Re: [PATCH v8 4/8] dt-bindings: pwm: Support new PWM_USAGE_POWER flag

From: Thierry Reding <thierry.reding@gmail.com>
To: "Uwe Kleine-König" <u.kleine-koenig@pengutronix.de>
Cc: Clemens Gruber <clemens.gruber@pqgruber.com>,
	linux-pwm@vger.kernel.org,
	Sven Van Asbroeck <TheSven73@gmail.com>,
	devicetree@vger.kernel.org, linux-kernel@vger.kernel.org,
	Rob Herring <robh+dt@kernel.org>
Subject: Re: [PATCH v8 4/8] dt-bindings: pwm: Support new PWM_USAGE_POWER flag
Date: Tue, 13 Apr 2021 13:51:15 +0200	[thread overview]
Message-ID: <YHWFs1f0XHkqbddp@orome.fritz.box> (raw)
In-Reply-To: <20210412162723.7hlhgqp6wlfbkeky@pengutronix.de>

[-- Attachment #1: Type: text/plain, Size: 2544 bytes --]

On Mon, Apr 12, 2021 at 06:27:23PM +0200, Uwe Kleine-König wrote:
> On Mon, Apr 12, 2021 at 03:27:41PM +0200, Clemens Gruber wrote:
> > Add the flag and corresponding documentation for PWM_USAGE_POWER.
> 
> My concern here in the previous round was that PWM_USAGE_POWER isn't a
> name that intuitively suggests its semantic. Do you disagree?

I suggested PWM_USAGE_POWER because I think it accurately captures what
we want here.

> > Cc: Rob Herring <robh+dt@kernel.org>
> > Signed-off-by: Clemens Gruber <clemens.gruber@pqgruber.com>
> > ---
> >  Documentation/devicetree/bindings/pwm/pwm.txt | 3 +++
> >  include/dt-bindings/pwm/pwm.h                 | 1 +
> >  2 files changed, 4 insertions(+)
> > 
> > diff --git a/Documentation/devicetree/bindings/pwm/pwm.txt b/Documentation/devicetree/bindings/pwm/pwm.txt
> > index 084886bd721e..fe3a28f887c0 100644
> > --- a/Documentation/devicetree/bindings/pwm/pwm.txt
> > +++ b/Documentation/devicetree/bindings/pwm/pwm.txt
> > @@ -46,6 +46,9 @@ period in nanoseconds.
> >  Optionally, the pwm-specifier can encode a number of flags (defined in
> >  <dt-bindings/pwm/pwm.h>) in a third cell:
> >  - PWM_POLARITY_INVERTED: invert the PWM signal polarity
> > +- PWM_USAGE_POWER: Only care about the power output of the signal. This
> > +  allows drivers (if supported) to optimize the signals, for example to
> > +  improve EMI and reduce current spikes.
> 
> IMHO there are too many open questions about which freedom this gives to
> the lowlevel driver. If the consumer requests .duty_cycle = 25ns +
> .period = 100ns, can the driver provide .duty_cycle = 25s + .period =
> 100s which nominally has the same power output? Let's not introduce more
> ambiguity than there already is.

The freedom given to the driver should be to adjust the signal within
reasonable bounds. Changing the time unit by a factor of 1000000000 is
not within reason, and I doubt anyone would interpret it that way, even
if we didn't document this at all.

To be frank I think that quest of yours to try and rid the PWM API of
all ambiguity is futile. I've been trying to be lenient because you seem
motivated, but I think you're taking this too far. There are always
going to be cases that aren't completely clear-cut and where drivers
need the flexibility to cheat in order to be useful at all. If we get to
a point where everything needs to be 100% accurate, the majority of the
PWM controllers won't be usable at all.

Don't let perfect be the enemy of good.

Thierry

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]