From mboxrd@z Thu Jan 1 00:00:00 1970 From: Mike Snitzer Subject: Re: dm-bufio: move dm-bufio.h to include/linux/ Date: Mon, 12 Mar 2018 11:57:13 -0400 Message-ID: <20180312155712.GB29395@redhat.com> References: <20180308144955.GA23262@infradead.org> <20180308203621.GA8133@redhat.com> <20180312074827.GA30695@infradead.org> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Return-path: Content-Disposition: inline In-Reply-To: <20180312074827.GA30695@infradead.org> List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: dm-devel-bounces@redhat.com Errors-To: dm-devel-bounces@redhat.com To: Christoph Hellwig Cc: dm-devel@redhat.com, Mikulas Patocka , Thomas Jaskiewicz List-Id: dm-devel.ids On Mon, Mar 12 2018 at 3:48am -0400, Christoph Hellwig wrote: > On Thu, Mar 08, 2018 at 03:36:21PM -0500, Mike Snitzer wrote: > > While I appreciate your interest in what constitutes a reasonable change > > for DM, why are you so emphatic about disallowing out-of-tree DM modules > > from using dm-bufio? > > Because I'd rather get them into the tree than providing escapes just > for them. This isn't about providing an escape. It may be unintuitive, or you may not believe me, but as I said in my previous email this change encourages out of tree DM targets to use dm-bufio rather than invent their own equivalent. It is very useful to me (speaking as the sucker who has to shepherd DM targets upstream) for others to get that code path right, without reinventing it, rather than have to disruptively change the target to use dm-bufio (where it makes sense) prior to submission for upstream inclusion. For DM targets that are extremely complex, external use of dm-bufio vs not is the tail wagging the dog. There are like 50+ other things that need work in addition to using dm-bufio. But every code transformation hurdle to upstream inclusion that can be done out of tree will help the DM target be more stable upon upstream inclusion. Mike