Merge branch 'main' into riscv-address-resolution

Author: arjunUpatel

.github/new-prs-labeler.yml

@@ -777,6 +777,10 @@ backend:NVPTX:
   - 'llvm/**/*nvptx*/**'
   - 'llvm/**/*NVPTX*/**'
 
+backend:MIPS:
+  - '**/*mips*'
+  - '**/*Mips*'
+
 backend:RISC-V:
   - clang/**/*riscv*
   - clang/**/*RISCV*

.github/workflows/email-check.yaml

@@ -32,7 +32,8 @@ jobs:
           COMMENT: >-
             ⚠️ We detected that you are using a GitHub private e-mail address to contribute to the repo.<br/>
             Please turn off [Keep my email addresses private](https://github.com/settings/emails) setting in your account.<br/>
-            See [LLVM Discourse](https://discourse.llvm.org/t/hidden-emails-on-github-should-we-do-something-about-it) for more information.
+            See [LLVM Developer Policy](https://llvm.org/docs/DeveloperPolicy.html#email-addresses) and
+            [LLVM Discourse](https://discourse.llvm.org/t/hidden-emails-on-github-should-we-do-something-about-it) for more information.
         run: |
           cat << EOF > comments
           [{"body" : "$COMMENT"}]

.github/workflows/libcxx-build-and-test.yaml

@@ -52,8 +52,8 @@ jobs:
         cxx: [ 'clang++-21' ]
         include:
           - config: 'generic-gcc'
-            cc: 'gcc-14'
-            cxx: 'g++-14'
+            cc: 'gcc-15'
+            cxx: 'g++-15'
     steps:
       - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
       - name: ${{ matrix.config }}.${{ matrix.cxx }}
@@ -92,8 +92,8 @@ jobs:
         cxx: [ 'clang++-21' ]
         include:
           - config: 'generic-gcc-cxx11'
-            cc: 'gcc-14'
-            cxx: 'g++-14'
+            cc: 'gcc-15'
+            cxx: 'g++-15'
           - config: 'generic-cxx26'
             cc: 'clang-20'
             cxx: 'clang++-20'

.github/workflows/libcxx-restart-preempted-jobs.yaml

@@ -33,7 +33,7 @@ jobs:
         with:
           script: |
             const failure_regex = /Process completed with exit code 1./
-            const preemption_regex = /The runner has received a shutdown signal/
+            const preemption_regex = /(The runner has received a shutdown signal)|(The operation was canceled)/
 
             const wf_run = context.payload.workflow_run
             core.notice(`Running on "${wf_run.display_title}" by @${wf_run.actor.login} (event: ${wf_run.event})\nWorkflow run URL: ${wf_run.html_url}`)

bolt/include/bolt/Core/BinaryFunction.h

@@ -388,6 +388,10 @@ class BinaryFunction {
   /// The profile data for the number of times the function was executed.
   uint64_t ExecutionCount{COUNT_NO_PROFILE};
 
+  /// Profile data for the number of times this function was entered from
+  /// external code (DSO, JIT, etc).
+  uint64_t ExternEntryCount{0};
+
   /// Profile match ratio.
   float ProfileMatchRatio{0.0f};
 
@@ -1877,6 +1881,10 @@ class BinaryFunction {
     return *this;
   }
 
+  /// Set the profile data for the number of times the function was entered from
+  /// external code (DSO/JIT).
+  void setExternEntryCount(uint64_t Count) { ExternEntryCount = Count; }
+
   /// Adjust execution count for the function by a given \p Count. The value
   /// \p Count will be subtracted from the current function count.
   ///
@@ -1904,6 +1912,10 @@ class BinaryFunction {
   /// Return COUNT_NO_PROFILE if there's no profile info.
   uint64_t getExecutionCount() const { return ExecutionCount; }
 
+  /// Return the profile information about the number of times the function was
+  /// entered from external code (DSO/JIT).
+  uint64_t getExternEntryCount() const { return ExternEntryCount; }
+
   /// Return the raw profile information about the number of branch
   /// executions corresponding to this function.
   uint64_t getRawSampleCount() const { return RawSampleCount; }

bolt/include/bolt/Passes/PAuthGadgetScanner.h

@@ -199,8 +199,7 @@ namespace PAuthGadgetScanner {
 // to distinguish intermediate and final results at the type level.
 //
 // Here is an overview of issue life-cycle:
-// * an analysis (SrcSafetyAnalysis at now, DstSafetyAnalysis will be added
-//   later to support the detection of authentication oracles) computes register
+// * an analysis (SrcSafetyAnalysis or DstSafetyAnalysis) computes register
 //   state for each instruction in the function.
 // * for each instruction, it is checked whether it is a gadget of some kind,
 //   taking the computed state into account. If a gadget is found, its kind
@@ -273,6 +272,11 @@ class ExtraInfo {
   virtual ~ExtraInfo() {}
 };
 
+/// The set of instructions writing to the affected register in an unsafe
+/// manner.
+///
+/// This is a hint to be printed alongside the report. It should be further
+/// analyzed by the user.
 class ClobberingInfo : public ExtraInfo {
   SmallVector<MCInstReference> ClobberingInstrs;
 
@@ -282,6 +286,20 @@ class ClobberingInfo : public ExtraInfo {
   void print(raw_ostream &OS, const MCInstReference Location) const override;
 };
 
+/// The set of instructions leaking the authenticated pointer before the
+/// result of authentication was checked.
+///
+/// This is a hint to be printed alongside the report. It should be further
+/// analyzed by the user.
+class LeakageInfo : public ExtraInfo {
+  SmallVector<MCInstReference> LeakingInstrs;
+
+public:
+  LeakageInfo(ArrayRef<MCInstReference> Instrs) : LeakingInstrs(Instrs) {}
+
+  void print(raw_ostream &OS, const MCInstReference Location) const override;
+};
+
 /// A brief version of a report that can be further augmented with the details.
 ///
 /// A half-baked report produced on the first run of the analysis. An extra,
@@ -322,6 +340,9 @@ class FunctionAnalysisContext {
   void findUnsafeUses(SmallVector<PartialReport<MCPhysReg>> &Reports);
   void augmentUnsafeUseReports(ArrayRef<PartialReport<MCPhysReg>> Reports);
 
+  void findUnsafeDefs(SmallVector<PartialReport<MCPhysReg>> &Reports);
+  void augmentUnsafeDefReports(ArrayRef<PartialReport<MCPhysReg>> Reports);
+
   /// Process the reports which do not have to be augmented, and remove them
   /// from Reports.
   void handleSimpleReports(SmallVector<PartialReport<MCPhysReg>> &Reports);

bolt/include/bolt/Profile/DataAggregator.h

@@ -85,6 +85,8 @@ class DataAggregator : public DataReader {
   };
   friend raw_ostream &operator<<(raw_ostream &OS, const LBREntry &);
 
+  friend struct PerfSpeEventsTestHelper;
+
   struct PerfBranchSample {
     SmallVector<LBREntry, 32> LBR;
   };
@@ -99,24 +101,29 @@ class DataAggregator : public DataReader {
     uint64_t Addr;
   };
 
+  /// Container for the unit of branch data, matching pre-aggregated trace type.
+  /// Backwards compatible with branch and fall-through types:
+  /// - if \p To is < 0, the trace only contains branch data (BR_ONLY),
+  /// - if \p Branch is < 0, the trace only contains fall-through data
+  ///   (FT_ONLY, FT_EXTERNAL_ORIGIN, or FT_EXTERNAL_RETURN).
   struct Trace {
+    static constexpr const uint64_t EXTERNAL = 0ULL;
+    static constexpr const uint64_t BR_ONLY = -1ULL;
+    static constexpr const uint64_t FT_ONLY = -1ULL;
+    static constexpr const uint64_t FT_EXTERNAL_ORIGIN = -2ULL;
+    static constexpr const uint64_t FT_EXTERNAL_RETURN = -3ULL;
+
+    uint64_t Branch;
     uint64_t From;
     uint64_t To;
-    Trace(uint64_t From, uint64_t To) : From(From), To(To) {}
-    bool operator==(const Trace &Other) const {
-      return From == Other.From && To == Other.To;
-    }
+    auto tie() const { return std::tie(Branch, From, To); }
+    bool operator==(const Trace &Other) const { return tie() == Other.tie(); }
+    bool operator<(const Trace &Other) const { return tie() < Other.tie(); }
   };
+  friend raw_ostream &operator<<(raw_ostream &OS, const Trace &);
 
   struct TraceHash {
-    size_t operator()(const Trace &L) const {
-      return std::hash<uint64_t>()(L.From << 32 | L.To);
-    }
-  };
-
-  struct FTInfo {
-    uint64_t InternCount{0};
-    uint64_t ExternCount{0};
+    size_t operator()(const Trace &L) const { return hash_combine(L.tie()); }
   };
 
   struct TakenBranchInfo {
@@ -126,8 +133,11 @@ class DataAggregator : public DataReader {
 
   /// Intermediate storage for profile data. We save the results of parsing
   /// and use them later for processing and assigning profile.
-  std::unordered_map<Trace, TakenBranchInfo, TraceHash> BranchLBRs;
-  std::unordered_map<Trace, FTInfo, TraceHash> FallthroughLBRs;
+  std::unordered_map<Trace, TakenBranchInfo, TraceHash> TraceMap;
+  std::vector<std::pair<Trace, TakenBranchInfo>> Traces;
+  /// Pre-populated addresses of returns, coming from pre-aggregated data or
+  /// disassembly. Used to disambiguate call-continuation fall-throughs.
+  std::unordered_set<uint64_t> Returns;
   std::unordered_map<uint64_t, uint64_t> BasicSamples;
   std::vector<PerfMemSample> MemSamples;
 
@@ -200,8 +210,8 @@ class DataAggregator : public DataReader {
   /// Return a vector of offsets corresponding to a trace in a function
   /// if the trace is valid, std::nullopt otherwise.
   std::optional<SmallVector<std::pair<uint64_t, uint64_t>, 16>>
-  getFallthroughsInTrace(BinaryFunction &BF, const LBREntry &First,
-                         const LBREntry &Second, uint64_t Count = 1) const;
+  getFallthroughsInTrace(BinaryFunction &BF, const Trace &Trace, uint64_t Count,
+                         bool IsReturn) const;
 
   /// Record external entry into the function \p BF.
   ///
@@ -261,12 +271,14 @@ class DataAggregator : public DataReader {
                      uint64_t From, uint64_t To, uint64_t Count,
                      uint64_t Mispreds);
 
+  /// Checks if \p Addr corresponds to a return instruction.
+  bool checkReturn(uint64_t Addr);
+
   /// Register a \p Branch.
   bool doBranch(uint64_t From, uint64_t To, uint64_t Count, uint64_t Mispreds);
 
   /// Register a trace between two LBR entries supplied in execution order.
-  bool doTrace(const LBREntry &First, const LBREntry &Second,
-               uint64_t Count = 1);
+  bool doTrace(const Trace &Trace, uint64_t Count, bool IsReturn);
 
   /// Parser helpers
   /// Return false if we exhausted our parser buffer and finished parsing
@@ -379,9 +391,9 @@ class DataAggregator : public DataReader {
   /// File format syntax:
   /// E <event>
   /// S <start> <count>
-  /// T <start> <end> <ft_end> <count>
+  /// [TR] <start> <end> <ft_end> <count>
   /// B <start> <end> <count> <mispred_count>
-  /// [Ff] <start> <end> <count>
+  /// [Ffr] <start> <end> <count>
   ///
   /// where <start>, <end>, <ft_end> have the format [<id>:]<offset>
   ///
@@ -392,8 +404,11 @@ class DataAggregator : public DataReader {
   /// f - an aggregated fall-through with external origin - used to disambiguate
   ///       between a return hitting a basic block head and a regular internal
   ///       jump to the block
+  /// r - an aggregated fall-through originating at an external return, no
+  ///       checks are performed for a fallthrough start
   /// T - an aggregated trace: branch from <start> to <end> with a fall-through
   ///       to <ft_end>
+  /// R - an aggregated trace originating at a return
   ///
   /// <id> - build id of the object containing the address. We can skip it for
   /// the main binary and use "X" for an unknown object. This will save some
@@ -516,6 +531,26 @@ inline raw_ostream &operator<<(raw_ostream &OS,
   OS << formatv("{0:x} -> {1:x}/{2}", L.From, L.To, L.Mispred ? 'M' : 'P');
   return OS;
 }
+
+inline raw_ostream &operator<<(raw_ostream &OS,
+                               const DataAggregator::Trace &T) {
+  switch (T.Branch) {
+  case DataAggregator::Trace::FT_ONLY:
+    break;
+  case DataAggregator::Trace::FT_EXTERNAL_ORIGIN:
+    OS << "X:0 -> ";
+    break;
+  case DataAggregator::Trace::FT_EXTERNAL_RETURN:
+    OS << "X:R -> ";
+    break;
+  default:
+    OS << Twine::utohexstr(T.Branch) << " -> ";
+  }
+  OS << Twine::utohexstr(T.From);
+  if (T.To != DataAggregator::Trace::BR_ONLY)
+    OS << " ... " << Twine::utohexstr(T.To);
+  return OS;
+}
 } // namespace bolt
 } // namespace llvm
 

bolt/include/bolt/Profile/DataReader.h

@@ -97,6 +97,9 @@ struct FuncBranchData {
   /// Total execution count for the function.
   int64_t ExecutionCount{0};
 
+  /// Total entry count from external code for the function.
+  uint64_t ExternEntryCount{0};
+
   /// Indicate if the data was used.
   bool Used{false};
 

bolt/include/bolt/Profile/ProfileYAMLMapping.h

@@ -206,6 +206,7 @@ struct BinaryFunctionProfile {
   uint32_t Id{0};
   llvm::yaml::Hex64 Hash{0};
   uint64_t ExecCount{0};
+  uint64_t ExternEntryCount{0};
   std::vector<BinaryBasicBlockProfile> Blocks;
   std::vector<InlineTreeNode> InlineTree;
   bool Used{false};
@@ -218,6 +219,7 @@ template <> struct MappingTraits<bolt::BinaryFunctionProfile> {
     YamlIO.mapRequired("fid", BFP.Id);
     YamlIO.mapRequired("hash", BFP.Hash);
     YamlIO.mapRequired("exec", BFP.ExecCount);
+    YamlIO.mapOptional("extern", BFP.ExternEntryCount, 0);
     YamlIO.mapRequired("nblocks", BFP.NumBasicBlocks);
     YamlIO.mapOptional("blocks", BFP.Blocks,
                        std::vector<bolt::BinaryBasicBlockProfile>());

bolt/include/bolt/Utils/CommandLineOpts.h

@@ -48,6 +48,7 @@ extern llvm::cl::OptionCategory BinaryAnalysisCategory;
 extern llvm::cl::opt<unsigned> AlignText;
 extern llvm::cl::opt<unsigned> AlignFunctions;
 extern llvm::cl::opt<bool> AggregateOnly;
+extern llvm::cl::opt<bool> ArmSPE;
 extern llvm::cl::opt<unsigned> BucketsPerLine;
 extern llvm::cl::opt<bool> CompactCodeModel;
 extern llvm::cl::opt<bool> DiffOnly;

bolt/lib/Core/BinaryFunction.cpp

@@ -471,6 +471,8 @@ void BinaryFunction::print(raw_ostream &OS, std::string Annotation) {
     OS << "\n  Sample Count: " << RawSampleCount;
     OS << "\n  Profile Acc : " << format("%.1f%%", ProfileMatchRatio * 100.0f);
   }
+  if (ExternEntryCount)
+    OS << "\n  Extern Entry Count: " << ExternEntryCount;
 
   if (opts::PrintDynoStats && !getLayout().block_empty()) {
     OS << '\n';