[AutoBump] Merge with 267de854 (May 22) (50)

.ci/generate-buildkite-pipeline-premerge

@@ -68,7 +68,7 @@ function compute-projects-to-test() {
       done
     ;;
     clang)
-      for p in clang-tools-extra compiler-rt flang lldb cross-project-tests; do
+      for p in clang-tools-extra compiler-rt lldb cross-project-tests; do
         echo $p
       done
     ;;

.github/workflows/issue-release-workflow.yml

@@ -53,7 +53,7 @@ jobs:
 
       - name: Setup Environment
         run: |
-          pip install -r ./llvm/utils/git/requirements.txt
+          pip install --require-hashes -r ./llvm/utils/git/requirements.txt
           ./llvm/utils/git/github-automation.py --token ${{ github.token }} setup-llvmbot-git
 
       - name: Backport Commits

.github/workflows/issue-subscriber.yml

@@ -22,7 +22,7 @@ jobs:
       - name: Setup Automation Script
         working-directory: ./llvm/utils/git/
         run: |
-          pip install -r requirements.txt
+          pip install --require-hashes -r requirements.txt
 
       - name: Update watchers
         working-directory: ./llvm/utils/git/

.github/workflows/llvm-bugs.yml

@@ -14,7 +14,7 @@ jobs:
     runs-on: ubuntu-latest
     if: github.repository == 'llvm/llvm-project'
     steps:
-      - uses: actions/setup-node@v3
+      - uses: actions/setup-node@v4
         with:
           node-version: 18
           check-latest: true

.github/workflows/merged-prs.yml

@@ -29,7 +29,7 @@ jobs:
       - name: Setup Automation Script
         working-directory: ./llvm/utils/git/
         run: |
-          pip install -r requirements.txt
+          pip install --require-hashes -r requirements.txt
 
       - name: Add Buildbot information comment
         working-directory: ./llvm/utils/git/

.github/workflows/new-prs.yml

@@ -43,7 +43,7 @@ jobs:
       - name: Setup Automation Script
         working-directory: ./llvm/utils/git/
         run: |
-          pip install -r requirements.txt
+          pip install --require-hashes -r requirements.txt
 
       - name: Greet Author
         working-directory: ./llvm/utils/git/

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

@@ -29,7 +29,7 @@ jobs:
 
       - name: Install Dependencies
         run: |
-          pip install -r llvm/utils/git/requirements.txt
+          pip install --require-hashes -r llvm/utils/git/requirements.txt
 
       - name: Request Release Note
         env:

.github/workflows/pr-subscriber.yml

@@ -22,7 +22,7 @@ jobs:
       - name: Setup Automation Script
         working-directory: ./llvm/utils/git/
         run: |
-          pip install -r requirements.txt
+          pip install --require-hashes -r requirements.txt
 
       - name: Update watchers
         working-directory: ./llvm/utils/git/

.github/workflows/release-binaries.yml

@@ -47,7 +47,7 @@ jobs:
 
     - name: Install Dependencies
       run: |
-        pip install -r ./llvm/utils/git/requirements.txt
+        pip install --require-hashes -r ./llvm/utils/git/requirements.txt
 
     - name: Check Permissions
       env:

.github/workflows/version-check.yml

@@ -23,7 +23,7 @@ jobs:
 
       - name: Install dependencies
         run: |
-          pip install -r ./llvm/utils/git/requirements.txt
+          pip install --require-hashes -r ./llvm/utils/git/requirements.txt
 
       - name: Version Check
         run: |

bolt/include/bolt/Core/BinaryContext.h

@@ -677,6 +677,9 @@ class BinaryContext {
   /// have an origin file name available.
   bool HasSymbolsWithFileName{false};
 
+  /// Does the binary have BAT section.
+  bool HasBATSection{false};
+
   /// Sum of execution count of all functions
   uint64_t SumExecutionCount{0};
 

bolt/include/bolt/Core/MCPlusBuilder.h

@@ -438,8 +438,8 @@ class MCPlusBuilder {
     return false;
   }
 
-  /// Check whether we support inverting this branch
-  virtual bool isUnsupportedBranch(const MCInst &Inst) const { return false; }
+  /// Check whether this conditional branch can be reversed
+  virtual bool isReversibleBranch(const MCInst &Inst) const { return true; }
 
   /// Return true of the instruction is of pseudo kind.
   virtual bool isPseudo(const MCInst &Inst) const {

bolt/include/bolt/Passes/BinaryPasses.h

@@ -16,6 +16,7 @@
 #include "bolt/Core/BinaryContext.h"
 #include "bolt/Core/BinaryFunction.h"
 #include "bolt/Core/DynoStats.h"
+#include "bolt/Profile/BoltAddressTranslation.h"
 #include "llvm/Support/CommandLine.h"
 #include <atomic>
 #include <set>
@@ -399,8 +400,11 @@ class PrintProfileStats : public BinaryFunctionPass {
 /// Prints a list of the top 100 functions sorted by a set of
 /// dyno stats categories.
 class PrintProgramStats : public BinaryFunctionPass {
+  BoltAddressTranslation *BAT = nullptr;
+
 public:
-  explicit PrintProgramStats() : BinaryFunctionPass(false) {}
+  explicit PrintProgramStats(BoltAddressTranslation *BAT = nullptr)
+      : BinaryFunctionPass(false), BAT(BAT) {}
 
   const char *getName() const override { return "print-stats"; }
   bool shouldPrint(const BinaryFunction &) const override { return false; }

bolt/include/bolt/Passes/StokeInfo.h

@@ -87,10 +87,10 @@ struct StokeFuncInfo {
               << "," << NumBlocks << "," << IsLoopFree << "," << NumLoops << ","
               << MaxLoopDepth << "," << HotSize << "," << TotalSize << ","
               << Score << "," << HasCall << ",\"{ ";
-      for (std::string S : DefIn)
+      for (const std::string &S : DefIn)
         Outfile << "%" << S << " ";
       Outfile << "}\",\"{ ";
-      for (std::string S : LiveOut)
+      for (const std::string &S : LiveOut)
         Outfile << "%" << S << " ";
       Outfile << "}\"," << HeapOut << "," << StackOut << "," << HasRipAddr
               << "," << Omitted << "\n";

bolt/include/bolt/Profile/DataAggregator.h

@@ -15,6 +15,7 @@
 #define BOLT_PROFILE_DATA_AGGREGATOR_H
 
 #include "bolt/Profile/DataReader.h"
+#include "bolt/Profile/YAMLProfileWriter.h"
 #include "llvm/ADT/StringRef.h"
 #include "llvm/Support/Error.h"
 #include "llvm/Support/Program.h"
@@ -122,14 +123,14 @@ class DataAggregator : public DataReader {
     uint64_t ExternCount{0};
   };
 
-  struct BranchInfo {
+  struct TakenBranchInfo {
     uint64_t TakenCount{0};
     uint64_t MispredCount{0};
   };
 
   /// Intermediate storage for profile data. We save the results of parsing
   /// and use them later for processing and assigning profile.
-  std::unordered_map<Trace, BranchInfo, TraceHash> BranchLBRs;
+  std::unordered_map<Trace, TakenBranchInfo, TraceHash> BranchLBRs;
   std::unordered_map<Trace, FTInfo, TraceHash> FallthroughLBRs;
   std::vector<AggregatedLBREntry> AggregatedLBRs;
   std::unordered_map<uint64_t, uint64_t> BasicSamples;
@@ -248,7 +249,7 @@ class DataAggregator : public DataReader {
   BinaryFunction *getBATParentFunction(const BinaryFunction &Func) const;
 
   /// Retrieve the location name to be used for samples recorded in \p Func.
-  StringRef getLocationName(const BinaryFunction &Func) const;
+  static StringRef getLocationName(const BinaryFunction &Func, bool BAT);
 
   /// Semantic actions - parser hooks to interpret parsed perf samples
   /// Register a sample (non-LBR mode), i.e. a new hit at \p Address
@@ -490,6 +491,8 @@ class DataAggregator : public DataReader {
   /// Parse the output generated by "perf buildid-list" to extract build-ids
   /// and return a file name matching a given \p FileBuildID.
   std::optional<StringRef> getFileNameForBuildID(StringRef FileBuildID);
+
+  friend class YAMLProfileWriter;
 };
 } // namespace bolt
 } // namespace llvm

bolt/lib/Core/BinaryContext.cpp

@@ -1322,7 +1322,9 @@ void BinaryContext::processInterproceduralReferences() {
        InterproceduralReferences) {
     BinaryFunction &Function = *It.first;
     uint64_t Address = It.second;
-    if (!Address || Function.isIgnored())
+    // Process interprocedural references from ignored functions in BAT mode
+    // (non-simple in non-relocation mode) to properly register entry points
+    if (!Address || (Function.isIgnored() && !HasBATSection))
       continue;
 
     BinaryFunction *TargetFunction =

bolt/lib/Core/BinaryFunction.cpp

@@ -1284,7 +1284,7 @@ Error BinaryFunction::disassemble() {
         const bool IsCondBranch = MIB->isConditionalBranch(Instruction);
         MCSymbol *TargetSymbol = nullptr;
 
-        if (BC.MIB->isUnsupportedBranch(Instruction)) {
+        if (!BC.MIB->isReversibleBranch(Instruction)) {
           setIgnored();
           if (BinaryFunction *TargetFunc =
                   BC.getBinaryFunctionContainingAddress(TargetAddress))
@@ -1666,7 +1666,8 @@ void BinaryFunction::postProcessEntryPoints() {
     // In non-relocation mode there's potentially an external undetectable
     // reference to the entry point and hence we cannot move this entry
     // point. Optimizing without moving could be difficult.
-    if (!BC.HasRelocations)
+    // In BAT mode, register any known entry points for CFG construction.
+    if (!BC.HasRelocations && !BC.HasBATSection)
       setSimple(false);
 
     const uint32_t Offset = KV.first;
@@ -3381,7 +3382,7 @@ void BinaryFunction::fixBranches() {
 
       // Reverse branch condition and swap successors.
       auto swapSuccessors = [&]() {
-        if (MIB->isUnsupportedBranch(*CondBranch)) {
+        if (!MIB->isReversibleBranch(*CondBranch)) {
           if (opts::Verbosity) {
             BC.outs() << "BOLT-INFO: unable to swap successors in " << *this
                       << '\n';

bolt/lib/Passes/BinaryPasses.cpp

@@ -674,7 +674,8 @@ static uint64_t fixDoubleJumps(BinaryFunction &Function, bool MarkInvalid) {
   MCPlusBuilder *MIB = Function.getBinaryContext().MIB.get();
   for (BinaryBasicBlock &BB : Function) {
     auto checkAndPatch = [&](BinaryBasicBlock *Pred, BinaryBasicBlock *Succ,
-                             const MCSymbol *SuccSym) {
+                             const MCSymbol *SuccSym,
+                             std::optional<uint32_t> Offset) {
       // Ignore infinite loop jumps or fallthrough tail jumps.
       if (Pred == Succ || Succ == &BB)
         return false;
@@ -715,9 +716,11 @@ static uint64_t fixDoubleJumps(BinaryFunction &Function, bool MarkInvalid) {
           Pred->removeSuccessor(&BB);
           Pred->eraseInstruction(Pred->findInstruction(Branch));
           Pred->addTailCallInstruction(SuccSym);
-          MCInst *TailCall = Pred->getLastNonPseudoInstr();
-          assert(TailCall);
-          MIB->setOffset(*TailCall, BB.getOffset());
+          if (Offset) {
+            MCInst *TailCall = Pred->getLastNonPseudoInstr();
+            assert(TailCall);
+            MIB->setOffset(*TailCall, *Offset);
+          }
         } else {
           return false;
         }
@@ -760,7 +763,8 @@ static uint64_t fixDoubleJumps(BinaryFunction &Function, bool MarkInvalid) {
       if (Pred->getSuccessor() == &BB ||
           (Pred->getConditionalSuccessor(true) == &BB && !IsTailCall) ||
           Pred->getConditionalSuccessor(false) == &BB)
-        if (checkAndPatch(Pred, Succ, SuccSym) && MarkInvalid)
+        if (checkAndPatch(Pred, Succ, SuccSym, MIB->getOffset(*Inst)) &&
+            MarkInvalid)
           BB.markValid(BB.pred_size() != 0 || BB.isLandingPad() ||
                        BB.isEntryPoint());
     }
@@ -1386,9 +1390,19 @@ Error PrintProgramStats::runOnFunctions(BinaryContext &BC) {
     if (Function.isPLTFunction())
       continue;
 
+    // Adjustment for BAT mode: the profile for BOLT split fragments is combined
+    // so only count the hot fragment.
+    const uint64_t Address = Function.getAddress();
+    bool IsHotParentOfBOLTSplitFunction = !Function.getFragments().empty() &&
+                                          BAT && BAT->isBATFunction(Address) &&
+                                          !BAT->fetchParentAddress(Address);
+
     ++NumRegularFunctions;
 
-    if (!Function.isSimple()) {
+    // In BOLTed binaries split functions are non-simple (due to non-relocation
+    // mode), but the original function is known to be simple and we have a
+    // valid profile for it.
+    if (!Function.isSimple() && !IsHotParentOfBOLTSplitFunction) {
       if (Function.hasProfile())
         ++NumNonSimpleProfiledFunctions;
       continue;

bolt/lib/Passes/Instrumentation.cpp

@@ -480,7 +480,7 @@ void Instrumentation::instrumentFunction(BinaryFunction &Function,
       else if (BC.MIB->isUnconditionalBranch(Inst))
         HasUnconditionalBranch = true;
       else if ((!BC.MIB->isCall(Inst) && !BC.MIB->isConditionalBranch(Inst)) ||
-               BC.MIB->isUnsupportedBranch(Inst))
+               !BC.MIB->isReversibleBranch(Inst))
         continue;
 
       const uint32_t FromOffset = *BC.MIB->getOffset(Inst);