PWM test : Make PWM pin configurable

[jeromecoutant]

The new function is deducting the PinIn depending on the PinOut value.

For ex, PWM_0 is no more hardcoded connected to D2.

[BlackstoneEngineering]

What is the point of this? The current configuration settings are in the config file mbed_app.json, this is where the config is supposed to live.

[jeromecoutant]

I agree, configuration should be set in the json file.

But it is not the case...
Ex:
PWM_Period_Test< MBED_CONF_APP_PWM_0, MBED_CONF_APP_DIO_2, 10, 1000 >
=> PWM_0 is connected to D0, whatever the value of PWM_0 in json...

With the correction, test is checking the PWM_0 value to determine which digital pin is conencted

[BlackstoneEngineering]

@0xc0170

[0xc0170]

@jeromecoutant I dont follow

This is the config:

        "PWM_0": "D3",
        "PWM_1": "D5",
        "PWM_2": "D6",
        "PWM_3": "D9",

Thus PWM_0 is D3. What else is needed that this PR fixes? PinName PWM_IN(PinName pwm_out_pin) does not contain any documentation. How are pins assigned (out to in conversion) ?

This is one of the cases there:

+        case MBED_CONF_APP_DIO_0:
+            return MBED_CONF_APP_DIO_1;

DIO0 to DIO 1 ?

[jeromecoutant]

Test needs 2 pins: 1 PWM out and 1 Interrupt In.
CI-shield is connecting them.
So in the original code, PWM_0 (D3 as you said) is connected to D2, which is correct.

What happened if user is setting PWM_0=D0 in the json file?
If you want to avoid external wires, you need to set the Interrupt In to D1
This is the patch

Thx

[0xc0170]

@mray19027 your latest PR - is not resolving similar problem or can you review this one at least?

[mray190]

PR #47 fixes this issue

The PWM test dynamically finds the paired pin connected to the PWM pin and sets the interrupt to that paired pin.
https://github.com/ARMmbed/ci-test-shield/pull/47/files#diff-8071f2cb677762e6cf5f556dac2d8a74R79

[jeromecoutant]

Rebase done as PWM tests have been split, but without this PR

Each test case selects only the pwm_out_pin,
int_in_pin is no more hardcoded but defined according the selected pwm_out_pin.

[jeromecoutant]

Is there any issue with that ?

[jeromecoutant]

Hi all
Is ci-test-shield abandoned ?

[BlackstoneEngineering]

@c1728p9 - care to comment here?

[jeromecoutant]

CI test shield is quite important in ST automatic test setup.
This is the way we verify SPI, I2C and all others things.

[c1728p9]

Hi @jeromecoutant the CI test shield isn't abandoned as far as I'm aware. There is a more advanced test shield coming, but this test shield should still be supported. @donatieng does this sound right and if so who will be supporting this?

[0xc0170]

@jeromecoutant Shall this be recreated (rebased) or closed? It was left in air unfortunately.

[jeromecoutant]

I am using this PR for each non regression tests :-)
so I can rebase it