[linux-sunxi] Re: [PATCH v3 01/10] mfd: AXP20x: Add mfd driver for AXP20x PMIC

[Carlo Caione]

On Fri, Mar 28, 2014 at 09:21:54AM +0000, Lee Jones wrote:
> > This patch introduces the preliminary support for PMICs X-Powers AXP202
> > and AXP209. The AXP209 and AXP202 are the PMUs (Power Management Unit)
> > used by A10, A13 and A20 SoCs and developed by X-Powers, a sister company
> > of Allwinner.
> > 
> > The core enables support for two subsystems:
> > - PEK (Power Enable Key)
> > - Regulators
> > 
> > Signed-off-by: Carlo Caione <carlo at caione.org>
> > ---
> >  drivers/mfd/Kconfig        |  12 +++
> >  drivers/mfd/Makefile       |   1 +
> >  drivers/mfd/axp20x.c       | 240 +++++++++++++++++++++++++++++++++++++++++++++
> >  include/linux/mfd/axp20x.h | 180 ++++++++++++++++++++++++++++++++++
> >  4 files changed, 433 insertions(+)
> >  create mode 100644 drivers/mfd/axp20x.c
> >  create mode 100644 include/linux/mfd/axp20x.h
> > 
> > diff --git a/drivers/mfd/Kconfig b/drivers/mfd/Kconfig
> > index 49bb445..24ba61a 100644
> > --- a/drivers/mfd/Kconfig
> > +++ b/drivers/mfd/Kconfig
> > @@ -59,6 +59,18 @@ config MFD_AAT2870_CORE
> >  	  additional drivers must be enabled in order to use the
> >  	  functionality of the device.
> >  
> > +config MFD_AXP20X
> > +	bool "X-Powers AXP20X"
> > +	select MFD_CORE
> > +	select REGMAP_I2C
> > +	select REGMAP_IRQ
> > +	depends on I2C=y
> > +	help
> > +	  If you say Y here you get support for the AXP20X.
> > +	  This driver provides common support for accessing the device,
> > +	  additional drivers must be enabled in order to use the
> > +	  functionality of the device.
> 
> Please tell us what this device is and what sub-devices are available?

Ok

> [...]
> 
> > diff --git a/drivers/mfd/axp20x.c b/drivers/mfd/axp20x.c
> > new file mode 100644
> > index 0000000..f77a663
> > --- /dev/null
> > +++ b/drivers/mfd/axp20x.c
> > @@ -0,0 +1,240 @@
> > +/*
> > + * axp20x.c - MFD core driver for the X-Powers AXP202 and AXP209
> 
> ... which consist of ...

Yeah, I should improve my code documentation :)

> [...]
> 
> > +static struct resource axp20x_pek_resources[] = {
> > +	{
> > +		.name	= "PEK_DBR",
> > +		.start	= AXP20X_IRQ_PEK_RIS_EDGE,
> > +		.end	= AXP20X_IRQ_PEK_RIS_EDGE,
> > +		.flags	= IORESOURCE_IRQ,
> > +	}, {
> > +		.name	= "PEK_DBF",
> > +		.start	= AXP20X_IRQ_PEK_FAL_EDGE,
> > +		.end	= AXP20X_IRQ_PEK_FAL_EDGE,
> > +		.flags	= IORESOURCE_IRQ,
> > +	},
> > +};
> 
> Have you considered doing this in the Device Tree? It's a lot less
> code/overhead.

Ok, I'll change it in v4.

> > +static const struct i2c_device_id axp20x_i2c_id[] = {
> > +	{ },
> > +};
> > +MODULE_DEVICE_TABLE(i2c, axp20x_i2c_id);
>  
> We really should consider changing the I2C subsystem!
> 
> Can you add a comment here describing why we have to add this
> seemingly pointless empty struct please?

Sure.

> > +static struct mfd_cell axp20x_cells[] = {
> > +	{
> > +		.name		= "axp20x-pek",
> > +		.num_resources	= ARRAY_SIZE(axp20x_pek_resources),
> > +		.resources	= axp20x_pek_resources,
> > +	}, {
> > +		.name		= "axp20x-regulator",
> > +	},
> > +};
> 
> Do these drivers don't look inside the DTB at all?

Only the regulators driver.

> > +	of_id = of_match_device(axp20x_of_match, &i2c->dev);
> > +	if (!of_id) {
> > +		dev_err(&i2c->dev, "Unable to setup AXP20X data\n");
> > +		return -ENODEV;
> > +	}
> > +	axp20x->variant = (int) of_id->data;
> 
> 'variant' needs to be a (unsigned?) long or it will break on 64bit
> architectures.

Ok, I really need to start thinking to 64bit also.

Thank you,

-- 
Carlo Caione