From owner-freebsd-current  Tue Dec  2 09:06:26 1997
Return-Path: <owner-freebsd-current>
Received: (from root@localhost)
          by hub.freebsd.org (8.8.7/8.8.7) id JAA14822
          for current-outgoing; Tue, 2 Dec 1997 09:06:26 -0800 (PST)
          (envelope-from owner-freebsd-current)
Received: from kithrup.com (kithrup.com [205.179.156.40])
          by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id JAA14814
          for <current@freebsd.org>; Tue, 2 Dec 1997 09:06:20 -0800 (PST)
          (envelope-from sef@kithrup.com)
Received: (from sef@localhost)
	by kithrup.com (8.8.8/8.8.7) id JAA28517;
	Tue, 2 Dec 1997 09:06:20 -0800 (PST)
	(envelope-from sef)
Date: Tue, 2 Dec 1997 09:06:20 -0800 (PST)
From: Sean Eric Fagan <sef@kithrup.com>
Message-Id: <199712021706.JAA28517@kithrup.com>
To: current@freebsd.org
Reply-To: current@freebsd.org
Subject: Re: cvs commit: src/sys/kern vfs_aio.c 
In-Reply-To: <1357.881065577.kithrup.freebsd.current@critter.freebsd.dk>
References: Your message of "Mon, 01 Dec 1997 15:55:30 PST."
             <199712012355.PAA11311@kithrup.com> 
Organization: Kithrup Enterprises, Ltd.
Sender: owner-freebsd-current@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk

In article <1357.881065577.kithrup.freebsd.current@critter.freebsd.dk> you write:
>>I questioned the change, and said I "strongly objected" to it.
>...with no further comments.  If you had bothered to say why, I could have
>engaged in a discussion with you about it...

I made a couple of comments.  Gosh, didn't know I was required to prove
myself to you, though.  But, instead, you just ignored it -- and now people
are asking why there was no discussion about it.

Who reviewed the changes?  Who submitted them?  What PR did they fix?  None
of those things was mentioned in the commit message.

>I think that the entire "retval" thing should go away entirely and be
>handled in the libc syscall stubs, this was the first step towards 
>that goal.

And handled *how*?

>And yes, the reason is performance.

Then you should back the changes out IMMEDIATELY, as

	*retval = 10;

is exactly the same speed as:

	p->p_retval[0] = 10;

yet takes up less total system space (two words on the kernel stack vs. two
words added to the proc structure of every process), requires less copying
when dealing with emulation support, and requires less changing when
incorporating code from other code bases (you *have* heard of things such as
NetBSD, BSDi, 4.4BSD, and so forth?).

>No you didn't, all you said was "I strongly object to this change".

And I still do.

*You* are the one who checked it in.  I challenge *you* to justify this
intrusive, complex, not-apparantly-worthwhile change.  I claim that it is
not worth it, and should be backed out.