From: Matthew Wilcox Subject: RE: [PATCH v2 2/9] ext2: tell DAX the size of allocation holes Date: Sun, 11 Sep 2016 00:42:09 +0000 Message-ID: References: <20160823220419.11717-1-ross.zwisler@linux.intel.com> <20160823220419.11717-3-ross.zwisler@linux.intel.com> <20160825075728.GA11235@infradead.org> <20160826212934.GA11265@linux.intel.com> <20160829074116.GA16491@infradead.org> <20160829125741.cdnbb2uaditcmnw2@thunk.org> <20160909164808.GC18554@linux.intel.com> <20160910073151.GB5295@infradead.org> <20160910174910.yyirb7smiob7evt5@thunk.org> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Cc: Andrew Morton , "linux-nvdimm-y27Ovi1pjclAfugRpC6u6w@public.gmane.org" , Dave Chinner , "linux-kernel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org" , "linux-mm-Bw31MaZKKs3YtjvyW6yDsg@public.gmane.org" , Andreas Dilger , Alexander Viro , Jan Kara , "linux-fsdevel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org" , "linux-ext4-u79uwXL29TY76Z2rM5mHXA@public.gmane.org" To: Theodore Ts'o , Christoph Hellwig Return-path: In-Reply-To: <20160910174910.yyirb7smiob7evt5-AKGzg7BKzIDYtjvyW6yDsg@public.gmane.org> Content-Language: en-US List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: linux-nvdimm-bounces-hn68Rpc1hR1g9hUCZPvPmw@public.gmane.org Sender: "Linux-nvdimm" List-Id: linux-ext4.vger.kernel.org From: Theodore Ts'o [mailto:tytso-3s7WtUTddSA@public.gmane.org] > On Sat, Sep 10, 2016 at 12:31:51AM -0700, Christoph Hellwig wrote: > > I've mentioned this before, but I'd like to repeat it. With all the > > work reqwuired in the file system I would prefer to drop DAX support > > in ext2 (and if people really cry for it reinstate the trivial old xip > > support). > > Why is so much work required to support the new DAX interfaces in > ext2? Is that unique to ext2, or is adding DAX support just going to > be painful for all file systems? Hopefully it's not the latter, > right? It's always been my goal to make supporting DAX as easy as possible for the filesystem. Hence the sharing of the DIO locking, and the (it turned out) premature reliance on helper functions. It's more complex than I wanted it to be right now, and I hope we get to simplify it again.