Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1030451AbWALQEr (ORCPT ); Thu, 12 Jan 2006 11:04:47 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1030453AbWALQEr (ORCPT ); Thu, 12 Jan 2006 11:04:47 -0500 Received: from mx.pathscale.com ([64.160.42.68]:9143 "EHLO mx.pathscale.com") by vger.kernel.org with ESMTP id S1030451AbWALQEr (ORCPT ); Thu, 12 Jan 2006 11:04:47 -0500 Subject: Re: [PATCH 2 of 3] memcpy32 for x86_64 From: "Bryan O'Sullivan" To: Denis Vlasenko Cc: akpm@osdl.org, linux-kernel@vger.kernel.org, hch@infradead.org, ak@suse.de, rdreier@cisco.com In-Reply-To: <200601121038.17764.vda@ilport.com.ua> References: <200601121038.17764.vda@ilport.com.ua> Content-Type: text/plain Organization: PathScale, Inc. Date: Thu, 12 Jan 2006 08:04:41 -0800 Message-Id: <1137081882.28011.1.camel@serpentine.pathscale.com> Mime-Version: 1.0 X-Mailer: Evolution 2.2.3 (2.2.3-2.fc4) Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 669 Lines: 17 On Thu, 2006-01-12 at 10:38 +0200, Denis Vlasenko wrote: > 2) On all current x86_64 hardware each 64bit access from/to > IO mapped addresses is always converted to two 32bit accesses. This is true for 64-bit writes over Hypertransport (reads don't get split up this way), but not for PCI-Express memory writes, which remain atomic 64-bit. I'll be converting the 64-bit accesses to 32-bit, as you and Andi suggested.