Arm backend: Fix Smollm2 model test

[gggekov]

• Export & lower the smollm2 via extensions/llm/export_llm
• Build the arm_executor_runner application
• Fix the propagation of select_ops_list in the CMakeLists.txt
• Test the application runs on FVP in fast mode

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

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/19812

• 📄 Preview Python docs built from this PR

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

❗ 2 Active SEVs

There are 2 currently active SEVs. If your PR is affected, please view them below:

• /easycla is not responding
• LF Fleet unable to register self-hosted runners with pytorch/pytorch and pytorch/test-infra

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: gggekov / name: George Gekov (d0401ab)

[github-actions]

This PR needs a release notes: label

If your change should be included in the release notes (i.e. would users of this library care about this change?), please use a label starting with release notes:. This helps us keep track and include your important work in the next release notes.

To add a label, you can comment to pytorchbot, for example
@pytorchbot label "release notes: none"

For more information, see
https://github.com/pytorch/pytorch/wiki/PyTorch-AutoLabel-Bot#why-categorize-for-release-notes-and-how-does-it-work.

[Copilot]

Pull request overview

This PR updates the Arm backend SmolLM2-135M test flow to export/lower via extension/llm/export_llm, build arm_executor_runner, and run on Corstone-320 FVP in fast mode, while also adjusting build scripts/CMake to better propagate selective-kernel configuration.

Changes:

• Update SmolLM2-135M Arm backend test to export a PTE, build arm_executor_runner with PTE preloaded at a fixed DDR address, and run it on FVP fast mode.
• Adjust arm_executor_runner CMake selective-oplist logic so user-provided EXECUTORCH_SELECT_OPS_LIST is not unconditionally cleared.
• Minor robustness tweaks to Arm scripts (Ethos-U scratch derivation output handling; conditional forwarding of extra CMake args).

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 1 comment.

File	Description
examples/arm/run.sh	Adjusts how Ethos-U scratch size is captured from PTE-derived output.
examples/arm/executor_runner/CMakeLists.txt	Changes selective-oplist propagation behavior for kernel registration.
backends/arm/test/test_arm_backend.sh	Reworks SmolLM2-135M test into export → build runner → FVP execution flow.
backends/arm/scripts/build_executorch.sh	Only appends extra CMake args when provided.

Comments suppressed due to low confidence (1)

examples/arm/executor_runner/CMakeLists.txt:356

• The else() branch no longer clears EXECUTORCH_SELECT_OPS_LIST, which can unintentionally keep a default or previously set list (e.g., from run.sh) even when the PTE has no undelegated ops. This contradicts the comment above (“If the pte contains no undelegated ops, use neither.”) and may register/build extra kernels unnecessarily. Consider only preserving the list for the ET_MODEL_PTE_ADDR flow (where auto-gen from model isn’t possible), and clearing it otherwise.

else()
  set(EXECUTORCH_SELECT_OPS_MODEL "")
  message(
    "gen_oplist: No non delagated ops was found in ${ET_PTE_FILE_PATH} no ops added to build"
  )
endif()

---

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

[Copilot]

Pull request overview

Copilot reviewed 4 out of 4 changed files in this pull request and generated no new comments.

Comments suppressed due to low confidence (1)

examples/arm/executor_runner/CMakeLists.txt:355

• In the else() branch, EXECUTORCH_SELECT_OPS_LIST is no longer cleared (which is needed to allow a user-provided list to propagate), but the log message still claims that no ops will be added. Also fix the typo "delagated" → "delegated". Consider making the message conditional on whether EXECUTORCH_SELECT_OPS_LIST is empty so the build output matches actual behavior.

else()
  set(EXECUTORCH_SELECT_OPS_MODEL "")
  message(
    "gen_oplist: No non delagated ops was found in ${ET_PTE_FILE_PATH} no ops added to build"
  )