Update

[ghstack-poisoned]

Author: swolchok

.ci/scripts/setup-emscripten.sh

@@ -0,0 +1,13 @@
+
+set -ex
+
+install_emscripten() {
+    git clone https://github.com/emscripten-core/emsdk.git
+    pushd emsdk || return
+    ./emsdk install 4.0.10
+    ./emsdk activate 4.0.10
+    source ./emsdk_env.sh
+    popd || return
+}
+
+install_emscripten

.github/workflows/android-release-artifacts.yml

@@ -47,15 +47,13 @@ jobs:
     name: build-aar
     needs: check-if-aar-exists
     if: ${{ !github.event.pull_request.head.repo.fork }}
-    uses: pytorch/test-infra/.github/workflows/linux_job_v2.yml@release/2.7
+    uses: pytorch/test-infra/.github/workflows/linux_job_v2.yml@main
     secrets: inherit
     permissions:
       id-token: write
       contents: read
     with:
       secrets-env: EXECUTORCH_MAVEN_SIGNING_KEYID EXECUTORCH_MAVEN_SIGNING_PASSWORD EXECUTORCH_MAVEN_CENTRAL_PASSWORD EXECUTORCH_MAVEN_CENTRAL_USERNAME EXECUTORCH_MAVEN_SIGNING_GPG_KEY_CONTENTS
-      # As this job has access to Maven credential, run this on a fresh ephemeral runner
-      runner: ephemeral.linux.2xlarge
       docker-image: executorch-ubuntu-22.04-clang12-android
       submodules: 'recursive'
       ref: ${{ github.sha }}

.github/workflows/pull.yml

@@ -734,3 +734,31 @@ jobs:
 
         PYTHON_EXECUTABLE=python bash .ci/scripts/setup-openvino.sh
         PYTHON_EXECUTABLE=python bash .ci/scripts/test_openvino.sh
+
+  test-build-wasm-linux:
+    name: test-build-wasm-linux
+    uses: pytorch/test-infra/.github/workflows/linux_job_v2.yml@main
+    permissions:
+      id-token: write
+      contents: read
+    strategy:
+      fail-fast: false
+    with:
+      runner: linux.2xlarge
+      docker-image: executorch-ubuntu-22.04-clang12
+      submodules: 'recursive'
+      ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
+      timeout: 90
+      script: |
+        # The generic Linux job chooses to use base env, not the one setup by the image
+        CONDA_ENV=$(conda env list --json | jq -r ".envs | .[-1]")
+        conda activate "${CONDA_ENV}"
+
+        BUILD_TOOL="cmake"
+        PYTHON_EXECUTABLE=python bash .ci/scripts/setup-linux.sh --build-tool "${BUILD_TOOL}"
+
+        # Install Node.js and Emscripten
+        source .ci/scripts/setup-emscripten.sh
+
+        # Test selective build
+        PYTHON_EXECUTABLE=python bash examples/wasm/test_build_wasm.sh

.gitmodules

@@ -27,7 +27,7 @@
 	url = https://github.com/pytorch/cpuinfo.git
 [submodule "backends/xnnpack/third-party/pthreadpool"]
 	path = backends/xnnpack/third-party/pthreadpool
-	url = https://github.com/Maratyszcza/pthreadpool.git
+	url = https://github.com/google/pthreadpool.git
 [submodule "extension/llm/tokenizers"]
 	path = extension/llm/tokenizers
 	url = https://github.com/pytorch-labs/tokenizers.git

CMakeLists.txt

@@ -112,11 +112,12 @@ set(CMAKE_SKIP_BUILD_RPATH OFF)
 set(CMAKE_BUILD_WITH_INSTALL_RPATH ON)
 # Automatically add all linked folders that are NOT in the build directory to
 # the rpath (per library?)
-# TODO: Doesn't work for us right now because we are
-# not installing .so's into the correct locations. For example we have
-# libcustom_ops_aot_lib.so depending on _portable_lib.so, which was eventually
-# put under <site-packages>/executorch/extension/pybindings/ but this rpath is
-# not automatically added because at build time it seems `portable_lib` is being
+#
+# TODO: Doesn't work for us right now because we are not installing .so's into
+# the correct locations. For example we have libcustom_ops_aot_lib.so depending
+# on _portable_lib.so, which was eventually put under
+# <site-packages>/executorch/extension/pybindings/ but this rpath is not
+# automatically added because at build time it seems `portable_lib` is being
 # built under the same directory, so no extra rpath is being added. To properly
 # fix this we need to install `portable_lib` into the correct path.
 set(CMAKE_INSTALL_RPATH_USE_LINK_PATH ON)
@@ -321,8 +322,9 @@ if(EXECUTORCH_USE_CPP_CODE_COVERAGE)
            " -fprofile-instr-generate -fcoverage-mapping"
     )
   else()
-    message(FATAL_ERROR
-            "Code coverage for compiler ${CMAKE_CXX_COMPILER_ID} is unsupported"
+    message(
+      FATAL_ERROR
+        "Code coverage for compiler ${CMAKE_CXX_COMPILER_ID} is unsupported"
     )
   endif()
 endif()
@@ -633,8 +635,8 @@ if(EXECUTORCH_BUILD_PYBIND)
   endif()
 
   if(EXECUTORCH_BUILD_XNNPACK)
-    # need to explicitly specify XNNPACK and xnnpack-microkernels-prod here otherwise
-    # uses XNNPACK and microkernel-prod symbols from libtorch_cpu
+    # need to explicitly specify XNNPACK and xnnpack-microkernels-prod here
+    # otherwise uses XNNPACK and microkernel-prod symbols from libtorch_cpu
     list(APPEND _dep_libs xnnpack_backend XNNPACK xnnpack-microkernels-prod)
   endif()
 
@@ -726,6 +728,17 @@ if(EXECUTORCH_BUILD_EXECUTOR_RUNNER)
   endif()
   target_link_libraries(executor_runner ${_executor_runner_libs})
   target_compile_options(executor_runner PUBLIC ${_common_compile_options})
+
+  # Automatically set when using `emcmake cmake` for Wasm build.
+  if(EMSCRIPTEN)
+    # Directory of model pte files to embed in the wasm binary.
+    if(NOT DEFINED WASM_MODEL_DIR)
+      set(WASM_MODEL_DIR "${CMAKE_SOURCE_DIR}/models/")
+    endif()
+
+    set(CMAKE_EXECUTABLE_SUFFIX ".html")
+    target_link_options(executor_runner PUBLIC -sALLOW_MEMORY_GROWTH --embed-file "${WASM_MODEL_DIR}@/")
+  endif()
 endif()
 
 if(EXECUTORCH_BUILD_VULKAN)

backends/apple/coreml/CMakeLists.txt

@@ -104,7 +104,9 @@ if(APPLE)
 endif()
 
 add_library(coreml_util ${UTIL_SOURCES})
-target_include_directories(coreml_util PUBLIC ${CMAKE_CURRENT_SOURCE_DIR}/runtime/util)
+target_include_directories(
+  coreml_util PUBLIC ${CMAKE_CURRENT_SOURCE_DIR}/runtime/util
+)
 if(APPLE)
   target_link_libraries(coreml_util PRIVATE ${FOUNDATION_FRAMEWORK})
 endif()
@@ -119,21 +121,25 @@ install(
 
 # CoreML inmemoryfs
 
-set(
-  INMEMORYFS_SOURCES
-  runtime/inmemoryfs/inmemory_filesystem.cpp
-  runtime/inmemoryfs/memory_buffer.cpp
-  runtime/inmemoryfs/memory_stream.cpp
-  runtime/inmemoryfs/reversed_memory_stream.cpp
+set(INMEMORYFS_SOURCES
+    runtime/inmemoryfs/inmemory_filesystem.cpp
+    runtime/inmemoryfs/memory_buffer.cpp runtime/inmemoryfs/memory_stream.cpp
+    runtime/inmemoryfs/reversed_memory_stream.cpp
 )
 if(APPLE)
-  list(APPEND INMEMORYFS_SOURCES runtime/inmemoryfs/inmemory_filesystem_utils.mm)
+  list(APPEND INMEMORYFS_SOURCES
+       runtime/inmemoryfs/inmemory_filesystem_utils.mm
+  )
 endif()
 
 add_library(coreml_inmemoryfs ${INMEMORYFS_SOURCES})
-target_include_directories(coreml_inmemoryfs PUBLIC ${CMAKE_CURRENT_SOURCE_DIR}/runtime/inmemoryfs)
+target_include_directories(
+  coreml_inmemoryfs PUBLIC ${CMAKE_CURRENT_SOURCE_DIR}/runtime/inmemoryfs
+)
 if(APPLE)
-  target_link_libraries(coreml_inmemoryfs PRIVATE coreml_util ${FOUNDATION_FRAMEWORK})
+  target_link_libraries(
+    coreml_inmemoryfs PRIVATE coreml_util ${FOUNDATION_FRAMEWORK}
+  )
 endif()
 target_compile_options(coreml_inmemoryfs PUBLIC -fPIC)
 
@@ -148,17 +154,12 @@ install(
 
 if(EXECUTORCH_BUILD_PYBIND)
   pybind11_add_module(
-    executorchcoreml
-    SHARED
-      runtime/inmemoryfs/inmemory_filesystem_py.cpp
-      runtime/inmemoryfs/inmemory_filesystem_utils.cpp
+    executorchcoreml SHARED runtime/inmemoryfs/inmemory_filesystem_py.cpp
+    runtime/inmemoryfs/inmemory_filesystem_utils.cpp
   )
   target_link_libraries(
-    executorchcoreml
-    PRIVATE
-      coreml_util
-      coreml_inmemoryfs
-      nlohmann_json::nlohmann_json
+    executorchcoreml PRIVATE coreml_util coreml_inmemoryfs
+                             nlohmann_json::nlohmann_json
   )
   target_compile_options(executorchcoreml PUBLIC -fPIC)
 endif()
@@ -179,8 +180,12 @@ if(APPLE)
     coremldelegate PRIVATE ${CMAKE_CURRENT_SOURCE_DIR}/runtime/delegate
   )
   target_include_directories(coremldelegate PRIVATE ${PROJECT_SOURCE_DIR}/..)
-  target_include_directories(coremldelegate PRIVATE ${PROJECT_SOURCE_DIR}/runtime/core/portable_type/c10)
-  target_compile_definitions(coremldelegate PRIVATE C10_USING_CUSTOM_GENERATED_MACROS)
+  target_include_directories(
+    coremldelegate PRIVATE ${PROJECT_SOURCE_DIR}/runtime/core/portable_type/c10
+  )
+  target_compile_definitions(
+    coremldelegate PRIVATE C10_USING_CUSTOM_GENERATED_MACROS
+  )
 
   if(EXECUTORCH_BUILD_DEVTOOLS)
     target_sources(coremldelegate PRIVATE ${SDK_SOURCES} ${PROTOBUF_SOURCES})
@@ -200,13 +205,9 @@ if(APPLE)
 
   target_link_libraries(
     coremldelegate
-    PUBLIC  coreml_util
-            coreml_inmemoryfs
-    PRIVATE executorch_core
-            ${ACCELERATE_FRAMEWORK}
-            ${COREML_FRAMEWORK}
-            ${FOUNDATION_FRAMEWORK}
-            ${SQLITE_LIBRARY}
+    PUBLIC coreml_util coreml_inmemoryfs
+    PRIVATE executorch_core ${ACCELERATE_FRAMEWORK} ${COREML_FRAMEWORK}
+            ${FOUNDATION_FRAMEWORK} ${SQLITE_LIBRARY}
   )
 
   target_link_options_shared_lib(coremldelegate)
@@ -218,13 +219,8 @@ if(APPLE)
   endif()
 
   target_compile_options(
-    coremldelegate
-    PRIVATE
-      -fobjc-arc
-      -fno-exceptions
-      -x objective-c++
-      -Wno-null-character
-      -Wno-receiver-expr
+    coremldelegate PRIVATE -fobjc-arc -fno-exceptions -x objective-c++
+                           -Wno-null-character -Wno-receiver-expr
   )
 
   if(EXECUTORCH_BUILD_DEVTOOLS)

backends/arm/scripts/mlsdk_utils.sh

@@ -0,0 +1,121 @@
+#!/usr/bin/env bash
+# Copyright 2025 Arm Limited and/or its affiliates.
+#
+# This source code is licensed under the BSD-style license found in the
+# LICENSE file in the root directory of this source tree.
+
+set -euo pipefail
+
+# TODO
+mlsdk_manifest_url=""
+
+script_dir=$(cd -- "$( dirname -- "${BASH_SOURCE[0]}" )" &> /dev/null && pwd)
+
+source ${script_dir}/utils.sh
+
+usage() { echo "Usage: $0 [-u <mlsdk-manifest-url>]" 1>&2; exit 1; }
+
+while getopts ":u:" opt; do
+    case "${opt}" in
+        u)
+            mlsdk_manifest_url=${OPTARG}
+            ;;
+        *)
+            usage
+            ;;
+    esac
+done
+
+function download_ai_mlsdk_manifest() {
+    local _dada_dir="$1"
+
+    if [[ -z "${_dada_dir}" ]]; then
+        echo "Error: _dada_dir parameter missing?"
+        return 1
+    fi
+
+    if [[ -z "${mlsdk_manifest_url}" ]]; then
+        echo "Error: mlsdk_manifest_url parameter missing?"
+        return 1
+    fi
+
+    if [[ ! -d "${_dada_dir}" ]]; then
+        mkdir -p "$_dada_dir"
+        pushd "$_dada_dir" || exit 1
+
+        curl https://storage.googleapis.com/git-repo-downloads/repo > repo
+        chmod u+x repo
+        ./repo init  --no-repo-verify --depth=1  --manifest-url  ${mlsdk_manifest_url} -g model-converter,emulation-layer,vgf-library
+        ./repo sync
+
+        popd
+    fi
+}
+
+function setup_model_converter() {
+    local work_dir="$1"
+    local manifest_dir="$2"
+    local enable_vgf_lib="$3"
+    local enable_emulation_layer="$4"
+
+    if [[ -z "$work_dir" ]]; then
+        echo "Error: work_dir parameter is required."
+        return 1
+    fi
+
+    if [[ -z "$manifest_dir" ]]; then
+        echo "Error: manifest_dir parameter is required."
+        return 1
+    fi
+
+    mkdir -p "$work_dir"
+    pushd "$work_dir" || exit 1
+
+    download_ai_mlsdk_manifest ${manifest_dir}
+
+    pushd "$manifest_dir"
+
+    # model-converter
+    # TODO: Remove macOS patch after mlsdk fully supports macOS
+    if [[ "$(uname)" == "Darwin" ]]; then
+    sed -i '' '/^ *print(f"Unsupported host platform/ i\
+            if system == "Darwin":\
+                # Use default Apple toolchain (Clang) on macOS\
+                return True\
+\
+' sw/model-converter/scripts/build.py
+    fi
+    python sw/model-converter/scripts/build.py -j$(nproc)
+
+    # libvgf
+    if [[ "${enable_vgf_lib}" -eq 1 ]]; then
+    # TODO: Remove macOS patch after mlsdk fully supports macOS
+        if [[ "$(uname)" == "Darwin" ]]; then
+    sed -i '' '/^ *print(f"ERROR: Unsupported host platform/ i\
+            if system == "Darwin":\
+                # Use default Apple toolchain (Clang) on macOS\
+                return True\
+\
+' sw/vgf-lib/scripts/build.py
+        fi
+        python sw/vgf-lib/scripts/build.py -j$(nproc)
+    fi
+
+    # emu layer
+    if [[ "${enable_emulation_layer}" -eq 1 ]]; then
+        pushd sw/emulation-layer
+        cmake -B build                                               \
+            -DGLSLANG_PATH=../../dependencies/glslang                \
+            -DSPIRV_CROSS_PATH=../../dependencies/SPIRV-Cross        \
+            -DSPIRV_HEADERS_PATH=../../dependencies/SPIRV-Headers    \
+            -DSPIRV_TOOLS_PATH=../../dependencies/SPIRV-Tools        \
+            -DVULKAN_HEADERS_PATH=../../dependencies/Vulkan-Headers
+        cmake --build build
+        popd
+    fi
+
+    popd
+}
+
+#setup_model_converter() $1
+# `"$manifest_dir"'

backends/arm/tosa_quant_utils.py

@@ -146,7 +146,15 @@ def quantize_value(self, x: torch.Tensor | float) -> Tensor:
         x = x.to(torch.float32)
         if self.per_channel:
             q_op = exir_ops.edge.quantized_decomposed.quantize_per_channel.default
-            args = (x, self.scale, self.zp, self.axis, self.qmin, self.qmax, self.dtype)
+            args = (
+                x,
+                torch.tensor(self.scale),
+                torch.tensor(self.zp),
+                self.axis,
+                self.qmin,
+                self.qmax,
+                self.dtype,
+            )
         else:
             q_op = exir_ops.edge.quantized_decomposed.quantize_per_tensor.default
             args = (x, self.scale, self.zp, self.qmin, self.qmax, self.dtype)  # type: ignore[assignment]
@@ -162,8 +170,8 @@ def dequantize_value(self, qx: torch.Tensor) -> torch.Tensor:
             dq_op = exir_ops.edge.quantized_decomposed.dequantize_per_channel.default
             args = (
                 qx,
-                self.scale,
-                self.zp,
+                torch.tensor(self.scale),
+                torch.tensor(self.zp),
                 self.axis,
                 self.qmin,
                 self.qmax,