esp32/Makefile: Add support for EIM environments. - #19657

Code size report:

Reference:  rp2: Keep machine.RTC ticking while in lightsleep(). [cc12057]
Comparison: mpy-cross/Makefile: Let mpy-cross build in EIM environments. [merge of b1f9119]
  mpy-cross:    +0 +0.000% 
   bare-arm:    +0 +0.000% 
minimal x86:    +0 +0.000% 
   unix x64:    +0 +0.000% standard
      stm32:    +0 +0.000% PYBV10
      esp32:    +0 +0.000% ESP32_GENERIC
     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

The issue with locally-built tools under EIM may have been found, see espressif/idf-im-ui#1067 (comment)

It'd be great if somebody could try this out on macOS after editing the PATH variable to remove the directories mentioned in the comment linked above, as EIM also installs its own version of clang. So I can at least update the documentation mentioning whether EIM can be used to build MicroPython on macOS :)

I've been told EIM does take over a default LLVM installation to use its own, but that's a Xtensa + RV32 cross compiler only.

So, unless Espressif updates its clang version to also target x86_64 and AArch64, this method is not usable on macOS or on systems where there's no GCC installed (*BSDs, maybe?).

Thinking a bit more about this, there may be a "cleaner" solution if Espressif can't (or won't) fix things on their end.

If $IDF_TOOLS_PATH is not set, congratulations, things will work as before. Otherwise:

• Make a copy of $PATH
• Strip away all entries that start with the value of $IDF_TOOLS_PATH
• Export the stripped version of $PATH
• Force a build of mpy-tool
• Export the original $PATH
• Build the firmware

then, in theory this should work for Linux, WSL, and macOS.

This should make it work under Linux, both using GCC and LLVM to build mpy-cross from an EIM environment. I assume this should also work on WSL, but it is not tested on macOS.

No idea about building this from Windows outside WSL, but hey, maybe they broke that too...

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 98.55%. Comparing base (cc12057) to head (b1f9119).

@@           Coverage Diff           @@
##           master   #19657   +/-   ##
=======================================
  Coverage   98.55%   98.55%           
=======================================
  Files         182      182           
  Lines       23335    23335           
  Branches        5        5           
=======================================
  Hits        22998    22998           
  Misses        336      336           
  Partials        1        1           

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

Octoprobe PR report

• Duration: 44min, 13124 tests run, 205ms/test
• Tested ports: esp32
• Summary Report , Logdirectory

Thanks for the review! I've updated the text where suggested and cleaned things up a bit.

The EIM installation section now also has some more information on what a working setup could look like, with what to select on certain EIM prompts.

ESP-IDF 5.3 and 5.4 set IDF_PATH as well, so the ESP32 Makefile was updated to check whether such a variable exists in the first place, and bail out if not as that indicates an environment that is probably not set up correctly. Following on that, there's also a check on whether IDF_PATH points to a valid path (as in, directory containing tools/idf.py, haven't gone further than that). There's more comments in the Makefile explaining that as well.

For mpy-cross, comments have been cleaned up, including mentioning a potential pitfall occurring if the compiler is located in a directory in the system PATH that has a space in the name. Extremely unlikely, but at least mentioned there for whoever will have to deal with that :)