>     uname(3) is broken - I've traced this down to the sysctl() for
>     	KERN_VERSION failing with ENOMEM...the source for uname(3) 
>     	looks good as best I can tell.  This of course breaks
>     	uname(1).

i fixed this the other day; you need the new utsname.h header file,
and then you need to recompile the uname.c header file.

>     kernfs file sizes are broken - They show up like this:
> 
> nirvana::mark-21> ls -l /kern
> total 0
> -r--r--r--  1 root  wheel            -4294967195 May 17 01:24 copyright
> -rw-r--r--  1 root  wheel            -4294967288 May 17 01:24 hostname
> -r--r--r--  1 root  wheel            -4294967292 May 17 01:24 hz
> -r--r--r--  1 root  wheel            -4294967276 May 17 01:24 loadavg
> -r--r--r--  1 root  wheel            -4294967291 May 17 01:24 physmem
> brw-r-----  1 root  operator    0,   0 Apr 20 01:41 rootdev
> -r--r--r--  1 root  wheel            -4294967279 May 17 01:24 time
> -r--r--r--  1 root  wheel            -4294967170 May 17 01:24 version
> nirvana::mark-22> 

charles fixed these a few minutes ago.

>     Finally, escaping $ in sh seems to have busted.  The following 
> line which used to work now tries to map to arg 6.  It may be a POSIX thing
> but it works under bash.  It only shows up with double escapes and 
> recursive shells.  This shows up in a script when trying to pass something
> like foo=`awk " print { \\$6 }" `, but is easily illustrated with the
> following example:

I'm not sure about this; perhaps jtc can shed some light on it...



later,

chris

------------------------------------------------------------------------------