From mboxrd@z Thu Jan 1 00:00:00 1970 From: Christoph Hellwig Subject: Re: [RFC PATCH 00/22] loop: Issue O_DIRECT aio with pages Date: Wed, 29 Feb 2012 04:08:03 -0500 Message-ID: <20120229090803.GA9995@infradead.org> References: <1330377576-3659-1-git-send-email-dave.kleikamp@oracle.com> <20120228092926.GA2766@infradead.org> <4F4CEF5A.1020207@zabbo.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Christoph Hellwig , Dave Kleikamp , linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org To: Zach Brown Return-path: Content-Disposition: inline In-Reply-To: <4F4CEF5A.1020207@zabbo.net> Sender: linux-kernel-owner@vger.kernel.org List-Id: linux-fsdevel.vger.kernel.org On Tue, Feb 28, 2012 at 10:14:34AM -0500, Zach Brown wrote: > Yeah, that's reasonable. Perhaps obviously, we started with new entry > points to minimize the amount of churn we'd have to go through to test > the change in behaviour. > > It's going to be messy to try and abstract away the pinning and dirtying > of the iter regions from direct IO through the iter interface, but maybe > not horribly so. I don't really care to much about the implementation inside fs/direct-io.c (at least for now - once I see it I might still scream "bloody murder!"). The point is to pass the iov_iter all the way down to a common entry point in fs/direct-io.c, so that the filesystems don't have to care for that difference.