From owner-freebsd-commit  Thu Dec 14 05:05:33 1995
Return-Path: owner-commit
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.3/8.7.3) id FAA00703
          for freebsd-commit-outgoing; Thu, 14 Dec 1995 05:05:33 -0800 (PST)
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.3/8.7.3) id FAA00557
          for cvs-all-outgoing; Thu, 14 Dec 1995 05:02:49 -0800 (PST)
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.3/8.7.3) id FAA00549
          for cvs-user-outgoing; Thu, 14 Dec 1995 05:02:46 -0800 (PST)
Received: from jhome.DIALix.COM (root@jhome.DIALix.COM [192.203.228.69])
          by freefall.freebsd.org (8.7.3/8.7.3) with ESMTP id FAA00536
          Thu, 14 Dec 1995 05:02:31 -0800 (PST)
Received: (from peter@localhost) by jhome.DIALix.COM (8.7.3/8.7.3) id VAA03988; Thu, 14 Dec 1995 21:00:32 +0800 (WST)
Date: Thu, 14 Dec 1995 21:00:32 +0800 (WST)
From: Peter Wemm <peter@jhome.DIALix.COM>
To: Paul Richards <p.richards@elsevier.co.uk>
cc: Bruce Evans <bde@zeta.org.au>, CVS-committers@freefall.freebsd.org,
        cvs-user@freefall.freebsd.org, phk@freefall.freebsd.org
Subject: Re: cvs commit: src/lkm/gnufpu Makefile
In-Reply-To: <199512141245.MAA22992@cadair.elsevier.co.uk>
Message-ID: <Pine.BSF.3.91.951214205455.3291A-100000@jhome.DIALix.COM>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Sender: owner-commit@FreeBSD.ORG
Precedence: bulk

On Thu, 14 Dec 1995, Paul Richards wrote:
> In reply to Bruce Evans who said
> > 
> > >  Modified:    lkm       Makefile
> > >  Added:       lkm/fpu   Makefile
> > >               lkm/gnufpu  Makefile
> > >  Log:
> > >  Two x87 emulators as LKMs.
> > >  If somebody with the right HW would make the change to /etc/rc to use
> > >  this, we could rip MATH_EMULATE from GENERIC...
> 
> The kernel should load it if the npx probe fails. Can we do this yet or will
> it require an in-kernel linker? It's something we need to be able do so
> that we can load device drivers on the fly following probes so all you have
> to have compiled in is the probe routines or actually you can load the device
> driver, do probe and then unload it if it fails. All you need then is a list
> of available drivers.

The math emulators are a special case. They have a _single_ entry point, 
and I dont think (from memory) that they reference anything else in the 
kernel.  We may be able to get away with a very simple a.out style loader 
and relocator for this special purpose.  (ie: do vnode IO to read in the 
a.out header, malloc some space, do vnode IO to bring in the code, 
then relocate it, then set the trap handler to point to the first bit of 
executable code.  Make it all do vnode IO to (say) /sbin/emulator.o and 
leave an a.out format object file there.

This is something like what SYSV does, except the SYSV one is PIC I 
think...

I created math emulator LKM's too, about a week ago, but once I realised 
they were useless, I dropped them.

-Peter