0.91g and Kingston KNE100TX problems

[Donald Becker]

On Wed, 28 Jul 1999, Kevin Myer wrote:

> A few months ago, I replaced some 3Com 905B-TX cards in two servers I had,
> ...Kingston KNE100TX.
...
> However, I am seeing massive errors on these cards and I don't now why.
> And although the linkbeat is there and the cards negotiate at 100/FD
> speeds with the switch, I am only getting around 220Kb/s transfer rates
> between the two machines.

You can verify that the Tulip chips are set to FDX by using the 'tulip-diag'
program
  http://cesdis.gsfc.nasa.gov/linux/diag/index.html
I doubt that this is the problem.

> eth0      Link encap:Ethernet  HWaddr 00:C0:F0:3B:D0:E8  
>           RX packets:28220223 errors:1 dropped:45 overruns:0 frame:1
>           TX packets:39811549 errors:22859571 dropped:0 overruns:1
> carrier:22859570

Acckkkk!!!  That's horribly bad.
Check that your cables are paired correctly.
Check /proc/net/dev to see what category the errors are in.

> [myer@marble myer]$ /sbin/ifconfig
> eth0      Link encap:Ethernet  HWaddr 00:C0:F0:3B:C0:5B  
>           RX packets:67576 errors:0 dropped:0 overruns:0 frame:0
>           TX packets:714 errors:50291 dropped:0 overruns:0 carrier:50291

Some is horribly, horribly, wrong.

> Its probably not coincidence that on both machines, the errors are always
> occuring when transmitting.  It almost seems like a problem with a
> mismatch in duplex between the cards and the switch but mii-diag reports
> on both machines:

Yes, I would normally guess a duplex mismatch, with the cards being left in
half duplex mode while the switch is in full.

You can test this with 'tulip-diag' and with 'ping -s 1400' on an idle
network. Since ping waits for responses, there will almost never be
collisions.

> ./mii[myer@gneiss myer]$ ./mii-diag
> Using the default interface 'eth0'.
> Basic registers of MII PHY #1:  3000 782d 0016 f831 01e1 41e1 ffff ffff.
>  Basic mode control register 0x3000: Auto-negotiation enabled.
>  You have link beat, and everything is working OK.
>  Your link partner can do 41e1: 100baseTx-FD 100baseTx 10baseT-FD 10baseT.

That look normal.

Donald Becker					  becker@cesdis.gsfc.nasa.gov
USRA-CESDIS, Center of Excellence in Space Data and Information Sciences.
Code 930.5, Goddard Space Flight Center,  Greenbelt, MD.  20771
301-286-0882	     http://cesdis.gsfc.nasa.gov/people/becker/whoiam.html