Re: [RESEND PATCH v27 11/15] leds: lp55xx: Add multicolor framework support to lp55xx

From: Dan Murphy
Date: Mon Jun 22 2020 - 09:05:28 EST


Jacek

On 6/21/20 3:24 PM, Jacek Anaszewski wrote:
Dan,

On 6/21/20 4:12 PM, Dan Murphy wrote:
Jacek

On 6/19/20 5:10 PM, Jacek Anaszewski wrote:
Dan,

On 6/19/20 6:35 PM, Dan Murphy wrote:
Jacek

On 6/18/20 6:26 PM, Jacek Anaszewski wrote:
On 6/19/20 12:09 AM, Jacek Anaszewski wrote:
Dan,

On 6/18/20 11:44 PM, Dan Murphy wrote:
Jacek

On 6/18/20 4:21 PM, Jacek Anaszewski wrote:
Dan,

On 6/18/20 12:33 AM, Dan Murphy wrote:
Jacek

On 6/17/20 4:41 PM, Jacek Anaszewski wrote:
Dan,

On 6/17/20 9:22 PM, Dan Murphy wrote:
Pavel/Jacek

On 6/17/20 11:28 AM, kernel test robot wrote:
Hi Dan,

I love your patch! Yet something to improve:

[auto build test ERROR on pavel-linux-leds/for-next]
[cannot apply to j.anaszewski-leds/for-next]
[if your patch is applied to the wrong git tree, please drop us a note to help
improve the system. BTW, we also suggest to use '--base' option to specify the
base tree in git format-patch, please see https://stackoverflow.com/a/37406982]

url: https://github.com/0day-ci/linux/commits/Dan-Murphy/Multicolor-Framework-v27/20200616-042217
base: git://git.kernel.org/pub/scm/linux/kernel/git/pavel/linux-leds.git for-next
config: ia64-randconfig-r015-20200617 (attached as .config)
compiler: ia64-linux-gcc (GCC) 9.3.0
reproduce (this is a W=1 build):
         wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross
         chmod +x ~/bin/make.cross
         # save the attached .config to linux build tree
         COMPILER_INSTALL_PATH=$HOME/0day COMPILER=gcc-9.3.0 make.cross ARCH=ia64

If you fix the issue, kindly add following tag as appropriate
Reported-by: kernel test robot <lkp@xxxxxxxxx>

All errors (new ones prefixed by >>, old ones prefixed by <<):

ia64-linux-ld: drivers/leds/leds-lp55xx-common.o: in function `lp55xx_set_mc_brightness':
drivers/leds/leds-lp55xx-common.c:146: undefined reference to `led_mc_calc_color_components'
ia64-linux-ld: drivers/leds/leds-lp55xx-common.o: in function `devm_led_classdev_multicolor_register':
include/linux/led-class-multicolor.h:74: undefined reference to `devm_led_classdev_multicolor_register_ext'
vim +146 drivers/leds/leds-lp55xx-common.c

    138
    139    static int lp55xx_set_mc_brightness(struct led_classdev *cdev,
    140                        enum led_brightness brightness)
    141    {
    142        struct led_classdev_mc *mc_dev = lcdev_to_mccdev(cdev);
    143        struct lp55xx_led *led = mcled_cdev_to_led(mc_dev);
    144        struct lp55xx_device_config *cfg = led->chip->cfg;
    145
  > 146 led_mc_calc_color_components(&led->mc_cdev, brightness);
    147        return cfg->multicolor_brightness_fn(led);
    148

Well this was a mess to figure out.

The only fix I can figure out here is to remove the

     depends on LEDS_CLASS_MULTI_COLOR || !LEDS_CLASS_MULTI_COLOR

from each child device and add

     select LEDS_CLASS_MULTI_COLOR

to the LP55XX_COMMON

This way the Multi color framework will inherit the symbol that was set by the COMMON flag which is inherited by majority from the child flags.

Did you try this?

--- a/drivers/leds/Kconfig
+++ b/drivers/leds/Kconfig
@@ -398,6 +398,7 @@ config LEDS_LP50XX
 config LEDS_LP55XX_COMMON
        tristate "Common Driver for TI/National LP5521/5523/55231/5562/8501"
        depends on LEDS_LP5521 || LEDS_LP5523 || LEDS_LP5562 || LEDS_LP8501
+       depends on LEDS_CLASS_MULTI_COLOR || !LEDS_CLASS_MULTI_COLOR
        depends on OF
        select FW_LOADER
        select FW_LOADER_USER_HELPER


Yes I did

That gave unmet dependencies.

WARNING: unmet direct dependencies detected for LEDS_LP55XX_COMMON
   Depends on [m]: NEW_LEDS [=y] && (LEDS_LP5521 [=n] || LEDS_LP5523 [=m] || LEDS_LP5562 [=y] || LEDS_LP8501 [=y]) && (LEDS_CLASS_MULTI_COLOR [=m] || !LEDS_CLASS_MULTI_COLOR [=m]) && OF [=y]
   Selected by [y]:
   - LEDS_LP5562 [=y] && NEW_LEDS [=y] && LEDS_CLASS [=y] && I2C [=y]
   - LEDS_LP8501 [=y] && NEW_LEDS [=y] && LEDS_CLASS [=y] && I2C [=y]
   Selected by [m]:
   - LEDS_LP5523 [=m] && NEW_LEDS [=y] && LEDS_CLASS [=y] && I2C [=y] && (LEDS_CLASS_MULTI_COLOR [=m] || !LEDS_CLASS_MULTI_COLOR [=m])


When I was testing that yesterday I also had the same warning at some
point of testing different Kconfig setups, but with what I showed above
it ceased to appear. Now every time I am doing "make oldconfig" the
CONFIG_LEDS_LP55XX_COMMON=y entry gets changed to =m with the config
from the test bot.

That is not what I saw in my testing especially after doing a distclean

Could you please give your exact steps after "make distclean" and
copying test bot config to the kernel root directory?

Also, please share the toolchain you're using for tests.

Actually at this stage the toolchain is of lesser relevance.

I've tried once more and indeed the problem shows up.

It is caused by the driver entries doing

"select LEDS_LP55XX_COMMON".

Select sets config to "y" so it conflicts with
"depends on LEDS_CLASS_MULTI_COLOR || !LEDS_CLASS_MULTI_COLOR"
in the "config LEDS_LP55XX_COMMON".

Your proposed fix will block the possibility of building
LED_CLASS_MULTI_COLOR as a module when LP55XX drivers
are enabled so this is also not an option.

Solving this riddle will require some more thinking.
I haven't analyzed it in detail but maybe "imply" statement from
kconfig-language.rst could help somehow here.

The multicolor framework will build as a module if the LED_CLASS is defined as a module.

See attached test_defconfig

But it will be impossible to enable CONFIG_LEDS_LP50XX without
CONFIG_LEDS_CLASS_MULTI_COLOR if you will remove

depends on LEDS_CLASS_MULTI_COLOR || !LEDS_CLASS_MULTI_COLOR.

I was not removing the dependency for the LP50xx only the LP55xx.

It was my typo, I meant LP55XX.


This is actually why the above entry was needed.

LP55XX drivers have to work also without multicolor class.

Well I am not sure how else to resolve this problem.  Because the LP55xx has multi level dependencies.

Only the LP55xx_common has the dependency on the MC framework. The device drivers do not.

The issue is the mixing and matching of the MC fw as a module vs the LP55XX_COMMON as a built-in.

The simplest solution is to avoid selecting LP55XX_COMMON in favor
of making child drivers depending on it. Of course, maybe there exists
some trickier option to keep the selection, but I would have to play
a bit more with it to be sure.

That actually makes more sense then what is there now.

The children depend on the COMMON and the COMMON depends on the MC.

Not sure why select was used originally.

But this will prompt another patch to update the u8500_defconfig to include the common.

Dan