Re: [edk2-devel] [edk2-platforms][PATCH v1 1/2] Platform/RaspberryPi: Add support for PWM1 in ACPI

["Mario Bălănică" <mariobalanica02@gmail.com>]

Hi Jeremy,

> On 11/5/21 15:34, Mario Bălănică via groups.io wrote:
>>
>> Also fix PWM0 on the Raspberry Pi 4, but we can't expose both yet.
>
> Why is that? The rpi4 needs both PWM devices to output stereo right? So
> with this patch is still mono?


Each PWM controller has two channels, so only one controller is needed
for stereo output.

> Assuming we are just aiming for mono on the rpi4, Do we need a separate
> set of buffers here for both the rpi3 and 4, can't the buffers be reused
> below?


> I would expect that the pi4 just reuses 17-30 here, or it uses both PWM
> channels.


Right, they could be reused, but the plan is to expose both
controllers on RPi 4 in the future.
We can't do it right now because of some conflicts in the Windows drivers.

The Clock Manager is shared, which may cause a race condition if both
devices try to change the PWM clock.

> But, that said. I think there might be a better, but much more complex
> way to do this.


> So, instead of trying to _DSD the base freq/etc in the next patch we
> just define a _DSM() method which sets the audio/pwm rate. Maybe it
> takes the audio sample rate and depth, sets up the PWM divisor and
> returns the pwm rate. (along with errors about boundaries, or we assume
> the sample rate is 4 bits and bound the sample rate/etc).


Yeah, a _DSM would be better. This way, I can also get rid of the
clock handling code in the PWM driver.
But then only the audio driver would be able to set the clock divisor.
I guess it's not really an issue, as long as we set an initial divisor
so that the PWM peripherals can be used without the audio driver being
installed.

Even though I don't really like having those clock values in the audio
_DSD, I think it's fine until we decide on / implement a better
solution.

--Mario