Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1030251AbVLMW2H (ORCPT ); Tue, 13 Dec 2005 17:28:07 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1030284AbVLMW2H (ORCPT ); Tue, 13 Dec 2005 17:28:07 -0500 Received: from mail.ocs.com.au ([202.147.117.210]:39619 "EHLO mail.ocs.com.au") by vger.kernel.org with ESMTP id S1030251AbVLMW2F (ORCPT ); Tue, 13 Dec 2005 17:28:05 -0500 X-Mailer: exmh version 2.6.3_20040314 03/14/2004 with nmh-1.1 From: Keith Owens To: paulmck@us.ibm.com Cc: Andi Kleen , ajwade@cpe001346162bf9-cm0011ae8cd564.cpe.net.cable.rogers.com, vatsa@in.ibm.com, Oleg Nesterov , linux-kernel@vger.kernel.org, Dipankar Sarma , Andrew Morton , Ingo Molnar Subject: Re: Semantics of smp_mb() [was : Re: [PATCH] Fix RCU race in access of nohz_cpu_mask ] In-reply-to: Your message of "Tue, 13 Dec 2005 08:20:27 -0800." <20051213162027.GA14158@us.ibm.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Wed, 14 Dec 2005 09:27:41 +1100 Message-ID: <17158.1134512861@ocs3.ocs.com.au> Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 830 Lines: 20 On Tue, 13 Dec 2005 08:20:27 -0800, "Paul E. McKenney" wrote: >If the variable p references MMIO rather than normal memory, then >wmb() and rmb() are needed instead of smp_wmb() and smp_rmb(). mmiowb(), not wmb(). IA64 has a different form of memory fence for I/O space compared to normal memory. MIPS also has a non-empty form of mmiowb(). >This is because the I/O device needs to see the accesses ordered >even in a UP system. Why does mmiowb() map to empty on most systems, including Alpha? Should it not map to wmb() for everything except IA64 and MIPS? - 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/