can not get IP for eth0

[re_box]

I am trying to get networking up on an old PII which I have successfully installed with Fedora Core 4.

There are link and activity lights on the onboard interface. The controller is a 3Com PCI 3c905 Boomerang 100baseTx using address 0xdc80.

The driver module, 3c59x, is loaded as well as the mii module that uses it. There is an ipv6 module loaded but it is not clear if there is or is needed an ipv4 module.

eth0 is sharing an interrupt with the UHCI controller. That should not be a problem.

When I drop to console and runlevel 3, and call dhclient, which is installed, I get message telling me there is no offer to the dhcp request. But I am fairly certain there is not a problem on the router/switch side.

Does anyone have an idea what might be amiss here?

[MikeTbob]

The first thing I'd do is grep dmesg for any errors related to eth0

dmesg |grep eth0

You also need to make sure the correct module is being loaded, you need to find out which chipset is being used by the card.

lspci -nn |grep -i net

Post back the results please
Have you tried using the network tool to connect? Is this not working?Which network tool is being used? network manager?

[re_box]

The only thing dmesg says about eth0 is " Dropping NETIF_F_SG since no checksum feature."

The ethernet controller is 3Com, as I enumerated above.

According to what I can find out, the 3c59c driver is the one for this controller. Although there is one objection in dmesg to the effect "Unknown param `irq'."

I am beginning to think this problem might be unsolvable. Could there be a burnt-out circuit in the south bridge controller?

[Lazydog]

The only thing dmesg says about eth0 is " Dropping NETIF_F_SG since no checksum feature."

The ethernet controller is 3Com, as I enumerated above.

According to what I can find out, the 3c59c driver is the one for this controller. Although there is one objection in dmesg to the effect "Unknown param `irq'."

I am beginning to think this problem might be unsolvable. Could there be a burnt-out circuit in the south bridge controller?

Give yourself an ip address on the same network as the switch. If the switch is setup as 192.168.1.1 then give yourself 192.168.1.10 and ping the switch. If that works then the problem is not your interface.

If the above worked then I would check to ensure that the switch is setup properly and ensure the DHCP is turned on.

[re_box]

The problem is that does not work. The ping replies "Destination Host unreachable"

I have tested the cable repeatedly. All four pairs light up the tester.

What is really weird -- this old Dell had one of those back-plane PCI/ISA expansion boards. I tried a plug-in PCI nic without any luck, to bypass the onboard nic -- same story.

The BIOS is a joke. There is a control for on/off, but that is it.

Which brings me back to some low-level hardware malfunction at the south bridge.

[Lazydog]

Do you have a loopback plug or could you make one? That would be a good test to see if the port is the problem.
Have you tried different ports on the switch?

[re_box]

That's a good idea! I assume the loopback would entail an rj45 and a length of cable with each pair wired to itself.

Then what?

[re_box]

Okay, I'm an idiot. It was the cable. That cable tester lied to me.