Subzero: Basic Block Profiler.

src/IceCfg.h

 //===- subzero/src/IceCfg.h - Control flow graph ----------------*- C++ -*-===//
 //
 //                        The Subzero Code Generator
 //
 // This file is distributed under the University of Illinois Open Source
 // License. See LICENSE.TXT for details.
 //
 //===----------------------------------------------------------------------===//
 //
 // This file declares the Cfg class, which represents the control flow
@@ skipping 110 matching lines @@
   const VarList &getImplicitArgs() const { return ImplicitArgs; }
 
   // Miscellaneous accessors.
   TargetLowering *getTarget() const { return Target.get(); }
   VariablesMetadata *getVMetadata() const { return VMetadata.get(); }
   Liveness *getLiveness() const { return Live.get(); }
   template <typename T = Assembler> T *getAssembler() const {
     return static_cast<T *>(TargetAssembler.get());
   }
   Assembler *releaseAssembler() { return TargetAssembler.release(); }
+  std::unique_ptr<VariableDeclarationList> getGlobalInits() {
+    return std::move(GlobalInits);
+  }
   bool hasComputedFrame() const;
   bool getFocusedTiming() const { return FocusedTiming; }
   void setFocusedTiming() { FocusedTiming = true; }
 
+  // Returns true if Var is a global variable that is used by the profiling
+  // code.
+  static bool isProfileGlobal(const VariableDeclaration &Var);
+
   // Passes over the CFG.
   void translate();
   // After the CFG is fully constructed, iterate over the nodes and
   // compute the predecessor and successor edges, in the form of
   // CfgNode::InEdges[] and CfgNode::OutEdges[].
   void computeInOutEdges();
   void renumberInstructions();
   void placePhiLoads();
   void placePhiStores();
   void deletePhis();
@@ skipping 36 matching lines @@
   }
 
   // Deallocate data that was allocated via allocateArrayOf<T>().
   template <typename T> void deallocateArrayOf(T *Array) {
     Allocator->Deallocate(Array);
   }
 
 private:
   Cfg(GlobalContext *Ctx, uint32_t SequenceNumber);
 
+  // Adds a call to the ProfileSummary runtime function as the first instruction
+  // in this CFG's entry block.
+  void addCallToProfileSummary();
+
+  // Iterates over the basic blocks in this CFG, adding profiling code to each
+  // one of them. It returns a list with all the globals that the profiling code
+  // needs to be defined.
+  void profileBlocks();
+
   GlobalContext *Ctx;
   uint32_t SequenceNumber; // output order for emission
   VerboseMask VMask;
   IceString FunctionName;
   Type ReturnType;
   bool IsInternalLinkage;
   bool HasError;
   bool FocusedTiming;
   IceString ErrorMessage;
   CfgNode *Entry; // entry basic block
   NodeList Nodes; // linearized node list; Entry should be first
   std::vector<IceString> IdentifierNames;
   InstNumberT NextInstNumber;
   VarList Variables;
   VarList Args;         // subset of Variables, in argument order
   VarList ImplicitArgs; // subset of Variables
   std::unique_ptr<ArenaAllocator<>> Allocator;
   std::unique_ptr<Liveness> Live;
   std::unique_ptr<TargetLowering> Target;
   std::unique_ptr<VariablesMetadata> VMetadata;
   std::unique_ptr<Assembler> TargetAssembler;
+  // Globals required by this CFG. Mostly used for the profiler's globals.
+  std::unique_ptr<VariableDeclarationList> GlobalInits;
 
   // CurrentNode is maintained during dumping/emitting just for
   // validating Variable::DefNode.  Normally, a traversal over
   // CfgNodes maintains this, but before global operations like
   // register allocation, resetCurrentNode() should be called to avoid
   // spurious validation failures.
   const CfgNode *CurrentNode;
 
   // Maintain a pointer in TLS to the current Cfg being translated.
   // This is primarily for accessing its allocator statelessly, but
   // other uses are possible.
   ICE_TLS_DECLARE_FIELD(const Cfg *, CurrentCfg);
 
 public:
   static void TlsInit() { ICE_TLS_INIT_FIELD(CurrentCfg); }
 };
 
 } // end of namespace Ice
 
 #endif // SUBZERO_SRC_ICECFG_H