Re: [5/5] i2c: davinci: use ICPFUNC to toggle I2C as gpio for bus recovery

[Grygorii Strashko <grygorii.strashko-l0cyMroinI0@public.gmane.org>]

On 11/24/2014 08:13 PM, Mike Looijmans wrote:
> On 24-11-2014 14:15, Grygorii Strashko wrote:
>> Hi Uwe,
>> On 11/23/2014 07:04 PM, Uwe Kleine-König wrote:
>>> On Thu, Nov 20, 2014 at 12:03:08PM +0200, Grygorii Strashko wrote:
>>>> @@ -664,6 +759,7 @@ static int davinci_i2c_probe(struct 
>>>> platform_device *pdev)
>>>>            if (!of_property_read_u32(pdev->dev.of_node, 
>>>> "clock-frequency",
>>>>                &prop))
>>>>                dev->pdata->bus_freq = prop / 1000;
>>>> +        dev->pdata->has_pfunc = true;
>>> I don't understand this. Why does this ICPFUNC recovery work if the bus
>>> is probed by oftree, but doesn't if not?
>> I've mentioned this in commit message:
>>   Allow platforms to indicate the presence of the ICPFUNC registers 
>> with a has_pfunc
>>   platform data flag and enable this mode for platforms which supports 
>> DT (da850 and
>>   Keystone 2 are two SoCs which support DT now and they also support 
>> ICPFUNC registers).
>>
>> I'll add proper comment here.
> 
> Just thinking: What happens if you try to use the ICPFUNC registers on 
> platforms that don't support it? If the answer is "nothing bad", then 
> you might as well assume that if the platform doesn't specify its own 
> GPIOs, you can always try using the ICPFUNC registers to shake the I2C 
> bus. Better to try and fail than to never try at all...
> 

I think the right answer is !"nothing bad".

My intention was to enable this feature by default, because current DT-compatible
SoCs support it and the possibility that older SoCs will migrate to DT is low.
But now I think that the right way will be to add proper compatible strings
and use them to detect if ICPFUNC registers are supported or not.

[...]

regards,
-grygorii