On Mon Jan 20, 2025 at 3:13 PM CET, Uwe Kleine-König wrote: > Hello Mathieu, > > On Fri, Jan 17, 2025 at 04:47:45PM +0100, Mathieu Dubois-Briand wrote: > > On Fri Jan 17, 2025 at 3:40 PM CET, Uwe Kleine-König wrote: > > > sounds great. Note that the detail in rounding that is different for > > > waveforms is that a value that cannot be round down to a valid value > > > (because it's too small) is round up. This is a bit ugly in the drivers > > > but simplifies usage considerably. So you never return -EINVAL because > > > the values don't fit. > > > > Sorry, I'm not sure I got it right. Does this affect the three members > > of pwm_waveform (period_length_ns, duty_offset_ns, duty_length_ns) ? So > > on this device where the period is fixed and I cannot define an offset, > > does that mean I will silently accept any value for period_length_ns and > > duty_offset_ns ? > > Yes. The fromhw callback obviously always fills the respective constants > into .period_length_ns and .duty_offset_ns and the tohw callback > essentially only looks at .duty_length_ns. > > Best regards > Uwe Ok, thanks! I will make these changes for the next version. Best regards, Mathieu -- Mathieu Dubois-Briand, Bootlin Embedded Linux and Kernel engineering https://bootlin.com
