ethernet checksum errors

[G. David Butler gdb@dbSystems.com]

Hello Plan9 fans!

I have a PC file server, PC cpu server and PC terminal servers.  They
are all connected via 10b2 using WD8013 cards.  When I cause large
amounts of I/O on the file server (using any command including
terminal servers paging to the file server), I get either of these:

from stil.c il: cksum error, pkt(...)
from stip.c ip: checksum error (from ... ?)

from both the file server and the machine doing the I/O (cpu server or
terminal server).

If enough of these happen it results in a machine panic.  Most times
the file server, but others also.

The question is, are the systems in question missing interrupts so that
the ethernet buffers are overwriten?  Could that get bad enough so
a "can't happen" bug causes the machine to panic?

I have not looked hard enough yet to determine the tools available
for kernel debugging, but can someone give me enough of a start to
find this one?

The system has all updates applied.

sum stip.c: c44ec0e0 14160
sum stil.c: 918ba6d0 15970

Thanks for any help.

David Butler
gdb@dbSystems.com