More detail to python setup, and fixed small typo (#4247)

- it's unused
- valgrind is basically a linux-only software.
  it isn't a good idea to make it a hard requirement.
  if we want to use valgrind, it's better to introduce
  a separate option to control it.

.dockerignore

@@ -1,16 +1,12 @@
 # for now, it is used to speed up wasi-nn tests only.
 # you shall adapt below rules to incoming requirements
 
-build
-*/build
-*/*/build
-*/*/*/build
-*/*/*/*/build
-*/*/*/*/*/build
-*/*/*/*/*/*/build
+**/build
+**/tmp.*
 .*
+**/*.gguf
 
-core/deps
-!core/deps/tensorflow-src
-samples
-tests
+/core/deps/
+!/core/deps/tensorflow-src
+/samples
+/tests

.github/actions/install-linux-sgx/action.yml

@@ -0,0 +1,47 @@
+# Copyright (C) 2019 Intel Corporation.  All rights reserved.
+# SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+
+# Always follow https://download.01.org/intel-sgx/latest/linux-latest/docs/
+
+name: "Install Intel SGX SDK"
+description: "Installs the Intel SGX SDK and necessary libraries for Ubuntu."
+author: "Intel Corporation"
+inputs:
+  os:
+    description: "Operating system to install on (ubuntu-22.04)"
+    required: true
+
+runs:
+  using: "composite"
+  steps:
+    - name: Check Runner OS
+      if: ${{ inputs.os != 'ubuntu-22.04' }}
+      shell: bash
+      run: |
+        echo "::error title=⛔ error hint::Only support ubuntu-22.04 for now"
+        exit 1
+
+    - name: Create installation directory
+      shell: bash
+      run: sudo mkdir -p /opt/intel
+
+    - name: Download and install SGX SDK on ubuntu-22.04
+      if: ${{ inputs.os == 'ubuntu-22.04' }}
+      shell: bash
+      run: |
+        sudo wget -O sgx_linux_x64_sdk.bin https://download.01.org/intel-sgx/sgx-linux/2.25/distro/ubuntu22.04-server/sgx_linux_x64_sdk_2.25.100.3.bin
+        sudo chmod +x sgx_linux_x64_sdk.bin
+        echo 'yes' | sudo ./sgx_linux_x64_sdk.bin
+      working-directory: /opt/intel
+
+    - name: Add SGX repository and install libraries
+      shell: bash
+      run: |
+        echo "deb [arch=amd64] https://download.01.org/intel-sgx/sgx_repo/ubuntu $(lsb_release -cs) main" | sudo tee /etc/apt/sources.list.d/intel-sgx.list
+        wget -qO - https://download.01.org/intel-sgx/sgx_repo/ubuntu/intel-sgx-deb.key | sudo apt-key add -
+        sudo apt update
+        sudo apt install -y libsgx-launch libsgx-urts
+
+    - name: Source SGX SDK environment
+      shell: bash
+      run: source /opt/intel/sgxsdk/environment

.github/actions/install-wasi-sdk-wabt/action.yml

@@ -0,0 +1,80 @@
+# Copyright (C) 2019 Intel Corporation.  All rights reserved.
+# SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+
+# Get URLs from:
+#   - https://github.com/WebAssembly/wasi-sdk/releases
+#   - https://github.com/WebAssembly/wabt/releases
+
+# Install WASI-SDK and WABT at /opt
+# /opt is the assumed location widely used in the project
+name: Install WASI-SDK and WABT
+
+description: A composite action to download and install wasi-sdk and wabt on Ubuntu, macOS.
+
+inputs:
+  os:
+    description: "Operating system to install on (ubuntu, macos)"
+    required: true
+
+runs:
+  using: "composite"
+  steps:
+    - name: Check Runner OS
+      if: ${{ !startsWith(inputs.os, 'ubuntu') && !startsWith(inputs.os, 'windows') && !startsWith(inputs.os, 'macos') }}
+      shell: bash
+      run: |
+        echo "::error title=⛔ error hint::Support Ubuntu, Windows, and macOS Only"
+        exit 1
+
+    - name: Set up wasi-sdk and wabt on Ubuntu
+      if: ${{ startsWith(inputs.os, 'ubuntu') }}
+      shell: bash
+      run: |
+        sudo wget -O wasi-sdk.tar.gz --progress=dot:giga https://github.com/WebAssembly/wasi-sdk/releases/download/wasi-sdk-25/wasi-sdk-25.0-x86_64-linux.tar.gz
+        sudo tar -xf wasi-sdk.tar.gz
+        sudo ln -sf  wasi-sdk-25.0-x86_64-linux/ wasi-sdk
+        sudo wget -O wabt.tar.gz     --progress=dot:giga https://github.com/WebAssembly/wabt/releases/download/1.0.37/wabt-1.0.37-ubuntu-20.04.tar.gz
+        sudo tar -xf wabt.tar.gz
+        sudo ln -sf  wabt-1.0.37 wabt
+        /opt/wasi-sdk/bin/clang --version
+        /opt/wabt/bin/wasm-interp --version
+        echo "::notice::wasi-sdk-25 and wabt-1.0.37 installed on ubuntu"
+      working-directory: /opt
+
+    - name: Set up wasi-sdk and wabt on macOS-13 (intel)
+      if: ${{ inputs.os == 'macos-13' }}
+      shell: bash
+      run: |
+        sudo wget -O wasi-sdk.tar.gz --progress=dot:giga https://github.com/WebAssembly/wasi-sdk/releases/download/wasi-sdk-25/wasi-sdk-25.0-x86_64-macos.tar.gz
+        sudo tar -xf wasi-sdk.tar.gz
+        sudo ln -sf  wasi-sdk-25.0-x86_64-macos wasi-sdk
+        sudo wget -O wabt.tar.gz     --progress=dot:giga https://github.com/WebAssembly/wabt/releases/download/1.0.36/wabt-1.0.36-macos-12.tar.gz
+        sudo tar -xf wabt.tar.gz
+        sudo ln -sf  wabt-1.0.36 wabt
+        /opt/wasi-sdk/bin/clang --version
+        /opt/wabt/bin/wasm-interp --version
+        echo "::notice::wasi-sdk-25 and wabt-1.0.36 installed on macos-13"
+      working-directory: /opt
+
+    - name: Set up wasi-sdk and wabt on macOS-14 (arm64)
+      if: ${{ inputs.os == 'macos-14' }}
+      shell: bash
+      run: |
+        sudo wget -O wasi-sdk.tar.gz --progress=dot:giga https://github.com/WebAssembly/wasi-sdk/releases/download/wasi-sdk-25/wasi-sdk-25.0-arm64-macos.tar.gz
+        sudo tar -xf wasi-sdk.tar.gz
+        sudo ln -sf  wasi-sdk-25.0-arm64-macos wasi-sdk
+        sudo wget -O wabt.tar.gz     --progress=dot:giga https://github.com/WebAssembly/wabt/releases/download/1.0.37/wabt-1.0.37-macos-14.tar.gz
+        sudo tar -xf wabt.tar.gz
+        sudo ln -sf  wabt-1.0.37 wabt
+        /opt/wasi-sdk/bin/clang --version
+        /opt/wabt/bin/wasm-interp --version
+        echo "::notice::wasi-sdk-25 and wabt-1.0.37 installed on macos-14"
+      working-directory: /opt
+
+    #TODO: Add support for Windows
+    - name: Set up wasi-sdk and wabt on Windows
+      if: ${{ startsWith(inputs.os, 'windows') }}
+      shell: powershell
+      run: |
+        echo "::notice::Support for Windows is not implemented yet"
+        exit 1

.github/scripts/codeql_buildscript.sh

@@ -126,6 +126,16 @@ if [[ $? != 0 ]]; then
     exit 1;
 fi
 
+# build iwasm with multi-memory enabled
+cd ${WAMR_DIR}/product-mini/platforms/linux
+rm -rf build && mkdir build && cd build
+cmake .. -DCMAKE_BUILD_TYPE=Debug -DWAMR_BUILD_MULTI_MEMORY=1
+make -j
+if [[ $? != 0 ]]; then
+    echo "Failed to build iwasm with multi-memory enabled!"
+    exit 1;
+fi
+
 # build iwasm with hardware boundary check disabled
 cd ${WAMR_DIR}/product-mini/platforms/linux
 rm -rf build && mkdir build && cd build
@@ -280,3 +290,23 @@ if [[ $? != 0 ]]; then
     echo "Failed to build iwasm with linux perf support enabled!"
     exit 1;
 fi
+
+# build iwasm with shared heap enabled
+cd ${WAMR_DIR}/product-mini/platforms/linux
+rm -rf build && mkdir build && cd build
+cmake .. -DCMAKE_BUILD_TYPE=Debug -DWAMR_BUILD_SHARED_HEAP=1
+make -j
+if [[ $? != 0 ]]; then
+    echo "Failed to build iwasm with shared heap enabled!"
+    exit 1;
+fi
+
+# build iwasm with dynamic aot debug enabled
+cd ${WAMR_DIR}/product-mini/platforms/linux
+rm -rf build && mkdir build && cd build
+cmake .. -DCMAKE_BUILD_TYPE=Debug -DWAMR_BUILD_DYNAMIC_AOT_DEBUG=1
+make -j
+if [[ $? != 0 ]]; then
+    echo "Failed to build iwasm dynamic aot debug enabled!"
+    exit 1;
+fi

.github/workflows/build_docker_images.yml

@@ -15,9 +15,14 @@ on:
         type: string
         required: true
 
+permissions:
+  contents: read
+
 jobs:
   build-and-push-images:
     runs-on: ubuntu-22.04
+    permissions:
+      contents: write # for uploading release artifacts
 
     steps:
       - name: Checkout repository

.github/workflows/build_iwasm_release.yml

@@ -31,9 +31,78 @@ on:
         type: string
         required: false
 
+env:
+  DEFAULT_BUILD_OPTIONS: 
+    "-DWAMR_BUILD_AOT=1 -DWAMR_BUILD_INTERP=1 -DWAMR_BUILD_FAST_JIT=1 -DWAMR_BUILD_JIT=1 \
+     -DWAMR_BUILD_CUSTOM_NAME_SECTION=0 \
+     -DWAMR_BUILD_DEBUG_INTERP=0 \
+     -DWAMR_BUILD_DEBUG_AOT=0 \
+     -DWAMR_BUILD_DUMP_CALL_STACK=0 \
+     -DWAMR_BUILD_LIBC_UVWASI=0 \
+     -DWAMR_BUILD_LIBC_EMCC=0 \
+     -DWAMR_BUILD_LIB_RATS=0 \
+     -DWAMR_BUILD_LOAD_CUSTOM_SECTION=0 \
+     -DWAMR_BUILD_MEMORY_PROFILING=0 \
+     -DWAMR_BUILD_MINI_LOADER=0 \
+     -DWAMR_BUILD_MULTI_MODULE=0 \
+     -DWAMR_BUILD_PERF_PROFILING=0 \
+     -DWAMR_BUILD_SPEC_TEST=0 \
+     -DWAMR_BUILD_BULK_MEMORY=1 \
+     -DWAMR_BUILD_LIB_PTHREAD=1 \
+     -DWAMR_BUILD_LIB_PTHREAD_SEMAPHORE=1 \
+     -DWAMR_BUILD_LIB_WASI_THREADS=1 \
+     -DWAMR_BUILD_LIBC_BUILTIN=1 \
+     -DWAMR_BUILD_LIBC_WASI=1 \
+     -DWAMR_BUILD_REF_TYPES=1 \
+     -DWAMR_BUILD_SIMD=1 \
+     -DWAMR_BUILD_SHARED_MEMORY=1 \
+     -DWAMR_BUILD_TAIL_CALL=1 \
+     -DWAMR_BUILD_THREAD_MGR=1"
+  GC_EH_BUILD_OPTIONS:
+    "-DWAMR_BUILD_AOT=1 -DWAMR_BUILD_INTERP=1 -DWAMR_BUILD_FAST_INTERP=0 -DWAMR_BUILD_FAST_JIT=0 -DWAMR_BUILD_JIT=0 \
+     -DWAMR_BUILD_CUSTOM_NAME_SECTION=0 \
+     -DWAMR_BUILD_DEBUG_INTERP=0 \
+     -DWAMR_BUILD_DEBUG_AOT=0 \
+     -DWAMR_BUILD_DUMP_CALL_STACK=0 \
+     -DWAMR_BUILD_LIBC_UVWASI=0 \
+     -DWAMR_BUILD_LIBC_EMCC=0 \
+     -DWAMR_BUILD_LIB_RATS=0 \
+     -DWAMR_BUILD_LOAD_CUSTOM_SECTION=0 \
+     -DWAMR_BUILD_MEMORY_PROFILING=0 \
+     -DWAMR_BUILD_MINI_LOADER=0 \
+     -DWAMR_BUILD_MULTI_MODULE=0 \
+     -DWAMR_BUILD_PERF_PROFILING=0 \
+     -DWAMR_BUILD_SPEC_TEST=0 \
+     -DWAMR_BUILD_BULK_MEMORY=1 \
+     -DWAMR_BUILD_LIB_PTHREAD=1 \
+     -DWAMR_BUILD_LIB_PTHREAD_SEMAPHORE=1 \
+     -DWAMR_BUILD_LIB_WASI_THREADS=1 \
+     -DWAMR_BUILD_LIBC_BUILTIN=1 \
+     -DWAMR_BUILD_LIBC_WASI=1 \
+     -DWAMR_BUILD_REF_TYPES=1 \
+     -DWAMR_BUILD_SIMD=1 \
+     -DWAMR_BUILD_SHARED_MEMORY=1 \
+     -DWAMR_BUILD_TAIL_CALL=1 \
+     -DWAMR_BUILD_THREAD_MGR=1 \
+     -DWAMR_BUILD_EXCE_HANDLING=1 \
+     -DWAMR_BUILD_GC=1"
+
+permissions:
+  contents: read
+
 jobs:
   build:
     runs-on: ${{ inputs.runner }}
+    strategy:
+      matrix:
+        include:
+          - build_options: $DEFAULT_BUILD_OPTIONS
+            suffix: ''
+          - build_options: $GC_EH_BUILD_OPTIONS
+            suffix: '-gc-eh'
+    permissions:
+      contents: write # for uploading release artifacts
+
     steps:
       - uses: actions/checkout@v4
 
@@ -51,40 +120,25 @@ jobs:
           fail-on-cache-miss: true
 
       - name: generate iwasm binary release
+        shell: bash
         run: |
-          cmake -S . -B build \
-            -DWAMR_BUILD_AOT=1 -DWAMR_BUILD_INTERP=1 -DWAMR_BUILD_FAST_JIT=1 -DWAMR_BUILD_JIT=1 \
-            -DWAMR_BUILD_CUSTOM_NAME_SECTION=0 \
-            -DWAMR_BUILD_DEBUG_INTERP=0 \
-            -DWAMR_BUILD_DEBUG_AOT=0 \
-            -DWAMR_BUILD_DUMP_CALL_STACK=0 \
-            -DWAMR_BUILD_LIBC_UVWASI=0 \
-            -DWAMR_BUILD_LIBC_EMCC=0 \
-            -DWAMR_BUILD_LIB_RATS=0 \
-            -DWAMR_BUILD_LOAD_CUSTOM_SECTION=0 \
-            -DWAMR_BUILD_MEMORY_PROFILING=0 \
-            -DWAMR_BUILD_MINI_LOADER=0 \
-            -DWAMR_BUILD_MULTI_MODULE=0 \
-            -DWAMR_BUILD_PERF_PROFILING=0 \
-            -DWAMR_BUILD_SPEC_TEST=0 \
-            -DWAMR_BUILD_BULK_MEMORY=1 \
-            -DWAMR_BUILD_LIB_PTHREAD=1 \
-            -DWAMR_BUILD_LIB_PTHREAD_SEMAPHORE=1 \
-            -DWAMR_BUILD_LIB_WASI_THREADS=1 \
-            -DWAMR_BUILD_LIBC_BUILTIN=1 \
-            -DWAMR_BUILD_LIBC_WASI=1 \
-            -DWAMR_BUILD_REF_TYPES=1 \
-            -DWAMR_BUILD_SIMD=1 \
-            -DWAMR_BUILD_SHARED_MEMORY=1 \
-            -DWAMR_BUILD_TAIL_CALL=1 \
-            -DWAMR_BUILD_THREAD_MGR=1
+          cmake -S . -B build ${{ matrix.build_options }}
           cmake --build build --config Release --parallel 4
         working-directory: ${{ inputs.cwd }}
 
-      - name: compress the binary
+      - name: Compress the binary on Windows
+        if: inputs.runner == 'windows-latest'
         run: |
-          tar czf iwasm-${{ inputs.ver_num }}-${{ inputs.runner }}.tar.gz iwasm
-          zip iwasm-${{ inputs.ver_num }}-${{ inputs.runner }}.zip iwasm
+          tar -czf iwasm${{ matrix.suffix }}-${{ inputs.ver_num }}-${{ inputs.runner }}.tar.gz iwasm.exe
+          Compress-Archive -Path iwasm.exe -DestinationPath iwasm${{ matrix.suffix }}-${{ inputs.ver_num }}-${{ inputs.runner }}.zip
+          mv iwasm${{ matrix.suffix }}-${{ inputs.ver_num }}-${{ inputs.runner }}.* ../
+        working-directory: ${{ inputs.cwd }}/build/Release
+
+      - name: compress the binary on non-Windows
+        if: inputs.runner != 'windows-latest'
+        run: |
+          tar czf iwasm${{ matrix.suffix }}-${{ inputs.ver_num }}-${{ inputs.runner }}.tar.gz iwasm
+          zip iwasm${{ matrix.suffix }}-${{ inputs.ver_num }}-${{ inputs.runner }}.zip iwasm
         working-directory: ${{ inputs.cwd }}/build
 
       - name: upload release tar.gz
@@ -93,8 +147,8 @@ jobs:
           GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
         with:
           upload_url: ${{ inputs.upload_url }}
-          asset_path: ${{ inputs.cwd }}/build/iwasm-${{ inputs.ver_num }}-${{ inputs.runner }}.tar.gz
-          asset_name: iwasm-${{ inputs.ver_num }}-${{ inputs.arch }}-${{ inputs.runner }}.tar.gz
+          asset_path: ${{ inputs.cwd }}/build/iwasm${{ matrix.suffix }}-${{ inputs.ver_num }}-${{ inputs.runner }}.tar.gz
+          asset_name: iwasm${{ matrix.suffix }}-${{ inputs.ver_num }}-${{ inputs.arch }}-${{ inputs.runner }}.tar.gz
           asset_content_type: application/x-gzip
 
       - name: upload release zip
@@ -103,6 +157,6 @@ jobs:
           GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
         with:
           upload_url: ${{ inputs.upload_url }}
-          asset_path: ${{ inputs.cwd }}/build/iwasm-${{ inputs.ver_num }}-${{ inputs.runner }}.zip
-          asset_name: iwasm-${{ inputs.ver_num }}-${{ inputs.arch }}-${{ inputs.runner }}.zip
+          asset_path: ${{ inputs.cwd }}/build/iwasm${{ matrix.suffix }}-${{ inputs.ver_num }}-${{ inputs.runner }}.zip
+          asset_name: iwasm${{ matrix.suffix }}-${{ inputs.ver_num }}-${{ inputs.arch }}-${{ inputs.runner }}.zip
           asset_content_type: application/zip

.github/workflows/build_llvm_libraries.yml

@@ -27,6 +27,9 @@ on:
         description: "A cached key of LLVM libraries"
         value: ${{ jobs.build_llvm_libraries.outputs.key}}
 
+permissions:
+  contents: read
+
 jobs:
   build_llvm_libraries:
     runs-on: ${{ inputs.os }}
@@ -36,6 +39,9 @@ jobs:
       image: ${{ inputs.container_image }}
     outputs:
       key: ${{ steps.create_lib_cache_key.outputs.key}}
+    permissions:
+      contents: read
+      actions: write # for uploading cached artifact
 
     steps:
       - name: checkout
@@ -43,6 +49,7 @@ jobs:
 
       - name: install dependencies for non macos-14
         if: inputs.os != 'macos-14'
+        shell: bash
         run: /usr/bin/env python3 -m pip install -r requirements.txt
         working-directory: build-scripts
 
@@ -51,9 +58,13 @@ jobs:
         run: /usr/bin/env python3 -m pip install -r requirements.txt --break-system-packages
         working-directory: build-scripts
 
-      - name: retrieve the last commit ID
+      - name: Retrieve the last commit ID
         id: get_last_commit
-        run: echo "last_commit=$(GH_TOKEN=${{ secrets.GITHUB_TOKEN }} /usr/bin/env python3 ./build_llvm.py ${{ inputs.extra_build_llvm_options }} --llvm-ver)" >> $GITHUB_OUTPUT
+        env:
+          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+        shell: bash
+        run: |
+          echo "last_commit=$(GH_TOKEN=${{ secrets.GITHUB_TOKEN }} /usr/bin/env python3 ./build_llvm.py ${{ inputs.extra_build_llvm_options }} --llvm-ver)" >> $GITHUB_OUTPUT
         working-directory: build-scripts
 
       # Bump the prefix number to evict all previous caches and
@@ -62,7 +73,9 @@ jobs:
       # suspect.
       - name: form the cache key of libraries
         id: create_lib_cache_key
-        run: echo "key=0-llvm-libraries-${{ inputs.os }}-${{ inputs.arch }}-${{ steps.get_last_commit.outputs.last_commit }}${{ inputs.cache_key_suffix }}" >> $GITHUB_OUTPUT
+        shell: bash
+        run: |
+          echo "key=0-llvm-libraries-${{ inputs.os }}-${{ inputs.arch }}-${{ steps.get_last_commit.outputs.last_commit }}${{ inputs.cache_key_suffix }}" >> $GITHUB_OUTPUT
 
       - name: Cache LLVM libraries
         id: retrieve_llvm_libs
@@ -76,14 +89,6 @@ jobs:
             ./core/deps/llvm/build/share
           key: ${{ steps.create_lib_cache_key.outputs.key}}
 
-      - uses: actions/cache@v4
-        with:
-          path: ~/.ccache
-          key: 0-ccache-${{ inputs.os }}-${{ steps.get_last_commit.outputs.last_commit }}
-          restore-keys: |
-            0-ccache-${{ inputs.os }}
-        if: steps.retrieve_llvm_libs.outputs.cache-hit != 'true' && inputs.os == 'ubuntu-20.04'
-
       - uses: actions/cache@v4
         with:
           path: ~/.cache/ccache
@@ -107,7 +112,20 @@ jobs:
       - run: brew install ccache ninja
         if: steps.retrieve_llvm_libs.outputs.cache-hit != 'true' && startsWith(inputs.os, 'macos')
 
+      - uses: actions/cache@v4
+        with:
+          path: ~/.cache/ccache
+          key: 0-ccache-${{ inputs.os }}-${{ steps.get_last_commit.outputs.last_commit }}
+          restore-keys: |
+            0-ccache-${{ inputs.os }}
+        if: steps.retrieve_llvm_libs.outputs.cache-hit != 'true' && inputs.os == 'windows-latest'
+
+      # Install tools on Windows
+      - run: choco install -y ccache ninja
+        if: steps.retrieve_llvm_libs.outputs.cache-hit != 'true' && inputs.os == 'windows-latest'
+
       - name: Build LLVM libraries
         if: steps.retrieve_llvm_libs.outputs.cache-hit != 'true'
+        shell: bash
         run: /usr/bin/env python3 ./build_llvm.py ${{ inputs.extra_build_llvm_options }} --arch ${{ inputs.arch }}
         working-directory: build-scripts

.github/workflows/build_wamr_lldb.yml

@@ -28,8 +28,13 @@ on:
         required: false
         default: "https://github.com/WebAssembly/wasi-sdk/releases/download/wasi-sdk-20/wasi-sdk-20.0-linux.tar.gz"
 
+permissions:
+  contents: read
+
 jobs:
   try_reuse:
+    permissions:
+      contents: write # for uploading release artifacts
     uses: ./.github/workflows/reuse_latest_release_binaries.yml
     with:
       binary_name_stem: "wamr-lldb-${{ inputs.ver_num }}-${{ inputs.arch }}-${{ inputs.runner }}"
@@ -46,6 +51,9 @@ jobs:
       PYTHON_VERSION: '3.10'
       PYTHON_UBUNTU_STANDALONE_BUILD: https://github.com/indygreg/python-build-standalone/releases/download/20230507/cpython-3.10.11+20230507-x86_64-unknown-linux-gnu-install_only.tar.gz
       PYTHON_MACOS_STANDALONE_BUILD: https://github.com/indygreg/python-build-standalone/releases/download/20230507/cpython-3.10.11+20230507-x86_64-apple-darwin-install_only.tar.gz
+    permissions:
+      contents: write # for uploading release artifacts
+
     steps:
       - uses: actions/checkout@v4
 

.github/workflows/build_wamr_sdk.yml

@@ -35,9 +35,15 @@ on:
         type: string
         required: true
 
+permissions:
+  contents: read
+
 jobs:
   build:
     runs-on: ${{ inputs.runner }}
+    permissions:
+      contents: write # for uploading release artifacts
+
     steps:
       - uses: actions/checkout@v4
 

.github/workflows/build_wamr_vscode_ext.yml

@@ -14,9 +14,15 @@ on:
         type: string
         required: true
 
+permissions:
+  contents: read
+
 jobs:
   build:
     runs-on: ubuntu-22.04
+    permissions:
+      contents: write # for uploading release artifacts
+
     steps:
       - uses: actions/checkout@v4
 

.github/workflows/build_wamrc.yml

@@ -31,9 +31,15 @@ on:
         type: string
         required: false
 
+permissions:
+  contents: read
+
 jobs:
   build:
     runs-on: ${{ inputs.runner }}
+    permissions:
+      contents: write # for uploading release artifacts
+
     steps:
       - uses: actions/checkout@v4
 
@@ -56,8 +62,16 @@ jobs:
           cmake --build build --config Release --parallel 4
         working-directory: wamr-compiler
 
-      - name: compress the binary
-        if: inputs.release
+      - name: Compress the binary on Windows
+        if: inputs.runner == 'windows-latest' && inputs.release
+        run: |
+          tar -czf wamrc-${{ inputs.ver_num }}-${{ inputs.runner }}.tar.gz wamrc.exe
+          Compress-Archive -Path wamrc.exe -DestinationPath wamrc-${{ inputs.ver_num }}-${{ inputs.runner }}.zip
+          mv wamrc-${{ inputs.ver_num }}-${{ inputs.runner }}.* ../
+        working-directory: wamr-compiler/build/Release
+
+      - name: compress the binary on non-Windows
+        if: inputs.runner != 'windows-latest' && inputs.release
         run: |
           tar czf wamrc-${{ inputs.ver_num }}-${{ inputs.runner }}.tar.gz wamrc
           zip wamrc-${{ inputs.ver_num }}-${{ inputs.runner }}.zip wamrc

.github/workflows/check_version_h.yml

@@ -0,0 +1,29 @@
+# Copyright (C) 2019 Intel Corporation.  All rights reserved.
+# SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+name: confirm version.h stay in sync
+
+on:
+  workflow_call:
+
+permissions:
+  contents: read
+
+jobs:
+  confirm_version:
+    runs-on: ubuntu-latest
+
+    steps:
+      - name: checkout
+        uses: actions/checkout@v4
+
+      - name: cmake execute to generate version.h
+        run: cmake -B build_version -S .
+
+      - name: confirm version.h
+        run: |
+          if [ -z "$(git status --porcelain | grep version.h)" ]; then
+            echo "version.h is in sync"
+          else
+            echo "version.h is not in sync"
+            exit 1
+          fi

.github/workflows/codeql.yml

@@ -19,6 +19,9 @@ on:
   # allow to be triggered manually
   workflow_dispatch:
 
+permissions:
+  contents: read
+
 jobs:
   analyze:
     if: github.repository == 'bytecodealliance/wasm-micro-runtime'
@@ -30,10 +33,6 @@ jobs:
     # Consider using larger runners for possible analysis time improvements.
     runs-on: ${{ (matrix.language == 'swift' && 'macos-13') || 'ubuntu-22.04' }}
     timeout-minutes: ${{ (matrix.language == 'swift' && 120) || 360 }}
-    permissions:
-      actions: read
-      contents: read
-      security-events: write
 
     strategy:
       fail-fast: false
@@ -41,6 +40,11 @@ jobs:
         language: [ 'cpp' ]
         # CodeQL supports [ 'cpp', 'csharp', 'go', 'java', 'javascript', 'python', 'ruby', 'swift' ]
 
+    permissions:
+      contents: read
+      actions: read
+      security-events: write
+
     steps:
     - name: Checkout repository
       uses: actions/checkout@v3
@@ -49,7 +53,7 @@ jobs:
 
     # Initializes the CodeQL tools for scanning.
     - name: Initialize CodeQL
-      uses: github/codeql-action/init@v3
+      uses: github/codeql-action/init@v3.28.17
       with:
         languages: ${{ matrix.language }}
 
@@ -66,7 +70,7 @@ jobs:
     - run: |
         ./.github/scripts/codeql_buildscript.sh
     - name: Perform CodeQL Analysis
-      uses: github/codeql-action/analyze@v3
+      uses: github/codeql-action/analyze@v3.28.17
       with:
         category: "/language:${{matrix.language}}"
         upload: false
@@ -95,14 +99,14 @@ jobs:
         output: ${{ steps.step1.outputs.sarif-output }}/cpp.sarif
 
     - name: Upload CodeQL results to code scanning
-      uses: github/codeql-action/upload-sarif@v3
+      uses: github/codeql-action/upload-sarif@v3.28.17
       with:
         sarif_file: ${{ steps.step1.outputs.sarif-output }}
         category: "/language:${{matrix.language}}"
 
     - name: Upload CodeQL results as an artifact
       if: success() || failure()
-      uses: actions/upload-artifact@v4
+      uses: actions/upload-artifact@v4.6.2
       with:
         name: codeql-results
         path: ${{ steps.step1.outputs.sarif-output }}

.github/workflows/coding_guidelines.yml

@@ -14,9 +14,12 @@ concurrency:
   group: ${{ github.workflow }}-${{ github.ref }}
   cancel-in-progress: true
 
+permissions:
+  contents: read
+
 jobs:
   compliance_job:
-    runs-on: ubuntu-20.04
+    runs-on: ubuntu-22.04
     steps:
       - name: checkout
         uses: actions/checkout@v4

.github/workflows/compilation_on_android_ubuntu.yml

@@ -68,9 +68,22 @@ env:
   WAMR_COMPILER_TEST_OPTIONS: "-s wamr_compiler -S -b -P"
   GC_TEST_OPTIONS: "-s spec -G -b -P"
   MEMORY64_TEST_OPTIONS: "-s spec -W -b -P"
+  MULTI_MEMORY_TEST_OPTIONS: "-s spec -E -b -P"
+
+permissions:
+  contents: read
 
 jobs:
+  check_version_h:
+    permissions:
+      contents: read
+      actions: write
+    uses: ./.github/workflows/check_version_h.yml
+
   build_llvm_libraries_on_ubuntu_2204:
+    permissions:
+      contents: read
+      actions: write
     uses: ./.github/workflows/build_llvm_libraries.yml
     with:
       os: "ubuntu-22.04"
@@ -145,9 +158,12 @@ jobs:
             "-DWAMR_BUILD_PERF_PROFILING=1",
             "-DWAMR_BUILD_REF_TYPES=1",
             "-DWAMR_BUILD_SIMD=1",
+            "-DWAMR_BUILD_LIB_SIMDE=1",
             "-DWAMR_BUILD_TAIL_CALL=1",
             "-DWAMR_DISABLE_HW_BOUND_CHECK=1",
             "-DWAMR_BUILD_MEMORY64=1",
+            "-DWAMR_BUILD_MULTI_MEMORY=1",
+            "-DWAMR_BUILD_SHARED=1",
           ]
         os: [ubuntu-22.04]
         platform: [android, linux]
@@ -163,11 +179,9 @@ jobs:
             make_options_feature: "-DWAMR_BUILD_MULTI_MODULE=1"
           - make_options_run_mode: $MULTI_TIER_JIT_BUILD_OPTIONS
             make_options_feature: "-DWAMR_BUILD_MULTI_MODULE=1"
-          # SIMD only on JIT/AOT mode
+          # SIMD only on JIT/AOT/fast interpreter mode
           - make_options_run_mode: $CLASSIC_INTERP_BUILD_OPTIONS
             make_options_feature: "-DWAMR_BUILD_SIMD=1"
-          - make_options_run_mode: $FAST_INTERP_BUILD_OPTIONS
-            make_options_feature: "-DWAMR_BUILD_SIMD=1"
           # DEBUG_INTERP only on CLASSIC INTERP mode
           - make_options_run_mode: $AOT_BUILD_OPTIONS
             make_options_feature: "-DWAMR_BUILD_DEBUG_INTERP=1"
@@ -206,11 +220,9 @@ jobs:
             make_options_feature: "-DWAMR_BUILD_MINI_LOADER=1"
           - make_options_run_mode: $MULTI_TIER_JIT_BUILD_OPTIONS
             make_options_feature: "-DWAMR_BUILD_MINI_LOADER=1"
-          # Memory64 only on CLASSIC INTERP mode, and only on 64-bit platform
+          # Memory64 only on CLASSIC INTERP and AOT mode, and only on 64-bit platform
           - make_options_feature: "-DWAMR_BUILD_MEMORY64=1"
             platform: android
-          - make_options_run_mode: $AOT_BUILD_OPTIONS
-            make_options_feature: "-DWAMR_BUILD_MEMORY64=1"
           - make_options_run_mode: $FAST_INTERP_BUILD_OPTIONS
             make_options_feature: "-DWAMR_BUILD_MEMORY64=1"
           - make_options_run_mode: $FAST_JIT_BUILD_OPTIONS
@@ -221,6 +233,21 @@ jobs:
             make_options_feature: "-DWAMR_BUILD_MEMORY64=1"
           - make_options_run_mode: $MULTI_TIER_JIT_BUILD_OPTIONS
             make_options_feature: "-DWAMR_BUILD_MEMORY64=1"
+          # Multi memory only on CLASSIC INTERP mode, and only on 64-bit platform
+          - make_options_feature: "-DWAMR_BUILD_MEMORY64=1"
+            platform: android
+          - make_options_run_mode: $AOT_BUILD_OPTIONS
+            make_options_feature: "-DWAMR_BUILD_MULTI_MEMORY=1"
+          - make_options_run_mode: $FAST_INTERP_BUILD_OPTIONS
+            make_options_feature: "-DWAMR_BUILD_MULTI_MEMORY=1"
+          - make_options_run_mode: $FAST_JIT_BUILD_OPTIONS
+            make_options_feature: "-DWAMR_BUILD_MULTI_MEMORY=1"
+          - make_options_run_mode: $LLVM_LAZY_JIT_BUILD_OPTIONS
+            make_options_feature: "-DWAMR_BUILD_MULTI_MEMORY=1"
+          - make_options_run_mode: $LLVM_EAGER_JIT_BUILD_OPTIONS
+            make_options_feature: "-DWAMR_BUILD_MULTI_MEMORY=1"
+          - make_options_run_mode: $MULTI_TIER_JIT_BUILD_OPTIONS
+            make_options_feature: "-DWAMR_BUILD_MULTI_MEMORY=1"
           # Fast-JIT and Multi-Tier-JIT mode don't support android
           - make_options_run_mode: $FAST_JIT_BUILD_OPTIONS
             platform: android
@@ -232,6 +259,9 @@ jobs:
             platform: android
           - make_options_run_mode: $LLVM_EAGER_JIT_BUILD_OPTIONS
             platform: android
+          # android does not support WAMR_BUILD_SHARED in its CMakeLists.txt.
+          - make_options_feature: "-DWAMR_BUILD_SHARED=1"
+            platform: android
         include:
           - os: ubuntu-22.04
             llvm_cache_key: ${{ needs.build_llvm_libraries_on_ubuntu_2204.outputs.cache_key }}
@@ -285,17 +315,10 @@ jobs:
       fail-fast: false
       matrix:
         os: [ubuntu-22.04]
-        wasi_sdk_release:
-          [
-            "https://github.com/WebAssembly/wasi-sdk/releases/download/wasi-sdk-20/wasi-sdk-20.0-linux.tar.gz",
-          ]
-        wabt_release:
-          [
-            "https://github.com/WebAssembly/wabt/releases/download/1.0.31/wabt-1.0.31-ubuntu.tar.gz",
-          ]
         include:
           - os: ubuntu-22.04
             llvm_cache_key: ${{ needs.build_llvm_libraries_on_ubuntu_2204.outputs.cache_key }}
+
     steps:
       - name: checkout
         uses: actions/checkout@v4
@@ -316,19 +339,10 @@ jobs:
         if: (steps.retrieve_llvm_libs.outputs.cache-hit != 'true')
         run: echo "::error::can not get prebuilt llvm libraries" && exit 1
 
-      - name: download and install wasi-sdk
-        run: |
-          cd /opt
-          sudo wget ${{ matrix.wasi_sdk_release }}
-          sudo tar -xzf wasi-sdk-*.tar.gz
-          sudo ln -sf wasi-sdk-20.0 wasi-sdk
-
-      - name: download and install wabt
-        run: |
-          cd /opt
-          sudo wget ${{ matrix.wabt_release }}
-          sudo tar -xzf wabt-1.0.31-*.tar.gz
-          sudo mv wabt-1.0.31 wabt
+      - name: install-wasi-sdk-wabt
+        uses: ./.github/actions/install-wasi-sdk-wabt
+        with:
+          os: ${{ matrix.os }}
 
       - name: Build wamrc
         run: |
@@ -367,14 +381,6 @@ jobs:
             $MULTI_TIER_JIT_BUILD_OPTIONS,
           ]
         os: [ubuntu-22.04]
-        wasi_sdk_release:
-          [
-            "https://github.com/WebAssembly/wasi-sdk/releases/download/wasi-sdk-20/wasi-sdk-20.0-linux.tar.gz",
-          ]
-        wabt_release:
-          [
-            "https://github.com/WebAssembly/wabt/releases/download/1.0.31/wabt-1.0.31-ubuntu.tar.gz",
-          ]
         include:
           - os: ubuntu-22.04
             llvm_cache_key: ${{ needs.build_llvm_libraries_on_ubuntu_2204.outputs.cache_key }}
@@ -400,12 +406,10 @@ jobs:
         if: (!endsWith(matrix.make_options, '_INTERP_BUILD_OPTIONS')) && (steps.retrieve_llvm_libs.outputs.cache-hit != 'true')
         run: echo "::error::can not get prebuilt llvm libraries" && exit 1
 
-      - name: download and install wabt
-        run: |
-          cd /opt
-          sudo wget ${{ matrix.wabt_release }}
-          sudo tar -xzf wabt-1.0.31-*.tar.gz
-          sudo mv wabt-1.0.31 wabt
+      - name: install-wasi-sdk-wabt
+        uses: ./.github/actions/install-wasi-sdk-wabt
+        with:
+          os: ${{ matrix.os }}
 
       - name: Build wamrc
         if: (!endsWith(matrix.make_options, '_INTERP_BUILD_OPTIONS'))
@@ -434,34 +438,14 @@ jobs:
     strategy:
       matrix:
         os: [ubuntu-22.04]
-        wasi_sdk_release:
-          [
-            "https://github.com/WebAssembly/wasi-sdk/releases/download/wasi-sdk-20/wasi-sdk-20.0-linux.tar.gz",
-          ]
-        wabt_release:
-          [
-            "https://github.com/WebAssembly/wabt/releases/download/1.0.31/wabt-1.0.31-ubuntu.tar.gz",
-          ]
         include:
           - os: ubuntu-22.04
             llvm_cache_key: ${{ needs.build_llvm_libraries_on_ubuntu_2204.outputs.cache_key }}
+
     steps:
       - name: checkout
         uses: actions/checkout@v4
 
-      - name: download and install wasi-sdk
-        run: |
-          cd /opt
-          sudo wget ${{ matrix.wasi_sdk_release }}
-          sudo tar -xzf wasi-sdk-*.tar.gz
-          sudo ln -sf wasi-sdk-20.0 wasi-sdk
-
-      - name: download and install wabt
-        run: |
-          cd /opt
-          sudo wget ${{ matrix.wabt_release }}
-          sudo tar -xzf wabt-1.0.31-*.tar.gz
-          sudo ln -sf wabt-1.0.31 wabt
       - name: Get LLVM libraries
         id: retrieve_llvm_libs
         uses: actions/cache@v4
@@ -473,12 +457,19 @@ jobs:
             ./core/deps/llvm/build/libexec
             ./core/deps/llvm/build/share
           key: ${{ matrix.llvm_cache_key }}
+
+      - name: install-wasi-sdk-wabt
+        uses: ./.github/actions/install-wasi-sdk-wabt
+        with:
+          os: ${{ matrix.os }}
+
       - name: Build wamrc
         run: |
           mkdir build && cd build
           cmake ..
           cmake --build . --config Release --parallel 4
         working-directory: wamr-compiler
+
       - name: Build Sample [basic]
         run: |
           cd samples/basic
@@ -563,6 +554,15 @@ jobs:
           ./run.sh test1
           ./run.sh test2
 
+      - name: Build Sample [shared-heap]
+        run: |
+          cd samples/shared-heap
+          mkdir build && cd build
+          cmake ..
+          cmake --build . --config Debug --parallel 4
+          ./shared_heap_test
+          ./shared_heap_test --aot
+
   test:
     needs:
       [
@@ -593,10 +593,7 @@ jobs:
             $WASI_TEST_OPTIONS,
             $GC_TEST_OPTIONS,
             $MEMORY64_TEST_OPTIONS,
-          ]
-        wasi_sdk_release:
-          [
-            "https://github.com/WebAssembly/wasi-sdk/releases/download/wasi-sdk-20/wasi-sdk-20.0-linux.tar.gz",
+            $MULTI_MEMORY_TEST_OPTIONS,
           ]
         include:
           - os: ubuntu-22.04
@@ -608,11 +605,9 @@ jobs:
             test_option: $WAMR_COMPILER_TEST_OPTIONS
         exclude:
           # incompatible modes and features
-          # classic-interp and fast-interp don't support simd
+          # classic-interp doesn't support simd
           - running_mode: "classic-interp"
             test_option: $SIMD_TEST_OPTIONS
-          - running_mode: "fast-interp"
-            test_option: $SIMD_TEST_OPTIONS
           # llvm jit doesn't support multi module
           - running_mode: "jit"
             test_option: $MULTI_MODULES_TEST_OPTIONS
@@ -640,27 +635,37 @@ jobs:
             test_option: $MEMORY64_TEST_OPTIONS
           - running_mode: "multi-tier-jit"
             test_option: $MEMORY64_TEST_OPTIONS
+          # aot, fast-interp, fast-jit, llvm-jit, multi-tier-jit don't support Multi Memory
+          - running_mode: "aot"
+            test_option: $MULTI_MEMORY_TEST_OPTIONS
+          - running_mode: "fast-interp"
+            test_option: $MULTI_MEMORY_TEST_OPTIONS
+          - running_mode: "fast-jit"
+            test_option: $MULTI_MEMORY_TEST_OPTIONS
+          - running_mode: "jit"
+            test_option: $MULTI_MEMORY_TEST_OPTIONS
+          - running_mode: "multi-tier-jit"
+            test_option: $MULTI_MEMORY_TEST_OPTIONS
+
     steps:
       - name: checkout
         uses: actions/checkout@v4
 
       - name: Set-up OCaml
         uses: ocaml/setup-ocaml@v3
-        if: matrix.test_option == '$GC_TEST_OPTIONS' || matrix.test_option == '$MEMORY64_TEST_OPTIONS'
+        if: matrix.test_option == '$GC_TEST_OPTIONS'
         with:
           ocaml-compiler: 4.13
 
       - name: Set-up Ocamlbuild
-        if: matrix.test_option == '$GC_TEST_OPTIONS' || matrix.test_option == '$MEMORY64_TEST_OPTIONS'
+        if: matrix.test_option == '$GC_TEST_OPTIONS'
         run: opam install ocamlbuild dune menhir
 
       - name: download and install wasi-sdk
         if: matrix.test_option == '$WASI_TEST_OPTIONS'
-        run: |
-          cd /opt
-          sudo wget ${{ matrix.wasi_sdk_release }}
-          sudo tar -xzf wasi-sdk-*.tar.gz
-          sudo mv wasi-sdk-20.0 wasi-sdk
+        uses: ./.github/actions/install-wasi-sdk-wabt
+        with:
+          os: ${{ matrix.os }}
 
       # It is a temporary solution until new wasi-sdk that includes bug fixes is released
       - name: build wasi-libc from source
@@ -717,13 +722,13 @@ jobs:
 
       - name: run tests
         timeout-minutes: 30
-        if: matrix.test_option != '$GC_TEST_OPTIONS' && matrix.test_option != '$MEMORY64_TEST_OPTIONS'
+        if: matrix.test_option != '$GC_TEST_OPTIONS'
         run: ./test_wamr.sh ${{ matrix.test_option }} -t ${{ matrix.running_mode }}
         working-directory: ./tests/wamr-test-suites
 
-      - name: run gc or memory64 tests
+      - name: run gc tests
         timeout-minutes: 20
-        if: matrix.test_option == '$GC_TEST_OPTIONS' || matrix.test_option == '$MEMORY64_TEST_OPTIONS'
+        if: matrix.test_option == '$GC_TEST_OPTIONS'
         run: |
           eval $(opam env)
           ./test_wamr.sh ${{ matrix.test_option }} -t ${{ matrix.running_mode }}
@@ -770,7 +775,7 @@ jobs:
 
       - name: install dependencies
         run: |
-          rustup target add wasm32-wasi
+          rustup target add wasm32-wasip1
           sudo apt update && sudo apt-get install -y lld ninja-build
           npm install
         working-directory: test-tools/wamr-ide/VSCode-Extension
@@ -785,7 +790,7 @@ jobs:
         run: |
           mkdir build
           cd build
-          cmake .. -DWAMR_BUILD_DEBUG_INTERP=1
+          cmake .. -DWAMR_BUILD_DEBUG_INTERP=1 -DWAMR_BUILD_REF_TYPES=1
           make
         working-directory: product-mini/platforms/linux
 

.github/workflows/compilation_on_macos.yml

@@ -52,13 +52,22 @@ env:
   LLVM_LAZY_JIT_BUILD_OPTIONS: "-DWAMR_BUILD_AOT=1 -DWAMR_BUILD_FAST_INTERP=0 -DWAMR_BUILD_INTERP=0 -DWAMR_BUILD_JIT=1 -DWAMR_BUILD_LAZY_JIT=1"
   LLVM_EAGER_JIT_BUILD_OPTIONS: "-DWAMR_BUILD_AOT=1 -DWAMR_BUILD_FAST_INTERP=0 -DWAMR_BUILD_INTERP=0 -DWAMR_BUILD_JIT=1 -DWAMR_BUILD_LAZY_JIT=0"
 
+permissions:
+  contents: read
+
 jobs:
   build_llvm_libraries_on_intel_macos:
+    permissions:
+      contents: read
+      actions: write
     uses: ./.github/workflows/build_llvm_libraries.yml
     with:
       os: "macos-13"
       arch: "X86"
   build_llvm_libraries_on_arm_macos:
+    permissions:
+      contents: read
+      actions: write
     uses: ./.github/workflows/build_llvm_libraries.yml
     with:
       os: "macos-14"
@@ -219,24 +228,14 @@ jobs:
             #$AOT_BUILD_OPTIONS,
           ]
         os: [macos-13]
-        wasi_sdk_release:
-          [
-            "https://github.com/WebAssembly/wasi-sdk/releases/download/wasi-sdk-20/wasi-sdk-20.0-macos.tar.gz",
-          ]
-        wabt_release:
-          [
-            "https://github.com/WebAssembly/wabt/releases/download/1.0.31/wabt-1.0.31-macos-12.tar.gz",
-          ]
     steps:
       - name: checkout
         uses: actions/checkout@v4
 
-      - name: download and install wabt
-        run: |
-          cd /opt
-          sudo wget ${{ matrix.wabt_release }}
-          sudo tar -xzf wabt-1.0.31-*.tar.gz
-          sudo mv wabt-1.0.31 wabt
+      - name: install-wasi-sdk-wabt
+        uses: ./.github/actions/install-wasi-sdk-wabt
+        with:
+          os: ${{ matrix.os }}
 
       - name: Build Sample [wasm-c-api]
         run: |
@@ -251,14 +250,6 @@ jobs:
     strategy:
       matrix:
         os: [macos-13, macos-14]
-        wasi_sdk_release:
-          [
-            "https://github.com/WebAssembly/wasi-sdk/releases/download/wasi-sdk-20/wasi-sdk-20.0-macos.tar.gz",
-          ]
-        wabt_release:
-          [
-            "https://github.com/WebAssembly/wabt/releases/download/1.0.31/wabt-1.0.31-macos-12.tar.gz",
-          ]
         include:
           - os: macos-13
             llvm_cache_key: ${{ needs.build_llvm_libraries_on_intel_macos.outputs.cache_key }}
@@ -268,19 +259,10 @@ jobs:
       - name: checkout
         uses: actions/checkout@v4
 
-      - name: download and install wasi-sdk
-        run: |
-          cd /opt
-          sudo wget ${{ matrix.wasi_sdk_release }}
-          sudo tar -xzf wasi-sdk-*.tar.gz
-          sudo ln -sf wasi-sdk-20.0 wasi-sdk
-
-      - name: download and install wabt
-        run: |
-          cd /opt
-          sudo wget ${{ matrix.wabt_release }}
-          sudo tar -xzf wabt-1.0.31-*.tar.gz
-          sudo ln -sf wabt-1.0.31 wabt
+      - name: install-wasi-sdk-wabt
+        uses: ./.github/actions/install-wasi-sdk-wabt
+        with:
+          os: ${{ matrix.os }}
 
       - name: Build Sample [basic]
         run: |
@@ -347,12 +329,13 @@ jobs:
           cmake --build . --config Release --parallel 4
         working-directory: wamr-compiler
 
+      # cmake --build . --config Debug --parallel 4
       - name: Build Sample [wasi-threads]
         run: |
           cd samples/wasi-threads
           mkdir build && cd build
           cmake ..
-          cmake --build . --config Debug --parallel 4
+          cmake --build . --config Debug --verbose
           ./iwasm wasm-apps/no_pthread.wasm
 
           ../../../wamr-compiler/build/wamrc --size-level=0 --enable-multi-thread -o wasm-apps/no_pthread.aot wasm-apps/no_pthread.wasm
@@ -386,3 +369,12 @@ jobs:
           ./build.sh
           ./run.sh test1
           ./run.sh test2
+
+      - name: Build Sample [shared-heap]
+        run: |
+          cd samples/shared-heap
+          mkdir build && cd build
+          cmake ..
+          cmake --build . --config Debug --parallel 4
+          ./shared_heap_test
+          ./shared_heap_test --aot

.github/workflows/compilation_on_nuttx.yml

@@ -46,6 +46,9 @@ concurrency:
 env:
   WASI_SDK_PATH: "/opt/wasi-sdk"
 
+permissions:
+  contents: read
+
 jobs:
   build_iwasm_on_nuttx:
     runs-on: ubuntu-latest
@@ -85,14 +88,14 @@ jobs:
         uses: actions/checkout@v4
         with:
           repository: apache/nuttx
-          ref: releases/12.6
+          ref: releases/12.9
           path: nuttx
 
       - name: Checkout NuttX Apps
         uses: actions/checkout@v4
         with:
           repository: apache/nuttx-apps
-          ref: releases/12.6
+          ref: releases/12.9
           path: apps
 
       - name: Checkout WAMR
@@ -119,11 +122,12 @@ jobs:
         run: make -j$(nproc) EXTRAFLAGS=-Werror
 
       - name: Checkout Bloaty
-        uses: actions/checkout@v3
+        uses: actions/checkout@v4
         with:
           repository: google/bloaty
           submodules: recursive
           path: bloaty
+          ref: 34f4a66559ad4938c1e629e9b5f54630b2b4d7b0
 
       - name: Build Bloaty
         run: |

.github/workflows/compilation_on_sgx.yml

@@ -49,16 +49,26 @@ env:
   # ref types enabled in wamrc by default, so we need to enable it for iwasm in AOT mode
   AOT_BUILD_OPTIONS: "-DWAMR_BUILD_AOT=1 -DWAMR_BUILD_FAST_INTERP=0 -DWAMR_BUILD_INTERP=0 -DWAMR_BUILD_JIT=0 -DWAMR_BUILD_LAZY_JIT=0 -DWAMR_BUILD_REF_TYPES=1"
   CLASSIC_INTERP_BUILD_OPTIONS: "-DWAMR_BUILD_AOT=0 -DWAMR_BUILD_FAST_INTERP=0 -DWAMR_BUILD_INTERP=1 -DWAMR_BUILD_JIT=0 -DWAMR_BUILD_LAZY_JIT=0"
-  FAST_INTERP_BUILD_OPTIONS: "-DWAMR_BUILD_AOT=0 -DWAMR_BUILD_FAST_INTERP=1 -DWAMR_BUILD_INTERP=1 -DWAMR_BUILD_JIT=0 -DWAMR_BUILD_LAZY_JIT=0"
+  FAST_INTERP_BUILD_OPTIONS: "-DWAMR_BUILD_AOT=0 -DWAMR_BUILD_FAST_INTERP=1 -DWAMR_BUILD_INTERP=1 -DWAMR_BUILD_JIT=0 -DWAMR_BUILD_LAZY_JIT=0 -DWAMR_BUILD_SIMD=0"
   FAST_JIT_BUILD_OPTIONS: "-DWAMR_BUILD_AOT=1 -DWAMR_BUILD_FAST_INTERP=0 -DWAMR_BUILD_INTERP=1 -DWAMR_BUILD_FAST_JIT=1 -DWAMR_BUILD_JIT=0 -DWAMR_BUILD_LAZY_JIT=1"
   LLVM_LAZY_JIT_BUILD_OPTIONS: "-DWAMR_BUILD_AOT=1 -DWAMR_BUILD_FAST_INTERP=0 -DWAMR_BUILD_INTERP=0 -DWAMR_BUILD_JIT=1 -DWAMR_BUILD_LAZY_JIT=1"
   LLVM_EAGER_JIT_BUILD_OPTIONS: "-DWAMR_BUILD_AOT=1 -DWAMR_BUILD_FAST_INTERP=0 -DWAMR_BUILD_INTERP=0 -DWAMR_BUILD_JIT=1 -DWAMR_BUILD_LAZY_JIT=0"
+  # For Spec Test
+  DEFAULT_TEST_OPTIONS: "-s spec -x -p -b"
+  SIMD_TEST_OPTIONS: "-s spec -x -p -b -S"
+  XIP_TEST_OPTIONS: "-s spec -x -p -b -X"
+
+permissions:
+  contents: read
 
 jobs:
   build_llvm_libraries:
+    permissions:
+      contents: read
+      actions: write
     uses: ./.github/workflows/build_llvm_libraries.yml
     with:
-      os: "ubuntu-20.04"
+      os: ubuntu-22.04
       arch: "X86"
 
   build_iwasm:
@@ -91,12 +101,12 @@ jobs:
             "-DWAMR_BUILD_PERF_PROFILING=1",
             "-DWAMR_BUILD_REF_TYPES=1",
             # doesn't support
-            # "-DWAMR_BUILD_SIMD=1",
+            "-DWAMR_BUILD_SIMD=0",
             "-DWAMR_BUILD_TAIL_CALL=1",
             "-DWAMR_DISABLE_HW_BOUND_CHECK=1",
             "-DWAMR_BUILD_SGX_IPFS=1",
           ]
-        os: [ubuntu-20.04]
+        os: [ubuntu-22.04]
         platform: [linux-sgx]
         exclude:
           # incompatible mode and feature
@@ -104,22 +114,14 @@ jobs:
           - make_options_run_mode: $AOT_BUILD_OPTIONS
             make_options_feature: "-DWAMR_BUILD_MINI_LOADER=1"
     steps:
-      - name: install SGX SDK and necessary libraries
-        run: |
-          mkdir -p /opt/intel
-          cd /opt/intel
-          wget https://download.01.org/intel-sgx/sgx-linux/2.15/distro/ubuntu20.04-server/sgx_linux_x64_sdk_2.15.100.3.bin
-          chmod +x sgx_linux_x64_sdk_2.15.100.3.bin
-          echo 'yes' | ./sgx_linux_x64_sdk_2.15.100.3.bin
-          echo 'deb [arch=amd64] https://download.01.org/intel-sgx/sgx_repo/ubuntu focal main' | sudo tee /etc/apt/sources.list.d/intel-sgx.list
-          wget -qO - https://download.01.org/intel-sgx/sgx_repo/ubuntu/intel-sgx-deb.key | sudo apt-key add -
-          sudo apt update
-          sudo apt install -y libsgx-launch libsgx-urts
-          source /opt/intel/sgxsdk/environment
-
       - name: checkout
         uses: actions/checkout@v4
 
+      - name: install SGX SDK and necessary libraries
+        uses: ./.github/actions/install-linux-sgx
+        with:
+          os: ${{ matrix.os }}
+
       - name: Build iwasm
         run: |
           mkdir build && cd build
@@ -144,69 +146,29 @@ jobs:
             #$LLVM_LAZY_JIT_BUILD_OPTIONS,
             #$LLVM_EAGER_JIT_BUILD_OPTIONS,
           ]
-        os: [ubuntu-20.04]
-        wasi_sdk_release:
-          [
-            "https://github.com/WebAssembly/wasi-sdk/releases/download/wasi-sdk-19/wasi-sdk-19.0-linux.tar.gz",
-          ]
-        wabt_release:
-          [
-            "https://github.com/WebAssembly/wabt/releases/download/1.0.31/wabt-1.0.31-ubuntu.tar.gz",
-          ]
+        os: [ubuntu-22.04]
         iwasm_make_options_feature: [
             # Features to be tested: IPFS
             "-DWAMR_BUILD_SGX_IPFS=1",
           ]
         platform: [linux-sgx]
         include:
-          - os: ubuntu-20.04
+          - os: ubuntu-22.04
             llvm_cache_key: ${{ needs.build_llvm_libraries.outputs.cache_key }}
 
     steps:
       - name: checkout
         uses: actions/checkout@v4
 
-      - name: download and install wasi-sdk
-        run: |
-          cd /opt
-          sudo wget ${{ matrix.wasi_sdk_release }}
-          sudo tar -xzf wasi-sdk-*.tar.gz
-          sudo mv wasi-sdk-19.0 wasi-sdk
-
-      - name: download and install wabt
-        run: |
-          cd /opt
-          sudo wget ${{ matrix.wabt_release }}
-          sudo tar -xzf wabt-1.0.31-*.tar.gz
-          sudo mv wabt-1.0.31 wabt
-
-      - name: build wasi-libc (needed for wasi-threads)
-        run: |
-          mkdir wasi-libc
-          cd wasi-libc
-          git init
-          # "Fix a_store operation in atomic.h" commit on main branch
-          git fetch https://github.com/WebAssembly/wasi-libc \
-            1dfe5c302d1c5ab621f7abf04620fae92700fd22
-          git checkout FETCH_HEAD
-          make \
-            AR=/opt/wasi-sdk/bin/llvm-ar \
-            NM=/opt/wasi-sdk/bin/llvm-nm \
-            CC=/opt/wasi-sdk/bin/clang \
-            THREAD_MODEL=posix
-        working-directory: core/deps
+      - name: install-wasi-sdk-wabt
+        uses: ./.github/actions/install-wasi-sdk-wabt
+        with:
+          os: ${{ matrix.os }}
 
       - name: install SGX SDK and necessary libraries
-        run: |
-          mkdir -p /opt/intel
-          cd /opt/intel
-          wget https://download.01.org/intel-sgx/sgx-linux/2.15/distro/ubuntu20.04-server/sgx_linux_x64_sdk_2.15.100.3.bin
-          chmod +x sgx_linux_x64_sdk_2.15.100.3.bin
-          echo 'yes' | ./sgx_linux_x64_sdk_2.15.100.3.bin
-          echo 'deb [arch=amd64] https://download.01.org/intel-sgx/sgx_repo/ubuntu focal main' | sudo tee /etc/apt/sources.list.d/intel-sgx.list
-          wget -qO - https://download.01.org/intel-sgx/sgx_repo/ubuntu/intel-sgx-deb.key | sudo apt-key add -
-          sudo apt update
-          sudo apt install -y libsgx-launch libsgx-urts
+        uses: ./.github/actions/install-linux-sgx
+        with:
+          os: ${{ matrix.os }}
 
       - name: Build iwasm for testing samples
         run: |
@@ -234,59 +196,61 @@ jobs:
       - name: Build wamrc only for testing samples in aot mode
         if: matrix.iwasm_make_options_run_mode == '$AOT_BUILD_OPTIONS'
         run: |
-          mkdir build && cd build
-          cmake ..
-          cmake --build . --config Release --parallel 4
-          cp wamrc `pwd`/../../product-mini/platforms/${{ matrix.platform }}/enclave-sample
+          cmake -S . -B build
+          cmake --build build --config Release --parallel 4
+          cp build/wamrc ../product-mini/platforms/${{ matrix.platform }}/enclave-sample
         working-directory: wamr-compiler
 
       - name: Build Sample [file]
         run: |
-          cd samples/file
-          mkdir build && cd build
-          cmake ..
-          cmake --build . --config Debug --parallel 4
-          cp wasm-app/file.wasm `pwd`/../../../product-mini/platforms/${{ matrix.platform }}/enclave-sample
+          cmake -S . -B build
+          cmake --build build --config Debug --parallel 4
+          cp build/wasm-app/file.wasm ../../product-mini/platforms/${{ matrix.platform }}/enclave-sample
+        working-directory: samples/file
 
       - name: Test Sample [file] in non-aot mode
         if: matrix.iwasm_make_options_run_mode != '$AOT_BUILD_OPTIONS'
         run: |
           source /opt/intel/sgxsdk/environment
-          ./iwasm --dir=. file.wasm
+          ./iwasm --dir=. ./file.wasm
         working-directory: product-mini/platforms/${{ matrix.platform }}/enclave-sample
 
       - name: Test Sample [file] in aot mode
         if: matrix.iwasm_make_options_run_mode == '$AOT_BUILD_OPTIONS'
         run: |
           source /opt/intel/sgxsdk/environment
-          ./wamrc -sgx -o file.aot file.wasm
-          ./iwasm --dir=. file.aot
+          ./wamrc -sgx -o ./file.aot ./file.wasm
+          ./iwasm --dir=. ./file.aot
         working-directory: product-mini/platforms/${{ matrix.platform }}/enclave-sample
 
   spec_test_default:
     needs: [build_iwasm, build_llvm_libraries]
-    runs-on: ubuntu-20.04
+    runs-on: ${{ matrix.os }}
     strategy:
       matrix:
-        running_mode: ["classic-interp", "fast-interp", "aot", "fast-jit"]
-        # FIXME: use binary release(adding -b) instead of building from source after upgrading to 22.04
-        test_option: ["-x -p -s spec -P", "-x -p -s spec -S -P", "-x -p -s spec -X -P"]
-        llvm_cache_key: ["${{ needs.build_llvm_libraries.outputs.cache_key }}"]
+        #(workaround) disable "fast-interp" because of SIMDE
+        running_mode: ["classic-interp", "aot", "fast-jit"]
+        test_option:
+          [$DEFAULT_TEST_OPTIONS, $SIMD_TEST_OPTIONS, $XIP_TEST_OPTIONS]
+        os: [ubuntu-22.04]
         exclude:
           # classic-interp, fast-interp and fast-jit don't support simd
           - running_mode: "classic-interp"
-            test_option: "-x -p -s spec -S -P"
+            test_option: $SIMD_TEST_OPTIONS
           - running_mode: "fast-interp"
-            test_option: "-x -p -s spec -S -P"
+            test_option: $SIMD_TEST_OPTIONS
           - running_mode: "fast-jit"
-            test_option: "-x -p -s spec -S -P"
+            test_option: $SIMD_TEST_OPTIONS
           # classic-interp, fast-interp and fast jit don't support XIP
           - running_mode: "classic-interp"
-            test_option: "-x -p -s spec -X -P"
+            test_option: $XIP_TEST_OPTIONS
           - running_mode: "fast-interp"
-            test_option: "-x -p -s spec -X -P"
+            test_option: $XIP_TEST_OPTIONS
           - running_mode: "fast-jit"
-            test_option: "-x -p -s spec -X -P"
+            test_option: $XIP_TEST_OPTIONS
+        include:
+          - os: ubuntu-22.04
+            llvm_cache_key: ${{ needs.build_llvm_libraries.outputs.cache_key }}
 
     steps:
       - name: checkout
@@ -310,19 +274,19 @@ jobs:
         run: echo "::error::can not get prebuilt llvm libraries" && exit 1
 
       - name: install SGX SDK and necessary libraries
-        run: |
-          mkdir -p /opt/intel
-          cd /opt/intel
-          wget https://download.01.org/intel-sgx/sgx-linux/2.15/distro/ubuntu20.04-server/sgx_linux_x64_sdk_2.15.100.3.bin
-          chmod +x sgx_linux_x64_sdk_2.15.100.3.bin
-          echo 'yes' | ./sgx_linux_x64_sdk_2.15.100.3.bin
-          echo 'deb [arch=amd64] https://download.01.org/intel-sgx/sgx_repo/ubuntu focal main' | sudo tee /etc/apt/sources.list.d/intel-sgx.list
-          wget -qO - https://download.01.org/intel-sgx/sgx_repo/ubuntu/intel-sgx-deb.key | sudo apt-key add -
-          sudo apt update
-          sudo apt install -y libsgx-launch libsgx-urts
+        uses: ./.github/actions/install-linux-sgx
+        with:
+          os: ${{ matrix.os }}
 
-      - name: install for wabt compilation
-        run: sudo apt update && sudo apt install -y ninja-build
+      #workaround about a https://github.com/actions/runner-images/issues/6680#issuecomment-2640923706
+      - name: Increase swapfile
+        run: |
+          sudo swapoff -a
+          sudo fallocate -l 15G /swapfile
+          sudo chmod 600 /swapfile
+          sudo mkswap /swapfile
+          sudo swapon /swapfile
+          sudo swapon --show
 
       - name: run spec tests
         run: |

.github/workflows/compilation_on_windows.yml

@@ -53,6 +53,9 @@ concurrency:
   group: ${{ github.workflow }}-${{ github.ref }}
   cancel-in-progress: true
 
+permissions:
+  contents: read
+
 jobs:
   build:
     runs-on: windows-latest

.github/workflows/create_tag.yml

@@ -15,6 +15,9 @@ on:
         description: "the new tag just created"
         value: ${{ jobs.create_tag.outputs.new_tag}}
 
+permissions:
+  contents: read
+
 jobs:
   create_tag:
     runs-on: ubuntu-latest
@@ -22,6 +25,8 @@ jobs:
       minor_version: ${{ steps.preparation.outputs.minor_version }}
       new_ver: ${{ steps.preparation.outputs.new_ver }}
       new_tag: ${{ steps.preparation.outputs.new_tag }}
+    permissions:
+      contents: write # create and push tags
 
     steps:
       - uses: actions/checkout@v4

.github/workflows/hadolint_dockerfiles.yml

@@ -28,6 +28,9 @@ concurrency:
   group: ${{ github.workflow }}-${{ github.ref }}
   cancel-in-progress: true
 
+permissions:
+  contents: read
+
 jobs:
   run-hadolint-on-dockerfiles:
     runs-on: ubuntu-22.04

.github/workflows/nightly_run.yml

@@ -44,29 +44,27 @@ env:
   X86_32_TARGET_TEST_OPTIONS: "-m x86_32 -P"
   WASI_TEST_OPTIONS: "-s wasi_certification -w"
 
+permissions:
+  contents: read
+
 jobs:
-  build_llvm_libraries_on_ubuntu_2004:
-    uses: ./.github/workflows/build_llvm_libraries.yml
-    with:
-      os: "ubuntu-20.04"
-      arch: "X86"
-  build_llvm_libraries_on_ubuntu_2204:
+  build_llvm_libraries_on_ubuntu:
+    permissions:
+      contents: read
+      actions: write
     uses: ./.github/workflows/build_llvm_libraries.yml
     with:
       os: "ubuntu-22.04"
       arch: "X86"
 
   build_wamrc:
-    needs:
-      [
-        build_llvm_libraries_on_ubuntu_2004,
-      ]
+    needs: build_llvm_libraries_on_ubuntu
     runs-on: ${{ matrix.os }}
     strategy:
       matrix:
         include:
-          - os: ubuntu-20.04
-            llvm_cache_key: ${{ needs.build_llvm_libraries_on_ubuntu_2004.outputs.cache_key }}
+          - os: ubuntu-22.04
+            llvm_cache_key: ${{ needs.build_llvm_libraries_on_ubuntu.outputs.cache_key }}
     steps:
       - name: checkout
         uses: actions/checkout@v4
@@ -97,10 +95,7 @@ jobs:
         working-directory: wamr-compiler
 
   build_iwasm:
-    needs:
-      [
-        build_llvm_libraries_on_ubuntu_2004,
-      ]
+    needs: build_llvm_libraries_on_ubuntu
     runs-on: ${{ matrix.os }}
     strategy:
       matrix:
@@ -132,8 +127,10 @@ jobs:
             "-DWAMR_BUILD_TAIL_CALL=1",
             "-DWAMR_DISABLE_HW_BOUND_CHECK=1",
             "-DWAMR_BUILD_MEMORY64=1",
+            "-DWAMR_BUILD_MULTI_MEMORY=1",
+            "-DWAMR_BUILD_SHARED=1",
           ]
-        os: [ubuntu-20.04]
+        os: [ubuntu-22.04]
         platform: [android, linux]
         exclude:
           # incompatible feature and platform
@@ -190,11 +187,9 @@ jobs:
             make_options_feature: "-DWAMR_BUILD_MINI_LOADER=1"
           - make_options_run_mode: $MULTI_TIER_JIT_BUILD_OPTIONS
             make_options_feature: "-DWAMR_BUILD_MINI_LOADER=1"
-          # Memory64 only on CLASSIC INTERP mode, and only on 64-bit platform
+          # Memory64 only on CLASSIC INTERP and AOT mode, and only on 64-bit platform
           - make_options_feature: "-DWAMR_BUILD_MEMORY64=1"
             platform: android
-          - make_options_run_mode: $AOT_BUILD_OPTIONS
-            make_options_feature: "-DWAMR_BUILD_MEMORY64=1"
           - make_options_run_mode: $FAST_INTERP_BUILD_OPTIONS
             make_options_feature: "-DWAMR_BUILD_MEMORY64=1"
           - make_options_run_mode: $FAST_JIT_BUILD_OPTIONS
@@ -205,6 +200,21 @@ jobs:
             make_options_feature: "-DWAMR_BUILD_MEMORY64=1"
           - make_options_run_mode: $MULTI_TIER_JIT_BUILD_OPTIONS
             make_options_feature: "-DWAMR_BUILD_MEMORY64=1"
+          # Multi memory only on CLASSIC INTERP mode, and only on 64-bit platform
+          - make_options_feature: "-DWAMR_BUILD_MEMORY64=1"
+            platform: android
+          - make_options_run_mode: $AOT_BUILD_OPTIONS
+            make_options_feature: "-DWAMR_BUILD_MULTI_MEMORY=1"
+          - make_options_run_mode: $FAST_INTERP_BUILD_OPTIONS
+            make_options_feature: "-DWAMR_BUILD_MULTI_MEMORY=1"
+          - make_options_run_mode: $FAST_JIT_BUILD_OPTIONS
+            make_options_feature: "-DWAMR_BUILD_MULTI_MEMORY=1"
+          - make_options_run_mode: $LLVM_LAZY_JIT_BUILD_OPTIONS
+            make_options_feature: "-DWAMR_BUILD_MULTI_MEMORY=1"
+          - make_options_run_mode: $LLVM_EAGER_JIT_BUILD_OPTIONS
+            make_options_feature: "-DWAMR_BUILD_MULTI_MEMORY=1"
+          - make_options_run_mode: $MULTI_TIER_JIT_BUILD_OPTIONS
+            make_options_feature: "-DWAMR_BUILD_MULTI_MEMORY=1"
           # Fast-JIT and Multi-Tier-JIT mode don't support android
           - make_options_run_mode: $FAST_JIT_BUILD_OPTIONS
             platform: android
@@ -217,8 +227,8 @@ jobs:
           - make_options_run_mode: $LLVM_EAGER_JIT_BUILD_OPTIONS
             platform: android
         include:
-          - os: ubuntu-20.04
-            llvm_cache_key: ${{ needs.build_llvm_libraries_on_ubuntu_2004.outputs.cache_key }}
+          - os: ubuntu-22.04
+            llvm_cache_key: ${{ needs.build_llvm_libraries_on_ubuntu.outputs.cache_key }}
 
     steps:
       - name: checkout
@@ -289,6 +299,7 @@ jobs:
             "-DWAMR_BUILD_TAIL_CALL=1",
             "-DWAMR_DISABLE_HW_BOUND_CHECK=1",
             "-DWAMR_BUILD_MEMORY64=1",
+            "-DWAMR_BUILD_MULTI_MEMORY=1",
           ]
         exclude:
           # incompatible feature and platform
@@ -322,6 +333,11 @@ jobs:
             make_options_feature: "-DWAMR_BUILD_MEMORY64=1"
           - make_options_run_mode: $FAST_JIT_BUILD_OPTIONS
             make_options_feature: "-DWAMR_BUILD_MEMORY64=1"
+          # Memory64 only on CLASSIC INTERP mode
+          - make_options_run_mode: $FAST_INTERP_BUILD_OPTIONS
+            make_options_feature: "-DWAMR_BUILD_MULTI_MEMORY=1"
+          - make_options_run_mode: $FAST_JIT_BUILD_OPTIONS
+            make_options_feature: "-DWAMR_BUILD_MULTI_MEMORY=1"
     steps:
       - name: Install dependencies
         run: |
@@ -345,12 +361,7 @@ jobs:
         working-directory: wamr/product-mini/platforms/linux
 
   build_samples_wasm_c_api:
-    needs:
-      [
-        build_iwasm,
-        build_llvm_libraries_on_ubuntu_2004,
-        build_wamrc,
-      ]
+    needs: [build_iwasm, build_llvm_libraries_on_ubuntu, build_wamrc]
     runs-on: ${{ matrix.os }}
     strategy:
       fail-fast: false
@@ -366,18 +377,10 @@ jobs:
             $LLVM_EAGER_JIT_BUILD_OPTIONS,
             $MULTI_TIER_JIT_BUILD_OPTIONS,
           ]
-        os: [ubuntu-20.04]
-        wasi_sdk_release:
-          [
-            "https://github.com/WebAssembly/wasi-sdk/releases/download/wasi-sdk-20/wasi-sdk-20.0-linux.tar.gz",
-          ]
-        wabt_release:
-          [
-            "https://github.com/WebAssembly/wabt/releases/download/1.0.31/wabt-1.0.31-ubuntu.tar.gz",
-          ]
+        os: [ubuntu-22.04]
         include:
-          - os: ubuntu-20.04
-            llvm_cache_key: ${{ needs.build_llvm_libraries_on_ubuntu_2004.outputs.cache_key }}
+          - os: ubuntu-22.04
+            llvm_cache_key: ${{ needs.build_llvm_libraries_on_ubuntu.outputs.cache_key }}
         exclude:
           - make_options: $MULTI_TIER_JIT_BUILD_OPTIONS
             sanitizer: asan
@@ -402,12 +405,11 @@ jobs:
         if: (!endsWith(matrix.make_options, '_INTERP_BUILD_OPTIONS')) && (steps.retrieve_llvm_libs.outputs.cache-hit != 'true')
         run: echo "::error::can not get prebuilt llvm libraries" && exit 1
 
-      - name: download and install wabt
-        run: |
-          cd /opt
-          sudo wget ${{ matrix.wabt_release }}
-          sudo tar -xzf wabt-1.0.31-*.tar.gz
-          sudo mv wabt-1.0.31 wabt
+      - name: install-wasi-sdk-wabt
+        uses: ./.github/actions/install-wasi-sdk-wabt
+        with:
+          os: ${{ matrix.os }}
+
       - name: Build wamrc
         if: (!endsWith(matrix.make_options, '_INTERP_BUILD_OPTIONS'))
         run: |
@@ -427,44 +429,23 @@ jobs:
         working-directory: samples/wasm-c-api
 
   build_samples_others:
-    needs:
-      [
-        build_iwasm,
-        build_llvm_libraries_on_ubuntu_2004,
-        build_wamrc,
-      ]
+    needs: [build_iwasm, build_llvm_libraries_on_ubuntu, build_wamrc]
     runs-on: ${{ matrix.os }}
     strategy:
       matrix:
-        os: [ubuntu-20.04]
-        wasi_sdk_release:
-          [
-            "https://github.com/WebAssembly/wasi-sdk/releases/download/wasi-sdk-20/wasi-sdk-20.0-linux.tar.gz",
-          ]
-        wabt_release:
-          [
-            "https://github.com/WebAssembly/wabt/releases/download/1.0.31/wabt-1.0.31-ubuntu.tar.gz",
-          ]
+        os: [ubuntu-22.04]
         include:
-          - os: ubuntu-20.04
-            llvm_cache_key: ${{ needs.build_llvm_libraries_on_ubuntu_2004.outputs.cache_key }}
+          - os: ubuntu-22.04
+            llvm_cache_key: ${{ needs.build_llvm_libraries_on_ubuntu.outputs.cache_key }}
     steps:
       - name: checkout
         uses: actions/checkout@v4
 
-      - name: download and install wasi-sdk
-        run: |
-          cd /opt
-          sudo wget ${{ matrix.wasi_sdk_release }}
-          sudo tar -xzf wasi-sdk-*.tar.gz
-          sudo ln -sf wasi-sdk-20.0 wasi-sdk
-      - name: download and install wabt
-        run: |
-          cd /opt
-          sudo wget ${{ matrix.wabt_release }}
-          sudo tar -xzf wabt-1.0.31-*.tar.gz
-          sudo ln -sf wabt-1.0.31 wabt
-
+      - name: install-wasi-sdk-wabt
+        uses: ./.github/actions/install-wasi-sdk-wabt
+        with:
+          os: ${{ matrix.os }}
+        
       - name: Get LLVM libraries
         id: retrieve_llvm_libs
         uses: actions/cache@v4
@@ -561,31 +542,38 @@ jobs:
           ./iwasm --native-lib=./libtest_add.so --native-lib=./libtest_sqrt.so --native-lib=./libtest_hello.so --native-lib=./libtest_hello2.so wasm-app/test.wasm
         working-directory: ./samples/native-lib
 
-      - name: checkout wamr-app-framework
-        run: git clone https://github.com/bytecodealliance/wamr-app-framework.git
-      - name: download wamr-app-framework dependencies
-        run: LVGL=0 LV_DRIVERS=0 ./download.sh
-        working-directory: ./wamr-app-framework/deps
-      - name: Build Sample [simple]
+      # FIXME: un-comment me after fix cmake minimum issue
+      # https://github.com/bytecodealliance/wamr-app-framework/pull/11
+      # - name: checkout wamr-app-framework
+      #   run: git clone https://github.com/bytecodealliance/wamr-app-framework.git
+
+      # - name: download wamr-app-framework dependencies
+      #   run: LVGL=0 LV_DRIVERS=0 ./download.sh
+      #   working-directory: ./wamr-app-framework/deps
+
+      # - name: Build Sample [simple]
+      #   run: |
+      #     ./build.sh -p host-interp
+      #     python3 ./sample_test_run.py $(pwd)/out
+      #     exit $?
+      #   working-directory: ./wamr-app-framework/samples/simple
+
+      - name: Build Sample [shared-heap]
         run: |
-          ./build.sh -p host-interp
-          python3 ./sample_test_run.py $(pwd)/out
-          exit $?
-        working-directory: ./wamr-app-framework/samples/simple
+          cd samples/shared-heap
+          mkdir build && cd build
+          cmake ..
+          cmake --build . --config Debug --parallel 4
+          ./shared_heap_test
+          ./shared_heap_test --aot
 
   test:
-    needs:
-      [
-        build_iwasm,
-        build_llvm_libraries_on_ubuntu_2004,
-        build_llvm_libraries_on_ubuntu_2204,
-        build_wamrc,
-      ]
+    needs: [build_iwasm, build_llvm_libraries_on_ubuntu, build_wamrc]
     runs-on: ${{ matrix.os }}
     strategy:
       fail-fast: false
       matrix:
-        os: [ubuntu-20.04, ubuntu-22.04]
+        os: [ubuntu-22.04]
         sanitizer: ["", "ubsan", "asan", "tsan"]
         running_mode:
           [
@@ -604,22 +592,12 @@ jobs:
             $THREADS_TEST_OPTIONS,
             $WASI_TEST_OPTIONS,
           ]
-        wasi_sdk_release:
-          [
-            "https://github.com/WebAssembly/wasi-sdk/releases/download/wasi-sdk-20/wasi-sdk-20.0-linux.tar.gz",
-          ]
         include:
-          - os: ubuntu-20.04
-            llvm_cache_key: ${{ needs.build_llvm_libraries_on_ubuntu_2004.outputs.cache_key }}
-            ubuntu_version: "20.04"
           - os: ubuntu-22.04
-            llvm_cache_key: ${{ needs.build_llvm_libraries_on_ubuntu_2204.outputs.cache_key }}
+            llvm_cache_key: ${{ needs.build_llvm_libraries_on_ubuntu.outputs.cache_key }}
             ubuntu_version: "22.04"
 
         exclude:
-          # incompatible modes and features
-          - os: ubuntu-20.04
-            sanitizer: tsan
           # asan works only for aot now
           - running_mode: "classic-interp"
             sanitizer: asan
@@ -661,13 +639,10 @@ jobs:
       - name: checkout
         uses: actions/checkout@v4
 
-      - name: download and install wasi-sdk
-        if: matrix.test_option == '$WASI_TEST_OPTIONS'
-        run: |
-          cd /opt
-          sudo wget ${{ matrix.wasi_sdk_release }}
-          sudo tar -xzf wasi-sdk-*.tar.gz
-          sudo mv wasi-sdk-20.0 wasi-sdk
+      - name: install-wasi-sdk-wabt
+        uses: ./.github/actions/install-wasi-sdk-wabt
+        with:
+          os: ${{ matrix.os }}
 
       # It is a temporary solution until new wasi-sdk that includes bug fixes is released
       - name: build wasi-libc from source

.github/workflows/release_process.yml

@@ -18,11 +18,18 @@ concurrency:
   group: ${{ github.workflow }}-${{ github.ref }}
   cancel-in-progress: true
 
+permissions:
+  contents: read
+
 jobs:
   create_tag:
+    permissions:
+      contents: write # create and push tags
     uses: ./.github/workflows/create_tag.yml
 
   create_release:
+    permissions:
+      contents: write # create release
     needs: [create_tag]
     runs-on: ubuntu-latest
     outputs:
@@ -51,14 +58,10 @@ jobs:
 
   #
   # LLVM_LIBRARIES
-  build_llvm_libraries_on_ubuntu_2004:
-    needs: [create_tag, create_release]
-    uses: ./.github/workflows/build_llvm_libraries.yml
-    with:
-      os: "ubuntu-20.04"
-      arch: "AArch64 ARM Mips RISCV X86"
-
   build_llvm_libraries_on_ubuntu_2204:
+    permissions:
+      contents: read
+      actions: write
     needs: [create_tag, create_release]
     uses: ./.github/workflows/build_llvm_libraries.yml
     with:
@@ -66,25 +69,30 @@ jobs:
       arch: "AArch64 ARM Mips RISCV X86"
 
   build_llvm_libraries_on_macos:
+    permissions:
+      contents: read
+      actions: write
     needs: [create_tag, create_release]
     uses: ./.github/workflows/build_llvm_libraries.yml
     with:
       os: "macos-13"
       arch: "AArch64 ARM Mips RISCV X86"
 
+  build_llvm_libraries_on_windows:
+    permissions:
+      contents: read
+      actions: write
+    needs: [create_tag, create_release]
+    uses: ./.github/workflows/build_llvm_libraries.yml
+    with:
+      os: "windows-latest"
+      arch: "AArch64 ARM Mips RISCV X86"
+
   #
   # WAMRC
-  release_wamrc_on_ubuntu_2004:
-    needs: [create_tag, create_release, build_llvm_libraries_on_ubuntu_2004]
-    uses: ./.github/workflows/build_wamrc.yml
-    with:
-      llvm_cache_key: ${{ needs.build_llvm_libraries_on_ubuntu_2004.outputs.cache_key }}
-      release: true
-      runner: ubuntu-20.04
-      upload_url: ${{ needs.create_release.outputs.upload_url }}
-      ver_num: ${{ needs.create_tag.outputs.new_ver}}
-
   release_wamrc_on_ubuntu_2204:
+    permissions:
+      contents: write # upload release artifact
     needs: [create_tag, create_release, build_llvm_libraries_on_ubuntu_2204 ]
     uses: ./.github/workflows/build_wamrc.yml
     with:
@@ -95,6 +103,8 @@ jobs:
       ver_num: ${{ needs.create_tag.outputs.new_ver }}
 
   release_wamrc_on_ubuntu_macos:
+    permissions:
+      contents: write # upload release artifact
     needs: [create_tag, create_release, build_llvm_libraries_on_macos]
     uses: ./.github/workflows/build_wamrc.yml
     with:
@@ -103,20 +113,24 @@ jobs:
       runner: macos-13
       upload_url: ${{ needs.create_release.outputs.upload_url }}
       ver_num: ${{ needs.create_tag.outputs.new_ver }}
+  
+  release_wamrc_on_windows:
+    permissions:
+      contents: write # upload release artifact
+    needs: [create_tag, create_release, build_llvm_libraries_on_windows]
+    uses: ./.github/workflows/build_wamrc.yml
+    with:
+      llvm_cache_key: ${{ needs.build_llvm_libraries_on_windows.outputs.cache_key }}
+      release: true
+      runner: windows-latest
+      upload_url: ${{ needs.create_release.outputs.upload_url }}
+      ver_num: ${{ needs.create_tag.outputs.new_ver }}
 
   #
   # IWASM
-  release_iwasm_on_ubuntu_2004:
-    needs: [create_tag, create_release, build_llvm_libraries_on_ubuntu_2004]
-    uses: ./.github/workflows/build_iwasm_release.yml
-    with:
-      cwd: product-mini/platforms/linux
-      llvm_cache_key: ${{ needs.build_llvm_libraries_on_ubuntu_2004.outputs.cache_key }}
-      runner: ubuntu-20.04
-      upload_url: ${{ needs.create_release.outputs.upload_url }}
-      ver_num: ${{ needs.create_tag.outputs.new_ver}}
-
   release_iwasm_on_ubuntu_2204:
+    permissions:
+      contents: write # upload release artifact
     needs: [create_tag, create_release, build_llvm_libraries_on_ubuntu_2204]
     uses: ./.github/workflows/build_iwasm_release.yml
     with:
@@ -127,6 +141,8 @@ jobs:
       ver_num: ${{ needs.create_tag.outputs.new_ver}}
 
   release_iwasm_on_macos:
+    permissions:
+      contents: write # upload release artifact
     needs: [create_tag, create_release, build_llvm_libraries_on_macos]
     uses: ./.github/workflows/build_iwasm_release.yml
     with:
@@ -136,20 +152,23 @@ jobs:
       upload_url: ${{ needs.create_release.outputs.upload_url }}
       ver_num: ${{ needs.create_tag.outputs.new_ver}}
 
-  #
-  # WAMR_SDK
-  release_wamr_sdk_on_ubuntu_2004:
-    needs: [create_tag, create_release]
-    uses: ./.github/workflows/build_wamr_sdk.yml
+  release_iwasm_on_windows:
+    permissions:
+      contents: write # upload release artifact
+    needs: [create_tag, create_release, build_llvm_libraries_on_windows]
+    uses: ./.github/workflows/build_iwasm_release.yml
     with:
-      config_file: wamr_config_ubuntu_release.cmake
-      runner: ubuntu-20.04
+      cwd: product-mini/platforms/windows
+      llvm_cache_key: ${{ needs.build_llvm_libraries_on_windows.outputs.cache_key }}
+      runner: windows-latest
       upload_url: ${{ needs.create_release.outputs.upload_url }}
       ver_num: ${{ needs.create_tag.outputs.new_ver}}
-      wasi_sdk_url: https://github.com/WebAssembly/wasi-sdk/releases/download/wasi-sdk-19/wasi-sdk-19.0-linux.tar.gz
-      wamr_app_framework_url: https://github.com/bytecodealliance/wamr-app-framework.git
 
+  #
+  # WAMR_SDK
   release_wamr_sdk_on_ubuntu_2204:
+    permissions:
+      contents: write # upload release artifact
     needs: [create_tag, create_release]
     uses: ./.github/workflows/build_wamr_sdk.yml
     with:
@@ -161,6 +180,8 @@ jobs:
       wamr_app_framework_url: https://github.com/bytecodealliance/wamr-app-framework.git
 
   release_wamr_sdk_on_macos:
+    permissions:
+      contents: write # upload release artifact
     needs: [create_tag, create_release]
     uses: ./.github/workflows/build_wamr_sdk.yml
     with:
@@ -174,6 +195,8 @@ jobs:
   #
   # vscode extension cross-platform
   release_wamr_ide_vscode_ext:
+    permissions:
+      contents: write # upload release artifact
     needs: [create_tag, create_release]
     uses: ./.github/workflows/build_wamr_vscode_ext.yml
     secrets: inherit
@@ -184,6 +207,8 @@ jobs:
   #
   # vscode extension docker images package
   release_wamr_ide_docker_images_package:
+    permissions:
+      contents: write # upload release artifact
     needs: [create_tag, create_release]
     uses: ./.github/workflows/build_docker_images.yml
     with:
@@ -192,15 +217,9 @@ jobs:
 
   #
   # WAMR_LLDB
-  release_wamr_lldb_on_ubuntu_2004:
-    needs: [create_tag, create_release]
-    uses: ./.github/workflows/build_wamr_lldb.yml
-    with:
-      runner: ubuntu-20.04
-      upload_url: ${{ needs.create_release.outputs.upload_url }}
-      ver_num: ${{ needs.create_tag.outputs.new_ver}}
-
   release_wamr_lldb_on_ubuntu_2204:
+    permissions:
+      contents: write # upload release artifact
     needs: [create_tag, create_release]
     uses: ./.github/workflows/build_wamr_lldb.yml
     with:
@@ -209,6 +228,8 @@ jobs:
       ver_num: ${{ needs.create_tag.outputs.new_ver}}
 
   release_wamr_lldb_on_macos_universal:
+    permissions:
+      contents: write # upload release artifact
     needs: [create_tag, create_release]
     uses: ./.github/workflows/build_wamr_lldb.yml
     with:

.github/workflows/reuse_latest_release_binaries.yml

@@ -22,11 +22,17 @@ on:
       result:
         value: ${{ jobs.build.outputs.result }}
 
+permissions:
+  contents: read
+
 jobs:
   reuse:
     runs-on: ubuntu-latest
     outputs:
       result: ${{ steps.try_reuse.outputs.result }}
+    permissions:
+      contents: write # for creating realease and uploading release artifacts
+
     steps:
       - uses: actions/checkout@v4
         # Full git history is needed to get a proper list of commits and tags

.github/workflows/spec_test_on_nuttx.yml

@@ -29,8 +29,14 @@ env:
   LLVM_CACHE_SUFFIX: "build-llvm_libraries_ex"
   WASI_SDK_PATH: "/opt/wasi-sdk"
 
+permissions:
+  contents: read
+
 jobs:
   build_llvm_libraries:
+    permissions:
+      contents: read
+      actions: write
     uses: ./.github/workflows/build_llvm_libraries.yml
     with:
       os: "ubuntu-22.04"
@@ -38,6 +44,9 @@ jobs:
       container_image: ghcr.io/no1wudi/nuttx/apache-nuttx-ci-linux@sha256:8c4e00b607d4d6d66ba8f51c4544819a616eac69d3a2ac669e2af2150e2eb0f9
 
   build_llvm_libraries_xtensa:
+    permissions:
+      contents: read
+      actions: write
     uses: ./.github/workflows/build_llvm_libraries.yml
     with:
       os: "ubuntu-22.04"
@@ -74,11 +83,11 @@ jobs:
             target: "riscv32",
             fpu_type: "none"
           },
-          {
-            config: "boards/risc-v/qemu-rv/rv-virt/configs/nsh",
-            target: "riscv32_ilp32f",
-            fpu_type: "fp"
-          },
+          #{
+          #  config: "boards/risc-v/qemu-rv/rv-virt/configs/nsh",
+          #  target: "riscv32_ilp32f",
+          #  fpu_type: "fp"
+          #},
           # {
           #   config: "boards/risc-v/qemu-rv/rv-virt/configs/nsh",
           #   target: "riscv32_ilp32d",
@@ -118,7 +127,8 @@ jobs:
         wamr_feature_option:
           # Empty option for default
           - { option: "", mode: "" }
-          - { option: "CONFIG_INTERPRETERS_WAMR_GC CONFIG_INTERPRETERS_WAMR_AOT_STACK_FRAME", mode: "-G" }
+          # need to install menhir
+          # - { option: "CONFIG_INTERPRETERS_WAMR_GC CONFIG_INTERPRETERS_WAMR_AOT_STACK_FRAME", mode: "-G" }
 
         exclude:
           # XIP is not fully supported yet on RISCV64, some relocations can not be resolved
@@ -136,14 +146,14 @@ jobs:
         uses: actions/checkout@v4
         with:
           repository: apache/nuttx
-          ref: ${{ matrix.target_config.target == 'xtensa' && '985d395b025cf2012b22f6bb4461959fa6d87645' || 'releases/12.6' }}
+          ref: ${{ matrix.target_config.target == 'xtensa' && '985d395b025cf2012b22f6bb4461959fa6d87645' || 'releases/12.9' }}
           path: nuttx
 
       - name: Checkout NuttX Apps
         uses: actions/checkout@v4
         with:
           repository: apache/nuttx-apps
-          ref: ${{ matrix.target_config.target == 'xtensa' && '2ef3eb25c0cec944b13792185f7e5d5a05990d5f' || 'releases/12.6' }}
+          ref: ${{ matrix.target_config.target == 'xtensa' && '2ef3eb25c0cec944b13792185f7e5d5a05990d5f' || 'releases/12.9' }}
           path: apps
 
       - name: Checkout WAMR
@@ -173,33 +183,11 @@ jobs:
         if: contains(matrix.wamr_test_option.mode, 'aot')
         run: cp -r core/deps/llvm apps/interpreters/wamr/wamr/core/deps/llvm
 
-      # Inject the config option to NuttX
-      # TODO: Merge this into NuttX once GC is generally available
-      #
-      # Note: the version of nuttx-apps we use for xtensa does have
-      # an equivalent. (the default of INTERPRETERS_WAMR_TAIL_CALL is
-      # different though.)
-      - name: Modify Kconfig
-        if: matrix.target_config.target != 'xtensa'
-        run: |
-          echo "\n" >> apps/interpreters/wamr/Kconfig
-          echo "config INTERPRETERS_WAMR_GC" >> apps/interpreters/wamr/Kconfig
-          echo "\tbool \"Enable GC\"" >> apps/interpreters/wamr/Kconfig
-          echo "\tdefault n" >> apps/interpreters/wamr/Kconfig
-          echo "\n" >> apps/interpreters/wamr/Kconfig
-          echo "config INTERPRETERS_WAMR_AOT_STACK_FRAME" >> apps/interpreters/wamr/Kconfig
-          echo "\tbool \"Enable AOT stack frame\"" >> apps/interpreters/wamr/Kconfig
-          echo "\tdefault n" >> apps/interpreters/wamr/Kconfig
-          echo "\n" >> apps/interpreters/wamr/Kconfig
-          echo "config INTERPRETERS_WAMR_TAIL_CALL" >> apps/interpreters/wamr/Kconfig
-          echo "\tbool \"Enable Tail Call\"" >> apps/interpreters/wamr/Kconfig
-          echo "\tdefault y" >> apps/interpreters/wamr/Kconfig
-
       - name: Build wamrc
         if: contains(matrix.wamr_test_option.mode, 'aot')
         working-directory: apps/interpreters/wamr/wamr/wamr-compiler
         run: |
-          cmake -Bbuild .
+          cmake -B build -DWAMR_BUILD_SHRUNK_MEMORY=0 -S .
           cmake --build build
 
       # the nuttx version we use for xtensa requires esptool.py newer than
@@ -327,19 +315,6 @@ jobs:
         working-directory: apps/interpreters/wamr/wamr
 
       - name: Test
-        if: matrix.target_config.target != 'xtensa'
-        run: |
-          cd apps/interpreters/wamr/wamr/tests/wamr-test-suites
-          ./test_wamr.sh -s spec ${{ matrix.wamr_test_option.mode }} -m ${{ matrix.target_config.target }} -b -Q -P -F ${{ steps.build_firmware_path.outputs.firmware }} ${{ matrix.wamr_feature_option.mode}}
-
-      # for xtensa, for some reasons, when running the tests
-      # with test_wamr.sh -P, nuttx occasionally hangs after
-      # "total segments stored 6" on the CI.
-      # i (yamamoto) couldn't reproduce it locally (macOS) even
-      # with the identical flash image.
-      # for now, run the tests without -P.
-      - name: Test
-        if: matrix.target_config.target == 'xtensa'
         run: |
           cd apps/interpreters/wamr/wamr/tests/wamr-test-suites
           ./test_wamr.sh -s spec ${{ matrix.wamr_test_option.mode }} -m ${{ matrix.target_config.target }} -b -Q -F ${{ steps.build_firmware_path.outputs.firmware }} ${{ matrix.wamr_feature_option.mode}}
@@ -354,7 +329,7 @@ jobs:
 
       - name: upload the log
         if: always()
-        uses: actions/upload-artifact@v4
+        uses: actions/upload-artifact@v4.6.2
         with:
           name: spec-test-log-${{ github.run_id }}-${{ strategy.job-index }}-${{ matrix.target_config.target }}
           path: log

.github/workflows/supply_chain.yml

@@ -0,0 +1,65 @@
+# This workflow uses actions that are not certified by GitHub. They are provided
+# by a third-party and are governed by separate terms of service, privacy
+# policy, and support documentation.
+
+# Check current WASM Micro Runtime results here: https://securityscorecards.dev/viewer/?uri=github.com/bytecodealliance/wasm-micro-runtime
+
+name: Scorecard supply-chain security
+on:
+  # For Branch-Protection check. Only the default branch is supported. See
+  # https://github.com/ossf/scorecard/blob/main/docs/checks.md#branch-protection
+  branch_protection_rule:
+  # To guarantee Maintained check is occasionally updated. See
+  # https://github.com/ossf/scorecard/blob/main/docs/checks.md#maintained
+  # midnight UTC
+  schedule:
+    - cron: "0 0 * * *"
+  # allow to be triggered manually
+  workflow_dispatch:
+
+# Declare default permissions as read only.
+permissions:
+  contents: read
+
+jobs:
+  analysis:
+    name: Scorecard analysis
+    runs-on: ubuntu-latest
+    if: github.repository == 'bytecodealliance/wasm-micro-runtime'
+    permissions:
+      # Needed to upload the results to code-scanning dashboard.
+      security-events: write
+      # Needed to publish results and get a badge (see publish_results below).
+      id-token: write      
+
+    steps:
+      - name: "Checkout code"
+        uses: actions/checkout@93ea575cb5d8a053eaa0ac8fa3b40d7e05a33cc8 # v3.1.0
+        with:
+          persist-credentials: false
+
+      - name: "Run analysis"
+        uses: ossf/scorecard-action@f49aabe0b5af0936a0987cfb85d86b75731b0186 # v2.4.1
+        with:
+          results_file: results.sarif
+          results_format: sarif
+
+          #   - Publish results to OpenSSF REST API for easy access by consumers
+          #   - Allows the repository to include the Scorecard badge.
+          #   - See https://github.com/ossf/scorecard-action#publishing-results.      
+          publish_results: true
+
+      # Upload the results as artifacts (optional). Commenting out will disable uploads of run results in SARIF
+      # format to the repository Actions tab.
+      - name: "Upload artifact"
+        uses: actions/upload-artifact@ea165f8d65b6e75b540449e92b4886f43607fa02 # v3.1.0
+        with:
+          name: SARIF file
+          path: results.sarif
+          retention-days: 5
+
+      # Upload the results to GitHub's code scanning dashboard.
+      - name: "Upload to code-scanning"
+        uses: github/codeql-action/upload-sarif@5eb3ed6614230b1931d5c08df9e096e4ba524f21 
+        with:
+          sarif_file: results.sarif

.gitignore

@@ -13,6 +13,7 @@
 .clangd
 .DS_Store
 *.o
+.aider*
 
 core/deps/**
 core/shared/mem-alloc/tlsf

ATTRIBUTIONS.md

@@ -60,7 +60,7 @@ The WAMR fast interpreter is a clean room development. We would acknowledge the
 
 ### llvm
 
-[LICENSE](./LICENCE.txt)
+[LICENSE](./LICENSE)
 
 ### wasm-c-api
 

CMakeLists.txt

@@ -1,14 +1,18 @@
 # Copyright (C) 2019 Intel Corporation.  All rights reserved.
 # SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 
-cmake_minimum_required (VERSION 3.0)
+cmake_minimum_required (VERSION 3.14)
+
+option(BUILD_SHARED_LIBS "Build using shared libraries" OFF)
 
 if(ESP_PLATFORM)
   include (${COMPONENT_DIR}/build-scripts/esp-idf/wamr/CMakeLists.txt)
   return()
 endif()
 
-project (iwasm)
+project (iwasm LANGUAGES C)
+
+set(CMAKE_CXX_STANDARD 17)
 
 set (CMAKE_VERBOSE_MAKEFILE OFF)
 
@@ -16,13 +20,6 @@ if (NOT DEFINED WAMR_BUILD_PLATFORM)
   string (TOLOWER ${CMAKE_HOST_SYSTEM_NAME} WAMR_BUILD_PLATFORM)
 endif ()
 
-if (NOT DEFINED WAMR_BUILD_STATIC)
-  set (WAMR_BUILD_STATIC 1)
-endif ()
-if (NOT DEFINED WAMR_BUILD_SHARED)
-  set (WAMR_BUILD_SHARED 1)
-endif ()
-
 # Reset default linker flags
 set (CMAKE_SHARED_LIBRARY_LINK_C_FLAGS "")
 set (CMAKE_SHARED_LIBRARY_LINK_CXX_FLAGS "")
@@ -102,6 +99,11 @@ if (NOT DEFINED WAMR_BUILD_LIB_WASI_THREADS)
   set (WAMR_BUILD_LIB_WASI_THREADS 0)
 endif ()
 
+if (NOT DEFINED WAMR_ENABLE_COPY_CALLSTACK)
+  # Disable copy callstack by default
+  set (WAMR_ENABLE_COPY_CALLSTACK 0)
+endif()
+
 if (NOT DEFINED WAMR_BUILD_MINI_LOADER)
   # Disable wasm mini loader by default
   set (WAMR_BUILD_MINI_LOADER 0)
@@ -113,18 +115,23 @@ if (NOT DEFINED WAMR_BUILD_SIMD)
 endif ()
 
 if (NOT DEFINED WAMR_BUILD_REF_TYPES)
-  # Disable reference types by default
-  set (WAMR_BUILD_REF_TYPES 0)
+  # Enable reference types by default
+  set (WAMR_BUILD_REF_TYPES 1)
 endif ()
 
 set (WAMR_ROOT_DIR ${CMAKE_CURRENT_SOURCE_DIR})
 
 include (${WAMR_ROOT_DIR}/build-scripts/runtime_lib.cmake)
 
-set (CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -Wall -Wextra -Wformat -Wformat-security -Wshadow -Wno-unused-parameter -fvisibility=hidden")
-# set (CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -Wconversion -Wsign-conversion")
-
-set (CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Wall -Wextra -Wformat -Wformat-security -Wno-unused")
+if (NOT WIN32)
+  set (CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -Wall -Wextra -Wformat -Wformat-security \
+                                       -ffunction-sections -fdata-sections \
+                                       -Wno-unused-parameter -Wno-pedantic \
+                                       -fvisibility=hidden")
+  # Remove the extra spaces for better make log
+  string (REGEX REPLACE "  *" " " CMAKE_C_FLAGS ${CMAKE_C_FLAGS})
+  set (CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Wall -Wextra -Wformat -Wformat-security -Wno-unused")
+endif()
 
 if (WAMR_BUILD_TARGET MATCHES "X86_.*" OR WAMR_BUILD_TARGET STREQUAL "AMD_64")
   if (NOT (CMAKE_C_COMPILER MATCHES ".*clang.*" OR CMAKE_C_COMPILER_ID MATCHES ".*Clang"))
@@ -145,39 +152,44 @@ include (${SHARED_DIR}/utils/uncommon/shared_uncommon.cmake)
 set (THREADS_PREFER_PTHREAD_FLAG ON)
 find_package(Threads REQUIRED)
 
-# STATIC LIBRARY
-if (WAMR_BUILD_STATIC)
-    add_library(iwasm_static STATIC ${WAMR_RUNTIME_LIB_SOURCE})
-    set_target_properties (iwasm_static PROPERTIES OUTPUT_NAME vmlib)
-    target_include_directories(iwasm_static INTERFACE ${WAMR_ROOT_DIR}/core/iwasm/include)
-    target_link_libraries (iwasm_static INTERFACE ${LLVM_AVAILABLE_LIBS} ${UV_A_LIBS} -lm -ldl ${CMAKE_THREAD_LIBS_INIT})
-    if (WAMR_BUILD_WASM_CACHE EQUAL 1)
-      target_link_libraries(iwasm_static INTERFACE boringssl_crypto)
-    endif ()
-
-    install (TARGETS iwasm_static ARCHIVE DESTINATION lib)
+if (MSVC)
+  add_definitions(-DCOMPILING_WASM_RUNTIME_API=1)
 endif ()
 
-# SHARED LIBRARY
-if (WAMR_BUILD_SHARED)
-    add_library (iwasm_shared SHARED ${WAMR_RUNTIME_LIB_SOURCE})
-    set_target_properties (iwasm_shared PROPERTIES OUTPUT_NAME iwasm)
-    target_include_directories(iwasm_shared INTERFACE ${WAMR_ROOT_DIR}/core/iwasm/include)
-    target_link_libraries (iwasm_shared INTERFACE ${LLVM_AVAILABLE_LIBS} ${UV_A_LIBS} -lm -ldl ${CMAKE_THREAD_LIBS_INIT})
-    if (WAMR_BUILD_WASM_CACHE EQUAL 1)
-      target_link_libraries(iwasm_shared INTERFACE boringssl_crypto)
-    endif ()
+add_library (vmlib ${WAMR_RUNTIME_LIB_SOURCE})
+set_target_properties (vmlib PROPERTIES OUTPUT_NAME iwasm)
+target_include_directories(vmlib INTERFACE
+  $<BUILD_INTERFACE:${WAMR_ROOT_DIR}/core/iwasm/include>
+  $<INSTALL_INTERFACE:include/iwasm>
+)
 
-    if (MINGW)
-      target_link_libraries (iwasm_shared INTERFACE -lWs2_32 -lwsock32)
-    endif ()
-
-    install (TARGETS iwasm_shared LIBRARY DESTINATION lib)
+target_link_libraries (vmlib PUBLIC ${LLVM_AVAILABLE_LIBS} ${UV_A_LIBS} -lm -ldl ${CMAKE_THREAD_LIBS_INIT})
+if (WAMR_BUILD_WASM_CACHE EQUAL 1)
+  target_link_libraries(vmlib INTERFACE boringssl_crypto)
 endif ()
 
-# HEADERS
-install (FILES
-    ${WAMR_ROOT_DIR}/core/iwasm/include/wasm_c_api.h
-    ${WAMR_ROOT_DIR}/core/iwasm/include/wasm_export.h
-    ${WAMR_ROOT_DIR}/core/iwasm/include/lib_export.h
-    DESTINATION include)
+if (MINGW)
+  target_link_libraries(vmlib INTERFACE -lWs2_32 -lwsock32)
+  target_link_libraries(vmlib PRIVATE ws2_32)
+endif ()
+
+if (WIN32)
+  target_link_libraries(vmlib PRIVATE ntdll)
+endif()
+
+set (WAMR_PUBLIC_HEADERS
+  ${WAMR_ROOT_DIR}/core/iwasm/include/wasm_c_api.h
+  ${WAMR_ROOT_DIR}/core/iwasm/include/wasm_export.h
+  ${WAMR_ROOT_DIR}/core/iwasm/include/lib_export.h
+)
+set_target_properties (vmlib PROPERTIES PUBLIC_HEADER "${WAMR_PUBLIC_HEADERS}")
+
+set_version_info (vmlib)
+
+install (TARGETS vmlib
+  EXPORT iwasmTargets
+  LIBRARY DESTINATION lib
+  PUBLIC_HEADER DESTINATION include/iwasm
+)
+
+install_iwasm_package ()

CODEOWNERS

@@ -0,0 +1,29 @@
+# In this project, we use CODEOWNERS to identify people who are likely to know
+# who should review a pull request.
+#
+# People listed in this file are committing to respond in a timely fashion to
+# PRs in the selected areas. However, that response doesn't have to be a full
+# code review; it could also take any of these forms:
+#
+# - "I intend to review this but I can't yet. Please leave me a message if I
+#   haven't responded by (a specific date in the near future)."
+#
+# - "I think (a specific other contributor) should review this." (Note that the
+#   best reviewer for a PR may not necessarily be listed in this file.)
+#
+# People must only be added to this file if they've agreed to provide one of
+# the above responses in a reasonable amount of time for every PR to which
+# they're assigned.
+#
+# We only ask for this commitment from people who are employed full-time to
+# work on this project. We gratefully welcome reviews from other contributors,
+# but we don't believe it's fair to ask volunteers to respond quickly.
+
+# If none of the later patterns match, assign to anyone. This team is the
+# parent of all the other teams and automatically includes everyone on those
+# teams.
+* @loganek @lum1n0us @no1wudi @TianlongLiang @wenyongh @xujuntwt95329 @yamt
+
+# Some parts of the project require more specialized knowledge. In those areas
+# we designate smaller groups who are more likely to be aware of who's working
+# in specific areas.

CONTRIBUTING.md

@@ -19,7 +19,7 @@ Code changes
 We Use Github Flow, So All Code Changes Happen Through Pull Requests. Pull requests are the best way to propose changes to the codebase. We actively welcome your pull requests:
 
 - If you've added code that should be tested, add tests. Ensure the test suite passes.
-- Avoid use macros for different platforms. Use seperate folder of source files to host diffeent platform logic.
+- Avoid use macros for different platforms. Use separate folder of source files to host different platform logic.
 - Put macro definitions inside share_lib/include/config.h if you have to use macro.
 - Make sure your code lints and compliant to our coding style.
 - Extend the application library is highly welcome.

README.md

@@ -58,8 +58,8 @@ The WAMR VMcore supports the following architectures:
 - XTENSA, MIPS, ARC
 
 The following platforms are supported, click each link below for how to build iwasm on that platform. Refer to [WAMR porting guide](./doc/port_wamr.md) for how to port WAMR to a new platform.
-- [Linux](./product-mini/README.md#linux),  [Linux SGX (Intel Software Guard Extension)](./doc/linux_sgx.md),  [MacOS](./product-mini/README.md#macos),  [Android](./product-mini/README.md#android), [Windows](./product-mini/README.md#windows), [Windows (MinGW)](./product-mini/README.md#mingw)
-- [Zephyr](./product-mini/README.md#zephyr),  [AliOS-Things](./product-mini/README.md#alios-things),  [VxWorks](./product-mini/README.md#vxworks), [NuttX](./product-mini/README.md#nuttx), [RT-Thread](./product-mini/README.md#RT-Thread), [ESP-IDF](./product-mini/README.md#esp-idf)
+- [Linux](./product-mini/README.md#linux),  [Linux SGX (Intel Software Guard Extension)](./doc/linux_sgx.md),  [MacOS](./product-mini/README.md#macos),  [Android](./product-mini/README.md#android), [Windows](./product-mini/README.md#windows), [Windows (MinGW, MSVC)](./product-mini/README.md#mingw)
+- [Zephyr](./product-mini/README.md#zephyr),  [AliOS-Things](./product-mini/README.md#alios-things),  [VxWorks](./product-mini/README.md#vxworks), [NuttX](./product-mini/README.md#nuttx), [RT-Thread](./product-mini/README.md#RT-Thread), [ESP-IDF(FreeRTOS)](./product-mini/README.md#esp-idf)
 
 
 ## Getting started

RELEASE_NOTES.md

@@ -1,3 +1,85 @@
+## WAMR-2.2.0
+
+### Breaking changes 
+
+### New features 
+- Add support for multi-memory proposal in classic interpreter (#3742) 
+- wasi-nn: Add a new target for llama.cpp as a wasi-nn backend (#3709)
+- Add memory instance support apis (#3786) 
+- Implement a first version of shared heap feature (#3789) 
+- Support dynamic aot debug (#3788) 
+- Implement shared heap for AOT (#3815)
+- Support table64 extension in classic-interp and AOT running modes (#3811)
+  
+
+### Bug fixes
+- Enable merged os_mmap for aot data sections (#3681) 
+- Fix arm64 issues on mac (#3688) 
+- aot loader: Call os_mmap with MMAP_MAP_32BIT only when target is x86-64 or riscv64 (#3755)
+- Fix building iwasm_shared and iwasm_static libs on win32 (#3762) 
+- Fix compile error when multi-module and tags are enabled (#3781)
+- Fix aot multi export memory support (#3791)
+- Fix Windows compile error when uvwasi is enabled (#3810)
+- Fix missing symbols when using aot mode on riscv platforms (#3812)
+- Fix mac build of libc_emcc_wrapper.c (#3836)
+- aot_comp_option.h: Add missing stdint.h header (#3834)
+- Fix compilation error found in tflite test (#3820) 
+- Fix exec_env_tls assertion in module instantiation (#3844) 
+- Fix issues of destroy_shared_heaps (#3847)
+
+### Enhancements
+- aot loader: Refine os_mmap related code (#3711)
+- Enable merged os_mmap for aot data sections and aot text (#3743)
+- Improve posix mmap retry logic (#3714)
+- Remove unnecessary code duplication in aot runtime (#3767)
+- Add wamrc parameter to configure stack frame features (#3763) 
+- refactoring: Re-use commit IP functionality between exception handling and other cases (#3768)
+- AOT call stack optimizations (#3773) 
+- Appease GCC strict prototypes warning (#3775)
+- Appease GCC -Wformat (#3783)
+- Fix compiler warnings (#3784)
+- Implement option for skipping function index in the callstack (#3785)
+- Fix a compile warning in aot_emit_function.c (#3793) 
+- Restore cmake hidden compile symbol visibility (#3796)
+- Refactor shared heap feature for interpreter mode (#3794)
+- Add no_resolve to LoadArgs and wasm_runtime_resolve_symbols (#3790)
+- shared heap: Fix some issues and add basic unit test case (#3801)
+- Add shared heap sample (#3806)
+- Fix unused param warning when GC is enabled (#3814)
+- Add scoreboard CI for supply-chain security (#3819)
+- Emit load_addr and load_size if WAMR_ENABLE_COMPILER is set (#3835)
+- libc-emcc: Use alternate method to check getrandom support (#3848)
+- Enable libc-wasi for windows msvc build (#3852) 
+- Remove unused folder samples/gui and samples/littlevgl (#3853) 
+- Fix some compile warnings and typos (#3854)
+- Allow to set native stack boundary to exec_env (#3862)
+- Refine wasm/aot function instance lookup (#3865)
+- Fix quadratic runtime for duplicate export name detection (#3861)
+
+  
+### Others
+- Add a comment on AOT_SECTION_TYPE_SIGNATURE (#3746) 
+- CI: Freeze version of bloaty for NuttX compilation (#3756)
+- aot compiler: Allow to control stack boundary check when boundary check is enabled (#3754)
+- Update ref to the multi-memory tests (#3764)
+- compilation_on_nuttx.yml: Update checkout action to suppress warnings (#3765)
+- CI: Disable parallel test in spectest for NuttX (#3780)
+- spec_test_on_nuttx.yml: Disable riscv32_ilp32f for now (#3777) 
+- Ignore temporary file from aider (#3787)
+- Add CODEOWNERS (#3822)
+- build(deps): bump github/codeql-action from 2.2.4 to 3.26.9 (#3826)
+- build(deps): bump actions/upload-artifact from 3.1.0 to 4.4.0 (#3827)
+- build(deps): bump ossf/scorecard-action from 2.3.1 to 2.4.0 (#3828)
+- build(deps): bump github/codeql-action from 3.26.9 to 3.26.11 (#3843) 
+- build(deps): bump actions/upload-artifact from 4.4.0 to 4.4.3 (#3855)
+- build(deps): bump github/codeql-action from 3.26.11 to 3.26.12 (#3856)
+- Add Windows wamrc and iwasm build in release CI (#3857)
+- Fix syntax error in codeql_buildscript.sh (#3864)
+- release CI: Add another iwasm binary that supports Garbage Collection and Exception Handling (#3866)
+- Fix lookup function issue reported in nightly run (#3868)
+ 
+---
+
 ## WAMR-2.1.2
 
 ### Breaking Changes

build-scripts/build_llvm.py

@@ -102,12 +102,21 @@ def build_llvm(llvm_dir, platform, backends, projects, use_clang=False, extra_fl
         "default": [],
     }
 
+    experimental_backends = ["ARC", "Xtensa"]
+    normal_backends = [s for s in backends if s not in experimental_backends]
+
     LLVM_TARGETS_TO_BUILD = [
-        '-DLLVM_TARGETS_TO_BUILD:STRING="' + ";".join(backends) + '"'
-        if backends
+        '-DLLVM_TARGETS_TO_BUILD:STRING="' + ";".join(normal_backends) + '"'
+        if normal_backends
         else '-DLLVM_TARGETS_TO_BUILD:STRING="AArch64;ARM;Mips;RISCV;X86"'
     ]
 
+    # if not on ARC platform, but want to add expeirmental backend ARC as target
+    if platform != "ARC" and "ARC" in backends:
+        LLVM_TARGETS_TO_BUILD.extend(
+            LLVM_EXTRA_COMPILE_OPTIONS["arc"]
+        )
+
     LLVM_PROJECTS_TO_BUILD = [
         '-DLLVM_ENABLE_PROJECTS:STRING="' + ";".join(projects) + '"' if projects else ""
     ]
@@ -125,9 +134,7 @@ def build_llvm(llvm_dir, platform, backends, projects, use_clang=False, extra_fl
     if not llvm_dir.exists():
         raise Exception(f"{llvm_dir} doesn't exist")
 
-    build_dir = llvm_dir.joinpath(
-        "win32build" if "windows" == platform else "build"
-    ).resolve()
+    build_dir = llvm_dir.joinpath("build").resolve()
     build_dir.mkdir(exist_ok=True)
 
     lib_llvm_core_library = build_dir.joinpath("lib/libLLVMCore.a").resolve()
@@ -178,6 +185,7 @@ def repackage_llvm(llvm_dir):
         raise Exception("Find more than one LLVM-*.tar.gz")
 
     if not packs:
+        raise Exception("Didn't find any LLVM-* package")
         return
 
     llvm_package = packs[0].name
@@ -193,6 +201,31 @@ def repackage_llvm(llvm_dir):
     # rm ./LLVM-1*.gz
     os.remove(llvm_dir.joinpath(llvm_package).resolve())
 
+def repackage_llvm_windows(llvm_dir):
+    build_dir = llvm_dir.joinpath("./build").resolve()
+
+    packs_path = [f for f in build_dir.glob("./_CPack_Packages/win64/NSIS/LLVM-*-win64")]
+    if len(packs_path) > 1:
+        raise Exception("Find more than one LLVM-* package")
+
+    if not packs_path:
+        raise Exception("Didn't find any LLVM-* package")
+        return
+
+    llvm_package_path = f"_CPack_Packages/win64/NSIS/{packs_path[0].name}"
+    windows_package_dir = build_dir.joinpath(llvm_package_path).resolve()
+
+    # mv package dir outside of build
+    shutil.move(str(windows_package_dir), str(llvm_dir))
+    # rm -r build
+    shutil.rmtree(str(build_dir))
+    # mkdir build
+    build_dir.mkdir()
+    # move back all the subdiretories under cpack directory(bin/include/lib) to build dir
+    moved_package_dir = llvm_dir.joinpath(packs_path[0].name)
+    for sub_dir in moved_package_dir.iterdir():
+        shutil.move(str(sub_dir), str(build_dir))
+    moved_package_dir.rmdir()
 
 def main():
     parser = argparse.ArgumentParser(description="build necessary LLVM libraries")
@@ -216,6 +249,7 @@ def main():
             "X86",
             "Xtensa",
         ],
+        default=[],
         help="identify LLVM supported backends, separate by space, like '--arch ARM Mips X86'",
     )
     parser.add_argument(
@@ -304,7 +338,11 @@ def main():
             )
             is not None
         ):
-            repackage_llvm(llvm_dir)
+            # TODO: repackage process may change in the future, this work for LLVM 15.x
+            if "windows" == platform:
+                repackage_llvm_windows(llvm_dir)
+            else:
+                repackage_llvm(llvm_dir)
 
         return True
     except subprocess.CalledProcessError:

build-scripts/config_common.cmake

@@ -102,9 +102,6 @@ if (WAMR_BUILD_JIT EQUAL 1)
   if (NOT DEFINED LLVM_DIR)
     set (LLVM_SRC_ROOT "${WAMR_ROOT_DIR}/core/deps/llvm")
     set (LLVM_BUILD_ROOT "${LLVM_SRC_ROOT}/build")
-    if (WAMR_BUILD_PLATFORM STREQUAL "windows")
-      set (LLVM_BUILD_ROOT "${LLVM_SRC_ROOT}/win32build")
-    endif ()
     if (NOT EXISTS "${LLVM_BUILD_ROOT}")
         message (FATAL_ERROR "Cannot find LLVM dir: ${LLVM_BUILD_ROOT}")
     endif ()
@@ -134,6 +131,12 @@ else ()
   unset (LLVM_AVAILABLE_LIBS)
 endif ()
 
+# Version
+include (${WAMR_ROOT_DIR}/build-scripts/version.cmake)
+
+# Package
+include (${WAMR_ROOT_DIR}/build-scripts/package.cmake)
+
 # Sanitizers
 
 if (NOT DEFINED WAMR_BUILD_SANITIZER)
@@ -154,6 +157,9 @@ elseif (WAMR_BUILD_SANITIZER STREQUAL "asan")
 elseif (WAMR_BUILD_SANITIZER STREQUAL "tsan")
   set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -g -O0 -fno-omit-frame-pointer -fsanitize=thread -fno-sanitize-recover=all" )
   set(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} -fsanitize=thread")
+elseif (WAMR_BUILD_SANITIZER STREQUAL "posan")
+  set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -g -O0 -fno-omit-frame-pointer -fsanitize=pointer-overflow -fno-sanitize-recover=all" )
+  set(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} -fsanitize=pointer-overflow")
 elseif (NOT (WAMR_BUILD_SANITIZER STREQUAL "") )
   message(SEND_ERROR "Unsupported sanitizer: ${WAMR_BUILD_SANITIZER}")
 endif()
@@ -165,16 +171,68 @@ if (WAMR_BUILD_LINUX_PERF EQUAL 1)
   endif ()
 endif ()
 
+if (NOT DEFINED WAMR_BUILD_SHRUNK_MEMORY)
+  # Enable shrunk memory by default
+  set (WAMR_BUILD_SHRUNK_MEMORY 1)
+endif ()
+
+########################################
+# Default values
+########################################
+if (NOT DEFINED WAMR_BUILD_BULK_MEMORY)
+  set (WAMR_BUILD_BULK_MEMORY 1)
+endif ()
+
+if (NOT DEFINED WAMR_BUILD_EXCE_HANDLING)
+  set (WAMR_BUILD_EXCE_HANDLING 0)
+endif ()
+
+if (NOT DEFINED WAMR_BUILD_GC)
+  set (WAMR_BUILD_GC 0)
+endif ()
+
+if (NOT DEFINED WAMR_BUILD_MEMORY64)
+  set (WAMR_BUILD_MEMORY64 0)
+endif ()
+
+if (NOT DEFINED WAMR_BUILD_MULTI_MEMORY)
+  set (WAMR_BUILD_MULTI_MEMORY 0)
+endif ()
+
+if (NOT DEFINED WAMR_BUILD_SHARED_MEMORY)
+  set(WAMR_BUILD_SHARED_MEMORY 0)
+endif ()
+
+if (NOT DEFINED WAMR_BUILD_STRINGREF)
+  set(WAMR_BUILD_STRINGREF 0)
+endif ()
+
+if (NOT DEFINED WAMR_BUILD_TAIL_CALL)
+  set (WAMR_BUILD_TAIL_CALL 0)
+endif ()
+
+########################################
+# Compilation options to marco
 ########################################
 
 message ("-- Build Configurations:")
 message ("     Build as target ${WAMR_BUILD_TARGET}")
+message ("     Build for platform ${WAMR_BUILD_PLATFORM}")
 message ("     CMAKE_BUILD_TYPE " ${CMAKE_BUILD_TYPE})
+message ("     BUILD_SHARED_LIBS " ${BUILD_SHARED_LIBS})
+################## running mode ##################
 if (WAMR_BUILD_INTERP EQUAL 1)
   message ("     WAMR Interpreter enabled")
 else ()
   message ("     WAMR Interpreter disabled")
 endif ()
+if ((WAMR_BUILD_FAST_INTERP EQUAL 1) AND (WAMR_BUILD_INTERP EQUAL 1))
+  add_definitions (-DWASM_ENABLE_FAST_INTERP=1)
+  message ("     Fast interpreter enabled")
+else ()
+  add_definitions (-DWASM_ENABLE_FAST_INTERP=0)
+  message ("     Fast interpreter disabled")
+endif ()
 if (WAMR_BUILD_AOT EQUAL 1)
   message ("     WAMR AOT enabled")
 else ()
@@ -205,6 +263,16 @@ if (WAMR_BUILD_FAST_JIT EQUAL 1 AND WAMR_BUILD_JIT EQUAL 1
     AND WAMR_BUILD_LAZY_JIT EQUAL 1)
   message ("     Multi-tier JIT enabled")
 endif ()
+################## test modes ##################
+if (WAMR_BUILD_SPEC_TEST EQUAL 1)
+  add_definitions (-DWASM_ENABLE_SPEC_TEST=1)
+  message ("     spec test compatible mode is on")
+endif ()
+if (WAMR_BUILD_WASI_TEST EQUAL 1)
+  add_definitions (-DWASM_ENABLE_WASI_TEST=1)
+  message ("     wasi test compatible mode is on")
+endif ()
+################## native ##################
 if (WAMR_BUILD_LIBC_BUILTIN EQUAL 1)
   message ("     Libc builtin enabled")
 else ()
@@ -217,54 +285,6 @@ elseif (WAMR_BUILD_LIBC_WASI EQUAL 1)
 else ()
   message ("     Libc WASI disabled")
 endif ()
-if ((WAMR_BUILD_FAST_INTERP EQUAL 1) AND (WAMR_BUILD_INTERP EQUAL 1))
-  add_definitions (-DWASM_ENABLE_FAST_INTERP=1)
-  message ("     Fast interpreter enabled")
-else ()
-  add_definitions (-DWASM_ENABLE_FAST_INTERP=0)
-  message ("     Fast interpreter disabled")
-endif ()
-if (WAMR_BUILD_MULTI_MODULE EQUAL 1)
-  add_definitions (-DWASM_ENABLE_MULTI_MODULE=1)
-  message ("     Multiple modules enabled")
-else ()
-  add_definitions (-DWASM_ENABLE_MULTI_MODULE=0)
-  message ("     Multiple modules disabled")
-endif ()
-if (WAMR_BUILD_SPEC_TEST EQUAL 1)
-  add_definitions (-DWASM_ENABLE_SPEC_TEST=1)
-  message ("     spec test compatible mode is on")
-endif ()
-if (WAMR_BUILD_WASI_TEST EQUAL 1)
-  add_definitions (-DWASM_ENABLE_WASI_TEST=1)
-  message ("     wasi test compatible mode is on")
-endif ()
-if (NOT DEFINED WAMR_BUILD_BULK_MEMORY)
-  # Enable bulk memory by default
-  set (WAMR_BUILD_BULK_MEMORY 1)
-endif ()
-if (WAMR_BUILD_BULK_MEMORY EQUAL 1)
-  add_definitions (-DWASM_ENABLE_BULK_MEMORY=1)
-  message ("     Bulk memory feature enabled")
-else ()
-  add_definitions (-DWASM_ENABLE_BULK_MEMORY=0)
-  message ("     Bulk memory feature disabled")
-endif ()
-if (WAMR_BUILD_SHARED_MEMORY EQUAL 1)
-  add_definitions (-DWASM_ENABLE_SHARED_MEMORY=1)
-  message ("     Shared memory enabled")
-else ()
-  add_definitions (-DWASM_ENABLE_SHARED_MEMORY=0)
-endif ()
-if (WAMR_BUILD_MEMORY64 EQUAL 1)
-  # if native is 32-bit or cross-compiled to 32-bit
-  if (NOT WAMR_BUILD_TARGET MATCHES ".*64.*")
-    message (FATAL_ERROR "-- Memory64 is only available on the 64-bit platform/target")
-  endif()
-  add_definitions (-DWASM_ENABLE_MEMORY64=1)
-  set (WAMR_DISABLE_HW_BOUND_CHECK 1)
-  message ("     Memory64 memory enabled")
-endif ()
 if (WAMR_BUILD_THREAD_MGR EQUAL 1)
   message ("     Thread manager enabled")
 endif ()
@@ -283,6 +303,53 @@ endif ()
 if (WAMR_BUILD_LIB_RATS EQUAL 1)
   message ("     Lib rats enabled")
 endif()
+if ((WAMR_BUILD_LIB_SIMDE EQUAL 1))
+  message ("     Lib simde enabled")
+endif()
+################## WAMR features ##################
+if (WAMR_BUILD_MULTI_MODULE EQUAL 1)
+  add_definitions (-DWASM_ENABLE_MULTI_MODULE=1)
+  message ("     Multiple modules enabled")
+else ()
+  add_definitions (-DWASM_ENABLE_MULTI_MODULE=0)
+  message ("     Multiple modules disabled")
+endif ()
+if (WAMR_BUILD_BULK_MEMORY EQUAL 1)
+  add_definitions (-DWASM_ENABLE_BULK_MEMORY=1)
+else ()
+  add_definitions (-DWASM_ENABLE_BULK_MEMORY=0)
+endif ()
+if (WAMR_BUILD_SHARED_MEMORY EQUAL 1)
+  add_definitions (-DWASM_ENABLE_SHARED_MEMORY=1)
+  message ("     Shared memory enabled")
+else ()
+  add_definitions (-DWASM_ENABLE_SHARED_MEMORY=0)
+endif ()
+if (WAMR_BUILD_SHARED_HEAP EQUAL 1)
+  add_definitions (-DWASM_ENABLE_SHARED_HEAP=1)
+  message ("     Shared heap enabled")
+endif()
+
+if (WAMR_ENABLE_COPY_CALLSTACK EQUAL 1)
+  add_definitions (-DWAMR_ENABLE_COPY_CALLSTACK=1)
+  message("     Copy callstack enabled")
+else ()
+  add_definitions (-DWAMR_ENABLE_COPY_CALLSTACK=0)
+  message("     Copy callstack disabled")
+endif()
+
+if (WAMR_BUILD_MEMORY64 EQUAL 1)
+  # if native is 32-bit or cross-compiled to 32-bit
+  if (NOT WAMR_BUILD_TARGET MATCHES ".*64.*")
+    message (FATAL_ERROR "-- Memory64 is only available on the 64-bit platform/target")
+  endif()
+  add_definitions (-DWASM_ENABLE_MEMORY64=1)
+  set (WAMR_DISABLE_HW_BOUND_CHECK 1)
+endif ()
+if (WAMR_BUILD_MULTI_MEMORY EQUAL 1)
+  add_definitions (-DWASM_ENABLE_MULTI_MEMORY=1)
+  set (WAMR_BUILD_DEBUG_INTERP 0)
+endif ()
 if (WAMR_BUILD_MINI_LOADER EQUAL 1)
   add_definitions (-DWASM_ENABLE_MINI_LOADER=1)
   message ("     WASM mini loader enabled")
@@ -310,12 +377,17 @@ else ()
   message ("     Wakeup of blocking operations enabled")
 endif ()
 if (WAMR_BUILD_SIMD EQUAL 1)
-  if (NOT WAMR_BUILD_TARGET MATCHES "RISCV64.*")
-    add_definitions (-DWASM_ENABLE_SIMD=1)
-    message ("     SIMD enabled")
-  else ()
+  if (WAMR_BUILD_FAST_INTERP EQUAL 1 AND WAMR_BUILD_SIMDE EQUAL 0)
+    set(SIMD_ENABLED 0)
+    message("     SIMD disabled for fast-interp as simde is not being built")
+  elseif (WAMR_BUILD_TARGET MATCHES "RISCV64.*")
+    set(SIMD_ENABLED 0)
     message ("     SIMD disabled due to not supported on target RISCV64")
+  else()
+    set(SIMD_ENABLED 1)
+    message ("     SIMD enabled")
   endif ()
+  add_definitions(-DWASM_ENABLE_SIMD=${SIMD_ENABLED})
 endif ()
 if (WAMR_BUILD_AOT_STACK_FRAME EQUAL 1)
   add_definitions (-DWASM_ENABLE_AOT_STACK_FRAME=1)
@@ -342,16 +414,11 @@ if (WAMR_BUILD_DUMP_CALL_STACK EQUAL 1)
 endif ()
 if (WAMR_BUILD_TAIL_CALL EQUAL 1)
   add_definitions (-DWASM_ENABLE_TAIL_CALL=1)
-  message ("     Tail call enabled")
 endif ()
 if (WAMR_BUILD_REF_TYPES EQUAL 1)
   add_definitions (-DWASM_ENABLE_REF_TYPES=1)
-  message ("     Reference types enabled")
-else ()
-  message ("     Reference types disabled")
 endif ()
 if (WAMR_BUILD_GC EQUAL 1)
-  message ("     GC enabled")
   if (WAMR_TEST_GC EQUAL 1)
     message("      GC testing enabled")
   endif()
@@ -363,7 +430,6 @@ else ()
   message ("     GC performance profiling disabled")
 endif ()
 if (WAMR_BUILD_STRINGREF EQUAL 1)
-  message ("     Stringref enabled")
   if (NOT DEFINED WAMR_STRINGREF_IMPL_SOURCE)
     message ("       Using WAMR builtin implementation for stringref")
   else ()
@@ -399,6 +465,10 @@ endif ()
 if (WAMR_BUILD_DEBUG_AOT EQUAL 1)
     message ("     Debug AOT enabled")
 endif ()
+if (WAMR_BUILD_DYNAMIC_AOT_DEBUG EQUAL 1)
+    add_definitions (-DWASM_ENABLE_DYNAMIC_AOT_DEBUG=1)
+    message ("     Dynamic AOT debug enabled")
+endif ()
 if (WAMR_BUILD_LOAD_CUSTOM_SECTION EQUAL 1)
     add_definitions (-DWASM_ENABLE_LOAD_CUSTOM_SECTION=1)
     message ("     Load custom section enabled")
@@ -437,7 +507,9 @@ if (WAMR_BUILD_WASI_NN EQUAL 1)
   message ("     WASI-NN enabled")
   add_definitions (-DWASM_ENABLE_WASI_NN=1)
   # Variant backends
-  if (NOT WAMR_BUILD_WASI_NN_TFLITE EQUAL 1 AND NOT WAMR_BUILD_WASI_NN_OPENVINO EQUAL 1)
+  if (NOT WAMR_BUILD_WASI_NN_TFLITE EQUAL 1 AND
+      NOT WAMR_BUILD_WASI_NN_OPENVINO EQUAL 1 AND
+      NOT WAMR_BUILD_WASI_NN_LLAMACPP EQUAL 1)
     message (FATAL_ERROR "   Need to select a backend for WASI-NN")
   endif ()
 
@@ -449,6 +521,10 @@ if (WAMR_BUILD_WASI_NN EQUAL 1)
     message ("     WASI-NN: backend openvino enabled")
     add_definitions (-DWASM_ENABLE_WASI_NN_OPENVINO)
   endif ()
+  if (WAMR_BUILD_WASI_NN_LLAMACPP EQUAL 1)
+    message ("     WASI-NN: backend llamacpp enabled")
+    add_definitions (-DWASM_ENABLE_WASI_NN_LLAMACPP)
+  endif ()
   # Variant devices
   if (WAMR_BUILD_WASI_NN_ENABLE_GPU EQUAL 1)
       message ("     WASI-NN: GPU enabled")
@@ -478,7 +554,7 @@ if (WAMR_BUILD_MODULE_INST_CONTEXT EQUAL 1)
   message ("     Module instance context enabled")
 endif ()
 if (WAMR_BUILD_GC_HEAP_VERIFY EQUAL 1)
-  add_definitions (-DWASM_ENABLE_GC_VERIFY=1)
+  add_definitions (-DBH_ENABLE_GC_VERIFY=1)
   message ("     GC heap verification enabled")
 endif ()
 if ("$ENV{COLLECT_CODE_COVERAGE}" STREQUAL "1" OR COLLECT_CODE_COVERAGE EQUAL 1)
@@ -582,3 +658,47 @@ endif()
 if (NOT WAMR_BUILD_SANITIZER STREQUAL "")
   message ("     Sanitizer ${WAMR_BUILD_SANITIZER} enabled")
 endif ()
+if (WAMR_BUILD_SHRUNK_MEMORY EQUAL 1)
+  add_definitions (-DWASM_ENABLE_SHRUNK_MEMORY=1)
+  message ("     Shrunk memory enabled")
+else ()
+  add_definitions (-DWASM_ENABLE_SHRUNK_MEMORY=0)
+  message ("     Shrunk memory disabled")
+endif()
+if (WAMR_BUILD_AOT_VALIDATOR EQUAL 1)
+  message ("     AOT validator enabled")
+  add_definitions (-DWASM_ENABLE_AOT_VALIDATOR=1)
+endif ()
+
+########################################
+# Show Phase4 Wasm proposals status.
+########################################
+
+message (
+"-- About Wasm Proposals:\n"
+"     Always-on:\n"
+"       \"Extended Constant Expressions\"\n"
+"       \"Multi-value\"\n"
+"       \"Non-trapping float-to-int conversions\"\n"
+"       \"Sign-extension operators\"\n"
+"       \"WebAssembly C and C++ API\"\n"
+"     Configurable. 0 is OFF. 1 is ON:\n"
+"       \"Bulk Memory Operation\" via WAMR_BUILD_BULK_MEMORY: ${WAMR_BUILD_BULK_MEMORY}\n"
+"       \"Fixed-width SIMD\" via WAMR_BUILD_SIMD: ${WAMR_BUILD_SIMD}\n"
+"       \"Garbage collection\" via WAMR_BUILD_GC: ${WAMR_BUILD_GC}\n"
+"       \"Legacy Exception handling\" via WAMR_BUILD_EXCE_HANDLING: ${WAMR_BUILD_EXCE_HANDLING}\n"
+"       \"Memory64\" via WAMR_BUILD_MEMORY64: ${WAMR_BUILD_MEMORY64}\n"
+"       \"Multiple memories\" via WAMR_BUILD_MULTI_MEMORY: ${WAMR_BUILD_MULTI_MEMORY}\n"
+"       \"Reference Types\" via WAMR_BUILD_REF_TYPES: ${WAMR_BUILD_REF_TYPES}\n"
+"       \"Reference-Typed Strings\" via WAMR_BUILD_STRINGREF: ${WAMR_BUILD_STRINGREF}\n"
+"       \"Tail call\" via WAMR_BUILD_TAIL_CALL: ${WAMR_BUILD_TAIL_CALL}\n"
+"       \"Threads\" via WAMR_BUILD_SHARED_MEMORY: ${WAMR_BUILD_SHARED_MEMORY}\n"
+"       \"Typed Function References\" via WAMR_BUILD_GC: ${WAMR_BUILD_GC}\n"
+"     Unsupported (>= Phase4):\n"
+"       \"Branch Hinting\"\n"
+"       \"Custom Annotation Syntax in the Text Format\"\n"
+"       \"Exception handling\"\n"
+"       \"Import/Export of Mutable Globals\"\n"
+"       \"JS String Builtins\"\n"
+"       \"Relaxed SIMD\"\n"
+)

build-scripts/esp-idf/README.md

@@ -11,7 +11,7 @@ You can build an ESP-IDF project with wasm-micro-runtime as a component:
 - In the newly created project folder edit the `CMakeList.txt`:
 
   ```
-  cmake_minimum_required(VERSION 3.5)
+  cmake_minimum_required(VERSION 3.14)
 
   include($ENV{IDF_PATH}/tools/cmake/project.cmake)
 

build-scripts/iwasmConfig.cmake.in

@@ -0,0 +1,6 @@
+# Copyright (C) 2019 Intel Corporation.  All rights reserved.
+# SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+
+@PACKAGE_INIT@
+
+include("${CMAKE_CURRENT_LIST_DIR}/iwasmTargets.cmake")

build-scripts/package.cmake

@@ -0,0 +1,30 @@
+# Copyright (C) 2019 Intel Corporation.  All rights reserved.
+# SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+
+set(_WAMR_BUILD_SCRIPTS_DIR "${CMAKE_CURRENT_LIST_DIR}")
+
+function(install_iwasm_package)
+    install (EXPORT iwasmTargets
+        FILE iwasmTargets.cmake
+        NAMESPACE iwasm::
+        DESTINATION lib/cmake/iwasm
+    )
+
+    include (CMakePackageConfigHelpers)
+    configure_package_config_file (${_WAMR_BUILD_SCRIPTS_DIR}/iwasmConfig.cmake.in
+        "${CMAKE_CURRENT_BINARY_DIR}/iwasmConfig.cmake"
+        INSTALL_DESTINATION lib/cmake/iwasm
+    )
+
+    write_basic_package_version_file(
+        "${CMAKE_CURRENT_BINARY_DIR}/iwasmConfigVersion.cmake"
+        VERSION ${WAMR_VERSION_MAJOR}.${WAMR_VERSION_MINOR}.${WAMR_VERSION_PATCH}
+        COMPATIBILITY SameMajorVersion
+    )
+
+    install (FILES
+        "${CMAKE_CURRENT_BINARY_DIR}/iwasmConfig.cmake"
+        "${CMAKE_CURRENT_BINARY_DIR}/iwasmConfigVersion.cmake"
+        DESTINATION lib/cmake/iwasm
+    )
+endfunction()

build-scripts/runtime_lib.cmake

@@ -14,7 +14,7 @@ if (NOT DEFINED DEPS_DIR)
     set (DEPS_DIR ${WAMR_ROOT_DIR}/core/deps)
 endif ()
 if (NOT DEFINED SHARED_PLATFORM_CONFIG)
-    # CMake file for platform configuration. The PLATFORM_SHARED_SOURCE varable
+    # CMake file for platform configuration. The PLATFORM_SHARED_SOURCE variable
     # should point to a list of platform-specfic source files to compile.
     set (SHARED_PLATFORM_CONFIG ${SHARED_DIR}/platform/${WAMR_BUILD_PLATFORM}/shared_platform.cmake)
 endif ()
@@ -59,7 +59,12 @@ if (WAMR_BUILD_INTERP EQUAL 1)
 endif ()
 
 if (WAMR_BUILD_FAST_JIT EQUAL 1)
-    include (${IWASM_DIR}/fast-jit/iwasm_fast_jit.cmake)
+    if (WAMR_BUILD_PLATFORM STREQUAL "windows")
+        message ("Fast JIT currently not supported on Windows")
+        set (WAMR_BUILD_FAST_JIT 0)
+    else ()
+        include (${IWASM_DIR}/fast-jit/iwasm_fast_jit.cmake)
+    endif ()
 endif ()
 
 if (WAMR_BUILD_JIT EQUAL 1)
@@ -104,6 +109,10 @@ if (WAMR_BUILD_WASI_NN EQUAL 1)
 endif ()
 
 if (WAMR_BUILD_LIB_PTHREAD EQUAL 1)
+    if (WAMR_BUILD_PLATFORM STREQUAL "windows")
+        set (WAMR_BUILD_LIB_PTHREAD_SEMAPHORE 0)
+        message ("Lib pthread semaphore currently not supported on Windows")
+    endif ()
     include (${IWASM_DIR}/libraries/lib-pthread/lib_pthread.cmake)
     # Enable the dependent feature if lib pthread is enabled
     set (WAMR_BUILD_THREAD_MGR 1)
@@ -119,6 +128,10 @@ if (WAMR_BUILD_LIB_WASI_THREADS EQUAL 1)
     set (WAMR_BUILD_SHARED_MEMORY 1)
 endif ()
 
+if (WAMR_BUILD_SHARED_HEAP EQUAL 1)
+    include (${IWASM_DIR}/libraries/shared-heap/shared_heap.cmake)
+endif ()
+
 if (WAMR_BUILD_DEBUG_INTERP EQUAL 1)
     set (WAMR_BUILD_THREAD_MGR 1)
     include (${IWASM_DIR}/libraries/debug-engine/debug_engine.cmake)
@@ -142,6 +155,16 @@ if (WAMR_BUILD_LIB_RATS EQUAL 1)
     include (${IWASM_DIR}/libraries/lib-rats/lib_rats.cmake)
 endif ()
 
+if (WAMR_BUILD_SIMD EQUAL 1 AND WAMR_BUILD_FAST_INTERP EQUAL 1)
+    if (WAMR_BUILD_PLATFORM STREQUAL "windows")
+        message(STATUS "SIMDe doesnt support platform " ${WAMR_BUILD_PLATFORM})
+        set(WAMR_BUILD_SIMDE 0)
+    else()
+        include (${IWASM_DIR}/libraries/simde/simde.cmake)
+        set (WAMR_BUILD_SIMDE 1)
+    endif()
+endif ()
+
 if (WAMR_BUILD_WASM_CACHE EQUAL 1)
     include (${WAMR_ROOT_DIR}/build-scripts/involve_boringssl.cmake)
 endif ()
@@ -193,6 +216,7 @@ set (source_all
     ${LIBC_EMCC_SOURCE}
     ${LIB_RATS_SOURCE}
     ${DEBUG_ENGINE_SOURCE}
+    ${LIB_SHARED_HEAP_SOURCE}
 )
 
 set (WAMR_RUNTIME_LIB_SOURCE ${source_all})

build-scripts/version.cmake

@@ -0,0 +1,28 @@
+# Copyright (C) 2019 Intel Corporation.  All rights reserved.
+# SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+
+if(NOT WAMR_ROOT_DIR)
+  # if from wamr-compiler
+  set(WAMR_ROOT_DIR ${CMAKE_CURRENT_LIST_DIR}/..)
+endif()
+
+set(WAMR_VERSION_MAJOR 2)
+set(WAMR_VERSION_MINOR 2)
+set(WAMR_VERSION_PATCH 0)
+
+message("-- WAMR version: ${WAMR_VERSION_MAJOR}.${WAMR_VERSION_MINOR}.${WAMR_VERSION_PATCH}")
+
+# Configure the version header file
+configure_file(
+  ${WAMR_ROOT_DIR}/core/version.h.in
+  ${WAMR_ROOT_DIR}/core/version.h
+)
+
+# Set the library version and SOVERSION
+function(set_version_info target)
+  set_target_properties(${target}
+    PROPERTIES
+      VERSION ${WAMR_VERSION_MAJOR}.${WAMR_VERSION_MINOR}.${WAMR_VERSION_PATCH}
+      SOVERSION ${WAMR_VERSION_MAJOR}
+)
+endfunction()

ci/coding_guidelines_check.py

@@ -13,8 +13,8 @@ import subprocess
 import sys
 import unittest
 
-CLANG_FORMAT_CMD = "clang-format-12"
-GIT_CLANG_FORMAT_CMD = "git-clang-format-12"
+CLANG_FORMAT_CMD = "clang-format-14"
+GIT_CLANG_FORMAT_CMD = "git-clang-format-14"
 
 # glob style patterns
 EXCLUDE_PATHS = [
@@ -32,7 +32,7 @@ EXCLUDE_PATHS = [
     "**/tests/wamr-test-suites/workspace/*",
 ]
 
-C_SUFFIXES = [".c", ".cpp", ".h"]
+C_SUFFIXES = [".c", ".cc", ".cpp", ".h"]
 INVALID_DIR_NAME_SEGMENT = r"([a-zA-Z0-9]+\_[a-zA-Z0-9]+)"
 INVALID_FILE_NAME_SEGMENT = r"([a-zA-Z0-9]+\-[a-zA-Z0-9]+)"
 
@@ -93,20 +93,29 @@ def run_clang_format(file_path: Path, root: Path) -> bool:
 
 def run_clang_format_diff(root: Path, commits: str) -> bool:
     """
-    Use `clang-format-12` or `git-clang-format-12` to check code format of
+    Use `clang-format-14` or `git-clang-format-14` to check code format of
     the PR, with a commit range specified. It is required to format the
     code before committing the PR, or it might fail to pass the CI check:
 
-    1. Install clang-format-12.0.0
-    Normally we can install it by `sudo apt-get install clang-format-12`,
-    or download the `clang+llvm-12.0.0-xxx-tar.xz` package from
-      https://github.com/llvm/llvm-project/releases/tag/llvmorg-12.0.0
-    and install it
+    1. Install clang-format-14.0.0
+
+    You can download the package from
+    https://github.com/llvm/llvm-project/releases
+    and install it.
+
+    For Debian/Ubuntu, we can probably use
+    `sudo apt-get install clang-format-14`.
+
+    Homebrew has it as a part of llvm@14.
+    ```shell
+    brew install llvm@14
+    /usr/local/opt/llvm@14/bin/clang-format
+    ```
 
     2. Format the C/C++ source file
     ``` shell
     cd path/to/wamr/root
-    clang-format-12 --style file -i path/to/file
+    clang-format-14 --style file -i path/to/file
     ```
 
     The code wrapped by `/* clang-format off */` and `/* clang-format on */`
@@ -145,7 +154,7 @@ def run_clang_format_diff(root: Path, commits: str) -> bool:
         found = False
         for summary in [x for x in diff_content if x.startswith("diff --git")]:
             # b/path/to/file -> path/to/file
-            with_invalid_format = re.split("\s+", summary)[-1][2:]
+            with_invalid_format = re.split(r"\s+", summary)[-1][2:]
             if not is_excluded(with_invalid_format):
                 print(f"--- {with_invalid_format} failed on code style checking.")
                 found = True
@@ -180,6 +189,7 @@ def check_file_name(path: Path) -> bool:
         "docker-compose",
         "package-lock",
         "vite-env.d",
+        "osv-scanner",
     ]:
         return True
 

core/config.h

@@ -75,12 +75,16 @@
 #define WASM_ENABLE_AOT 0
 #endif
 
+#ifndef WASM_ENABLE_DYNAMIC_AOT_DEBUG
+#define WASM_ENABLE_DYNAMIC_AOT_DEBUG 0
+#endif
+
 #ifndef WASM_ENABLE_WORD_ALIGN_READ
 #define WASM_ENABLE_WORD_ALIGN_READ 0
 #endif
 
 #define AOT_MAGIC_NUMBER 0x746f6100
-#define AOT_CURRENT_VERSION 3
+#define AOT_CURRENT_VERSION 4
 
 #ifndef WASM_ENABLE_JIT
 #define WASM_ENABLE_JIT 0
@@ -189,6 +193,10 @@
 #error "Heap aux stack allocation must be enabled for WASI threads"
 #endif
 
+#ifndef WAMR_ENABLE_COPY_CALLSTACK
+#define WAMR_ENABLE_COPY_CALLSTACK 0
+#endif
+
 #ifndef WASM_ENABLE_BASE_LIB
 #define WASM_ENABLE_BASE_LIB 0
 #endif
@@ -314,6 +322,12 @@
 #define WASM_ENABLE_SIMD 0
 #endif
 
+/* Disable SIMDe (used in the fast interpreter for SIMD opcodes)
+unless used elsewhere */
+#ifndef WASM_ENABLE_SIMDE
+#define WASM_ENABLE_SIMDE 0
+#endif
+
 /* GC performance profiling */
 #ifndef WASM_ENABLE_GC_PERF_PROFILING
 #define WASM_ENABLE_GC_PERF_PROFILING 0
@@ -392,7 +406,9 @@
 #define APP_HEAP_SIZE_DEFAULT (8 * 1024)
 #endif
 #define APP_HEAP_SIZE_MIN (256)
-#define APP_HEAP_SIZE_MAX (512 * 1024 * 1024)
+/* The ems memory allocator supports maximal heap size 1GB,
+   see ems_gc_internal.h */
+#define APP_HEAP_SIZE_MAX (1024 * 1024 * 1024)
 
 /* Default min/max gc heap size of each app */
 #ifndef GC_HEAP_SIZE_DEFAULT
@@ -664,6 +680,11 @@
 #define WASM_ENABLE_MEMORY64 0
 #endif
 
+/* Disable multi-memory by default */
+#ifndef WASM_ENABLE_MULTI_MEMORY
+#define WASM_ENABLE_MULTI_MEMORY 0
+#endif
+
 #ifndef WASM_TABLE_MAX_SIZE
 #define WASM_TABLE_MAX_SIZE 1024
 #endif
@@ -683,4 +704,16 @@
 #endif
 #endif /* WASM_ENABLE_FUZZ_TEST != 0 */
 
+#ifndef WASM_ENABLE_SHARED_HEAP
+#define WASM_ENABLE_SHARED_HEAP 0
+#endif
+
+#ifndef WASM_ENABLE_SHRUNK_MEMORY
+#define WASM_ENABLE_SHRUNK_MEMORY 1
+#endif
+
+#ifndef WASM_ENABLE_AOT_VALIDATOR
+#define WASM_ENABLE_AOT_VALIDATOR 0
+#endif
+
 #endif /* end of _CONFIG_H_ */

core/iwasm/aot/aot_intrinsic.c

@@ -194,7 +194,7 @@ aot_intrinsic_clz_i32(uint32 type)
     return num;
 }
 
-uint32
+uint64
 aot_intrinsic_clz_i64(uint64 type)
 {
     uint32 num = 0;
@@ -220,7 +220,7 @@ aot_intrinsic_ctz_i32(uint32 type)
     return num;
 }
 
-uint32
+uint64
 aot_intrinsic_ctz_i64(uint64 type)
 {
     uint32 num = 0;
@@ -244,7 +244,7 @@ aot_intrinsic_popcnt_i32(uint32 u)
     return ret;
 }
 
-uint32
+uint64
 aot_intrinsic_popcnt_i64(uint64 u)
 {
     uint32 ret = 0;
@@ -485,6 +485,30 @@ aot_intrinsic_i64_bit_and(uint64 l, uint64 r)
     return l & r;
 }
 
+uint64
+aot_intrinsic_i64_mul(uint64 l, uint64 r)
+{
+    return l * r;
+}
+
+uint64
+aot_intrinsic_i64_shl(uint64 l, uint64 r)
+{
+    return l << r;
+}
+
+uint64
+aot_intrinsic_i64_shr_s(uint64 l, uint64 r)
+{
+    return (int64)l >> r;
+}
+
+uint64
+aot_intrinsic_i64_shr_u(uint64 l, uint64 r)
+{
+    return l >> r;
+}
+
 #if WASM_ENABLE_WAMR_COMPILER != 0 || WASM_ENABLE_JIT != 0
 
 typedef struct {
@@ -561,6 +585,10 @@ static const aot_intrinsic g_intrinsic_mapping[] = {
     { "i64.rem_u", "aot_intrinsic_i64_rem_u", AOT_INTRINSIC_FLAG_I64_REM_U},
     { "i64.or", "aot_intrinsic_i64_bit_or", AOT_INTRINSIC_FLAG_I64_BIT_OR},
     { "i64.and", "aot_intrinsic_i64_bit_and", AOT_INTRINSIC_FLAG_I64_BIT_AND},
+    { "i64.mul", "aot_intrinsic_i64_mul", AOT_INTRINSIC_FLAG_I64_MUL},
+    { "i64.shl", "aot_intrinsic_i64_shl", AOT_INTRINSIC_FLAG_I64_SHL},
+    { "i64.shr_s", "aot_intrinsic_i64_shr_s", AOT_INTRINSIC_FLAG_I64_SHR_S},
+    { "i64.shr_u", "aot_intrinsic_i64_shr_u", AOT_INTRINSIC_FLAG_I64_SHR_U},
 };
 /* clang-format on */
 
@@ -601,6 +629,10 @@ add_i64_common_intrinsics(AOTCompContext *comp_ctx)
     add_intrinsic_capability(comp_ctx, AOT_INTRINSIC_FLAG_I64_REM_U);
     add_intrinsic_capability(comp_ctx, AOT_INTRINSIC_FLAG_I64_BIT_OR);
     add_intrinsic_capability(comp_ctx, AOT_INTRINSIC_FLAG_I64_BIT_AND);
+    add_intrinsic_capability(comp_ctx, AOT_INTRINSIC_FLAG_I64_MUL);
+    add_intrinsic_capability(comp_ctx, AOT_INTRINSIC_FLAG_I64_SHL);
+    add_intrinsic_capability(comp_ctx, AOT_INTRINSIC_FLAG_I64_SHR_S);
+    add_intrinsic_capability(comp_ctx, AOT_INTRINSIC_FLAG_I64_SHR_U);
 }
 
 static void
@@ -866,6 +898,17 @@ aot_intrinsic_fill_capability_flags(AOTCompContext *comp_ctx)
         if (!strncmp(comp_ctx->target_arch, "riscv32", 7)) {
             add_i64_common_intrinsics(comp_ctx);
         }
+        /*
+         * LLVM 16 and later expands cttz intrinsic to a table lookup,
+         * which involves some relocations. (unless ZBB is available,
+         * in which case the native instructions are preferred over
+         * the table-based lowering.)
+         * https://reviews.llvm.org/D128911
+         */
+#if LLVM_VERSION_MAJOR >= 16
+        add_intrinsic_capability(comp_ctx, AOT_INTRINSIC_FLAG_I32_CTZ);
+        add_intrinsic_capability(comp_ctx, AOT_INTRINSIC_FLAG_I64_CTZ);
+#endif
     }
     else if (!strncmp(comp_ctx->target_arch, "xtensa", 6)) {
         /*
@@ -878,9 +921,6 @@ aot_intrinsic_fill_capability_flags(AOTCompContext *comp_ctx)
         add_i64_common_intrinsics(comp_ctx);
         add_common_float_integer_conversion(comp_ctx);
         add_intrinsic_capability(comp_ctx, AOT_INTRINSIC_FLAG_F32_CONST);
-        add_intrinsic_capability(comp_ctx, AOT_INTRINSIC_FLAG_F64_CONST);
-        add_intrinsic_capability(comp_ctx, AOT_INTRINSIC_FLAG_I32_CONST);
-        add_intrinsic_capability(comp_ctx, AOT_INTRINSIC_FLAG_I64_CONST);
     }
     else {
         /*

core/iwasm/aot/aot_intrinsic.h

@@ -98,6 +98,10 @@ extern "C" {
 #define AOT_INTRINSIC_FLAG_I64_REM_U    AOT_INTRINSIC_FLAG(1, 31)
 #define AOT_INTRINSIC_FLAG_I64_BIT_OR   AOT_INTRINSIC_FLAG(1, 32)
 #define AOT_INTRINSIC_FLAG_I64_BIT_AND  AOT_INTRINSIC_FLAG(1, 33)
+#define AOT_INTRINSIC_FLAG_I64_MUL      AOT_INTRINSIC_FLAG(1, 34)
+#define AOT_INTRINSIC_FLAG_I64_SHL      AOT_INTRINSIC_FLAG(1, 35)
+#define AOT_INTRINSIC_FLAG_I64_SHR_S    AOT_INTRINSIC_FLAG(1, 36)
+#define AOT_INTRINSIC_FLAG_I64_SHR_U    AOT_INTRINSIC_FLAG(1, 37)
 
 /* clang-format on */
 
@@ -182,19 +186,19 @@ aot_intrinsic_fmax_f64(float64 a, float64 b);
 uint32
 aot_intrinsic_clz_i32(uint32 type);
 
-uint32
+uint64
 aot_intrinsic_clz_i64(uint64 type);
 
 uint32
 aot_intrinsic_ctz_i32(uint32 type);
 
-uint32
+uint64
 aot_intrinsic_ctz_i64(uint64 type);
 
 uint32
 aot_intrinsic_popcnt_i32(uint32 u);
 
-uint32
+uint64
 aot_intrinsic_popcnt_i64(uint64 u);
 
 float32
@@ -287,6 +291,18 @@ aot_intrinsic_i64_bit_or(uint64 l, uint64 r);
 uint64
 aot_intrinsic_i64_bit_and(uint64 l, uint64 r);
 
+uint64
+aot_intrinsic_i64_mul(uint64 l, uint64 r);
+
+uint64
+aot_intrinsic_i64_shl(uint64 l, uint64 r);
+
+uint64
+aot_intrinsic_i64_shr_s(uint64 l, uint64 r);
+
+uint64
+aot_intrinsic_i64_shr_u(uint64 l, uint64 r);
+
 #if WASM_ENABLE_WAMR_COMPILER != 0 || WASM_ENABLE_JIT != 0
 const char *
 aot_intrinsic_get_symbol(const char *llvm_intrinsic);

core/iwasm/aot/aot_loader.c

@@ -4,13 +4,15 @@
  */
 
 #include "aot_runtime.h"
-#include "bh_common.h"
-#include "bh_log.h"
 #include "aot_reloc.h"
+#include "bh_platform.h"
 #include "../common/wasm_runtime_common.h"
 #include "../common/wasm_native.h"
 #include "../common/wasm_loader_common.h"
 #include "../compilation/aot.h"
+#if WASM_ENABLE_AOT_VALIDATOR != 0
+#include "aot_validator.h"
+#endif
 
 #if WASM_ENABLE_DEBUG_AOT != 0
 #include "debug/elf_parser.h"
@@ -294,6 +296,47 @@ loader_malloc(uint64 size, char *error_buf, uint32 error_buf_size)
     return mem;
 }
 
+static void *
+loader_mmap(uint32 size, bool prot_exec, char *error_buf, uint32 error_buf_size)
+{
+    int map_prot =
+        MMAP_PROT_READ | MMAP_PROT_WRITE | (prot_exec ? MMAP_PROT_EXEC : 0);
+    int map_flags;
+    void *mem;
+
+#if defined(BUILD_TARGET_X86_64) || defined(BUILD_TARGET_AMD_64) \
+    || defined(BUILD_TARGET_RISCV64_LP64D)                       \
+    || defined(BUILD_TARGET_RISCV64_LP64)
+#if !defined(__APPLE__) && !defined(BH_PLATFORM_LINUX_SGX)
+    /* The mmapped AOT data and code in 64-bit targets had better be in
+       range 0 to 2G, or aot loader may fail to apply some relocations,
+       e.g., R_X86_64_32/R_X86_64_32S/R_X86_64_PC32/R_RISCV_32.
+       We try to mmap with MMAP_MAP_32BIT flag first, and if fails, mmap
+       again without the flag. */
+    /* sgx_tprotect_rsrv_mem() and sgx_alloc_rsrv_mem() will ignore flags */
+    map_flags = MMAP_MAP_32BIT;
+    if ((mem = os_mmap(NULL, size, map_prot, map_flags,
+                       os_get_invalid_handle()))) {
+        /* Test whether the mmapped memory in the first 2 Gigabytes of the
+           process address space */
+        if ((uintptr_t)mem >= INT32_MAX)
+            LOG_WARNING(
+                "Warning: loader mmap memory address is not in the first 2 "
+                "Gigabytes of the process address space.");
+        return mem;
+    }
+#endif
+#endif
+
+    map_flags = MMAP_MAP_NONE;
+    if (!(mem = os_mmap(NULL, size, map_prot, map_flags,
+                        os_get_invalid_handle()))) {
+        set_error_buf(error_buf, error_buf_size, "allocate memory failed");
+        return NULL;
+    }
+    return mem;
+}
+
 static char *
 load_string(uint8 **p_buf, const uint8 *buf_end, AOTModule *module,
             bool is_load_from_file_buf,
@@ -546,6 +589,11 @@ load_target_info_section(const uint8 *buf, const uint8 *buf_end,
         return false;
     }
 
+    /* for backwards compatibility with previous wamrc aot files */
+    if (!strcmp(target_info.arch, "arm64")
+        || !strcmp(target_info.arch, "aarch64"))
+        bh_strcpy_s(target_info.arch, sizeof(target_info.arch), "aarch64v8");
+
     /* Check machine info */
     if (!check_machine_info(&target_info, error_buf, error_buf_size)) {
         return false;
@@ -556,6 +604,10 @@ load_target_info_section(const uint8 *buf, const uint8 *buf_end,
         return false;
     }
 
+#if WASM_ENABLE_DUMP_CALL_STACK != 0
+    module->feature_flags = target_info.feature_flags;
+#endif
+
     /* Finally, check feature flags */
     return check_feature_flags(error_buf, error_buf_size,
                                target_info.feature_flags);
@@ -589,73 +641,6 @@ str2uint32(const char *buf, uint32 *p_res);
 static bool
 str2uint64(const char *buf, uint64 *p_res);
 
-#if WASM_ENABLE_MULTI_MODULE != 0
-static void *
-aot_loader_resolve_function(const AOTModule *module, const char *function_name,
-                            const AOTFuncType *expected_function_type,
-                            char *error_buf, uint32 error_buf_size);
-
-static void *
-aot_loader_resolve_function_ex(const char *module_name,
-                               const char *function_name,
-                               const AOTFuncType *expected_function_type,
-                               char *error_buf, uint32 error_buf_size)
-{
-    WASMModuleCommon *module_reg;
-
-    module_reg = wasm_runtime_find_module_registered(module_name);
-    if (!module_reg || module_reg->module_type != Wasm_Module_AoT) {
-        LOG_DEBUG("can not find a module named %s for function %s", module_name,
-                  function_name);
-        set_error_buf(error_buf, error_buf_size, "unknown import");
-        return NULL;
-    }
-    return aot_loader_resolve_function((AOTModule *)module_reg, function_name,
-                                       expected_function_type, error_buf,
-                                       error_buf_size);
-}
-
-static void *
-aot_loader_resolve_function(const AOTModule *module, const char *function_name,
-                            const AOTFuncType *expected_function_type,
-                            char *error_buf, uint32 error_buf_size)
-{
-    void *function = NULL;
-    AOTExport *export = NULL;
-    AOTFuncType *target_function_type = NULL;
-
-    export = loader_find_export((WASMModuleCommon *)module, module->name,
-                                function_name, EXPORT_KIND_FUNC, error_buf,
-                                error_buf_size);
-    if (!export) {
-        return NULL;
-    }
-
-    /* resolve function type and function */
-    if (export->index < module->import_func_count) {
-        target_function_type = module->import_funcs[export->index].func_type;
-        function = module->import_funcs[export->index].func_ptr_linked;
-    }
-    else {
-        target_function_type =
-            (AOTFuncType *)module
-                ->types[module->func_type_indexes[export->index
-                                                  - module->import_func_count]];
-        function =
-            (module->func_ptrs[export->index - module->import_func_count]);
-    }
-    /* check function type */
-    if (!wasm_type_equal((WASMType *)expected_function_type,
-                         (WASMType *)target_function_type, module->types,
-                         module->type_count)) {
-        LOG_DEBUG("%s.%s failed the type check", module->name, function_name);
-        set_error_buf(error_buf, error_buf_size, "incompatible import type");
-        return NULL;
-    }
-    return function;
-}
-#endif /* end of WASM_ENABLE_MULTI_MODULE */
-
 static bool
 load_native_symbol_section(const uint8 *buf, const uint8 *buf_end,
                            AOTModule *module, bool is_load_from_file_buf,
@@ -989,13 +974,29 @@ destroy_import_memories(AOTImportMemory *import_memories)
     wasm_runtime_free(import_memories);
 }
 
+/**
+ * Free memory initialization data segments.
+ *
+ * @param module the AOT module containing the data
+ * @param data_list array of memory initialization data segments to free
+ * @param count number of segments in the data_list array
+ */
+
 static void
-destroy_mem_init_data_list(AOTMemInitData **data_list, uint32 count)
+destroy_mem_init_data_list(AOTModule *module, AOTMemInitData **data_list,
+                           uint32 count)
 {
     uint32 i;
+    /* Free each memory initialization data segment */
     for (i = 0; i < count; i++)
-        if (data_list[i])
+        if (data_list[i]) {
+            /* If the module owns the binary data, free the bytes buffer */
+            if (module->is_binary_freeable && data_list[i]->bytes)
+                wasm_runtime_free(data_list[i]->bytes);
+            /* Free the data segment structure itself */
             wasm_runtime_free(data_list[i]);
+        }
+    /* Free the array of data segment pointers */
     wasm_runtime_free(data_list);
 }
 
@@ -1004,6 +1005,22 @@ load_init_expr(const uint8 **p_buf, const uint8 *buf_end, AOTModule *module,
                InitializerExpression *expr, char *error_buf,
                uint32 error_buf_size);
 
+/**
+ * Load memory initialization data segments from the AOT module.
+ *
+ * This function reads memory initialization data segments from the buffer and
+ * creates AOTMemInitData structures for each segment. The data can either be
+ * cloned into new memory or referenced directly from the buffer.
+ *
+ * @param p_buf pointer to buffer containing memory init data
+ * @param buf_end end of buffer
+ * @param module the AOT module being loaded
+ * @param error_buf buffer for error messages
+ * @param error_buf_size size of error buffer
+ *
+ * @return true if successful, false if error occurred
+ */
+
 static bool
 load_mem_init_data_list(const uint8 **p_buf, const uint8 *buf_end,
                         AOTModule *module, char *error_buf,
@@ -1035,8 +1052,8 @@ load_mem_init_data_list(const uint8 **p_buf, const uint8 *buf_end,
             return false;
         }
         read_uint32(buf, buf_end, byte_count);
-        size = offsetof(AOTMemInitData, bytes) + (uint64)byte_count;
-        if (!(data_list[i] = loader_malloc(size, error_buf, error_buf_size))) {
+        if (!(data_list[i] = loader_malloc(sizeof(AOTMemInitData), error_buf,
+                                           error_buf_size))) {
             return false;
         }
 
@@ -1048,8 +1065,22 @@ load_mem_init_data_list(const uint8 **p_buf, const uint8 *buf_end,
         data_list[i]->offset.init_expr_type = init_value.init_expr_type;
         data_list[i]->offset.u = init_value.u;
         data_list[i]->byte_count = byte_count;
-        read_byte_array(buf, buf_end, data_list[i]->bytes,
-                        data_list[i]->byte_count);
+        data_list[i]->bytes = NULL;
+        /* If the module owns the binary data, clone the bytes buffer */
+        if (module->is_binary_freeable) {
+            if (byte_count > 0) {
+                if (!(data_list[i]->bytes = loader_malloc(byte_count, error_buf,
+                                                          error_buf_size))) {
+                    return false;
+                }
+                read_byte_array(buf, buf_end, data_list[i]->bytes,
+                                data_list[i]->byte_count);
+            }
+        }
+        else {
+            data_list[i]->bytes = (uint8 *)buf;
+            buf += byte_count;
+        }
     }
 
     *p_buf = buf;
@@ -1058,6 +1089,21 @@ fail:
     return false;
 }
 
+/**
+ * Load memory information from the AOT module.
+ *
+ * This function reads memory-related data including import memory count,
+ * memory count, memory flags, page sizes, and memory initialization data.
+ *
+ * @param p_buf pointer to buffer containing memory info
+ * @param buf_end end of buffer
+ * @param module the AOT module being loaded
+ * @param error_buf buffer for error messages
+ * @param error_buf_size size of error buffer
+ *
+ * @return true if successful, false if error occurred
+ */
+
 static bool
 load_memory_info(const uint8 **p_buf, const uint8 *buf_end, AOTModule *module,
                  char *error_buf, uint32 error_buf_size)
@@ -1067,9 +1113,6 @@ load_memory_info(const uint8 **p_buf, const uint8 *buf_end, AOTModule *module,
     const uint8 *buf = *p_buf;
 
     read_uint32(buf, buf_end, module->import_memory_count);
-    /* We don't support import_memory_count > 0 currently */
-    if (module->import_memory_count > 0)
-        return false;
 
     read_uint32(buf, buf_end, module->memory_count);
     total_size = sizeof(AOTMemory) * (uint64)module->memory_count;
@@ -1212,6 +1255,7 @@ load_init_expr(const uint8 **p_buf, const uint8 *buf_end, AOTModule *module,
             }
             free_if_fail = true;
             init_values->count = field_count;
+            init_values->type_idx = type_idx;
             expr->u.data = init_values;
 
             if (type_idx >= module->type_count) {
@@ -1355,7 +1399,12 @@ load_import_table_list(const uint8 **p_buf, const uint8 *buf_end,
         if (wasm_is_type_multi_byte_type(import_table->table_type.elem_type)) {
             read_uint8(buf, buf_end, ref_type.ref_ht_common.nullable);
         }
+        else
 #endif
+        {
+            /* Skip 1 byte */
+            buf += 1;
+        }
         read_uint32(buf, buf_end, import_table->table_type.init_size);
         read_uint32(buf, buf_end, import_table->table_type.max_size);
 #if WASM_ENABLE_GC != 0
@@ -1404,12 +1453,23 @@ load_table_list(const uint8 **p_buf, const uint8 *buf_end, AOTModule *module,
     for (i = 0; i < module->table_count; i++, table++) {
         read_uint8(buf, buf_end, table->table_type.elem_type);
         read_uint8(buf, buf_end, table->table_type.flags);
+
+        if (!wasm_table_check_flags(table->table_type.flags, error_buf,
+                                    error_buf_size, true)) {
+            return false;
+        }
+
         read_uint8(buf, buf_end, table->table_type.possible_grow);
 #if WASM_ENABLE_GC != 0
         if (wasm_is_type_multi_byte_type(table->table_type.elem_type)) {
             read_uint8(buf, buf_end, ref_type.ref_ht_common.nullable);
         }
+        else
 #endif
+        {
+            /* Skip 1 byte */
+            buf += 1;
+        }
         read_uint32(buf, buf_end, table->table_type.init_size);
         read_uint32(buf, buf_end, table->table_type.max_size);
 #if WASM_ENABLE_GC != 0
@@ -1497,7 +1557,7 @@ load_table_init_data_list(const uint8 **p_buf, const uint8 *buf_end,
         else
 #endif
         {
-            /* Skip 8 byte for ref type info */
+            /* Skip 8 byte(2+2+4) for ref type info */
             buf += 8;
         }
 
@@ -1655,7 +1715,7 @@ load_types(const uint8 **p_buf, const uint8 *buf_end, AOTModule *module,
         read_uint16(buf, buf_end, type_flag);
 
         read_uint8(buf, buf_end, is_equivalence_type);
-        /* If there is an equivalence type, re-use it */
+        /* If there is an equivalence type, reuse it */
         if (is_equivalence_type) {
             uint8 u8;
             /* padding */
@@ -2240,19 +2300,13 @@ destroy_import_funcs(AOTImportFunc *import_funcs)
 
 static bool
 load_import_funcs(const uint8 **p_buf, const uint8 *buf_end, AOTModule *module,
-                  bool is_load_from_file_buf, char *error_buf,
+                  bool is_load_from_file_buf, bool no_resolve, char *error_buf,
                   uint32 error_buf_size)
 {
-    char *module_name, *field_name;
     const uint8 *buf = *p_buf;
     AOTImportFunc *import_funcs;
     uint64 size;
     uint32 i;
-#if WASM_ENABLE_MULTI_MODULE != 0
-    AOTModule *sub_module = NULL;
-    AOTFunc *linked_func = NULL;
-    AOTFuncType *declare_func_type = NULL;
-#endif
 
     /* Allocate memory */
     size = sizeof(AOTImportFunc) * (uint64)module->import_func_count;
@@ -2269,53 +2323,17 @@ load_import_funcs(const uint8 **p_buf, const uint8 *buf_end, AOTModule *module,
             return false;
         }
 
-#if WASM_ENABLE_MULTI_MODULE != 0
-        declare_func_type =
-            (AOTFuncType *)module->types[import_funcs[i].func_type_index];
-        read_string(buf, buf_end, module_name);
-        read_string(buf, buf_end, field_name);
-
-        import_funcs[i].module_name = module_name;
-        import_funcs[i].func_name = field_name;
-        linked_func = wasm_native_resolve_symbol(
-            module_name, field_name, declare_func_type,
-            &import_funcs[i].signature, &import_funcs[i].attachment,
-            &import_funcs[i].call_conv_raw);
-        if (!linked_func) {
-            sub_module = NULL;
-            if (!wasm_runtime_is_built_in_module(module_name)) {
-                sub_module = (AOTModule *)wasm_runtime_load_depended_module(
-                    (WASMModuleCommon *)module, module_name, error_buf,
-                    error_buf_size);
-                if (!sub_module) {
-                    LOG_ERROR("failed to load sub module: %s", error_buf);
-                    return false;
-                }
-            }
-            if (!sub_module)
-                linked_func = aot_loader_resolve_function_ex(
-                    module_name, field_name, declare_func_type, error_buf,
-                    error_buf_size);
-            else
-                linked_func = aot_loader_resolve_function(
-                    sub_module, field_name, declare_func_type, error_buf,
-                    error_buf_size);
-        }
-        import_funcs[i].func_ptr_linked = linked_func;
-        import_funcs[i].func_type = declare_func_type;
-
-#else
         import_funcs[i].func_type =
             (AOTFuncType *)module->types[import_funcs[i].func_type_index];
         read_string(buf, buf_end, import_funcs[i].module_name);
         read_string(buf, buf_end, import_funcs[i].func_name);
-        module_name = import_funcs[i].module_name;
-        field_name = import_funcs[i].func_name;
-        import_funcs[i].func_ptr_linked = wasm_native_resolve_symbol(
-            module_name, field_name, import_funcs[i].func_type,
-            &import_funcs[i].signature, &import_funcs[i].attachment,
-            &import_funcs[i].call_conv_raw);
-#endif
+        import_funcs[i].attachment = NULL;
+        import_funcs[i].signature = NULL;
+        import_funcs[i].call_conv_raw = false;
+
+        if (!no_resolve) {
+            aot_resolve_import_func(module, &import_funcs[i]);
+        }
 
 #if WASM_ENABLE_LIBC_WASI != 0
         if (!strcmp(import_funcs[i].module_name, "wasi_unstable")
@@ -2333,7 +2351,7 @@ fail:
 static bool
 load_import_func_info(const uint8 **p_buf, const uint8 *buf_end,
                       AOTModule *module, bool is_load_from_file_buf,
-                      char *error_buf, uint32 error_buf_size)
+                      bool no_resolve, char *error_buf, uint32 error_buf_size)
 {
     const uint8 *buf = *p_buf;
 
@@ -2342,7 +2360,7 @@ load_import_func_info(const uint8 **p_buf, const uint8 *buf_end,
     /* load import funcs */
     if (module->import_func_count > 0
         && !load_import_funcs(&buf, buf_end, module, is_load_from_file_buf,
-                              error_buf, error_buf_size))
+                              no_resolve, error_buf, error_buf_size))
         return false;
 
     *p_buf = buf;
@@ -2378,7 +2396,6 @@ destroy_object_data_sections(AOTObjectDataSection *data_sections,
                 }
             }
 #endif
-            os_munmap(data_section->data, data_section->size);
         }
     wasm_runtime_free(data_sections);
 }
@@ -2392,6 +2409,9 @@ load_object_data_sections(const uint8 **p_buf, const uint8 *buf_end,
     AOTObjectDataSection *data_sections;
     uint64 size;
     uint32 i;
+    uint64 total_size = 0;
+    uint32 page_size = os_getpagesize();
+    uint8 *merged_sections = NULL;
 
     /* Allocate memory */
     size = sizeof(AOTObjectDataSection) * (uint64)module->data_section_count;
@@ -2400,41 +2420,40 @@ load_object_data_sections(const uint8 **p_buf, const uint8 *buf_end,
         return false;
     }
 
-    /* Create each data section */
+    /* First iteration: read data from buf, and calculate total memory needed */
     for (i = 0; i < module->data_section_count; i++) {
-        int map_prot = MMAP_PROT_READ | MMAP_PROT_WRITE;
-#if defined(BUILD_TARGET_X86_64) || defined(BUILD_TARGET_AMD_64) \
-    || defined(BUILD_TARGET_RISCV64_LP64D)                       \
-    || defined(BUILD_TARGET_RISCV64_LP64)
-        /* aot code and data in x86_64 must be in range 0 to 2G due to
-           relocation for R_X86_64_32/32S/PC32 */
-        int map_flags = MMAP_MAP_32BIT;
-#else
-        int map_flags = MMAP_MAP_NONE;
-#endif
-
         read_string(buf, buf_end, data_sections[i].name);
         read_uint32(buf, buf_end, data_sections[i].size);
-
+        CHECK_BUF(buf, buf_end, data_sections[i].size);
+        /* Temporary record data ptr for merge, will be replaced after the
+           merged_data_sections is mmapped */
+        if (data_sections[i].size > 0)
+            data_sections[i].data = (uint8 *)buf;
+        buf += data_sections[i].size;
+        total_size += align_uint64((uint64)data_sections[i].size, page_size);
+    }
+    if (total_size > UINT32_MAX) {
+        set_error_buf(error_buf, error_buf_size, "data sections too large");
+        return false;
+    }
+    if (total_size > 0) {
         /* Allocate memory for data */
-        if (data_sections[i].size > 0
-            && !(data_sections[i].data =
-                     os_mmap(NULL, data_sections[i].size, map_prot, map_flags,
-                             os_get_invalid_handle()))) {
-            set_error_buf(error_buf, error_buf_size, "allocate memory failed");
+        merged_sections = module->merged_data_sections =
+            loader_mmap((uint32)total_size, false, error_buf, error_buf_size);
+        if (!merged_sections) {
             return false;
         }
-#if defined(BUILD_TARGET_X86_64) || defined(BUILD_TARGET_AMD_64)
-#if !defined(BH_PLATFORM_LINUX_SGX) && !defined(BH_PLATFORM_WINDOWS) \
-    && !defined(BH_PLATFORM_DARWIN)
-        /* address must be in the first 2 Gigabytes of
-           the process address space */
-        bh_assert((uintptr_t)data_sections[i].data < INT32_MAX);
-#endif
-#endif
+        module->merged_data_sections_size = (uint32)total_size;
+    }
 
-        read_byte_array(buf, buf_end, data_sections[i].data,
-                        data_sections[i].size);
+    /* Second iteration: Create each data section */
+    for (i = 0; i < module->data_section_count; i++) {
+        if (data_sections[i].size > 0) {
+            bh_memcpy_s(merged_sections, data_sections[i].size,
+                        data_sections[i].data, data_sections[i].size);
+            data_sections[i].data = merged_sections;
+            merged_sections += align_uint(data_sections[i].size, page_size);
+        }
     }
 
     *p_buf = buf;
@@ -2468,7 +2487,7 @@ fail:
 static bool
 load_init_data_section(const uint8 *buf, const uint8 *buf_end,
                        AOTModule *module, bool is_load_from_file_buf,
-                       char *error_buf, uint32 error_buf_size)
+                       bool no_resolve, char *error_buf, uint32 error_buf_size)
 {
     const uint8 *p = buf, *p_end = buf_end;
 
@@ -2479,7 +2498,7 @@ load_init_data_section(const uint8 *buf, const uint8 *buf_end,
                                     error_buf, error_buf_size)
         || !load_global_info(&p, p_end, module, error_buf, error_buf_size)
         || !load_import_func_info(&p, p_end, module, is_load_from_file_buf,
-                                  error_buf, error_buf_size))
+                                  no_resolve, error_buf, error_buf_size))
         return false;
 
     /* load function count and start function index */
@@ -2532,6 +2551,91 @@ fail:
     return false;
 }
 
+#if !defined(BH_PLATFORM_NUTTX) && !defined(BH_PLATFORM_ESP_IDF)
+static bool
+try_merge_data_and_text(const uint8 **buf, const uint8 **buf_end,
+                        AOTModule *module, char *error_buf,
+                        uint32 error_buf_size)
+{
+    uint8 *old_buf = (uint8 *)*buf;
+    uint8 *old_end = (uint8 *)*buf_end;
+    size_t code_size = (size_t)(old_end - old_buf);
+    uint32 page_size = os_getpagesize();
+    uint64 total_size = 0;
+    uint32 i;
+    uint8 *sections;
+
+    if (code_size == 0) {
+        return true;
+    }
+
+    /* calculate the total memory needed */
+    total_size += align_uint64((uint64)code_size, page_size);
+    for (i = 0; i < module->data_section_count; ++i) {
+        total_size +=
+            align_uint64((uint64)module->data_sections[i].size, page_size);
+    }
+    /* distance between .data and .text should not be greater than 4GB
+       for some targets (e.g. arm64 reloc need < 4G distance) */
+    if (total_size > UINT32_MAX) {
+        return false;
+    }
+    /* code_size was checked and must be larger than 0 here */
+    bh_assert(total_size > 0);
+
+    sections = loader_mmap((uint32)total_size, false, NULL, 0);
+    if (!sections) {
+        /* merge failed but may be not critical for some targets */
+        return false;
+    }
+
+#ifdef BH_PLATFORM_WINDOWS
+    if (!os_mem_commit(sections, code_size,
+                       MMAP_PROT_READ | MMAP_PROT_WRITE | MMAP_PROT_EXEC)) {
+        os_munmap(sections, (uint32)total_size);
+        return false;
+    }
+#endif
+
+    /* change the code part to be executable */
+    if (os_mprotect(sections, code_size,
+                    MMAP_PROT_READ | MMAP_PROT_WRITE | MMAP_PROT_EXEC)
+        != 0) {
+        os_munmap(sections, (uint32)total_size);
+        return false;
+    }
+
+    module->merged_data_text_sections = sections;
+    module->merged_data_text_sections_size = (uint32)total_size;
+
+    /* order not essential just as compiler does: .text section first */
+    *buf = sections;
+    *buf_end = sections + code_size;
+    bh_memcpy_s(sections, (uint32)code_size, old_buf, (uint32)code_size);
+    os_munmap(old_buf, code_size);
+    sections += align_uint((uint32)code_size, page_size);
+
+    /* then migrate .data sections */
+    for (i = 0; i < module->data_section_count; ++i) {
+        AOTObjectDataSection *data_section = module->data_sections + i;
+        uint8 *old_data = data_section->data;
+        data_section->data = sections;
+        bh_memcpy_s(data_section->data, data_section->size, old_data,
+                    data_section->size);
+        sections += align_uint(data_section->size, page_size);
+    }
+    /* free the original data sections */
+    if (module->merged_data_sections) {
+        os_munmap(module->merged_data_sections,
+                  module->merged_data_sections_size);
+        module->merged_data_sections = NULL;
+        module->merged_data_sections_size = 0;
+    }
+
+    return true;
+}
+#endif /* ! defined(BH_PLATFORM_NUTTX) && !defined(BH_PLATFORM_ESP_IDF) */
+
 static bool
 load_text_section(const uint8 *buf, const uint8 *buf_end, AOTModule *module,
                   char *error_buf, uint32 error_buf_size)
@@ -2725,6 +2829,48 @@ destroy_exports(AOTExport *exports)
     wasm_runtime_free(exports);
 }
 
+static int
+cmp_export_name(const void *a, const void *b)
+{
+    return strcmp(*(char **)a, *(char **)b);
+}
+
+static bool
+check_duplicate_exports(AOTModule *module, char *error_buf,
+                        uint32 error_buf_size)
+{
+    uint32 i;
+    bool result = false;
+    char *names_buf[32], **names = names_buf;
+    if (module->export_count > 32) {
+        names = loader_malloc(module->export_count * sizeof(char *), error_buf,
+                              error_buf_size);
+        if (!names) {
+            return result;
+        }
+    }
+
+    for (i = 0; i < module->export_count; i++) {
+        names[i] = module->exports[i].name;
+    }
+
+    qsort(names, module->export_count, sizeof(char *), cmp_export_name);
+
+    for (i = 1; i < module->export_count; i++) {
+        if (!strcmp(names[i], names[i - 1])) {
+            set_error_buf(error_buf, error_buf_size, "duplicate export name");
+            goto cleanup;
+        }
+    }
+
+    result = true;
+cleanup:
+    if (module->export_count > 32) {
+        wasm_runtime_free(names);
+    }
+    return result;
+}
+
 static bool
 load_exports(const uint8 **p_buf, const uint8 *buf_end, AOTModule *module,
              bool is_load_from_file_buf, char *error_buf, uint32 error_buf_size)
@@ -2746,14 +2892,58 @@ load_exports(const uint8 **p_buf, const uint8 *buf_end, AOTModule *module,
         read_uint32(buf, buf_end, exports[i].index);
         read_uint8(buf, buf_end, exports[i].kind);
         read_string(buf, buf_end, exports[i].name);
-#if 0 /* TODO: check kind and index */
-        if (export_funcs[i].index >=
-              module->func_count + module->import_func_count) {
-            set_error_buf(error_buf, error_buf_size,
-                          "function index is out of range");
+
+        /* Check export kind and index */
+        switch (exports[i].kind) {
+            case EXPORT_KIND_FUNC:
+                if (exports[i].index
+                    >= module->import_func_count + module->func_count) {
+                    set_error_buf(error_buf, error_buf_size,
+                                  "unknown function");
+                    return false;
+                }
+                break;
+            case EXPORT_KIND_TABLE:
+                if (exports[i].index
+                    >= module->import_table_count + module->table_count) {
+                    set_error_buf(error_buf, error_buf_size, "unknown table");
+                    return false;
+                }
+                break;
+            case EXPORT_KIND_MEMORY:
+                if (exports[i].index
+                    >= module->import_memory_count + module->memory_count) {
+                    set_error_buf(error_buf, error_buf_size, "unknown memory");
+                    return false;
+                }
+                break;
+            case EXPORT_KIND_GLOBAL:
+                if (exports[i].index
+                    >= module->import_global_count + module->global_count) {
+                    set_error_buf(error_buf, error_buf_size, "unknown global");
+                    return false;
+                }
+                break;
+#if WASM_ENABLE_TAGS != 0
+                /* TODO
+                case EXPORT_KIND_TAG:
+                    if (index >= module->import_tag_count + module->tag_count) {
+                        set_error_buf(error_buf, error_buf_size, "unknown tag");
+                        return false;
+                    }
+                    break;
+                */
+#endif
+            default:
+                set_error_buf(error_buf, error_buf_size, "invalid export kind");
+                return false;
+        }
+    }
+
+    if (module->export_count > 0) {
+        if (!check_duplicate_exports(module, error_buf, error_buf_size)) {
             return false;
         }
-#endif
     }
 
     *p_buf = buf;
@@ -2999,10 +3189,12 @@ do_text_relocation(AOTModule *module, AOTRelocationGroup *group,
             symbol_addr = module->code;
         }
         else if (!strcmp(symbol, ".data") || !strcmp(symbol, ".sdata")
-                 || !strcmp(symbol, ".rdata")
-                 || !strcmp(symbol, ".rodata")
+                 || !strcmp(symbol, ".rdata") || !strcmp(symbol, ".rodata")
+                 || !strcmp(symbol, ".srodata")
                  /* ".rodata.cst4/8/16/.." */
                  || !strncmp(symbol, ".rodata.cst", strlen(".rodata.cst"))
+                 /* ".srodata.cst4/8/16/.." */
+                 || !strncmp(symbol, ".srodata.cst", strlen(".srodata.cst"))
                  /* ".rodata.strn.m" */
                  || !strncmp(symbol, ".rodata.str", strlen(".rodata.str"))
                  || !strcmp(symbol, AOT_STACK_SIZES_SECTION_NAME)
@@ -3391,16 +3583,9 @@ load_relocation_section(const uint8 *buf, const uint8 *buf_end,
            + sizeof(uint64) * module->real_plt_count
            + sizeof(uint32) * module->float_plt_count;
     if (size > 0) {
-        map_prot = MMAP_PROT_READ | MMAP_PROT_WRITE | MMAP_PROT_EXEC;
-        /* aot code and data in x86_64 must be in range 0 to 2G due to
-           relocation for R_X86_64_32/32S/PC32 */
-        map_flags = MMAP_MAP_32BIT;
-
         if (size > UINT32_MAX
-            || !(module->extra_plt_data =
-                     os_mmap(NULL, (uint32)size, map_prot, map_flags,
-                             os_get_invalid_handle()))) {
-            set_error_buf(error_buf, error_buf_size, "mmap memory failed");
+            || !(module->extra_plt_data = loader_mmap(
+                     (uint32)size, true, error_buf, error_buf_size))) {
             goto fail;
         }
         module->extra_plt_data_size = (uint32)size;
@@ -3512,19 +3697,12 @@ load_relocation_section(const uint8 *buf, const uint8 *buf_end,
         GOTItem *got_item = module->got_item_list;
         uint32 got_item_idx = 0;
 
-        map_prot = MMAP_PROT_READ | MMAP_PROT_WRITE;
-        /* aot code and data in x86_64 must be in range 0 to 2G due to
-           relocation for R_X86_64_32/32S/PC32 */
-        map_flags = MMAP_MAP_32BIT;
-
         /* Create the GOT for func_ptrs, note that it is different from
            the .got section of a dynamic object file */
         size = (uint64)sizeof(void *) * got_item_count;
         if (size > UINT32_MAX
-            || !(module->got_func_ptrs =
-                     os_mmap(NULL, (uint32)size, map_prot, map_flags,
-                             os_get_invalid_handle()))) {
-            set_error_buf(error_buf, error_buf_size, "mmap memory failed");
+            || !(module->got_func_ptrs = loader_mmap(
+                     (uint32)size, false, error_buf, error_buf_size))) {
             goto fail;
         }
 
@@ -3711,7 +3889,7 @@ has_module_memory64(AOTModule *module)
 
 static bool
 load_from_sections(AOTModule *module, AOTSection *sections,
-                   bool is_load_from_file_buf, char *error_buf,
+                   bool is_load_from_file_buf, bool no_resolve, char *error_buf,
                    uint32 error_buf_size)
 {
     AOTSection *section = sections;
@@ -3744,11 +3922,22 @@ load_from_sections(AOTModule *module, AOTSection *sections,
                 break;
             case AOT_SECTION_TYPE_INIT_DATA:
                 if (!load_init_data_section(buf, buf_end, module,
-                                            is_load_from_file_buf, error_buf,
-                                            error_buf_size))
+                                            is_load_from_file_buf, no_resolve,
+                                            error_buf, error_buf_size))
                     return false;
                 break;
             case AOT_SECTION_TYPE_TEXT:
+#if !defined(BH_PLATFORM_NUTTX) && !defined(BH_PLATFORM_ESP_IDF)
+                /* try to merge .data and .text, with exceptions:
+                 * 1. XIP mode
+                 * 2. pre-mmapped module load from aot_load_from_sections()
+                 * 3. nuttx & esp-idf: have separate region for MMAP_PROT_EXEC
+                 */
+                if (!module->is_indirect_mode && is_load_from_file_buf)
+                    if (!try_merge_data_and_text(&buf, &buf_end, module,
+                                                 error_buf, error_buf_size))
+                        LOG_WARNING("merge .data and .text sections failed");
+#endif /* ! defined(BH_PLATFORM_NUTTX) && !defined(BH_PLATFORM_ESP_IDF) */
                 if (!load_text_section(buf, buf_end, module, error_buf,
                                        error_buf_size))
                     return false;
@@ -3957,7 +4146,7 @@ aot_load_from_sections(AOTSection *section_list, char *error_buf,
     if (!module)
         return NULL;
 
-    if (!load_from_sections(module, section_list, false, error_buf,
+    if (!load_from_sections(module, section_list, false, false, error_buf,
                             error_buf_size)) {
         aot_unload(module);
         return NULL;
@@ -4065,37 +4254,16 @@ create_sections(AOTModule *module, const uint8 *buf, uint32 size,
 
             if (section_type == AOT_SECTION_TYPE_TEXT) {
                 if ((section_size > 0) && !module->is_indirect_mode) {
-                    int map_prot =
-                        MMAP_PROT_READ | MMAP_PROT_WRITE | MMAP_PROT_EXEC;
-#if defined(BUILD_TARGET_X86_64) || defined(BUILD_TARGET_AMD_64) \
-    || defined(BUILD_TARGET_RISCV64_LP64D)                       \
-    || defined(BUILD_TARGET_RISCV64_LP64)
-                    /* aot code and data in x86_64 must be in range 0 to 2G due
-                       to relocation for R_X86_64_32/32S/PC32 */
-                    int map_flags = MMAP_MAP_32BIT;
-#else
-                    int map_flags = MMAP_MAP_NONE;
-#endif
                     total_size =
                         (uint64)section_size + aot_get_plt_table_size();
                     total_size = (total_size + 3) & ~((uint64)3);
                     if (total_size >= UINT32_MAX
                         || !(aot_text =
-                                 os_mmap(NULL, (uint32)total_size, map_prot,
-                                         map_flags, os_get_invalid_handle()))) {
+                                 loader_mmap((uint32)total_size, true,
+                                             error_buf, error_buf_size))) {
                         wasm_runtime_free(section);
-                        set_error_buf(error_buf, error_buf_size,
-                                      "mmap memory failed");
                         goto fail;
                     }
-#if defined(BUILD_TARGET_X86_64) || defined(BUILD_TARGET_AMD_64)
-#if !defined(BH_PLATFORM_LINUX_SGX) && !defined(BH_PLATFORM_WINDOWS) \
-    && !defined(BH_PLATFORM_DARWIN)
-                    /* address must be in the first 2 Gigabytes of
-                       the process address space */
-                    bh_assert((uintptr_t)aot_text < INT32_MAX);
-#endif
-#endif
 
 #if (WASM_MEM_DUAL_BUS_MIRROR != 0)
                     mirrored_text = os_get_dbus_mirror(aot_text);
@@ -4146,9 +4314,20 @@ fail:
     return false;
 }
 
+static bool
+aot_compatible_version(uint32 version)
+{
+    /*
+     * refer to "AoT-compiled module compatibility among WAMR versions" in
+     * ./doc/biuld_wasm_app.md
+     */
+    return version == 4 || version == 3;
+}
+
 static bool
 load(const uint8 *buf, uint32 size, AOTModule *module,
-     bool wasm_binary_freeable, char *error_buf, uint32 error_buf_size)
+     bool wasm_binary_freeable, bool no_resolve, char *error_buf,
+     uint32 error_buf_size)
 {
     const uint8 *buf_end = buf + size;
     const uint8 *p = buf, *p_end = buf_end;
@@ -4163,7 +4342,7 @@ load(const uint8 *buf, uint32 size, AOTModule *module,
     }
 
     read_uint32(p, p_end, version);
-    if (version != AOT_CURRENT_VERSION) {
+    if (!aot_compatible_version(version)) {
         set_error_buf(error_buf, error_buf_size, "unknown binary version");
         return false;
     }
@@ -4175,11 +4354,15 @@ load(const uint8 *buf, uint32 size, AOTModule *module,
         return false;
 
     ret = load_from_sections(module, section_list, !wasm_binary_freeable,
-                             error_buf, error_buf_size);
+                             no_resolve, error_buf, error_buf_size);
     if (!ret) {
         /* If load_from_sections() fails, then aot text is destroyed
            in destroy_sections() */
-        destroy_sections(section_list, module->is_indirect_mode ? false : true);
+        destroy_sections(section_list,
+                         module->is_indirect_mode
+                                 || module->merged_data_text_sections
+                             ? false
+                             : true);
         /* aot_unload() won't destroy aot text again */
         module->code = NULL;
     }
@@ -4219,14 +4402,21 @@ aot_load_from_aot_file(const uint8 *buf, uint32 size, const LoadArgs *args,
         return NULL;
 
     os_thread_jit_write_protect_np(false); /* Make memory writable */
-    if (!load(buf, size, module, args->wasm_binary_freeable, error_buf,
-              error_buf_size)) {
+    if (!load(buf, size, module, args->wasm_binary_freeable, args->no_resolve,
+              error_buf, error_buf_size)) {
         aot_unload(module);
         return NULL;
     }
     os_thread_jit_write_protect_np(true); /* Make memory executable */
     os_icache_flush(module->code, module->code_size);
 
+#if WASM_ENABLE_AOT_VALIDATOR != 0
+    if (!aot_module_validate(module, error_buf, error_buf_size)) {
+        aot_unload(module);
+        return NULL;
+    }
+#endif /* WASM_ENABLE_AOT_VALIDATOR != 0 */
+
     LOG_VERBOSE("Load module success.\n");
     return module;
 }
@@ -4241,7 +4431,7 @@ aot_unload(AOTModule *module)
         wasm_runtime_free(module->memories);
 
     if (module->mem_init_data_list)
-        destroy_mem_init_data_list(module->mem_init_data_list,
+        destroy_mem_init_data_list(module, module->mem_init_data_list,
                                    module->mem_init_data_count);
 
     if (module->native_symbol_list)
@@ -4329,7 +4519,8 @@ aot_unload(AOTModule *module)
     }
 #endif
 
-    if (module->code && !module->is_indirect_mode) {
+    if (module->code && !module->is_indirect_mode
+        && !module->merged_data_text_sections) {
         /* The layout is: literal size + literal + code (with plt table) */
         uint8 *mmap_addr = module->literal - sizeof(uint32);
         uint32 total_size =
@@ -4364,6 +4555,14 @@ aot_unload(AOTModule *module)
         destroy_object_data_sections(module->data_sections,
                                      module->data_section_count);
 
+    if (module->merged_data_sections)
+        os_munmap(module->merged_data_sections,
+                  module->merged_data_sections_size);
+
+    if (module->merged_data_text_sections)
+        os_munmap(module->merged_data_text_sections,
+                  module->merged_data_text_sections_size);
+
 #if WASM_ENABLE_DEBUG_AOT != 0
     jit_code_entry_destroy(module->elf_hdr);
 #endif

core/iwasm/aot/aot_perf_map.c

@@ -7,7 +7,6 @@
 #include "bh_log.h"
 #include "bh_platform.h"
 
-#if WASM_ENABLE_LINUX_PERF != 0
 struct func_info {
     uint32 idx;
     void *ptr;
@@ -32,8 +31,15 @@ get_func_size(const AOTModule *module, struct func_info *sorted_func_ptrs,
 static int
 compare_func_ptrs(const void *f1, const void *f2)
 {
-    return (intptr_t)((struct func_info *)f1)->ptr
-           - (intptr_t)((struct func_info *)f2)->ptr;
+    uintptr_t ptr1 = (uintptr_t)((struct func_info *)f1)->ptr;
+    uintptr_t ptr2 = (uintptr_t)((struct func_info *)f2)->ptr;
+
+    if (ptr1 < ptr2)
+        return -1;
+    else if (ptr1 > ptr2)
+        return 1;
+    else
+        return 0;
 }
 
 static struct func_info *
@@ -46,8 +52,8 @@ sort_func_ptrs(const AOTModule *module, char *error_buf, uint32 error_buf_size)
     content_len = (uint64)sizeof(struct func_info) * module->func_count;
     sorted_func_ptrs = wasm_runtime_malloc(content_len);
     if (!sorted_func_ptrs) {
-        snprintf(error_buf, error_buf_size,
-                 "allocate memory failed when creating perf map");
+        (void)snprintf(error_buf, error_buf_size,
+                       "allocate memory failed when creating perf map");
         return NULL;
     }
 
@@ -78,7 +84,8 @@ aot_create_perf_map(const AOTModule *module, char *error_buf,
     if (!sorted_func_ptrs)
         goto quit;
 
-    snprintf(perf_map_path, sizeof(perf_map_path) - 1, "/tmp/perf-%d.map", pid);
+    (void)snprintf(perf_map_path, sizeof(perf_map_path) - 1, "/tmp/perf-%d.map",
+                   pid);
     perf_map = fopen(perf_map_path, "a");
     if (!perf_map) {
         LOG_WARNING("warning: can't create /tmp/perf-%d.map, because %s", pid,
@@ -89,19 +96,23 @@ aot_create_perf_map(const AOTModule *module, char *error_buf,
     const char *module_name = aot_get_module_name((AOTModule *)module);
     for (i = 0; i < module->func_count; i++) {
         memset(perf_map_info, 0, 128);
-        if (strlen(module_name) > 0)
-            snprintf(perf_map_info, 128, "%lx  %x  [%s]#aot_func#%u\n",
-                     (uintptr_t)sorted_func_ptrs[i].ptr,
-                     get_func_size(module, sorted_func_ptrs, i), module_name,
-                     sorted_func_ptrs[i].idx);
-        else
-            snprintf(perf_map_info, 128, "%lx  %x  aot_func#%u\n",
-                     (uintptr_t)sorted_func_ptrs[i].ptr,
-                     get_func_size(module, sorted_func_ptrs, i),
-                     sorted_func_ptrs[i].idx);
+        if (strlen(module_name) > 0) {
+            (void)snprintf(perf_map_info, 128,
+                           "%" PRIxPTR "  %x  [%s]#aot_func#%u\n",
+                           (uintptr_t)sorted_func_ptrs[i].ptr,
+                           get_func_size(module, sorted_func_ptrs, i),
+                           module_name, sorted_func_ptrs[i].idx);
+        }
+        else {
+            (void)snprintf(perf_map_info, 128,
+                           "%" PRIxPTR "  %x  aot_func#%u\n",
+                           (uintptr_t)sorted_func_ptrs[i].ptr,
+                           get_func_size(module, sorted_func_ptrs, i),
+                           sorted_func_ptrs[i].idx);
+        }
 
         /* fwrite() is thread safe */
-        fwrite(perf_map_info, 1, strlen(perf_map_info), perf_map);
+        (void)fwrite(perf_map_info, 1, strlen(perf_map_info), perf_map);
     }
 
     LOG_VERBOSE("write map information from %s into /tmp/perf-%d.map",
@@ -113,8 +124,7 @@ quit:
         wasm_runtime_free(sorted_func_ptrs);
 
     if (perf_map)
-        fclose(perf_map);
+        (void)fclose(perf_map);
 
     return ret;
 }
-#endif /* WASM_ENABLE_LINUX_PERF != 0 */

core/iwasm/aot/aot_reloc.h

@@ -122,6 +122,10 @@ typedef struct {
     REG_SYM(aot_intrinsic_i64_rem_u),     \
     REG_SYM(aot_intrinsic_i64_bit_or),    \
     REG_SYM(aot_intrinsic_i64_bit_and),   \
+    REG_SYM(aot_intrinsic_i64_mul),       \
+    REG_SYM(aot_intrinsic_i64_shl),       \
+    REG_SYM(aot_intrinsic_i64_shr_s),     \
+    REG_SYM(aot_intrinsic_i64_shr_u),     \
     REG_SYM(aot_intrinsic_i32_div_s),     \
     REG_SYM(aot_intrinsic_i32_div_u),     \
     REG_SYM(aot_intrinsic_i32_rem_s),     \
@@ -226,7 +230,7 @@ SymbolMap *
 get_target_symbol_map(uint32 *sym_num);
 
 uint32
-get_plt_table_size();
+get_plt_table_size(void);
 
 void
 init_plt_table(uint8 *plt);

core/iwasm/aot/aot_runtime.h

@@ -25,12 +25,16 @@ extern "C" {
 #define WASM_FEATURE_REF_TYPES (1 << 3)
 #define WASM_FEATURE_GARBAGE_COLLECTION (1 << 4)
 #define WASM_FEATURE_EXCEPTION_HANDLING (1 << 5)
-#define WASM_FEATURE_MEMORY64 (1 << 6)
+#define WASM_FEATURE_TINY_STACK_FRAME (1 << 6)
 #define WASM_FEATURE_MULTI_MEMORY (1 << 7)
 #define WASM_FEATURE_DYNAMIC_LINKING (1 << 8)
 #define WASM_FEATURE_COMPONENT_MODEL (1 << 9)
 #define WASM_FEATURE_RELAXED_SIMD (1 << 10)
 #define WASM_FEATURE_FLEXIBLE_VECTORS (1 << 11)
+/* Stack frame is created at the beginning of the function,
+ * and not at the beginning of each function call */
+#define WASM_FEATURE_FRAME_PER_FUNCTION (1 << 12)
+#define WASM_FEATURE_FRAME_NO_FUNC_IDX (1 << 13)
 
 typedef enum AOTSectionType {
     AOT_SECTION_TYPE_TARGET_INFO = 0,
@@ -39,6 +43,10 @@ typedef enum AOTSectionType {
     AOT_SECTION_TYPE_FUNCTION = 3,
     AOT_SECTION_TYPE_EXPORT = 4,
     AOT_SECTION_TYPE_RELOCATION = 5,
+    /*
+     * Note: We haven't had anything to use AOT_SECTION_TYPE_SIGNATURE.
+     * It's just reserved for possible module signing features.
+     */
     AOT_SECTION_TYPE_SIGNATURE = 6,
     AOT_SECTION_TYPE_CUSTOM = 100,
 } AOTSectionType;
@@ -101,9 +109,31 @@ typedef struct AOTFunctionInstance {
     } u;
 } AOTFunctionInstance;
 
+/* Map of a function index to the element ith in
+   the export functions array */
+typedef struct ExportFuncMap {
+    uint32 func_idx;
+    uint32 export_idx;
+} ExportFuncMap;
+
 typedef struct AOTModuleInstanceExtra {
     DefPointer(const uint32 *, stack_sizes);
+    /*
+     * Adjusted shared heap based addr to simple the calculation
+     * in the aot code. The value is:
+     *   shared_heap->base_addr - shared_heap->start_off
+     */
+    DefPointer(uint8 *, shared_heap_base_addr_adj);
+    MemBound shared_heap_start_off;
+
     WASMModuleInstanceExtraCommon common;
+
+    /**
+     * maps of func indexes to export func indexes, which
+     * is sorted by func index for a quick lookup and is
+     * created only when first time used.
+     */
+    ExportFuncMap *export_func_maps;
     AOTFunctionInstance **functions;
     uint32 function_count;
 #if WASM_ENABLE_MULTI_MODULE != 0
@@ -111,6 +141,10 @@ typedef struct AOTModuleInstanceExtra {
     bh_list *sub_module_inst_list;
     WASMModuleInstanceCommon **import_func_module_insts;
 #endif
+
+#if WASM_ENABLE_SHARED_HEAP != 0
+    WASMSharedHeap *shared_heap;
+#endif
 } AOTModuleInstanceExtra;
 
 #if defined(BUILD_TARGET_X86_64) || defined(BUILD_TARGET_AMD_64)
@@ -315,6 +349,17 @@ typedef struct AOTModule {
 
     /* Whether the underlying wasm binary buffer can be freed */
     bool is_binary_freeable;
+
+    /* `.data` sections merged into one mmaped to reduce the tlb cache miss */
+    uint8 *merged_data_sections;
+    uint32 merged_data_sections_size;
+    /* `.data` and `.text` sections merged into one large mmaped section */
+    uint8 *merged_data_text_sections;
+    uint32 merged_data_text_sections_size;
+
+#if WASM_ENABLE_AOT_STACK_FRAME != 0
+    uint32 feature_flags;
+#endif
 } AOTModule;
 
 #define AOTMemoryInstance WASMMemoryInstance
@@ -473,6 +518,18 @@ aot_load_from_sections(AOTSection *section_list, char *error_buf,
 void
 aot_unload(AOTModule *module);
 
+/**
+ * Resolve symbols for an AOT module
+ */
+bool
+aot_resolve_symbols(AOTModule *module);
+
+/**
+ * Helper function to resolve a single function
+ */
+bool
+aot_resolve_import_func(AOTModule *module, AOTImportFunc *import_func);
+
 /**
  * Instantiate a AOT module.
  *
@@ -513,6 +570,22 @@ aot_deinstantiate(AOTModuleInstance *module_inst, bool is_sub_inst);
 AOTFunctionInstance *
 aot_lookup_function(const AOTModuleInstance *module_inst, const char *name);
 
+/**
+ * Lookup an exported function in the AOT module instance with
+ * the function index.
+ */
+AOTFunctionInstance *
+aot_lookup_function_with_idx(AOTModuleInstance *module_inst, uint32 func_idx);
+
+AOTMemoryInstance *
+aot_lookup_memory(AOTModuleInstance *module_inst, char const *name);
+
+AOTMemoryInstance *
+aot_get_default_memory(AOTModuleInstance *module_inst);
+
+AOTMemoryInstance *
+aot_get_memory_with_idx(AOTModuleInstance *module_inst, uint32 mem_idx);
+
 /**
  * Get a function in the AOT module instance.
  *
@@ -522,7 +595,7 @@ aot_lookup_function(const AOTModuleInstance *module_inst, const char *name);
  * @return the function instance found
  */
 AOTFunctionInstance *
-aot_get_function_instance(AOTModuleInstance *module_inst, uint32_t func_idx);
+aot_get_function_instance(AOTModuleInstance *module_inst, uint32 func_idx);
 
 /**
  * Call the given AOT function of a AOT module instance with
@@ -605,6 +678,10 @@ aot_module_dup_data(AOTModuleInstance *module_inst, const char *src,
 bool
 aot_enlarge_memory(AOTModuleInstance *module_inst, uint32 inc_page_count);
 
+bool
+aot_enlarge_memory_with_idx(AOTModuleInstance *module_inst,
+                            uint32 inc_page_count, uint32 memidx);
+
 /**
  * Invoke native function from aot code
  */
@@ -626,7 +703,7 @@ aot_check_app_addr_and_convert(AOTModuleInstance *module_inst, bool is_str,
                                void **p_native_addr);
 
 uint32
-aot_get_plt_table_size();
+aot_get_plt_table_size(void);
 
 void *
 aot_memmove(void *dest, const void *src, size_t n);
@@ -700,6 +777,13 @@ aot_frame_update_profile_info(WASMExecEnv *exec_env, bool alloc_frame);
 bool
 aot_create_call_stack(struct WASMExecEnv *exec_env);
 
+#if WAMR_ENABLE_COPY_CALLSTACK != 0
+uint32
+aot_copy_callstack(WASMExecEnv *exec_env, wasm_frame_t *buffer,
+                   const uint32 length, const uint32 skip_n, char *error_buf,
+                   uint32_t error_buf_size);
+#endif // WAMR_ENABLE_COPY_CALLSTACK
+
 /**
  * @brief Dump wasm call stack or get the size
  *

core/iwasm/aot/aot_validator.c

@@ -0,0 +1,45 @@
+/*
+ * Copyright (C) 2025 Intel Corporation. All rights reserved.
+ * SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+ */
+
+#include "aot_validator.h"
+
+static void
+set_error_buf(char *error_buf, uint32 error_buf_size, const char *string)
+{
+    if (error_buf != NULL) {
+        snprintf(error_buf, error_buf_size,
+                 "AOT module load failed: from validator. %s", string);
+    }
+}
+
+static bool
+aot_memory_info_validate(const AOTModule *module, char *error_buf,
+                         uint32 error_buf_size)
+{
+    if (module->import_memory_count > 0) {
+        set_error_buf(error_buf, error_buf_size,
+                      "import memory is not supported");
+        return false;
+    }
+
+    if (module->memory_count < 1) {
+        set_error_buf(error_buf, error_buf_size,
+                      "there should be >=1 memory in one aot module");
+        return false;
+    }
+
+    return true;
+}
+
+bool
+aot_module_validate(const AOTModule *module, char *error_buf,
+                    uint32 error_buf_size)
+{
+    if (!aot_memory_info_validate(module, error_buf, error_buf_size)) {
+        return false;
+    }
+
+    return true;
+}

core/iwasm/aot/aot_validator.h

@@ -0,0 +1,15 @@
+/*
+ * Copyright (C) 2025 Intel Corporation. All rights reserved.
+ * SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+ */
+
+#ifndef _AOT_VALIDATOR_H_
+#define _AOT_VALIDATOR_H_
+
+#include "aot_runtime.h"
+
+bool
+aot_module_validate(const AOTModule *module, char *error_buf,
+                    uint32 error_buf_size);
+
+#endif /* _AOT_VALIDATOR_H_ */

core/iwasm/aot/arch/aot_reloc_aarch64.c

@@ -53,12 +53,6 @@ get_target_symbol_map(uint32 *sym_num)
     return target_sym_map;
 }
 
-#if (defined(__APPLE__) || defined(__MACH__)) && defined(__arm64__)
-#define BUILD_TARGET_AARCH64_DEFAULT "arm64"
-#else
-#define BUILD_TARGET_AARCH64_DEFAULT "aarch64v8"
-#endif
-
 void
 get_current_target(char *target_buf, uint32 target_buf_size)
 {
@@ -68,8 +62,8 @@ get_current_target(char *target_buf, uint32 target_buf_size)
 
     /* Set to "aarch64v8" by default if sub version isn't specified */
     if (strcmp(s, "AARCH64") == 0) {
-        s = BUILD_TARGET_AARCH64_DEFAULT;
-        s_size = sizeof(BUILD_TARGET_AARCH64_DEFAULT);
+        s = "aarch64v8";
+        s_size = 10; /* sizeof("aarch64v8"); */
     }
     if (target_buf_size < s_size) {
         s_size = target_buf_size;
@@ -83,10 +77,9 @@ get_current_target(char *target_buf, uint32 target_buf_size)
     /* Ensure the string is null byte ('\0') terminated */
     *d = '\0';
 }
-#undef BUILD_TARGET_AARCH64_DEFAULT
 
 static uint32
-get_plt_item_size()
+get_plt_item_size(void)
 {
     /* 6*4 bytes instructions and 8 bytes symbol address */
     return 32;

core/iwasm/aot/arch/aot_reloc_arm.c

@@ -12,102 +12,102 @@
 #define R_ARM_MOVT_ABS 44
 
 /* clang-format off */
-void __adddf3();
-void __addsf3();
-void __aeabi_d2f();
-void __aeabi_d2iz();
-void __aeabi_d2lz();
-void __aeabi_d2uiz();
-void __aeabi_d2ulz();
-void __aeabi_dadd();
-void __aeabi_dcmpeq();
-void __aeabi_dcmpge();
-void __aeabi_dcmpgt();
-void __aeabi_dcmple();
-void __aeabi_dcmplt();
-void __aeabi_dcmpun();
-void __aeabi_ddiv();
-void __aeabi_dmul();
-void __aeabi_dsub();
-void __aeabi_f2d();
-void __aeabi_f2iz();
-void __aeabi_f2lz();
-void __aeabi_f2ulz();
-void __aeabi_fadd();
-void __aeabi_fcmpeq();
-void __aeabi_fcmpge();
-void __aeabi_fcmpgt();
-void __aeabi_fcmple();
-void __aeabi_fcmplt();
-void __aeabi_fcmpun();
-void __aeabi_fdiv();
-void __aeabi_fmul();
-void __aeabi_fsub();
-void __aeabi_i2d();
-void __aeabi_i2f();
-void __aeabi_idiv();
-void __aeabi_idivmod();
-void __aeabi_l2d();
-void __aeabi_l2f();
-void __aeabi_ldivmod();
-void __aeabi_memclr();
-void __aeabi_memcpy();
-void __aeabi_memmove();
-void __aeabi_memset();
-void __aeabi_ui2d();
-void __aeabi_ui2f();
-void __aeabi_uidiv();
-void __aeabi_uidivmod();
-void __aeabi_ul2d();
-void __aeabi_ul2f();
-void __aeabi_uldivmod();
-void __clzsi2();
-void __divdf3();
-void __divdi3();
-void __divsf3();
-void __divsi3();
-void __eqdf2();
-void __eqsf2();
-void __extendsfdf2();
-void __fixdfdi();
-void __fixdfsi();
-void __fixsfdi();
-void __fixsfsi();
-void __fixunsdfdi();
-void __fixunsdfsi();
-void __fixunssfdi();
-void __floatdidf();
-void __floatdisf();
-void __floatsidf();
-void __floatsisf();
-void __floatundidf();
-void __floatundisf();
-void __floatunsidf();
-void __floatunsisf();
-void __gedf2();
-void __gesf2();
-void __gtdf2();
-void __gtsf2();
-void __ledf2();
-void __lesf2();
-void __ltdf2();
-void __ltsf2();
-void __moddi3();
-void __modsi3();
-void __muldf3();
-void __mulsf3();
-void __nedf2();
-void __nesf2();
-void __subdf3();
-void __subsf3();
-void __truncdfsf2();
-void __udivdi3();
-void __udivmoddi4();
-void __udivsi3();
-void __umoddi3();
-void __umodsi3();
-void __unorddf2();
-void __unordsf2();
+void __adddf3(void);
+void __addsf3(void);
+void __aeabi_d2f(void);
+void __aeabi_d2iz(void);
+void __aeabi_d2lz(void);
+void __aeabi_d2uiz(void);
+void __aeabi_d2ulz(void);
+void __aeabi_dadd(void);
+void __aeabi_dcmpeq(void);
+void __aeabi_dcmpge(void);
+void __aeabi_dcmpgt(void);
+void __aeabi_dcmple(void);
+void __aeabi_dcmplt(void);
+void __aeabi_dcmpun(void);
+void __aeabi_ddiv(void);
+void __aeabi_dmul(void);
+void __aeabi_dsub(void);
+void __aeabi_f2d(void);
+void __aeabi_f2iz(void);
+void __aeabi_f2lz(void);
+void __aeabi_f2ulz(void);
+void __aeabi_fadd(void);
+void __aeabi_fcmpeq(void);
+void __aeabi_fcmpge(void);
+void __aeabi_fcmpgt(void);
+void __aeabi_fcmple(void);
+void __aeabi_fcmplt(void);
+void __aeabi_fcmpun(void);
+void __aeabi_fdiv(void);
+void __aeabi_fmul(void);
+void __aeabi_fsub(void);
+void __aeabi_i2d(void);
+void __aeabi_i2f(void);
+void __aeabi_idiv(void);
+void __aeabi_idivmod(void);
+void __aeabi_l2d(void);
+void __aeabi_l2f(void);
+void __aeabi_ldivmod(void);
+void __aeabi_memclr(void);
+void __aeabi_memcpy(void);
+void __aeabi_memmove(void);
+void __aeabi_memset(void);
+void __aeabi_ui2d(void);
+void __aeabi_ui2f(void);
+void __aeabi_uidiv(void);
+void __aeabi_uidivmod(void);
+void __aeabi_ul2d(void);
+void __aeabi_ul2f(void);
+void __aeabi_uldivmod(void);
+void __clzsi2(void);
+void __divdf3(void);
+void __divdi3(void);
+void __divsf3(void);
+void __divsi3(void);
+void __eqdf2(void);
+void __eqsf2(void);
+void __extendsfdf2(void);
+void __fixdfdi(void);
+void __fixdfsi(void);
+void __fixsfdi(void);
+void __fixsfsi(void);
+void __fixunsdfdi(void);
+void __fixunsdfsi(void);
+void __fixunssfdi(void);
+void __floatdidf(void);
+void __floatdisf(void);
+void __floatsidf(void);
+void __floatsisf(void);
+void __floatundidf(void);
+void __floatundisf(void);
+void __floatunsidf(void);
+void __floatunsisf(void);
+void __gedf2(void);
+void __gesf2(void);
+void __gtdf2(void);
+void __gtsf2(void);
+void __ledf2(void);
+void __lesf2(void);
+void __ltdf2(void);
+void __ltsf2(void);
+void __moddi3(void);
+void __modsi3(void);
+void __muldf3(void);
+void __mulsf3(void);
+void __nedf2(void);
+void __nesf2(void);
+void __subdf3(void);
+void __subsf3(void);
+void __truncdfsf2(void);
+void __udivdi3(void);
+void __udivmoddi4(void);
+void __udivsi3(void);
+void __umoddi3(void);
+void __umodsi3(void);
+void __unorddf2(void);
+void __unordsf2(void);
 /* clang-format on */
 
 static SymbolMap target_sym_map[] = {
@@ -255,7 +255,7 @@ get_current_target(char *target_buf, uint32 target_buf_size)
 #undef BUILD_TARGET_ARM_DEFAULT
 
 uint32
-get_plt_item_size()
+get_plt_item_size(void)
 {
     /* 8 bytes instructions and 4 bytes symbol address */
     return 12;

core/iwasm/aot/arch/aot_reloc_dummy.c

@@ -0,0 +1,39 @@
+/*
+ * Copyright (C) 2020 Intel Corporation. All rights reserved.
+ * SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+ */
+
+#include "aot_reloc.h"
+
+SymbolMap *
+get_target_symbol_map(uint32 *sym_num)
+{
+    abort();
+}
+
+uint32
+get_plt_table_size(void)
+{
+    abort();
+}
+
+void
+init_plt_table(uint8 *plt)
+{
+    abort();
+}
+
+void
+get_current_target(char *target_buf, uint32 target_buf_size)
+{
+    abort();
+}
+
+bool
+apply_relocation(AOTModule *module, uint8 *target_section_addr,
+                 uint32 target_section_size, uint64 reloc_offset,
+                 int64 reloc_addend, uint32 reloc_type, void *symbol_addr,
+                 int32 symbol_index, char *error_buf, uint32 error_buf_size)
+{
+    abort();
+}

core/iwasm/aot/arch/aot_reloc_mips.c

@@ -28,7 +28,7 @@ get_current_target(char *target_buf, uint32 target_buf_size)
 }
 
 static uint32
-get_plt_item_size()
+get_plt_item_size(void)
 {
     return 0;
 }

core/iwasm/aot/arch/aot_reloc_riscv.c

@@ -24,6 +24,7 @@
 #undef NEED_SOFT_I32_DIV
 #undef NEED_SOFT_I64_MUL
 #undef NEED_SOFT_I64_DIV
+#undef NEED_SOFT_ATOMIC
 
 #ifdef __riscv_flen
 #if __riscv_flen == 32
@@ -48,59 +49,66 @@
 #define NEED_SOFT_I64_DIV
 #endif
 
+#ifndef __riscv_atomic
+#define NEED_SOFT_ATOMIC
+#endif
+
 /* clang-format off */
-void __adddf3();
-void __addsf3();
-void __divdf3();
-void __divdi3();
-void __divsf3();
-void __divsi3();
-void __eqdf2();
-void __eqsf2();
-void __extendsfdf2();
-void __fixdfdi();
-void __fixdfsi();
-void __fixsfdi();
-void __fixsfsi();
-void __fixunsdfdi();
-void __fixunsdfsi();
-void __fixunssfdi();
-void __fixunssfsi();
-void __floatdidf();
-void __floatdisf();
-void __floatsidf();
-void __floatsisf();
-void __floatundidf();
-void __floatundisf();
-void __floatunsidf();
-void __floatunsisf();
-void __gedf2();
-void __gesf2();
-void __gtdf2();
-void __gtsf2();
-void __ledf2();
-void __lesf2();
-void __ltdf2();
-void __ltsf2();
-void __moddi3();
-void __modsi3();
-void __muldf3();
-void __muldi3();
-void __mulsf3();
-void __mulsi3();
-void __nedf2();
-void __negdf2();
-void __negsf2();
-void __nesf2();
-void __subdf3();
-void __subsf3();
-void __truncdfsf2();
-void __udivdi3();
-void __udivsi3();
-void __umoddi3();
-void __umodsi3();
-void __unorddf2();
-void __unordsf2();
+void __adddf3(void);
+void __addsf3(void);
+void __divdf3(void);
+void __divdi3(void);
+void __divsf3(void);
+void __divsi3(void);
+void __eqdf2(void);
+void __eqsf2(void);
+void __extendsfdf2(void);
+void __fixdfdi(void);
+void __fixdfsi(void);
+void __fixsfdi(void);
+void __fixsfsi(void);
+void __fixunsdfdi(void);
+void __fixunsdfsi(void);
+void __fixunssfdi(void);
+void __fixunssfsi(void);
+void __floatdidf(void);
+void __floatdisf(void);
+void __floatsidf(void);
+void __floatsisf(void);
+void __floatundidf(void);
+void __floatundisf(void);
+void __floatunsidf(void);
+void __floatunsisf(void);
+void __gedf2(void);
+void __gesf2(void);
+void __gtdf2(void);
+void __gtsf2(void);
+void __ledf2(void);
+void __lesf2(void);
+void __ltdf2(void);
+void __ltsf2(void);
+void __moddi3(void);
+void __modsi3(void);
+void __muldf3(void);
+void __muldi3(void);
+void __mulsf3(void);
+void __mulsi3(void);
+void __nedf2(void);
+void __negdf2(void);
+void __negsf2(void);
+void __nesf2(void);
+void __subdf3(void);
+void __subsf3(void);
+void __truncdfsf2(void);
+void __udivdi3(void);
+void __udivsi3(void);
+void __umoddi3(void);
+void __umodsi3(void);
+void __unorddf2(void);
+void __unordsf2(void);
+bool __atomic_compare_exchange_4(volatile void *, void *, unsigned int,
+                                 bool, int, int);
+void __atomic_store_4(volatile void *, unsigned int, int);
 /* clang-format on */
 
 static SymbolMap target_sym_map[] = {
@@ -127,6 +135,7 @@ static SymbolMap target_sym_map[] = {
      * to convert float and long long
      */
     REG_SYM(__floatundisf),
+    REG_SYM(__floatdisf),
 #endif
 #ifdef NEED_SOFT_DP
     REG_SYM(__adddf3),
@@ -175,6 +184,10 @@ static SymbolMap target_sym_map[] = {
     REG_SYM(__moddi3),
     REG_SYM(__udivdi3),
     REG_SYM(__umoddi3),
+#endif
+#ifdef NEED_SOFT_ATOMIC
+    REG_SYM(__atomic_compare_exchange_4),
+    REG_SYM(__atomic_store_4),
 #endif
     /* clang-format on */
 };
@@ -193,7 +206,7 @@ get_current_target(char *target_buf, uint32 target_buf_size)
 }
 
 uint32
-get_plt_item_size()
+get_plt_item_size(void)
 {
 #if __riscv_xlen == 64
     /* auipc + ld + jalr + nop + addr */

core/iwasm/aot/arch/aot_reloc_thumb.c

@@ -14,102 +14,102 @@
 #define R_ARM_THM_MOVT_PREL 50
 
 /* clang-format off */
-void __adddf3();
-void __addsf3();
-void __aeabi_d2f();
-void __aeabi_d2iz();
-void __aeabi_d2lz();
-void __aeabi_d2uiz();
-void __aeabi_d2ulz();
-void __aeabi_dadd();
-void __aeabi_dcmpeq();
-void __aeabi_dcmpge();
-void __aeabi_dcmpgt();
-void __aeabi_dcmple();
-void __aeabi_dcmplt();
-void __aeabi_dcmpun();
-void __aeabi_ddiv();
-void __aeabi_dmul();
-void __aeabi_dsub();
-void __aeabi_f2d();
-void __aeabi_f2iz();
-void __aeabi_f2lz();
-void __aeabi_f2ulz();
-void __aeabi_fadd();
-void __aeabi_fcmpeq();
-void __aeabi_fcmpge();
-void __aeabi_fcmpgt();
-void __aeabi_fcmple();
-void __aeabi_fcmplt();
-void __aeabi_fcmpun();
-void __aeabi_fdiv();
-void __aeabi_fmul();
-void __aeabi_fsub();
-void __aeabi_i2d();
-void __aeabi_i2f();
-void __aeabi_idiv();
-void __aeabi_idivmod();
-void __aeabi_l2d();
-void __aeabi_l2f();
-void __aeabi_ldivmod();
-void __aeabi_llsl();
-void __aeabi_llsr();
-void __aeabi_lmul();
-void __aeabi_ui2d();
-void __aeabi_ui2f();
-void __aeabi_uidiv();
-void __aeabi_uidivmod();
-void __aeabi_ul2d();
-void __aeabi_ul2f();
-void __aeabi_uldivmod();
-void __ashldi3();
-void __clzsi2();
-void __divdf3();
-void __divdi3();
-void __divsi3();
-void __eqdf2();
-void __eqsf2();
-void __extendsfdf2();
-void __fixdfdi();
-void __fixdfsi();
-void __fixsfdi();
-void __fixunsdfdi();
-void __fixunsdfsi();
-void __fixunssfdi();
-void __floatdidf();
-void __floatdisf();
-void __floatsidf();
-void __floatsisf();
-void __floatundidf();
-void __floatundisf();
-void __floatunsidf();
-void __floatunsisf();
-void __gedf2();
-void __gesf2();
-void __gtdf2();
-void __gtsf2();
-void __ledf2();
-void __lesf2();
-void __lshrdi3();
-void __ltdf2();
-void __ltsf2();
-void __moddi3();
-void __modsi3();
-void __muldf3();
-void __muldi3();
-void __mulsf3();
-void __nedf2();
-void __nesf2();
-void __subdf3();
-void __subsf3();
-void __truncdfsf2();
-void __udivdi3();
-void __udivmoddi4();
-void __udivsi3();
-void __umoddi3();
-void __umodsi3();
-void __unorddf2();
-void __unordsf2();
+void __adddf3(void);
+void __addsf3(void);
+void __aeabi_d2f(void);
+void __aeabi_d2iz(void);
+void __aeabi_d2lz(void);
+void __aeabi_d2uiz(void);
+void __aeabi_d2ulz(void);
+void __aeabi_dadd(void);
+void __aeabi_dcmpeq(void);
+void __aeabi_dcmpge(void);
+void __aeabi_dcmpgt(void);
+void __aeabi_dcmple(void);
+void __aeabi_dcmplt(void);
+void __aeabi_dcmpun(void);
+void __aeabi_ddiv(void);
+void __aeabi_dmul(void);
+void __aeabi_dsub(void);
+void __aeabi_f2d(void);
+void __aeabi_f2iz(void);
+void __aeabi_f2lz(void);
+void __aeabi_f2ulz(void);
+void __aeabi_fadd(void);
+void __aeabi_fcmpeq(void);
+void __aeabi_fcmpge(void);
+void __aeabi_fcmpgt(void);
+void __aeabi_fcmple(void);
+void __aeabi_fcmplt(void);
+void __aeabi_fcmpun(void);
+void __aeabi_fdiv(void);
+void __aeabi_fmul(void);
+void __aeabi_fsub(void);
+void __aeabi_i2d(void);
+void __aeabi_i2f(void);
+void __aeabi_idiv(void);
+void __aeabi_idivmod(void);
+void __aeabi_l2d(void);
+void __aeabi_l2f(void);
+void __aeabi_ldivmod(void);
+void __aeabi_llsl(void);
+void __aeabi_llsr(void);
+void __aeabi_lmul(void);
+void __aeabi_ui2d(void);
+void __aeabi_ui2f(void);
+void __aeabi_uidiv(void);
+void __aeabi_uidivmod(void);
+void __aeabi_ul2d(void);
+void __aeabi_ul2f(void);
+void __aeabi_uldivmod(void);
+void __ashldi3(void);
+void __clzsi2(void);
+void __divdf3(void);
+void __divdi3(void);
+void __divsi3(void);
+void __eqdf2(void);
+void __eqsf2(void);
+void __extendsfdf2(void);
+void __fixdfdi(void);
+void __fixdfsi(void);
+void __fixsfdi(void);
+void __fixunsdfdi(void);
+void __fixunsdfsi(void);
+void __fixunssfdi(void);
+void __floatdidf(void);
+void __floatdisf(void);
+void __floatsidf(void);
+void __floatsisf(void);
+void __floatundidf(void);
+void __floatundisf(void);
+void __floatunsidf(void);
+void __floatunsisf(void);
+void __gedf2(void);
+void __gesf2(void);
+void __gtdf2(void);
+void __gtsf2(void);
+void __ledf2(void);
+void __lesf2(void);
+void __lshrdi3(void);
+void __ltdf2(void);
+void __ltsf2(void);
+void __moddi3(void);
+void __modsi3(void);
+void __muldf3(void);
+void __muldi3(void);
+void __mulsf3(void);
+void __nedf2(void);
+void __nesf2(void);
+void __subdf3(void);
+void __subsf3(void);
+void __truncdfsf2(void);
+void __udivdi3(void);
+void __udivmoddi4(void);
+void __udivsi3(void);
+void __umoddi3(void);
+void __umodsi3(void);
+void __unorddf2(void);
+void __unordsf2(void);
 /* clang-format on */
 
 static SymbolMap target_sym_map[] = {
@@ -259,7 +259,7 @@ get_current_target(char *target_buf, uint32 target_buf_size)
 #undef BUILD_TARGET_THUMB_V4T
 
 uint32
-get_plt_item_size()
+get_plt_item_size(void)
 {
     /* 16 bytes instructions and 4 bytes symbol address */
     return 20;

core/iwasm/aot/arch/aot_reloc_x86_32.c

@@ -30,36 +30,39 @@ void __umoddi3();
 #pragma function(floor)
 #pragma function(ceil)
 
-static int64
-__divdi3(int64 a, int64 b)
+static int64 __stdcall __divdi3(int64 a, int64 b)
 {
     return a / b;
 }
 
-static uint64
-__udivdi3(uint64 a, uint64 b)
+static uint64 __stdcall __udivdi3(uint64 a, uint64 b)
 {
     return a / b;
 }
 
-static int64
-__moddi3(int64 a, int64 b)
+static int64 __stdcall __moddi3(int64 a, int64 b)
 {
     return a % b;
 }
 
-static uint64
-__umoddi3(uint64 a, uint64 b)
+static uint64 __stdcall __umoddi3(uint64 a, uint64 b)
 {
     return a % b;
 }
-#endif
 
-static uint64
-__aulldiv(uint64 a, uint64 b)
+static uint64 __stdcall __aulldiv(uint64 a, uint64 b)
 {
     return a / b;
 }
+static int64 __stdcall __alldiv(int64 a, int64 b)
+{
+    return a / b;
+}
+static int64 __stdcall __allrem(int64 a, int64 b)
+{
+    return a % b;
+}
+#endif /* !defined(_WIN32) && !defined(_WIN32_) */
 
 /* clang-format off */
 static SymbolMap target_sym_map[] = {
@@ -69,7 +72,11 @@ static SymbolMap target_sym_map[] = {
     REG_SYM(__udivdi3),
     REG_SYM(__moddi3),
     REG_SYM(__umoddi3),
-    REG_SYM(__aulldiv)
+#if defined(_WIN32) || defined(_WIN32_)
+    REG_SYM(__aulldiv),
+    REG_SYM(__alldiv),
+    REG_SYM(__allrem)
+#endif /* defined(_WIN32) || defined(_WIN32_) */
 };
 /* clang-format on */
 

core/iwasm/aot/arch/aot_reloc_x86_64.c

@@ -58,7 +58,7 @@ get_current_target(char *target_buf, uint32 target_buf_size)
 }
 
 static uint32
-get_plt_item_size()
+get_plt_item_size(void)
 {
     /* size of mov instruction and jmp instruction */
     return 12;

core/iwasm/aot/arch/aot_reloc_xtensa.c

@@ -10,44 +10,44 @@
 
 /* clang-format off */
 /* for soft-float */
-void __floatsidf();
-void __divdf3();
-void __ltdf2();
+void __floatsidf(void);
+void __divdf3(void);
+void __ltdf2(void);
 
 /* for mul32 */
-void __mulsi3();
-void __muldi3();
+void __mulsi3(void);
+void __muldi3(void);
 
-void __modsi3();
+void __modsi3(void);
 
-void __divdi3();
+void __divdi3(void);
 
-void __udivdi3();
-void __unorddf2();
-void __adddf3();
-void __eqdf2();
-void __muldf3();
-void __gedf2();
-void __ledf2();
-void __fixunsdfsi();
-void __floatunsidf();
-void __subdf3();
-void __nedf2();
-void __fixdfsi();
-void __moddi3();
-void __extendsfdf2();
-void __truncdfsf2();
-void __gtdf2();
-void __umoddi3();
-void __floatdidf();
-void __divsf3();
-void __fixdfdi();
-void __floatundidf();
-void __fixsfdi();
-void __fixunssfdi();
-void __fixunsdfdi();
-void __floatdisf();
-void __floatundisf();
+void __udivdi3(void);
+void __unorddf2(void);
+void __adddf3(void);
+void __eqdf2(void);
+void __muldf3(void);
+void __gedf2(void);
+void __ledf2(void);
+void __fixunsdfsi(void);
+void __floatunsidf(void);
+void __subdf3(void);
+void __nedf2(void);
+void __fixdfsi(void);
+void __moddi3(void);
+void __extendsfdf2(void);
+void __truncdfsf2(void);
+void __gtdf2(void);
+void __umoddi3(void);
+void __floatdidf(void);
+void __divsf3(void);
+void __fixdfdi(void);
+void __floatundidf(void);
+void __fixsfdi(void);
+void __fixunssfdi(void);
+void __fixunsdfdi(void);
+void __floatdisf(void);
+void __floatundisf(void);
 
 
 static SymbolMap target_sym_map[] = {
@@ -119,7 +119,7 @@ get_current_target(char *target_buf, uint32 target_buf_size)
 }
 
 static uint32
-get_plt_item_size()
+get_plt_item_size(void)
 {
     return 0;
 }
@@ -154,7 +154,7 @@ check_reloc_offset(uint32 target_section_size, uint64 reloc_offset,
  * CPU like esp32 can read and write data through the instruction bus, but only
  * in a word aligned manner; non-word-aligned access will cause a CPU exception.
  * This function uses a world aligned manner to write 16bit value to instruction
- * addreess.
+ * address.
  */
 static void
 put_imm16_to_addr(int16 imm16, int16 *addr)

core/iwasm/aot/debug/jit_debug.c

@@ -69,10 +69,10 @@ typedef struct JITDescriptor {
  * and inline assembler statement inside.
  */
 void attribute_noinline
-__jit_debug_register_code();
+__jit_debug_register_code(void);
 
 void attribute_noinline
-__jit_debug_register_code()
+__jit_debug_register_code(void)
 {
     int x;
     *(char *)&x = '\0';
@@ -96,7 +96,7 @@ extern JITDescriptor __jit_debug_descriptor;
  * This gives the debugger an easy way to inject custom code to
  * handle the events.
  */
-void (*__jit_debug_register_code_ptr)() = __jit_debug_register_code;
+void (*__jit_debug_register_code_ptr)(void) = __jit_debug_register_code;
 
 #ifdef __cplusplus
 }
@@ -171,7 +171,7 @@ DestroyJITCodeEntryInternal(JITCodeEntry *entry)
 }
 
 bool
-jit_debug_engine_init()
+jit_debug_engine_init(void)
 {
     if (jit_debug_engine) {
         return true;
@@ -194,7 +194,7 @@ jit_debug_engine_init()
 }
 
 void
-jit_debug_engine_destroy()
+jit_debug_engine_destroy(void)
 {
     if (jit_debug_engine) {
         WASMJITEntryNode *node, *node_next;

core/iwasm/aot/debug/jit_debug.h

@@ -11,10 +11,10 @@ extern "C" {
 #endif
 
 bool
-jit_debug_engine_init();
+jit_debug_engine_init(void);
 
 void
-jit_debug_engine_destroy();
+jit_debug_engine_destroy(void);
 
 bool
 jit_code_entry_create(const uint8 *symfile_addr, uint64 symfile_size);

core/iwasm/aot/iwasm_aot.cmake

@@ -1,15 +1,37 @@
 # Copyright (C) 2019 Intel Corporation.  All rights reserved.
 # SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 
+# Yes. To solve the compatibility issue with CMAKE (>= 4.0), we need to update
+# our `cmake_minimum_required()` to 3.5. However, there are CMakeLists.txt
+# from 3rd parties that we should not alter. Therefore, in addition to
+# changing the `cmake_minimum_required()`, we should also add a configuration
+# here that is compatible with earlier versions.
+set(CMAKE_POLICY_VERSION_MINIMUM 3.5 FORCE)
+
 set (IWASM_AOT_DIR ${CMAKE_CURRENT_LIST_DIR})
 
 add_definitions (-DWASM_ENABLE_AOT=1)
 
 include_directories (${IWASM_AOT_DIR})
 
-file (GLOB c_source_all ${IWASM_AOT_DIR}/*.c)
+list (APPEND c_source_all
+  ${IWASM_AOT_DIR}/aot_intrinsic.c
+  ${IWASM_AOT_DIR}/aot_loader.c
+  ${IWASM_AOT_DIR}/aot_runtime.c
+)
 
-if (WAMR_BUILD_TARGET STREQUAL "X86_64" OR WAMR_BUILD_TARGET STREQUAL "AMD_64")
+if (WAMR_BUILD_LINUX_PERF EQUAL 1)
+  list (APPEND c_source_all ${IWASM_AOT_DIR}/aot_perf_map.c)
+endif ()
+
+if (WAMR_BUILD_AOT_VALIDATOR EQUAL 1)
+  list (APPEND c_source_all ${IWASM_AOT_DIR}/aot_validator.c)
+endif ()
+
+if (WAMR_BUILD_WAMR_COMPILER EQUAL 1)
+  # AOT reloc functions are not used during AOT compilation
+  set (arch_source ${IWASM_AOT_DIR}/arch/aot_reloc_dummy.c)
+elseif (WAMR_BUILD_TARGET STREQUAL "X86_64" OR WAMR_BUILD_TARGET STREQUAL "AMD_64")
   set (arch_source ${IWASM_AOT_DIR}/arch/aot_reloc_x86_64.c)
 elseif (WAMR_BUILD_TARGET STREQUAL "X86_32")
   set (arch_source ${IWASM_AOT_DIR}/arch/aot_reloc_x86_32.c)

core/iwasm/common/SConscript

@@ -13,13 +13,10 @@ cwd     = GetCurrentDir()
 
 src = Glob('*.c')
 
-if rtconfig.ARCH == 'arm':
-    if re.match('^cortex-m.*', rtconfig.CPU):
-        src += ['arch/invokeNative_thumb.s']
-    elif re.match('^cortex-a.*', rtconfig.CPU):
-        src += ['arch/invokeNative_arm.s']
-elif rtconfig.ARCH == 'ia32':
-    src += ['arch/invokeNative_ia32.s']
+if rtconfig.ARCH == 'arm' and re.match('^cortex-m.*', rtconfig.CPU):
+    src += ['arch/invokeNative_thumb.s']
+else:
+    src.append(f"arch/invokeNative_{rtconfig.ARCH}.s")
 
 CPPPATH = [cwd, cwd + '/../include']
 

core/iwasm/common/arch/invokeNative_aarch64.s

@@ -45,7 +45,7 @@ _invokeNative:
 
         /* Now x20 points to stack args */
 
-        /* Directly call the fucntion if no args in stack */
+        /* Directly call the function if no args in stack */
         cmp     x21, #0
         beq     call_func
 
@@ -69,7 +69,7 @@ loop_stack_args:                 /* copy stack arguments to stack */
 call_func:
         mov     x20, x30         /* save x30(lr) */
         blr     x19
-        mov     sp, x22          /* restore sp which is saved before calling fuction*/
+        mov     sp, x22          /* restore sp which is saved before calling function*/
 
 return:
         mov     x30,  x20              /* restore x30(lr) */

core/iwasm/common/arch/invokeNative_aarch64_simd.s

@@ -43,7 +43,7 @@ _invokeNative:
 
         /* Now x20 points to stack args */
 
-        /* Directly call the fucntion if no args in stack */
+        /* Directly call the function if no args in stack */
         cmp     x21, #0
         beq     call_func
 
@@ -67,7 +67,7 @@ loop_stack_args:                 /* copy stack arguments to stack */
 call_func:
         mov     x20, x30         /* save x30(lr) */
         blr     x19
-        mov     sp, x22          /* restore sp which is saved before calling fuction*/
+        mov     sp, x22          /* restore sp which is saved before calling function*/
 
 return:
         mov     x30,  x20              /* restore x30(lr) */

core/iwasm/common/arch/invokeNative_arm_vfp.s

@@ -53,7 +53,7 @@ _invokeNative:
         vldr    s13, [r4, #52]
         vldr    s14, [r4, #56]
         vldr    s15, [r4, #60]
-        /* Directly call the fucntion if no args in stack */
+        /* Directly call the function if no args in stack */
         cmp     r5, #0
         beq     call_func
 

core/iwasm/common/arch/invokeNative_riscv.S

@@ -4,7 +4,7 @@
  */
 
 /*
- * The float abi macros used bellow are from risc-v c api:
+ * The float abi macros used below are from risc-v c api:
  * https://github.com/riscv/riscv-c-api-doc/blob/master/riscv-c-api.md
  *
  */
@@ -130,7 +130,7 @@ _invokeNative:
 loop_stack_args:
         beq              t2, x0, call_func
         RV_OP_LOADREG    t5, 0(t1)                  /* load stack argument, t5 = argv[i] */
-        RV_OP_STOREREG   t5, 0(t4)                  /* store t5 to reseved stack, sp[j] = t5 */
+        RV_OP_STOREREG   t5, 0(t4)                  /* store t5 to reserved stack, sp[j] = t5 */
         addi             t1, t1, RV_REG_SIZE        /* move to next stack argument */
         addi             t4, t4, RV_REG_SIZE        /* move to next stack pointer */
         addi             t2, t2, -1                 /* decrease t2 every loop, nstacks = nstacks -1 */
@@ -142,7 +142,7 @@ call_func:
         /* restore registers pushed in stack or saved in another register */
 return:
         mv               sp, fp                     /* restore sp saved in fp before function call */
-        RV_OP_LOADREG    fp, 0 * RV_REG_SIZE(sp)    /* load previous frame poniter to fp register */
+        RV_OP_LOADREG    fp, 0 * RV_REG_SIZE(sp)    /* load previous frame pointer to fp register */
         RV_OP_LOADREG    ra, 1 * RV_REG_SIZE(sp)    /* load previous return address to ra register */
         addi             sp, sp, 2 * RV_REG_SIZE    /* pop frame, restore sp */
         jr               ra

core/iwasm/common/arch/invokeNative_thumb_vfp.s

@@ -55,7 +55,7 @@ _invokeNative:
         vldr    s13, [r4, #52]
         vldr    s14, [r4, #56]
         vldr    s15, [r4, #60]
-        /* Directly call the fucntion if no args in stack */
+        /* Directly call the function if no args in stack */
         cmp     r5, #0
         beq     call_func
 

core/iwasm/common/gc/gc_type.c

@@ -1145,6 +1145,14 @@ wasm_reftype_is_subtype_of(uint8 type1, const WASMRefType *ref_type1,
                 return true;
             else {
                 int32 heap_type = ref_type1->ref_ht_common.heap_type;
+                // We dont care whether type2 is nullable or not. So
+                // we normalize it into its related one-byte type.
+                if (type2 == REF_TYPE_HT_NULLABLE
+                    || type2 == REF_TYPE_HT_NON_NULLABLE) {
+                    bh_assert(ref_type2);
+                    type2 = (uint8)(ref_type2->ref_ht_common.heap_type
+                                    + REF_TYPE_FUNCREF - HEAP_TYPE_FUNC);
+                }
                 if (heap_type == HEAP_TYPE_ANY) {
                     /* (ref any) <: anyref */
                     return type2 == REF_TYPE_ANYREF ? true : false;
@@ -1188,19 +1196,15 @@ wasm_reftype_is_subtype_of(uint8 type1, const WASMRefType *ref_type1,
                 }
 #endif
                 else if (heap_type == HEAP_TYPE_NONE) {
-                    /* (ref none) */
-                    /* TODO */
-                    bh_assert(0);
+                    return wasm_is_reftype_supers_of_none(type2, NULL, types,
+                                                          type_count);
                 }
                 else if (heap_type == HEAP_TYPE_NOEXTERN) {
-                    /* (ref noextern) */
-                    /* TODO */
-                    bh_assert(0);
+                    return wasm_is_reftype_supers_of_noextern(type2);
                 }
                 else if (heap_type == HEAP_TYPE_NOFUNC) {
-                    /* (ref nofunc) */
-                    /* TODO */
-                    bh_assert(0);
+                    return wasm_is_reftype_supers_of_nofunc(type2, NULL, types,
+                                                            type_count);
                 }
                 else {
                     bh_assert(0);

core/iwasm/common/wasm_application.c

@@ -105,7 +105,8 @@ execute_main(WASMModuleInstanceCommon *module_inst, int32 argc, char *argv[])
     bool ret, is_import_func = true, is_memory64 = false;
 #if WASM_ENABLE_MEMORY64 != 0
     WASMModuleInstance *wasm_module_inst = (WASMModuleInstance *)module_inst;
-    is_memory64 = wasm_module_inst->memories[0]->is_memory64;
+    if (wasm_module_inst->memory_count > 0)
+        is_memory64 = wasm_module_inst->memories[0]->is_memory64;
 #endif
 
     exec_env = wasm_runtime_get_exec_env_singleton(module_inst);
@@ -117,7 +118,7 @@ execute_main(WASMModuleInstanceCommon *module_inst, int32 argc, char *argv[])
 
 #if WASM_ENABLE_LIBC_WASI != 0
     /* In wasi mode, we should call the function named "_start"
-       which initializes the wasi envrionment and then calls
+       which initializes the wasi environment and then calls
        the actual main function. Directly calling main function
        may cause exception thrown. */
     if ((func = wasm_runtime_lookup_wasi_start_function(module_inst))) {

core/iwasm/common/wasm_c_api.c

@@ -2257,7 +2257,7 @@ wasm_module_new_ex(wasm_store_t *store, wasm_byte_vec_t *binary, LoadArgs *args)
     if (!store || !binary || binary->size == 0 || binary->size > UINT32_MAX)
         goto quit;
 
-    /* whether the combination of compilation flags are compatable with the
+    /* whether the combination of compilation flags are compatible with the
      * package type */
     {
         PackageType pkg_type;
@@ -3330,6 +3330,7 @@ argv_to_results(const uint32 *argv, const wasm_valtype_vec_t *result_defs,
                 break;
 #if WASM_ENABLE_GC == 0 && WASM_ENABLE_REF_TYPES != 0
             case WASM_EXTERNREF:
+            case WASM_FUNCREF:
                 result->of.ref = (struct wasm_ref_t *)(*(uintptr_t *)argv);
                 argv += sizeof(uintptr_t) / sizeof(uint32);
                 break;
@@ -3383,22 +3384,8 @@ wasm_func_call(const wasm_func_t *func, const wasm_val_vec_t *params,
         if (!(func_comm_rt = func->func_comm_rt)) {
             AOTModuleInstance *inst_aot =
                 (AOTModuleInstance *)func->inst_comm_rt;
-            AOTModule *module_aot = (AOTModule *)inst_aot->module;
-            uint32 export_i = 0, export_func_j = 0;
-
-            for (; export_i < module_aot->export_count; ++export_i) {
-                AOTExport *export = module_aot->exports + export_i;
-                if (export->kind == EXPORT_KIND_FUNC) {
-                    if (export->index == func->func_idx_rt) {
-                        func_comm_rt =
-                            (AOTFunctionInstance *)inst_aot->export_functions
-                            + export_func_j;
-                        ((wasm_func_t *)func)->func_comm_rt = func_comm_rt;
-                        break;
-                    }
-                    export_func_j++;
-                }
-            }
+            func_comm_rt = ((wasm_func_t *)func)->func_comm_rt =
+                aot_lookup_function_with_idx(inst_aot, func->func_idx_rt);
         }
 #endif
     }
@@ -3455,6 +3442,8 @@ wasm_func_call(const wasm_func_t *func, const wasm_val_vec_t *params,
     if (result_count) {
         if (!argv_to_results(argv, wasm_functype_results(func->type),
                              results)) {
+            wasm_runtime_set_exception(func->inst_comm_rt,
+                                       "argv_to_results failed");
             goto failed;
         }
         results->num_elems = result_count;

core/iwasm/common/wasm_exec_env.c

@@ -202,7 +202,7 @@ wasm_exec_env_destroy(WASMExecEnv *exec_env)
         wasm_cluster_wait_for_all_except_self(cluster, exec_env);
 #if WASM_ENABLE_DEBUG_INTERP != 0
         /* Must fire exit event after other threads exits, otherwise
-           the stopped thread will be overriden by other threads */
+           the stopped thread will be overridden by other threads */
         wasm_cluster_thread_exited(exec_env);
 #endif
         /* We have waited for other threads, this is the only alive thread, so
@@ -282,8 +282,13 @@ wasm_exec_env_set_thread_info(WASMExecEnv *exec_env)
     os_mutex_lock(&exec_env->wait_lock);
 #endif
     exec_env->handle = os_self_thread();
-    exec_env->native_stack_boundary =
-        stack_boundary ? stack_boundary + WASM_STACK_GUARD_SIZE : NULL;
+    if (exec_env->user_native_stack_boundary)
+        /* WASM_STACK_GUARD_SIZE isn't added for flexibility to developer,
+           he must ensure that enough guard bytes are kept. */
+        exec_env->native_stack_boundary = exec_env->user_native_stack_boundary;
+    else
+        exec_env->native_stack_boundary =
+            stack_boundary ? stack_boundary + WASM_STACK_GUARD_SIZE : NULL;
     exec_env->native_stack_top_min = (void *)UINTPTR_MAX;
 #if WASM_ENABLE_THREAD_MGR != 0
     os_mutex_unlock(&exec_env->wait_lock);

core/iwasm/common/wasm_exec_env.h

@@ -136,6 +136,10 @@ typedef struct WASMExecEnv {
 
     void *user_data;
 
+    /* The boundary of native stack set by host embedder. It is used
+       if it is not NULL when calling wasm functions. */
+    uint8 *user_native_stack_boundary;
+
     /* The native thread handle of current thread */
     korp_tid handle;
 

core/iwasm/common/wasm_loader_common.c

@@ -19,6 +19,37 @@ wasm_loader_set_error_buf(char *error_buf, uint32 error_buf_size,
     }
 }
 
+#if WASM_ENABLE_MEMORY64 != 0
+bool
+check_memory64_flags_consistency(WASMModule *module, char *error_buf,
+                                 uint32 error_buf_size, bool is_aot)
+{
+    uint32 i;
+    bool wasm64_flag, all_wasm64 = true, none_wasm64 = true;
+
+    for (i = 0; i < module->import_memory_count; ++i) {
+        wasm64_flag =
+            module->import_memories[i].u.memory.mem_type.flags & MEMORY64_FLAG;
+        all_wasm64 &= wasm64_flag;
+        none_wasm64 &= !wasm64_flag;
+    }
+
+    for (i = 0; i < module->memory_count; ++i) {
+        wasm64_flag = module->memories[i].flags & MEMORY64_FLAG;
+        all_wasm64 &= wasm64_flag;
+        none_wasm64 &= !wasm64_flag;
+    }
+
+    if (!(all_wasm64 || none_wasm64)) {
+        wasm_loader_set_error_buf(
+            error_buf, error_buf_size,
+            "inconsistent limits wasm64 flags for memory sections", is_aot);
+        return false;
+    }
+    return true;
+}
+#endif
+
 bool
 wasm_memory_check_flags(const uint8 mem_flag, char *error_buf,
                         uint32 error_buf_size, bool is_aot)
@@ -60,6 +91,37 @@ wasm_memory_check_flags(const uint8 mem_flag, char *error_buf,
     return true;
 }
 
+bool
+wasm_table_check_flags(const uint8 table_flag, char *error_buf,
+                       uint32 error_buf_size, bool is_aot)
+{
+    /* Check whether certain features indicated by mem_flag are enabled in
+     * runtime */
+    if (table_flag > MAX_TABLE_SIZE_FLAG) {
+        if (table_flag & SHARED_TABLE_FLAG) {
+            wasm_loader_set_error_buf(error_buf, error_buf_size,
+                                      "tables cannot be shared", is_aot);
+        }
+#if WASM_ENABLE_MEMORY64 == 0
+        if (table_flag & TABLE64_FLAG) {
+            wasm_loader_set_error_buf(error_buf, error_buf_size,
+                                      "invalid limits flags(table64 flag was "
+                                      "found, please enable memory64)",
+                                      is_aot);
+            return false;
+        }
+#endif
+    }
+
+    if (table_flag > MAX_TABLE_SIZE_FLAG + TABLE64_FLAG) {
+        wasm_loader_set_error_buf(error_buf, error_buf_size,
+                                  "invalid limits flags", is_aot);
+        return false;
+    }
+
+    return true;
+}
+
 /*
  * compare with a bigger type set in `wasm_value_type_size_internal()`,
  * this function will only cover global value type, function's param
@@ -89,7 +151,8 @@ is_valid_value_type(uint8 type)
 bool
 is_valid_value_type_for_interpreter(uint8 value_type)
 {
-#if (WASM_ENABLE_WAMR_COMPILER == 0) && (WASM_ENABLE_JIT == 0)
+#if (WASM_ENABLE_WAMR_COMPILER == 0) && (WASM_ENABLE_JIT == 0) \
+    && (WASM_ENABLE_FAST_INTERP == 0)
     /*
      * Note: regardless of WASM_ENABLE_SIMD, our interpreters don't have
      * SIMD implemented. It's safer to reject v128, especially for the

core/iwasm/common/wasm_loader_common.h

@@ -13,10 +13,22 @@
 extern "C" {
 #endif
 
+#if WASM_ENABLE_MEMORY64 != 0
+/* check consistency of memory64 flags across all memories,
+ * they must be either all wasm64 or all wasm32 */
+bool
+check_memory64_flags_consistency(WASMModule *module, char *error_buf,
+                                 uint32 error_buf_size, bool is_aot);
+#endif
+
 bool
 wasm_memory_check_flags(const uint8 mem_flag, char *error_buf,
                         uint32 error_buf_size, bool is_aot);
 
+bool
+wasm_table_check_flags(const uint8 table_flag, char *error_buf,
+                       uint32 error_buf_size, bool is_aot);
+
 bool
 is_valid_value_type(uint8 value_tpye);
 

core/iwasm/common/wasm_memory.c

@@ -13,6 +13,10 @@
 #include "../common/wasm_shared_memory.h"
 #endif
 
+#if WASM_ENABLE_THREAD_MGR != 0
+#include "../libraries/thread-mgr/thread_manager.h"
+#endif
+
 typedef enum Memory_Mode {
     MEMORY_MODE_UNKNOWN = 0,
     MEMORY_MODE_POOL,
@@ -24,6 +28,11 @@ static Memory_Mode memory_mode = MEMORY_MODE_UNKNOWN;
 
 static mem_allocator_t pool_allocator = NULL;
 
+#if WASM_ENABLE_SHARED_HEAP != 0
+static WASMSharedHeap *shared_heap_list = NULL;
+static korp_mutex shared_heap_list_lock;
+#endif
+
 static enlarge_memory_error_callback_t enlarge_memory_error_cb;
 static void *enlarge_memory_error_user_data;
 
@@ -132,16 +141,389 @@ is_bounds_checks_enabled(WASMModuleInstanceCommon *module_inst)
 #endif
 }
 
+#if WASM_ENABLE_SHARED_HEAP != 0
+static void *
+wasm_mmap_linear_memory(uint64_t map_size, uint64 commit_size);
+static void
+wasm_munmap_linear_memory(void *mapped_mem, uint64 commit_size,
+                          uint64 map_size);
+
+static void *
+runtime_malloc(uint64 size)
+{
+    void *mem;
+
+    if (size >= UINT32_MAX || !(mem = wasm_runtime_malloc((uint32)size))) {
+        LOG_WARNING("Allocate memory failed");
+        return NULL;
+    }
+
+    memset(mem, 0, (uint32)size);
+    return mem;
+}
+
+WASMSharedHeap *
+wasm_runtime_create_shared_heap(SharedHeapInitArgs *init_args)
+{
+    uint64 heap_struct_size = sizeof(WASMSharedHeap), map_size;
+    uint32 size = init_args->size;
+    WASMSharedHeap *heap;
+
+    if (size == 0) {
+        goto fail1;
+    }
+
+    if (!(heap = runtime_malloc(heap_struct_size))) {
+        goto fail1;
+    }
+
+    if (!(heap->heap_handle =
+              runtime_malloc(mem_allocator_get_heap_struct_size()))) {
+        goto fail2;
+    }
+
+    size = align_uint(size, os_getpagesize());
+    heap->size = size;
+    heap->start_off_mem64 = UINT64_MAX - heap->size + 1;
+    heap->start_off_mem32 = UINT32_MAX - heap->size + 1;
+
+    if (size > APP_HEAP_SIZE_MAX || size < APP_HEAP_SIZE_MIN) {
+        LOG_WARNING("Invalid size of shared heap");
+        goto fail3;
+    }
+
+#ifndef OS_ENABLE_HW_BOUND_CHECK
+    map_size = size;
+#else
+    /* Totally 8G is mapped, the opcode load/store address range is 0 to 8G:
+     *   ea = i + memarg.offset
+     * both i and memarg.offset are u32 in range 0 to 4G
+     * so the range of ea is 0 to 8G
+     */
+    map_size = 8 * (uint64)BH_GB;
+#endif
+
+    if (!(heap->base_addr = wasm_mmap_linear_memory(map_size, size))) {
+        goto fail3;
+    }
+    if (!mem_allocator_create_with_struct_and_pool(
+            heap->heap_handle, heap_struct_size, heap->base_addr, size)) {
+        LOG_WARNING("init share heap failed");
+        goto fail4;
+    }
+
+    os_mutex_lock(&shared_heap_list_lock);
+    if (shared_heap_list == NULL) {
+        shared_heap_list = heap;
+    }
+    else {
+        heap->next = shared_heap_list;
+        shared_heap_list = heap;
+    }
+    os_mutex_unlock(&shared_heap_list_lock);
+    return heap;
+
+fail4:
+    wasm_munmap_linear_memory(heap->base_addr, size, map_size);
+fail3:
+    wasm_runtime_free(heap->heap_handle);
+fail2:
+    wasm_runtime_free(heap);
+fail1:
+    return NULL;
+}
+
+bool
+wasm_runtime_attach_shared_heap_internal(WASMModuleInstanceCommon *module_inst,
+                                         WASMSharedHeap *shared_heap)
+{
+    WASMMemoryInstance *memory =
+        wasm_get_default_memory((WASMModuleInstance *)module_inst);
+    uint64 linear_mem_size;
+
+    if (!memory)
+        return false;
+
+    linear_mem_size = memory->memory_data_size;
+
+    /* check if linear memory and shared heap are overlapped */
+    if ((memory->is_memory64 && linear_mem_size > shared_heap->start_off_mem64)
+        || (!memory->is_memory64
+            && linear_mem_size > shared_heap->start_off_mem32)) {
+        LOG_WARNING("Linear memory address is overlapped with shared heap");
+        return false;
+    }
+
+#if WASM_ENABLE_INTERP != 0
+    if (module_inst->module_type == Wasm_Module_Bytecode) {
+        WASMModuleInstanceExtra *e =
+            (WASMModuleInstanceExtra *)((WASMModuleInstance *)module_inst)->e;
+        if (e->shared_heap) {
+            LOG_WARNING("A shared heap is already attached");
+            return false;
+        }
+        e->shared_heap = shared_heap;
+#if WASM_ENABLE_JIT != 0
+#if UINTPTR_MAX == UINT64_MAX
+        if (memory->is_memory64)
+            e->shared_heap_start_off.u64 = shared_heap->start_off_mem64;
+        else
+            e->shared_heap_start_off.u64 = shared_heap->start_off_mem32;
+        e->shared_heap_base_addr_adj =
+            shared_heap->base_addr - e->shared_heap_start_off.u64;
+#else
+        e->shared_heap_start_off.u32[0] = (uint32)shared_heap->start_off_mem32;
+        e->shared_heap_base_addr_adj =
+            shared_heap->base_addr - e->shared_heap_start_off.u32[0];
+#endif
+#endif /* end of WASM_ENABLE_JIT != 0 */
+    }
+#endif /* end of WASM_ENABLE_INTERP != 0 */
+#if WASM_ENABLE_AOT != 0
+    if (module_inst->module_type == Wasm_Module_AoT) {
+        AOTModuleInstanceExtra *e =
+            (AOTModuleInstanceExtra *)((AOTModuleInstance *)module_inst)->e;
+        if (e->shared_heap) {
+            LOG_WARNING("A shared heap is already attached");
+            return false;
+        }
+        e->shared_heap = shared_heap;
+#if UINTPTR_MAX == UINT64_MAX
+        if (memory->is_memory64)
+            e->shared_heap_start_off.u64 = shared_heap->start_off_mem64;
+        else
+            e->shared_heap_start_off.u64 = shared_heap->start_off_mem32;
+        e->shared_heap_base_addr_adj =
+            shared_heap->base_addr - e->shared_heap_start_off.u64;
+#else
+        e->shared_heap_start_off.u32[0] = (uint32)shared_heap->start_off_mem32;
+        e->shared_heap_base_addr_adj =
+            shared_heap->base_addr - e->shared_heap_start_off.u32[0];
+#endif
+    }
+#endif /* end of WASM_ENABLE_AOT != 0 */
+
+    return true;
+}
+
+bool
+wasm_runtime_attach_shared_heap(WASMModuleInstanceCommon *module_inst,
+                                WASMSharedHeap *shared_heap)
+{
+#if WASM_ENABLE_THREAD_MGR != 0
+    return wasm_cluster_attach_shared_heap(module_inst, shared_heap);
+#else
+    return wasm_runtime_attach_shared_heap_internal(module_inst, shared_heap);
+#endif
+}
+
+void
+wasm_runtime_detach_shared_heap_internal(WASMModuleInstanceCommon *module_inst)
+{
+#if WASM_ENABLE_INTERP != 0
+    if (module_inst->module_type == Wasm_Module_Bytecode) {
+        WASMModuleInstanceExtra *e =
+            (WASMModuleInstanceExtra *)((WASMModuleInstance *)module_inst)->e;
+        e->shared_heap = NULL;
+#if WASM_ENABLE_JIT != 0
+#if UINTPTR_MAX == UINT64_MAX
+        e->shared_heap_start_off.u64 = UINT64_MAX;
+#else
+        e->shared_heap_start_off.u32[0] = UINT32_MAX;
+#endif
+        e->shared_heap_base_addr_adj = NULL;
+#endif
+    }
+#endif /* end of WASM_ENABLE_INTERP != 0 */
+#if WASM_ENABLE_AOT != 0
+    if (module_inst->module_type == Wasm_Module_AoT) {
+        AOTModuleInstanceExtra *e =
+            (AOTModuleInstanceExtra *)((AOTModuleInstance *)module_inst)->e;
+        e->shared_heap = NULL;
+#if UINTPTR_MAX == UINT64_MAX
+        e->shared_heap_start_off.u64 = UINT64_MAX;
+#else
+        e->shared_heap_start_off.u32[0] = UINT32_MAX;
+#endif
+        e->shared_heap_base_addr_adj = NULL;
+    }
+#endif /* end of WASM_ENABLE_AOT != 0 */
+}
+
+void
+wasm_runtime_detach_shared_heap(WASMModuleInstanceCommon *module_inst)
+{
+#if WASM_ENABLE_THREAD_MGR != 0
+    wasm_cluster_detach_shared_heap(module_inst);
+#else
+    wasm_runtime_detach_shared_heap_internal(module_inst);
+#endif
+}
+
+static WASMSharedHeap *
+get_shared_heap(WASMModuleInstanceCommon *module_inst_comm)
+{
+#if WASM_ENABLE_INTERP != 0
+    if (module_inst_comm->module_type == Wasm_Module_Bytecode) {
+        return ((WASMModuleInstance *)module_inst_comm)->e->shared_heap;
+    }
+#endif
+#if WASM_ENABLE_AOT != 0
+    if (module_inst_comm->module_type == Wasm_Module_AoT) {
+        AOTModuleInstanceExtra *e =
+            (AOTModuleInstanceExtra *)((AOTModuleInstance *)module_inst_comm)
+                ->e;
+        return e->shared_heap;
+    }
+#endif
+    return NULL;
+}
+
+WASMSharedHeap *
+wasm_runtime_get_shared_heap(WASMModuleInstanceCommon *module_inst_comm)
+{
+    return get_shared_heap(module_inst_comm);
+}
+
+static bool
+is_app_addr_in_shared_heap(WASMModuleInstanceCommon *module_inst,
+                           bool is_memory64, uint64 app_offset, uint32 bytes)
+{
+    WASMSharedHeap *heap = get_shared_heap(module_inst);
+
+    if (!heap) {
+        return false;
+    }
+
+    if (bytes == 0) {
+        bytes = 1;
+    }
+
+    if (!is_memory64) {
+        if (app_offset >= heap->start_off_mem32
+            && app_offset <= UINT32_MAX - bytes + 1) {
+            return true;
+        }
+    }
+    else {
+        if (app_offset >= heap->start_off_mem64
+            && app_offset <= UINT64_MAX - bytes + 1) {
+            return true;
+        }
+    }
+
+    return false;
+}
+
+static bool
+is_native_addr_in_shared_heap(WASMModuleInstanceCommon *module_inst,
+                              uint8 *addr, uint32 bytes)
+{
+    WASMSharedHeap *heap = get_shared_heap(module_inst);
+    uintptr_t base_addr;
+    uintptr_t addr_int;
+    uintptr_t end_addr;
+
+    if (!heap) {
+        return false;
+    }
+
+    base_addr = (uintptr_t)heap->base_addr;
+    addr_int = (uintptr_t)addr;
+    if (addr_int < base_addr) {
+        return false;
+    }
+
+    end_addr = addr_int + bytes;
+    /* Check for overflow */
+    if (end_addr <= addr_int) {
+        return false;
+    }
+
+    if (end_addr > base_addr + heap->size) {
+        return false;
+    }
+
+    return true;
+}
+
+uint64
+wasm_runtime_shared_heap_malloc(WASMModuleInstanceCommon *module_inst,
+                                uint64_t size, void **p_native_addr)
+{
+    WASMMemoryInstance *memory =
+        wasm_get_default_memory((WASMModuleInstance *)module_inst);
+    WASMSharedHeap *shared_heap = get_shared_heap(module_inst);
+    void *native_addr = NULL;
+
+    if (!memory || !shared_heap)
+        return 0;
+
+    native_addr = mem_allocator_malloc(shared_heap->heap_handle, size);
+    if (!native_addr)
+        return 0;
+
+    if (p_native_addr) {
+        *p_native_addr = native_addr;
+    }
+
+    if (memory->is_memory64)
+        return shared_heap->start_off_mem64
+               + ((uint8 *)native_addr - shared_heap->base_addr);
+    else
+        return shared_heap->start_off_mem32
+               + ((uint8 *)native_addr - shared_heap->base_addr);
+}
+
+void
+wasm_runtime_shared_heap_free(WASMModuleInstanceCommon *module_inst, uint64 ptr)
+{
+    WASMMemoryInstance *memory =
+        wasm_get_default_memory((WASMModuleInstance *)module_inst);
+    WASMSharedHeap *shared_heap = get_shared_heap(module_inst);
+    uint8 *addr = NULL;
+
+    if (!memory || !shared_heap) {
+        return;
+    }
+
+    if (memory->is_memory64) {
+        if (ptr < shared_heap->start_off_mem64) { /* ptr can not > UINT64_MAX */
+            LOG_WARNING("The address to free isn't in shared heap");
+            return;
+        }
+        addr = shared_heap->base_addr + (ptr - shared_heap->start_off_mem64);
+    }
+    else {
+        if (ptr < shared_heap->start_off_mem32 || ptr > UINT32_MAX) {
+            LOG_WARNING("The address to free isn't in shared heap");
+            return;
+        }
+        addr = shared_heap->base_addr + (ptr - shared_heap->start_off_mem32);
+    }
+
+    mem_allocator_free(shared_heap->heap_handle, addr);
+}
+#endif /* end of WASM_ENABLE_SHARED_HEAP != 0 */
+
 bool
 wasm_runtime_memory_init(mem_alloc_type_t mem_alloc_type,
                          const MemAllocOption *alloc_option)
 {
+    bool ret = false;
+
+#if WASM_ENABLE_SHARED_HEAP != 0
+    if (os_mutex_init(&shared_heap_list_lock)) {
+        return false;
+    }
+#endif
+
     if (mem_alloc_type == Alloc_With_Pool) {
-        return wasm_memory_init_with_pool(alloc_option->pool.heap_buf,
-                                          alloc_option->pool.heap_size);
+        ret = wasm_memory_init_with_pool(alloc_option->pool.heap_buf,
+                                         alloc_option->pool.heap_size);
     }
     else if (mem_alloc_type == Alloc_With_Allocator) {
-        return wasm_memory_init_with_allocator(
+        ret = wasm_memory_init_with_allocator(
 #if WASM_MEM_ALLOC_WITH_USER_DATA != 0
             alloc_option->allocator.user_data,
 #endif
@@ -151,16 +533,58 @@ wasm_runtime_memory_init(mem_alloc_type_t mem_alloc_type,
     }
     else if (mem_alloc_type == Alloc_With_System_Allocator) {
         memory_mode = MEMORY_MODE_SYSTEM_ALLOCATOR;
-        return true;
+        ret = true;
     }
     else {
-        return false;
+        ret = false;
     }
+
+#if WASM_ENABLE_SHARED_HEAP != 0
+    if (!ret) {
+        os_mutex_destroy(&shared_heap_list_lock);
+    }
+#endif
+
+    return ret;
 }
 
-void
-wasm_runtime_memory_destroy()
+#if WASM_ENABLE_SHARED_HEAP != 0
+static void
+destroy_shared_heaps()
 {
+    WASMSharedHeap *heap;
+    WASMSharedHeap *cur;
+    uint64 map_size;
+
+    os_mutex_lock(&shared_heap_list_lock);
+    heap = shared_heap_list;
+    shared_heap_list = NULL;
+    os_mutex_unlock(&shared_heap_list_lock);
+
+    while (heap) {
+        cur = heap;
+        heap = heap->next;
+        mem_allocator_destroy(cur->heap_handle);
+        wasm_runtime_free(cur->heap_handle);
+#ifndef OS_ENABLE_HW_BOUND_CHECK
+        map_size = cur->size;
+#else
+        map_size = 8 * (uint64)BH_GB;
+#endif
+        wasm_munmap_linear_memory(cur->base_addr, cur->size, map_size);
+        wasm_runtime_free(cur);
+    }
+    os_mutex_destroy(&shared_heap_list_lock);
+}
+#endif
+
+void
+wasm_runtime_memory_destroy(void)
+{
+#if WASM_ENABLE_SHARED_HEAP != 0
+    destroy_shared_heaps();
+#endif
+
     if (memory_mode == MEMORY_MODE_POOL) {
 #if BH_ENABLE_GC_VERIFY == 0
         (void)mem_allocator_destroy(pool_allocator);
@@ -176,7 +600,7 @@ wasm_runtime_memory_destroy()
 }
 
 unsigned
-wasm_runtime_memory_pool_size()
+wasm_runtime_memory_pool_size(void)
 {
     if (memory_mode == MEMORY_MODE_POOL)
         return global_pool_size;
@@ -335,6 +759,13 @@ wasm_runtime_validate_app_addr(WASMModuleInstanceCommon *module_inst_comm,
         goto fail;
     }
 
+#if WASM_ENABLE_SHARED_HEAP != 0
+    if (is_app_addr_in_shared_heap(module_inst_comm, memory_inst->is_memory64,
+                                   app_offset, size)) {
+        return true;
+    }
+#endif
+
 #if WASM_ENABLE_MEMORY64 != 0
     if (memory_inst->is_memory64)
         max_linear_memory_size = MAX_LINEAR_MEM64_MEMORY_SIZE;
@@ -364,6 +795,7 @@ wasm_runtime_validate_app_str_addr(WASMModuleInstanceCommon *module_inst_comm,
                                    uint64 app_str_offset)
 {
     WASMModuleInstance *module_inst = (WASMModuleInstance *)module_inst_comm;
+    WASMMemoryInstance *memory_inst;
     uint64 app_end_offset, max_linear_memory_size = MAX_LINEAR_MEMORY_SIZE;
     char *str, *str_end;
 
@@ -374,22 +806,42 @@ wasm_runtime_validate_app_str_addr(WASMModuleInstanceCommon *module_inst_comm,
         return true;
     }
 
-    if (!wasm_runtime_get_app_addr_range(module_inst_comm, app_str_offset, NULL,
-                                         &app_end_offset))
+    memory_inst = wasm_get_default_memory(module_inst);
+    if (!memory_inst) {
         goto fail;
+    }
+
+#if WASM_ENABLE_SHARED_HEAP != 0
+    if (is_app_addr_in_shared_heap(module_inst_comm, memory_inst->is_memory64,
+                                   app_str_offset, 1)) {
+        WASMSharedHeap *shared_heap = get_shared_heap(module_inst_comm);
+        str = (char *)shared_heap->base_addr
+              + (memory_inst->is_memory64
+                     ? (app_str_offset - shared_heap->start_off_mem64)
+                     : (app_str_offset - shared_heap->start_off_mem32));
+        str_end = (char *)shared_heap->base_addr + shared_heap->size;
+    }
+    else
+#endif
+    {
+        if (!wasm_runtime_get_app_addr_range(module_inst_comm, app_str_offset,
+                                             NULL, &app_end_offset))
+            goto fail;
 
 #if WASM_ENABLE_MEMORY64 != 0
-    if (module_inst->memories[0]->is_memory64)
-        max_linear_memory_size = MAX_LINEAR_MEM64_MEMORY_SIZE;
+        if (memory_inst->is_memory64)
+            max_linear_memory_size = MAX_LINEAR_MEM64_MEMORY_SIZE;
 #endif
-    /* boundary overflow check, max start offset can only be size - 1, while end
-     * offset can be size */
-    if (app_str_offset >= max_linear_memory_size
-        || app_end_offset > max_linear_memory_size)
-        goto fail;
+        /* boundary overflow check, max start offset can be size - 1, while end
+           offset can be size */
+        if (app_str_offset >= max_linear_memory_size
+            || app_end_offset > max_linear_memory_size)
+            goto fail;
+
+        str = wasm_runtime_addr_app_to_native(module_inst_comm, app_str_offset);
+        str_end = str + (app_end_offset - app_str_offset);
+    }
 
-    str = wasm_runtime_addr_app_to_native(module_inst_comm, app_str_offset);
-    str_end = str + (app_end_offset - app_str_offset);
     while (str < str_end && *str != '\0')
         str++;
     if (str == str_end)
@@ -431,6 +883,12 @@ wasm_runtime_validate_native_addr(WASMModuleInstanceCommon *module_inst_comm,
         goto fail;
     }
 
+#if WASM_ENABLE_SHARED_HEAP != 0
+    if (is_native_addr_in_shared_heap(module_inst_comm, native_ptr, size)) {
+        return true;
+    }
+#endif
+
     SHARED_MEMORY_LOCK(memory_inst);
 
     if (memory_inst->memory_data <= addr
@@ -465,6 +923,23 @@ wasm_runtime_addr_app_to_native(WASMModuleInstanceCommon *module_inst_comm,
         return NULL;
     }
 
+#if WASM_ENABLE_SHARED_HEAP != 0
+    if (is_app_addr_in_shared_heap(module_inst_comm, memory_inst->is_memory64,
+                                   app_offset, 1)) {
+        WASMSharedHeap *shared_heap = get_shared_heap(module_inst_comm);
+        uint64 shared_heap_start = 0;
+
+        if (memory_inst && !memory_inst->is_memory64) {
+            shared_heap_start = shared_heap->start_off_mem32;
+        }
+        else if (memory_inst && memory_inst->is_memory64) {
+            shared_heap_start = shared_heap->start_off_mem64;
+        }
+
+        return shared_heap->base_addr + app_offset - shared_heap_start;
+    }
+#endif
+
     SHARED_MEMORY_LOCK(memory_inst);
 
     addr = memory_inst->memory_data + (uintptr_t)app_offset;
@@ -499,11 +974,32 @@ wasm_runtime_addr_native_to_app(WASMModuleInstanceCommon *module_inst_comm,
 
     bounds_checks = is_bounds_checks_enabled(module_inst_comm);
 
+#if WASM_ENABLE_SHARED_HEAP != 0
+    /* If shared heap is enabled, bounds check is always needed */
+    bounds_checks = true;
+#endif
+
     memory_inst = wasm_get_default_memory(module_inst);
     if (!memory_inst) {
         return 0;
     }
 
+#if WASM_ENABLE_SHARED_HEAP != 0
+    if (is_native_addr_in_shared_heap(module_inst_comm, addr, 1)) {
+        WASMSharedHeap *shared_heap = get_shared_heap(module_inst_comm);
+        uint64 shared_heap_start = 0;
+
+        if (memory_inst && !memory_inst->is_memory64) {
+            shared_heap_start = shared_heap->start_off_mem32;
+        }
+        else if (memory_inst && memory_inst->is_memory64) {
+            shared_heap_start = shared_heap->start_off_mem64;
+        }
+
+        return shared_heap_start + (addr - shared_heap->base_addr);
+    }
+#endif
+
     SHARED_MEMORY_LOCK(memory_inst);
 
     if (bounds_checks) {
@@ -601,6 +1097,10 @@ wasm_check_app_addr_and_convert(WASMModuleInstance *module_inst, bool is_str,
     WASMMemoryInstance *memory_inst = wasm_get_default_memory(module_inst);
     uint8 *native_addr;
     bool bounds_checks;
+#if WASM_ENABLE_SHARED_HEAP != 0
+    WASMSharedHeap *shared_heap;
+    bool is_in_shared_heap = false;
+#endif
 
     bh_assert(app_buf_addr <= UINTPTR_MAX && app_buf_size <= UINTPTR_MAX);
 
@@ -609,9 +1109,25 @@ wasm_check_app_addr_and_convert(WASMModuleInstance *module_inst, bool is_str,
         return false;
     }
 
-    native_addr = memory_inst->memory_data + (uintptr_t)app_buf_addr;
+#if WASM_ENABLE_SHARED_HEAP != 0
+    if (is_app_addr_in_shared_heap((WASMModuleInstanceCommon *)module_inst,
+                                   memory_inst->is_memory64, app_buf_addr,
+                                   app_buf_size)) {
+        shared_heap = get_shared_heap((WASMModuleInstanceCommon *)module_inst);
+        native_addr = shared_heap->base_addr
+                      + (memory_inst->is_memory64
+                             ? (app_buf_addr - shared_heap->start_off_mem64)
+                             : (app_buf_addr - shared_heap->start_off_mem32));
+        is_in_shared_heap = true;
+    }
+    else
+#endif
+    {
+        native_addr = memory_inst->memory_data + (uintptr_t)app_buf_addr;
+    }
 
-    bounds_checks = is_bounds_checks_enabled((wasm_module_inst_t)module_inst);
+    bounds_checks =
+        is_bounds_checks_enabled((WASMModuleInstanceCommon *)module_inst);
 
     if (!bounds_checks) {
         if (app_buf_addr == 0) {
@@ -620,6 +1136,24 @@ wasm_check_app_addr_and_convert(WASMModuleInstance *module_inst, bool is_str,
         goto success;
     }
 
+#if WASM_ENABLE_SHARED_HEAP != 0
+    if (is_in_shared_heap) {
+        const char *str, *str_end;
+
+        /* The whole string must be in the linear memory */
+        str = (const char *)native_addr;
+        str_end = (const char *)shared_heap->base_addr + shared_heap->size;
+        while (str < str_end && *str != '\0')
+            str++;
+        if (str == str_end) {
+            wasm_set_exception(module_inst, "out of bounds memory access");
+            return false;
+        }
+        else
+            goto success;
+    }
+#endif
+
     /* No need to check the app_offset and buf_size if memory access
        boundary check with hardware trap is enabled */
 #ifndef OS_ENABLE_HW_BOUND_CHECK
@@ -670,6 +1204,14 @@ wasm_get_default_memory(WASMModuleInstance *module_inst)
         return NULL;
 }
 
+WASMMemoryInstance *
+wasm_get_memory_with_idx(WASMModuleInstance *module_inst, uint32 index)
+{
+    if ((index >= module_inst->memory_count) || !module_inst->memories)
+        return NULL;
+    return module_inst->memories[index];
+}
+
 void
 wasm_runtime_set_mem_bound_check_bytes(WASMMemoryInstance *memory,
                                        uint64 memory_data_size)
@@ -711,6 +1253,14 @@ wasm_mremap_linear_memory(void *mapped_mem, uint64 old_size, uint64 new_size,
     bh_assert(new_size > 0);
     bh_assert(new_size > old_size);
 
+#if UINTPTR_MAX == UINT32_MAX
+    if (new_size == 4 * (uint64)BH_GB) {
+        LOG_WARNING("On 32 bit platform, linear memory can't reach maximum "
+                    "size of 4GB\n");
+        return NULL;
+    }
+#endif
+
     if (mapped_mem) {
         new_mem = os_mremap(mapped_mem, old_size, new_size);
     }
@@ -741,15 +1291,18 @@ wasm_mremap_linear_memory(void *mapped_mem, uint64 old_size, uint64 new_size,
 }
 
 static void *
-wasm_mmap_linear_memory(uint64_t map_size, uint64 commit_size)
+wasm_mmap_linear_memory(uint64 map_size, uint64 commit_size)
 {
     return wasm_mremap_linear_memory(NULL, 0, map_size, commit_size);
 }
 
-bool
-wasm_enlarge_memory_internal(WASMModuleInstance *module, uint32 inc_page_count)
+static bool
+wasm_enlarge_memory_internal(WASMModuleInstanceCommon *module,
+                             WASMMemoryInstance *memory, uint32 inc_page_count)
 {
-    WASMMemoryInstance *memory = wasm_get_default_memory(module);
+#if WASM_ENABLE_SHARED_HEAP != 0
+    WASMSharedHeap *shared_heap;
+#endif
     uint8 *memory_data_old, *memory_data_new, *heap_data_old;
     uint32 num_bytes_per_page, heap_size;
     uint32 cur_page_count, max_page_count, total_page_count;
@@ -797,6 +1350,24 @@ wasm_enlarge_memory_internal(WASMModuleInstance *module, uint32 inc_page_count)
         goto return_func;
     }
 
+#if WASM_ENABLE_SHARED_HEAP != 0
+    shared_heap = get_shared_heap(module);
+    if (shared_heap) {
+        if (memory->is_memory64
+            && total_size_new > shared_heap->start_off_mem64) {
+            LOG_WARNING("Linear memory address is overlapped with shared heap");
+            ret = false;
+            goto return_func;
+        }
+        else if (!memory->is_memory64
+                 && total_size_new > shared_heap->start_off_mem32) {
+            LOG_WARNING("Linear memory address is overlapped with shared heap");
+            ret = false;
+            goto return_func;
+        }
+    }
+#endif
+
     bh_assert(total_size_new
               <= GET_MAX_LINEAR_MEMORY_SIZE(memory->is_memory64));
 
@@ -804,7 +1375,7 @@ wasm_enlarge_memory_internal(WASMModuleInstance *module, uint32 inc_page_count)
     if (!(memory_data_new =
               realloc_func(Alloc_For_LinearMemory, full_size_mmaped,
 #if WASM_MEM_ALLOC_WITH_USER_DATA != 0
-                           NULL,
+                           allocator_user_data,
 #endif
                            memory_data_old, total_size_new))) {
         ret = false;
@@ -826,7 +1397,7 @@ wasm_enlarge_memory_internal(WASMModuleInstance *module, uint32 inc_page_count)
     if (full_size_mmaped) {
 #ifdef BH_PLATFORM_WINDOWS
         if (!os_mem_commit(memory->memory_data_end,
-                           (mem_offset_t)(total_size_new - total_size_old),
+                           total_size_new - total_size_old,
                            MMAP_PROT_READ | MMAP_PROT_WRITE)) {
             ret = false;
             goto return_func;
@@ -834,12 +1405,12 @@ wasm_enlarge_memory_internal(WASMModuleInstance *module, uint32 inc_page_count)
 #endif
 
         if (os_mprotect(memory->memory_data_end,
-                        (mem_offset_t)(total_size_new - total_size_old),
+                        total_size_new - total_size_old,
                         MMAP_PROT_READ | MMAP_PROT_WRITE)
             != 0) {
 #ifdef BH_PLATFORM_WINDOWS
             os_mem_decommit(memory->memory_data_end,
-                            (mem_offset_t)(total_size_new - total_size_old));
+                            total_size_new - total_size_old);
 #endif
             ret = false;
             goto return_func;
@@ -898,7 +1469,7 @@ wasm_enlarge_memory_internal(WASMModuleInstance *module, uint32 inc_page_count)
     wasm_runtime_set_mem_bound_check_bytes(memory, total_size_new);
 
 return_func:
-    if (!ret && enlarge_memory_error_cb) {
+    if (!ret && module && enlarge_memory_error_cb) {
         WASMExecEnv *exec_env = NULL;
 
 #if WASM_ENABLE_INTERP != 0
@@ -911,8 +1482,7 @@ return_func:
 #endif
 
         enlarge_memory_error_cb(inc_page_count, total_size_old, 0,
-                                failure_reason,
-                                (WASMModuleInstanceCommon *)module, exec_env,
+                                failure_reason, module, exec_env,
                                 enlarge_memory_error_user_data);
     }
 
@@ -956,15 +1526,136 @@ wasm_enlarge_memory(WASMModuleInstance *module, uint32 inc_page_count)
 {
     bool ret = false;
 
+    if (module->memory_count > 0) {
 #if WASM_ENABLE_SHARED_MEMORY != 0
-    if (module->memory_count > 0)
         shared_memory_lock(module->memories[0]);
 #endif
-    ret = wasm_enlarge_memory_internal(module, inc_page_count);
+        ret = wasm_enlarge_memory_internal((WASMModuleInstanceCommon *)module,
+                                           module->memories[0], inc_page_count);
 #if WASM_ENABLE_SHARED_MEMORY != 0
-    if (module->memory_count > 0)
         shared_memory_unlock(module->memories[0]);
 #endif
+    }
+
+    return ret;
+}
+
+bool
+wasm_enlarge_memory_with_idx(WASMModuleInstance *module, uint32 inc_page_count,
+                             uint32 memidx)
+{
+    bool ret = false;
+
+    if (memidx < module->memory_count) {
+#if WASM_ENABLE_SHARED_MEMORY != 0
+        shared_memory_lock(module->memories[memidx]);
+#endif
+        ret = wasm_enlarge_memory_internal((WASMModuleInstanceCommon *)module,
+                                           module->memories[memidx],
+                                           inc_page_count);
+#if WASM_ENABLE_SHARED_MEMORY != 0
+        shared_memory_unlock(module->memories[memidx]);
+#endif
+    }
+
+    return ret;
+}
+
+WASMMemoryInstance *
+wasm_runtime_lookup_memory(WASMModuleInstanceCommon *module_inst,
+                           const char *name)
+{
+#if WASM_ENABLE_INTERP != 0
+    if (module_inst->module_type == Wasm_Module_Bytecode)
+        return wasm_lookup_memory((WASMModuleInstance *)module_inst, name);
+#endif
+
+#if WASM_ENABLE_AOT != 0
+    if (module_inst->module_type == Wasm_Module_AoT)
+        return aot_lookup_memory((WASMModuleInstance *)module_inst, name);
+#endif
+
+    return NULL;
+}
+
+WASMMemoryInstance *
+wasm_runtime_get_default_memory(WASMModuleInstanceCommon *module_inst)
+{
+#if WASM_ENABLE_INTERP != 0
+    if (module_inst->module_type == Wasm_Module_Bytecode)
+        return wasm_get_default_memory((WASMModuleInstance *)module_inst);
+#endif
+
+#if WASM_ENABLE_AOT != 0
+    if (module_inst->module_type == Wasm_Module_AoT)
+        return aot_get_default_memory((AOTModuleInstance *)module_inst);
+#endif
+
+    return NULL;
+}
+
+WASMMemoryInstance *
+wasm_runtime_get_memory(WASMModuleInstanceCommon *module_inst, uint32 index)
+{
+#if WASM_ENABLE_INTERP != 0
+    if (module_inst->module_type == Wasm_Module_Bytecode)
+        return wasm_get_memory_with_idx((WASMModuleInstance *)module_inst,
+                                        index);
+#endif
+
+#if WASM_ENABLE_AOT != 0
+    if (module_inst->module_type == Wasm_Module_AoT)
+        return aot_get_memory_with_idx((AOTModuleInstance *)module_inst, index);
+#endif
+
+    return NULL;
+}
+
+uint64
+wasm_memory_get_cur_page_count(WASMMemoryInstance *memory)
+{
+    return memory->cur_page_count;
+}
+
+uint64
+wasm_memory_get_max_page_count(WASMMemoryInstance *memory)
+{
+    return memory->max_page_count;
+}
+
+uint64
+wasm_memory_get_bytes_per_page(WASMMemoryInstance *memory)
+{
+    return memory->num_bytes_per_page;
+}
+
+bool
+wasm_memory_get_shared(WASMMemoryInstance *memory)
+{
+    return memory->is_shared_memory;
+}
+
+void *
+wasm_memory_get_base_address(WASMMemoryInstance *memory)
+{
+    return memory->memory_data;
+}
+
+bool
+wasm_memory_enlarge(WASMMemoryInstance *memory, uint64 inc_page_count)
+{
+    bool ret = false;
+
+    if (memory) {
+#if WASM_ENABLE_SHARED_MEMORY != 0
+        shared_memory_lock(memory);
+#endif
+        ret =
+            wasm_enlarge_memory_internal(NULL, memory, (uint32)inc_page_count);
+#if WASM_ENABLE_SHARED_MEMORY != 0
+        shared_memory_unlock(memory);
+#endif
+    }
 
     return ret;
 }
@@ -997,7 +1688,7 @@ wasm_deallocate_linear_memory(WASMMemoryInstance *memory_inst)
     (void)map_size;
     free_func(Alloc_For_LinearMemory,
 #if WASM_MEM_ALLOC_WITH_USER_DATA != 0
-              NULL,
+              allocator_user_data,
 #endif
               memory_inst->memory_data);
 #else
@@ -1050,7 +1741,7 @@ wasm_allocate_linear_memory(uint8 **data, bool is_shared_memory,
         (void)wasm_mmap_linear_memory;
         if (!(*data = malloc_func(Alloc_For_LinearMemory,
 #if WASM_MEM_ALLOC_WITH_USER_DATA != 0
-                                  NULL,
+                                  allocator_user_data,
 #endif
                                   *memory_data_size))) {
             return BHT_ERROR;

core/iwasm/common/wasm_memory.h

@@ -41,15 +41,44 @@ SET_LINEAR_MEMORY_SIZE(WASMMemoryInstance *memory, uint64 size)
 #define SET_LINEAR_MEMORY_SIZE(memory, size) memory->memory_data_size = size
 #endif
 
+#if WASM_ENABLE_SHARED_HEAP != 0
+WASMSharedHeap *
+wasm_runtime_create_shared_heap(SharedHeapInitArgs *init_args);
+
+bool
+wasm_runtime_attach_shared_heap(WASMModuleInstanceCommon *module_inst,
+                                WASMSharedHeap *shared_heap);
+bool
+wasm_runtime_attach_shared_heap_internal(WASMModuleInstanceCommon *module_inst,
+                                         WASMSharedHeap *shared_heap);
+
+void
+wasm_runtime_detach_shared_heap(WASMModuleInstanceCommon *module_inst);
+
+void
+wasm_runtime_detach_shared_heap_internal(WASMModuleInstanceCommon *module_inst);
+
+WASMSharedHeap *
+wasm_runtime_get_shared_heap(WASMModuleInstanceCommon *module_inst_comm);
+
+uint64
+wasm_runtime_shared_heap_malloc(WASMModuleInstanceCommon *module_inst,
+                                uint64 size, void **p_native_addr);
+
+void
+wasm_runtime_shared_heap_free(WASMModuleInstanceCommon *module_inst,
+                              uint64 ptr);
+#endif
+
 bool
 wasm_runtime_memory_init(mem_alloc_type_t mem_alloc_type,
                          const MemAllocOption *alloc_option);
 
 void
-wasm_runtime_memory_destroy();
+wasm_runtime_memory_destroy(void);
 
 unsigned
-wasm_runtime_memory_pool_size();
+wasm_runtime_memory_pool_size(void);
 
 void
 wasm_runtime_set_mem_bound_check_bytes(WASMMemoryInstance *memory,

core/iwasm/common/wasm_native.c

@@ -33,6 +33,11 @@ uint32
 get_spectest_export_apis(NativeSymbol **p_libc_builtin_apis);
 #endif
 
+#if WASM_ENABLE_SHARED_HEAP != 0
+uint32
+get_lib_shared_heap_export_apis(NativeSymbol **p_shared_heap_apis);
+#endif
+
 uint32
 get_libc_wasi_export_apis(NativeSymbol **p_libc_wasi_apis);
 
@@ -469,7 +474,7 @@ wasi_context_dtor(WASMModuleInstanceCommon *inst, void *ctx)
 
 #if WASM_ENABLE_QUICK_AOT_ENTRY != 0
 static bool
-quick_aot_entry_init();
+quick_aot_entry_init(void);
 #endif
 
 bool
@@ -512,6 +517,14 @@ wasm_native_init()
         goto fail;
 #endif
 
+#if WASM_ENABLE_SHARED_HEAP != 0
+    n_native_symbols = get_lib_shared_heap_export_apis(&native_symbols);
+    if (n_native_symbols > 0
+        && !wasm_native_register_natives("env", native_symbols,
+                                         n_native_symbols))
+        goto fail;
+#endif
+
 #if WASM_ENABLE_BASE_LIB != 0
     n_native_symbols = get_base_lib_export_apis(&native_symbols);
     if (n_native_symbols > 0
@@ -1461,7 +1474,7 @@ quick_aot_entry_cmp(const void *quick_aot_entry1, const void *quick_aot_entry2)
 }
 
 static bool
-quick_aot_entry_init()
+quick_aot_entry_init(void)
 {
     qsort(quick_aot_entries, sizeof(quick_aot_entries) / sizeof(QuickAOTEntry),
           sizeof(QuickAOTEntry), quick_aot_entry_cmp);

core/iwasm/common/wasm_native.h

@@ -100,10 +100,10 @@ wasm_native_inherit_contexts(struct WASMModuleInstanceCommon *child,
 #endif /* WASM_ENABLE_MODULE_INST_CONTEXT */
 
 bool
-wasm_native_init();
+wasm_native_init(void);
 
 void
-wasm_native_destroy();
+wasm_native_destroy(void);
 
 #if WASM_ENABLE_QUICK_AOT_ENTRY != 0
 void *

core/iwasm/common/wasm_runtime_common.c

@@ -86,7 +86,7 @@ static bh_list registered_module_list_head;
 static bh_list *const registered_module_list = &registered_module_list_head;
 static korp_mutex registered_module_list_lock;
 static void
-wasm_runtime_destroy_registered_module_list();
+wasm_runtime_destroy_registered_module_list(void);
 #endif /* WASM_ENABLE_MULTI_MODULE */
 
 #define E_TYPE_XIP 4
@@ -97,11 +97,11 @@ val_type_to_val_kind(uint8 value_type);
 #if WASM_ENABLE_GC == 0 && WASM_ENABLE_REF_TYPES != 0
 /* Initialize externref hashmap */
 static bool
-wasm_externref_map_init();
+wasm_externref_map_init(void);
 
 /* Destroy externref hashmap */
 static void
-wasm_externref_map_destroy();
+wasm_externref_map_destroy(void);
 #endif /* end of WASM_ENABLE_GC == 0 && WASM_ENABLE_REF_TYPES != 0 */
 
 static void
@@ -181,15 +181,54 @@ static RunningMode runtime_running_mode = Mode_Default;
    of signal handler */
 static os_thread_local_attribute WASMExecEnv *exec_env_tls = NULL;
 
+static bool
+is_sig_addr_in_guard_pages(void *sig_addr, WASMModuleInstance *module_inst)
+{
+    WASMMemoryInstance *memory_inst;
+#if WASM_ENABLE_SHARED_HEAP != 0
+    WASMSharedHeap *shared_heap;
+#endif
+    uint8 *mapped_mem_start_addr = NULL;
+    uint8 *mapped_mem_end_addr = NULL;
+    uint32 i;
+
+    for (i = 0; i < module_inst->memory_count; ++i) {
+        /* To be compatible with multi memory, get the ith memory instance */
+        memory_inst = wasm_get_memory_with_idx(module_inst, i);
+        mapped_mem_start_addr = memory_inst->memory_data;
+        mapped_mem_end_addr = memory_inst->memory_data + 8 * (uint64)BH_GB;
+        if (mapped_mem_start_addr <= (uint8 *)sig_addr
+            && (uint8 *)sig_addr < mapped_mem_end_addr) {
+            /* The address which causes segmentation fault is inside
+               the memory instance's guard regions */
+            return true;
+        }
+    }
+
+#if WASM_ENABLE_SHARED_HEAP != 0
+    shared_heap =
+        wasm_runtime_get_shared_heap((WASMModuleInstanceCommon *)module_inst);
+    if (shared_heap) {
+        mapped_mem_start_addr = shared_heap->base_addr;
+        mapped_mem_end_addr = shared_heap->base_addr + 8 * (uint64)BH_GB;
+        if (mapped_mem_start_addr <= (uint8 *)sig_addr
+            && (uint8 *)sig_addr < mapped_mem_end_addr) {
+            /* The address which causes segmentation fault is inside
+               the shared heap's guard regions */
+            return true;
+        }
+    }
+#endif
+
+    return false;
+}
+
 #ifndef BH_PLATFORM_WINDOWS
 static void
 runtime_signal_handler(void *sig_addr)
 {
     WASMModuleInstance *module_inst;
-    WASMMemoryInstance *memory_inst;
     WASMJmpBuf *jmpbuf_node;
-    uint8 *mapped_mem_start_addr = NULL;
-    uint8 *mapped_mem_end_addr = NULL;
     uint32 page_size = os_getpagesize();
 #if WASM_DISABLE_STACK_HW_BOUND_CHECK == 0
     uint8 *stack_min_addr;
@@ -201,23 +240,13 @@ runtime_signal_handler(void *sig_addr)
         && (jmpbuf_node = exec_env_tls->jmpbuf_stack_top)) {
         /* Get mapped mem info of current instance */
         module_inst = (WASMModuleInstance *)exec_env_tls->module_inst;
-        /* Get the default memory instance */
-        memory_inst = wasm_get_default_memory(module_inst);
-        if (memory_inst) {
-            mapped_mem_start_addr = memory_inst->memory_data;
-            mapped_mem_end_addr = memory_inst->memory_data + 8 * (uint64)BH_GB;
-        }
 
 #if WASM_DISABLE_STACK_HW_BOUND_CHECK == 0
         /* Get stack info of current thread */
         stack_min_addr = os_thread_get_stack_boundary();
 #endif
 
-        if (memory_inst
-            && (mapped_mem_start_addr <= (uint8 *)sig_addr
-                && (uint8 *)sig_addr < mapped_mem_end_addr)) {
-            /* The address which causes segmentation fault is inside
-               the memory instance's guard regions */
+        if (is_sig_addr_in_guard_pages(sig_addr, module_inst)) {
             wasm_set_exception(module_inst, "out of bounds memory access");
             os_longjmp(jmpbuf_node->jmpbuf, 1);
         }
@@ -329,7 +358,6 @@ runtime_exception_handler(EXCEPTION_POINTERS *exce_info)
     PEXCEPTION_RECORD ExceptionRecord = exce_info->ExceptionRecord;
     uint8 *sig_addr = (uint8 *)ExceptionRecord->ExceptionInformation[1];
     WASMModuleInstance *module_inst;
-    WASMMemoryInstance *memory_inst;
     WASMJmpBuf *jmpbuf_node;
     uint8 *mapped_mem_start_addr = NULL;
     uint8 *mapped_mem_end_addr = NULL;
@@ -340,16 +368,7 @@ runtime_exception_handler(EXCEPTION_POINTERS *exce_info)
         && (jmpbuf_node = exec_env_tls->jmpbuf_stack_top)) {
         module_inst = (WASMModuleInstance *)exec_env_tls->module_inst;
         if (ExceptionRecord->ExceptionCode == EXCEPTION_ACCESS_VIOLATION) {
-            /* Get the default memory instance */
-            memory_inst = wasm_get_default_memory(module_inst);
-            if (memory_inst) {
-                mapped_mem_start_addr = memory_inst->memory_data;
-                mapped_mem_end_addr =
-                    memory_inst->memory_data + 8 * (uint64)BH_GB;
-            }
-
-            if (memory_inst && mapped_mem_start_addr <= (uint8 *)sig_addr
-                && (uint8 *)sig_addr < mapped_mem_end_addr) {
+            if (is_sig_addr_in_guard_pages(sig_addr, module_inst)) {
                 /* The address which causes segmentation fault is inside
                    the memory instance's guard regions.
                    Set exception and let the wasm func continue to run, when
@@ -436,7 +455,7 @@ wasm_runtime_get_exec_env_tls()
 #endif /* end of OS_ENABLE_HW_BOUND_CHECK */
 
 static bool
-wasm_runtime_env_init()
+wasm_runtime_env_init(void)
 {
     if (bh_platform_init() != 0)
         return false;
@@ -582,7 +601,7 @@ static korp_mutex runtime_lock = OS_THREAD_MUTEX_INITIALIZER;
 static int32 runtime_ref_count = 0;
 
 static bool
-wasm_runtime_init_internal()
+wasm_runtime_init_internal(void)
 {
     if (!wasm_runtime_memory_init(Alloc_With_System_Allocator, NULL))
         return false;
@@ -620,7 +639,7 @@ wasm_runtime_init()
 }
 
 static void
-wasm_runtime_destroy_internal()
+wasm_runtime_destroy_internal(void)
 {
 #if WASM_ENABLE_GC == 0 && WASM_ENABLE_REF_TYPES != 0
     wasm_externref_map_destroy();
@@ -738,7 +757,10 @@ wasm_runtime_full_init_internal(RuntimeInitArgs *init_args)
 #endif
 
 #if WASM_ENABLE_GC != 0
-    gc_heap_size_default = init_args->gc_heap_size;
+    uint32 gc_heap_size = init_args->gc_heap_size;
+    if (gc_heap_size > 0) {
+        gc_heap_size_default = gc_heap_size;
+    }
 #endif
 
 #if WASM_ENABLE_JIT != 0
@@ -1482,6 +1504,22 @@ wasm_runtime_load_ex(uint8 *buf, uint32 size, const LoadArgs *args,
                                           error_buf_size);
 }
 
+WASM_RUNTIME_API_EXTERN bool
+wasm_runtime_resolve_symbols(WASMModuleCommon *module)
+{
+#if WASM_ENABLE_INTERP != 0
+    if (module->module_type == Wasm_Module_Bytecode) {
+        return wasm_resolve_symbols((WASMModule *)module);
+    }
+#endif
+#if WASM_ENABLE_AOT != 0
+    if (module->module_type == Wasm_Module_AoT) {
+        return aot_resolve_symbols((AOTModule *)module);
+    }
+#endif
+    return false;
+}
+
 WASMModuleCommon *
 wasm_runtime_load(uint8 *buf, uint32 size, char *error_buf,
                   uint32 error_buf_size)
@@ -1705,6 +1743,45 @@ wasm_runtime_destroy_exec_env(WASMExecEnv *exec_env)
     wasm_exec_env_destroy(exec_env);
 }
 
+#if WAMR_ENABLE_COPY_CALLSTACK != 0
+uint32
+wasm_copy_callstack(const wasm_exec_env_t exec_env, wasm_frame_t *buffer,
+                    const uint32 length, const uint32 skip_n, char *error_buf,
+                    uint32_t error_buf_size)
+{
+    /*
+     * Note for devs: please refrain from such modifications inside of
+     * wasm_copy_callstack to preserve async-signal-safety
+     * - any allocations/freeing memory
+     * - dereferencing any pointers other than: exec_env, exec_env->module_inst,
+     * exec_env->module_inst->module, pointers between stack's bottom and
+     * top_boundary For more details check wasm_copy_callstack in
+     * wasm_export.h
+     */
+#if WASM_ENABLE_DUMP_CALL_STACK
+    WASMModuleInstance *module_inst =
+        (WASMModuleInstance *)get_module_inst(exec_env);
+
+#if WASM_ENABLE_INTERP != 0
+    if (module_inst->module_type == Wasm_Module_Bytecode) {
+        return wasm_interp_copy_callstack(exec_env, buffer, length, skip_n,
+                                          error_buf, error_buf_size);
+    }
+#endif
+
+#if WASM_ENABLE_AOT != 0
+    if (module_inst->module_type == Wasm_Module_AoT) {
+        return aot_copy_callstack(exec_env, buffer, length, skip_n, error_buf,
+                                  error_buf_size);
+    }
+#endif
+#endif
+    char *err_msg = "No copy_callstack API was actually executed";
+    strncpy(error_buf, err_msg, error_buf_size);
+    return 0;
+}
+#endif // WAMR_ENABLE_COPY_CALLSTACK
+
 bool
 wasm_runtime_init_thread_env(void)
 {
@@ -2190,6 +2267,13 @@ wasm_runtime_get_user_data(WASMExecEnv *exec_env)
     return exec_env->user_data;
 }
 
+void
+wasm_runtime_set_native_stack_boundary(WASMExecEnv *exec_env,
+                                       uint8 *native_stack_boundary)
+{
+    exec_env->user_native_stack_boundary = native_stack_boundary;
+}
+
 #ifdef OS_ENABLE_HW_BOUND_CHECK
 void
 wasm_runtime_access_exce_check_guard_page()
@@ -2972,9 +3056,9 @@ static const char *exception_msgs[] = {
     "wasm operand stack overflow",    /* EXCE_OPERAND_STACK_OVERFLOW */
     "failed to compile fast jit function", /* EXCE_FAILED_TO_COMPILE_FAST_JIT_FUNC */
     /* GC related exceptions */
-    "null function object",           /* EXCE_NULL_FUNC_OBJ */
-    "null structure object",          /* EXCE_NULL_STRUCT_OBJ */
-    "null array reference",              /* EXCE_NULL_ARRAY_OBJ */
+    "null function reference",        /* EXCE_NULL_FUNC_OBJ */
+    "null structure reference",       /* EXCE_NULL_STRUCT_OBJ */
+    "null array reference",           /* EXCE_NULL_ARRAY_OBJ */
     "null i31 reference",             /* EXCE_NULL_I31_OBJ */
     "null reference",                 /* EXCE_NULL_REFERENCE */
     "create rtt type failed",         /* EXCE_FAILED_TO_CREATE_RTT_TYPE */
@@ -2982,7 +3066,7 @@ static const char *exception_msgs[] = {
     "create array object failed",     /* EXCE_FAILED_TO_CREATE_ARRAY_OBJ */
     "create externref object failed", /* EXCE_FAILED_TO_CREATE_EXTERNREF_OBJ */
     "cast failure",                   /* EXCE_CAST_FAILURE */
-    "out of bounds array access",      /* EXCE_ARRAY_IDX_OOB */
+    "out of bounds array access",     /* EXCE_ARRAY_IDX_OOB */
     /* stringref related exceptions */
     "create string object failed",    /* EXCE_FAILED_TO_CREATE_STRING */
     "create stringref failed",        /* EXCE_FAILED_TO_CREATE_STRINGREF */
@@ -3575,7 +3659,8 @@ wasm_runtime_init_wasi(WASMModuleInstanceCommon *module_inst,
         char mapping_copy_buf[256];
         char *mapping_copy = mapping_copy_buf;
         char *map_mapped = NULL, *map_host = NULL;
-        const unsigned long max_len = strlen(map_dir_list[i]) * 2 + 3;
+        const unsigned long max_len =
+            (unsigned long)strlen(map_dir_list[i]) * 2 + 3;
 
         /* Allocation limit for runtime environments with reduced stack size */
         if (max_len > 256) {
@@ -3588,8 +3673,14 @@ wasm_runtime_init_wasi(WASMModuleInstanceCommon *module_inst,
 
         bh_memcpy_s(mapping_copy, max_len, map_dir_list[i],
                     (uint32)(strlen(map_dir_list[i]) + 1));
-        map_mapped = strtok(mapping_copy, "::");
-        map_host = strtok(NULL, "::");
+
+        const char *delim = "::";
+        char *delim_pos = strstr(mapping_copy, delim);
+        if (delim_pos) {
+            *delim_pos = '\0';
+            map_mapped = mapping_copy;
+            map_host = delim_pos + strlen(delim);
+        }
 
         if (!map_mapped || !map_host) {
             if (error_buf)
@@ -3865,23 +3956,18 @@ wasm_runtime_is_wasi_mode(WASMModuleInstanceCommon *module_inst)
 WASMFunctionInstanceCommon *
 wasm_runtime_lookup_wasi_start_function(WASMModuleInstanceCommon *module_inst)
 {
-    uint32 i;
-
 #if WASM_ENABLE_INTERP != 0
     if (module_inst->module_type == Wasm_Module_Bytecode) {
         WASMModuleInstance *wasm_inst = (WASMModuleInstance *)module_inst;
-        WASMFunctionInstance *func;
-        for (i = 0; i < wasm_inst->export_func_count; i++) {
-            if (!strcmp(wasm_inst->export_functions[i].name, "_start")) {
-                func = wasm_inst->export_functions[i].function;
-                if (func->u.func->func_type->param_count != 0
-                    || func->u.func->func_type->result_count != 0) {
-                    LOG_ERROR("Lookup wasi _start function failed: "
-                              "invalid function type.\n");
-                    return NULL;
-                }
-                return (WASMFunctionInstanceCommon *)func;
+        WASMFunctionInstance *func = wasm_lookup_function(wasm_inst, "_start");
+        if (func) {
+            if (func->u.func->func_type->param_count != 0
+                || func->u.func->func_type->result_count != 0) {
+                LOG_ERROR("Lookup wasi _start function failed: "
+                          "invalid function type.\n");
+                return NULL;
             }
+            return (WASMFunctionInstanceCommon *)func;
         }
         return NULL;
     }
@@ -3890,19 +3976,15 @@ wasm_runtime_lookup_wasi_start_function(WASMModuleInstanceCommon *module_inst)
 #if WASM_ENABLE_AOT != 0
     if (module_inst->module_type == Wasm_Module_AoT) {
         AOTModuleInstance *aot_inst = (AOTModuleInstance *)module_inst;
-        AOTFunctionInstance *export_funcs =
-            (AOTFunctionInstance *)aot_inst->export_functions;
-        for (i = 0; i < aot_inst->export_func_count; i++) {
-            if (!strcmp(export_funcs[i].func_name, "_start")) {
-                AOTFuncType *func_type = export_funcs[i].u.func.func_type;
-                if (func_type->param_count != 0
-                    || func_type->result_count != 0) {
-                    LOG_ERROR("Lookup wasi _start function failed: "
-                              "invalid function type.\n");
-                    return NULL;
-                }
-                return (WASMFunctionInstanceCommon *)&export_funcs[i];
+        AOTFunctionInstance *func = aot_lookup_function(aot_inst, "_start");
+        if (func) {
+            AOTFuncType *func_type = func->u.func.func_type;
+            if (func_type->param_count != 0 || func_type->result_count != 0) {
+                LOG_ERROR("Lookup wasi _start function failed: "
+                          "invalid function type.\n");
+                return NULL;
             }
+            return func;
         }
         return NULL;
     }
@@ -4226,31 +4308,68 @@ wasm_runtime_get_export_type(WASMModuleCommon *const module, int32 export_index,
         export_type->kind = aot_export->kind;
         switch (export_type->kind) {
             case WASM_IMPORT_EXPORT_KIND_FUNC:
-                export_type->u.func_type =
-                    (AOTFuncType *)aot_module
-                        ->types[aot_module->func_type_indexes
-                                    [aot_export->index
-                                     - aot_module->import_func_count]];
+            {
+                if (aot_export->index < aot_module->import_func_count) {
+                    export_type->u.func_type =
+                        (AOTFuncType *)aot_module
+                            ->import_funcs[aot_export->index]
+                            .func_type;
+                }
+                else {
+                    export_type->u.func_type =
+                        (AOTFuncType *)aot_module
+                            ->types[aot_module->func_type_indexes
+                                        [aot_export->index
+                                         - aot_module->import_func_count]];
+                }
                 break;
+            }
             case WASM_IMPORT_EXPORT_KIND_GLOBAL:
-                export_type->u.global_type =
-                    &aot_module
-                         ->globals[aot_export->index
-                                   - aot_module->import_global_count]
-                         .type;
+            {
+                if (aot_export->index < aot_module->import_global_count) {
+                    export_type->u.global_type =
+                        &aot_module->import_globals[aot_export->index].type;
+                }
+                else {
+                    export_type->u.global_type =
+                        &aot_module
+                             ->globals[aot_export->index
+                                       - aot_module->import_global_count]
+                             .type;
+                }
                 break;
+            }
             case WASM_IMPORT_EXPORT_KIND_TABLE:
-                export_type->u.table_type =
-                    &aot_module
-                         ->tables[aot_export->index
-                                  - aot_module->import_table_count]
-                         .table_type;
+            {
+                if (aot_export->index < aot_module->import_table_count) {
+                    export_type->u.table_type =
+                        &aot_module->import_tables[aot_export->index]
+                             .table_type;
+                }
+                else {
+                    export_type->u.table_type =
+                        &aot_module
+                             ->tables[aot_export->index
+                                      - aot_module->import_table_count]
+                             .table_type;
+                }
                 break;
+            }
             case WASM_IMPORT_EXPORT_KIND_MEMORY:
-                export_type->u.memory_type =
-                    &aot_module->memories[aot_export->index
-                                          - aot_module->import_memory_count];
+            {
+                if (aot_export->index < aot_module->import_memory_count) {
+                    export_type->u.memory_type =
+                        &aot_module->import_memories[aot_export->index]
+                             .mem_type;
+                }
+                else {
+                    export_type->u.memory_type =
+                        &aot_module
+                             ->memories[aot_export->index
+                                        - aot_module->import_memory_count];
+                }
                 break;
+            }
             default:
                 bh_assert(0);
                 break;
@@ -4272,31 +4391,76 @@ wasm_runtime_get_export_type(WASMModuleCommon *const module, int32 export_index,
         export_type->kind = wasm_export->kind;
         switch (export_type->kind) {
             case WASM_IMPORT_EXPORT_KIND_FUNC:
-                export_type->u.func_type =
-                    wasm_module
-                        ->functions[wasm_export->index
-                                    - wasm_module->import_function_count]
-                        ->func_type;
+            {
+                if (wasm_export->index < wasm_module->import_function_count) {
+                    export_type->u.func_type =
+                        (WASMFuncType *)wasm_module
+                            ->import_functions[wasm_export->index]
+                            .u.function.func_type;
+                }
+                else {
+                    export_type->u.func_type =
+                        wasm_module
+                            ->functions[wasm_export->index
+                                        - wasm_module->import_function_count]
+                            ->func_type;
+                }
+
                 break;
+            }
             case WASM_IMPORT_EXPORT_KIND_GLOBAL:
-                export_type->u.global_type =
-                    &wasm_module
-                         ->globals[wasm_export->index
-                                   - wasm_module->import_global_count]
-                         .type;
+            {
+                if (wasm_export->index < wasm_module->import_global_count) {
+                    export_type->u.global_type =
+                        (WASMGlobalType *)&wasm_module
+                            ->import_globals[wasm_export->index]
+                            .u.global.type;
+                }
+                else {
+                    export_type->u.global_type =
+                        &wasm_module
+                             ->globals[wasm_export->index
+                                       - wasm_module->import_global_count]
+                             .type;
+                }
+
                 break;
+            }
             case WASM_IMPORT_EXPORT_KIND_TABLE:
-                export_type->u.table_type =
-                    &wasm_module
-                         ->tables[wasm_export->index
-                                  - wasm_module->import_table_count]
-                         .table_type;
+            {
+                if (wasm_export->index < wasm_module->import_table_count) {
+                    export_type->u.table_type =
+                        (WASMTableType *)&wasm_module
+                            ->import_tables[wasm_export->index]
+                            .u.table.table_type;
+                }
+                else {
+                    export_type->u.table_type =
+                        &wasm_module
+                             ->tables[wasm_export->index
+                                      - wasm_module->import_table_count]
+                             .table_type;
+                }
+
                 break;
+            }
             case WASM_IMPORT_EXPORT_KIND_MEMORY:
-                export_type->u.memory_type =
-                    &wasm_module->memories[wasm_export->index
-                                           - wasm_module->import_memory_count];
+            {
+                if (wasm_export->index < wasm_module->import_memory_count) {
+                    export_type->u.memory_type =
+                        (WASMMemoryType *)&wasm_module
+                            ->import_memories[wasm_export->index]
+                            .u.memory.mem_type;
+                }
+                else {
+                    export_type->u.memory_type =
+                        &wasm_module
+                             ->memories[wasm_export->index
+                                        - wasm_module->import_memory_count];
+                }
+
                 break;
+            }
             default:
                 bh_assert(0);
                 break;
@@ -4336,8 +4500,9 @@ wasm_func_type_get_param_valkind(WASMFuncType *const func_type,
             return WASM_V128;
         case VALUE_TYPE_FUNCREF:
             return WASM_FUNCREF;
-
         case VALUE_TYPE_EXTERNREF:
+            return WASM_EXTERNREF;
+
         case VALUE_TYPE_VOID:
         default:
         {
@@ -4496,9 +4661,14 @@ wasm_runtime_invoke_native_raw(WASMExecEnv *exec_env, void *func_ptr,
                                uint32 *argv, uint32 argc, uint32 *argv_ret)
 {
     WASMModuleInstanceCommon *module = wasm_runtime_get_module_inst(exec_env);
+#if WASM_ENABLE_MEMORY64 != 0
+    WASMMemoryInstance *memory =
+        wasm_get_default_memory((WASMModuleInstance *)module);
+    bool is_memory64 = memory ? memory->is_memory64 : false;
+#endif
     typedef void (*NativeRawFuncPtr)(WASMExecEnv *, uint64 *);
     NativeRawFuncPtr invoke_native_raw = (NativeRawFuncPtr)func_ptr;
-    uint64 argv_buf[16] = { 0 }, *argv1 = argv_buf, *argv_dst, size, arg_i64;
+    uint64 argv_buf[16] = { 0 }, *argv1 = argv_buf, *argv_dst, size;
     uint32 *argv_src = argv, i, argc1, ptr_len;
     uint32 arg_i32;
     bool ret = false;
@@ -4523,11 +4693,11 @@ wasm_runtime_invoke_native_raw(WASMExecEnv *exec_env, void *func_ptr,
 #endif
             {
                 *(uint32 *)argv_dst = arg_i32 = *argv_src++;
-                /* TODO: memory64 if future there is a way for supporting
-                 * wasm64 and wasm32 in libc at the same time, remove the
-                 * macro control */
-#if WASM_ENABLE_MEMORY64 == 0
-                if (signature) {
+                if (signature
+#if WASM_ENABLE_MEMORY64 != 0
+                    && !is_memory64
+#endif
+                ) {
                     if (signature[i + 1] == '*') {
                         /* param is a pointer */
                         if (signature[i + 2] == '~')
@@ -4556,17 +4726,18 @@ wasm_runtime_invoke_native_raw(WASMExecEnv *exec_env, void *func_ptr,
                                 module, (uint64)arg_i32);
                     }
                 }
-#endif
                 break;
             }
             case VALUE_TYPE_I64:
 #if WASM_ENABLE_MEMORY64 != 0
             {
+                uint64 arg_i64;
+
                 PUT_I64_TO_ADDR((uint32 *)argv_dst,
                                 GET_I64_FROM_ADDR(argv_src));
                 argv_src += 2;
                 arg_i64 = *argv_dst;
-                if (signature) {
+                if (signature && is_memory64) {
                     /* TODO: memory64 pointer with length need a new symbol
                      * to represent type i64, with '~' still represent i32
                      * length */
@@ -4727,9 +4898,6 @@ wasm_runtime_invoke_native_raw(WASMExecEnv *exec_env, void *func_ptr,
 fail:
     if (argv1 != argv_buf)
         wasm_runtime_free(argv1);
-#if WASM_ENABLE_MEMORY64 == 0
-    (void)arg_i64;
-#endif
     return ret;
 }
 
@@ -4745,7 +4913,7 @@ fail:
     || defined(BUILD_TARGET_RISCV32_ILP32D)                          \
     || defined(BUILD_TARGET_RISCV32_ILP32F)                          \
     || defined(BUILD_TARGET_RISCV32_ILP32) || defined(BUILD_TARGET_ARC)
-typedef void (*GenericFunctionPointer)();
+typedef void (*GenericFunctionPointer)(void);
 void
 invokeNative(GenericFunctionPointer f, uint32 *args, uint32 n_stacks);
 
@@ -5310,7 +5478,7 @@ fail:
 #if defined(BUILD_TARGET_X86_32) || defined(BUILD_TARGET_ARM)    \
     || defined(BUILD_TARGET_THUMB) || defined(BUILD_TARGET_MIPS) \
     || defined(BUILD_TARGET_XTENSA)
-typedef void (*GenericFunctionPointer)();
+typedef void (*GenericFunctionPointer)(void);
 void
 invokeNative(GenericFunctionPointer f, uint32 *args, uint32 sz);
 
@@ -5595,7 +5763,7 @@ typedef uint32x4_t __m128i;
 
 #endif /* end of WASM_ENABLE_SIMD != 0 */
 
-typedef void (*GenericFunctionPointer)();
+typedef void (*GenericFunctionPointer)(void);
 void
 invokeNative(GenericFunctionPointer f, uint64 *args, uint64 n_stacks);
 
@@ -5653,6 +5821,11 @@ wasm_runtime_invoke_native(WASMExecEnv *exec_env, void *func_ptr,
                            uint32 *argv_ret)
 {
     WASMModuleInstanceCommon *module = wasm_runtime_get_module_inst(exec_env);
+#if WASM_ENABLE_MEMORY64 != 0
+    WASMMemoryInstance *memory =
+        wasm_get_default_memory((WASMModuleInstance *)module);
+    bool is_memory64 = memory ? memory->is_memory64 : false;
+#endif
     uint64 argv_buf[32] = { 0 }, *argv1 = argv_buf, *ints, *stacks, size,
            arg_i64;
     uint32 *argv_src = argv, i, argc1, n_ints = 0, n_stacks = 0;
@@ -5718,11 +5891,11 @@ wasm_runtime_invoke_native(WASMExecEnv *exec_env, void *func_ptr,
             {
                 arg_i32 = *argv_src++;
                 arg_i64 = arg_i32;
-                /* TODO: memory64 if future there is a way for supporting
-                 * wasm64 and wasm32 in libc at the same time, remove the
-                 * macro control */
-#if WASM_ENABLE_MEMORY64 == 0
-                if (signature) {
+                if (signature
+#if WASM_ENABLE_MEMORY64 != 0
+                    && !is_memory64
+#endif
+                ) {
                     if (signature[i + 1] == '*') {
                         /* param is a pointer */
                         if (signature[i + 2] == '~')
@@ -5749,7 +5922,6 @@ wasm_runtime_invoke_native(WASMExecEnv *exec_env, void *func_ptr,
                             module, (uint64)arg_i32);
                     }
                 }
-#endif
                 if (n_ints < MAX_REG_INTS)
                     ints[n_ints++] = arg_i64;
                 else
@@ -5761,7 +5933,7 @@ wasm_runtime_invoke_native(WASMExecEnv *exec_env, void *func_ptr,
             {
                 arg_i64 = GET_I64_FROM_ADDR(argv_src);
                 argv_src += 2;
-                if (signature) {
+                if (signature && is_memory64) {
                     /* TODO: memory64 pointer with length need a new symbol
                      * to represent type i64, with '~' still represent i32
                      * length */

core/iwasm/common/wasm_runtime_common.h

@@ -37,6 +37,10 @@ extern "C" {
     do {                                   \
         *(int64 *)(addr) = (int64)(value); \
     } while (0)
+#define PUT_V128_TO_ADDR(addr, value) \
+    do {                              \
+        *(V128 *)(addr) = (value);    \
+    } while (0)
 #define PUT_F64_TO_ADDR(addr, value)           \
     do {                                       \
         *(float64 *)(addr) = (float64)(value); \
@@ -49,6 +53,7 @@ extern "C" {
 #define GET_I64_FROM_ADDR(addr) (*(int64 *)(addr))
 #define GET_F64_FROM_ADDR(addr) (*(float64 *)(addr))
 #define GET_REF_FROM_ADDR(addr) (*(void **)(addr))
+#define GET_V128_FROM_ADDR(addr) (*(V128 *)(addr))
 
 /* For STORE opcodes */
 #define STORE_I64 PUT_I64_TO_ADDR
@@ -68,6 +73,12 @@ STORE_U8(void *addr, uint8_t value)
     *(uint8 *)addr = value;
 }
 
+static inline void
+STORE_V128(void *addr, V128 value)
+{
+    *(V128 *)addr = value;
+}
+
 /* For LOAD opcodes */
 #define LOAD_I64(addr) (*(int64 *)(addr))
 #define LOAD_F64(addr) (*(float64 *)(addr))
@@ -75,6 +86,7 @@ STORE_U8(void *addr, uint8_t value)
 #define LOAD_U32(addr) (*(uint32 *)(addr))
 #define LOAD_I16(addr) (*(int16 *)(addr))
 #define LOAD_U16(addr) (*(uint16 *)(addr))
+#define LOAD_V128(addr) (*(V128 *)(addr))
 
 #define STORE_PTR(addr, ptr)          \
     do {                              \
@@ -83,6 +95,15 @@ STORE_U8(void *addr, uint8_t value)
 
 #else /* WASM_CPU_SUPPORTS_UNALIGNED_ADDR_ACCESS != 0 */
 
+#define PUT_V128_TO_ADDR(addr, value)        \
+    do {                                     \
+        uint32 *addr_u32 = (uint32 *)(addr); \
+        addr_u32[0] = (value).i32x4[0];      \
+        addr_u32[1] = (value).i32x4[1];      \
+        addr_u32[2] = (value).i32x4[2];      \
+        addr_u32[3] = (value).i32x4[3];      \
+    } while (0)
+
 #define PUT_I64_TO_ADDR(addr, value)         \
     do {                                     \
         uint32 *addr_u32 = (uint32 *)(addr); \
@@ -124,6 +145,17 @@ STORE_U8(void *addr, uint8_t value)
     } while (0)
 #endif
 
+static inline V128
+GET_V128_FROM_ADDR(uint32 *addr)
+{
+    V128 ret;
+    ret.i32x4[0] = addr[0];
+    ret.i32x4[1] = addr[1];
+    ret.i32x4[2] = addr[2];
+    ret.i32x4[3] = addr[3];
+    return ret;
+}
+
 static inline int64
 GET_I64_FROM_ADDR(uint32 *addr)
 {
@@ -239,7 +271,94 @@ STORE_U16(void *addr, uint16_t value)
     ((uint8_t *)(addr))[0] = u.u8[0];
     ((uint8_t *)(addr))[1] = u.u8[1];
 }
+
+static inline void
+STORE_V128(void *addr, V128 value)
+{
+    uintptr_t addr_ = (uintptr_t)(addr);
+    union {
+        V128 val;
+        uint64 u64[2];
+        uint32 u32[4];
+        uint16 u16[8];
+        uint8 u8[16];
+    } u;
+
+    if ((addr_ & (uintptr_t)15) == 0) {
+        *(V128 *)addr = value;
+    }
+    else if ((addr_ & (uintptr_t)7) == 0) {
+        u.val = value;
+        ((uint64 *)(addr))[0] = u.u64[0];
+        ((uint64 *)(addr))[1] = u.u64[1];
+    }
+    else if ((addr_ & (uintptr_t)3) == 0) {
+        u.val = value;
+        ((uint32 *)addr)[0] = u.u32[0];
+        ((uint32 *)addr)[1] = u.u32[1];
+        ((uint32 *)addr)[2] = u.u32[2];
+        ((uint32 *)addr)[3] = u.u32[3];
+    }
+    else if ((addr_ & (uintptr_t)1) == 0) {
+        u.val = value;
+        ((uint16 *)addr)[0] = u.u16[0];
+        ((uint16 *)addr)[1] = u.u16[1];
+        ((uint16 *)addr)[2] = u.u16[2];
+        ((uint16 *)addr)[3] = u.u16[3];
+        ((uint16 *)addr)[4] = u.u16[4];
+        ((uint16 *)addr)[5] = u.u16[5];
+        ((uint16 *)addr)[6] = u.u16[6];
+        ((uint16 *)addr)[7] = u.u16[7];
+    }
+    else {
+        u.val = value;
+        for (int i = 0; i < 16; i++)
+            ((uint8 *)addr)[i] = u.u8[i];
+    }
+}
+
 /* For LOAD opcodes */
+static inline V128
+LOAD_V128(void *addr)
+{
+    uintptr_t addr1 = (uintptr_t)addr;
+    union {
+        V128 val;
+        uint64 u64[2];
+        uint32 u32[4];
+        uint16 u16[8];
+        uint8 u8[16];
+    } u;
+    if ((addr1 & (uintptr_t)15) == 0)
+        return *(V128 *)addr;
+
+    if ((addr1 & (uintptr_t)7) == 0) {
+        u.u64[0] = ((uint64 *)addr)[0];
+        u.u64[1] = ((uint64 *)addr)[1];
+    }
+    else if ((addr1 & (uintptr_t)3) == 0) {
+        u.u32[0] = ((uint32 *)addr)[0];
+        u.u32[1] = ((uint32 *)addr)[1];
+        u.u32[2] = ((uint32 *)addr)[2];
+        u.u32[3] = ((uint32 *)addr)[3];
+    }
+    else if ((addr1 & (uintptr_t)1) == 0) {
+        u.u16[0] = ((uint16 *)addr)[0];
+        u.u16[1] = ((uint16 *)addr)[1];
+        u.u16[2] = ((uint16 *)addr)[2];
+        u.u16[3] = ((uint16 *)addr)[3];
+        u.u16[4] = ((uint16 *)addr)[4];
+        u.u16[5] = ((uint16 *)addr)[5];
+        u.u16[6] = ((uint16 *)addr)[6];
+        u.u16[7] = ((uint16 *)addr)[7];
+    }
+    else {
+        for (int i = 0; i < 16; i++)
+            u.u8[i] = ((uint8 *)addr)[i];
+    }
+    return u.val;
+}
+
 static inline int64
 LOAD_I64(void *addr)
 {
@@ -464,19 +583,6 @@ typedef struct WASMRegisteredModule {
 typedef package_type_t PackageType;
 typedef wasm_section_t WASMSection, AOTSection;
 
-typedef struct wasm_frame_t {
-    /*  wasm_instance_t */
-    void *instance;
-    uint32 module_offset;
-    uint32 func_index;
-    uint32 func_offset;
-    const char *func_name_wp;
-
-    uint32 *sp;
-    uint8 *frame_ref;
-    uint32 *lp;
-} WASMCApiFrame;
-
 #if WASM_ENABLE_JIT != 0
 typedef struct LLVMJITOptions {
     uint32 opt_level;
@@ -652,6 +758,13 @@ wasm_runtime_create_exec_env(WASMModuleInstanceCommon *module_inst,
 WASM_RUNTIME_API_EXTERN void
 wasm_runtime_destroy_exec_env(WASMExecEnv *exec_env);
 
+#if WAMR_ENABLE_COPY_CALLSTACK != 0
+WASM_RUNTIME_API_EXTERN uint32_t
+wasm_copy_callstack(const wasm_exec_env_t exec_env, wasm_frame_t *buffer,
+                    const uint32 length, const uint32 skip_n, char *error_buf,
+                    uint32 error_buf_size);
+#endif // WAMR_ENABLE_COPY_CALLSTACK
+
 /* See wasm_export.h for description */
 WASM_RUNTIME_API_EXTERN WASMModuleInstanceCommon *
 wasm_runtime_get_module_inst(WASMExecEnv *exec_env);
@@ -673,6 +786,11 @@ wasm_runtime_set_user_data(WASMExecEnv *exec_env, void *user_data);
 WASM_RUNTIME_API_EXTERN void *
 wasm_runtime_get_user_data(WASMExecEnv *exec_env);
 
+/* See wasm_export.h for description */
+WASM_RUNTIME_API_EXTERN void
+wasm_runtime_set_native_stack_boundary(WASMExecEnv *exec_env,
+                                       uint8 *native_stack_boundary);
+
 #if WASM_CONFIGURABLE_BOUNDS_CHECKS != 0
 /* See wasm_export.h for description */
 WASM_RUNTIME_API_EXTERN void
@@ -852,10 +970,10 @@ wasm_runtime_set_module_reader(const module_reader reader,
                                const module_destroyer destroyer);
 
 module_reader
-wasm_runtime_get_module_reader();
+wasm_runtime_get_module_reader(void);
 
 module_destroyer
-wasm_runtime_get_module_destroyer();
+wasm_runtime_get_module_destroyer(void);
 
 bool
 wasm_runtime_register_module_internal(const char *module_name,
@@ -881,7 +999,7 @@ bool
 wasm_runtime_is_loading_module(const char *module_name);
 
 void
-wasm_runtime_destroy_loading_module_list();
+wasm_runtime_destroy_loading_module_list(void);
 
 WASMModuleCommon *
 wasm_runtime_search_sub_module(const WASMModuleCommon *parent_module,
@@ -1168,7 +1286,7 @@ wasm_runtime_quick_invoke_c_api_native(WASMModuleInstanceCommon *module_inst,
                                        uint32 result_count);
 
 void
-wasm_runtime_show_app_heap_corrupted_prompt();
+wasm_runtime_show_app_heap_corrupted_prompt(void);
 
 #if WASM_ENABLE_LOAD_CUSTOM_SECTION != 0
 void

core/iwasm/common/wasm_shared_memory.c

@@ -8,6 +8,9 @@
 #if WASM_ENABLE_THREAD_MGR != 0
 #include "../libraries/thread-mgr/thread_manager.h"
 #endif
+#if WASM_ENABLE_AOT != 0
+#include "../aot/aot_runtime.h"
+#endif
 
 /*
  * Note: this lock can be per memory.
@@ -243,6 +246,31 @@ map_try_release_wait_info(HashMap *wait_hash_map, AtomicWaitInfo *wait_info,
     destroy_wait_info(wait_info);
 }
 
+#if WASM_ENABLE_SHARED_HEAP != 0
+static bool
+is_native_addr_in_shared_heap(WASMModuleInstanceCommon *module_inst,
+                              uint8 *addr, uint32 bytes)
+{
+    WASMSharedHeap *shared_heap = NULL;
+
+#if WASM_ENABLE_INTERP != 0
+    if (module_inst->module_type == Wasm_Module_Bytecode) {
+        shared_heap = ((WASMModuleInstance *)module_inst)->e->shared_heap;
+    }
+#endif
+#if WASM_ENABLE_AOT != 0
+    if (module_inst->module_type == Wasm_Module_AoT) {
+        AOTModuleInstanceExtra *e =
+            (AOTModuleInstanceExtra *)((AOTModuleInstance *)module_inst)->e;
+        shared_heap = e->shared_heap;
+    }
+#endif
+
+    return shared_heap && addr >= shared_heap->base_addr
+           && addr + bytes <= shared_heap->base_addr + shared_heap->size;
+}
+#endif
+
 uint32
 wasm_runtime_atomic_wait(WASMModuleInstanceCommon *module, void *address,
                          uint64 expect, int64 timeout, bool wait64)
@@ -271,9 +299,17 @@ wasm_runtime_atomic_wait(WASMModuleInstanceCommon *module, void *address,
     }
 
     shared_memory_lock(module_inst->memories[0]);
-    if ((uint8 *)address < module_inst->memories[0]->memory_data
-        || (uint8 *)address + (wait64 ? 8 : 4)
-               > module_inst->memories[0]->memory_data_end) {
+    if (
+#if WASM_ENABLE_SHARED_HEAP != 0
+        /* not in shared heap */
+        !is_native_addr_in_shared_heap((WASMModuleInstanceCommon *)module_inst,
+                                       address, wait64 ? 8 : 4)
+        &&
+#endif
+        /* and not in linear memory */
+        ((uint8 *)address < module_inst->memories[0]->memory_data
+         || (uint8 *)address + (wait64 ? 8 : 4)
+                > module_inst->memories[0]->memory_data_end)) {
         shared_memory_unlock(module_inst->memories[0]);
         wasm_runtime_set_exception(module, "out of bounds memory access");
         return -1;
@@ -397,6 +433,11 @@ wasm_runtime_atomic_notify(WASMModuleInstanceCommon *module, void *address,
 
     shared_memory_lock(module_inst->memories[0]);
     out_of_bounds =
+#if WASM_ENABLE_SHARED_HEAP != 0
+        /* not in shared heap */
+        !is_native_addr_in_shared_heap(module, address, 4) &&
+#endif
+        /* and not in linear memory */
         ((uint8 *)address < module_inst->memories[0]->memory_data
          || (uint8 *)address + 4 > module_inst->memories[0]->memory_data_end);
     shared_memory_unlock(module_inst->memories[0]);

core/iwasm/common/wasm_shared_memory.h

@@ -17,10 +17,10 @@ extern "C" {
 extern korp_mutex g_shared_memory_lock;
 
 bool
-wasm_shared_memory_init();
+wasm_shared_memory_init(void);
 
 void
-wasm_shared_memory_destroy();
+wasm_shared_memory_destroy(void);
 
 uint16
 shared_memory_inc_reference(WASMMemoryInstance *memory);

core/iwasm/compilation/aot.c

@@ -36,8 +36,11 @@ aot_destroy_mem_init_data_list(AOTMemInitData **data_list, uint32 count)
 {
     uint32 i;
     for (i = 0; i < count; i++)
-        if (data_list[i])
+        if (data_list[i]) {
+            if (data_list[i]->bytes)
+                wasm_runtime_free(data_list[i]->bytes);
             wasm_runtime_free(data_list[i]);
+        }
     wasm_runtime_free(data_list);
 }
 
@@ -60,8 +63,7 @@ aot_create_mem_init_data_list(const WASMModule *module)
 
     /* Create each memory data segment */
     for (i = 0; i < module->data_seg_count; i++) {
-        size = offsetof(AOTMemInitData, bytes)
-               + (uint64)module->data_segments[i]->data_length;
+        size = sizeof(AOTMemInitData);
         if (size >= UINT32_MAX
             || !(data_list[i] = wasm_runtime_malloc((uint32)size))) {
             aot_set_last_error("allocate memory failed.");
@@ -69,18 +71,31 @@ aot_create_mem_init_data_list(const WASMModule *module)
         }
 
 #if WASM_ENABLE_BULK_MEMORY != 0
+        /* Set bulk memory specific properties if enabled */
         data_list[i]->is_passive = module->data_segments[i]->is_passive;
         data_list[i]->memory_index = module->data_segments[i]->memory_index;
 #endif
         data_list[i]->offset = module->data_segments[i]->base_offset;
         data_list[i]->byte_count = module->data_segments[i]->data_length;
-        memcpy(data_list[i]->bytes, module->data_segments[i]->data,
-               module->data_segments[i]->data_length);
+        data_list[i]->bytes = NULL;
+        /* Allocate memory for AOT compiler is OK, because the data segment
+         * is small and the host memory is enough */
+        if (data_list[i]->byte_count > 0) {
+            data_list[i]->bytes = wasm_runtime_malloc(data_list[i]->byte_count);
+            if (!data_list[i]->bytes) {
+                aot_set_last_error("allocate memory failed.");
+                goto fail;
+            }
+            /* Copy the actual data bytes from the WASM module */
+            memcpy(data_list[i]->bytes, module->data_segments[i]->data,
+                   module->data_segments[i]->data_length);
+        }
     }
 
     return data_list;
 
 fail:
+    /* Clean up allocated memory in case of failure */
     aot_destroy_mem_init_data_list(data_list, module->data_seg_count);
     return NULL;
 }
@@ -487,58 +502,69 @@ calculate_struct_field_sizes_offsets(AOTCompData *comp_data, bool is_target_x86,
 }
 #endif
 
-AOTCompData *
-aot_create_comp_data(WASMModule *module, const char *target_arch,
-                     bool gc_enabled)
+/**
+ * Checks if target architecture is 64-bit based on target_arch string.
+ *
+ * @param target_arch The target architecture string (e.g. "x86_64", "aarch64")
+ * @return true if target is 64-bit architecture, false otherwise
+ *
+ * If target_arch is NULL, detection is based on UINTPTR_MAX.
+ * Otherwise looks for "64" in target_arch string.
+ */
+static bool
+arch_is_64bit(const char *target_arch)
 {
-    AOTCompData *comp_data;
-    uint32 import_global_data_size_64bit = 0, global_data_size_64bit = 0, i, j;
-    uint32 import_global_data_size_32bit = 0, global_data_size_32bit = 0;
-    uint64 size;
-    bool is_64bit_target = false;
-#if WASM_ENABLE_GC != 0
-    bool is_target_x86 = false;
+    if (!target_arch) {
+#if UINTPTR_MAX == UINT64_MAX
+        return true;
+#else
+        return false;
 #endif
+    }
+    /* All 64bit targets contains "64" string in their target name */
+    return strstr(target_arch, "64") != NULL;
+}
 
-#if WASM_ENABLE_GC != 0
+/**
+ * Checks if target architecture is x86/x64 based on target_arch string.
+ *
+ * @param target_arch The target architecture string (e.g. "x86_64", "i386")
+ * @return true if target is x86/x64 architecture, false otherwise
+ *
+ * If target_arch is NULL, detection is based on build-time definitions.
+ * Otherwise checks for x86_64 or i386 in target_arch string.
+ */
+static bool
+arch_is_x86(const char *target_arch)
+{
     if (!target_arch) {
 #if defined(BUILD_TARGET_X86_64) || defined(BUILD_TARGET_AMD_64) \
     || defined(BUILD_TARGET_X86_32)
-        is_target_x86 = true;
+        return true;
+#else
+        return false;
 #endif
     }
-    else {
-        if (!strncmp(target_arch, "x86_64", 6)
-            || !strncmp(target_arch, "i386", 4))
-            is_target_x86 = true;
-    }
-#endif
+    return !strncmp(target_arch, "x86_64", 6)
+           || !strncmp(target_arch, "i386", 4);
+}
 
-    if (!target_arch) {
-#if UINTPTR_MAX == UINT64_MAX
-        is_64bit_target = true;
-#endif
-    }
-    else {
-        /* All 64bit targets contains "64" string in their target name */
-        if (strstr(target_arch, "64") != NULL) {
-            is_64bit_target = true;
-        }
-    }
-
-    /* Allocate memory */
-    if (!(comp_data = wasm_runtime_malloc(sizeof(AOTCompData)))) {
-        aot_set_last_error("create compile data failed.\n");
-        return NULL;
-    }
-
-    memset(comp_data, 0, sizeof(AOTCompData));
+/**
+ * Initialize memory information in AOT compilation data from WASM module.
+ *
+ * @param comp_data the AOT compilation data structure to initialize
+ * @param module the source WASM module containing memory information
+ * @return true if initialization succeeded, false otherwise
+ */
+static bool
+aot_init_memories(AOTCompData *comp_data, WASMModule *module)
+{
+    uint32 i, j;
+    uint64 size;
 
     comp_data->memory_count =
         module->import_memory_count + module->memory_count;
 
-    /* TODO: create import memories */
-
     /* Allocate memory for memory array, reserve one AOTMemory space at least */
     if (!comp_data->memory_count)
         comp_data->memory_count = 1;
@@ -547,7 +573,7 @@ aot_create_comp_data(WASMModule *module, const char *target_arch,
     if (size >= UINT32_MAX
         || !(comp_data->memories = wasm_runtime_malloc((uint32)size))) {
         aot_set_last_error("create memories array failed.\n");
-        goto fail;
+        return false;
     }
     memset(comp_data->memories, 0, size);
 
@@ -579,22 +605,30 @@ aot_create_comp_data(WASMModule *module, const char *target_arch,
         }
     }
 
-    /* Create memory data segments */
-    comp_data->mem_init_data_count = module->data_seg_count;
-    if (comp_data->mem_init_data_count > 0
-        && !(comp_data->mem_init_data_list =
-                 aot_create_mem_init_data_list(module)))
-        goto fail;
+    return true;
+}
+
+/**
+ * Initialize table information in AOT compilation data from WASM module.
+ *
+ * @param comp_data the AOT compilation data structure to initialize
+ * @param module the source WASM module containing table information
+ * @return true if initialization succeeded, false otherwise
+ */
+static bool
+aot_init_tables(AOTCompData *comp_data, WASMModule *module)
+{
+    uint32 i, j;
+    uint64 size;
 
-    /* Create tables */
     comp_data->table_count = module->import_table_count + module->table_count;
 
     if (comp_data->table_count > 0) {
         size = sizeof(AOTTable) * (uint64)comp_data->table_count;
         if (size >= UINT32_MAX
             || !(comp_data->tables = wasm_runtime_malloc((uint32)size))) {
-            aot_set_last_error("create memories array failed.\n");
-            goto fail;
+            aot_set_last_error("create tables array failed.\n");
+            return false;
         }
         memset(comp_data->tables, 0, size);
         for (i = 0; i < comp_data->table_count; i++) {
@@ -640,64 +674,150 @@ aot_create_comp_data(WASMModule *module, const char *target_arch,
         }
     }
 
-    /* Create table data segments */
+    return true;
+}
+
+/**
+ * Initialize memory segment information in AOT compilation data.
+ *
+ * @param comp_data the AOT compilation data structure to initialize
+ * @param module the source WASM module containing memory segments
+ * @return true if initialization succeeded, false otherwise
+ */
+static bool
+aot_init_memory_segments(AOTCompData *comp_data, WASMModule *module)
+{
+    comp_data->mem_init_data_count = module->data_seg_count;
+    if (comp_data->mem_init_data_count > 0
+        && !(comp_data->mem_init_data_list =
+                 aot_create_mem_init_data_list(module))) {
+        return false;
+    }
+    return true;
+}
+
+/**
+ * Initialize table segment information in AOT compilation data.
+ *
+ * @param comp_data the AOT compilation data structure to initialize
+ * @param module the source WASM module containing table segments
+ * @return true if initialization succeeded, false otherwise
+ */
+static bool
+aot_init_table_segments(AOTCompData *comp_data, WASMModule *module)
+{
     comp_data->table_init_data_count = module->table_seg_count;
     if (comp_data->table_init_data_count > 0
         && !(comp_data->table_init_data_list =
-                 aot_create_table_init_data_list(module)))
-        goto fail;
+                 aot_create_table_init_data_list(module))) {
+        return false;
+    }
+    return true;
+}
 
-    /* Create import globals */
+/**
+ * Initialize global variable information in AOT compilation data.
+ *
+ * @param comp_data the AOT compilation data structure to initialize
+ * @param module the source WASM module containing global information
+ * @param gc_enabled whether garbage collection is enabled
+ * @param import_global_data_size_64bit [out] size of imported global data for
+ * 64-bit
+ * @param import_global_data_size_32bit [out] size of imported global data for
+ * 32-bit
+ * @param global_data_size_64bit [out] size of global data for 64-bit
+ * @param global_data_size_32bit [out] size of global data for 32-bit
+ * @return true if initialization succeeded, false otherwise
+ */
+static bool
+aot_init_globals(AOTCompData *comp_data, WASMModule *module, bool gc_enabled,
+                 uint32 *import_global_data_size_64bit,
+                 uint32 *import_global_data_size_32bit,
+                 uint32 *global_data_size_64bit, uint32 *global_data_size_32bit)
+{
     comp_data->import_global_count = module->import_global_count;
     if (comp_data->import_global_count > 0
         && !(comp_data->import_globals = aot_create_import_globals(
-                 module, gc_enabled, &import_global_data_size_64bit,
-                 &import_global_data_size_32bit)))
-        goto fail;
+                 module, gc_enabled, import_global_data_size_64bit,
+                 import_global_data_size_32bit))) {
+        return false;
+    }
 
-    /* Create globals */
     comp_data->global_count = module->global_count;
     if (comp_data->global_count
         && !(comp_data->globals = aot_create_globals(
-                 module, gc_enabled, import_global_data_size_64bit,
-                 import_global_data_size_32bit, &global_data_size_64bit,
-                 &global_data_size_32bit)))
-        goto fail;
+                 module, gc_enabled, *import_global_data_size_64bit,
+                 *import_global_data_size_32bit, global_data_size_64bit,
+                 global_data_size_32bit))) {
+        return false;
+    }
 
     comp_data->global_data_size_64bit =
-        import_global_data_size_64bit + global_data_size_64bit;
+        *import_global_data_size_64bit + *global_data_size_64bit;
     comp_data->global_data_size_32bit =
-        import_global_data_size_32bit + global_data_size_32bit;
+        *import_global_data_size_32bit + *global_data_size_32bit;
 
-    /* Create types, they are checked by wasm loader */
+    return true;
+}
+
+/**
+ * Initialize type information in AOT compilation data.
+ *
+ * @param comp_data the AOT compilation data structure to initialize
+ * @param module the source WASM module containing type information
+ * @param is_target_x86 whether the target architecture is x86
+ * @param gc_enabled whether garbage collection is enabled
+ * @return true if initialization succeeded, false otherwise
+ */
+static bool
+aot_init_types(AOTCompData *comp_data, WASMModule *module, bool is_target_x86,
+               bool gc_enabled)
+{
     comp_data->type_count = module->type_count;
     comp_data->types = module->types;
 #if WASM_ENABLE_GC != 0
-    /* Calculate the field sizes and field offsets for 64-bit and 32-bit
-       targets since they may vary in 32-bit target and 64-bit target */
     calculate_struct_field_sizes_offsets(comp_data, is_target_x86, gc_enabled);
 #endif
+    return true;
+}
 
-    /* Create import functions */
+/**
+ * Initialize function information in AOT compilation data.
+ *
+ * @param comp_data the AOT compilation data structure to initialize
+ * @param module the source WASM module containing function information
+ * @param is_64bit_target whether the target architecture is 64-bit
+ * @return true if initialization succeeded, false otherwise
+ */
+static bool
+aot_init_functions(AOTCompData *comp_data, WASMModule *module,
+                   bool is_64bit_target)
+{
     comp_data->import_func_count = module->import_function_count;
     if (comp_data->import_func_count
-        && !(comp_data->import_funcs = aot_create_import_funcs(module)))
-        goto fail;
+        && !(comp_data->import_funcs = aot_create_import_funcs(module))) {
+        return false;
+    }
 
-    /* Create functions */
     comp_data->func_count = module->function_count;
     if (comp_data->func_count
         && !(comp_data->funcs =
-                 aot_create_funcs(module, is_64bit_target ? 8 : 4)))
-        goto fail;
+                 aot_create_funcs(module, is_64bit_target ? 8 : 4))) {
+        return false;
+    }
 
-#if WASM_ENABLE_CUSTOM_NAME_SECTION != 0
-    /* Create custom name section */
-    comp_data->name_section_buf = module->name_section_buf;
-    comp_data->name_section_buf_end = module->name_section_buf_end;
-#endif
+    return true;
+}
 
-    /* Create aux data/heap/stack information */
+/**
+ * Initialize auxiliary data in AOT compilation data.
+ *
+ * @param comp_data the AOT compilation data structure to initialize
+ * @param module the source WASM module containing auxiliary data
+ */
+static void
+aot_init_aux_data(AOTCompData *comp_data, WASMModule *module)
+{
     comp_data->aux_data_end_global_index = module->aux_data_end_global_index;
     comp_data->aux_data_end = module->aux_data_end;
     comp_data->aux_heap_base_global_index = module->aux_heap_base_global_index;
@@ -716,6 +836,43 @@ aot_create_comp_data(WASMModule *module, const char *target_arch,
     comp_data->string_literal_ptrs_wp = module->string_literal_ptrs;
     comp_data->string_literal_lengths_wp = module->string_literal_lengths;
 #endif
+}
+
+AOTCompData *
+aot_create_comp_data(WASMModule *module, const char *target_arch,
+                     bool gc_enabled)
+{
+    AOTCompData *comp_data;
+    uint32 import_global_data_size_64bit = 0, global_data_size_64bit = 0;
+    uint32 import_global_data_size_32bit = 0, global_data_size_32bit = 0;
+    bool is_64bit_target = arch_is_64bit(target_arch);
+    bool is_target_x86 = arch_is_x86(target_arch);
+
+    if (!(comp_data = wasm_runtime_malloc(sizeof(AOTCompData)))) {
+        aot_set_last_error("create compile data failed.\n");
+        return NULL;
+    }
+    memset(comp_data, 0, sizeof(AOTCompData));
+
+    if (!aot_init_memories(comp_data, module)
+        || !aot_init_memory_segments(comp_data, module)
+        || !aot_init_tables(comp_data, module)
+        || !aot_init_table_segments(comp_data, module)
+        || !aot_init_globals(comp_data, module, gc_enabled,
+                             &import_global_data_size_64bit,
+                             &import_global_data_size_32bit,
+                             &global_data_size_64bit, &global_data_size_32bit)
+        || !aot_init_types(comp_data, module, is_target_x86, gc_enabled)
+        || !aot_init_functions(comp_data, module, is_64bit_target)) {
+        goto fail;
+    }
+
+#if WASM_ENABLE_CUSTOM_NAME_SECTION != 0
+    comp_data->name_section_buf = module->name_section_buf;
+    comp_data->name_section_buf_end = module->name_section_buf_end;
+#endif
+
+    aot_init_aux_data(comp_data, module);
 
     comp_data->wasm_module = module;
 

core/iwasm/compilation/aot.h

@@ -103,7 +103,7 @@ typedef struct AOTMemInitData {
     /* Byte count */
     uint32 byte_count;
     /* Byte array */
-    uint8 bytes[1];
+    uint8 *bytes;
 } AOTMemInitData;
 
 /**
@@ -312,7 +312,7 @@ void
 aot_destroy_comp_data(AOTCompData *comp_data);
 
 char *
-aot_get_last_error();
+aot_get_last_error(void);
 
 void
 aot_set_last_error(const char *error);

core/iwasm/compilation/aot_compiler.c

@@ -16,6 +16,7 @@
 #include "aot_emit_parametric.h"
 #include "aot_emit_table.h"
 #include "aot_emit_gc.h"
+#include "aot_stack_frame_comp.h"
 #include "simd/simd_access_lanes.h"
 #include "simd/simd_bitmask_extracts.h"
 #include "simd/simd_bit_shifts.h"
@@ -145,9 +146,20 @@ aot_validate_wasm(AOTCompContext *comp_ctx)
     }
 
 #if WASM_ENABLE_MEMORY64 != 0
-    if (comp_ctx->pointer_size < sizeof(uint64) && IS_MEMORY64) {
-        aot_set_last_error("Compiling wasm64 to 32bit platform is not allowed");
-        return false;
+    if (comp_ctx->pointer_size < sizeof(uint64)) {
+        if (IS_MEMORY64) {
+            aot_set_last_error("Compiling wasm64(contains i64 memory section) "
+                               "to 32bit platform is not allowed");
+            return false;
+        }
+
+        for (uint32 i = 0; i < comp_ctx->comp_data->table_count; ++i) {
+            if (IS_TABLE64(i)) {
+                aot_set_last_error("Compiling wasm64(contains i64 table "
+                                   "section) to 32bit platform is not allowed");
+                return false;
+            }
+        }
     }
 #endif
 
@@ -253,6 +265,13 @@ store_value(AOTCompContext *comp_ctx, LLVMValueRef value, uint8 value_type,
     return true;
 }
 
+void
+aot_call_stack_features_init_default(AOTCallStackFeatures *features)
+{
+    memset(features, 1, sizeof(AOTCallStackFeatures));
+    features->frame_per_function = false;
+}
+
 bool
 aot_frame_store_value(AOTCompContext *comp_ctx, LLVMValueRef value,
                       uint8 value_type, LLVMValueRef cur_frame, uint32 offset)
@@ -337,6 +356,10 @@ aot_gen_commit_values(AOTCompFrame *frame)
     LLVMValueRef value;
     uint32 n;
 
+    if (!frame->comp_ctx->call_stack_features.values) {
+        return true;
+    }
+
     /* First, commit reference flags
      * For LLVM JIT, iterate all local and stack ref flags
      * For AOT, ignore local(params + locals) ref flags */
@@ -569,6 +592,64 @@ aot_gen_commit_values(AOTCompFrame *frame)
     return true;
 }
 
+static bool
+aot_standard_frame_gen_commit_ip(AOTCompContext *comp_ctx,
+                                 AOTFuncContext *func_ctx,
+                                 LLVMValueRef ip_value, bool is_64bit)
+{
+    LLVMValueRef cur_frame = func_ctx->cur_frame;
+    LLVMValueRef value_offset, value_addr, value_ptr;
+    uint32 offset_ip;
+
+    if (!comp_ctx->is_jit_mode)
+        offset_ip = comp_ctx->pointer_size * 4;
+    else
+        offset_ip = offsetof(WASMInterpFrame, ip);
+
+    if (!(value_offset = I32_CONST(offset_ip))) {
+        aot_set_last_error("llvm build const failed");
+        return false;
+    }
+
+    if (!(value_addr =
+              LLVMBuildInBoundsGEP2(comp_ctx->builder, INT8_TYPE, cur_frame,
+                                    &value_offset, 1, "ip_addr"))) {
+        aot_set_last_error("llvm build in bounds gep failed");
+        return false;
+    }
+
+    if (!(value_ptr = LLVMBuildBitCast(
+              comp_ctx->builder, value_addr,
+              is_64bit ? INT64_PTR_TYPE : INT32_PTR_TYPE, "ip_ptr"))) {
+        aot_set_last_error("llvm build bit cast failed");
+        return false;
+    }
+
+    if (!LLVMBuildStore(comp_ctx->builder, ip_value, value_ptr)) {
+        aot_set_last_error("llvm build store failed");
+        return false;
+    }
+
+    return true;
+}
+
+bool
+aot_gen_commit_ip(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
+                  LLVMValueRef ip_value, bool is_64bit)
+{
+    switch (comp_ctx->aux_stack_frame_type) {
+        case AOT_STACK_FRAME_TYPE_STANDARD:
+            return aot_standard_frame_gen_commit_ip(comp_ctx, func_ctx,
+                                                    ip_value, is_64bit);
+        case AOT_STACK_FRAME_TYPE_TINY:
+            return aot_tiny_frame_gen_commit_ip(comp_ctx, func_ctx, ip_value);
+        default:
+            aot_set_last_error(
+                "unsupported mode when generating commit_ip code");
+            return false;
+    }
+}
+
 bool
 aot_gen_commit_sp_ip(AOTCompFrame *frame, bool commit_sp, bool commit_ip)
 {
@@ -577,40 +658,19 @@ aot_gen_commit_sp_ip(AOTCompFrame *frame, bool commit_sp, bool commit_ip)
     LLVMValueRef cur_frame = func_ctx->cur_frame;
     LLVMValueRef value_offset, value_addr, value_ptr, value;
     LLVMTypeRef int8_ptr_ptr_type;
-    uint32 offset_ip, offset_sp, n;
+    uint32 offset_sp, n;
     bool is_64bit = (comp_ctx->pointer_size == sizeof(uint64)) ? true : false;
     const AOTValueSlot *sp = frame->sp;
     const uint8 *ip = frame->frame_ip;
 
     if (!comp_ctx->is_jit_mode) {
-        offset_ip = frame->comp_ctx->pointer_size * 4;
         offset_sp = frame->comp_ctx->pointer_size * 5;
     }
     else {
-        offset_ip = offsetof(WASMInterpFrame, ip);
         offset_sp = offsetof(WASMInterpFrame, sp);
     }
 
-    if (commit_ip) {
-        if (!(value_offset = I32_CONST(offset_ip))) {
-            aot_set_last_error("llvm build const failed");
-            return false;
-        }
-
-        if (!(value_addr =
-                  LLVMBuildInBoundsGEP2(comp_ctx->builder, INT8_TYPE, cur_frame,
-                                        &value_offset, 1, "ip_addr"))) {
-            aot_set_last_error("llvm build in bounds gep failed");
-            return false;
-        }
-
-        if (!(value_ptr = LLVMBuildBitCast(
-                  comp_ctx->builder, value_addr,
-                  is_64bit ? INT64_PTR_TYPE : INT32_PTR_TYPE, "ip_ptr"))) {
-            aot_set_last_error("llvm build bit cast failed");
-            return false;
-        }
-
+    if (commit_ip && comp_ctx->call_stack_features.ip) {
         if (!comp_ctx->is_jit_mode) {
             WASMModule *module = comp_ctx->comp_data->wasm_module;
             if (is_64bit)
@@ -630,13 +690,12 @@ aot_gen_commit_sp_ip(AOTCompFrame *frame, bool commit_sp, bool commit_ip)
             return false;
         }
 
-        if (!LLVMBuildStore(comp_ctx->builder, value, value_ptr)) {
-            aot_set_last_error("llvm build store failed");
+        if (!aot_gen_commit_ip(comp_ctx, func_ctx, value, is_64bit)) {
             return false;
         }
     }
 
-    if (commit_sp) {
+    if (commit_sp && comp_ctx->call_stack_features.values) {
         n = (uint32)(sp - frame->lp);
         value = I32_CONST(offset_of_local(comp_ctx, n));
         if (!value) {
@@ -940,6 +999,7 @@ static bool
 aot_compile_func(AOTCompContext *comp_ctx, uint32 func_index)
 {
     AOTFuncContext *func_ctx = comp_ctx->func_ctxes[func_index];
+    LLVMValueRef func_index_ref;
     uint8 *frame_ip = func_ctx->aot_func->code, opcode, *p_f32, *p_f64;
     uint8 *frame_ip_end = frame_ip + func_ctx->aot_func->code_size;
     uint8 *param_types = NULL;
@@ -962,16 +1022,27 @@ aot_compile_func(AOTCompContext *comp_ctx, uint32 func_index)
     LLVMMetadataRef location;
 #endif
 
-    if (comp_ctx->enable_aux_stack_frame) {
+    /* Start to translate the opcodes */
+    LLVMPositionBuilderAtEnd(
+        comp_ctx->builder,
+        func_ctx->block_stack.block_list_head->llvm_entry_block);
+
+    if (comp_ctx->aux_stack_frame_type
+        && comp_ctx->call_stack_features.frame_per_function) {
+        INT_CONST(func_index_ref,
+                  func_index + comp_ctx->comp_data->import_func_count, I32_TYPE,
+                  true);
+        if (!aot_alloc_frame_per_function_frame_for_aot_func(comp_ctx, func_ctx,
+                                                             func_index_ref)) {
+            return false;
+        }
+    }
+    if (comp_ctx->aux_stack_frame_type) {
         if (!init_comp_frame(comp_ctx, func_ctx, func_index)) {
             return false;
         }
     }
 
-    /* Start to translate the opcodes */
-    LLVMPositionBuilderAtEnd(
-        comp_ctx->builder,
-        func_ctx->block_stack.block_list_head->llvm_entry_block);
     while (frame_ip < frame_ip_end) {
         opcode = *frame_ip++;
 
@@ -4022,39 +4093,6 @@ aot_compile_wasm(AOTCompContext *comp_ctx)
     return true;
 }
 
-#if !(defined(_WIN32) || defined(_WIN32_))
-char *
-aot_generate_tempfile_name(const char *prefix, const char *extension,
-                           char *buffer, uint32 len)
-{
-    int fd, name_len;
-
-    name_len = snprintf(buffer, len, "%s-XXXXXX", prefix);
-
-    if ((fd = mkstemp(buffer)) <= 0) {
-        aot_set_last_error("make temp file failed.");
-        return NULL;
-    }
-
-    /* close and remove temp file */
-    close(fd);
-    unlink(buffer);
-
-    /* Check if buffer length is enough */
-    /* name_len + '.' + extension + '\0' */
-    if (name_len + 1 + strlen(extension) + 1 > len) {
-        aot_set_last_error("temp file name too long.");
-        return NULL;
-    }
-
-    snprintf(buffer + name_len, len - name_len, ".%s", extension);
-    return buffer;
-}
-#else
-
-errno_t
-_mktemp_s(char *nameTemplate, size_t sizeInChars);
-
 char *
 aot_generate_tempfile_name(const char *prefix, const char *extension,
                            char *buffer, uint32 len)
@@ -4063,7 +4101,8 @@ aot_generate_tempfile_name(const char *prefix, const char *extension,
 
     name_len = snprintf(buffer, len, "%s-XXXXXX", prefix);
 
-    if (_mktemp_s(buffer, name_len + 1) != 0) {
+    if (!bh_mkstemp(buffer, name_len + 1)) {
+        aot_set_last_error("make temp file failed.");
         return NULL;
     }
 
@@ -4077,7 +4116,6 @@ aot_generate_tempfile_name(const char *prefix, const char *extension,
     snprintf(buffer + name_len, len - name_len, ".%s", extension);
     return buffer;
 }
-#endif /* end of !(defined(_WIN32) || defined(_WIN32_)) */
 
 bool
 aot_emit_llvm_file(AOTCompContext *comp_ctx, const char *file_name)
@@ -4156,7 +4194,6 @@ aot_emit_object_file(AOTCompContext *comp_ctx, char *file_name)
 
     bh_print_time("Begin to emit object file");
 
-#if !(defined(_WIN32) || defined(_WIN32_))
     if (comp_ctx->external_llc_compiler || comp_ctx->external_asm_compiler) {
         char cmd[1024];
         int ret;
@@ -4199,7 +4236,7 @@ aot_emit_object_file(AOTCompContext *comp_ctx, char *file_name)
                      file_name, bc_file_name);
             LOG_VERBOSE("invoking external LLC compiler:\n\t%s", cmd);
 
-            ret = system(cmd);
+            ret = bh_system(cmd);
             /* remove temp bitcode file */
             unlink(bc_file_name);
 
@@ -4252,7 +4289,7 @@ aot_emit_object_file(AOTCompContext *comp_ctx, char *file_name)
                      file_name, asm_file_name);
             LOG_VERBOSE("invoking external ASM compiler:\n\t%s", cmd);
 
-            ret = system(cmd);
+            ret = bh_system(cmd);
             /* remove temp assembly file */
             unlink(asm_file_name);
 
@@ -4265,7 +4302,6 @@ aot_emit_object_file(AOTCompContext *comp_ctx, char *file_name)
 
         return true;
     }
-#endif /* end of !(defined(_WIN32) || defined(_WIN32_)) */
 
     if (!strncmp(LLVMGetTargetName(target), "arc", 3))
         /* Emit to assembly file instead for arc target

core/iwasm/compilation/aot_compiler.h

@@ -195,6 +195,15 @@ aot_gen_commit_values(AOTCompFrame *frame);
 bool
 aot_gen_commit_sp_ip(AOTCompFrame *frame, bool commit_sp, bool commit_ip);
 
+/**
+ * Generate instructions to commit IP pointer to the frame.
+ *
+ * @param frame the frame information
+ */
+bool
+aot_gen_commit_ip(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
+                  LLVMValueRef ip_value, bool is_64bit);
+
 bool
 aot_frame_store_value(AOTCompContext *comp_ctx, LLVMValueRef value,
                       uint8 value_type, LLVMValueRef cur_frame, uint32 offset);
@@ -523,8 +532,13 @@ set_local_gc_ref(AOTCompFrame *frame, int n, LLVMValueRef value, uint8 ref_type)
 #define IS_MEMORY64 (comp_ctx->comp_data->memories[0].flags & MEMORY64_FLAG)
 #define MEMORY64_COND_VALUE(VAL_IF_ENABLED, VAL_IF_DISABLED) \
     (IS_MEMORY64 ? VAL_IF_ENABLED : VAL_IF_DISABLED)
+#define IS_TABLE64(i) \
+    (comp_ctx->comp_data->tables[i].table_type.flags & TABLE64_FLAG)
+#define TABLE64_COND_VALUE(i, VAL_IF_ENABLED, VAL_IF_DISABLED) \
+    (IS_TABLE64(i) ? VAL_IF_ENABLED : VAL_IF_DISABLED)
 #else
 #define MEMORY64_COND_VALUE(VAL_IF_ENABLED, VAL_IF_DISABLED) (VAL_IF_DISABLED)
+#define TABLE64_COND_VALUE(i, VAL_IF_ENABLED, VAL_IF_DISABLED) (VAL_IF_DISABLED)
 #endif
 
 #define POP_I32(v) POP(v, VALUE_TYPE_I32)
@@ -539,6 +553,9 @@ set_local_gc_ref(AOTCompFrame *frame, int n, LLVMValueRef value, uint8 ref_type)
     POP(v, MEMORY64_COND_VALUE(VALUE_TYPE_I64, VALUE_TYPE_I32))
 #define POP_PAGE_COUNT(v) \
     POP(v, MEMORY64_COND_VALUE(VALUE_TYPE_I64, VALUE_TYPE_I32))
+#define POP_TBL_ELEM_IDX(v) \
+    POP(v, TABLE64_COND_VALUE(tbl_idx, VALUE_TYPE_I64, VALUE_TYPE_I32))
+#define POP_TBL_ELEM_LEN(v) POP_TBL_ELEM_IDX(v)
 
 #define POP_COND(llvm_value)                                                   \
     do {                                                                       \
@@ -604,6 +621,9 @@ set_local_gc_ref(AOTCompFrame *frame, int n, LLVMValueRef value, uint8 ref_type)
 #define PUSH_GC_REF(v) PUSH(v, VALUE_TYPE_GC_REF)
 #define PUSH_PAGE_COUNT(v) \
     PUSH(v, MEMORY64_COND_VALUE(VALUE_TYPE_I64, VALUE_TYPE_I32))
+#define PUSH_TBL_ELEM_IDX(v) \
+    PUSH(v, TABLE64_COND_VALUE(tbl_idx, VALUE_TYPE_I64, VALUE_TYPE_I32))
+#define PUSH_TBL_ELEM_LEN(v) PUSH_TBL_ELEM_IDX(v)
 
 #define SET_CONST(v)                                                          \
     do {                                                                      \
@@ -652,6 +672,15 @@ set_local_gc_ref(AOTCompFrame *frame, int n, LLVMValueRef value, uint8 ref_type)
 #define F64_CONST(v) LLVMConstReal(F64_TYPE, v)
 #define I8_CONST(v) LLVMConstInt(INT8_TYPE, v, true)
 
+#define INT_CONST(variable, value, type, is_signed)        \
+    do {                                                   \
+        variable = LLVMConstInt(type, value, is_signed);   \
+        if (!variable) {                                   \
+            aot_set_last_error("llvm build const failed"); \
+            return false;                                  \
+        }                                                  \
+    } while (0)
+
 #define LLVM_CONST(name) (comp_ctx->llvm_consts.name)
 #define I1_ZERO LLVM_CONST(i1_zero)
 #define I1_ONE LLVM_CONST(i1_one)
@@ -761,7 +790,7 @@ set_local_gc_ref(AOTCompFrame *frame, int n, LLVMValueRef value, uint8 ref_type)
         }                                                                   \
         else {                                                              \
             char *func_name = #name;                                        \
-            /* AOT mode, delcare the function */                            \
+            /* AOT mode, declare the function */                            \
             if (!(func = LLVMGetNamedFunction(func_ctx->module, func_name)) \
                 && !(func = LLVMAddFunction(func_ctx->module, func_name,    \
                                             func_type))) {                  \

core/iwasm/compilation/aot_emit_aot_file.c

@@ -920,9 +920,11 @@ get_relocations_size(AOTObjectData *obj_data,
         /* ignore the relocations to aot_func_internal#n in text section
            for windows platform since they will be applied in
            aot_emit_text_section */
+
+        const char *name = relocation->symbol_name;
         if ((!strcmp(relocation_group->section_name, ".text")
              || !strcmp(relocation_group->section_name, ".ltext"))
-            && !strncmp(relocation->symbol_name, AOT_FUNC_INTERNAL_PREFIX,
+            && !strncmp(name, AOT_FUNC_INTERNAL_PREFIX,
                         strlen(AOT_FUNC_INTERNAL_PREFIX))
             && ((!strncmp(obj_data->comp_ctx->target_arch, "x86_64", 6)
                  /* Windows AOT_COFF64_BIN_TYPE */
@@ -1790,7 +1792,9 @@ aot_emit_mem_info(uint8 *buf, uint8 *buf_end, uint32 *p_offset,
                                 &init_datas[i]->offset))
             return false;
         EMIT_U32(init_datas[i]->byte_count);
-        EMIT_BUF(init_datas[i]->bytes, init_datas[i]->byte_count);
+        if (init_datas[i]->byte_count) {
+            EMIT_BUF(init_datas[i]->bytes, init_datas[i]->byte_count);
+        }
     }
 
     if (offset - *p_offset != get_mem_info_size(comp_ctx, comp_data)) {
@@ -2487,8 +2491,8 @@ aot_emit_text_section(uint8 *buf, uint8 *buf_end, uint32 *p_offset,
                 relocation_count = relocation_group->relocation_count;
                 for (j = 0; j < relocation_count; j++) {
                     /* relocation to aot_func_internal#n */
-                    if (str_starts_with(relocation->symbol_name,
-                                        AOT_FUNC_INTERNAL_PREFIX)
+                    const char *name = relocation->symbol_name;
+                    if (str_starts_with(name, AOT_FUNC_INTERNAL_PREFIX)
                         && ((obj_data->target_info.bin_type
                                  == 6 /* AOT_COFF64_BIN_TYPE */
                              && relocation->relocation_type
@@ -2498,8 +2502,7 @@ aot_emit_text_section(uint8 *buf, uint8 *buf_end, uint32 *p_offset,
                                 && relocation->relocation_type
                                        == 20 /* IMAGE_REL_I386_REL32 */))) {
                         uint32 func_idx =
-                            atoi(relocation->symbol_name
-                                 + strlen(AOT_FUNC_INTERNAL_PREFIX));
+                            atoi(name + strlen(AOT_FUNC_INTERNAL_PREFIX));
                         uint64 text_offset, reloc_offset, reloc_addend;
 
                         bh_assert(func_idx < obj_data->func_count);
@@ -3050,6 +3053,27 @@ typedef struct elf64_rela {
 #define SET_TARGET_INFO_FIELD(f, v, type, little) \
     SET_TARGET_INFO_VALUE(f, elf_header->v, type, little)
 
+/* in windows 32, the symbol name may start with '_' */
+static char *
+LLVMGetSymbolNameAndUnDecorate(LLVMSymbolIteratorRef si,
+                               AOTTargetInfo target_info)
+{
+    char *original_name = (char *)LLVMGetSymbolName(si);
+    if (!original_name) {
+        return NULL;
+    }
+
+    if (target_info.bin_type != AOT_COFF32_BIN_TYPE) {
+        return original_name;
+    }
+
+    if (*original_name == '_') {
+        return ++original_name;
+    }
+
+    return original_name;
+}
+
 static bool
 aot_resolve_target_info(AOTCompContext *comp_ctx, AOTObjectData *obj_data)
 {
@@ -3246,8 +3270,17 @@ is_data_section(AOTObjectData *obj_data, LLVMSectionIteratorRef sec_itr,
 
     return (!strcmp(section_name, ".data") || !strcmp(section_name, ".sdata")
             || !strcmp(section_name, ".rodata")
+#if LLVM_VERSION_MAJOR >= 19
+            /* https://github.com/llvm/llvm-project/pull/82214 */
+            || !strcmp(section_name, ".srodata")
+#endif
             /* ".rodata.cst4/8/16/.." */
             || !strncmp(section_name, ".rodata.cst", strlen(".rodata.cst"))
+#if LLVM_VERSION_MAJOR >= 19
+            /* https://github.com/llvm/llvm-project/pull/82214
+             * ".srodata.cst4/8/16/.." */
+            || !strncmp(section_name, ".srodata.cst", strlen(".srodata.cst"))
+#endif
             /* ".rodata.strn.m" */
             || !strncmp(section_name, ".rodata.str", strlen(".rodata.str"))
             || (!strcmp(section_name, ".rdata")
@@ -3524,12 +3557,9 @@ aot_resolve_stack_sizes(AOTCompContext *comp_ctx, AOTObjectData *obj_data)
     }
 
     while (!LLVMObjectFileIsSymbolIteratorAtEnd(obj_data->binary, sym_itr)) {
-        if ((name = LLVMGetSymbolName(sym_itr))
-            && (!strcmp(name, aot_stack_sizes_alias_name)
-                /* symbol of COFF32 starts with "_" */
-                || (obj_data->target_info.bin_type == AOT_COFF32_BIN_TYPE
-                    && !strncmp(name, "_", 1)
-                    && !strcmp(name + 1, aot_stack_sizes_alias_name)))) {
+        if ((name =
+                 LLVMGetSymbolNameAndUnDecorate(sym_itr, obj_data->target_info))
+            && (!strcmp(name, aot_stack_sizes_alias_name))) {
 #if 0 /* cf. https://github.com/llvm/llvm-project/issues/67765 */
             uint64 sz = LLVMGetSymbolSize(sym_itr);
             if (sz != sizeof(uint32) * obj_data->func_count
@@ -3693,8 +3723,8 @@ aot_resolve_functions(AOTCompContext *comp_ctx, AOTObjectData *obj_data)
     }
 
     while (!LLVMObjectFileIsSymbolIteratorAtEnd(obj_data->binary, sym_itr)) {
-        if ((name = (char *)LLVMGetSymbolName(sym_itr))
-            && str_starts_with(name, prefix)) {
+        name = LLVMGetSymbolNameAndUnDecorate(sym_itr, obj_data->target_info);
+        if (name && str_starts_with(name, prefix)) {
             /* symbol aot_func#n */
             func_index = (uint32)atoi(name + strlen(prefix));
             if (func_index < obj_data->func_count) {
@@ -3732,8 +3762,7 @@ aot_resolve_functions(AOTCompContext *comp_ctx, AOTObjectData *obj_data)
                 }
             }
         }
-        else if ((name = (char *)LLVMGetSymbolName(sym_itr))
-                 && str_starts_with(name, AOT_FUNC_INTERNAL_PREFIX)) {
+        else if (name && str_starts_with(name, AOT_FUNC_INTERNAL_PREFIX)) {
             /* symbol aot_func_internal#n */
             func_index = (uint32)atoi(name + strlen(AOT_FUNC_INTERNAL_PREFIX));
             if (func_index < obj_data->func_count) {
@@ -3844,7 +3873,7 @@ aot_resolve_object_relocation_group(AOTObjectData *obj_data,
         }
     }
 
-    /* pares each relocation */
+    /* parse each relocation */
     if (!(rel_itr = LLVMGetRelocations(rel_sec))) {
         aot_set_last_error("llvm get relocations failed.");
         return false;
@@ -3881,7 +3910,8 @@ aot_resolve_object_relocation_group(AOTObjectData *obj_data,
 
         /* set relocation fields */
         relocation->relocation_type = (uint32)type;
-        relocation->symbol_name = (char *)LLVMGetSymbolName(rel_sym);
+        relocation->symbol_name =
+            LLVMGetSymbolNameAndUnDecorate(rel_sym, obj_data->target_info);
         relocation->relocation_offset = offset;
         if (!strcmp(group->section_name, ".rela.text.unlikely.")
             || !strcmp(group->section_name, ".rel.text.unlikely.")) {
@@ -3908,12 +3938,7 @@ aot_resolve_object_relocation_group(AOTObjectData *obj_data,
          * Note: aot_stack_sizes_section_name section only contains
          * stack_sizes table.
          */
-        if (!strcmp(relocation->symbol_name, aot_stack_sizes_name)
-            /* in windows 32, the symbol name may start with '_' */
-            || (strlen(relocation->symbol_name) > 0
-                && relocation->symbol_name[0] == '_'
-                && !strcmp(relocation->symbol_name + 1,
-                           aot_stack_sizes_name))) {
+        if (!strcmp(relocation->symbol_name, aot_stack_sizes_name)) {
             /* discard const */
             relocation->symbol_name = (char *)aot_stack_sizes_section_name;
         }
@@ -3982,8 +4007,12 @@ aot_resolve_object_relocation_group(AOTObjectData *obj_data,
             && (str_starts_with(relocation->symbol_name, ".LCPI")
                 || str_starts_with(relocation->symbol_name, ".LJTI")
                 || str_starts_with(relocation->symbol_name, ".LBB")
-                || str_starts_with(relocation->symbol_name,
-                                   ".Lswitch.table."))) {
+                || str_starts_with(relocation->symbol_name, ".Lswitch.table.")
+#if LLVM_VERSION_MAJOR >= 16
+                /* cf. https://reviews.llvm.org/D123264 */
+                || str_starts_with(relocation->symbol_name, ".Lpcrel_hi")
+#endif
+                    )) {
             /* change relocation->relocation_addend and
                relocation->symbol_name */
             LLVMSectionIteratorRef contain_section;
@@ -4292,10 +4321,6 @@ aot_obj_data_create(AOTCompContext *comp_ctx)
 
     bh_print_time("Begin to emit object file");
     if (comp_ctx->external_llc_compiler || comp_ctx->external_asm_compiler) {
-#if defined(_WIN32) || defined(_WIN32_)
-        aot_set_last_error("external toolchain not supported on Windows");
-        goto fail;
-#else
         /* Generate a temp file name */
         int ret;
         char obj_file_name[64];
@@ -4323,27 +4348,18 @@ aot_obj_data_create(AOTCompContext *comp_ctx)
             aot_set_last_error("create mem buffer with file failed.");
             goto fail;
         }
-#endif /* end of defined(_WIN32) || defined(_WIN32_) */
     }
     else if (!strncmp(LLVMGetTargetName(target), "arc", 3)) {
-#if defined(_WIN32) || defined(_WIN32_)
-        aot_set_last_error("emit object file on Windows is unsupported.");
-        goto fail;
-#else
         /* Emit to assembly file instead for arc target
            as it cannot emit to object file */
         char file_name[] = "wasm-XXXXXX", buf[128];
-        int fd, ret;
+        int ret;
 
-        if ((fd = mkstemp(file_name)) <= 0) {
+        if (!bh_mkstemp(file_name, sizeof(file_name))) {
             aot_set_last_error("make temp file failed.");
             goto fail;
         }
 
-        /* close and remove temp file */
-        close(fd);
-        unlink(file_name);
-
         snprintf(buf, sizeof(buf), "%s%s", file_name, ".s");
         if (LLVMTargetMachineEmitToFile(comp_ctx->target_machine,
                                         comp_ctx->module, buf, LLVMAssemblyFile,
@@ -4364,7 +4380,7 @@ aot_obj_data_create(AOTCompContext *comp_ctx)
                  "/opt/zephyr-sdk/arc-zephyr-elf/bin/arc-zephyr-elf-gcc ",
                  "-mcpu=arcem -o ", file_name, ".o -c ", file_name, ".s");
         /* TODO: use try..catch to handle possible exceptions */
-        ret = system(buf);
+        ret = bh_system(buf);
         /* remove temp assembly file */
         snprintf(buf, sizeof(buf), "%s%s", file_name, ".s");
         unlink(buf);
@@ -4391,7 +4407,6 @@ aot_obj_data_create(AOTCompContext *comp_ctx)
             aot_set_last_error("create mem buffer with file failed.");
             goto fail;
         }
-#endif /* end of defined(_WIN32) || defined(_WIN32_) */
     }
     else {
         if (LLVMTargetMachineEmitToMemoryBuffer(
@@ -4433,6 +4448,15 @@ aot_obj_data_create(AOTCompContext *comp_ctx)
     if (comp_ctx->enable_gc) {
         obj_data->target_info.feature_flags |= WASM_FEATURE_GARBAGE_COLLECTION;
     }
+    if (comp_ctx->aux_stack_frame_type == AOT_STACK_FRAME_TYPE_TINY) {
+        obj_data->target_info.feature_flags |= WASM_FEATURE_TINY_STACK_FRAME;
+    }
+    if (comp_ctx->call_stack_features.frame_per_function) {
+        obj_data->target_info.feature_flags |= WASM_FEATURE_FRAME_PER_FUNCTION;
+    }
+    if (!comp_ctx->call_stack_features.func_idx) {
+        obj_data->target_info.feature_flags |= WASM_FEATURE_FRAME_NO_FUNC_IDX;
+    }
 
     bh_print_time("Begin to resolve object file info");
 

core/iwasm/compilation/aot_emit_control.c

@@ -6,6 +6,7 @@
 #include "aot_emit_control.h"
 #include "aot_compiler.h"
 #include "aot_emit_exception.h"
+#include "aot_stack_frame_comp.h"
 #if WASM_ENABLE_GC != 0
 #include "aot_emit_gc.h"
 #endif
@@ -38,13 +39,24 @@ format_block_name(char *name, uint32 name_size, uint32 block_index,
         snprintf(name, name_size, "%s", "func_end");
 }
 
-#define CREATE_BLOCK(new_llvm_block, name)                      \
-    do {                                                        \
-        if (!(new_llvm_block = LLVMAppendBasicBlockInContext(   \
-                  comp_ctx->context, func_ctx->func, name))) {  \
-            aot_set_last_error("add LLVM basic block failed."); \
-            goto fail;                                          \
-        }                                                       \
+#define CREATE_BLOCK(new_llvm_block, name)                                   \
+    do {                                                                     \
+        if (!(new_llvm_block = LLVMAppendBasicBlockInContext(                \
+                  comp_ctx->context, func_ctx->func, name))) {               \
+            aot_set_last_error("add LLVM basic block failed.");              \
+            goto fail;                                                       \
+        }                                                                    \
+        if (!strcmp(name, "func_end") && comp_ctx->aux_stack_frame_type      \
+            && comp_ctx->call_stack_features.frame_per_function) {           \
+            LLVMBasicBlockRef cur_block =                                    \
+                LLVMGetInsertBlock(comp_ctx->builder);                       \
+            SET_BUILDER_POS(new_llvm_block);                                 \
+            if (!aot_free_frame_per_function_frame_for_aot_func(comp_ctx,    \
+                                                                func_ctx)) { \
+                goto fail;                                                   \
+            }                                                                \
+            SET_BUILDER_POS(cur_block);                                      \
+        }                                                                    \
     } while (0)
 
 #define CURR_BLOCK() LLVMGetInsertBlock(comp_ctx->builder)
@@ -93,6 +105,11 @@ format_block_name(char *name, uint32 name_size, uint32 block_index,
                 goto fail;                                                  \
             }                                                               \
             SET_BUILDER_POS(block->llvm_end_block);                         \
+            LLVMValueRef first_instr =                                      \
+                get_first_non_phi(block->llvm_end_block);                   \
+            if (first_instr) {                                              \
+                LLVMPositionBuilderBefore(comp_ctx->builder, first_instr);  \
+            }                                                               \
             for (_i = 0; _i < block->result_count; _i++) {                  \
                 if (!(block->result_phis[_i] = LLVMBuildPhi(                \
                           comp_ctx->builder,                                \
@@ -158,6 +175,18 @@ get_target_block(AOTFuncContext *func_ctx, uint32 br_depth)
     return block;
 }
 
+LLVMValueRef
+get_first_non_phi(LLVMBasicBlockRef block)
+{
+    LLVMValueRef instr = LLVMGetFirstInstruction(block);
+
+    while (instr && LLVMIsAPHINode(instr)) {
+        instr = LLVMGetNextInstruction(instr);
+    }
+
+    return instr;
+}
+
 static void
 clear_frame_locals(AOTCompFrame *aot_frame)
 {
@@ -883,7 +912,7 @@ check_suspend_flags(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
         aot_set_last_error("llvm build LOAD failed");
         return false;
     }
-    /* Set terminate_flags memory accecc to volatile, so that the value
+    /* Set terminate_flags memory access to volatile, so that the value
         will always be loaded from memory rather than register */
     LLVMSetVolatile(terminate_flags, true);
 
@@ -1189,6 +1218,28 @@ aot_compile_op_br_table(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
         return aot_handle_next_reachable_block(comp_ctx, func_ctx, p_frame_ip);
     }
 
+    /*
+     * if (value_cmp > br_count)
+     *   value_cmp = br_count;
+     */
+    LLVMValueRef br_count_value = I32_CONST(br_count);
+    CHECK_LLVM_CONST(br_count_value);
+
+    LLVMValueRef clap_value_cmp_cond =
+        LLVMBuildICmp(comp_ctx->builder, LLVMIntUGT, value_cmp, br_count_value,
+                      "cmp_w_br_count");
+    if (!clap_value_cmp_cond) {
+        aot_set_last_error("llvm build icmp failed.");
+        return false;
+    }
+
+    value_cmp = LLVMBuildSelect(comp_ctx->builder, clap_value_cmp_cond,
+                                br_count_value, value_cmp, "clap_value_cmp");
+    if (!value_cmp) {
+        aot_set_last_error("llvm build select failed.");
+        return false;
+    }
+
     if (!LLVMIsEfficientConstInt(value_cmp)) {
         if (comp_ctx->aot_frame) {
             if (comp_ctx->enable_gc
@@ -1361,6 +1412,13 @@ aot_compile_op_return(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
         (*p_frame_ip - 1) - comp_ctx->comp_data->wasm_module->buf_code);
 #endif
 
+    if (comp_ctx->aux_stack_frame_type
+        && comp_ctx->call_stack_features.frame_per_function
+        && !aot_free_frame_per_function_frame_for_aot_func(comp_ctx,
+                                                           func_ctx)) {
+        return false;
+    }
+
     if (block_func->result_count) {
         /* Store extra result values to function parameters */
         for (i = 0; i < block_func->result_count - 1; i++) {

core/iwasm/compilation/aot_emit_conversion.c

@@ -347,17 +347,8 @@ aot_compile_op_i32_trunc_f32(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
 
     POP_F32(value);
 
-    if (!comp_ctx->is_indirect_mode) {
-        if (sign) {
-            min_value = F32_CONST(-2147483904.0f);
-            max_value = F32_CONST(2147483648.0f);
-        }
-        else {
-            min_value = F32_CONST(-1.0f);
-            max_value = F32_CONST(4294967296.0f);
-        }
-    }
-    else {
+    if (comp_ctx->is_indirect_mode
+        && aot_intrinsic_check_capability(comp_ctx, "f32.const")) {
         WASMValue wasm_value;
         if (sign) {
             wasm_value.f32 = -2147483904.0f;
@@ -376,6 +367,16 @@ aot_compile_op_i32_trunc_f32(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
                 comp_ctx, func_ctx->native_symbol, &wasm_value, VALUE_TYPE_F32);
         }
     }
+    else {
+        if (sign) {
+            min_value = F32_CONST(-2147483904.0f);
+            max_value = F32_CONST(2147483648.0f);
+        }
+        else {
+            min_value = F32_CONST(-1.0f);
+            max_value = F32_CONST(4294967296.0f);
+        }
+    }
     CHECK_LLVM_CONST(min_value);
     CHECK_LLVM_CONST(max_value);
 
@@ -400,17 +401,8 @@ aot_compile_op_i32_trunc_f64(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
 
     POP_F64(value);
 
-    if (!comp_ctx->is_indirect_mode) {
-        if (sign) {
-            min_value = F64_CONST(-2147483649.0);
-            max_value = F64_CONST(2147483648.0);
-        }
-        else {
-            min_value = F64_CONST(-1.0);
-            max_value = F64_CONST(4294967296.0);
-        }
-    }
-    else {
+    if (comp_ctx->is_indirect_mode
+        && aot_intrinsic_check_capability(comp_ctx, "f64.const")) {
         WASMValue wasm_value;
         if (sign) {
             wasm_value.f64 = -2147483649.0;
@@ -429,6 +421,16 @@ aot_compile_op_i32_trunc_f64(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
                 comp_ctx, func_ctx->native_symbol, &wasm_value, VALUE_TYPE_F64);
         }
     }
+    else {
+        if (sign) {
+            min_value = F64_CONST(-2147483649.0);
+            max_value = F64_CONST(2147483648.0);
+        }
+        else {
+            min_value = F64_CONST(-1.0);
+            max_value = F64_CONST(4294967296.0);
+        }
+    }
     CHECK_LLVM_CONST(min_value);
     CHECK_LLVM_CONST(max_value);
 
@@ -554,17 +556,8 @@ aot_compile_op_i64_trunc_f32(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
 
     POP_F32(value);
 
-    if (!comp_ctx->is_indirect_mode) {
-        if (sign) {
-            min_value = F32_CONST(-9223373136366403584.0f);
-            max_value = F32_CONST(9223372036854775808.0f);
-        }
-        else {
-            min_value = F32_CONST(-1.0f);
-            max_value = F32_CONST(18446744073709551616.0f);
-        }
-    }
-    else {
+    if (comp_ctx->is_indirect_mode
+        && aot_intrinsic_check_capability(comp_ctx, "f32.const")) {
         WASMValue wasm_value;
         if (sign) {
             wasm_value.f32 = -9223373136366403584.0f;
@@ -583,6 +576,16 @@ aot_compile_op_i64_trunc_f32(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
                 comp_ctx, func_ctx->native_symbol, &wasm_value, VALUE_TYPE_F32);
         }
     }
+    else {
+        if (sign) {
+            min_value = F32_CONST(-9223373136366403584.0f);
+            max_value = F32_CONST(9223372036854775808.0f);
+        }
+        else {
+            min_value = F32_CONST(-1.0f);
+            max_value = F32_CONST(18446744073709551616.0f);
+        }
+    }
     CHECK_LLVM_CONST(min_value);
     CHECK_LLVM_CONST(max_value);
 
@@ -607,17 +610,8 @@ aot_compile_op_i64_trunc_f64(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
 
     POP_F64(value);
 
-    if (!comp_ctx->is_indirect_mode) {
-        if (sign) {
-            min_value = F64_CONST(-9223372036854777856.0);
-            max_value = F64_CONST(9223372036854775808.0);
-        }
-        else {
-            min_value = F64_CONST(-1.0);
-            max_value = F64_CONST(18446744073709551616.0);
-        }
-    }
-    else {
+    if (comp_ctx->is_indirect_mode
+        && aot_intrinsic_check_capability(comp_ctx, "f64.const")) {
         WASMValue wasm_value;
         if (sign) {
             wasm_value.f64 = -9223372036854777856.0;
@@ -636,6 +630,16 @@ aot_compile_op_i64_trunc_f64(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
                 comp_ctx, func_ctx->native_symbol, &wasm_value, VALUE_TYPE_F64);
         }
     }
+    else {
+        if (sign) {
+            min_value = F64_CONST(-9223372036854777856.0);
+            max_value = F64_CONST(9223372036854775808.0);
+        }
+        else {
+            min_value = F64_CONST(-1.0);
+            max_value = F64_CONST(18446744073709551616.0);
+        }
+    }
     CHECK_LLVM_CONST(min_value);
     CHECK_LLVM_CONST(max_value);
 

core/iwasm/compilation/aot_emit_exception.c

@@ -4,49 +4,10 @@
  */
 
 #include "aot_emit_exception.h"
+#include "aot_compiler.h"
 #include "../interpreter/wasm_runtime.h"
 #include "../aot/aot_runtime.h"
 
-static bool
-commit_ip(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
-          LLVMValueRef exce_ip, bool is_64bit)
-{
-    LLVMValueRef cur_frame = func_ctx->cur_frame;
-    LLVMValueRef value_offset, value_addr, value_ptr;
-    uint32 offset_ip;
-
-    if (!comp_ctx->is_jit_mode)
-        offset_ip = comp_ctx->pointer_size * 4;
-    else
-        offset_ip = offsetof(WASMInterpFrame, ip);
-
-    if (!(value_offset = I32_CONST(offset_ip))) {
-        aot_set_last_error("llvm build const failed");
-        return false;
-    }
-
-    if (!(value_addr =
-              LLVMBuildInBoundsGEP2(comp_ctx->builder, INT8_TYPE, cur_frame,
-                                    &value_offset, 1, "ip_addr"))) {
-        aot_set_last_error("llvm build in bounds gep failed");
-        return false;
-    }
-
-    if (!(value_ptr = LLVMBuildBitCast(
-              comp_ctx->builder, value_addr,
-              is_64bit ? INT64_PTR_TYPE : INT32_PTR_TYPE, "ip_ptr"))) {
-        aot_set_last_error("llvm build bit cast failed");
-        return false;
-    }
-
-    if (!LLVMBuildStore(comp_ctx->builder, exce_ip, value_ptr)) {
-        aot_set_last_error("llvm build store failed");
-        return false;
-    }
-
-    return true;
-}
-
 bool
 aot_emit_exception(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
                    int32 exception_id, bool is_cond_br, LLVMValueRef cond_br_if,
@@ -80,7 +41,7 @@ aot_emit_exception(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
             return false;
         }
 
-        if (comp_ctx->aot_frame) {
+        if (comp_ctx->aot_frame && comp_ctx->call_stack_features.trap_ip) {
             /* Create exception ip phi */
             if (!(func_ctx->exception_ip_phi = LLVMBuildPhi(
                       comp_ctx->builder, is_64bit ? I64_TYPE : I32_TYPE,
@@ -90,8 +51,8 @@ aot_emit_exception(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
             }
 
             /* Commit ip to current frame */
-            if (!commit_ip(comp_ctx, func_ctx, func_ctx->exception_ip_phi,
-                           is_64bit)) {
+            if (!aot_gen_commit_ip(comp_ctx, func_ctx,
+                                   func_ctx->exception_ip_phi, is_64bit)) {
                 return false;
             }
         }
@@ -173,7 +134,7 @@ aot_emit_exception(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
     /* Add phi incoming value to got_exception block */
     LLVMAddIncoming(func_ctx->exception_id_phi, &exce_id, &block_curr, 1);
 
-    if (comp_ctx->aot_frame) {
+    if (comp_ctx->aot_frame && comp_ctx->call_stack_features.trap_ip) {
         const uint8 *ip = comp_ctx->aot_frame->frame_ip;
         LLVMValueRef exce_ip = NULL;
 

core/iwasm/compilation/aot_emit_function.c

@@ -7,6 +7,7 @@
 #include "aot_emit_exception.h"
 #include "aot_emit_control.h"
 #include "aot_emit_table.h"
+#include "aot_stack_frame_comp.h"
 #include "../aot/aot_runtime.h"
 #if WASM_ENABLE_GC != 0
 #include "aot_emit_gc.h"
@@ -682,24 +683,29 @@ alloc_frame_for_aot_func(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
 
     new_frame = wasm_stack_top;
 
-    if (!(check_wasm_stack_succ = LLVMAppendBasicBlockInContext(
-              comp_ctx->context, func_ctx->func, "check_wasm_stack_succ"))) {
-        aot_set_last_error("llvm add basic block failed.");
-        return false;
-    }
+    if (comp_ctx->call_stack_features.bounds_checks) {
+        if (!(check_wasm_stack_succ = LLVMAppendBasicBlockInContext(
+                  comp_ctx->context, func_ctx->func,
+                  "check_wasm_stack_succ"))) {
+            aot_set_last_error("llvm add basic block failed.");
+            return false;
+        }
 
-    LLVMMoveBasicBlockAfter(check_wasm_stack_succ,
-                            LLVMGetInsertBlock(comp_ctx->builder));
+        LLVMMoveBasicBlockAfter(check_wasm_stack_succ,
+                                LLVMGetInsertBlock(comp_ctx->builder));
 
-    if (!(cmp = LLVMBuildICmp(comp_ctx->builder, LLVMIntUGT, wasm_stack_top_max,
-                              wasm_stack_top_bound, "cmp"))) {
-        aot_set_last_error("llvm build icmp failed");
-        return false;
-    }
+        if (!(cmp = LLVMBuildICmp(comp_ctx->builder, LLVMIntUGT,
+                                  wasm_stack_top_max, wasm_stack_top_bound,
+                                  "cmp"))) {
+            aot_set_last_error("llvm build icmp failed");
+            return false;
+        }
 
-    if (!(aot_emit_exception(comp_ctx, func_ctx, EXCE_OPERAND_STACK_OVERFLOW,
-                             true, cmp, check_wasm_stack_succ))) {
-        return false;
+        if (!(aot_emit_exception(comp_ctx, func_ctx,
+                                 EXCE_OPERAND_STACK_OVERFLOW, true, cmp,
+                                 check_wasm_stack_succ))) {
+            return false;
+        }
     }
 
 #if WASM_ENABLE_GC != 0
@@ -879,25 +885,28 @@ alloc_frame_for_aot_func(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
     }
 
     if (!comp_ctx->is_jit_mode) {
-        /* aot mode: new_frame->func_idx = func_idx */
-        func_idx_val = comp_ctx->pointer_size == sizeof(uint64)
-                           ? I64_CONST(func_idx)
-                           : I32_CONST(func_idx);
-        offset = I32_CONST(comp_ctx->pointer_size);
-        CHECK_LLVM_CONST(func_idx_val);
-        CHECK_LLVM_CONST(offset);
-        if (!(func_idx_ptr =
-                  LLVMBuildInBoundsGEP2(comp_ctx->builder, INT8_TYPE, new_frame,
-                                        &offset, 1, "func_idx_addr"))
-            || !(func_idx_ptr =
-                     LLVMBuildBitCast(comp_ctx->builder, func_idx_ptr,
-                                      INTPTR_T_PTR_TYPE, "func_idx_ptr"))) {
-            aot_set_last_error("llvm get func_idx_ptr failed");
-            return false;
-        }
-        if (!LLVMBuildStore(comp_ctx->builder, func_idx_val, func_idx_ptr)) {
-            aot_set_last_error("llvm build store failed");
-            return false;
+        if (comp_ctx->call_stack_features.func_idx) {
+            /* aot mode: new_frame->func_idx = func_idx */
+            func_idx_val = comp_ctx->pointer_size == sizeof(uint64)
+                               ? I64_CONST(func_idx)
+                               : I32_CONST(func_idx);
+            offset = I32_CONST(comp_ctx->pointer_size);
+            CHECK_LLVM_CONST(func_idx_val);
+            CHECK_LLVM_CONST(offset);
+            if (!(func_idx_ptr = LLVMBuildInBoundsGEP2(
+                      comp_ctx->builder, INT8_TYPE, new_frame, &offset, 1,
+                      "func_idx_addr"))
+                || !(func_idx_ptr =
+                         LLVMBuildBitCast(comp_ctx->builder, func_idx_ptr,
+                                          INTPTR_T_PTR_TYPE, "func_idx_ptr"))) {
+                aot_set_last_error("llvm get func_idx_ptr failed");
+                return false;
+            }
+            if (!LLVMBuildStore(comp_ctx->builder, func_idx_val,
+                                func_idx_ptr)) {
+                aot_set_last_error("llvm build store failed");
+                return false;
+            }
         }
     }
     else {
@@ -1285,6 +1294,10 @@ commit_params_to_frame_of_import_func(AOTCompContext *comp_ctx,
 {
     uint32 i, n;
 
+    if (!comp_ctx->call_stack_features.values) {
+        return true;
+    }
+
     for (i = 0, n = 0; i < func_type->param_count; i++, n++) {
         switch (func_type->types[i]) {
             case VALUE_TYPE_I32:
@@ -1394,6 +1407,9 @@ aot_compile_op_call(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
     LLVMValueRef *param_values = NULL, value_ret = NULL, func;
     LLVMValueRef import_func_idx, res;
     LLVMValueRef ext_ret, ext_ret_ptr, ext_ret_idx;
+#if WASM_ENABLE_AOT_STACK_FRAME != 0
+    LLVMValueRef func_idx_ref;
+#endif
     int32 i, j = 0, param_count, result_count, ext_ret_count;
     uint64 total_size;
     uint8 wasm_ret_type;
@@ -1438,12 +1454,28 @@ aot_compile_op_call(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
             return false;
     }
 
-    if (comp_ctx->enable_aux_stack_frame) {
 #if WASM_ENABLE_AOT_STACK_FRAME != 0
-        if (!alloc_frame_for_aot_func(comp_ctx, func_ctx, func_idx))
-            return false;
-#endif
+    if (comp_ctx->aux_stack_frame_type) {
+        if (func_idx < import_func_count
+            && comp_ctx->call_stack_features.frame_per_function) {
+            INT_CONST(func_idx_ref, func_idx, I32_TYPE, true);
+            if (!aot_alloc_frame_per_function_frame_for_aot_func(
+                    comp_ctx, func_ctx, func_idx_ref)) {
+                return false;
+            }
+        }
+        else if (!comp_ctx->call_stack_features.frame_per_function) {
+            if (comp_ctx->aux_stack_frame_type
+                != AOT_STACK_FRAME_TYPE_STANDARD) {
+                aot_set_last_error("unsupported mode");
+                return false;
+            }
+            if (!alloc_frame_for_aot_func(comp_ctx, func_ctx, func_idx)) {
+                return false;
+            }
+        }
     }
+#endif
 
     /* Get param cell number */
     param_cell_num = func_type->param_cell_num;
@@ -1513,7 +1545,7 @@ aot_compile_op_call(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
     }
 
     if (func_idx < import_func_count) {
-        if (comp_ctx->enable_aux_stack_frame
+        if (comp_ctx->aux_stack_frame_type == AOT_STACK_FRAME_TYPE_STANDARD
             && !commit_params_to_frame_of_import_func(
                 comp_ctx, func_ctx, func_type, param_values + 1)) {
             goto fail;
@@ -1800,16 +1832,31 @@ aot_compile_op_call(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
                 aot_set_last_error("llvm build load failed.");
                 goto fail;
             }
+            LLVMSetAlignment(ext_ret, 4);
             PUSH(ext_ret, ext_ret_types[i]);
         }
     }
 
-    if (comp_ctx->enable_aux_stack_frame) {
 #if WASM_ENABLE_AOT_STACK_FRAME != 0
-        if (!free_frame_for_aot_func(comp_ctx, func_ctx))
-            goto fail;
-#endif
+    if (comp_ctx->aux_stack_frame_type) {
+        if (func_idx < import_func_count
+            && comp_ctx->call_stack_features.frame_per_function) {
+            if (!aot_free_frame_per_function_frame_for_aot_func(comp_ctx,
+                                                                func_ctx)) {
+                goto fail;
+            }
+        }
+        else if (!comp_ctx->call_stack_features.frame_per_function) {
+            if (comp_ctx->aux_stack_frame_type
+                != AOT_STACK_FRAME_TYPE_STANDARD) {
+                aot_set_last_error("unsupported mode");
+            }
+            if (!free_frame_for_aot_func(comp_ctx, func_ctx)) {
+                goto fail;
+            }
+        }
     }
+#endif
 
     /* Insert suspend check point */
     if (comp_ctx->enable_thread_mgr) {
@@ -2022,6 +2069,7 @@ call_aot_call_indirect_func(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
             aot_set_last_error("llvm build load failed.");
             return false;
         }
+        LLVMSetAlignment(value_rets[i], 4);
         cell_num += wasm_value_type_cell_num_internal(wasm_ret_types[i],
                                                       comp_ctx->pointer_size);
     }
@@ -2043,6 +2091,9 @@ aot_compile_op_call_indirect(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
     LLVMValueRef ext_ret_offset, ext_ret_ptr, ext_ret, res;
     LLVMValueRef *param_values = NULL, *value_rets = NULL;
     LLVMValueRef *result_phis = NULL, value_ret, import_func_count;
+#if WASM_ENABLE_MEMORY64 != 0
+    LLVMValueRef u32_max, u32_cmp_result = NULL;
+#endif
     LLVMTypeRef *param_types = NULL, ret_type;
     LLVMTypeRef llvm_func_type, llvm_func_ptr_type;
     LLVMTypeRef ext_ret_ptr_type;
@@ -2107,7 +2158,7 @@ aot_compile_op_call_indirect(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
     func_param_count = func_type->param_count;
     func_result_count = func_type->result_count;
 
-    POP_I32(elem_idx);
+    POP_TBL_ELEM_IDX(elem_idx);
 
     /* get the cur size of the table instance */
     if (!(offset = I32_CONST(get_tbl_inst_offset(comp_ctx, func_ctx, tbl_idx)
@@ -2136,6 +2187,27 @@ aot_compile_op_call_indirect(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
         goto fail;
     }
 
+#if WASM_ENABLE_MEMORY64 != 0
+    /* Check if elem index >= UINT32_MAX */
+    if (IS_TABLE64(tbl_idx)) {
+        if (!(u32_max = I64_CONST(UINT32_MAX))) {
+            aot_set_last_error("llvm build const failed");
+            goto fail;
+        }
+        if (!(u32_cmp_result =
+                  LLVMBuildICmp(comp_ctx->builder, LLVMIntUGE, elem_idx,
+                                u32_max, "cmp_elem_idx_u32_max"))) {
+            aot_set_last_error("llvm build icmp failed.");
+            goto fail;
+        }
+        if (!(elem_idx = LLVMBuildTrunc(comp_ctx->builder, elem_idx, I32_TYPE,
+                                        "elem_idx_i32"))) {
+            aot_set_last_error("llvm build trunc failed.");
+            goto fail;
+        }
+    }
+#endif
+
     /* Check if (uint32)elem index >= table size */
     if (!(cmp_elem_idx = LLVMBuildICmp(comp_ctx->builder, LLVMIntUGE, elem_idx,
                                        table_size_const, "cmp_elem_idx"))) {
@@ -2143,7 +2215,19 @@ aot_compile_op_call_indirect(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
         goto fail;
     }
 
-    /* Throw exception if elem index >= table size */
+#if WASM_ENABLE_MEMORY64 != 0
+    if (IS_TABLE64(tbl_idx)) {
+        if (!(cmp_elem_idx =
+                  LLVMBuildOr(comp_ctx->builder, cmp_elem_idx, u32_cmp_result,
+                              "larger_than_u32_max_or_cur_size"))) {
+            aot_set_last_error("llvm build or failed.");
+            goto fail;
+        }
+    }
+#endif
+
+    /* Throw exception if elem index >= table size or elem index >= UINT32_MAX
+     */
     if (!(check_elem_idx_succ = LLVMAppendBasicBlockInContext(
               comp_ctx->context, func_ctx->func, "check_elem_idx_succ"))) {
         aot_set_last_error("llvm add basic block failed.");
@@ -2430,7 +2514,8 @@ aot_compile_op_call_indirect(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
         goto fail;
     }
 
-    if (comp_ctx->enable_aux_stack_frame) {
+    if (comp_ctx->aux_stack_frame_type
+        && !comp_ctx->call_stack_features.frame_per_function) {
 #if WASM_ENABLE_AOT_STACK_FRAME != 0
         /*  TODO: use current frame instead of allocating new frame
                   for WASM_OP_RETURN_CALL_INDIRECT */
@@ -2499,7 +2584,13 @@ aot_compile_op_call_indirect(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
     /* Translate call import block */
     LLVMPositionBuilderAtEnd(comp_ctx->builder, block_call_import);
 
-    if (comp_ctx->enable_aux_stack_frame
+    if (comp_ctx->aot_frame && comp_ctx->call_stack_features.frame_per_function
+        && !aot_alloc_frame_per_function_frame_for_aot_func(comp_ctx, func_ctx,
+                                                            func_idx)) {
+        goto fail;
+    }
+
+    if (comp_ctx->aux_stack_frame_type == AOT_STACK_FRAME_TYPE_STANDARD
         && !commit_params_to_frame_of_import_func(comp_ctx, func_ctx, func_type,
                                                   param_values + 1)) {
         goto fail;
@@ -2536,6 +2627,12 @@ aot_compile_op_call_indirect(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
         && !check_call_return(comp_ctx, func_ctx, res))
         goto fail;
 
+    if (comp_ctx->aot_frame && comp_ctx->call_stack_features.frame_per_function
+        && !aot_free_frame_per_function_frame_for_aot_func(comp_ctx,
+                                                           func_ctx)) {
+        goto fail;
+    }
+
     block_curr = LLVMGetInsertBlock(comp_ctx->builder);
     for (i = 0; i < func_result_count; i++) {
         LLVMAddIncoming(result_phis[i], &value_rets[i], &block_curr, 1);
@@ -2604,6 +2701,7 @@ aot_compile_op_call_indirect(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
                 aot_set_last_error("llvm build load failed.");
                 goto fail;
             }
+            LLVMSetAlignment(ext_ret, 4);
             LLVMAddIncoming(result_phis[i], &ext_ret, &block_curr, 1);
         }
     }
@@ -2620,7 +2718,8 @@ aot_compile_op_call_indirect(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
         PUSH(result_phis[i], func_type->types[func_param_count + i]);
     }
 
-    if (comp_ctx->enable_aux_stack_frame) {
+    if (comp_ctx->aux_stack_frame_type
+        && !comp_ctx->call_stack_features.frame_per_function) {
 #if WASM_ENABLE_AOT_STACK_FRAME != 0
         if (!free_frame_for_aot_func(comp_ctx, func_ctx))
             goto fail;
@@ -2927,7 +3026,8 @@ aot_compile_op_call_ref(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
         goto fail;
     }
 
-    if (comp_ctx->enable_aux_stack_frame) {
+    if (comp_ctx->aux_stack_frame_type
+        && !comp_ctx->call_stack_features.frame_per_function) {
 #if WASM_ENABLE_AOT_STACK_FRAME != 0
         /*  TODO: use current frame instead of allocating new frame
                   for WASM_OP_RETURN_CALL_REF */
@@ -2996,7 +3096,7 @@ aot_compile_op_call_ref(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
     /* Translate call import block */
     LLVMPositionBuilderAtEnd(comp_ctx->builder, block_call_import);
 
-    if (comp_ctx->enable_aux_stack_frame
+    if (comp_ctx->aux_stack_frame_type == AOT_STACK_FRAME_TYPE_STANDARD
         && !commit_params_to_frame_of_import_func(comp_ctx, func_ctx, func_type,
                                                   param_values + 1)) {
         goto fail;
@@ -3033,6 +3133,7 @@ aot_compile_op_call_ref(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
                 aot_set_last_error("llvm build load failed.");
                 goto fail;
             }
+            LLVMSetAlignment(ext_ret, 4);
             LLVMAddIncoming(result_phis[i], &ext_ret, &block_curr, 1);
         }
     }
@@ -3108,6 +3209,7 @@ aot_compile_op_call_ref(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
                 aot_set_last_error("llvm build load failed.");
                 goto fail;
             }
+            LLVMSetAlignment(ext_ret, 4);
             LLVMAddIncoming(result_phis[i], &ext_ret, &block_curr, 1);
         }
     }
@@ -3124,7 +3226,8 @@ aot_compile_op_call_ref(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
         PUSH(result_phis[i], func_type->types[func_param_count + i]);
     }
 
-    if (comp_ctx->enable_aux_stack_frame) {
+    if (comp_ctx->aux_stack_frame_type
+        && !comp_ctx->call_stack_features.frame_per_function) {
 #if WASM_ENABLE_AOT_STACK_FRAME != 0
         if (!free_frame_for_aot_func(comp_ctx, func_ctx))
             goto fail;

core/iwasm/compilation/aot_emit_memory.c

@@ -118,10 +118,10 @@ aot_check_memory_overflow(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
 {
     LLVMValueRef offset_const =
         MEMORY64_COND_VALUE(I64_CONST(offset), I32_CONST(offset));
-    LLVMValueRef addr, maddr, offset1, cmp1, cmp2, cmp;
+    LLVMValueRef addr, maddr, maddr_phi = NULL, offset1, cmp1, cmp2, cmp;
     LLVMValueRef mem_base_addr, mem_check_bound;
     LLVMBasicBlockRef block_curr = LLVMGetInsertBlock(comp_ctx->builder);
-    LLVMBasicBlockRef check_succ;
+    LLVMBasicBlockRef check_succ, block_maddr_phi = NULL;
     AOTValue *aot_value_top;
     uint32 local_idx_of_aot_value = 0;
     uint64 const_value;
@@ -131,6 +131,11 @@ aot_check_memory_overflow(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
     bool is_shared_memory =
         comp_ctx->comp_data->memories[0].flags & SHARED_MEMORY_FLAG;
 #endif
+#if WASM_ENABLE_MEMORY64 == 0
+    bool is_memory64 = false;
+#else
+    bool is_memory64 = IS_MEMORY64;
+#endif
 
     is_target_64bit = (comp_ctx->pointer_size == sizeof(uint64)) ? true : false;
 
@@ -270,6 +275,149 @@ aot_check_memory_overflow(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
     /* offset1 = offset + addr; */
     BUILD_OP(Add, offset_const, addr, offset1, "offset1");
 
+    if (is_memory64 && comp_ctx->enable_bound_check) {
+        /* Check whether integer overflow occurs in offset + addr */
+        LLVMBasicBlockRef check_integer_overflow_end;
+        ADD_BASIC_BLOCK(check_integer_overflow_end,
+                        "check_integer_overflow_end");
+        LLVMMoveBasicBlockAfter(check_integer_overflow_end, block_curr);
+
+        BUILD_ICMP(LLVMIntULT, offset1, offset_const, cmp1, "cmp1");
+        if (!aot_emit_exception(comp_ctx, func_ctx,
+                                EXCE_OUT_OF_BOUNDS_MEMORY_ACCESS, true, cmp1,
+                                check_integer_overflow_end)) {
+            goto fail;
+        }
+        SET_BUILD_POS(check_integer_overflow_end);
+    }
+
+    if (comp_ctx->enable_shared_heap /* TODO: && mem_idx == 0 */) {
+        LLVMBasicBlockRef app_addr_in_shared_heap, app_addr_in_linear_mem;
+        LLVMValueRef is_in_shared_heap, shared_heap_check_bound = NULL;
+
+        /* Add basic blocks */
+        ADD_BASIC_BLOCK(app_addr_in_shared_heap, "app_addr_in_shared_heap");
+        ADD_BASIC_BLOCK(app_addr_in_linear_mem, "app_addr_in_linear_mem");
+        ADD_BASIC_BLOCK(block_maddr_phi, "maddr_phi");
+
+        LLVMMoveBasicBlockAfter(app_addr_in_shared_heap, block_curr);
+        LLVMMoveBasicBlockAfter(app_addr_in_linear_mem,
+                                app_addr_in_shared_heap);
+        LLVMMoveBasicBlockAfter(block_maddr_phi, app_addr_in_linear_mem);
+
+        LLVMPositionBuilderAtEnd(comp_ctx->builder, block_maddr_phi);
+        if (!(maddr_phi =
+                  LLVMBuildPhi(comp_ctx->builder,
+                               enable_segue ? INT8_PTR_TYPE_GS : INT8_PTR_TYPE,
+                               "maddr_phi"))) {
+            aot_set_last_error("llvm build phi failed");
+            goto fail;
+        }
+
+        LLVMPositionBuilderAtEnd(comp_ctx->builder, block_curr);
+
+        if (!is_target_64bit) {
+            /* Check whether integer overflow occurs in addr + offset */
+            LLVMBasicBlockRef check_integer_overflow_end;
+            ADD_BASIC_BLOCK(check_integer_overflow_end,
+                            "check_integer_overflow_end");
+            LLVMMoveBasicBlockAfter(check_integer_overflow_end, block_curr);
+
+            BUILD_ICMP(LLVMIntULT, offset1, addr, cmp1, "cmp1");
+            if (!aot_emit_exception(comp_ctx, func_ctx,
+                                    EXCE_OUT_OF_BOUNDS_MEMORY_ACCESS, true,
+                                    cmp1, check_integer_overflow_end)) {
+                goto fail;
+            }
+            SET_BUILD_POS(check_integer_overflow_end);
+        }
+
+        shared_heap_check_bound =
+            is_memory64 ? I64_CONST(UINT64_MAX - bytes + 1)
+                        : (comp_ctx->pointer_size == sizeof(uint64)
+                               ? I64_CONST(UINT32_MAX - bytes + 1)
+                               : I32_CONST(UINT32_MAX - bytes + 1));
+        CHECK_LLVM_CONST(shared_heap_check_bound);
+
+        /* Check whether the bytes to access are in shared heap */
+        if (!comp_ctx->enable_bound_check) {
+            /* Use IntUGT but not IntUGE to compare, since (1) in the ems
+               memory allocator, the hmu node includes hmu header and hmu
+               memory, only the latter is returned to the caller as the
+               allocated memory, the hmu header isn't returned so the
+               first byte of the shared heap won't be accessed, (2) using
+               IntUGT gets better performance than IntUGE in some cases */
+            BUILD_ICMP(LLVMIntUGT, offset1, func_ctx->shared_heap_start_off,
+                       is_in_shared_heap, "is_in_shared_heap");
+            /* We don't check the shared heap's upper boundary if boundary
+               check isn't enabled, the runtime may also use the guard pages
+               of shared heap to check the boundary if hardware boundary
+               check feature is enabled. */
+        }
+        else {
+            /* Use IntUGT but not IntUGE to compare, same as above */
+            BUILD_ICMP(LLVMIntUGT, offset1, func_ctx->shared_heap_start_off,
+                       cmp1, "cmp1");
+            /* Check the shared heap's upper boundary if boundary check is
+               enabled */
+            BUILD_ICMP(LLVMIntULE, offset1, shared_heap_check_bound, cmp2,
+                       "cmp2");
+            BUILD_OP(And, cmp1, cmp2, is_in_shared_heap, "is_in_shared_heap");
+        }
+
+        if (!LLVMBuildCondBr(comp_ctx->builder, is_in_shared_heap,
+                             app_addr_in_shared_heap, app_addr_in_linear_mem)) {
+            aot_set_last_error("llvm build cond br failed");
+            goto fail;
+        }
+
+        LLVMPositionBuilderAtEnd(comp_ctx->builder, app_addr_in_shared_heap);
+
+        /* Get native address inside shared heap */
+        if (!(maddr =
+                  LLVMBuildInBoundsGEP2(comp_ctx->builder, INT8_TYPE,
+                                        func_ctx->shared_heap_base_addr_adj,
+                                        &offset1, 1, "maddr_shared_heap"))) {
+            aot_set_last_error("llvm build inbounds gep failed");
+            goto fail;
+        }
+
+        if (enable_segue) {
+            LLVMValueRef mem_base_addr_u64, maddr_u64, offset_to_mem_base;
+
+            if (!(maddr_u64 = LLVMBuildPtrToInt(comp_ctx->builder, maddr,
+                                                I64_TYPE, "maddr_u64"))
+                || !(mem_base_addr_u64 =
+                         LLVMBuildPtrToInt(comp_ctx->builder, mem_base_addr,
+                                           I64_TYPE, "mem_base_addr_u64"))) {
+                aot_set_last_error("llvm build ptr to int failed");
+                goto fail;
+            }
+            if (!(offset_to_mem_base =
+                      LLVMBuildSub(comp_ctx->builder, maddr_u64,
+                                   mem_base_addr_u64, "offset_to_mem_base"))) {
+                aot_set_last_error("llvm build sub failed");
+                goto fail;
+            }
+            if (!(maddr = LLVMBuildIntToPtr(
+                      comp_ctx->builder, offset_to_mem_base, INT8_PTR_TYPE_GS,
+                      "maddr_shared_heap_segue"))) {
+                aot_set_last_error("llvm build int to ptr failed.");
+                goto fail;
+            }
+        }
+
+        LLVMAddIncoming(maddr_phi, &maddr, &app_addr_in_shared_heap, 1);
+
+        if (!LLVMBuildBr(comp_ctx->builder, block_maddr_phi)) {
+            aot_set_last_error("llvm build br failed");
+            goto fail;
+        }
+
+        LLVMPositionBuilderAtEnd(comp_ctx->builder, app_addr_in_linear_mem);
+        block_curr = LLVMGetInsertBlock(comp_ctx->builder);
+    }
+
     if (comp_ctx->enable_bound_check
         && !(is_local_of_aot_value
              && aot_checked_addr_list_find(func_ctx, local_idx_of_aot_value,
@@ -305,10 +453,16 @@ aot_check_memory_overflow(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
             BUILD_ICMP(LLVMIntUGT, offset1, mem_check_bound, cmp, "cmp");
         }
         else {
-            /* Check integer overflow */
-            BUILD_ICMP(LLVMIntULT, offset1, addr, cmp1, "cmp1");
-            BUILD_ICMP(LLVMIntUGT, offset1, mem_check_bound, cmp2, "cmp2");
-            BUILD_OP(Or, cmp1, cmp2, cmp, "cmp");
+            if (comp_ctx->enable_shared_heap /* TODO: && mem_idx == 0 */) {
+                /* Check integer overflow has been checked above */
+                BUILD_ICMP(LLVMIntUGT, offset1, mem_check_bound, cmp, "cmp");
+            }
+            else {
+                /* Check integer overflow */
+                BUILD_ICMP(LLVMIntULT, offset1, addr, cmp1, "cmp1");
+                BUILD_ICMP(LLVMIntUGT, offset1, mem_check_bound, cmp2, "cmp2");
+                BUILD_OP(Or, cmp1, cmp2, cmp, "cmp");
+            }
         }
 
         /* Add basic blocks */
@@ -354,7 +508,19 @@ aot_check_memory_overflow(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
             goto fail;
         }
     }
-    return maddr;
+
+    if (comp_ctx->enable_shared_heap /* TODO: && mem_idx == 0 */) {
+        block_curr = LLVMGetInsertBlock(comp_ctx->builder);
+        LLVMAddIncoming(maddr_phi, &maddr, &block_curr, 1);
+        if (!LLVMBuildBr(comp_ctx->builder, block_maddr_phi)) {
+            aot_set_last_error("llvm build br failed");
+            goto fail;
+        }
+        LLVMPositionBuilderAtEnd(comp_ctx->builder, block_maddr_phi);
+        return maddr_phi;
+    }
+    else
+        return maddr;
 fail:
     return NULL;
 }
@@ -895,6 +1061,7 @@ aot_compile_op_memory_grow(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx)
 
     POP_PAGE_COUNT(delta);
 
+    /* TODO: multi-memory aot_enlarge_memory_with_idx() */
     /* Function type of aot_enlarge_memory() */
     param_types[0] = INT8_PTR_TYPE;
     param_types[1] = I32_TYPE;
@@ -934,7 +1101,7 @@ aot_compile_op_memory_grow(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx)
     }
     else {
         char *func_name = "aot_enlarge_memory";
-        /* AOT mode, delcare the function */
+        /* AOT mode, declare the function */
         if (!(func = LLVMGetNamedFunction(func_ctx->module, func_name))
             && !(func =
                      LLVMAddFunction(func_ctx->module, func_name, func_type))) {
@@ -984,10 +1151,15 @@ check_bulk_memory_overflow(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
                            LLVMValueRef offset, LLVMValueRef bytes)
 {
     LLVMValueRef maddr, max_addr, cmp;
-    LLVMValueRef mem_base_addr;
+    LLVMValueRef mem_base_addr, maddr_phi = NULL;
     LLVMBasicBlockRef block_curr = LLVMGetInsertBlock(comp_ctx->builder);
-    LLVMBasicBlockRef check_succ;
+    LLVMBasicBlockRef check_succ, block_maddr_phi = NULL;
     LLVMValueRef mem_size;
+#if WASM_ENABLE_MEMORY64 == 0
+    bool is_memory64 = false;
+#else
+    bool is_memory64 = IS_MEMORY64;
+#endif
 
     /* Get memory base address and memory data size */
 #if WASM_ENABLE_SHARED_MEMORY != 0
@@ -1012,7 +1184,7 @@ check_bulk_memory_overflow(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
      * Note: not throw the integer-overflow-exception here since it must
      * have been thrown when converting float to integer before
      */
-    /* return addres directly if constant offset and inside memory space */
+    /* return address directly if constant offset and inside memory space */
     if (LLVMIsEfficientConstInt(offset) && LLVMIsEfficientConstInt(bytes)) {
         uint64 mem_offset = (uint64)LLVMConstIntGetZExtValue(offset);
         uint64 mem_len = (uint64)LLVMConstIntGetZExtValue(bytes);
@@ -1052,9 +1224,110 @@ check_bulk_memory_overflow(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
     offset =
         LLVMBuildZExt(comp_ctx->builder, offset, I64_TYPE, "extend_offset");
     bytes = LLVMBuildZExt(comp_ctx->builder, bytes, I64_TYPE, "extend_len");
+    if (!offset || !bytes) {
+        aot_set_last_error("llvm build zext failed.");
+        goto fail;
+    }
 
     BUILD_OP(Add, offset, bytes, max_addr, "max_addr");
+
+    if (is_memory64 && comp_ctx->enable_bound_check) {
+        /* Check whether integer overflow occurs in offset + addr */
+        LLVMBasicBlockRef check_integer_overflow_end;
+        ADD_BASIC_BLOCK(check_integer_overflow_end,
+                        "check_integer_overflow_end");
+        LLVMMoveBasicBlockAfter(check_integer_overflow_end, block_curr);
+
+        BUILD_ICMP(LLVMIntULT, max_addr, offset, cmp, "cmp");
+        if (!aot_emit_exception(comp_ctx, func_ctx,
+                                EXCE_OUT_OF_BOUNDS_MEMORY_ACCESS, true, cmp,
+                                check_integer_overflow_end)) {
+            goto fail;
+        }
+        SET_BUILD_POS(check_integer_overflow_end);
+    }
+
+    if (comp_ctx->enable_shared_heap /* TODO: && mem_idx == 0 */) {
+        LLVMBasicBlockRef app_addr_in_shared_heap, app_addr_in_linear_mem;
+        LLVMValueRef shared_heap_start_off, shared_heap_check_bound;
+        LLVMValueRef max_offset, cmp1, cmp2, is_in_shared_heap;
+
+        /* Add basic blocks */
+        ADD_BASIC_BLOCK(app_addr_in_shared_heap, "app_addr_in_shared_heap");
+        ADD_BASIC_BLOCK(app_addr_in_linear_mem, "app_addr_in_linear_mem");
+        ADD_BASIC_BLOCK(block_maddr_phi, "maddr_phi");
+
+        LLVMMoveBasicBlockAfter(app_addr_in_shared_heap, block_curr);
+        LLVMMoveBasicBlockAfter(app_addr_in_linear_mem,
+                                app_addr_in_shared_heap);
+        LLVMMoveBasicBlockAfter(block_maddr_phi, check_succ);
+
+        LLVMPositionBuilderAtEnd(comp_ctx->builder, block_maddr_phi);
+        if (!(maddr_phi = LLVMBuildPhi(comp_ctx->builder, INT8_PTR_TYPE,
+                                       "maddr_phi"))) {
+            aot_set_last_error("llvm build phi failed");
+            goto fail;
+        }
+
+        LLVMPositionBuilderAtEnd(comp_ctx->builder, block_curr);
+
+        shared_heap_start_off = func_ctx->shared_heap_start_off;
+        if (comp_ctx->pointer_size == sizeof(uint32)) {
+            if (!(shared_heap_start_off =
+                      LLVMBuildZExt(comp_ctx->builder, shared_heap_start_off,
+                                    I64_TYPE, "shared_heap_start_off_u64"))) {
+                aot_set_last_error("llvm build zext failed");
+                goto fail;
+            }
+        }
+        shared_heap_check_bound =
+            is_memory64 ? I64_CONST(UINT64_MAX) : I64_CONST(UINT32_MAX);
+        CHECK_LLVM_CONST(shared_heap_check_bound);
+
+        /* Check whether the bytes to access are in shared heap */
+        if (!comp_ctx->enable_bound_check) {
+            /* Use IntUGT but not IntUGE to compare, same as the check
+               in aot_check_memory_overflow */
+            BUILD_ICMP(LLVMIntUGT, offset, func_ctx->shared_heap_start_off,
+                       is_in_shared_heap, "is_in_shared_heap");
+        }
+        else {
+            BUILD_ICMP(LLVMIntUGT, offset, func_ctx->shared_heap_start_off,
+                       cmp1, "cmp1");
+            BUILD_OP(Add, max_addr, I64_NEG_ONE, max_offset, "max_offset");
+            BUILD_ICMP(LLVMIntULE, max_offset, shared_heap_check_bound, cmp2,
+                       "cmp2");
+            BUILD_OP(And, cmp1, cmp2, is_in_shared_heap, "is_in_shared_heap");
+        }
+
+        if (!LLVMBuildCondBr(comp_ctx->builder, is_in_shared_heap,
+                             app_addr_in_shared_heap, app_addr_in_linear_mem)) {
+            aot_set_last_error("llvm build cond br failed");
+            goto fail;
+        }
+
+        LLVMPositionBuilderAtEnd(comp_ctx->builder, app_addr_in_shared_heap);
+
+        /* Get native address inside shared heap */
+        if (!(maddr = LLVMBuildInBoundsGEP2(comp_ctx->builder, INT8_TYPE,
+                                            func_ctx->shared_heap_base_addr_adj,
+                                            &offset, 1, "maddr_shared_heap"))) {
+            aot_set_last_error("llvm build inbounds gep failed");
+            goto fail;
+        }
+        LLVMAddIncoming(maddr_phi, &maddr, &app_addr_in_shared_heap, 1);
+
+        if (!LLVMBuildBr(comp_ctx->builder, block_maddr_phi)) {
+            aot_set_last_error("llvm build br failed");
+            goto fail;
+        }
+
+        LLVMPositionBuilderAtEnd(comp_ctx->builder, app_addr_in_linear_mem);
+        block_curr = LLVMGetInsertBlock(comp_ctx->builder);
+    }
+
     BUILD_ICMP(LLVMIntUGT, max_addr, mem_size, cmp, "cmp_max_mem_addr");
+
     if (!aot_emit_exception(comp_ctx, func_ctx,
                             EXCE_OUT_OF_BOUNDS_MEMORY_ACCESS, true, cmp,
                             check_succ)) {
@@ -1067,11 +1340,23 @@ check_bulk_memory_overflow(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
         aot_set_last_error("llvm build add failed.");
         goto fail;
     }
-    return maddr;
+
+    if (comp_ctx->enable_shared_heap /* TODO: && mem_idx == 0 */) {
+        block_curr = LLVMGetInsertBlock(comp_ctx->builder);
+        LLVMAddIncoming(maddr_phi, &maddr, &block_curr, 1);
+        if (!LLVMBuildBr(comp_ctx->builder, block_maddr_phi)) {
+            aot_set_last_error("llvm build br failed");
+            goto fail;
+        }
+        LLVMPositionBuilderAtEnd(comp_ctx->builder, block_maddr_phi);
+        return maddr_phi;
+    }
+    else
+        return maddr;
 fail:
     return NULL;
 }
-#endif /* end of WASM_ENABLE_BULK_MEMORY != 0 or WASM_ENABLE_STRINGREF != 0 */
+#endif /* end of WASM_ENABLE_BULK_MEMORY != 0 || WASM_ENABLE_STRINGREF != 0 */
 
 #if WASM_ENABLE_BULK_MEMORY != 0
 bool

core/iwasm/compilation/aot_emit_numberic.c

@@ -653,15 +653,22 @@ compile_int_sub(AOTCompContext *comp_ctx, LLVMValueRef left, LLVMValueRef right,
 }
 
 static LLVMValueRef
-compile_int_mul(AOTCompContext *comp_ctx, LLVMValueRef left, LLVMValueRef right,
-                bool is_i32)
+compile_int_mul(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
+                LLVMValueRef left, LLVMValueRef right, bool is_i32)
 {
     /* If one of the operands is 0, just return constant 0 */
     if (IS_CONST_ZERO(left) || IS_CONST_ZERO(right))
         return is_i32 ? I32_ZERO : I64_ZERO;
 
     /* Build mul */
-    return LLVMBuildMul(comp_ctx->builder, left, right, "mul");
+    LLVMTypeRef param_types[2];
+    param_types[1] = param_types[0] = is_i32 ? I32_TYPE : I64_TYPE;
+
+    LLVMValueRef res;
+    LLVM_BUILD_OP_OR_INTRINSIC(Mul, left, right, res,
+                               is_i32 ? "i32.mul" : "i64.mul", "mul", false);
+
+    return res;
 }
 
 static bool
@@ -679,8 +686,9 @@ compile_op_int_arithmetic(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
                               "compile int sub fail.");
             return true;
         case INT_MUL:
-            DEF_INT_BINARY_OP(compile_int_mul(comp_ctx, left, right, is_i32),
-                              "compile int mul fail.");
+            DEF_INT_BINARY_OP(
+                compile_int_mul(comp_ctx, func_ctx, left, right, is_i32),
+                "compile int mul fail.");
             return true;
         case INT_DIV_S:
         case INT_DIV_U:
@@ -726,43 +734,57 @@ fail:
 }
 
 static LLVMValueRef
-compile_int_shl(AOTCompContext *comp_ctx, LLVMValueRef left, LLVMValueRef right,
-                bool is_i32)
+compile_int_shl(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
+                LLVMValueRef left, LLVMValueRef right, bool is_i32)
 {
     LLVMValueRef res;
 
     SHIFT_COUNT_MASK;
 
     /* Build shl */
-    LLVM_BUILD_OP(Shl, left, right, res, "shl", NULL);
+    LLVMTypeRef param_types[2];
+    param_types[1] = param_types[0] = is_i32 ? I32_TYPE : I64_TYPE;
+
+    LLVM_BUILD_OP_OR_INTRINSIC(Shl, left, right, res,
+                               is_i32 ? "i32.shl" : "i64.shl", "shl", false);
 
     return res;
 }
 
 static LLVMValueRef
-compile_int_shr_s(AOTCompContext *comp_ctx, LLVMValueRef left,
-                  LLVMValueRef right, bool is_i32)
+compile_int_shr_s(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
+                  LLVMValueRef left, LLVMValueRef right, bool is_i32)
 {
     LLVMValueRef res;
 
     SHIFT_COUNT_MASK;
 
     /* Build shl */
-    LLVM_BUILD_OP(AShr, left, right, res, "shr_s", NULL);
+    LLVMTypeRef param_types[2];
+    param_types[1] = param_types[0] = is_i32 ? I32_TYPE : I64_TYPE;
+
+    LLVM_BUILD_OP_OR_INTRINSIC(AShr, left, right, res,
+                               is_i32 ? "i32.shr_s" : "i64.shr_s", "shr_s",
+                               false);
 
     return res;
 }
 
 static LLVMValueRef
-compile_int_shr_u(AOTCompContext *comp_ctx, LLVMValueRef left,
-                  LLVMValueRef right, bool is_i32)
+compile_int_shr_u(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
+                  LLVMValueRef left, LLVMValueRef right, bool is_i32)
 {
     LLVMValueRef res;
 
     SHIFT_COUNT_MASK;
 
     /* Build shl */
-    LLVM_BUILD_OP(LShr, left, right, res, "shr_u", NULL);
+    LLVMTypeRef param_types[2];
+    param_types[1] = param_types[0] = is_i32 ? I32_TYPE : I64_TYPE;
+
+    LLVM_BUILD_OP_OR_INTRINSIC(LShr, left, right, res,
+                               is_i32 ? "i32.shr_u" : "i64.shr_u", "shr_u",
+                               false);
 
     return res;
 }
@@ -814,16 +836,18 @@ compile_op_int_shift(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
 {
     switch (shift_op) {
         case INT_SHL:
-            DEF_INT_BINARY_OP(compile_int_shl(comp_ctx, left, right, is_i32),
-                              NULL);
+            DEF_INT_BINARY_OP(
+                compile_int_shl(comp_ctx, func_ctx, left, right, is_i32), NULL);
             return true;
         case INT_SHR_S:
-            DEF_INT_BINARY_OP(compile_int_shr_s(comp_ctx, left, right, is_i32),
-                              NULL);
+            DEF_INT_BINARY_OP(
+                compile_int_shr_s(comp_ctx, func_ctx, left, right, is_i32),
+                NULL);
             return true;
         case INT_SHR_U:
-            DEF_INT_BINARY_OP(compile_int_shr_u(comp_ctx, left, right, is_i32),
-                              NULL);
+            DEF_INT_BINARY_OP(
+                compile_int_shr_u(comp_ctx, func_ctx, left, right, is_i32),
+                NULL);
             return true;
         case INT_ROTL:
             DEF_INT_BINARY_OP(

core/iwasm/compilation/aot_emit_table.c

@@ -10,6 +10,78 @@
 #include "aot_emit_gc.h"
 #endif
 
+#if WASM_ENABLE_REF_TYPES != 0 || WASM_ENABLE_GC != 0
+#if WASM_ENABLE_MEMORY64 != 0
+static bool
+zero_extend_u64(AOTCompContext *comp_ctx, LLVMValueRef *value, const char *name)
+{
+    if (comp_ctx->pointer_size == sizeof(uint64)) {
+        /* zero extend to uint64 if the target is 64-bit */
+        *value = LLVMBuildZExt(comp_ctx->builder, *value, I64_TYPE, name);
+        if (!*value) {
+            aot_set_last_error("llvm build zero extend failed.");
+            return false;
+        }
+    }
+    return true;
+}
+#endif
+
+/* check whether a table64 elem idx is greater than UINT32_MAX, if so, throw
+ * exception, otherwise trunc it to uint32 */
+static bool
+check_tbl_elem_idx_and_trunc(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
+                             LLVMValueRef *elem_idx, uint32 tbl_idx)
+{
+#if WASM_ENABLE_MEMORY64 != 0
+    LLVMValueRef u32_max, u32_cmp_result;
+    LLVMBasicBlockRef check_elem_idx_succ;
+
+    if (!IS_TABLE64(tbl_idx)) {
+        return true;
+    }
+
+    /* Check if elem index >= UINT32_MAX */
+    if (!(u32_max = I64_CONST(UINT32_MAX))) {
+        aot_set_last_error("llvm build const failed");
+        goto fail;
+    }
+    if (!(u32_cmp_result =
+              LLVMBuildICmp(comp_ctx->builder, LLVMIntUGE, *elem_idx, u32_max,
+                            "cmp_elem_idx_u32_max"))) {
+        aot_set_last_error("llvm build icmp failed.");
+        goto fail;
+    }
+    if (!(*elem_idx = LLVMBuildTrunc(comp_ctx->builder, *elem_idx, I32_TYPE,
+                                     "elem_idx_i32"))) {
+        aot_set_last_error("llvm build trunc failed.");
+        goto fail;
+    }
+
+    /* Throw exception if elem index >= UINT32_MAX*/
+    if (!(check_elem_idx_succ = LLVMAppendBasicBlockInContext(
+              comp_ctx->context, func_ctx->func, "check_elem_idx_succ"))) {
+        aot_set_last_error("llvm add basic block failed.");
+        goto fail;
+    }
+
+    LLVMMoveBasicBlockAfter(check_elem_idx_succ,
+                            LLVMGetInsertBlock(comp_ctx->builder));
+
+    if (!(aot_emit_exception(comp_ctx, func_ctx,
+                             EXCE_OUT_OF_BOUNDS_TABLE_ACCESS, true,
+                             u32_cmp_result, check_elem_idx_succ)))
+        goto fail;
+
+    return true;
+fail:
+    return false;
+#else
+    return true;
+#endif
+}
+#endif /* WASM_ENABLE_REF_TYPES != 0 || WASM_ENABLE_GC !=0 */
+
 uint64
 get_tbl_inst_offset(const AOTCompContext *comp_ctx,
                     const AOTFuncContext *func_ctx, uint32 tbl_idx)
@@ -158,6 +230,10 @@ aot_check_table_access(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
         goto fail;
     }
 
+    if (!check_tbl_elem_idx_and_trunc(comp_ctx, func_ctx, &elem_idx, tbl_idx)) {
+        goto fail;
+    }
+
     /* Check if (uint32)elem index >= table size */
     if (!(cmp_elem_idx = LLVMBuildICmp(comp_ctx->builder, LLVMIntUGE, elem_idx,
                                        tbl_sz, "cmp_elem_idx"))) {
@@ -192,7 +268,7 @@ aot_compile_op_table_get(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
     LLVMValueRef elem_idx, offset, func_idx;
     LLVMValueRef table_elem_base, table_elem_addr, table_elem;
 
-    POP_I32(elem_idx);
+    POP_TBL_ELEM_IDX(elem_idx);
 
     if (!aot_check_table_access(comp_ctx, func_ctx, tbl_idx, elem_idx)) {
         goto fail;
@@ -289,7 +365,7 @@ aot_compile_op_table_set(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
         }
     }
 
-    POP_I32(elem_idx);
+    POP_TBL_ELEM_IDX(elem_idx);
 
     if (!aot_check_table_access(comp_ctx, func_ctx, tbl_idx, elem_idx)) {
         goto fail;
@@ -388,7 +464,11 @@ aot_compile_op_table_init(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
     /* s */
     POP_I32(param_values[4]);
     /* d */
-    POP_I32(param_values[5]);
+    POP_TBL_ELEM_IDX(param_values[5]);
+    if (!check_tbl_elem_idx_and_trunc(comp_ctx, func_ctx, &param_values[5],
+                                      tbl_idx)) {
+        goto fail;
+    }
 
     /* "" means return void */
     if (!(LLVMBuildCall2(comp_ctx->builder, func_type, func, param_values, 6,
@@ -408,6 +488,7 @@ aot_compile_op_table_copy(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
 {
     LLVMTypeRef param_types[6], ret_type, func_type, func_ptr_type;
     LLVMValueRef func, param_values[6], value;
+    uint32 tbl_idx;
 
     param_types[0] = INT8_PTR_TYPE;
     param_types[1] = I32_TYPE;
@@ -434,12 +515,34 @@ aot_compile_op_table_copy(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
         goto fail;
     }
 
+    /* In table64, the length should be i32 type if any one of src/dst table
+     * is i32 type, set the table index to the lesser-or-equal table when
+     * popping length n */
+    if (!(comp_ctx->comp_data->tables[src_tbl_idx].table_type.flags
+          & TABLE64_FLAG))
+        tbl_idx = src_tbl_idx;
+    else
+        tbl_idx = dst_tbl_idx;
     /* n */
-    POP_I32(param_values[3]);
+    POP_TBL_ELEM_LEN(param_values[3]);
+    if (!check_tbl_elem_idx_and_trunc(comp_ctx, func_ctx, &param_values[3],
+                                      tbl_idx)) {
+        goto fail;
+    }
     /* s */
-    POP_I32(param_values[4]);
+    tbl_idx = src_tbl_idx;
+    POP_TBL_ELEM_IDX(param_values[4]);
+    if (!check_tbl_elem_idx_and_trunc(comp_ctx, func_ctx, &param_values[4],
+                                      tbl_idx)) {
+        goto fail;
+    }
     /* d */
-    POP_I32(param_values[5]);
+    tbl_idx = dst_tbl_idx;
+    POP_TBL_ELEM_IDX(param_values[5]);
+    if (!check_tbl_elem_idx_and_trunc(comp_ctx, func_ctx, &param_values[5],
+                                      tbl_idx)) {
+        goto fail;
+    }
 
     /* "" means return void */
     if (!(LLVMBuildCall2(comp_ctx->builder, func_type, func, param_values, 6,
@@ -484,7 +587,14 @@ aot_compile_op_table_size(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
         goto fail;
     }
 
-    PUSH_I32(tbl_sz);
+#if WASM_ENABLE_MEMORY64 != 0
+    if (IS_TABLE64(tbl_idx)) {
+        if (!zero_extend_u64(comp_ctx, &tbl_sz, "length64")) {
+            goto fail;
+        }
+    }
+#endif
+    PUSH_TBL_ELEM_IDX(tbl_sz);
 
     return true;
 fail:
@@ -517,7 +627,11 @@ aot_compile_op_table_grow(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
     }
 
     /* n */
-    POP_I32(param_values[2]);
+    POP_TBL_ELEM_LEN(param_values[2]);
+    if (!check_tbl_elem_idx_and_trunc(comp_ctx, func_ctx, &param_values[2],
+                                      tbl_idx)) {
+        goto fail;
+    }
     /* v */
 
     if (comp_ctx->enable_gc) {
@@ -545,7 +659,14 @@ aot_compile_op_table_grow(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
         goto fail;
     }
 
-    PUSH_I32(ret);
+#if WASM_ENABLE_MEMORY64 != 0
+    if (IS_TABLE64(tbl_idx)) {
+        if (!zero_extend_u64(comp_ctx, &ret, "table_size64")) {
+            goto fail;
+        }
+    }
+#endif
+    PUSH_TBL_ELEM_LEN(ret);
 
     return true;
 fail:
@@ -579,7 +700,11 @@ aot_compile_op_table_fill(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
     }
 
     /* n */
-    POP_I32(param_values[2]);
+    POP_TBL_ELEM_LEN(param_values[2]);
+    if (!check_tbl_elem_idx_and_trunc(comp_ctx, func_ctx, &param_values[2],
+                                      tbl_idx)) {
+        goto fail;
+    }
     /* v */
 
     if (comp_ctx->enable_gc) {
@@ -601,7 +726,11 @@ aot_compile_op_table_fill(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
         }
     }
     /* i */
-    POP_I32(param_values[4]);
+    POP_TBL_ELEM_IDX(param_values[4]);
+    if (!check_tbl_elem_idx_and_trunc(comp_ctx, func_ctx, &param_values[4],
+                                      tbl_idx)) {
+        goto fail;
+    }
 
     /* "" means return void */
     if (!(LLVMBuildCall2(comp_ctx->builder, func_type, func, param_values, 5,
@@ -615,4 +744,4 @@ fail:
     return false;
 }
 
-#endif /*  WASM_ENABLE_REF_TYPES != 0 || WASM_ENABLE_GC !=0 */
+#endif /* WASM_ENABLE_REF_TYPES != 0 || WASM_ENABLE_GC !=0 */

core/iwasm/compilation/aot_llvm.c

@@ -711,8 +711,7 @@ aot_add_llvm_func(AOTCompContext *comp_ctx, LLVMModuleRef module,
                                     prefix)))
         goto fail;
 
-    if (comp_ctx->is_indirect_mode) {
-        /* avoid LUT relocations ("switch-table") */
+    if (comp_ctx->disable_llvm_jump_tables) {
         LLVMAttributeRef attr_no_jump_tables = LLVMCreateStringAttribute(
             comp_ctx->context, "no-jump-tables",
             (uint32)strlen("no-jump-tables"), "true", (uint32)strlen("true"));
@@ -1518,6 +1517,75 @@ create_memory_info(const AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
     return true;
 }
 
+static bool
+create_shared_heap_info(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx)
+{
+    LLVMValueRef offset, base_addr_p, start_off_p, cmp;
+    uint32 offset_u32;
+
+    /* Load aot_inst->e->shared_heap_base_addr_adj */
+    offset_u32 = get_module_inst_extra_offset(comp_ctx);
+#if WASM_ENABLE_JIT != 0 && WASM_ENABLE_SHARED_HEAP != 0
+    if (comp_ctx->is_jit_mode)
+        offset_u32 +=
+            offsetof(WASMModuleInstanceExtra, shared_heap_base_addr_adj);
+    else
+#endif
+        offset_u32 +=
+            offsetof(AOTModuleInstanceExtra, shared_heap_base_addr_adj);
+    offset = I32_CONST(offset_u32);
+    CHECK_LLVM_CONST(offset);
+
+    if (!(base_addr_p = LLVMBuildInBoundsGEP2(comp_ctx->builder, INT8_TYPE,
+                                              func_ctx->aot_inst, &offset, 1,
+                                              "shared_heap_base_addr_adj_p"))) {
+        aot_set_last_error("llvm build inbounds gep failed");
+        return false;
+    }
+    if (!(func_ctx->shared_heap_base_addr_adj =
+              LLVMBuildLoad2(comp_ctx->builder, INT8_PTR_TYPE, base_addr_p,
+                             "shared_heap_base_addr_adj"))) {
+        aot_set_last_error("llvm build load failed");
+        return false;
+    }
+
+    /* Load aot_inst->e->shared_heap_start_off */
+    offset_u32 = get_module_inst_extra_offset(comp_ctx);
+#if WASM_ENABLE_JIT != 0 && WASM_ENABLE_SHARED_HEAP != 0
+    if (comp_ctx->is_jit_mode)
+        offset_u32 += offsetof(WASMModuleInstanceExtra, shared_heap_start_off);
+    else
+#endif
+        offset_u32 += offsetof(AOTModuleInstanceExtra, shared_heap_start_off);
+    offset = I32_CONST(offset_u32);
+    CHECK_LLVM_CONST(offset);
+
+    if (!(start_off_p = LLVMBuildInBoundsGEP2(comp_ctx->builder, INT8_TYPE,
+                                              func_ctx->aot_inst, &offset, 1,
+                                              "shared_heap_start_off_p"))) {
+        aot_set_last_error("llvm build inbounds gep failed");
+        return false;
+    }
+    if (!(func_ctx->shared_heap_start_off = LLVMBuildLoad2(
+              comp_ctx->builder,
+              comp_ctx->pointer_size == sizeof(uint64) ? I64_TYPE : I32_TYPE,
+              start_off_p, "shared_heap_start_off"))) {
+        aot_set_last_error("llvm build load failed");
+        return false;
+    }
+
+    if (!(cmp = LLVMBuildIsNotNull(comp_ctx->builder,
+                                   func_ctx->shared_heap_base_addr_adj,
+                                   "has_shared_heap"))) {
+        aot_set_last_error("llvm build is not null failed");
+        return false;
+    }
+
+    return true;
+fail:
+    return false;
+}
+
 static bool
 create_cur_exception(const AOTCompContext *comp_ctx, AOTFuncContext *func_ctx)
 {
@@ -1651,7 +1719,7 @@ aot_create_stack_sizes(const AOTCompData *comp_data, AOTCompContext *comp_ctx)
          * This value is a placeholder, which will be replaced
          * after the corresponding functions are compiled.
          *
-         * Don't use zeros becasue LLVM can optimize them to
+         * Don't use zeros because LLVM can optimize them to
          * zeroinitializer.
          */
         values[i] = I32_NEG_ONE;
@@ -1771,7 +1839,7 @@ aot_create_func_context(const AOTCompData *comp_data, AOTCompContext *comp_ctx,
         goto fail;
     }
 
-    if (comp_ctx->enable_aux_stack_frame
+    if (comp_ctx->aux_stack_frame_type
         && !create_aux_stack_frame(comp_ctx, func_ctx)) {
         goto fail;
     }
@@ -1808,6 +1876,12 @@ aot_create_func_context(const AOTCompData *comp_data, AOTCompContext *comp_ctx,
         goto fail;
     }
 
+    /* Load shared heap, shared heap start off mem32 or mem64 */
+    if (comp_ctx->enable_shared_heap
+        && !create_shared_heap_info(comp_ctx, func_ctx)) {
+        goto fail;
+    }
+
     return func_ctx;
 
 fail:
@@ -2279,7 +2353,7 @@ create_target_machine_detect_host(AOTCompContext *comp_ctx)
     }
 
     if (!LLVMTargetHasJIT(target)) {
-        aot_set_last_error("unspported JIT on this platform.");
+        aot_set_last_error("unsupported JIT on this platform.");
         goto fail;
     }
 
@@ -2446,7 +2520,8 @@ aot_compiler_init(void)
     LLVMInitializeCore(LLVMGetGlobalPassRegistry());
 #endif
 
-#if WASM_ENABLE_WAMR_COMPILER != 0
+/* fuzzing only use host targets for simple */
+#if WASM_ENABLE_WAMR_COMPILER != 0 && WASM_ENABLE_FUZZ_TEST == 0
     /* Init environment of all targets for AOT compiler */
     LLVMInitializeAllTargetInfos();
     LLVMInitializeAllTargets();
@@ -2577,8 +2652,8 @@ aot_create_comp_context(const AOTCompData *comp_data, aot_comp_option_t option)
     if (option->enable_ref_types)
         comp_ctx->enable_ref_types = true;
 
-    if (option->enable_aux_stack_frame)
-        comp_ctx->enable_aux_stack_frame = true;
+    comp_ctx->aux_stack_frame_type = option->aux_stack_frame_type;
+    comp_ctx->call_stack_features = option->call_stack_features;
 
     if (option->enable_perf_profiling)
         comp_ctx->enable_perf_profiling = true;
@@ -2589,12 +2664,18 @@ aot_create_comp_context(const AOTCompData *comp_data, aot_comp_option_t option)
     if (option->enable_aux_stack_check)
         comp_ctx->enable_aux_stack_check = true;
 
-    if (option->is_indirect_mode)
+    if (option->is_indirect_mode) {
         comp_ctx->is_indirect_mode = true;
+        /* avoid LUT relocations ("switch-table") */
+        comp_ctx->disable_llvm_jump_tables = true;
+    }
 
     if (option->disable_llvm_intrinsics)
         comp_ctx->disable_llvm_intrinsics = true;
 
+    if (option->disable_llvm_jump_tables)
+        comp_ctx->disable_llvm_jump_tables = true;
+
     if (option->disable_llvm_lto)
         comp_ctx->disable_llvm_lto = true;
 
@@ -2619,6 +2700,9 @@ aot_create_comp_context(const AOTCompData *comp_data, aot_comp_option_t option)
     if (option->enable_gc)
         comp_ctx->enable_gc = true;
 
+    if (option->enable_shared_heap)
+        comp_ctx->enable_shared_heap = true;
+
     comp_ctx->opt_level = option->opt_level;
     comp_ctx->size_level = option->size_level;
 
@@ -2658,20 +2742,29 @@ aot_create_comp_context(const AOTCompData *comp_data, aot_comp_option_t option)
     }
     else {
         /* Create LLVM target machine */
-        arch = option->target_arch;
-        abi = option->target_abi;
-        cpu = option->target_cpu;
-        features = option->cpu_features;
+        if (!option->target_arch || !strstr(option->target_arch, "-")) {
+            /* Retrieve the target triple based on user input */
+            triple = NULL;
+            arch = option->target_arch;
+            abi = option->target_abi;
+            cpu = option->target_cpu;
+            features = option->cpu_features;
+        }
+        else {
+            /* Form a target triple */
+            triple = option->target_arch;
+            arch = NULL;
+            abi = NULL;
+            cpu = NULL;
+            features = NULL;
+        }
+
         opt_level = option->opt_level;
         size_level = option->size_level;
 
         /* verify external llc compiler */
         comp_ctx->external_llc_compiler = getenv("WAMRC_LLC_COMPILER");
         if (comp_ctx->external_llc_compiler) {
-#if defined(_WIN32) || defined(_WIN32_)
-            comp_ctx->external_llc_compiler = NULL;
-            LOG_WARNING("External LLC compiler not supported on Windows.");
-#else
             if (access(comp_ctx->external_llc_compiler, X_OK) != 0) {
                 LOG_WARNING("WAMRC_LLC_COMPILER [%s] not found, fallback to "
                             "default pipeline",
@@ -2683,17 +2776,12 @@ aot_create_comp_context(const AOTCompData *comp_data, aot_comp_option_t option)
                 LOG_VERBOSE("Using external LLC compiler [%s]",
                             comp_ctx->external_llc_compiler);
             }
-#endif
         }
 
         /* verify external asm compiler */
         if (!comp_ctx->external_llc_compiler) {
             comp_ctx->external_asm_compiler = getenv("WAMRC_ASM_COMPILER");
             if (comp_ctx->external_asm_compiler) {
-#if defined(_WIN32) || defined(_WIN32_)
-                comp_ctx->external_asm_compiler = NULL;
-                LOG_WARNING("External ASM compiler not supported on Windows.");
-#else
                 if (access(comp_ctx->external_asm_compiler, X_OK) != 0) {
                     LOG_WARNING(
                         "WAMRC_ASM_COMPILER [%s] not found, fallback to "
@@ -2706,7 +2794,6 @@ aot_create_comp_context(const AOTCompData *comp_data, aot_comp_option_t option)
                     LOG_VERBOSE("Using external ASM compiler [%s]",
                                 comp_ctx->external_asm_compiler);
                 }
-#endif
             }
         }
 
@@ -2790,6 +2877,15 @@ aot_create_comp_context(const AOTCompData *comp_data, aot_comp_option_t option)
                 bh_assert(vendor_sys);
                 bh_memcpy_s(default_arch, sizeof(default_arch), default_triple,
                             (uint32)(vendor_sys - default_triple));
+                /**
+                 * On Mac M[1-9]+ LLVM will report arm64 as the
+                 * architecture, for the purposes of wamr this is the
+                 * same as aarch64v8 so we'll normalize it here.
+                 */
+                if (!strcmp(default_arch, "arm64")) {
+                    bh_strcpy_s(default_arch, sizeof(default_arch),
+                                "aarch64v8");
+                }
                 arch1 = default_arch;
 
                 LLVMDisposeMessage(default_triple);
@@ -2903,6 +2999,7 @@ aot_create_comp_context(const AOTCompData *comp_data, aot_comp_option_t option)
                 aot_set_last_error(buf);
                 goto fail;
             }
+            LOG_VERBOSE("triple: %s => normailized: %s", triple, triple_norm);
             if (!cpu)
                 cpu = "";
         }
@@ -2960,12 +3057,12 @@ aot_create_comp_context(const AOTCompData *comp_data, aot_comp_option_t option)
                                     sizeof(comp_ctx->target_arch));
 
         if (option->bounds_checks == 1 || option->bounds_checks == 0) {
-            /* Set by user */
+            /* Set by the user */
             comp_ctx->enable_bound_check =
                 (option->bounds_checks == 1) ? true : false;
         }
         else {
-            /* Unset by user, use default value */
+            /* Unset by the user, use the default value */
             if (strstr(comp_ctx->target_arch, "64")
                 && !option->is_sgx_platform) {
                 comp_ctx->enable_bound_check = false;
@@ -2975,17 +3072,17 @@ aot_create_comp_context(const AOTCompData *comp_data, aot_comp_option_t option)
             }
         }
 
-        if (comp_ctx->enable_bound_check) {
-            /* Always enable stack boundary check if `bounds-checks`
-               is enabled */
-            comp_ctx->enable_stack_bound_check = true;
-        }
-        else {
-            /* When `bounds-checks` is disabled, we set stack boundary
-               check status according to the input option */
+        if (option->stack_bounds_checks == 1
+            || option->stack_bounds_checks == 0) {
+            /* Set by the user */
             comp_ctx->enable_stack_bound_check =
                 (option->stack_bounds_checks == 1) ? true : false;
         }
+        else {
+            /* Unset by the user, use the default value, it will be the same
+             * value as the bound check */
+            comp_ctx->enable_stack_bound_check = comp_ctx->enable_bound_check;
+        }
 
         if ((comp_ctx->enable_stack_bound_check
              || comp_ctx->enable_stack_estimation)
@@ -3257,6 +3354,11 @@ aot_destroy_comp_context(AOTCompContext *comp_ctx)
     if (comp_ctx->builder)
         LLVMDisposeBuilder(comp_ctx->builder);
 
+#if WASM_ENABLE_DEBUG_AOT != 0
+    if (comp_ctx->debug_builder)
+        LLVMDisposeDIBuilder(comp_ctx->debug_builder);
+#endif
+
     if (comp_ctx->orc_thread_safe_context)
         LLVMOrcDisposeThreadSafeContext(comp_ctx->orc_thread_safe_context);
 
@@ -3330,7 +3432,7 @@ aot_get_native_symbol_index(AOTCompContext *comp_ctx, const char *symbol)
 
     sym = bh_list_first_elem(&comp_ctx->native_symbols);
 
-    /* Lookup an existing symobl record */
+    /* Lookup an existing symbol record */
 
     while (sym) {
         if (strcmp(sym->symbol, symbol) == 0) {

core/iwasm/compilation/aot_llvm.h

@@ -37,6 +37,18 @@
 #include "aot_orc_extra.h"
 #include "aot_comp_option.h"
 
+#if defined(_WIN32) || defined(_WIN32_)
+#include <io.h>
+#define access _access
+/* On windows there is no X_OK flag to check for executablity, only check for
+ * existence */
+#ifdef X_OK
+#undef X_OK
+#endif
+#define X_OK 00
+#define unlink _unlink
+#endif
+
 #ifdef __cplusplus
 extern "C" {
 #endif
@@ -242,6 +254,9 @@ typedef struct AOTFuncContext {
     bool mem_space_unchanged;
     AOTCheckedAddrList checked_addr_list;
 
+    LLVMValueRef shared_heap_base_addr_adj;
+    LLVMValueRef shared_heap_start_off;
+
     LLVMBasicBlockRef got_exception_block;
     LLVMBasicBlockRef func_return_block;
     LLVMValueRef exception_id_phi;
@@ -410,7 +425,10 @@ typedef struct AOTCompContext {
     bool enable_aux_stack_check;
 
     /* Generate auxiliary stack frame */
-    bool enable_aux_stack_frame;
+    AOTStackFrameType aux_stack_frame_type;
+
+    /* Auxiliary call stack features */
+    AOTCallStackFeatures call_stack_features;
 
     /* Function performance profiling */
     bool enable_perf_profiling;
@@ -430,6 +448,9 @@ typedef struct AOTCompContext {
     /* Disable LLVM built-in intrinsics */
     bool disable_llvm_intrinsics;
 
+    /* Disable LLVM jump tables */
+    bool disable_llvm_jump_tables;
+
     /* Disable LLVM link time optimization */
     bool disable_llvm_lto;
 
@@ -464,6 +485,8 @@ typedef struct AOTCompContext {
     /* Enable GC */
     bool enable_gc;
 
+    bool enable_shared_heap;
+
     uint32 opt_level;
     uint32 size_level;
 

core/iwasm/compilation/aot_llvm_extra.cpp

@@ -318,10 +318,15 @@ aot_apply_llvm_new_pass_manager(AOTCompContext *comp_ctx, LLVMModuleRef module)
     ModulePassManager MPM;
 
     if (comp_ctx->is_jit_mode) {
+#if LLVM_VERSION_MAJOR >= 18
+#define INSTCOMBINE "instcombine<no-verify-fixpoint>"
+#else
+#define INSTCOMBINE "instcombine"
+#endif
         const char *Passes =
             "loop-vectorize,slp-vectorizer,"
             "load-store-vectorizer,vector-combine,"
-            "mem2reg,instcombine,simplifycfg,jump-threading,indvars";
+            "mem2reg," INSTCOMBINE ",simplifycfg,jump-threading,indvars";
         ExitOnErr(PB.parsePassPipeline(MPM, Passes));
     }
     else {

core/iwasm/compilation/aot_llvm_extra2.cpp

@@ -159,6 +159,17 @@ LLVMCreateTargetMachineWithOpts(LLVMTargetRef ctarget, const char *triple,
     auto cm = convert(code_model, &jit);
     auto targetmachine = target->createTargetMachine(triple, cpu, features,
                                                      opts, rm, cm, ol, jit);
+#if LLVM_VERSION_MAJOR >= 18
+    // always place data in normal data section.
+    //
+    // note that:
+    // - our aot file emitter/loader doesn't support x86-64 large data
+    //   sections. (eg .lrodata)
+    // - for our purposes, "data" is usually something the compiler
+    //   generated. (eg. jump tables) we probably never benefit from
+    //   large data sections.
+    targetmachine->setLargeDataThreshold(UINT64_MAX);
+#endif
     return reinterpret_cast<LLVMTargetMachineRef>(targetmachine);
 }