[stringrefs] Add wtf8_policy immediate to string.measure_wtf8

Following change in https://github.com/WebAssembly/stringref/pull/22.

Bug: v8:12868
Change-Id: Ic7728bff5d03ab547cb26ff41d6966f95bfb6b62
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3717986
Commit-Queue: Andy Wingo <wingo@igalia.com>
Reviewed-by: Jakob Kummerow <jkummerow@chromium.org>
Cr-Commit-Position: refs/heads/main@{#81308}

src/wasm/baseline/liftoff-compiler.cc

@@ -6078,34 +6078,25 @@ class LiftoffCompiler {
     __ PushRegister(kRef, result_reg);
   }
 
-  void StringMeasureUtf8(FullDecoder* decoder, const Value& str,
-                         Value* result) {
-    LiftoffRegList pinned;
-    LiftoffRegister string_reg = pinned.set(__ PopToRegister(pinned));
-    MaybeEmitNullCheck(decoder, string_reg.gp(), pinned, str.type);
-    LiftoffAssembler::VarState string_var(kRef, string_reg, 0);
-
-    CallRuntimeStub(WasmCode::kWasmStringMeasureUtf8,
-                    MakeSig::Returns(kI32).Params(kRef),
-                    {
-                        string_var,
-                    },
-                    decoder->position());
-    RegisterDebugSideTableEntry(decoder, DebugSideTableBuilder::kDidSpill);
-
-    LiftoffRegister result_reg(kReturnRegister0);
-    __ PushRegister(kI32, result_reg);
-  }
-
-  void StringMeasureWtf8(FullDecoder* decoder, const Value& str,
-                         Value* result) {
+  void StringMeasureWtf8(FullDecoder* decoder,
+                         const Wtf8PolicyImmediate<validate>& imm,
+                         const Value& str, Value* result) {
     LiftoffRegList pinned;
     LiftoffRegister string_reg = pinned.set(__ PopToRegister(pinned));
     MaybeEmitNullCheck(decoder, string_reg.gp(), pinned, str.type);
     LiftoffAssembler::VarState string_var(kRef, string_reg, 0);
 
-    CallRuntimeStub(WasmCode::kWasmStringMeasureWtf8,
-                    MakeSig::Returns(kI32).Params(kRef),
+    WasmCode::RuntimeStubId stub_id;
+    switch (imm.value) {
+      case kWtf8PolicyReject:
+        stub_id = WasmCode::kWasmStringMeasureUtf8;
+        break;
+      case kWtf8PolicyAccept:
+      case kWtf8PolicyReplace:
+        stub_id = WasmCode::kWasmStringMeasureWtf8;
+        break;
+    }
+    CallRuntimeStub(stub_id, MakeSig::Returns(kI32).Params(kRef),
                     {
                         string_var,
                     },

src/wasm/function-body-decoder-impl.h

@@ -1137,8 +1137,8 @@ struct ControlBase : public PcForErrors<validate> {
     const Value& offset, const Value& size, Value* result)                     \
   F(StringNewWtf16, const MemoryIndexImmediate<validate>& imm,                 \
     const Value& offset, const Value& size, Value* result)                     \
-  F(StringMeasureUtf8, const Value& str, Value* result)                        \
-  F(StringMeasureWtf8, const Value& str, Value* result)                        \
+  F(StringMeasureWtf8, const Wtf8PolicyImmediate<validate>& imm,               \
+    const Value& str, Value* result)                                           \
   F(StringMeasureWtf16, const Value& str, Value* result)                       \
   F(StringEncodeWtf8, const EncodeWtf8Immediate<validate>& memory,             \
     const Value& str, const Value& address)                                    \
@@ -2028,8 +2028,10 @@ class WasmDecoder : public Decoder {
             StringConstImmediate<validate> imm(decoder, pc + length);
             return length + imm.length;
           }
-          case kExprStringMeasureUtf8:
-          case kExprStringMeasureWtf8:
+          case kExprStringMeasureWtf8: {
+            Wtf8PolicyImmediate<validate> imm(decoder, pc + length);
+            return length + imm.length;
+          }
           case kExprStringMeasureWtf16:
           case kExprStringConcat:
           case kExprStringEq:
@@ -2250,7 +2252,6 @@ class WasmDecoder : public Decoder {
           }
           case kExprStringConst:
             return { 0, 1 };
-          case kExprStringMeasureUtf8:
           case kExprStringMeasureWtf8:
           case kExprStringMeasureWtf16:
           case kExprStringIsUSVSequence:
@@ -5223,23 +5224,16 @@ class WasmFullDecoder : public WasmDecoder<validate, decoding_mode> {
         Push(result);
         return opcode_length + imm.length;
       }
-      case kExprStringMeasureUtf8: {
-        NON_CONST_ONLY
-        Value str = Peek(0, 0, kWasmStringRef);
-        Value result = CreateValue(kWasmI32);
-        CALL_INTERFACE_IF_OK_AND_REACHABLE(StringMeasureUtf8, str, &result);
-        Drop(str);
-        Push(result);
-        return opcode_length;
-      }
       case kExprStringMeasureWtf8: {
         NON_CONST_ONLY
+        Wtf8PolicyImmediate<validate> imm(this, this->pc_ + opcode_length);
         Value str = Peek(0, 0, kWasmStringRef);
         Value result = CreateValue(kWasmI32);
-        CALL_INTERFACE_IF_OK_AND_REACHABLE(StringMeasureWtf8, str, &result);
+        CALL_INTERFACE_IF_OK_AND_REACHABLE(StringMeasureWtf8, imm, str,
+                                           &result);
         Drop(str);
         Push(result);
-        return opcode_length;
+        return opcode_length + imm.length;
       }
       case kExprStringMeasureWtf16: {
         NON_CONST_ONLY

src/wasm/graph-builder-interface.cc

@@ -1413,16 +1413,20 @@ class WasmGraphBuildingInterface {
     result->node = builder_->StringConst(imm.index);
   }
 
-  void StringMeasureUtf8(FullDecoder* decoder, const Value& str,
-                         Value* result) {
-    result->node = builder_->StringMeasureUtf8(str.node, NullCheckFor(str.type),
-                                               decoder->position());
-  }
-
-  void StringMeasureWtf8(FullDecoder* decoder, const Value& str,
-                         Value* result) {
-    result->node = builder_->StringMeasureWtf8(str.node, NullCheckFor(str.type),
-                                               decoder->position());
+  void StringMeasureWtf8(FullDecoder* decoder,
+                         const Wtf8PolicyImmediate<validate>& imm,
+                         const Value& str, Value* result) {
+    switch (imm.value) {
+      case kWtf8PolicyReject:
+        result->node = builder_->StringMeasureUtf8(
+            str.node, NullCheckFor(str.type), decoder->position());
+        break;
+      case kWtf8PolicyAccept:
+      case kWtf8PolicyReplace:
+        result->node = builder_->StringMeasureWtf8(
+            str.node, NullCheckFor(str.type), decoder->position());
+        break;
+    }
   }
 
   void StringMeasureWtf16(FullDecoder* decoder, const Value& str,

src/wasm/wasm-constants.h

@@ -151,7 +151,8 @@ enum NameSectionKindCode : uint8_t {
 enum StringRefWtf8Policy : uint8_t {
   kWtf8PolicyReject = 0,   // Strict UTF-8; no isolated surrogates allowed.
   kWtf8PolicyAccept = 1,   // Follow WTF-8 encoding of isolates surrogates.
-  kWtf8PolicyReplace = 2,  // Replace isolated surrogates with U+FFFD.
+  kWtf8PolicyReplace = 2,  // Replace isolated surrogates and decoding errors
+                           // with U+FFFD.
   kLastWtf8Policy = kWtf8PolicyReplace
 };
 

src/wasm/wasm-opcodes.h

@@ -742,7 +742,6 @@ bool V8_EXPORT_PRIVATE IsJSCompatibleSignature(const FunctionSig* sig,
   V(StringNewWtf8, 0xfb80, _, "string.new_wtf8")                               \
   V(StringNewWtf16, 0xfb81, _, "string.new_wtf16")                             \
   V(StringConst, 0xfb82, _, "string.const")                                    \
-  V(StringMeasureUtf8, 0xfb83, _, "string.measure_utf8")                       \
   V(StringMeasureWtf8, 0xfb84, _, "string.measure_wtf8")                       \
   V(StringMeasureWtf16, 0xfb85, _, "string.measure_wtf16")                     \
   V(StringEncodeWtf8, 0xfb86, _, "string.encode_wtf8")                         \

test/mjsunit/wasm/stringrefs-exec.js

@@ -186,34 +186,50 @@ function HasIsolatedSurrogate(str) {
     .exportFunc()
     .addBody([
       kExprLocalGet, 0,
-      kGCPrefix, kExprStringMeasureUtf8
+      kGCPrefix, kExprStringMeasureWtf8, 0
     ]);
 
   builder.addFunction("string_measure_wtf8", kSig_i_w)
     .exportFunc()
     .addBody([
       kExprLocalGet, 0,
-      kGCPrefix, kExprStringMeasureWtf8
+      kGCPrefix, kExprStringMeasureWtf8, 1
+    ]);
+
+  builder.addFunction("string_measure_wtf8_replace", kSig_i_w)
+    .exportFunc()
+    .addBody([
+      kExprLocalGet, 0,
+      kGCPrefix, kExprStringMeasureWtf8, 2
     ]);
 
   builder.addFunction("string_measure_utf8_null", kSig_i_v)
     .exportFunc()
     .addBody([
       kExprRefNull, kStringRefCode,
-      kGCPrefix, kExprStringMeasureUtf8
+      kGCPrefix, kExprStringMeasureWtf8, 0
     ]);
 
   builder.addFunction("string_measure_wtf8_null", kSig_i_v)
     .exportFunc()
     .addBody([
       kExprRefNull, kStringRefCode,
-      kGCPrefix, kExprStringMeasureWtf8
+      kGCPrefix, kExprStringMeasureWtf8, 1
+    ]);
+
+  builder.addFunction("string_measure_wtf8_replace_null", kSig_i_v)
+    .exportFunc()
+    .addBody([
+      kExprRefNull, kStringRefCode,
+      kGCPrefix, kExprStringMeasureWtf8, 2
     ]);
 
   let instance = builder.instantiate();
   for (let str of interestingStrings) {
     let wtf8 = encodeWtf8(str);
     assertEquals(wtf8.length, instance.exports.string_measure_wtf8(str));
+    assertEquals(wtf8.length,
+                 instance.exports.string_measure_wtf8_replace(str));
     if (HasIsolatedSurrogate(str)) {
       assertEquals(-1, instance.exports.string_measure_utf8(str));
     } else {
@@ -225,6 +241,8 @@ function HasIsolatedSurrogate(str) {
                WebAssembly.RuntimeError, "dereferencing a null pointer");
   assertThrows(() => instance.exports.string_measure_wtf8_null(),
                WebAssembly.RuntimeError, "dereferencing a null pointer");
+  assertThrows(() => instance.exports.string_measure_wtf8_replace_null(),
+               WebAssembly.RuntimeError, "dereferencing a null pointer");
 })();
 
 (function TestStringMeasureWtf16() {

test/mjsunit/wasm/stringrefs-valid.js

@@ -91,16 +91,22 @@ let kSig_w_zi = makeSig([kWasmStringViewIter, kWasmI32],
       kGCPrefix, kExprStringConst, 0
     ]);
 
-  builder.addFunction("string.measure_utf8", kSig_i_w)
+  builder.addFunction("string.measure_wtf8/utf-8", kSig_i_w)
     .addBody([
       kExprLocalGet, 0,
-      kGCPrefix, kExprStringMeasureUtf8
+      kGCPrefix, kExprStringMeasureWtf8, 0
     ]);
 
-  builder.addFunction("string.measure_wtf8", kSig_i_w)
+  builder.addFunction("string.measure_wtf8/wtf-8", kSig_i_w)
     .addBody([
       kExprLocalGet, 0,
-      kGCPrefix, kExprStringMeasureWtf8
+      kGCPrefix, kExprStringMeasureWtf8, 1
+    ]);
+
+  builder.addFunction("string.measure_wtf8/replace", kSig_i_w)
+    .addBody([
+      kExprLocalGet, 0,
+      kGCPrefix, kExprStringMeasureWtf8, 2
     ]);
 
   builder.addFunction("string.measure_wtf16", kSig_i_w)
@@ -309,3 +315,14 @@ assertInvalid(
   },
   "Compiling function #0:\"string.encode_wtf8/bad-policy\" failed: " +
     "expected wtf8 policy 0, 1, or 2, but found 3 @+37");
+
+assertInvalid(
+  builder => {
+    builder.addFunction("string.measure_wtf8/bad-policy", kSig_i_w)
+      .addBody([
+        kExprLocalGet, 0,
+        kGCPrefix, kExprStringMeasureWtf8, 3
+      ]);
+  },
+  "Compiling function #0:\"string.measure_wtf8/bad-policy\" failed: " +
+    "expected wtf8 policy 0, 1, or 2, but found 3 @+29");

test/mjsunit/wasm/wasm-module-builder.js

@@ -536,7 +536,6 @@ let kExprBrOnNonArray = 0x67;
 let kExprStringNewWtf8 = 0x80;
 let kExprStringNewWtf16 = 0x81;
 let kExprStringConst = 0x82;
-let kExprStringMeasureUtf8 = 0x83;
 let kExprStringMeasureWtf8 = 0x84;
 let kExprStringMeasureWtf16 = 0x85;
 let kExprStringEncodeWtf8 = 0x86;