Merge from 'main' to 'sycl-web' (402 commits)

  CONFLICT (content): Merge conflict in llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp

Author: fineg74

.ci/monolithic-linux.sh

@@ -49,7 +49,8 @@ cmake -S ${MONOREPO_ROOT}/llvm -B ${BUILD_DIR} \
       -D LLVM_ENABLE_LLD=ON \
       -D CMAKE_CXX_FLAGS=-gmlt \
       -D BOLT_CLANG_EXE=/usr/bin/clang \
-      -D LLVM_CCACHE_BUILD=ON
+      -D LLVM_CCACHE_BUILD=ON \
+      -D MLIR_ENABLE_BINDINGS_PYTHON=ON
 
 echo "--- ninja"
 # Targets are not escaped as they are passed as separate arguments.

.ci/monolithic-windows.sh

@@ -48,7 +48,8 @@ cmake -S ${MONOREPO_ROOT}/llvm -B ${BUILD_DIR} \
       -D LLVM_LIT_ARGS="-v --xunit-xml-output ${BUILD_DIR}/test-results.xml" \
       -D COMPILER_RT_BUILD_ORC=OFF \
       -D CMAKE_C_COMPILER_LAUNCHER=sccache \
-      -D CMAKE_CXX_COMPILER_LAUNCHER=sccache
+      -D CMAKE_CXX_COMPILER_LAUNCHER=sccache \
+      -D MLIR_ENABLE_BINDINGS_PYTHON=ON
 
 echo "--- ninja"
 # Targets are not escaped as they are passed as separate arguments.

.github/workflows/docs.yml

@@ -15,6 +15,9 @@ on:
     paths:
       - 'llvm/docs/**'
       - 'clang/docs/**'
+      - 'clang/include/clang/Basic/AttrDocs.td'
+      - 'clang/include/clang/Driver/ClangOptionDocs.td'
+      - 'clang/include/clang/Basic/DiagnosticDocs.td'
       - 'clang-tools-extra/docs/**'
       - 'lldb/docs/**'
       - 'libunwind/docs/**'
@@ -29,6 +32,9 @@ on:
     paths:
       - 'llvm/docs/**'
       - 'clang/docs/**'
+      - 'clang/include/clang/Basic/AttrDocs.td'
+      - 'clang/include/clang/Driver/ClangOptionDocs.td'
+      - 'clang/include/clang/Basic/DiagnosticDocs.td'
       - 'clang-tools-extra/docs/**'
       - 'lldb/docs/**'
       - 'libunwind/docs/**'
@@ -64,6 +70,9 @@ jobs:
               - 'llvm/docs/**'
             clang:
               - 'clang/docs/**'
+              - 'clang/include/clang/Basic/AttrDocs.td'
+              - 'clang/include/clang/Driver/ClangOptionDocs.td'
+              - 'clang/include/clang/Basic/DiagnosticDocs.td'
             clang-tools-extra:
               - 'clang-tools-extra/docs/**'
             lldb:

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

@@ -43,6 +43,7 @@ env:
 
 jobs:
   stage1:
+    if: github.repository_owner == 'llvm'
     runs-on:
       group: libcxx-runners-8
     continue-on-error: false
@@ -81,6 +82,7 @@ jobs:
             **/CMakeOutput.log
             **/crash_diagnostics/*
   stage2:
+    if: github.repository_owner == 'llvm'
     runs-on:
       group: libcxx-runners-8
     needs: [ stage1 ]
@@ -130,6 +132,7 @@ jobs:
             **/CMakeOutput.log
             **/crash_diagnostics/*
   stage3:
+    if: github.repository_owner == 'llvm'
     needs: [ stage1, stage2 ]
     continue-on-error: false
     strategy:

.github/workflows/pr-code-format.yml

@@ -40,9 +40,11 @@ jobs:
           path: code-format-tools
 
       - name: "Listed files"
+        env:
+          CHANGED_FILES: ${{ steps.changed-files.outputs.all_changed_files }}
         run: |
           echo "Formatting files:"
-          echo "${{ steps.changed-files.outputs.all_changed_files }}"
+          echo "$CHANGED_FILES"
 
       - name: Install clang-format
         uses: aminya/setup-cpp@v1

bolt/include/bolt/Core/BinaryContext.h

@@ -1230,6 +1230,9 @@ class BinaryContext {
   ///
   /// Return the pair where the first size is for the main part, and the second
   /// size is for the cold one.
+  /// Modify BinaryBasicBlock::OutputAddressRange for each basic block in the
+  /// function in place so that BinaryBasicBlock::getOutputSize() gives the
+  /// emitted size of the basic block.
   std::pair<size_t, size_t> calculateEmittedSize(BinaryFunction &BF,
                                                  bool FixBranches = true);
 

bolt/lib/Core/BinaryContext.cpp

@@ -2331,14 +2331,36 @@ BinaryContext::calculateEmittedSize(BinaryFunction &BF, bool FixBranches) {
   MCAsmLayout Layout(Assembler);
   Assembler.layout(Layout);
 
+  // Obtain fragment sizes.
+  std::vector<uint64_t> FragmentSizes;
+  // Main fragment size.
   const uint64_t HotSize =
       Layout.getSymbolOffset(*EndLabel) - Layout.getSymbolOffset(*StartLabel);
-  const uint64_t ColdSize =
-      std::accumulate(SplitLabels.begin(), SplitLabels.end(), 0ULL,
-                      [&](const uint64_t Accu, const LabelRange &Labels) {
-                        return Accu + Layout.getSymbolOffset(*Labels.second) -
-                               Layout.getSymbolOffset(*Labels.first);
-                      });
+  FragmentSizes.push_back(HotSize);
+  // Split fragment sizes.
+  uint64_t ColdSize = 0;
+  for (const auto &Labels : SplitLabels) {
+    uint64_t Size = Layout.getSymbolOffset(*Labels.second) -
+                    Layout.getSymbolOffset(*Labels.first);
+    FragmentSizes.push_back(Size);
+    ColdSize += Size;
+  }
+
+  // Populate new start and end offsets of each basic block.
+  uint64_t FragmentIndex = 0;
+  for (FunctionFragment &FF : BF.getLayout().fragments()) {
+    BinaryBasicBlock *PrevBB = nullptr;
+    for (BinaryBasicBlock *BB : FF) {
+      const uint64_t BBStartOffset = Layout.getSymbolOffset(*(BB->getLabel()));
+      BB->setOutputStartAddress(BBStartOffset);
+      if (PrevBB)
+        PrevBB->setOutputEndAddress(BBStartOffset);
+      PrevBB = BB;
+    }
+    if (PrevBB)
+      PrevBB->setOutputEndAddress(FragmentSizes[FragmentIndex]);
+    FragmentIndex++;
+  }
 
   // Clean-up the effect of the code emission.
   for (const MCSymbol &Symbol : Assembler.symbols()) {

bolt/lib/Core/BinaryFunction.cpp

@@ -40,6 +40,7 @@
 #include "llvm/Support/Regex.h"
 #include "llvm/Support/Timer.h"
 #include "llvm/Support/raw_ostream.h"
+#include "llvm/Support/xxhash.h"
 #include <functional>
 #include <limits>
 #include <numeric>
@@ -108,6 +109,13 @@ cl::opt<bool>
                             cl::desc("try to preserve basic block alignment"),
                             cl::cat(BoltOptCategory));
 
+static cl::opt<bool> PrintOutputAddressRange(
+    "print-output-address-range",
+    cl::desc(
+        "print output address range for each basic block in the function when"
+        "BinaryFunction::print is called"),
+    cl::Hidden, cl::cat(BoltOptCategory));
+
 cl::opt<bool>
 PrintDynoStats("dyno-stats",
   cl::desc("print execution info based on profile"),
@@ -510,6 +518,11 @@ void BinaryFunction::print(raw_ostream &OS, std::string Annotation) {
       OS << BB->getName() << " (" << BB->size()
          << " instructions, align : " << BB->getAlignment() << ")\n";
 
+      if (opts::PrintOutputAddressRange)
+        OS << formatv("  Output Address Range: [{0:x}, {1:x}) ({2} bytes)\n",
+                      BB->getOutputAddressRange().first,
+                      BB->getOutputAddressRange().second, BB->getOutputSize());
+
       if (isEntryPoint(*BB)) {
         if (MCSymbol *EntrySymbol = getSecondaryEntryPointSymbol(*BB))
           OS << "  Secondary Entry Point: " << EntrySymbol->getName() << '\n';
@@ -3624,7 +3637,7 @@ size_t BinaryFunction::computeHash(bool UseDFS,
   for (const BinaryBasicBlock *BB : Order)
     HashString.append(hashBlock(BC, *BB, OperandHashFunc));
 
-  return Hash = std::hash<std::string>{}(HashString);
+  return Hash = llvm::xxh3_64bits(HashString);
 }
 
 void BinaryFunction::insertBasicBlocks(
@@ -4131,10 +4144,6 @@ void BinaryFunction::updateOutputValues(const BOLTLinker &Linker) {
   if (!requiresAddressMap())
     return;
 
-  // Output ranges should match the input if the body hasn't changed.
-  if (!isSimple() && !BC.HasRelocations)
-    return;
-
   // AArch64 may have functions that only contains a constant island (no code).
   if (getLayout().block_empty())
     return;
@@ -4182,6 +4191,12 @@ void BinaryFunction::updateOutputValues(const BOLTLinker &Linker) {
                                     ? FF.getAddress() + FF.getImageSize()
                                     : getOutputAddress() + getOutputSize());
   }
+
+  // Reset output addresses for deleted blocks.
+  for (BinaryBasicBlock *BB : DeletedBasicBlocks) {
+    BB->setOutputStartAddress(0);
+    BB->setOutputEndAddress(0);
+  }
 }
 
 DebugAddressRangesVector BinaryFunction::getOutputAddressRanges() const {

bolt/lib/Core/HashUtilities.cpp

@@ -18,15 +18,6 @@
 namespace llvm {
 namespace bolt {
 
-/// Hashing a 64-bit integer to a 16-bit one.
-uint16_t hash_64_to_16(const uint64_t Hash) {
-  uint16_t Res = (uint16_t)(Hash & 0xFFFF);
-  Res ^= (uint16_t)((Hash >> 16) & 0xFFFF);
-  Res ^= (uint16_t)((Hash >> 32) & 0xFFFF);
-  Res ^= (uint16_t)((Hash >> 48) & 0xFFFF);
-  return Res;
-}
-
 std::string hashInteger(uint64_t Value) {
   std::string HashString;
   if (Value == 0)

bolt/lib/Profile/StaleProfileMatching.cpp

@@ -30,6 +30,7 @@
 #include "llvm/ADT/Bitfields.h"
 #include "llvm/ADT/Hashing.h"
 #include "llvm/Support/CommandLine.h"
+#include "llvm/Support/xxhash.h"
 #include "llvm/Transforms/Utils/SampleProfileInference.h"
 
 #include <queue>
@@ -240,12 +241,12 @@ void BinaryFunction::computeBlockHashes() const {
     // Hashing complete instructions.
     std::string InstrHashStr = hashBlock(
         BC, *BB, [&](const MCOperand &Op) { return hashInstOperand(BC, Op); });
-    uint64_t InstrHash = std::hash<std::string>{}(InstrHashStr);
-    BlendedHashes[I].InstrHash = (uint16_t)hash_value(InstrHash);
+    uint64_t InstrHash = llvm::xxh3_64bits(InstrHashStr);
+    BlendedHashes[I].InstrHash = (uint16_t)InstrHash;
     // Hashing opcodes.
     std::string OpcodeHashStr = hashBlockLoose(BC, *BB);
-    OpcodeHashes[I] = std::hash<std::string>{}(OpcodeHashStr);
-    BlendedHashes[I].OpcodeHash = (uint16_t)hash_value(OpcodeHashes[I]);
+    OpcodeHashes[I] = llvm::xxh3_64bits(OpcodeHashStr);
+    BlendedHashes[I].OpcodeHash = (uint16_t)OpcodeHashes[I];
   }
 
   // Initialize neighbor hash.
@@ -257,15 +258,15 @@ void BinaryFunction::computeBlockHashes() const {
       uint64_t SuccHash = OpcodeHashes[SuccBB->getIndex()];
       Hash = hashing::detail::hash_16_bytes(Hash, SuccHash);
     }
-    BlendedHashes[I].SuccHash = (uint8_t)hash_value(Hash);
+    BlendedHashes[I].SuccHash = (uint8_t)Hash;
 
     // Append hashes of predecessors.
     Hash = 0;
     for (BinaryBasicBlock *PredBB : BB->predecessors()) {
       uint64_t PredHash = OpcodeHashes[PredBB->getIndex()];
       Hash = hashing::detail::hash_16_bytes(Hash, PredHash);
     }
-    BlendedHashes[I].PredHash = (uint8_t)hash_value(Hash);
+    BlendedHashes[I].PredHash = (uint8_t)Hash;
   }
 
   //  Assign hashes.
@@ -405,6 +406,8 @@ void matchWeightsByHashes(BinaryContext &BC,
         ++BC.Stats.NumMatchedBlocks;
         BC.Stats.MatchedSampleCount += YamlBB.ExecCount;
         LLVM_DEBUG(dbgs() << "  exact match\n");
+      } else {
+        LLVM_DEBUG(dbgs() << "  loose match\n");
       }
     } else {
       LLVM_DEBUG(