> In many cases, the actual disk *does* extend beyond what the
> `geometry' implies; even after you discount reserved sectors.

> It seems more appropriate to either:

> a) stop pretending we have any idea what the geometry of a SCSI disk
>    is, and just make it one large track,

I like this idea.  Modern disks don't fit the model any longer.  Some
of them even come with doc that outright says that the
sectors-per-track value they give you is an average, obtained by
dividing the disk size by the number of cylinders and number of heads,
and rounding down to an integer.  Certainly the sectors-per-track value
varies depending on the cylinder number, in ways that we have to
consider unpredictable.

What breaks in extreme cases, like one-sector cylinders, or
one-cylinder disks with an enormous sectors-per-track value?  Perhaps
we can just use such a label....

>    or
> b) make the warning from disklabel(8) non-fatal.

This should probably be done anyway.

					der Mouse

			    mouse@collatz.mcrcim.mcgill.edu