Re: [PATCH 1/3] dt: Add DT bindings documentation for SUNXI Security System

[Tomasz Figa]

On 24.05.2014 21:43, Marek Vasut wrote:
> On Saturday, May 24, 2014 at 09:20:03 PM, Tomasz Figa wrote:
>> Hi Marek,
>>
>> On 24.05.2014 13:21, Marek Vasut wrote:
>>> On Thursday, May 22, 2014 at 05:09:54 PM, LABBE Corentin wrote:
>>>
>>> Missing commit message. Please fix this and send a V2.
>>>
>>>> Signed-off-by: LABBE Corentin <clabbe.montjoie@xxxxxxxxx>
>>>> ---
>>>>
>>>>  Documentation/devicetree/bindings/crypto/sunxi-ss.txt | 9 +++++++++
>>>>  1 file changed, 9 insertions(+)
>>>>  create mode 100644
>>>>  Documentation/devicetree/bindings/crypto/sunxi-ss.txt
>>>>
>>>> diff --git a/Documentation/devicetree/bindings/crypto/sunxi-ss.txt
>>>> b/Documentation/devicetree/bindings/crypto/sunxi-ss.txt new file mode
>>>> 100644
>>>> index 0000000..356563b
>>>> --- /dev/null
>>>> +++ b/Documentation/devicetree/bindings/crypto/sunxi-ss.txt
>>>> @@ -0,0 +1,9 @@
>>>> +* Allwinner Security System found on A20 SoC
>>>> +
>>>> +Required properties:
>>>> +- compatible : Should be "allwinner,sun7i-a20-crypto".
>>>
>>> Why sun7i-a20 ? Is the crypto unit different in other sunxi chips ? Can
>>> that not be described by DT props ?
>>
>> A widely used convention is to define compatible strings after first
>> SoCs on which particular IP blocks appear. It is quite common among IP
>> blocks for which there is no well defined versioning scheme.
> 
> Well yeah, that's fine. But in this case, "sun7i" is the entire group of CPUs 
> manufactured by AW. I find that information redundant, the "allwinner,a20-
> crypto" would suffice. But I wonder if that IP block might have appeared even 
> earlier ? Or if it is CPU family specific, thus "allwinner,sun7i-crypto" would 
> be a better string ?

I'm not aware of Allwinner naming schemes too much, so please correct me
if I'm wrong, but if A20 implies sun7i, then "allwinner,a20-crypto"
would be better indeed.

Whether it was really the first SoC is another thing. Obviously this
needs to be checked, although it isn't really that important. For this
particular naming scheme you need to specify all the SoCs for which
given compatible string can be used for this IP anyway, because there is
usually no other source of information about this available (except
directly comparing two datasheets...).

Best regards,
Tomasz
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/