Hey,
  After upgrading from -test11 to 2.4.0, I find that under heavy network
load the eth0 interface seems to lockup... with the following output in
dmesg:

NETDEV WATCHDOG: eth0: transmit timed out
eth0: Tx timed out, lost interrupt? TSR=0x3, ISR=0x97, t=18556.
NETDEV WATCHDOG: eth0: transmit timed out
eth0: Tx timed out, lost interrupt? TSR=0x3, ISR=0x3, t=283.
NETDEV WATCHDOG: eth0: transmit timed out
eth0: Tx timed out, lost interrupt? TSR=0x3, ISR=0x3, t=56.
NETDEV WATCHDOG: eth0: transmit timed out
eth0: Tx timed out, lost interrupt? TSR=0x3, ISR=0x3, t=56.

  This just happened while I was scp'ing a 50meg file across the network,
but has happened a couple of times in the last few days just after opening
netscape navigator (i have an isdn line, so the traffic wouldnt have been
particularly heavy in this case). If I ifdown/ifup eth0, it makes no
difference... nor if i remove & re-insert the ne2k-pci module as well.

dual-proc p3 600, 512meg ram

Linux fusion 2.4.0 #4 SMP Wed Jan 10 11:36:36 GMT 2001 i686 unknown

gcc version 2.95.3 20001229 (prerelease)

Jan 10 19:59:37 fusion kernel: eth0: RealTek RTL-8029 found at 0xb800, IRQ 18, 00:00:E8:EF:36:E2.

[jon@fusion ~] $ cat /proc/interrupts 
           CPU0       CPU1
...
 18:      37311      37308   IO-APIC-level  eth0

.config is attached... any ideas on how I could better debug this problem
when it happens again?

-- 
Jon / Cybah@IRCNet