> A little UTSLing makes it clear what's going on here.  swread(), the
> read routine for the swap-space driver (in vm/vm_swap.c), reads

> But when I look at physio() (sys/kern_physio.c), it starts

Nice catch by der Mouse, but unfortunately, fixing this has increased
my crash frequency to about every 4 hours.  I am attaching a list
of files on my system that have changed since July 4.  I am most
suspicious of the changes in vfs_subr.c and vfs_bio.c.  Just a 
hunch.  Can anyone supply files prior to these dates?

Hmmm.. Must go see if the tarballs are older....

./arch/i386/i386/conf.c
./arch/i386/include/psl.h
./arch/i386/include/types.h
./arch/i386/isa/isa_machdep.c
./arch/i386/isa/fd.c
./dev/vnd.c
./dev/ccd.c
./dev/isa/aha1542.c
./dev/isa/bt742a.c
./dev/isa/com.c
./dev/isa/comreg.h
./dev/isa/mcd.c
./dev/isa/mcdreg.h
./dev/isa/wd.c
./dev/isa/wt.c
./dev/isa/ad1848.c
./dev/isa/ad1848var.h
./dev/isa/pas.c
./dev/isa/pss.c
./dev/isa/sb.c
./dev/isa/sbdsp.c
./dev/isa/wss.c
./dev/isa/wssreg.h
./dev/isa/cs4231var.h
./dev/ic/ad1848reg.h
./dev/ic/cs4231reg.h
./dev/ic/ncr5380.c
./dev/ic/ncr5380reg.h
./dev/ic/ncr5380var.h
./dev/audio.c
./dev/audio_if.h
./kern/vfs_bio.c
./kern/kern_physio.c
./kern/vfs_subr.c
./miscfs/specfs/spec_vnops.c
./miscfs/union/union_vnops.c
./net/netisr.h
./net/if_ppp.h
./net/slcompress.c
./net/slcompress.h
./net/if_ppp.c
./net/if_arcsubr.c
./net/bsd-comp.c
./net/if_pppvar.h
./net/ppp-comp.h
./net/ppp_defs.h
./net/ppp_tty.c
./netinet/ip_input.c
./netinet/raw_ip.c
./netiso/tuba_usrreq.c
./scsi/st.c
./scsi/sd.c
./scsi/scsiconf.c
./scsi/scsiconf.h
./scsi/cd.c
./sys/conf.h
./sys/msg.h
./sys/audioio.h
./vm/vm_object.c
./vm/vm_meter.c
./vm/vm_swap.c