Return-Path: linux-nfs-owner@vger.kernel.org Received: from mail-pa0-f50.google.com ([209.85.220.50]:53009 "EHLO mail-pa0-f50.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751153AbaLOWLI (ORCPT ); Mon, 15 Dec 2014 17:11:08 -0500 Received: by mail-pa0-f50.google.com with SMTP id bj1so12766185pad.9 for ; Mon, 15 Dec 2014 14:11:07 -0800 (PST) Date: Mon, 15 Dec 2014 14:11:00 -0800 From: Omar Sandoval To: Christoph Hellwig Cc: Jan Kara , Alexander Viro , Andrew Morton , Trond Myklebust , David Sterba , linux-fsdevel@vger.kernel.org, linux-mm@kvack.org, linux-nfs@vger.kernel.org, linux-kernel@vger.kernel.org Subject: Re: [PATCH 2/8] swap: lock i_mutex for swap_writepage direct_IO Message-ID: <20141215221100.GA4637@mew> References: <20141215162705.GA23887@quack.suse.cz> <20141215165615.GA19041@infradead.org> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii In-Reply-To: <20141215165615.GA19041@infradead.org> Sender: linux-nfs-owner@vger.kernel.org List-ID: On Mon, Dec 15, 2014 at 08:56:15AM -0800, Christoph Hellwig wrote: > On Mon, Dec 15, 2014 at 05:27:05PM +0100, Jan Kara wrote: > > On Sun 14-12-14 21:26:56, Omar Sandoval wrote: > > > The generic write code locks i_mutex for a direct_IO. Swap-over-NFS > > > doesn't grab the mutex because nfs_direct_IO doesn't expect i_mutex to > > > be held, but most direct_IO implementations do. > > I think you are speaking about direct IO writes only, aren't you? For DIO > > reads we don't hold i_mutex AFAICS. And also for DIO writes we don't > > necessarily hold i_mutex - see for example XFS which doesn't take i_mutex > > for direct IO writes. It uses it's internal rwlock for this (see > > xfs_file_dio_aio_write()). So I think this is just wrong. > > The problem is that the use of ->direct_IO by the swap code is a gross > layering violation. ->direct_IO is a callback for the filesystem, and > the swap code need to call ->read_iter instead of ->readpage and > ->write_tier instead of ->direct_IO, and leave the locking to the > filesystem. > Ok, I got the swap code working with ->read_iter/->write_iter without too much trouble. I wanted to double check before I submit if there's any gotchas involved with adding the O_DIRECT flag to a file pointer after it has been opened -- swapon opens the swapfile before we know if we're using the SWP_FILE infrastructure, and we need to add O_DIRECT so ->{read,write}_iter use direct I/O, but we can't add O_DIRECT to the original open without excluding filesystems that support the old bmap path but not direct I/O. -- Omar