Date: Thu, 9 Nov 1995 12:35:24 +0100 (MET) From: Luigi Rizzo <luigi@labinfo.iet.unipi.it> To: phk@critter.tfs.com (Poul-Henning Kamp) Cc: hasty@rah.star-gate.com, koshy@blr.novell.com, freebsd-hackers@freebsd.org Subject: Re: Load/Store using FPU regs ... Message-ID: <199511091135.MAA08283@labinfo.iet.unipi.it> In-Reply-To: <6539.815911372@critter.tfs.com> from "Poul-Henning Kamp" at Nov 9, 95 11:02:33 am
next in thread | previous in thread | raw e-mail | index | archive | help
> > > > Not to mention that you might not have a FPU.
> > >
> > > So what ? You still have the FPU emulator :)
> >
> > Just ignore Pohl's comment we can probably figure out a clever way
> > to find out if we have a FPU .
> sure, there's a flag in the kernel, and it's a sysctl var too.
I was just kidding, but the discussion has become serious!
One question: what is the behaviour of our FP library if there is not a
coprocessor: do we just trap into the emulator, or the code is
magically replaced (e.g. at the first execution) with a call to the
proper code ?
Anyways using the same technique here seems expensive: the code
should look like
if (kern.have.fpu) {
... fragment of code using FP regs
} else {
... fragment of code using standard techniques
}
and I don't believe something like this can be done easily by the
compiler in a machine-independent way.
> The place to use it would be in the Copy-On-Write and Demand-Zero
> parts of the VM system.
If the scope is limited to such cases, then the kernel could include
appropriate code.
Luigi
====================================================================
Luigi Rizzo Dip. di Ingegneria dell'Informazione
email: luigi@iet.unipi.it Universita' di Pisa
tel: +39-50-568533 via Diotisalvi 2, 56126 PISA (Italy)
fax: +39-50-568522 http://www.iet.unipi.it/~luigi/
====================================================================
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199511091135.MAA08283>