On Thu, 4 Jul 1996 06:48:50 -0400 
 der Mouse <mouse@Collatz.McRCIM.McGill.EDU> wrote:

 > I suppose then my question becomes, why does ccd care?  I certainly
 > can't see any excuse for it to be poking around with disk labels; it
 > seems to me it should do what it's told with a minimum of fuss.  Why
 > would anyone want a ccd or vnd to have anything to do with disklabels
 > anyway?  Again, I must be missing something; someone must have
 > deliberately made ccd care about disklabels, and I must confess I am
 > completely unable to understand why.

Originally, Mike Hibler (when he wrote the first N versions) made it care.

This is what it does; it uses the DIOCGPART ioctl, which is used only 
inside the kernel for various things.  This returns a pointer to the 
partition requested and a pointer to the disklabel.  This information is 
used to obtain:

	- component partition size

	- component partition fstype

The former is more important than the latter, though the ccd disallows 
use of partitions != FS_BSDFFS (we've had this discussion on 
current-users before, regarding ccd and fstypes.

It should be pretty obvious why component partition size is important :-)

I later added support for disklables in the ccd itself, but this is 
different from inquiring the components' partition layouts.

The ccd _does_ do what it's told to with a minimum amount of fuss... It's 
_very_ simple to use with components that support all of the necessary 
features.  Now, when I check in some changes to the vnd "soon", it will 
do the same ... an image of a filesystem will have no disklabel, so a 
"fake" in-core label will be created (which is exactly what happens with 
the ccd, sd, cd, etc.)  You don't have to explicitly put a label on a vnd 
(that would be rediculous :-)

One could argue that the ccd could be hacked to just use "all" of a vnd.  
My response to that is:

	- should make the "disk" drivers somewhat consistent

	- There are many benefits to supporting real disklabels in
	  the vnd.

The latter was used to save the data on sun-lamp (remember that box? :-)
when one of its disks started to fail.  Chris was able to get a disk 
image to me, I was able to read the actual disklabel and access the 
individual filesystems as if it were a real disk.  I imagine this could 
be useful in other applications (like making disk images, etc).

Anyhow, hope that answers your question.

 -- save the ancient forests - http://www.bayarea.net/~thorpej/forest/ -- 
Jason R. Thorpe                                       thorpej@nas.nasa.gov
NASA Ames Research Center                               Home: 408.866.1912
NAS: M/S 258-6                                          Work: 415.604.0935
Moffett Field, CA 94035                                Pager: 415.428.6939