Zephyr: Update to Zephyr 4.4

[zingo]

Update CI install and testing, samples and docs
Zephyr 4.4 also added support for ensemble_e8_dk
so it can be used out of the box instead of the
alif zephyr git fork.

Test plan

Mosted tested by Github CI , Alif readme tested by hand.

cc @digantdesai @freddan80 @per @oscarandersson8218 @mansnils @Sebastian-Larsson @robell @rascani

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/19807

• 📄 Preview Python docs built from this PR

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

❗ 1 Active SEVs

There are 1 currently active SEVs. If your PR is affected, please view them below:

• [ROCm] MI350 CI jobs will have longer queue times due to CI migration

❌ 3 New Failures, 6 Pending, 1 Unrelated Failure, 3 Unclassified Failures

As of commit a5e1281 with merge base 6795a41 ():

NEW FAILURES - The following jobs have failed:

• pull / test-arm-cortex-m-size-test (zephyr-preset) / linux-job (gh)
  undefined reference to _exit'`
• pull / test-sqnr-static-llm-qnn-linux (smollm2_135m) / linux-job (gh)
  RuntimeError: Command docker exec -t 1488604e32007ff5253db69f3de8549037f04e6ba8b168ad38f3c4ab645bafc2 /exec failed with exit code 92
• trunk / unittest-release / windows / windows-job (gh)
  Process completed with exit code 1.

UNCLASSIFIED FAILURES - DrCI could not classify the following jobs because the workflow did not run on the merge base. The failures may be pre-existing on trunk or introduced by this PR:

• Cadence Build & Test / hifi-build / hifi4 (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)
  Input required and not supplied: aws-region
• Cadence Build & Test / vision-build / vision (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)
  Input required and not supplied: aws-region
• trunk / test-models-macos-coreml (resnet50) / macos-job (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)
  ##[error]The operation was canceled.

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

• pull / test-qnn-testsuite-linux / test-backend-linux (qnn, models) / linux-job (gh) (matched linux rule in flaky-rules.json)
  The runner has received a shutdown signal. This can happen when the runner service is stopped, or a manually started runner is canceled.

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

[linux-foundation-easycla]

The committers listed above are authorized under a signed CLA.

• ✅ login: Copilot / name: Copilot Autofix powered by AI (11b9daa, 61d6819, bcf6a0b)
• ✅ login: zingo / name: Zingo Andersen (a6a7a36, 11b9daa, 61d6819, bcf6a0b)

[Copilot]

Pull request overview

Updates the ExecuTorch Zephyr integration to target Zephyr v4.4.0, refreshes the Zephyr setup flow used by CI/docs, and adds out-of-the-box support for the ensemble_e8_dk Alif E8 board variants.

Changes:

• Bump Zephyr manifest revision to v4.4.0 and switch setup docs/CI to west packages pip --install + west sdk install.
• Update Zephyr sample docs (FVP + Alif E8) and tutorials for Python 3.12+, new board naming, and improved flashing/run instructions.
• Improve Zephyr CI setup: shared workspace directory (zephyr_dev_root) and a local Zephyr SDK releases proxy fallback to mitigate rate limits/network flakiness.

Reviewed changes

Copilot reviewed 16 out of 17 changed files in this pull request and generated 3 comments.

Show a summary per file

File	Description
zephyr/samples/mv2-ethosu/README.md	Updates prerequisites (Python 3.12+), FVP env vars, and Alif E8 build/flash/run instructions.
zephyr/samples/mv2-ethosu/boards/ensemble_e8_dk_ae822fa0e5597ls0_rtss_hp.overlay	New DTS overlay to expose a larger contiguous SRAM region and enable Ethos-U for RTSS-HP.
zephyr/samples/mv2-ethosu/boards/ensemble_e8_dk_ae822fa0e5597ls0_rtss_hp.conf	New board config enabling Ethos-U and DMA-accessible PTE handling (HP).
zephyr/samples/mv2-ethosu/boards/ensemble_e8_dk_ae822fa0e5597ls0_rtss_he.overlay	New DTS overlay for RTSS-HE (larger SRAM + Ethos-U node).
zephyr/samples/mv2-ethosu/boards/ensemble_e8_dk_ae822fa0e5597ls0_rtss_he.conf	New board config enabling Ethos-U and DMA-accessible PTE handling (HE).
zephyr/samples/mv2-ethosu/boards/alif_e8_dk_ae822fa0e5597xx0_rtss_hp.conf	Removes old Alif E8 board config in favor of upstream ensemble_e8_dk.
zephyr/samples/hello-executorch/README.md	Adds safer LD_LIBRARY_PATH export for FVP usage.
zephyr/README.md	Updates to Zephyr 4.4 and adds explicit west packages + west sdk install steps (used by CI).
docs/source/zephyr_alif_tutorial.md	Updates Alif tutorial for Zephyr 4.4, Python 3.12+, new model naming, and expanded flashing guidance.
backends/arm/test/test_github_trunk_zephyr.sh	Adds --zephyr-dev-root and reuses a single setup across README runs.
backends/arm/scripts/toolchain_utils.sh	Updates Zephyr toolchain download URLs/checksums to SDK v1.0.1 assets.
.gitignore	Ignores the new zephyr_dev_root/ workspace and its backups.
.github/workflows/trunk.yml	Ensures Zephyr CI runs under Python 3.12 by creating/activating a dedicated conda env.
.claude/skills/zephyr/SKILL.md	Updates internal Zephyr skill doc to Zephyr 4.4 / west sdk install / west packages.
.ci/scripts/zephyr_sdk_release_proxy.py	Adds a local synthetic GitHub Releases API/proxy for Zephyr SDK assets with caching + checksum verification.
.ci/scripts/test_zephyr.sh	Refactors Zephyr setup to a configurable workspace, adds setup-only/skip-setup, and SDK proxy fallback.
.ci/docker/common/install_zephyr.sh	Removes pre-downloading Zephyr SDK from the image build step.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[zingo]

@psiddh & @rascani this probably need a Meta review.

[Copilot]

Pull request overview

Copilot reviewed 16 out of 17 changed files in this pull request and generated 4 comments.

[Copilot]

Pull request overview

Copilot reviewed 16 out of 17 changed files in this pull request and generated 3 comments.

[Copilot]

Pull request overview

Copilot reviewed 16 out of 17 changed files in this pull request and generated 2 comments.

[rascani]

It looks like there are a few zephyr jobs failing, can you take a look @zingo?

Also, I'd like to get a better understanding for the rationale behind the SDK caching/proxying. In general, I'd like to minimize downloads in CI (causes flakes), but to me the ideal is caching most downloads in the docker image.

[zingo]

Seem to run into this problem #20092

[Copilot]

Pull request overview

Copilot reviewed 18 out of 19 changed files in this pull request and generated 2 comments.

Comments suppressed due to low confidence (1)

docs/source/zephyr_alif_tutorial.md:391

• The text after the output block still claims all predictions show 0.0000 because mv2_untrained has random weights, but the tutorial now exports --model_name=mv2 and the example output shows non-zero scores. Update this paragraph to describe the mv2_untrained behavior without contradicting the current instructions/output.

I [executorch:main.cpp] Inference time: 30 ms
I [executorch:main.cpp] ========================================

All predictions show 0.0000 because mv2_untrained has random weights.

</details>

[Copilot]

Pull request overview

Copilot reviewed 18 out of 19 changed files in this pull request and generated 2 comments.

[Copilot]

Pull request overview

Copilot reviewed 18 out of 19 changed files in this pull request and generated 1 comment.

[Copilot]

Pull request overview

Copilot reviewed 18 out of 19 changed files in this pull request and generated 2 comments.