LLVM 19 Support

.github/workflows/basic-ci.yml

@@ -43,7 +43,7 @@ jobs:
       fail-fast: false
       matrix:
         os: [ ubuntu-22.04, ubuntu-24.04 ]
-        llvm-version: [ 14, 18 ]
+        llvm-version: [ 14, 18, 19 ]
         typeart-typegen-legacy: [ 0, 1 ]
         preset:
           - name: ci-thread-safe-safeptr
@@ -65,12 +65,22 @@ jobs:
             os: ubuntu-22.04
           - llvm-version: 18
             typeart-typegen-legacy: 1
+          - llvm-version: 19
+            os: ubuntu-22.04
+          - llvm-version: 19
+            typeart-typegen-legacy: 1
 
     runs-on: ${{ matrix.os }}
 
     steps:
       - uses: actions/checkout@v4
 
+      - name: LLVM apt
+        if: ${{ matrix.llvm-version == 19 }}
+        run: |
+          wget -O - https://apt.llvm.org/llvm-snapshot.gpg.key | sudo apt-key add -
+          echo "deb http://apt.llvm.org/noble/ llvm-toolchain-noble-19 main" | sudo tee /etc/apt/sources.list.d/llvm-19.list
+
       - name: Update apt
         run: sudo apt-get update
 

.github/workflows/ext-ci.yml

@@ -15,7 +15,7 @@ jobs:
       fail-fast: false
       matrix:
         os: [ ubuntu-22.04, ubuntu-24.04 ]
-        llvm-version: [ 14, 18 ]
+        llvm-version: [ 14, 18, 19 ]
         preset:
           - name: release-counter
           - name: release-safeptr-counter
@@ -25,6 +25,8 @@ jobs:
             os: ubuntu-24.04
           - llvm-version: 18
             os: ubuntu-22.04
+          - llvm-version: 19
+            os: ubuntu-22.04
 
     runs-on: ${{ matrix.os }}
 
@@ -38,6 +40,12 @@ jobs:
           ssh-key: ${{ secrets.AUTH_SSH_CI_EXT }}
           path: test-bench
 
+      - name: LLVM apt
+        if: ${{ matrix.llvm-version == 19 }}
+        run: |
+          wget -O - https://apt.llvm.org/llvm-snapshot.gpg.key | sudo apt-key add -
+          echo "deb http://apt.llvm.org/noble/ llvm-toolchain-noble-19 main" | sudo tee /etc/apt/sources.list.d/llvm-19.list
+
       - name: Update apt
         run: sudo apt-get update
 
@@ -106,7 +114,7 @@ jobs:
       fail-fast: false
       matrix:
         os: [ ubuntu-22.04, ubuntu-24.04 ]
-        llvm-version: [ 14, 18 ]
+        llvm-version: [ 14, 18, 19 ]
         preset:
           - name: release-counter
           - name: release-unsafe-counter
@@ -115,6 +123,8 @@ jobs:
             os: ubuntu-24.04
           - llvm-version: 18
             os: ubuntu-22.04
+          - llvm-version: 19
+            os: ubuntu-22.04
 
     runs-on: ${{ matrix.os }}
 
@@ -129,6 +139,12 @@ jobs:
           ref: feat/ci
           path: ad-test-bench
 
+      - name: LLVM apt
+        if: ${{ matrix.llvm-version == 19 }}
+        run: |
+          wget -O - https://apt.llvm.org/llvm-snapshot.gpg.key | sudo apt-key add -
+          echo "deb http://apt.llvm.org/noble/ llvm-toolchain-noble-19 main" | sudo tee /etc/apt/sources.list.d/llvm-19.list
+
       - name: Update apt
         run: sudo apt-get update
 
@@ -184,7 +200,7 @@ jobs:
       fail-fast: false
       matrix:
         os: [ ubuntu-22.04, ubuntu-24.04 ]
-        llvm-version: [ 14, 18 ]
+        llvm-version: [ 14, 18, 19 ]
         preset:
           - name: release-counter
           - name: release-safeptr-counter
@@ -193,6 +209,8 @@ jobs:
             os: ubuntu-24.04
           - llvm-version: 18
             os: ubuntu-22.04
+          - llvm-version: 19
+            os: ubuntu-22.04
 
     runs-on: ${{ matrix.os }}
 
@@ -207,6 +225,12 @@ jobs:
           ref: ci/omp
           path: omp-test-bench
 
+      - name: LLVM apt
+        if: ${{ matrix.llvm-version == 19 }}
+        run: |
+          wget -O - https://apt.llvm.org/llvm-snapshot.gpg.key | sudo apt-key add -
+          echo "deb http://apt.llvm.org/noble/ llvm-toolchain-noble-19 main" | sudo tee /etc/apt/sources.list.d/llvm-19.list
+
       - name: Update apt
         run: sudo apt-get update
 

CMakeLists.txt

@@ -1,7 +1,7 @@
 cmake_minimum_required(VERSION 3.20)
 
 project(typeart
-  VERSION 1.9
+  VERSION 2.0
   HOMEPAGE_URL https://github.com/tudasc/TypeART
   DESCRIPTION "LLVM-based type and memory allocation tracking sanitizer"
 )

externals/dimeta/CMakeLists.txt

@@ -1,7 +1,7 @@
 FetchContent_Declare(
   llvm-dimeta
   GIT_REPOSITORY https://github.com/ahueck/llvm-dimeta
-  GIT_TAG devel
+  GIT_TAG v0.2.0
   GIT_SHALLOW 1
 )
 

lib/passes/instrumentation/MemOpInstrumentation.cpp

@@ -35,6 +35,7 @@
 #include "llvm/IR/Instruction.h"
 #include "llvm/IR/Instructions.h"
 #include "llvm/IR/IntrinsicInst.h"
+#include "llvm/IR/Module.h"
 #include "llvm/IR/Type.h"
 #include "llvm/Support/Casting.h"
 #include "llvm/Support/raw_ostream.h"

lib/passes/typegen/dimeta/DimetaTypeGen.cpp

@@ -54,13 +54,25 @@ template <typename Type>
 std::pair<std::optional<typeart_builtin_type>, int> typeid_if_ptr(const Type& type) {
   using namespace dimeta;
   const auto& quals = type.qual;
-  int count{0};
-  if (!quals.empty()) {
-    count = llvm::count_if(quals, [](auto& qual) { return qual == Qualifier::kPtr || qual == Qualifier::kRef; });
-    if (count > 0) {
-      return {{TYPEART_POINTER}, count};
+  const int count =
+      llvm::count_if(quals, [](auto& qual) { return qual == Qualifier::kPtr || qual == Qualifier::kRef; });
+
+  if constexpr (std::is_same_v<Type, typename dimeta::QualifiedFundamental>) {
+    switch (type.type.encoding) {
+      case FundamentalType::Encoding::kVtablePtr:
+        return {TYPEART_VTABLE_POINTER, count};
+      case FundamentalType::Encoding::kNullptr:
+        return {TYPEART_NULLPOINTER, count};
+      case FundamentalType::Encoding::kVoid:
+        return {TYPEART_VOID, count};
+      default:
+        break;
     }
   }
+  if (count > 0) {
+    return {{TYPEART_POINTER}, count};
+  }
+
   return {{}, count};
 }
 
@@ -220,6 +232,8 @@ std::optional<typeart_builtin_type> get_builtin_typeid(const dimeta::QualifiedFu
   const auto encoding = type.type.encoding;
 
   switch (encoding) {
+    case FundamentalType::Encoding::kVtablePtr:
+      return TYPEART_VTABLE_POINTER;
     case FundamentalType::Encoding::kUnknown:
       return TYPEART_UNKNOWN_TYPE;
     case FundamentalType::Encoding::kVoid:
@@ -279,11 +293,11 @@ std::optional<typeart_builtin_type> get_builtin_typeid(const dimeta::QualifiedFu
     }
     case FundamentalType::Encoding::kComplex: {
       switch (extent) {
-        case 2:
+        case 8:
           return TYPEART_COMPLEX_64;
-        case 4:
+        case 16:
           return TYPEART_COMPLEX_128;
-        case 1:
+        case 32:
           return TYPEART_COMPLEX_256;
         default:
           return TYPEART_UNKNOWN_TYPE;
@@ -398,17 +412,31 @@ class DimetaTypeManager final : public TypeIDGenerator {
     if (q.is_forward_decl) {
       struct_info.flag = StructTypeFlag::FWD_DECL;
     } else {
-      struct_info.flag = StructTypeFlag::USER_DEFINED;
+      if (q.type.type == CompoundType::Tag::kUnion) {
+        struct_info.flag = StructTypeFlag::UNION;
+      } else {
+        struct_info.flag = StructTypeFlag::USER_DEFINED;
+      }
     }
 
-    struct_info.extent      = compound.extent;
-    struct_info.offsets     = compound.offsets;
-    struct_info.num_members = compound.bases.size() + compound.members.size();
+    struct_info.extent = compound.extent;
 
+    size_t num_bases{0};
     for (const auto& base : compound.bases) {
+      if (base->is_empty_base_class) {
+        continue;
+      }
+      num_bases++;
       struct_info.member_types.push_back(getOrRegister(base->base));
       struct_info.array_sizes.push_back(array_size(base->base));
+      struct_info.offsets.push_back(base->offset);
     }
+
+    struct_info.num_members = compound.members.size() + num_bases;
+    struct_info.offsets.insert(std::end(struct_info.offsets),  //
+                               std::begin(compound.offsets),   //
+                               std::end(compound.offsets));
+
     for (const auto& member : compound.members) {
       struct_info.member_types.push_back(getOrRegister(member->member));
       struct_info.array_sizes.push_back(array_size(member->member));

lib/runtime/RuntimeInterface.h

@@ -267,13 +267,21 @@ bool typeart_is_struct_type(int type_id);
 
 /**
  * Returns true if the type ID is a user-defined structure type
- * (struct, class etc.)
+ * (struct, class, union etc.)
  *
  * \param[in] type_id The type ID.
  * \return true, false
  */
 bool typeart_is_userdefined_type(int type_id);
 
+/**
+ * Returns true if the type ID is a (user-defined) union
+ *
+ * \param[in] type_id The type ID.
+ * \return true, false
+ */
+bool typeart_is_union(int type_id);
+
 /**
  * Returns true if the type ID is a LLVM SIMD vector type
  *

lib/runtime/TypeResolution.cpp

@@ -451,6 +451,11 @@ bool typeart_is_userdefined_type(int type_id) {
   return typeart::RuntimeSystem::get().typeResolution.db().isUserDefinedType(type_id);
 }
 
+bool typeart_is_union(int type_id) {
+  typeart::RTGuard guard;
+  return typeart::RuntimeSystem::get().typeResolution.db().isUnion(type_id);
+}
+
 size_t typeart_get_type_size(int type_id) {
   typeart::RTGuard guard;
   return typeart::RuntimeSystem::get().typeResolution.db().getTypeSize(type_id);

lib/typelib/TypeDB.cpp

@@ -36,6 +36,7 @@ inline constexpr auto size_complex_long_double = sizeof(std::complex<long double
 #define FOR_EACH_TYPEART_BUILTIN(X)                             \
   X(TYPEART_UNKNOWN_TYPE, "typeart_unknown_type", 0)            \
   X(TYPEART_POINTER, "ptr", sizeof(void*))                      \
+  X(TYPEART_VTABLE_POINTER, "vtable_ptr", sizeof(void*))        \
   X(TYPEART_VOID, "void*", sizeof(void*))                       \
   X(TYPEART_NULLPOINTER, "nullptr_t", sizeof(void*))            \
   X(TYPEART_BOOL, "bool", sizeof(bool))                         \
@@ -114,9 +115,11 @@ bool TypeDB::isPointerType(int type_id) const {
 bool TypeDB::isUserDefinedType(int type_id) const {
   const auto* structInfo = getStructInfo(type_id);
   LOG_DEBUG(structInfo->name << " " << static_cast<int>(structInfo->flag) << " "
-                             << (static_cast<int>(structInfo->flag) == static_cast<int>(StructTypeFlag::USER_DEFINED)))
+                             << (static_cast<int>(structInfo->flag) == static_cast<int>(StructTypeFlag::USER_DEFINED))
+                             << " " << (static_cast<int>(structInfo->flag) == static_cast<int>(StructTypeFlag::UNION)))
   return (structInfo != nullptr) &&
-         (static_cast<int>(structInfo->flag) == static_cast<int>(StructTypeFlag::USER_DEFINED));
+         (static_cast<int>(structInfo->flag) == static_cast<int>(StructTypeFlag::USER_DEFINED) ||
+          static_cast<int>(structInfo->flag) == static_cast<int>(StructTypeFlag::UNION));
 }
 
 bool TypeDB::isVectorType(int type_id) const {
@@ -127,6 +130,13 @@ bool TypeDB::isVectorType(int type_id) const {
          (static_cast<int>(structInfo->flag) == static_cast<int>(StructTypeFlag::LLVM_VECTOR));
 }
 
+bool TypeDB::isUnion(int type_id) const {
+  const auto* structInfo = getStructInfo(type_id);
+  LOG_DEBUG(structInfo->name << " " << static_cast<int>(structInfo->flag) << " "
+                             << (static_cast<int>(structInfo->flag) == static_cast<int>(StructTypeFlag::UNION)))
+  return (structInfo != nullptr) && (static_cast<int>(structInfo->flag) == static_cast<int>(StructTypeFlag::UNION));
+}
+
 bool TypeDB::isValid(int type_id) const {
   if (isBuiltinType(type_id)) {
     return true;

lib/typelib/TypeDB.h

@@ -91,6 +91,8 @@ class TypeDB final : public TypeDatabase {
 
   bool isVectorType(int type_id) const override;
 
+  bool isUnion(int type_id) const override;
+
   const std::string& getTypeName(int type_id) const override;
 
   const StructTypeInfo* getStructInfo(int type_id) const override;

lib/typelib/TypeDatabase.h

@@ -22,7 +22,7 @@
 
 namespace typeart {
 
-enum class StructTypeFlag : int { USER_DEFINED = 1, LLVM_VECTOR = 2, FWD_DECL = 4 };
+enum class StructTypeFlag : int { USER_DEFINED = 1, LLVM_VECTOR = 2, FWD_DECL = 4, UNION = 8 };
 
 struct StructTypeInfo {
   int type_id;
@@ -57,6 +57,8 @@ class TypeDatabase {
 
   [[nodiscard]] virtual bool isVectorType(int type_id) const = 0;
 
+  [[nodiscard]] virtual bool isUnion(int type_id) const = 0;
+
   [[nodiscard]] virtual const std::string& getTypeName(int type_id) const = 0;
 
   [[nodiscard]] virtual const StructTypeInfo* getStructInfo(int type_id) const = 0;

lib/typelib/TypeIO.cpp

@@ -54,7 +54,7 @@ struct llvm::yaml::ScalarTraits<typeart::StructTypeFlag> {
     } else {
       value = static_cast<typeart::StructTypeFlag>(flag);
     }
-    return StringRef();
+    return {};
   }
 
   // Determine if this scalar needs quotes.

lib/typelib/TypeInterface.h

@@ -21,6 +21,7 @@ typedef enum typeart_builtin_type_t {  // NOLINT
   TYPEART_UNKNOWN_TYPE = 0,
 
   TYPEART_POINTER,
+  TYPEART_VTABLE_POINTER,
   TYPEART_VOID,
   TYPEART_NULLPOINTER,
 

test/pass/arrays/01_simple_array.c

@@ -15,6 +15,6 @@ void test() {
 
 // ALLOC: [[POINTER:%[0-9a-z]+]] = alloca [100 x i32]
 // ALLOC-NEXT: [[POINTER2:%[0-9a-z]+]] = bitcast [100 x i32]* [[POINTER]] to i8*
-// ALLOC-NEXT: call void @__typeart_alloc_stack(i8* [[POINTER2]], i32 12, i64 100)
+// ALLOC-NEXT: call void @__typeart_alloc_stack(i8* [[POINTER2]], i32 13, i64 100)
 
 // CHECK: call void @__typeart_leave_scope(i32 %__ta_counter_load)

test/pass/arrays/02_array_to_pointer.c

@@ -17,6 +17,6 @@ void test() {
 
 // CHECK: [[POINTER:%[0-9a-z]+]] = alloca [100 x i32]
 // CHECK-NEXT: [[POINTER2:%[0-9a-z]+]] = bitcast [100 x i32]* [[POINTER]] to i8*
-// CHECK-NEXT: call void @__typeart_alloc_stack(i8* [[POINTER2]], i32 12, i64 100)
+// CHECK-NEXT: call void @__typeart_alloc_stack(i8* [[POINTER2]], i32 13, i64 100)
 
 // CHECK: call void @__typeart_leave_scope(i32 %__ta_counter_load)

test/pass/arrays/03_multidim_array.c

@@ -14,7 +14,7 @@ void test() {
 // CHECK: Free{{[ ]*}}:{{[ ]*}}0
 // CHECK: Alloca{{[ ]*}}:{{[ ]*}}6
 
-// CHECK: call void @__typeart_alloc_stack({{i8\*|ptr}} %{{[0-9a-z]+}}, i32 12, i64 64)
-// CHECK: call void @__typeart_alloc_stack({{i8\*|ptr}} %{{[0-9a-z]+}}, i32 12, i64 4096)
-// CHECK: call void @__typeart_alloc_stack({{i8\*|ptr}} %{{[0-9a-z]+}}, i32 12, i64 8192)
-// CHECK: call void @__typeart_alloc_stack({{i8\*|ptr}} %{{[0-9a-z]+}}, i32 12, i64 524288)
+// CHECK: call void @__typeart_alloc_stack({{i8\*|ptr}} %{{[0-9a-z]+}}, i32 13, i64 64)
+// CHECK: call void @__typeart_alloc_stack({{i8\*|ptr}} %{{[0-9a-z]+}}, i32 13, i64 4096)
+// CHECK: call void @__typeart_alloc_stack({{i8\*|ptr}} %{{[0-9a-z]+}}, i32 13, i64 8192)
+// CHECK: call void @__typeart_alloc_stack({{i8\*|ptr}} %{{[0-9a-z]+}}, i32 13, i64 524288)

test/pass/arrays/06_malloc_of_array.c

@@ -6,4 +6,4 @@ void foo(int n) {
   int(*array)[3] = malloc(2 * sizeof(int[3]));
 }
 
-// CHECK: @__typeart_alloc({{i8\*|ptr}} %{{[0-9a-z]+}}, i32 12, i64 6)
+// CHECK: @__typeart_alloc({{i8\*|ptr}} %{{[0-9a-z]+}}, i32 13, i64 6)