From mboxrd@z Thu Jan 1 00:00:00 1970 From: "H. Peter Anvin" Subject: Re: [PATCH] BISECTED x86: avoid qword access in memcpy_*io Date: Wed, 21 Jul 2010 20:59:48 -0700 Message-ID: <4C47C234.7040400@zytor.com> References: <4C464BA0.20609@jp.fujitsu.com> <4C465FF0.1030407@zytor.com> <4C479D7F.2070204@jp.fujitsu.com> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 7bit Return-path: Received: from terminus.zytor.com ([198.137.202.10]:57783 "EHLO mail.zytor.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750760Ab0GVEAQ (ORCPT ); Thu, 22 Jul 2010 00:00:16 -0400 In-Reply-To: <4C479D7F.2070204@jp.fujitsu.com> Sender: linux-scsi-owner@vger.kernel.org List-Id: linux-scsi@vger.kernel.org To: Hidetoshi Seto Cc: linux-kernel@vger.kernel.org, linux-scsi@vger.kernel.org, James Smart , Thomas Gleixner , Ingo Molnar , x86@kernel.org, Brian Gerst On 07/21/2010 06:23 PM, Hidetoshi Seto wrote: > > It seems that lpfc already implemented lpfc_memcpy_{to,from}_slim() > as such memcpy_*io32, but limited use of it to on big endian platforms > only. Now lpfc can move to use it always, right? > What it probably should do is instead of open-coding this stuff use __iowrite32_copy and we probably should have a version without double underscores which byteswaps on bigendian. Personally I would not object to seeing a patch renaming __iowrite32_copy() to memcpy_toio32()... especially since it doesn't have the "memory or I/O space" property of iowrite and friends. -hpa -- H. Peter Anvin, Intel Open Source Technology Center I work for Intel. I don't speak on their behalf.