Re: [PATCH] mux: mmio: select REGMAP as the code has that dependency

[Peter Rosin <peda@axentia.se>]

Hi!

[sorry about the delay, busy-busy]

2023-02-10 at 12:56, Philipp Zabel wrote:
> On Fri, Feb 10, 2023 at 11:46:42AM +0100, Peter Rosin wrote:
>> If CONFIG_REGMAP is not active, the build errors out like this:
> 
> The driver used to have an indirect dependency on CONFIG_REGMAP via
> CONFIG_MFD_SYSCON -> CONFIG_REGMAP_MMIO -> CONFIG_REGMAP until
> commit 8ecfaca7926f ("mux: mmio: add generic regmap bitfield-based
> multiplexer").
> 
>> ../drivers/mux/mmio.c: In function ‘mux_mmio_probe’:
>> ../drivers/mux/mmio.c:76:34: error: storage size of ‘field’ isn’t known
>>    76 |                 struct reg_field field;
>>       |                                  ^~~~~
>> In file included from ../include/linux/bits.h:21,
>>                  from ../include/linux/bitops.h:6,
>>                  from ../drivers/mux/mmio.c:8:
>> ../include/linux/bits.h:23:28: error: first argument to ‘__builtin_choose_expr’ not a constant
>>    23 |         (BUILD_BUG_ON_ZERO(__builtin_choose_expr( \
>>       |                            ^~~~~~~~~~~~~~~~~~~~~
>> ../include/linux/build_bug.h:16:62: note: in definition of macro ‘BUILD_BUG_ON_ZERO’
>>    16 | #define BUILD_BUG_ON_ZERO(e) ((int)(sizeof(struct { int:(-!!(e)); })))
>>       |                                                              ^
>> ../include/linux/bits.h:37:10: note: in expansion of macro ‘GENMASK_INPUT_CHECK’
>>    37 |         (GENMASK_INPUT_CHECK(h, l) + __GENMASK(h, l))
>>       |          ^~~~~~~~~~~~~~~~~~~
>> ../drivers/mux/mmio.c:96:29: note: in expansion of macro ‘GENMASK’
>>    96 |                 if (mask != GENMASK(field.msb, field.lsb)) {
>>       |                             ^~~~~~~
>> ../include/linux/build_bug.h:16:51: error: bit-field ‘<anonymous>’ width not an integer constant
>>    16 | #define BUILD_BUG_ON_ZERO(e) ((int)(sizeof(struct { int:(-!!(e)); })))
>>       |                                                   ^
>> ../include/linux/bits.h:23:10: note: in expansion of macro ‘BUILD_BUG_ON_ZERO’
>>    23 |         (BUILD_BUG_ON_ZERO(__builtin_choose_expr( \
>>       |          ^~~~~~~~~~~~~~~~~
>> ../include/linux/bits.h:37:10: note: in expansion of macro ‘GENMASK_INPUT_CHECK’
>>    37 |         (GENMASK_INPUT_CHECK(h, l) + __GENMASK(h, l))
>>       |          ^~~~~~~~~~~~~~~~~~~
>> ../drivers/mux/mmio.c:96:29: note: in expansion of macro ‘GENMASK’
>>    96 |                 if (mask != GENMASK(field.msb, field.lsb)) {
>>       |                             ^~~~~~~
>> ../drivers/mux/mmio.c:102:29: error: implicit declaration of function ‘devm_regmap_field_alloc’ [-Werror=implicit-function-declaration]
>>   102 |                 fields[i] = devm_regmap_field_alloc(dev, regmap, field);
>>       |                             ^~~~~~~~~~~~~~~~~~~~~~~
>> ../drivers/mux/mmio.c:76:34: warning: unused variable ‘field’ [-Wunused-variable]
>>    76 |                 struct reg_field field;
>>       |                                  ^~~~~
>>
> 
> Fixes: 8ecfaca7926f ("mux: mmio: add generic regmap bitfield-based multiplexer")

Right, I'll add that.

>> Reported by: Randy Dunlap <rdunlap@infradead.org>
>> Link: https://lore.kernel.org/lkml/64cf625b-2495-2382-0331-519d1cab0adf@infradead.org/
>> Signed-off-by: Peter Rosin <peda@axentia.se>
>> ---
>>  drivers/mux/Kconfig | 1 +
>>  1 file changed, 1 insertion(+)
>>
>> Note, there is a patch in linux-next that removes "|| COMPILE_TEST" from depends.

Still true, haven't gotten around to pushing that upstream. Sorry. Busy-busy...

>>
>> Cheers,
>> Peter
>>
>> diff --git a/drivers/mux/Kconfig b/drivers/mux/Kconfig
>> index 80f015cf6e54..9234f8c400ca 100644
>> --- a/drivers/mux/Kconfig
>> +++ b/drivers/mux/Kconfig
>> @@ -48,6 +48,7 @@ config MUX_GPIO
>>  config MUX_MMIO
>>  	tristate "MMIO/Regmap register bitfield-controlled Multiplexer"
>>  	depends on OF
>> +	select REGMAP
> 
> I get a dependency loop with this:
> 
> drivers/net/ethernet/arc/Kconfig:19:error: recursive dependency detected!
> drivers/net/ethernet/arc/Kconfig:19:symbol ARC_EMAC_CORE is selected by ARC_EMAC
> drivers/net/ethernet/arc/Kconfig:26:symbol ARC_EMAC depends on OF_IRQ
> drivers/of/Kconfig:69:symbol OF_IRQ depends on IRQ_DOMAIN
> kernel/irq/Kconfig:60:symbol IRQ_DOMAIN is selected by REGMAP
> drivers/base/regmap/Kconfig:6:symbol REGMAP is selected by MUX_MMIO
> drivers/mux/Kconfig:48:symbol MUX_MMIO depends on MULTIPLEXER
> drivers/mux/Kconfig:6:symbol MULTIPLEXER is selected by MDIO_BUS_MUX_MULTIPLEXER
> drivers/net/mdio/Kconfig:261:symbol MDIO_BUS_MUX_MULTIPLEXER depends on MDIO_DEVICE
> drivers/net/mdio/Kconfig:6:symbol MDIO_DEVICE is selected by PHYLIB
> drivers/net/phy/Kconfig:16:symbol PHYLIB is selected by ARC_EMAC_CORE
> 
> (on next-20230210).

Right. It seemed so obvious that I failed to even compile test before
sending the patch. LART applied to self. Sorry.

I'll follow up with a v2 shortly.

Cheers,
Peter