Merge remote-tracking branch 'origin/main' into disable-block-signatu…

…re-string

.github/CODEOWNERS

@@ -120,6 +120,9 @@ clang/test/AST/Interp/ @tbaederr
 /mlir/**/LLVMIR/**/BasicPtxBuilderInterface* @grypp
 /mlir/**/NVVM* @grypp
 
+# MLIR Index Dialect
+/mlir/**/Index* @mogball
+
 # MLIR Python Bindings
 /mlir/test/python/ @ftynse @makslevental @stellaraccident
 /mlir/python/ @ftynse @makslevental @stellaraccident

.github/new-prs-labeler.yml

@@ -775,6 +775,29 @@ backend:AArch64:
   - clang/include/clang/Sema/SemaARM.h
   - clang/lib/Sema/SemaARM.cpp
 
+backend:Hexagon:
+  - clang/include/clang/Basic/BuiltinsHexagon*.def
+  - clang/include/clang/Sema/SemaHexagon.h
+  - clang/lib/Basic/Targets/Hexagon.*
+  - clang/lib/CodeGen/Targets/Hexagon.cpp
+  - clang/lib/Driver/ToolChains/Hexagon.*
+  - clang/lib/Sema/SemaHexagon.cpp
+  - lld/ELF/Arch/Hexagon.cpp
+  - lldb/source/Plugins/ABI/Hexagon/**
+  - lldb/source/Plugins/DynamicLoader/Hexagon-DYLD/**
+  - llvm/include/llvm/BinaryFormat/ELFRelocs/Hexagon.def
+  - llvm/include/llvm/IR/IntrinsicsHexagon*
+  - llvm/include/llvm/Support/Hexagon*
+  - llvm/lib/Support/Hexagon*
+  - llvm/lib/Target/Hexagon/**
+  - llvm/test/CodeGen/Hexagon/**
+  - llvm/test/CodeGen/*/Hexagon/**
+  - llvm/test/DebugInfo/*/Hexagon/**
+  - llvm/test/Transforms/*/Hexagon
+  - llvm/test/MC/Disassembler/Hexagon/**
+  - llvm/test/MC/Hexagon/**
+  - llvm/test/tools/llvm-objdump/ELF/Hexagon/**
+
 backend:loongarch:
   - llvm/include/llvm/IR/IntrinsicsLoongArch.td
   - llvm/test/MC/LoongArch/**

.github/workflows/issue-write.yml

@@ -5,6 +5,7 @@ on:
     workflows:
       - "Check code formatting"
       - "Check for private emails used in PRs"
+      - "PR Request Release Note"
     types:
       - completed
 
@@ -17,7 +18,11 @@ jobs:
     permissions:
       pull-requests: write
     if: >
-      github.event.workflow_run.event == 'pull_request'
+      github.event.workflow_run.event == 'pull_request' &&
+      (
+        github.event.workflow_run.conclusion == 'success' ||
+        github.event.workflow_run.conclusion == 'failure'
+      )
     steps:
       - name: 'Download artifact'
         uses: actions/download-artifact@6b208ae046db98c579e8a3aa621ab581ff575935 # v4.1.1
@@ -92,7 +97,11 @@ jobs:
 
             var pr_number = 0;
             gql_result.repository.ref.associatedPullRequests.nodes.forEach((pr) => {
-              if (pr.baseRepository.owner.login = context.repo.owner && pr.state == 'OPEN') {
+
+              // The largest PR number is the one we care about.  The only way
+              // to have more than one associated pull requests is if all the
+              // old pull requests are in the closed state.
+              if (pr.baseRepository.owner.login = context.repo.owner && pr.number > pr_number) {
                 pr_number = pr.number;
               }
             });

.github/workflows/libcxx-build-and-test.yaml

@@ -63,8 +63,8 @@ jobs:
         cxx: [ 'clang++-19' ]
         include:
           - config: 'generic-gcc'
-            cc: 'gcc-13'
-            cxx: 'g++-13'
+            cc: 'gcc-14'
+            cxx: 'g++-14'
     steps:
       - uses: actions/checkout@v4
       - name: ${{ matrix.config }}.${{ matrix.cxx }}
@@ -101,8 +101,8 @@ jobs:
         cxx: [ 'clang++-19' ]
         include:
           - config: 'generic-gcc-cxx11'
-            cc: 'gcc-13'
-            cxx: 'g++-13'
+            cc: 'gcc-14'
+            cxx: 'g++-14'
           - config: 'generic-cxx23'
             cc: 'clang-17'
             cxx: 'clang++-17'

.github/workflows/pr-request-release-note.yml

@@ -2,7 +2,6 @@ name: PR Request Release Note
 
 permissions:
   contents: read
-  pull-requests: write
 
 on:
   pull_request:
@@ -41,3 +40,10 @@ jobs:
             --token "$GITHUB_TOKEN" \
             request-release-note \
             --pr-number ${{ github.event.pull_request.number}}
+
+      - uses: actions/upload-artifact@26f96dfa697d77e81fd5907df203aa23a56210a8 #v4.3.0
+        if: always()
+        with:
+          name: workflow-args
+          path: |
+            comments

bolt/docs/CommandLineArgumentReference.md

@@ -259,6 +259,10 @@
 
   Always use long jumps/nops for Linux kernel static keys
 
+- `--match-profile-with-function-hash`
+
+  Match profile with function hash
+
 - `--max-data-relocations=<uint>`
 
   Maximum number of data relocations to process
@@ -684,6 +688,10 @@
 
   Use a modified clustering algorithm geared towards minimizing branches
 
+- `--name-similarity-function-matching-threshold=<uint>`
+
+  Match functions using namespace and edit distance.
+
 - `--no-inline`
 
   Disable all inlining (overrides other inlining options)

bolt/docs/OptimizingLinux.md

@@ -44,6 +44,11 @@ $ perf2bolt -p perf.data -o perf.fdata vmlinux
 
 Under a high load, `perf.data` should be several gigabytes in size and you should expect the converted `perf.fdata` not to exceed 100 MB.
 
+Profiles collected from multiple workloads could be joined into a single profile using `merge-fdata` utility:
+```bash
+$ merge-fdata perf.1.fdata perf.2.fdata ... perf.<N>.fdata > perf.merged.fdata
+```
+
 Two changes are required for the kernel build. The first one is optional but highly recommended. It introduces a BOLT-reserved space into `vmlinux` code section:
 
 

bolt/include/bolt/Core/BinaryBasicBlock.h

@@ -842,15 +842,6 @@ class BinaryBasicBlock {
   bool analyzeBranch(const MCSymbol *&TBB, const MCSymbol *&FBB,
                      MCInst *&CondBranch, MCInst *&UncondBranch);
 
-  /// Return true if iterator \p I is pointing to the first instruction in
-  /// a pair that could be macro-fused.
-  bool isMacroOpFusionPair(const_iterator I) const;
-
-  /// If the basic block has a pair of instructions suitable for macro-fusion,
-  /// return iterator to the first instruction of the pair.
-  /// Otherwise return end().
-  const_iterator getMacroOpFusionPair() const;
-
   /// Printer required for printing dominator trees.
   void printAsOperand(raw_ostream &OS, bool PrintType = true) {
     if (PrintType)

bolt/include/bolt/Core/BinaryContext.h

@@ -698,10 +698,6 @@ class BinaryContext {
 
   /// Binary-wide aggregated stats.
   struct BinaryStats {
-    /// Stats for macro-fusion.
-    uint64_t MissedMacroFusionPairs{0};
-    uint64_t MissedMacroFusionExecCount{0};
-
     /// Stats for stale profile matching:
     ///   the total number of basic blocks in the profile
     uint32_t NumStaleBlocks{0};

bolt/include/bolt/Core/BinaryFunction.h

@@ -835,10 +835,6 @@ class BinaryFunction {
   /// them.
   void calculateLoopInfo();
 
-  /// Calculate missed macro-fusion opportunities and update BinaryContext
-  /// stats.
-  void calculateMacroOpFusionStats();
-
   /// Returns if BinaryDominatorTree has been constructed for this function.
   bool hasDomTree() const { return BDT != nullptr; }
 
@@ -930,6 +926,10 @@ class BinaryFunction {
     return const_cast<BinaryFunction *>(this)->getInstructionAtOffset(Offset);
   }
 
+  /// When the function is in disassembled state, return an instruction that
+  /// contains the \p Offset.
+  MCInst *getInstructionContainingOffset(uint64_t Offset);
+
   std::optional<MCInst> disassembleInstructionAtOffset(uint64_t Offset) const;
 
   /// Return offset for the first instruction. If there is data at the

bolt/include/bolt/Passes/BinaryFunctionCallGraph.h → bolt/include/bolt/Core/BinaryFunctionCallGraph.h

@@ -1,4 +1,4 @@
-//===- bolt/Passes/CallGraph.h ----------------------------------*- C++ -*-===//
+//===- bolt/Core/CallGraph.h ----------------------------------*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
@@ -9,7 +9,7 @@
 #ifndef BOLT_PASSES_BINARY_FUNCTION_CALLGRAPH_H
 #define BOLT_PASSES_BINARY_FUNCTION_CALLGRAPH_H
 
-#include "bolt/Passes/CallGraph.h"
+#include "bolt/Core/CallGraph.h"
 #include <deque>
 #include <functional>
 #include <unordered_map>

bolt/include/bolt/Passes/CallGraph.h → bolt/include/bolt/Core/CallGraph.h

@@ -1,4 +1,4 @@
-//===- bolt/Passes/CallGraph.h ----------------------------------*- C++ -*-===//
+//===- bolt/Core/CallGraph.h ----------------------------------*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.

bolt/include/bolt/Passes/CallGraphWalker.h → bolt/include/bolt/Core/CallGraphWalker.h

@@ -1,4 +1,4 @@
-//===- bolt/Passes/CallGraphWalker.h ----------------------------*- C++ -*-===//
+//===- bolt/Core/CallGraphWalker.h ----------------------------*- C++ -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.

bolt/include/bolt/Core/DebugData.h

@@ -210,6 +210,15 @@ class DebugRangesSectionWriter {
   static bool classof(const DebugRangesSectionWriter *Writer) {
     return Writer->getKind() == RangesWriterKind::DebugRangesWriter;
   }
+
+  /// Append a range to the main buffer.
+  void appendToRangeBuffer(const DebugBufferVector &CUBuffer);
+
+  /// Sets Unit DIE to be updated for CU.
+  void setDie(DIE *Die) { this->Die = Die; }
+
+  /// Returns Unit DIE to be updated for CU.
+  DIE *getDie() const { return Die; }
 
   /// Writes out range lists for a current CU being processed.
   void virtual finalizeSection(){};
@@ -232,6 +241,9 @@ class DebugRangesSectionWriter {
   static constexpr uint64_t EmptyRangesOffset{0};
 
 private:
+  /// Stores Unit DIE to be updated for CU.
+  DIE *Die{0};
+
   RangesWriterKind Kind;
 };
 

bolt/include/bolt/Core/DebugNames.h

@@ -91,6 +91,10 @@ class DWARF5AcceleratorTable {
   uint64_t CurrentUnitOffset = 0;
   const DWARFUnit *CurrentUnit = nullptr;
   std::unordered_map<uint32_t, uint32_t> AbbrevTagToIndexMap;
+  /// Contains a map of TU hashes to a Foreign TU indecies.
+  /// This is used to reduce the size of Foreign TU list since there could be
+  /// multiple TUs with the same hash.
+  DenseMap<uint64_t, uint32_t> TUHashToIndexMap;
 
   /// Represents a group of entries with identical name (and hence, hash value).
   struct HashData {

bolt/include/bolt/Core/MCPlusBuilder.h

@@ -439,7 +439,20 @@ class MCPlusBuilder {
   }
 
   /// Check whether this conditional branch can be reversed
-  virtual bool isReversibleBranch(const MCInst &Inst) const { return true; }
+  virtual bool isReversibleBranch(const MCInst &Inst) const {
+    assert(!isUnsupportedInstruction(Inst) && isConditionalBranch(Inst) &&
+           "Instruction is not known conditional branch");
+
+    if (isDynamicBranch(Inst))
+      return false;
+    return true;
+  }
+
+  /// Return true if this instruction inhibits analysis of the containing
+  /// function.
+  virtual bool isUnsupportedInstruction(const MCInst &Inst) const {
+    return false;
+  }
 
   /// Return true of the instruction is of pseudo kind.
   virtual bool isPseudo(const MCInst &Inst) const {
@@ -917,13 +930,6 @@ class MCPlusBuilder {
   /// Return true if the instruction is encoded using EVEX (AVX-512).
   virtual bool hasEVEXEncoding(const MCInst &Inst) const { return false; }
 
-  /// Return true if a pair of instructions represented by \p Insts
-  /// could be fused into a single uop.
-  virtual bool isMacroOpFusionPair(ArrayRef<MCInst> Insts) const {
-    llvm_unreachable("not implemented");
-    return false;
-  }
-
   struct X86MemOperand {
     unsigned BaseRegNum;
     int64_t ScaleImm;

bolt/include/bolt/Passes/HFSort.h

@@ -19,7 +19,7 @@
 #ifndef BOLT_PASSES_HFSORT_H
 #define BOLT_PASSES_HFSORT_H
 
-#include "bolt/Passes/CallGraph.h"
+#include "bolt/Core/CallGraph.h"
 
 #include <string>
 #include <vector>

bolt/include/bolt/Passes/RegReAssign.h

@@ -9,7 +9,7 @@
 #ifndef BOLT_PASSES_REGREASSIGN_H
 #define BOLT_PASSES_REGREASSIGN_H
 
-#include "bolt/Passes/BinaryFunctionCallGraph.h"
+#include "bolt/Core/BinaryFunctionCallGraph.h"
 #include "bolt/Passes/BinaryPasses.h"
 #include "bolt/Passes/RegAnalysis.h"
 

bolt/include/bolt/Passes/ReorderFunctions.h

@@ -9,7 +9,7 @@
 #ifndef BOLT_PASSES_REORDER_FUNCTIONS_H
 #define BOLT_PASSES_REORDER_FUNCTIONS_H
 
-#include "bolt/Passes/BinaryFunctionCallGraph.h"
+#include "bolt/Core/BinaryFunctionCallGraph.h"
 #include "bolt/Passes/BinaryPasses.h"
 
 namespace llvm {

bolt/include/bolt/Profile/YAMLProfileReader.h

@@ -73,13 +73,29 @@ class YAMLProfileReader : public ProfileReaderBase {
   bool parseFunctionProfile(BinaryFunction &Function,
                             const yaml::bolt::BinaryFunctionProfile &YamlBF);
 
+  /// Checks if a function profile matches a binary function.
+  bool profileMatches(const yaml::bolt::BinaryFunctionProfile &Profile,
+                      const BinaryFunction &BF);
+
   /// Infer function profile from stale data (collected on older binaries).
   bool inferStaleProfile(BinaryFunction &Function,
                          const yaml::bolt::BinaryFunctionProfile &YamlBF);
 
   /// Initialize maps for profile matching.
   void buildNameMaps(BinaryContext &BC);
 
+  /// Matches functions using exact name.
+  size_t matchWithExactName();
+
+  /// Matches function using LTO comomon name.
+  size_t matchWithLTOCommonName();
+
+  /// Matches functions using exact hash.
+  size_t matchWithHash(BinaryContext &BC);
+
+  /// Matches functions with similarly named profiled functions.
+  size_t matchWithNameSimilarity(BinaryContext &BC);
+
   /// Update matched YAML -> BinaryFunction pair.
   void matchProfileToFunction(yaml::bolt::BinaryFunctionProfile &YamlBF,
                               BinaryFunction &BF) {

bolt/include/bolt/Rewrite/DWARFRewriter.h

@@ -16,7 +16,6 @@
 #include "llvm/ADT/StringRef.h"
 #include "llvm/CodeGen/DIE.h"
 #include "llvm/DWP/DWP.h"
-#include "llvm/MC/MCAsmLayout.h"
 #include "llvm/MC/MCContext.h"
 #include "llvm/Support/ToolOutputFile.h"
 #include <cstdint>
@@ -90,6 +89,10 @@ class DWARFRewriter {
   /// Store Rangelists writer for each DWO CU.
   RangeListsDWOWriers RangeListsWritersByCU;
 
+  /// Stores ranges writer for each DWO CU.
+  std::unordered_map<uint64_t, std::unique_ptr<DebugRangesSectionWriter>>
+      LegacyRangesWritersByCU;
+
   std::mutex LocListDebugInfoPatchesMutex;
 
   /// Dwo id specific its RangesBase.
@@ -183,7 +186,7 @@ class DWARFRewriter {
   void updateDebugInfo();
 
   /// Update stmt_list for CUs based on the new .debug_line \p Layout.
-  void updateLineTableOffsets(const MCAsmLayout &Layout);
+  void updateLineTableOffsets(const MCAssembler &Asm);
 
   uint64_t getDwoRangesBase(uint64_t DWOId) { return DwoRangesBase[DWOId]; }