Re: [PATCH v2 3/4] gpio: adp5585: Add Analog Devices ADP5585 support

[Laurent Pinchart <laurent.pinchart@xxxxxxxxxxxxxxxx>]

On Tue, May 28, 2024 at 11:20:45PM +0300, Laurent Pinchart wrote:
> Hi Andy,
> 
> Thank you for the patch.
> 
> On Tue, May 28, 2024 at 10:36:06PM +0300, Andy Shevchenko wrote:
> > Tue, May 28, 2024 at 10:03:13PM +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 GPIO function using the platform device
> > > registered by the core MFD driver.
> > > 
> > > The driver is derived from an initial implementation from NXP, available
> > > in commit 451f61b46b76 ("MLK-25917-2 gpio: adp5585-gpio: add
> > > adp5585-gpio support") in their BSP kernel tree. It has been extensively
> > > rewritten.
> > 
> > Why is this not using gpio-regmap?

I forgot to answer this:

I don't think it's a good match for the hardware.

> > ...
> > 
> > > +#include <linux/device.h>
> > > +#include <linux/gpio/driver.h>
> > > +#include <linux/mfd/adp5585.h>
> > > +#include <linux/module.h>
> > > +#include <linux/platform_device.h>
> > > +#include <linux/regmap.h>
> > 
> > + types.h
> > 
> > ...
> > 
> > > +	bit = off * 2 + (off > 5 ? 4 : 0);
> > 
> > Right, but can you use >= 6 here which immediately follows to the next
> > question, i.e. why not use bank in this conditional?
> 
> The ADP5585_BANK() macro is meant to be used with ADP5585_BIT(), for a
> set of registers with the same layout. Here the layout is different, the
> registers contain multi-bit fields. I can't use ADP5585_BIT(), so I'd
> rather not use ADP5585_BANK() either. I have decided to use > 5 instead
> of >= 6 to match the R5 field name in the comment above:
> 
>         /*
>          * The bias configuration fields are 2 bits wide and laid down in
>          * consecutive registers ADP5585_RPULL_CONFIG_*, with a hole of 4 bits
>          * after R5.
>          */
> 
> > ...
> > 
> > > +	struct adp5585_dev *adp5585 = dev_get_drvdata(pdev->dev.parent);
> > 
> > (see below)
> > 
> > > +	struct adp5585_gpio_dev *adp5585_gpio;
> > > +	struct device *dev = &pdev->dev;
> > 
> > 	struct adp5585_dev *adp5585 = dev_get_drvdata(dev->parent);
> 
> I prefer keeping the current ordering, with long lines first, I think
> that's more readable.
> 
> > > +	struct gpio_chip *gc;
> > > +	int ret;
> > 
> > ...
> > 
> > > +	platform_set_drvdata(pdev, adp5585_gpio);
> > 
> > Any use of driver data?
> 
> In v1, not v2. I'll drop it.
> 
> > ...
> > 
> > > +	device_set_of_node_from_dev(dev, dev->parent);
> > 
> > Why not device_set_node()?
> 
> Because device_set_of_node_from_dev() is meant for this exact use case,
> where the same node is used for multiple devices. It also puts any
> previous dev->of_node, ensuring proper refcounting when devices are
> unbound and rebound, without being deleted.
> 
> -- 
> Regards,
> 
> Laurent Pinchart

-- 
Regards,

Laurent Pinchart