How to configure 2 ethernet devices

How to configure 2 ethernet devices

[OTAVIO ISAMU SUGENO]

Hi,
Can anyone help me to configure two Ethernet devices in my custom board.
We use a MPC860T with FEC and SCC1 as Ethernet devices. FEC is eth0 and
SCC1 is eth1. We are also using Monta Vista HardHat Linux 2.0.
Linux kernel goes up with no problem, also the Ethernet devices but when
configuring the devices using bootp/rarpd in message IP-Config only
appears:

IP-Config: Got RARP answer form 10.0.0.10, my address is 10.0.0.100

I guess this is to the eth0 interface.
I suppose also the the Linux should ask an ip address to each interface
but it does not happen. By the way, each Ethernet interface has a its
own MAC address.
What do I need to configure in the HardHat Linux to receive a second ip
address to the eth1?
Many thanks.
Otavio Sugeno

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: How to configure 2 ethernet devices

[Dan Malek]

OTAVIO ISAMU SUGENO wrote:

> ......but when
> configuring the devices using bootp/rarpd in message IP-Config only
> appears:

The purpose of IP-Config is to provide a minimal network configuration
as to allow a network booting feature.  It will not completely initialize
a network environment suitable for little more than the NFS connection itself.

> IP-Config: Got RARP answer form 10.0.0.10, my address is 10.0.0.100
>
> I guess this is to the eth0 interface.

The IP-Config sends requests on all configured interfaces and accepts the
first one that replies.

> What do I need to configure in the HardHat Linux to receive a second ip
> address to the eth1?

You need to run the real user-land network configuration scripts and utilities
to properly configure a network on any system.  It should also be done even
when you are using IP-Config and network booting.


	-- Dan


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: How to configure 2 ethernet devices

[Wolfgang Denk]

Hi,

in message <3CE28613.8090206@embeddededge.com> Dan Malek wrote:
>
> The purpose of IP-Config is to provide a minimal network configuration
> as to allow a network booting feature.  It will not completely initialize
> a network environment suitable for little more than the NFS connection itself.

...but it comes in very handy for embedded systems  that  don't  need
the  overhead  (memory  footprint) for tools like ifconfig and route.
And for most simple configurations (where you just  need  to  set  IP
address, netmask, and default router) it is completely sufficient.

> > IP-Config: Got RARP answer form 10.0.0.10, my address is 10.0.0.100
> >
> > I guess this is to the eth0 interface.
>
> The IP-Config sends requests on all configured interfaces and accepts the
> first one that replies.

You can restrict the chioce of interfaces  by  using  an  "ip="  boot
argument.

> You need to run the real user-land network configuration scripts and utilities
> to properly configure a network on any system.  It should also be done even
> when you are using IP-Config and network booting.

Ummm... what for? I never do this, and the systems are  just  working
fine.  I  never  saw any deficiency in the IP autoconfiguration stuff
(except that it doesn't accept more than one "ip=" option, so we  can
configure only one interface that way). What am I missing?

Wolfgang Denk

--
Software Engineering:  Embedded and Realtime Systems,  Embedded Linux
Phone: (+49)-8142-4596-87  Fax: (+49)-8142-4596-88  Email: wd@denx.de
"Out of register space (ugh)"
- vi

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

Re: How to configure 2 ethernet devices

[Wolfgang Denk]

In message <3CE2C6CB.20402@embeddededge.com> Dan Malek wrote:
>
> If you like to use DNS, routing or other networking features you have
> to use some kind of userland configuration to make that happen.

Rigth, for more complicated configurations this is the way to go.

> I think it's kind of amusing when you use "dynamic" configuration to
> get a network configuration to boot a kernel, then have a bunch of
> static configuration files to support the application environment :-)
> If you are always booting from the same server on the same static
> network, why not just boot a ramdisk with some static configuration
> and network set up?  It will boot up faster.

But we don't have a static configuration. We use something like this:

=> setenv bootargs $(bootargs) ip=$(ipaddr):$(serverip):$(gatewayip):$(netmask):$(hostname):$(netdev):off

Right,  we  use  PPCBoot  as  firmware  and  get  all  the  "dynamic"
configuration   by  PPCBoot's  feature  to  define,  store,  and  use
variables.

> If you are truly using a dynamic environment where you could plug in
> anywhere and have the application automatically adapt, you need to
> follow through with dynamic userland configuration.

... not when we can pass all the necessary parameters on the  command
line (or in bi_recs :-)

Wolfgang Denk

--
Software Engineering:  Embedded and Realtime Systems,  Embedded Linux
Phone: (+49)-8142-4596-87  Fax: (+49)-8142-4596-88  Email: wd@denx.de
I have made mistakes, but have never made the mistake of  claiming  I
never made one.                                     - James G. Bennet

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/