[Strings] string.encode (#4776)

Author: kripken

scripts/gen-s-parser.py

@@ -619,6 +619,8 @@
     ("string.const",         "makeStringConst(s)"),
     ("string.measure_wtf8",  "makeStringMeasure(s, StringMeasureWTF8)"),
     ("string.measure_wtf16", "makeStringMeasure(s, StringMeasureWTF16)"),
+    ("string.encode_wtf8",   "makeStringEncode(s, StringEncodeWTF8)"),
+    ("string.encode_wtf16",  "makeStringEncode(s, StringEncodeWTF16)"),
 ]
 
 

src/gen-s-parser.inc

@@ -3132,6 +3132,17 @@ switch (op[0]) {
               case 'c':
                 if (strcmp(op, "string.const") == 0) { return makeStringConst(s); }
                 goto parse_error;
+              case 'e': {
+                switch (op[17]) {
+                  case '1':
+                    if (strcmp(op, "string.encode_wtf16") == 0) { return makeStringEncode(s, StringEncodeWTF16); }
+                    goto parse_error;
+                  case '8':
+                    if (strcmp(op, "string.encode_wtf8") == 0) { return makeStringEncode(s, StringEncodeWTF8); }
+                    goto parse_error;
+                  default: goto parse_error;
+                }
+              }
               case 'm': {
                 switch (op[18]) {
                   case '1':

src/ir/ReFinalize.cpp

@@ -175,6 +175,7 @@ void ReFinalize::visitRefAs(RefAs* curr) { curr->finalize(); }
 void ReFinalize::visitStringNew(StringNew* curr) { curr->finalize(); }
 void ReFinalize::visitStringConst(StringConst* curr) { curr->finalize(); }
 void ReFinalize::visitStringMeasure(StringMeasure* curr) { curr->finalize(); }
+void ReFinalize::visitStringEncode(StringEncode* curr) { curr->finalize(); }
 
 void ReFinalize::visitFunction(Function* curr) {
   // we may have changed the body from unreachable to none, which might be bad

src/ir/cost.h

@@ -678,6 +678,9 @@ struct CostAnalyzer : public OverriddenVisitor<CostAnalyzer, CostType> {
   CostType visitStringMeasure(StringMeasure* curr) {
     return 6 + visit(curr->ref);
   }
+  CostType visitStringEncode(StringEncode* curr) {
+    return 6 + visit(curr->ref) + visit(curr->ptr);
+  }
 
 private:
   CostType nullCheckCost(Expression* ref) {

src/ir/effects.h

@@ -734,7 +734,14 @@ class EffectAnalyzer {
     }
     void visitStringNew(StringNew* curr) {}
     void visitStringConst(StringConst* curr) {}
-    void visitStringMeasure(StringMeasure* curr) {}
+    void visitStringMeasure(StringMeasure* curr) {
+      // traps when ref is null.
+      parent.implicitTrap = true;
+    }
+    void visitStringEncode(StringEncode* curr) {
+      // traps when ref is null or we write out of bounds.
+      parent.implicitTrap = true;
+    }
   };
 
 public:

src/ir/possible-contents.cpp

@@ -685,6 +685,10 @@ struct InfoCollector
     // TODO: optimize when possible
     addRoot(curr);
   }
+  void visitStringEncode(StringEncode* curr) {
+    // TODO: optimize when possible
+    addRoot(curr);
+  }
 
   // TODO: Model which throws can go to which catches. For now, anything thrown
   //       is sent to the location of that tag, and any catch of that tag can

src/passes/Print.cpp

@@ -2257,6 +2257,21 @@ struct PrintExpressionContents
         WASM_UNREACHABLE("invalid string.measure*");
     }
   }
+  void visitStringEncode(StringEncode* curr) {
+    switch (curr->op) {
+      case StringEncodeUTF8:
+        printMedium(o, "string.encode_wtf8 utf8");
+        break;
+      case StringEncodeWTF8:
+        printMedium(o, "string.encode_wtf8 wtf8");
+        break;
+      case StringEncodeWTF16:
+        printMedium(o, "string.encode_wtf16");
+        break;
+      default:
+        WASM_UNREACHABLE("invalid string.encode*");
+    }
+  }
 };
 
 // Prints an expression in s-expr format, including both the

src/wasm-binary.h

@@ -1142,6 +1142,8 @@ enum ASTNodes {
   StringConst = 0x82,
   StringMeasureWTF8 = 0x84,
   StringMeasureWTF16 = 0x85,
+  StringEncodeWTF8 = 0x86,
+  StringEncodeWTF16 = 0x87,
 };
 
 enum MemoryAccess {
@@ -1725,6 +1727,7 @@ class WasmBinaryBuilder {
   bool maybeVisitStringNew(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);
   void visitSelect(Select* curr, uint8_t code);
   void visitReturn(Return* curr);
   void visitMemorySize(MemorySize* curr);

src/wasm-builder.h

@@ -1011,6 +1011,15 @@ class Builder {
     ret->finalize();
     return ret;
   }
+  StringEncode*
+  makeStringEncode(StringEncodeOp op, Expression* ref, Expression* ptr) {
+    auto* ret = wasm.allocator.alloc<StringEncode>();
+    ret->op = op;
+    ret->ref = ref;
+    ret->ptr = ptr;
+    ret->finalize();
+    return ret;
+  }
 
   // Additional helpers
 

src/wasm-delegations-fields.def

@@ -734,6 +734,14 @@ switch (DELEGATE_ID) {
     DELEGATE_END(StringMeasure);
     break;
   }
+  case Expression::Id::StringEncodeId: {
+    DELEGATE_START(StringEncode);
+    DELEGATE_FIELD_INT(StringEncode, op);
+    DELEGATE_FIELD_CHILD(StringEncode, ptr);
+    DELEGATE_FIELD_CHILD(StringEncode, ref);
+    DELEGATE_END(StringEncode);
+    break;
+  }
 }
 
 #undef DELEGATE_ID