[RFC PATCH v1] EP93XX: Add ADC support

[Christian Gagneraud]

H Hartley Sweeten wrote:
> On Monday, October 12, 2009 7:24 AM, Christian Gagneraud wrote:
>> This patch add support for the ADC found in the EP93XX.
>>
>> This work is based on S3C platform ADC, apart from hardware related
>> stuff, the following modifications have been done:
>> - Remove touchscreen support:
>>   On S3C the TS is a "normal" ADC client, but it has priority over all
>>   other clients. On EP93XX the TS controller is built-in and offer
>>   advanced features.
>> - Remove convert and select callbacks:
>>   This was done for the shake of simplicity, it can be added easily.
>> - Channel definition:
>>   On S3c, channel is just an index (unsigned char). On EP93xx channel
>>   is the analog switch configuration (unsigned long), it gives the
>>   client full freedom on how to make the analog conversion (including
>>   routing VRef+ and VRef-, activationg PU/PD resistors, connecting
>>   pins to VDD/GND, ...)
>>
>>
>> This is a first draft. Comments and criticism welcome.
>>
>> Signed-off-by: Christian Gagneraud <cgagneraud at techworks.ie>
>> ---
> 
> [snip]
> 
>> diff --git a/arch/arm/mach-ep93xx/adc.c b/arch/arm/mach-ep93xx/adc.c
>> new file mode 100644
>> index 0000000..78dc074
>> --- /dev/null
>> +++ b/arch/arm/mach-ep93xx/adc.c
> 
> [snip]
> 
>> +	adc->irq = platform_get_irq(pdev, 0);
>> +	if (adc->irq <= 0) {
>> +		dev_err(dev, "failed to get adc irq\n");
>> +		ret = -ENOENT;
>> +		goto err_alloc;
>> +	}
>> +
>> +	ret = request_irq(adc->irq, ep93xx_adc_irq, IRQF_DISABLED,
>> +			  dev_name(dev), adc);
>> +	if (ret < 0) {
>> +		dev_err(dev, "failed to attach adc irq\n");
>> +		goto err_alloc;
>> +	}
> 
> The interrupt should be the last thing you request.  If an interrupt
> occurs before the device is completely ready will crash the kernel.

OK, will do. Actually now that you pointed me this out, there's even 
more things to do to ensure everything is initialized correctly, same 
  for ep93xx_adc_remove()

> 
> Most of these dev_err messages are not needed.  The upper layers will
> print the error code.  You can keep them around for early debugging
> but they should be removed before the driver is finally merged.

OK, will do.

> 
> Regards,
> Hartley

Thanks,
Chris