Vectorized scanning for ARM Neon

.github/workflows/benchmark.yml

@@ -0,0 +1,41 @@
+name: Benchmark
+
+on: workflow_dispatch
+
+jobs:
+  windows:
+    strategy:
+      matrix:
+        toolset: [ v145, ClangCL ]
+    runs-on: windows-11-vs2026-arm
+    steps:
+      - uses: actions/checkout@v6
+
+      - name: CPM Cache
+        uses: actions/cache@v5
+        with:
+          path: ${{github.workspace}}/.cpmcache
+          key: cpm-${{runner.os}}-ARM64-${{hashFiles('test/CMakeLists.txt')}}
+          restore-keys: |
+            cpm-${{runner.os}}-ARM64-
+
+      - name: Check Hardware
+        run: (Get-CimInstance Win32_Processor).Name
+
+      - name: Locate Visual Studio
+        run: |
+          $vsInstall = & "${env:ProgramFiles(x86)}\Microsoft Visual Studio\Installer\vswhere.exe" -latest -property installationPath
+          echo "VS_INSTALL_DIR=$vsInstall" >> $env:GITHUB_ENV
+
+      - name: Configure
+        run: cmake -B ${{github.workspace}}/build -DCMAKE_CXX_STANDARD=23 -DLIBHAT_SHARED_C_LIB=ON -DLIBHAT_TESTING_SDE=OFF -A ARM64 -T ${{matrix.toolset}}
+
+      - name: Build
+        run: cmake --build ${{github.workspace}}/build -j 4 --config Release --target libhat_benchmark_compare_impl libhat_benchmark_chromium
+
+      - name: Test
+        working-directory: ${{github.workspace}}/build
+        shell: cmd
+        run: |
+          call "${{env.VS_INSTALL_DIR}}\VC\Auxiliary\Build\vcvarsarm64.bat"
+          ctest --verbose -C Release -R "(libhat_benchmark_compare_impl|libhat_benchmark_chromium)"

.github/workflows/cmake.yml

@@ -18,20 +18,38 @@ jobs:
     steps:
       - uses: actions/checkout@v6
 
+      - name: CPM Cache
+        uses: actions/cache@v5
+        with:
+          path: ${{github.workspace}}/.cpmcache
+          key: cpm-${{runner.os}}-${{hashFiles('test/CMakeLists.txt')}}
+          restore-keys: |
+            cpm-${{runner.os}}-
+
       - name: Configure
-        run: cmake -B ${{github.workspace}}/build -DCMAKE_CXX_STANDARD=${{matrix.cxx_standard}} -DCMAKE_BUILD_TYPE=${{env.BUILD_TYPE}} -DLIBHAT_TESTING=OFF
+        run: cmake -B ${{github.workspace}}/build -DCPM_SOURCE_CACHE=${{github.workspace}}/.cpmcache -DCMAKE_CXX_STANDARD=${{matrix.cxx_standard}} -DCMAKE_BUILD_TYPE=${{env.BUILD_TYPE}} -DLIBHAT_TESTING_SAMPLE_BIN=OFF -DLIBHAT_TESTING_SDE=OFF
 
       - name: Build
         run: cmake --build ${{github.workspace}}/build -j 4
 
+      - name: Test
+        working-directory: ${{github.workspace}}/build
+        run: ctest --verbose -C ${{env.BUILD_TYPE}} -R libhat_test_.*
+
   linux:
     strategy:
       matrix:
+        target: [ x64, ARM64 ]
+        cxx_standard: [ 20, 23 ]
         compiler:
           - { pkg: g++, exe: g++, version: 14 }
           - { pkg: clang, exe: clang++, version: 18 }
-        cxx_standard: [ 20, 23 ]
-    runs-on: ubuntu-24.04
+        include:
+          - target: x64
+            os: ubuntu-26.04
+          - target: ARM64
+            os: ubuntu-26.04-arm
+    runs-on: ${{matrix.os}}
     steps:
       - uses: actions/checkout@v6
 
@@ -51,7 +69,7 @@ jobs:
       - name: Configure
         env:
           CXX: ${{matrix.compiler.exe}}-${{matrix.compiler.version}}
-        run: cmake -B ${{github.workspace}}/build -DCPM_SOURCE_CACHE=${{github.workspace}}/.cpmcache -DCMAKE_CXX_STANDARD=${{matrix.cxx_standard}} -DCMAKE_BUILD_TYPE=${{env.BUILD_TYPE}} -DLIBHAT_TESTING=ON -DLIBHAT_TESTING_SAMPLE_BIN=OFF -DLIBHAT_TESTING_SANITIZE=OFF
+        run: cmake -B ${{github.workspace}}/build -DCPM_SOURCE_CACHE=${{github.workspace}}/.cpmcache -DCMAKE_CXX_STANDARD=${{matrix.cxx_standard}} -DCMAKE_BUILD_TYPE=${{env.BUILD_TYPE}} -DLIBHAT_TESTING_SDE=${{startsWith(matrix.target, 'ARM') && 'OFF' || 'ON'}} -DLIBHAT_TESTING_SAMPLE_BIN=OFF -DLIBHAT_TESTING_SANITIZE=OFF
 
       - name: Build
         run: cmake --build ${{github.workspace}}/build -j 4
@@ -60,6 +78,57 @@ jobs:
         working-directory: ${{github.workspace}}/build
         run: ctest --verbose -C ${{env.BUILD_TYPE}} -R libhat_test_.*
 
+  android:
+    strategy:
+      matrix:
+        cxx_standard: [ 20 ]
+        abi: [ armeabi-v7a, arm64-v8a, x86, x86_64 ]
+        sdk: [ 24 ]
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v6
+
+      - name: AVD Cache
+        if: ${{!startsWith(matrix.abi, 'arm')}}
+        uses: actions/cache@v5
+        id: avd-cache
+        with:
+          path: |
+            ~/.android/avd/*
+            ~/.android/adb*
+          key: avd-${{matrix.sdk}}-${{matrix.abi}}
+
+      - name: Configure
+        run: cmake -B ${{github.workspace}}/build -DCMAKE_TOOLCHAIN_FILE=$ANDROID_NDK/build/cmake/android.toolchain.cmake -DANDROID_ABI=${{matrix.abi}} -DANDROID_PLATFORM=${{matrix.sdk}} -DCPM_SOURCE_CACHE=${{github.workspace}}/.cpmcache -DCMAKE_CXX_STANDARD=${{matrix.cxx_standard}} -DCMAKE_BUILD_TYPE=${{env.BUILD_TYPE}} -DLIBHAT_TESTING_SAMPLE_BIN=OFF -DLIBHAT_TESTING_SDE=OFF -DLIBHAT_TESTING_SANITIZE=OFF
+
+      - name: Build
+        run: cmake --build ${{github.workspace}}/build -j 4
+
+      - name: Create AVD and generate snapshot for caching
+        if: ${{!startsWith(matrix.abi, 'arm') && steps.avd-cache.outputs.cache-hit != 'true'}}
+        uses: reactivecircus/android-emulator-runner@v2
+        with:
+          api-level: ${{matrix.sdk}}
+          arch: ${{matrix.abi}}
+          force-avd-creation: false
+          emulator-options: -no-window -gpu swiftshader_indirect -noaudio -no-boot-anim -camera-back none
+          disable-animations: false
+          script: echo "Generated AVD snapshot for caching."
+
+      - name: Test
+        if: ${{!startsWith(matrix.abi, 'arm')}}
+        uses: reactivecircus/android-emulator-runner@v2
+        with:
+          api-level: ${{matrix.sdk}}
+          arch: ${{matrix.abi}}
+          force-avd-creation: false
+          emulator-options: -no-snapshot-save -no-window -gpu swiftshader_indirect -noaudio -no-boot-anim -camera-back none
+          disable-animations: true
+          script: |
+            adb push ${{github.workspace}}/build/test /data/local/tmp/
+            adb shell chmod +x /data/local/tmp/test/libhat_test_process
+            adb shell /data/local/tmp/test/libhat_test_process
+
   windows:
     strategy:
       matrix:

CMakeLists.txt

@@ -28,6 +28,7 @@ option(LIBHAT_USE_STD_MODULE "Compile the module target using the std module" OF
 option(LIBHAT_EXAMPLES "Include example targets" ${PROJECT_IS_TOP_LEVEL})
 
 option(LIBHAT_HINT_X86_64 "Enables support for the x86_64 scan hint, requires a small (2KB) data table" ON)
+option(LIBHAT_HINT_AARCH64 "Enables support for the aarch64 scan hint, requires a small (2KB) data table" ON)
 
 if(LIBHAT_TESTING AND LIBHAT_TESTING_SANITIZE)
     if(MSVC)
@@ -42,6 +43,8 @@ set(LIBHAT_SRC
     src/Scanner.cpp
     src/System.cpp
 
+    src/os/mac/Process.cpp
+
     src/os/linux/MemoryProtector.cpp
     src/os/linux/Process.cpp
 
@@ -57,7 +60,9 @@ set(LIBHAT_SRC
     src/arch/x86/AVX512.cpp
     src/arch/x86/System.cpp
 
-    src/arch/arm/System.cpp)
+    src/arch/arm/Neon.cpp
+    src/arch/arm/System.cpp
+)
 
 add_library(libhat STATIC ${LIBHAT_SRC})
 add_library(libhat::libhat ALIAS libhat)
@@ -75,10 +80,7 @@ if (MSVC)
         target_compile_options(libhat PRIVATE /W3 /WX)
     endif()
 elseif(CMAKE_CXX_COMPILER_ID MATCHES "GNU|Clang")
-    target_compile_options(libhat PRIVATE -Wall -Wextra -Wconversion -Werror
-            # temp fix for macOS CI failing due to incorrect LIBHAT_COMPILER_X86_OPTIONS value
-            -Wno-unused-command-line-argument
-    )
+    target_compile_options(libhat PRIVATE -Wall -Wextra -Wconversion -Werror)
 endif()
 
 target_include_directories(libhat PUBLIC
@@ -90,6 +92,7 @@ target_compile_definitions(libhat PUBLIC
     "$<$<BOOL:${LIBHAT_DISABLE_SSE}>:LIBHAT_DISABLE_SSE>"
     "$<$<BOOL:${LIBHAT_DISABLE_AVX512}>:LIBHAT_DISABLE_AVX512>"
     "$<$<BOOL:${LIBHAT_HINT_X86_64}>:LIBHAT_HINT_X86_64>"
+    "$<$<BOOL:${LIBHAT_HINT_AARCH64}>:LIBHAT_HINT_AARCH64>"
 )
 
 if(LIBHAT_STATIC_C_LIB OR LIBHAT_SHARED_C_LIB)

README.md

@@ -2,14 +2,15 @@
 A modern, high-performance library for C++20 designed around game hacking
 
 ## Feature overview
-- Windows x86/x64 support
-- Partial Linux and macOS support
 - Vectorized scanning for byte patterns
   - SSE 4.1 and AVX2 on x86/x64
   - AVX-512 on x64
+  - Neon on ARM/ARM64
 - RAII memory protector
 - Convenience wrappers over OS APIs
 - Language bindings (C, C#, Java)
+- Full Windows support
+- Partial (WIP) Linux, macOS, and Android support
 
 ## Versioning
 This project adheres to [semantic versioning](https://semver.org/spec/v2.0.0.html). Any declaration that
@@ -100,19 +101,21 @@ BM_Throughput_UC2/256MiB                 261157833 ns    261160714 ns
 
 Below is a summary of the current support for libhat's platform-dependent APIs:
 
-|                                | Windows | Linux | macOS |
-|--------------------------------|:-------:|:-----:|:-----:|
-| `hat::get_system`              |    ✅    |   ✅   |   ✅   |
-| `hat::memory_protector`        |    ✅    |   ✅   |       |
-| `hp::get_process_module`       |    ✅    |   ✅   |       |
-| `hp::get_module`               |    ✅    |   ✅   |       |
-| `hp::module_at`                |    ✅    |       |       |
-| `hp::is_readable`              |    ✅    |   ✅   |       |
-| `hp::is_writable`              |    ✅    |   ✅   |       |
-| `hp::is_executable`            |    ✅    |   ✅   |       |
-| `hp::module::get_module_data`  |    ✅    |   ✅   |       |
-| `hp::module::get_section_data` |    ✅    |       |       |
-| `hp::module::for_each_segment` |    ✅    |   ✅   |       |
+### APIs
+
+|                                | Windows | Linux | macOS | Android |
+|--------------------------------|:-------:|:-----:|:-----:|:-------:|
+| `hat::get_system`              |    ✅    |   ✅   |   ✅   |    ✅    |
+| `hat::memory_protector`        |    ✅    |   ✅   |       |    ✅    |
+| `hp::get_process_module`       |    ✅    |   ✅   |   ✅   |    ✅    |
+| `hp::get_module`               |    ✅    |   ✅   |   ✅   |    ✅    |
+| `hp::module_at`                |    ✅    |       |       |         |
+| `hp::is_readable`              |    ✅    |   ✅   |       |    ✅    |
+| `hp::is_writable`              |    ✅    |   ✅   |       |    ✅    |
+| `hp::is_executable`            |    ✅    |   ✅   |       |    ✅    |
+| `hp::module::get_module_data`  |    ✅    |   ✅   |       |    ✅    |
+| `hp::module::get_section_data` |    ✅    |       |       |         |
+| `hp::module::for_each_segment` |    ✅    |   ✅   |   ✅   |    ✅    |
 
 ## Quick start
 ### Defining patterns
@@ -189,19 +192,24 @@ const std::byte* address = result.get();
 const std::byte* relative_address = result.rel(3);
 ```
 
-libhat has a few optimizations for searching for patterns in `x86_64` machine code:
+libhat has a few optimizations for searching for patterns in `x86_64` and `AArch64` machine code:
 ```cpp
 #include <libhat/scanner.hpp>
 
-// If a byte pattern matches at the start of a function, the result will be aligned on 16-bytes.
-// This can be indicated via the defaulted `alignment` parameter (all overloads have this parameter):
+// Compilers will often align the start address of a function on 16-bytes. Scanning for patterns that
+// match the start of a function can take advantage of this by specifying the defaulted `alignment`
+// parameter (all overloads have this parameter):
 std::span<std::byte> range   = /* ... */;
 hat::signature_view  pattern = /* ... */;
 hat::scan_result     result  = hat::find_pattern(range, pattern, hat::scan_alignment::X16);
 
-// Additionally, x86_64 contains a non-uniform distribution of byte pairs. By passing the `x86_64`
-// scan hint, the search can be based on the least common byte pair that is found in the pattern.
-hat::scan_result result  = hat::find_pattern(range, pattern, hat::scan_alignment::X1, hat::scan_hint::x86_64);
+// Or, if the architecture has byte-aligned instructions (such as ARM and AArch64):
+hat::scan_result result = hat::find_pattern(range, pattern, hat::scan_alignment::X4);
+
+// Additionally, machine code contains a non-uniform distribution of bytes. By passing the respective
+// scan hint (either `x86_64` or `aarch64`), the search anchor can be tuned to the least frequent
+// bytes that are present in the pattern.
+hat::scan_result result = hat::find_pattern(range, pattern, hat::scan_alignment::X1, hat::scan_hint::x86_64);
 ```
 
 ### Accessing members

bindings/java/src/main/java/me/zero/libhat/Hat.java

@@ -127,7 +127,7 @@ public static OptionalInt findPattern(@NotNull final Signature signature, @NotNu
             Objects.requireNonNull(signature.handle),
             new Pointer(start),
             count,
-            alignment.ordinal()
+            alignment.alignment()
         );
 
         if (result == Pointer.NULL) {
@@ -175,7 +175,7 @@ public static Optional<Pointer> findPattern(@NotNull final Signature signature,
             Objects.requireNonNull(signature.handle),
             module.handle,
             section,
-            alignment.ordinal()
+            alignment.alignment()
         );
 
         return Optional.ofNullable(result);

bindings/java/src/main/java/me/zero/libhat/ScanAlignment.java

@@ -7,10 +7,28 @@ public enum ScanAlignment {
     /**
      * No byte alignment
      */
-    X1,
+    X1(1),
+
+    /**
+     * 4 byte alignment
+     */
+    X4(4),
 
     /**
      * 16 byte alignment
      */
-    X16
+    X16(16);
+
+    private final int alignment;
+
+    ScanAlignment(final int alignment) {
+        this.alignment = alignment;
+    }
+
+    /**
+     * @return The scan result alignment requirement, in bytes
+     */
+    public final int alignment() {
+        return this.alignment;
+    }
 }

include/libhat/c/libhat.h

@@ -37,8 +37,9 @@ typedef enum libhat_status_t {
 } libhat_status_t;
 
 typedef enum scan_alignment {
-    scan_alignment_x1,
-    scan_alignment_x16,
+    scan_alignment_x1 = 1,
+    scan_alignment_x4 = 4,
+    scan_alignment_x16 = 16,
 } scan_alignment_t;
 
 typedef struct signature {

include/libhat/platform.h

@@ -13,12 +13,21 @@
     #error Unsupported Architecture
 #endif
 
+#if defined(__LP64__) || defined(_M_X64)
+    #define LIBHAT_LP64
+#else
+    #define LIBHAT_LP32
+#endif
+
 // Detect Operating System
 #if defined(_WIN32)
     #define LIBHAT_WINDOWS
 #elif defined(linux) || defined(__linux__) || defined(__linux)
     #define LIBHAT_UNIX
     #define LIBHAT_LINUX
+    #if defined(__ANDROID__)
+        #define LIBHAT_ANDROID
+    #endif
 #elif defined(__APPLE__) && defined(__MACH__)
     #define LIBHAT_UNIX
     #define LIBHAT_MAC