Re: [PATCH RESEND v2 2/4] mfd: intel_soc_pmic: I2C interface

["Zhu, Lejun" <lejun.zhu@linux.intel.com>]

On 5/24/2014 1:53 AM, Mark Brown wrote:
> On Fri, May 23, 2014 at 08:40:27AM +0800, Zhu, Lejun wrote:
> 
>> +static int pmic_i2c_lookup_gpio(struct device *dev, int acpi_index)
>> +{
>> +	struct gpio_desc *desc;
>> +	int gpio;
>> +
>> +	desc = gpiod_get_index(dev, KBUILD_MODNAME, acpi_index);
>> +	if (IS_ERR(desc))
>> +		return PTR_ERR(desc);
>> +
>> +	gpio = desc_to_gpio(desc);
>> +
>> +	gpiod_put(desc);
>> +
>> +	return gpio;
>> +}
> 
> Why not just have the driver work with the gpiod API, is there any real
> need to convert to a GPIO number?

The whole function will be removed since the gpio set code is removed
during probe.

>> +static const struct i2c_device_id pmic_i2c_id[] = {
>> +	{ "crystal_cove", (kernel_ulong_t)&crystal_cove_pmic},
>> +	{ "INT33FD", (kernel_ulong_t)&crystal_cove_pmic},
>> +	{ "INT33FD:00", (kernel_ulong_t)&crystal_cove_pmic},
>> +	{ }
>> +};
>> +MODULE_DEVICE_TABLE(i2c, pmic_i2c_id);
> 
> The INT33FD ones here look like they should only be in the ACPI table.

You are right. Only INT33FD:00 should be here.

>> +static int __init pmic_i2c_init(void)
>> +{
>> +	int ret;
>> +
>> +	ret = i2c_add_driver(&pmic_i2c_driver);
>> +	if (ret != 0)
>> +		pr_err("Failed to register pmic I2C driver: %d\n", ret);
>> +
>> +	return ret;
>> +}
>> +subsys_initcall(pmic_i2c_init);
> 
> module_i2c_driver() - you shouldn't need subsys_initcall().

I'll change it to module_init.

Best Regards
Lejun