From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jens Axboe Subject: Re: [PATCH] libata: increase LBA48 max sectors to 65535 Date: Mon, 20 Feb 2006 14:58:10 +0100 Message-ID: <20060220135809.GK8852@suse.de> References: <43EFD73D.4070207@gmail.com> <43F011DF.6020203@pobox.com> <20060215072449.GA31691@htj.dyndns.org> <20060215130725.GP4203@suse.de> <43F342E6.9060909@gmail.com> <20060215151200.GW4203@suse.de> <43F38D49.4080601@pobox.com> <43F3B3CE.5060105@pobox.com> <43F9A6A8.4020606@pobox.com> <43F9CA10.8000206@pobox.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Received: from ns.virtualhost.dk ([195.184.98.160]:7538 "EHLO virtualhost.dk") by vger.kernel.org with ESMTP id S1030228AbWBTN6W (ORCPT ); Mon, 20 Feb 2006 08:58:22 -0500 Content-Disposition: inline In-Reply-To: <43F9CA10.8000206@pobox.com> Sender: linux-ide-owner@vger.kernel.org List-Id: linux-ide@vger.kernel.org To: Mark Lord Cc: Jeff Garzik , Tejun Heo , albertcc@tw.ibm.com, linux-ide@vger.kernel.org On Mon, Feb 20 2006, Mark Lord wrote: > Jeff Garzik wrote: > > > >Indeed. I am paranoid about controllers assuming that no more than 256 > >PRD entries will be submitted. But I have never seen any document that > >says PCI IDE BMDMA hardware can -only- do 256 PRD entries. Never tested > >it, either... > > The hardware vendors I have worked with have never had any limit > on PRD count -- they'd happily keep reading through an infinite > list of PRDs if we could supply one. > > A number of them did have issues with 64KB boundary crossings, > and it is generally VERY unsafe to have a single PRD "wrap" > into a fresh 64KB segment mid-stride. Both IDE and libata take care not to make that happen, so we should be fine there. The block layer also supports this property. -- Jens Axboe