From mboxrd@z Thu Jan 1 00:00:00 1970 From: Christoph Hellwig Subject: Re: [PATCH 2/3] tmscsim: remove bogus endianness conversions Date: Sat, 5 May 2007 10:37:51 +0100 Message-ID: <20070505093751.GB9422@infradead.org> References: Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Received: from pentafluge.infradead.org ([213.146.154.40]:36783 "EHLO pentafluge.infradead.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1423241AbXEEJhw (ORCPT ); Sat, 5 May 2007 05:37:52 -0400 Content-Disposition: inline In-Reply-To: Sender: linux-scsi-owner@vger.kernel.org List-Id: linux-scsi@vger.kernel.org To: Guennadi Liakhovetski Cc: linux-scsi@vger.kernel.org, James Bottomley On Fri, May 04, 2007 at 10:59:40PM +0200, Guennadi Liakhovetski wrote: > cpu_to_le32 endianness conversions in tmscsim.c, followed by > arithmetic operations don't look correct. Besides, {in,out}[wl] > already perform the necessary conversions. Further, bus addresses > of request buffers are guaranteed to be (mapped) under 4G by > current scsi- and block-layer defaults. This could be explicitly > enforced by using blk_queue_bounce_limit(), which, however, > doesn't seem to be the common practice among SCSI drivers. Looks good to me. If there are structures in dma memory in the driver it would help to give the __le annotation to get the whole endianess handling completely right.