chore: upgrade clang-format and clang-tidy version

Author: XuehaiPan

.clang-tidy

@@ -1,6 +1,5 @@
 ---
 InheritParentConfig: true
-ExtraArgs: ['-v']
 FormatStyle: file
 UseColor: true
 WarningsAsErrors: '*'

.github/workflows/ci.yml

@@ -226,6 +226,7 @@ jobs:
             echo "PIP_EXTRA_INDEX_URL=${PIP_EXTRA_INDEX_URL}" | tee -a "${GITHUB_ENV}"
             echo "UV_INDEX=${UV_INDEX}" | tee -a "${GITHUB_ENV}"
           fi
+
           export CLANG_TIDY_CMAKE_OPTIONS="${CLANG_TIDY_CMAKE_OPTIONS} -DUSE_METAL=ON"
 
           echo "USE_METAL=ON" | tee -a "${GITHUB_ENV}"
@@ -289,21 +290,33 @@ jobs:
 
       - name: Run clang-tidy
         id: clang-tidy
+        if: runner.os == 'Linux'
         run: |
-          clang-tidy --version
+          echo "\$ $(command -v clang-tidy) --version" && clang-tidy --version
+
+          if [[ -x "$(command -v run-clang-tidy)" ]]; then
+            echo "Using run-clang-tidy from $(command -v run-clang-tidy)"
+            CLANG_TIDY=(run-clang-tidy)
+          else
+            echo "Downloading run-clang-tidy script"
+            wget -O run-clang-tidy.py https://raw.githubusercontent.com/llvm/llvm-project/refs/heads/release/21.x/clang-tools-extra/clang-tidy/tool/run-clang-tidy.py
+            CLANG_TIDY=(uv run --no-project --script -- run-clang-tidy.py)
+          fi
+          if [[ -x "$(command -v clang-apply-replacements)" ]]; then
+            echo "Using clang-apply-replacements from $(command -v clang-apply-replacements)"
+            CLANG_TIDY+=(-fix -clang-apply-replacements-binary="$(command -v clang-apply-replacements)")
+          else
+            echo "::warning::clang-apply-replacements not found in PATH, automatic fixing disabled."
+          fi
 
           # Run cmake to create the build directory with compile_commands.json
           cmake -S . -B cmake-build --fresh ${CLANG_TIDY_CMAKE_OPTIONS}  # no quotes here
 
           CXX_FILES=$(find src -type f -iname "*.[ch]pp" -o -iname "*.cc" -o -iname "*.c" -o -iname "*.h")
           rc=0
-          if [[ -x "$(command -v run-clang-tidy)" ]]; then
-            run-clang-tidy -clang-tidy-binary="$(command -v clang-tidy)" \
-              -fix -p="cmake-build" ${CXX_FILES} || rc="$?"
-          else
-            clang-tidy --fix -p="cmake-build" ${CXX_FILES} || rc="$?"
-          fi
-          rm -rf cmake-build
+          "${CLANG_TIDY[@]}" -clang-tidy-binary="$(command -v clang-tidy)" \
+            -p="cmake-build" ${CXX_FILES} || rc="$?"
+          rm -rf cmake-build run-clang-tidy.py
           if (( rc != 0 )); then
             echo "::error::clang-tidy found issues (exit code: ${rc}). Please run 'clang-tidy --fix' locally to fix them."
             git diff --color=always || true

.pre-commit-config.yaml

@@ -32,7 +32,7 @@ repos:
         args: [--ignore-case]
         files: ^docs/spelling_wordlist\.txt$
   - repo: https://github.com/pre-commit/mirrors-clang-format
-    rev: v15.0.7  # sync with requirements-lint.txt
+    rev: v21.1.2  # sync with requirements-lint.txt
     hooks:
       - id: clang-format
         exclude: |

requirements-lint.txt

@@ -1,7 +1,7 @@
 # Format and lint requirements
 pre-commit
-clang-format==15.0.7
-clang-tidy==18.1.8
+clang-format==21.1.2
+clang-tidy==21.1.1
 codespell[toml]==2.4.1
 ruff==0.14.0
 yapf==0.43.0

src/op/gemm.cc

@@ -27,9 +27,7 @@ static inline std::pair<bool, TCGEN5MMAMeta>
 GetTCGEN5MMAMeta(int M, int N, int K, DataType ab_dtype, DataType c_dtype) {
 // TODO (lei) Currently not all shapes / dtypes are supported for TCGEN5MMA.
 #define FAIL                                                                   \
-  return {                                                                     \
-    false, TCGEN5MMAMeta { 0, 0, 0 }                                           \
-  }
+  return { false, TCGEN5MMAMeta{0, 0, 0} }
 #define SUCCESS(atom_m, atom_n, atom_k)                                        \
   return {                                                                     \
     true, TCGEN5MMAMeta { atom_m, atom_n, atom_k }                             \

src/op/parallel.h

@@ -42,7 +42,7 @@ class ParallelOpNode;
 
 class ParallelLoopNestVisitor : public StmtExprVisitor {
 private:
-  ParallelLoopNestVisitor(ParallelOpNode *op) : p(op){};
+  ParallelLoopNestVisitor(ParallelOpNode *op) : p(op) {};
   void VisitStmt_(const ForNode *op) override;
   void VisitStmt_(const BufferStoreNode *op) override;
   void VisitExpr_(const BufferLoadNode *op) override;

src/target/codegen_cuda.cc

@@ -1749,8 +1749,8 @@ void CodeGenTileLangCUDA::VisitExpr_(const CallNode *op, std::ostream &os) {
       os << "}\n";
     } else {
       os << "for (int local_id = 0; local_id < 8; ++local_id) {\n";
-      os << dst << "[" + this->PrintExpr(dst_ind) + "]"
-         << " = " << src << "[" << src_offset << " + local_id];\n";
+      os << dst << "[" + this->PrintExpr(dst_ind) + "]" << " = " << src << "["
+         << src_offset << " + local_id];\n";
       os << "}\n";
     }
 

src/target/cuda.h

@@ -5023,12 +5023,12 @@ typedef struct CUgraphNodeParams_st {
 /**
  * Device that represents the CPU
  */
-#define CU_DEVICE_CPU ((CUdevice)-1)
+#define CU_DEVICE_CPU ((CUdevice) - 1)
 
 /**
  * Device that represents an invalid device
  */
-#define CU_DEVICE_INVALID ((CUdevice)-2)
+#define CU_DEVICE_INVALID ((CUdevice) - 2)
 
 /**
  * Bitmasks for ::CU_DEVICE_ATTRIBUTE_GPU_DIRECT_RDMA_FLUSH_WRITES_OPTIONS

src/tl_templates/cpp/half.hpp

@@ -1192,8 +1192,8 @@ unsigned int float2half_impl(T value, ...) {
 template <std::float_round_style R, typename T>
 unsigned int float2half(T value) {
   return float2half_impl<R>(
-      value, bool_type < std::numeric_limits<T>::is_iec559 &&
-                 sizeof(typename bits<T>::type) == sizeof(T) > ());
+      value, bool_type<std::numeric_limits<T>::is_iec559 &&
+                       sizeof(typename bits<T>::type) == sizeof(T)>());
 }
 
 /// Convert integer to half-precision floating-point.
@@ -1665,9 +1665,10 @@ template <typename T> T half2float_impl(unsigned int value, T, ...) {
 /// \param value half-precision value to convert
 /// \return floating-point value
 template <typename T> T half2float(unsigned int value) {
-  return half2float_impl(value, T(),
-                         bool_type < std::numeric_limits<T>::is_iec559 &&
-                             sizeof(typename bits<T>::type) == sizeof(T) > ());
+  return half2float_impl(
+      value, T(),
+      bool_type<std::numeric_limits<T>::is_iec559 &&
+                sizeof(typename bits<T>::type) == sizeof(T)>());
 }
 
 /// Convert half-precision floating-point to integer.

src/tl_templates/cuda/common.h

@@ -244,8 +244,8 @@ union GmmaDescriptor {
     uint16_t stride_byte_offset_ : 14, : 2; // 14 bits [0,14), 2 bits unused
     // base_offset, bit [49,52)
     // Valid only for SWIZZLE_128B and SWIZZLE_64B
-    uint8_t : 1,
-        base_offset_ : 3, : 4; // 1 bit unused, 3 bits [1,4), 4 bits unused
+    uint8_t : 1, base_offset_ : 3,
+        : 4; // 1 bit unused, 3 bits [1,4), 4 bits unused
     // layout type, bit [62,64)
     // SWIZZLE_NONE = 0, SWIZZLE_32B = 3, SWIZZLE_64B = 2, SWIZZLE_128B = 1
     uint8_t : 6, layout_type_ : 2; // 6 bits unused, 2 bits [6,8)