?
> (MAC address is printed as 0000C54B11F4 on this card.
> Honestly, I almost don't understand what I'm doing..)

Ok, so maybe this isn't the best job for you  However,
I'll  
describe what needs to happen just in case you feeling
adventurous.  
You'll need Macsbug to search the card's address space.
There is a  
help feature which is good in Macsbug. As I recall, Macsbug
uses $  
instead of 0x as a prefix to indicate hexidecimal numbers.

Not sure what the SResource is telling you. Maybe this is
the offset  
in the NuBus address space of the chip? That would be very  
convenient. Take a look at the card's address space at that
offset  
and see if there's anything of interest there. Look at the
driver and  
see what the offset of the MAC address is. If it's at
<card base> +  
4000100FFFFD4 + <offset of MAC address register> then
that is the  
offset! Though, we're probably not that lucky 

If the MAC address is 0000C54B11F4 then you could search the
address  
space of the card for this pattern or its little endian
equivalent  
(does this chip store this stuff in LE?). Once you know
this, you  
would look at the NetBSD driver for this chip and see where
in the  
chip's register map the MAC address lives. Then you would
subtract  
that offset from the address you found, and that should be
the base  
address of the chip in slot space. Then you'd mask off the
top bits  
of the address since that's the slot's offset in physical
address  
space. That would give you the offset of the chip when the
card is in  
any slot. This is the value that you need for the driver.

One problem, however, might be that the MAC address isn't
stored in a  
register. On some cards, the MAC address might be stored in
some  
EEPROM that gets accessed by poking at some registers. So
you may not  
find the MAC address at all.

Another way to find the offset might be to look for other
patterns  
that might be likely in the card's address space other than
the MAC  
address. You'll have to look at the NetBSD driver for these.

Finally, another way to look for the base address is to just
look for  
any non-regular patterns in the card's address space. Many
cards  
don't show anything of interest in the card's address space
except  
ROMs and whatever on-board chips exist. So you'll see some
bit  
pattern like 0x0 or 0xFF for thousands and thousands of
addresses and  
then, suddenly, you'll see something different. However,
some cards  
alias the chip across the slot address space. So the Mac
address may  
show up many, many times. Any one of them should be fine for
use.  
Though, it might be hard to figure out where the chip's
address space  
actually starts.

If you can't do all this, don't worry about it. There may be
other  
folks with this card out there with the requisite technical
knowledge  
to do it... or you may have to wait for a long time for
somebody like  
that to show up 




-- 
----------------------------------------------
  Michael Zucca - mrz5149acm.org
----------------------------------------------
  "I'm too old to use Emacs." -- Rod MacDonald
----------------------------------------------