From: Andreas Dilger Subject: Re: [PATCH 1/4] vfs: vfs-level fiemap interface Date: Mon, 15 Sep 2008 23:49:43 -0700 Message-ID: <20080916064514.GH3241@webber.adilger.int> References: <1221331767-16870-1-git-send-email-tytso@mit.edu> <20080914134711.GA21746@infradead.org> <20080914180132.GC13074@mit.edu> <20080914180843.GA31649@infradead.org> <20080914195811.GE13074@mit.edu> <20080915144754.GA16491@infradead.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7BIT Cc: Theodore Tso , linux-fsdevel@vger.kernel.org, linux-ext4@vger.kernel.org, akpm@linuxfoundation.org, Mark Fasheh , mtk.manpages@gmail.com To: Christoph Hellwig Return-path: Received: from sca-es-mail-1.Sun.COM ([192.18.43.132]:46458 "EHLO sca-es-mail-1.sun.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750754AbYIPGtz (ORCPT ); Tue, 16 Sep 2008 02:49:55 -0400 In-reply-to: <20080915144754.GA16491@infradead.org> Content-disposition: inline Sender: linux-ext4-owner@vger.kernel.org List-ID: On Sep 15, 2008 10:47 -0400, Christoph Hellwig wrote: > Again, I'd love it when people would read my mails :( Can we please > first agree on what the flag is actually supposed to mean? I made one > guess in this mail, but I'm not sure if that's even what Andreas > intended. And of course the description for it is even more important > than the name. The intent of this flag was a "catch-all" to indicate it isn't safe to try and read this block from disk, either because it is encrypted, compressed, on a remote system (HSM or over a network), or maybe not even written to disk yet (delalloc). In some cases (e.g. dump on a snapshot, or boot with LILO) it IS ok to read directly from a block device underneath the filesystem, but that would completely fail for the above cases. Note that the NO_BYPASS (formerly NO_DIRECT) flag is meant to be used in conjunction with other flags that specify more clearly the reason that this block is not directly accessible. Having a "generic" flag cover these different flag allows simple applications to know whether the block is readable or not, without having to understand each flag itself, and allows the flags to be expanded in the future (e.g. HSM or whatever that doesn't get included right away). Cheers, Andreas -- Andreas Dilger Sr. Staff Engineer, Lustre Group Sun Microsystems of Canada, Inc.