Date: Sat, 1 Sep 2001 16:10:54 -0700 From: Marcel Moolenaar <marcel@xcllnt.net> To: John Baldwin <jhb@FreeBSD.org> Cc: freebsd-stable@FreeBSD.org, qa@FreeBSD.org, Eric Masson <e-masson@kisoft-services.com>, Larry Rosenman <ler@lerctr.org> Subject: Re: cputype=486 Message-ID: <20010901161054.B13047@athlon.pn.xcllnt.net> In-Reply-To: <XFMail.010901132212.jhb@FreeBSD.org> References: <20010901114903.D11062@athlon.pn.xcllnt.net> <XFMail.010901132212.jhb@FreeBSD.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sat, Sep 01, 2001 at 01:22:12PM -0700, John Baldwin wrote: > > >> > >> Is *ANYTHING* going to be done about this ever? It appears nothing > >> was done at all during the 4.4 cycle. > > > > If I have some time (yeah, right :-) I'll take a look at it. The > > quickest fix would be to define CPUTYPE as empty when building > > cross- and build-tools. This should override any definitions that > > are present in /etc/make.conf. > > That doesn't help. The problem is that the cross-tools built during buildworld > are linked against the hosts' libc.a, and then run on the target machine in > installworld. Thus, in this case you end up running a 686 libc.a in the cross > tool on a 486 and it blows up. > Ah yes. In that case it only works if you link shared, but then you have different problems. I guess this rules out CPUTYPE as a generic tunable. If you want the highest possible performance, you give up on portibility. You can't have it both... -- Marcel Moolenaar USPA: A-39004 marcel@xcllnt.net To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-qa" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20010901161054.B13047>