Merge branch main into dev/wasi-libc-windows

2025-02-06 15:05:19 +00:00 · 2023-08-23 17:05:19 +08:00 · 2023-08-23 17:05:19 +08:00 · b7a9da1620
commit b7a9da1620
parent 834e0314aa fa2f29fd8a
77 changed files with 1903 additions and 588 deletions
--- a/.github/workflows/build_wamr_lldb.yml
+++ b/.github/workflows/build_wamr_lldb.yml
@ -27,7 +27,6 @@ on:
        type: string
        required: false
        default: "https://github.com/WebAssembly/wasi-sdk/releases/download/wasi-sdk-20/wasi-sdk-20.0-linux.tar.gz"
-      

 jobs:
  try_reuse:
@ -35,7 +34,7 @@ jobs:
    with:
      binary_name_stem: "wamr-lldb-${{ inputs.ver_num }}-${{ inputs.arch }}-${{ inputs.runner }}"
      last_commit: "ea63ba4bd010c2285623ad4acc0262a4d63bcfea"
-      the_path: "./build-scripts/lldb-wasm.patch"
+      the_path: "./build-scripts/lldb_wasm.patch"
      upload_url: ${{ inputs.upload_url }}

  build:
@ -49,7 +48,7 @@ jobs:
      PYTHON_MACOS_STANDALONE_BUILD: https://github.com/indygreg/python-build-standalone/releases/download/20230507/cpython-3.10.11+20230507-x86_64-apple-darwin-install_only.tar.gz
    steps:
      - uses: actions/checkout@v3
-      
+
      - name: download and install wasi-sdk
        run: |
          cd /opt
@ -84,8 +83,8 @@ jobs:
        if: steps.lldb_build_cache.outputs.cache-hit != 'true' && contains(inputs.runner, 'macos')
        run: |
          brew remove swig
-          brew install swig@3 cmake ninja libedit
-          brew link --overwrite swig@3
+          brew install swig@4.1 cmake ninja libedit
+          brew link --overwrite swig@4.1
          sudo rm -rf /Library/Developer/CommandLineTools

      - name: install utils ubuntu
@ -107,7 +106,7 @@ jobs:
          git init
          git config user.email "action@github.com"
          git config user.name "github action"
-          git apply ../../../build-scripts/lldb-wasm.patch
+          git apply ../../../build-scripts/lldb_wasm.patch
        working-directory: core/deps/llvm-project

      - name: get stand-alone python ubuntu
@ -163,6 +162,7 @@ jobs:
          mkdir -p wamr-debug
          cmake -S product-mini/platforms/linux -B wamr-debug -DWAMR_BUILD_DEBUG_INTERP=1
          cmake --build wamr-debug --parallel $(nproc)
+          export LD_LIBRARY_PATH=$(pwd)/core/deps/python/lib:${LD_LIBRARY_PATH}
          python3 ci/validate_lldb.py --port 1239 --lldb core/deps/wamr-lldb/bin/lldb --wamr wamr-debug/iwasm --verbose
        working-directory: .

@ -205,7 +205,7 @@ jobs:
          mkdir -p wamr-lldb/lib
          cp build/bin/lldb* wamr-lldb/bin
          cp lldb/tools/lldb-vscode/package.json wamr-lldb
-          cp -r lldb/tools/lldb-vscode/syntaxes/ wamr-lldb 
+          cp -r lldb/tools/lldb-vscode/syntaxes/ wamr-lldb
        working-directory: core/deps/llvm-project

      - name: pack ubuntu specific libraries
@ -226,7 +226,7 @@ jobs:
          cp -R ../python/lib/python* wamr-lldb/lib
          cp ../python/lib/libpython*.dylib wamr-lldb/lib
          install_name_tool -change /install/lib/libpython${{ env.PYTHON_VERSION }}.dylib @rpath/libpython${{ env.PYTHON_VERSION }}.dylib wamr-lldb/lib/liblldb.*.dylib
-        # Patch path of python library -> https://github.com/indygreg/python-build-standalone/blob/85923ca3911784e6978b85d56e06e9ae75cb2dc4/docs/quirks.rst?plain=1#L412-L446  
+        # Patch path of python library -> https://github.com/indygreg/python-build-standalone/blob/85923ca3911784e6978b85d56e06e9ae75cb2dc4/docs/quirks.rst?plain=1#L412-L446
        working-directory: core/deps/llvm-project

      - name: compress the binary
--- a/.github/workflows/compilation_on_android_ubuntu.yml
+++ b/.github/workflows/compilation_on_android_ubuntu.yml
@ -483,6 +483,16 @@ jobs:
          sudo tar -xzf wasi-sdk-*.tar.gz
          sudo mv wasi-sdk-20.0 wasi-sdk

+      # It is a temporary solution until new wasi-sdk that includes bug fixes is released 
+      - name: build wasi-libc from source
+        if: matrix.test_option == '$WASI_TEST_OPTIONS'
+        run: |
+          git clone https://github.com/WebAssembly/wasi-libc
+          cd wasi-libc
+          make -j AR=/opt/wasi-sdk/bin/llvm-ar NM=/opt/wasi-sdk/bin/llvm-nm CC=/opt/wasi-sdk/bin/clang THREAD_MODEL=posix
+          echo "SYSROOT_PATH=$PWD/sysroot" >> $GITHUB_ENV
+
+
      - name: set env variable(if llvm are used)
        if: matrix.running_mode == 'aot' || matrix.running_mode == 'jit' || matrix.running_mode == 'multi-tier-jit'
        run: echo "USE_LLVM=true" >> $GITHUB_ENV
@ -518,7 +528,7 @@ jobs:

      - name: Build WASI thread tests
        if: matrix.test_option == '$WASI_TEST_OPTIONS'
-        run: bash build.sh
+        run: bash build.sh --sysroot "$SYSROOT_PATH"
        working-directory: ./core/iwasm/libraries/lib-wasi-threads/test/

      - name: build socket api tests
@ -527,7 +537,7 @@ jobs:
        working-directory: ./core/iwasm/libraries/lib-socket/test/

      - name: run tests
-        timeout-minutes: 20
+        timeout-minutes: 30
        run: ./test_wamr.sh ${{ matrix.test_option }} -t ${{ matrix.running_mode }}
        working-directory: ./tests/wamr-test-suites

@ -543,7 +553,7 @@ jobs:
          sudo apt install -y g++-multilib lib32gcc-9-dev

      - name: run tests x86_32
-        timeout-minutes: 20
+        timeout-minutes: 30
        if: env.TEST_ON_X86_32 == 'true'
        run: ./test_wamr.sh ${{ env.X86_32_TARGET_TEST_OPTIONS }} ${{ matrix.test_option }} -t ${{ matrix.running_mode }}
        working-directory: ./tests/wamr-test-suites
@ -584,7 +594,7 @@ jobs:
          cache-name: cache-lldb-vscode
        with:
          path: test-tools/wamr-ide/VSCode-Extension/resource/debug/linux
-          key: ${{ env.cache-name }}-${{ hashFiles('build-scripts/lldb-wasm.patch') }}-${{ env.PYTHON_UBUNTU_STANDALONE_BUILD }}
+          key: ${{ env.cache-name }}-${{ hashFiles('build-scripts/lldb_wasm.patch') }}-${{ env.PYTHON_UBUNTU_STANDALONE_BUILD }}

      - if: ${{ steps.cache-lldb.outputs.cache-hit != 'true' }}
        name: get stand-alone python ubuntu
@ -607,7 +617,7 @@ jobs:
          git init
          git config user.email "action@github.com"
          git config user.name "github action"
-          git apply ../../../build-scripts/lldb-wasm.patch
+          git apply ../../../build-scripts/lldb_wasm.patch
        working-directory: core/deps/llvm-project

      - if: ${{ steps.cache-lldb.outputs.cache-hit != 'true' }}
--- a/.github/workflows/nightly_run.yml
+++ b/.github/workflows/nightly_run.yml
@ -548,6 +548,16 @@ jobs:
          sudo wget ${{ matrix.wasi_sdk_release }}
          sudo tar -xzf wasi-sdk-*.tar.gz
          sudo mv wasi-sdk-20.0 wasi-sdk
+
+      # It is a temporary solution until new wasi-sdk that includes bug fixes is released 
+      - name: build wasi-libc from source
+        if: matrix.test_option == '$WASI_TEST_OPTIONS'
+        run: |
+          git clone https://github.com/WebAssembly/wasi-libc
+          cd wasi-libc
+          make -j AR=/opt/wasi-sdk/bin/llvm-ar NM=/opt/wasi-sdk/bin/llvm-nm CC=/opt/wasi-sdk/bin/clang THREAD_MODEL=posix
+          echo "SYSROOT_PATH=$PWD/sysroot" >> $GITHUB_ENV
+
      - name: set env variable(if llvm are used)
        if: matrix.running_mode == 'aot' || matrix.running_mode == 'jit' || matrix.running_mode == 'multi-tier-jit'
        run: echo "USE_LLVM=true" >> $GITHUB_ENV
@ -586,7 +596,7 @@ jobs:

      - name: Build WASI thread tests
        if: matrix.test_option == '$WASI_TEST_OPTIONS'
-        run: bash build.sh
+        run: bash build.sh --sysroot "$SYSROOT_PATH"
        working-directory: ./core/iwasm/libraries/lib-wasi-threads/test/

      - name: build socket api tests
@ -595,7 +605,7 @@ jobs:
        working-directory: ./core/iwasm/libraries/lib-socket/test/

      - name: run tests
-        timeout-minutes: 20
+        timeout-minutes: 40
        run: ./test_wamr.sh ${{ matrix.test_option }} -t ${{ matrix.running_mode }}
        working-directory: ./tests/wamr-test-suites

@ -611,7 +621,7 @@ jobs:
          sudo apt install -y g++-multilib lib32gcc-9-dev

      - name: run tests x86_32
-        timeout-minutes: 20
+        timeout-minutes: 40
        if: env.TEST_ON_X86_32 == 'true'
        run: ./test_wamr.sh ${{ env.X86_32_TARGET_TEST_OPTIONS }} ${{ matrix.test_option }} -t ${{ matrix.running_mode }}
        working-directory: ./tests/wamr-test-suites
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@ -107,6 +107,11 @@ endif ()

 set (WAMR_ROOT_DIR ${CMAKE_CURRENT_SOURCE_DIR})

+# Set the strip command based on the system (GNU or Clang)
+if (CMAKE_STRIP)
+    set (CMAKE_STRIP_FLAGS "--strip-all")
+endif ()
+
 include (${WAMR_ROOT_DIR}/build-scripts/runtime_lib.cmake)

 set (CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -Wall -Wextra -Wformat -Wformat-security -Wshadow -Wno-unused-parameter")
@ -141,6 +146,15 @@ endif ()

 install (TARGETS iwasm_static ARCHIVE DESTINATION lib)

+# If it's a Release build, strip the static library
+if (CMAKE_STRIP AND CMAKE_BUILD_TYPE STREQUAL "Release")
+  # Strip static library
+  message (STATUS "Stripping static library after build!")
+  add_custom_command (TARGET iwasm_static POST_BUILD
+      COMMAND ${CMAKE_STRIP} ${CMAKE_STRIP_FLAGS} $<TARGET_FILE:iwasm_static>
+  )
+endif ()
+
 # SHARED LIBRARY
 add_library (iwasm_shared SHARED ${WAMR_RUNTIME_LIB_SOURCE})
 set_target_properties (iwasm_shared PROPERTIES OUTPUT_NAME iwasm)
@ -162,3 +176,12 @@ install (FILES
    ${WAMR_ROOT_DIR}/core/iwasm/include/wasm_export.h
    ${WAMR_ROOT_DIR}/core/iwasm/include/lib_export.h
    DESTINATION include)
+
+# If it's a Release build, strip the shared library
+if (CMAKE_STRIP AND CMAKE_BUILD_TYPE STREQUAL "Release")
+  # Strip shared library
+  message (STATUS "Stripping shared library after build!")
+  add_custom_command (TARGET iwasm_shared POST_BUILD
+      COMMAND ${CMAKE_STRIP} ${CMAKE_STRIP_FLAGS} $<TARGET_FILE:iwasm_shared>
+  )
+endif ()
--- a/RELEASE_NOTES.md
+++ b/RELEASE_NOTES.md
@ -1,3 +1,139 @@
+## WAMR-1.2.3
+
+### Breaking Changes
+- Increase default native stack size (#2332)
+
+### New Features
+- Implement the segue optimization for LLVM AOT/JIT (#2230)
+- Implement AOT static PGO (#2243)
+- Enable static PGO for Linux SGX (#2270)
+- Add Rust Formatters to Debugger (Vector, Map etc.) (#2219)
+
+### Bug Fixes
+- The Python language-binding needs python>=3.9 (#2228)
+- aot_compile_op_call: Remove a wrong optimization (#2233)
+- Fix typo in samples/ref-types (#2236)
+- Update thread proposal ignore cases (#2246)
+- Disable writting GS register on linux-sgx platform (#2255)
+- Fix compile error of wamrc with llvm-13/llvm-14 (#2261)
+- aot/jit: Set module layout (#2260)
+- Fix build error with LLVM 16 (#2259)
+- spec-test-script: Disable conversions.wast on i386 (#2269)
+- Fix a heap corruption bug in ems realloc (#2279)
+- Fix fast-interp issue of LAST_OP_OUTPUT_I32/64 check (#2295)
+- Fix wamrc build issues with LLVM 13 and LLVM 16 (#2313)
+- aot: Move stack_sizes table to a dedicated section (#2317)
+- product-mini/platforms/linux: Mark vmlib POSITION_INDEPENDENT_CODE (#2323)
+- aot: Avoid possible relocations around "stack_sizes" for XIP mode (#2322)
+- Avoid switch lowering to lookup tables for XIP (#2339)
+- Fix typo in zephyr's Dockerfile.old (#2354)
+- Fix typo (dwarf) in the codebase (#2367)
+- Implement suspend flags as atomic variable (#2361)
+- Fix llvm jit failed to lookup aot_stack_sizes symbol issue (#2384)
+- Fix some check issues on table operations (#2392)
+- Fix ExpandMemoryOpPass doesn't work properly (#2399)
+- Fix non-builtin BH_ATOMIC_32_FETCH_OR and BH_ATOMIC_32_FETCH_AND (#2400)
+- Fix wasi-sockets tests (#2389)
+- Fix result arity check on select_t opcode (#2406)
+- Re-organize intrinsics in aot_reloc_riscv.c to fix some FPU issues (#2414)
+- Fix lib-pthread issues (#2410)
+- Fix typo in test_wamr.sh (#2421)
+- Fix memory sharing (#2415)
+- wasm_export.h: Fix struct wasm_val_t (#2435)
+- Fix typos in wamrc print_help() (#2442)
+- iwasm: Fix native lib cleanup after error occurs (#2443)
+- Correct --heap-size option in messages (#2458)
+- wasm_instantiate: Fix a potential integer overflow issue (#2459)
+- Fix windows link error and clear windows warnings (#2463)
+- aot: Disable musttail for mips (#2457)
+- Fix opcode overwrite issue in fast interp (#2476)
+- wamrc: Fix windows relocation to `aot_func_internal#n` (#2474)
+- Fix windows AOT hw bound check (#2475)
+- Fix typo in aot_emit_aot_file.c (#2478)
+
+### Enhancements
+- A few changes related to WAMRC_LLC_COMPILER (#2218)
+- Enhance linux-sgx CI (#2102)
+- Add asan and ubsan to WAMR CI (#2161)
+- Update doc on WAMR_DISABLE_HW_BOUND_CHECK 32-bit (#2262)
+- wamrc: Add an incompatibility note in the help message (#2276)
+- Add cmake variable to disable writing gs register (#2284)
+- Make hmu_tree_node 4 byte aligned to reduce compiler warning (#2268)
+- Appease unused warning on min_uint64 (#2277)
+- Fix format warning by PRIu32 in [wasm|aot] dump call stack  (#2251)
+- Fix a compile warning due to missing include (#2293)
+- Fix dockerfile linter warnings (#2291)
+- Enable windows x86-32 AOT relocations (#2285)
+- wamr-ide: Add vscode extension tests (#2292)
+- AOT/JIT native stack bound check improvement (#2244)
+- Add retries to flaky step in nightly run CI (#2306)
+- Use system libuv if available (#1861)
+- wasi-nn: Simplify cmake and headers' location (#2308)
+- wasi-nn: Improve tests paths for local dev (#2309)
+- aot: Implement a few more relocation types for riscv (#2318)
+- wasi-nn: Add support of wasi-nn as shared lib (#2310)
+- Add a few more assertions on structures to which aot abi is sensitive (#2326)
+- Fix sanitizer errors in posix socket  (#2331)
+- Add "--xip" option for wamrc (#2336)
+- Add "--enable-llvm-passes=<passes>" option to wamrc (#2335)
+- Make memory access boundary check behavior configurable (#2289)
+- Migrate ExpandMemoryOpPass to llvm new pass manager (#2334)
+- Allow defining hints without exact socket type or address family (#2337)
+- wamrc: Warn on text relocations for XIP (#2340)
+- Add scripts to validate lldb source debugger (#2150)
+- Add docker file to fix Zephy ESP32 linking issue (#2314)
+- Add "--native-lib=<lib>" option to wamrc (#2342)
+- Fix unused warnings on disable_bounds_checks (#2347)
+- Add "--enable-builtin-intrinsics=<flags>" option to wamrc (#2341)
+- nuttx: Add a kconfig for wasi-threads (#2343)
+- iwasm: Disable app heap by default if wasi is enabled (#2346)
+- Fix some static scan issues (#2362)
+- Bring up WAMR on esp32-s3 device (#2348)
+- ESP-IDF platform supports to load AOT to PSRAM and run it (#2385)
+- Add hadolint CI for Dockerfile linting (#2387)
+- Move generic parts of wasm_suspend_flags.h to bh_atomic.h (#2393)
+- bh_atomic.h: Add comments (#2398)
+- bh_atomic.h: Add BH_ATOMIC_32_FETCH_ADD/BH_ATOMIC_32_FETCH_SUB (#2408)
+- Update libuv version to v1.46.0 (#2405)
+- Remove a few unused functions (#2409)
+- Add initial stress test (#2364)
+- Move wasm_runtime_destroy_wasi and wasi_nn_destroy calls together (#2418)
+- embed_wamr.md: Improvements about threads (#2420)
+- Add runtime inited checks in Enclave command handlings to improve security (#2416)
+- Add some relocation symbols for xtensa target (#2422)
+- Remove unnecessary and extra zero length check in mem functions' macro (#2428)
+- Introduce WASMModuleInstanceExtraCommon (#2429)
+- Strip static and shared libraries of iwasm to reduce the binary size (#2431)
+- Auto-check wrgsbase in cmake script (#2437)
+- iwasm: call native lib init/deinit if exists (#2439)
+- wasi-nn: Support uint8 quantized networks (#2433)
+- Implement `wasm_externref_objdel` and `wasm_externref_set_cleanup` (#2455)
+- wasi-nn: Improve TPU support (#2447)
+- wamr-python: Enable debugging WASM and grant dir access (#2449)
+- Build wasi-libc from source in WAMR CI (#2465)
+- wamrc: More friendly to print help info (#2451)
+- Add another wamr test (#2411)
+- Fix issues reported by Coverity and clear windows warnings (#2467)
+- Clone the input binary during wasm_module_validate (#2483)
+
+### Others
+- Nuttx CI: Ignore the expired certificate for riscv gcc toolchain (#2222)
+- core/iwasm/compilation: constify a bit (#2223)
+- Bump requests from 2.28.2 to 2.31.0 in /build-scripts (#2229)
+- dwarf_extractor: Constify a bit (#2278)
+- AOTFuncContext: Remove a stale comment (#2283)
+- Add performance tunning document (#2286)
+- Reduce CI jobs number (#2296)
+- CI: Update used node version to 16 (#2303)
+- Update Docker image for latest version of external libraries & tools (#2374)
+- Upgrade cJSON version to v1.7.16 (#2404)
+- Upgrade XNNPACK workload (#2394)
+- Build more benchmarks in workload XNNPACK (#2417)
+- Upgrade SGX-RA integration for 0.1.2 and Ubuntu 20.04 (#2454)
+- Add sample pre-commit hook (#2470)
+
+---
+
 ## WAMR-1.2.2

 ### Breaking Changes
--- a/build-scripts/config_common.cmake
+++ b/build-scripts/config_common.cmake
@ -359,16 +359,16 @@ endif ()
 if (WAMR_BUILD_WASI_NN EQUAL 1)
  message ("     WASI-NN enabled")
  add_definitions (-DWASM_ENABLE_WASI_NN=1)
-  if (WASI_NN_ENABLE_GPU EQUAL 1)
+  if (WAMR_BUILD_WASI_NN_ENABLE_GPU EQUAL 1)
      message ("     WASI-NN: GPU enabled")
-      add_definitions (-DWASI_NN_ENABLE_GPU=1)
+      add_definitions (-DWASM_ENABLE_WASI_NN_GPU=1)
  endif ()
-  if (WAMR_BUILD_WASI_NN_ENABLE_EXT EQUAL 1)
+  if (WAMR_BUILD_WASI_NN_ENABLE_EXTERNAL_DELEGATE EQUAL 1)
      message ("     WASI-NN: External Delegation enabled")
-      add_definitions (-DWASI_NN_ENABLE_EXTERNAL_DELEGATE=1)
+      add_definitions (-DWASM_ENABLE_WASI_NN_EXTERNAL_DELEGATE=1)
  endif ()
-  if (DEFINED WASI_NN_EXT_DELEGATE_PATH)
-      add_definitions (-DWASI_NN_EXT_DELEGATE_PATH="${WASI_NN_EXT_DELEGATE_PATH}")
+  if (DEFINED WAMR_BUILD_WASI_NN_EXTERNAL_DELEGATE_PATH)
+      add_definitions (-DWASM_WASI_NN_EXTERNAL_DELEGATE_PATH="${WAMR_BUILD_WASI_NN_EXTERNAL_DELEGATE_PATH}")
  endif ()
 endif ()
 if (WAMR_BUILD_ALLOC_WITH_USER_DATA EQUAL 1)
@ -395,6 +395,30 @@ endif ()
 if (WAMR_DISABLE_WRITE_GS_BASE EQUAL 1)
  add_definitions (-DWASM_DISABLE_WRITE_GS_BASE=1)
  message ("     Write linear memory base addr to x86 GS register disabled")
+elseif (WAMR_BUILD_TARGET STREQUAL "X86_64"
+        AND WAMR_BUILD_PLATFORM STREQUAL "linux")
+  set (TEST_WRGSBASE_SOURCE "${CMAKE_BINARY_DIR}/test_wrgsbase.c")
+  file (WRITE "${TEST_WRGSBASE_SOURCE}" "
+  #include <stdio.h>
+  #include <stdint.h>
+  int main() {
+      uint64_t value;
+      asm volatile (\"wrgsbase %0\" : : \"r\"(value));
+      printf(\"WRGSBASE instruction is available.\\n\");
+      return 0;
+  }")
+  # Try to compile and run the test program
+  try_run (TEST_WRGSBASE_RESULT
+    TEST_WRGSBASE_COMPILED
+    ${CMAKE_BINARY_DIR}/test_wrgsbase
+    SOURCES ${TEST_WRGSBASE_SOURCE}
+    CMAKE_FLAGS -DCMAKE_C_COMPILER=${CMAKE_C_COMPILER}
+  )
+  #message("${TEST_WRGSBASE_COMPILED}, ${TEST_WRGSBASE_RESULT}")
+  if (NOT TEST_WRGSBASE_RESULT EQUAL 0)
+    add_definitions (-DWASM_DISABLE_WRITE_GS_BASE=1)
+    message ("     Write linear memory base addr to x86 GS register disabled")
+  endif ()
 endif ()
 if (WAMR_CONFIGUABLE_BOUNDS_CHECKS EQUAL 1)
  add_definitions (-DWASM_CONFIGURABLE_BOUNDS_CHECKS=1)
--- a/build-scripts/lldb_wasm.patch
+++ b/build-scripts/lldb_wasm.patch
@ -1,5 +1,44 @@
+diff --git a/lldb/bindings/CMakeLists.txt b/lldb/bindings/CMakeLists.txt
+index 9759b069fdc4..25b427f8bcf2 100644
+--- a/lldb/bindings/CMakeLists.txt
+++ b/lldb/bindings/CMakeLists.txt
+@@ -26,8 +26,6 @@ set(SWIG_COMMON_FLAGS
+   -features autodoc
+   -I${LLDB_SOURCE_DIR}/include
+   -I${CMAKE_CURRENT_SOURCE_DIR}
+-  -D__STDC_LIMIT_MACROS
+-  -D__STDC_CONSTANT_MACROS
+   ${DARWIN_EXTRAS}
+ )
+ 
+diff --git a/lldb/bindings/interfaces.swig b/lldb/bindings/interfaces.swig
+index c9a6d0f06056..021c7683d170 100644
+--- a/lldb/bindings/interfaces.swig
+++ b/lldb/bindings/interfaces.swig
+@@ -1,8 +1,5 @@
+ /* Various liblldb typedefs that SWIG needs to know about.  */
+ #define __extension__ /* Undefine GCC keyword to make Swig happy when processing glibc's stdint.h. */
+-/* The ISO C99 standard specifies that in C++ implementations limit macros such
+-   as INT32_MAX should only be defined if __STDC_LIMIT_MACROS is. */
+-#define __STDC_LIMIT_MACROS
+ %include "stdint.i"
+ 
+ %include "lldb/lldb-defines.h"
+diff --git a/lldb/bindings/python/python-typemaps.swig b/lldb/bindings/python/python-typemaps.swig
+index b1ace4ff3b1e..5f8f4aa678c4 100644
+--- a/lldb/bindings/python/python-typemaps.swig
+++ b/lldb/bindings/python/python-typemaps.swig
+@@ -439,7 +439,7 @@ bool SetNumberFromPyObject<double>(double &number, PyObject *obj) {
+ 
+ %typemap(out) lldb::FileSP {
+   $result = nullptr;
+-  lldb::FileSP &sp = $1;
+  const lldb::FileSP &sp = $1;
+   if (sp) {
+     PythonFile pyfile = unwrapOrSetPythonException(PythonFile::FromFile(*sp));
+     if (!pyfile.IsValid())
 diff --git a/lldb/include/lldb/Breakpoint/Breakpoint.h b/lldb/include/lldb/Breakpoint/Breakpoint.h
-index f2e2a0d22..426d1129b 100644
+index f2e2a0d22784..426d1129bd10 100644
 --- a/lldb/include/lldb/Breakpoint/Breakpoint.h
 +++ b/lldb/include/lldb/Breakpoint/Breakpoint.h
@@ -9,6 +9,7 @@
@ -11,7 +50,7 @@ index f2e2a0d22..426d1129b 100644
 #include <string>
 #include <unordered_set>
 diff --git a/lldb/include/lldb/Core/Module.h b/lldb/include/lldb/Core/Module.h
-index dd7100c46..97d70daad 100644
+index dd7100c4616c..97d70daadbdc 100644
 --- a/lldb/include/lldb/Core/Module.h
 +++ b/lldb/include/lldb/Core/Module.h
@@ -41,6 +41,7 @@
@ -41,7 +80,7 @@ index dd7100c46..97d70daad 100644
   ///
   /// Tries to resolve \a vm_addr as a file address (if \a
 diff --git a/lldb/include/lldb/Core/PluginManager.h b/lldb/include/lldb/Core/PluginManager.h
-index be91929c6..8d876fc1f 100644
+index be91929c62e1..8d876fc1fa2f 100644
 --- a/lldb/include/lldb/Core/PluginManager.h
 +++ b/lldb/include/lldb/Core/PluginManager.h
@@ -508,6 +508,17 @@ public:
@ -64,7 +103,7 @@ index be91929c6..8d876fc1f 100644
 } // namespace lldb_private
 diff --git a/lldb/include/lldb/Expression/DWARFEvaluator.h b/lldb/include/lldb/Expression/DWARFEvaluator.h
 new file mode 100644
-index 000000000..6811cbeae
+index 000000000000..6811cbeae3d3
 --- /dev/null
 +++ b/lldb/include/lldb/Expression/DWARFEvaluator.h
@@ -0,0 +1,110 @@
@ -180,7 +219,7 @@ index 000000000..6811cbeae
 +#endif // LLDB_EXPRESSION_DWARFEVALUATOR_H
 diff --git a/lldb/include/lldb/Expression/DWARFEvaluatorFactory.h b/lldb/include/lldb/Expression/DWARFEvaluatorFactory.h
 new file mode 100644
-index 000000000..f3b496c58
+index 000000000000..f3b496c580e4
 --- /dev/null
 +++ b/lldb/include/lldb/Expression/DWARFEvaluatorFactory.h
@@ -0,0 +1,56 @@
@ -241,7 +280,7 @@ index 000000000..f3b496c58
 +
 +#endif // LLDB_EXPRESSION_DWARFEVALUATORFACTORY_H
 diff --git a/lldb/include/lldb/Expression/DWARFExpression.h b/lldb/include/lldb/Expression/DWARFExpression.h
-index 1490ac2d6..35c741d4e 100644
+index 1490ac2d614a..35c741d4e6ba 100644
 --- a/lldb/include/lldb/Expression/DWARFExpression.h
 +++ b/lldb/include/lldb/Expression/DWARFExpression.h
@@ -120,6 +120,10 @@ public:
@ -275,7 +314,7 @@ index 1490ac2d6..35c741d4e 100644
   GetLocationExpression(lldb::addr_t load_function_start,
                         lldb::addr_t addr) const;
 diff --git a/lldb/include/lldb/Target/Process.h b/lldb/include/lldb/Target/Process.h
-index aaa2470d2..c15f2db52 100644
+index aaa2470d2931..c15f2db52fbc 100644
 --- a/lldb/include/lldb/Target/Process.h
 +++ b/lldb/include/lldb/Target/Process.h
@@ -1434,7 +1434,7 @@ public:
@ -288,7 +327,7 @@ index aaa2470d2..c15f2db52 100644
   /// Read of memory from a process.
   ///
 diff --git a/lldb/include/lldb/Target/ProcessTrace.h b/lldb/include/lldb/Target/ProcessTrace.h
-index 7b9d6b13d..9525fc975 100644
+index 7b9d6b13dd6f..9525fc9750fd 100644
 --- a/lldb/include/lldb/Target/ProcessTrace.h
 +++ b/lldb/include/lldb/Target/ProcessTrace.h
@@ -59,7 +59,7 @@ public:
@ -301,7 +340,7 @@ index 7b9d6b13d..9525fc975 100644
   size_t DoReadMemory(lldb::addr_t addr, void *buf, size_t size,
                       Status &error) override;
 diff --git a/lldb/include/lldb/lldb-forward.h b/lldb/include/lldb/lldb-forward.h
-index ad5298151..5a3c0b27a 100644
+index ad5298151e4a..5a3c0b27a738 100644
 --- a/lldb/include/lldb/lldb-forward.h
 +++ b/lldb/include/lldb/lldb-forward.h
@@ -74,6 +74,7 @@ class Disassembler;
@ -313,7 +352,7 @@ index ad5298151..5a3c0b27a 100644
 class EmulateInstruction;
 class Environment;
 diff --git a/lldb/include/lldb/lldb-private-interfaces.h b/lldb/include/lldb/lldb-private-interfaces.h
-index 2ed083ec8..f4d500d19 100644
+index 2ed083ec8ae9..f4d500d198e8 100644
 --- a/lldb/include/lldb/lldb-private-interfaces.h
 +++ b/lldb/include/lldb/lldb-private-interfaces.h
@@ -113,6 +113,8 @@ typedef lldb::REPLSP (*REPLCreateInstance)(Status &error,
@ -326,7 +365,7 @@ index 2ed083ec8..f4d500d19 100644
 /// \{
 typedef llvm::Expected<lldb::TraceSP> (*TraceCreateInstanceForSessionFile)(
 diff --git a/lldb/source/Core/Module.cpp b/lldb/source/Core/Module.cpp
-index 19c97be15..1647f93ec 100644
+index 19c97be15066..1647f93ec4f3 100644
 --- a/lldb/source/Core/Module.cpp
 +++ b/lldb/source/Core/Module.cpp
@@ -16,6 +16,7 @@
@ -348,7 +387,7 @@ index 19c97be15..1647f93ec 100644
 +  return m_dwarf_evaluator_factory.get();
 +}
 diff --git a/lldb/source/Core/PluginManager.cpp b/lldb/source/Core/PluginManager.cpp
-index fcaa868b0..59a404d4a 100644
+index fcaa868b083e..59a404d4a7e1 100644
 --- a/lldb/source/Core/PluginManager.cpp
 +++ b/lldb/source/Core/PluginManager.cpp
@@ -1597,3 +1597,32 @@ bool PluginManager::CreateSettingForStructuredDataPlugin(
@ -385,7 +424,7 @@ index fcaa868b0..59a404d4a 100644
 +  return GetDWARFEvaluatorFactoryInstances().GetCallbackAtIndex(idx);
 +}
 diff --git a/lldb/source/Core/Value.cpp b/lldb/source/Core/Value.cpp
-index fb57c0fed..f92d6a54d 100644
+index fb57c0fedf04..f92d6a54de94 100644
 --- a/lldb/source/Core/Value.cpp
 +++ b/lldb/source/Core/Value.cpp
@@ -538,7 +538,7 @@ Status Value::GetValueAsData(ExecutionContext *exe_ctx, DataExtractor &data,
@ -398,7 +437,7 @@ index fb57c0fed..f92d6a54d 100644
             error.SetErrorStringWithFormat(
                 "read memory from 0x%" PRIx64 " failed (%u of %u bytes read)",
 diff --git a/lldb/source/Core/ValueObject.cpp b/lldb/source/Core/ValueObject.cpp
-index 9c1ba99da..b15b214b2 100644
+index 9c1ba99da1d0..b15b214b2a2f 100644
 --- a/lldb/source/Core/ValueObject.cpp
 +++ b/lldb/source/Core/ValueObject.cpp
@@ -735,7 +735,7 @@ size_t ValueObject::GetPointeeData(DataExtractor &data, uint32_t item_idx,
@ -411,7 +450,7 @@ index 9c1ba99da..b15b214b2 100644
           data.SetData(data_sp);
           return bytes_read;
 diff --git a/lldb/source/Expression/CMakeLists.txt b/lldb/source/Expression/CMakeLists.txt
-index bf94361dd..4e76d547a 100644
+index bf94361dd6c1..4e76d547aeaf 100644
 --- a/lldb/source/Expression/CMakeLists.txt
 +++ b/lldb/source/Expression/CMakeLists.txt
@@ -1,5 +1,7 @@
@ -424,7 +463,7 @@ index bf94361dd..4e76d547a 100644
   ExpressionVariable.cpp
 diff --git a/lldb/source/Expression/DWARFEvaluator.cpp b/lldb/source/Expression/DWARFEvaluator.cpp
 new file mode 100644
-index 000000000..06107e136
+index 000000000000..06107e136197
 --- /dev/null
 +++ b/lldb/source/Expression/DWARFEvaluator.cpp
@@ -0,0 +1,1952 @@
@ -2382,7 +2421,7 @@ index 000000000..06107e136
 +}
 diff --git a/lldb/source/Expression/DWARFEvaluatorFactory.cpp b/lldb/source/Expression/DWARFEvaluatorFactory.cpp
 new file mode 100644
-index 000000000..c06126412
+index 000000000000..c0612641204a
 --- /dev/null
 +++ b/lldb/source/Expression/DWARFEvaluatorFactory.cpp
@@ -0,0 +1,57 @@
@ -2444,7 +2483,7 @@ index 000000000..c06126412
 +                                          object_address_ptr);
 +}
 diff --git a/lldb/source/Expression/DWARFExpression.cpp b/lldb/source/Expression/DWARFExpression.cpp
-index a10546c1d..4d13e4642 100644
+index a10546c1deae..4d13e4642af3 100644
 --- a/lldb/source/Expression/DWARFExpression.cpp
 +++ b/lldb/source/Expression/DWARFExpression.cpp
@@ -15,6 +15,8 @@
@ -4261,7 +4300,7 @@ index a10546c1d..4d13e4642 100644
 
 static DataExtractor ToDataExtractor(const llvm::DWARFLocationExpression &loc,
 diff --git a/lldb/source/Interpreter/CommandInterpreter.cpp b/lldb/source/Interpreter/CommandInterpreter.cpp
-index 00e9ccb76..2137a1ac8 100644
+index 00e9ccb762c3..2137a1ac8324 100644
 --- a/lldb/source/Interpreter/CommandInterpreter.cpp
 +++ b/lldb/source/Interpreter/CommandInterpreter.cpp
@@ -759,6 +759,24 @@ void CommandInterpreter::LoadCommandDictionary() {
@ -4290,7 +4329,7 @@ index 00e9ccb76..2137a1ac8 100644
       new CommandObjectRegexCommand(
           *this, "kdp-remote",
 diff --git a/lldb/source/Plugins/CMakeLists.txt b/lldb/source/Plugins/CMakeLists.txt
-index 9181a4e47..2be6ec365 100644
+index 9181a4e47675..2be6ec3657c0 100644
 --- a/lldb/source/Plugins/CMakeLists.txt
 +++ b/lldb/source/Plugins/CMakeLists.txt
@@ -2,6 +2,7 @@ add_subdirectory(ABI)
@ -4320,14 +4359,14 @@ index 9181a4e47..2be6ec365 100644
   endif()
 diff --git a/lldb/source/Plugins/DWARFEvaluator/CMakeLists.txt b/lldb/source/Plugins/DWARFEvaluator/CMakeLists.txt
 new file mode 100644
-index 000000000..73fad41e1
+index 000000000000..73fad41e1a72
 --- /dev/null
 +++ b/lldb/source/Plugins/DWARFEvaluator/CMakeLists.txt
@@ -0,0 +1 @@
 +add_subdirectory(wasm)
 diff --git a/lldb/source/Plugins/DWARFEvaluator/wasm/CMakeLists.txt b/lldb/source/Plugins/DWARFEvaluator/wasm/CMakeLists.txt
 new file mode 100644
-index 000000000..e50b1bef7
+index 000000000000..e50b1bef7e69
 --- /dev/null
 +++ b/lldb/source/Plugins/DWARFEvaluator/wasm/CMakeLists.txt
@@ -0,0 +1,10 @@
@ -4343,7 +4382,7 @@ index 000000000..e50b1bef7
 +  )
 diff --git a/lldb/source/Plugins/DWARFEvaluator/wasm/WasmDWARFEvaluator.cpp b/lldb/source/Plugins/DWARFEvaluator/wasm/WasmDWARFEvaluator.cpp
 new file mode 100644
-index 000000000..fdda1991d
+index 000000000000..fdda1991d19f
 --- /dev/null
 +++ b/lldb/source/Plugins/DWARFEvaluator/wasm/WasmDWARFEvaluator.cpp
@@ -0,0 +1,126 @@
@ -4475,7 +4514,7 @@ index 000000000..fdda1991d
 +}
 diff --git a/lldb/source/Plugins/DWARFEvaluator/wasm/WasmDWARFEvaluator.h b/lldb/source/Plugins/DWARFEvaluator/wasm/WasmDWARFEvaluator.h
 new file mode 100644
-index 000000000..a01159064
+index 000000000000..a01159064a39
 --- /dev/null
 +++ b/lldb/source/Plugins/DWARFEvaluator/wasm/WasmDWARFEvaluator.h
@@ -0,0 +1,47 @@
@ -4528,7 +4567,7 @@ index 000000000..a01159064
 +#endif // LLDB_SOURCE_PLUGINS_DWARFEVALUATOR_WASM_WASMDWARFEVALUATOR_H
 diff --git a/lldb/source/Plugins/DWARFEvaluator/wasm/WasmDWARFEvaluatorFactory.cpp b/lldb/source/Plugins/DWARFEvaluator/wasm/WasmDWARFEvaluatorFactory.cpp
 new file mode 100644
-index 000000000..d43e96a34
+index 000000000000..d43e96a34d37
 --- /dev/null
 +++ b/lldb/source/Plugins/DWARFEvaluator/wasm/WasmDWARFEvaluatorFactory.cpp
@@ -0,0 +1,64 @@
@ -4598,7 +4637,7 @@ index 000000000..d43e96a34
 +}
 diff --git a/lldb/source/Plugins/DWARFEvaluator/wasm/WasmDWARFEvaluatorFactory.h b/lldb/source/Plugins/DWARFEvaluator/wasm/WasmDWARFEvaluatorFactory.h
 new file mode 100644
-index 000000000..8a946592a
+index 000000000000..8a946592a09a
 --- /dev/null
 +++ b/lldb/source/Plugins/DWARFEvaluator/wasm/WasmDWARFEvaluatorFactory.h
@@ -0,0 +1,55 @@
@ -4658,7 +4697,7 @@ index 000000000..8a946592a
 +
 +#endif // LLDB_SOURCE_PLUGINS_DWARFEVALUATOR_WASM_WASMDWARFEVALUATORFACTORY_H
 diff --git a/lldb/source/Plugins/DynamicLoader/wasm-DYLD/DynamicLoaderWasmDYLD.cpp b/lldb/source/Plugins/DynamicLoader/wasm-DYLD/DynamicLoaderWasmDYLD.cpp
-index ae7e011ea..24ea75d19 100644
+index ae7e011eaa52..24ea75d1971c 100644
 --- a/lldb/source/Plugins/DynamicLoader/wasm-DYLD/DynamicLoaderWasmDYLD.cpp
 +++ b/lldb/source/Plugins/DynamicLoader/wasm-DYLD/DynamicLoaderWasmDYLD.cpp
@@ -62,6 +62,15 @@ void DynamicLoaderWasmDYLD::DidAttach() {
@ -4678,7 +4717,7 @@ index ae7e011ea..24ea75d19 100644
 
 ThreadPlanSP DynamicLoaderWasmDYLD::GetStepThroughTrampolinePlan(Thread &thread,
 diff --git a/lldb/source/Plugins/ObjectFile/wasm/ObjectFileWasm.cpp b/lldb/source/Plugins/ObjectFile/wasm/ObjectFileWasm.cpp
-index 5272da9ab..abc5523bf 100644
+index 5272da9ab33a..abc5523bfd70 100644
 --- a/lldb/source/Plugins/ObjectFile/wasm/ObjectFileWasm.cpp
 +++ b/lldb/source/Plugins/ObjectFile/wasm/ObjectFileWasm.cpp
@@ -23,6 +23,7 @@
@ -4718,7 +4757,7 @@ index 5272da9ab..abc5523bf 100644
     }
   }
 diff --git a/lldb/source/Plugins/Platform/CMakeLists.txt b/lldb/source/Plugins/Platform/CMakeLists.txt
-index 5f284e517..6084cbc93 100644
+index 5f284e517dca..6084cbc9378d 100644
 --- a/lldb/source/Plugins/Platform/CMakeLists.txt
 +++ b/lldb/source/Plugins/Platform/CMakeLists.txt
@@ -15,3 +15,4 @@
@ -4728,7 +4767,7 @@ index 5f284e517..6084cbc93 100644
 +add_subdirectory(wasm-remote)
 diff --git a/lldb/source/Plugins/Platform/wasm-remote/CMakeLists.txt b/lldb/source/Plugins/Platform/wasm-remote/CMakeLists.txt
 new file mode 100644
-index 000000000..4a65765a5
+index 000000000000..4a65765a5659
 --- /dev/null
 +++ b/lldb/source/Plugins/Platform/wasm-remote/CMakeLists.txt
@@ -0,0 +1,10 @@
@ -4744,7 +4783,7 @@ index 000000000..4a65765a5
 +  )
 diff --git a/lldb/source/Plugins/Platform/wasm-remote/PlatformRemoteWasmServer.cpp b/lldb/source/Plugins/Platform/wasm-remote/PlatformRemoteWasmServer.cpp
 new file mode 100644
-index 000000000..f26d11f00
+index 000000000000..f26d11f00e5c
 --- /dev/null
 +++ b/lldb/source/Plugins/Platform/wasm-remote/PlatformRemoteWasmServer.cpp
@@ -0,0 +1,139 @@
@ -4890,7 +4929,7 @@ index 000000000..f26d11f00
 \ No newline at end of file
 diff --git a/lldb/source/Plugins/Platform/wasm-remote/PlatformRemoteWasmServer.h b/lldb/source/Plugins/Platform/wasm-remote/PlatformRemoteWasmServer.h
 new file mode 100644
-index 000000000..f306a79d3
+index 000000000000..f306a79d3f4f
 --- /dev/null
 +++ b/lldb/source/Plugins/Platform/wasm-remote/PlatformRemoteWasmServer.h
@@ -0,0 +1,37 @@
@ -4933,7 +4972,7 @@ index 000000000..f306a79d3
 +#endif
 \ No newline at end of file
 diff --git a/lldb/source/Plugins/Plugins.def.in b/lldb/source/Plugins/Plugins.def.in
-index bf54598fb..b0bd7b996 100644
+index bf54598fb2f3..b0bd7b9965fe 100644
 --- a/lldb/source/Plugins/Plugins.def.in
 +++ b/lldb/source/Plugins/Plugins.def.in
@@ -31,6 +31,7 @@
@ -4945,7 +4984,7 @@ index bf54598fb..b0bd7b996 100644
 
 #undef LLDB_PLUGIN
 diff --git a/lldb/source/Plugins/Process/CMakeLists.txt b/lldb/source/Plugins/Process/CMakeLists.txt
-index bea5bac9e..7a0855e02 100644
+index bea5bac9eb21..7a0855e02ca2 100644
 --- a/lldb/source/Plugins/Process/CMakeLists.txt
 +++ b/lldb/source/Plugins/Process/CMakeLists.txt
@@ -18,3 +18,4 @@ add_subdirectory(Utility)
@ -4954,7 +4993,7 @@ index bea5bac9e..7a0855e02 100644
 add_subdirectory(minidump)
 +add_subdirectory(wasm)
 diff --git a/lldb/source/Plugins/Process/elf-core/ProcessElfCore.cpp b/lldb/source/Plugins/Process/elf-core/ProcessElfCore.cpp
-index 12bc7390c..707ab85e5 100644
+index 12bc7390c729..707ab85e5615 100644
 --- a/lldb/source/Plugins/Process/elf-core/ProcessElfCore.cpp
 +++ b/lldb/source/Plugins/Process/elf-core/ProcessElfCore.cpp
@@ -285,7 +285,7 @@ bool ProcessElfCore::IsAlive() { return true; }
@ -4967,7 +5006,7 @@ index 12bc7390c..707ab85e5 100644
   // in core files we have it all cached our our core file anyway.
   return DoReadMemory(addr, buf, size, error);
 diff --git a/lldb/source/Plugins/Process/elf-core/ProcessElfCore.h b/lldb/source/Plugins/Process/elf-core/ProcessElfCore.h
-index d8e3cc9ae..f0bf9c4d3 100644
+index d8e3cc9ae3e1..f0bf9c4d3b00 100644
 --- a/lldb/source/Plugins/Process/elf-core/ProcessElfCore.h
 +++ b/lldb/source/Plugins/Process/elf-core/ProcessElfCore.h
@@ -84,7 +84,8 @@ public:
@ -4981,7 +5020,7 @@ index d8e3cc9ae..f0bf9c4d3 100644
   size_t DoReadMemory(lldb::addr_t addr, void *buf, size_t size,
                       lldb_private::Status &error) override;
 diff --git a/lldb/source/Plugins/Process/gdb-remote/ProcessGDBRemote.cpp b/lldb/source/Plugins/Process/gdb-remote/ProcessGDBRemote.cpp
-index 6914b3734..bb8a05604 100644
+index 6914b37348ea..bb8a056049f3 100644
 --- a/lldb/source/Plugins/Process/gdb-remote/ProcessGDBRemote.cpp
 +++ b/lldb/source/Plugins/Process/gdb-remote/ProcessGDBRemote.cpp
@@ -334,6 +334,11 @@ ConstString ProcessGDBRemote::GetPluginName() { return GetPluginNameStatic(); }
@ -5015,7 +5054,7 @@ index 6914b3734..bb8a05604 100644
       }
     }
 diff --git a/lldb/source/Plugins/Process/gdb-remote/ProcessGDBRemote.h b/lldb/source/Plugins/Process/gdb-remote/ProcessGDBRemote.h
-index fe04cdddd..e4a14c645 100644
+index fe04cdddd0f5..e4a14c64579a 100644
 --- a/lldb/source/Plugins/Process/gdb-remote/ProcessGDBRemote.h
 +++ b/lldb/source/Plugins/Process/gdb-remote/ProcessGDBRemote.h
@@ -237,6 +237,8 @@ protected:
@ -5028,7 +5067,7 @@ index fe04cdddd..e4a14c645 100644
   enum {
     eBroadcastBitAsyncContinue = (1 << 0),
 diff --git a/lldb/source/Plugins/Process/mach-core/ProcessMachCore.cpp b/lldb/source/Plugins/Process/mach-core/ProcessMachCore.cpp
-index 84548edb5..0ae6f7e4a 100644
+index 84548edb5caa..0ae6f7e4a177 100644
 --- a/lldb/source/Plugins/Process/mach-core/ProcessMachCore.cpp
 +++ b/lldb/source/Plugins/Process/mach-core/ProcessMachCore.cpp
@@ -596,7 +596,7 @@ bool ProcessMachCore::WarnBeforeDetach() const { return false; }
@ -5041,7 +5080,7 @@ index 84548edb5..0ae6f7e4a 100644
   // in core files we have it all cached our our core file anyway.
   return DoReadMemory(addr, buf, size, error);
 diff --git a/lldb/source/Plugins/Process/mach-core/ProcessMachCore.h b/lldb/source/Plugins/Process/mach-core/ProcessMachCore.h
-index db77e96f1..1c930896c 100644
+index db77e96f1072..1c930896c743 100644
 --- a/lldb/source/Plugins/Process/mach-core/ProcessMachCore.h
 +++ b/lldb/source/Plugins/Process/mach-core/ProcessMachCore.h
@@ -65,7 +65,8 @@ public:
@ -5055,7 +5094,7 @@ index db77e96f1..1c930896c 100644
   size_t DoReadMemory(lldb::addr_t addr, void *buf, size_t size,
                       lldb_private::Status &error) override;
 diff --git a/lldb/source/Plugins/Process/minidump/ProcessMinidump.cpp b/lldb/source/Plugins/Process/minidump/ProcessMinidump.cpp
-index 385557422..d8bb21581 100644
+index 385557422758..d8bb21581086 100644
 --- a/lldb/source/Plugins/Process/minidump/ProcessMinidump.cpp
 +++ b/lldb/source/Plugins/Process/minidump/ProcessMinidump.cpp
@@ -374,7 +374,7 @@ bool ProcessMinidump::IsAlive() { return true; }
@ -5068,7 +5107,7 @@ index 385557422..d8bb21581 100644
   // we have it all cached in our dump file anyway.
   return DoReadMemory(addr, buf, size, error);
 diff --git a/lldb/source/Plugins/Process/minidump/ProcessMinidump.h b/lldb/source/Plugins/Process/minidump/ProcessMinidump.h
-index 27b0da004..e94ecab43 100644
+index 27b0da0047a5..e94ecab430c1 100644
 --- a/lldb/source/Plugins/Process/minidump/ProcessMinidump.h
 +++ b/lldb/source/Plugins/Process/minidump/ProcessMinidump.h
@@ -69,8 +69,8 @@ public:
@ -5084,7 +5123,7 @@ index 27b0da004..e94ecab43 100644
                       Status &error) override;
 diff --git a/lldb/source/Plugins/Process/wasm/CMakeLists.txt b/lldb/source/Plugins/Process/wasm/CMakeLists.txt
 new file mode 100644
-index 000000000..61efb933f
+index 000000000000..61efb933fa62
 --- /dev/null
 +++ b/lldb/source/Plugins/Process/wasm/CMakeLists.txt
@@ -0,0 +1,12 @@
@ -5102,7 +5141,7 @@ index 000000000..61efb933f
 +  )
 diff --git a/lldb/source/Plugins/Process/wasm/ProcessWasm.cpp b/lldb/source/Plugins/Process/wasm/ProcessWasm.cpp
 new file mode 100644
-index 000000000..9c0fc7b7f
+index 000000000000..9c0fc7b7f270
 --- /dev/null
 +++ b/lldb/source/Plugins/Process/wasm/ProcessWasm.cpp
@@ -0,0 +1,261 @@
@ -5369,7 +5408,7 @@ index 000000000..9c0fc7b7f
 +}
 diff --git a/lldb/source/Plugins/Process/wasm/ProcessWasm.h b/lldb/source/Plugins/Process/wasm/ProcessWasm.h
 new file mode 100644
-index 000000000..d3aece7a6
+index 000000000000..d3aece7a6554
 --- /dev/null
 +++ b/lldb/source/Plugins/Process/wasm/ProcessWasm.h
@@ -0,0 +1,128 @@
@ -5503,7 +5542,7 @@ index 000000000..d3aece7a6
 +#endif // LLDB_SOURCE_PLUGINS_PROCESS_WASM_PROCESSWASM_H
 diff --git a/lldb/source/Plugins/Process/wasm/ThreadWasm.cpp b/lldb/source/Plugins/Process/wasm/ThreadWasm.cpp
 new file mode 100644
-index 000000000..fa02073e7
+index 000000000000..fa02073e7a52
 --- /dev/null
 +++ b/lldb/source/Plugins/Process/wasm/ThreadWasm.cpp
@@ -0,0 +1,35 @@
@ -5544,7 +5583,7 @@ index 000000000..fa02073e7
 +}
 diff --git a/lldb/source/Plugins/Process/wasm/ThreadWasm.h b/lldb/source/Plugins/Process/wasm/ThreadWasm.h
 new file mode 100644
-index 000000000..0a33c07de
+index 000000000000..0a33c07de994
 --- /dev/null
 +++ b/lldb/source/Plugins/Process/wasm/ThreadWasm.h
@@ -0,0 +1,41 @@
@ -5591,7 +5630,7 @@ index 000000000..0a33c07de
 +#endif // LLDB_SOURCE_PLUGINS_PROCESS_WASM_THREADWASM_H
 diff --git a/lldb/source/Plugins/Process/wasm/UnwindWasm.cpp b/lldb/source/Plugins/Process/wasm/UnwindWasm.cpp
 new file mode 100644
-index 000000000..1a195cb93
+index 000000000000..1a195cb9361a
 --- /dev/null
 +++ b/lldb/source/Plugins/Process/wasm/UnwindWasm.cpp
@@ -0,0 +1,74 @@
@ -5672,7 +5711,7 @@ index 000000000..1a195cb93
 \ No newline at end of file
 diff --git a/lldb/source/Plugins/Process/wasm/UnwindWasm.h b/lldb/source/Plugins/Process/wasm/UnwindWasm.h
 new file mode 100644
-index 000000000..9bd1dac9a
+index 000000000000..9bd1dac9a98a
 --- /dev/null
 +++ b/lldb/source/Plugins/Process/wasm/UnwindWasm.h
@@ -0,0 +1,55 @@
@ -5732,7 +5771,7 @@ index 000000000..9bd1dac9a
 +
 +#endif // lldb_UnwindWasm_h_
 diff --git a/lldb/source/Plugins/SymbolFile/DWARF/SymbolFileDWARF.cpp b/lldb/source/Plugins/SymbolFile/DWARF/SymbolFileDWARF.cpp
-index ccaf31317..c3ef5aebd 100644
+index ccaf31317d75..c3ef5aebd46d 100644
 --- a/lldb/source/Plugins/SymbolFile/DWARF/SymbolFileDWARF.cpp
 +++ b/lldb/source/Plugins/SymbolFile/DWARF/SymbolFileDWARF.cpp
@@ -3212,8 +3212,13 @@ VariableSP SymbolFileDWARF::ParseVariableDIE(const SymbolContext &sc,
@ -5751,7 +5790,7 @@ index ccaf31317..c3ef5aebd 100644
     // DWARF doesn't specify if a DW_TAG_variable is a local, global
     // or static variable, so we have to do a little digging:
 diff --git a/lldb/source/Target/PathMappingList.cpp b/lldb/source/Target/PathMappingList.cpp
-index b660c310e..cd76421ce 100644
+index b660c310ef31..cd76421cec18 100644
 --- a/lldb/source/Target/PathMappingList.cpp
 +++ b/lldb/source/Target/PathMappingList.cpp
@@ -218,7 +218,12 @@ bool PathMappingList::ReverseRemapPath(const FileSpec &file, FileSpec &fixed) co
@ -5769,7 +5808,7 @@ index b660c310e..cd76421ce 100644
 
   return {};
 diff --git a/lldb/source/Target/Platform.cpp b/lldb/source/Target/Platform.cpp
-index a77ecddfb..e257f9350 100644
+index a77ecddfbab6..e257f93508f6 100644
 --- a/lldb/source/Target/Platform.cpp
 +++ b/lldb/source/Target/Platform.cpp
@@ -1970,6 +1970,12 @@ size_t Platform::GetSoftwareBreakpointTrapOpcode(Target &target,
@ -5786,7 +5825,7 @@ index a77ecddfb..e257f9350 100644
     return 0;
   }
 diff --git a/lldb/source/Target/Process.cpp b/lldb/source/Target/Process.cpp
-index 8ecc66b59..f14898791 100644
+index 8ecc66b592ea..f148987915de 100644
 --- a/lldb/source/Target/Process.cpp
 +++ b/lldb/source/Target/Process.cpp
@@ -1892,7 +1892,8 @@ Status Process::DisableSoftwareBreakpoint(BreakpointSite *bp_site) {
@ -5800,7 +5839,7 @@ index 8ecc66b59..f14898791 100644
   if (!GetDisableMemoryCache()) {
 #if defined(VERIFY_MEMORY_READS)
 diff --git a/lldb/source/Target/ProcessTrace.cpp b/lldb/source/Target/ProcessTrace.cpp
-index c878a2ac4..ad5945b0a 100644
+index c878a2ac4eb9..ad5945b0ad1f 100644
 --- a/lldb/source/Target/ProcessTrace.cpp
 +++ b/lldb/source/Target/ProcessTrace.cpp
@@ -88,7 +88,7 @@ void ProcessTrace::RefreshStateAfterStop() {}
@ -5813,7 +5852,7 @@ index c878a2ac4..ad5945b0a 100644
   // we have it all cached in the trace files.
   return DoReadMemory(addr, buf, size, error);
 diff --git a/lldb/source/Target/ThreadPlanStepRange.cpp b/lldb/source/Target/ThreadPlanStepRange.cpp
-index 896e647bb..f76307016 100644
+index 896e647bbb52..f76307016102 100644
 --- a/lldb/source/Target/ThreadPlanStepRange.cpp
 +++ b/lldb/source/Target/ThreadPlanStepRange.cpp
@@ -334,7 +334,10 @@ bool ThreadPlanStepRange::SetNextBranchBreakpoint() {
@ -5829,7 +5868,7 @@ index 896e647bb..f76307016 100644
             instructions->GetInstructionAtIndex(last_index);
         size_t last_inst_size = last_inst->GetOpcode().GetByteSize();
 diff --git a/lldb/source/Target/UnixSignals.cpp b/lldb/source/Target/UnixSignals.cpp
-index 4ec2e25c7..24c88fe9a 100644
+index 4ec2e25c7e3b..24c88fe9ae4f 100644
 --- a/lldb/source/Target/UnixSignals.cpp
 +++ b/lldb/source/Target/UnixSignals.cpp
@@ -46,6 +46,8 @@ lldb::UnixSignalsSP UnixSignals::Create(const ArchSpec &arch) {
@ -5842,7 +5881,7 @@ index 4ec2e25c7..24c88fe9a 100644
     return std::make_shared<UnixSignals>();
   }
 diff --git a/llvm/include/llvm/ExecutionEngine/Orc/OrcRPCExecutorProcessControl.h b/llvm/include/llvm/ExecutionEngine/Orc/OrcRPCExecutorProcessControl.h
-index 4310ba9ce..297b33879 100644
+index 4310ba9ce9e0..297b3387999d 100644
 --- a/llvm/include/llvm/ExecutionEngine/Orc/OrcRPCExecutorProcessControl.h
 +++ b/llvm/include/llvm/ExecutionEngine/Orc/OrcRPCExecutorProcessControl.h
@@ -13,6 +13,7 @@
@ -5854,7 +5893,7 @@ index 4310ba9ce..297b33879 100644
 #include "llvm/ExecutionEngine/Orc/Shared/RPCUtils.h"
 #include "llvm/ExecutionEngine/Orc/Shared/RawByteChannel.h"
 diff --git a/llvm/include/llvm/Support/MathExtras.h b/llvm/include/llvm/Support/MathExtras.h
-index 753b1998c..27370c62d 100644
+index 753b1998c40c..27370c62dd6e 100644
 --- a/llvm/include/llvm/Support/MathExtras.h
 +++ b/llvm/include/llvm/Support/MathExtras.h
@@ -16,6 +16,7 @@
--- a/ci/pre_commit_hook_sample
+++ b/ci/pre_commit_hook_sample
@ -0,0 +1,32 @@
+#!/bin/bash
+
+# Copyright (C) 2023 Amazon.com Inc. or its affiliates. All rights reserved.
+# SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+
+# This is a sample of pre-commit hook that can be used to make your code fit the WAMR CI code style requirements.
+# You need to have clang-format-12 installed to use this hook.
+# To add this pre-commit hook, copy it to <path_to_wamr>/.git/hooks/pre-commit
+# (you don't need any extensions here)
+
+# Function to check if a file has a C or C++ extension
+
+is_c_or_cpp_file() {
+    file="$1"
+    if [[ "$filename" =~ \.(h|c|cpp)$  ]]; then
+        return 0
+    else
+        return 1
+    fi
+}
+
+# Loop through staged files and apply command "abc" to C and C++ files
+for staged_file in $(git diff --cached --name-only); do
+    if is_c_or_cpp_file "$staged_file"; then
+        clang-format-12 -Werror --style file --dry-run "$staged_file" 2>/dev/null
+        if [ $? -ne 0 ]; then 
+            echo "Issues are found in $staged_file. Applying the fix" 
+            clang-format-12 --style file -i "$staged_file"
+        fi
+        git add "$staged_file"  # Add the modified file back to staging
+    fi
+done
--- a/ci/setup.sh
+++ b/ci/setup.sh
@ -0,0 +1,15 @@
+#!/bin/bash
+
+# Copyright (C) 2023 Amazon.com Inc. or its affiliates. All rights reserved.
+# SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+
+# This script executes some commands to make your onboarding with WAMR easier.
+# For example, setting pre-commit hook that will make your code complaint with the
+# code style requirements checked in WAMR CI
+
+echo "Copy the pre-commit hook to your hooks folder"
+cp pre_commit_hook_sample ../.git/hooks/pre-commit
+
+# Feel free to propose your commands to this script to make developing WAMR easier
+
+echo "Setup is done"
--- a/ci/validate_lldb.py
+++ b/ci/validate_lldb.py
@ -32,7 +32,6 @@ parser.add_argument(

 options = parser.parse_args()

-lldb_command_prologue = f'{options.lldb} -o "process connect -p wasm connect://127.0.0.1:{options.port}"'
 lldb_command_epilogue = '-o q'

 test_cases = {
@ -64,8 +63,13 @@ def print_process_output(p):
        print("Failed to get process output")

 # Step2: Launch WAMR in debug mode and validate lldb commands
-wamr_cmd = f'{options.wamr} -g=127.0.0.1:{options.port} {WASM_OUT_FILE}'
+
+iteration = 0
 for case, cmd in test_cases.items():
+    lldb_command_prologue = f'{options.lldb} -o "process connect -p wasm connect://127.0.0.1:{int(options.port) + iteration}"'
+    wamr_cmd = f'{options.wamr} -g=127.0.0.1:{int(options.port) + iteration} {WASM_OUT_FILE}'
+    iteration += 1
+
    has_error = False
    print(f'validating case [{case}] ...', end='', flush=True)
    lldb_cmd = f'{lldb_command_prologue} {cmd} {lldb_command_epilogue}'
--- a/core/config.h
+++ b/core/config.h
@ -144,6 +144,14 @@
 #define WASM_ENABLE_WASI_NN 0
 #endif

+#ifndef WASM_ENABLE_WASI_NN_GPU
+#define WASM_ENABLE_WASI_NN_GPU 0
+#endif
+
+#ifndef WASM_ENABLE_WASI_NN_EXTERNAL_DELEGATE
+#define WASM_ENABLE_WASI_NN_EXTERNAL_DELEGATE 0
+#endif
+
 /* Default disable libc emcc */
 #ifndef WASM_ENABLE_LIBC_EMCC
 #define WASM_ENABLE_LIBC_EMCC 0
--- a/core/iwasm/aot/aot_loader.c
+++ b/core/iwasm/aot/aot_loader.c
@ -1644,27 +1644,6 @@ load_function_section(const uint8 *buf, const uint8 *buf_end, AOTModule *module,
    const uint8 *p = buf, *p_end = buf_end;
    uint32 i;
    uint64 size, text_offset;
-#if defined(OS_ENABLE_HW_BOUND_CHECK) && defined(BH_PLATFORM_WINDOWS)
-    RUNTIME_FUNCTION *rtl_func_table;
-    AOTUnwindInfo *unwind_info;
-    uint32 unwind_info_offset = module->code_size - sizeof(AOTUnwindInfo);
-    uint32 unwind_code_offset = unwind_info_offset - PLT_ITEM_SIZE;
-#endif
-
-#if defined(OS_ENABLE_HW_BOUND_CHECK) && defined(BH_PLATFORM_WINDOWS)
-    unwind_info = (AOTUnwindInfo *)((uint8 *)module->code + module->code_size
-                                    - sizeof(AOTUnwindInfo));
-    unwind_info->Version = 1;
-    unwind_info->Flags = UNW_FLAG_NHANDLER;
-    *(uint32 *)&unwind_info->UnwindCode[0] = unwind_code_offset;
-
-    size = sizeof(RUNTIME_FUNCTION) * (uint64)module->func_count;
-    if (size > 0
-        && !(rtl_func_table = module->rtl_func_table =
-                 loader_malloc(size, error_buf, error_buf_size))) {
-        return false;
-    }
-#endif

    size = sizeof(void *) * (uint64)module->func_count;
    if (size > 0
@ -1691,33 +1670,9 @@ load_function_section(const uint8 *buf, const uint8 *buf_end, AOTModule *module,
 #if defined(BUILD_TARGET_THUMB) || defined(BUILD_TARGET_THUMB_VFP)
        /* bits[0] of thumb function address must be 1 */
        module->func_ptrs[i] = (void *)((uintptr_t)module->func_ptrs[i] | 1);
-#endif
-#if defined(OS_ENABLE_HW_BOUND_CHECK) && defined(BH_PLATFORM_WINDOWS)
-        rtl_func_table[i].BeginAddress = (DWORD)text_offset;
-        if (i > 0) {
-            rtl_func_table[i - 1].EndAddress = rtl_func_table[i].BeginAddress;
-        }
-        rtl_func_table[i].UnwindInfoAddress = (DWORD)unwind_info_offset;
 #endif
    }

-#if defined(OS_ENABLE_HW_BOUND_CHECK) && defined(BH_PLATFORM_WINDOWS)
-    if (module->func_count > 0) {
-        uint32 plt_table_size =
-            module->is_indirect_mode ? 0 : get_plt_table_size();
-        rtl_func_table[module->func_count - 1].EndAddress =
-            (DWORD)(module->code_size - plt_table_size);
-
-        if (!RtlAddFunctionTable(rtl_func_table, module->func_count,
-                                 (DWORD64)(uintptr_t)module->code)) {
-            set_error_buf(error_buf, error_buf_size,
-                          "add dynamic function table failed");
-            return false;
-        }
-        module->rtl_func_table_registered = true;
-    }
-#endif
-
    /* Set start function when function pointers are resolved */
    if (module->start_func_index != (uint32)-1) {
        if (module->start_func_index >= module->import_func_count)
@ -3261,14 +3216,6 @@ aot_unload(AOTModule *module)
    }
 #endif

-#if defined(OS_ENABLE_HW_BOUND_CHECK) && defined(BH_PLATFORM_WINDOWS)
-    if (module->rtl_func_table) {
-        if (module->rtl_func_table_registered)
-            RtlDeleteFunctionTable(module->rtl_func_table);
-        wasm_runtime_free(module->rtl_func_table);
-    }
-#endif
-
 #if (defined(BUILD_TARGET_X86_64) || defined(BUILD_TARGET_AMD_64)) \
    && !defined(BH_PLATFORM_WINDOWS)
    {
--- a/core/iwasm/aot/aot_runtime.c
+++ b/core/iwasm/aot/aot_runtime.c
@ -420,7 +420,7 @@ memory_instantiate(AOTModuleInstance *module_inst, AOTModuleInstance *parent,
        if (num_bytes_per_page < heap_size) {
            set_error_buf(error_buf, error_buf_size,
                          "failed to insert app heap into linear memory, "
-                          "try using `--heap_size=0` option");
+                          "try using `--heap-size=0` option");
            return NULL;
        }
    }
@ -478,7 +478,7 @@ memory_instantiate(AOTModuleInstance *module_inst, AOTModuleInstance *parent,
        if (init_page_count > DEFAULT_MAX_PAGES) {
            set_error_buf(error_buf, error_buf_size,
                          "failed to insert app heap into linear memory, "
-                          "try using `--heap_size=0` option");
+                          "try using `--heap-size=0` option");
            return NULL;
        }
        else if (init_page_count == DEFAULT_MAX_PAGES) {
@ -1271,9 +1271,9 @@ aot_deinstantiate(AOTModuleInstance *module_inst, bool is_sub_inst)
    if (module_inst->func_type_indexes)
        wasm_runtime_free(module_inst->func_type_indexes);

-    if (((AOTModuleInstanceExtra *)module_inst->e)->c_api_func_imports)
-        wasm_runtime_free(
-            ((AOTModuleInstanceExtra *)module_inst->e)->c_api_func_imports);
+    if (((AOTModuleInstanceExtra *)module_inst->e)->common.c_api_func_imports)
+        wasm_runtime_free(((AOTModuleInstanceExtra *)module_inst->e)
+                              ->common.c_api_func_imports);

    if (!is_sub_inst) {
 #if WASM_ENABLE_LIBC_WASI != 0
@ -1925,8 +1925,8 @@ aot_invoke_native(WASMExecEnv *exec_env, uint32 func_idx, uint32 argc,
    AOTModuleInstanceExtra *module_inst_extra =
        (AOTModuleInstanceExtra *)module_inst->e;
    CApiFuncImport *c_api_func_import =
-        module_inst_extra->c_api_func_imports
-            ? module_inst_extra->c_api_func_imports + func_idx
+        module_inst_extra->common.c_api_func_imports
+            ? module_inst_extra->common.c_api_func_imports + func_idx
            : NULL;
    uint32 *func_type_indexes = module_inst->func_type_indexes;
    uint32 func_type_idx = func_type_indexes[func_idx];
--- a/core/iwasm/aot/aot_runtime.h
+++ b/core/iwasm/aot/aot_runtime.h
@ -89,37 +89,9 @@ typedef struct AOTFunctionInstance {

 typedef struct AOTModuleInstanceExtra {
    DefPointer(const uint32 *, stack_sizes);
-    CApiFuncImport *c_api_func_imports;
-#if WASM_CONFIGUABLE_BOUNDS_CHECKS != 0
-    /* Disable bounds checks or not */
-    bool disable_bounds_checks;
-#endif
+    WASMModuleInstanceExtraCommon common;
 } AOTModuleInstanceExtra;

-#if defined(OS_ENABLE_HW_BOUND_CHECK) && defined(BH_PLATFORM_WINDOWS)
-/* clang-format off */
-typedef struct AOTUnwindInfo {
-    uint8 Version       : 3;
-    uint8 Flags         : 5;
-    uint8 SizeOfProlog;
-    uint8 CountOfCodes;
-    uint8 FrameRegister : 4;
-    uint8 FrameOffset   : 4;
-    struct {
-        struct {
-            uint8 CodeOffset;
-            uint8 UnwindOp : 4;
-            uint8 OpInfo   : 4;
-        };
-        uint16 FrameOffset;
-    } UnwindCode[1];
-} AOTUnwindInfo;
-/* clang-format on */
-
-/* size of mov instruction and jmp instruction */
-#define PLT_ITEM_SIZE 12
-#endif
-
 #if defined(BUILD_TARGET_X86_64) || defined(BUILD_TARGET_AMD_64)
 typedef struct GOTItem {
    uint32 func_idx;
@ -215,14 +187,6 @@ typedef struct AOTModule {
    uint32 float_plt_count;
 #endif

-#if defined(OS_ENABLE_HW_BOUND_CHECK) && defined(BH_PLATFORM_WINDOWS)
-    /* dynamic function table to be added by RtlAddFunctionTable(),
-       used to unwind the call stack and register exception handler
-       for AOT functions */
-    RUNTIME_FUNCTION *rtl_func_table;
-    bool rtl_func_table_registered;
-#endif
-
 #if defined(BUILD_TARGET_X86_64) || defined(BUILD_TARGET_AMD_64)
    uint32 got_item_count;
    GOTItemList got_item_list;
--- a/core/iwasm/aot/arch/aot_reloc_x86_32.c
+++ b/core/iwasm/aot/arch/aot_reloc_x86_32.c
@ -164,6 +164,9 @@ apply_relocation(AOTModule *module, uint8 *target_section_addr,
                (uint32)((uintptr_t)symbol_addr + (intptr_t)reloc_addend
                         - (uintptr_t)(target_section_addr
                                       + (uint32)reloc_offset)
+#if defined(BH_PLATFORM_WINDOWS)
+                         - sizeof(int32)
+#endif
                         + value); /* S + A - P */
            break;
        }
--- a/core/iwasm/aot/arch/aot_reloc_x86_64.c
+++ b/core/iwasm/aot/arch/aot_reloc_x86_64.c
@ -69,9 +69,6 @@ get_plt_table_size()
 {
    uint32 size =
        get_plt_item_size() * (sizeof(target_sym_map) / sizeof(SymbolMap));
-#if defined(OS_ENABLE_HW_BOUND_CHECK) && defined(BH_PLATFORM_WINDOWS)
-    size += get_plt_item_size() + sizeof(AOTUnwindInfo);
-#endif
    return size;
 }

@ -93,18 +90,6 @@ init_plt_table(uint8 *plt)
        *p++ = 0xE0;
        plt += get_plt_item_size();
    }
-
-#if defined(OS_ENABLE_HW_BOUND_CHECK) && defined(BH_PLATFORM_WINDOWS)
-    p = plt;
-    /* mov exception_handler, rax */
-    *p++ = 0x48;
-    *p++ = 0xB8;
-    *(uint64 *)p = 0; /*(uint64)(uintptr_t)aot_exception_handler;*/
-    p += sizeof(uint64);
-    /* jmp rax */
-    *p++ = 0xFF;
-    *p++ = 0xE0;
-#endif
 }

 static bool
@ -242,7 +227,7 @@ apply_relocation(AOTModule *module, uint8 *target_section_addr,
                     - (uintptr_t)(target_section_addr + reloc_offset));
            }
            else {
-                target_addr = (intptr_t) /* L + A - P */
+                target_addr = (intptr_t) /* S + A - P */
                    ((uintptr_t)symbol_addr + reloc_addend
                     - (uintptr_t)(target_section_addr + reloc_offset));
            }
--- a/core/iwasm/aot/iwasm_aot.cmake
+++ b/core/iwasm/aot/iwasm_aot.cmake
@ -36,5 +36,49 @@ if (WAMR_BUILD_DEBUG_AOT EQUAL 1)
  file(GLOB debug_source ${IWASM_AOT_DIR}/debug/*.c)
 endif()

-set (IWASM_AOT_SOURCE ${c_source_all} ${arch_source} ${debug_source})
+if ((WAMR_BUILD_TARGET STREQUAL "X86_64" OR WAMR_BUILD_TARGET STREQUAL "AMD_64")
+    AND (WAMR_BUILD_PLATFORM STREQUAL "windows")
+    AND (NOT WAMR_DISABLE_HW_BOUND_CHECK EQUAL 1))
+  include(FetchContent)

+  FetchContent_Declare(
+    zycore
+    GIT_REPOSITORY https://github.com/zyantific/zycore-c.git
+  )
+  FetchContent_GetProperties(zycore)
+  if (NOT zycore_POPULATED)
+    message ("-- Fetching zycore ..")
+    FetchContent_Populate(zycore)
+    include_directories("${zycore_SOURCE_DIR}/include")
+    include_directories("${zycore_BINARY_DIR}")
+    add_definitions(-DZYCORE_STATIC_BUILD=1)
+    add_subdirectory(${zycore_SOURCE_DIR} ${zycore_BINARY_DIR} EXCLUDE_FROM_ALL)
+    file (GLOB_RECURSE c_source_zycore ${zycore_SOURCE_DIR}/src/*.c)
+  endif ()
+
+  FetchContent_Declare(
+    zydis
+    GIT_REPOSITORY https://github.com/zyantific/zydis.git
+    GIT_TAG e14a07895136182a5b53e181eec3b1c6e0b434de
+  )
+  FetchContent_GetProperties(zydis)
+  if (NOT zydis_POPULATED)
+    message ("-- Fetching zydis ..")
+    FetchContent_Populate(zydis)
+    option(ZYDIS_FEATURE_ENCODER "" OFF)
+    option(ZYDIS_BUILD_TOOLS "" OFF)
+    option(ZYDIS_BUILD_EXAMPLES "" OFF)
+    option(ZYDIS_BUILD_MAN "" OFF)
+    option(ZYDIS_BUILD_DOXYGEN "" OFF)
+    include_directories("${zydis_BINARY_DIR}")
+    include_directories("${zydis_SOURCE_DIR}/include")
+    include_directories("${zydis_SOURCE_DIR}/src")
+    add_definitions(-DZYDIS_STATIC_BUILD=1)
+    add_subdirectory(${zydis_SOURCE_DIR} ${zydis_BINARY_DIR} EXCLUDE_FROM_ALL)
+    file (GLOB_RECURSE c_source_zydis ${zydis_SOURCE_DIR}/src/*.c)
+  endif ()
+endif ()
+
+
+set (IWASM_AOT_SOURCE ${c_source_all} ${arch_source} ${debug_source}
+	              ${c_source_zycore} ${c_source_zydis})
--- a/core/iwasm/common/wasm_c_api.c
+++ b/core/iwasm/common/wasm_c_api.c
@ -2290,8 +2290,10 @@ quit:
 bool
 wasm_module_validate(wasm_store_t *store, const wasm_byte_vec_t *binary)
 {
+    wasm_byte_vec_t local_binary = { 0 };
    struct WASMModuleCommon *module_rt;
    char error_buf[128] = { 0 };
+    bool ret;

    bh_assert(singleton_engine);

@ -2300,15 +2302,25 @@ wasm_module_validate(wasm_store_t *store, const wasm_byte_vec_t *binary)
        return false;
    }

-    if ((module_rt = wasm_runtime_load((uint8 *)binary->data,
-                                       (uint32)binary->size, error_buf, 128))) {
+    /* make a copy of binary */
+    wasm_byte_vec_copy(&local_binary, binary);
+
+    if (binary->size && !local_binary.data)
+        return false;
+
+    module_rt = wasm_runtime_load((uint8 *)local_binary.data,
+                                  (uint32)local_binary.size, error_buf, 128);
+    wasm_byte_vec_delete(&local_binary);
+    if (module_rt) {
        wasm_runtime_unload(module_rt);
-        return true;
+        ret = true;
    }
    else {
+        ret = false;
        LOG_VERBOSE(error_buf);
-        return false;
    }
+
+    return ret;
 }

 static void
@ -4858,7 +4870,7 @@ wasm_instance_new_with_args(wasm_store_t *store, const wasm_module_t *module,
    if (instance->inst_comm_rt->module_type == Wasm_Module_Bytecode) {
        WASMModuleInstanceExtra *e =
            ((WASMModuleInstance *)instance->inst_comm_rt)->e;
-        p_func_imports = &(e->c_api_func_imports);
+        p_func_imports = &(e->common.c_api_func_imports);
        import_func_count = MODULE_INTERP(module)->import_function_count;
    }
 #endif
@ -4868,7 +4880,7 @@ wasm_instance_new_with_args(wasm_store_t *store, const wasm_module_t *module,
            (AOTModuleInstanceExtra *)((AOTModuleInstance *)
                                           instance->inst_comm_rt)
                ->e;
-        p_func_imports = &(e->c_api_func_imports);
+        p_func_imports = &(e->common.c_api_func_imports);
        import_func_count = MODULE_AOT(module)->import_func_count;
    }
 #endif
--- a/core/iwasm/common/wasm_runtime_common.c
+++ b/core/iwasm/common/wasm_runtime_common.c
@ -199,7 +199,90 @@ runtime_signal_handler(void *sig_addr)
        }
    }
 }
-#else
+#else /* else of BH_PLATFORM_WINDOWS */
+
+#if WASM_ENABLE_AOT != 0
+#include <Zydis/Zydis.h>
+
+static uint32
+decode_insn(uint8 *insn)
+{
+    uint8 *data = (uint8 *)insn;
+    uint32 length = 32; /* reserve enough size */
+
+    /* Initialize decoder context */
+    ZydisDecoder decoder;
+    ZydisDecoderInit(&decoder, ZYDIS_MACHINE_MODE_LONG_64,
+                     ZYDIS_STACK_WIDTH_64);
+
+    /* Initialize formatter */
+    ZydisFormatter formatter;
+    ZydisFormatterInit(&formatter, ZYDIS_FORMATTER_STYLE_INTEL);
+
+    /* Loop over the instructions in our buffer */
+    ZyanU64 runtime_address = (ZyanU64)(uintptr_t)data;
+    ZyanUSize offset = 0;
+    ZydisDecodedInstruction instruction;
+    ZydisDecodedOperand operands[ZYDIS_MAX_OPERAND_COUNT_VISIBLE];
+    char buffer[256];
+
+    if (ZYAN_SUCCESS(ZydisDecoderDecodeFull(
+            &decoder, data + offset, length - offset, &instruction, operands,
+            ZYDIS_MAX_OPERAND_COUNT_VISIBLE,
+            ZYDIS_DFLAG_VISIBLE_OPERANDS_ONLY))) {
+
+        /* Format & print the binary instruction structure to
+           human readable format */
+        ZydisFormatterFormatInstruction(&formatter, &instruction, operands,
+                                        instruction.operand_count_visible,
+                                        buffer, sizeof(buffer),
+                                        runtime_address);
+
+        /* Print current instruction */
+        /*
+        os_printf("%012" PRIX64 "  ", runtime_address);
+        puts(buffer);
+        */
+
+        return instruction.length;
+    }
+
+    /* Decode failed */
+    return 0;
+}
+#endif /* end of WASM_ENABLE_AOT != 0 */
+
+static LONG
+next_action(WASMModuleInstance *module_inst, EXCEPTION_POINTERS *exce_info)
+{
+#if WASM_ENABLE_AOT != 0
+    uint32 insn_size;
+#endif
+
+    if (module_inst->module_type == Wasm_Module_Bytecode
+        && module_inst->e->running_mode == Mode_Interp) {
+        /* Continue to search next exception handler for
+           interpreter mode as it can be caught by
+           `__try { .. } __except { .. }` sentences in
+           wasm_runtime.c */
+        return EXCEPTION_CONTINUE_SEARCH;
+    }
+
+#if WASM_ENABLE_AOT != 0
+    /* Skip current instruction and continue to run for AOT/JIT mode.
+       TODO: implement unwind support for AOT/JIT code in Windows platform */
+    insn_size = decode_insn((uint8 *)exce_info->ContextRecord->Rip);
+    if (insn_size > 0) {
+        exce_info->ContextRecord->Rip += insn_size;
+        return EXCEPTION_CONTINUE_EXECUTION;
+    }
+#endif
+
+    /* return different value from EXCEPTION_CONTINUE_SEARCH (= 0)
+       and EXCEPTION_CONTINUE_EXECUTION (= -1) */
+    return -2;
+}
+
 static LONG
 runtime_exception_handler(EXCEPTION_POINTERS *exce_info)
 {
@ -211,6 +294,7 @@ runtime_exception_handler(EXCEPTION_POINTERS *exce_info)
    uint8 *mapped_mem_start_addr = NULL;
    uint8 *mapped_mem_end_addr = NULL;
    uint32 page_size = os_getpagesize();
+    LONG ret;

    if (exec_env_tls && exec_env_tls->handle == os_self_thread()
        && (jmpbuf_node = exec_env_tls->jmpbuf_stack_top)) {
@ -232,32 +316,19 @@ runtime_exception_handler(EXCEPTION_POINTERS *exce_info)
                   the wasm func returns, the caller will check whether the
                   exception is thrown and return to runtime. */
                wasm_set_exception(module_inst, "out of bounds memory access");
-                if (module_inst->module_type == Wasm_Module_Bytecode) {
-                    /* Continue to search next exception handler for
-                       interpreter mode as it can be caught by
-                       `__try { .. } __except { .. }` sentences in
-                       wasm_runtime.c */
-                    return EXCEPTION_CONTINUE_SEARCH;
-                }
-                else {
-                    /* Skip current instruction and continue to run for
-                       AOT mode. TODO: implement unwind support for AOT
-                       code in Windows platform */
-                    exce_info->ContextRecord->Rip++;
-                    return EXCEPTION_CONTINUE_EXECUTION;
-                }
+                ret = next_action(module_inst, exce_info);
+                if (ret == EXCEPTION_CONTINUE_SEARCH
+                    || ret == EXCEPTION_CONTINUE_EXECUTION)
+                    return ret;
            }
            else if (exec_env_tls->exce_check_guard_page <= (uint8 *)sig_addr
                     && (uint8 *)sig_addr
                            < exec_env_tls->exce_check_guard_page + page_size) {
                bh_assert(wasm_copy_exception(module_inst, NULL));
-                if (module_inst->module_type == Wasm_Module_Bytecode) {
-                    return EXCEPTION_CONTINUE_SEARCH;
-                }
-                else {
-                    exce_info->ContextRecord->Rip++;
-                    return EXCEPTION_CONTINUE_EXECUTION;
-                }
+                ret = next_action(module_inst, exce_info);
+                if (ret == EXCEPTION_CONTINUE_SEARCH
+                    || ret == EXCEPTION_CONTINUE_EXECUTION)
+                    return ret;
            }
        }
 #if WASM_DISABLE_STACK_HW_BOUND_CHECK == 0
@ -267,12 +338,10 @@ runtime_exception_handler(EXCEPTION_POINTERS *exce_info)
               whether the exception is thrown and return to runtime, and
               the damaged stack will be recovered by _resetstkoflw(). */
            wasm_set_exception(module_inst, "native stack overflow");
-            if (module_inst->module_type == Wasm_Module_Bytecode) {
-                return EXCEPTION_CONTINUE_SEARCH;
-            }
-            else {
-                return EXCEPTION_CONTINUE_EXECUTION;
-            }
+            ret = next_action(module_inst, exce_info);
+            if (ret == EXCEPTION_CONTINUE_SEARCH
+                || ret == EXCEPTION_CONTINUE_EXECUTION)
+                return ret;
        }
 #endif
    }
@ -2491,14 +2560,14 @@ wasm_runtime_set_bounds_checks(WASMModuleInstanceCommon *module_inst,
 #if WASM_ENABLE_INTERP != 0
    if (module_inst->module_type == Wasm_Module_Bytecode) {
        ((WASMModuleInstanceExtra *)((WASMModuleInstance *)module_inst)->e)
-            ->disable_bounds_checks = enable ? false : true;
+            ->common.disable_bounds_checks = enable ? false : true;
    }
 #endif

 #if WASM_ENABLE_AOT != 0
    if (module_inst->module_type == Wasm_Module_AoT) {
        ((AOTModuleInstanceExtra *)((AOTModuleInstance *)module_inst)->e)
-            ->disable_bounds_checks = enable ? false : true;
+            ->common.disable_bounds_checks = enable ? false : true;
    }
 #endif
 }
@ -2511,7 +2580,7 @@ wasm_runtime_is_bounds_checks_enabled(WASMModuleInstanceCommon *module_inst)
    if (module_inst->module_type == Wasm_Module_Bytecode) {
        return !((WASMModuleInstanceExtra *)((WASMModuleInstance *)module_inst)
                     ->e)
-                    ->disable_bounds_checks;
+                    ->common.disable_bounds_checks;
    }
 #endif

@ -2519,7 +2588,7 @@ wasm_runtime_is_bounds_checks_enabled(WASMModuleInstanceCommon *module_inst)
    if (module_inst->module_type == Wasm_Module_AoT) {
        return !((AOTModuleInstanceExtra *)((WASMModuleInstance *)module_inst)
                     ->e)
-                    ->disable_bounds_checks;
+                    ->common.disable_bounds_checks;
    }
 #endif

@ -4692,6 +4761,8 @@ typedef struct ExternRefMapNode {
    bool retained;
    /* Whether it is marked by runtime */
    bool marked;
+    /* cleanup function called when the externref is freed */
+    void (*cleanup)(void *);
 } ExternRefMapNode;

 static uint32
@ -4754,6 +4825,81 @@ lookup_extobj_callback(void *key, void *value, void *user_data)
    }
 }

+static void
+delete_externref(void *key, ExternRefMapNode *node)
+{
+    bh_hash_map_remove(externref_map, key, NULL, NULL);
+    if (node->cleanup) {
+        (*node->cleanup)(node->extern_obj);
+    }
+    wasm_runtime_free(node);
+}
+
+static void
+delete_extobj_callback(void *key, void *value, void *user_data)
+{
+    ExternRefMapNode *node = (ExternRefMapNode *)value;
+    LookupExtObj_UserData *lookup_user_data =
+        (LookupExtObj_UserData *)user_data;
+
+    if (node->extern_obj == lookup_user_data->node.extern_obj
+        && node->module_inst == lookup_user_data->node.module_inst) {
+        lookup_user_data->found = true;
+        delete_externref(key, node);
+    }
+}
+
+bool
+wasm_externref_objdel(WASMModuleInstanceCommon *module_inst, void *extern_obj)
+{
+    LookupExtObj_UserData lookup_user_data = { 0 };
+    bool ok = false;
+
+    /* in a wrapper, extern_obj could be any value */
+    lookup_user_data.node.extern_obj = extern_obj;
+    lookup_user_data.node.module_inst = module_inst;
+    lookup_user_data.found = false;
+
+    os_mutex_lock(&externref_lock);
+    /* Lookup hashmap firstly */
+    bh_hash_map_traverse(externref_map, delete_extobj_callback,
+                         (void *)&lookup_user_data);
+    if (lookup_user_data.found) {
+        ok = true;
+    }
+    os_mutex_unlock(&externref_lock);
+
+    return ok;
+}
+
+bool
+wasm_externref_set_cleanup(WASMModuleInstanceCommon *module_inst,
+                           void *extern_obj, void (*extern_obj_cleanup)(void *))
+{
+
+    LookupExtObj_UserData lookup_user_data = { 0 };
+    bool ok = false;
+
+    /* in a wrapper, extern_obj could be any value */
+    lookup_user_data.node.extern_obj = extern_obj;
+    lookup_user_data.node.module_inst = module_inst;
+    lookup_user_data.found = false;
+
+    os_mutex_lock(&externref_lock);
+    /* Lookup hashmap firstly */
+    bh_hash_map_traverse(externref_map, lookup_extobj_callback,
+                         (void *)&lookup_user_data);
+    if (lookup_user_data.found) {
+        void *key = (void *)(uintptr_t)lookup_user_data.externref_idx;
+        ExternRefMapNode *node = bh_hash_map_find(externref_map, key);
+        node->cleanup = extern_obj_cleanup;
+        ok = true;
+    }
+    os_mutex_unlock(&externref_lock);
+
+    return ok;
+}
+
 bool
 wasm_externref_obj2ref(WASMModuleInstanceCommon *module_inst, void *extern_obj,
                       uint32 *p_externref_idx)
@ -4803,6 +4949,7 @@ wasm_externref_obj2ref(WASMModuleInstanceCommon *module_inst, void *extern_obj,
    memset(node, 0, sizeof(ExternRefMapNode));
    node->extern_obj = extern_obj;
    node->module_inst = module_inst;
+    node->cleanup = NULL;

    externref_idx = externref_global_id;

@ -4853,8 +5000,7 @@ reclaim_extobj_callback(void *key, void *value, void *user_data)

    if (node->module_inst == module_inst) {
        if (!node->marked && !node->retained) {
-            bh_hash_map_remove(externref_map, key, NULL, NULL);
-            wasm_runtime_free(value);
+            delete_externref(key, node);
        }
        else {
            node->marked = false;
@ -4969,8 +5115,7 @@ cleanup_extobj_callback(void *key, void *value, void *user_data)
        (WASMModuleInstanceCommon *)user_data;

    if (node->module_inst == module_inst) {
-        bh_hash_map_remove(externref_map, key, NULL, NULL);
-        wasm_runtime_free(value);
+        delete_externref(key, node);
    }
 }

--- a/core/iwasm/compilation/aot_compiler.c
+++ b/core/iwasm/compilation/aot_compiler.c
@ -2735,6 +2735,33 @@ aot_generate_tempfile_name(const char *prefix, const char *extension,
    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)
+{
+    int name_len;
+
+    name_len = snprintf(buffer, len, "%s-XXXXXX", prefix);
+
+    if (_mktemp_s(buffer, name_len + 1) != 0) {
+        return NULL;
+    }
+
+    /* 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;
+}
 #endif /* end of !(defined(_WIN32) || defined(_WIN32_)) */

 bool
--- a/core/iwasm/compilation/aot_emit_aot_file.c
+++ b/core/iwasm/compilation/aot_emit_aot_file.c
@ -93,7 +93,10 @@ check_utf8_str(const uint8 *str, uint32 len)
 /* Internal function in object file */
 typedef struct AOTObjectFunc {
    char *func_name;
+    /* text offset of aot_func#n */
    uint64 text_offset;
+    /* text offset of aot_func_internal#n */
+    uint64 text_offset_of_aot_func_internal;
 } AOTObjectFunc;

 /* Symbol table list node */
@ -637,13 +640,33 @@ get_relocation_size(AOTRelocation *relocation, bool is_32bin)
 }

 static uint32
-get_relocations_size(AOTRelocation *relocations, uint32 relocation_count,
+get_relocations_size(AOTObjectData *obj_data,
+                     AOTRelocationGroup *relocation_group,
+                     AOTRelocation *relocations, uint32 relocation_count,
                     bool is_32bin)
 {
    AOTRelocation *relocation = relocations;
    uint32 size = 0, i;

    for (i = 0; i < relocation_count; i++, relocation++) {
+        /* ignore the relocations to aot_func_internal#n in text section
+           for windows platform since they will be applied in
+           aot_emit_text_section */
+        if (!strcmp(relocation_group->section_name, ".text")
+            && !strncmp(relocation->symbol_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 */
+                 && obj_data->target_info.bin_type == 6
+                 /* IMAGE_REL_AMD64_REL32 in windows x86_64 */
+                 && relocation->relocation_type == 4)
+                || (!strncmp(obj_data->comp_ctx->target_arch, "i386", 4)
+                    /* Windows AOT_COFF32_BIN_TYPE */
+                    && obj_data->target_info.bin_type == 4
+                    /* IMAGE_REL_I386_REL32 in windows x86_32 */
+                    && relocation->relocation_type == 20))) {
+            continue;
+        }
        size = align_uint(size, 4);
        size += get_relocation_size(relocation, is_32bin);
    }
@ -651,19 +674,22 @@ get_relocations_size(AOTRelocation *relocations, uint32 relocation_count,
 }

 static uint32
-get_relocation_group_size(AOTRelocationGroup *relocation_group, bool is_32bin)
+get_relocation_group_size(AOTObjectData *obj_data,
+                          AOTRelocationGroup *relocation_group, bool is_32bin)
 {
    uint32 size = 0;
    /* section name index + relocation count + relocations */
    size += (uint32)sizeof(uint32);
    size += (uint32)sizeof(uint32);
-    size += get_relocations_size(relocation_group->relocations,
+    size += get_relocations_size(obj_data, relocation_group,
+                                 relocation_group->relocations,
                                 relocation_group->relocation_count, is_32bin);
    return size;
 }

 static uint32
-get_relocation_groups_size(AOTRelocationGroup *relocation_groups,
+get_relocation_groups_size(AOTObjectData *obj_data,
+                           AOTRelocationGroup *relocation_groups,
                           uint32 relocation_group_count, bool is_32bin)
 {
    AOTRelocationGroup *relocation_group = relocation_groups;
@ -671,7 +697,7 @@ get_relocation_groups_size(AOTRelocationGroup *relocation_groups,

    for (i = 0; i < relocation_group_count; i++, relocation_group++) {
        size = align_uint(size, 4);
-        size += get_relocation_group_size(relocation_group, is_32bin);
+        size += get_relocation_group_size(obj_data, relocation_group, is_32bin);
    }
    return size;
 }
@ -864,7 +890,7 @@ get_relocation_section_size(AOTCompContext *comp_ctx, AOTObjectData *obj_data)

    /* relocation group count + symbol_table + relocation groups */
    return (uint32)sizeof(uint32) + symbol_table_size
-           + get_relocation_groups_size(relocation_groups,
+           + get_relocation_groups_size(obj_data, relocation_groups,
                                        relocation_group_count,
                                        is_32bit_binary(obj_data));
 }
@ -1734,6 +1760,10 @@ aot_emit_text_section(uint8 *buf, uint8 *buf_end, uint32 *p_offset,
    uint32 section_size = get_text_section_size(obj_data);
    uint32 offset = *p_offset;
    uint8 placeholder = 0;
+    AOTRelocationGroup *relocation_group;
+    AOTRelocation *relocation;
+    uint32 i, j, relocation_count;
+    uint8 *text;

    *p_offset = offset = align_uint(offset, 4);

@ -1747,6 +1777,8 @@ aot_emit_text_section(uint8 *buf, uint8 *buf_end, uint32 *p_offset,
            EMIT_BUF(&placeholder, 1);
    }

+    text = buf + offset;
+
    if (obj_data->text_size > 0) {
        EMIT_BUF(obj_data->text, obj_data->text_size);
        while (offset & 3)
@ -1768,6 +1800,67 @@ aot_emit_text_section(uint8 *buf, uint8 *buf_end, uint32 *p_offset,
        return false;
    }

+    /* apply relocations to aot_func_internal#n in text section for
+       windows platform */
+    if ((!strncmp(obj_data->comp_ctx->target_arch, "x86_64", 6)
+         /* Windows AOT_COFF64_BIN_TYPE */
+         && obj_data->target_info.bin_type == 6)
+        || (!strncmp(obj_data->comp_ctx->target_arch, "i386", 4)
+            /* Windows AOT_COFF32_BIN_TYPE */
+            && obj_data->target_info.bin_type == 4)) {
+        relocation_group = obj_data->relocation_groups;
+        for (i = 0; i < obj_data->relocation_group_count;
+             i++, relocation_group++) {
+            /* relocation in text section */
+            if (!strcmp(relocation_group->section_name, ".text")) {
+                relocation = relocation_group->relocations;
+                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)
+                        && ((obj_data->target_info.bin_type
+                                 == 6 /* AOT_COFF64_BIN_TYPE */
+                             && relocation->relocation_type
+                                    == 4 /* IMAGE_REL_AMD64_REL32 */)
+                            || (obj_data->target_info.bin_type
+                                    == 4 /* AOT_COFF32_BIN_TYPE */
+                                && relocation->relocation_type
+                                       == 20 /* IMAGE_REL_I386_REL32 */))) {
+                        uint32 func_idx =
+                            atoi(relocation->symbol_name
+                                 + strlen(AOT_FUNC_INTERNAL_PREFIX));
+                        uint64 text_offset, reloc_offset, reloc_addend;
+
+                        bh_assert(func_idx < obj_data->func_count);
+
+                        text_offset = obj_data->funcs[func_idx]
+                                          .text_offset_of_aot_func_internal;
+                        reloc_offset = relocation->relocation_offset;
+                        reloc_addend = relocation->relocation_addend;
+                        /* S + A - P */
+                        *(uint32 *)(text + reloc_offset) =
+                            (uint32)(text_offset + reloc_addend - reloc_offset
+                                     - 4);
+
+                        /* remove current relocation as it has been applied */
+                        if (j < relocation_count - 1) {
+                            uint32 move_size =
+                                (uint32)(sizeof(AOTRelocation)
+                                         * (relocation_count - 1 - j));
+                            bh_memmove_s(relocation, move_size, relocation + 1,
+                                         move_size);
+                        }
+                        relocation_group->relocation_count--;
+                    }
+                    else {
+                        relocation++;
+                    }
+                }
+            }
+        }
+    }
+
    *p_offset = offset;

    return true;
@ -2403,7 +2496,7 @@ aot_resolve_object_data_sections(AOTObjectData *obj_data)
                    && !strcmp(name, "__llvm_prf_cnts")) {
                    snprintf(buf, sizeof(buf), "%s%u", name,
                             llvm_prf_cnts_idx++);
-                    size = strlen(buf) + 1;
+                    size = (uint32)(strlen(buf) + 1);
                    if (!(data_section->name = wasm_runtime_malloc(size))) {
                        aot_set_last_error(
                            "allocate memory for data section name failed.");
@ -2416,7 +2509,7 @@ aot_resolve_object_data_sections(AOTObjectData *obj_data)
                         && !strcmp(name, "__llvm_prf_data")) {
                    snprintf(buf, sizeof(buf), "%s%u", name,
                             llvm_prf_data_idx++);
-                    size = strlen(buf) + 1;
+                    size = (uint32)(strlen(buf) + 1);
                    if (!(data_section->name = wasm_runtime_malloc(size))) {
                        aot_set_last_error(
                            "allocate memory for data section name failed.");
@ -2520,15 +2613,15 @@ read_stack_usage_file(const AOTCompContext *comp_ctx, const char *filename,
        }
        if (prefix == aot_func_prefix) {
            if (sz < precheck_stack_size_min) {
-                precheck_stack_size_min = sz;
+                precheck_stack_size_min = (uint32)sz;
            }
            if (sz > precheck_stack_size_max) {
-                precheck_stack_size_max = sz;
+                precheck_stack_size_max = (uint32)sz;
            }
            precheck_found++;
            continue;
        }
-        sizes[func_idx] = sz;
+        sizes[func_idx] = (uint32)sz;
        found++;
    }
    fclose(fp);
@ -2605,9 +2698,16 @@ 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)) {
+            && (!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)))) {
            uint64 sz = LLVMGetSymbolSize(sym_itr);
-            if (sz != sizeof(uint32) * obj_data->func_count) {
+            if (sz != sizeof(uint32) * obj_data->func_count
+                /* sz of COFF64/COFF32 is 0, ignore the check */
+                && obj_data->target_info.bin_type != AOT_COFF64_BIN_TYPE
+                && obj_data->target_info.bin_type != AOT_COFF32_BIN_TYPE) {
                aot_set_last_error("stack_sizes had unexpected size.");
                goto fail;
            }
@ -2642,16 +2742,12 @@ aot_resolve_stack_sizes(AOTCompContext *comp_ctx, AOTObjectData *obj_data)
                    goto fail;
                }
            }
-            if (addr > UINT32_MAX) {
-                aot_set_last_error("too large stack_sizes offset.");
-                goto fail;
-            }
            /*
             * Record section/offset and construct a copy of stack_sizes.
             * aot_emit_object_data_section_info will emit this copy.
             */
            obj_data->stack_sizes_section_name = sec_name;
-            obj_data->stack_sizes_offset = addr;
+            obj_data->stack_sizes_offset = (uint32)addr;
            obj_data->stack_sizes = wasm_runtime_malloc(
                obj_data->func_count * sizeof(*obj_data->stack_sizes));
            if (obj_data->stack_sizes == NULL) {
@ -2770,6 +2866,7 @@ 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)) {
+            /* symbol aot_func#n */
            func_index = (uint32)atoi(name + strlen(prefix));
            if (func_index < obj_data->func_count) {
                LLVMSectionIteratorRef contain_section;
@ -2804,6 +2901,44 @@ aot_resolve_functions(AOTCompContext *comp_ctx, AOTObjectData *obj_data)
                }
            }
        }
+        else if ((name = (char *)LLVMGetSymbolName(sym_itr))
+                 && 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) {
+                LLVMSectionIteratorRef contain_section;
+                char *contain_section_name;
+
+                func = obj_data->funcs + func_index;
+
+                if (!(contain_section = LLVMObjectFileCopySectionIterator(
+                          obj_data->binary))) {
+                    aot_set_last_error("llvm get section iterator failed.");
+                    LLVMDisposeSymbolIterator(sym_itr);
+                    return false;
+                }
+                LLVMMoveToContainingSection(contain_section, sym_itr);
+                contain_section_name =
+                    (char *)LLVMGetSectionName(contain_section);
+                LLVMDisposeSectionIterator(contain_section);
+
+                if (!strcmp(contain_section_name, ".text.unlikely.")) {
+                    func->text_offset_of_aot_func_internal =
+                        align_uint(obj_data->text_size, 4)
+                        + LLVMGetSymbolAddress(sym_itr);
+                }
+                else if (!strcmp(contain_section_name, ".text.hot.")) {
+                    func->text_offset_of_aot_func_internal =
+                        align_uint(obj_data->text_size, 4)
+                        + align_uint(obj_data->text_unlikely_size, 4)
+                        + LLVMGetSymbolAddress(sym_itr);
+                }
+                else {
+                    func->text_offset_of_aot_func_internal =
+                        LLVMGetSymbolAddress(sym_itr);
+                }
+            }
+        }
        LLVMMoveToNextSymbol(sym_itr);
    }
    LLVMDisposeSymbolIterator(sym_itr);
@ -2975,7 +3110,7 @@ aot_resolve_object_relocation_group(AOTObjectData *obj_data,
                || !strcmp(group->section_name, ".rel.text")) {
                snprintf(buf, sizeof(buf), "%s%u", relocation->symbol_name,
                         prof_section_idx);
-                size = strlen(buf) + 1;
+                size = (uint32)(strlen(buf) + 1);
                if (!(relocation->symbol_name = wasm_runtime_malloc(size))) {
                    aot_set_last_error(
                        "allocate memory for relocation symbol name failed.");
@ -2990,7 +3125,7 @@ aot_resolve_object_relocation_group(AOTObjectData *obj_data,
                                 19)) {
                snprintf(buf, sizeof(buf), "%s%u", relocation->symbol_name,
                         prof_section_idx);
-                size = strlen(buf) + 1;
+                size = (uint32)(strlen(buf) + 1);
                if (!(relocation->symbol_name = wasm_runtime_malloc(size))) {
                    aot_set_last_error(
                        "allocate memory for relocation symbol name failed.");
@ -3156,7 +3291,7 @@ aot_resolve_object_relocation_groups(AOTObjectData *obj_data)
                    || !strcmp(name, ".rel__llvm_prf_data"))) {
                char buf[32];
                snprintf(buf, sizeof(buf), "%s%u", name, llvm_prf_data_idx);
-                size = strlen(buf) + 1;
+                size = (uint32)(strlen(buf) + 1);
                if (!(relocation_group->section_name =
                          wasm_runtime_malloc(size))) {
                    aot_set_last_error(
--- a/core/iwasm/compilation/aot_emit_function.c
+++ b/core/iwasm/compilation/aot_emit_function.c
@ -18,6 +18,17 @@
        }                                                                     \
    } while (0)

+static bool
+is_win_platform(AOTCompContext *comp_ctx)
+{
+    char *triple = LLVMGetTargetMachineTriple(comp_ctx->target_machine);
+
+    bh_assert(triple);
+    if (strstr(triple, "win32") || strstr(triple, "win"))
+        return true;
+    return false;
+}
+
 static bool
 create_func_return_block(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx)
 {
@ -458,7 +469,7 @@ check_app_addr_and_convert(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
    }

    /* Check whether exception was thrown when executing the function */
-    if (comp_ctx->enable_bound_check
+    if ((comp_ctx->enable_bound_check || is_win_platform(comp_ctx))
        && !check_call_return(comp_ctx, func_ctx, res)) {
        return false;
    }
@ -696,7 +707,7 @@ aot_compile_op_call(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
                goto fail;
            /* Check whether there was exception thrown when executing
               the function */
-            if (comp_ctx->enable_bound_check
+            if ((comp_ctx->enable_bound_check || is_win_platform(comp_ctx))
                && !check_call_return(comp_ctx, func_ctx, res))
                goto fail;
        }
@ -849,7 +860,8 @@ aot_compile_op_call(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,

        /* Check whether there was exception thrown when executing
           the function */
-        if (!tail_call && comp_ctx->enable_bound_check
+        if (!tail_call
+            && (comp_ctx->enable_bound_check || is_win_platform(comp_ctx))
            && !check_exception_thrown(comp_ctx, func_ctx))
            goto fail;
    }
@ -1431,7 +1443,7 @@ aot_compile_op_call_indirect(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
        goto fail;

    /* Check whether exception was thrown when executing the function */
-    if (comp_ctx->enable_bound_check
+    if ((comp_ctx->enable_bound_check || is_win_platform(comp_ctx))
        && !check_call_return(comp_ctx, func_ctx, res))
        goto fail;

@ -1483,7 +1495,7 @@ aot_compile_op_call_indirect(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
    }

    /* Check whether exception was thrown when executing the function */
-    if (comp_ctx->enable_bound_check
+    if ((comp_ctx->enable_bound_check || is_win_platform(comp_ctx))
        && !check_exception_thrown(comp_ctx, func_ctx))
        goto fail;

--- a/core/iwasm/compilation/aot_llvm.c
+++ b/core/iwasm/compilation/aot_llvm.c
@ -146,6 +146,13 @@ aot_target_precheck_can_use_musttail(const AOTCompContext *comp_ctx)
         */
        return false;
    }
+    if (!strcmp(comp_ctx->target_arch, "mips")) {
+        /*
+         * cf.
+         * https://github.com/bytecodealliance/wasm-micro-runtime/issues/2412
+         */
+        return false;
+    }
    /*
     * x86-64/i386: true
     *
@ -237,9 +244,10 @@ get_inst_extra_offset(AOTCompContext *comp_ctx)
    const AOTCompData *comp_data = comp_ctx->comp_data;
    uint32 table_count = comp_data->import_table_count + comp_data->table_count;
    uint64 offset = get_tbl_inst_offset(comp_ctx, NULL, table_count);
-    bh_assert(offset <= UINT_MAX);
-    offset = align_uint(offset, 8);
-    return offset;
+    uint32 offset_32 = (uint32)offset;
+    bh_assert(offset <= UINT32_MAX);
+    offset_32 = align_uint((uint32)offset_32, 8);
+    return offset_32;
 }

 /*
@ -309,8 +317,8 @@ aot_add_precheck_function(AOTCompContext *comp_ctx, LLVMModuleRef module,
        goto fail;
    }

-    unsigned int param_count = LLVMCountParams(precheck_func);
-    uint64 sz = param_count * sizeof(LLVMValueRef);
+    uint32 param_count = LLVMCountParams(precheck_func);
+    uint32 sz = param_count * (uint32)sizeof(LLVMValueRef);
    params = wasm_runtime_malloc(sz);
    if (params == NULL) {
        goto fail;
@ -626,8 +634,8 @@ aot_add_llvm_func(AOTCompContext *comp_ctx, LLVMModuleRef module,
    if (comp_ctx->is_indirect_mode) {
        /* avoid LUT relocations ("switch-table") */
        LLVMAttributeRef attr_no_jump_tables = LLVMCreateStringAttribute(
-            comp_ctx->context, "no-jump-tables", strlen("no-jump-tables"),
-            "true", strlen("true"));
+            comp_ctx->context, "no-jump-tables",
+            (uint32)strlen("no-jump-tables"), "true", (uint32)strlen("true"));
        LLVMAddAttributeAtIndex(func, LLVMAttributeFunctionIndex,
                                attr_no_jump_tables);
    }
@ -2081,7 +2089,7 @@ jit_stack_size_callback(void *user_data, const char *name, size_t namelen,
        return;
    }
    /* ensure NUL termination */
-    bh_memcpy_s(buf, sizeof(buf), name, namelen);
+    bh_memcpy_s(buf, (uint32)sizeof(buf), name, (uint32)namelen);
    buf[namelen] = 0;

    ret = sscanf(buf, AOT_FUNC_INTERNAL_PREFIX "%" SCNu32, &func_idx);
@ -2102,7 +2110,7 @@ jit_stack_size_callback(void *user_data, const char *name, size_t namelen,

    /* Note: -1 == AOT_NEG_ONE from aot_create_stack_sizes */
    bh_assert(comp_ctx->jit_stack_sizes[func_idx] == (uint32)-1);
-    comp_ctx->jit_stack_sizes[func_idx] = stack_size + call_size;
+    comp_ctx->jit_stack_sizes[func_idx] = (uint32)stack_size + call_size;
 }

 static bool
@ -2744,6 +2752,16 @@ aot_create_comp_context(const AOTCompData *comp_data, aot_comp_option_t option)
            aot_set_last_error("create LLVM target machine failed.");
            goto fail;
        }
+
+        /* If only to create target machine for querying information, early stop
+         */
+        if ((arch && !strcmp(arch, "help")) || (abi && !strcmp(abi, "help"))
+            || (cpu && !strcmp(cpu, "help"))
+            || (features && !strcmp(features, "+help"))) {
+            LOG_DEBUG(
+                "create LLVM target machine only for printing help info.");
+            goto fail;
+        }
    }

    triple = LLVMGetTargetMachineTriple(comp_ctx->target_machine);
--- a/core/iwasm/include/wasm_export.h
+++ b/core/iwasm/include/wasm_export.h
@ -186,6 +186,7 @@ enum wasm_valkind_enum {

 #ifndef WASM_VAL_T_DEFINED
 #define WASM_VAL_T_DEFINED
+struct wasm_ref_t;

 typedef struct wasm_val_t {
    wasm_valkind_t kind;
@ -197,6 +198,7 @@ typedef struct wasm_val_t {
        double f64;
        /* represent a foreign object, aka externref in .wat */
        uintptr_t foreign;
+        struct wasm_ref_t *ref;
    } of;
 } wasm_val_t;
 #endif
@ -1290,6 +1292,32 @@ WASM_RUNTIME_API_EXTERN bool
 wasm_externref_obj2ref(wasm_module_inst_t module_inst,
                       void *extern_obj, uint32_t *p_externref_idx);

+/**
+ * Delete external object registered by `wasm_externref_obj2ref`.
+ *
+ * @param module_inst the WASM module instance that the extern object
+ *        belongs to
+ * @param extern_obj the external object to be deleted
+ *
+ * @return true if success, false otherwise
+ */
+WASM_RUNTIME_API_EXTERN bool
+wasm_externref_objdel(wasm_module_inst_t module_inst, void *extern_obj);
+
+/**
+ * Set cleanup callback to release external object.
+ *
+ * @param module_inst the WASM module instance that the extern object
+ *        belongs to
+ * @param extern_obj the external object to which to set the `extern_obj_cleanup` cleanup callback.
+ * @param extern_obj_cleanup a callback to release `extern_obj`
+ *
+ * @return true if success, false otherwise
+ */
+WASM_RUNTIME_API_EXTERN bool
+wasm_externref_set_cleanup(wasm_module_inst_t module_inst, void *extern_obj,
+                           void (*extern_obj_cleanup)(void *));
+
 /**
 * Retrieve the external object from an internal externref index
 *
--- a/core/iwasm/interpreter/wasm_interp_classic.c
+++ b/core/iwasm/interpreter/wasm_interp_classic.c
@ -905,8 +905,9 @@ wasm_interp_call_func_native(WASMModuleInstance *module_inst,
    if (!func_import->call_conv_wasm_c_api) {
        native_func_pointer = module_inst->import_func_ptrs[cur_func_index];
    }
-    else if (module_inst->e->c_api_func_imports) {
-        c_api_func_import = module_inst->e->c_api_func_imports + cur_func_index;
+    else if (module_inst->e->common.c_api_func_imports) {
+        c_api_func_import =
+            module_inst->e->common.c_api_func_imports + cur_func_index;
        native_func_pointer = c_api_func_import->func_ptr_linked;
    }

--- a/core/iwasm/interpreter/wasm_interp_fast.c
+++ b/core/iwasm/interpreter/wasm_interp_fast.c
@ -938,8 +938,9 @@ wasm_interp_call_func_native(WASMModuleInstance *module_inst,
    if (!func_import->call_conv_wasm_c_api) {
        native_func_pointer = module_inst->import_func_ptrs[cur_func_index];
    }
-    else if (module_inst->e->c_api_func_imports) {
-        c_api_func_import = module_inst->e->c_api_func_imports + cur_func_index;
+    else if (module_inst->e->common.c_api_func_imports) {
+        c_api_func_import =
+            module_inst->e->common.c_api_func_imports + cur_func_index;
        native_func_pointer = c_api_func_import->func_ptr_linked;
    }

--- a/core/iwasm/interpreter/wasm_loader.c
+++ b/core/iwasm/interpreter/wasm_loader.c
@ -8064,6 +8064,9 @@ re_scan:
            case WASM_OP_SELECT_T:
            {
                uint8 vec_len, ref_type;
+#if WASM_ENABLE_FAST_INTERP != 0
+                uint8 *p_code_compiled_tmp = loader_ctx->p_code_compiled;
+#endif

                read_leb_uint32(p, p_end, vec_len);
                if (vec_len != 1) {
@ -8086,8 +8089,6 @@ re_scan:
 #if WASM_ENABLE_FAST_INTERP != 0
                if (loader_ctx->p_code_compiled) {
                    uint8 opcode_tmp = WASM_OP_SELECT;
-                    uint8 *p_code_compiled_tmp =
-                        loader_ctx->p_code_compiled - 2;

                    if (ref_type == VALUE_TYPE_V128) {
 #if (WASM_ENABLE_SIMD == 0) \
--- a/core/iwasm/interpreter/wasm_mini_loader.c
+++ b/core/iwasm/interpreter/wasm_mini_loader.c
@ -6233,6 +6233,9 @@ re_scan:
            case WASM_OP_SELECT_T:
            {
                uint8 vec_len, ref_type;
+#if WASM_ENABLE_FAST_INTERP != 0
+                uint8 *p_code_compiled_tmp = loader_ctx->p_code_compiled;
+#endif

                read_leb_uint32(p, p_end, vec_len);
                if (vec_len != 1) {
@ -6255,8 +6258,6 @@ re_scan:
 #if WASM_ENABLE_FAST_INTERP != 0
                if (loader_ctx->p_code_compiled) {
                    uint8 opcode_tmp = WASM_OP_SELECT;
-                    uint8 *p_code_compiled_tmp =
-                        loader_ctx->p_code_compiled - 2;

                    if (ref_type == VALUE_TYPE_F64
                        || ref_type == VALUE_TYPE_I64)
--- a/core/iwasm/interpreter/wasm_runtime.c
+++ b/core/iwasm/interpreter/wasm_runtime.c
@ -202,7 +202,7 @@ memory_instantiate(WASMModuleInstance *module_inst, WASMModuleInstance *parent,
        if (num_bytes_per_page < heap_size) {
            set_error_buf(error_buf, error_buf_size,
                          "failed to insert app heap into linear memory, "
-                          "try using `--heap_size=0` option");
+                          "try using `--heap-size=0` option");
            return NULL;
        }
    }
@ -261,7 +261,7 @@ memory_instantiate(WASMModuleInstance *module_inst, WASMModuleInstance *parent,
        if (init_page_count > DEFAULT_MAX_PAGES) {
            set_error_buf(error_buf, error_buf_size,
                          "failed to insert app heap into linear memory, "
-                          "try using `--heap_size=0` option");
+                          "try using `--heap-size=0` option");
            return NULL;
        }
        else if (init_page_count == DEFAULT_MAX_PAGES) {
@ -1840,7 +1840,7 @@ wasm_instantiate(WASMModule *module, WASMModuleInstance *parent,
    for (i = 0; i < module->data_seg_count; i++) {
        WASMMemoryInstance *memory = NULL;
        uint8 *memory_data = NULL;
-        uint32 memory_size = 0;
+        uint64 memory_size = 0;
        WASMDataSeg *data_seg = module->data_segments[i];

 #if WASM_ENABLE_BULK_MEMORY != 0
@ -1853,7 +1853,8 @@ wasm_instantiate(WASMModule *module, WASMModuleInstance *parent,
        bh_assert(memory);

        memory_data = memory->memory_data;
-        memory_size = memory->num_bytes_per_page * memory->cur_page_count;
+        memory_size =
+            (uint64)memory->num_bytes_per_page * memory->cur_page_count;
        bh_assert(memory_data || memory_size == 0);

        bh_assert(data_seg->base_offset.init_expr_type
@ -1899,7 +1900,7 @@ wasm_instantiate(WASMModule *module, WASMModuleInstance *parent,

        /* check offset + length(could be zero) */
        length = data_seg->data_length;
-        if (base_offset + length > memory_size) {
+        if ((uint64)base_offset + length > memory_size) {
            LOG_DEBUG("base_offset(%d) + length(%d) > memory_size(%d)",
                      base_offset, length, memory_size);
 #if WASM_ENABLE_REF_TYPES != 0
@ -1913,8 +1914,9 @@ wasm_instantiate(WASMModule *module, WASMModuleInstance *parent,
        }

        if (memory_data) {
-            bh_memcpy_s(memory_data + base_offset, memory_size - base_offset,
-                        data_seg->data, length);
+            bh_memcpy_s(memory_data + base_offset,
+                        (uint32)memory_size - base_offset, data_seg->data,
+                        length);
        }
    }

@ -2228,8 +2230,8 @@ wasm_deinstantiate(WASMModuleInstance *module_inst, bool is_sub_inst)
    }
 #endif

-    if (module_inst->e->c_api_func_imports)
-        wasm_runtime_free(module_inst->e->c_api_func_imports);
+    if (module_inst->e->common.c_api_func_imports)
+        wasm_runtime_free(module_inst->e->common.c_api_func_imports);

    if (!is_sub_inst) {
 #if WASM_ENABLE_LIBC_WASI != 0
@ -3094,11 +3096,7 @@ llvm_jit_call_indirect(WASMExecEnv *exec_env, uint32 tbl_idx, uint32 elem_idx,
 {
    bool ret;

-#if WASM_ENABLE_JIT != 0
-    if (Wasm_Module_AoT == exec_env->module_inst->module_type) {
-        return aot_call_indirect(exec_env, tbl_idx, elem_idx, argc, argv);
-    }
-#endif
+    bh_assert(exec_env->module_inst->module_type == Wasm_Module_Bytecode);

    ret = call_indirect(exec_env, tbl_idx, elem_idx, argc, argv, false, 0);
 #ifdef OS_ENABLE_HW_BOUND_CHECK
@ -3125,11 +3123,7 @@ llvm_jit_invoke_native(WASMExecEnv *exec_env, uint32 func_idx, uint32 argc,
    char buf[96];
    bool ret = false;

-#if WASM_ENABLE_JIT != 0
-    if (Wasm_Module_AoT == exec_env->module_inst->module_type) {
-        return aot_invoke_native(exec_env, func_idx, argc, argv);
-    }
-#endif
+    bh_assert(exec_env->module_inst->module_type == Wasm_Module_Bytecode);

    module_inst = (WASMModuleInstance *)wasm_runtime_get_module_inst(exec_env);
    module = module_inst->module;
@ -3142,8 +3136,9 @@ llvm_jit_invoke_native(WASMExecEnv *exec_env, uint32 func_idx, uint32 argc,

    import_func = &module->import_functions[func_idx].u.function;
    if (import_func->call_conv_wasm_c_api) {
-        if (module_inst->e->c_api_func_imports) {
-            c_api_func_import = module_inst->e->c_api_func_imports + func_idx;
+        if (module_inst->e->common.c_api_func_imports) {
+            c_api_func_import =
+                module_inst->e->common.c_api_func_imports + func_idx;
            func_ptr = c_api_func_import->func_ptr_linked;
        }
        else {
@ -3198,11 +3193,7 @@ llvm_jit_memory_init(WASMModuleInstance *module_inst, uint32 seg_index,
    uint8 *maddr;
    uint64 seg_len = 0;

-#if WASM_ENABLE_JIT != 0
-    if (Wasm_Module_AoT == module_inst->module_type) {
-        return aot_memory_init(module_inst, seg_index, offset, len, dst);
-    }
-#endif
+    bh_assert(module_inst->module_type == Wasm_Module_Bytecode);

    memory_inst = wasm_get_default_memory(module_inst);
    module = module_inst->module;
@ -3228,11 +3219,7 @@ llvm_jit_memory_init(WASMModuleInstance *module_inst, uint32 seg_index,
 bool
 llvm_jit_data_drop(WASMModuleInstance *module_inst, uint32 seg_index)
 {
-#if WASM_ENABLE_JIT != 0
-    if (Wasm_Module_AoT == module_inst->module_type) {
-        return aot_data_drop(module_inst, seg_index);
-    }
-#endif
+    bh_assert(module_inst->module_type == Wasm_Module_Bytecode);

    module_inst->module->data_segments[seg_index]->data_length = 0;
    /* Currently we can't free the dropped data segment
@ -3247,11 +3234,7 @@ llvm_jit_drop_table_seg(WASMModuleInstance *module_inst, uint32 tbl_seg_idx)
 {
    WASMTableSeg *tbl_segs;

-#if WASM_ENABLE_JIT != 0
-    if (Wasm_Module_AoT == module_inst->module_type) {
-        return aot_drop_table_seg(module_inst, tbl_seg_idx);
-    }
-#endif
+    bh_assert(module_inst->module_type == Wasm_Module_Bytecode);

    tbl_segs = module_inst->module->table_segments;
    tbl_segs[tbl_seg_idx].is_dropped = true;
@ -3265,12 +3248,7 @@ llvm_jit_table_init(WASMModuleInstance *module_inst, uint32 tbl_idx,
    WASMTableInstance *tbl_inst;
    WASMTableSeg *tbl_seg;

-#if WASM_ENABLE_JIT != 0
-    if (Wasm_Module_AoT == module_inst->module_type) {
-        return aot_table_init(module_inst, tbl_idx, tbl_seg_idx, length,
-                              src_offset, dst_offset);
-    }
-#endif
+    bh_assert(module_inst->module_type == Wasm_Module_Bytecode);

    tbl_inst = wasm_get_table_inst(module_inst, tbl_idx);
    tbl_seg = module_inst->module->table_segments + tbl_seg_idx;
@ -3313,13 +3291,7 @@ llvm_jit_table_copy(WASMModuleInstance *module_inst, uint32 src_tbl_idx,
    WASMTableInstance *src_tbl_inst;
    WASMTableInstance *dst_tbl_inst;

-#if WASM_ENABLE_JIT != 0
-    if (Wasm_Module_AoT == module_inst->module_type) {
-        aot_table_copy(module_inst, src_tbl_idx, dst_tbl_idx, length,
-                       src_offset, dst_offset);
-        return;
-    }
-#endif
+    bh_assert(module_inst->module_type == Wasm_Module_Bytecode);

    src_tbl_inst = wasm_get_table_inst(module_inst, src_tbl_idx);
    dst_tbl_inst = wasm_get_table_inst(module_inst, dst_tbl_idx);
@ -3350,12 +3322,7 @@ llvm_jit_table_fill(WASMModuleInstance *module_inst, uint32 tbl_idx,
 {
    WASMTableInstance *tbl_inst;

-#if WASM_ENABLE_JIT != 0
-    if (Wasm_Module_AoT == module_inst->module_type) {
-        aot_table_fill(module_inst, tbl_idx, length, val, data_offset);
-        return;
-    }
-#endif
+    bh_assert(module_inst->module_type == Wasm_Module_Bytecode);

    tbl_inst = wasm_get_table_inst(module_inst, tbl_idx);
    bh_assert(tbl_inst);
@ -3377,11 +3344,7 @@ llvm_jit_table_grow(WASMModuleInstance *module_inst, uint32 tbl_idx,
    WASMTableInstance *tbl_inst;
    uint32 i, orig_size, total_size;

-#if WASM_ENABLE_JIT != 0
-    if (Wasm_Module_AoT == module_inst->module_type) {
-        return aot_table_grow(module_inst, tbl_idx, inc_size, init_val);
-    }
-#endif
+    bh_assert(module_inst->module_type == Wasm_Module_Bytecode);

    tbl_inst = wasm_get_table_inst(module_inst, tbl_idx);
    if (!tbl_inst) {
@ -3421,11 +3384,7 @@ llvm_jit_alloc_frame(WASMExecEnv *exec_env, uint32 func_index)
    WASMInterpFrame *frame;
    uint32 size;

-#if WASM_ENABLE_JIT != 0
-    if (Wasm_Module_AoT == exec_env->module_inst->module_type) {
-        return aot_alloc_frame(exec_env, func_index);
-    }
-#endif
+    bh_assert(exec_env->module_inst->module_type == Wasm_Module_Bytecode);

    module_inst = (WASMModuleInstance *)exec_env->module_inst;
    size = wasm_interp_interp_frame_size(0);
@ -3454,12 +3413,7 @@ llvm_jit_free_frame(WASMExecEnv *exec_env)
    WASMInterpFrame *frame;
    WASMInterpFrame *prev_frame;

-#if WASM_ENABLE_JIT != 0
-    if (Wasm_Module_AoT == exec_env->module_inst->module_type) {
-        aot_free_frame(exec_env);
-        return;
-    }
-#endif
+    bh_assert(exec_env->module_inst->module_type == Wasm_Module_Bytecode);

    frame = wasm_exec_env_get_cur_frame(exec_env);
    prev_frame = frame->prev_frame;
--- a/core/iwasm/interpreter/wasm_runtime.h
+++ b/core/iwasm/interpreter/wasm_runtime.h
@ -210,8 +210,19 @@ typedef struct CApiFuncImport {
    void *env_arg;
 } CApiFuncImport;

+/* The common part of WASMModuleInstanceExtra and AOTModuleInstanceExtra */
+typedef struct WASMModuleInstanceExtraCommon {
+    CApiFuncImport *c_api_func_imports;
+#if WASM_CONFIGUABLE_BOUNDS_CHECKS != 0
+    /* Disable bounds checks or not */
+    bool disable_bounds_checks;
+#endif
+} WASMModuleInstanceExtraCommon;
+
 /* Extra info of WASM module instance for interpreter/jit mode */
 typedef struct WASMModuleInstanceExtra {
+    WASMModuleInstanceExtraCommon common;
+
    WASMGlobalInstance *globals;
    WASMFunctionInstance *functions;

@ -223,7 +234,6 @@ typedef struct WASMModuleInstanceExtra {
    WASMFunctionInstance *free_function;
    WASMFunctionInstance *retain_function;

-    CApiFuncImport *c_api_func_imports;
    RunningMode running_mode;

 #if WASM_ENABLE_MULTI_MODULE != 0
@ -242,10 +252,6 @@ typedef struct WASMModuleInstanceExtra {
        && WASM_ENABLE_LAZY_JIT != 0)
    WASMModuleInstance *next;
 #endif
-#if WASM_CONFIGUABLE_BOUNDS_CHECKS != 0
-    /* Disable bounds checks or not */
-    bool disable_bounds_checks;
-#endif
 } WASMModuleInstanceExtra;

 struct AOTFuncPerfProfInfo;
--- a/core/iwasm/libraries/lib-rats/lib_rats.cmake
+++ b/core/iwasm/libraries/lib-rats/lib_rats.cmake
@ -23,6 +23,7 @@ include(FetchContent)
 set(RATS_BUILD_MODE "sgx"
    CACHE INTERNAL "Select build mode for librats(host|occlum|sgx｜wasm)")
 set(RATS_INSTALL_PATH  "${CMAKE_BINARY_DIR}/librats" CACHE INTERNAL "")
+set(BUILD_SAMPLES OFF)

 FetchContent_Declare(
    librats
@ -34,8 +35,17 @@ if (NOT librats_POPULATED)
    message("-- Fetching librats ..")
    FetchContent_Populate(librats)
    include_directories("${librats_SOURCE_DIR}/include")
+    
+    # Prevent the propagation of the CMAKE_C_FLAGS of WAMR into librats
+    set(SAVED_CMAKE_C_FLAGS ${CMAKE_C_FLAGS})
+    set(CMAKE_C_FLAGS "")
+
+    # Import the building scripts of librats
    add_subdirectory(${librats_SOURCE_DIR} ${librats_BINARY_DIR} EXCLUDE_FROM_ALL)

+    # Restore the CMAKE_C_FLAGS of WAMR
+    set(CMAKE_C_FLAGS ${SAVED_CMAKE_C_FLAGS})
+
 endif()

 file (GLOB source_all ${LIB_RATS_DIR}/*.c)
--- a/core/iwasm/libraries/lib-wasi-threads/test/build.sh
+++ b/core/iwasm/libraries/lib-wasi-threads/test/build.sh
@ -9,8 +9,32 @@ set -eo pipefail
 CC=${CC:=/opt/wasi-sdk/bin/clang}
 WAMR_DIR=../../../../..

+show_usage() {
+    echo "Usage: $0 [--sysroot PATH_TO_SYSROOT]"
+    echo "--sysroot PATH_TO_SYSROOT specify to build with custom sysroot for wasi-libc"
+}
+
+while [[ $# -gt 0 ]]; do
+    key="$1"
+    case $key in
+        --sysroot)
+            sysroot_path="$2"
+            shift
+            shift
+            ;;
+        --help)
+            show_usage
+            exit
+            ;;
+        *)
+            echo "Unknown option: $1"
+            exit 1
+            ;;
+    esac
+done
+
 # Stress tests names
-thread_start_file_exclusions=("spawn_stress_test.wasm" "linear_memory_size_update.wasm")
+thread_start_file_exclusions=("spawn_stress_test.wasm" "linear_memory_size_update.wasm" "stress_test_threads_creation.wasm")

 for test_c in *.c; do
    test_wasm="$(basename $test_c .c).wasm"
@ -21,9 +45,18 @@ for test_c in *.c; do
        thread_start_file=$WAMR_DIR/samples/wasi-threads/wasm-apps/wasi_thread_start.S
    fi

+    if [[ -n "$sysroot_path" ]]; then 
+        if [ ! -d "$sysroot_path" ]; then 
+            echo "Directory $sysroot_path  doesn't exist. Aborting"
+            exit 1
+        fi
+        sysroot_command="--sysroot $sysroot_path"
+    fi
+    
    echo "Compiling $test_c to $test_wasm"
    $CC \
        -target wasm32-wasi-threads \
+        -O2 \
        -pthread -ftls-model=local-exec \
        -z stack-size=32768 \
        -Wl,--export=__heap_base \
@ -33,6 +66,7 @@ for test_c in *.c; do
        -Wl,--export=malloc \
        -Wl,--export=free \
        -I $WAMR_DIR/samples/wasi-threads/wasm-apps \
+        $sysroot_command \
        $thread_start_file \
        $test_c -o $test_wasm
 done
--- a/core/iwasm/libraries/lib-wasi-threads/test/skip.json
+++ b/core/iwasm/libraries/lib-wasi-threads/test/skip.json
@ -1,5 +1,6 @@
 {
    "lib-wasi-threads tests": {
-        "spawn_stress_test": "Stress tests are incompatible with the other part and executed differently"
+        "spawn_stress_test": "Stress tests are incompatible with the other part and executed differently",
+        "stress_test_threads_creation": "Stress tests are incompatible with the other part and executed differently"
    }
 }
--- a/core/iwasm/libraries/lib-wasi-threads/test/spawn_stress_test.c
+++ b/core/iwasm/libraries/lib-wasi-threads/test/spawn_stress_test.c
@ -18,8 +18,9 @@

 enum CONSTANTS {
    NUM_ITER = 100000,
-    NUM_RETRY = 5,
+    NUM_RETRY = 8,
    MAX_NUM_THREADS = 8,
+    RETRY_SLEEP_TIME_US = 2000,
 };

 unsigned prime_numbers_count = 0;
@ -62,11 +63,13 @@ void
 spawn_thread(pthread_t *thread, unsigned int *arg)
 {
    int status_code = -1;
+    int timeout_us = RETRY_SLEEP_TIME_US;
    for (int tries = 0; status_code != 0 && tries < NUM_RETRY; ++tries) {
        status_code = pthread_create(thread, NULL, &check_if_prime, arg);
        assert(status_code == 0 || status_code == EAGAIN);
        if (status_code == EAGAIN) {
-            usleep(2000);
+            usleep(timeout_us);
+            timeout_us *= 2;
        }
    }

@ -95,7 +98,7 @@ main(int argc, char **argv)

        args[thread_num] = factorised_number;

-        usleep(2000);
+        usleep(RETRY_SLEEP_TIME_US);
        spawn_thread(&threads[thread_num], &args[thread_num]);
        assert(threads[thread_num] != 0);
    }
--- a/core/iwasm/libraries/lib-wasi-threads/test/stress_test_threads_creation.c
+++ b/core/iwasm/libraries/lib-wasi-threads/test/stress_test_threads_creation.c
@ -0,0 +1,93 @@
+/*
+ * Copyright (C) 2023 Amazon.com Inc. or its affiliates. All rights reserved.
+ * SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+ */
+
+#include <assert.h>
+#include <errno.h>
+#include <pthread.h>
+#include <stdio.h>
+#include <unistd.h>
+
+enum CONSTANTS {
+    NUM_ITER = 200000,
+    NUM_RETRY = 8,
+    MAX_NUM_THREADS = 8,
+    RETRY_SLEEP_TIME_US = 4000,
+    SECOND = 1000 * 1000 * 1000
+};
+
+int threads_executed = 0;
+unsigned int threads_creation_tried = 0;
+unsigned int threads_in_use = 0;
+
+void *
+thread_func(void *arg)
+{
+    (void)(arg);
+    __atomic_fetch_add(&threads_executed, 1, __ATOMIC_RELAXED);
+    __atomic_fetch_sub(&threads_in_use, 1, __ATOMIC_SEQ_CST);
+    return NULL;
+}
+
+void
+spawn_thread(pthread_t *thread)
+{
+    int status_code = -1;
+    int timeout_us = RETRY_SLEEP_TIME_US;
+    for (int tries = 0; status_code != 0 && tries < NUM_RETRY; ++tries) {
+        status_code = pthread_create(thread, NULL, &thread_func, NULL);
+        __atomic_fetch_add(&threads_creation_tried, 1, __ATOMIC_RELAXED);
+
+        assert(status_code == 0 || status_code == EAGAIN);
+        if (status_code == EAGAIN) {
+            usleep(timeout_us);
+            timeout_us *= 2;
+        }
+    }
+
+    assert(status_code == 0 && "Thread creation should succeed");
+}
+
+int
+main(int argc, char **argv)
+{
+    double percentage = 0.1;
+
+    for (int iter = 0; iter < NUM_ITER; ++iter) {
+        if (iter > NUM_ITER * percentage) {
+            fprintf(stderr, "Spawning stress test is %d%% finished\n",
+                    (unsigned int)(percentage * 100));
+            percentage += 0.1;
+        }
+        while (__atomic_load_n(&threads_in_use, __ATOMIC_SEQ_CST)
+               == MAX_NUM_THREADS) {
+            usleep(100);
+        }
+
+        __atomic_fetch_add(&threads_in_use, 1, __ATOMIC_SEQ_CST);
+        pthread_t tmp;
+        spawn_thread(&tmp);
+        pthread_detach(tmp);
+    }
+
+    while ((__atomic_load_n(&threads_in_use, __ATOMIC_SEQ_CST) != 0)) {
+        __builtin_wasm_memory_atomic_wait32(&threads_in_use, 0, SECOND);
+    }
+
+    assert(__atomic_load_n(&threads_in_use, __ATOMIC_SEQ_CST) == 0);
+
+    // Validation
+    assert(threads_creation_tried >= threads_executed
+           && "Test executed more threads than were created");
+    assert((1. * threads_creation_tried) / threads_executed < 2.5
+           && "Ensuring that we're retrying thread creation less than 2.5 "
+              "times on average ");
+
+    fprintf(stderr,
+            "Spawning stress test finished successfully executed %d threads "
+            "with retry ratio %f\n",
+            threads_creation_tried,
+            (1. * threads_creation_tried) / threads_executed);
+    return 0;
+}
--- a/core/iwasm/libraries/lib-wasi-threads/tid_allocator.c
+++ b/core/iwasm/libraries/lib-wasi-threads/tid_allocator.c
@ -21,7 +21,8 @@ tid_allocator_init(TidAllocator *tid_allocator)
        return false;

    for (int64 i = tid_allocator->pos - 1; i >= 0; i--)
-        tid_allocator->ids[i] = TID_MIN + (tid_allocator->pos - 1 - i);
+        tid_allocator->ids[i] =
+            (uint32)(TID_MIN + (tid_allocator->pos - 1 - i));

    return true;
 }
@ -54,7 +55,8 @@ tid_allocator_get_tid(TidAllocator *tid_allocator)
            LOG_ERROR("Overflow detected during realloc");
            return -1;
        }
-        int32 *tmp = wasm_runtime_realloc(tid_allocator->ids, realloc_size);
+        int32 *tmp =
+            wasm_runtime_realloc(tid_allocator->ids, (uint32)realloc_size);
        if (tmp == NULL) {
            LOG_ERROR("Thread ID allocator realloc failed");
            return -1;
@ -64,7 +66,8 @@ tid_allocator_get_tid(TidAllocator *tid_allocator)
        tid_allocator->pos = new_size - old_size;
        tid_allocator->ids = tmp;
        for (int64 i = tid_allocator->pos - 1; i >= 0; i--)
-            tid_allocator->ids[i] = TID_MIN + (tid_allocator->size - 1 - i);
+            tid_allocator->ids[i] =
+                (uint32)(TID_MIN + (tid_allocator->size - 1 - i));
    }

    // Pop available thread identifier from the stack
@ -77,4 +80,4 @@ tid_allocator_release_tid(TidAllocator *tid_allocator, int32 thread_id)
    // Release thread identifier by pushing it into the stack
    bh_assert(tid_allocator->pos < tid_allocator->size);
    tid_allocator->ids[tid_allocator->pos++] = thread_id;
-}
+}
--- a/core/iwasm/libraries/thread-mgr/thread_manager.c
+++ b/core/iwasm/libraries/thread-mgr/thread_manager.c
@ -746,10 +746,10 @@ wasm_cluster_dup_c_api_imports(WASMModuleInstanceCommon *module_inst_dst,

 #if WASM_ENABLE_INTERP != 0
    if (module_inst_src->module_type == Wasm_Module_Bytecode) {
-        new_c_api_func_imports =
-            &(((WASMModuleInstance *)module_inst_dst)->e->c_api_func_imports);
+        new_c_api_func_imports = &(((WASMModuleInstance *)module_inst_dst)
+                                       ->e->common.c_api_func_imports);
        c_api_func_imports = ((const WASMModuleInstance *)module_inst_src)
-                                 ->e->c_api_func_imports;
+                                 ->e->common.c_api_func_imports;
        import_func_count =
            ((WASMModule *)(((const WASMModuleInstance *)module_inst_src)
                                ->module))
@ -760,10 +760,10 @@ wasm_cluster_dup_c_api_imports(WASMModuleInstanceCommon *module_inst_dst,
    if (module_inst_src->module_type == Wasm_Module_AoT) {
        AOTModuleInstanceExtra *e =
            (AOTModuleInstanceExtra *)((AOTModuleInstance *)module_inst_dst)->e;
-        new_c_api_func_imports = &(e->c_api_func_imports);
+        new_c_api_func_imports = &(e->common.c_api_func_imports);

        e = (AOTModuleInstanceExtra *)((AOTModuleInstance *)module_inst_src)->e;
-        c_api_func_imports = e->c_api_func_imports;
+        c_api_func_imports = e->common.c_api_func_imports;

        import_func_count =
            ((AOTModule *)(((AOTModuleInstance *)module_inst_src)->module))
--- a/core/iwasm/libraries/wasi-nn/.gitignore
+++ b/core/iwasm/libraries/wasi-nn/.gitignore
@ -0,0 +1,2 @@
+**/*.wasm
+**/*.tflite
--- a/core/iwasm/libraries/wasi-nn/README.md
+++ b/core/iwasm/libraries/wasi-nn/README.md
@ -25,6 +25,7 @@ Build the runtime image for your execution target type.
 * `cpu`
 * `nvidia-gpu`
 * `vx-delegate`
+* `tpu`

 ```
 EXECUTION_TYPE=cpu
@ -64,6 +65,8 @@ docker run \
 ```

 * (NVIDIA) GPU
+    * Requirements:
+        * [NVIDIA docker](https://github.com/NVIDIA/nvidia-docker).

 ```
 docker run \
@ -76,25 +79,36 @@ docker run \
    /assets/test_tensorflow.wasm
 ```

-* vx-delegate for NPU (x86 simulater)
+* vx-delegate for NPU (x86 simulator)

 ```
 docker run \
-    -v $PWD/core/iwasm/libraries/wasi-nn/test:/assets wasi-nn-vx-delegate \
-    --dir=/assets \
+    -v $PWD/core/iwasm/libraries/wasi-nn/test:/assets \
+    wasi-nn-vx-delegate \
+    --dir=/ \
    --env="TARGET=gpu" \
-    /assets/test_tensorflow.wasm
+    /assets/test_tensorflow_quantized.wasm
 ```

+* (Coral) TPU
+    * Requirements:
+        * [Coral USB](https://coral.ai/products/accelerator/).

-
-Requirements:
-* [NVIDIA docker](https://github.com/NVIDIA/nvidia-docker).
+```
+docker run \
+    --privileged \
+    --device=/dev/bus/usb:/dev/bus/usb \
+    -v $PWD/core/iwasm/libraries/wasi-nn/test:/assets \
+    wasi-nn-tpu \
+    --dir=/ \
+    --env="TARGET=tpu" \
+    /assets/test_tensorflow_quantized.wasm
+```

 ## What is missing

 Supported:

 * Graph encoding: `tensorflowlite`.
-* Execution target: `cpu` and `gpu`.
+* Execution target: `cpu`, `gpu` and `tpu`.
 * Tensor type: `fp32`.
--- a/core/iwasm/libraries/wasi-nn/cmake/Findtensorflow_lite.cmake
+++ b/core/iwasm/libraries/wasi-nn/cmake/Findtensorflow_lite.cmake
@ -18,12 +18,16 @@ if(NOT EXISTS ${TENSORFLOW_LITE})

  set(TENSORFLOW_SOURCE_DIR "${WAMR_ROOT_DIR}/core/deps/tensorflow-src")

-  if(WASI_NN_ENABLE_GPU EQUAL 1)
+  if(WAMR_BUILD_WASI_NN_ENABLE_GPU EQUAL 1)
    # Tensorflow specific:
    # * https://www.tensorflow.org/lite/guide/build_cmake#available_options_to_build_tensorflow_lite
    set (TFLITE_ENABLE_GPU ON)
  endif()

+  if (CMAKE_SIZEOF_VOID_P EQUAL 4)
+    set (TFLITE_ENABLE_XNNPACK OFF)
+  endif()
+
  add_subdirectory(
    "${TENSORFLOW_SOURCE_DIR}/tensorflow/lite"
    "${CMAKE_CURRENT_BINARY_DIR}/tensorflow-lite"
--- a/core/iwasm/libraries/wasi-nn/src/wasi_nn_tensorflowlite.cpp
+++ b/core/iwasm/libraries/wasi-nn/src/wasi_nn_tensorflowlite.cpp
@ -16,11 +16,11 @@
 #include <tensorflow/lite/optional_debug_tools.h>
 #include <tensorflow/lite/error_reporter.h>

-#if defined(WASI_NN_ENABLE_GPU)
+#if WASM_ENABLE_WASI_NN_GPU != 0
 #include <tensorflow/lite/delegates/gpu/delegate.h>
 #endif

-#if defined(WASI_NN_ENABLE_EXTERNAL_DELEGATE)
+#if WASM_ENABLE_WASI_NN_EXTERNAL_DELEGATE != 0
 #include <tensorflow/lite/delegates/external/external_delegate.h>
 #endif

@ -130,8 +130,8 @@ tensorflowlite_load(void *tflite_ctx, graph_builder_array *builder,
        return invalid_argument;
    }

-    if (target != cpu && target != gpu) {
-        NN_ERR_PRINTF("Only CPU and GPU target is supported.");
+    if (target != cpu && target != gpu && target != tpu) {
+        NN_ERR_PRINTF("Only CPU, GPU and TPU target is supported.");
        return invalid_argument;
    }

@ -195,7 +195,7 @@ tensorflowlite_init_execution_context(void *tflite_ctx, graph g,
    switch (tfl_ctx->models[g].target) {
        case gpu:
        {
-#if defined(WASI_NN_ENABLE_GPU)
+#if WASM_ENABLE_WASI_NN_GPU != 0
            NN_WARN_PRINTF("GPU enabled.");
            // https://www.tensorflow.org/lite/performance/gpu
            TfLiteGpuDelegateOptionsV2 options =
@ -216,10 +216,19 @@ tensorflowlite_init_execution_context(void *tflite_ctx, graph g,
                NN_ERR_PRINTF("Error when enabling GPU delegate.");
                use_default = true;
            }
-#elif defined(WASI_NN_ENABLE_EXTERNAL_DELEGATE)
+#else
+            NN_WARN_PRINTF("GPU not enabled.");
+            use_default = true;
+#endif
+            break;
+        }
+        case tpu:
+        {
+#if WASM_ENABLE_WASI_NN_EXTERNAL_DELEGATE != 0
            NN_WARN_PRINTF("external delegation enabled.");
            TfLiteExternalDelegateOptions options =
-                TfLiteExternalDelegateOptionsDefault(WASI_NN_EXT_DELEGATE_PATH);
+                TfLiteExternalDelegateOptionsDefault(
+                    WASM_WASI_NN_EXTERNAL_DELEGATE_PATH);
            tfl_ctx->delegate = TfLiteExternalDelegateCreate(&options);
            if (tfl_ctx->delegate == NULL) {
                NN_ERR_PRINTF("Error when generating External delegate.");
@ -233,7 +242,7 @@ tensorflowlite_init_execution_context(void *tflite_ctx, graph g,
                use_default = true;
            }
 #else
-            NN_WARN_PRINTF("GPU not enabled.");
+            NN_WARN_PRINTF("External delegate not enabled.");
            use_default = true;
 #endif
            break;
@ -285,14 +294,37 @@ tensorflowlite_set_input(void *tflite_ctx, graph_execution_context ctx,
        return invalid_argument;
    }

-    auto *input =
-        tfl_ctx->interpreters[ctx].interpreter->typed_input_tensor<float>(
-            index);
-    if (input == NULL)
-        return missing_memory;
+    if (tensor->quantization.type == kTfLiteNoQuantization) {
+        NN_DBG_PRINTF("No quantization information. Using float as default");
+        float *it =
+            tfl_ctx->interpreters[ctx].interpreter->typed_input_tensor<float>(
+                index);
+
+        int size = model_tensor_size * sizeof(float);
+        bh_memcpy_s(it, size, input_tensor->data, size);
+    }
+    else { // TODO: Assumming uint8 quantized networks.
+        TfLiteAffineQuantization *quant_info =
+            (TfLiteAffineQuantization *)tensor->quantization.params;
+        if (quant_info->scale->size != 1 || quant_info->zero_point->size != 1) {
+            NN_ERR_PRINTF("Quantization per channel is not supported");
+            return runtime_error;
+        }
+        uint8_t *it =
+            tfl_ctx->interpreters[ctx].interpreter->typed_input_tensor<uint8_t>(
+                index);
+
+        float scale = quant_info->scale->data[0];
+        float zero_point = (float)quant_info->zero_point->data[0];
+        NN_DBG_PRINTF("input tensor: (scale, offset) = (%f, %f)", scale,
+                      zero_point);
+
+        float *input_tensor_f = (float *)input_tensor->data;
+        for (uint32_t i = 0; i < model_tensor_size; ++i) {
+            it[i] = (uint8_t)(input_tensor_f[i] / scale + zero_point);
+        }
+    }

-    bh_memcpy_s(input, model_tensor_size * sizeof(float), input_tensor->data,
-                model_tensor_size * sizeof(float));
    return success;
 }

@ -325,6 +357,7 @@ tensorflowlite_get_output(void *tflite_ctx, graph_execution_context ctx,
    NN_DBG_PRINTF("Number of tensors (%d)", num_output_tensors);

    if (index + 1 > num_output_tensors) {
+        NN_ERR_PRINTF("Index %d is invalid.", index);
        return runtime_error;
    }

@ -343,15 +376,37 @@ tensorflowlite_get_output(void *tflite_ctx, graph_execution_context ctx,
        return missing_memory;
    }

-    float *tensor_f =
-        tfl_ctx->interpreters[ctx].interpreter->typed_output_tensor<float>(
-            index);
-    for (uint32_t i = 0; i < model_tensor_size; ++i)
-        NN_DBG_PRINTF("output: %f", tensor_f[i]);
+    if (tensor->quantization.type == kTfLiteNoQuantization) {
+        NN_DBG_PRINTF("No quantization information");
+        float *ot =
+            tfl_ctx->interpreters[ctx].interpreter->typed_output_tensor<float>(
+                index);
+
+        int size = model_tensor_size * sizeof(float);
+        bh_memcpy_s(output_tensor, size, ot, size);
+    }
+    else { // TODO: Assumming uint8 quantized networks.
+        TfLiteAffineQuantization *quant_info =
+            (TfLiteAffineQuantization *)tensor->quantization.params;
+        if (quant_info->scale->size != 1 || quant_info->zero_point->size != 1) {
+            NN_ERR_PRINTF("Quantization per channel is not supported");
+            return runtime_error;
+        }
+        uint8_t *ot = tfl_ctx->interpreters[ctx]
+                          .interpreter->typed_output_tensor<uint8_t>(index);
+
+        float scale = quant_info->scale->data[0];
+        float zero_point = (float)quant_info->zero_point->data[0];
+        NN_DBG_PRINTF("output tensor: (scale, offset) = (%f, %f)", scale,
+                      zero_point);
+
+        float *output_tensor_f = (float *)output_tensor;
+        for (uint32_t i = 0; i < model_tensor_size; ++i) {
+            output_tensor_f[i] = (ot[i] - zero_point) * scale;
+        }
+    }

    *output_tensor_size = model_tensor_size;
-    bh_memcpy_s(output_tensor, model_tensor_size * sizeof(float), tensor_f,
-                model_tensor_size * sizeof(float));
    return success;
 }

@ -392,19 +447,35 @@ tensorflowlite_destroy(void *tflite_ctx)
    */
    TFLiteContext *tfl_ctx = (TFLiteContext *)tflite_ctx;

-    if (tfl_ctx->delegate != NULL) {
-#if defined(WASI_NN_ENABLE_GPU)
-        TfLiteGpuDelegateV2Delete(tfl_ctx->delegate);
-#elif defined(WASI_NN_ENABLE_EXTERNAL_DELEGATE)
-        TfLiteExternalDelegateDelete(tfl_ctx->delegate);
-#endif
-    }
-
    NN_DBG_PRINTF("Freeing memory.");
    for (int i = 0; i < MAX_GRAPHS_PER_INST; ++i) {
        tfl_ctx->models[i].model.reset();
-        if (tfl_ctx->models[i].model_pointer)
+        if (tfl_ctx->models[i].model_pointer) {
+            if (tfl_ctx->delegate) {
+                switch (tfl_ctx->models[i].target) {
+                    case gpu:
+                    {
+#if WASM_ENABLE_WASI_NN_GPU != 0
+                        TfLiteGpuDelegateV2Delete(tfl_ctx->delegate);
+#else
+                        NN_ERR_PRINTF("GPU delegate delete but not enabled.");
+#endif
+                        break;
+                    }
+                    case tpu:
+                    {
+#if WASM_ENABLE_WASI_NN_EXTERNAL_DELEGATE != 0
+                        TfLiteExternalDelegateDelete(tfl_ctx->delegate);
+#else
+                        NN_ERR_PRINTF(
+                            "External delegate delete but not enabled.");
+#endif
+                        break;
+                    }
+                }
+            }
            wasm_runtime_free(tfl_ctx->models[i].model_pointer);
+        }
        tfl_ctx->models[i].model_pointer = NULL;
    }
    for (int i = 0; i < MAX_GRAPH_EXEC_CONTEXTS_PER_INST; ++i) {
--- a/core/iwasm/libraries/wasi-nn/test/Dockerfile.cpu
+++ b/core/iwasm/libraries/wasi-nn/test/Dockerfile.cpu
@ -30,7 +30,6 @@ RUN make -j "$(grep -c ^processor /proc/cpuinfo)"

 FROM ubuntu:22.04

-COPY --from=base /home/wamr/product-mini/platforms/linux/build/libvmlib.so /libvmlib.so
 COPY --from=base /home/wamr/product-mini/platforms/linux/build/iwasm /iwasm

 ENTRYPOINT [ "/iwasm" ]
--- a/core/iwasm/libraries/wasi-nn/test/Dockerfile.nvidia-gpu
+++ b/core/iwasm/libraries/wasi-nn/test/Dockerfile.nvidia-gpu
@ -24,7 +24,7 @@ RUN apt-get install -y wget ca-certificates --no-install-recommends \

 RUN cmake \
    -DWAMR_BUILD_WASI_NN=1 \
-    -DWASI_NN_ENABLE_GPU=1 \
+    -DWAMR_BUILD_WASI_NN_ENABLE_GPU=1 \
    ..

 RUN make -j "$(grep -c ^processor /proc/cpuinfo)"
@ -44,7 +44,6 @@ RUN mkdir -p /etc/OpenCL/vendors && \
 ENV NVIDIA_VISIBLE_DEVICES=all
 ENV NVIDIA_DRIVER_CAPABILITIES=compute,utility

-COPY --from=base /home/wamr/product-mini/platforms/linux/build/libvmlib.so /libvmlib.so
 COPY --from=base /home/wamr/product-mini/platforms/linux/build/iwasm /iwasm

 ENTRYPOINT [ "/iwasm" ]
--- a/core/iwasm/libraries/wasi-nn/test/Dockerfile.tpu
+++ b/core/iwasm/libraries/wasi-nn/test/Dockerfile.tpu
@ -0,0 +1,37 @@
+# Copyright (C) 2019 Intel Corporation.  All rights reserved.
+# SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+
+FROM ubuntu:20.04 AS base
+
+ENV DEBIAN_FRONTEND=noninteractive
+
+# hadolint ignore=DL3008
+RUN apt-get update && apt-get install -y \
+    cmake build-essential git curl gnupg --no-install-recommends && \
+    rm -rf /var/lib/apt/lists/*
+
+# hadolint ignore=DL3008,DL4006
+RUN echo "deb https://packages.cloud.google.com/apt coral-edgetpu-stable main" | tee /etc/apt/sources.list.d/coral-edgetpu.list && \
+    curl https://packages.cloud.google.com/apt/doc/apt-key.gpg | apt-key add - && \
+    apt-get update && apt-get install -y libedgetpu1-std --no-install-recommends && \
+    rm -rf /var/lib/apt/lists/*
+
+WORKDIR /home/wamr
+
+COPY . .
+
+WORKDIR /home/wamr/product-mini/platforms/linux/build
+
+RUN cmake \
+  -DWAMR_BUILD_WASI_NN=1 \
+  -DWAMR_BUILD_WASI_NN_ENABLE_EXTERNAL_DELEGATE=1 \
+  -DWAMR_BUILD_WASI_NN_EXTERNAL_DELEGATE_PATH="libedgetpu.so.1.0" \
+  -DWAMR_BUILD_WASI_NN_ENABLE_GPU=1 \
+  ..
+
+RUN make -j "$(grep -c ^processor /proc/cpuinfo)" && \
+    cp /home/wamr/product-mini/platforms/linux/build/iwasm /iwasm
+
+WORKDIR /assets
+
+ENTRYPOINT [ "/iwasm" ]
--- a/core/iwasm/libraries/wasi-nn/test/build.sh
+++ b/core/iwasm/libraries/wasi-nn/test/build.sh
@ -1,6 +1,10 @@
+#!/bin/sh
+
 # Copyright (C) 2019 Intel Corporation.  All rights reserved.
 # SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception

+CURR_PATH=$(cd $(dirname $0) && pwd -P)
+
 # WASM application that uses WASI-NN

 /opt/wasi-sdk/bin/clang \
@ -13,9 +17,25 @@

 # TFLite models to use in the tests

-cd models
+cd ${CURR_PATH}/models
 python3 average.py
 python3 max.py
 python3 mult_dimension.py
 python3 mult_outputs.py
 python3 sum.py
+
+# Specific tests for TPU
+
+cd ${CURR_PATH}
+/opt/wasi-sdk/bin/clang \
+    -Wl,--allow-undefined \
+    -Wl,--strip-all,--no-entry \
+    --sysroot=/opt/wasi-sdk/share/wasi-sysroot \
+    -I../include -I../src/utils \
+    -o test_tensorflow_quantized.wasm \
+    test_tensorflow_quantized.c utils.c
+
+cd ${CURR_PATH}/models
+python3 quantized.py
+
+cd ${CURR_PATH}
--- a/core/iwasm/libraries/wasi-nn/test/models/quantized.py
+++ b/core/iwasm/libraries/wasi-nn/test/models/quantized.py
@ -0,0 +1,30 @@
+# Copyright (C) 2019 Intel Corporation.  All rights reserved.
+# SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+
+import tensorflow as tf
+import numpy as np
+import pathlib
+
+model = tf.keras.Sequential([
+    tf.keras.layers.InputLayer(input_shape=[5, 5, 1]),
+    tf.keras.layers.AveragePooling2D(
+        pool_size=(5, 5), strides=None, padding="valid", data_format=None)
+
+])
+
+def representative_dataset():
+    for _ in range(1000):
+      data = np.random.randint(0, 25, (1, 5, 5, 1))
+      yield [data.astype(np.float32)]
+
+converter = tf.lite.TFLiteConverter.from_keras_model(model)
+converter.optimizations = [tf.lite.Optimize.DEFAULT]
+converter.representative_dataset = representative_dataset
+converter.target_spec.supported_ops = [tf.lite.OpsSet.TFLITE_BUILTINS_INT8]
+converter.inference_input_type = tf.uint8  # or tf.int8
+converter.inference_output_type = tf.uint8  # or tf.int8
+tflite_model = converter.convert()
+
+tflite_models_dir = pathlib.Path("./")
+tflite_model_file = tflite_models_dir / "quantized_model.tflite"
+tflite_model_file.write_bytes(tflite_model)
--- a/core/iwasm/libraries/wasi-nn/test/test_tensorflow_quantized.c
+++ b/core/iwasm/libraries/wasi-nn/test/test_tensorflow_quantized.c
@ -0,0 +1,63 @@
+/*
+ * Copyright (C) 2019 Intel Corporation.  All rights reserved.
+ * SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <assert.h>
+#include <string.h>
+#include <math.h>
+
+#include "utils.h"
+#include "logger.h"
+
+#undef EPSILON
+#define EPSILON 1e-2
+
+void
+test_average_quantized(execution_target target)
+{
+    int dims[] = { 1, 5, 5, 1 };
+    input_info input = create_input(dims);
+
+    uint32_t output_size = 0;
+    float *output =
+        run_inference(target, input.input_tensor, input.dim, &output_size,
+                      "./models/quantized_model.tflite", 1);
+
+    NN_INFO_PRINTF("Output size: %d", output_size);
+    NN_INFO_PRINTF("Result: average is %f", output[0]);
+    // NOTE: 11.95 instead of 12 because of errors due quantization
+    assert(fabs(output[0] - 11.95) < EPSILON);
+
+    free(input.dim);
+    free(input.input_tensor);
+    free(output);
+}
+
+int
+main()
+{
+    char *env = getenv("TARGET");
+    if (env == NULL) {
+        NN_INFO_PRINTF("Usage:\n--env=\"TARGET=[cpu|gpu|tpu]\"");
+        return 1;
+    }
+    execution_target target;
+    if (strcmp(env, "cpu") == 0)
+        target = cpu;
+    else if (strcmp(env, "gpu") == 0)
+        target = gpu;
+    else if (strcmp(env, "tpu") == 0)
+        target = tpu;
+    else {
+        NN_ERR_PRINTF("Wrong target!");
+        return 1;
+    }
+    NN_INFO_PRINTF("################### Testing quantized model...");
+    test_average_quantized(target);
+
+    NN_INFO_PRINTF("Tests: passed!");
+    return 0;
+}
--- a/core/iwasm/libraries/wasi-nn/test/utils.c
+++ b/core/iwasm/libraries/wasi-nn/test/utils.c
@ -132,8 +132,8 @@ run_inference(execution_target target, float *input, uint32_t *input_size,
        *output_size = MAX_OUTPUT_TENSOR_SIZE - *output_size;
        if (wasm_get_output(ctx, i, &out_tensor[offset], output_size)
            != success) {
-            NN_ERR_PRINTF("Error when getting output.");
-            exit(1);
+            NN_ERR_PRINTF("Error when getting index %d.", i);
+            break;
        }

        offset += *output_size;
--- a/core/iwasm/libraries/wasi-nn/test/utils.h
+++ b/core/iwasm/libraries/wasi-nn/test/utils.h
@ -11,7 +11,7 @@
 #include "wasi_nn.h"

 #define MAX_MODEL_SIZE 85000000
-#define MAX_OUTPUT_TENSOR_SIZE 200
+#define MAX_OUTPUT_TENSOR_SIZE 1000000
 #define INPUT_TENSOR_DIMS 4
 #define EPSILON 1e-8

--- a/core/shared/platform/common/posix/posix_socket.c
+++ b/core/shared/platform/common/posix/posix_socket.c
@ -275,7 +275,7 @@ os_socket_recv_from(bh_socket_t socket, void *buf, unsigned int len, int flags,
            return -1;
        }
    }
-    else {
+    else if (src_addr) {
        memset(src_addr, 0, sizeof(*src_addr));
    }

--- a/core/shared/utils/bh_common.c
+++ b/core/shared/utils/bh_common.c
@ -31,6 +31,10 @@ b_memcpy_wa(void *s1, unsigned int s1max, const void *s2, unsigned int n)
    unsigned int *p;
    char *ps;

+    if (n == 0) {
+        return 0;
+    }
+
    if (pa > src) {
        pa -= 4;
    }
--- a/core/shared/utils/bh_common.h
+++ b/core/shared/utils/bh_common.h
@ -12,25 +12,25 @@
 extern "C" {
 #endif

-#define bh_memcpy_s(dest, dlen, src, slen)                            \
-    do {                                                              \
-        int _ret = slen == 0 ? 0 : b_memcpy_s(dest, dlen, src, slen); \
-        (void)_ret;                                                   \
-        bh_assert(_ret == 0);                                         \
+#define bh_memcpy_s(dest, dlen, src, slen)            \
+    do {                                              \
+        int _ret = b_memcpy_s(dest, dlen, src, slen); \
+        (void)_ret;                                   \
+        bh_assert(_ret == 0);                         \
    } while (0)

-#define bh_memcpy_wa(dest, dlen, src, slen)                            \
-    do {                                                               \
-        int _ret = slen == 0 ? 0 : b_memcpy_wa(dest, dlen, src, slen); \
-        (void)_ret;                                                    \
-        bh_assert(_ret == 0);                                          \
+#define bh_memcpy_wa(dest, dlen, src, slen)            \
+    do {                                               \
+        int _ret = b_memcpy_wa(dest, dlen, src, slen); \
+        (void)_ret;                                    \
+        bh_assert(_ret == 0);                          \
    } while (0)

-#define bh_memmove_s(dest, dlen, src, slen)                            \
-    do {                                                               \
-        int _ret = slen == 0 ? 0 : b_memmove_s(dest, dlen, src, slen); \
-        (void)_ret;                                                    \
-        bh_assert(_ret == 0);                                          \
+#define bh_memmove_s(dest, dlen, src, slen)            \
+    do {                                               \
+        int _ret = b_memmove_s(dest, dlen, src, slen); \
+        (void)_ret;                                    \
+        bh_assert(_ret == 0);                          \
    } while (0)

 #define bh_strcat_s(dest, dlen, src)            \
--- a/core/version.h
+++ b/core/version.h
@ -7,5 +7,5 @@
 #define _WAMR_VERSION_H_
 #define WAMR_VERSION_MAJOR 1
 #define WAMR_VERSION_MINOR 2
-#define WAMR_VERSION_PATCH 2
+#define WAMR_VERSION_PATCH 3
 #endif
--- a/doc/build_wamr.md
+++ b/doc/build_wamr.md
@ -94,7 +94,12 @@ cmake -DWAMR_BUILD_PLATFORM=linux -DWAMR_BUILD_TARGET=ARM
 - **WAMR_BUILD_WASI_NN**=1/0, default to disable if not set

 #### **Enable lib wasi-nn GPU mode**
- **WASI_NN_ENABLE_GPU**=1/0, default to disable if not set
+- **WAMR_BUILD_WASI_NN_ENABLE_GPU**=1/0, default to disable if not set
+
+#### **Enable lib wasi-nn external delegate mode**
+- **WAMR_BUILD_WASI_NN_ENABLE_EXTERNAL_DELEGATE**=1/0, default to disable if not set
+
+- **WAMR_BUILD_WASI_NN_EXTERNAL_DELEGATE_PATH**=Path to the external delegate shared library (e.g. `libedgetpu.so.1.0` for Coral USB)

 #### **Disable boundary check with hardware trap**
 - **WAMR_DISABLE_HW_BOUND_CHECK**=1/0, default to enable if not set and supported by platform
--- a/doc/source_debugging.md
+++ b/doc/source_debugging.md
@ -44,7 +44,7 @@ iwasm -g=127.0.0.1:1234 test.wasm
 ``` bash
 git clone --branch release/13.x --depth=1 https://github.com/llvm/llvm-project
 cd llvm-project
-git apply ${WAMR_ROOT}/build-scripts/lldb-wasm.patch
+git apply ${WAMR_ROOT}/build-scripts/lldb_wasm.patch
 mkdir build-lldb
 cmake -S ./llvm -B build-lldb \
    -G Ninja \
--- a/language-bindings/python/src/wamr/wamrapi/wamr.py
+++ b/language-bindings/python/src/wamr/wamrapi/wamr.py
@ -2,6 +2,7 @@
 # SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception

 from ctypes import Array
+from ctypes import addressof
 from ctypes import c_char
 from ctypes import c_uint
 from ctypes import c_uint8
@ -10,6 +11,8 @@ from ctypes import cast
 from ctypes import create_string_buffer
 from ctypes import POINTER
 from ctypes import pointer
+from typing import List
+from typing import Tuple
 from wamr.wamrapi.iwasm import String
 from wamr.wamrapi.iwasm import Alloc_With_Pool
 from wamr.wamrapi.iwasm import RuntimeInitArgs
@ -31,6 +34,14 @@ from wamr.wamrapi.iwasm import wasm_runtime_module_malloc
 from wamr.wamrapi.iwasm import wasm_runtime_module_free
 from wamr.wamrapi.iwasm import wasm_runtime_register_natives
 from wamr.wamrapi.iwasm import NativeSymbol
+from wamr.wamrapi.iwasm import wasm_runtime_start_debug_instance
+from wamr.wamrapi.iwasm import wasm_runtime_call_indirect
+from wamr.wamrapi.iwasm import wasm_runtime_get_module_inst
+from wamr.wamrapi.iwasm import wasm_runtime_addr_app_to_native
+from wamr.wamrapi.iwasm import wasm_runtime_addr_native_to_app
+from wamr.wamrapi.iwasm import wasm_runtime_set_wasi_args
+
+ID_TO_EXEC_ENV_MAPPING = {}


 class Engine:
@ -43,16 +54,26 @@ class Engine:
        print("deleting Engine")
        wasm_runtime_destroy()

-    def _get_init_args(self, heap_size: int = 1024 * 512) -> RuntimeInitArgs:
+    def _get_init_args(
+        self,
+        heap_size: int = 1024 * 1024 * 2,
+        ip_addr: str = "127.0.0.1",
+        instance_port: int = 1234,
+    ) -> RuntimeInitArgs:
        init_args = RuntimeInitArgs()
        init_args.mem_alloc_type = Alloc_With_Pool
        init_args.mem_alloc_option.pool.heap_buf = cast(
            (c_char * heap_size)(), c_void_p
        )
        init_args.mem_alloc_option.pool.heap_size = heap_size
+        # Debug port setting
+        init_args.ip_addr = bytes(ip_addr, "utf-8")
+        init_args.instance_port = instance_port
        return init_args

-    def register_natives(self, module_name: str, native_symbols: list[NativeSymbol]) -> None:
+    def register_natives(
+        self, module_name: str, native_symbols: List[NativeSymbol]
+    ) -> None:
        module_name = String.from_param(module_name)
        # WAMR does not copy the symbols. We must store them.
        for native in native_symbols:
@ -62,12 +83,13 @@ class Engine:
            module_name,
            cast(
                (NativeSymbol * len(native_symbols))(*native_symbols),
-                POINTER(NativeSymbol)
+                POINTER(NativeSymbol),
            ),
-            len(native_symbols)
+            len(native_symbols),
        ):
            raise Exception("Error while registering symbols")

+
 class Module:
    __create_key = object()

@ -86,7 +108,7 @@ class Module:
        print("deleting Module")
        wasm_runtime_unload(self.module)

-    def _create_module(self, fp: str) -> tuple[wasm_module_t, Array[c_uint]]:
+    def _create_module(self, fp: str) -> Tuple[wasm_module_t, "Array[c_uint]"]:
        with open(fp, "rb") as f:
            data = f.read()
            data = (c_uint8 * len(data))(*data)
@ -99,14 +121,52 @@ class Module:


 class Instance:
-    def __init__(self, module: Module, stack_size: int = 65536, heap_size: int = 16384):
+    def __init__(
+        self,
+        module: Module,
+        stack_size: int = 65536,
+        heap_size: int = 16384,
+        dir_list: List[str] | None = None,
+        preinitialized_module_inst: wasm_module_inst_t | None = None,
+    ):
+        # Store module ensures GC does not remove it
        self.module = module
-        self.module_inst = self._create_module_inst(module, stack_size, heap_size)
+        if dir_list:
+            self._set_wasi_args(module, dir_list)
+        if preinitialized_module_inst is None:
+            self.module_inst = self._create_module_inst(module, stack_size, heap_size)
+        else:
+            self.module_inst = preinitialized_module_inst

    def __del__(self):
        print("deleting Instance")
        wasm_runtime_deinstantiate(self.module_inst)

+    def _set_wasi_args(self, module: Module, dir_list: List[str]) -> None:
+        LP_c_char = POINTER(c_char)
+        LP_LP_c_char = POINTER(LP_c_char)
+
+        p = (LP_c_char * len(dir_list))()
+        for i, dir in enumerate(dir_list):
+            enc_dir = dir.encode("utf-8")
+            p[i] = create_string_buffer(enc_dir)
+
+        na = cast(p, LP_LP_c_char)
+        wasm_runtime_set_wasi_args(
+            module.module, na, len(dir_list), None, 0, None, 0, None, 0
+        )
+
+    def _create_module_inst(
+        self, module: Module, stack_size: int, heap_size: int
+    ) -> wasm_module_inst_t:
+        error_buf = create_string_buffer(128)
+        module_inst = wasm_runtime_instantiate(
+            module.module, stack_size, heap_size, error_buf, len(error_buf)
+        )
+        if not module_inst:
+            raise Exception("Error while creating module instance")
+        return module_inst
+
    def malloc(self, nbytes: int, native_handler) -> c_uint:
        return wasm_runtime_module_malloc(self.module_inst, nbytes, native_handler)

@ -119,31 +179,70 @@ class Instance:
            raise Exception("Error while looking-up function")
        return func

-    def _create_module_inst(self, module: Module, stack_size: int, heap_size: int) -> wasm_module_inst_t:
-        error_buf = create_string_buffer(128)
-        module_inst = wasm_runtime_instantiate(
-            module.module, stack_size, heap_size, error_buf, len(error_buf)
-        )
-        if not module_inst:
-            raise Exception("Error while creating module instance")
-        return module_inst
+    def native_addr_to_app_addr(self, native_addr) -> c_void_p:
+        return wasm_runtime_addr_native_to_app(self.module_inst, native_addr)
+
+    def app_addr_to_native_addr(self, app_addr) -> c_void_p:
+        return wasm_runtime_addr_app_to_native(self.module_inst, app_addr)


 class ExecEnv:
    def __init__(self, module_inst: Instance, stack_size: int = 65536):
        self.module_inst = module_inst
        self.exec_env = self._create_exec_env(module_inst, stack_size)
+        self.env = addressof(self.exec_env.contents)
+        self.own_c = True
+
+        ID_TO_EXEC_ENV_MAPPING[str(self.env)] = self

    def __del__(self):
-        print("deleting ExecEnv")
-        wasm_runtime_destroy_exec_env(self.exec_env)
+        if self.own_c:
+            print("deleting ExecEnv")
+            wasm_runtime_destroy_exec_env(self.exec_env)
+            del ID_TO_EXEC_ENV_MAPPING[str(self.env)]
+
+    def _create_exec_env(
+        self, module_inst: Instance, stack_size: int
+    ) -> wasm_exec_env_t:
+        exec_env = wasm_runtime_create_exec_env(module_inst.module_inst, stack_size)
+        if not exec_env:
+            raise Exception("Error while creating execution environment")
+        return exec_env

    def call(self, func: wasm_function_inst_t, argc: int, argv: "POINTER[c_uint]"):
        if not wasm_runtime_call_wasm(self.exec_env, func, argc, argv):
            raise Exception("Error while calling function")

-    def _create_exec_env(self, module_inst: Instance, stack_size: int) -> wasm_exec_env_t:
-        exec_env = wasm_runtime_create_exec_env(module_inst.module_inst, stack_size)
-        if not exec_env:
-            raise Exception("Error while creating execution environment")
-        return exec_env
+    def get_module_inst(self) -> Instance:
+        return self.module_inst
+
+    def start_debugging(self) -> int:
+        return wasm_runtime_start_debug_instance(self.exec_env)
+
+    def call_indirect(self, element_index: int, argc: int, argv: "POINTER[c_uint]"):
+        if not wasm_runtime_call_indirect(self.exec_env, element_index, argc, argv):
+            raise Exception("Error while calling function")
+
+    @staticmethod
+    def wrap(env: int) -> "ExecEnv":
+        if str(env) in ID_TO_EXEC_ENV_MAPPING:
+            return ID_TO_EXEC_ENV_MAPPING[str(env)]
+        return InternalExecEnv(env)
+
+
+class InternalExecEnv(ExecEnv):
+    """
+    Generate Python ExecEnv-like object from a `wasm_exec_env_t` index.
+    """
+
+    def __init__(self, env: int):
+        self.env = env
+        self.exec_env = cast(env, wasm_exec_env_t)
+        self.module_inst = Instance(
+            module=object(),
+            preinitialized_module_inst=wasm_runtime_get_module_inst(self.exec_env),
+        )
+        ID_TO_EXEC_ENV_MAPPING[str(env)] = self
+
+    def __del__(self):
+        del ID_TO_EXEC_ENV_MAPPING[str(self.env)]
--- a/language-bindings/python/utils/create_lib.sh
+++ b/language-bindings/python/utils/create_lib.sh
@ -12,7 +12,12 @@ WAMR_BUILD_PLATFORM=${WAMR_BUILD_PLATFORM:-${UNAME}}
 cd ${ROOT_DIR}/product-mini/platforms/${WAMR_BUILD_PLATFORM}

 mkdir -p build && cd build
-cmake ..
+cmake \
+    -DWAMR_BUILD_DEBUG_INTERP=1 \
+    -DWAMR_BUILD_LIB_PTHREAD=1 \
+    -DWAMR_BUILD_LIB_WASI_THREADS=1 \
+    -DWAMR_BUILD_LIB_WASI=1 \
+    ..
 make -j

 case ${UNAME} in
--- a/language-bindings/python/wamr-api/README.md
+++ b/language-bindings/python/wamr-api/README.md
@ -22,10 +22,7 @@ bash language-bindings/python/utils/create_lib.sh

 This will build and copy libiwasm into the package.

-## Examples
+## Samples

-There is a [simple example](./samples/main.py) to show how to use bindings.
-
-```
-python samples/main.py
-```
+- **[basic](./samples/basic)**: Demonstrating how to use basic python bindings.
+- **[native-symbol](./samples/native-symbol)**: Desmostrate how to call WASM from Python and how to export Python functions into WASM.
--- a/language-bindings/python/wamr-api/samples/basic/compile.sh
+++ b/language-bindings/python/wamr-api/samples/basic/compile.sh
--- a/language-bindings/python/wamr-api/samples/basic/main.py
+++ b/language-bindings/python/wamr-api/samples/basic/main.py
--- a/language-bindings/python/wamr-api/samples/basic/sum.c
+++ b/language-bindings/python/wamr-api/samples/basic/sum.c
--- a/language-bindings/python/wamr-api/samples/native-symbol/README.md
+++ b/language-bindings/python/wamr-api/samples/native-symbol/README.md
@ -0,0 +1,44 @@
+# Native Symbol
+
+This sample demonstrates how to declare a Python function as `NativeSymbol`.
+
+Steps of the example:
+1. Load WASM from Python
+2. Call `c_func` from WASM.
+3. `c_func` calls `python_func` from Python.
+4. `python_func` calls `add` from WASM.
+5. Result shown by Python.
+
+## Build
+
+Follow instructions [build wamr Python package](../../README.md). 
+
+Compile WASM app example,
+
+```sh
+./compile.sh
+```
+
+## Run sample
+
+```sh
+python main.py
+```
+
+Output:
+
+```
+python: calling c_func(10)
+c: in c_func with input: 10
+c: calling python_func(11)
+python: in python_func with input: 11
+python: calling add(11, 1000)
+python: result from add: 1011
+c: result from python_func: 1012
+c: returning 1013
+python: result from c_func: 1013
+deleting ExecEnv
+deleting Instance
+deleting Module
+deleting Engine
+```
--- a/language-bindings/python/wamr-api/samples/native-symbol/compile.sh
+++ b/language-bindings/python/wamr-api/samples/native-symbol/compile.sh
@ -0,0 +1,14 @@
+#!/bin/sh
+
+# Copyright (C) 2019 Intel Corporation.  All rights reserved.
+# SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+
+/opt/wasi-sdk/bin/clang     \
+    -O0 -z stack-size=4096 -Wl,--initial-memory=65536 \
+    -Wl,--export=main -Wl,--export=__main_argc_argv \
+    -Wl,--export=__data_end -Wl,--export=__heap_base \
+    -Wl,--strip-all,--no-entry \
+    -Wl,--allow-undefined \
+    -Wl,--export=c_func\
+    -Wl,--export=add\
+    -o func.wasm func.c
--- a/language-bindings/python/wamr-api/samples/native-symbol/func.c
+++ b/language-bindings/python/wamr-api/samples/native-symbol/func.c
@ -0,0 +1,30 @@
+/*
+ * Copyright (C) 2019 Intel Corporation.  All rights reserved.
+ * SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+ */
+
+#include <stdio.h>
+
+int
+python_func(int val);
+
+int
+add(int val1, int val2)
+{
+    return val1 + val2;
+}
+
+int
+c_func(int val)
+{
+    printf("c: in c_func with input: %d\n", val);
+    printf("c: calling python_func(%d)\n", val + 1);
+    int res = python_func(val + 1);
+    printf("c: result from python_func: %d\n", res);
+    printf("c: returning %d\n", res + 1);
+    return res + 1;
+}
+
+int
+main()
+{}
--- a/language-bindings/python/wamr-api/samples/native-symbol/main.py
+++ b/language-bindings/python/wamr-api/samples/native-symbol/main.py
@ -0,0 +1,59 @@
+# Copyright (C) 2019 Intel Corporation.  All rights reserved.
+# SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+
+from wamr.wamrapi.wamr import Engine, Module, Instance, ExecEnv
+from ctypes import c_uint
+import pathlib
+from ctypes import c_int32
+from ctypes import c_uint
+from ctypes import c_void_p
+from ctypes import cast
+from ctypes import CFUNCTYPE
+
+from wamr.wamrapi.iwasm import NativeSymbol
+from wamr.wamrapi.iwasm import String
+from wamr.wamrapi.wamr import ExecEnv
+
+def python_func(env: int, value: int) -> int:
+    print("python: in python_func with input:", value)
+    # Example of generating ExecEnv from `wasm_exec_env_t``
+    exec_env = ExecEnv.wrap(env)
+    add = exec_env.get_module_inst().lookup_function("add")
+    const = 1000
+    argv = (c_uint * 2)(value, const)
+    print(f"python: calling add({value}, {const})")
+    exec_env.call(add, 2, argv)
+    res = argv[0]
+    print("python: result from add:", res)
+    return res + 1
+
+
+native_symbols = (NativeSymbol * 1)(
+    *[
+        NativeSymbol(
+            symbol=String.from_param("python_func"),
+            func_ptr=cast(
+                CFUNCTYPE(c_int32, c_void_p, c_int32)(python_func), c_void_p
+            ),
+            signature=String.from_param("(i)i"),
+        )
+    ]
+)
+
+def main():
+    engine = Engine()
+    engine.register_natives("env", native_symbols)
+    module = Module.from_file(engine, pathlib.Path(__file__).parent / "func.wasm")
+    module_inst = Instance(module)
+    exec_env = ExecEnv(module_inst)
+
+    func = module_inst.lookup_function("c_func")
+
+    inp = 10
+    print(f"python: calling c_func({inp})")
+    argv = (c_uint)(inp)
+    exec_env.call(func, 1, argv)
+    print("python: result from c_func:", argv.value)
+
+if __name__ == "__main__":
+    main()
--- a/product-mini/platforms/linux-sgx/enclave-sample/Makefile
+++ b/product-mini/platforms/linux-sgx/enclave-sample/Makefile
@ -143,10 +143,10 @@ else
 endif

 ifeq ($(WAMR_BUILD_LIB_RATS), 1)
-	Rats_Lib_Link_Dirs := -L$(LIB_RATS_INSTALL_DIR) -L$(LIB_RATS_INSTALL_DIR)/attesters -L$(LIB_RATS_INSTALL_DIR)/verifiers -L$(SGX_SSL)/lib64
+	Rats_Lib_Link_Dirs := -L$(LIB_RATS_INSTALL_DIR) -L$(LIB_RATS_INSTALL_DIR)/attesters -L$(LIB_RATS_INSTALL_DIR)/verifiers -L$(SGX_SSL)/lib64 -L$(VMLIB_BUILD_DIR)/external/libcbor/src/libcbor/lib -L$(LIB_RATS_INSTALL_DIR)/crypto_wrappers
 	Rats_Lib_W_Link_libs := -lattester_nullattester -lattester_sgx_ecdsa -lattester_sgx_la \
-						-lverifier_nullverifier -lverifier_sgx_ecdsa -lverifier_sgx_la -lverifier_sgx_ecdsa_qve \
-						-lrats_lib -lsgx_tsgxssl
+						-lverifier_nullverifier -lverifier_sgx_la -lverifier_sgx_ecdsa_qve -lcbor \
+						-lrats_lib -lsgx_tsgxssl -lcrypto_wrapper_nullcrypto -lcrypto_wrapper_openssl
 	Rats_Lib_NW_Link_libs := -lsgx_dcap_tvl -lsgx_tsgxssl_crypto
 endif

--- a/product-mini/platforms/posix/main.c
+++ b/product-mini/platforms/posix/main.c
@ -253,93 +253,124 @@ resolve_segue_flags(char *str_flags)
 #endif /* end of WASM_ENABLE_JIT != 0 */

 #if BH_HAS_DLFCN
-typedef uint32 (*get_native_lib_func)(char **p_module_name,
-                                      NativeSymbol **p_native_symbols);
+struct native_lib {
+    void *handle;
+
+    uint32 (*get_native_lib)(char **p_module_name,
+                             NativeSymbol **p_native_symbols);
+    int (*init_native_lib)(void);
+    void (*deinit_native_lib)(void);
+
+    char *module_name;
+    NativeSymbol *native_symbols;
+    uint32 n_native_symbols;
+};
+
+struct native_lib *
+load_native_lib(const char *name)
+{
+    struct native_lib *lib = wasm_runtime_malloc(sizeof(*lib));
+    if (lib == NULL) {
+        LOG_WARNING("warning: failed to load native library %s because of "
+                    "allocation failure",
+                    name);
+        goto fail;
+    }
+    memset(lib, 0, sizeof(*lib));
+
+    /* open the native library */
+    if (!(lib->handle = dlopen(name, RTLD_NOW | RTLD_GLOBAL))
+        && !(lib->handle = dlopen(name, RTLD_LAZY))) {
+        LOG_WARNING("warning: failed to load native library %s", name);
+        goto fail;
+    }
+
+    lib->init_native_lib = dlsym(lib->handle, "init_native_lib");
+    lib->get_native_lib = dlsym(lib->handle, "get_native_lib");
+    lib->deinit_native_lib = dlsym(lib->handle, "deinit_native_lib");
+
+    if (!lib->get_native_lib) {
+        LOG_WARNING("warning: failed to lookup `get_native_lib` function "
+                    "from native lib %s",
+                    name);
+        goto fail;
+    }
+
+    if (lib->init_native_lib) {
+        int ret = lib->init_native_lib();
+        if (ret != 0) {
+            LOG_WARNING("warning: `init_native_lib` function from native "
+                        "lib %s failed with %d",
+                        name, ret);
+            goto fail;
+        }
+    }
+
+    lib->n_native_symbols =
+        lib->get_native_lib(&lib->module_name, &lib->native_symbols);
+
+    /* register native symbols */
+    if (!(lib->n_native_symbols > 0 && lib->module_name && lib->native_symbols
+          && wasm_runtime_register_natives(
+              lib->module_name, lib->native_symbols, lib->n_native_symbols))) {
+        LOG_WARNING("warning: failed to register native lib %s", name);
+        if (lib->deinit_native_lib) {
+            lib->deinit_native_lib();
+        }
+        goto fail;
+    }
+    return lib;
+fail:
+    if (lib != NULL) {
+        if (lib->handle != NULL) {
+            dlclose(lib->handle);
+        }
+        wasm_runtime_free(lib);
+    }
+    return NULL;
+}

 static uint32
 load_and_register_native_libs(const char **native_lib_list,
                              uint32 native_lib_count,
-                              void **native_handle_list)
+                              struct native_lib **native_lib_loaded_list)
 {
-    uint32 i, native_handle_count = 0, n_native_symbols;
-    NativeSymbol *native_symbols;
-    char *module_name;
-    void *handle;
+    uint32 i, native_lib_loaded_count = 0;

    for (i = 0; i < native_lib_count; i++) {
-        /* open the native library */
-        if (!(handle = dlopen(native_lib_list[i], RTLD_NOW | RTLD_GLOBAL))
-            && !(handle = dlopen(native_lib_list[i], RTLD_LAZY))) {
-            LOG_WARNING("warning: failed to load native library %s",
-                        native_lib_list[i]);
+        struct native_lib *lib = load_native_lib(native_lib_list[i]);
+        if (lib == NULL) {
            continue;
        }
-
-        /* lookup get_native_lib func */
-        get_native_lib_func get_native_lib = dlsym(handle, "get_native_lib");
-        if (!get_native_lib) {
-            LOG_WARNING("warning: failed to lookup `get_native_lib` function "
-                        "from native lib %s",
-                        native_lib_list[i]);
-            dlclose(handle);
-            continue;
-        }
-
-        n_native_symbols = get_native_lib(&module_name, &native_symbols);
-
-        /* register native symbols */
-        if (!(n_native_symbols > 0 && module_name && native_symbols
-              && wasm_runtime_register_natives(module_name, native_symbols,
-                                               n_native_symbols))) {
-            LOG_WARNING("warning: failed to register native lib %s",
-                        native_lib_list[i]);
-            dlclose(handle);
-            continue;
-        }
-
-        native_handle_list[native_handle_count++] = handle;
+        native_lib_loaded_list[native_lib_loaded_count++] = lib;
    }

-    return native_handle_count;
+    return native_lib_loaded_count;
 }

 static void
 unregister_and_unload_native_libs(uint32 native_lib_count,
-                                  void **native_handle_list)
+                                  struct native_lib **native_lib_loaded_list)
 {
-    uint32 i, n_native_symbols;
-    NativeSymbol *native_symbols;
-    char *module_name;
-    void *handle;
+    uint32 i;

    for (i = 0; i < native_lib_count; i++) {
-        handle = native_handle_list[i];
-
-        /* lookup get_native_lib func */
-        get_native_lib_func get_native_lib = dlsym(handle, "get_native_lib");
-        if (!get_native_lib) {
-            LOG_WARNING("warning: failed to lookup `get_native_lib` function "
-                        "from native lib %p",
-                        handle);
-            continue;
-        }
-
-        n_native_symbols = get_native_lib(&module_name, &native_symbols);
-        if (n_native_symbols == 0 || module_name == NULL
-            || native_symbols == NULL) {
-            LOG_WARNING("warning: get_native_lib returned different values for "
-                        "native lib %p",
-                        handle);
-            continue;
-        }
+        struct native_lib *lib = native_lib_loaded_list[i];

        /* unregister native symbols */
-        if (!wasm_runtime_unregister_natives(module_name, native_symbols)) {
-            LOG_WARNING("warning: failed to unregister native lib %p", handle);
+        if (!wasm_runtime_unregister_natives(lib->module_name,
+                                             lib->native_symbols)) {
+            LOG_WARNING("warning: failed to unregister native lib %p",
+                        lib->handle);
            continue;
        }

-        dlclose(handle);
+        if (lib->deinit_native_lib) {
+            lib->deinit_native_lib();
+        }
+
+        dlclose(lib->handle);
+        wasm_runtime_free(lib);
    }
 }
 #endif /* BH_HAS_DLFCN */
@ -469,8 +500,8 @@ main(int argc, char *argv[])
 #if BH_HAS_DLFCN
    const char *native_lib_list[8] = { NULL };
    uint32 native_lib_count = 0;
-    void *native_handle_list[8] = { NULL };
-    uint32 native_handle_count = 0;
+    struct native_lib *native_lib_loaded_list[8];
+    uint32 native_lib_loaded_count = 0;
 #endif
 #if WASM_ENABLE_DEBUG_INTERP != 0
    char *ip_addr = NULL;
@ -708,8 +739,8 @@ main(int argc, char *argv[])
 #endif

 #if BH_HAS_DLFCN
-    native_handle_count = load_and_register_native_libs(
-        native_lib_list, native_lib_count, native_handle_list);
+    native_lib_loaded_count = load_and_register_native_libs(
+        native_lib_list, native_lib_count, native_lib_loaded_list);
 #endif

    /* load WASM byte buffer from WASM bin file */
@ -838,7 +869,8 @@ fail2:
 fail1:
 #if BH_HAS_DLFCN
    /* unload the native libraries */
-    unregister_and_unload_native_libs(native_handle_count, native_handle_list);
+    unregister_and_unload_native_libs(native_lib_loaded_count,
+                                      native_lib_loaded_list);
 #endif

    /* destroy runtime environment */
--- a/samples/bh-atomic/CMakeLists.txt
+++ b/samples/bh-atomic/CMakeLists.txt
--- a/samples/bh-atomic/main.c
+++ b/samples/bh-atomic/main.c
--- a/samples/mem-allocator/CMakeLists.txt
+++ b/samples/mem-allocator/CMakeLists.txt
--- a/samples/mem-allocator/main.c
+++ b/samples/mem-allocator/main.c
--- a/samples/native-lib/README.md
+++ b/samples/native-lib/README.md
@ -62,6 +62,7 @@ cd build
 The output is:

 ```bash
+init_native_lib in test_hello2.c called
 Hello World!
 10 + 20 = 30
 sqrt(10, 20) = 500
@ -72,5 +73,6 @@ Message from test_hello: Hello, main. This is test_hello_wrapper!
 test_hello2("main", 0x0, 0) = 85
 malloc(86) = 0x24e8
 test_hello2("main", 0x24e8, 86) = 85
-Message from test_hello2: Hello, main. This is test_hello2_wrapper! Your wasm_module_inst_t is 0x7fd443704990.
+Message from test_hello2: Hello, main. This is test_hello2_wrapper! Your wasm_module_inst_t is 0x7fe0e6804280.
+deinit_native_lib in test_hello2.c called
 ```
--- a/samples/native-lib/test_hello2.c
+++ b/samples/native-lib/test_hello2.c
@ -57,3 +57,16 @@ get_native_lib(char **p_module_name, NativeSymbol **p_native_symbols)
    *p_native_symbols = native_symbols;
    return sizeof(native_symbols) / sizeof(NativeSymbol);
 }
+
+int
+init_native_lib()
+{
+    printf("%s in test_hello2.c called\n", __func__);
+    return 0;
+}
+
+void
+deinit_native_lib()
+{
+    printf("%s in test_hello2.c called\n", __func__);
+}
--- a/samples/sgx-ra/README.md
+++ b/samples/sgx-ra/README.md
@ -7,52 +7,52 @@ This sample demonstrates how to execute Remote Attestation on SGX with [librats]

 SGX-RA requires to have installed:
 - the WASI-SDK, located in `/opt/wasi-sdk`
- - CMake >= 3.11, which is not provided on Ubuntu 18.04 (use [Kitware APT Repository](https://apt.kitware.com/))

 ### Intel SGX dependencies

 Before starting, we need to download and install [SGX SDK](https://download.01.org/intel-sgx/latest/linux-latest/distro) and [SGX DCAP Library](https://download.01.org/intel-sgx/latest/dcap-latest) referring to this [guide](https://download.01.org/intel-sgx/sgx-dcap/1.8/linux/docs/Intel_SGX_DCAP_Linux_SW_Installation_Guide.pdf).

-The following commands are an example of the SGX environment installation on Ubuntu 18.04.
+The following commands are an example of the SGX environment installation on Ubuntu 20.04.
 ``` shell
 # Set your platform, you can get the platforms list on
 # https://download.01.org/intel-sgx/latest/linux-latest/distro
 $ cd $HOME
-$ SGX_PLATFORM=ubuntu18.04-server
-$ SGX_SDK_VERSION=2.17.100.3
+$ OS_PLATFORM=ubuntu20.04
+$ SGX_PLATFORM=$OS_PLATFORM-server
+$ SGX_RELEASE_VERSION=1.17
 $ SGX_DRIVER_VERSION=1.41
+$ SGX_SDK_VERSION=2.20.100.4

 # install the dependencies
 $ sudo apt-get update
-$ sudo apt-get install -y dkms
+$ sudo apt-get install -y build-essential ocaml automake autoconf libtool wget python3 libssl-dev dkms zip cmake
+$ sudo update-alternatives --install /usr/bin/python python /usr/bin/python3 1

 # install SGX Driver
-$ wget https://download.01.org/intel-sgx/latest/linux-latest/distro/$SGX_PLATFORM/sgx_linux_x64_driver_$SGX_DRIVER_VERSION.bin
+$ wget https://download.01.org/intel-sgx/sgx-dcap/$SGX_RELEASE_VERSION/linux/distro/$SGX_PLATFORM/sgx_linux_x64_driver_$SGX_DRIVER_VERSION.bin
 $ chmod +x sgx_linux_x64_driver_$SGX_DRIVER_VERSION.bin
 $ sudo ./sgx_linux_x64_driver_$SGX_DRIVER_VERSION.bin

 # install SGX SDK
-$ wget https://download.01.org/intel-sgx/latest/linux-latest/distro/$SGX_PLATFORM/sgx_linux_x64_sdk_$SGX_SDK_VERSION.bin
+$ wget https://download.01.org/intel-sgx/sgx-dcap/$SGX_RELEASE_VERSION/linux/distro/$SGX_PLATFORM/sgx_linux_x64_sdk_$SGX_SDK_VERSION.bin
 $ chmod +x sgx_linux_x64_sdk_$SGX_SDK_VERSION.bin
-$ sudo ./sgx_linux_x64_sdk_$SGX_SDK_VERSION.bin
+$ sudo ./sgx_linux_x64_sdk_$SGX_SDK_VERSION.bin --prefix /opt/intel

 # install SGX DCAP Library
-$ echo 'deb [arch=amd64] https://download.01.org/intel-sgx/sgx_repo/ubuntu bionic main' | sudo tee /etc/apt/sources.list.d/intel-sgx.list > /dev/null
-$ wget -O - https://download.01.org/intel-sgx/sgx_repo/ubuntu/intel-sgx-deb.key | sudo apt-key add -
+$ 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 -O - https://download.01.org/intel-sgx/sgx_repo/ubuntu/intel-sgx-deb.key | sudo apt-key add
 $ sudo apt-get update
-$ sudo apt-get install -y libsgx-uae-service libsgx-dcap-default-qpl-dev libsgx-dcap-ql-dev libsgx-dcap-quote-verify-dev
+$ sudo apt-get install -y libsgx-epid libsgx-quote-ex libsgx-dcap-ql libsgx-enclave-common-dev libsgx-dcap-ql-dev libsgx-dcap-default-qpl-dev libsgx-dcap-quote-verify-dev

 # install SGX SSL Library
 $ git clone https://github.com/intel/linux-sgx.git
 $ cd linux-sgx && make preparation
-$ sudo cp external/toolset/{current_distr}/* /usr/local/bin
+$ sudo cp external/toolset/$OS_PLATFORM/* /usr/local/bin
 $ # Verify that the paths are correctly set
 $ which ar as ld objcopy objdump ranlib
 $ cd ../
 $ git clone https://github.com/intel/intel-sgx-ssl.git
-$ wget https://www.openssl.org/source/openssl-1.1.1q.tar.gz
-$ cp openssl-1.1.1q.tar.gz intel-sgx-ssl/openssl_source
-$ rm -f openssl-1.1.1q.tar.gz
+$ wget https://www.openssl.org/source/openssl-1.1.1v.tar.gz -O intel-sgx-ssl/openssl_source/openssl-1.1.1v.tar.gz
 $ cd intel-sgx-ssl/Linux
 $ source /opt/intel/sgxsdk/environment
 $ make all
@ -70,7 +70,7 @@ sudo usermod -a -G sgx_prv <username>

 Intel DCAP connects to Intel PCS to download the attestation collateral for SGX-enabled machines.
 Intel provides a [quick install guide](https://www.intel.com/content/www/us/en/developer/articles/guide/intel-software-guard-extensions-data-center-attestation-primitives-quick-install-guide.html) to set up a simplified environment.
-This section summarizes the commands to issue for setting up a working environment on Ubuntu 18.04.
+This section summarizes the commands to issue for setting up a working environment on Ubuntu 20.04.

 ### Subscribe to Intel PCS Web services

@ -86,11 +86,9 @@ Intel provides an implementation of the cache mechanism.
 The following commands set up Intel PCCS.
 ```shell
 # install Node.js
-$ curl -o setup.sh -sL https://deb.nodesource.com/setup_14.x
-$ chmod a+x setup.sh
-$ sudo ./setup.sh
+$ curl -fsSL https://deb.nodesource.com/setup_20.x | sudo -E bash - && sudo apt-get install -y nodejs
 # install PCCS software
-$ sudo apt-get install -y cracklib-runtime sqlite3 python build-essential
+$ sudo apt-get install -y cracklib-runtime
 $ sudo apt-get install -y sgx-dcap-pccs
 ```

@ -160,9 +158,10 @@ Adapt the configuration file of `PCKIDRetrievalTool` located in `/opt/intel/sgx-
 Save your changes and run the provisioning tool.

 ```shell
-$ PCKIDRetrievalTool
-Intel(R) Software Guard Extensions PCK Cert ID Retrieval Tool Version 1.14.100.3
+$ sudo PCKIDRetrievalTool
+Intel(R) Software Guard Extensions PCK Cert ID Retrieval Tool Version 1.17.100.4

+Registration status has been set to completed status.
 the data has been sent to cache server successfully and pckid_retrieval.csv has been generated successfully!
 ```

--- a/samples/wasm-c-api/src/hello.c
+++ b/samples/wasm-c-api/src/hello.c
@ -65,6 +65,12 @@ int main(int argc, const char* argv[]) {
  }
  fclose(file);

+  if (!wasm_module_validate(store, &binary)) {
+    printf("> Error validate module!\n");
+    wasm_byte_vec_delete(&binary);
+    return 1;
+  }
+
  // Compile.
  printf("Compiling module...\n");
  own wasm_module_t* module = wasm_module_new(store, &binary);
--- a/tests/wamr-test-suites/wasi-test-script/run_wasi_tests.sh
+++ b/tests/wamr-test-suites/wasi-test-script/run_wasi_tests.sh
@ -27,6 +27,7 @@ readonly WAMR_DIR="${WORK_DIR}/../../../.."
 readonly IWASM_CMD="${IWASM_EXE} \
    --allow-resolve=google-public-dns-a.google.com \
    --addr-pool=::1/128,127.0.0.1/32"
+
 readonly IWASM_CMD_STRESS="${IWASM_CMD} --max-threads=8"
 readonly WAMRC_CMD="${WORK_DIR}/../../../../wamr-compiler/build/wamrc"
 readonly C_TESTS="tests/c/testsuite/"
@ -34,14 +35,18 @@ readonly ASSEMBLYSCRIPT_TESTS="tests/assemblyscript/testsuite/"
 readonly THREAD_PROPOSAL_TESTS="tests/proposals/wasi-threads/"
 readonly THREAD_INTERNAL_TESTS="${WAMR_DIR}/core/iwasm/libraries/lib-wasi-threads/test/"
 readonly LIB_SOCKET_TESTS="${WAMR_DIR}/core/iwasm/libraries/lib-socket/test/"
+readonly STRESS_TESTS=("spawn_stress_test.wasm" "stress_test_threads_creation.wasm")

 run_aot_tests () {
    local tests=("$@")
+    local iwasm="${IWASM_CMD}"
    for test_wasm in ${tests[@]}; do
        local extra_stress_flags=""
-        if [[ "$test_wasm" =~ "stress" ]]; then
-            extra_stress_flags="--max-threads=8"
-        fi
+        for stress_test in "${STRESS_TESTS[@]}"; do
+            if [ "$test_wasm" == "$stress_test" ]; then
+                iwasm="${IWASM_CMD_STRESS}"
+            fi
+        done

        test_aot="${test_wasm%.wasm}.aot"
        test_json="${test_wasm%.wasm}.json"
@ -61,7 +66,6 @@ run_aot_tests () {
        fi

        ${IWASM_CMD} $extra_stress_flags $test_aot
-
        ret=${PIPESTATUS[0]}

        echo "expected=$expected, actual=$ret"
@ -75,15 +79,18 @@ if [[ $MODE != "aot" ]];then
    $PYTHON_EXE -m venv wasi-env && source wasi-env/${VENV_BIN_DIR}/activate
    $PYTHON_EXE -m pip install -r test-runner/requirements.txt

-    # Stress test requires max-threads=8 so it's run separately
-    if [[ -e "${THREAD_INTERNAL_TESTS}spawn_stress_test.wasm" ]]; then
-        ${IWASM_CMD_STRESS} ${THREAD_INTERNAL_TESTS}spawn_stress_test.wasm
-        ret=${PIPESTATUS[0]}
-        if [ "${ret}" -ne 0 ]; then
-            echo "Stress test spawn_stress_test FAILED with code " ${ret}
-            exit_code=${ret}
+    # Stress tests require max-threads=8 so they're executed separately
+    for stress_test in "${STRESS_TESTS[@]}"; do
+        if [[ -e "${THREAD_INTERNAL_TESTS}${stress_test}" ]]; then
+            echo "${stress_test}" is a stress test
+            ${IWASM_CMD_STRESS} ${THREAD_INTERNAL_TESTS}${stress_test}
+            ret=${PIPESTATUS[0]}
+            if [ "${ret}" -ne 0 ]; then
+                echo "Stress test ${stress_test} FAILED with code " ${ret}
+                exit_code=${ret}
+            fi
        fi
-    fi
+    done

    TEST_RUNTIME_EXE="${IWASM_CMD}" $PYTHON_EXE test-runner/wasi_test_runner.py \
            -r adapters/wasm-micro-runtime.py \
@ -114,4 +121,4 @@ else
    done
 fi

-exit ${exit_code}
+exit ${exit_code}
--- a/wamr-compiler/main.c
+++ b/wamr-compiler/main.c
@ -126,7 +126,7 @@ print_help()
    printf("                            Use --cpu-features=+help to list all the features supported\n");
    printf("  --opt-level=n             Set the optimization level (0 to 3, default is 3)\n");
    printf("  --size-level=n            Set the code size level (0 to 3, default is 3)\n");
-    printf("  -sgx                      Generate code for SGX platform (Intel Software Guard Extention)\n");
+    printf("  -sgx                      Generate code for SGX platform (Intel Software Guard Extensions)\n");
    printf("  --bounds-checks=1/0       Enable or disable the bounds checks for memory access:\n");
    printf("                              by default it is disabled in all 64-bit platforms except SGX and\n");
    printf("                              in these platforms runtime does bounds checks with hardware trap,\n");
@ -159,8 +159,8 @@ print_help()
    printf("  --enable-dump-call-stack  Enable stack trace feature\n");
    printf("  --enable-perf-profiling   Enable function performance profiling\n");
    printf("  --enable-memory-profiling Enable memory usage profiling\n");
-    printf("  --xip                     A shorthand of --enalbe-indirect-mode --disable-llvm-intrinsics\n");
-    printf("  --enable-indirect-mode    Enalbe call function through symbol table but not direct call\n");
+    printf("  --xip                     A shorthand of --enable-indirect-mode --disable-llvm-intrinsics\n");
+    printf("  --enable-indirect-mode    Enable call function through symbol table but not direct call\n");
    printf("  --disable-llvm-intrinsics Disable the LLVM built-in intrinsics\n");
    printf("  --enable-builtin-intrinsics=<flags>\n");
    printf("                            Enable the specified built-in intrinsics, it will override the default\n");
@ -193,6 +193,8 @@ print_help()
    printf("Examples: wamrc -o test.aot test.wasm\n");
    printf("          wamrc --target=i386 -o test.aot test.wasm\n");
    printf("          wamrc --target=i386 --format=object -o test.o test.wasm\n");
+    printf("          wamrc --target-abi=help\n");
+    printf("          wamrc --target=x86_64 --cpu=help\n");
 }
 /* clang-format on */

@ -296,6 +298,12 @@ resolve_segue_flags(char *str_flags)
    return segue_flags;
 }

+/* When print help info for target/cpu/target-abi/cpu-features, load this dummy
+ * wasm file content rather than from an input file, the dummy wasm file content
+ * is: magic header + version number */
+static unsigned char dummy_wasm_file[8] = { 0x00, 0x61, 0x73, 0x6D,
+                                            0x01, 0x00, 0x00, 0x00 };
+
 int
 main(int argc, char *argv[])
 {
@ -309,7 +317,7 @@ main(int argc, char *argv[])
    AOTCompOption option = { 0 };
    char error_buf[128];
    int log_verbose_level = 2;
-    bool sgx_mode = false, size_level_set = false;
+    bool sgx_mode = false, size_level_set = false, use_dummy_wasm = false;
    int exit_status = EXIT_FAILURE;
 #if BH_HAS_DLFCN
    const char *native_lib_list[8] = { NULL };
@ -342,21 +350,33 @@ main(int argc, char *argv[])
            if (argv[0][9] == '\0')
                PRINT_HELP_AND_EXIT();
            option.target_arch = argv[0] + 9;
+            if (!strcmp(option.target_arch, "help")) {
+                use_dummy_wasm = true;
+            }
        }
        else if (!strncmp(argv[0], "--target-abi=", 13)) {
            if (argv[0][13] == '\0')
                PRINT_HELP_AND_EXIT();
            option.target_abi = argv[0] + 13;
+            if (!strcmp(option.target_abi, "help")) {
+                use_dummy_wasm = true;
+            }
        }
        else if (!strncmp(argv[0], "--cpu=", 6)) {
            if (argv[0][6] == '\0')
                PRINT_HELP_AND_EXIT();
            option.target_cpu = argv[0] + 6;
+            if (!strcmp(option.target_cpu, "help")) {
+                use_dummy_wasm = true;
+            }
        }
        else if (!strncmp(argv[0], "--cpu-features=", 15)) {
            if (argv[0][15] == '\0')
                PRINT_HELP_AND_EXIT();
            option.cpu_features = argv[0] + 15;
+            if (!strcmp(option.cpu_features, "+help")) {
+                use_dummy_wasm = true;
+            }
        }
        else if (!strncmp(argv[0], "--opt-level=", 12)) {
            if (argv[0][12] == '\0')
@ -516,7 +536,7 @@ main(int argc, char *argv[])
            PRINT_HELP_AND_EXIT();
    }

-    if (argc == 0 || !out_file_name)
+    if (!use_dummy_wasm && (argc == 0 || !out_file_name))
        PRINT_HELP_AND_EXIT();

    if (!size_level_set) {
@ -544,11 +564,13 @@ main(int argc, char *argv[])
        option.is_sgx_platform = true;
    }

-    wasm_file_name = argv[0];
+    if (!use_dummy_wasm) {
+        wasm_file_name = argv[0];

-    if (!strcmp(wasm_file_name, out_file_name)) {
-        printf("Error: input file and output file are the same");
-        return -1;
+        if (!strcmp(wasm_file_name, out_file_name)) {
+            printf("Error: input file and output file are the same");
+            return -1;
+        }
    }

    memset(&init_args, 0, sizeof(RuntimeInitArgs));
@ -574,10 +596,17 @@ main(int argc, char *argv[])

    bh_print_time("Begin to load wasm file");

-    /* load WASM byte buffer from WASM bin file */
-    if (!(wasm_file =
-              (uint8 *)bh_read_file_to_buffer(wasm_file_name, &wasm_file_size)))
-        goto fail1;
+    if (use_dummy_wasm) {
+        /* load WASM byte buffer from dummy buffer */
+        wasm_file_size = sizeof(dummy_wasm_file);
+        wasm_file = dummy_wasm_file;
+    }
+    else {
+        /* load WASM byte buffer from WASM bin file */
+        if (!(wasm_file = (uint8 *)bh_read_file_to_buffer(wasm_file_name,
+                                                          &wasm_file_size)))
+            goto fail1;
+    }

    if (get_package_type(wasm_file, wasm_file_size) != Wasm_Module_Bytecode) {
        printf("Invalid file type: expected wasm file but got other\n");
@ -659,7 +688,9 @@ fail3:

 fail2:
    /* free the file buffer */
-    wasm_runtime_free(wasm_file);
+    if (!use_dummy_wasm) {
+        wasm_runtime_free(wasm_file);
+    }

 fail1:
 #if BH_HAS_DLFCN