From owner-freebsd-hackers  Tue Jul 29 20:45:08 1997
Return-Path: <owner-freebsd-hackers>
Received: (from root@localhost)
          by hub.freebsd.org (8.8.5/8.8.5) id UAA04333
          for hackers-outgoing; Tue, 29 Jul 1997 20:45:08 -0700 (PDT)
Received: from kithrup.com (kithrup.com [205.179.156.40])
          by hub.freebsd.org (8.8.5/8.8.5) with SMTP id UAA04328
          for <hackers@freebsd.org>; Tue, 29 Jul 1997 20:45:03 -0700 (PDT)
Received: (from sef@localhost) by kithrup.com (8.6.8/8.6.6) id UAA16264; Tue, 29 Jul 1997 20:44:55 -0700
Date: Tue, 29 Jul 1997 20:44:55 -0700
From: Sean Eric Fagan <sef@Kithrup.COM>
Message-Id: <199707300344.UAA16264@kithrup.com>
To: hackers@freebsd.org
Subject: Re: Location of copyin() and copyout().. 
In-Reply-To: <199707300310.UAA12221.kithrup.freebsd.hackers@implode.root.com>
References: Your message of "Wed, 30 Jul 1997 12:19:48 +0930."
             <199707300249.MAA17292@genesis.atrad.adelaide.edu.au> 
Organization: Kithrup Enterprises, Ltd.
Sender: owner-freebsd-hackers@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk

In article <199707300310.UAA12221.kithrup.freebsd.hackers@implode.root.com> you write:
>If on the other hand you're copying from kernel<->kernel, then you can just
>use bcopy. I'm afraid that I started reading this thread rather late into it
>and don't have the original context...so what are we talking about? :-)

Many kernel functions already use copyin/copyout.  Consider fstat().  The
last thing it does (before a return) is:

	error = copyout((caddr_t)&ub, (caddr_t)uap->sb, sizeof (ub));

Now consider ibcs2_fstat().  Rather than duplicate the work of fstat
(which would really be a bad thing, because fstat may change and
ibcs2_fstat may not be updated), ibcs2_fstat goes through some
*HORRIBLE* hacks to allocate some space in user space, using the
stackgap_alloc() function.

If fstat used uiomove(), then it could go to a kernel address.
(Admittedly, it would still have to be changed -- it would have to have a
struct uio passed in, probably.  But then ibcs2_fstat could call that
common function, with less pain.)

Mind you, I'm probably not going to stop using copyin()/copyout() myself...
but it is something to consider.  And it's a valid point.  (Unless, of
course, you don't think that the emulation support is worthwhile :).)

Please let me know if this still isn't clear... I've been working on some
other stuff, and am tired, and my explanation ability goes downhill quickly
when I'm tired :).