Use IRBuilder in the binary parser

src/ir/subtype-exprs.h

@@ -122,7 +122,7 @@ struct SubtypingDiscoverer : public OverriddenVisitor<SubType> {
     }
   }
   void visitIf(If* curr) {
-    if (curr->ifFalse) {
+    if (curr->ifFalse && curr->type != Type::unreachable) {
       self()->noteSubtype(curr->ifTrue, curr);
       self()->noteSubtype(curr->ifFalse, curr);
     }

src/passes/CodeFolding.cpp

@@ -234,6 +234,13 @@ struct CodeFolding
     if (!curr->ifFalse) {
       return;
     }
+    if (curr->condition->type == Type::unreachable) {
+      // If the arms are foldable and concrete, we would be replacing an
+      // unreachable If with a concrete block, which may or may not be valid,
+      // depending on the context. Leave this for DCE rather than trying to
+      // handle that.
+      return;
+    }
     // If both are blocks, look for a tail we can merge.
     auto* left = curr->ifTrue->dynCast<Block>();
     auto* right = curr->ifFalse->dynCast<Block>();

src/passes/Flatten.cpp

@@ -147,20 +147,24 @@ struct Flatten
         // arm preludes go in the arms. we must also remove an if value
         auto* originalIfTrue = iff->ifTrue;
         auto* originalIfFalse = iff->ifFalse;
-        auto type = iff->type;
+        auto type = iff->ifFalse ? Type::getLeastUpperBound(iff->ifTrue->type,
+                                                            iff->ifFalse->type)
+                                 : Type::none;
         Expression* prelude = nullptr;
         if (type.isConcrete()) {
           Index temp = builder.addVar(getFunction(), type);
           if (iff->ifTrue->type.isConcrete()) {
             iff->ifTrue = builder.makeLocalSet(temp, iff->ifTrue);
           }
-          if (iff->ifFalse && iff->ifFalse->type.isConcrete()) {
+          if (iff->ifFalse->type.isConcrete()) {
             iff->ifFalse = builder.makeLocalSet(temp, iff->ifFalse);
           }
-          // the whole if (+any preludes from the condition) is now a prelude
-          prelude = rep;
-          // and we leave just a get of the value
-          rep = builder.makeLocalGet(temp, type);
+          if (curr->type.isConcrete()) {
+            // the whole if (+any preludes from the condition) is now a prelude
+            prelude = rep;
+            // and we leave just a get of the value
+            rep = builder.makeLocalGet(temp, type);
+          }
         }
         iff->ifTrue = getPreludesWithExpression(originalIfTrue, iff->ifTrue);
         if (iff->ifFalse) {

src/passes/Print.cpp

@@ -460,9 +460,16 @@ struct PrintExpressionContents
   }
   void visitIf(If* curr) {
     printMedium(o, "if");
-    if (curr->type.isConcrete()) {
+    // Ifs are unreachable if their condition is unreachable, but in that case
+    // the arms might have some concrete type we have to account for to produce
+    // valid wat.
+    auto type = curr->type;
+    if (curr->condition->type == Type::unreachable && curr->ifFalse) {
+      type = Type::getLeastUpperBound(curr->ifTrue->type, curr->ifFalse->type);
+    }
+    if (type.isConcrete()) {
       o << ' ';
-      printBlockType(Signature(Type::none, curr->type));
+      printBlockType(Signature(Type::none, type));
     }
   }
   void visitLoop(Loop* curr) {

src/wasm-binary.h

@@ -29,6 +29,7 @@
 #include "ir/module-utils.h"
 #include "parsing.h"
 #include "wasm-builder.h"
+#include "wasm-ir-builder.h"
 #include "wasm-traversal.h"
 #include "wasm-validator.h"
 #include "wasm.h"
@@ -1543,8 +1544,6 @@ class WasmBinaryReader {
   Signature getSignatureByTypeIndex(Index index);
   Signature getSignatureByFunctionIndex(Index index);
 
-  size_t nextLabel;
-
   Name getNextLabel();
 
   // We read the names section first so we know in advance what names various
@@ -1573,67 +1572,19 @@ class WasmBinaryReader {
   void readVars();
   void setLocalNames(Function& func, Index i);
 
+  Result<> readInst();
+
   void readExports();
 
   // The strings in the strings section (which are referred to by StringConst).
   std::vector<Name> strings;
   void readStrings();
+  Name getIndexedString();
 
   Expression* readExpression();
   void readGlobals();
 
-  struct BreakTarget {
-    Name name;
-    Type type;
-    BreakTarget(Name name, Type type) : name(name), type(type) {}
-  };
-  std::vector<BreakTarget> breakStack;
-  // the names that breaks target. this lets us know if a block has breaks to it
-  // or not.
-  std::unordered_set<Name> breakTargetNames;
-  // the names that delegates target.
-  std::unordered_set<Name> exceptionTargetNames;
-
-  std::vector<Expression*> expressionStack;
-
-  // Control flow structure parsing: these have not just the normal binary
-  // data for an instruction, but also some bytes later on like "end" or "else".
-  // We must be aware of the connection between those things, for debug info.
-  std::vector<Expression*> controlFlowStack;
-
-  // Called when we parse the beginning of a control flow structure.
-  void startControlFlow(Expression* curr);
-
-  // set when we know code is unreachable in the sense of the wasm spec: we are
-  // in a block and after an unreachable element. this helps parse stacky wasm
-  // code, which can be unsuitable for our IR when unreachable.
-  bool unreachableInTheWasmSense;
-
-  // set when the current code being processed will not be emitted in the
-  // output, which is the case when it is literally unreachable, for example,
-  // (block $a
-  //   (unreachable)
-  //   (block $b
-  //     ;; code here is reachable in the wasm sense, even though $b as a whole
-  //     ;; is not
-  //     (unreachable)
-  //     ;; code here is unreachable in the wasm sense
-  //   )
-  // )
-  bool willBeIgnored;
-
-  BinaryConsts::ASTNodes lastSeparator = BinaryConsts::End;
-
-  // process a block-type scope, until an end or else marker, or the end of the
-  // function
-  void processExpressions();
-  void skipUnreachableCode();
-
-  void pushExpression(Expression* curr);
-  Expression* popExpression();
-  Expression* popNonVoidExpression();
-  Expression* popTuple(size_t numElems);
-  Expression* popTypedExpression(Type type);
+  IRBuilder builder;
 
   // validations that cannot be performed on the Module
   void validateBinary();
@@ -1663,127 +1614,12 @@ class WasmBinaryReader {
   void readNextDebugLocation();
   void readSourceMapHeader();
 
-  // AST reading
-  int depth = 0; // only for debugging
-
-  BinaryConsts::ASTNodes readExpression(Expression*& curr);
-  void pushBlockElements(Block* curr, Type type, size_t start);
-  void visitBlock(Block* curr);
-
-  // Gets a block of expressions. If it's just one, return that singleton.
-  Expression* getBlockOrSingleton(Type type);
-
-  BreakTarget getBreakTarget(int32_t offset);
-  Name getExceptionTargetName(int32_t offset);
-
   Index readMemoryAccess(Address& alignment, Address& offset);
+  std::tuple<Name, Address, Address> getMemarg();
 
-  void visitIf(If* curr);
-  void visitLoop(Loop* curr);
-  void visitBreak(Break* curr, uint8_t code);
-  void visitSwitch(Switch* curr);
-  void visitCall(Call* curr);
-  void visitCallIndirect(CallIndirect* curr);
-  void visitLocalGet(LocalGet* curr);
-  void visitLocalSet(LocalSet* curr, uint8_t code);
-  void visitGlobalGet(GlobalGet* curr);
-  void visitGlobalSet(GlobalSet* curr);
-  bool maybeVisitLoad(Expression*& out,
-                      uint8_t code,
-                      std::optional<BinaryConsts::ASTNodes> prefix);
-  bool maybeVisitStore(Expression*& out,
-                       uint8_t code,
-                       std::optional<BinaryConsts::ASTNodes> prefix);
-  bool maybeVisitNontrappingTrunc(Expression*& out, uint32_t code);
-  bool maybeVisitAtomicRMW(Expression*& out, uint8_t code);
-  bool maybeVisitAtomicCmpxchg(Expression*& out, uint8_t code);
-  bool maybeVisitAtomicWait(Expression*& out, uint8_t code);
-  bool maybeVisitAtomicNotify(Expression*& out, uint8_t code);
-  bool maybeVisitAtomicFence(Expression*& out, uint8_t code);
-  bool maybeVisitConst(Expression*& out, uint8_t code);
-  bool maybeVisitUnary(Expression*& out, uint8_t code);
-  bool maybeVisitBinary(Expression*& out, uint8_t code);
-  bool maybeVisitTruncSat(Expression*& out, uint32_t code);
-  bool maybeVisitSIMDBinary(Expression*& out, uint32_t code);
-  bool maybeVisitSIMDUnary(Expression*& out, uint32_t code);
-  bool maybeVisitSIMDConst(Expression*& out, uint32_t code);
-  bool maybeVisitSIMDStore(Expression*& out, uint32_t code);
-  bool maybeVisitSIMDExtract(Expression*& out, uint32_t code);
-  bool maybeVisitSIMDReplace(Expression*& out, uint32_t code);
-  bool maybeVisitSIMDShuffle(Expression*& out, uint32_t code);
-  bool maybeVisitSIMDTernary(Expression*& out, uint32_t code);
-  bool maybeVisitSIMDShift(Expression*& out, uint32_t code);
-  bool maybeVisitSIMDLoad(Expression*& out, uint32_t code);
-  bool maybeVisitSIMDLoadStoreLane(Expression*& out, uint32_t code);
-  bool maybeVisitMemoryInit(Expression*& out, uint32_t code);
-  bool maybeVisitDataDrop(Expression*& out, uint32_t code);
-  bool maybeVisitMemoryCopy(Expression*& out, uint32_t code);
-  bool maybeVisitMemoryFill(Expression*& out, uint32_t code);
-  bool maybeVisitTableSize(Expression*& out, uint32_t code);
-  bool maybeVisitTableGrow(Expression*& out, uint32_t code);
-  bool maybeVisitTableFill(Expression*& out, uint32_t code);
-  bool maybeVisitTableCopy(Expression*& out, uint32_t code);
-  bool maybeVisitTableInit(Expression*& out, uint32_t code);
-  bool maybeVisitRefI31(Expression*& out, uint32_t code);
-  bool maybeVisitI31Get(Expression*& out, uint32_t code);
-  bool maybeVisitRefTest(Expression*& out, uint32_t code);
-  bool maybeVisitRefCast(Expression*& out, uint32_t code);
-  bool maybeVisitBrOn(Expression*& out, uint32_t code);
-  bool maybeVisitStructNew(Expression*& out, uint32_t code);
-  bool maybeVisitStructGet(Expression*& out, uint32_t code);
-  bool maybeVisitStructSet(Expression*& out, uint32_t code);
-  bool maybeVisitArrayNewData(Expression*& out, uint32_t code);
-  bool maybeVisitArrayNewElem(Expression*& out, uint32_t code);
-  bool maybeVisitArrayNewFixed(Expression*& out, uint32_t code);
-  bool maybeVisitArrayGet(Expression*& out, uint32_t code);
-  bool maybeVisitArraySet(Expression*& out, uint32_t code);
-  bool maybeVisitArrayLen(Expression*& out, uint32_t code);
-  bool maybeVisitArrayCopy(Expression*& out, uint32_t code);
-  bool maybeVisitArrayFill(Expression*& out, uint32_t code);
-  bool maybeVisitArrayInit(Expression*& out, uint32_t code);
-  bool maybeVisitStringNew(Expression*& out, uint32_t code);
-  bool maybeVisitStringAsWTF16(Expression*& out, uint32_t code);
-  bool maybeVisitStringConst(Expression*& out, uint32_t code);
-  bool maybeVisitStringMeasure(Expression*& out, uint32_t code);
-  bool maybeVisitStringEncode(Expression*& out, uint32_t code);
-  bool maybeVisitStringConcat(Expression*& out, uint32_t code);
-  bool maybeVisitStringEq(Expression*& out, uint32_t code);
-  bool maybeVisitStringWTF16Get(Expression*& out, uint32_t code);
-  bool maybeVisitStringSliceWTF(Expression*& out, uint32_t code);
-  void visitSelect(Select* curr, uint8_t code);
-  void visitReturn(Return* curr);
-  void visitMemorySize(MemorySize* curr);
-  void visitMemoryGrow(MemoryGrow* curr);
-  void visitNop(Nop* curr);
-  void visitUnreachable(Unreachable* curr);
-  void visitDrop(Drop* curr);
-  void visitRefNull(RefNull* curr);
-  void visitRefIsNull(RefIsNull* curr);
-  void visitRefFunc(RefFunc* curr);
-  void visitRefEq(RefEq* curr);
-  void visitTableGet(TableGet* curr);
-  void visitTableSet(TableSet* curr);
-  void visitTryOrTryInBlock(Expression*& out);
-  void visitTryTable(TryTable* curr);
-  void visitThrow(Throw* curr);
-  void visitRethrow(Rethrow* curr);
-  void visitThrowRef(ThrowRef* curr);
-  void visitCallRef(CallRef* curr);
-  void visitRefAsCast(RefCast* curr, uint32_t code);
-  void visitRefAs(RefAs* curr, uint8_t code);
-  void visitContNew(ContNew* curr);
-  void visitContBind(ContBind* curr);
-  void visitResume(Resume* curr);
-  void visitSuspend(Suspend* curr);
-
-  [[noreturn]] void throwError(std::string text);
-
-  // Struct/Array instructions have an unnecessary heap type that is just for
-  // validation (except for the case of unreachability, but that's not a problem
-  // anyhow, we can ignore it there). That is, we also have a reference typed
-  // child from which we can infer the type anyhow, and we just need to check
-  // that type is the same.
-  void validateHeapTypeUsingChild(Expression* child, HeapType heapType);
+  [[noreturn]] void throwError(std::string text) {
+    throw ParseException(text, 0, pos);
+  }
 
 private:
   bool hasDWARFSections();

src/wasm-ir-builder.h

@@ -47,6 +47,10 @@ class IRBuilder : public UnifiedExpressionVisitor<IRBuilder, Result<>> {
   // of instructions after this is called.
   Result<Expression*> build();
 
+  // If the IRBuilder is empty, then it's ready to parse a new self-contained
+  // sequence of instructions.
+  [[nodiscard]] bool empty() { return scopeStack.empty(); }
+
   // Call visit() on an existing Expression with its non-child fields
   // initialized to initialize the child fields and refinalize it.
   Result<> visit(Expression*);
@@ -59,6 +63,15 @@ class IRBuilder : public UnifiedExpressionVisitor<IRBuilder, Result<>> {
   // pushed instruction.
   void setDebugLocation(const std::optional<Function::DebugLocation>&);
 
+  // Give the builder a pointer to the counter tracking the current location in
+  // the binary. If this pointer is non-null, the builder will record the binary
+  // locations relative to the given code section offset for all instructions
+  // and delimiters inside functions.
+  void setBinaryLocation(size_t* binaryPos, size_t codeSectionOffset) {
+    this->binaryPos = binaryPos;
+    this->codeSectionOffset = codeSectionOffset;
+  }
+
   // Set the function used to add scratch locals when constructing an isolated
   // sequence of IR.
   void setFunction(Function* func) { this->func = func; }
@@ -232,6 +245,11 @@ class IRBuilder : public UnifiedExpressionVisitor<IRBuilder, Result<>> {
   Function* func = nullptr;
   Builder builder;
 
+  // Used for setting DWARF expression locations.
+  size_t* binaryPos = nullptr;
+  size_t lastBinaryPos = 0;
+  size_t codeSectionOffset = 0;
+
   // The location lacks debug info as it was marked as not having it.
   struct NoDebug : public std::monostate {};
   // The location lacks debug info, but was not marked as not having
@@ -316,6 +334,9 @@ class IRBuilder : public UnifiedExpressionVisitor<IRBuilder, Result<>> {
     // stack-polymorphic unreachable mode.
     bool unreachable = false;
 
+    // The binary location of the start of the scope, used to set debug info.
+    size_t startPos = 0;
+
     ScopeCtx() : scope(NoScope{}) {}
     ScopeCtx(Scope scope) : scope(scope) {}
     ScopeCtx(Scope scope, Name label, bool labelUsed)
@@ -529,6 +550,10 @@ class IRBuilder : public UnifiedExpressionVisitor<IRBuilder, Result<>> {
       // Record the original label to handle references to it correctly.
       labelDepths[label].push_back(scopeStack.size() + 1);
     }
+    if (binaryPos) {
+      scope.startPos = lastBinaryPos;
+      lastBinaryPos = *binaryPos;
+    }
     scopeStack.push_back(scope);
   }