Re: [PATCH 2/2] ARM: shmobile: r8a7794: alt: Enable ethernet controller

[Sergei Shtylyov <sergei.shtylyov@cogentembedded.com>]

Hello.

On 2/5/2015 3:17 PM, Laurent Pinchart wrote:

>>>>>> Enable the ethernet controller for the Alt board. Pin muxing entries
>>>>>> are currently left out as r8a7794 pin control support isn't available
>>>>>> yet. We thus rely on the boot loader to configure ethernet pins for
>>>>>> now.

>>>>>> Signed-off-by: Laurent Pinchart
>>>>>> <laurent.pinchart+renesas@ideasonboard.com>
>>>>>> ---

>>>>>>     arch/arm/boot/dts/r8a7794-alt.dts | 13 +++++++++++++
>>>>>>     1 file changed, 13 insertions(+)

>>>>>> diff --git a/arch/arm/boot/dts/r8a7794-alt.dts
>>>>>> b/arch/arm/boot/dts/r8a7794-alt.dts index 0d848e605071..25bf434433b1
>>>>>> 100644
>>>>>> --- a/arch/arm/boot/dts/r8a7794-alt.dts
>>>>>> +++ b/arch/arm/boot/dts/r8a7794-alt.dts
>>>>>> @@ -43,6 +43,19 @@
>>>>>>         status = "okay";
>>>>>>     };
>>>>>>
>>>>>> +&ether {
>>>>>> +    phy-handle = <&phy1>;
>>>>>> +    renesas,ether-link-active-low;
>>>>>> +    status = "okay";
>>>>>> +
>>>>>> +    phy1: ethernet-phy@1 {
>>>>>> +        reg = <1>;
>>>>>> +        interrupt-parent = <&irqc0>;
>>>>>> +        interrupts = <0 IRQ_TYPE_LEVEL_LOW>;

>>>>> The hardware manual tells me the PHY uses IRQ8#, not IRQ0#.

>>>> I don't have the complete Alt schematics.

>>> I don't have *any* Alt schematics, only SILK.

>> And SILK also has the PHY IRQ connected to IRQ8#.

>>>> I've double-checked the block
>>>> diagram and it mentions "GP1_24(IRQ)"

>>> GP1_24 is connected to the PHY's RST# pin, according to the hardware
>>> manual. Which seems bad to me -- we hardly needed such kind of reset with
>>> the device tree. :-/ Hopefully, U-Boot leaves it high...

>>>> next to the PHY. This patch is thus
>>>> incorrect, but where did you find a mention of IRQ8 ?

>>> In the Alt hardware manual, rev. 0.04. :-)

>> Mentioned in several places there.

>> Probably it makes sense to compare the dates of our documents. The
>> aforementioned manual is dated June 4, 2014.

> The block diagram I have mentions "Rev0.04 Nov 22, 2013", as well as
> "PRELIMINARY". Let's use IRQ8 then. IRQ0 is clearly wrong anyway, and we can
> always fix this later if the hardware manual is wrong.

    Note that the PHY IRQ can be verified by looking at /proc/interrupts. If 
you e.g. use NFS root with the Ethernet cable already inserted, there should 
be exactly one interrupt registered, corresponding to the link-up event. I 
have verified IRQ8# on SILK this way.

WBR, Sergei