Merge remote-tracking branch 'origin/main' into vplan-all-regions-as-transform

Author: fhahn

.ci/monolithic-linux.sh

@@ -53,9 +53,9 @@ targets="${2}"
 lit_args="-v --xunit-xml-output ${BUILD_DIR}/test-results.xml --use-unique-output-file-name --timeout=1200 --time-tests"
 
 echo "--- cmake"
-pip install -q -r "${MONOREPO_ROOT}"/mlir/python/requirements.txt
-pip install -q -r "${MONOREPO_ROOT}"/lldb/test/requirements.txt
-pip install -q -r "${MONOREPO_ROOT}"/.ci/requirements.txt
+pip install --break-system-packages -q -r "${MONOREPO_ROOT}"/mlir/python/requirements.txt
+pip install --break-system-packages -q -r "${MONOREPO_ROOT}"/lldb/test/requirements.txt
+pip install --break-system-packages -q -r "${MONOREPO_ROOT}"/.ci/requirements.txt
 cmake -S "${MONOREPO_ROOT}"/llvm -B "${BUILD_DIR}" \
       -D LLVM_ENABLE_PROJECTS="${projects}" \
       -G Ninja \

.github/workflows/containers/github-action-ci-windows/Dockerfile

@@ -108,7 +108,7 @@ RUN choco install -y handle
 
 RUN pip3 install pywin32 buildbot-worker==2.8.4
 
-ARG RUNNER_VERSION=2.322.0
+ARG RUNNER_VERSION=2.323.0
 ENV RUNNER_VERSION=$RUNNER_VERSION
 
 RUN powershell -Command \

.github/workflows/containers/github-action-ci/Dockerfile

@@ -95,7 +95,7 @@ WORKDIR /home/gha
 
 FROM ci-container as ci-container-agent
 
-ENV GITHUB_RUNNER_VERSION=2.322.0
+ENV GITHUB_RUNNER_VERSION=2.323.0
 
 RUN mkdir actions-runner && \
     cd actions-runner && \

bolt/include/bolt/Core/Relocation.h

@@ -86,6 +86,9 @@ class Relocation {
   /// Adjust value depending on relocation type (make it PC relative or not).
   static uint64_t encodeValue(uint32_t Type, uint64_t Value, uint64_t PC);
 
+  /// Return true if there are enough bits to encode the relocation value.
+  static bool canEncodeValue(uint32_t Type, uint64_t Value, uint64_t PC);
+
   /// Extract current relocated value from binary contents. This is used for
   /// RISC architectures where values are encoded in specific bits depending
   /// on the relocation value. For X86, we limit to sign extending the value

bolt/lib/Core/BinaryFunction.cpp

@@ -1795,6 +1795,8 @@ bool BinaryFunction::scanExternalRefs() {
     // Create relocation for every fixup.
     for (const MCFixup &Fixup : Fixups) {
       std::optional<Relocation> Rel = BC.MIB->createRelocation(Fixup, *BC.MAB);
+      // Can be skipped in case of overlow during relocation value encoding.
+      Rel->setOptional();
       if (!Rel) {
         Success = false;
         continue;

bolt/lib/Core/BinarySection.cpp

@@ -12,6 +12,7 @@
 
 #include "bolt/Core/BinarySection.h"
 #include "bolt/Core/BinaryContext.h"
+#include "bolt/Utils/CommandLineOpts.h"
 #include "bolt/Utils/Utils.h"
 #include "llvm/MC/MCStreamer.h"
 #include "llvm/Support/CommandLine.h"
@@ -22,8 +23,8 @@ using namespace llvm;
 using namespace bolt;
 
 namespace opts {
-extern cl::opt<bool> PrintRelocations;
 extern cl::opt<bool> HotData;
+extern cl::opt<bool> PrintRelocations;
 } // namespace opts
 
 uint64_t BinarySection::Count = 0;
@@ -174,11 +175,30 @@ void BinarySection::flushPendingRelocations(raw_pwrite_stream &OS,
     OS.pwrite(Patch.Bytes.data(), Patch.Bytes.size(),
               SectionFileOffset + Patch.Offset);
 
+  uint64_t SkippedPendingRelocations = 0;
   for (Relocation &Reloc : PendingRelocations) {
     uint64_t Value = Reloc.Addend;
     if (Reloc.Symbol)
       Value += Resolver(Reloc.Symbol);
 
+    // Safely skip any optional pending relocation that cannot be encoded.
+    if (Reloc.isOptional() &&
+        !Relocation::canEncodeValue(Reloc.Type, Value,
+                                    SectionAddress + Reloc.Offset)) {
+
+      // A successful run of 'scanExternalRefs' means that all pending
+      // relocations are flushed. Otherwise, PatchEntries should run.
+      if (!opts::ForcePatch) {
+        BC.errs()
+            << "BOLT-ERROR: cannot encode relocation for symbol "
+            << Reloc.Symbol->getName()
+            << " as it is out-of-range. To proceed must use -force-patch\n";
+        exit(1);
+      }
+
+      ++SkippedPendingRelocations;
+      continue;
+    }
     Value = Relocation::encodeValue(Reloc.Type, Value,
                                     SectionAddress + Reloc.Offset);
 
@@ -197,6 +217,11 @@ void BinarySection::flushPendingRelocations(raw_pwrite_stream &OS,
   }
 
   clearList(PendingRelocations);
+
+  if (SkippedPendingRelocations > 0 && opts::Verbosity >= 1) {
+    BC.outs() << "BOLT-INFO: skipped " << SkippedPendingRelocations
+              << " out-of-range optional relocations\n";
+  }
 }
 
 BinarySection::~BinarySection() { updateContents(nullptr, 0); }

bolt/lib/Core/Relocation.cpp

@@ -271,6 +271,16 @@ static uint64_t encodeValueX86(uint32_t Type, uint64_t Value, uint64_t PC) {
   return Value;
 }
 
+static bool canEncodeValueAArch64(uint32_t Type, uint64_t Value, uint64_t PC) {
+  switch (Type) {
+  default:
+    llvm_unreachable("unsupported relocation");
+  case ELF::R_AARCH64_CALL26:
+  case ELF::R_AARCH64_JUMP26:
+    return isInt<28>(Value - PC);
+  }
+}
+
 static uint64_t encodeValueAArch64(uint32_t Type, uint64_t Value, uint64_t PC) {
   switch (Type) {
   default:
@@ -303,6 +313,16 @@ static uint64_t encodeValueAArch64(uint32_t Type, uint64_t Value, uint64_t PC) {
   return Value;
 }
 
+static uint64_t canEncodeValueRISCV(uint32_t Type, uint64_t Value,
+                                    uint64_t PC) {
+  switch (Type) {
+  default:
+    llvm_unreachable("unsupported relocation");
+  case ELF::R_RISCV_64:
+    return true;
+  }
+}
+
 static uint64_t encodeValueRISCV(uint32_t Type, uint64_t Value, uint64_t PC) {
   switch (Type) {
   default:
@@ -739,6 +759,19 @@ uint64_t Relocation::encodeValue(uint32_t Type, uint64_t Value, uint64_t PC) {
   }
 }
 
+bool Relocation::canEncodeValue(uint32_t Type, uint64_t Value, uint64_t PC) {
+  switch (Arch) {
+  default:
+    llvm_unreachable("Unsupported architecture");
+  case Triple::aarch64:
+    return canEncodeValueAArch64(Type, Value, PC);
+  case Triple::riscv64:
+    return canEncodeValueRISCV(Type, Value, PC);
+  case Triple::x86_64:
+    return true;
+  }
+}
+
 uint64_t Relocation::extractValue(uint32_t Type, uint64_t Contents,
                                   uint64_t PC) {
   switch (Arch) {

bolt/lib/Rewrite/RewriteInstance.cpp

@@ -2827,9 +2827,10 @@ void RewriteInstance::handleRelocation(const SectionRef &RelocatedSection,
     if (SymbolAddress == 0)
       ReferencedSymbol = BC->registerNameAtAddress(SymbolName, 0, 0, 0);
 
-    LLVM_DEBUG(dbgs() << "BOLT-DEBUG: forcing relocation against symbol "
-                      << ReferencedSymbol->getName() << " with addend "
-                      << Addend << '\n');
+    LLVM_DEBUG(
+        dbgs() << "BOLT-DEBUG: forcing relocation against symbol "
+               << (ReferencedSymbol ? ReferencedSymbol->getName() : "<none>")
+               << " with addend " << Addend << '\n');
   } else if (ReferencedBF) {
     ReferencedSymbol = ReferencedBF->getSymbol();
     uint64_t RefFunctionOffset = 0;

bolt/lib/Target/AArch64/AArch64MCSymbolizer.cpp

@@ -119,26 +119,39 @@ AArch64MCSymbolizer::adjustRelocation(const Relocation &Rel,
     // The ADRP+LDR sequence was converted into ADRP+ADD. We are looking at the
     // second instruction and have to use the relocation type for ADD.
     AdjustedRel.Type = ELF::R_AARCH64_ADD_ABS_LO12_NC;
-  } else {
-    // For instructions that reference GOT, ignore the referenced symbol and
-    // use value at the relocation site. FixRelaxationPass will look at
-    // instruction pairs and will perform necessary adjustments.
+    return AdjustedRel;
+  }
+
+  // ADRP is a special case since the linker can leave the instruction opcode
+  // intact and modify only the operand. We are doing our best to detect when
+  // such conversion has happened without looking at the next instruction.
+  //
+  // If we detect that a page referenced by the ADRP cannot belong to GOT, and
+  // that it matches the symbol from the relocation, then we can be certain
+  // that the linker converted the GOT reference into the local one. Otherwise,
+  // we leave the disambiguation resolution to FixRelaxationPass.
+  //
+  // Note that ADRP relaxation described above cannot happen for TLS relocation.
+  // Since TLS relocations may not even have a valid symbol (not supported by
+  // BOLT), we explicitly exclude them from the check.
+  if (BC.MIB->isADRP(Inst) && Rel.Addend == 0 && !Relocation::isTLS(Rel.Type)) {
     ErrorOr<uint64_t> SymbolValue = BC.getSymbolValue(*Rel.Symbol);
     assert(SymbolValue && "Symbol value should be set");
     const uint64_t SymbolPageAddr = *SymbolValue & ~0xfffULL;
 
-    // Check if defined symbol and GOT are on the same page. If they are not,
-    // disambiguate the operand.
-    if (BC.MIB->isADRP(Inst) && Rel.Addend == 0 &&
-        SymbolPageAddr == Rel.Value &&
+    if (SymbolPageAddr == Rel.Value &&
         !isPageAddressValidForGOT(SymbolPageAddr)) {
       AdjustedRel.Type = ELF::R_AARCH64_ADR_PREL_PG_HI21;
-    } else {
-      AdjustedRel.Symbol = BC.registerNameAtAddress("__BOLT_got_zero", 0, 0, 0);
-      AdjustedRel.Addend = Rel.Value;
+      return AdjustedRel;
     }
   }
 
+  // For instructions that reference GOT, ignore the referenced symbol and
+  // use value at the relocation site. FixRelaxationPass will look at
+  // instruction pairs and will perform necessary adjustments.
+  AdjustedRel.Symbol = BC.registerNameAtAddress("__BOLT_got_zero", 0, 0, 0);
+  AdjustedRel.Addend = Rel.Value;
+
   return AdjustedRel;
 }
 

bolt/test/AArch64/tls.c

@@ -34,3 +34,11 @@ int main() {
 // RUN:   -target aarch64-linux -fuse-ld=lld \
 // RUN:   -nostdlib
 // RUN: llvm-bolt %t_pie.exe -o %t.bolt
+
+// RUN: %clang %cflags -fPIC -shared %s -o %t.so -Wl,-q -fuse-ld=lld
+// RUN: llvm-objdump -d -r --disassemble-symbols=main %t.so | FileCheck %s
+// RUN: llvm-bolt %t.so -o %t.bolt.so
+
+// Verify that unoptimized TLS access was generated for shared object.
+// CHECK:      adrp    x0
+// CHECK-NEXT: R_AARCH64_TLSDESC_ADR_PAGE21     tbssstruct

bolt/unittests/Core/BinaryContext.cpp

@@ -7,6 +7,7 @@
 //===----------------------------------------------------------------------===//
 
 #include "bolt/Core/BinaryContext.h"
+#include "bolt/Utils/CommandLineOpts.h"
 #include "llvm/BinaryFormat/ELF.h"
 #include "llvm/DebugInfo/DWARF/DWARFContext.h"
 #include "llvm/Support/TargetSelect.h"
@@ -161,6 +162,67 @@ TEST_P(BinaryContextTester, FlushPendingRelocJUMP26) {
       << "Wrong forward branch value\n";
 }
 
+TEST_P(BinaryContextTester,
+       FlushOptionalOutOfRangePendingRelocCALL26_ForcePatchOff) {
+  if (GetParam() != Triple::aarch64)
+    GTEST_SKIP();
+
+  // Tests that flushPendingRelocations exits if any pending relocation is out
+  // of range and PatchEntries hasn't run. Pending relocations are added by
+  // scanExternalRefs, so this ensures that either all scanExternalRefs
+  // relocations were flushed or PatchEntries ran.
+
+  BinarySection &BS = BC->registerOrUpdateSection(
+      ".text", ELF::SHT_PROGBITS, ELF::SHF_EXECINSTR | ELF::SHF_ALLOC);
+  // Create symbol 'Func0x4'
+  MCSymbol *RelSymbol = BC->getOrCreateGlobalSymbol(4, "Func");
+  ASSERT_TRUE(RelSymbol);
+  Relocation Reloc{8, RelSymbol, ELF::R_AARCH64_CALL26, 0, 0};
+  Reloc.setOptional();
+  BS.addPendingRelocation(Reloc);
+
+  SmallVector<char> Vect;
+  raw_svector_ostream OS(Vect);
+
+  // Resolve relocation symbol to a high value so encoding will be out of range.
+  EXPECT_EXIT(BS.flushPendingRelocations(
+                  OS, [&](const MCSymbol *S) { return 0x800000F; }),
+              ::testing::ExitedWithCode(1),
+              "BOLT-ERROR: cannot encode relocation for symbol Func0x4 as it is"
+              " out-of-range. To proceed must use -force-patch");
+}
+
+TEST_P(BinaryContextTester,
+       FlushOptionalOutOfRangePendingRelocCALL26_ForcePatchOn) {
+  if (GetParam() != Triple::aarch64)
+    GTEST_SKIP();
+
+  // Tests that flushPendingRelocations can skip flushing any optional pending
+  // relocations that cannot be encoded, given that PatchEntries runs.
+  opts::ForcePatch = true;
+
+  opts::Verbosity = 1;
+  testing::internal::CaptureStdout();
+
+  BinarySection &BS = BC->registerOrUpdateSection(
+      ".text", ELF::SHT_PROGBITS, ELF::SHF_EXECINSTR | ELF::SHF_ALLOC);
+  MCSymbol *RelSymbol = BC->getOrCreateGlobalSymbol(4, "Func");
+  ASSERT_TRUE(RelSymbol);
+  Relocation Reloc{8, RelSymbol, ELF::R_AARCH64_CALL26, 0, 0};
+  Reloc.setOptional();
+  BS.addPendingRelocation(Reloc);
+
+  SmallVector<char> Vect;
+  raw_svector_ostream OS(Vect);
+
+  // Resolve relocation symbol to a high value so encoding will be out of range.
+  BS.flushPendingRelocations(OS, [&](const MCSymbol *S) { return 0x800000F; });
+  outs().flush();
+  std::string CapturedStdOut = testing::internal::GetCapturedStdout();
+  EXPECT_EQ(CapturedStdOut,
+            "BOLT-INFO: skipped 1 out-of-range optional relocations\n");
+}
+
 #endif
 
 TEST_P(BinaryContextTester, BaseAddress) {

bolt/unittests/Core/CMakeLists.txt

@@ -19,6 +19,7 @@ target_link_libraries(CoreTests
   LLVMBOLTCore
   LLVMBOLTRewrite
   LLVMBOLTProfile
+  LLVMBOLTUtils
   LLVMTestingSupport
   )
 

clang-tools-extra/clang-tidy/ExpandModularHeadersPPCallbacks.cpp

@@ -89,15 +89,14 @@ ExpandModularHeadersPPCallbacks::ExpandModularHeadersPPCallbacks(
   HeaderInfo = std::make_unique<HeaderSearch>(HSOpts, Sources, Diags, LangOpts,
                                               &Compiler.getTarget());
 
-  auto PO = std::make_shared<PreprocessorOptions>();
-  *PO = Compiler.getPreprocessorOpts();
-
-  PP = std::make_unique<clang::Preprocessor>(PO, Diags, LangOpts, Sources,
-                                              *HeaderInfo, ModuleLoader,
-                                              /*IILookup=*/nullptr,
-                                              /*OwnsHeaderSearch=*/false);
+  PP = std::make_unique<clang::Preprocessor>(Compiler.getPreprocessorOpts(),
+                                             Diags, LangOpts, Sources,
+                                             *HeaderInfo, ModuleLoader,
+                                             /*IILookup=*/nullptr,
+                                             /*OwnsHeaderSearch=*/false);
   PP->Initialize(Compiler.getTarget(), Compiler.getAuxTarget());
-  InitializePreprocessor(*PP, *PO, Compiler.getPCHContainerReader(),
+  InitializePreprocessor(*PP, Compiler.getPreprocessorOpts(),
+                         Compiler.getPCHContainerReader(),
                          Compiler.getFrontendOpts(), Compiler.getCodeGenOpts());
   ApplyHeaderSearchOptions(*HeaderInfo, HSOpts, LangOpts,
                            Compiler.getTarget().getTriple());

clang-tools-extra/clang-tidy/misc/RedundantExpressionCheck.cpp

@@ -927,18 +927,6 @@ static bool areExprsSameMacroOrLiteral(const BinaryOperator *BinOp,
   if (Lil && Ril)
     return Lil->getValue() == Ril->getValue();
 
-  const auto *LStrl = dyn_cast<StringLiteral>(Lhs);
-  const auto *RStrl = dyn_cast<StringLiteral>(Rhs);
-  if (Lil && Ril) {
-    const llvm::StringRef L = Lexer::getSourceText(
-        CharSourceRange::getTokenRange(LStrl->getBeginLoc()), SM,
-        Context->getLangOpts(), 0);
-    const llvm::StringRef R = Lexer::getSourceText(
-        CharSourceRange::getTokenRange(RStrl->getBeginLoc()), SM,
-        Context->getLangOpts(), 0);
-    return L.compare(R) == 0;
-  }
-
   const auto *Lbl = dyn_cast<CXXBoolLiteralExpr>(Lhs);
   const auto *Rbl = dyn_cast<CXXBoolLiteralExpr>(Rhs);
   if (Lbl && Rbl)

clang-tools-extra/clangd/ModulesBuilder.cpp

@@ -202,9 +202,10 @@ bool IsModuleFileUpToDate(PathRef ModuleFilePath,
   HeaderSearch HeaderInfo(HSOpts, SourceMgr, *Diags, LangOpts,
                           /*Target=*/nullptr);
 
+  PreprocessorOptions PPOpts;
   TrivialModuleLoader ModuleLoader;
-  Preprocessor PP(std::make_shared<PreprocessorOptions>(), *Diags, LangOpts,
-                  SourceMgr, HeaderInfo, ModuleLoader);
+  Preprocessor PP(PPOpts, *Diags, LangOpts, SourceMgr, HeaderInfo,
+                  ModuleLoader);
 
   IntrusiveRefCntPtr<ModuleCache> ModCache = createCrossProcessModuleCache();
   PCHContainerOperations PCHOperations;