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/workflows/build_llvm_libraries.yml

@@ -65,6 +65,7 @@ jobs:
         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
       # enforce a clean build, in the unlikely case that some
@@ -88,14 +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

.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

@@ -53,7 +53,7 @@ jobs:
 
     # Initializes the CodeQL tools for scanning.
     - name: Initialize CodeQL
-      uses: github/codeql-action/init@v3.26.13
+      uses: github/codeql-action/init@v3.28.17
       with:
         languages: ${{ matrix.language }}
 
@@ -70,7 +70,7 @@ jobs:
     - run: |
         ./.github/scripts/codeql_buildscript.sh
     - name: Perform CodeQL Analysis
-      uses: github/codeql-action/analyze@v3.26.13
+      uses: github/codeql-action/analyze@v3.28.17
       with:
         category: "/language:${{matrix.language}}"
         upload: false
@@ -99,14 +99,14 @@ jobs:
         output: ${{ steps.step1.outputs.sarif-output }}/cpp.sarif
 
     - name: Upload CodeQL results to code scanning
-      uses: github/codeql-action/upload-sarif@v3.26.13
+      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@v4.4.3
+      uses: actions/upload-artifact@v4.6.2
       with:
         name: codeql-results
         path: ${{ steps.step1.outputs.sarif-output }}

.github/workflows/coding_guidelines.yml

@@ -19,7 +19,7 @@ permissions:
 
 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

@@ -74,6 +74,12 @@ 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
@@ -152,10 +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]
@@ -171,11 +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"
@@ -253,6 +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 }}
@@ -306,17 +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
@@ -337,19 +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: |
@@ -388,14 +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 }}
@@ -421,12 +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'))
@@ -455,34 +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
@@ -494,12 +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
@@ -625,10 +595,6 @@ jobs:
             $MEMORY64_TEST_OPTIONS,
             $MULTI_MEMORY_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-22.04
             llvm_cache_key: ${{ needs.build_llvm_libraries_on_ubuntu_2204.outputs.cache_key }}
@@ -639,11 +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
@@ -671,7 +635,7 @@ 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 
+          # 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"
@@ -699,11 +663,9 @@ jobs:
 
       - 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
@@ -813,7 +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
@@ -828,7 +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

@@ -228,24 +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: |
@@ -260,14 +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 }}
@@ -277,19 +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: |
@@ -356,12 +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

.github/workflows/compilation_on_nuttx.yml

@@ -88,14 +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

.github/workflows/compilation_on_sgx.yml

@@ -49,10 +49,14 @@ 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
@@ -64,7 +68,7 @@ jobs:
       actions: write
     uses: ./.github/workflows/build_llvm_libraries.yml
     with:
-      os: "ubuntu-20.04"
+      os: ubuntu-22.04
       arch: "X86"
 
   build_iwasm:
@@ -97,12 +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
@@ -110,22 +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: ${{ matrix.os }}
+
       - name: Build iwasm
         run: |
           mkdir build && cd build
@@ -150,69 +146,29 @@ jobs:
             #$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: |
@@ -240,59 +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
@@ -316,19 +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/nightly_run.yml

@@ -48,15 +48,7 @@ permissions:
   contents: read
 
 jobs:
-  build_llvm_libraries_on_ubuntu_2004:
-    permissions:
-      contents: read
-      actions: write
-    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
@@ -66,16 +58,13 @@ jobs:
       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
@@ -106,10 +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:
@@ -142,8 +128,9 @@ jobs:
             "-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
@@ -240,8 +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
@@ -374,12 +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
@@ -395,18 +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
@@ -431,12 +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: |
@@ -456,44 +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
@@ -590,17 +542,21 @@ 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]
-        run: |
-          ./build.sh -p host-interp
-          python3 ./sample_test_run.py $(pwd)/out
-          exit $?
-        working-directory: ./wamr-app-framework/samples/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: |
@@ -612,18 +568,12 @@ jobs:
           ./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:
           [
@@ -642,22 +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
@@ -699,13 +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

@@ -58,16 +58,6 @@ jobs:
 
   #
   # LLVM_LIBRARIES
-  build_llvm_libraries_on_ubuntu_2004:
-    permissions:
-      contents: read
-      actions: write
-    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
@@ -100,18 +90,6 @@ jobs:
 
   #
   # WAMRC
-  release_wamrc_on_ubuntu_2004:
-    permissions:
-      contents: write # upload release artifact
-    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
@@ -150,18 +128,6 @@ jobs:
 
   #
   # IWASM
-  release_iwasm_on_ubuntu_2004:
-    permissions:
-      contents: write # upload release artifact
-    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
@@ -200,19 +166,6 @@ jobs:
 
   #
   # WAMR_SDK
-  release_wamr_sdk_on_ubuntu_2004:
-    permissions:
-      contents: write # upload release artifact
-    needs: [create_tag, create_release]
-    uses: ./.github/workflows/build_wamr_sdk.yml
-    with:
-      config_file: wamr_config_ubuntu_release.cmake
-      runner: ubuntu-20.04
-      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
-
   release_wamr_sdk_on_ubuntu_2204:
     permissions:
       contents: write # upload release artifact
@@ -264,16 +217,6 @@ jobs:
 
   #
   # WAMR_LLDB
-  release_wamr_lldb_on_ubuntu_2004:
-    permissions:
-      contents: write # upload release artifact
-    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

.github/workflows/spec_test_on_nuttx.yml

@@ -127,7 +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
@@ -145,14 +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
@@ -182,33 +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
@@ -350,7 +329,7 @@ jobs:
 
       - name: upload the log
         if: always()
-        uses: actions/upload-artifact@v4.4.3
+        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

@@ -39,7 +39,7 @@ jobs:
           persist-credentials: false
 
       - name: "Run analysis"
-        uses: ossf/scorecard-action@62b2cac7ed8198b15735ed49ab1e5cf35480ba46 # v2.4.0
+        uses: ossf/scorecard-action@f49aabe0b5af0936a0987cfb85d86b75731b0186 # v2.4.1
         with:
           results_file: results.sarif
           results_format: sarif
@@ -52,7 +52,7 @@ jobs:
       # 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@184d73b71b93c222403b2e7f1ffebe4508014249 # v3.1.0
+        uses: actions/upload-artifact@ea165f8d65b6e75b540449e92b4886f43607fa02 # v3.1.0
         with:
           name: SARIF file
           path: results.sarif
@@ -60,6 +60,6 @@ jobs:
 
       # Upload the results to GitHub's code scanning dashboard.
       - name: "Upload to code-scanning"
-        uses: github/codeql-action/upload-sarif@af56b044b5d41c317aef5d19920b3183cb4fbbec # v2.2.4
+        uses: github/codeql-action/upload-sarif@5eb3ed6614230b1931d5c08df9e096e4ba524f21 
         with:
           sarif_file: results.sarif

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)
@@ -113,8 +115,8 @@ 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})
@@ -154,52 +156,40 @@ if (MSVC)
   add_definitions(-DCOMPILING_WASM_RUNTIME_API=1)
 endif ()
 
-# 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 ()
+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_static PRIVATE ws2_32)
-    endif ()
-
-    if (WIN32)
-      target_link_libraries(iwasm_static PRIVATE ntdll)
-    endif()
-
-    install (TARGETS iwasm_static ARCHIVE 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 ()
 
-# 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 ()
-
-    if (MINGW)
-      target_link_libraries(iwasm_shared INTERFACE -lWs2_32 -lwsock32)
-      target_link_libraries(iwasm_shared PRIVATE ws2_32)
-    endif ()
-
-    if (WIN32)
-      target_link_libraries(iwasm_shared PRIVATE ntdll)
-    endif()
-
-    install (TARGETS iwasm_shared LIBRARY DESTINATION lib)
+if (MINGW)
+  target_link_libraries(vmlib INTERFACE -lWs2_32 -lwsock32)
+  target_link_libraries(vmlib PRIVATE ws2_32)
 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 (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

@@ -22,7 +22,7 @@
 # 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 @wenyongh @xujuntwt95329 @yamt
+* @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

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

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 ""
     ]
@@ -202,11 +211,11 @@ def repackage_llvm_windows(llvm_dir):
     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 
+    # mv package dir outside of build
     shutil.move(str(windows_package_dir), str(llvm_dir))
     # rm -r build
     shutil.rmtree(str(build_dir))
@@ -216,7 +225,7 @@ def repackage_llvm_windows(llvm_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()  
+    moved_package_dir.rmdir()
 
 def main():
     parser = argparse.ArgumentParser(description="build necessary LLVM libraries")
@@ -240,6 +249,7 @@ def main():
             "X86",
             "Xtensa",
         ],
+        default=[],
         help="identify LLVM supported backends, separate by space, like '--arch ARM Mips X86'",
     )
     parser.add_argument(

build-scripts/config_common.cmake

@@ -131,6 +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)
@@ -151,6 +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()
@@ -162,16 +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 ()
@@ -202,6 +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 ()
@@ -214,64 +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_SHARED_HEAP EQUAL 1)
-  add_definitions (-DWASM_ENABLE_SHARED_HEAP=1)
-  message ("     Shared heap enabled")
-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_MULTI_MEMORY EQUAL 1)
-  add_definitions (-DWASM_ENABLE_MULTI_MEMORY=1)
-  message ("     Multi memory enabled")
-  set (WAMR_BUILD_DEBUG_INTERP 0)
-endif ()
 if (WAMR_BUILD_THREAD_MGR EQUAL 1)
   message ("     Thread manager enabled")
 endif ()
@@ -290,6 +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")
@@ -317,12 +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)
@@ -349,16 +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()
@@ -370,7 +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 ()
@@ -599,3 +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 ()
@@ -155,6 +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 ()

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

@@ -84,7 +84,7 @@
 #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
@@ -193,6 +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
@@ -318,6 +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
@@ -698,4 +708,12 @@
 #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

@@ -10,6 +10,9 @@
 #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"
@@ -304,18 +307,22 @@ loader_mmap(uint32 size, bool prot_exec, char *error_buf, uint32 error_buf_size)
 #if defined(BUILD_TARGET_X86_64) || defined(BUILD_TARGET_AMD_64) \
     || defined(BUILD_TARGET_RISCV64_LP64D)                       \
     || defined(BUILD_TARGET_RISCV64_LP64)
-#ifndef __APPLE__
+#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()))) {
-        /* The mmapped memory must be in the first 2 Gigabytes of the
+        /* Test whether the mmapped memory in the first 2 Gigabytes of the
            process address space */
-        bh_assert((uintptr_t)mem < INT32_MAX);
+        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
@@ -583,7 +590,8 @@ load_target_info_section(const uint8 *buf, const uint8 *buf_end,
     }
 
     /* for backwards compatibility with previous wamrc aot files */
-    if (!strcmp(target_info.arch, "arm64"))
+    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 */
@@ -966,13 +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);
 }
 
@@ -981,6 +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,
@@ -1012,8 +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;
         }
 
@@ -1025,8 +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;
@@ -1035,6 +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)
@@ -1044,9 +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;
@@ -1189,6 +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) {
@@ -1332,7 +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
@@ -1392,7 +1464,12 @@ load_table_list(const uint8 **p_buf, const uint8 *buf_end, AOTModule *module,
         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
@@ -1480,7 +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;
         }
 
@@ -1638,7 +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 */
@@ -3112,10 +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)
@@ -4235,6 +4314,16 @@ 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, bool no_resolve, char *error_buf,
@@ -4253,7 +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;
     }
@@ -4321,6 +4410,13 @@ aot_load_from_aot_file(const uint8 *buf, uint32 size, const LoadArgs *args,
     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;
 }
@@ -4335,7 +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)

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),     \

core/iwasm/aot/aot_runtime.c

@@ -689,20 +689,23 @@ tables_instantiate(AOTModuleInstance *module_inst, AOTModule *module,
             tbl_inst->cur_size = import_table->table_type.init_size;
             tbl_inst->max_size =
                 aot_get_imp_tbl_data_slots(import_table, false);
-            tbl_inst->elem_type = module->tables[i].table_type.elem_type;
+            tbl_inst->elem_type = import_table->table_type.elem_type;
+            tbl_inst->is_table64 =
+                import_table->table_type.flags & TABLE64_FLAG;
 #if WASM_ENABLE_GC != 0
             tbl_inst->elem_ref_type.elem_ref_type =
-                module->tables[i].table_type.elem_ref_type;
+                import_table->table_type.elem_ref_type;
 #endif
         }
         else {
             AOTTable *table = module->tables + (i - module->import_table_count);
             tbl_inst->cur_size = table->table_type.init_size;
             tbl_inst->max_size = aot_get_tbl_data_slots(table, false);
-            tbl_inst->elem_type = module->tables[i].table_type.elem_type;
+            tbl_inst->elem_type = table->table_type.elem_type;
+            tbl_inst->is_table64 = table->table_type.flags & TABLE64_FLAG;
 #if WASM_ENABLE_GC != 0
             tbl_inst->elem_ref_type.elem_ref_type =
-                module->tables[i].table_type.elem_ref_type;
+                table->table_type.elem_ref_type;
 #endif
         }
 
@@ -1096,11 +1099,11 @@ aot_get_default_memory(AOTModuleInstance *module_inst)
 }
 
 AOTMemoryInstance *
-aot_get_memory_with_index(AOTModuleInstance *module_inst, uint32 index)
+aot_get_memory_with_idx(AOTModuleInstance *module_inst, uint32 mem_idx)
 {
-    if ((index >= module_inst->memory_count) || !module_inst->memories)
+    if ((mem_idx >= module_inst->memory_count) || !module_inst->memories)
         return NULL;
-    return module_inst->memories[index];
+    return module_inst->memories[mem_idx];
 }
 
 static bool
@@ -1282,20 +1285,77 @@ init_func_ptrs(AOTModuleInstance *module_inst, AOTModule *module,
     return true;
 }
 
+static int
+cmp_export_func_map(const void *a, const void *b)
+{
+    uint32 func_idx1 = ((const ExportFuncMap *)a)->func_idx;
+    uint32 func_idx2 = ((const ExportFuncMap *)b)->func_idx;
+    return func_idx1 < func_idx2 ? -1 : (func_idx1 > func_idx2 ? 1 : 0);
+}
+
+AOTFunctionInstance *
+aot_lookup_function_with_idx(AOTModuleInstance *module_inst, uint32 func_idx)
+{
+    AOTModuleInstanceExtra *extra = (AOTModuleInstanceExtra *)module_inst->e;
+    AOTFunctionInstance *export_funcs =
+        (AOTFunctionInstance *)module_inst->export_functions;
+    AOTFunctionInstance *func_inst = NULL;
+    ExportFuncMap *export_func_maps, *export_func_map, key;
+    uint64 size;
+    uint32 i;
+
+    if (module_inst->export_func_count == 0)
+        return NULL;
+
+    exception_lock(module_inst);
+
+    /* create the func_idx to export_idx maps if it hasn't been created */
+    if (!extra->export_func_maps) {
+        size = sizeof(ExportFuncMap) * (uint64)module_inst->export_func_count;
+        if (!(export_func_maps = extra->export_func_maps =
+                  runtime_malloc(size, NULL, 0))) {
+            /* allocate memory failed, lookup the export function one by one */
+            for (i = 0; i < module_inst->export_func_count; i++) {
+                if (export_funcs[i].func_index == func_idx) {
+                    func_inst = &export_funcs[i];
+                    break;
+                }
+            }
+            goto unlock_and_return;
+        }
+
+        for (i = 0; i < module_inst->export_func_count; i++) {
+            export_func_maps[i].func_idx = export_funcs[i].func_index;
+            export_func_maps[i].export_idx = i;
+        }
+
+        qsort(export_func_maps, module_inst->export_func_count,
+              sizeof(ExportFuncMap), cmp_export_func_map);
+    }
+
+    /* lookup the map to get the export_idx of the func_idx */
+    key.func_idx = func_idx;
+    export_func_map =
+        bsearch(&key, extra->export_func_maps, module_inst->export_func_count,
+                sizeof(ExportFuncMap), cmp_export_func_map);
+    if (export_func_map)
+        func_inst = &export_funcs[export_func_map->export_idx];
+
+unlock_and_return:
+    exception_unlock(module_inst);
+    return func_inst;
+}
+
 AOTFunctionInstance *
 aot_get_function_instance(AOTModuleInstance *module_inst, uint32 func_idx)
 {
     AOTModule *module = (AOTModule *)module_inst->module;
     AOTModuleInstanceExtra *extra = (AOTModuleInstanceExtra *)module_inst->e;
-    AOTFunctionInstance *export_funcs =
-        (AOTFunctionInstance *)module_inst->export_functions;
-    uint32 i;
+    AOTFunctionInstance *func_inst;
 
-    /* export functions are pre-instantiated */
-    for (i = 0; i < module_inst->export_func_count; i++) {
-        if (export_funcs[i].func_index == func_idx)
-            return &export_funcs[i];
-    }
+    /* lookup from export functions first */
+    if ((func_inst = aot_lookup_function_with_idx(module_inst, func_idx)))
+        return func_inst;
 
     exception_lock(module_inst);
 
@@ -1728,7 +1788,7 @@ aot_instantiate(AOTModule *module, AOTModuleInstance *parent,
     bool ret = false;
 #endif
 
-    /* Check heap size */
+    /* Align and validate heap size */
     heap_size = align_uint(heap_size, 8);
     if (heap_size > APP_HEAP_SIZE_MAX)
         heap_size = APP_HEAP_SIZE_MAX;
@@ -1848,7 +1908,9 @@ aot_instantiate(AOTModule *module, AOTModuleInstance *parent,
             goto fail;
         }
         for (i = 0; i < module->table_init_data_count; i++) {
-            if (wasm_elem_is_active(module->table_init_data_list[i]->mode))
+            if (wasm_elem_is_active(module->table_init_data_list[i]->mode)
+                || wasm_elem_is_declarative(
+                    module->table_init_data_list[i]->mode))
                 bh_bitmap_set_bit(common->elem_dropped, i);
         }
     }
@@ -1944,7 +2006,11 @@ aot_instantiate(AOTModule *module, AOTModuleInstance *parent,
         AOTTableInstance *table_inst;
         table_elem_type_t *table_data;
 
-        table = &module->tables[i];
+        /* bypass imported table since AOTImportTable doesn't have init_expr */
+        if (i < module->import_table_count)
+            continue;
+
+        table = &module->tables[i - module->import_table_count];
         bh_assert(table);
 
         if (table->init_expr.init_expr_type == INIT_EXPR_NONE) {
@@ -2168,6 +2234,9 @@ aot_deinstantiate(AOTModuleInstance *module_inst, bool is_sub_inst)
     if (module_inst->export_functions)
         wasm_runtime_free(module_inst->export_functions);
 
+    if (extra->export_func_maps)
+        wasm_runtime_free(extra->export_func_maps);
+
 #if WASM_ENABLE_MULTI_MEMORY != 0
     if (module_inst->export_memories)
         wasm_runtime_free(module_inst->export_memories);
@@ -2246,13 +2315,6 @@ invoke_native_with_hw_bound_check(WASMExecEnv *exec_env, void *func_ptr,
 #endif
     bool ret;
 
-    /* Check native stack overflow firstly to ensure we have enough
-       native stack to run the following codes before actually calling
-       the aot function in invokeNative function. */
-    if (!wasm_runtime_detect_native_stack_overflow(exec_env)) {
-        return false;
-    }
-
     if (!exec_env_tls) {
         if (!os_thread_signal_inited()) {
             aot_set_exception(module_inst, "thread signal env not inited");
@@ -2271,6 +2333,13 @@ invoke_native_with_hw_bound_check(WASMExecEnv *exec_env, void *func_ptr,
         }
     }
 
+    /* Check native stack overflow firstly to ensure we have enough
+       native stack to run the following codes before actually calling
+       the aot function in invokeNative function. */
+    if (!wasm_runtime_detect_native_stack_overflow(exec_env)) {
+        return false;
+    }
+
     wasm_exec_env_push_jmpbuf(exec_env, &jmpbuf_node);
 
     if (os_setjmp(jmpbuf_node.jmpbuf) == 0) {
@@ -2561,7 +2630,7 @@ aot_call_function(WASMExecEnv *exec_env, AOTFunctionInstance *function,
         ret = invoke_native_internal(exec_env, func_ptr, func_type, NULL,
                                      attachment, argv, argc, argv);
 
-        if (aot_copy_exception(module_inst, NULL)) {
+        if (!ret) {
 #ifdef AOT_STACK_FRAME_DEBUG
             if (aot_stack_frame_callback) {
                 aot_stack_frame_callback(exec_env);
@@ -2582,7 +2651,7 @@ aot_call_function(WASMExecEnv *exec_env, AOTFunctionInstance *function,
             aot_free_frame(exec_env);
 #endif
 
-        return ret && !aot_copy_exception(module_inst, NULL) ? true : false;
+        return ret;
     }
 }
 
@@ -3216,8 +3285,25 @@ aot_call_indirect(WASMExecEnv *exec_env, uint32 tbl_idx, uint32 table_elem_idx,
             cell_num += wasm_value_type_cell_num(ext_ret_types[i]);
         }
 
+#if WASM_ENABLE_AOT_STACK_FRAME != 0
+        void *prev_frame = get_top_frame(exec_env);
+        if (!is_frame_per_function(exec_env)
+            && !aot_alloc_frame(exec_env, func_idx)) {
+            if (argv1 != argv1_buf)
+                wasm_runtime_free(argv1);
+            return false;
+        }
+#endif
         ret = invoke_native_internal(exec_env, func_ptr, func_type, signature,
                                      attachment, argv1, argc, argv);
+#if WASM_ENABLE_AOT_STACK_FRAME != 0
+        /* Free all frames allocated, note that some frames
+           may be allocated in AOT code and haven't been
+           freed if exception occurred */
+        while (get_top_frame(exec_env) != prev_frame)
+            aot_free_frame(exec_env);
+#endif
+
         if (!ret) {
             if (argv1 != argv1_buf)
                 wasm_runtime_free(argv1);
@@ -3258,8 +3344,25 @@ aot_call_indirect(WASMExecEnv *exec_env, uint32 tbl_idx, uint32 table_elem_idx,
         return true;
     }
     else {
+#if WASM_ENABLE_AOT_STACK_FRAME != 0
+        void *prev_frame = get_top_frame(exec_env);
+        /* Only allocate frame for frame-per-call mode; in the
+           frame-per-function mode the frame is allocated at the
+           beginning of the function. */
+        if (!is_frame_per_function(exec_env)
+            && !aot_alloc_frame(exec_env, func_idx)) {
+            return false;
+        }
+#endif
         ret = invoke_native_internal(exec_env, func_ptr, func_type, signature,
                                      attachment, argv, argc, argv);
+#if WASM_ENABLE_AOT_STACK_FRAME != 0
+        /* Free all frames allocated, note that some frames
+           may be allocated in AOT code and haven't been
+           freed if exception occurred */
+        while (get_top_frame(exec_env) != prev_frame)
+            aot_free_frame(exec_env);
+#endif
         if (!ret)
             goto fail;
 
@@ -4034,6 +4137,136 @@ aot_frame_update_profile_info(WASMExecEnv *exec_env, bool alloc_frame)
 }
 #endif /* end of WASM_ENABLE_AOT_STACK_FRAME != 0 */
 
+#if WAMR_ENABLE_COPY_CALLSTACK != 0
+uint32
+aot_copy_callstack_tiny_frame(WASMExecEnv *exec_env, wasm_frame_t *buffer,
+                              const uint32 length, const uint32 skip_n,
+                              char *error_buf, uint32 error_buf_size)
+{
+    /*
+     * Note for devs: please refrain from such modifications inside of
+     * aot_copy_callstack_tiny_frame
+     * - 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
+     */
+    uint8 *top_boundary = exec_env->wasm_stack.top_boundary;
+    uint8 *top = exec_env->wasm_stack.top;
+    uint8 *bottom = exec_env->wasm_stack.bottom;
+    uint32 count = 0;
+
+    bool is_top_index_in_range =
+        top_boundary >= top && top >= (bottom + sizeof(AOTTinyFrame));
+    if (!is_top_index_in_range) {
+        char *err_msg =
+            "Top of the stack pointer is outside of the stack boundaries";
+        strncpy(error_buf, err_msg, error_buf_size);
+        return 0;
+    }
+    bool is_top_aligned_with_bottom =
+        (unsigned long)(top - bottom) % sizeof(AOTTinyFrame) == 0;
+    if (!is_top_aligned_with_bottom) {
+        char *err_msg = "Top of the stack is not aligned with the bottom";
+        strncpy(error_buf, err_msg, error_buf_size);
+        return 0;
+    }
+
+    AOTTinyFrame *frame = (AOTTinyFrame *)(top - sizeof(AOTTinyFrame));
+    WASMCApiFrame record_frame;
+    while (frame && (uint8_t *)frame >= bottom && count < (skip_n + length)) {
+        if (count < skip_n) {
+            ++count;
+            frame -= 1;
+            continue;
+        }
+        record_frame.instance = exec_env->module_inst;
+        record_frame.module_offset = 0;
+        record_frame.func_index = frame->func_index;
+        record_frame.func_offset = frame->ip_offset;
+        buffer[count - skip_n] = record_frame;
+        frame -= 1;
+        ++count;
+    }
+    return count >= skip_n ? count - skip_n : 0;
+}
+
+uint32
+aot_copy_callstack_standard_frame(WASMExecEnv *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
+     * aot_iterate_callstack_standard_frame
+     * - 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_iterate_callstack in
+     * wasm_export.h
+     */
+
+    uint32 count = 0;
+#if WASM_ENABLE_GC == 0
+    WASMModuleInstance *module_inst =
+        (WASMModuleInstance *)wasm_exec_env_get_module_inst(exec_env);
+    AOTFrame *cur_frame = (AOTFrame *)wasm_exec_env_get_cur_frame(exec_env);
+    uint8 *top_boundary = exec_env->wasm_stack.top_boundary;
+    uint8 *bottom = exec_env->wasm_stack.bottom;
+    uint32 frame_size = (uint32)offsetof(AOTFrame, lp);
+
+    WASMCApiFrame record_frame;
+    while (cur_frame && (uint8_t *)cur_frame >= bottom
+           && (uint8_t *)cur_frame + frame_size <= top_boundary
+           && count < (skip_n + length)) {
+        if (count < skip_n) {
+            ++count;
+            cur_frame = cur_frame->prev_frame;
+            continue;
+        }
+        record_frame.instance = module_inst;
+        record_frame.module_offset = 0;
+        record_frame.func_index = (uint32)cur_frame->func_index;
+        record_frame.func_offset = (uint32)cur_frame->ip_offset;
+        buffer[count - skip_n] = record_frame;
+        cur_frame = cur_frame->prev_frame;
+        ++count;
+    }
+#else
+/*
+ * TODO: add support for standard frames when GC is enabled
+ * now it poses a risk due to variable size of the frame
+ */
+#endif
+    return count >= skip_n ? count - skip_n : 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)
+{
+    /*
+     * Note for devs: please refrain from such modifications inside of
+     * aot_iterate_callstack
+     * - 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_iterate_callstack in
+     * wasm_export.h
+     */
+    if (!is_tiny_frame(exec_env)) {
+        return aot_copy_callstack_standard_frame(
+            exec_env, buffer, length, skip_n, error_buf, error_buf_size);
+    }
+    else {
+        return aot_copy_callstack_tiny_frame(exec_env, buffer, length, skip_n,
+                                             error_buf, error_buf_size);
+    }
+}
+#endif // WAMR_ENABLE_COPY_CALLSTACK
+
 #if WASM_ENABLE_DUMP_CALL_STACK != 0
 bool
 aot_create_call_stack(struct WASMExecEnv *exec_env)

core/iwasm/aot/aot_runtime.h

@@ -109,6 +109,13 @@ 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);
     /*
@@ -120,6 +127,13 @@ typedef struct AOTModuleInstanceExtra {
     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
@@ -556,6 +570,13 @@ 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);
 
@@ -563,7 +584,7 @@ AOTMemoryInstance *
 aot_get_default_memory(AOTModuleInstance *module_inst);
 
 AOTMemoryInstance *
-aot_get_memory_with_index(AOTModuleInstance *module_inst, uint32 index);
+aot_get_memory_with_idx(AOTModuleInstance *module_inst, uint32 mem_idx);
 
 /**
  * Get a function in the AOT module instance.
@@ -574,7 +595,7 @@ aot_get_memory_with_index(AOTModuleInstance *module_inst, uint32 index);
  * @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
@@ -756,6 +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

@@ -63,7 +63,7 @@ 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 = "aarch64v8";
-        s_size = 9; /* strlen("aarch64v8"); */
+        s_size = 10; /* sizeof("aarch64v8"); */
     }
     if (target_buf_size < s_size) {
         s_size = target_buf_size;

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_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_xtensa.c

@@ -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/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;
@@ -3383,21 +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 =
-                            aot_lookup_function(inst_aot, export->name);
-                        ((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
     }
@@ -3454,6 +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

core/iwasm/common/wasm_loader_common.c

@@ -151,7 +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_memory.c

@@ -420,13 +420,31 @@ 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 && addr >= heap->base_addr
-        && addr + bytes <= heap->base_addr + heap->size
-        && addr + bytes > addr) {
-        return true;
+    if (!heap) {
+        return false;
     }
-    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
@@ -1235,6 +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);
     }
@@ -1349,7 +1375,7 @@ wasm_enlarge_memory_internal(WASMModuleInstanceCommon *module,
     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;
@@ -1371,7 +1397,7 @@ wasm_enlarge_memory_internal(WASMModuleInstanceCommon *module,
     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;
@@ -1379,12 +1405,12 @@ wasm_enlarge_memory_internal(WASMModuleInstanceCommon *module,
 #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;
@@ -1579,8 +1605,7 @@ wasm_runtime_get_memory(WASMModuleInstanceCommon *module_inst, uint32 index)
 
 #if WASM_ENABLE_AOT != 0
     if (module_inst->module_type == Wasm_Module_AoT)
-        return aot_get_memory_with_index((AOTModuleInstance *)module_inst,
-                                         index);
+        return aot_get_memory_with_idx((AOTModuleInstance *)module_inst, index);
 #endif
 
     return NULL;
@@ -1663,7 +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
@@ -1716,7 +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_runtime_common.c

@@ -757,7 +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
@@ -1740,6 +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)
 {
@@ -3014,9 +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 */
@@ -3024,7 +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 */
@@ -3631,8 +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)
@@ -4260,31 +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;
@@ -4306,31 +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;
@@ -4370,8 +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:
         {

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);

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;
 }
@@ -487,60 +502,70 @@ 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 */
-    /* TODO: multi-memory */
     if (!comp_data->memory_count)
         comp_data->memory_count = 1;
 
@@ -548,7 +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);
 
@@ -580,22 +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++) {
@@ -641,64 +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;
@@ -717,6 +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;
 
 /**

core/iwasm/compilation/aot_compiler.c

@@ -4093,39 +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)
@@ -4134,7 +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;
     }
 
@@ -4148,7 +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)
@@ -4227,7 +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;
@@ -4270,7 +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);
 
@@ -4323,7 +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);
 
@@ -4336,7 +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

@@ -790,7 +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(

core/iwasm/compilation/aot_emit_control.c

@@ -1218,6 +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

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_function.c

@@ -1832,6 +1832,7 @@ 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]);
         }
     }
@@ -2068,6 +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);
     }
@@ -2699,6 +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);
         }
     }
@@ -3130,6 +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);
         }
     }
@@ -3205,6 +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);
         }
     }

core/iwasm/compilation/aot_emit_memory.c

@@ -273,10 +273,24 @@ aot_check_memory_overflow(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
     }
 
     /* offset1 = offset + addr; */
-    /* TODO: check whether integer overflow occurs when memory is 64-bit
-             and boundary check is enabled */
     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;
@@ -303,7 +317,7 @@ aot_check_memory_overflow(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
         LLVMPositionBuilderAtEnd(comp_ctx->builder, block_curr);
 
         if (!is_target_64bit) {
-            /* Check whether interger overflow occurs in addr + offset */
+            /* Check whether integer overflow occurs in addr + offset */
             LLVMBasicBlockRef check_integer_overflow_end;
             ADD_BASIC_BLOCK(check_integer_overflow_end,
                             "check_integer_overflow_end");
@@ -331,7 +345,7 @@ aot_check_memory_overflow(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
                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 accesed, (2) using
+               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");
@@ -1087,7 +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))) {
@@ -1170,7 +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);
@@ -1215,10 +1229,24 @@ check_bulk_memory_overflow(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
         goto fail;
     }
 
-    /* TODO: check whether integer overflow occurs when memory is 64-bit
-             and boundary check is enabled */
     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;

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_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"));
@@ -1720,7 +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;
@@ -2354,7 +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;
     }
 
@@ -2521,7 +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();
@@ -2664,12 +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;
 
@@ -2736,20 +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",
@@ -2761,17 +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 "
@@ -2784,7 +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
             }
         }
 
@@ -2990,6 +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 = "";
         }
@@ -3344,6 +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);
 
@@ -3417,7 +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
@@ -436,6 +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;
 

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);
 }
 

core/iwasm/compilation/aot_orc_extra.cpp

@@ -177,7 +177,7 @@ LLVMOrcLLLazyJITBuilderSetJITTargetMachineBuilder(
     LLVMOrcDisposeJITTargetMachineBuilder(JTMP);
 }
 
-static Optional<CompileOnDemandLayer::GlobalValueSet>
+static Optional<GlobalValueSet>
 PartitionFunction(GlobalValueSet Requested)
 {
     std::vector<const GlobalValue *> GVsToAdd;

core/iwasm/compilation/debug/dwarf_extractor.cpp

@@ -311,14 +311,18 @@ lldb_function_to_function_dbi(const AOTCompContext *comp_ctx,
         case eLanguageTypeC:
         case eLanguageTypeC99:
         case eLanguageTypeC11:
+#if LLVM_VERSION_MAJOR >= 17
         case eLanguageTypeC17:
+#endif
             break;
         case eLanguageTypeC_plus_plus:
         case eLanguageTypeC_plus_plus_03:
         case eLanguageTypeC_plus_plus_11:
         case eLanguageTypeC_plus_plus_14:
+#if LLVM_VERSION_MAJOR >= 17
         case eLanguageTypeC_plus_plus_17:
         case eLanguageTypeC_plus_plus_20:
+#endif
             cplusplus = true;
             break;
         default:

core/iwasm/compilation/iwasm_compl.cmake

@@ -1,6 +1,7 @@
 set (IWASM_COMPL_DIR ${CMAKE_CURRENT_LIST_DIR})
 
 include_directories(${IWASM_COMPL_DIR})
+enable_language(CXX)
 
 if (WAMR_BUILD_DEBUG_AOT EQUAL 1)
     file (GLOB_RECURSE source_all
@@ -16,7 +17,7 @@ endif()
 
 set (IWASM_COMPL_SOURCE ${source_all})
 
-# Disalbe rtti to works with LLVM
+# Disable rtti to works with LLVM
 
 if (MSVC)
   set (CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /GR-")

core/iwasm/compilation/simd/simd_int_arith.c

@@ -32,7 +32,7 @@ simd_integer_arith(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
             result = LLVMBuildMul(comp_ctx->builder, lhs, rhs, "product");
             break;
         default:
-            HANDLE_FAILURE("Unsupport arith_op");
+            HANDLE_FAILURE("Unsupported arith_op");
             break;
     }
 

core/iwasm/fast-jit/cg/x86-64/jit_codegen_x86_64.cpp

@@ -423,7 +423,7 @@ jmp_from_label_to_label(x86::Assembler &a, bh_list *jmp_info_list,
 
 /**
  * Encode detecting compare result register according to condition code
- * and then jumping to suitable label when the condtion is met
+ * and then jumping to suitable label when the condition is met
  *
  * @param cc the compiler context
  * @param a the assembler to emit the code
@@ -431,7 +431,7 @@ jmp_from_label_to_label(x86::Assembler &a, bh_list *jmp_info_list,
  * @param label_src the index of src label
  * @param op the opcode of condition operation
  * @param r1 the label info when condition is met
- * @param r2 the label info when condition is unmet, do nonthing if VOID
+ * @param r2 the label info when condition is unmet, do nothing if VOID
  * @param is_last_insn if current insn is the last insn of current block
  *
  * @return true if success, false if failed
@@ -2589,7 +2589,7 @@ alu_r_r_r_i32(x86::Assembler &a, ALU_OP op, int32 reg_no_dst, int32 reg_no1_src,
                 if (reg_no2_src == REG_EDX_IDX) {
                     /* convert `REM_S edx, eax, edx` into
                        `mov esi, edx` and `REM_S edx eax, rsi` to
-                       avoid overwritting edx when a.cdq() */
+                       avoid overwriting edx when a.cdq() */
                     a.mov(regs_i32[REG_I32_FREE_IDX], regs_i32[REG_EDX_IDX]);
                     reg_no2_src = REG_I32_FREE_IDX;
                 }
@@ -2609,7 +2609,7 @@ alu_r_r_r_i32(x86::Assembler &a, ALU_OP op, int32 reg_no_dst, int32 reg_no1_src,
                 if (reg_no2_src == REG_EDX_IDX) {
                     /* convert `REM_U edx, eax, edx` into
                        `mov esi, edx` and `REM_U edx eax, rsi` to
-                       avoid overwritting edx when unsigned extend
+                       avoid overwriting edx when unsigned extend
                        eax to edx:eax */
                     a.mov(regs_i32[REG_I32_FREE_IDX], regs_i32[REG_EDX_IDX]);
                     reg_no2_src = REG_I32_FREE_IDX;
@@ -5602,7 +5602,7 @@ fail:
             a.jmp(imm);                                                    \
             if (!err_handler->err) {                                       \
                 /* The offset written by asmjit is always 0, we patch it   \
-                   again, 6 is the size of jmp instruciton */              \
+                   again, 6 is the size of jmp instruction */              \
                 stream = (char *)a.code()->sectionById(0)->buffer().data() \
                          + a.code()->sectionById(0)->buffer().size() - 6;  \
                 _offset = label_offsets[label_dst]                         \
@@ -6169,7 +6169,7 @@ fail:
  * Replace all the jmp address pre-saved when the code cache hasn't been
  * allocated with actual address after code cache allocated
  *
- * @param cc compiler context containting the allocated code cacha info
+ * @param cc compiler context containing the allocated code cacha info
  * @param jmp_info_list the jmp info list
  */
 static void
@@ -6557,7 +6557,7 @@ at_cmpxchg_r_ra_base_r_offset_imm(x86::Assembler &a, uint32 bytes_dst,
  * @param a the assembler to emit the code
  * @param bytes_dst the bytes number of the data to actual operated on(load,
  * compare, replacement) could be 1(byte), 2(short), 4(int32), 8(int64)
- * @param data_xchg the immediate data for exchange(conditionally replacment
+ * @param data_xchg the immediate data for exchange(conditionally replacement
  * value)
  * @param reg_no_base the no of register that stores the base address
  *        of src&dst memory
@@ -6587,7 +6587,7 @@ at_cmpxchg_imm_ra_base_r_offset_r(x86::Assembler &a, uint32 bytes_dst,
  * @param a the assembler to emit the code
  * @param bytes_dst the bytes number of the data to actual operated on(load,
  * compare, replacement) could be 1(byte), 2(short), 4(int32), 8(int64)
- * @param data_xchg the immediate data for exchange(conditionally replacment
+ * @param data_xchg the immediate data for exchange(conditionally replacement
  * value)
  * @param reg_no_base the no of register that stores the base address
  *        of src&dst memory
@@ -8820,7 +8820,7 @@ jit_codegen_compile_call_to_fast_jit(const WASMModule *module, uint32 func_idx)
 
         /* If yes, set eax to 0, return to caller */
 
-        /* Pop all integer arument registers */
+        /* Pop all integer argument registers */
         for (i = 0; i < MAX_REG_INTS; i++) {
             a.pop(regs_i64[reg_idx_of_int_args[i]]);
         }
@@ -9084,7 +9084,7 @@ jit_codegen_compile_call_to_fast_jit(const WASMModule *module, uint32 func_idx)
         a.mov(m, x86::rdx);
     }
 
-    /* Pop all integer arument registers */
+    /* Pop all integer argument registers */
     for (i = 0; i < MAX_REG_INTS; i++) {
         a.pop(regs_i64[reg_idx_of_int_args[i]]);
     }

core/iwasm/fast-jit/fe/jit_emit_compare.c

@@ -14,7 +14,7 @@ jit_compile_op_compare_integer(JitCompContext *cc, IntCond cond, bool is64Bit)
     JitReg lhs, rhs, res, const_zero, const_one;
 
     if (cond < INT_EQZ || cond > INT_GE_U) {
-        jit_set_last_error(cc, "unsupported comparation operation");
+        jit_set_last_error(cc, "unsupported comparison operation");
         goto fail;
     }
 

core/iwasm/fast-jit/fe/jit_emit_control.c

@@ -1230,7 +1230,7 @@ jit_compile_op_br_table(JitCompContext *cc, uint32 *br_depths, uint32 br_count,
         copy_arities = check_copy_arities(block_dst, cc->jit_frame);
 
         if (!copy_arities) {
-            /* No need to create new basic block, direclty jump to
+            /* No need to create new basic block, directly jump to
                the existing basic block when no need to copy arities */
             if (i == br_count) {
                 if (block_dst->label_type == LABEL_TYPE_LOOP) {

core/iwasm/fast-jit/iwasm_fast_jit.cmake

@@ -1,14 +1,21 @@
 # Copyright (C) 2019 Intel Corporation.  All rights reserved.
 # SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 
-set (IWASM_FAST_JIT_DIR ${CMAKE_CURRENT_LIST_DIR})
+# 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_FAST_JIT_DIR ${CMAKE_CURRENT_LIST_DIR})
 add_definitions(-DWASM_ENABLE_FAST_JIT=1)
 if (WAMR_BUILD_FAST_JIT_DUMP EQUAL 1)
     add_definitions(-DWASM_ENABLE_FAST_JIT_DUMP=1)
 endif ()
 
 include_directories (${IWASM_FAST_JIT_DIR})
+enable_language(CXX)
 
 if (WAMR_BUILD_TARGET STREQUAL "X86_64" OR WAMR_BUILD_TARGET STREQUAL "AMD_64")
     include(FetchContent)

core/iwasm/fast-jit/jit_frontend.c

@@ -31,7 +31,7 @@ get_global_base_offset(const WASMModule *module)
         * (module->import_memory_count + module->memory_count);
 
 #if WASM_ENABLE_JIT != 0
-    /* If the module dosen't have memory, reserve one mem_info space
+    /* If the module doesn't have memory, reserve one mem_info space
        with empty content to align with llvm jit compiler */
     if (mem_inst_size == 0)
         mem_inst_size = (uint32)sizeof(WASMMemoryInstance);
@@ -1169,7 +1169,7 @@ init_func_translation(JitCompContext *cc)
     time_started = jit_cc_new_reg_I64(cc);
     /* Call os_time_thread_cputime_us() to get time_started firstly
        as there is stack frame switching below, calling native in them
-       may cause register spilling work inproperly */
+       may cause register spilling work improperly */
     if (!jit_emit_callnative(cc, os_time_thread_cputime_us, time_started, NULL,
                              0)) {
         return NULL;

core/iwasm/fast-jit/jit_ir.h

@@ -94,7 +94,7 @@ typedef uint32 JitReg;
 /*
  * Constant index flag of non-constant-value (constant value flag is
  * not set in register no. field) integer, floating point and vector
- * regisers. If this flag is set, the rest bits of the register
+ * registers. If this flag is set, the rest bits of the register
  * no. represent an index to the constant value table of the
  * corresponding type of the register and the register is read-only.
  */
@@ -1084,7 +1084,7 @@ typedef struct JitCompContext {
         /* Capacity of register annotations of each kind. */
         uint32 _capacity[JIT_REG_KIND_L32];
 
-        /* Constant vallues of each kind. */
+        /* Constant values of each kind. */
         uint8 *_value[JIT_REG_KIND_L32];
 
         /* Next element on the list of values with the same hash code. */
@@ -1145,7 +1145,7 @@ typedef struct JitCompContext {
         JitInsn **_table;
     } _insn_hash_table;
 
-    /* indicate if the last comparision is about floating-point numbers or not
+    /* indicate if the last comparison is about floating-point numbers or not
      */
     bool last_cmp_on_fp;
 } JitCompContext;
@@ -1203,7 +1203,7 @@ typedef struct JitCompContext {
  * Annotation disabling functions jit_annl_disable_NAME,
  * jit_anni_disable_NAME and jit_annr_disable_NAME, which release
  * memory of the annotations.  Before calling these functions,
- * resources owned by the annotations must be explictely released.
+ * resources owned by the annotations must be explicitly released.
  */
 #define ANN_LABEL(TYPE, NAME) void jit_annl_disable_##NAME(JitCompContext *cc);
 #define ANN_INSN(TYPE, NAME) void jit_anni_disable_##NAME(JitCompContext *cc);
@@ -1559,7 +1559,7 @@ _jit_cc_new_insn_norm(JitCompContext *cc, JitReg *result, JitInsn *insn);
  *
  * @param cc the compilationo context
  * @param result returned result of the instruction. If the value is
- * non-zero, it is the result of the constant-folding or an exsiting
+ * non-zero, it is the result of the constant-folding or an existing
  * equivalent instruction, in which case no instruction is added into
  * the compilation context. Otherwise, a new normalized instruction
  * has been added into the compilation context.

core/iwasm/fast-jit/jit_regalloc.c

@@ -221,7 +221,7 @@ get_reg_stride(JitReg reg)
  * @param rc the regalloc context
  * @param vreg the virtual register
  *
- * @return the spill slot encoded in a consant register
+ * @return the spill slot encoded in a constant register
  */
 static JitReg
 rc_alloc_spill_slot(RegallocContext *rc, JitReg vreg)
@@ -478,7 +478,7 @@ reload_vreg(RegallocContext *rc, JitReg vreg, JitInsn *cur_insn)
         JitReg fp_reg = rc->cc->fp_reg, offset;
 
         if (!vr->slot && !(vr->slot = rc_alloc_spill_slot(rc, vreg)))
-            /* Cannot allocte spill slot (due to OOM or frame size limit).  */
+            /* Cannot allocate spill slot (due to OOM or frame size limit).  */
             return NULL;
 
         offset = offset_of_spill_slot(rc->cc, vr->slot);
@@ -579,7 +579,7 @@ spill_vreg(RegallocContext *rc, JitReg vreg, JitInsn *cur_insn)
 
 /**
  * Allocate a hard register for the virtual register.  Necessary
- * reloade instruction will be inserted after the given instruction.
+ * reload instruction will be inserted after the given instruction.
  *
  * @param rc the regalloc context
  * @param vreg the virtual register
@@ -665,7 +665,7 @@ allocate_hreg(RegallocContext *rc, JitReg vreg, JitInsn *insn, int distance)
 
 /**
  * Allocate a hard register for the virtual register if not allocated
- * yet.  Necessary spill and reloade instructions will be inserted
+ * yet.  Necessary spill and reload instructions will be inserted
  * before/after and after the given instruction.  This operation will
  * convert the virtual register's state from 1 or 3 to 2.
  *

core/iwasm/include/aot_comp_option.h

@@ -8,6 +8,10 @@
 
 #include <stdint.h>
 
+#ifdef __cplusplus
+extern "C" {
+#endif
+
 typedef struct {
     /* Enables or disables bounds checks for stack frames. When enabled, the AOT
      * compiler generates code to check if the stack pointer is within the
@@ -69,6 +73,7 @@ typedef struct AOTCompOption {
     bool enable_perf_profiling;
     bool enable_memory_profiling;
     bool disable_llvm_intrinsics;
+    bool disable_llvm_jump_tables;
     bool disable_llvm_lto;
     bool enable_llvm_pgo;
     bool enable_stack_estimation;
@@ -88,4 +93,8 @@ typedef struct AOTCompOption {
     const char *builtin_intrinsics;
 } AOTCompOption, *aot_comp_option_t;
 
+#ifdef __cplusplus
+}
 #endif
+
+#endif /* end of __AOT_COMP_OPTION_H__ */

core/iwasm/include/wasm_c_api.h

@@ -46,7 +46,7 @@ extern "C" {
 // Auxiliaries
 
 // Machine types
-#if (__STDC_VERSION__) > 199901L
+#if defined(__STDC_VERSION__) && (__STDC_VERSION__) > 199901L
 inline void assertions(void) {
   static_assert(sizeof(float) == sizeof(uint32_t), "incompatible float type");
   static_assert(sizeof(double) == sizeof(uint64_t), "incompatible double type");

core/iwasm/include/wasm_export.h

@@ -23,6 +23,8 @@
 #else
 #define WASM_RUNTIME_API_EXTERN __declspec(dllimport)
 #endif
+#elif defined(__GNUC__) || defined(__clang__)
+#define WASM_RUNTIME_API_EXTERN __attribute__((visibility("default")))
 #else
 #define WASM_RUNTIME_API_EXTERN
 #endif
@@ -124,6 +126,21 @@ typedef WASMFunctionInstanceCommon *wasm_function_inst_t;
 struct WASMMemoryInstance;
 typedef struct WASMMemoryInstance *wasm_memory_inst_t;
 
+typedef struct wasm_frame_t {
+    /*  wasm_instance_t */
+    void *instance;
+    uint32_t module_offset;
+    uint32_t func_index;
+    uint32_t func_offset;
+    const char *func_name_wp;
+
+    uint32_t *sp;
+    uint8_t *frame_ref;
+    uint32_t *lp;
+} WASMCApiFrame;
+
+typedef WASMCApiFrame wasm_frame_t;
+
 /* WASM section */
 typedef struct wasm_section_t {
     struct wasm_section_t *next;
@@ -862,6 +879,35 @@ wasm_runtime_create_exec_env(wasm_module_inst_t module_inst,
 WASM_RUNTIME_API_EXTERN void
 wasm_runtime_destroy_exec_env(wasm_exec_env_t exec_env);
 
+/**
+ * @brief Copy callstack frames.
+ *
+ * Caution: This is not a thread-safe function. Ensure the exec_env
+ * is suspended before calling it from another thread.
+ *
+ * Usage: In the callback to read frames fields use APIs
+ * for wasm_frame_t from wasm_c_api.h
+ *
+ * Note: The function is async-signal-safe if called with verified arguments.
+ * Meaning it's safe to call it from a signal handler even on a signal
+ * interruption from another thread if next variables hold valid pointers
+ * - exec_env
+ * - exec_env->module_inst
+ * - exec_env->module_inst->module
+ *
+ * @param exec_env the execution environment that containes frames
+ * @param buffer the buffer of size equal length * sizeof(wasm_frame_t) to copy
+ * frames to
+ * @param length the number of frames to copy
+ * @param skip_n the number of frames to skip from the top of the stack
+ *
+ * @return number of copied frames
+ */
+WASM_RUNTIME_API_EXTERN uint32_t
+wasm_copy_callstack(const wasm_exec_env_t exec_env, wasm_frame_t *buffer,
+                    const uint32_t length, const uint32_t skip_n,
+                    char *error_buf, uint32_t error_buf_size);
+
 /**
  * Get the singleton execution environment for the instance.
  *
@@ -2281,7 +2327,7 @@ wasm_runtime_detach_shared_heap(wasm_module_inst_t module_inst);
  * @param size required memory size
  * @param p_native_addr native address of allocated memory
  *
- * @return return the allocated memory address, which re-uses part of the wasm
+ * @return return the allocated memory address, which reuses part of the wasm
  * address space and is in the range of [UINT32 - shared_heap_size + 1, UINT32]
  * (when the wasm memory is 32-bit) or [UINT64 - shared_heap_size + 1, UINT64]
  * (when the wasm memory is 64-bit). Note that it is not an absolute address.

core/iwasm/interpreter/wasm_interp_classic.c

@@ -593,8 +593,8 @@ wasm_interp_get_frame_ref(WASMInterpFrame *frame)
 #endif
 
 #if WASM_ENABLE_MEMORY64 != 0
-#define POP_MEM_OFFSET() (is_memory64 ? POP_I64() : POP_I32())
-#define POP_TBL_ELEM_IDX() (is_table64 ? POP_I64() : POP_I32())
+#define POP_MEM_OFFSET() (is_memory64 ? POP_I64() : (uint32)POP_I32())
+#define POP_TBL_ELEM_IDX() (is_table64 ? POP_I64() : (uint32)POP_I32())
 #else
 #define POP_MEM_OFFSET() POP_I32()
 #define POP_TBL_ELEM_IDX() POP_I32()
@@ -1934,7 +1934,7 @@ wasm_interp_call_func_bytecode(WASMModuleInstance *module,
                             UNWIND_CSP(relative_depth, LABEL_TYPE_FUNCTION);
                             /* push exception values for catch
                              * The values are copied to the CALLER FRAME
-                             * (prev_frame->sp) same behvior ad WASM_OP_RETURN
+                             * (prev_frame->sp) same behavior ad WASM_OP_RETURN
                              */
                             if (cell_num_to_copy > 0) {
                                 word_copy(prev_frame->sp,
@@ -2649,7 +2649,7 @@ wasm_interp_call_func_bytecode(WASMModuleInstance *module,
                 read_leb_uint32(frame_ip, frame_ip_end, type_index);
                 func_obj = POP_REF();
                 if (!func_obj) {
-                    wasm_set_exception(module, "null function object");
+                    wasm_set_exception(module, "null function reference");
                     goto got_exception;
                 }
 
@@ -2666,7 +2666,7 @@ wasm_interp_call_func_bytecode(WASMModuleInstance *module,
                 read_leb_uint32(frame_ip, frame_ip_end, type_index);
                 func_obj = POP_REF();
                 if (!func_obj) {
-                    wasm_set_exception(module, "null function object");
+                    wasm_set_exception(module, "null function reference");
                     goto got_exception;
                 }
 
@@ -2813,7 +2813,8 @@ wasm_interp_call_func_bytecode(WASMModuleInstance *module,
                         struct_obj = POP_REF();
 
                         if (!struct_obj) {
-                            wasm_set_exception(module, "null structure object");
+                            wasm_set_exception(module,
+                                               "null structure reference");
                             goto got_exception;
                         }
 
@@ -2869,7 +2870,8 @@ wasm_interp_call_func_bytecode(WASMModuleInstance *module,
 
                         struct_obj = POP_REF();
                         if (!struct_obj) {
-                            wasm_set_exception(module, "null structure object");
+                            wasm_set_exception(module,
+                                               "null structure reference");
                             goto got_exception;
                         }
 
@@ -4961,7 +4963,7 @@ wasm_interp_call_func_bytecode(WASMModuleInstance *module,
                 HANDLE_OP_END();
             }
 
-            /* numberic instructions of i32 */
+            /* numeric instructions of i32 */
             HANDLE_OP(WASM_OP_I32_CLZ)
             {
                 DEF_OP_BIT_COUNT(uint32, I32, clz32);
@@ -5118,7 +5120,7 @@ wasm_interp_call_func_bytecode(WASMModuleInstance *module,
                 HANDLE_OP_END();
             }
 
-            /* numberic instructions of i64 */
+            /* numeric instructions of i64 */
             HANDLE_OP(WASM_OP_I64_CLZ)
             {
                 DEF_OP_BIT_COUNT(uint64, I64, clz64);
@@ -5275,7 +5277,7 @@ wasm_interp_call_func_bytecode(WASMModuleInstance *module,
                 HANDLE_OP_END();
             }
 
-            /* numberic instructions of f32 */
+            /* numeric instructions of f32 */
             HANDLE_OP(WASM_OP_F32_ABS)
             {
                 DEF_OP_MATH(float32, F32, fabsf);
@@ -5379,7 +5381,7 @@ wasm_interp_call_func_bytecode(WASMModuleInstance *module,
                 HANDLE_OP_END();
             }
 
-            /* numberic instructions of f64 */
+            /* numeric instructions of f64 */
             HANDLE_OP(WASM_OP_F64_ABS)
             {
                 DEF_OP_MATH(float64, F64, fabs);
@@ -5782,7 +5784,6 @@ wasm_interp_call_func_bytecode(WASMModuleInstance *module,
                     {
                         mem_offset_t dst, src, len;
                         uint8 *mdst, *msrc;
-                        uint64 dlen;
 
                         len = POP_MEM_OFFSET();
                         src = POP_MEM_OFFSET();
@@ -5795,24 +5796,17 @@ wasm_interp_call_func_bytecode(WASMModuleInstance *module,
                         /* skip dst memidx */
                         frame_ip += 1;
 #endif
+                        // TODO: apply memidx
 #if WASM_ENABLE_THREAD_MGR != 0
                         linear_mem_size = get_linear_mem_size();
 #endif
-
-                        dlen = linear_mem_size - dst;
-
                         /* dst boundary check */
 #ifndef OS_ENABLE_HW_BOUND_CHECK
                         CHECK_BULK_MEMORY_OVERFLOW(dst, len, mdst);
-#if WASM_ENABLE_SHARED_HEAP != 0
-                        if (app_addr_in_shared_heap((uint64)dst, len))
-                            dlen = shared_heap_end_off - dst + 1;
-#endif
 #else /* else of OS_ENABLE_HW_BOUND_CHECK */
 #if WASM_ENABLE_SHARED_HEAP != 0
                         if (app_addr_in_shared_heap((uint64)dst, len)) {
                             shared_heap_addr_app_to_native((uint64)dst, mdst);
-                            dlen = shared_heap_end_off - dst + 1;
                         }
                         else
 #endif
@@ -5830,6 +5824,7 @@ wasm_interp_call_func_bytecode(WASMModuleInstance *module,
                         /* skip src memidx */
                         frame_ip += 1;
 #endif
+                        // TODO: apply memidx
 #if WASM_ENABLE_THREAD_MGR != 0
                         linear_mem_size = get_linear_mem_size();
 #endif
@@ -5849,15 +5844,21 @@ wasm_interp_call_func_bytecode(WASMModuleInstance *module,
                         }
 #endif
 
-#if WASM_ENABLE_MEMORY64 == 0
-                        /* allowing the destination and source to overlap */
-                        bh_memmove_s(mdst, (uint32)dlen, msrc, (uint32)len);
-#else
-                        /* use memmove when memory64 is enabled since len
-                           may be larger than UINT32_MAX */
-                        memmove(mdst, msrc, len);
-                        (void)dlen;
-#endif
+                        /*
+                         * avoid unnecessary operations
+                         *
+                         * since dst and src both are valid indexes in the
+                         * linear memory, mdst and msrc can't be NULL
+                         *
+                         * The spec. converts memory.copy into i32.load8 and
+                         * i32.store8; the following are runtime-specific
+                         * optimizations.
+                         *
+                         */
+                        if (len && mdst != msrc) {
+                            /* allowing the destination and source to overlap */
+                            memmove(mdst, msrc, len);
+                        }
                         break;
                     }
                     case WASM_OP_MEMORY_FILL:
@@ -6678,7 +6679,7 @@ wasm_interp_call_func_bytecode(WASMModuleInstance *module,
                          tag++, t++) {
 
                         /* compare the module and the external index with the
-                         * imort tag data */
+                         * import tag data */
                         if ((cur_func->u.func_import->import_module
                              == tag->u.tag_import->import_module)
                             && (ext_exception
@@ -6727,7 +6728,7 @@ wasm_interp_call_func_bytecode(WASMModuleInstance *module,
     || WASM_CPU_SUPPORTS_UNALIGNED_ADDR_ACCESS == 0 \
     || WASM_ENABLE_BULK_MEMORY != 0
             if (memory)
-                linear_mem_size = get_linear_mem_size();
+                linear_mem_size = GET_LINEAR_MEMORY_SIZE(memory);
 #endif
             if (wasm_copy_exception(module, NULL)) {
 #if WASM_ENABLE_EXCE_HANDLING != 0

core/iwasm/interpreter/wasm_mini_loader.c

@@ -48,7 +48,9 @@ is_table_64bit(WASMModule *module, uint32 table_idx)
         return !!(module->import_tables[table_idx].u.table.table_type.flags
                   & TABLE64_FLAG);
     else
-        return !!(module->tables[table_idx].table_type.flags & TABLE64_FLAG);
+        return !!(module->tables[table_idx - module->import_table_count]
+                      .table_type.flags
+                  & TABLE64_FLAG);
 
     return false;
 }
@@ -206,9 +208,14 @@ memory_realloc(void *mem_old, uint32 size_old, uint32 size_new, char *error_buf,
 {
     uint8 *mem_new;
     bh_assert(size_new > size_old);
+
+    if ((mem_new = wasm_runtime_realloc(mem_old, size_new))) {
+        memset(mem_new + size_old, 0, size_new - size_old);
+        return mem_new;
+    }
+
     if ((mem_new = loader_malloc(size_new, error_buf, error_buf_size))) {
         bh_memcpy_s(mem_new, size_new, mem_old, size_old);
-        memset(mem_new + size_old, 0, size_new - size_old);
         wasm_runtime_free(mem_old);
     }
     return mem_new;
@@ -658,7 +665,7 @@ load_table_import(const uint8 **p_buf, const uint8 *buf_end,
                   const char *table_name, WASMTableImport *table,
                   char *error_buf, uint32 error_buf_size)
 {
-    const uint8 *p = *p_buf, *p_end = buf_end;
+    const uint8 *p = *p_buf, *p_end = buf_end, *p_org;
     uint32 declare_elem_type = 0, table_flag = 0, declare_init_size = 0,
            declare_max_size = 0;
 
@@ -671,7 +678,12 @@ load_table_import(const uint8 **p_buf, const uint8 *buf_end,
 #endif
     );
 
+    /* the table flag can't exceed one byte, only check in debug build given
+     * the nature of mini-loader */
+    p_org = p;
     read_leb_uint32(p, p_end, table_flag);
+    bh_assert(p - p_org <= 1);
+    (void)p_org;
 
     if (!wasm_table_check_flags(table_flag, error_buf, error_buf_size, false)) {
         return false;
@@ -704,7 +716,7 @@ load_memory_import(const uint8 **p_buf, const uint8 *buf_end,
                    const char *memory_name, WASMMemoryImport *memory,
                    char *error_buf, uint32 error_buf_size)
 {
-    const uint8 *p = *p_buf, *p_end = buf_end;
+    const uint8 *p = *p_buf, *p_end = buf_end, *p_org;
 #if WASM_ENABLE_APP_FRAMEWORK != 0
     uint32 pool_size = wasm_runtime_memory_pool_size();
     uint32 max_page_count = pool_size * APP_MEMORY_MAX_GLOBAL_HEAP_PERCENT
@@ -717,7 +729,13 @@ load_memory_import(const uint8 **p_buf, const uint8 *buf_end,
     uint32 declare_init_page_count = 0;
     uint32 declare_max_page_count = 0;
 
+    /* the memory flag can't exceed one byte, only check in debug build given
+     * the nature of mini-loader */
+    p_org = p;
     read_leb_uint32(p, p_end, mem_flag);
+    bh_assert(p - p_org <= 1);
+    (void)p_org;
+
     if (!wasm_memory_check_flags(mem_flag, error_buf, error_buf_size, false)) {
         return false;
     }
@@ -808,6 +826,8 @@ load_table(const uint8 **p_buf, const uint8 *buf_end, WASMTable *table,
 #endif
     );
 
+    /* the table flag can't exceed one byte, only check in debug build given
+     * the nature of mini-loader */
     p_org = p;
     read_leb_uint32(p, p_end, table->table_type.flags);
     bh_assert(p - p_org <= 1);
@@ -847,6 +867,8 @@ load_memory(const uint8 **p_buf, const uint8 *buf_end, WASMMemory *memory,
     bool is_memory64 = false;
 #endif
 
+    /* the memory flag can't exceed one byte, only check in debug build given
+     * the nature of mini-loader */
     p_org = p;
     read_leb_uint32(p, p_end, memory->flags);
     bh_assert(p - p_org <= 1);
@@ -1183,6 +1205,8 @@ load_function_section(const uint8 *buf, const uint8 *buf_end,
                 local_count += sub_local_count;
             }
 
+            bh_assert(p_code_end > p_code && *(p_code_end - 1) == WASM_OP_END);
+
             /* Alloc memory, layout: function structure + local types */
             code_size = (uint32)(p_code_end - p_code);
 
@@ -2448,7 +2472,7 @@ orcjit_thread_callback(void *arg)
                     i + j * group_stride + module->import_function_count,
                     (void *)func_addr);
 
-                /* Try to switch to call this llvm jit funtion instead of
+                /* Try to switch to call this llvm jit function instead of
                    fast jit function from fast jit jitted code */
                 jit_compiler_set_call_to_llvm_jit(
                     module,
@@ -2564,7 +2588,7 @@ get_table_elem_type(const WASMModule *module, uint32 table_idx,
                 module->import_tables[table_idx].u.table.table_type.elem_type;
         else
             *p_elem_type =
-                module->tables[module->import_table_count + table_idx]
+                module->tables[table_idx - module->import_table_count]
                     .table_type.elem_type;
     }
     return true;
@@ -2725,6 +2749,11 @@ load_from_sections(WASMModule *module, WASMSection *sections,
         section = section->next;
     }
 
+#if WASM_ENABLE_BULK_MEMORY != 0
+    bh_assert(!has_datacount_section
+              || module->data_seg_count == module->data_seg_count1);
+#endif
+
     module->aux_data_end_global_index = (uint32)-1;
     module->aux_heap_base_global_index = (uint32)-1;
     module->aux_stack_top_global_index = (uint32)-1;
@@ -2734,6 +2763,12 @@ load_from_sections(WASMModule *module, WASMSection *sections,
     for (i = 0; i < module->export_count; i++, export ++) {
         if (export->kind == EXPORT_KIND_GLOBAL) {
             if (!strcmp(export->name, "__heap_base")) {
+                if (export->index < module->import_global_count) {
+                    LOG_DEBUG("Skip the process if __heap_base is imported "
+                              "instead of being a local global");
+                    continue;
+                }
+
                 global_index = export->index - module->import_global_count;
                 global = module->globals + global_index;
                 if (global->type.val_type == VALUE_TYPE_I32
@@ -2748,6 +2783,12 @@ load_from_sections(WASMModule *module, WASMSection *sections,
                 }
             }
             else if (!strcmp(export->name, "__data_end")) {
+                if (export->index < module->import_global_count) {
+                    LOG_DEBUG("Skip the process if __data_end is imported "
+                              "instead of being a local global");
+                    continue;
+                }
+
                 global_index = export->index - module->import_global_count;
                 global = module->globals + global_index;
                 if (global->type.val_type == VALUE_TYPE_I32
@@ -2942,9 +2983,7 @@ load_from_sections(WASMModule *module, WASMSection *sections,
     }
 
     if (!module->possible_memory_grow) {
-        WASMMemoryImport *memory_import;
-        WASMMemory *memory;
-
+#if WASM_ENABLE_SHRUNK_MEMORY != 0
         if (aux_data_end_global && aux_heap_base_global
             && aux_stack_top_global) {
             uint64 init_memory_size;
@@ -2954,7 +2993,8 @@ load_from_sections(WASMModule *module, WASMSection *sections,
              * valid range of uint32 */
             if (shrunk_memory_size <= UINT32_MAX) {
                 if (module->import_memory_count) {
-                    memory_import = &module->import_memories[0].u.memory;
+                    WASMMemoryImport *memory_import =
+                        &module->import_memories[0].u.memory;
                     init_memory_size =
                         (uint64)memory_import->mem_type.num_bytes_per_page
                         * memory_import->mem_type.init_page_count;
@@ -2969,7 +3009,7 @@ load_from_sections(WASMModule *module, WASMSection *sections,
                 }
 
                 if (module->memory_count) {
-                    memory = &module->memories[0];
+                    WASMMemory *memory = &module->memories[0];
                     init_memory_size = (uint64)memory->num_bytes_per_page
                                        * memory->init_page_count;
                     if (shrunk_memory_size <= init_memory_size) {
@@ -2982,9 +3022,11 @@ load_from_sections(WASMModule *module, WASMSection *sections,
                 }
             }
         }
+#endif /* WASM_ENABLE_SHRUNK_MEMORY != 0 */
 
         if (module->import_memory_count) {
-            memory_import = &module->import_memories[0].u.memory;
+            WASMMemoryImport *memory_import =
+                &module->import_memories[0].u.memory;
             if (memory_import->mem_type.init_page_count < DEFAULT_MAX_PAGES) {
                 memory_import->mem_type.num_bytes_per_page *=
                     memory_import->mem_type.init_page_count;
@@ -2998,7 +3040,7 @@ load_from_sections(WASMModule *module, WASMSection *sections,
         }
 
         if (module->memory_count) {
-            memory = &module->memories[0];
+            WASMMemory *memory = &module->memories[0];
             if (memory->init_page_count < DEFAULT_MAX_PAGES) {
                 memory->num_bytes_per_page *= memory->init_page_count;
                 if (memory->init_page_count > 0)
@@ -4030,11 +4072,16 @@ typedef struct WASMLoaderContext {
     /* preserved local offset */
     int16 preserved_local_offset;
 
-    /* const buffer */
-    uint8 *const_buf;
-    uint16 num_const;
-    uint16 const_cell_num;
-    uint32 const_buf_size;
+    /* const buffer for i64 and f64 consts, note that the raw bytes
+     * of i64 and f64 are the same, so we read an i64 value from an
+     * f64 const with its raw bytes, something like `*(int64 *)&f64 */
+    int64 *i64_consts;
+    uint32 i64_const_max_num;
+    uint32 i64_const_num;
+    /* const buffer for i32 and f32 consts */
+    int32 *i32_consts;
+    uint32 i32_const_max_num;
+    uint32 i32_const_num;
 
     /* processed code */
     uint8 *p_code_compiled;
@@ -4047,12 +4094,6 @@ typedef struct WASMLoaderContext {
 #endif
 } WASMLoaderContext;
 
-typedef struct Const {
-    WASMValue value;
-    uint16 slot_index;
-    uint8 value_type;
-} Const;
-
 #define CHECK_CSP_PUSH()                                                  \
     do {                                                                  \
         if (ctx->frame_csp >= ctx->frame_csp_boundary) {                  \
@@ -4207,8 +4248,10 @@ wasm_loader_ctx_destroy(WASMLoaderContext *ctx)
 #if WASM_ENABLE_FAST_INTERP != 0
         if (ctx->frame_offset_bottom)
             wasm_runtime_free(ctx->frame_offset_bottom);
-        if (ctx->const_buf)
-            wasm_runtime_free(ctx->const_buf);
+        if (ctx->i64_consts)
+            wasm_runtime_free(ctx->i64_consts);
+        if (ctx->i32_consts)
+            wasm_runtime_free(ctx->i32_consts);
 #endif
         wasm_runtime_free(ctx);
     }
@@ -4242,10 +4285,15 @@ wasm_loader_ctx_init(WASMFunction *func, char *error_buf, uint32 error_buf_size)
         goto fail;
     loader_ctx->frame_offset_boundary = loader_ctx->frame_offset_bottom + 32;
 
-    loader_ctx->num_const = 0;
-    loader_ctx->const_buf_size = sizeof(Const) * 8;
-    if (!(loader_ctx->const_buf = loader_malloc(loader_ctx->const_buf_size,
-                                                error_buf, error_buf_size)))
+    loader_ctx->i64_const_max_num = 8;
+    if (!(loader_ctx->i64_consts =
+              loader_malloc(sizeof(int64) * loader_ctx->i64_const_max_num,
+                            error_buf, error_buf_size)))
+        goto fail;
+    loader_ctx->i32_const_max_num = 8;
+    if (!(loader_ctx->i32_consts =
+              loader_malloc(sizeof(int32) * loader_ctx->i32_const_max_num,
+                            error_buf, error_buf_size)))
         goto fail;
 
     if (func->param_cell_num >= (int32)INT16_MAX - func->local_cell_num) {
@@ -4730,6 +4778,11 @@ preserve_referenced_local(WASMLoaderContext *loader_ctx, uint8 opcode,
                         loader_ctx->preserved_local_offset += 2;
                     emit_label(EXT_OP_COPY_STACK_TOP_I64);
                 }
+
+                /* overflow */
+                bh_assert(preserved_offset
+                          <= loader_ctx->preserved_local_offset);
+
                 emit_operand(loader_ctx, local_index);
                 emit_operand(loader_ctx, preserved_offset);
                 emit_label(opcode);
@@ -5068,107 +5121,116 @@ wasm_loader_push_pop_frame_ref_offset(WASMLoaderContext *ctx, uint8 pop_cnt,
     return true;
 }
 
+static int
+cmp_i64_const(const void *p_i64_const1, const void *p_i64_const2)
+{
+    int64 i64_const1 = *(int64 *)p_i64_const1;
+    int64 i64_const2 = *(int64 *)p_i64_const2;
+
+    return (i64_const1 < i64_const2) ? -1 : (i64_const1 > i64_const2) ? 1 : 0;
+}
+
+static int
+cmp_i32_const(const void *p_i32_const1, const void *p_i32_const2)
+{
+    int32 i32_const1 = *(int32 *)p_i32_const1;
+    int32 i32_const2 = *(int32 *)p_i32_const2;
+
+    return (i32_const1 < i32_const2) ? -1 : (i32_const1 > i32_const2) ? 1 : 0;
+}
+
 static bool
 wasm_loader_get_const_offset(WASMLoaderContext *ctx, uint8 type, void *value,
                              int16 *offset, char *error_buf,
                              uint32 error_buf_size)
 {
-    int8 bytes_to_increase;
-    int16 operand_offset = 0;
-    Const *c;
+    if (!ctx->p_code_compiled) {
+        /* Treat i64 and f64 as the same by reading i64 value from
+           the raw bytes */
+        if (type == VALUE_TYPE_I64 || type == VALUE_TYPE_F64) {
+            /* No slot left, emit const instead */
+            if (ctx->i64_const_num * 2 + ctx->i32_const_num > INT16_MAX - 2) {
+                *offset = 0;
+                return true;
+            }
 
-    /* Search existing constant */
-    for (c = (Const *)ctx->const_buf;
-         (uint8 *)c < ctx->const_buf + ctx->num_const * sizeof(Const); c++) {
-        if ((type == c->value_type)
-            && ((type == VALUE_TYPE_I64 && *(int64 *)value == c->value.i64)
-                || (type == VALUE_TYPE_I32 && *(int32 *)value == c->value.i32)
-#if WASM_ENABLE_REF_TYPES != 0
-                || (type == VALUE_TYPE_FUNCREF
-                    && *(int32 *)value == c->value.i32)
-                || (type == VALUE_TYPE_EXTERNREF
-                    && *(int32 *)value == c->value.i32)
-#endif
-                || (type == VALUE_TYPE_F64
-                    && (0 == memcmp(value, &(c->value.f64), sizeof(float64))))
-                || (type == VALUE_TYPE_F32
-                    && (0
-                        == memcmp(value, &(c->value.f32), sizeof(float32)))))) {
-            operand_offset = c->slot_index;
-            break;
-        }
-        if (c->value_type == VALUE_TYPE_I64 || c->value_type == VALUE_TYPE_F64)
-            operand_offset += 2;
-        else
-            operand_offset += 1;
-    }
+            /* Traverse the list if the const num is small */
+            if (ctx->i64_const_num < 10) {
+                for (uint32 i = 0; i < ctx->i64_const_num; i++) {
+                    if (ctx->i64_consts[i] == *(int64 *)value) {
+                        *offset = -1;
+                        return true;
+                    }
+                }
+            }
 
-    if ((uint8 *)c == ctx->const_buf + ctx->num_const * sizeof(Const)) {
-        /* New constant, append to the const buffer */
-        if ((type == VALUE_TYPE_F64) || (type == VALUE_TYPE_I64)) {
-            bytes_to_increase = 2;
+            if (ctx->i64_const_num >= ctx->i64_const_max_num) {
+                MEM_REALLOC(ctx->i64_consts,
+                            sizeof(int64) * ctx->i64_const_max_num,
+                            sizeof(int64) * (ctx->i64_const_max_num * 2));
+                ctx->i64_const_max_num *= 2;
+            }
+            ctx->i64_consts[ctx->i64_const_num++] = *(int64 *)value;
         }
         else {
-            bytes_to_increase = 1;
+            /* Treat i32 and f32 as the same by reading i32 value from
+               the raw bytes */
+            bh_assert(type == VALUE_TYPE_I32 || type == VALUE_TYPE_F32);
+
+            /* No slot left, emit const instead */
+            if (ctx->i64_const_num * 2 + ctx->i32_const_num > INT16_MAX - 1) {
+                *offset = 0;
+                return true;
+            }
+
+            /* Traverse the list if the const num is small */
+            if (ctx->i32_const_num < 10) {
+                for (uint32 i = 0; i < ctx->i32_const_num; i++) {
+                    if (ctx->i32_consts[i] == *(int32 *)value) {
+                        *offset = -1;
+                        return true;
+                    }
+                }
+            }
+
+            if (ctx->i32_const_num >= ctx->i32_const_max_num) {
+                MEM_REALLOC(ctx->i32_consts,
+                            sizeof(int32) * ctx->i32_const_max_num,
+                            sizeof(int32) * (ctx->i32_const_max_num * 2));
+                ctx->i32_const_max_num *= 2;
+            }
+            ctx->i32_consts[ctx->i32_const_num++] = *(int32 *)value;
         }
 
-        /* The max cell num of const buffer is 32768 since the valid index range
-         * is -32768 ~ -1. Return an invalid index 0 to indicate the buffer is
-         * full */
-        if (ctx->const_cell_num > INT16_MAX - bytes_to_increase + 1) {
-            *offset = 0;
-            return true;
-        }
-
-        if ((uint8 *)c == ctx->const_buf + ctx->const_buf_size) {
-            MEM_REALLOC(ctx->const_buf, ctx->const_buf_size,
-                        ctx->const_buf_size + 4 * sizeof(Const));
-            ctx->const_buf_size += 4 * sizeof(Const);
-            c = (Const *)(ctx->const_buf + ctx->num_const * sizeof(Const));
-        }
-        c->value_type = type;
-        switch (type) {
-            case VALUE_TYPE_F64:
-                bh_memcpy_s(&(c->value.f64), sizeof(WASMValue), value,
-                            sizeof(float64));
-                ctx->const_cell_num += 2;
-                /* The const buf will be reversed, we use the second cell */
-                /* of the i64/f64 const so the finnal offset is corrent */
-                operand_offset++;
-                break;
-            case VALUE_TYPE_I64:
-                c->value.i64 = *(int64 *)value;
-                ctx->const_cell_num += 2;
-                operand_offset++;
-                break;
-            case VALUE_TYPE_F32:
-                bh_memcpy_s(&(c->value.f32), sizeof(WASMValue), value,
-                            sizeof(float32));
-                ctx->const_cell_num++;
-                break;
-            case VALUE_TYPE_I32:
-                c->value.i32 = *(int32 *)value;
-                ctx->const_cell_num++;
-                break;
-#if WASM_ENABLE_REF_TYPES != 0
-            case VALUE_TYPE_EXTERNREF:
-            case VALUE_TYPE_FUNCREF:
-                c->value.i32 = *(int32 *)value;
-                ctx->const_cell_num++;
-                break;
-#endif
-            default:
-                break;
-        }
-        c->slot_index = operand_offset;
-        ctx->num_const++;
-        LOG_OP("#### new const [%d]: %ld\n", ctx->num_const,
-               (int64)c->value.i64);
+        *offset = -1;
+        return true;
+    }
+    else {
+        if (type == VALUE_TYPE_I64 || type == VALUE_TYPE_F64) {
+            int64 key = *(int64 *)value, *i64_const;
+            i64_const = bsearch(&key, ctx->i64_consts, ctx->i64_const_num,
+                                sizeof(int64), cmp_i64_const);
+            if (!i64_const) { /* not found, emit const instead */
+                *offset = 0;
+                return true;
+            }
+            *offset = -(uint32)(ctx->i64_const_num * 2 + ctx->i32_const_num)
+                      + (uint32)(i64_const - ctx->i64_consts) * 2;
+        }
+        else {
+            int32 key = *(int32 *)value, *i32_const;
+            i32_const = bsearch(&key, ctx->i32_consts, ctx->i32_const_num,
+                                sizeof(int32), cmp_i32_const);
+            if (!i32_const) { /* not found, emit const instead */
+                *offset = 0;
+                return true;
+            }
+            *offset = -(uint32)(ctx->i32_const_num)
+                      + (uint32)(i32_const - ctx->i32_consts);
+        }
+
+        return true;
     }
-    /* use negetive index for const */
-    operand_offset = -(operand_offset + 1);
-    *offset = operand_offset;
-    return true;
 fail:
     return false;
 }
@@ -5559,7 +5621,12 @@ reserve_block_ret(WASMLoaderContext *loader_ctx, uint8 opcode,
             else {
                 loader_ctx->frame_offset = frame_offset;
                 loader_ctx->dynamic_offset = dynamic_offset;
-                PUSH_OFFSET_TYPE(return_types[i]);
+                if (!(wasm_loader_push_frame_offset(
+                        loader_ctx, return_types[i], disable_emit,
+                        operand_offset, error_buf, error_buf_size))) {
+                    wasm_runtime_free(emit_data);
+                    goto fail;
+                }
                 wasm_loader_emit_backspace(loader_ctx, sizeof(int16));
                 loader_ctx->frame_offset = frame_offset_org;
                 loader_ctx->dynamic_offset = dynamic_offset_org;
@@ -5590,13 +5657,6 @@ fail:
 
 #endif /* WASM_ENABLE_FAST_INTERP */
 
-#define RESERVE_BLOCK_RET()                                                 \
-    do {                                                                    \
-        if (!reserve_block_ret(loader_ctx, opcode, disable_emit, error_buf, \
-                               error_buf_size))                             \
-            goto fail;                                                      \
-    } while (0)
-
 #define PUSH_TYPE(type)                                               \
     do {                                                              \
         if (!(wasm_loader_push_frame_ref(loader_ctx, type, error_buf, \
@@ -6033,6 +6093,86 @@ fail:
         }                                                             \
     } while (0)
 
+#if WASM_ENABLE_FAST_INTERP == 0
+
+#define pb_read_leb_uint32 read_leb_uint32
+#define pb_read_leb_int32 read_leb_int32
+#define pb_read_leb_int64 read_leb_int64
+#define pb_read_leb_memarg read_leb_memarg
+#define pb_read_leb_mem_offset read_leb_mem_offset
+#define pb_read_leb_memidx read_leb_memidx
+
+#else
+
+/* Read leb without malformed format check */
+static uint64
+read_leb_quick(uint8 **p_buf, uint32 maxbits, bool sign)
+{
+    uint8 *buf = *p_buf;
+    uint64 result = 0, byte = 0;
+    uint32 shift = 0;
+
+    do {
+        byte = *buf++;
+        result |= ((byte & 0x7f) << shift);
+        shift += 7;
+    } while (byte & 0x80);
+
+    if (sign && (shift < maxbits) && (byte & 0x40)) {
+        /* Sign extend */
+        result |= (~((uint64)0)) << shift;
+    }
+
+    *p_buf = buf;
+    return result;
+}
+
+#define pb_read_leb_uint32(p, p_end, res)                 \
+    do {                                                  \
+        if (!loader_ctx->p_code_compiled)                 \
+            /* Enable format check in the first scan */   \
+            read_leb_uint32(p, p_end, res);               \
+        else                                              \
+            /* Disable format check in the second scan */ \
+            res = (uint32)read_leb_quick(&p, 32, false);  \
+    } while (0)
+
+#define pb_read_leb_int32(p, p_end, res)                  \
+    do {                                                  \
+        if (!loader_ctx->p_code_compiled)                 \
+            /* Enable format check in the first scan */   \
+            read_leb_int32(p, p_end, res);                \
+        else                                              \
+            /* Disable format check in the second scan */ \
+            res = (int32)read_leb_quick(&p, 32, true);    \
+    } while (0)
+
+#define pb_read_leb_int64(p, p_end, res)                  \
+    do {                                                  \
+        if (!loader_ctx->p_code_compiled)                 \
+            /* Enable format check in the first scan */   \
+            read_leb_int64(p, p_end, res);                \
+        else                                              \
+            /* Disable format check in the second scan */ \
+            res = (int64)read_leb_quick(&p, 64, true);    \
+    } while (0)
+
+#if WASM_ENABLE_MULTI_MEMORY != 0
+#define pb_read_leb_memarg read_leb_memarg
+#else
+#define pb_read_leb_memarg pb_read_leb_uint32
+#endif
+
+#if WASM_ENABLE_MEMORY64 != 0
+#define pb_read_leb_mem_offset read_leb_mem_offset
+#else
+#define pb_read_leb_mem_offset pb_read_leb_uint32
+#endif
+
+#define pb_read_leb_memidx pb_read_leb_uint32
+
+#endif /* end of WASM_ENABLE_FAST_INTERP != 0 */
+
 static bool
 wasm_loader_prepare_bytecode(WASMModule *module, WASMFunction *func,
                              uint32 cur_func_idx, char *error_buf,
@@ -6056,11 +6196,9 @@ wasm_loader_prepare_bytecode(WASMModule *module, WASMFunction *func,
     uint32 segment_index;
 #endif
 #if WASM_ENABLE_FAST_INTERP != 0
-    uint8 *func_const_end, *func_const = NULL;
     int16 operand_offset = 0;
     uint8 last_op = 0;
     bool disable_emit, preserve_local = false, if_condition_available = true;
-    ;
     float32 f32_const;
     float64 f64_const;
 
@@ -6111,6 +6249,68 @@ re_scan:
         p = func->code;
         func->code_compiled = loader_ctx->p_code_compiled;
         func->code_compiled_size = loader_ctx->code_compiled_size;
+
+        if (loader_ctx->i64_const_num > 0) {
+            int64 *i64_consts_old = loader_ctx->i64_consts;
+
+            /* Sort the i64 consts */
+            qsort(i64_consts_old, loader_ctx->i64_const_num, sizeof(int64),
+                  cmp_i64_const);
+
+            /* Remove the duplicated i64 consts */
+            uint32 k = 1;
+            for (i = 1; i < loader_ctx->i64_const_num; i++) {
+                if (i64_consts_old[i] != i64_consts_old[i - 1]) {
+                    i64_consts_old[k++] = i64_consts_old[i];
+                }
+            }
+
+            if (k < loader_ctx->i64_const_num) {
+                int64 *i64_consts_new;
+                /* Try to reallocate memory with a smaller size */
+                if ((i64_consts_new =
+                         wasm_runtime_malloc((uint32)sizeof(int64) * k))) {
+                    bh_memcpy_s(i64_consts_new, (uint32)sizeof(int64) * k,
+                                i64_consts_old, (uint32)sizeof(int64) * k);
+                    /* Free the old memory */
+                    wasm_runtime_free(i64_consts_old);
+                    loader_ctx->i64_consts = i64_consts_new;
+                    loader_ctx->i64_const_max_num = k;
+                }
+                loader_ctx->i64_const_num = k;
+            }
+        }
+
+        if (loader_ctx->i32_const_num > 0) {
+            int32 *i32_consts_old = loader_ctx->i32_consts;
+
+            /* Sort the i32 consts */
+            qsort(i32_consts_old, loader_ctx->i32_const_num, sizeof(int32),
+                  cmp_i32_const);
+
+            /* Remove the duplicated i32 consts */
+            uint32 k = 1;
+            for (i = 1; i < loader_ctx->i32_const_num; i++) {
+                if (i32_consts_old[i] != i32_consts_old[i - 1]) {
+                    i32_consts_old[k++] = i32_consts_old[i];
+                }
+            }
+
+            if (k < loader_ctx->i32_const_num) {
+                int32 *i32_consts_new;
+                /* Try to reallocate memory with a smaller size */
+                if ((i32_consts_new =
+                         wasm_runtime_malloc((uint32)sizeof(int32) * k))) {
+                    bh_memcpy_s(i32_consts_new, (uint32)sizeof(int32) * k,
+                                i32_consts_old, (uint32)sizeof(int32) * k);
+                    /* Free the old memory */
+                    wasm_runtime_free(i32_consts_old);
+                    loader_ctx->i32_consts = i32_consts_new;
+                    loader_ctx->i32_const_max_num = k;
+                }
+                loader_ctx->i32_const_num = k;
+            }
+        }
     }
 #endif
 
@@ -6180,7 +6380,7 @@ re_scan:
                     int32 type_index;
                     /* Resolve the leb128 encoded type index as block type */
                     p--;
-                    read_leb_int32(p, p_end, type_index);
+                    pb_read_leb_int32(p, p_end, type_index);
                     bh_assert((uint32)type_index < module->type_count);
                     block_type.is_value_type = false;
                     block_type.u.type = module->types[type_index];
@@ -6364,7 +6564,10 @@ re_scan:
 #if WASM_ENABLE_FAST_INTERP != 0
                 /* if the result of if branch is in local or const area, add a
                  * copy op */
-                RESERVE_BLOCK_RET();
+                if (!reserve_block_ret(loader_ctx, opcode, disable_emit,
+                                       error_buf, error_buf_size)) {
+                    goto fail;
+                }
 
                 emit_empty_label_addr_and_frame_ip(PATCH_END);
                 apply_label_patch(loader_ctx, 1, PATCH_ELSE);
@@ -6424,7 +6627,11 @@ re_scan:
 #if WASM_ENABLE_FAST_INTERP != 0
                 skip_label();
                 /* copy the result to the block return address */
-                RESERVE_BLOCK_RET();
+                if (!reserve_block_ret(loader_ctx, opcode, disable_emit,
+                                       error_buf, error_buf_size)) {
+                    free_label_patch_list(loader_ctx->frame_csp);
+                    goto fail;
+                }
 
                 apply_label_patch(loader_ctx, 0, PATCH_END);
                 free_label_patch_list(loader_ctx->frame_csp);
@@ -6486,7 +6693,7 @@ re_scan:
                 uint32 j;
 #endif
 
-                read_leb_uint32(p, p_end, count);
+                pb_read_leb_uint32(p, p_end, count);
 #if WASM_ENABLE_FAST_INTERP != 0
                 emit_uint32(loader_ctx, count);
 #endif
@@ -6495,7 +6702,7 @@ re_scan:
                 /* Get each depth and check it */
                 p_org = p;
                 for (i = 0; i <= count; i++) {
-                    read_leb_uint32(p, p_end, depth);
+                    pb_read_leb_uint32(p, p_end, depth);
                     bh_assert(loader_ctx->csp_num > 0);
                     bh_assert(loader_ctx->csp_num - 1 >= depth);
                     (void)depth;
@@ -6593,7 +6800,7 @@ re_scan:
                 uint32 func_idx;
                 int32 idx;
 
-                read_leb_uint32(p, p_end, func_idx);
+                pb_read_leb_uint32(p, p_end, func_idx);
 #if WASM_ENABLE_FAST_INTERP != 0
                 /* we need to emit func_idx before arguments */
                 emit_uint32(loader_ctx, func_idx);
@@ -6666,10 +6873,10 @@ re_scan:
 
                 bh_assert(module->import_table_count + module->table_count > 0);
 
-                read_leb_uint32(p, p_end, type_idx);
+                pb_read_leb_uint32(p, p_end, type_idx);
 
 #if WASM_ENABLE_REF_TYPES != 0
-                read_leb_uint32(p, p_end, table_idx);
+                pb_read_leb_uint32(p, p_end, table_idx);
 #else
                 CHECK_BUF(p, p_end, 1);
                 table_idx = read_uint8(p);
@@ -6679,6 +6886,15 @@ re_scan:
                     goto fail;
                 }
 
+                bh_assert(
+                    (table_idx < module->import_table_count
+                         ? module->import_tables[table_idx]
+                               .u.table.table_type.elem_type
+                         : module
+                               ->tables[table_idx - module->import_table_count]
+                               .table_type.elem_type)
+                    == VALUE_TYPE_FUNCREF);
+
 #if WASM_ENABLE_FAST_INTERP != 0
                 /* we need to emit before arguments */
                 emit_uint32(loader_ctx, type_idx);
@@ -6900,7 +7116,7 @@ re_scan:
                 uint8 *p_code_compiled_tmp = loader_ctx->p_code_compiled;
 #endif
 
-                read_leb_uint32(p, p_end, vec_len);
+                pb_read_leb_uint32(p, p_end, vec_len);
                 if (vec_len != 1) {
                     /* typed select must have exactly one result */
                     set_error_buf(error_buf, error_buf_size,
@@ -6975,7 +7191,7 @@ re_scan:
                 uint8 decl_ref_type;
                 uint32 table_idx;
 
-                read_leb_uint32(p, p_end, table_idx);
+                pb_read_leb_uint32(p, p_end, table_idx);
                 if (!get_table_elem_type(module, table_idx, &decl_ref_type,
                                          error_buf, error_buf_size))
                     goto fail;
@@ -7069,7 +7285,7 @@ re_scan:
             case WASM_OP_REF_FUNC:
             {
                 uint32 func_idx = 0;
-                read_leb_uint32(p, p_end, func_idx);
+                pb_read_leb_uint32(p, p_end, func_idx);
 
                 if (!check_function_index(module, func_idx, error_buf,
                                           error_buf_size)) {
@@ -7286,7 +7502,7 @@ re_scan:
             case WASM_OP_GET_GLOBAL:
             {
                 p_org = p - 1;
-                read_leb_uint32(p, p_end, global_idx);
+                pb_read_leb_uint32(p, p_end, global_idx);
                 bh_assert(global_idx < global_count);
 
                 global_type = global_idx < module->import_global_count
@@ -7320,7 +7536,7 @@ re_scan:
                 bool is_mutable = false;
 
                 p_org = p - 1;
-                read_leb_uint32(p, p_end, global_idx);
+                pb_read_leb_uint32(p, p_end, global_idx);
                 bh_assert(global_idx < global_count);
 
                 is_mutable = global_idx < module->import_global_count
@@ -7417,8 +7633,8 @@ re_scan:
                 }
 #endif
                 CHECK_MEMORY();
-                read_leb_memarg(p, p_end, align);          /* align */
-                read_leb_mem_offset(p, p_end, mem_offset); /* offset */
+                pb_read_leb_memarg(p, p_end, align);          /* align */
+                pb_read_leb_mem_offset(p, p_end, mem_offset); /* offset */
 #if WASM_ENABLE_FAST_INTERP != 0
                 emit_uint32(loader_ctx, mem_offset);
 #endif
@@ -7479,7 +7695,7 @@ re_scan:
 
             case WASM_OP_MEMORY_SIZE:
                 CHECK_MEMORY();
-                read_leb_memidx(p, p_end, memidx);
+                pb_read_leb_memidx(p, p_end, memidx);
                 check_memidx(module, memidx);
                 PUSH_PAGE_COUNT();
 
@@ -7491,7 +7707,7 @@ re_scan:
 
             case WASM_OP_MEMORY_GROW:
                 CHECK_MEMORY();
-                read_leb_memidx(p, p_end, memidx);
+                pb_read_leb_memidx(p, p_end, memidx);
                 check_memidx(module, memidx);
                 POP_AND_PUSH(mem_offset_type, mem_offset_type);
 
@@ -7506,7 +7722,7 @@ re_scan:
                 break;
 
             case WASM_OP_I32_CONST:
-                read_leb_int32(p, p_end, i32_const);
+                pb_read_leb_int32(p, p_end, i32_const);
 #if WASM_ENABLE_FAST_INTERP != 0
                 skip_label();
                 disable_emit = true;
@@ -7524,7 +7740,7 @@ re_scan:
                 break;
 
             case WASM_OP_I64_CONST:
-                read_leb_int64(p, p_end, i64_const);
+                pb_read_leb_int64(p, p_end, i64_const);
 #if WASM_ENABLE_FAST_INTERP != 0
                 skip_label();
                 disable_emit = true;
@@ -7806,7 +8022,7 @@ re_scan:
             {
                 uint32 opcode1;
 
-                read_leb_uint32(p, p_end, opcode1);
+                pb_read_leb_uint32(p, p_end, opcode1);
 #if WASM_ENABLE_FAST_INTERP != 0
                 emit_byte(loader_ctx, ((uint8)opcode1));
 #endif
@@ -7831,11 +8047,11 @@ re_scan:
                     case WASM_OP_MEMORY_INIT:
                     {
                         CHECK_MEMORY();
-                        read_leb_uint32(p, p_end, segment_index);
+                        pb_read_leb_uint32(p, p_end, segment_index);
 #if WASM_ENABLE_FAST_INTERP != 0
                         emit_uint32(loader_ctx, segment_index);
 #endif
-                        read_leb_memidx(p, p_end, memidx);
+                        pb_read_leb_memidx(p, p_end, memidx);
                         check_memidx(module, memidx);
 
                         bh_assert(segment_index < module->data_seg_count);
@@ -7851,7 +8067,7 @@ re_scan:
                     }
                     case WASM_OP_DATA_DROP:
                     {
-                        read_leb_uint32(p, p_end, segment_index);
+                        pb_read_leb_uint32(p, p_end, segment_index);
 #if WASM_ENABLE_FAST_INTERP != 0
                         emit_uint32(loader_ctx, segment_index);
 #endif
@@ -7867,9 +8083,9 @@ re_scan:
                         CHECK_MEMORY();
                         CHECK_BUF(p, p_end, sizeof(int16));
                         /* check both src and dst memory index */
-                        read_leb_memidx(p, p_end, memidx);
+                        pb_read_leb_memidx(p, p_end, memidx);
                         check_memidx(module, memidx);
-                        read_leb_memidx(p, p_end, memidx);
+                        pb_read_leb_memidx(p, p_end, memidx);
                         check_memidx(module, memidx);
 
                         POP_MEM_OFFSET();
@@ -7883,7 +8099,7 @@ re_scan:
                     case WASM_OP_MEMORY_FILL:
                     {
                         CHECK_MEMORY();
-                        read_leb_memidx(p, p_end, memidx);
+                        pb_read_leb_memidx(p, p_end, memidx);
                         check_memidx(module, memidx);
 
                         POP_MEM_OFFSET();
@@ -7901,8 +8117,8 @@ re_scan:
                         uint8 seg_ref_type, tbl_ref_type;
                         uint32 table_seg_idx, table_idx;
 
-                        read_leb_uint32(p, p_end, table_seg_idx);
-                        read_leb_uint32(p, p_end, table_idx);
+                        pb_read_leb_uint32(p, p_end, table_seg_idx);
+                        pb_read_leb_uint32(p, p_end, table_idx);
 
                         if (!get_table_elem_type(module, table_idx,
                                                  &tbl_ref_type, error_buf,
@@ -7937,7 +8153,7 @@ re_scan:
                     case WASM_OP_ELEM_DROP:
                     {
                         uint32 table_seg_idx;
-                        read_leb_uint32(p, p_end, table_seg_idx);
+                        pb_read_leb_uint32(p, p_end, table_seg_idx);
                         if (!get_table_seg_elem_type(module, table_seg_idx,
                                                      NULL, error_buf,
                                                      error_buf_size))
@@ -7953,13 +8169,13 @@ re_scan:
                         uint32 src_tbl_idx, dst_tbl_idx, src_tbl_idx_type,
                             dst_tbl_idx_type, min_tbl_idx_type;
 
-                        read_leb_uint32(p, p_end, src_tbl_idx);
+                        pb_read_leb_uint32(p, p_end, src_tbl_idx);
                         if (!get_table_elem_type(module, src_tbl_idx,
                                                  &src_ref_type, error_buf,
                                                  error_buf_size))
                             goto fail;
 
-                        read_leb_uint32(p, p_end, dst_tbl_idx);
+                        pb_read_leb_uint32(p, p_end, dst_tbl_idx);
                         if (!get_table_elem_type(module, dst_tbl_idx,
                                                  &dst_ref_type, error_buf,
                                                  error_buf_size))
@@ -8006,7 +8222,7 @@ re_scan:
                     {
                         uint32 table_idx;
 
-                        read_leb_uint32(p, p_end, table_idx);
+                        pb_read_leb_uint32(p, p_end, table_idx);
                         /* TODO: shall we create a new function to check
                                  table idx instead of using below function? */
                         if (!get_table_elem_type(module, table_idx, NULL,
@@ -8031,7 +8247,7 @@ re_scan:
                         uint8 decl_ref_type;
                         uint32 table_idx;
 
-                        read_leb_uint32(p, p_end, table_idx);
+                        pb_read_leb_uint32(p, p_end, table_idx);
                         if (!get_table_elem_type(module, table_idx,
                                                  &decl_ref_type, error_buf,
                                                  error_buf_size))
@@ -8083,15 +8299,15 @@ re_scan:
             {
                 uint32 opcode1;
 
-                read_leb_uint32(p, p_end, opcode1);
+                pb_read_leb_uint32(p, p_end, opcode1);
 
 #if WASM_ENABLE_FAST_INTERP != 0
                 emit_byte(loader_ctx, opcode1);
 #endif
                 if (opcode1 != WASM_OP_ATOMIC_FENCE) {
                     CHECK_MEMORY();
-                    read_leb_uint32(p, p_end, align);          /* align */
-                    read_leb_mem_offset(p, p_end, mem_offset); /* offset */
+                    pb_read_leb_uint32(p, p_end, align);          /* align */
+                    pb_read_leb_mem_offset(p, p_end, mem_offset); /* offset */
 #if WASM_ENABLE_FAST_INTERP != 0
                     emit_uint32(loader_ctx, mem_offset);
 #endif
@@ -8241,29 +8457,25 @@ re_scan:
     if (loader_ctx->p_code_compiled == NULL)
         goto re_scan;
 
-    func->const_cell_num = loader_ctx->const_cell_num;
+    func->const_cell_num =
+        loader_ctx->i64_const_num * 2 + loader_ctx->i32_const_num;
     if (func->const_cell_num > 0) {
-        int32 j;
-
-        if (!(func->consts = func_const = loader_malloc(
-                  func->const_cell_num * 4, error_buf, error_buf_size)))
+        if (!(func->consts =
+                  loader_malloc((uint64)sizeof(uint32) * func->const_cell_num,
+                                error_buf, error_buf_size)))
             goto fail;
-
-        func_const_end = func->consts + func->const_cell_num * 4;
-        /* reverse the const buf */
-        for (j = loader_ctx->num_const - 1; j >= 0; j--) {
-            Const *c = (Const *)(loader_ctx->const_buf + j * sizeof(Const));
-            if (c->value_type == VALUE_TYPE_F64
-                || c->value_type == VALUE_TYPE_I64) {
-                bh_memcpy_s(func_const, (uint32)(func_const_end - func_const),
-                            &(c->value.f64), (uint32)sizeof(int64));
-                func_const += sizeof(int64);
-            }
-            else {
-                bh_memcpy_s(func_const, (uint32)(func_const_end - func_const),
-                            &(c->value.f32), (uint32)sizeof(int32));
-                func_const += sizeof(int32);
-            }
+        if (loader_ctx->i64_const_num > 0) {
+            bh_memcpy_s(func->consts,
+                        (uint32)sizeof(int64) * loader_ctx->i64_const_num,
+                        loader_ctx->i64_consts,
+                        (uint32)sizeof(int64) * loader_ctx->i64_const_num);
+        }
+        if (loader_ctx->i32_const_num > 0) {
+            bh_memcpy_s(func->consts
+                            + sizeof(int64) * loader_ctx->i64_const_num,
+                        (uint32)sizeof(int32) * loader_ctx->i32_const_num,
+                        loader_ctx->i32_consts,
+                        (uint32)sizeof(int32) * loader_ctx->i32_const_num);
         }
     }
 

core/iwasm/interpreter/wasm_opcode.h

@@ -278,6 +278,15 @@ typedef enum WASMOpcode {
     DEBUG_OP_BREAK = 0xdc, /* debug break point */
 #endif
 
+#if WASM_ENABLE_JIT != 0 \
+    || WASM_ENABLE_FAST_INTERP != 0 && WASM_ENABLE_SIMD != 0
+    EXT_OP_SET_LOCAL_FAST_V128 = 0xdd,
+    EXT_OP_TEE_LOCAL_FAST_V128 = 0xde,
+    EXT_OP_COPY_STACK_TOP_V128 = 0xdf,
+    WASM_OP_GET_GLOBAL_V128 = 0xe0,
+    WASM_OP_SET_GLOBAL_V128 = 0xe1,
+#endif
+
     /* Post-MVP extend op prefix */
     WASM_OP_GC_PREFIX = 0xfb,
     WASM_OP_MISC_PREFIX = 0xfc,
@@ -779,16 +788,27 @@ typedef enum WASMAtomicEXTOpcode {
 #else
 #define DEF_DEBUG_BREAK_HANDLE()
 #endif
-
 #define SET_GOTO_TABLE_ELEM(opcode) [opcode] = HANDLE_OPCODE(opcode)
 
-#if WASM_ENABLE_JIT != 0 && WASM_ENABLE_SIMD != 0
+#if (WASM_ENABLE_JIT != 0 || WASM_ENABLE_FAST_INTERP != 0) \
+    && WASM_ENABLE_SIMD != 0
 #define SET_GOTO_TABLE_SIMD_PREFIX_ELEM() \
     SET_GOTO_TABLE_ELEM(WASM_OP_SIMD_PREFIX),
 #else
 #define SET_GOTO_TABLE_SIMD_PREFIX_ELEM()
 #endif
 
+#if (WASM_ENABLE_FAST_INTERP != 0) && WASM_ENABLE_SIMD != 0
+#define DEF_EXT_V128_HANDLE()                                       \
+    SET_GOTO_TABLE_ELEM(EXT_OP_SET_LOCAL_FAST_V128),     /* 0xdd */ \
+        SET_GOTO_TABLE_ELEM(EXT_OP_TEE_LOCAL_FAST_V128), /* 0xde */ \
+        SET_GOTO_TABLE_ELEM(EXT_OP_COPY_STACK_TOP_V128), /* 0xdf */ \
+        SET_GOTO_TABLE_ELEM(WASM_OP_GET_GLOBAL_V128),    /* 0xe0 */ \
+        SET_GOTO_TABLE_ELEM(WASM_OP_SET_GLOBAL_V128),    /* 0xe1 */
+
+#else
+#define DEF_EXT_V128_HANDLE()
+#endif
 /*
  * Macro used to generate computed goto tables for the C interpreter.
  */
@@ -1020,7 +1040,7 @@ typedef enum WASMAtomicEXTOpcode {
         SET_GOTO_TABLE_ELEM(WASM_OP_MISC_PREFIX),    /* 0xfc */ \
         SET_GOTO_TABLE_SIMD_PREFIX_ELEM()            /* 0xfd */ \
         SET_GOTO_TABLE_ELEM(WASM_OP_ATOMIC_PREFIX),  /* 0xfe */ \
-        DEF_DEBUG_BREAK_HANDLE()                                \
+        DEF_DEBUG_BREAK_HANDLE() DEF_EXT_V128_HANDLE()          \
     };
 
 #ifdef __cplusplus

core/iwasm/interpreter/wasm_runtime.c

@@ -1208,9 +1208,8 @@ globals_instantiate(WASMModule *module, WASMModuleInstance *module_inst,
 
             /* The linked global instance has been initialized, we
                just need to copy the value. */
-            bh_memcpy_s(&(global->initial_value), sizeof(WASMValue),
-                        &(global_import->import_global_linked->init_expr),
-                        sizeof(WASMValue));
+            global->initial_value =
+                global_import->import_global_linked->init_expr.u;
         }
         else
 #endif
@@ -2467,7 +2466,8 @@ wasm_instantiate(WASMModule *module, WASMModuleInstance *parent,
             goto fail;
         }
         for (i = 0; i < module->table_seg_count; i++) {
-            if (wasm_elem_is_active(module->table_segments[i].mode))
+            if (wasm_elem_is_active(module->table_segments[i].mode)
+                || wasm_elem_is_declarative(module->table_segments[i].mode))
                 bh_bitmap_set_bit(module_inst->e->common.elem_dropped, i);
         }
     }
@@ -3523,13 +3523,6 @@ call_wasm_with_hw_bound_check(WASMModuleInstance *module_inst,
 #endif
     bool ret = true;
 
-    /* Check native stack overflow firstly to ensure we have enough
-       native stack to run the following codes before actually calling
-       the aot function in invokeNative function. */
-    if (!wasm_runtime_detect_native_stack_overflow(exec_env)) {
-        return;
-    }
-
     if (!exec_env_tls) {
         if (!os_thread_signal_inited()) {
             wasm_set_exception(module_inst, "thread signal env not inited");
@@ -3548,6 +3541,13 @@ call_wasm_with_hw_bound_check(WASMModuleInstance *module_inst,
         }
     }
 
+    /* Check native stack overflow firstly to ensure we have enough
+       native stack to run the following codes before actually calling
+       the aot function in invokeNative function. */
+    if (!wasm_runtime_detect_native_stack_overflow(exec_env)) {
+        return;
+    }
+
     wasm_exec_env_push_jmpbuf(exec_env, &jmpbuf_node);
 
     if (os_setjmp(jmpbuf_node.jmpbuf) == 0) {
@@ -4195,6 +4195,55 @@ wasm_get_module_inst_mem_consumption(const WASMModuleInstance *module_inst,
 #endif /* end of (WASM_ENABLE_MEMORY_PROFILING != 0) \
                  || (WASM_ENABLE_MEMORY_TRACING != 0) */
 
+#if WAMR_ENABLE_COPY_CALLSTACK != 0
+uint32
+wasm_interp_copy_callstack(WASMExecEnv *exec_env, wasm_frame_t *buffer,
+                           uint32 length, uint32 skip_n, char *error_buf,
+                           uint32_t error_buf_size)
+{
+    /*
+     * Note for devs: please refrain from such modifications inside of
+     * wasm_interp_copy_callstack
+     * - 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
+     */
+    WASMModuleInstance *module_inst =
+        (WASMModuleInstance *)wasm_exec_env_get_module_inst(exec_env);
+    WASMInterpFrame *cur_frame = wasm_exec_env_get_cur_frame(exec_env);
+    uint8 *top_boundary = exec_env->wasm_stack.top_boundary;
+    uint8 *bottom = exec_env->wasm_stack.bottom;
+    uint32 count = 0;
+
+    WASMCApiFrame record_frame;
+    while (cur_frame && (uint8_t *)cur_frame >= bottom
+           && (uint8_t *)cur_frame + sizeof(WASMInterpFrame) <= top_boundary
+           && count < (skip_n + length)) {
+        if (!cur_frame->function) {
+            cur_frame = cur_frame->prev_frame;
+            continue;
+        }
+        if (count < skip_n) {
+            ++count;
+            cur_frame = cur_frame->prev_frame;
+            continue;
+        }
+        record_frame.instance = module_inst;
+        record_frame.module_offset = 0;
+        // It's safe to dereference module_inst->e because "e" is asigned only
+        // once in wasm_instantiate
+        record_frame.func_index =
+            (uint32)(cur_frame->function - module_inst->e->functions);
+        buffer[count - skip_n] = record_frame;
+        cur_frame = cur_frame->prev_frame;
+        ++count;
+    }
+    return count >= skip_n ? count - skip_n : 0;
+}
+#endif // WAMR_ENABLE_COPY_CALLSTACK
+
 #if WASM_ENABLE_DUMP_CALL_STACK != 0
 bool
 wasm_interp_create_call_stack(struct WASMExecEnv *exec_env)
@@ -4660,7 +4709,7 @@ llvm_jit_table_init(WASMModuleInstance *module_inst, uint32 tbl_idx,
             if (!(func_obj = wasm_create_func_obj(module_inst,
                                                   init_values[i].u.ref_index,
                                                   true, NULL, 0))) {
-                wasm_set_exception(module_inst, "null function object");
+                wasm_set_exception(module_inst, "null function reference");
                 return;
             }
             table_elems[i] = func_obj;

core/iwasm/interpreter/wasm_runtime.h

@@ -730,6 +730,14 @@ wasm_get_table_inst(const WASMModuleInstance *module_inst, uint32 tbl_idx)
 }
 
 #if WASM_ENABLE_DUMP_CALL_STACK != 0
+
+#if WAMR_ENABLE_COPY_CALLSTACK != 0
+uint32
+wasm_interp_copy_callstack(WASMExecEnv *exec_env, wasm_frame_t *buffer,
+                           uint32 length, uint32 skip_n, char *error_buf,
+                           uint32_t error_buf_size);
+#endif // WAMR_ENABLE_COPY_CALLSTACK
+
 bool
 wasm_interp_create_call_stack(struct WASMExecEnv *exec_env);
 

core/iwasm/libraries/debug-engine/debug_engine.c

@@ -58,7 +58,7 @@ static WASMDebugEngine *g_debug_engine;
 static uint32 current_instance_id = 1;
 
 static uint32
-allocate_instance_id()
+allocate_instance_id(void)
 {
     uint32 id;
 
@@ -302,7 +302,7 @@ wasm_debug_control_thread_destroy(WASMDebugInstance *debug_instance)
 }
 
 static WASMDebugEngine *
-wasm_debug_engine_create()
+wasm_debug_engine_create(void)
 {
     WASMDebugEngine *engine;
 
@@ -326,7 +326,7 @@ wasm_debug_engine_create()
 }
 
 void
-wasm_debug_engine_destroy()
+wasm_debug_engine_destroy(void)
 {
     if (g_debug_engine) {
         wasm_debug_handler_deinit();

core/iwasm/libraries/debug-engine/debug_engine.h

@@ -133,7 +133,7 @@ bool
 wasm_debug_engine_init(char *ip_addr, int32 process_port);
 
 void
-wasm_debug_engine_destroy();
+wasm_debug_engine_destroy(void);
 
 WASMExecEnv *
 wasm_debug_instance_get_current_env(WASMDebugInstance *instance);

core/iwasm/libraries/debug-engine/gdbserver.c

@@ -38,7 +38,7 @@ static const struct packet_handler_elem packet_handler_table[255] = {
 };
 
 WASMGDBServer *
-wasm_create_gdbserver(const char *host, int32 *port)
+wasm_create_gdbserver(const char *host, int *port)
 {
     bh_socket_t listen_fd = (bh_socket_t)-1;
     WASMGDBServer *server;

core/iwasm/libraries/debug-engine/gdbserver.h

@@ -51,7 +51,7 @@ typedef struct WASMGDBServer {
 } WASMGDBServer;
 
 WASMGDBServer *
-wasm_create_gdbserver(const char *host, int32 *port);
+wasm_create_gdbserver(const char *host, int *port);
 
 bool
 wasm_gdbserver_listen(WASMGDBServer *server);

core/iwasm/libraries/debug-engine/handler.c

@@ -34,7 +34,7 @@ static char *tmpbuf;
 static korp_mutex tmpbuf_lock;
 
 int
-wasm_debug_handler_init()
+wasm_debug_handler_init(void)
 {
     int ret;
     tmpbuf = wasm_runtime_malloc(MAX_PACKET_SIZE);
@@ -51,7 +51,7 @@ wasm_debug_handler_init()
 }
 
 void
-wasm_debug_handler_deinit()
+wasm_debug_handler_deinit(void)
 {
     wasm_runtime_free(tmpbuf);
     tmpbuf = NULL;
@@ -204,8 +204,7 @@ handle_general_query(WASMGDBServer *server, char *payload)
     if (!strcmp(name, "Supported")) {
         os_mutex_lock(&tmpbuf_lock);
         snprintf(tmpbuf, MAX_PACKET_SIZE,
-                 "qXfer:libraries:read+;PacketSize=%" PRIx32 ";",
-                 MAX_PACKET_SIZE);
+                 "qXfer:libraries:read+;PacketSize=%x;", MAX_PACKET_SIZE);
         write_packet(server, tmpbuf);
         os_mutex_unlock(&tmpbuf_lock);
     }
@@ -384,7 +383,7 @@ send_thread_stop_status(WASMGDBServer *server, uint32 status, korp_tid tid)
 
     if (status == 0) {
         os_mutex_lock(&tmpbuf_lock);
-        snprintf(tmpbuf, MAX_PACKET_SIZE, "W%02x", status);
+        snprintf(tmpbuf, MAX_PACKET_SIZE, "W%02" PRIx32, status);
         write_packet(server, tmpbuf);
         os_mutex_unlock(&tmpbuf_lock);
         return;
@@ -400,8 +399,9 @@ send_thread_stop_status(WASMGDBServer *server, uint32 status, korp_tid tid)
 
     os_mutex_lock(&tmpbuf_lock);
     // TODO: how name a wasm thread?
-    len += snprintf(tmpbuf, MAX_PACKET_SIZE, "T%02xthread:%" PRIx64 ";name:%s;",
-                    gdb_status, (uint64)(uintptr_t)tid, "nobody");
+    len += snprintf(tmpbuf, MAX_PACKET_SIZE,
+                    "T%02" PRIx32 "thread:%" PRIx64 ";name:%s;", gdb_status,
+                    (uint64)(uintptr_t)tid, "nobody");
     if (tids_count > 0) {
         len += snprintf(tmpbuf + len, MAX_PACKET_SIZE - len, "threads:");
         while (i < tids_count) {
@@ -624,7 +624,8 @@ void
 handle_get_write_memory(WASMGDBServer *server, char *payload)
 {
     size_t hex_len;
-    int32 offset, act_len;
+    int offset;
+    int32 act_len;
     uint64 maddr, mlen;
     char *buff;
     bool ret;

core/iwasm/libraries/debug-engine/handler.h

@@ -9,10 +9,10 @@
 #include "gdbserver.h"
 
 int
-wasm_debug_handler_init();
+wasm_debug_handler_init(void);
 
 void
-wasm_debug_handler_deinit();
+wasm_debug_handler_deinit(void);
 
 void
 handle_interrupt(WASMGDBServer *server);

core/iwasm/libraries/lib-rats/lib_rats.cmake

@@ -2,6 +2,13 @@
 # Copyright (c) 2020-2021 Alibaba Cloud
 # 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 (LIB_RATS_DIR ${CMAKE_CURRENT_LIST_DIR})
 
 if ("$ENV{SGX_SSL_DIR}" STREQUAL "")

core/iwasm/libraries/lib-socket/inc/wasi_socket_ext.h

@@ -991,7 +991,7 @@ __wasi_sock_get_ipv6_only(__wasi_fd_t fd, bool *option)
 /**
  * TODO: modify recv() and send()
  * since don't want to re-compile the wasi-libc,
- * we tend to keep original implentations of recv() and send().
+ * we tend to keep original implementations of recv() and send().
  */
 
 #ifdef __cplusplus

core/iwasm/libraries/lib-socket/lib_socket_wasi.cmake

@@ -1,7 +1,7 @@
 # Copyright (C) 2019 Intel Corporation.  All rights reserved.
 # SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 
-cmake_minimum_required (VERSION 2.8...3.16)
+cmake_minimum_required (VERSION 3.14)
 
 project(socket_wasi_ext LANGUAGES C)