On Mon, Apr 10, 2017 at 11:36 PM, Oleksij Rempel <o...@pengutronix.de> wrote: > Hi, > > > On 04/10/2017 08:22 PM, Rob Herring wrote: >> >> On Thu, Apr 06, 2017 at 09:31:07AM +0200, Oleksij Rempel wrote: >>> >>> Documenation bindings for the Low Power General Purpose Registe >> >> >> s/Registe/Register/ >> >>> available on i.MX6 SoCs in the Secure Non-Volatile Storage. >>> >>> Signed-off-by: Oleksij Rempel <o.rem...@pengutronix.de> >>> Cc: Srinivas Kandagatla <srinivas.kandaga...@linaro.org> >>> Cc: Maxime Ripard <maxime.rip...@free-electrons.com> >>> Cc: Rob Herring <robh...@kernel.org> >>> Cc: Mark Rutland <mark.rutl...@arm.com> >>> Cc: devicet...@vger.kernel.org >>> Cc: linux-kernel@vger.kernel.org >>> --- >>> Documentation/devicetree/bindings/nvmem/snvs-lpgpr.txt | 15 >>> +++++++++++++++ >>> 1 file changed, 15 insertions(+) >>> create mode 100644 >>> Documentation/devicetree/bindings/nvmem/snvs-lpgpr.txt >>> >>> diff --git a/Documentation/devicetree/bindings/nvmem/snvs-lpgpr.txt >>> b/Documentation/devicetree/bindings/nvmem/snvs-lpgpr.txt >>> new file mode 100644 >>> index 000000000000..9a8be1a2d12e >>> --- /dev/null >>> +++ b/Documentation/devicetree/bindings/nvmem/snvs-lpgpr.txt >>> @@ -0,0 +1,15 @@ >>> +Device tree bindings for Low Power General Purpose Registe found in >>> i.MX6Q/D >>> +Secure Non-Volatile Storage. >>> + >>> +Required properties: >>> +- compatible: should be one of >>> + "fsl,imx6q-snvs-lpgpr" (i.MX6Q/D/DL/S). >>> +- offset: Should contain the offset relative to syscon parrent node. >> >> >> typo > > > ok. > >> +- regmap: Should contain a phandle pointing to syscon. >>> >>> + >>> +Example: >>> + snvs_lpgpr: snvs-lpgpr { >>> + compatible = "fsl,imx6q-snvs-lpgpr"; >>> + regmap = <&snvs>; >>> + offset = <0x68>; >> >> >> Why does this need to be in DT? Is something going to refer to this >> node? If not, the &snvs node should be enough information for the OS. > > > Jes, it is refereed by other driver.
What I mean is snvs-lpgpr referenced elsewhere in DT, not by some driver? You are not using the nvmem binding here so it doesn't seem like it is. If that's the case, then you don't need this node. The only information here is the offset which can be part of a driver for the parent node. To put it another way, we don't want to fill DT with a node per register. Rob