I had a minor run-in with vnd and ccd together; I constructed the
following as an example showing what goes wrong.

    [Callisto] 21> dd if=/dev/zero bs=1048576 count=10 of=file.0
    10+0 records in
    10+0 records out
    10485760 bytes transferred in 12 secs (873813 bytes/sec)
    [Callisto] 22> dd if=/dev/zero bs=1048576 count=10 of=file.1
    10+0 records in
    10+0 records out
    10485760 bytes transferred in 12 secs (873813 bytes/sec)
    [Callisto] 23> dd if=/dev/zero bs=1048576 count=10 of=file.2
    10+0 records in
    10+0 records out
    10485760 bytes transferred in 11 secs (953250 bytes/sec)
    [Callisto] 24> vnconfig -c /dev/rvnd0c file.0
    [Callisto] 25> vnconfig -c /dev/rvnd1c file.1
    [Callisto] 26> vnconfig -c /dev/rvnd2c file.2
    [Callisto] 27> ccdconfig ccd0 0 0 /dev/vnd0c /dev/vnd1c /dev/vnd2c
    ccdconfig: ioctl (CCDIOCSET): /dev/ccd0c: Inappropriate ioctl for device
    [Callisto] 28> 

What am I missing here?  Why won't ccd and vnd work together?

(In my initial use, the three files were different sizes and spread
across three filesystems; I wanted to store a huge file temporarily.
No single filesystem had enough space, but put together they did; I was
willing to pay the performance penalty....)

And yes, /dev/ccd0c is block special, with the correct major number for
the ccd driver.  The kernel has 4 each vnd and ccd configured in.

In case it matters, this was on NetBSD/sparc, everything built from
June 10th sources.  Somewhat patched, but my patches don't go near
either ccd or vnd....

					der Mouse

			    mouse@collatz.mcrcim.mcgill.edu
		    01 EE 31 F6 BB 0C 34 36  00 F3 7C 5A C1 A0 67 1D