Some API CS20 notes

[John Klos <john%ziaspace.com@localhost>]

Hi,

I recently picked up two API CS20 machines from Eric Schnoebelen (thanks, 
Eric!), and I thought I'd share some notes about NetBSD on these machines.

Disks on the IDE bus don't show up after a warm reboot.

The installer (from -current) doesn't make a proper disklabel on IDE 
disks. No idea what's going on, but one can run the installer or manually 
create / edit a disklabel, but it will never get read back from the IDE 
disk.

BTW - this is both with disks larger than 128 gigs and smaller than 128 
gigs.

On a SCSI disk, when the first partition doesn't start at sector zero, the 
disk can still be made bootable (even with FFSv2), but the first stage 
bootloader can never find /boot and the system won't be bootable.

The machine panics when the installer newfses the disk when the disk 
already had data. It doesn't do this when one dds /dev/zero to the first 
few gigabytes:

    Command: /sbin/newfs -V2 -O 2 -b 8192 -f 1024 /dev/rsd0a

--------------------------------------------------------------------------------
/dev/rsd0a: 32955.1MB (67492000 sectors) block size 8192, fragment size 1024
        using 720 cylinder groups of 45.77MB, 5859 blks, 11040 inodes.
...............................................................................
ffs_newvnode: dup alloc ino=3 on : mode ffff/ffff gen ffffffff/ffffffff size fff
fffffffffffff blocks ffffffffffffffff
Stopped in pid 16.1 (mount_ffs) at      fffffc0000a51510:       ret     zero,(ra)

This, by the way, is with the -current installer from 25-January-2017. The 
toolchain, incidentally, didn't work, but the toolchain from -current on 
24-January-2017 works just fine.

Just like my DS25 from last March, running multiprocessor is not stable. 
While last year it locked up, now it panics, but the debugger seems 
useless (bt and other commands show nothing).

Would anyone like serial console access to a CS20 running multiprocessor?

Running single processor on this CS20 seems to be working well and the 
system is stable - I've been doing all sorts of things and compiling lots 
of things for twenty four hours straight with no problems.

John