Re: [PATCH v4 1/3] mfd: mc13xxx: Add mc34708 adc support

[Lee Jones <lee.jones@linaro.org>]

On Thu, 25 Jul 2019, Lukasz Majewski wrote:
> > On Thu, 18 Jul 2019, Lukasz Majewski wrote:
> > 
> > > From: Sascha Hauer <s.hauer@pengutronix.de>
> > > 
> > > The mc34708 has an improved adc. The older variants will always
> > > convert a fixed order of channels. The mc34708 can do up to eight
> > > conversions in arbitrary channel order. Currently this extended
> > > feature is not supported. We only support touchscreen conversions
> > > now, which will be sampled in a data format compatible to the older
> > > chips in order to keep the API between the mfd and the touchscreen
> > > driver.
> > > 
> > > Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
> > > Signed-off-by: Lukasz Majewski <lukma@denx.de>
> > > 
> > > ---
> > > Changes for v4:
> > > - None
> > > 
> > > Changes for v3:
> > > - None
> > > 
> > > Changes for v2:
> > > - Change the return code patch when the mc13xxx ADC is performing
> > > conversion
> > > - Introduce new include/linux/mfd/mc34708.h header file for mc34708
> > > specific defines
> > > 
> > > Changes from the original patches:
> > > - ADC conversion functions prototypes added to fix build error
> > > - Adjustments to make checkpatch clean (-ENOSYS, line over 80 char)
> > > 
> > > This patch applies on top of v5.2 - SHA1:
> > > 0ecfebd2b52404ae0c54a878c872bb93363ada36 ---
> > >  drivers/mfd/mc13xxx-core.c  | 102
> > > +++++++++++++++++++++++++++++++++++++++++++-
> > > drivers/mfd/mc13xxx.h       |   3 ++ include/linux/mfd/mc34708.h |
> > > 37 ++++++++++++++++ 3 files changed, 141 insertions(+), 1
> > > deletion(-) create mode 100644 include/linux/mfd/mc34708.h
> > > 
> > > diff --git a/drivers/mfd/mc13xxx-core.c b/drivers/mfd/mc13xxx-core.c
> > > index 1abe7432aad8..01473d6fda21 100644
> > > --- a/drivers/mfd/mc13xxx-core.c
> > > +++ b/drivers/mfd/mc13xxx-core.c
> > > @@ -12,6 +12,7 @@
> > >  #include <linux/of_device.h>
> > >  #include <linux/platform_device.h>
> > >  #include <linux/mfd/core.h>
> > > +#include <linux/mfd/mc34708.h>
> > >  
> > >  #include "mc13xxx.h"
> > >  
> > > @@ -45,6 +46,8 @@
> > >  
> > >  #define MC13XXX_ADC2		45
> > >  
> > > +#define MC13XXX_ADC_WORKING		(1 << 0)  
> > 
> > BIT(0) ?
> 
> The same convention - i.e. (1 << 0) is used in the rest of the file.

Very well, but please consider converting it subsequently.

> > >  void mc13xxx_lock(struct mc13xxx *mc13xxx)
> > >  {
> > >  	if (!mutex_trylock(&mc13xxx->lock)) {
> > > @@ -198,22 +201,30 @@ static void mc34708_print_revision(struct
> > > mc13xxx *mc13xxx, u32 revision) maskval(revision,
> > > MC34708_REVISION_FAB)); }
> > >  
> > > +static int mc13xxx_adc_conversion(struct mc13xxx *, unsigned int,
> > > +				  unsigned int, u8, bool, unsigned
> > > int *); +static int mc34708_adc_conversion(struct mc13xxx *,
> > > unsigned int,
> > > +				  unsigned int, u8, bool, unsigned
> > > int *); +
> > >  /* These are only exported for mc13xxx-i2c and mc13xxx-spi */
> > >  struct mc13xxx_variant mc13xxx_variant_mc13783 = {
> > >  	.name = "mc13783",
> > >  	.print_revision = mc13xxx_print_revision,
> > > +	.adc_do_conversion = mc13xxx_adc_conversion,
> > >  };
> > >  EXPORT_SYMBOL_GPL(mc13xxx_variant_mc13783);  
> > 
> > I'd prefer to keep the call-back functions as close to zero as
> > possible.
> 
> If I may ask - what is wrong with having per device callback(s) ?

Call-backs are ugly and hard to read/debug.

If they can be avoided, they should be, IMHO>

> > It would be better to turn mc13xxx_adc_conversion() in to the catch
> > function
> 
> Could you share any example? 

Just put the code doing the conversion into mc13xxx_adc_conversion()
and remove the call-back.

> > choose an execution route based on some platform matching.
> > 
> 
> Could you help me with giving a hint of how shall I do the "platform
> matching" in this particular driver ? 

We normally match on some platform ID, rather than passing around
pointers to structures containing pointers to device specific
functions.

> The mc13xxx driver seems rather complex with SPI and I2C support and in
> which the subdevices are added (e.g. rtc, adc, etc).

Not sure I follow your point?

> This particular patch just follows current driver design and fixes its
> usability for mc13708 drvice.

Right, but it is the current driver design that I'm trying to change.

> > If you could do the same for print_revision too, that would be even
> > better.
> 
> I would prefer to fix the driver (for mc13708) without the need to
> change the working code.

Okay, but no additional call-backs please.

-- 
Lee Jones [李琼斯]
Linaro Services Technical Lead
Linaro.org │ Open source software for ARM SoCs
Follow Linaro: Facebook | Twitter | Blog