NetBSD 8 on Alpha

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

Hi,

I've written about my experiences with NetBSD on two API CS20 machines 
which are similar to AlphaServer DS20L machines. I referred to them as 
DS20L before, but they're definitely the API machines.

I've never gotten to the bottom of the reasons why disks become unbootable 
or eventually has filesystem issues:

http://mail-index.netbsd.org/port-alpha/2017/07/13/msg000830.html

Or why installboot currently kills the disklabel:

http://mail-index.netbsd.org/port-alpha/2018/07/02/msg000884.html

These two things may be related, but all corruption has been seen when 
booting and using the same disk, and at least anecdotally when using disks 
on one of the internal busses heavily while using motherboard ethernet, so 
now I plan to keep the boot disk and filesystem disks separate.


But at least the time issues are fixed by this, although the patch still 
must be applied manually (this doesn't appear to have been committed):

http://mail-index.netbsd.org/port-alpha/2017/11/02/msg000879.html


Now I'll just boot the kernel from a CF card attached to the system's IDE 
interface and have the root filesystem on a large SATA disk attached to a 
USB 3 card or to a SATA card.

Some notes:

USB 3 works perfectly out of the box just by adding xhci to the kernel, 
and although I get this warning, so far it works perfectly at > 80 MB/sec 
on a 32 bit, 33 MHz PCI bus:

xhci0: xHCI version 1.10 not known to be supported

When booting from IDE, if the IDE bus is configured to just use PIO via 
"flags 0x0002", either to support larger drives or to have fewer bus 
errors when communicating with drives, the system hangs on reboot when the 
firmware tries to read from the IDE. Removing "flags 0x0002" from aceride* 
makes it so the system can reboot unattended.

So far, so good!

John Klos