[tulip] PCI bus fatal errors with Linksys LNE100tx

[Donald Becker]

On Wed, 20 Sep 2000, Mario Simoni wrote:

> I'm trying to run a Linksys 10/100 ethernet card using the tulip driver.
> I'm running an old P90 machine with 3 PCI ports and 3 ISA ports.
...
>     Ethernet controller: Unknown vendor Unknown device (rev 17).
>       Vendor id=1317. Device id=985.

This is the ADMtek Pegasus/P chip, which is a PCI v2.1 chip.

Your old P90 system is likely PCI v1.0.  This *might* be the cause of the
reported bus errors.

> I am new to Linux but have read a lot on the tulip driver. I have
> downloaded and successfully installed the .rpm file, but I have to force
> the driver to install using 
> 
> insmod pci-scan
> insmod tulip

Does 'modprobe tulip' work?
If so, you should configure your driver and interface using the tools
provided in your distribution.
If not, please report it.

> When I then type ifconfig eth0 I get: 
> 
> eth0      Link encap:Ethernet  HWaddr 00:20:78:1E:53:F4  
>           UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
>           RX packets:144 errors:0 dropped:0 overruns:0 frame:0
>           TX packets:0 errors:0 dropped:0 overruns:0 carrier:0
>           collisions:0 txqueuelen:100 
>           Interrupt:11 Base address:0x3c00 

This looks fine, although you have not assigned an IP address.

> If I try to use ifup eth0 I get:
> 
> SIOCSIFMAP: Operation not supported
> SIOCSIFMAP: Operation not supported

You are trying to set some driver parameter that is not supported.  You must
have configured some bogus parameter...

> I don't understand why the io_address for eth0 in /proc/pci (0xfc00) is
> different from ifconfig (0x3c00). Does the error mean a hardware problem
> with my network card or PCI bus? or Is there a problem with my kernel
> loading the pci-scan.o module? 

... are you trying to force the I/O address?  If so, that would explain the
driver not loading correct and the IFMAP errors.

> So then I checked out the /var/log/messages and saw that I was getting
> eth0:PCI fatal bus errors:
> 
> Sep 20 01:40:01 rivendel kernel: eth0: PCI Fatal Bus Error, fc67e050. 
> Sep 20 01:41:36 rivendel kernel: PCI: PCI BIOS revision 2.00 entry at
> 0xf0e40 
> Sep 20 01:41:36 rivendel kernel: PCI: Using configuration type 2 
> Sep 20 01:41:37 rivendel kernel: PCI: Probing PCI hardware 
> Sep 20 01:41:37 rivendel kernel: RZ1000: IDE controller on PCI bus 00
> dev 08 
> Sep 20 01:44:52 rivendel kernel: eth0: PCI Fatal Bus Error, fc66e010. 
> Sep 20 01:44:52 rivendel kernel: eth0: PCI Fatal Bus Error, fc67e050. 
> Sep 20 01:45:28 rivendel kernel: eth0: PCI Fatal Bus Error, fc67e050. 

Run 'tulip-diag' after this happens.

This is reporting a PCI parity error while transferring a received packet
into main memory.  The driver should resume normal operation, but I wouldn't
trust the data transfered on the system.

Donald Becker				becker@scyld.com
Scyld Computing Corporation		http://www.scyld.com
410 Severn Ave. Suite 210		Beowulf-II Cluster Distribution
Annapolis MD 21403