Return-Path: Date: Mon, 13 Apr 2015 19:22:25 +0200 From: Christoph Hellwig Subject: Re: [GIT PULL] PMEM driver for v4.1 Message-ID: <20150413172225.GC14243@lst.de> References: <20150413093309.GA30219@gmail.com> <20150413093541.GA5147@lst.de> <20150413104531.GB30556@gmail.com> <552BB4CA.6060808@plexistor.com> <20150413123535.GA2666@gmail.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20150413123535.GA2666@gmail.com> Sender: linux-kernel-owner@vger.kernel.org To: Ingo Molnar Cc: Boaz Harrosh , Christoph Hellwig , Linus Torvalds , linux-kernel@vger.kernel.org, linux-nvdimm@lists.01.org, Ross Zwisler , Dan Williams , Matthew Wilcox List-ID: On Mon, Apr 13, 2015 at 02:35:35PM +0200, Ingo Molnar wrote: > How does splice work with DAX files? By falling back to default_file_splice_read/default_file_splice_write which doesn't use the iter ops, but instead requires a copy in the splice code. But given that the actual underlying reads and writes bypass the pagecache it's not any less effiecient than the normal pagecache based splice. > AFAICS vmsplice() won't work, as > it uses get_user_pages(), which needs struct page backing. Exactly. > Also, how > will f_op->sendpage() work? That too needs page backing. default_file_splice_read allocates it's own kernel pages, which are then passed to ->sendpage.