Date: Wed, 15 Dec 1999 00:00:00 +0100 From: Wilko Bulte <wilko@yedi.iaf.nl> To: Jason Thorpe <thorpej@nas.nasa.gov> Cc: FreeBSD-alpha mailing list <freebsd-alpha@FreeBSD.ORG> Subject: Re: use of -mcpu=21164a Message-ID: <19991215000000.B77327@yedi.iaf.nl> In-Reply-To: <199912142214.OAA00903@lestat.nas.nasa.gov>; from thorpej@nas.nasa.gov on Tue, Dec 14, 1999 at 02:14:45PM -0800 References: <199912142214.OAA00903@lestat.nas.nasa.gov>
next in thread | previous in thread | raw e-mail | index | archive | help
On Tue, Dec 14, 1999 at 02:14:45PM -0800, Jason Thorpe wrote:
> On Tue, 14 Dec 1999 22:47:29 +0100
> Wilko Bulte <wilko@yedi.iaf.nl> wrote:
>
> > > Also, your assembler has to do the right thing; gas will treat the BWX
> > > instructions are macros and open-code the equivalent in EV4 instructions
> > > unless it is explicitly in EV56 (or greater) mode.
> >
> > So, in case of -mcpu=21164a gas will generate BWX and if -mcpu=21164a
> > is omitted it will open-code in EV4 instructions?
>
> Yes. (Of course, I'd rather it emitted an error, or a warning, at least...)
>
> > what if I generate 'netboot' on a Miata with mcpu=m21164a and want to
> > use the generated netboot on a Sandpiper (EV4)? There is no kernel (yet)
> > to emulate BWX so I lose, right?
>
> Right, if the compiler/assembler emitted BWX, you lose.
This is definitely a nice pitfall. The only obvious way to avoid it (that
I see right now) is to have the build environment of the standalone
parts of the system set their own cpu-independent flags (read:
the least-common denominator cpu). The standalone bits then also include
the kernel itself (or can it emulate missing BWX while they are being
used inside the kernel? I doubt it.)
W/
--
Wilko Bulte Arnhem, The Netherlands - The FreeBSD Project
WWW : http://www.tcja.nl http://www.freebsd.org
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-alpha" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?19991215000000.B77327>