BOOTP problems on Embedded Planet CLLF with Monta Vista kernel

Paul Fagerburg PaulFagerburg at
Fri Aug 4 09:43:18 EST 2000

I'm working with Embedded Planet's CLLF board (an 860T), and the Hard
Hat Linux Journeyman edition from Monta Vista.  My host system is an
intel celeron box running Mandrake 7.

I've been referring to the Hard Hat Linux docs (cdk.pdf) and the
getting_started.rpx file that is available on several of the linuxppc sites.

The mandrake host (named "penguin") is part of our NT network, and is a
properly-behaved DHCP client.  Minicom works fine talking to the CLLF

Once I started dhcpd (the DHCP *server*), nothing went right from then
on; the TFTP client on the CLLF timed out with ARP failures.  Once I
stopped dhcpd, the CLLF was able to TFTP its kernel, uncompress it, and
boot.  However, after booting the kernel, it wants to use BOOTP to get
more info and then NFS mount the root filesystem.  This is where it all
goes downhill.

Thinking that there was some kind of issue with our NT box that is
serving up DHCP, I tried putting the CLLF and penguin on their own
network.  I put a second NIC in penguin and hooked up a dumb hub.  Since
we're using 192.168 addresses in our NT network, I decided to use a 10
network on the second NIC. I gave penguin the IP of, and told
the CLLF (in its EEPROM config) that it was

I boot the RPX, tell it to load and execute from TFTP, and it does in
fact boot the kernel.  After the kernel boots, it tries to use BOOTP,
and gets no responses.  I used tcpdump to see the traffic on eth1, and
here's the amended version:

13:30:16.443429 arp who-has tell rpx
13:30:16.443465 arp reply is-at 0:c0:f0:5b:9e:23
13:30:16.557802 rpx.1 > 22 RRQ "vmlinuz.srec"
13:30:16.576315 > rpx.1: udp 516
13:30:16.577263 rpx.1 > udp 22
[ ... ]
13:30:22.830821 > rpx.1: udp 516
13:30:22.833848 rpx.1 > udp 22
13:30:22.833915 > rpx.1: udp 493
13:30:22.836905 rpx.1 > udp 22
13:30:23.569352 arp who-has rpx tell
13:30:24.569343 arp who-has rpx tell
13:30:25.569352 arp who-has rpx tell
13:30:26.569348 arp who-has rpx tell

When the four arp packets show up at the end, I see
    Linux/PPC load: root=/dev/nfs rw
I hit ENTER, it goes through the kernel load, and then gets stuck
waiting for BOOTP.

Here's what I hope will be useful data:

# route
Kernel IP routing table
Destination     Gateway         Genmask         Flags Metric Ref    Use
Iface        *      UH    0      0        0 eth1     *        U     0      0        0 eth0        *            U     0      0        0 eth1       *            U     0      0        0 lo
default         venus          UG    0      0        0 eth0

# cat /etc/hosts   penguin localhost venus    rpxhost    rpx

# cat /etc/exports
/opt/hardhat/devkit/ppc/8xx/target (rw,no_root_squash)

# cat /etc/dhcpd.conf
allow bootp;
subnet netmask {

subnet netmask {
    default-lease-time 1209600; # two weeks
    max-lease-time 31557600;    # one year
    group {
        host rpx {
            hardware ethernet 00:10:EC:00:19:F4;
            option root-path "/opt/hardhat/devkit/ppc/8xx/target";

I had to add that first subnet to dhcpd.conf, or else the daemon
wouldn't start, complaining about not knowing what to do with eth0.

So, that's where I'm stuck for now.

Surely I can't be the only person trying to hook up one of these boards
to a linux system that is sitting in an NT network.  Separate NIC and
hub 's is not an issue to me; show me the way to do it with either 1 or
2 NIC's and I'll be happy to do it that way.  I'll be quite grateful for
any help that anyone can provide.


Paul Fagerburg
Renaissance Network Technology
829 De La Vina St., Suite 200
Santa Barbara, CA 93101
805-899-4260 x 115
plf at

** Sent via the linuxppc-embedded mail list. See

More information about the Linuxppc-embedded mailing list