From mboxrd@z Thu Jan 1 00:00:00 1970 From: Daniel Baluta Subject: Re: [PATCH 2/2] arm64: dts: imx8qxp: Add EDMA0/EDMA1 nodes Date: Wed, 27 Mar 2019 18:36:30 +0200 Message-ID: References: <20190326094239.5910-1-daniel.baluta@nxp.com> <20190326094239.5910-3-daniel.baluta@nxp.com> <1553679236.2561.27.camel@pengutronix.de> Mime-Version: 1.0 Content-Type: text/plain; charset="UTF-8" Return-path: In-Reply-To: <1553679236.2561.27.camel@pengutronix.de> Sender: linux-kernel-owner@vger.kernel.org To: Lucas Stach Cc: Aisheng Dong , "mark.rutland@arm.com" , "devicetree@vger.kernel.org" , Peng Fan , "festevam@gmail.com" , Anson Huang , Teo Hall , Daniel Baluta , "s.hauer@pengutronix.de" , "linux-kernel@vger.kernel.org" , "robh+dt@kernel.org" , dl-linux-imx , "kernel@pengutronix.de" , "shawnguo@kernel.org" , "S.j. Wang" , "linux-arm-kernel@lists.infradead.org" List-Id: devicetree@vger.kernel.org On Wed, Mar 27, 2019 at 11:33 AM Lucas Stach wrote: > > Hi Daniel, > > Am Mittwoch, den 27.03.2019, 10:51 +0200 schrieb Daniel Baluta: > [...] > > > > > or > > > "fsl,imx8qxp-edma", "fsl,imx8qm-edma"? > > > > One thing that it is not clear for me is why there are places > > where we use two compatible strings? > > > > I understand the situation where are two distinct drivers, but is there > > any other reason to add multiple compatible strings for a node in dts? > > We use 2 compatible string where there should not be any differences > between the IP blocks of this SoC and a version the driver already > supports. > > So if the eDMA driver already supports the software interface for > "fsl,imx8qm-edma" and the IP block is compatible with this, we add this > to the DT, so the we don't need any driver changes just to support a > new SoC. But as you can never be sure if there are subtle differences > in the IP block and/or SOC integration when adding the DT support, we > also add a more specific compatible to the DT. If it turns out that > there are software visible differences, we only need to adapt the > driver to check for the more specific compatible to trigger the changed > behavior, allowing to keep the DT stable. Excellent explanation Lucas! For the moment upstream eDMA driver doesn't have support for any i.MX8 so I think using "fsl,imx8qxp-edma" should be acceptable. Thanks, Daniel.