Subject: Re: Floating point in the kernel
To: Ignatios Souvatzis <is@jocelyn.rhein.de>
From: Dr. Bill Studenmund <wrstuden@loki.stanford.edu>
List: tech-kern
Date: 09/18/1998 15:23:08
On Fri, 18 Sep 1998, Ignatios Souvatzis wrote:

> On Fri, Sep 18, 1998 at 02:05:58PM -0700, Andrei Petrov wrote:
> > 
> > I would expect that kernel saves/restores user's FPU state so it should be
> > safe just use FP in kernel mode assuming that you are the only part of kernel
> > who does that. Don't know what to do with interrupts from FP in kernel mode.
> 
> Many NetBSD kernels (in terms of cpu types) only save and restore the FPU
> at process switch time, or not even then (if the new process doesnt use 
> the FPU), so... no.

Right!

My whole point is that if we only use FP in one place in the kernel (in
the DSP driver), and that one place coordinates things so that process
switching works right, then we're ok.

Though some sort of real-time scheduling and in-kernel processes might
work easier (like aren't nfsiod's actually always in-kernel?).

Take care,

Bill