On Thu, Oct 27, 2016 at 10:03:14AM +0900, Eric Jeong wrote:
> config REGULATOR_PV88080
> - tristate "Powerventure Semiconductor PV88080 regulator"
> - depends on I2C
> - select REGMAP_I2C
> + bool "Powerventure Semiconductor PV88080 regulator"
> + depends on MFD_PV88080
Forcing the driver to be built in looks like a regression, why would we
want to do that?
> + irq = platform_get_irq_byname(pdev, "regulator-irq");
> + if (irq < 0) {
> + dev_err(&pdev->dev, "Failed to get IRQ.\n");
> + return irq;
> }
What's the _byname() adding here given that the name is so generic? It
feels like if the name ever becomes important then this particular name
is going to be a problem.
> -module_i2c_driver(pv88080_regulator_driver);
> +static int __init pv88080_regulator_init(void)
> +{
> + return platform_driver_register(&pv88080_regulator_driver);
> +}
> +subsys_initcall(pv88080_regulator_init);
Why are you converting this to subsys_initcall()? This looks like
another regression.
Attachment:
signature.asc
Description: PGP signature
