Thus spake David Laight ("DL> ") sometime Yesterday...

DL> > Can I inquire, while we're here, *why* a user-tunable cpg has vanished?
DL>
DL> Probably because the value read from the disklabel had a nasty habit
DL> of being completely wrong.

Umno.  newfs never read the CPG from the disklabel, one thing I had
previously commented on as having been rather annoying.  I could set CPG
in the disklabel and newfs would blithlely ignore it.

DL> Why do you want to set cpg?
DL>
DL> Would a parameter to set the minimum number of 'cylinder groups'
DL> be more reasonable?  It defaults to 4 (unless that would generate
DL> too much red tape).
DL>
DL> Note that the current newfs doesn't use the physical cylinder size
DL> when formatting the filesystem.  Each ffs 'cylinder group' contains
DL> only one 'cylinder' and may be arbitrarly aligned on the disk.

You know, maybe it's my lack of caffeine, but the above is just not making
sense to me at all.  Defaults to 4 cg/partition, only allocating more
as necessary.  Hm.  Odd.  And one cylinder per group?  That doesn't jibe
with the output from newfs.

DL> (Without this change, the number of cylinders per group causes the
DL> red tape to expand to the point where you can't create a large filesystem.)
DL>
DL> This might be a loss if you have a real rk05 [1] or rl01, but it probably
DL> not noticable on any modern disks with variable geometry.

DL> 	David
DL>
DL> Mmm a massive 2.5MB.

A more realistic look at this might be the old outdated SMD devices
-- if, indeed, any are still running, energy hogs/mechanical beasts that
they are.

				--*greywolf;
--
NetBSD: We Stay Up.