From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Date: Wed, 10 Jun 2015 09:33:16 +0200 From: Christoph Hellwig Subject: Re: [PATCH v5 05/21] libnvdimm: control (ioctl) messages for libnvdimm bus and dimm devices Message-ID: <20150610073316.GB3369@lst.de> References: <20150602001134.4506.45867.stgit@dwillia2-desk3.amr.corp.intel.com> <20150602001435.4506.42614.stgit@dwillia2-desk3.amr.corp.intel.com> <20150609063418.GB9804@lst.de> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: Sender: linux-kernel-owner@vger.kernel.org To: Dan Williams Cc: Christoph Hellwig , "linux-nvdimm@lists.01.org" , Jens Axboe , Stephen Rothwell , "Rafael J. Wysocki" , Neil Brown , Greg KH , "linux-kernel@vger.kernel.org" , Ingo Molnar , Linux ACPI , jmoyer , linux-api@vger.kernel.org, Andrew Morton List-ID: On Mon, Jun 08, 2015 at 11:57:42PM -0700, Dan Williams wrote: > Data payload size for one, these commands transfer more than a page > worth of data at a time. > > Even if we killed the ioctl interface to userspace we still need all > the ugly data marshaling code in the kernel to craft properly > formatted ACPI _DSM messages. I try to share as much common > infrastructure from the ACPI _DSM implementation to the ioctl > interface (nd_cmd_dimm_desc() + nd_cmd_bus_desc()). Ok. Not a fan of these dual interfaces but in this case they might be justified.