>IIRC NetBSD 1.2 does not use the clean mark.

i think it was much earlier than that that it was actually added.  i'm
running 1.2 on a few machines and none of them exhibit this
behavior...

>> Any ideas what this might be?
>
>At boot time, fsck -p checks the raw devices (/dev/rwd0a in this case)
>and mark it clean, but mount -a later tries to mount block devices
>(/dev/wd0a) and svears about block device not being marked clean. Mount
>will not complain if you do fsck /dev/wd0a by hand and then remount it.
>
>Perhaps could be fsck changed to mark clean both raw and block device?
>Or is there some deeper reason for this behavior?

sorry, no.  the block and char (or raw) devices are just different
views of the same media.  thus, if the raw (or char) device (which is
what fsck uses so that it is unaffected by kernel disk buffering?) is
marked clean, then the block device should also be clean.

-- 
|-----< "CODE WARRIOR" >-----|
andrew@echonyc.com (TheMan)        * "ah!  i see you have the internet
codewarrior@daemon.org                               that goes *ping*!"
warfare@graffiti.com      * "information is power -- share the wealth."