In article <slrnb4j0it.124.apost@water.amer.interwoven.com>,
Alan Post <apost@interwoven.com> wrote:
>In article <slrnb4b0iq.q0q.apost@water.amer.interwoven.com>, Alan Post wrote:
>> In article <HA02zI.4A7@tac.nyc.ny.us>, Christos Zoulas wrote:
>>> 
>>> Try changing:
>>> 
>>>         if ((nmp->nm_flag & NFSMNT_NQNFS) && np->n_timer.cqe_next != 0) {
>>> 
>>> to:
>>>         if (np->n_timer.cqe_next != 0 && (nmp->nm_flag & NFSMNT_NQNFS)) {
>>> 
>>> 
>>> [yes this looks stupid, but I have the suspicion that nmp is invalid somehow
>>>  and since you are not using leases, the timer will always be empty...]
>>> 
>> 
>> Thanks.  I'll give this a shot on Monday when I have physical access
>> to the box (panic on remote machine == bad :).
>
>So far, so good.  With the above patch, last night the machine
>survived /etc/daily, including find_core, with amd enabled.

Grr, I would be lying to you if I said I would be happy that this is the
case. It seems that we are forgetting to lock the mountlist somewhere, it
gets free'd and then we access it. Did you get a core dump of the last
crash?

christos