[SYCL][Graph] Update doc for UR PR moving reset commands to a dedicated cmd-list

.ci/generate-buildkite-pipeline-premerge

@@ -233,7 +233,10 @@ linux_projects=$(add-dependencies ${linux_projects_to_test} | sort | uniq)
 
 windows_projects_to_test=$(exclude-windows $(compute-projects-to-test ${modified_projects}))
 windows_check_targets=$(check-targets ${windows_projects_to_test} | sort | uniq)
-windows_projects=$(add-dependencies ${windows_projects_to_test} | sort | uniq)
+# Temporary disable the windows job.
+# See https://discourse.llvm.org/t/rfc-future-of-windows-pre-commit-ci/76840
+#windows_projects=$(add-dependencies ${windows_projects_to_test} | sort | uniq)
+windows_projects=""
 
 # Generate the appropriate pipeline
 if [[ "${linux_projects}" != "" ]]; then

.git-blame-ignore-revs

@@ -78,3 +78,6 @@ f6d557ee34b6bbdb1dc32f29e34b4a4a8ad35e81
 082b89b25faae3e45a023caf51b65ca0f02f377f
 0ba22f51d128bee9d69756c56c4678097270e10b
 84da0e1bb75f8666cf222d2f600f37bebb9ea389
+
+# [NFC] clang-format utils/TableGen (#80973)
+b9079baaddfed5e604fbfaa1d81a7a1c38e78c26

.github/CODEOWNERS

@@ -50,8 +50,9 @@ sycl/plugins/**/hip/ @intel/llvm-reviewers-cuda
 # CUDA specific runtime implementations
 sycl/include/sycl/ext/oneapi/experimental/cuda/ @intel/llvm-reviewers-cuda
 
-# CUDA device code tests
+# CUDA and HIP device code tests
 sycl/test/check_device_code/cuda/ @intel/llvm-reviewers-cuda
+sycl/test/check_device_code/hip/ @intel/llvm-reviewers-cuda
 
 # XPTI instrumentation utilities
 xpti/ @intel/llvm-reviewers-runtime

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

@@ -65,4 +65,5 @@ jobs:
           release-workflow \
           --branch-repo-token ${{ secrets.RELEASE_WORKFLOW_PUSH_SECRET }} \
           --issue-number ${{ github.event.issue.number }} \
+          --requested-by ${{ github.event.issue.user.login }} \
           auto

.github/workflows/llvm-project-workflow-tests.yml

@@ -0,0 +1,32 @@
+# This workflow will test the llvm-project-tests workflow in PRs
+# targetting the main branch.  Since this workflow doesn't normally
+# run on main PRs, we need some way to test it to ensure new updates
+# don't break it.
+
+name: LLVM Workflow Test
+
+permissions:
+  contents: read
+
+on:
+  pull_request:
+    branches:
+      - 'main'
+    paths:
+      - '.github/workflows/llvm-project-tests.yml'
+      - '.github/workflows/llvm-project-workflow-tests.yml'
+
+concurrency:
+  # Skip intermediate builds: always.
+  # Cancel intermediate builds: only if it is a pull request build.
+  group: ${{ github.workflow }}-${{ github.ref }}
+  cancel-in-progress: ${{ startsWith(github.ref, 'refs/pull/') }}
+
+jobs:
+  llvm-test:
+    if: github.repository_owner == 'llvm'
+    name: Build and Test
+    uses: ./.github/workflows/llvm-project-tests.yml
+    with:
+      build_target: check-all
+      projects: clang;lld;libclc;lldb

bolt/include/bolt/Core/BinaryContext.h

@@ -145,6 +145,35 @@ class FilterIterator {
   }
 };
 
+/// BOLT-exclusive errors generated in core BOLT libraries, optionally holding a
+/// string message and whether it is fatal or not. In case it is fatal and if
+/// BOLT is running as a standalone process, the process might be killed as soon
+/// as the error is checked.
+class BOLTError : public ErrorInfo<BOLTError> {
+public:
+  static char ID;
+
+  BOLTError(bool IsFatal, const Twine &S = Twine());
+  void log(raw_ostream &OS) const override;
+  bool isFatal() const { return IsFatal; }
+
+  const std::string &getMessage() const { return Msg; }
+  std::error_code convertToErrorCode() const override;
+
+private:
+  bool IsFatal;
+  std::string Msg;
+};
+
+/// Streams used by BOLT to log regular or error events
+struct JournalingStreams {
+  raw_ostream &Out;
+  raw_ostream &Err;
+};
+
+Error createNonFatalBOLTError(const Twine &S);
+Error createFatalBOLTError(const Twine &S);
+
 class BinaryContext {
   BinaryContext() = delete;
 
@@ -237,7 +266,8 @@ class BinaryContext {
 public:
   static Expected<std::unique_ptr<BinaryContext>>
   createBinaryContext(const ObjectFile *File, bool IsPIC,
-                      std::unique_ptr<DWARFContext> DwCtx);
+                      std::unique_ptr<DWARFContext> DwCtx,
+                      JournalingStreams Logger);
 
   /// Superset of compiler units that will contain overwritten code that needs
   /// new debug info. In a few cases, functions may end up not being
@@ -605,6 +635,10 @@ class BinaryContext {
 
   std::unique_ptr<MCAsmBackend> MAB;
 
+  /// Allows BOLT to print to log whenever it is necessary (with or without
+  /// const references)
+  mutable JournalingStreams Logger;
+
   /// Indicates if the binary is Linux kernel.
   bool IsLinuxKernel{false};
 
@@ -737,7 +771,8 @@ class BinaryContext {
                 std::unique_ptr<const MCInstrAnalysis> MIA,
                 std::unique_ptr<MCPlusBuilder> MIB,
                 std::unique_ptr<const MCRegisterInfo> MRI,
-                std::unique_ptr<MCDisassembler> DisAsm);
+                std::unique_ptr<MCDisassembler> DisAsm,
+                JournalingStreams Logger);
 
   ~BinaryContext();
 
@@ -1349,8 +1384,12 @@ class BinaryContext {
     return Offset;
   }
 
-  void exitWithBugReport(StringRef Message,
-                         const BinaryFunction &Function) const;
+  /// Log BOLT errors to journaling streams and quit process with non-zero error
+  /// code 1 if error is fatal.
+  void logBOLTErrorsAndQuitOnFatal(Error E);
+
+  std::string generateBugReportMessage(StringRef Message,
+                                       const BinaryFunction &Function) const;
 
   struct IndependentCodeEmitter {
     std::unique_ptr<MCObjectFileInfo> LocalMOFI;
@@ -1398,6 +1437,10 @@ class BinaryContext {
     assert(IOAddressMap && "Address map not set yet");
     return *IOAddressMap;
   }
+
+  raw_ostream &outs() const { return Logger.Out; }
+
+  raw_ostream &errs() const { return Logger.Err; }
 };
 
 template <typename T, typename = std::enable_if_t<sizeof(T) == 1>>

bolt/include/bolt/Core/BinaryFunction.h

@@ -1910,12 +1910,11 @@ class BinaryFunction {
 
   /// Support dynamic relocations in constant islands, which may happen if
   /// binary is linked with -z notext option.
-  void markIslandDynamicRelocationAtAddress(uint64_t Address) {
-    if (!isInConstantIsland(Address)) {
-      errs() << "BOLT-ERROR: dynamic relocation found for text section at 0x"
-             << Twine::utohexstr(Address) << "\n";
-      exit(1);
-    }
+  Error markIslandDynamicRelocationAtAddress(uint64_t Address) {
+    if (!isInConstantIsland(Address))
+      return createFatalBOLTError(
+          Twine("dynamic relocation found for text section at 0x") +
+          Twine::utohexstr(Address) + Twine("\n"));
 
     // Mark island to have dynamic relocation
     Islands->HasDynamicRelocations = true;
@@ -1924,6 +1923,7 @@ class BinaryFunction {
     // move binary data during updateOutputValues, making us emit
     // dynamic relocation with the right offset value.
     getOrCreateIslandAccess(Address);
+    return Error::success();
   }
 
   bool hasDynamicRelocationAtIsland() const {
@@ -2054,9 +2054,10 @@ class BinaryFunction {
   /// state to State:Disassembled.
   ///
   /// Returns false if disassembly failed.
-  bool disassemble();
+  Error disassemble();
 
-  void handlePCRelOperand(MCInst &Instruction, uint64_t Address, uint64_t Size);
+  Error handlePCRelOperand(MCInst &Instruction, uint64_t Address,
+                           uint64_t Size);
 
   MCSymbol *handleExternalReference(MCInst &Instruction, uint64_t Size,
                                     uint64_t Offset, uint64_t TargetAddress,
@@ -2100,7 +2101,7 @@ class BinaryFunction {
   ///
   /// Returns true on success and update the current function state to
   /// State::CFG. Returns false if CFG cannot be built.
-  bool buildCFG(MCPlusBuilder::AllocatorIdTy);
+  Error buildCFG(MCPlusBuilder::AllocatorIdTy);
 
   /// Perform post-processing of the CFG.
   void postProcessCFG();
@@ -2217,7 +2218,7 @@ class BinaryFunction {
   }
 
   /// Process LSDA information for the function.
-  void parseLSDA(ArrayRef<uint8_t> LSDAData, uint64_t LSDAAddress);
+  Error parseLSDA(ArrayRef<uint8_t> LSDAData, uint64_t LSDAAddress);
 
   /// Update exception handling ranges for the function.
   void updateEHRanges();

bolt/include/bolt/Core/BinarySection.h

@@ -112,7 +112,7 @@ class BinarySection {
   static StringRef getName(SectionRef Section) {
     return cantFail(Section.getName());
   }
-  static StringRef getContents(SectionRef Section) {
+  static StringRef getContentsOrQuit(SectionRef Section) {
     if (Section.getObject()->isELF() &&
         ELFSectionRef(Section).getType() == ELF::SHT_NOBITS)
       return StringRef();
@@ -159,7 +159,7 @@ class BinarySection {
 
   BinarySection(BinaryContext &BC, SectionRef Section)
       : BC(BC), Name(getName(Section)), Section(Section),
-        Contents(getContents(Section)), Address(Section.getAddress()),
+        Contents(getContentsOrQuit(Section)), Address(Section.getAddress()),
         Size(Section.getSize()), Alignment(Section.getAlignment().value()),
         OutputName(Name), SectionNumber(++Count) {
     if (isELF()) {

bolt/include/bolt/Core/DIEBuilder.h

@@ -15,6 +15,7 @@
 #ifndef BOLT_CORE_DIE_BUILDER_H
 #define BOLT_CORE_DIE_BUILDER_H
 
+#include "bolt/Core/BinaryContext.h"
 #include "llvm/CodeGen/DIE.h"
 #include "llvm/DebugInfo/DWARF/DWARFAbbreviationDeclaration.h"
 #include "llvm/DebugInfo/DWARF/DWARFDie.h"
@@ -32,6 +33,7 @@
 namespace llvm {
 
 namespace bolt {
+
 class DIEStreamer;
 class DebugStrOffsetsWriter;
 
@@ -120,6 +122,7 @@ class DIEBuilder {
   std::unique_ptr<State> BuilderState;
   FoldingSet<DIEAbbrev> AbbreviationsSet;
   std::vector<std::unique_ptr<DIEAbbrev>> Abbreviations;
+  BinaryContext &BC;
   DWARFContext *DwarfContext{nullptr};
   bool IsDWO{false};
   uint64_t UnitSize{0};
@@ -219,9 +222,10 @@ class DIEBuilder {
     if (getState().CloneUnitCtxMap[UnitId].DieInfoVector.size() > DIEId)
       return *getState().CloneUnitCtxMap[UnitId].DieInfoVector[DIEId].get();
 
-    errs() << "BOLT-WARNING: [internal-dwarf-error]: The DIE is not allocated "
-              "before looking up, some"
-           << "unexpected corner cases happened.\n";
+    BC.errs()
+        << "BOLT-WARNING: [internal-dwarf-error]: The DIE is not allocated "
+           "before looking up, some"
+        << "unexpected corner cases happened.\n";
     return *getState().CloneUnitCtxMap[UnitId].DieInfoVector.front().get();
   }
 
@@ -261,7 +265,7 @@ class DIEBuilder {
   DIE *constructDIEFast(DWARFDie &DDie, DWARFUnit &U, uint32_t UnitId);
 
 public:
-  DIEBuilder(DWARFContext *DwarfContext, bool IsDWO = false);
+  DIEBuilder(BinaryContext &BC, DWARFContext *DwarfContext, bool IsDWO = false);
 
   /// Returns enum to what we are currently processing.
   ProcessingType getCurrentProcessingState() { return getState().Type; }
@@ -295,8 +299,9 @@ class DIEBuilder {
     if (getState().TypeDIEMap.count(&DU))
       return getState().TypeDIEMap[&DU];
 
-    errs() << "BOLT-ERROR: unable to find TypeUnit for Type Unit at offset 0x"
-           << DU.getOffset() << "\n";
+    BC.errs()
+        << "BOLT-ERROR: unable to find TypeUnit for Type Unit at offset 0x"
+        << DU.getOffset() << "\n";
     return nullptr;
   }
 

bolt/include/bolt/Core/DynoStats.h

@@ -159,8 +159,9 @@ inline DynoStats getDynoStats(FuncsType &Funcs, bool IsAArch64) {
 
 /// Call a function with optional before and after dynostats printing.
 template <typename FnType, typename FuncsType>
-inline void callWithDynoStats(FnType &&Func, FuncsType &Funcs, StringRef Phase,
-                              const bool Flag, bool IsAArch64) {
+inline void callWithDynoStats(raw_ostream &OS, FnType &&Func, FuncsType &Funcs,
+                              StringRef Phase, const bool Flag,
+                              bool IsAArch64) {
   DynoStats DynoStatsBefore(IsAArch64);
   if (Flag)
     DynoStatsBefore = getDynoStats(Funcs, IsAArch64);
@@ -170,12 +171,12 @@ inline void callWithDynoStats(FnType &&Func, FuncsType &Funcs, StringRef Phase,
   if (Flag) {
     const DynoStats DynoStatsAfter = getDynoStats(Funcs, IsAArch64);
     const bool Changed = (DynoStatsAfter != DynoStatsBefore);
-    outs() << "BOLT-INFO: program-wide dynostats after running " << Phase
-           << (Changed ? "" : " (no change)") << ":\n\n"
-           << DynoStatsBefore << '\n';
+    OS << "BOLT-INFO: program-wide dynostats after running " << Phase
+       << (Changed ? "" : " (no change)") << ":\n\n"
+       << DynoStatsBefore << '\n';
     if (Changed)
-      DynoStatsAfter.print(outs(), &DynoStatsBefore);
-    outs() << '\n';
+      DynoStatsAfter.print(OS, &DynoStatsBefore);
+    OS << '\n';
   }
 }
 

bolt/include/bolt/Core/Exceptions.h

@@ -30,13 +30,14 @@ class FDE;
 
 namespace bolt {
 
+class BinaryContext;
 class BinaryFunction;
 
 /// \brief Wraps up information to read all CFI instructions and feed them to a
 /// BinaryFunction, as well as rewriting CFI sections.
 class CFIReaderWriter {
 public:
-  explicit CFIReaderWriter(const DWARFDebugFrame &EHFrame);
+  explicit CFIReaderWriter(BinaryContext &BC, const DWARFDebugFrame &EHFrame);
 
   bool fillCFIInfoFor(BinaryFunction &Function) const;
 
@@ -59,6 +60,7 @@ class CFIReaderWriter {
   const FDEsMap &getFDEs() const { return FDEs; }
 
 private:
+  BinaryContext &BC;
   FDEsMap FDEs;
 };
 

bolt/include/bolt/Passes/ADRRelaxationPass.h

@@ -30,7 +30,7 @@ class ADRRelaxationPass : public BinaryFunctionPass {
   const char *getName() const override { return "adr-relaxation"; }
 
   /// Pass entry point
-  void runOnFunctions(BinaryContext &BC) override;
+  Error runOnFunctions(BinaryContext &BC) override;
   void runOnFunction(BinaryFunction &BF);
 };
 

bolt/include/bolt/Passes/Aligner.h

@@ -39,7 +39,7 @@ class AlignerPass : public BinaryFunctionPass {
   const char *getName() const override { return "aligner"; }
 
   /// Pass entry point
-  void runOnFunctions(BinaryContext &BC) override;
+  Error runOnFunctions(BinaryContext &BC) override;
 };
 
 } // namespace bolt

bolt/include/bolt/Passes/AllocCombiner.h

@@ -33,7 +33,7 @@ class AllocCombinerPass : public BinaryFunctionPass {
   }
 
   /// Pass entry point
-  void runOnFunctions(BinaryContext &BC) override;
+  Error runOnFunctions(BinaryContext &BC) override;
 };
 
 } // namespace bolt

bolt/include/bolt/Passes/AsmDump.h

@@ -28,7 +28,7 @@ class AsmDumpPass : public BinaryFunctionPass {
   bool shouldPrint(const BinaryFunction &BF) const override { return false; }
 
   /// Pass entry point
-  void runOnFunctions(BinaryContext &BC) override;
+  Error runOnFunctions(BinaryContext &BC) override;
 };
 
 } // namespace bolt