From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jamie Lokier Subject: Re: [PATCH 1 of 3] block: Export I/O hints for block devices and partitions Date: Fri, 6 Jun 2008 15:21:03 +0100 Message-ID: <20080606142103.GA22158@shareable.org> References: <3e5d8520d247223a236e.1212643370@sermon.lab.mkp.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: linux-scsi@vger.kernel.org, linux-fsdevel@vger.kernel.org To: "Martin K. Petersen" Return-path: Received: from mail2.shareable.org ([80.68.89.115]:43062 "EHLO mail2.shareable.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755153AbYFFOVF (ORCPT ); Fri, 6 Jun 2008 10:21:05 -0400 Content-Disposition: inline In-Reply-To: <3e5d8520d247223a236e.1212643370@sermon.lab.mkp.net> Sender: linux-fsdevel-owner@vger.kernel.org List-ID: Martin K. Petersen wrote: > + sector_t phys_offset; > + unsigned int optimal_io_block; > + unsigned int optimal_io_length; Wouldn't off_t or sector_t be more appropriate? I can (vaguely) imagine devices with stripe sizes >= 2GiB in the not too distant future. Especially clusters with high speed interconnect and RAM-backed devices. There's no harm in an application doing smaller I/O. It should always choose appropriate values itself anyway. But it seems cleaner to report the numbers derived from actual drive topology. -- Jamie