Arm Zephyr cmake Preset

[BujSet]

Summary

This change adds a cmake preset for Arm Zephyr RTOS toolchain. This requires the toolchain cmake compilation target to be added to PATH (see #12077 for more details). This change includes update to CI flows, modifying the CI build_size_test to run the test using either arm bare metal or Zephyr toolchains. Additionally, the CI for building presets is updated to target the new cmake Zephyr preset.

Test plan

Tested with test/build_size_test.sh to determine the correct threshold for running the size test for the arm zephyr toolchain. When the CI runs the size test for the arm zephyr toolchain, the new cmake preset is used directly. The following commands are run by the CI, and can be manually invoked to confirm functionality:

CXXFLAGS="-fno-exceptions -fno-rtti -Wall -Werror -Wno-int-in-bool-context -DET_HAVE_PREAD=0" cmake --preset zephyr -DCMAKE_BUILD_TYPE=Release -DEXECUTORCH_OPTIMIZE_SIZE=ON -DCMAKE_INSTALL_PREFIX=cmake-out -Bcmake-out .
cmake --build cmake-out -j9 --target install --config Release
CXXFLAGS="-fno-exceptions -fno-rtti -Wall -Werror -Wno-int-in-bool-context -DET_HAVE_PREAD=0"  cmake -DCMAKE_BUILD_TYPE=Release -DCMAKE_TOOLCHAIN_FILE=$(realpath examples/zephyr/x86_64-linux-arm-zephyr-eabi-gcc.cmake) -DCMAKE_BUILD_TYPE=Release -DCMAKE_INSTALL_PREFIX=cmake-out -Bcmake-out/test test
cmake --build cmake-out/test -j9 --config Release
arm-zephyr-eabi-strip cmake-out/test/size_test
ls -la cmake-out/test/size_test

Output should corroborate CI test threshold, reporting a resultant size_test binary size of ~125Kb (threshold set to 130KB toin CI test tolerate variability).

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/11923

• 📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

❌ 1 Cancelled Job, 3 Unrelated Failures

As of commit f404ad8 with merge base de24e18 ():

CANCELLED JOB - The following job was cancelled. Please retry:

• pull / test-models-linux (mobilebert, xnnpack-quantization-delegation, linux.2xlarge) / linux-job (gh)
  ##[error]The operation was canceled.

FLAKY - The following job failed but was likely due to flakiness present on trunk:

• trunk / test-models-linux-aarch64 (mobilebert, portable, linux.arm64.2xlarge) / linux-job (gh) (similar failure)
  ##[error]The operation was canceled.

BROKEN TRUNK - The following jobs failed but was present on the merge base:

👉 Rebase onto the `viable/strict` branch to avoid these failures

• pull / test-models-linux (mobilebert, portable, linux.2xlarge) / linux-job (gh) (trunk failure)
  ##[error]The operation was canceled.
• trunk / test-models-linux-aarch64 (mobilebert, xnnpack-quantization-delegation, linux.arm64.2xlarge) / linux-job (gh) (trunk failure)
  ##[error]The operation was canceled.

This comment was automatically generated by Dr. CI and updates every 15 minutes.

[BujSet]

@pytorchbot label "release notes: none"

[digantdesai]

examples/arm/ethos-u-setup/arm-none-eabi-gcc.cmake can we overload this file, and introduce perhaps PREFIX or TRIPLET i.e. arm-none-eabi- vs. arm-zephyr-eabi-?

See $ANDROID_NDK/build/cmake/android.toolchain.cmake as an example.

[BujSet]

Hm, I'm been trying to make changes to support this, but in doing so, the changes are becoming quite a bit more invasive. Every arm script (e.g. examples/arm/setup.sh, examples/arm/run.sh, backends/arm/scripts/build_executorch.sh ...) would need to take additional flags to use the arm-none-eabi-gcc.cmake but overloaded with a flag to specify that we shouldn't use the bare metal toolchain, but use the zephyr one instead. It is a bit of file duplication, but changes to the existing scripts are much more minimal if we retain arm-zephyr-eabi-gcc.cmake file separately. What do you think? I'm of the preference of keeping the arm-zephyr-eabi-gcc.cmake file for now, and maybe a later PR could consolidate the two?

[digantdesai]

@zingo - FYI