From mboxrd@z Thu Jan 1 00:00:00 1970 From: Darren Hart Subject: Re: [alsa-devel] [PATCH 1/2] ASoC: rt5677: Add ACPI device probing Date: Tue, 25 Nov 2014 08:01:14 -0800 Message-ID: <5474A7CA.3020106@linux.intel.com> References: <1416034608-24238-1-git-send-email-benzh@chromium.org> <1416925686.7260.28.camel@loki> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <1416925686.7260.28.camel@loki> Sender: linux-kernel-owner@vger.kernel.org To: Liam Girdwood , Grant Likely Cc: Ben Zhang , Oder Chiou , alsa-devel , "Rafael J. Wysocki" , Liam Girdwood , Linux Kernel Mailing List , Mark Brown , Anatol Pomozov , Bard Liao , Dylan Reid , flove@realtek.com List-Id: alsa-devel@alsa-project.org On 11/25/14 06:28, Liam Girdwood wrote: > On Tue, 2014-11-25 at 12:11 +0000, Grant Likely wrote: >> On Sat, Nov 15, 2014 at 6:56 AM, Ben Zhang wrote: >>> The rt5677 codec driver looks for ACPI device ID "RT5677CE", >>> which is specified in coreboot. This patch allows platform >>> data to be obtained via ACPI >>> >>> Signed-off-by: Ben Zhang >> >> This looks like an ideal time to talk about shared DT and ACPI driver >> bindings. This driver /already/ has a firmware binding. It is >> documented in the kernel under >> Documentation/bindings/sound/rt5677.txt. We now have a standard method >> for sharing bindings between DT and ACPI in the _DSD method[1]. >> Support for DSD is in linux-next and getting merged into v3.19. This >> is exactly the case that _DSD should be used for passing additional >> data to the driver, and it should use the existing binding. >> >> [1] http://www.uefi.org/sites/default/files/resources/_DSD-device-properties-UUID.pdf >> >> For a long time we've had the rule on DT that new bindings must be >> documented before we merge a patch. That rule I think has been a good >> one, even if it is a little chaoitc. I think when it comes to ACPI >> drivers that we should be requiring the same: Document the binding, >> either in the kernel as a DT binding, or point to somewhere else that >> has the binding documented. >> >> Also, since this patch is targeted at v3.19 or later, the >> device-properties API should be used. Don't create something custom. >> > > My sentiments exactly, there would be little point having bespoke device > properties for every single device. Btw, we also need to align here with > Windows too ! > The Windows folks definitely know about _DSD (and helped define it), so this is a good opportunity to work through that process. Liam, do you have a good contact to start that discussion? -- Darren Hart Intel Open Source Technology Center