From 41ca68aa196d93e0ffecabb958ab4b8a4128fe3d Mon Sep 17 00:00:00 2001
From: Jake Stevens <jrstevens@meta.com>
Date: Fri, 30 Oct 2026 06:47:08 -0700
Subject: [PATCH] Add OSS CI for the Native CPU engine (#23937)

Summary:

Add a CPU job to the Native Engines workflow that builds `run_classification_cpu` and runs `native_cpu_engine_test` with XNNPACK and optimized kernels enabled. Update dependency triggers, documentation, and xplat mirrors.

Authored with OpenAI Codex.

Reviewed By: digantdesai

Differential Revision: D125722080
---
 .ci/scripts/test-native.sh        | 12 +++++++++++-
 .github/workflows/test-native.yml | 11 ++++++++++-
 backends/native/README.md         | 16 +++++++++-------
 3 files changed, 30 insertions(+), 9 deletions(-)

diff --git a/.ci/scripts/test-native.sh b/.ci/scripts/test-native.sh
index 2f08fc6dabf..6f3b4c21a01 100644
--- a/.ci/scripts/test-native.sh
+++ b/.ci/scripts/test-native.sh
@@ -7,7 +7,7 @@
 
 set -euxo pipefail
 
-component="${1:?Usage: test-native.sh <runtime|ggml|vulkan> [cmake options]}"
+component="${1:?Usage: test-native.sh <runtime|cpu|ggml|vulkan> [cmake options]}"
 shift
 build_dir="cmake-out/native-${component}"
 build_jobs="${CMAKE_BUILD_PARALLEL_LEVEL:-$(nproc)}"
@@ -19,6 +19,16 @@ case "${component}" in
     targets+=(native_runtime_test)
     test_regex='^native_runtime_test$'
     ;;
+  cpu)
+    cmake_args+=(
+      -DEXECUTORCH_BUILD_NATIVE_CPU=ON
+      -DEXECUTORCH_BUILD_XNNPACK=ON
+      -DEXECUTORCH_BUILD_KERNELS_OPTIMIZED=ON
+      -DEXECUTORCH_BUILD_NATIVE_VISION_RUNNER=ON
+    )
+    targets+=(native_cpu_engine_test run_classification_cpu)
+    test_regex='^native_cpu_engine_test$'
+    ;;
   ggml)
     cmake_args+=(
       -DEXECUTORCH_BUILD_NATIVE_GGML=ON
diff --git a/.github/workflows/test-native.yml b/.github/workflows/test-native.yml
index c5e0f0b7c76..c7a7aae2f0c 100644
--- a/.github/workflows/test-native.yml
+++ b/.github/workflows/test-native.yml
@@ -14,6 +14,8 @@ on:
       - install_requirements.sh
       - requirements-dev.txt
       - requirements-examples.txt
+      - backends/cpu/CMakeLists.txt
+      - backends/cpu/runtime/**
       - backends/native/**
       - backends/vulkan/CMakeLists.txt
       - backends/vulkan/cmake/**
@@ -27,13 +29,20 @@ on:
       - backends/vulkan/third-party/volk
       - backends/vulkan_shared/CMakeLists.txt
       - backends/vulkan_shared/runtime/**
+      - backends/xnnpack/CMakeLists.txt
+      - backends/xnnpack/cmake/**
       - backends/xnnpack/third-party/FXdiv
+      - backends/xnnpack/third-party/XNNPACK
       - backends/xnnpack/third-party/cpuinfo
       - backends/xnnpack/third-party/pthreadpool
       - examples/models/llama/tokenizer/**
       - extension/llm/tokenizers
       - extension/llm/export/**
       - exir/**
+      - configurations/**
+      - extension/memory_allocator/**
+      - extension/threadpool/**
+      - kernels/**
       - runtime/**
       - third-party/**
       - tools/cmake/**
@@ -67,7 +76,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        component: [runtime, ggml, vulkan]
+        component: [runtime, cpu, ggml, vulkan]
     defaults:
       run:
         shell: bash
diff --git a/backends/native/README.md b/backends/native/README.md
index 5f156dad590..4fce4717128 100644
--- a/backends/native/README.md
+++ b/backends/native/README.md
@@ -202,16 +202,18 @@ target_link_libraries(my_app PRIVATE native_api native_vulkan_engine)
 ```
 
 With `EXECUTORCH_BUILD_TESTS=ON`, `native_runtime_test` covers the shared runtime
-and API. `native_vulkan_passes_test` covers the Vulkan graph rewrites without
-requiring a GPU. When either runner below is enabled,
-`native_runner_utils_test` covers their shared runner utilities. Run these with
-`ctest --test-dir cmake-out -R native_`.
+and API. `native_cpu_engine_test` covers the CPU engine when
+`EXECUTORCH_BUILD_NATIVE_CPU` is enabled. `native_vulkan_passes_test` covers the
+Vulkan graph rewrites without requiring a GPU. When either runner below is
+enabled, `native_runner_utils_test` covers their shared runner utilities. Run these
+with `ctest --test-dir cmake-out -R native_`.
 
 The `Test Native Engines` CI workflow has separate Linux jobs for the shared
-runtime, GGML, and Vulkan, with each test suite run once. To reproduce with
+runtime, CPU, GGML, and Vulkan, with each test suite run once. To reproduce with
 dependencies installed, run `bash .ci/scripts/test-native.sh runtime`,
-`bash .ci/scripts/test-native.sh ggml`, or
-`bash .ci/scripts/test-native.sh vulkan` from the OSS checkout. The GGML job
+`bash .ci/scripts/test-native.sh cpu`, `bash .ci/scripts/test-native.sh ggml`,
+or `bash .ci/scripts/test-native.sh vulkan` from the OSS checkout. The CPU job
+runs `native_cpu_engine_test` and builds `run_classification_cpu`. The GGML job
 also runs `native_llm_runner_utils_test` and executes `run_hf_llm_ggml` on a tiny
 seeded Hugging Face Llama with a generated byte tokenizer. Its eight generated
 token IDs must match eager PyTorch, covering prefill and KV-cache decoding without