Panic condition 'bad rmfree'

[KEVINS at s63.prime.com]

We have recently recieved a 3/280 running SunOS 3.5. Currrently there are
10 diskless 3/60 clients and 2 3/60 standalones on standard eithernet.
Otherwise the net is empty. All entries to the net are Muxed ( except the
3/280).

Our problem is random panics (3/280) with a condition of 'bad rmfree'.
Virtually everything on the 3/280 has been replaced, including the
backplane.

The kernel is generic and has been rebuilt from other 3/280's in house.

Sun's suggestions have been to replace the memory boards, cpu, increase
swap space to 50M and run a generic kernel. All of which we have done to
no avail. Subsequently we have  replaced just about everything.  Our only
non-Sun element of the system is Aviv 450/451 controller adaptors.  Which
have also been swapped out (with more Aviv's).

We have also tried eliminating all clients.

Occasionaly we have lost /usr. Mounted ro to all clients. When we rebuild
/usr and regenerate our kernel the system usually stays up and error free
for 30 - 36 hrs. Otherwise we are lucky to stay up for 2 hrs.

Any help would be greatly appreciated.

Kevin Sullivan
(508) 879-2960 ext. 3580 (other home) 4389
email: kevins at s63.prime.com