>> heres a latest manifestation of crashes on i386 with recent (july srcs).
>> nfs running between two i386 on the same kernel: 3c509 and WD8013 cards,
>> nfs running with -r/-w 4096 (would lock up w/o a -r/-w limit).  
>> 
>> Jul 11 17:34:36 atrium /netbsd: Data modified on freelist: word 0 of object 0xf8
>> 6cc980 size 128 previous type free (0x0 != 0xdeadbeef)
>> Jul 11 17:34:36 atrium /netbsd: Data modified on freelist: word 0 of object 0xf8
>> 6cc200 size 128 previous type free (0x0 != 0xdeadbeef)
>> Jul 11 17:34:36 atrium /netbsd: Data modified on freelist: word 0 of object 0xf8
>> 6cc214 size 128 previous type ??? (0xa040210 != 0xdeadbeef)
>> Jul 11 17:34:36 atrium /netbsd: nfsd send error 47
>> 
>
>are these new panics, or old?  are you using PPP?  do they go away if
>you stop using PPP?

Prior to mid-June/July updates, my i386 was rather stable.  Now the machine
which runs a slip connection crashes ~4-5 times a day.  I use nfs (though I
may need to try without it) and in the latest kernel set NMBUF... = 1024.

I've been at console once at crash and it claimed a vm fault.  Mostly I only
get: Jul 13 07:05:41 atrium savecore: reboot after panic: trap
but today I got:
Jul 13 12:47:47 atrium savecore: reboot after panic: free: unaligned addr 0x%x, size %d, type %s, mask %d 

These problems occur ONLY on the machine running slip.  The other machine 
running X and nfs (on the same kernel) appears to be perfectly happy.  MBUF
use does not seem excessive on either machine.  I may shift the slip line to
eliminate the possibility of flaky hardware.

-john