Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 98.54%. Comparing base (91cff8e) to head (037f2da).
Report is 77 commits behind head on master.

@@           Coverage Diff           @@
##           master   #17050   +/-   ##
=======================================
  Coverage   98.54%   98.54%           
=======================================
  Files         169      169           
  Lines       21890    21890           
=======================================
  Hits        21571    21571           
  Misses        319      319           

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

Code size report:

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

Commits that can be squashed:

• All changes to board config should be squashed into 2 commits and moved last (I see at least 28 commits).
• Gradual changes to mpconfigport.h that enable small features and fix for object representation, stack entry size etc.. could just be squashed into the initial port support.
• USB DCD updates, make USB IRQ handler weak etc.. can be squashed into the initial port.
• Add IRQ priorities, Lower CSI priority, add pendSV, softtimer etc.. can be squashed into the initial port.
• Minor Makefile flags fixes (like single precision, arch fix, add deploy-jlink target, fix jlink command. etc..)
• I see 3-4 machine_spi commits, should probably squash them all with the initial machine_spi support, or at least in one commit. This might be tricky though if the af unit changes were introduced in between.
• Many commits add new SE service wrappers, I'd squash them, along with small changes to se_services, ex: "Synchronize with SE after initialisation".
• Open-AMP support, and followup fixes, could be squashed and moved after MPU support.

This should be enough for a first pass, will take another look after these are squashed.

@dpgeorge - YAY! It's almost merged!

@dpgeorge - Any update on getting this merged?

@dpgeorge I see we're down to 127 commits. Note you could drop some of the commits here if you merge #16848 and rebase.

Also, some more commits could be easily squashed if you like:

Initial port:

pick alif: Add initial port to Alif Ensemble MCUs.
squash alif/tinyusb_port: Add Alif TinyUSB DCD driver.
squash alif/tinyusb_port: Disable USB IRQ on deinit.
squash alif/tinyusb_port: Implement SOF event.
squash alif/mpconfigport: Enable machine.SoftI2C. 
squash alif/mpconfigport: Enable os.sep, os.sync, os.uname.
squash alif/mpconfigport: Enable os.urandom(). 
squash alif/mpconfigport: Enable MICROPY_PY_RANDOM_SEED_INIT_FUNC. 
squash  alif/modalif: Add alif.info() function.
squash  alif/modmachine: Implement machine.unique_id(), fix machine.reset(). 
squash  alif: Update to work with latest DFP 1.2.0. 
squash alif: Enable dupterm. 
squash alif: Define more IRQ priorities.
squash alif/irq.h: Lower CSI priority.
squash alif: Update to work with latest DFP 1.3.0.

Systick:

pick alif/system_tick: Implement system_tick_wfe_with_timeout_us(). 
squash alif/system_tick: Use a UTIMER for system ticks and timing. 

Makefile:

pick alif: Misc Makefile updates and fixes.
squash alif/Makefile: Fix location of output build of alif-dfp files. 
squash alif/Makefile: Rename firmware.toc to firmware_toc.bin.

SE services:

pick alif/se_services: Add SE services interface, with rand64 func.
squash alif/se_services: Add services to reset SoC, get unique-id, dump info.
squash alif: Add more SE service calls, including CPU control services.
squash alif/se_services: SE services initialization race condition.
squash alif/se_services: Add SE select PLL source service call.
squash alif/services: Add off_profile.
squash alif/se_services: Fix UID.

ADC:

pick alif/machine_adc: Add basic ADC support.
squash alif/mcu: Add ADC12 information to pin struct.

Thanks @iabdalkader for looking at the commits. It's now down to 88 and I think I've applied almost all your suggestions.

Do you think we can squash down Support building the port for HE or HP or both into the main initial commit Add initial port to Alif Ensemble MCUs.?

I think you should squash these too on initial port or in a single "Add DCD driver" commit.

pick alif/tinyusb_port: Add Alif TinyUSB DCD driver.
squash alif/tinyusb_port: Disable USB IRQ on deinit.
squash alif/tinyusb_port: Implement SOF event.

Note that the second commit just disables USB IRQ in deinit and that all three precede the initial port, which doesn't make sense. Or maybe you want to see changes to the original DCD driver? In this case, should probably move them after initial port?

Could possibly also squash the following on initial port:

alif/mpconfigport: Enable os.urandom(). 
alif/modalif: Add alif.info() function. 
alif/modmachine: Implement machine.unique_id(), fix machine.reset(). 
alif/usbd: Implement proper USB serial number. 

Do you think we can squash down Support building the port for HE or HP or both into the main initial commit Add initial port to Alif Ensemble MCUs.?

Might be tricky, but we could probably squash the following commits:

squash alif: Add ToC config for dual images.
pick alif: Support building the port for HE or HP or both.
squash alif: Support running the port on the HE core.

In any case, I think it's good enough as it is.

Or maybe you want to see changes to the original DCD driver?

Yes, that's the intention. That driver is copied from upstream Alif GitHub and then modified by those two small commits.

I put this before the actual alif port commit because it's like adding the SDK and security toolkit submodules, only they are files rather than submodules. (They could go in lib/alif-tinuysb/* but I think it's simpler to have them live in the alif port directory.)

Could possibly also squash the following on initial port:

alif/mpconfigport: Enable os.urandom(). 
alif/modalif: Add alif.info() function. 
alif/modmachine: Implement machine.unique_id(), fix machine.reset(). 
alif/usbd: Implement proper USB serial number. 

They all rely on adding SE services, which is currently done as a separate commit.

Okay, then this should be good to go.

I now have this down to 76 commits after squashing a few smaller ones that were fixing things.

I also fixed cyw43 BLE on stm32 and mimxrt ports (they will continue to use the old drivers/cyw43/cywbt.c driver until after v1.25.0 is released).

Testing this on OPENMV_AE3, the WiFi cannot connect... not sure exactly what is wrong. It can associated with the AP but sending out a DHCP discover packet the board gets no response.

Testing this on OPENMV_AE3, the WiFi cannot connect... not sure exactly what is wrong. It can associated with the AP but sending out a DHCP discover packet the board gets no response.

I rebooted all the routers here and now it works...! I also tested the OPENMV_AE3 in STA mode connecting to a PYBD in AP mode, and that worked.

So, the WLAN support seems fine.

I tested it, seems to be working fine. However, I noticed this was dropped at some point:

 mp_obj_t machine_spi_make_new(const mp_obj_type_t *type, size_t n_args, size_t n_kw, const mp_obj_t *all_args) {
@@ -279,8 +284,6 @@
     machine_spi_obj_t *self = (machine_spi_obj_t *)self_in;
     // Disable all interrupts.
     spi_mask_interrupts(self->inst);
-    // Disable SCLK clock
-    spi_set_bus_speed(self->inst, 0, self->is_lp ? GetSystemCoreClock() : GetSystemAHBClock());

Not sure if it's important.

However, I noticed this was dropped at some point:

I had to remove that, it was causing a crash when deiniting the SPI due to a divide by zero.

Merged!!

Yay!

Congratulations, a wonderful effort & development. We posted about it
https://blog.adafruit.com/2025/04/09/micropythons-new-port-for-the-super-interesting-alif-ensemble-microcontroller/ to celebrate.

Note! This MCU blows others away. You have 11MB of RAM onboard at 3.2GB/s peak speed.

It's truly the first MCU where you can just abuse the heap using MicroPython.

Meaning, you can write code like how you would on the PC.

Note! This MCU blows others away. You have 11MB of RAM onboard at 3.2GB/s peak speed.

I'm really looking forward to experimenting with it! @kwagyeman do you know of any affordable development boards for the micro? I'm only aware of the official E7-based board which is ~$400, a little above the buy-to-experiment cost... (and I'm waiting in anticipation for an AE3!)

[Sorry, this is off-topic but a few folks have asked me the same question.]

We're kinda the first. So, the OpenMV N6 and AE3 Kickstarter is your best bet. We will probably ship the AE3 before December.

https://www.kickstarter.com/projects/openmv/openmv-n6-and-ae3-low-power-python-programmable-ai-cameras