So I've been a slackware user for almost 10 years now and I've never come across this. I recently installed Slack 11 on a system that previously ran Slack 8.1 with a 2.4.20 kernel. My network cards did not work under the Slack 11 base kernel (tulip and 8139too modules).

I took the old kernel config and ran it through make oldconfig on 2.4.33.3 Slack 11 installed kernel. The modules built and modprobed just fine. None of the cards would transmit when plugged into any of the PCI slots.

After performing some network testing I found out that the cards are sending out ARP requests. The cards supposedly are not receiving any ARP replies. I've manually set ARP addresses but that has not resolved the problem either.

So I took a 2.4.29 kernel, built it off of the 2.4.20 config. Networking worked immediately.

My Hardware:
Mobo: Asus P3V4X
Proc: P3 733MHz
Mem: 512 MB
Net: Netgear and Linksys 10/100 BaseT/TX (using tulip module), some cheap card using the 8139too module

Does anybody have any idea why this is happening? I'd use a 2.4.29 kernel, but I want feature enhancements from the 2.6 series and this system is going to face the Internet eventually.

Those modules have worked fine for me in every kernel I used, so I guess the difference is somewhere else.
I'd suggest comparing the .configs of a working and a non-working kernel (in this case the 2.4.29 that works and the 2.4.33.3) and see what's different.

Well I did several comparisons, and effectively the same modules are selected between the two config files. I'm just set up Slackware 10.2 over the 11 install and I plan to test this shortly. One thing to mention about the 10.2 setup: it hangs on running rc.hotplug, so I'm already excited.

I installed 11.0 a while ago on my old computer, before giving it to someone else. Its ethernet card ran with the tulip driver, and it did not work. Well, it did, but each time the connection died out a few minutes after connection, usually during a high speed traffic (downloads etc).

This was on the 2.6 kernels. I used the testing/ and extra/ kernels, then compiled one myself but all was the same. The 2.4 kernel worked well however.

This computer was running 10.1 (i think, or 10.0) before 11.0, and I was able to use it with custom compiled 2.6 kernels. What is more, the latest Ubuntu on the same computer worked OK. It seems unlikely to be a 2.6 kernel version problem, but I can't imagine what network seeting in 11.0 could be causing this.

Wow. After cursing for the past hour I decided to post my solution.

First, I had the odd problems with Slack 11, then after reinstalling with Slack 10.2 my system would hang (actually rc.hotplug didn't cause the problems, running ifconfig eth0 up or ifconfig eth0 <addr> did). My solution? Disable "PnP OS" in my BIOS config. Everything works perfectly in both Slack 10.2 and 11.

Anyone else ever hear of something like this?

Well, then your problem should be different than mine. Glad that you've found a solution. I haven't heard anything like that before. In my case, I googled a lot but it didn't help either.

Yes, it happened to me this weekend. I have a new computer I've been struggling with and one of the problems was the e1000 module for the onboard network not working. I couldn't load it with modprobe. Disabling plug and play fixed it.

Yeah, I talked with a coworker today about it. He said that he's heard of people needing to disable PnP in their BIOS because it misbehaved. I figured that maybe the kernel developers changed the PnP code, or support for my motherboard. I booted with a copy of Slax I pulled down a month ago, and a copy of Knoppix from about the same time, both with newer kernels than 2.4.29 and both worked just fine. Maybe it's gremilins.