Re: [PATCH v4 06/15] mfd: axp20x: Add support for AXP192

[Aidan MacDonald <aidanmacdonald.0x0@xxxxxxxxx>]

Lee Jones <lee.jones@xxxxxxxxxx> writes:

> On Tue, 05 Jul 2022, Aidan MacDonald wrote:
>
>> 
>> Lee Jones <lee.jones@xxxxxxxxxx> writes:
>> 
>> > On Wed, 29 Jun 2022, Aidan MacDonald wrote:
>> >
>> >> The AXP192 PMIC is similar to the AXP202/AXP209, but with different
>> >> regulators, additional GPIOs, and a different IRQ register layout.
>> >> 
>> >> Acked-for-MFD-by: Lee Jones <lee.jones@xxxxxxxxxx>
>> >> Signed-off-by: Aidan MacDonald <aidanmacdonald.0x0@xxxxxxxxx>
>> >> ---
>> >>  drivers/mfd/axp20x-i2c.c   |   2 +
>> >>  drivers/mfd/axp20x.c       | 152 +++++++++++++++++++++++++++++++++++++
>> >>  include/linux/mfd/axp20x.h |  84 ++++++++++++++++++++
>> >>  3 files changed, 238 insertions(+)
>> >> [...]
>> >> +static const struct regmap_irq_chip axp192_regmap_irq_chip = {
>> >> +	.name			= "axp192_irq_chip",
>> >> +	.status_base		= AXP192_IRQ1_STATE,
>> >> +	.ack_base		= AXP192_IRQ1_STATE,
>> >> +	.unmask_base		= AXP192_IRQ1_EN,
>> >> +	.init_ack_masked	= true,
>> >> +	.irqs			= axp192_regmap_irqs,
>> >> +	.num_irqs		= ARRAY_SIZE(axp192_regmap_irqs),
>> >> +	.num_regs		= 5,
>> >> +	.get_irq_reg		= axp192_get_irq_reg,
>> >
>> > Fix me:
>> >
>> > drivers/mfd/axp20x.c:634:35: error: initialization of 'unsigned int (*)(struct regmap_irq_chip_data *, unsigned int,  int)' from incompatible pointer type 'int (*)(unsigned int,  int)' [-Werror=incompatible-pointer-types]
>> >   634 |         .get_irq_reg            = axp192_get_irq_reg,
>> >       |                                   ^~~~~~~~~~~~~~~~~~
>> > drivers/mfd/axp20x.c:634:35: note: (near initialization for 'axp192_regmap_irq_chip.get_irq_reg')
>> >
>> 
>> This doesn't need fixing. You are missing the prerequisite patches.
>> 
>> -- snip --
>> Changes in v4:
>> * Drop regmap-irq patches and rebase on top of the regmap-irq
>>   refactoring series[1], which implements the same functionality.
>> 
>> [1]: https://lore.kernel.org/lkml/20220623211420.918875-1-aidanmacdonald.0x0@xxxxxxxxx/
>> -- snip --
>
>> Perhaps I wasn't clear enough that it was a dependency, sorry for
>> any confusion.
>
> Any idea why the regmap-irq patches weren't in -next?

Uh, sorry, I didn't pay attention and assumed this was the same error
the kernel test robot was reporting against mfd-next (missing field).
But the field does exist; this error is a type mismatch:

  initialization of 'unsigned int (*)(struct regmap_irq_chip_data *,
                                      unsigned int,  int)'
  from incompatible pointer type 'int (*)(unsigned int,  int)'

So the regmap-irq patches were in -next, the problem is you were
using the old patch from v3.

FYI, I just sent a v5 which drops the GPIO driver, so please use
the patches from that version. Thanks.

Regards,
Aidan