pwm-ctrl: fix disable if duty cycle is 0

[sstepansky]

No description provided.

[gtxaspec]

this won't work, on some devices 0 duty is full blast. which is why we have reverse. i'll have to dig through later and find the problem commit.

[sstepansky]

Maybe it's just me, I'm a little confused about the "reversed" thing... I think I assumed it was related to "active low" pins, to get the correct behavior.

I'll try to explain my understanding of this.

Let's assume we're using REVERSE=0, so the pin is reversed and we enter the if at L#211.
Next, L#213 has an -e argument, thus enabling the channel, not disabling it (as seen by running pwm -c 1 -q). It looks like it's disabled because the previous line sets the duty cycle to 0. To disable the channel (as seen in pwm -c 1 -q) we have to use the -d argument. So, if that's to be changed, then both branches if and else are the same.

For example, my cam's white pin is 50o. To turn it on at a faint 5% I run pwm-ctrl -r 50 5. Not using the "reversed" argument results in a very bright light, not the faint light. After that, to turn it off I used pwm-ctrl -r 50 0, again using -r.
Now run pwm -c 1 -q to check:

PWM Channel 1 Status:
Enabled: Enabled
Polarity: Inversed
Duty: 0 ns
Period: 1000000 ns

It's enabled.

Run pwm -c 1 -d to manually disable it, and then check status with pwm -c 1 -q

PWM Channel 1 Status:
Enabled: Disabled
Polarity: Inversed
Duty: 0 ns
Period: 1000000 ns

Running the command to set 0% duty cycle but with pin 49O results in

PWM Channel 0 Status:
Enabled: Disabled
Polarity: Normal
Duty: 0 ns
Period: 1000000 ns

It's disabled, ok here.

The part to enable pwm, from L#219 on, seems a little bit strange, because if REVERSE=0 it is first disabled, and then enabled in the last call to pwm. If REVERSE=1 it's enabled twice.

Please shed some light on this (pun intended).