Change tracking of basic blocks (within function) to use a vector.

src/PNaClTranslator.cpp

 //===- subzero/src/PNaClTranslator.cpp - ICE from bitcode -----------------===//
 //
 //                        The Subzero Code Generator
 //
 // This file is distributed under the University of Illinois Open Source
 // License. See LICENSE.TXT for details.
 //
 //===----------------------------------------------------------------------===//
 ///
 /// \file
@@ skipping 1233 matching lines @@
     if (isIRGenerationDisabled()) {
       CurrentNode = nullptr;
       for (Ice::Type ArgType : Signature.getArgList()) {
         (void)ArgType;
         setNextLocalInstIndex(nullptr);
       }
     } else {
       Func->setFunctionName(FuncDecl->getName());
       Func->setReturnType(Signature.getReturnType());
       Func->setInternal(FuncDecl->getLinkage() == GlobalValue::InternalLinkage);
-      constexpr NaClBcIndexSize_t EntryBlock = 0;
-      CurrentNode = getBasicBlock(EntryBlock);
+      CurrentNode = installNextBasicBlock();
       Func->setEntryNode(CurrentNode);
       for (Ice::Type ArgType : Signature.getArgList()) {
         Func->addArg(getNextInstVar(ArgType));
       }
     }
     bool ParserResult = ParseThisBlock();
 
     // Temporarily end per-function timing, which will be resumed by
     // the translator function.  This is because translation may be
     // done asynchronously in a separate thread.
@@ skipping 41 matching lines @@
       if (isIRGenerationDisabled())
         return nullptr;
       std::string Buffer;
       raw_string_ostream StrBuf(Buffer);
       StrBuf << "Value index " << Index << " not defined!";
       Fatal(StrBuf.str());
     }
     return Op;
   }
 
-  // Returns the Index-th basic block in the list of basic blocks.
-  Ice::CfgNode *getBasicBlock(NaClBcIndexSize_t Index) {
-    assert(!isIRGenerationDisabled());
-    Ice::CfgNode *&Node = BbMap[Index];
-    if (Node == nullptr)
-      Node = Func->makeNode();
-    return Node;
-  }
-
 private:
   typedef std::unordered_map<NaClBcIndexSize_t, Ice::Operand *> OperandMap;
-  typedef std::unordered_map<NaClBcIndexSize_t, Ice::CfgNode *> CfgNodeMap;
+  typedef std::vector<Ice::CfgNode *> BlockNodesType;
 
   Ice::TimerMarker Timer;
+  // The number of words in the bitstream defining the function block.
+  uint64_t NumBytesDefiningFunction = 0;
   // The corresponding ICE function defined by the function block.
   std::unique_ptr<Ice::Cfg> Func;
-  // The specified number of basic blocks in the bitcode file.
-  NaClBcIndexSize_t SpecifiedNumBbs = 0;
   // The index to the current basic block being built.
   NaClBcIndexSize_t CurrentBbIndex = 0;
+  // The number of basic blocks declared for the function block.
+  NaClBcIndexSize_t DeclaredNumberBbs = 0;
   // The basic block being built.
   Ice::CfgNode *CurrentNode = nullptr;
-  // Map from basic block id (as defined in the bitcode file) to
-  // the corresponding basic block that implements it.
-  CfgNodeMap BbMap;
+  // Vector of basic blocks. Used instead of indexing Func->getNodes() for
+  // faster access (constant vs linear).

[Jim Stichnoth, 2015/08/13 23:16:25]

This comment is wrong -- Func->getNodes() returns a NodeList which is actually a
std::vector.

[Karl, 2015/08/14 19:18:21]

Your right. I was confusing this with LLVM code. Removing BlockNodes.

+  BlockNodesType BlockNodes;
   // The ID for the function.
   NaClBcIndexSize_t FcnId;
   // The corresponding function declaration.
   Ice::FunctionDeclaration *FuncDecl;
   // Holds the dividing point between local and global absolute value indices.
   size_t CachedNumGlobalValueIDs;
   // Holds operands local to the function block, based on indices
   // defined in the bitcode file.
   OperandMap LocalOperands;
   // Holds the index within LocalOperands corresponding to the next
@@ skipping 20 matching lines @@
            << ". Found: 2**" << (AlignPower - 1);
     Error(StrBuf.str());
     // Error recover with value that is always acceptable.
     Alignment = 1;
   }
 
   bool ParseBlock(unsigned BlockID) override;
 
   void ProcessRecord() override;
 
+  void EnterBlock(unsigned NumWords) final {
+    // Note: Bitstream defines words as 32-bit values.
+    NumBytesDefiningFunction = NumWords * 4;
+  }
+
   void ExitBlock() override;
 
-  bool verifyAndRenameBasicBlocks();
+  bool verifyAllForwardRefsDefined();
 
-  bool verifyAllForwardRefsDefined();
+  // Creates and appends a new basic block to the list of basic blocks.
+  Ice::CfgNode *installNextBasicBlock() {
+    assert(!isIRGenerationDisabled());
+    Ice::CfgNode *Node = Func->makeNode();
+    BlockNodes.push_back(Node);
+    return Node;
+  }
+
+  // Returns the Index-th basic block in the list of basic blocks.
+  Ice::CfgNode *getBasicBlock(NaClBcIndexSize_t Index) {
+    assert(!isIRGenerationDisabled());
+    if (Index >= BlockNodes.size()) {
+      std::string Buffer;
+      raw_string_ostream StrBuf(Buffer);
+      StrBuf << "Reference to basic block " << Index
+             << " not found. Must be less than " << BlockNodes.size();
+      Error(StrBuf.str());
+      // TODO(kschimpf) Remove error recovery once implementation complete.

[Jim Stichnoth, 2015/08/13 23:16:25]

Remove this TODO like in the previous CL?

[Karl, 2015/08/14 19:18:21]

Done.

+      Index = 0;
+    }
+    return BlockNodes[Index];
+  }
 
   // Returns the Index-th basic block in the list of basic blocks.
   // Assumes Index corresponds to a branch instruction. Hence, if
   // the branch references the entry block, it also generates a
   // corresponding error.
   Ice::CfgNode *getBranchBasicBlock(NaClBcIndexSize_t Index) {
     assert(!isIRGenerationDisabled());
     if (Index == 0) {
       Error("Branch to entry block not allowed");
     }
@@ skipping 592 matching lines @@
   std::sort(UndefinedFwdRefs.begin(), UndefinedFwdRefs.end());
   for (const NaClBcIndexSize_t Index : UndefinedFwdRefs) {
     std::string Buffer;
     raw_string_ostream StrBuf(Buffer);
     StrBuf << "Instruction forward reference not defined: " << Index;
     Error(StrBuf.str());
   }
   return false;
 }
 
-bool FunctionParser::verifyAndRenameBasicBlocks() {
-  const size_t NumFoundBbs = BbMap.size();
-  // Verify number of basic blocks found match amount specified in function.
-  if (NumFoundBbs != SpecifiedNumBbs) {
-    std::string Buffer;
-    raw_string_ostream StrBuf(Buffer);
-    StrBuf << "Function specified " << SpecifiedNumBbs
-           << "basic blocks. Found: " << NumFoundBbs;
-    Error(StrBuf.str());
-    return false;
-  }
-  // Verify size limit allowed for basic blocks.
-  if (NumFoundBbs > NaClBcIndexSize_t_Max) {
-    std::string Buffer;
-    raw_string_ostream StrBuf(Buffer);
-    StrBuf << "Functions can't define more than " << NaClBcIndexSize_t_Max
-           << "basic blocks. Found: " << NumFoundBbs;
-    Error(StrBuf.str());
-    return false;
-  }
-  // Sort list of Bbs, verifying that no basic blocks are missing.
-  Ice::NodeList SortedBbs;
-  for (size_t i = 0; i < NumFoundBbs; ++i) {
-    CfgNodeMap::iterator pos = BbMap.find(i);
-    if (pos == BbMap.end()) {
-      std::string Buffer;
-      raw_string_ostream StrBuf(Buffer);
-      StrBuf << "Can't find definition for basic block " << i << ".";
-      Error(StrBuf.str());
-      return false;
-    }
-    SortedBbs.push_back(pos->second);
-  }
-  // Install sorted basic blocks.
-  Func->swapNodes(SortedBbs);
-  return true;
-}
-
 void FunctionParser::ExitBlock() {
   // Check if the last instruction in the function was terminating.
   if (!InstIsTerminating) {
     Error("Last instruction in function not terminator");
     if (isIRGenerationDisabled())
       return;
     // Recover by inserting an unreachable instruction.
     CurrentNode->appendInst(Ice::InstUnreachable::create(Func.get()));
   }
+  ++CurrentBbIndex;
+  if (CurrentBbIndex != DeclaredNumberBbs) {
+    std::string Buffer;
+    raw_string_ostream StrBuf(Buffer);
+    StrBuf << "Function declared " << DeclaredNumberBbs
+           << " basic blocks, but defined " << CurrentBbIndex << ".";
+    Error(StrBuf.str());
+  }
   if (isIRGenerationDisabled())
     return;
   if (!verifyAllForwardRefsDefined())
     return;
-  if (!verifyAndRenameBasicBlocks())
-    return;
   // Before translating, check for blocks without instructions, and
   // insert unreachable. This shouldn't happen, but be safe.
   size_t Index = 0;
   for (Ice::CfgNode *Node : Func->getNodes()) {
     if (Node->getInsts().empty()) {
       std::string Buffer;
       raw_string_ostream StrBuf(Buffer);
       StrBuf << "Basic block " << Index << " contains no instructions";
       Error(StrBuf.str());
       Node->appendInst(Ice::InstUnreachable::create(Func.get()));
@@ skipping 13 matching lines @@
 }
 
 void FunctionParser::ProcessRecord() {
   // Note: To better separate parse/IR generation times, when IR generation
   // is disabled we do the following:
   // 1) Delay exiting until after we extract operands.
   // 2) return before we access operands, since all operands will be a nullptr.
   const NaClBitcodeRecord::RecordVector &Values = Record.GetValues();
   if (InstIsTerminating) {
     InstIsTerminating = false;
+    ++CurrentBbIndex;
     if (!isIRGenerationDisabled())
-      CurrentNode = getBasicBlock(++CurrentBbIndex);
+      CurrentNode = getBasicBlock(CurrentBbIndex);
   }
   // The base index for relative indexing.
   NaClBcIndexSize_t BaseIndex = getNextInstIndex();
   switch (Record.GetCode()) {
   case naclbitc::FUNC_CODE_DECLAREBLOCKS: {
     // DECLAREBLOCKS: [n]
     if (!isValidRecordSize(1, "count"))
       return;
-    uint64_t NumBbsRaw = Values[0];
-    if (NumBbsRaw == 0) {
-      Error("Functions must contain at least one basic block.");
-      NumBbsRaw = 1;
-    } else if (NumBbsRaw > NaClBcIndexSize_t_Max) {
-      std::string Buffer;
-      raw_string_ostream StrBuf(Buffer);
-      StrBuf << "To many basic blocks specified: " << NumBbsRaw;
-      Error(StrBuf.str());
-      NumBbsRaw = NaClBcIndexSize_t_Max;
-    }
-    if (isIRGenerationDisabled())
-      return;
-    if (SpecifiedNumBbs != 0) {
+    if (DeclaredNumberBbs > 0) {
       Error("Duplicate function block count record");
       return;
     }
-    SpecifiedNumBbs = NumBbsRaw;
+
+    uint64_t NumBbs = Values[0];
+
+    // Check for bad large sizes, since they can make ridiculous memory
+    // requests and hang the user for large amounts of time.  Note: We know
+    // that each basic block must have a terminator instruction, and each
+    // instruction is minimally defined by a two-bit abreviation.
+    uint64_t MaxBbs = NumBytesDefiningFunction * (CHAR_BIT >> 1);
+    if (NumBbs > MaxBbs) {
+      std::string Buffer;
+      raw_string_ostream StrBuf(Buffer);
+      StrBuf << "Function defines " << NumBbs
+             << " basic blocks, which is too big for a function containing "
+             << NumBytesDefiningFunction << " bytes";
+      Error(StrBuf.str());
+      NumBbs = MaxBbs;
+    }
+
+    if (NumBbs == 0) {
+      Error("Functions must contain at least one basic block.");
+      NumBbs = 1;
+    }
+
+    DeclaredNumberBbs = NumBbs;
+    if (isIRGenerationDisabled())
+      return;
+    // Install the basic blocks, skipping bb0 which was created in the
+    // constructor.
+    BlockNodes.reserve(NumBbs);
+    for (size_t i = 1; i < NumBbs; ++i)
+      installNextBasicBlock();
     return;
   }
   case naclbitc::FUNC_CODE_INST_BINOP: {
     // BINOP: [opval, opval, opcode]
     if (!isValidRecordSize(3, "binop"))
       return;
     Ice::Operand *Op1 = getRelativeOperand(Values[0], BaseIndex);
     Ice::Operand *Op2 = getRelativeOperand(Values[1], BaseIndex);
     if (isIRGenerationDisabled()) {
       assert(Op1 == nullptr && Op2 == nullptr);
@@ skipping 803 matching lines @@
     reportUnableToAssign("variable", Index, Name);
   }
 }
 
 void FunctionValuesymtabParser::setBbName(NaClBcIndexSize_t Index,
                                           StringType &Name) {
   if (!Ice::BuildDefs::dump())
     return;
   if (isIRGenerationDisabled())
     return;
-  Ice::CfgNode *Bb = getFunctionParser()->getBasicBlock(Index);
+  if (Index >= getFunctionParser()->getFunc()->getNumNodes()) {
+    reportUnableToAssign("block", Index, Name);
+    return;
+  }
   std::string Nm(Name.data(), Name.size());
-  Bb->setName(Nm);
+  if (Ice::BuildDefs::dump())
+    getFunctionParser()->getFunc()->getNodes()[Index]->setName(Nm);
 }
 
 bool FunctionParser::ParseBlock(unsigned BlockID) {
   switch (BlockID) {
   case naclbitc::CONSTANTS_BLOCK_ID: {
     ConstantsParser Parser(BlockID, this);
     return Parser.ParseThisBlock();
   }
   case naclbitc::VALUE_SYMTAB_BLOCK_ID: {
     if (PNaClAllowLocalSymbolTables) {
@@ skipping 225 matching lines @@
   }
   if (InputStreamFile.getBitcodeBytes().getExtent() % 4 != 0) {
     ErrStream
         << IRFilename
         << ": Bitcode stream should be a multiple of 4 bytes in length.\n";
     llvm::report_fatal_error("Bitcode stream should be a multiple of 4 bytes");
   }
 }
 
 } // end of namespace Ice