Re: PWM and user space API

[Jason Thorpe <thorpej%me.com@localhost>]

> On May 28, 2018, at 11:09 AM, Jason Thorpe <thorpej%me.com@localhost> wrote:
> 
> I’m going to be writing a driver for the NXP PCA9685, and want to glue it into the PWM kernel API jmcneill@ added… but I have a bunch of uses for this device that won’t really be covered by FDT, so I think there needs to be some sort of user space API as well.

Actually, while I’m at it, there are some other problems with the current API, trying to think of how I would adjust it…

The PCA9685 is fairly simple in that it basically has 2 comparators per channel that reference a 12-bit counter that’s either driven by the internal 25MHz oscillator, or an external clock source that can run up to 50MHz.

There an “ON” value and an “OFF” value.  When the counter is equal to the ON value, the output is asserted.  When the counter is equal to the OFF value, the counter is de-asserted.  The advantage of this is that it, if you’re referencing multiple PCA9685s to the same external clock, you can do any phase-shift compensation as necessary by setting the ON time for a group of channels to non-zero.  The current API doesn’t have any way to express this.

The PCA9685 also doesn’t have a variable clock UNLESS it’s driven by an external clock source, in which case a pre-scaler value can be applied (though, I’m not 100% certain on this — need to re-read the data sheet).  So having to configure the period as the current API requires is kind of inconvenient.

-- thorpej