New port for Infineon PSoC6 controllers #16705

jaenrig-ifx · 2025-02-05T12:38:17Z

Hi @mattytrentini @dpgeorge,

We have kept working on further enabling the PSoC6 controller in MPY. This pull request deprecates #15642.

Here a list of the features enabled:
machine classes:

ADC
ADCBlock
I2C. Master and slave mode.
I2S
Pin
PWM
RTC
SDCard
SPI. Master and slave mode.
Timer
UART
WDT

Additional modules:

network.WLAN
Virtual File System (LFS2 and FAT format)

PSoC6 evaluation boards available:

CY8CPROTO-062-4343W
CY8CPROTO-063-BLE
CY8CKIT-062S2-AI
Tests running in a hardware in the loop setup under ci/cd are provided for (at least) the main API functionalities of the enabled classes.

Looking forward to discuss the next steps regarding this PR with our teams.

The following modules are enabled: - machine.ADC - machine.ADCBlock - machine.I2C - machine.I2S - machine.Pin - machine.PWM - machine.RTC - machine.SDCard - machine.SPI - machine.Timer - machine.UART - machine.WDT - network.WLAN - os.vfs (LFS2 and FAT) Evaluation boards initially integrated in the port: - CY8CPROTO-062-4343W - CY8CPROTO-063-BLE - CY8CKIT-062S2-AI Signed-off-by: enriquezgarc <enriquezgarcia.external@infineon.com>

Signed-off-by: enriquezgarc <enriquezgarcia.external@infineon.com>

codecov · 2025-02-05T12:40:52Z

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 98.53%. Comparing base (112f657) to head (fe81830).
Report is 147 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master   #16705      +/-   ##
==========================================
- Coverage   98.59%   98.53%   -0.06%     
==========================================
  Files         167      169       +2     
  Lines       21599    21822     +223     
==========================================
+ Hits        21295    21502     +207     
- Misses        304      320      +16

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

❄ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

github-actions · 2025-02-05T12:48:43Z

Code size report:

   bare-arm:    +0 +0.000% 
minimal x86:    +0 +0.000% 
   unix x64:    +0 +0.000% standard
      stm32:   +48 +0.012% PYBV10
     mimxrt:   +44 +0.012% TEENSY40
        rp2:   +56 +0.006% RPI_PICO_W
       samd:   +56 +0.021% ADAFRUIT_ITSYBITSY_M4_EXPRESS
  qemu rv32:    +0 +0.000% VIRT_RV32

IhorNehrutsa · 2025-02-05T23:08:26Z

Great work.

Modify micropython\tests\extmod_hardware\machine_pwm.py
to pass the test of machine.PWM().

Try to check code formatting by
micropython\tools\codeformat.py

jaenrig-ifx · 2025-02-06T10:02:53Z

Great work.

Modify micropython\tests\extmod_hardware\machine_pwm.py to pass the test of machine.PWM().

Try to check code formatting by micropython\tools\codeformat.py

Thanks for you review @IhorNehrutsa !

You mean to add the psoc port and boards to that tests no? That we can do.

With the code formatting you mean after modifying the test?
We are running it in our ci and as pre-commit. The github action workflow for code formatting has passed for this PR.
Any error or failure we have missed or overlooked?

ports/psoc6/machine_pwm.c

IhorNehrutsa · 2025-02-05T22:46:59Z

ports/psoc6/machine_pwm.c

+
+static void mp_machine_pwm_print(const mp_print_t *print, mp_obj_t self_in, mp_print_kind_t kind) {
+    machine_pwm_obj_t *self = MP_OBJ_TO_PTR(self_in);
+    mp_printf(print, "frequency=%u duty_cycle=%f", self->fz, (double)self->duty);


mp_printf(print, "PWM(Pin(%s),freq=%u, duty_u16=%f", self->pin, self->fz, (double)self->duty);

As a test code something like:

>>> pwm = PWM(Pin(XXX), freq=5000, duty_u16=3272) >>> pwm2 = eval(str(pwm)) >>> pwm PWM(Pin(XXX), freq=5000, duty_u16=3272) >>> pwm2 PWM(Pin(XXX), freq=5000, duty_u16=3272)

EDIT: Use similar test codes for other objects. Pin, ADC, UART etc.

Is this a requirement/standard for machine classes that the human readable string representation of an object needs to match its constructor call?

In our case, this would use string pin labels (i.e. "P12_6") instead of pin objects (Pin(x)):

Virtual File System: mounted at '/' with LFS2 format in internal flash. MicroPython 104dbf82c-dirty on 2025-02-06; CY8CPROTO-063-BLE with PSoC63 Type "help()" for more information. >>> from machine import PWM >>> pwm_string= "PWM(\"P12_6\", freq=5000, duty_u16=3272)" >>> pwm =eval(str(pwm_string)) >>> pwm PWM("P12_6",freq=5000, duty_u16=3272.000000) >>>

Please take a look at the discussion in #4379.

Is this a requirement/standard for machine classes that the human-readable string representation of an object needs to match its constructor call?

IMHO: Yes. This allows save/restore, send/receive the objects.

Also:
PWM(Pin("P12_6"), freq=5000, duty_u16=3272) - is preferred.
PWM("P12_6", freq=5000, duty_u16=3272) - is possible.

Understood. Then we have to review all the rest of machine classes, if we want to be consistent.

Regarding the Pin() vs label. We will stick to "the possible" as long as possible 😜, as the the Pin() object creation as in our port was rather artificial... due to the GPIO handling of the underlying PSOC6 apis for the different hw (machine) peripherals.
We thought on the impact in portability (across ports), but at the end of the day, for each port, the pin definition is specific... Therefore, we did not see the benefit on making our port code more intricated.
If this is a requirement then we have to refactor and redesign the machine classes... If we could skip that, at least for now, it would be great 🤞

ports/psoc6/machine_pwm.c

IhorNehrutsa · 2025-02-06T10:16:40Z

You mean to add the psoc port and boards to that tests no?

Yes, add.

jaenrig-ifx · 2025-02-06T12:07:32Z

Perfect. We will start working on those changes.
For now I will add them as additional commits. Before merging we can squash some of these changes if that is required. Is that okay?

@IhorNehrutsa is your review completed? Or just have started from the PWM module?

IhorNehrutsa · 2025-02-06T13:17:44Z

ports/psoc6/machine_wdt.c

+void mod_wdt_deinit() {
+    cyhal_wdt_stop(&psoc6_wdt);
+    cyhal_wdt_free(&psoc6_wdt);
+}


wdt_deinit was rejected here :
ESP32/machine_wdt.c: Add WDT.deinit() method. #6631

The deinit is not exposed at user API level.

This function is called once the device is gets out of the REPL loop in the main.c (usually upon soft-reset):
https://github.com/Infineon/micropython/blob/ports-psoc6-pr-new/ports/psoc6/main.c#L225
As part of the macro MICROPY_PORT_DEINIT_FUNC:
https://github.com/Infineon/micropython/blob/ports-psoc6-pr-new/ports/psoc6/mpconfigport.h#L207-L208
Which calls them machine_deinit():
https://github.com/Infineon/micropython/blob/ports-psoc6-pr-new/ports/psoc6/modmachine.c#L124-L139

Its removal will prevent having a cleaned up post-reset operating system for the users, in which the watchdog could only be disabled through hard-reset.
I would need to understand how this is handled at other ports better, and the overall implications, before just getting rid of it.

Don't rush to delete; may wait for maintainers.

ports/psoc6/machine_pwm.c

IhorNehrutsa · 2025-02-06T14:12:00Z

Before merging we can squash some of these changes if that is required. Is that okay?

IMHO, okay. I have reviewed PWM and WDT.
Maintainers of the repo will review the full code.

IhorNehrutsa · 2025-02-06T15:07:45Z

ports/psoc6/machine_pwm.c

+        self->duty_type = DUTY_U16;
+    }
+
+    if (args[ARG_duty_ns].u_int != VALUE_NOT_SET) {


else if (args[ARG_duty_ns].u_int != VALUE_NOT_SET) {

HERE

jaenrig-ifx · 2025-02-06T16:04:54Z

You mean to add the psoc port and boards to that tests no?

Yes, add.

I tried, but we would need to implement machine.time_pulse_us() first.

Anyway, we have our own ci test for pmw:
https://github.com/Infineon/micropython/blob/ports-psoc6-pr-new/tests/ports/psoc6/board_ext_hw/single/pwm.py

In the long run I assume the goal is to generalize and reuse as much the common tests for all ports.
If this is a currently a long term goal, maybe we can add this is as improvement.

If this is a must for the PR to be approved, we can/should implement it. It should be rather quick unless we fail to achieve the pwm and time_pulse_us performance required for that test.

IhorNehrutsa · 2025-02-07T07:40:41Z

Are all 183 files needed in PR?

jaenrig-ifx · 2025-02-07T08:29:50Z

Are all 183 files needed in PR?

I would say so. We could argue about a few of them, but they are all used.

For example, the "mtb-libs" is used for integration of the toolchain and the dependencies of the PSOC6. We could make that a submodule, but it needs to be retrieved (initialized and updated as submodule) to build the port.

The tests are all running under our CI HIL (hardware-in-the-loop). But of course, they aren´t needed for building the PSOC6 MicroPython binary.

Some files under "tools/psoc6" might be only used by our fork ci hardware-in-the-loop workflow. If some of those are bothering, we could keep them in our fork only...

Signed-off-by: enriquezgarc <enriquezgarcia.external@infineon.com>

jaenrig-ifx · 2025-02-07T12:25:48Z

I won´t yet push to this branch until you consider the first review iteration done.
You can keep track of the changes already implemented here:
Infineon/micropython@ports-psoc6-pr-new...Infineon:micropython:pr-feedback

ports/psoc6/machine_pwm.c

IhorNehrutsa · 2025-02-07T13:36:14Z

ports/psoc6/machine_pwm.c

+static cy_rslt_t pwm_freq_duty_set(cyhal_pwm_t *pwm_obj, uint32_t fz, float duty_cycle) {
+    return cyhal_pwm_set_duty_cycle(pwm_obj, duty_cycle * 100, fz); // duty_cycle in percentage
+}
+
+static inline cy_rslt_t pwm_duty_set_ns(cyhal_pwm_t *pwm_obj, uint32_t fz, uint32_t pulse_width) {
+    return cyhal_pwm_set_period(pwm_obj, 1000000 / fz, pulse_width / 1000); // !# * --> /
+}


It may be more convenient to choose one of these functions cyhal_pwm_set_duty_cycle or cyhal_pwm_set_period and use only that one in both cases.

Also, it is possible to recalculate a duty at the last minute before calling cyhal_pwm_set_XXX and remove float and double in other places.

Understand it might be a bit confusing. But, taking reference from here: https://ifx-micropython.readthedocs.io/en/latest/library/machine.PWM.html#machine.PWM

When duty_u16 is set, it is set as raw value and cyhal_pwm_set_duty_cycle allows to pass duty_cycle as percentage.

When duty_ns is set, it is set as time in ns and cyhal_pwm_set_period allows to directly pass time.

It helps to reuse the low-level API's as much as intended functionality is provided through them. Does that explains?

Yes.
Anyhow, I think as long as possible is a good idea to restrict the conversion to one place. @IhorNehrutsa let us check if we can refactor without breaking any functionality :)

As mentioned in the other comment:
Infineon@3c14c39

I have kept the 2 ways of setting freq/duty. But now abstracted and encapsulated under pwm_config().

IhorNehrutsa · 2025-02-07T14:05:01Z

mp_raise_ValueError(MP_ERROR_TEXT("PWM duty should be specified in either ns or u16"));

There was a big discussion about what to do when duty_u16 or duty_ns are not specified.
Some ports do not start until both frequency and duty are specified, but do not raise an exception.
ESP32 port starts with a default 5000Hz & 50% duty.

robert-hh · 2025-02-07T15:24:06Z

Some ports do not start until both frequency and duty are specified, but do not raise an exception.

Actually that is the common and intended behavior: not PWM signal unless both freq and duty cycle have been specified. There are no reasonable default values. The ESP32 does not yet meet that rule.

IhorNehrutsa · 2025-02-11T18:50:17Z

I have no restriction about the

You can keep track of the changes already implemented here: Infineon/micropython@ports-psoc6-pr-new...Infineon:micropython:pr-feedback

You may merge

IhorNehrutsa · 2025-02-12T10:23:56Z

You may interested in #16743

jaenrig-ifx · 2025-02-13T09:45:49Z

mp_raise_ValueError(MP_ERROR_TEXT("PWM duty should be specified in either ns or u16"));

There was a big discussion about what to do when duty_u16 or duty_ns are not specified. Some ports do not start until both frequency and duty are specified, but do not raise an exception. ESP32 port starts with a default 5000Hz & 50% duty.

In the PR feedback changes we are currently making mandatory that freq and only one of the duty modes are set.

freq mandatory: https://github.com/Infineon/micropython/blob/pr-feedback/ports/psoc6/machine_pwm.c#L94
mutual exclusion of duty formats: https://github.com/Infineon/micropython/blob/pr-feedback/ports/psoc6/machine_pwm.c#L108-L111
At least one duty format: https://github.com/Infineon/micropython/blob/pr-feedback/ports/psoc6/machine_pwm.c#L113-L124

Any issue with this approach?

Something like in the esp32 port:
https://github.com/micropython/micropython/blob/master/ports/esp32/machine_pwm.c#L480-L482

jaenrig-ifx · 2025-02-13T10:06:47Z

You may interested in #16743

Yes, let us put that in our backlog but with less prio.

jaenrig-ifx · 2025-02-17T17:29:53Z

I have no restriction about the

You can keep track of the changes already implemented here: Infineon/micropython@ports-psoc6-pr-new...Infineon:micropython:pr-feedback

You may merge

If you approve the changes in Infineon@3c14c39
I will merge.

Signed-off-by: enriquezgarc <enriquezgarcia.external@infineon.com>

jaenrig-ifx · 2025-02-18T08:28:30Z

Feedback changes discussed merged now.

IhorNehrutsa · 2025-02-20T13:09:58Z

Difficult to review due to the large number of files
https://github.com/micropython/micropython/pull/16705/files

jaenrig-ifx added 9 commits February 5, 2025 13:23

docs/psoc6: Added user documentation for PSOC6 port.

0361599

Signed-off-by: enriquezgarc <enriquezgarcia.external@infineon.com>

tools/psoc6: Added PSOC6 port support tools.

5c8bf1f

Signed-off-by: enriquezgarc <enriquezgarcia.external@infineon.com>

tests/ports/psoc6: Added machine and network hardware-in-the-loop tests.

4d31f2e

Signed-off-by: enriquezgarc <enriquezgarcia.external@infineon.com>

.github/workflows/ports_psoc6.yml: Added PSOC6 port CI workflow.

1d541b3

Signed-off-by: enriquezgarc <enriquezgarcia.external@infineon.com>

tools/ci.sh: Added PSOC6 port CI support functions.

5f96164

Signed-off-by: enriquezgarc <enriquezgarcia.external@infineon.com>

extmod/machine_i2c: Added deinit function to i2c API.

3745ed8

Signed-off-by: enriquezgarc <enriquezgarcia.external@infineon.com>

extmod/modlwip.c: Enabled arc_(un)protected only if not provided by OS.

c557d67

Signed-off-by: enriquezgarc <enriquezgarcia.external@infineon.com>

README.md: Added PSOC6 to list of ports.

104dbf8

Signed-off-by: enriquezgarc <enriquezgarcia.external@infineon.com>

jaenrig-ifx mentioned this pull request Feb 5, 2025

New port for Infineon PSoC6 controllers #15642

Closed

jaenrig-ifx changed the title ~~Ports psoc6 pr new~~ New port for Infineon PSoC6 controllers Feb 5, 2025

IhorNehrutsa reviewed Feb 6, 2025

View reviewed changes

ports/psoc6/machine_pwm.c Outdated Show resolved Hide resolved

IhorNehrutsa reviewed Feb 6, 2025

View reviewed changes

ports/psoc6/machine_pwm.c: Refined constructor, str and cleanup.

6c51875

Signed-off-by: enriquezgarc <enriquezgarcia.external@infineon.com>

IhorNehrutsa reviewed Feb 7, 2025

View reviewed changes

ports/psoc6/machine_pwm.c Outdated Show resolved Hide resolved

IhorNehrutsa reviewed Feb 7, 2025

View reviewed changes

dpgeorge added the ports Relates to multiple ports, or a new/proposed port label Feb 11, 2025

jaenrig-ifx added 2 commits February 18, 2025 09:11

ports/psoc6: Renamed machine classes deinit all.

3159ab8

Signed-off-by: enriquezgarc <enriquezgarcia.external@infineon.com>

ports/psoc6/machine_pwm.c: Clean up refactor and fixes on arithmetic.

fe81830

Signed-off-by: enriquezgarc <enriquezgarcia.external@infineon.com>

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

New port for Infineon PSoC6 controllers #16705

New port for Infineon PSoC6 controllers #16705

jaenrig-ifx commented Feb 5, 2025

codecov bot commented Feb 5, 2025 •

edited

Loading

github-actions bot commented Feb 5, 2025

IhorNehrutsa commented Feb 5, 2025

jaenrig-ifx commented Feb 6, 2025

IhorNehrutsa Feb 5, 2025 •

edited

Loading

jaenrig-ifx Feb 6, 2025

IhorNehrutsa Feb 6, 2025

jaenrig-ifx Feb 6, 2025

IhorNehrutsa commented Feb 6, 2025 •

edited

Loading

jaenrig-ifx commented Feb 6, 2025

IhorNehrutsa Feb 6, 2025

jaenrig-ifx Feb 6, 2025

IhorNehrutsa Feb 6, 2025

IhorNehrutsa commented Feb 6, 2025

IhorNehrutsa Feb 6, 2025

jaenrig-ifx Feb 6, 2025

jaenrig-ifx commented Feb 6, 2025

IhorNehrutsa commented Feb 7, 2025

jaenrig-ifx commented Feb 7, 2025

jaenrig-ifx commented Feb 7, 2025

IhorNehrutsa Feb 7, 2025

NikhitaR-IFX Feb 10, 2025

jaenrig-ifx Feb 13, 2025

jaenrig-ifx Feb 17, 2025

IhorNehrutsa commented Feb 7, 2025

robert-hh commented Feb 7, 2025

IhorNehrutsa commented Feb 11, 2025

IhorNehrutsa commented Feb 12, 2025

jaenrig-ifx commented Feb 13, 2025

jaenrig-ifx commented Feb 13, 2025

jaenrig-ifx commented Feb 17, 2025

jaenrig-ifx commented Feb 18, 2025

IhorNehrutsa commented Feb 20, 2025

New port for Infineon PSoC6 controllers #16705

Are you sure you want to change the base?

New port for Infineon PSoC6 controllers #16705

Conversation

jaenrig-ifx commented Feb 5, 2025

codecov bot commented Feb 5, 2025 • edited Loading

Codecov Report

github-actions bot commented Feb 5, 2025

IhorNehrutsa commented Feb 5, 2025

jaenrig-ifx commented Feb 6, 2025

IhorNehrutsa Feb 5, 2025 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

IhorNehrutsa commented Feb 6, 2025 • edited Loading

jaenrig-ifx commented Feb 6, 2025

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

IhorNehrutsa commented Feb 6, 2025

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jaenrig-ifx commented Feb 6, 2025

IhorNehrutsa commented Feb 7, 2025

jaenrig-ifx commented Feb 7, 2025

jaenrig-ifx commented Feb 7, 2025

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

IhorNehrutsa commented Feb 7, 2025

robert-hh commented Feb 7, 2025

IhorNehrutsa commented Feb 11, 2025

IhorNehrutsa commented Feb 12, 2025

jaenrig-ifx commented Feb 13, 2025

jaenrig-ifx commented Feb 13, 2025

jaenrig-ifx commented Feb 17, 2025

jaenrig-ifx commented Feb 18, 2025

IhorNehrutsa commented Feb 20, 2025

codecov bot commented Feb 5, 2025 •

edited

Loading

IhorNehrutsa Feb 5, 2025 •

edited

Loading

IhorNehrutsa commented Feb 6, 2025 •

edited

Loading