From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <linux-kernel-owner@vger.kernel.org>
Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand
	id S1753032AbbIKPB1 (ORCPT <rfc822;w@1wt.eu>);
	Fri, 11 Sep 2015 11:01:27 -0400
Received: from smtp02.smtpout.orange.fr ([80.12.242.124]:20894 "EHLO
	smtp.smtpout.orange.fr" rhost-flags-OK-OK-OK-OK) by vger.kernel.org
	with ESMTP id S1752114AbbIKPBZ (ORCPT
	<rfc822;linux-kernel@vger.kernel.org>);
	Fri, 11 Sep 2015 11:01:25 -0400
X-ME-Helo: belgarion
X-ME-Auth: amFyem1pay5yb2JlcnRAb3JhbmdlLmZy
X-ME-Date: Fri, 11 Sep 2015 17:01:24 +0200
X-ME-IP: 109.220.179.182
From: Robert Jarzmik <robert.jarzmik@free.fr>
To: Russell King <rmk+kernel@arm.linux.org.uk>
Cc: Dave Martin <Dave.Martin@arm.com>, linux-kernel@vger.kernel.org,
        linux-arm-kernel@lists.infradead.org
Subject: Re: [PATCH 2/2] ARM: domains: add memory dependencies to get_domain/set_domain
References: <20150911095438.GL21084@n2100.arm.linux.org.uk>
	<E1ZaL45-0003fs-1k@rmk-PC.arm.linux.org.uk>
X-URL: http://belgarath.falguerolles.org/
Date: Fri, 11 Sep 2015 16:56:27 +0200
In-Reply-To: <E1ZaL45-0003fs-1k@rmk-PC.arm.linux.org.uk> (Russell King's
	message of "Fri, 11 Sep 2015 10:56:17 +0100")
Message-ID: <87si6lt36s.fsf@belgarion.home>
User-Agent: Gnus/5.130008 (Ma Gnus v0.8) Emacs/24.4 (gnu/linux)
MIME-Version: 1.0
Content-Type: text/plain
Sender: linux-kernel-owner@vger.kernel.org
List-ID: <linux-kernel.vger.kernel.org>
X-Mailing-List: linux-kernel@vger.kernel.org

Russell King <rmk+kernel@arm.linux.org.uk> writes:

> We need to have memory dependencies on get_domain/set_domain to avoid
> the compiler over-optimising these inline assembly instructions.
>
> Loads/stores must not be reordered across a set_domain(), so introduce
> a compiler barrier for that assembly.
>
> The value of get_domain() must not be cached across a set_domain(), but
> we still want to allow the compiler to optimise it away.  Introduce a
> dependency on current_thread_info()->cpu_domain to avoid this; the new
> memory clobber in set_domain() should therefore cause the compiler to
> re-load this.  The other advantage of using this is we should have its
> address in the register set already, or very soon after at most call
> sites.
>
> Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Tested-by: Robert Jarzmik <robert.jarzmik@free.fr>

The test is failing without these 2 patches, while with them, an unaligned
access is fixed, and the generated code looks good :
     7ac:       ee13cf10        mrc     15, 0, ip, cr3, cr0, {0}
     7b0:       e3cc300c        bic     r3, ip, #12
     7b4:       e58dc014        str     ip, [sp, #20]
     7b8:       e3833004        orr     r3, r3, #4
     7bc:       ee033f10        mcr     15, 0, r3, cr3, cr0, {0}

Cheers.

--
Robert