On Sat, Feb 12, 2005 at 10:13:46AM +0000, David Laight wrote:
> > [snip ... ]
> > Or, perhaps there should be code to switch calculation methods based on
> > the size of the disk, anything less than about 4GB is CHS with partitions
> > on cylinder boundaries, anything over 8GB uses sector-based calculations.
> 
> Sysinst tries quite hard to align partitions on cylinder boundaries.
> [snip ... ]
> Failure of the last few cylinders could be caused by that part of the

What's freaking me out is that 31 extra cylinders seems more than just
a "few".  I also don't understand why the NetBSD probe at bootup seems
to be contradictory, i.e., it reports more sectors than can be carved
out of the number of cylinders reported.  It is not just my particular
machine or disk; the following URLs report the same geometry:
http://mail-index.netbsd.org/port-macppc/2001/06/15/0002.html
http://www.netsys.com/netbsd-port-sparc/2004/06/msg00031.html

From
sd0: 1222 MB, 6810 cyl, 2 head, 183 sec, 512 bytes/sect x 2503872 sectors
you either calculate
   2503872/183/2 = 6841 cylinders  (which is what sysinst did with the
                                    unformated disk)
or you calculate
   6810*183*2 = 2492460 sectors    (which is more along the lines of what
                                    Solaris does -- except 6808 cylinders)

Another thing that is strange to me is that after formating the disk
using Solaris's format command, sysinst reports "You have 2491728 sec
left on your disk", while the first time around with the raw disk it
let me use 2503872 sectors and labeled the disk with 6841 cylinders.

AFAIKT, in both instances NetBSD was using 2*183 to calculate cylinder
boundaries, so it seems to me something is haywire.

henry nelson