Date: Wed, 21 Feb 2001 13:27:54 -0500 (EST) From: mi@aldan.algebra.com To: Bruce Evans <bde@zeta.org.au> Cc: Warner Losh <imp@village.org>, Kris Kennaway <kris@FreeBSD.org>, cvs-committers@FreeBSD.org, cvs-all@FreeBSD.org Subject: Re: cvs commit: src/share/mk sys.mk Message-ID: <200102211827.f1LIRuv32516@misha.privatelabs.com> In-Reply-To: <Pine.BSF.4.21.0102211415050.25100-100000@besplex.bde.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On 21 Feb, Bruce Evans wrote:
= On Tue, 20 Feb 2001, Warner Losh wrote:
=
= > <200102200837.f1K8bpj89576@freefall.freebsd.org> Kris Kennaway writes:
= > : We now set up the reasonable default for i386 and alpha here --
= > : given this it probably makes sense to remove the corresponding
= > : code from make(1).
= >
= > Probably not. We really do not want to pollute sys.mk with any more
= > variables than we absolutely have to. It is included for all
= > Makefiles on the system, not just the ones to build the system.
= > Adding things there can cause problems elsewhere.
=
= But polluting make(1) is almost equivalent to polluting sys.mk. The
= pollution is just easier to implement and harder to remove in make(1).
=
= I have changed my mind a bit about this. MACHINE_CPU is not (yet?)
= nearly as fundamental as the other MACHINE_* variables. It is
= currently just a build option for libcrypto, so it could be handled
= like other build options.
BTW, may be -mcpu=${MACHINE_CPU} can be added to the default CFLAGS now?
And a comment, that changing -mcpu to -march will, probably, be even
better, but will prevent the binaries from being usable on earlier CPUs.
-mi
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe cvs-all" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200102211827.f1LIRuv32516>