Hi! > >I have not yet gone through the series in great detail. > > > >This will change the userland ABI, right? Now, I understand that old ABI is bad, but > >will it break someone's code? > > It will not break anyone since the generic support for composing > LED names is used only if struct init_data is provided. Otherwise > led_cdev->name property is taken as is for the LED class device name. > > Moreover, it is safe to switch many of existing DT (or fwnode in > general) based drivers to using generic mechanism. > If the DT provides label property it will be concatenated with the > led_hw_name provided by the driver. The problem would be only with > the drivers whose bindings erroneously require devicename section > in the label. They will need to be left intact. Ok. Plus it seems series causes some compile issues: 924 kbuild tes [j.anaszewski-leds:led_naming_v3 13/37] drivers/leds/leds-pwm.c:117:8: error: 925 N kbuild tes [j.anaszewski-leds:led_naming_v3 17/37] drivers/leds/led-core.o:undefined ref 926 N kbuild tes [j.anaszewski-leds:led_naming_v3 13/37] drivers/leds/leds-pwm.c:117:44: spars 927 N kbuild tes [j.anaszewski-leds:led_naming_v3 13/37] drivers/leds/leds-pwm.c:117:8: error: Pavel -- (english) http://www.livejournal.com/~pavelmachek (cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
Attachment:
signature.asc
Description: Digital signature
