Subject: Re: __archname__ [as C feature-test ]
To: Jonathan Stone <jonathan@DSG.Stanford.EDU>
From: Chris G. Demetriou <cgd@netbsd.org>
List: current-users
Date: 01/15/1999 17:44:20
cgd@netbsd.org (Chris G. Demetriou) writes:
> Jonathan Stone <jonathan@DSG.Stanford.EDU> writes:
> > I'd thought Core (Jason, at the time, iirc) had indicated we should.
> > I know ld.elf_so used to use #ifdef {__alpha__,__i386__,__mips__}.
> > 
> > Whats the new rule?
> 
> those are the user-land symbols used to distinguish MACHINE_ARCH.
> 
> in the kernel, $MACHINE is supposed to be defined by the kernel
> makefile, in addition to whatever other defines exist.

(and, to add a bit to that):

__${MACHINE}__ should not be defined (except perhaps as a side effect of
-D$MACHINE (i dunno).


Note that: __alpha__,__i386__,__mips__

are MACHINE_ARCH values, but __arc__ and __pmax__ ar MACHINE values.




cgd
-- 
Chris Demetriou - cgd@netbsd.org - http://www.netbsd.org/People/Pages/cgd.html
Disclaimer: Not speaking for NetBSD, just expressing my own opinion.