Issue running with latest micropython

[sheindel]

Hi Russ! Would you be able to provide the CMAKE configs you are using for your binaries? We have successfully used your compiled binary for a generic ESP32 with 8MB SPIRAM for our ESP32-WROVER-E-N8R8 .

However, I wanted to make some tweaks to the base image.

I cloned

• v1.25.0 of micropython
• esp-idf v5.2.2 (per the MicroPython setup docs)
• the master branch of your repo

I had to make a tweak to one .cmake file in micropython, and I had to change the relative paths for the manifest. But I did get an image that all (seemingly) worked correctly with the latest micropython!

I used the following tweaked makecommands based on your documentation.

make \
    BOARD=ESP32_GENERIC \
    BOARD_VARIANT=SPIRAM \
    USER_C_MODULES=../../../../gc9a01_mpy/src/micropython.cmake \
    FROZEN_MANIFEST=../../../../../../gc9a01_mpy/manifest.py \
    clean submodules all

make \
    BOARD=ESP32_GENERIC \
    BOARD_VARIANT=SPIRAM \
    USER_C_MODULES=../../../../gc9a01_mpy/src/micropython.cmake \
    FROZEN_MANIFEST=../../../../../../gc9a01_mpy/manifest.py \
    erase deploy

But I did still run into some runtime issues, specifically with the spi frequency. I had originally been driving it at machine.freq() // 2 per the docs, so 120Mhz when I set the clock frequency to 240Mhz. Unfortunately, with the above stated branches, it would give me a runtime crash and forced me to lower it to 26.7MHz.

So all that being said, I have a few (hopefully easy) asks

• Can you confirm the make configs you are using the build the above referenced image?
• Can you share your merge-bin config (assuming that is what you are using) to combine the bootloader, partition table, and application binaries into the unified binary above?
• Do you have any guesses/ideas on the SPI issue? This feels like it would just be a configuration issue on my part if this worked previously

Thanks in advance for any insight you have!

[sheindel]

For anyone interested, I adjusted the usermode.cmake to just not check the USER_C_MODULES path (obviously this isn't the best solution, I was just looking for a quick workaround). For some reason, it wasn't working as expected. I'm wondering if this is an issue with the make macros, which does make me question my entire build environment setup...

diff --git a/py/usermod.cmake b/py/usermod.cmake
index 4a8b99ff3..2d548e3f1 100644
--- a/py/usermod.cmake
+++ b/py/usermod.cmake
@@ -47,10 +47,6 @@ if (USER_C_MODULES)
             set(USER_C_MODULE_PATH "${USER_C_MODULE_PATH}/micropython.cmake")
         endif()
         # Confirm the provided path exists, show abspath if not to make it clearer to fix.
-        if (NOT EXISTS ${USER_C_MODULE_PATH})
-            get_filename_component(USER_C_MODULES_ABS "${USER_C_MODULE_PATH}" ABSOLUTE)
-            message(FATAL_ERROR "USER_C_MODULES doesn't exist: ${USER_C_MODULES_ABS}")
-        endif()
 
         message("Including User C Module(s) from ${USER_C_MODULE_PATH}")
         include(${USER_C_MODULE_PATH})