> On Thu, 29 Jan 1998, David Seifert wrote:
> 
> : What if the utsname structure became a per-process structure,
> : inherited across fork?  Then we add setuname() to change the
> : various fields to whatever we want.
> : 
> : Then we write a small wrapper program to set the desired uname info
> : and run the real app.
> 
> Toss in a field for `hostid' (sparc/m68k platforms) and I'll bow before you. 
> (Changing the hostid per-process is something I've been meaning to write
> code for.)  This is an idea I was hesitating to propose, but since you bit
> the bullet,... 

hw.machine = amiga
hw.model = DraCo rev.4 (68060 rev.1 CPU/MMU/FPU)
hw.ncpu = 1
hw.byteorder = 4321
hw.physmem = 41943040
hw.usermem = 37412864
hw.pagesize = 8192
hw.machine_arch = m68k

you mean, sysctl hw.cpumodel (which doesn't exist yet)?

We need this natively, btw. There are, depending on application class,
at least three sorts of 68k cpus we'd like to support (which is of
similar importance as the sun4/sun4c/sun4d/sun4m/sun4u distinction 
to normal userland applications):

		64bit intmuldiv		expensive FP in HW
-m68020/030	yes			yes
-m68040		yes			no
-m68060		no			no

... not even thinking about FPU-less machines.

so I've thought about something like this, too.

Regards,
	Ignatios