On Tue, May 28, 2024 at 10:27:34PM +0300, Andy Shevchenko wrote:
> Tue, May 28, 2024 at 10:03:12PM +0300, Laurent Pinchart kirjoitti:
> > From: Haibo Chen <haibo.chen@xxxxxxx>
> >
> > The ADP5585 is a 10/11 input/output port expander with a built in keypad
> > matrix decoder, programmable logic, reset generator, and PWM generator.
> > This driver supports the chip by modelling it as an MFD device, with two
> > child devices for the GPIO and PWM functions.
> >
> > The driver is derived from an initial implementation from NXP, available
> > in commit 8059835bee19 ("MLK-25917-1 mfd: adp5585: add ADI adp5585 core
> > support") in their BSP kernel tree. It has been extensively rewritten.
>
> ...
>
> > + tristate "Analog Devices ADP5585 MFD driver"
> > + select MFD_CORE
> > + select REGMAP_I2C
> > + depends on I2C && OF
>
> Why OF?
Because the driver works on OF systems only.
> No COMPILE_TEST?
The driver won't compile without CONFIG_I2C, so I can use
depends on I2C
depends on OF || COMPILE_TEST
Do you think that's better ?
>
> ...
>
> + array_size.h
> + device.h // e.g., devm_kzalloc()
>
> > +#include <linux/module.h>
> > +#include <linux/moduleparam.h>
> > +#include <linux/init.h>
> > +#include <linux/slab.h>
I'll drop those 3 headers, there's not needed anymore.
> > +#include <linux/i2c.h>
>
> > +#include <linux/of.h>
> > +#include <linux/of_device.h>
>
> You don't need them, instead of proxying...
of.h for of_device_get_match_data() and of_match_ptr(). I'll drop the
former, but I need the latter, so I'll keep of.h
of_device.h for historical reasons probably, I'll drop it.
> > +#include <linux/mfd/core.h>
> > +#include <linux/mfd/adp5585.h>
>
> m is earlier than 'o', but with above drop no more issue :-)
>
> ...just include mod_devicetable.h.
>
> > +#include <linux/regmap.h>
>
> + types.h // e.g., u8
>
> ...
>
> > + regmap_config = of_device_get_match_data(&i2c->dev);
>
> We have i2c_get_match_data().
Sounds good.
> ...
>
> > +#ifndef __LINUX_MFD_ADP5585_H_
> > +#define __LINUX_MFD_ADP5585_H_
> > +
> > +#include <linux/bits.h>
>
> ...
>
> > +#define ADP5585_MAN_ID(v) (((v) & 0xf0) >> 4)
>
> GENMASK()
This is not a mask. Or do you mean
(((v) & GENMASK(7, 4)) >> 4)
? I think that's overkill.
> ...
>
> > +#define ADP5585_Rx_PULL_CFG_MASK (3)
>
> GENMASK()
Not here, as this value is meant to be passed to ADP5585_Rx_PULL_CFG().
> Why parentheses in all of them, btw?
Probably for consistency, but I don't mind dropping them.
> ...
>
> > +#define ADP5585_C4_EXTEND_CFG_MASK (1U << 6)
>
> > +#define ADP5585_R4_EXTEND_CFG_MASK (1U << 5)
>
> > +#define ADP5585_R3_EXTEND_CFG_MASK (3U << 2)
>
> > +#define ADP5585_R0_EXTEND_CFG_MASK (1U << 0)
>
> > +#define ADP5585_OSC_FREQ_MASK (3U << 5)
>
> BIT() / GENMASK()
I'll use GENMASK for the masks.
> > +#endif
--
Regards,
Laurent Pinchart
