Merge branch 'main' into fill_segmented

Author: NoumanAmir657

.ci/generate-buildkite-pipeline-premerge

@@ -191,14 +191,17 @@ function keep-modified-projects() {
 }
 
 function check-targets() {
+  # Do not use "check-all" here because if there is "check-all" plus a
+  # project specific target like "check-clang", that project's tests
+  # will be run twice.
   projects=${@}
   for project in ${projects}; do
     case ${project} in
     clang-tools-extra)
       echo "check-clang-tools"
     ;;
     compiler-rt)
-      echo "check-all"
+      echo "check-compiler-rt"
     ;;
     cross-project-tests)
       echo "check-cross-project"
@@ -216,10 +219,10 @@ function check-targets() {
       echo "check-lldb"
     ;;
     pstl)
-      echo "check-all"
+      # Currently we do not run pstl tests in CI.
     ;;
     libclc)
-      echo "check-all"
+      # Currently there is no testing for libclc.
     ;;
     *)
       echo "check-${project}"

.github/new-prs-labeler.yml

@@ -69,7 +69,7 @@ PGO:
   - llvm/**/llvm-profdata/**/*
   - llvm/**/llvm-profgen/**/*
 
-vectorization:
+vectorizers:
   - llvm/lib/Transforms/Vectorize/**/*
   - llvm/include/llvm/Transforms/Vectorize/**/*
 

.github/workflows/release-binaries.yml

@@ -328,7 +328,7 @@ jobs:
       run: |
         # Build some of the mlir tools that take a long time to link
         if [ "${{ needs.prepare.outputs.build-flang }}" = "true" ]; then
-          ninja -C ${{ steps.setup-stage.outputs.build-prefix }}/build/tools/clang/stage2-bins/ -j2 flang-new bbc
+          ninja -C ${{ steps.setup-stage.outputs.build-prefix }}/build/tools/clang/stage2-bins/ -j2 flang bbc
         fi
         ninja -C ${{ steps.setup-stage.outputs.build-prefix }}/build/tools/clang/stage2-bins/ \
             mlir-bytecode-parser-fuzzer \

bolt/include/bolt/Core/DIEBuilder.h

@@ -314,7 +314,7 @@ class DIEBuilder {
 
     BC.errs()
         << "BOLT-ERROR: unable to find TypeUnit for Type Unit at offset 0x"
-        << DU.getOffset() << "\n";
+        << Twine::utohexstr(DU.getOffset()) << "\n";
     return nullptr;
   }
 

bolt/include/bolt/Passes/ContinuityStats.h

@@ -0,0 +1,61 @@
+//===- bolt/Passes/ContinuityStats.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.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+//
+// This pass checks how well the BOLT input profile satisfies the following
+// "CFG continuity" property of a perfect profile:
+//
+//        Each positive-execution-count block in the function’s CFG
+//        should be *reachable* from a positive-execution-count function
+//        entry block through a positive-execution-count path.
+//
+// More specifically, for each of the hottest 1000 functions, the pass
+// calculates the function’s fraction of basic block execution counts
+// that is *unreachable*. It then reports the 95th percentile of the
+// distribution of the 1000 unreachable fractions in a single BOLT-INFO line.
+// The smaller the reported value is, the better the BOLT profile
+// satisfies the CFG continuity property.
+
+// The default value of 1000 above can be changed via the hidden BOLT option
+// `-num-functions-for-continuity-check=[N]`.
+// If more detailed stats are needed, `-v=1` can be used: the hottest N
+// functions will be grouped into 5 equally-sized buckets, from the hottest
+// to the coldest; for each bucket, various summary statistics of the
+// distribution of the unreachable fractions and the raw unreachable execution
+// counts will be reported.
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef BOLT_PASSES_CONTINUITYSTATS_H
+#define BOLT_PASSES_CONTINUITYSTATS_H
+
+#include "bolt/Passes/BinaryPasses.h"
+#include <vector>
+
+namespace llvm {
+
+class raw_ostream;
+
+namespace bolt {
+class BinaryContext;
+
+/// Compute and report to the user the function CFG continuity quality
+class PrintContinuityStats : public BinaryFunctionPass {
+public:
+  explicit PrintContinuityStats(const cl::opt<bool> &PrintPass)
+      : BinaryFunctionPass(PrintPass) {}
+
+  bool shouldOptimize(const BinaryFunction &BF) const override;
+  const char *getName() const override { return "continuity-stats"; }
+  bool shouldPrint(const BinaryFunction &) const override { return false; }
+  Error runOnFunctions(BinaryContext &BC) override;
+};
+
+} // namespace bolt
+} // namespace llvm
+
+#endif // BOLT_PASSES_CONTINUITYSTATS_H

bolt/include/bolt/Profile/BoltAddressTranslation.h

@@ -141,15 +141,13 @@ class BoltAddressTranslation {
                          uint64_t FuncOutputAddress) const;
 
   /// Write the serialized address translation table for a function.
-  template <bool Cold>
-  void writeMaps(std::map<uint64_t, MapTy> &Maps, uint64_t &PrevAddress,
-                 raw_ostream &OS);
+  template <bool Cold> void writeMaps(uint64_t &PrevAddress, raw_ostream &OS);
 
   /// Read the serialized address translation table for a function.
   /// Return a parse error if failed.
   template <bool Cold>
-  void parseMaps(std::vector<uint64_t> &HotFuncs, uint64_t &PrevAddress,
-                 DataExtractor &DE, uint64_t &Offset, Error &Err);
+  void parseMaps(uint64_t &PrevAddress, DataExtractor &DE, uint64_t &Offset,
+                 Error &Err);
 
   /// Returns the bitmask with set bits corresponding to indices of BRANCHENTRY
   /// entries in function address translation map.
@@ -161,6 +159,9 @@ class BoltAddressTranslation {
 
   std::map<uint64_t, MapTy> Maps;
 
+  /// Ordered vector with addresses of hot functions.
+  std::vector<uint64_t> HotFuncs;
+
   /// Map a function to its basic blocks count
   std::unordered_map<uint64_t, size_t> NumBasicBlocksMap;
 

bolt/lib/Core/BinaryContext.cpp

@@ -1294,8 +1294,8 @@ bool BinaryContext::handleAArch64Veneer(uint64_t Address, bool MatchOnly) {
     Veneer->getOrCreateLocalLabel(Address);
     Veneer->setMaxSize(TotalSize);
     Veneer->updateState(BinaryFunction::State::Disassembled);
-    LLVM_DEBUG(dbgs() << "BOLT-DEBUG: handling veneer function at 0x" << Address
-                      << "\n");
+    LLVM_DEBUG(dbgs() << "BOLT-DEBUG: handling veneer function at 0x"
+                      << Twine::utohexstr(Address) << "\n");
     return true;
   };
 

bolt/lib/Core/BinaryFunction.cpp

@@ -3684,9 +3684,8 @@ BinaryFunction::BasicBlockListType BinaryFunction::dfs() const {
     BinaryBasicBlock *BB = Stack.top();
     Stack.pop();
 
-    if (Visited.find(BB) != Visited.end())
+    if (!Visited.insert(BB).second)
       continue;
-    Visited.insert(BB);
     DFS.push_back(BB);
 
     for (BinaryBasicBlock *SuccBB : BB->landing_pads()) {
@@ -3879,11 +3878,8 @@ void BinaryFunction::disambiguateJumpTables(
       JumpTable *JT = getJumpTable(Inst);
       if (!JT)
         continue;
-      auto Iter = JumpTables.find(JT);
-      if (Iter == JumpTables.end()) {
-        JumpTables.insert(JT);
+      if (JumpTables.insert(JT).second)
         continue;
-      }
       // This instruction is an indirect jump using a jump table, but it is
       // using the same jump table of another jump. Try all our tricks to
       // extract the jump table symbol and make it point to a new, duplicated JT

bolt/lib/Core/DIEBuilder.cpp

@@ -57,11 +57,9 @@ getDWOName(llvm::DWARFUnit &CU,
          "DW_AT_dwo_name/DW_AT_GNU_dwo_name does not exist.");
   if (DwarfOutputPath) {
     DWOName = std::string(sys::path::filename(DWOName));
-    auto Iter = NameToIndexMap.find(DWOName);
-    if (Iter == NameToIndexMap.end())
-      Iter = NameToIndexMap.insert({DWOName, 0}).first;
-    DWOName.append(std::to_string(Iter->second));
-    ++Iter->second;
+    uint32_t &Index = NameToIndexMap[DWOName];
+    DWOName.append(std::to_string(Index));
+    ++Index;
   }
   DWOName.append(".dwo");
   return DWOName;

bolt/lib/Passes/ADRRelaxationPass.cpp

@@ -56,13 +56,14 @@ void ADRRelaxationPass::runOnFunction(BinaryFunction &BF) {
           continue;
       }
 
-      // Don't relax adr if it points to the same function and it is not split
-      // and BF initial size is < 1MB.
+      // Don't relax ADR if it points to the same function and is in the main
+      // fragment and BF initial size is < 1MB.
       const unsigned OneMB = 0x100000;
       if (BF.getSize() < OneMB) {
         BinaryFunction *TargetBF = BC.getFunctionForSymbol(Symbol);
-        if (TargetBF == &BF && !BF.isSplit())
+        if (TargetBF == &BF && !BB.isSplit())
           continue;
+
         // No relaxation needed if ADR references a basic block in the same
         // fragment.
         if (BinaryBasicBlock *TargetBB = BF.getBasicBlockForLabel(Symbol))