Problem with TLAN ethernet driver in 12.1, 12.2 stable
 

I have Compaq Proliant 1600's with integrated TLAN 0xae43 10/100 UTP lan. Something is not right with the LAN starting with Slackware 12 / Kernel 2.6.

I have run Slack 3, 8, 10, and 11/SCSI.s on them with no problems. I have been unable to run any version of Slack 12. I actually have 2 problems, the second being no mouse, but that is minor and not needed. These boxes are mostly for DNS and http.

I have SCSI drives I can pop in and out of their bay. With several disks, I am able to boot into any version by the disk that I have loaded. So I can quickly test things and dd things back to clean stock "full" installs... I have both a hugesmp.s and a custom built 12.2 version with tlan bulit in.

When I boot into 12.x and run TCPDUMP, I see STP, ARP, RIPV2, etc packets coming IN. When I try to ping out, I see plenty of ARP "who has gateway..." requests going out, but nothing is responding. On the term screen I basically get a message from my eth0 IP address saying destination is unreachable. These servers have static IP's connected to Cisco Catalyst switches.

When I switch from a 12.x boot to a 11.0 boot, I have to let the box completely power down and rest for a couple of minutes or the problem sometimes shows up in the 11.0 booted version. And I can't set the IRQ above 9, or 7 really, because the card won't initialize.

I should also point out that in the earlier versions of tlan.h, there was a definition for this Ti chipset as 0xae43. However, this definition, and a few others are missing. May not be relevant, since the TLAN driver is loading. Just does not appear to be operating. I'm not a programmer.

I've not spent a lot of time with Slackware lately, and there are lots of new features. So if anyone has any suggestions, even the obvious, or something I can try, I would certainly appreciate hearing from you.

Thanks,
Steve Kinkaid

Hi,

It sounds like a hardware failure. You did check the cables? Do you have a spare LAN card to see if that will work for you? The indicator seems to be the failure when you roll back to a stable 11 that worked but now fails.

Thank you for posting your thoughts. I am certain it is not a hardware failure. (I'm using it right now.) The failure replicates to 11.0 only when I pull the hot-plug scsi drive from slot 0:1 during a ctrl-alt-del reboot, which allows scsi slot 0:2 to boot into slack 11.0 on warm boot. Since this nic is integrated, it is not possible to replace with an identical chipset. However I have confirmed this problem on 3 of the Compaq Proliant Servers with this chipset (1600, 1600r, 5500) I have.

Still Stumped.

Ok, after doing some more searching, looks like several have had the same problem around the time the driver was updated to version 1.16, which this distro has, even though it displays 1.15. But I never saw a solution.

Hopefully some more information will help jog some grey cells somewhere.

File 1 - Everything I can think of - dmesg, ifconfig, iptables, etc

File 2 - Tcpdump, which shows a link exists, just no data is getting out.

File 3 - Works - Using the older Slackware 11.0, everything works on the same hardware.

Thanks in advance!
Steve

As per the request for a move, moved to Hardware.

Latest TLAN driver in 2.6.29 works !
 

I found out the tlan driver is being worked on from the linuxhq site. So I installed the latest I could find (2.6.29) and I'm happy to say whatever was wrong with the driver has been fixed. I've tested it on both the Compaq integrated and the PCI Card Netelligent 10/100's.

Thanks to Sakari Ailus, wherever you are :D

Thought I would close this and provide the following for anyone else that runs into this. Looks like this same kernel release also has over 200 driver/net changes.

Steve