Richard Earnshaw wrote:
> 
> Just to be sure can you send me (not the list) the output of running
> 
> 	cc -E /usr/src/usr.sbin/sa/main.c
> 
> and I will check it with the current sources.

I'm afraid not as was I was successful in shafting my machine. I had
just supped the sources from sup.netbsd.org on Wednesday - hard to
check because

# mount
root_device on / type ffs (local, read-only)
# mount /dev/wd1a /
[hangs]
as does csh. sh works. So now I have the interesting question of how
to get a working mount binary on to mount a floppy to get a working
mount binary on (maybe unixfs will come to the rescue..)

Chris Demetriou wrote:
> Contains sources for a working(-ish) gdb and a working assembler for
> the arm32 port.  (Sorry it's so much to download...  8-)
>
> (Note that most of the NetBSD sources in those tarballs are of a
> NetBSD 1.2.1 vintage, with lots of local ingredients.  I'd not trust
> the rest of them on a -current system, but gdb, gas, gcc, and probably
> the linker/shared loader should be pretty safe.  "Be careful!" if
> you're trying to build your whole system from scratch; if you manage
> to muck things up, well, sorry, but it's not my fault.  FWIW, we hope
> to be shipping some 1.3-based sources + binaries in a few weeks. 8-)

All I did was carry on the build without gdb to completion and *error*
installed the world without keeping copies of the old binaries. This
was all using the comp set from ftp.ph.kcl.ac.uk. Maybe someone could
check my method, as I don't see what to do differently next time
around:

- grab kernel and sets from ftp.ph.kcl.ac.uk
- sup -current source
- build (config) kernel and reboot
- build (rpcgen) world (without -O on dhcp and sa, and without gdb)
- build kernel (this time with new compiler)
- build world

Wish me luck,

Patrick