sysinst is dying at the begining of an upgrade or install with the
	recent sparc64 snapshot.  I'm following the bootstrap from Solaris
	instruction in the sparc64 FAQ from netbsd.org on a Ultra5 with
	Solaris 8.  I format my disk and label in from within Solaris and then
	I dd over the ramdisk image to the swap slice I created.  I then boot
	from that swap slice and after giving it my TERM value (used sun-ss5
	since that's all I heard about on this list) and sysinst is started.
	  First sysinst complained because I was using an old inode format for
	ffs so it wanted to run fsck_ffs -c /dev/rsd1a.  However that is
	incorrect because the -c argument requires a number, so I got :

	      Status: Failed
	     Command: /sbin/fsck_ffs -c /dev/rsd1a
	      Press any key to continue
	-------------------------------------------------------------------
	fsck_ffs: -c flag requires a conversion level

	  I read the man page for fsck_ffs on another system and decided to
	try and run fsck_ffs -c N /dev/rsd1a where n was a number between 1-4
	from a shell in sysinst.  When I tried to run this command it failed
	with errors about / being full.

	  Any ideas?  I'm getting close to installing this beast so I would
	definately appreciate it if someone could help me get past the fsck
	errors in sysinst.  Thanks.

I must be missing something.

1) If you are using the built-in ide controller then you should
not be using sd anything.  Those devices are wd0 and wd1.  When
promped for a root partition you should be specifying `wd0*' or
`wd1*'.

2) If you're using a PCI scsi controller then the miniruut should
be going into partition `b' (slice 1) and when prompted for a 
root partition you should specify `sd1*'.

3) Formatting the root should be done by the ramdisk.  Since it's
creating a new partition it should not be running fsck on it.

Eduardo