When using profiling information and weak descriptors use uniquely named

sections to allow for GC in the linker. These sections can be merged again after GC.

PiperOrigin-RevId: 602848779

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

@@ -1489,9 +1489,10 @@ bool UsingImplicitWeakDescriptor(const FileDescriptor* file,
          !options.opensource_runtime;
 }
 
-std::string WeakDefaultInstanceSection(const Descriptor* descriptor,
-                                       int index_in_file_messages,
-                                       const Options& options) {
+std::string WeakDescriptorDataSection(absl::string_view prefix,
+                                      const Descriptor* descriptor,
+                                      int index_in_file_messages,
+                                      const Options& options) {
   const auto* file = descriptor->file();
 
   // To make a compact name we use the index of the object in its file
@@ -1499,8 +1500,8 @@ std::string WeakDefaultInstanceSection(const Descriptor* descriptor,
   // So the name could be `pb_def_3_HASH` instead of
   // `pd_def_VeryLongClassName_WithNesting_AndMoreNames_HASH`
   // We need a know common prefix to merge the sections later on.
-  return UniqueName(absl::StrCat("pb_def_", index_in_file_messages), file,
-                    options);
+  return UniqueName(absl::StrCat("pb_", prefix, "_", index_in_file_messages),
+                    file, options);
 }
 
 bool UsingImplicitWeakFields(const FileDescriptor* file,

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

@@ -778,11 +778,23 @@ void ListAllTypesForServices(const FileDescriptor* fd,
 bool UsingImplicitWeakDescriptor(const FileDescriptor* file,
                                  const Options& options);
 
+// Generate the section name to be used for a data object when using implicit
+// weak descriptors. The prefix determines the kind of object and the section it
+// will be merged into afterwards.
+// See `UsingImplicitWeakDescriptor` above.
+std::string WeakDescriptorDataSection(absl::string_view prefix,
+                                      const Descriptor* descriptor,
+                                      int index_in_file_messages,
+                                      const Options& options);
+
 // Section name to be used for the default instance for implicit weak descriptor
 // objects. See `UsingImplicitWeakDescriptor` above.
-std::string WeakDefaultInstanceSection(const Descriptor* descriptor,
-                                       int index_in_file_messages,
-                                       const Options& options);
+inline std::string WeakDefaultInstanceSection(const Descriptor* descriptor,
+                                              int index_in_file_messages,
+                                              const Options& options) {
+  return WeakDescriptorDataSection("def", descriptor, index_in_file_messages,
+                                   options);
+}
 
 // Indicates whether we should use implicit weak fields for this file.
 bool UsingImplicitWeakFields(const FileDescriptor* file,

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

@@ -573,7 +573,8 @@ MessageGenerator::MessageGenerator(
 
   parse_function_generator_ = std::make_unique<ParseFunctionGenerator>(
       descriptor_, max_has_bit_index_, has_bit_indices_,
-      inlined_string_indices_, options_, scc_analyzer_, variables_);
+      inlined_string_indices_, options_, scc_analyzer_, variables_,
+      index_in_file_messages_);
 }
 
 size_t MessageGenerator::HasBitsSize() const {

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

@@ -103,14 +103,16 @@ ParseFunctionGenerator::ParseFunctionGenerator(
     const std::vector<int>& has_bit_indices,
     const std::vector<int>& inlined_string_indices, const Options& options,
     MessageSCCAnalyzer* scc_analyzer,
-    const absl::flat_hash_map<absl::string_view, std::string>& vars)
+    const absl::flat_hash_map<absl::string_view, std::string>& vars,
+    int index_in_file_messages)
     : descriptor_(descriptor),
       scc_analyzer_(scc_analyzer),
       options_(options),
       variables_(vars),
       inlined_string_indices_(inlined_string_indices),
       ordered_fields_(GetOrderedFields(descriptor_, options_)),
-      num_hasbits_(max_has_bit_index) {
+      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_,
@@ -228,16 +230,23 @@ void ParseFunctionGenerator::GenerateDataDecls(io::Printer* p) {
           {"SECTION",
            [&] {
              if (!IsProfileDriven(options_)) return;
+             std::string section_name;
              // Since most (>80%) messages are never present, messages that are
              // present are considered hot enough to be clustered together.
-             if (IsPresentMessage(descriptor_, options_)) {
-               p->Emit(
-                   "ABSL_ATTRIBUTE_SECTION_VARIABLE(proto_parse_table_hot)");
+             // When using weak descriptors we use unique sections for each
+             // table to allow for GC to work. pth/ptl names must be in sync
+             // with the linker script.
+             if (UsingImplicitWeakDescriptor(descriptor_->file(), options_)) {
+               section_name = WeakDescriptorDataSection(
+                   IsPresentMessage(descriptor_, options_) ? "pth" : "ptl",
+                   descriptor_, index_in_file_messages_, options_);
+             } else if (IsPresentMessage(descriptor_, options_)) {
+               section_name = "proto_parse_table_hot";
              } else {
-               p->Emit(
-                   "ABSL_ATTRIBUTE_SECTION_VARIABLE(proto_parse_table_"
-                   "lukewarm)");
+               section_name = "proto_parse_table_lukewarm";
              }
+             p->Emit({{"section_name", section_name}},
+                     "ABSL_ATTRIBUTE_SECTION_VARIABLE($section_name$)");
            }},
           {"table_size_log2", tc_table_info_->table_size_log2},
           {"num_field_entries", ordered_fields_.size()},

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

@@ -33,7 +33,8 @@ class ParseFunctionGenerator {
       const std::vector<int>& has_bit_indices,
       const std::vector<int>& inlined_string_indices, const Options& options,
       MessageSCCAnalyzer* scc_analyzer,
-      const absl::flat_hash_map<absl::string_view, std::string>& vars);
+      const absl::flat_hash_map<absl::string_view, std::string>& vars,
+      int index_in_file_messages);
 
   // Emits class-level method declarations to `printer`:
   void GenerateMethodDecls(io::Printer* printer);
@@ -70,6 +71,7 @@ class ParseFunctionGenerator {
   std::vector<int> inlined_string_indices_;
   const std::vector<const FieldDescriptor*> ordered_fields_;
   int num_hasbits_;
+  int index_in_file_messages_;
 };
 
 }  // namespace cpp