[ECOS] CS8900A

[ECOS] CS8900A

[Carl van Schaik]

Hi

I have just got ecos running on my own design board using the AT91M40400
ARM processor. I have included  CS8900A ethernet chip and I was wondering
if anyone knows how to find it's MAC address. We need to add it to the
dhcp server before it is allowed to use the network here.

thanks

Carl van Schaik   <carl@leg.uct.ac.za>
-----------------------------------------------------
Dept. Electrical Engineering, UCT

621 Menzies Building
University of Cape Town
Rondebosch   
Western Cape
7700
South Africa

Tel: +27 21 650-3467
Fax: +27 21 650-3465

Re: [ECOS] CS8900A]

[Andrew Lunn]

> I have just got ecos running on my own design board using the AT91M40400
> ARM processor. I have included  CS8900A ethernet chip and I was wondering
> if anyone knows how to find it's MAC address. We need to add it to the
> dhcp server before it is allowed to use the network here.
> 
> thanks

Start it up so it makes a DHCP request. Use tcpdump to watch for the
packet. There is a parameter you can pass to make it print the link
level headers, ie the mac address. It will use the ip address 0.0.0.0
which should make it easier to spot.

Another option could be to look at the DHCP server. If it gets a
request from a machine it does not know the mac address of it may
print an error message with the MAC address.

        Andrew

Re: [ECOS] CS8900A]

[Gary Thomas]

This chip does not "know" its ESA (Etherenet Station Address) by default.
It has to be programmed in.  

If you're building/designing the board yourself, you'll need to get one
from the appropriate authorities.  Or, if the board will never see the
"real" internet, just make one up.

On 31-Aug-2000 Andrew Lunn wrote:
>> I have just got ecos running on my own design board using the AT91M40400
>> ARM processor. I have included  CS8900A ethernet chip and I was wondering
>> if anyone knows how to find it's MAC address. We need to add it to the
>> dhcp server before it is allowed to use the network here.
>> 
>> thanks
> 
> Start it up so it makes a DHCP request. Use tcpdump to watch for the
> packet. There is a parameter you can pass to make it print the link
> level headers, ie the mac address. It will use the ip address 0.0.0.0
> which should make it easier to spot.
> 
> Another option could be to look at the DHCP server. If it gets a
> request from a machine it does not know the mac address of it may
> print an error message with the MAC address.
> 
>         Andrew
> 

RE: [ECOS] CS8900A

[Gary Thomas]

On 31-Aug-2000 Carl van Schaik wrote:
> Hi
> 
> I have just got ecos running on my own design board using the AT91M40400
> ARM processor. I have included  CS8900A ethernet chip and I was wondering
> if anyone knows how to find it's MAC address. We need to add it to the
> dhcp server before it is allowed to use the network here.
> 

BTW, if you're using a driver derived from the standard one, you'll see
that the MAC (more properly, ESA) address is pretty hard-wired.  If you
get RedBoot running on your board you can have this information stored
in local flash configuration.  Alternatively, you can attach a serial
EEPROM to the device and keep it there.  In that case though you'd be
on your own since we've never had working hardware with the serial EEPROM
to test/support.

Re: [ECOS] CS8900A

[Shaun Jackman]

For development purposes we use a bunch of gibberish ESAs (MACs).
0x000102030405 is my personal favourite.
(so if you get some internet trafic from said ESA, drop me an e-mail because
something's gone terribly wrong)   ; )
For production purposes, like others have said, you'll need to get an EUI
from the IEEE.

On another (somewhat related) note...
I have an EB40 (w/ AT91R40807, esentially identical to the 40400 with 136KB
of internal memory instead of 4K) with a RT8019 etherchip (NE2k (NS8390)
clone).
There's a known issue with the packaged Angel monitor. I can't connect at
faster than 9600, and debugging is *DEAD* slow. (read: over two minutes to
get a 1k memory dump)
My preferred plan is to get a Jeeni, but they're a little pricey (although
right now I would shell out the bucks) and there's no stock!
So, is there another monitor I could flash into it that isn't broken? (gdb
stub, Cygmon, RedBoot, ...)
If there isn't one ported already, and I actually find some time to start
porting, which should I use? Where do I get the source? What's required to
complete the port? A "RTF-FAQ reply" would be appreciated when acompanied
with an URL.
I have TCP/IP (over ethernet) working on my board. Would it be reasonable to
consider using this for debugging rather than serial?

Cheers,
Shaun

----- Original Message -----
From: Carl van Schaik <carl@leg.uct.ac.za>
To: Ecos Mailing List <ecos-discuss@sourceware.cygnus.com>
Sent: Thursday, 2000 August 31 06.10
Subject: [ECOS] CS8900A


> Hi
>
> I have just got ecos running on my own design board using the AT91M40400
> ARM processor. I have included  CS8900A ethernet chip and I was wondering
> if anyone knows how to find it's MAC address. We need to add it to the
> dhcp server before it is allowed to use the network here.
>
> thanks
>
> Carl van Schaik   <carl@leg.uct.ac.za>
> -----------------------------------------------------
> Dept. Electrical Engineering, UCT
>
> 621 Menzies Building
> University of Cape Town
> Rondebosch
> Western Cape
> 7700
> South Africa
>
> Tel: +27 21 650-3467
> Fax: +27 21 650-3465
>
>

Re: [ECOS] CS8900A

[Gary Thomas]

If you have eCos running on this board, RedBoot is just a step away.
Implementing RedBoot will get you (amonst other things):
  * Network downloading and debugging
  * FLASH image and configuration support

If this interests you, please feel free to contact me.  I'd be happy
to help.

On 31-Aug-2000 Shaun Jackman wrote:
> For development purposes we use a bunch of gibberish ESAs (MACs).
> 0x000102030405 is my personal favourite.
> (so if you get some internet trafic from said ESA, drop me an e-mail because
> something's gone terribly wrong)   ; )
> For production purposes, like others have said, you'll need to get an EUI
> from the IEEE.
> 
> On another (somewhat related) note...
> I have an EB40 (w/ AT91R40807, esentially identical to the 40400 with 136KB
> of internal memory instead of 4K) with a RT8019 etherchip (NE2k (NS8390)
> clone).
> There's a known issue with the packaged Angel monitor. I can't connect at
> faster than 9600, and debugging is *DEAD* slow. (read: over two minutes to
> get a 1k memory dump)
> My preferred plan is to get a Jeeni, but they're a little pricey (although
> right now I would shell out the bucks) and there's no stock!
> So, is there another monitor I could flash into it that isn't broken? (gdb
> stub, Cygmon, RedBoot, ...)
> If there isn't one ported already, and I actually find some time to start
> porting, which should I use? Where do I get the source? What's required to
> complete the port? A "RTF-FAQ reply" would be appreciated when acompanied
> with an URL.
> I have TCP/IP (over ethernet) working on my board. Would it be reasonable to
> consider using this for debugging rather than serial?
> 
> Cheers,
> Shaun
> 
> ----- Original Message -----
> From: Carl van Schaik <carl@leg.uct.ac.za>
> To: Ecos Mailing List <ecos-discuss@sourceware.cygnus.com>
> Sent: Thursday, 2000 August 31 06.10
> Subject: [ECOS] CS8900A
> 
> 
>> Hi
>>
>> I have just got ecos running on my own design board using the AT91M40400
>> ARM processor. I have included  CS8900A ethernet chip and I was wondering
>> if anyone knows how to find it's MAC address. We need to add it to the
>> dhcp server before it is allowed to use the network here.
>>
>> thanks
>>
>> Carl van Schaik   <carl@leg.uct.ac.za>
>> -----------------------------------------------------
>> Dept. Electrical Engineering, UCT
>>
>> 621 Menzies Building
>> University of Cape Town
>> Rondebosch
>> Western Cape
>> 7700
>> South Africa
>>
>> Tel: +27 21 650-3467
>> Fax: +27 21 650-3465
>>
>>
> 

[ECOS] CS8900A

[Carl van Schaik]

Hi

OK, it seems the ethernet chip is working, but bootp/dhcp is not yet. When
I enable net_debug, I get sent back all the packets that are picked up on
the network so it seems the chip is functional.

Initially I am running in polled more (INTS_DONT_WORK mode) as I need to
fix my interrupt controller.

this is the chip setting and a tcpdump log
// sar10.ee:tc=.uct-53:ha=0x0040d800429c:ip=137.158.133.235
static unsigned char enaddr[] = { 0x00, 0x40, 0xd8, 0x00, 0x42, 0x9c };

tcpdump: listening on eth0
17:31:37.401959 0:40:d8:0:42:9c Broadcast arp 60: arp who-has 0.0.0.0 tell
0.0.0.0
17:31:37.402353 0:0:21:c8:bf:74 0:40:d8:0:42:9c arp 60: arp reply 0.0.0.0
is-at 0:0:21:c8:bf:74
17:31:37.402462 0:50:bf:12:af:3b 0:40:d8:0:42:9c arp 60: arp reply 0.0.0.0
is-at 0:50:bf:12:af:3b
17:31:37.402545 0:0:21:6a:93:7a 0:40:d8:0:42:9c arp 60: arp reply 0.0.0.0
is-at 0:0:21:6a:93:7a
17:31:37.402647 0:40:f6:d4:89:4e 0:40:d8:0:42:9c arp 60: arp reply 0.0.0.0
is-at 0:40:f6:d4:89:4e
17:31:37.403374 0:50:99:c0:a5:58 0:40:d8:0:42:9c arp 60: arp reply 0.0.0.0
is-at 0:50:99:c0:a5:58

Is it possible that we are just missing the packets in polled mode? The
network currently has a LOT of traffic on it, I'll have to try set up a
private network as well.

Can anyone tell me if the above responses look OK.

thanks

Carl van Schaik   <carl@leg.uct.ac.za>
-----------------------------------------------------
Dept. Electrical Engineering, UCT

621 Menzies Building
University of Cape Town
Rondebosch   
Western Cape
7700
South Africa

Tel: +27 21 650-3467
Fax: +27 21 650-3465