Merge branch 'main' into sparse-tensor-loop-ordering-infrastructure

Author: gmalasan

.ci/all_requirements.txt

@@ -194,9 +194,9 @@ ml-dtypes==0.5.1 ; python_version < "3.13" \
     --hash=sha256:d13755f8e8445b3870114e5b6240facaa7cb0c3361e54beba3e07fa912a6e12b \
     --hash=sha256:fd918d4e6a4e0c110e2e05be7a7814d10dc1b95872accbf6512b80a109b71ae1
     # via -r mlir/python/requirements.txt
-nanobind==2.9.2 \
-    --hash=sha256:c37957ffd5eac7eda349cff3622ecd32e5ee1244ecc912c99b5bc8188bafd16e \
-    --hash=sha256:e7608472de99d375759814cab3e2c94aba3f9ec80e62cfef8ced495ca5c27d6e
+nanobind==2.7.0 \
+    --hash=sha256:73b12d0e751d140d6c1bf4b215e18818a8debfdb374f08dc3776ad208d808e74 \
+    --hash=sha256:f9f1b160580c50dcf37b6495a0fd5ec61dc0d95dae5f8004f87dd9ad7eb46b34
     # via -r mlir/python/requirements.txt
 numpy==2.0.2 \
     --hash=sha256:0123ffdaa88fa4ab64835dcbde75dcdf89c453c922f18dced6e27c90d1d0ec5a \
@@ -383,10 +383,6 @@ swig==4.3.1 \
     --hash=sha256:efec16327029f682f649a26da726bb0305be8800bd0f1fa3e81bf0769cf5b476 \
     --hash=sha256:fc496c0d600cf1bb2d91e28d3d6eae9c4301e5ea7a0dec5a4281b5efed4245a8
     # via -r lldb/test/requirements.txt
-typing-extensions==4.15.0 \
-    --hash=sha256:0cea48d173cc12fa28ecabc3b837ea3cf6f38c6d1136f85cbaaf598984861466 \
-    --hash=sha256:f0fa19c6845758ab08074a0cfa8b7aecb71c999ca73d62883bc25cc018c4e548
-    # via -r mlir/python/requirements.txt
 urllib3==2.5.0 \
     --hash=sha256:3fc47733c7e419d4bc3f6b3dc2b4f890bb743906a30d56ba4a5bfa4bbff92760 \
     --hash=sha256:e6b01673c0fa6a13e374b50871808eb3bf7046c4b125b216f6bf1cc604cff0dc

.github/new-prs-labeler.yml

@@ -1090,6 +1090,14 @@ clang:openmp:
   - llvm/unittests/Frontend/OpenMP*
   - llvm/test/Transforms/OpenMP/**
 
+clang:temporal-safety:
+  - clang/include/clang/Analysis/Analyses/LifetimeSafety*
+  - clang/lib/Analysis/LifetimeSafety*
+  - clang/unittests/Analysis/LifetimeSafety*
+  - clang/test/Sema/*lifetime-safety*
+  - clang/test/Sema/*lifetime-analysis*
+  - clang/test/Analysis/LifetimeSafety/**
+
 clang:as-a-library:
   - clang/tools/libclang/**
   - clang/bindings/**

.github/workflows/libcxx-run-benchmarks.yml

@@ -0,0 +1,110 @@
+# This file defines a workflow that runs the libc++ benchmarks when a comment is added to the PR.
+#
+# The comment is of the form:
+#
+#   /libcxx-bot benchmark <path-to-benchmarks-to-run>
+#
+# That will cause the specified benchmarks to be run on the PR and on the pull-request target, and
+# their results to be compared.
+
+name: Benchmark libc++
+
+permissions:
+  contents: read
+
+on:
+  issue_comment:
+    types:
+      - created
+      - edited
+
+env:
+  CC: clang-22
+  CXX: clang++-22
+
+jobs:
+  run-benchmarks:
+    permissions:
+      pull-requests: write
+
+    if: >-
+      github.event.issue.pull_request &&
+      contains(github.event.comment.body, '/libcxx-bot benchmark')
+
+    runs-on: llvm-premerge-libcxx-next-runners # TODO: This should run on a dedicated set of machines
+    steps:
+      - uses: actions/setup-python@v6
+        with:
+          python-version: '3.10'
+
+      - name: Extract information from the PR
+        id: vars
+        run: |
+          python3 -m venv .venv
+          source .venv/bin/activate
+          python -m pip install pygithub
+
+          cat <<EOF | python >> ${GITHUB_OUTPUT}
+          import github
+          repo = github.Github("${{ github.token }}").get_repo("${{ github.repository }}")
+          pr = repo.get_pull(${{ github.event.issue.number }})
+          print(f"pr_base={pr.base.sha}")
+          print(f"pr_head={pr.head.sha}")
+          EOF
+          BENCHMARKS=$(echo "${{ github.event.comment.body }}" | sed -nE 's/\/libcxx-bot benchmark (.+)/\1/p')
+          echo "benchmarks=${BENCHMARKS}" >> ${GITHUB_OUTPUT}
+
+      - uses: actions/checkout@08c6903cd8c0fde910a37f88322edcfb5dd907a8 # v5.0.0
+        with:
+          ref: ${{ steps.vars.outputs.pr_head }}
+          fetch-depth: 0
+          fetch-tags: true # This job requires access to all the Git branches so it can diff against (usually) main
+          path: repo # Avoid nuking the workspace, where we have the Python virtualenv
+
+      - name: Run baseline
+        run: |
+          source .venv/bin/activate && cd repo
+          python -m pip install -r libcxx/utils/requirements.txt
+          baseline_commit=$(git merge-base ${{ steps.vars.outputs.pr_base }} ${{ steps.vars.outputs.pr_head }})
+          ./libcxx/utils/test-at-commit --commit ${baseline_commit} -B build/baseline -- -sv -j1 --param optimization=speed ${{ steps.vars.outputs.benchmarks }}
+          ./libcxx/utils/consolidate-benchmarks build/baseline | tee baseline.lnt
+
+      - name: Run candidate
+        run: |
+          source .venv/bin/activate && cd repo
+          ./libcxx/utils/test-at-commit --commit ${{ steps.vars.outputs.pr_head }} -B build/candidate -- -sv -j1 --param optimization=speed ${{ steps.vars.outputs.benchmarks }}
+          ./libcxx/utils/consolidate-benchmarks build/candidate | tee candidate.lnt
+
+      - name: Compare baseline and candidate runs
+        run: |
+          source .venv/bin/activate && cd repo
+          ./libcxx/utils/compare-benchmarks baseline.lnt candidate.lnt | tee results.txt
+
+      - name: Update comment with results
+        run: |
+          source .venv/bin/activate && cd repo
+          cat <<EOF | python
+          import github
+          repo = github.Github("${{ github.token }}").get_repo("${{ github.repository }}")
+          pr = repo.get_pull(${{ github.event.issue.number }})
+          comment = pr.get_issue_comment(${{ github.event.comment.id }})
+          with open('results.txt', 'r') as f:
+            benchmark_results = f.read()
+
+          new_comment_text = f"""
+          {comment.body}
+
+          <details>
+          <summary>
+          Benchmark results:
+          </summary>
+
+          \`\`\`
+          {benchmark_results}
+          \`\`\`
+
+          </details>
+          """
+
+          comment.edit(new_comment_text)
+          EOF

bolt/include/bolt/Core/BinaryContext.h

@@ -288,6 +288,12 @@ class BinaryContext {
   /// overwritten, but it is okay to re-generate debug info for them.
   std::set<const DWARFUnit *> ProcessedCUs;
 
+  /// DWARF-related container to manage lifecycle of groups of rows from line
+  /// tables associated with instructions. Since binary functions can span
+  /// multiple compilation units, instructions may reference debug line
+  /// information from multiple CUs.
+  ClusteredRowsContainer ClusteredRows;
+
   // Setup MCPlus target builder
   void initializeTarget(std::unique_ptr<MCPlusBuilder> TargetBuilder) {
     MIB = std::move(TargetBuilder);

bolt/include/bolt/Core/BinaryFunction.h

@@ -35,6 +35,7 @@
 #include "bolt/Core/JumpTable.h"
 #include "bolt/Core/MCPlus.h"
 #include "bolt/Utils/NameResolver.h"
+#include "llvm/ADT/DenseMap.h"
 #include "llvm/ADT/STLExtras.h"
 #include "llvm/ADT/SmallString.h"
 #include "llvm/ADT/StringRef.h"
@@ -423,8 +424,9 @@ class BinaryFunction {
   /// Original LSDA type encoding
   unsigned LSDATypeEncoding{dwarf::DW_EH_PE_omit};
 
-  /// Containing compilation unit for the function.
-  DWARFUnit *DwarfUnit{nullptr};
+  /// All compilation units this function belongs to.
+  /// Maps DWARF unit offset to the unit pointer.
+  DenseMap<uint64_t, DWARFUnit *> DwarfUnitMap;
 
   /// Last computed hash value. Note that the value could be recomputed using
   /// different parameters by every pass.
@@ -2409,15 +2411,21 @@ class BinaryFunction {
   void
   computeBlockHashes(HashFunction HashFunction = HashFunction::Default) const;
 
-  void setDWARFUnit(DWARFUnit *Unit) { DwarfUnit = Unit; }
+  void addDWARFUnit(DWARFUnit *Unit) { DwarfUnitMap[Unit->getOffset()] = Unit; }
 
-  /// Return DWARF compile unit for this function.
-  DWARFUnit *getDWARFUnit() const { return DwarfUnit; }
+  void removeDWARFUnit(DWARFUnit *Unit) {
+    DwarfUnitMap.erase(Unit->getOffset());
+  }
+
+  /// Return DWARF compile units for this function.
+  /// Returns a reference to the map of DWARF unit offsets to units.
+  const DenseMap<uint64_t, DWARFUnit *> &getDWARFUnits() const {
+    return DwarfUnitMap;
+  }
 
-  /// Return line info table for this function.
-  const DWARFDebugLine::LineTable *getDWARFLineTable() const {
-    return getDWARFUnit() ? BC.DwCtx->getLineTableForUnit(getDWARFUnit())
-                          : nullptr;
+  const DWARFDebugLine::LineTable *
+  getDWARFLineTableForUnit(DWARFUnit *Unit) const {
+    return BC.DwCtx->getLineTableForUnit(Unit);
   }
 
   /// Finalize profile for the function.

bolt/include/bolt/Core/DebugData.h

@@ -135,8 +135,6 @@ struct DebugLineTableRowRef {
   uint32_t DwCompileUnitIndex;
   uint32_t RowIndex;
 
-  const static DebugLineTableRowRef NULL_ROW;
-
   bool operator==(const DebugLineTableRowRef &Rhs) const {
     return DwCompileUnitIndex == Rhs.DwCompileUnitIndex &&
            RowIndex == Rhs.RowIndex;
@@ -145,24 +143,6 @@ struct DebugLineTableRowRef {
   bool operator!=(const DebugLineTableRowRef &Rhs) const {
     return !(*this == Rhs);
   }
-
-  static DebugLineTableRowRef fromSMLoc(const SMLoc &Loc) {
-    union {
-      decltype(Loc.getPointer()) Ptr;
-      DebugLineTableRowRef Ref;
-    } U;
-    U.Ptr = Loc.getPointer();
-    return U.Ref;
-  }
-
-  SMLoc toSMLoc() const {
-    union {
-      decltype(SMLoc().getPointer()) Ptr;
-      DebugLineTableRowRef Ref;
-    } U;
-    U.Ref = *this;
-    return SMLoc::getFromPointer(U.Ptr);
-  }
 };
 
 /// Common buffer vector used for debug info handling.
@@ -210,7 +190,7 @@ class DebugRangesSectionWriter {
   static bool classof(const DebugRangesSectionWriter *Writer) {
     return Writer->getKind() == RangesWriterKind::DebugRangesWriter;
   }
-  
+
   /// Append a range to the main buffer.
   void appendToRangeBuffer(const DebugBufferVector &CUBuffer);
 
@@ -852,6 +832,97 @@ class DwarfLineTable {
   // Returns DWARF Version for this line table.
   uint16_t getDwarfVersion() const { return DwarfVersion; }
 };
+
+/// ClusteredRows represents a collection of debug line table row references.
+///
+/// MEMORY LAYOUT AND DESIGN:
+/// This class uses a flexible array member pattern to store all
+/// DebugLineTableRowRef elements in a single contiguous memory allocation.
+/// The memory layout is:
+///
+/// +------------------+
+/// | ClusteredRows    |  <- Object header (Size + first element)
+/// | - Size           |
+/// | - Rows (element) |  <- First DebugLineTableRowRef element
+/// +------------------+
+/// | element[1]       |  <- Additional DebugLineTableRowRef elements
+/// | element[2]       |     stored immediately after the object
+/// | ...              |
+/// | element[Size-1]  |
+/// +------------------+
+///
+/// The 'Rows' member serves as both the first element storage and the base
+/// address for pointer arithmetic to access subsequent elements.
+class ClusteredRows {
+public:
+  ArrayRef<DebugLineTableRowRef> getRows() const {
+    return ArrayRef<DebugLineTableRowRef>(beginPtrConst(), Size);
+  }
+
+  /// Returns the number of elements in the array.
+  uint64_t size() const { return Size; }
+
+  /// We re-purpose SMLoc inside MCInst to store the pointer
+  /// to ClusteredRows. fromSMLoc() and toSMLoc() are helper
+  /// functions to convert between SMLoc and ClusteredRows.
+
+  static const ClusteredRows *fromSMLoc(const SMLoc &Loc) {
+    return reinterpret_cast<const ClusteredRows *>(Loc.getPointer());
+  }
+  SMLoc toSMLoc() const {
+    return SMLoc::getFromPointer(reinterpret_cast<const char *>(this));
+  }
+
+  /// Given a vector of DebugLineTableRowRef, this method
+  /// copies the elements into pre-allocated memory.
+  template <typename T> void populate(const T Vec) {
+    assert(Vec.size() == Size && "Sizes must match");
+    DebugLineTableRowRef *CurRawPtr = beginPtr();
+    for (DebugLineTableRowRef RowRef : Vec) {
+      *CurRawPtr = RowRef;
+      ++CurRawPtr;
+    }
+  }
+
+private:
+  uint64_t Size;
+  DebugLineTableRowRef Rows;
+
+  ClusteredRows(uint64_t Size) : Size(Size) {}
+
+  /// Total size of the object including the array.
+  static uint64_t getTotalSize(uint64_t Size) {
+    assert(Size > 0 && "Size must be greater than 0");
+    return sizeof(ClusteredRows) + (Size - 1) * sizeof(DebugLineTableRowRef);
+  }
+  const DebugLineTableRowRef *beginPtrConst() const {
+    return reinterpret_cast<const DebugLineTableRowRef *>(&Rows);
+  }
+  DebugLineTableRowRef *beginPtr() {
+    return reinterpret_cast<DebugLineTableRowRef *>(&Rows);
+  }
+
+  friend class ClusteredRowsContainer;
+};
+
+/// ClusteredRowsContainer manages the lifecycle of ClusteredRows objects.
+class ClusteredRowsContainer {
+public:
+  ClusteredRows *createClusteredRows(uint64_t Size) {
+    auto *CR = new (std::malloc(ClusteredRows::getTotalSize(Size)))
+        ClusteredRows(Size);
+    Clusters.push_back(CR);
+    return CR;
+  }
+  ~ClusteredRowsContainer() {
+    for (auto *CR : Clusters)
+      std::free(CR);
+  }
+
+private:
+  std::vector<ClusteredRows *> Clusters;
+};
+
 } // namespace bolt
 } // namespace llvm
 

bolt/include/bolt/Core/MCPlusBuilder.h

@@ -2216,7 +2216,8 @@ class MCPlusBuilder {
   }
 
   /// Print each annotation attached to \p Inst.
-  void printAnnotations(const MCInst &Inst, raw_ostream &OS) const;
+  void printAnnotations(const MCInst &Inst, raw_ostream &OS,
+                        bool PrintMemData = false) const;
 
   /// Remove annotation with a given \p Index.
   ///