Re: fs/tmpfs/t_vnd/basic fails

[David Laight <david%l8s.co.uk@localhost>]

On Sat, Sep 08, 2012 at 02:30:24PM +0900, Izumi Tsutsui wrote:
> msaitoh@ wrote:
> 
> > > The test should probably be "newfs -I /dev/rvnd3[raw partition]",
> > > or explicitly write disklabel.
> > 
> > Should we fill partition a in arm/disksubr.c?
> 
> IMO subr_disk_mbr.c should rather be fixed.
> 
> No reason to assume 4.2BSD on disks without disklabel,
> and newfs(8) (and newfs_ext2fs(8) etc) provides -I option
> for disks without disklabel.

IIRC i386 had code added (a long time ago) to attempt to genarate
a usable disk label for disks that don't have an explicit netbsd label.
As well as processing any mbr info (which would be useful on all archs)
it tends to create an 'a' partition covering the whole disk.

Guessing the partition type is shightly problematic! But in a lot of
cases the type doesn't really matter.

        David

-- 
David Laight: david%l8s.co.uk@localhost