I've been having trouble with these panics ever since I upgraded to post-1.2
current.  I found a reference to this problem in the new options(4) man page
(thanks, Perry!) which claimed that raising the value of the kernel config
option NKMEMCLUSTERS can make this go away.  I've been running a kernel with
this option doubled (to 1024 from the mac68k port default of 512) for a
few days, and I just saw the panic again.  It seems that raising the number
delayed the problem, but didn't really fix it.

This brings up two questions.

- What does this option actually indicate?  I assume it's a number of entries
    in a table somewhere.  Would someone care to elaborate?  Just curious...

- What number should I set it to to avoid this panic completely?  Is it 
    based on some system resource?  I'd really rather not just blindly 
    keep doubling the number -- I hate that kind of voodoo solution.  The
    machine in quesion has 20M of memory and is acting as an ethernet/slip
    gateway, if it matters.  (Having one's router panic when one is trying 
    to access one's site from a remote location can ruin one's whole day, 
    if you know what I mean...  ;-)

(BTW, is there a reason that _none_ of the sample config files for the
mac68k port have the option MROUTING included?  I wanted to build a kernel
with multicast routing enabled, and it took me a while to figure out what
the option was called.  I _knew_ there was a reason I supped the other arch
directories...)

-- monroe
----------------------------------------------------------------------------
Monroe Williams                                          monroe@teleport.com