> 
> I compiled the kernel overnight and all went except when
> it came to linking. One file called "zs.c" wants a symbol
> called "_abs" (probably wants stdlib). 
No! A kernel should _not_ rely on these files, nor should it have
libc linked in or such. I bug you saw was introduced because 
the syscall namespace needed cleanup. I added the abs() function
to ..../src/sys/lib/libkern/libkern.h . You can either try to
get this file from ftp.netbsd.org or add the following to libkern.h
yourself:
	static __inline int abs __P((int));
	
	static __inline int
	abs(j)
		int j;
	{
		return(j < 0 ? -j : j);
	}
> I think I will be able to fix this on my own, but as a 
> 'betatester' or somesuch I wanted to note this.
Yes, all bug reports are apreciated ;-)
> 
> Also I have to note that when I run systat or vmstat, none
> of my drive statistics show up. When I try :numbers when
> running :iostat in systat (crypto-speach :) ) I get
> a float exception. Most probably this is caused because 
> there are no disk drives connected and the program does a 
> divide by zero. (Just my guesses)
I think this bug is already in the gnats database. I'll check it
out.
Speaking about problem reports, you can also use send-pr to do the
job, for both atari-specific and general bugs. Checkout man 'send-pr'.
I have to amdmit however that the latest binary distribution for the
atari didn't know the category 'port-atari', you can add this in the
file '/usr/share/gnats/netbsd'. This will be fixed for NetBSD1.1 .
> 
> After the crash my terminal was a little fucked (no echo)
try 'stty sane'
> 
> Does anyone know what to tune in order to speed up a 12MB
> 36MHz Falcon ? 
> Subjectively I think NetBSD is slower than LINUX, any 
> truth to this statement ?
> Where's the source for the slow Atari terminal output 
> driver ?
/usr/src/sys/arch/atari/dev/ite.c ;-) I agree that it can use
some optimization here and there.
> 
> That's it for now, the rest runs smoothly...
Wait until you see Murpy ;-)

Leo.