[Replying to two messages at once here.]

>>> Disks really *should* have unique ids.
>> No. File systems should have.
> I don't think I completely agree.

I certainly don't.  Filesystems should be *capable of* having unique
IDs.  There should not be any *requirement* that they do.

Disks too should ideally be capable of having unique IDs.  They too
should not be required to do so.  (It's probably less important for
disks than it is for filesystems.)

> Well, I just vote for kernfs extension or adding a dedicated fs that
> will map devices detected in the kernel boot stage into [filesystem
> objects]

Before dragging this out, please read the archives; this has been
discussed before.  (Probably even on this list, though I don't recall
specifically.)

> 		/kern/hardware/mainbus0/pci0/ncr0/scsibus0/target1/lun0/

This has problems, notably that this is not what the device attachment
tree looks like.  This path would correspond to

scsibus0 at ncr0 ...
target1 at scsibus0 ...
lun0 at target1 ...

but that's not what you have.  Perhaps it should be; perhaps that's
what you want to advocate.  Or perhaps you are proposing a new twist on
devfs, in that you are suggesting a filesystem hierarchy that doesn't
quite mirror the device attachment hierarchy.  But it's not clear, and
in any case much of it has been discussed before.

					der Mouse

			       mouse@rodents.montreal.qc.ca
		     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B