From mboxrd@z Thu Jan 1 00:00:00 1970 From: Maxime Ripard Subject: Re: [PATCH] ARM: sun7i: dt: Add new MK808C device Date: Mon, 9 Mar 2015 15:50:44 +0100 Message-ID: <20150309145044.GS5085@lukather> References: <1425131004-8551-1-git-send-email-codekipper@gmail.com> <20150308174721.GL5085@lukather> <9483321.y7yrfWRaG5@wuerfel> Reply-To: linux-sunxi-/JYPxA39Uh5TLH3MbocFFw@public.gmane.org Mime-Version: 1.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="PkEWctFf+8E2rcii" Return-path: Content-Disposition: inline In-Reply-To: <9483321.y7yrfWRaG5@wuerfel> List-Post: , List-Help: , List-Archive: , List-Unsubscribe: , To: Arnd Bergmann Cc: Code Kipper , Rob Herring , Mark Rutland , linux-arm-kernel-IAPFreCvJWM7uuMidbF8XUB+6BGkLq7r@public.gmane.org, devicetree , linux-sunxi List-Id: devicetree@vger.kernel.org --PkEWctFf+8E2rcii Content-Type: text/plain; charset=UTF-8 Content-Disposition: inline On Sun, Mar 08, 2015 at 10:18:30PM +0100, Arnd Bergmann wrote: > On Sunday 08 March 2015 18:47:21 Maxime Ripard wrote: > > > > Not enough information to check signature validity. Show Details > > Hi, > > > > On Sat, Mar 07, 2015 at 09:41:54AM +0100, Code Kipper wrote: > > > > Don't your device has any brand on the case or the PCB? > > > > > > There is nothing on the PCB to reference a manufacturer, the actual packaging > > > and case are from the earlier MK808B and mention A9. There was only a > > > small sticker on the side to show that the contents was a MK808C. > > > I guess 'unknown' isn't a valid vendor name, > > > > Not really. > > > > I really don't know what's the policy to apply in such a case when we > > have a device that has no identified vendor. > > > > We need to give a vendor name, because the compatible is used to tell > > two devices apart, in the case where we would have to apply quirks for > > a particular devices. > > > > And if we have two devices (say a Marvlell and an Allwinner one), with > > the same name, and without any vendor, we're screwed. > > > > Maybe falling back to the SoC vendor, in our case, would make sense? > > > > Arnd? Rob? Mark? An opinion on this? > > > > I don't really have a good idea. In this case the SoC vendor might > not be the worst choice, because it's likely to be very close to > some reference design they actually did. > > Another possibility would be to use a string that refers to the > organization that does the upstream kernel support, but that's > harder for individuals that do not work for a company that wants > its name used that way. Yeah, I don't really think that would be applicable for individuals, and would understand if people wouldn't like to follow such policy (I'm not even sure I would). At least, the SoC vendor name would be consistent, without being very touchy. Maxime -- Maxime Ripard, Free Electrons Embedded Linux, Kernel and Android engineering http://free-electrons.com --PkEWctFf+8E2rcii--