[PATCH v4 3/3] ARM: dts: add I2C device nodes for Broadcom Cygnus

[Ray Jui]

On 1/16/2015 11:48 AM, Uwe Kleine-König wrote:
> Hello,
> 
> On Fri, Jan 16, 2015 at 11:24:09AM -0800, Ray Jui wrote:
>>>> +	i2c0: i2c at 18008000 {
>>>> +		compatible = "brcm,iproc-i2c";
>>> in patch 2 you wrote the driver is for a family of SoCs, right? Then I'd
>>> make this:
>>>
>>> 	compatible = "brcm,$mysoc-iproc-i2c", "brcm,iproc-i2c";
>>>
>> Sorry could you please help to explain the intention here? Note the
>> iProc I2C IP can be found in various iProc family of SoCs, but to my
>> best knowledge, there hasn't been any changes of the IP in any of those
>> SoCs.
> This is just for making the device tree stable in the future. Consider
> your gentle hardware engineers "fix" a small issue for the next
> generation iproc SoC "pony" that needs an incompatible software change.
> 
> Then you can fix the driver without updating the device trees by
> switching to the SoC specific compatible string for "pony". And in case
> the hardware engineers didn't tell you that there is a change and the
> need for the software change is only detected when the machines are
> already shipped, you're happy if you can fix your kernel without needing
> to change the bootloader that provides the dtb.
> So start already today to add the (for now unused) compatible string.
> 
>> Is the compatible ID "brcm,$mysoc-iproc-i2c" only to clarify that it's
>> for a specific SoC? If so, what should the compatible ID array look
>> like? Should it be changed to the following?
>>
>> static const struct of_device_id bcm_iproc_i2c_of_match[] = {
>> 	{ .compatible = "brcm,iproc-i2c" },
>> 	{ .compatible = "brcm,$mysoc-iproc-i2c" },
>> 	{},
>> };
> No, there is no need, see above.
> 
> If something is still unclear, don't hesitate to ask.
> 
> Best regards
> Uwe
> 
Okay got it. Thanks for the explanation!