[PATCH V4 3/4] ARM: bcm2835: add thermal node to device-tree of bcm283x

[Stefan Wahren]

> Martin Sperl <kernel at martin.sperl.org> hat am 9. September 2016 um 16:58
> geschrieben:
> 
> 
> 
> > On 09.09.2016, at 16:25, Stefan Wahren <stefan.wahren at i2se.com> wrote:
> > 
> >> Am 09.09.2016 um 09:49 schrieb kernel at martin.sperl.org:
> >> From: Martin Sperl <kernel at martin.sperl.org>
> >> 
> >> Add the node for the thermal sensor of the bcm2835-soc
> >> to the device tree.
> >> 
> >> Signed-off-by: Martin Sperl <kernel at martin.sperl.org>
> >> Reviewed-by: Eric Anholt <eric at anholt.net>
> >> ---
> >> arch/arm/boot/dts/bcm283x.dtsi | 6 ++++++
> >> 1 file changed, 6 insertions(+)
> >> 
> >> diff --git a/arch/arm/boot/dts/bcm283x.dtsi
> >> b/arch/arm/boot/dts/bcm283x.dtsi
> >> index b982522..e2e3a46 100644
> >> --- a/arch/arm/boot/dts/bcm283x.dtsi
> >> +++ b/arch/arm/boot/dts/bcm283x.dtsi
> >> @@ -186,6 +186,12 @@
> >>            interrupts = <2 14>; /* pwa1 */
> >>        };
> >> 
> >> +        thermal: thermal at 0x7e212000 {
> >> +            compatible = "brcm,bcm2835-thermal";
> >> +            reg = <0x7e212000 0x8>;
> >> +            clocks = <&clocks BCM2835_CLOCK_TSENS>;
> >> +        };
> >> +
> > 
> > Since the driver handles 3 different SoC (2835, 2836, 2837). This node
> > should be defined in the SoC specific dtsi files, because the BCM2836
> > includes bcm283x.dtsi too.
> > 
> > Be aware the patch for bcm2837 must go to ARM64.
> 
> I can not really follow:
> * the node is defined in the dtsi included by all 3 soc,
>    and it is available on all so it sits where for example
>    spi0 or uart0 is located
> * as for arm64: this describes the registers that are
>    identical for arm and arm64 and the bcm2837.dtsi
>    is also including ../../../../arm/boot/dts/bcm283x.dtsi
> 
> So what is the problem?

The thermal driver specifies 3 different compatibles with partially different
settings.
If this patch is applied, all SoC (bcm2835, bcm2836, bcm2837) would use the
bcm2835 settings.
I can't believe this is intended.

Why does the binding contains 3 different compatibles if only one is used?

> 
> Martin
> 
> P.s the patches apply cleanly against master.
>