On Wed, Nov 22, 2023 at 6:14 PM Peter Rosin <peda@xxxxxxxxxx> wrote: > > > 2023-11-22 at 17:57, Bartosz Golaszewski wrote: > > On Wed, 22 Nov 2023 at 17:41, Peter Rosin <peda@xxxxxxxxxx> wrote: > >> > >> Hi! > >> > >> 2023-11-22 at 16:59, Bartosz Golaszewski wrote: > >>> - .of_match_table = at24_of_match, > >>> + .of_match_table = of_match_ptr(at24_of_match), > >> > >> If you do that, you will need to also add #ifdef CONFIG_OF around the > >> at24_of_match definition, of you'll trigger a warning about an unused > >> const variable (for some configs). I think. > >> > >> Cheers, > >> Peter > > > > Nope, no warnings even with W=2. This is true for unused functions but > > unused data structures are just silently removed. > > Then I wonder what the difference is from the following? > > https://lore.kernel.org/all/202311161306.opyfcoCY-lkp@xxxxxxxxx/T/#m3a33dc4c3221ae167563bcff70757af776cf07b1 > > It sure looks like the exact same pattern to me. > I.e. a static const struct of_device_id paired with of_match_ptr(). > > CHeers, > Peter Ah, we need __maybe_unused here. Bart
