Turn on TDP table generation for messages with weak=true fields.

We still use the reflection based parser for it, but we have an otherwise
normal table generated.

PiperOrigin-RevId: 603481451

src/google/protobuf/compiler/cpp/parse_function_generator.cc

@@ -51,8 +51,7 @@ bool UseDirectTcParserTable(const FieldDescriptor* field,
                             const Options& options) {
   if (field->cpp_type() != field->CPPTYPE_MESSAGE) return false;
   auto* m = field->message_type();
-  return !m->options().message_set_wire_format() && !HasTracker(m, options) &&
-         !HasWeakFields(m)
+  return !m->options().message_set_wire_format() && !HasTracker(m, options)
       ;  // NOLINT(whitespace/semicolon)
 }
 
@@ -113,26 +112,18 @@ ParseFunctionGenerator::ParseFunctionGenerator(
       ordered_fields_(GetOrderedFields(descriptor_, options_)),
       num_hasbits_(max_has_bit_index),
       index_in_file_messages_(index_in_file_messages) {
-  if (should_generate_tctable()) {
-    tc_table_info_.reset(new TailCallTableInfo(
-        descriptor_, ordered_fields_,
-        {/* is_lite */ GetOptimizeFor(descriptor->file(), options_) ==
-             FileOptions::LITE_RUNTIME,
-         /* uses_codegen */ true, options_.profile_driven_cluster_aux_subtable},
-        GeneratedOptionProvider(this), has_bit_indices,
-        inlined_string_indices));
-  }
+  tc_table_info_.reset(new TailCallTableInfo(
+      descriptor_, ordered_fields_,
+      {/* is_lite */ GetOptimizeFor(descriptor->file(), options_) ==
+           FileOptions::LITE_RUNTIME,
+       /* uses_codegen */ true, options_.profile_driven_cluster_aux_subtable},
+      GeneratedOptionProvider(this), has_bit_indices, inlined_string_indices));
   SetCommonMessageDataVariables(descriptor_, &variables_);
   SetUnknownFieldsVariable(descriptor_, options_, &variables_);
   variables_["classname"] = ClassName(descriptor, false);
 }
 
 void ParseFunctionGenerator::GenerateMethodDecls(io::Printer* printer) {
-  if (HasWeakFields(descriptor_)) {
-    // We use the reflection based one.
-    ABSL_CHECK(HasDescriptorMethods(descriptor_->file(), options_));
-    return;
-  }
   Formatter format(printer, variables_);
   format(
       "const char* _InternalParse(const char* ptr, "
@@ -157,25 +148,10 @@ void ParseFunctionGenerator::GenerateMethodImpls(io::Printer* printer) {
         "}\n");
     return;
   }
-  if (HasWeakFields(descriptor_)) {
-    // We use the reflection based one.
-    ABSL_CHECK(HasDescriptorMethods(descriptor_->file(), options_));
-    return;
-  }
-  ABSL_CHECK(should_generate_tctable());
   GenerateTailcallParseFunction(format);
 }
 
-bool ParseFunctionGenerator::should_generate_tctable() const {
-  if (HasWeakFields(descriptor_)) {
-    return false;
-  }
-  return true;
-}
-
 void ParseFunctionGenerator::GenerateTailcallParseFunction(Formatter& format) {
-  ABSL_CHECK(should_generate_tctable());
-
   // Generate an `_InternalParse` that starts the tail-calling loop.
   format(
       "const char* $classname$::_InternalParse(\n"
@@ -220,9 +196,6 @@ static int FieldNameDataSize(const std::vector<uint8_t>& data) {
 }
 
 void ParseFunctionGenerator::GenerateDataDecls(io::Printer* p) {
-  if (!should_generate_tctable()) {
-    return;
-  }
   auto v = p->WithVars(variables_);
   auto field_num_to_entry_table = MakeNumToEntryTable(ordered_fields_);
   p->Emit(
@@ -266,9 +239,6 @@ void ParseFunctionGenerator::GenerateDataDecls(io::Printer* p) {
 }
 
 void ParseFunctionGenerator::GenerateDataDefinitions(io::Printer* printer) {
-  if (!should_generate_tctable()) {
-    return;
-  }
   GenerateTailCallTable(printer);
 }
 
@@ -336,12 +306,16 @@ static NumToEntryTable MakeNumToEntryTable(
 
 void ParseFunctionGenerator::GenerateTailCallTable(io::Printer* printer) {
   Formatter format(printer, variables_);
-  ABSL_CHECK(should_generate_tctable());
   // All entries without a fast-path parsing function need a fallback.
   std::string fallback = "::_pbi::TcParser::GenericFallback";
   if (GetOptimizeFor(descriptor_->file(), options_) ==
       FileOptions::LITE_RUNTIME) {
     absl::StrAppend(&fallback, "Lite");
+  } else if (HasWeakFields(descriptor_)) {
+    // weak=true fields are handled with the reflection fallback, but we don't
+    // want to install that for normal messages because it has more overhead.
+    ABSL_CHECK(HasDescriptorMethods(descriptor_->file(), options_));
+    fallback = "::_pbi::TcParser::ReflectionFallback";
   }
 
   // For simplicity and speed, the table is not covering all proto
@@ -654,7 +628,7 @@ void ParseFunctionGenerator::GenerateFieldEntries(Formatter& format) {
     PrintFieldComment(format, field, options_);
     format("{");
     if (IsWeak(field, options_)) {
-      // Weak fields are handled by the generated fallback function.
+      // Weak fields are handled by the reflection fallback function.
       // (These are handled by legacy Google-internal logic.)
       format("/* weak */ 0, 0, 0, 0");
     } else {

src/google/protobuf/compiler/cpp/parse_function_generator.h

@@ -51,9 +51,6 @@ class ParseFunctionGenerator {
  private:
   class GeneratedOptionProvider;
 
-  // Returns true if tailcall table code should be generated.
-  bool should_generate_tctable() const;
-
   // Generates a tail-calling `_InternalParse` function.
   void GenerateTailcallParseFunction(Formatter& format);
 

src/google/protobuf/generated_message_reflection.cc

@@ -3277,7 +3277,6 @@ void Reflection::PopulateTcParseEntries(
     TcParseTableBase::FieldEntry* entries) const {
   for (const auto& entry : table_info.field_entries) {
     const FieldDescriptor* field = entry.field;
-    ABSL_CHECK(!field->options().weak());
     if (field->type() == field->TYPE_ENUM &&
         table_info.aux_entries[entry.aux_idx].type ==
             internal::TailCallTableInfo::kEnumValidator) {
@@ -3359,12 +3358,6 @@ const internal::TcParseTableBase* Reflection::CreateTcParseTable() const {
     return CreateTcParseTableReflectionOnly();
   }
 
-  for (int i = 0; i < descriptor_->field_count(); ++i) {
-    if (descriptor_->field(i)->options().weak()) {
-      return CreateTcParseTableReflectionOnly();
-    }
-  }
-
   std::vector<const FieldDescriptor*> fields;
   constexpr int kNoHasbit = -1;
   std::vector<int> has_bit_indices(

src/google/protobuf/generated_message_tctable_gen.cc

@@ -225,10 +225,10 @@ bool IsFieldEligibleForFastParsing(
     const TailCallTableInfo::OptionProvider& option_provider) {
   const auto* field = entry.field;
   const auto options = option_provider.GetForField(field);
-  ABSL_CHECK(!field->options().weak());
   // Map, oneof, weak, and split fields are not handled on the fast path.
   if (field->is_map() || field->real_containing_oneof() ||
-      options.is_implicitly_weak || options.should_split) {
+      field->options().weak() || options.is_implicitly_weak ||
+      options.should_split) {
     return false;
   }
 
@@ -794,7 +794,7 @@ TailCallTableInfo::TailCallTableInfo(
         field->type() == FieldDescriptor::TYPE_GROUP) {
       // Message-typed fields have a FieldAux with the default instance pointer.
       if (field->is_map()) {
-        field_entries.back().aux_idx = aux_entries.size();
+        entry.aux_idx = aux_entries.size();
         aux_entries.push_back({kMapAuxInfo, {field}});
         if (message_options.uses_codegen) {
           // If we don't use codegen we can't add these.
@@ -807,9 +807,12 @@ TailCallTableInfo::TailCallTableInfo(
             aux_entries.push_back({kEnumValidator, {map_value}});
           }
         }
+      } else if (field->options().weak()) {
+        // Disable the type card for this entry to force the fallback.
+        entry.type_card = 0;
       } else if (HasLazyRep(field, options)) {
         if (message_options.uses_codegen) {
-          field_entries.back().aux_idx = aux_entries.size();
+          entry.aux_idx = aux_entries.size();
           aux_entries.push_back({kSubMessage, {field}});
           if (options.lazy_opt == field_layout::kTvEager) {
             aux_entries.push_back({kMessageVerifyFunc, {field}});
@@ -817,8 +820,7 @@ TailCallTableInfo::TailCallTableInfo(
             aux_entries.push_back({kNothing});
           }
         } else {
-          field_entries.back().aux_idx =
-              TcParseTableBase::FieldEntry::kNoAuxIdx;
+          entry.aux_idx = TcParseTableBase::FieldEntry::kNoAuxIdx;
         }
       } else {
         AuxType type = options.is_implicitly_weak          ? kSubMessageWeak
@@ -827,10 +829,10 @@ TailCallTableInfo::TailCallTableInfo(
         if (message_options.should_profile_driven_cluster_aux_subtable &&
             type == kSubTable && is_non_cold(options)) {
           aux_entries[subtable_aux_idx] = {type, {field}};
-          field_entries.back().aux_idx = subtable_aux_idx;
+          entry.aux_idx = subtable_aux_idx;
           ++subtable_aux_idx;
         } else {
-          field_entries.back().aux_idx = aux_entries.size();
+          entry.aux_idx = aux_entries.size();
           aux_entries.push_back({type, {field}});
         }
       }