On Thu, Mar 19, 2020 at 2:11 PM Uwe Kleine-König <u.kleine-koenig@xxxxxxxxxxxxxx> wrote: > > Hello, > > [dropping Tony Prisk <linux@xxxxxxxxxxxxxxx> from recipients] > > On Thu, Mar 19, 2020 at 01:40:28PM +0200, Oleksandr Suvorov wrote: > > Thierry, I see the PWM core converts the bit field "third cell" into > > the polarity variable. > > Now I probably understand your sight and agree that we shouldn't give > > the same names to bits in bitfield (dts) and values of a variable. > > > > But there are lots of useless "0" values of third cell of "pwms" > > option in dts files. > > > > I see 2 ways now: > > - just remove all "0" "third cell" from "pwms" options in dts files. I > > see this "0" confuses some people. > > Then you have to overwrite pwm-cells of the provider. If there are two > PWMs used from the same provider and only one is inverted this won't > work. (Not entirely sure I understood your suggestion.) So I don't like > this suggestion. Good point, agree. But we still have the unnamed "0". What about renaming the dt-bindings macro PWM_POLARITY_INVERTED and add the new one for the normal polarity? Like PWM_FLAG_POLARITY_NORMAL / PWM_FLAG_POLARITY_INVERTED or DT_PWM_POLARITY_NORMAL / DT_PWM_POLARITY_INVERTED? Using different prefix will prevent interfering names of enum and macros in the future and will allow using the named nop-flag PWM_FLAG_POLARITY_NORMAL in dts. > And also in my eyes this isn't clearer, just more complicated to use. > > > - convert pwm_state.polarity into pwm_state.flags and use bitfield > > directly from dtb. > > It simplifies the parsing logic and makes adding new flags easier. > > *shrug*, I don't care much. > > Best regards > Uwe > > -- > Pengutronix e.K. | Uwe Kleine-König | > Industrial Linux Solutions | https://www.pengutronix.de/ | -- Best regards Oleksandr Suvorov Toradex AG Ebenaustrasse 10 | 6048 Horw | Switzerland | T: +41 41 500 48 00
