Paul Goyette writes:
> Looks like I may have been a bit over-optimistic when I reported that
> my amd64 system was running fine with a kernel built from Nov 1 .
> 
> When I woke up today, all my sessions were completely locked up, and
> there was no response, not even to a request to switch my wscons back
> to console/ttye0.  Yet, it still responded normally to a 'ping' from
> another box on the same network.  This is the second day in a row
> with identical symptoms.
> 
> It _may_ be disk I/O related, as both times my nightly Amanda backup
> was in progress at the time of the hang.  A i386 system right next to
> it, built from the same sources, had no trouble, but it wasn't doing
> any heavy I/O.
> 
> Since this is repeatable, I'd certainly like to resolve it.  It's
> beyond my expertise level, but if anyone has any good ideas to try,
> I'm more than happy to cooperate.

Try a -current kernel from today, or at least one with everthing up 
to and including this fix: 

 cvs rdiff -r1.172 -r1.173 src/sys/kern/kern_synch.c

With this change (and other recent fixes to kern_sig.c and 
kern_synch.c), I've been unable to reproduce the hangs I was 
seeing on my Core Duo laptop... 

Later...

Greg Oster