On Wed, Dec 14, 2005 at 03:18:27PM -0600, James Chacon wrote:
> On Wed, Dec 14, 2005 at 09:06:56PM +0000, David Laight wrote:
> > On Wed, Dec 14, 2005 at 03:01:03PM +0000, Reinoud Zandijk wrote:
> > > 
> > > Module Name:	src
> > > Committed By:	reinoud
> > > Date:		Wed Dec 14 15:01:03 UTC 2005
> > > 
> > > Modified Files:
> > > 	src/sys/kern: subr_disk_mbr.c
> > > 
> > > Log Message:
> > > Fix disklabel recognition code for mbr based systems like i386. It would
> > > return NULL even though no disklabel was found making callers assume that a
> > > valid disklabel WAS found but instead were presented by the dummy disklabel
> > > that is created.
> > > 
> > > If the rval is SCAN_CONTINUE it now returns a standard error that no
> > > disklabel was found instead of the NULL.
> > 
> > Hummm - that was more or less deliberate, IIRC some of the callers
> > need a disklabel filled in with partition 'a' and 'd' when there
> > is nothing found on the disk.
> > 
> 
> This is why for changes in critical areas people should be sending mail
> to the appropriate tech-X list first asking for review before just committing
> what they think is right.
> 
> If you think this breaks something I'd revert it.

I've reverted it, it breaks the default label for disks that have neither
MBR nor NetBSD label.
There may be alternative fixes to retain more of the default label...

	David

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