> I have seen this during shutdowns on my i386 box running -current too.
> I suspect that some of the most recent changes to the VM might be the
> cause of the trouble.  My machine hangs in vm_collaps_object()...

FWIW (and there is no 'evidence' here 8-):

Since the VM fixes, i've noticed that in certain cases a commonly-used
shell script on my 4M i386 system will sometimes wait 'forever',
blocked with a "thrd_sleep" wait message.  It had never happened
before the VM fixes.

Since the system has only 4M, it used to 'leak' swap quite heavily,
and so probably exercises the VM object collapse fixes a fair bit.


My guess would be that something is missing a wakeup...



chris