Re: [PATCH v16] pwm: opencores: Add PWM driver support

[Uwe Kleine-König]

On Tue, Dec 24, 2024 at 01:53:03AM +0000, William Qiu wrote:
> > -----Original Message-----
> > From: Uwe Kleine-König <ukleinek@xxxxxxxxxx>
> > Sent: 2024年12月24日 1:12
> > To: William Qiu <william.qiu@xxxxxxxxxxxxxxxx>
> > Cc: linux-kernel@xxxxxxxxxxxxxxx; linux-pwm@xxxxxxxxxxxxxxx; Hal Feng
> > <hal.feng@xxxxxxxxxxxxxxxx>; Philipp Zabel <p.zabel@xxxxxxxxxxxxxx>
> > Subject: Re: [PATCH v16] pwm: opencores: Add PWM driver support
> > 
> > Hello William,
> > 
> > On Mon, Dec 23, 2024 at 02:47:59AM +0000, William Qiu wrote:
> > > > > +	duty_data = mul_u64_u32_div(state->duty_cycle, ddata->clk_rate,
> > > > NSEC_PER_SEC);
> > > > > +	if (!duty_data)
> > > > > +		return -EINVAL;
> > > >
> > > > I can understand that period_data == 0 is an error, but duty_data ==
> > > > 0 could/should just work?!
> > >
> > > It means no need to check whether the duty is valid?
> > 
> > No, it means that I expect that duty_data == 0 is a valid setting and most
> > controllers support it.
> > 
> > Best regards
> > Uwe
> 
> So we just need to check duty < 0?

I'm not sure I understand this question. You can assume that
state->duty_cycle is >= 0 in the driver callback.

If the hardware doesn't support duty_cycle == 0, add a comment and
proper error handling. (i.e. return with -EINVAL if the requested
duty_cycle is too low to be implemented.)

Best regards
Uwe

Attachment: signature.asc
Description: PGP signature