[Strings] Add experimental string.hash instruction (#5480)

See WebAssembly/stringref#60

Author: kripken

scripts/gen-s-parser.py

@@ -618,6 +618,7 @@
     ("string.measure_wtf8",  "makeStringMeasure(s, StringMeasureWTF8)"),
     ("string.measure_wtf16", "makeStringMeasure(s, StringMeasureWTF16)"),
     ("string.is_usv_sequence", "makeStringMeasure(s, StringMeasureIsUSV)"),
+    ("string.hash",          "makeStringMeasure(s, StringMeasureHash)"),
     ("string.encode_wtf8",   "makeStringEncode(s, StringEncodeWTF8)"),
     ("string.encode_wtf16",  "makeStringEncode(s, StringEncodeWTF16)"),
     ("string.encode_wtf8_array",   "makeStringEncode(s, StringEncodeWTF8Array)"),

src/gen-s-parser.inc

@@ -3194,6 +3194,9 @@ switch (buf[0]) {
                   case 'f':
                     if (op == "string.from_code_point"sv) { return makeStringNew(s, StringNewFromCodePoint, false); }
                     goto parse_error;
+                  case 'h':
+                    if (op == "string.hash"sv) { return makeStringMeasure(s, StringMeasureHash); }
+                    goto parse_error;
                   case 'i':
                     if (op == "string.is_usv_sequence"sv) { return makeStringMeasure(s, StringMeasureIsUSV); }
                     goto parse_error;
@@ -8931,6 +8934,13 @@ switch (buf[0]) {
                       return *ret;
                     }
                     goto parse_error;
+                  case 'h':
+                    if (op == "string.hash"sv) {
+                      auto ret = makeStringMeasure(ctx, pos, StringMeasureHash);
+                      CHECK_ERR(ret);
+                      return *ret;
+                    }
+                    goto parse_error;
                   case 'i':
                     if (op == "string.is_usv_sequence"sv) {
                       auto ret = makeStringMeasure(ctx, pos, StringMeasureIsUSV);

src/passes/Print.cpp

@@ -2404,6 +2404,9 @@ struct PrintExpressionContents
       case StringMeasureWTF16View:
         printMedium(o, "stringview_wtf16.length");
         break;
+      case StringMeasureHash:
+        printMedium(o, "string.hash");
+        break;
       default:
         WASM_UNREACHABLE("invalid string.measure*");
     }

src/wasm-binary.h

@@ -1165,6 +1165,7 @@ enum ASTNodes {
   StringViewIterSlice = 0xa4,
   StringCompare = 0xa8,
   StringFromCodePoint = 0xa9,
+  StringHash = 0xaa,
   StringNewWTF8Array = 0xb0,
   StringNewWTF16Array = 0xb1,
   StringEncodeWTF8Array = 0xb2,

src/wasm.h

@@ -596,6 +596,7 @@ enum StringMeasureOp {
   StringMeasureWTF16,
   StringMeasureIsUSV,
   StringMeasureWTF16View,
+  StringMeasureHash,
 };
 
 enum StringEncodeOp {

src/wasm/wasm-binary.cpp

@@ -7286,6 +7286,8 @@ bool WasmBinaryBuilder::maybeVisitStringMeasure(Expression*& out,
     op = StringMeasureIsUSV;
   } else if (code == BinaryConsts::StringViewWTF16Length) {
     op = StringMeasureWTF16View;
+  } else if (code == BinaryConsts::StringHash) {
+    op = StringMeasureHash;
   } else {
     return false;
   }

src/wasm/wasm-stack.cpp

@@ -2338,6 +2338,9 @@ void BinaryInstWriter::visitStringMeasure(StringMeasure* curr) {
     case StringMeasureWTF16View:
       o << U32LEB(BinaryConsts::StringViewWTF16Length);
       break;
+    case StringMeasureHash:
+      o << U32LEB(BinaryConsts::StringHash);
+      break;
     default:
       WASM_UNREACHABLE("invalid string.new*");
   }

test/lit/strings.wast

@@ -37,6 +37,8 @@
 
   ;; CHECK:      (type $ref|$array|_=>_none (func (param (ref $array))))
 
+  ;; CHECK:      (type $stringref_=>_i32 (func (param stringref) (result i32)))
+
   ;; CHECK:      (global $string-const stringref (string.const "string in a global \01\ff\00\t\t\n\n\r\r\"\"\'\'\\\\"))
   (global $string-const stringref (string.const "string in a global \01\ff\00\t\09\n\0a\r\0d\"\22\'\27\\\5c"))
 
@@ -626,4 +628,15 @@
       )
     )
   )
+
+  ;; CHECK:      (func $string.hash (type $stringref_=>_i32) (param $ref stringref) (result i32)
+  ;; CHECK-NEXT:  (string.hash
+  ;; CHECK-NEXT:   (local.get $ref)
+  ;; CHECK-NEXT:  )
+  ;; CHECK-NEXT: )
+  (func $string.hash (param $ref stringref) (result i32)
+    (string.hash
+      (local.get $ref)
+    )
+  )
 )