On 12 November 2012 01:12, Rob Herring <robherri...@gmail.com> wrote:
> On 11/11/2012 11:27 AM, Viresh Kumar wrote:
>> On 11 November 2012 19:42, Rob Herring <robherri...@gmail.com> wrote:
>>> On 11/06/2012 10:22 PM, viresh kumar wrote:
>>
>>>> cluster0: cluster@0 {
>>>> + data1 = <0x50 0x60 0x70>;
>>>> + data2 = <0x5000 0x6000 0x7000>;
>>>> + data3 = <0x50000000 0x60000000 0x70000000>;
>>>
>>> So there is a mismatch in our assumptions. You are just truncating
>>> 32-bit values. I assumed you were using the 8 and 16 bit sizes that are
>>> now supported in dts. I don't think we should just truncate values
>>> blindly. We have support for specifying 8 and 16 values now so you
>>> should use that and define that as part of a binding.
>>
>> Sorry couldn't get your point at all :(
>> What did you mean by "truncating 32 bit values" and how should we
>> tell via DT, that the value passed is 8 bit, 16 bit or 32 bit?
>>
>
> You are trying to retrieve an array of 8 or 16-bit values which are
> stored as 32-bit values in dtb. Why not define them in the binding as 8
> or 16 bit to begin with. Then there is never any ambiguity about their size.
>
> I don't think the size is stored in the dtb. It is only in the dts. You
> need to define the size in the binding definitions and use '/bits/'
> annotation. With this the data is packed. Then the array function used
> should match what the binding defines.
Aha, and in that case incrementing address by 4 in my patch will fail. Right?
Will fix it. Thanks for increasing my knowledge on this :)
--
viresh
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
