From owner-freebsd-commit  Sat Oct  7 23:42:52 1995
Return-Path: owner-commit
Received: (from root@localhost)
          by freefall.freebsd.org (8.6.12/8.6.6) id XAA20740
          for freebsd-commit-outgoing; Sat, 7 Oct 1995 23:42:52 -0700
Received: (from root@localhost)
          by freefall.freebsd.org (8.6.12/8.6.6) id XAA20728
          for cvs-all-outgoing; Sat, 7 Oct 1995 23:42:43 -0700
Received: (from root@localhost)
          by freefall.freebsd.org (8.6.12/8.6.6) id XAA20718
          for cvs-sys-outgoing; Sat, 7 Oct 1995 23:42:39 -0700
Received: from meter.eng.uci.edu (root@meter.eng.uci.edu [128.200.85.3])
          by freefall.freebsd.org (8.6.12/8.6.6) with ESMTP id XAA20713
          ; Sat, 7 Oct 1995 23:42:28 -0700
Received: from newport.ece.uci.edu by meter.eng.uci.edu (8.7) id XAA20350; Sat, 7 Oct 1995 23:42:26 -0700 (PDT)
Received: from localhost by newport.ece.uci.edu (8.7) id XAA20642; Sat, 7 Oct 1995 23:42:24 -0700 (PDT)
Message-Id: <199510080642.XAA20642@newport.ece.uci.edu>
To: Bruce Evans <bde@zeta.org.au>
cc: CVS-commiters@freefall.freebsd.org, cvs-sys@freefall.freebsd.org
Subject: Re: cvs commit: src/sys/sys systm.h 
In-reply-to: Your message of "Sun, 08 Oct 1995 11:54:51 +1000."
             <199510080154.LAA11459@godzilla.zeta.org.au> 
Date: Sat, 07 Oct 1995 23:42:18 -0700
From: Steven Wallace <swallace@ece.uci.edu>
Sender: owner-commit@FreeBSD.org
Precedence: bulk

>>  Add appropriate #include <sys/sysproto.h> to files that needed
>>  protos from systm.h.
> 
> This duplicates work that I was waiting to commit.  <sys/sysproto.h> has
> to be included in 48 headers altogether.  Sigh.
> 
>>  Add structure definitions to appropriate files that relied on sys/systm.h,
>>  right before system call definition, as in the rest of the kernel source.
> 
> Aren't the definitions in <sys/sysproto.h>?
> 

<sys/systm.h> had protos in it which our now removed since they are in
<sys/sysproto.h> now.  I only added #include <sys/sysproto.h> that
were making EXTERNAL calls and needed the struct/proto.
I suppose the real way to do stuff is to #include <sys/sysproto.h> everywhere
and remove the local struct definition (I thought they could be left
in place as comments since it is nice to look right at the function
instead of searching through a header file).  But I did not want
to commit such large changes.  I was simply conforming my changes
to fit the apparent standard:  define structure near system call definiton
and use <sys/systm.h> (which I have now redefined to use <sys/sysproto.h>
when an outside function needs the structure/proto to make an internal
system call.

Should we move to include <sys/sysproto.h> everywhere and remove
then unnecessary "local" arg structure definitions?  Do you want
to work on such changes?

Also, I would like to use "void *uap" instead of "struct dummy_arg *uap"
for system calls that do not need an argument.   Why do we need
to define a dummy structure with a single dummy int in it?  Won't
void * work just as good?  (Then I can remove all that dummy struct
generation)

BTW, regarding IBCS2 stuff, I now have new working code (taken from
NetBSD) and fixed everything up so that it compiles cleanly with
no warning messages.  It includes <sys/sysproto.h> everywhere
and also its local <i386/ibcs2/ibcs2_proto.h> (generated using
ibcs2 syscalls.master and new makesyscalls.sh).

Steven