From owner-freebsd-mips@FreeBSD.ORG  Thu Feb 25 02:31:56 2010
Return-Path: <owner-freebsd-mips@FreeBSD.ORG>
Delivered-To: freebsd-mips@freebsd.org
Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34])
	by hub.freebsd.org (Postfix) with ESMTP id 2BD7B106564A;
	Thu, 25 Feb 2010 02:31:56 +0000 (UTC) (envelope-from mahan@mahan.org)
Received: from ns.mahan.org (ns.mahan.org [67.116.10.138])
	by mx1.freebsd.org (Postfix) with ESMTP id 05FED8FC18;
	Thu, 25 Feb 2010 02:31:55 +0000 (UTC)
Received: from mahan.org (crowTrobot [67.116.10.140])
	by ns.mahan.org (8.13.6/8.13.6) with ESMTP id o1P2a2oX024250;
	Wed, 24 Feb 2010 18:36:02 -0800 (PST) (envelope-from mahan@mahan.org)
Message-Id: <201002250236.o1P2a2oX024250@ns.mahan.org>
To: Juli Mallett <jmallett@freebsd.org>
In-reply-to: <eaa228be1002241807q2613ccecy9773155e68ccda62@mail.gmail.com>
References: <17060.1267061906@mahan.org>
	<eaa228be1002241807q2613ccecy9773155e68ccda62@mail.gmail.com>
Comments: In-reply-to Juli Mallett <jmallett@freebsd.org>
	message dated "Wed, 24 Feb 2010 18:07:23 -0800."
Date: Wed, 24 Feb 2010 18:31:50 -0800
From: Patrick Mahan <mahan@mahan.org>
Cc: freebsd-mips@freebsd.org
Subject: Re: Writing MIPS assembler instructions in C
X-BeenThere: freebsd-mips@freebsd.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Porting FreeBSD to MIPS <freebsd-mips.freebsd.org>
List-Unsubscribe: <http://lists.freebsd.org/mailman/listinfo/freebsd-mips>,
	<mailto:freebsd-mips-request@freebsd.org?subject=unsubscribe>
List-Archive: <http://lists.freebsd.org/pipermail/freebsd-mips>
List-Post: <mailto:freebsd-mips@freebsd.org>
List-Help: <mailto:freebsd-mips-request@freebsd.org?subject=help>
List-Subscribe: <http://lists.freebsd.org/mailman/listinfo/freebsd-mips>,
	<mailto:freebsd-mips-request@freebsd.org?subject=subscribe>
X-List-Received-Date: Thu, 25 Feb 2010 02:31:56 -0000


See inline ...

> Hi Patrick,
> 
> On Wed, Feb 24, 2010 at 17:38, Patrick Mahan <mahan@mahan.org> wrote:
> > part of some code built on the linux platform and
> > there they are using "%[rt]" which I cannot find
> > a description.
> 
> %[foo] can be used to refer to named parameters (input or output,
> maybe even clobbered) of inline assembly instructions rather than
> using %0, %1, etc., which can be hard to read.  Consider:
> 
> %%%
> static inline bool
> atomic_cmpset64(volatile uint64_t *p, uint64_t o, uint64_t v)
> {
> 	uint64_t temp;
> 	int res;
> 
> 	asm volatile (
> 	"1:\n\t"
> 	"move	%[res], $0\n\t"
> 	"lld	%[temp], %[p]\n\t"
> 	"bne	%[temp], %[o], 2f\n\t"
> 	"move	%[temp], %[v]\n\t"
> 	"li	%[res], 1\n\t"
> 	"scd	%[temp], %[p]\n\t"
> 	"beqz	%[temp], 1b\n\t"
> 	"2:\n\t"
> 	: [res] "=&r"(res), [temp] "=&r"(temp), [p] "+m"(*p)
> 	: [o] "r"(o), [v] "r"(v)
> 	: "memory"
> 	);
> 
> 	return (res != 0);
> }
> %%%
> 
> The brackets with the input and output parameters specify what names
> to use to refer to them in the assembly listing (here they happen to
> be mostly the same as the variable names, but that's not necessary.)
>

Ah Sooo <*whack!*>......

I did not even twig on that... I had gotten it in my head that %[rt]
was something special to the compiler that caused a certain behavior
to occur.

Now my problem is I still need to force the value pointed to "addr" into
a specific register because there is a jalr to a function else where
that I only have binary access too and it expects it's a value in
that register.  Can I coerce this?

Thanks for the education.

Patrick