From owner-cvs-all  Thu Oct  1 14:08:47 1998
Return-Path: <owner-cvs-all>
Received: (from daemon@localhost)
          by hub.freebsd.org (8.8.8/8.8.8) id OAA06195
          for cvs-all-outgoing; Thu, 1 Oct 1998 14:08:47 -0700 (PDT)
          (envelope-from owner-cvs-all)
Received: from sumatra.americantv.com (sumatra.americantv.com [207.170.17.37])
          by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id OAA06164;
          Thu, 1 Oct 1998 14:08:29 -0700 (PDT)
          (envelope-from jlemon@americantv.com)
Received: from right.PCS (right.PCS [148.105.10.31])
	by sumatra.americantv.com (8.8.5/8.8.5) with ESMTP id QAA10130;
	Thu, 1 Oct 1998 16:08:12 -0500 (CDT)
Received: (from jlemon@localhost)
	by right.PCS (8.6.13/8.6.4) id QAA17672;
	Thu, 1 Oct 1998 16:07:40 -0500
Message-ID: <19981001160739.01031@right.PCS>
Date: Thu, 1 Oct 1998 16:07:39 -0500
From: Jonathan Lemon <jlemon@americantv.com>
To: Mike Smith <mike@smith.net.au>
Cc: sos@FreeBSD.ORG, adhir@worldbank.org, yokota@FreeBSD.ORG,
        cvs-committers@FreeBSD.ORG, cvs-all@FreeBSD.ORG
Subject: Re: cvs commit: src/sys/i386/include console.h src/sys/i386/isa scvesactl.c scvidctl.c syscons.c syscons.h vesa.c videoio.c
References: <19981001154049.29400@right.PCS> <199810012059.NAA01433@dingo.cdrom.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Transfer-Encoding: 8bit
X-Mailer: Mutt 0.61.1
In-Reply-To: <199810012059.NAA01433@dingo.cdrom.com>; from Mike Smith on Oct 10, 1998 at 01:59:53PM -0700
Sender: owner-cvs-all@FreeBSD.ORG
X-Loop: FreeBSD.org
Precedence: bulk

On Oct 10, 1998 at 01:59:53PM -0700, Mike Smith wrote:
> > On Oct 10, 1998 at 10:32:33PM +0200, Søren Schmidt wrote:
> > > In reply to Mike Smith who wrote:
> > > 
> > > It works here, but I still have a patch in kern_clock, but I dont think
> > > its needed anymore...
> > > 
> > > > There were some significant VM86-and-SMP fixes committed yesterday.  
> > > > I'd be inclined to try it again, certainly.
> > > > 
> > > > > Any idea if this works in SMP now?
> > 
> > I have one more patch to vm86bios.s, which I'll commit in a moment.
> > The patch to kern_clock.c is needed if you are using the statistics
> > clock.  (Actually, Tor suggested patching machine/cpu.h, in order to
> > make it dependent on opt_vm86.h, so there is no overhead on the alpha.)
> 
> Is the issue likely to apply to SMP alpha systems?  If so, it's 
> probably a reasonable investment in the future.

No, it's specific to vm86, not SMP.  Calling the BIOS implies that we
are running in usermode, and that there is a current process context.
This is not true if, for example, syscons happens to make a VESA call
when the processor is in the idle loop.

This causes problems with the clock code, which says, "okay, you were
in user mode, I'm going to dereference p-> now to update some stats." :-(
--
Jonathan