i'm trying to get familiar with cwfs, and i'm having issues. i'm
trying to use plain files for playing around, which cwfs(4) indicates
should be fine. i've got a directory /tmp/cwfs with a bunch of files
in it: confdisk0, fakedisk[0-3], and devmap. devmap has lines like:
w0 /tmp/cwfs/fakedisk0
w4 /tmp/cwfs/confdisk0
for all of w[0-4]. invoking cwfs as 'cwfs -m devmap w4' yields the following:
config diag: unknown type --
for each line in devmap. from playing around, it seems to only assume
the second argument on the line is a device type, never a plain file
(at least not there). changing the filename to fakedisk0 from
/tmp/cwfs/fakedisk0 causes it to complain that "a" is an unknown type;
it's trying to read it as an fs(4) device.

i'm then dumped into the config system, just before which it gives me
the promising line "map: mapped wren w4 to existing file
/tmp/cwfs/confdisk0". i enter the config i'd like to use:
service arkive
filsys main cw4f[w<0-3>]
filsys dump o
ream main
after the main line, i get one line of the above "map: mapped wren..."
form for each of w[0-4], all of which look correct. the ream seems to
have no effect. when i type end, however, i get this:
sysinit
config w4
map: mapped wren4 to existing file /tmp/cwfs/confdisk0
devinit w4
cwfs 50437: suicide: sys: trap: fault read addr=0x0 pc=0x00023efc
and a console print "50437 cwfs: checked 30822 page table entries".

the man page certainly implies with the dev-map example that this is
possible, although there's no example of it. i see some code for cases
when using plain files vs. sdXX files (with /data), but couldn't
follow where the condition was set. is this usage known to work? are
there further constraints i should be aware of?