I have a backtrace finally, and this is repeatable, always
with essentially the same postamble after the interrupt (see below).

I forgot about reboot 0x104 -> generate dump; I will do another crash 
and use gdb.

Before the interrupt I have seen various other things,
including below and this one, very badly transcribed onto paper.

 --- interrupt ---
softdep_disk_io_initiation+0x18c
spec_strategy+0x26
ufs_stratgegy
bwrite
vn_bwrite
bawrite
ffs_fsync
ffs_sync
sync_fsync
sched_sync

which started me on some experimentation.  Latest crash:

(copied by hand from a kernel based on (i386) DR-EVIL.4MB)

cpu_Debuger(c4a6ac50,de000efcc,400,d9ecabf0,c01c8e48) at cpu_Debugger+0x4
panic(c0278880,c4a6ac50,d000efcc,400,d9c45aec) at panic+0x65
softdep_disk_write_complete(c4a6ac50) at softdep_disk_write_complete+0x1c
biodone(c4a6ac50,c07e0c14,2ed71c,c073082c,400) at biodone+0x4b
wddone(c0730800) at wddone+01d6
wdc_ata_bio_done(c072e8b0,c072f000,c072e8b0,c072f000,d9c416158) at wdc_ata_bio_done+0x8b
wdc_ata_bio_intr(c072e8b0,c072f000,1,c0719ba0,c4aa3884) at wdc_ata_bio_intr+0x3a0
wdcintr(c072e8b0,d9ecada0,c01019c8,c072e8b0,0) at wdcintr+0xb5
pciide_compat_intr(c072e8b0) at pciide_compat_intr+0x22
Xintr14() at Xintr14+0x70
 --- interrupt ---
softdep_update_inodeblock(d9c46158,c4aa3884,0) at softdep_update_inodeblock+0xc6
ffs_update(d9ecae5c) at ffs_update+0x1f9
ufs_inactive(d9ecae88,d9c45aec,c04082c0,d9c45aec,d9eba1a8) at ufs_inactive+0x12c
vput(d9c45aec,d9981274,d9981274,d9eba1a8,c0407d40) at vput+0xd5
vn_closefile(d9981274,d9eba1a8,d997d748,d9981274,6) at vn_closefile+0x67
closef(d9981274,d9eba1a8,d9ecaf88,d9eba1a8,d9ecaf80) at closef+0x1f2
fdrelease(d9eba1a8,6,d9ecafa8,c01f80c2,d9eba18a) at fdrelease+0xb6
sys_close(dieba1a8,d9ecaf88,d9ecaf80,0,0) at sys_close+0x1a
syscall() at syscall+0x206
 --- syscall (number 6)
0x4811060b:

db> ps/w
...
7137    cpp     netbsd  79      0.0     0.0