From mboxrd@z Thu Jan 1 00:00:00 1970 From: Greg Banks Subject: Re: Proposal to improve filesystem/block snapshot interaction Date: Wed, 31 Oct 2007 14:44:36 +1100 Message-ID: <20071031034436.GB9041@sgi.com> References: <20070927063113.GD2989@sgi.com> <20071030010453.GF27385@sgi.com> <7fe698080710300235v3ce49613nfe3c5e733f1b6f5b@mail.gmail.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Linux Filesystem Mailing List , David Chinner , Donald Douwsma , Christoph Hellwig , Roger Strassburg , Mark Goodwin , Brett Jon Grandbois , Arnd Bergmann To: Dongjun Shin Return-path: Received: from netops-testserver-3-out.sgi.com ([192.48.171.28]:59808 "EHLO relay.sgi.com" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1752410AbXJaDgf (ORCPT ); Tue, 30 Oct 2007 23:36:35 -0400 Content-Disposition: inline In-Reply-To: <7fe698080710300235v3ce49613nfe3c5e733f1b6f5b@mail.gmail.com> Sender: linux-fsdevel-owner@vger.kernel.org List-Id: linux-fsdevel.vger.kernel.org On Tue, Oct 30, 2007 at 06:35:08PM +0900, Dongjun Shin wrote: > On 10/30/07, Greg Banks wrote: > > > > BIO_HINT_RELEASE > > The bio's block extent is no longer in use by the filesystem > > and will not be read in the future. Any storage used to back > > the extent may be released without any threat to filesystem > > or data integrity. > > > > I'd like to second the proposal, but it would be more useful to bring the hint > down to the physical devices. > > There is an ongoing discussion about adding 'Trim' ATA command for notifying > the drive about the deleted blocks. > > http://www.t13.org/Documents/UploadedDocuments/docs2007/e07154r3-Data_Set_Management_Proposal_for_ATA-ACS2.pdf What an interesting document. Am I reading the change markup correctly, did it get *simpler* in the last revision? Wow. I agree that BIO_HINT_RELEASE would be a good match for the proposed Trim command. But I don't think we'll ever be issuing Trims with more than a single LBA Range Entry, that feature seems unhelpful. The Trim proposal doesn't specify what happens when a sector which is already deallocated is deallocated again, presumably this is supposed to be harmless? Greg. -- Greg Banks, R&D Software Engineer, SGI Australian Software Group. Apparently, I'm Bedevere. Which MPHG character are you? I don't speak for SGI.