[wasm] Refactoring ahead of element segment changes

See related CL for context.

Changes:
- In InitExprInterface, add the ability to evaluate function references
  as index only. Remove the global buffers and use the ones passed with
  the instance object instead.
- In WasmElemSegment, add a field indicating if elements should be
  parsed as expressions or indices. Change module-decoder.cc to reflect
  this change.
- In module-instantiate, change the signatures of LoadElemSegment,
  LoadElemSegmentImpl, and EvaluateInitExpr. Move the latter out of
  InstanceBuilder.

Change-Id: I1df54393b2005fba49380654bdd40429bd4869dd
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/3364081Reviewed-by: Andreas Haas <ahaas@chromium.org>
Commit-Queue: Manos Koukoutos <manoskouk@chromium.org>
Cr-Commit-Position: refs/heads/main@{#78470}

src/wasm/init-expr-interface.cc

@@ -52,14 +52,20 @@ void InitExprInterface::RefNull(FullDecoder* decoder, ValueType type,
 
 void InitExprInterface::RefFunc(FullDecoder* decoder, uint32_t function_index,
                                 Value* result) {
-  if (isolate_ != nullptr) {
-    auto internal = WasmInstanceObject::GetOrCreateWasmInternalFunction(
-        isolate_, instance_, function_index);
-    result->runtime_value = WasmValue(
-        internal, ValueType::Ref(module_->functions[function_index].sig_index,
-                                 kNonNullable));
-  } else {
+  if (isolate_ == nullptr) {
     outer_module_->functions[function_index].declared = true;
+    return;
+  }
+  ValueType type = ValueType::Ref(module_->functions[function_index].sig_index,
+                                  kNonNullable);
+  if (function_strictness_ == kStrictFunctions) {
+    Handle<WasmInternalFunction> internal =
+        WasmInstanceObject::GetOrCreateWasmInternalFunction(isolate_, instance_,
+                                                            function_index);
+    result->runtime_value = WasmValue(internal, type);
+  } else {
+    result->runtime_value =
+        WasmValue(handle(Smi::FromInt(function_index), isolate_), type);
   }
 }
 
@@ -67,10 +73,17 @@ void InitExprInterface::GlobalGet(FullDecoder* decoder, Value* result,
                                   const GlobalIndexImmediate<validate>& imm) {
   if (isolate_ == nullptr) return;
   const WasmGlobal& global = module_->globals[imm.index];
+  DCHECK(!global.mutability);
   result->runtime_value =
       global.type.is_numeric()
-          ? WasmValue(GetRawUntaggedGlobalPtr(global), global.type)
-          : WasmValue(handle(tagged_globals_->get(global.offset), isolate_),
+          ? WasmValue(
+                reinterpret_cast<byte*>(
+                    instance_->untagged_globals_buffer().backing_store()) +
+                    global.offset,
+                global.type)
+          : WasmValue(
+                handle(instance_->tagged_globals_buffer().get(global.offset),
+                       isolate_),
                 global.type);
 }
 
@@ -175,11 +188,6 @@ void InitExprInterface::DoReturn(FullDecoder* decoder,
   if (isolate_ != nullptr) result_ = decoder->stack_value(1)->runtime_value;
 }
 
-byte* InitExprInterface::GetRawUntaggedGlobalPtr(const WasmGlobal& global) {
-  return reinterpret_cast<byte*>(untagged_globals_->backing_store()) +
-         global.offset;
-}
-
 }  // namespace wasm
 }  // namespace internal
 }  // namespace v8

src/wasm/init-expr-interface.h

@@ -44,16 +44,20 @@ class InitExprInterface {
   using FullDecoder =
       WasmFullDecoder<validate, InitExprInterface, decoding_mode>;
 
+  // Controls how function references are computed. {kLazyFunctions} means that
+  // only the index of the function will be computed, as a Smi.
+  // {kStrictFunctions} means we fully compute the function reference through
+  // {WasmInstanceObject::GetOrCreateWasmInternalFunction}.
+  enum FunctionStrictness { kLazyFunctions, kStrictFunctions };
+
   InitExprInterface(const WasmModule* module, Isolate* isolate,
                     Handle<WasmInstanceObject> instance,
-                    Handle<FixedArray> tagged_globals,
-                    Handle<JSArrayBuffer> untagged_globals)
+                    FunctionStrictness function_strictness)
       : module_(module),
         outer_module_(nullptr),
         isolate_(isolate),
         instance_(instance),
-        tagged_globals_(tagged_globals),
-        untagged_globals_(untagged_globals) {
+        function_strictness_(function_strictness) {
     DCHECK_NOT_NULL(isolate);
   }
 
@@ -81,16 +85,13 @@ class InitExprInterface {
   bool end_found() { return end_found_; }
 
  private:
-  byte* GetRawUntaggedGlobalPtr(const WasmGlobal& global);
-
   bool end_found_ = false;
   WasmValue result_;
   const WasmModule* module_;
   WasmModule* outer_module_;
   Isolate* isolate_;
   Handle<WasmInstanceObject> instance_;
-  Handle<FixedArray> tagged_globals_;
-  Handle<JSArrayBuffer> untagged_globals_;
+  FunctionStrictness function_strictness_;
 };
 
 }  // namespace wasm

src/wasm/module-decoder.cc

@@ -990,9 +990,7 @@ class ModuleDecoderImpl : public Decoder {
         consume_count("element count", FLAG_wasm_max_table_size);
 
     for (uint32_t i = 0; i < element_count; ++i) {
-      bool expressions_as_elements;
-      WasmElemSegment segment =
-          consume_element_segment_header(&expressions_as_elements);
+      WasmElemSegment segment = consume_element_segment_header();
       if (failed()) return;
       DCHECK_NE(segment.type, kWasmBottom);
 
@@ -1001,7 +999,7 @@ class ModuleDecoderImpl : public Decoder {
 
       for (uint32_t j = 0; j < num_elem; j++) {
         WasmElemSegment::Entry init =
-            expressions_as_elements
+            segment.element_type == WasmElemSegment::kExpressionElements
                 ? consume_element_expr()
                 : WasmElemSegment::Entry(WasmElemSegment::Entry::kRefFuncEntry,
                                          consume_element_func_index());
@@ -1910,8 +1908,7 @@ class ModuleDecoderImpl : public Decoder {
     return attribute;
   }
 
-  WasmElemSegment consume_element_segment_header(
-      bool* expressions_as_elements) {
+  WasmElemSegment consume_element_segment_header() {
     const byte* pos = pc();
 
     // The mask for the bit in the flag which indicates if the segment is
@@ -1941,7 +1938,10 @@ class ModuleDecoderImpl : public Decoder {
                                 : WasmElemSegment::kStatusActive;
     const bool is_active = status == WasmElemSegment::kStatusActive;
 
-    *expressions_as_elements = flag & kExpressionsAsElementsMask;
+    WasmElemSegment::ElementType element_type =
+        flag & kExpressionsAsElementsMask
+            ? WasmElemSegment::kExpressionElements
+            : WasmElemSegment::kFunctionIndexElements;
 
     const bool has_table_index =
         is_active && (flag & kHasTableIndexOrIsDeclarativeMask);
@@ -1965,7 +1965,7 @@ class ModuleDecoderImpl : public Decoder {
     const bool backwards_compatible_mode =
         is_active && !(flag & kHasTableIndexOrIsDeclarativeMask);
     ValueType type;
-    if (*expressions_as_elements) {
+    if (element_type == WasmElemSegment::kExpressionElements) {
       type =
           backwards_compatible_mode ? kWasmFuncRef : consume_reference_type();
       if (is_active && !IsSubtypeOf(type, table_type, this->module_.get())) {
@@ -2008,9 +2008,9 @@ class ModuleDecoderImpl : public Decoder {
     }
 
     if (is_active) {
-      return {type, table_index, std::move(offset)};
+      return {type, table_index, std::move(offset), element_type};
     } else {
-      return {type, status == WasmElemSegment::kStatusDeclarative};
+      return {type, status, element_type};
     }
   }
 

src/wasm/module-instantiate.cc

@@ -349,6 +349,9 @@ class InstanceBuilder {
   std::vector<Handle<WasmTagObject>> tags_wrappers_;
   Handle<WasmExportedFunction> start_function_;
   std::vector<SanitizedImport> sanitized_imports_;
+  // We pass this {Zone} to the temporary {WasmFullDecoder} we allocate during
+  // each call to {EvaluateInitExpression}. This has been found to improve
+  // performance a bit over allocating a new {Zone} each time.
   Zone init_expr_zone_;
 
 // Helper routines to print out errors with imports.
@@ -448,9 +451,6 @@ class InstanceBuilder {
   // Process initialization of globals.
   void InitGlobals(Handle<WasmInstanceObject> instance);
 
-  WasmValue EvaluateInitExpression(WireBytesRef init, ValueType expected,
-                                   Handle<WasmInstanceObject> instance);
-
   // Process the exports, creating wrappers for functions, tables, memories,
   // and globals.
   void ProcessExports(Handle<WasmInstanceObject> instance);
@@ -914,6 +914,28 @@ bool HasDefaultToNumberBehaviour(Isolate* isolate,
   // Just a default function, which will convert to "Nan". Accept this.
   return true;
 }
+
+WasmValue EvaluateInitExpression(
+    Zone* zone, WireBytesRef init, ValueType expected, Isolate* isolate,
+    Handle<WasmInstanceObject> instance,
+    InitExprInterface::FunctionStrictness function_strictness =
+        InitExprInterface::kStrictFunctions) {
+  base::Vector<const byte> module_bytes =
+      instance->module_object().native_module()->wire_bytes();
+  FunctionBody body(FunctionSig::Build(zone, {expected}, {}), init.offset(),
+                    module_bytes.begin() + init.offset(),
+                    module_bytes.begin() + init.end_offset());
+  WasmFeatures detected;
+  // We use kFullValidation so we do not have to create another instance of
+  // WasmFullDecoder, which would cost us >50Kb binary code size.
+  WasmFullDecoder<Decoder::kFullValidation, InitExprInterface, kInitExpression>
+      decoder(zone, instance->module(), WasmFeatures::All(), &detected, body,
+              instance->module(), isolate, instance, function_strictness);
+
+  decoder.DecodeFunctionBody();
+
+  return decoder.interface().result();
+}
 }  // namespace
 
 // Look up an import value in the {ffi_} object specifically for linking an
@@ -975,7 +997,8 @@ void InstanceBuilder::LoadDataSegments(Handle<WasmInstanceObject> instance) {
     size_t dest_offset;
     if (module_->is_memory64) {
       uint64_t dest_offset_64 =
-          EvaluateInitExpression(segment.dest_addr, kWasmI64, instance)
+          EvaluateInitExpression(&init_expr_zone_, segment.dest_addr, kWasmI64,
+                                 isolate_, instance)
               .to_u64();
       // Clamp to {std::numeric_limits<size_t>::max()}, which is always an
       // invalid offset.
@@ -983,8 +1006,8 @@ void InstanceBuilder::LoadDataSegments(Handle<WasmInstanceObject> instance) {
       dest_offset = static_cast<size_t>(std::min(
           dest_offset_64, uint64_t{std::numeric_limits<size_t>::max()}));
     } else {
-      dest_offset =
-          EvaluateInitExpression(segment.dest_addr, kWasmI32, instance)
+      dest_offset = EvaluateInitExpression(&init_expr_zone_, segment.dest_addr,
+                                           kWasmI32, isolate_, instance)
                         .to_u32();
     }
 
@@ -1651,26 +1674,6 @@ T* InstanceBuilder::GetRawUntaggedGlobalPtr(const WasmGlobal& global) {
   return reinterpret_cast<T*>(raw_buffer_ptr(untagged_globals_, global.offset));
 }
 
-WasmValue InstanceBuilder::EvaluateInitExpression(
-    WireBytesRef init, ValueType expected,
-    Handle<WasmInstanceObject> instance) {
-  base::Vector<const byte> module_bytes =
-      instance->module_object().native_module()->wire_bytes();
-  FunctionBody body(FunctionSig::Build(&init_expr_zone_, {expected}, {}),
-                    init.offset(), module_bytes.begin() + init.offset(),
-                    module_bytes.begin() + init.end_offset());
-  WasmFeatures detected;
-  // We use kFullValidation so we do not have to create another instance of
-  // WasmFullDecoder, which would cost us >50Kb binary code size.
-  WasmFullDecoder<Decoder::kFullValidation, InitExprInterface, kInitExpression>
-      decoder(&init_expr_zone_, module_, WasmFeatures::All(), &detected, body,
-              module_, isolate_, instance, tagged_globals_, untagged_globals_);
-
-  decoder.DecodeFunctionBody();
-
-  return decoder.interface().result();
-}
-
 // Process initialization of globals.
 void InstanceBuilder::InitGlobals(Handle<WasmInstanceObject> instance) {
   for (const WasmGlobal& global : module_->globals) {
@@ -1678,8 +1681,8 @@ void InstanceBuilder::InitGlobals(Handle<WasmInstanceObject> instance) {
     // Happens with imported globals.
     if (!global.init.is_set()) continue;
 
-    WasmValue value =
-        EvaluateInitExpression(global.init, global.type, instance);
+    WasmValue value = EvaluateInitExpression(&init_expr_zone_, global.init,
+                                             global.type, isolate_, instance);
 
     if (global.type.is_reference()) {
       tagged_globals_->set(global.offset, *value.to_ref());
@@ -1960,7 +1963,8 @@ void InstanceBuilder::InitializeIndirectFunctionTables(
       auto table_object = handle(
           WasmTableObject::cast(instance->tables().get(table_index)), isolate_);
       Handle<Object> value =
-          EvaluateInitExpression(table.initial_value, table.type, instance)
+          EvaluateInitExpression(&init_expr_zone_, table.initial_value,
+                                 table.type, isolate_, instance)
               .to_ref();
       if (value.is_null()) {
         for (uint32_t entry_index = 0; entry_index < table.initial_size;
@@ -1994,7 +1998,9 @@ void InstanceBuilder::InitializeIndirectFunctionTables(
   }
 }
 
-bool LoadElemSegmentImpl(Isolate* isolate, Handle<WasmInstanceObject> instance,
+namespace {
+bool LoadElemSegmentImpl(Zone* zone, Isolate* isolate,
+                         Handle<WasmInstanceObject> instance,
                          Handle<WasmTableObject> table_object,
                          uint32_t table_index, uint32_t segment_index,
                          uint32_t dst, uint32_t src, size_t count) {
@@ -2049,6 +2055,7 @@ bool LoadElemSegmentImpl(Isolate* isolate, Handle<WasmInstanceObject> instance,
   }
   return true;
 }
+}  // namespace
 
 void InstanceBuilder::LoadTableSegments(Handle<WasmInstanceObject> instance) {
   for (uint32_t segment_index = 0;
@@ -2058,14 +2065,14 @@ void InstanceBuilder::LoadTableSegments(Handle<WasmInstanceObject> instance) {
     if (elem_segment.status != WasmElemSegment::kStatusActive) continue;
 
     uint32_t table_index = elem_segment.table_index;
-    uint32_t dst =
-        EvaluateInitExpression(elem_segment.offset, kWasmI32, instance)
+    uint32_t dst = EvaluateInitExpression(&init_expr_zone_, elem_segment.offset,
+                                          kWasmI32, isolate_, instance)
                        .to_u32();
     uint32_t src = 0;
     size_t count = elem_segment.entries.size();
 
     bool success = LoadElemSegmentImpl(
-        isolate_, instance,
+        &init_expr_zone_, isolate_, instance,
         handle(WasmTableObject::cast(
                    instance->tables().get(elem_segment.table_index)),
                isolate_),
@@ -2108,8 +2115,13 @@ void InstanceBuilder::InitializeTags(Handle<WasmInstanceObject> instance) {
 bool LoadElemSegment(Isolate* isolate, Handle<WasmInstanceObject> instance,
                      uint32_t table_index, uint32_t segment_index, uint32_t dst,
                      uint32_t src, uint32_t count) {
+  AccountingAllocator allocator;
+  // This {Zone} will be used only by the temporary WasmFullDecoder allocated
+  // down the line from this call. Therefore it is safe to stack-allocate it
+  // here.
+  Zone zone(&allocator, "LoadElemSegment");
   return LoadElemSegmentImpl(
-      isolate, instance,
+      &zone, isolate, instance,
       handle(WasmTableObject::cast(instance->tables().get(table_index)),
              isolate),
       table_index, segment_index, dst, src, count);

src/wasm/module-instantiate.h

@@ -23,6 +23,7 @@ class JSReceiver;
 class WasmInitExpr;
 class WasmModuleObject;
 class WasmInstanceObject;
+class Zone;
 
 template <typename T>
 class Handle;

src/wasm/wasm-module.h

@@ -116,45 +116,53 @@ struct WasmDataSegment {
 
 // Static representation of wasm element segment (table initializer).
 struct WasmElemSegment {
+  enum Status {
+    kStatusActive,      // copied automatically during instantiation.
+    kStatusPassive,     // copied explicitly after instantiation.
+    kStatusDeclarative  // purely declarative and never copied.
+  };
+  struct Entry {
+    enum Kind { kGlobalGetEntry, kRefFuncEntry, kRefNullEntry } kind;
+    uint32_t index;
+    Entry(Kind kind, uint32_t index) : kind(kind), index(index) {}
+    Entry() : kind(kRefNullEntry), index(0) {}
+  };
+  enum ElementType { kFunctionIndexElements, kExpressionElements };
+
   // Construct an active segment.
-  WasmElemSegment(ValueType type, uint32_t table_index, WireBytesRef offset)
-      : type(type),
+  WasmElemSegment(ValueType type, uint32_t table_index, WireBytesRef offset,
+                  ElementType element_type)
+      : status(kStatusActive),
+        type(type),
         table_index(table_index),
         offset(std::move(offset)),
-        status(kStatusActive) {}
+        element_type(element_type) {}
 
   // Construct a passive or declarative segment, which has no table index or
   // offset.
-  WasmElemSegment(ValueType type, bool declarative)
-      : type(type),
-        table_index(0),
-        status(declarative ? kStatusDeclarative : kStatusPassive) {}
+  WasmElemSegment(ValueType type, Status status, ElementType element_type)
+      : status(status), type(type), table_index(0), element_type(element_type) {
+    DCHECK_NE(status, kStatusActive);
+  }
 
-  // Construct a passive or declarative segment, which has no table index or
-  // offset.
+  // Default constructor. Constucts an invalid segment.
   WasmElemSegment()
-      : type(kWasmBottom), table_index(0), status(kStatusActive) {}
+      : status(kStatusActive),
+        type(kWasmBottom),
+        table_index(0),
+        element_type(kFunctionIndexElements) {}
 
   WasmElemSegment(const WasmElemSegment&) = delete;
   WasmElemSegment(WasmElemSegment&&) V8_NOEXCEPT = default;
   WasmElemSegment& operator=(const WasmElemSegment&) = delete;
   WasmElemSegment& operator=(WasmElemSegment&&) V8_NOEXCEPT = default;
 
+  Status status;
   ValueType type;
   uint32_t table_index;
   WireBytesRef offset;
-  struct Entry {
-    enum Kind { kGlobalGetEntry, kRefFuncEntry, kRefNullEntry } kind;
-    uint32_t index;
-    Entry(Kind kind, uint32_t index) : kind(kind), index(index) {}
-    Entry() : kind(kRefNullEntry), index(0) {}
-  };
+  ElementType element_type;
   std::vector<Entry> entries;
-  enum Status {
-    kStatusActive,      // copied automatically during instantiation.
-    kStatusPassive,     // copied explicitly after instantiation.
-    kStatusDeclarative  // purely declarative and never copied.
-  } status;
 };
 
 // Static representation of a wasm import.

test/cctest/wasm/wasm-run-utils.cc

@@ -326,7 +326,9 @@ uint32_t TestingModuleBuilder::AddPassiveElementSegment(
   uint32_t index = static_cast<uint32_t>(test_module_->elem_segments.size());
   DCHECK_EQ(index, dropped_elem_segments_.size());
 
-  test_module_->elem_segments.emplace_back(kWasmFuncRef, false);
+  test_module_->elem_segments.emplace_back(
+      kWasmFuncRef, WasmElemSegment::kStatusPassive,
+      WasmElemSegment::kFunctionIndexElements);
   auto& elem_segment = test_module_->elem_segments.back();
   for (uint32_t entry : entries) {
     elem_segment.entries.push_back(

test/unittests/wasm/function-body-decoder-unittest.cc

@@ -159,20 +159,15 @@ class TestModuleBuilder {
   }
 
   byte AddPassiveElementSegment(wasm::ValueType type) {
-    mod.elem_segments.emplace_back(type, false);
-    auto& init = mod.elem_segments.back();
-    // Add 5 empty elements.
-    for (uint32_t j = 0; j < 5; j++) {
-      init.entries.push_back(WasmElemSegment::Entry(
-          WasmElemSegment::Entry::kRefNullEntry, type.heap_representation()));
-    }
+    mod.elem_segments.emplace_back(type, WasmElemSegment::kStatusPassive,
+                                   WasmElemSegment::kExpressionElements);
     return static_cast<byte>(mod.elem_segments.size() - 1);
   }
 
   byte AddDeclarativeElementSegment() {
-    mod.elem_segments.emplace_back(kWasmFuncRef, true);
-    mod.elem_segments.back().entries.push_back(WasmElemSegment::Entry(
-        WasmElemSegment::Entry::kRefNullEntry, HeapType::kFunc));
+    mod.elem_segments.emplace_back(kWasmFuncRef,
+                                   WasmElemSegment::kStatusDeclarative,
+                                   WasmElemSegment::kExpressionElements);
     return static_cast<byte>(mod.elem_segments.size() - 1);
   }