From owner-freebsd-current  Thu Jan  2 18:25:46 1997
Return-Path: <owner-current>
Received: (from root@localhost)
          by freefall.freebsd.org (8.8.4/8.8.4) id SAA22772
          for current-outgoing; Thu, 2 Jan 1997 18:25:46 -0800 (PST)
Received: from godzilla.zeta.org.au (godzilla.zeta.org.au [203.2.228.19])
          by freefall.freebsd.org (8.8.4/8.8.4) with ESMTP id SAA22767
          for <freebsd-current@freebsd.org>; Thu, 2 Jan 1997 18:25:39 -0800 (PST)
Received: (from bde@localhost) by godzilla.zeta.org.au (8.8.3/8.6.9) id NAA13793; Fri, 3 Jan 1997 13:15:47 +1100
Date: Fri, 3 Jan 1997 13:15:47 +1100
From: Bruce Evans <bde@zeta.org.au>
Message-Id: <199701030215.NAA13793@godzilla.zeta.org.au>
To: bde@zeta.org.au, cracauer@wavehh.hanse.de
Subject: Re: i586-optimized copyin/out still broken
Cc: freebsd-current@freebsd.org
Sender: owner-current@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk

>>>  Modified:    sys/i386/isa  npx.c
>>>  Log:
>>>  Disabled i586-optimized copyin and copyout again.  The fault handler
>>>  is still broken - it doesn't restore the floating point state.

>Could this problem result in NaN being produced at wrong places.

No, it causes kernel panics.

>I filled a PR (2142) about it, maybe it could be closed then.

That is much harder to fix.  It is caused by the floating point state
not being preserved across signal handlers.  There are few, if any,
valid and useful uses for floating point in signal handlers, because
an ANSI signal handler must not make any accesses to a global object
other than assignment to ones of type `volatile sig_atomic_t'.  Thus
preserving the state would mainly slow down signal handlers.

Bruce