Hi there! After (fairly) carefully installing 1.0 on a SS2 so that 
I could boot either SunOS or NetBSD with a little PROM diddling, I 
approached the installation of the May 3rd -current (from a local
source, who had it running with no problems) too casually. I can't get
a kernel built that recognizes my boot disks and swap
partition. Here's the disk layout:

target 0: NetBSD root, swap, and /usr (slices 0, 1, and 6)
target 2: NetBSD source tree
target 1: SunOS /var and homedirs
target 3: SunOS root, swap, and /usr

 For 1.0, I went ahead and used the "straight" kernel (no SunOS
crosseyed stuff) and things were fine. I assumed (ahhh, that word!)
that I would be able to do the same for thecurrent I have. Hah! 

 I tried "config"ing the kernel with the (apparently) working generic
config directive. Interestingly, it failed after booting the kernel
because it couldn't find swap space on sd3. I then tried directing
specific devices as in 1.0 - config netbsd root on sd0 swap on
sd0. This hung the machine HARD at the point when root and swap
devices should be mounted and enabled. I groveled through the
autoconfig stuff a bit, and found little of direct interest.

  Some questions/comments:

a) Is the generic SCSI autoconfig for sun4c system replicating the
   SunOS twister-mappiing? If so, WHY? Sun had reasonable marketing
   motives for munging autoconfig and logical device names. I don't
   think that behavior should be the default for NetBSD.
b) If that's not the reason that the kernel tried to make sd3a the
   root partition, then why doesn't the system use the device it
   acquires the bootblock from? That seems like the intuitive behavior -
   boot block off target 0, root on target 0 initial swap on target 0.
   If I want to do something more exotic, I can tell config about it.
c) Any idea why specifying the devices for root and swap hung the
   system up?

	Greg K.