v100 and the tlp driver

[Robin-David Hammond %KB3IEN <sparc64%kb3ien.us@localhost>]

i have several netbsd4.0 machines with both onboard ethernets configured. 
Everything seemed to be going just fine, then i noticed a corrolation 
between visits to the datacentres and systems requiring a 3-wire reboot.

(On site dude pulls the 3 pronged cable from the back of the chasis)

Took one home and began playing with it to turn the crashes into 
reproduceable phenomena, here's what i found:

1 setup tlp0 and tlp1 with ip addresses

2 connect both devices to respective switches.

3 ssh into tlp0

4 unplug tlp1

5 ssh hangs (so does EVERYTHING else)

6 connect to console

get this crap:

tlp1: incoming packet spilled, resetting
tlp1: receive ring overrun

repeated, ad nauseum, reconnecting the cable seems to let the kernel 
unfreeze, but i'd apreciate some help narrowing down the busted code 
because i use the 2nd ethernet jack to provide redundance if the first 
switch fails in some installs.

I do have altq compiled in, but not activated. even with ip forwarding =0, 
no routeing much less NAT, the problem remains.

If the kernel boots with no sync on a configured port, everything is fine 
until the first Sync->NoSync states change.

TIA

kb3ien