From: Christoph Hellwig <hch-wEGCiKHe2LqWVfeAwA7xHQ@public.gmane.org>
Subject: Re: [PATCH 02/11] ext4: Allow unaligned unlocked DAX IO
Date: Wed, 2 Nov 2016 07:27:46 -0700
Message-ID: <20161102142746.GA18387@infradead.org>
References: <1478034381-19037-1-git-send-email-jack@suse.cz>
 <1478034381-19037-3-git-send-email-jack@suse.cz>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit
Cc: linux-fsdevel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, Ted Tso <tytso-3s7WtUTddSA@public.gmane.org>,
 linux-ext4-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, Dave Chinner <david-FqsqvQoI3Ljby3iVrkZq2A@public.gmane.org>,
 linux-nvdimm-y27Ovi1pjclAfugRpC6u6w@public.gmane.org
To: Jan Kara <jack-AlSwsSmVLrQ@public.gmane.org>
Return-path: <linux-nvdimm-bounces-hn68Rpc1hR1g9hUCZPvPmw@public.gmane.org>
Content-Disposition: inline
In-Reply-To: <1478034381-19037-3-git-send-email-jack-AlSwsSmVLrQ@public.gmane.org>
List-Unsubscribe: <https://lists.01.org/mailman/options/linux-nvdimm>,
 <mailto:linux-nvdimm-request-hn68Rpc1hR1g9hUCZPvPmw@public.gmane.org?subject=unsubscribe>
List-Archive: <http://lists.01.org/pipermail/linux-nvdimm/>
List-Post: <mailto:linux-nvdimm-hn68Rpc1hR1g9hUCZPvPmw@public.gmane.org>
List-Help: <mailto:linux-nvdimm-request-hn68Rpc1hR1g9hUCZPvPmw@public.gmane.org?subject=help>
List-Subscribe: <https://lists.01.org/mailman/listinfo/linux-nvdimm>,
 <mailto:linux-nvdimm-request-hn68Rpc1hR1g9hUCZPvPmw@public.gmane.org?subject=subscribe>
Errors-To: linux-nvdimm-bounces-hn68Rpc1hR1g9hUCZPvPmw@public.gmane.org
Sender: "Linux-nvdimm" <linux-nvdimm-bounces-hn68Rpc1hR1g9hUCZPvPmw@public.gmane.org>
List-Id: linux-ext4.vger.kernel.org

On Tue, Nov 01, 2016 at 10:06:12PM +0100, Jan Kara wrote:
> Currently we don't allow unaligned writes without inode_lock. This is
> because zeroing of partial blocks could cause data corruption for racing
> unaligned writes to the same block. However DAX handles zeroing during
> block allocation and thus zeroing of partial blocks cannot race. Allow
> DAX unaligned IO to run without inode_lock.

DAX writes should always take the inode lock.  Without that you break
the expectations of existing applications that did not specify O_DIRECT.