Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1756239AbXLJSBL (ORCPT ); Mon, 10 Dec 2007 13:01:11 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1753940AbXLJSA5 (ORCPT ); Mon, 10 Dec 2007 13:00:57 -0500 Received: from [198.99.130.12] ([198.99.130.12]:55196 "EHLO saraswathi.solana.com" rhost-flags-FAIL-FAIL-OK-OK) by vger.kernel.org with ESMTP id S1753729AbXLJSA4 (ORCPT ); Mon, 10 Dec 2007 13:00:56 -0500 Date: Mon, 10 Dec 2007 13:00:09 -0500 From: Jeff Dike To: Andrew Morton Cc: LKML , uml-devel Subject: [PATCH 5/5] UML - use barrier() instead of mb() Message-ID: <20071210180009.GA9333@c2.user-mode-linux.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.4.2.3i Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 2791 Lines: 84 signals_enabled and pending have requirements on the order in which they are modified. This used to be done by declaring them volatile and putting an mb() where the ordering requirements were in effect. After getting a better (I hope) understanding of how to do this correctly, the volatile declarations are gone and the mb()'s replaced by barrier()'s. One of the mb()'s was deleted because I see no problematic writes that could be re-ordered past that point. Signed-off-by: Jeff Dike --- arch/um/os-Linux/signal.c | 28 ++++++++-------------------- 1 file changed, 8 insertions(+), 20 deletions(-) Index: linux-2.6-git/arch/um/os-Linux/signal.c =================================================================== --- linux-2.6-git.orig/arch/um/os-Linux/signal.c 2007-12-07 16:53:18.000000000 -0500 +++ linux-2.6-git/arch/um/os-Linux/signal.c 2007-12-07 17:37:10.000000000 -0500 @@ -15,6 +15,9 @@ #include "sysdep/sigcontext.h" #include "user.h" +/* Copied from linux/compiler-gcc.h since we can't include it directly */ +#define barrier() __asm__ __volatile__("": : :"memory") + /* * These are the asynchronous signals. SIGPROF is excluded because we want to * be able to profile all of UML, not just the non-critical sections. If @@ -27,13 +30,8 @@ #define SIGVTALRM_BIT 1 #define SIGVTALRM_MASK (1 << SIGVTALRM_BIT) -/* - * These are used by both the signal handlers and - * block/unblock_signals. I don't want modifications cached in a - * register - they must go straight to memory. - */ -static volatile int signals_enabled = 1; -static volatile int pending = 0; +static int signals_enabled; +static unsigned int pending; void sig_handler(int sig, struct sigcontext *sc) { @@ -198,7 +196,7 @@ void block_signals(void) * This might matter if gcc figures out how to inline this and * decides to shuffle this code into the caller. */ - mb(); + barrier(); } void unblock_signals(void) @@ -224,21 +222,11 @@ void unblock_signals(void) * Setting signals_enabled and reading pending must * happen in this order. */ - mb(); + barrier(); save_pending = pending; - if (save_pending == 0) { - /* - * This must return with signals enabled, so - * this barrier ensures that writes are - * flushed out before the return. This might - * matter if gcc figures out how to inline - * this (unlikely, given its size) and decides - * to shuffle this code into the caller. - */ - mb(); + if (save_pending == 0) return; - } pending = 0; -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/