From mboxrd@z Thu Jan 1 00:00:00 1970 From: l.stach@pengutronix.de (Lucas Stach) Date: Wed, 18 Nov 2015 12:36:16 +0100 Subject: [PATCH 2/2] ARM: dts: imx6q: add Novena board In-Reply-To: <201511181232.42446.marex@denx.de> References: <1447840120-8513-1-git-send-email-marex@denx.de> <201511181135.40171.marex@denx.de> <1447845958.3144.61.camel@pengutronix.de> <201511181232.42446.marex@denx.de> Message-ID: <1447846576.3144.63.camel@pengutronix.de> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org Am Mittwoch, den 18.11.2015, 12:32 +0100 schrieb Marek Vasut: > On Wednesday, November 18, 2015 at 12:25:58 PM, Lucas Stach wrote: > > Am Mittwoch, den 18.11.2015, 11:35 +0100 schrieb Marek Vasut: > > > On Wednesday, November 18, 2015 at 11:10:12 AM, Lucas Stach wrote: > > > > Hi Marek, [...] > > > [...] > > > > > > > > +&ecspi3 { > > > > > + pinctrl-names = "default"; > > > > > + pinctrl-0 = <&pinctrl_ecspi3_novena>; > > > > > + fsl,spi-num-chipselects = <3>; > > > > > + status = "okay"; > > > > > + > > > > > + spidev at 0 { > > > > > + compatible = "spidev"; > > > > > > > > This will explode on a new kernel. Specifying spidev without using a > > > > more specific compatible will trigger a WARN_ON(), as it's considered > > > > bad style. > > > > > > Oh, looks like imx_v6_v7_defconfig didn't have SPIDEV active, which is > > > why I didn't catch it during my last test, dang. > > > > > > This SPI interface is routed into the FPGA, so what do you suggest I put > > > in the compatible string? There is no other sensible driver for the > > > peripheral, since the peripheral can be anything. > > > > I'm not really the right person to ask for that, but I would guess that > > inventing a "novena,fpga-spi" compatible or something like that and > > adding it to the spidev driver would be the right thing to do. > > So what's the rationale behind generic "spidev" being bad ? In my mind, it > is much better to use generic driver and generic compatible prop "spidev" > than to generate many one-off compatible props and have a driver with a list > of all of those. > >>From the commit log: "Since spidev is a detail of how Linux controls a device rather than a description of the hardware in the system we should never have a node described as "spidev" in DT, any SPI device could be a spidev so this is just not a useful description." If you want to know about the details you would have to discuss this with Mark Brown. Regards, Lucas -- Pengutronix e.K. | Lucas Stach | Industrial Linux Solutions | http://www.pengutronix.de/ |