Re: [PATCH] powerpc/p1010rdb:remove interrupts of ethernet-phy in device tree

[Kumar Gala <galak@kernel.crashing.org>]

On Sep 12, 2013, at 1:54 AM, Liu Shengzhou-B36685 wrote:

>=20
>=20
>> -----Original Message-----
>> From: Kumar Gala [mailto:galak@kernel.crashing.org]
>> Sent: Wednesday, September 11, 2013 11:13 PM
>> To: Zhao Qiang-B45475
>> Cc: linuxppc-dev@lists.ozlabs.org; Liu Shengzhou-B36685
>> Subject: Re: [PATCH] powerpc/p1010rdb:remove interrupts of =
ethernet-phy in
>> device tree
>>=20
>>=20
>> On Sep 10, 2013, at 10:49 PM, Zhao Qiang wrote:
>>=20
>>> Since P1010RDB-PA and P1010RDB-PB boards use different external PHY
>>> interrupt signals.
>>> And actually the PHY interrupt is not used effectively with
>>> corresponding interrupt handler.
>>> So we can remove the interrupts node without side-effect to comply
>>> with both P1010RDB-PA and P1010RDB-PB.
>>>=20
>>> Signed-off-by: Shengzhou Liu <Shengzhou.Liu@freescale.com>
>>> Signed-off-by: Zhao Qiang <B45475@freescale.com>
>>> ---
>>> arch/powerpc/boot/dts/p1010rdb.dtsi | 3 ---
>>> 1 file changed, 3 deletions(-)
>>>=20
>>=20
>> NAK.  The device tree should represent the HW not what drivers decide =
to do with
>> it.
>>=20
>> If different board revs have different interrupt signals than create =
dts's to
>> handle the 2 board revs.
>>=20
>> - k
>>=20
> You mean we need to create p1010rdb-pa.dtsi and p1010rdb-pb.dtsi =
replacing current p1010rdb.dtsi just because of the unused phy =
interrupt?
> and phy interrupt is not present in those dts of P3/P4/P5 platforms.
> Actually currently many hardware are not present in dts, such as a lot =
of i2c devices, temperature monitor, etc.
>=20
> -Shengzhou
>=20

I'm saying of the board revs are different w/regards to how the PHY =
interrupt is wired, than create two .dts one for each of the board revs.

If the p3/p4/p5 platforms are missing the phy interrupt in the .dts than =
its an error.

Other devices like i2c, temp mon, etc should be added.  There is a =
difference between something not existing because people haven't gotten =
around to it / there isn't a binding vs a using the lack of information =
as a configuration mechanism.

- k