On Fri, Mar 31, 2017 at 04:10:30PM +0200, Neil Armstrong wrote: > On 03/31/2017 03:44 PM, Arnd Bergmann wrote: > > On Fri, Mar 31, 2017 at 10:47 AM, Neil Armstrong > > <narmstr...@baylibre.com> wrote: > >> Add bindings for the SoC information register of the Amlogic SoCs. > >> > >> Signed-off-by: Neil Armstrong <narmstr...@baylibre.com> > >> --- > >> Documentation/devicetree/bindings/arm/amlogic.txt | 20 > >> ++++++++++++++++++++ > >> 1 file changed, 20 insertions(+) > >> > >> diff --git a/Documentation/devicetree/bindings/arm/amlogic.txt > >> b/Documentation/devicetree/bindings/arm/amlogic.txt > >> index bfd5b55..b850985 100644 > >> --- a/Documentation/devicetree/bindings/arm/amlogic.txt > >> +++ b/Documentation/devicetree/bindings/arm/amlogic.txt > >> @@ -52,3 +52,23 @@ Board compatible values: > >> - "amlogic,q201" (Meson gxm s912) > >> - "nexbox,a95x" (Meson gxbb or Meson gxl s905x) > >> - "nexbox,a1" (Meson gxm s912) > >> + > >> +Amlogic Meson GX SoCs Information > >> +---------------------------------- > >> + > >> +The Meson SoCs have a Product Register that allows to retrieve SoC type, > >> +package and revision information. If present, a device node for this > >> register > >> +should be added. > >> + > >> +Required properties: > >> + - compatible: For Meson GX SoCs, must be "amlogic,meson-gx-socinfo". > >> + - reg: Base address and length of the register block. > >> + > >> +Examples > >> +-------- > >> + > >> + chipid@220 { > >> + compatible = "amlogic,meson-gx-socinfo"; > >> + reg = <0x0 0x00220 0x0 0x4>; > >> + }; > >> + > > > > The register location would hint that this is in the middle of some block of > > random registers, i.e. a syscon or some unrelated device. > > > > Are you sure that "socinfo" is the actual name of the IP block and that > > it only has a single 32-bit register? > > > > Arnd > > > > Hi Arnd, > > I'm sorry I did not find any relevant registers in the docs or source code > describing > it in a specific block of registers, and no close enough register definitions > either. > They may be used by the secure firmware I imagine. > > For the register name, Amlogic refers it to "cpu_version" in their code, but > it really > gives some details on the whole SoC and package, and socinfo seems better.
A register at address 0x220 seems a bit strange (unless there's ranges you're not showing), but ROM code at this address would be fairly typical. And putting version information into the ROM is also common. Rob