From mboxrd@z Thu Jan 1 00:00:00 1970 From: Christoph Hellwig Subject: Re: [RFC][PATCH] direct-io: btrfs: avoid splitting dio requests for non-btrfs filesystems Date: Tue, 2 Nov 2010 14:58:33 -0400 Message-ID: <20101102185833.GA7404@infradead.org> References: <4CD001A2.4000408@linux.vnet.ibm.com> <20101102145717.GA2531@localhost.localdomain> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Christian Ehrhardt , Jeff Moyer , Josef Bacik , Chris Mason , "linux-kernel@vger.kernel.org" , linux-btrfs@vger.kernel.org, "linux-fsdevel@vger.kernel.org" To: Josef Bacik Return-path: Content-Disposition: inline In-Reply-To: <20101102145717.GA2531@localhost.localdomain> Sender: linux-btrfs-owner@vger.kernel.org List-Id: linux-fsdevel.vger.kernel.org On Tue, Nov 02, 2010 at 10:57:18AM -0400, Josef Bacik wrote: > (which I did anyway because of my bug). So maybe the right idea is to rip out > my logical offset tests altogether and fix dio so we treat buffer_boundary() > like gospel. That way Btrfs can get what it needs without having this weird > special code, and then we can look at how other fs's set buffer_boundary (I'm > pretty sure ext2/3 are the only ones) and make sure they are setting it when > they really mean to. That sounds pretty reasonable to me. I really don't like the flag in the kiocb in this patch, and handling it as part of the get_blocks callback sounds much better to me. I don't know enough about the bounary blocks to know if we can reuse them - if we can it's perfect, if not another buffer flag seems like the way to go.