google-protobuf 3.22.0.rc.2 → 3.22.0.rc.3

data/ext/google/protobuf_c/ruby-upb.c

@@ -641,14 +641,12 @@ static int (*const compar[kUpb_FieldType_SizeOf])(const void*, const void*) = {
     [kUpb_FieldType_Bytes] = _upb_mapsorter_cmpstr,
 };
 
-bool _upb_mapsorter_pushmap(_upb_mapsorter* s, upb_FieldType key_type,
-                            const upb_Map* map, _upb_sortedmap* sorted) {
-  int map_size = _upb_Map_Size(map);
+static bool _upb_mapsorter_resize(_upb_mapsorter* s, _upb_sortedmap* sorted,
+                                  int size) {
   sorted->start = s->size;
   sorted->pos = sorted->start;
-  sorted->end = sorted->start + map_size;
+  sorted->end = sorted->start + size;
 
-  // Grow s->entries if necessary.
   if (sorted->end > s->cap) {
     s->cap = upb_Log2CeilingSize(sorted->end);
     s->entries = realloc(s->entries, s->cap * sizeof(*s->entries));
@@ -656,9 +654,17 @@ bool _upb_mapsorter_pushmap(_upb_mapsorter* s, upb_FieldType key_type,
   }
 
   s->size = sorted->end;
+  return true;
+}
+
+bool _upb_mapsorter_pushmap(_upb_mapsorter* s, upb_FieldType key_type,
+                            const upb_Map* map, _upb_sortedmap* sorted) {
+  int map_size = _upb_Map_Size(map);
+
+  if (!_upb_mapsorter_resize(s, sorted, map_size)) return false;
 
   // Copy non-empty entries from the table to s->entries.
-  upb_tabent const** dst = &s->entries[sorted->start];
+  const void** dst = &s->entries[sorted->start];
   const upb_tabent* src = map->table.t.entries;
   const upb_tabent* end = src + upb_table_size(&map->table.t);
   for (; src < end; src++) {
@@ -674,6 +680,29 @@ bool _upb_mapsorter_pushmap(_upb_mapsorter* s, upb_FieldType key_type,
         compar[key_type]);
   return true;
 }
+
+static int _upb_mapsorter_cmpext(const void* _a, const void* _b) {
+  const upb_Message_Extension* const* a = _a;
+  const upb_Message_Extension* const* b = _b;
+  uint32_t a_num = (*a)->ext->field.number;
+  uint32_t b_num = (*b)->ext->field.number;
+  assert(a_num != b_num);
+  return a_num < b_num ? -1 : 1;
+}
+
+bool _upb_mapsorter_pushexts(_upb_mapsorter* s,
+                             const upb_Message_Extension* exts, size_t count,
+                             _upb_sortedmap* sorted) {
+  if (!_upb_mapsorter_resize(s, sorted, count)) return false;
+
+  for (size_t i = 0; i < count; i++) {
+    s->entries[sorted->start + i] = &exts[i];
+  }
+
+  qsort(&s->entries[sorted->start], count, sizeof(*s->entries),
+        _upb_mapsorter_cmpext);
+  return true;
+}
 /* This file was generated by upbc (the upb compiler) from the input
  * file:
  *
@@ -1862,7 +1891,7 @@ char* upb_strdup2(const char* s, size_t len, upb_Arena* a) {
   n = len + 1;
   p = upb_Arena_Malloc(a, n);
   if (p) {
-    memcpy(p, s, len);
+    if (len != 0) memcpy(p, s, len);
     p[len] = 0;
   }
   return p;
@@ -5448,6 +5477,7 @@ upb_FindUnknownRet upb_MiniTable_FindUnknown(const upb_Message* msg,
   return ret;
 }
 
+// Warning: See TODO(b/267655898)
 upb_UnknownToMessageRet upb_MiniTable_PromoteUnknownToMessage(
     upb_Message* msg, const upb_MiniTable* mini_table,
     const upb_MiniTableField* field, const upb_MiniTable* sub_mini_table,
@@ -5458,7 +5488,10 @@ upb_UnknownToMessageRet upb_MiniTable_PromoteUnknownToMessage(
   // Callers should check that message is not set first before calling
   // PromotoUnknownToMessage.
   UPB_ASSERT(mini_table->subs[field->submsg_index].submsg == sub_mini_table);
-  UPB_ASSERT(upb_Message_GetMessage(msg, field, NULL) == NULL);
+  bool is_oneof = _upb_MiniTableField_InOneOf(field);
+  if (!is_oneof || _upb_getoneofcase_field(msg, field) == field->number) {
+    UPB_ASSERT(upb_Message_GetMessage(msg, field, NULL) == NULL);
+  }
   upb_UnknownToMessageRet ret;
   ret.status = kUpb_UnknownToMessage_Ok;
   do {
@@ -5486,6 +5519,9 @@ upb_UnknownToMessageRet upb_MiniTable_PromoteUnknownToMessage(
     }
   } while (unknown.status == kUpb_FindUnknown_Ok);
   if (message) {
+    if (is_oneof) {
+      *_upb_oneofcase_field(msg, field) = field->number;
+    }
     upb_Message_SetMessage(msg, mini_table, field, message);
     ret.message = message;
   }
@@ -5807,6 +5843,38 @@ upb_FieldType upb_MiniTableField_Type(const upb_MiniTableField* field) {
   return field->descriptortype;
 }
 
+static bool upb_MiniTable_Is_Oneof(const upb_MiniTableField* f) {
+  return f->presence < 0;
+}
+
+const upb_MiniTableField* upb_MiniTable_GetOneof(const upb_MiniTable* m,
+                                                 const upb_MiniTableField* f) {
+  if (UPB_UNLIKELY(!upb_MiniTable_Is_Oneof(f))) {
+    return NULL;
+  }
+  const upb_MiniTableField* ptr = &m->fields[0];
+  const upb_MiniTableField* end = &m->fields[m->field_count];
+  while (++ptr < end) {
+    if (ptr->presence == (*f).presence) {
+      return ptr;
+    }
+  }
+  return NULL;
+}
+
+bool upb_MiniTable_NextOneofField(const upb_MiniTable* m,
+                                  const upb_MiniTableField** f) {
+  const upb_MiniTableField* ptr = *f;
+  const upb_MiniTableField* end = &m->fields[m->field_count];
+  while (++ptr < end) {
+    if (ptr->presence == (*f)->presence) {
+      *f = ptr;
+      return true;
+    }
+  }
+  return false;
+}
+
 
 #include <inttypes.h>
 #include <stdlib.h>
@@ -6438,33 +6506,38 @@ static void upb_MtDecoder_AssignOffsets(upb_MtDecoder* d) {
   d->table->size = UPB_ALIGN_UP(d->table->size, 8);
 }
 
-static void upb_MtDecoder_ParseMap(upb_MtDecoder* d, const char* data,
-                                   size_t len) {
-  if (len < 2) {
-    upb_MtDecoder_ErrorFormat(d, "Invalid map encode length: %zu", len);
-    UPB_UNREACHABLE();
+static void upb_MtDecoder_ValidateEntryField(upb_MtDecoder* d,
+                                             const upb_MiniTableField* f,
+                                             int expected_num) {
+  const char* name = expected_num == 1 ? "key" : "val";
+  if (f->number != expected_num) {
+    upb_MtDecoder_ErrorFormat(d,
+                              "map %s did not have expected number (%d vs %d)",
+                              name, expected_num, (int)f->number);
   }
-  const upb_EncodedType key_type = _upb_FromBase92(data[0]);
-  switch (key_type) {
-    case kUpb_EncodedType_Fixed32:
-    case kUpb_EncodedType_Fixed64:
-    case kUpb_EncodedType_SFixed32:
-    case kUpb_EncodedType_SFixed64:
-    case kUpb_EncodedType_Int32:
-    case kUpb_EncodedType_UInt32:
-    case kUpb_EncodedType_SInt32:
-    case kUpb_EncodedType_Int64:
-    case kUpb_EncodedType_UInt64:
-    case kUpb_EncodedType_SInt64:
-    case kUpb_EncodedType_Bool:
-    case kUpb_EncodedType_String:
-      break;
 
-    default:
-      upb_MtDecoder_ErrorFormat(d, "Invalid map key field type: %d", key_type);
-      UPB_UNREACHABLE();
+  if (upb_IsRepeatedOrMap(f)) {
+    upb_MtDecoder_ErrorFormat(
+        d, "map %s cannot be repeated or map, or be in oneof", name);
+  }
+
+  uint32_t not_ok_types;
+  if (expected_num == 1) {
+    not_ok_types = (1 << kUpb_FieldType_Float) | (1 << kUpb_FieldType_Double) |
+                   (1 << kUpb_FieldType_Message) | (1 << kUpb_FieldType_Group) |
+                   (1 << kUpb_FieldType_Bytes) | (1 << kUpb_FieldType_Enum);
+  } else {
+    not_ok_types = 1 << kUpb_FieldType_Group;
   }
 
+  if ((1 << upb_MiniTableField_Type(f)) & not_ok_types) {
+    upb_MtDecoder_ErrorFormat(d, "map %s cannot have type %d", name,
+                              (int)f->descriptortype);
+  }
+}
+
+static void upb_MtDecoder_ParseMap(upb_MtDecoder* d, const char* data,
+                                   size_t len) {
   upb_MtDecoder_ParseMessage(d, data, len);
   upb_MtDecoder_AssignHasbits(d->table);
 
@@ -6473,29 +6546,15 @@ static void upb_MtDecoder_ParseMap(upb_MtDecoder* d, const char* data,
     UPB_UNREACHABLE();
   }
 
-  const int num0 = d->table->fields[0].number;
-  if (UPB_UNLIKELY(num0 != 1)) {
-    upb_MtDecoder_ErrorFormat(d, "field %d in map key", num0);
-    UPB_UNREACHABLE();
-  }
-
-  const int num1 = d->table->fields[1].number;
-  if (UPB_UNLIKELY(num1 != 2)) {
-    upb_MtDecoder_ErrorFormat(d, "field %d in map val", num1);
-    UPB_UNREACHABLE();
-  }
-
-  const int off0 = d->table->fields[0].offset;
-  if (UPB_UNLIKELY(off0 != kNoPresence && off0 != kHasbitPresence)) {
-    upb_MtDecoder_ErrorFormat(d, "bad offset %d in map key", off0);
-    UPB_UNREACHABLE();
+  upb_LayoutItem* end = UPB_PTRADD(d->vec.data, d->vec.size);
+  for (upb_LayoutItem* item = d->vec.data; item < end; item++) {
+    if (item->type == kUpb_LayoutItemType_OneofCase) {
+      upb_MtDecoder_ErrorFormat(d, "Map entry cannot have oneof");
+    }
   }
 
-  const int off1 = d->table->fields[1].offset;
-  if (UPB_UNLIKELY(off1 != kNoPresence && off1 != kHasbitPresence)) {
-    upb_MtDecoder_ErrorFormat(d, "bad offset %d in map val", off1);
-    UPB_UNREACHABLE();
-  }
+  upb_MtDecoder_ValidateEntryField(d, &d->table->fields[0], 1);
+  upb_MtDecoder_ValidateEntryField(d, &d->table->fields[1], 2);
 
   // Map entries have a pre-determined layout, regardless of types.
   // NOTE: sync with mini_table/message_internal.h.
@@ -6752,14 +6811,28 @@ bool upb_MiniTable_SetSubMessage(upb_MiniTable* table,
   UPB_ASSERT((uintptr_t)table->fields <= (uintptr_t)field &&
              (uintptr_t)field <
                  (uintptr_t)(table->fields + table->field_count));
-  // TODO: check these type invariants at runtime and return error to the
-  // caller if they are violated, instead of using an assert.
-  UPB_ASSERT(field->descriptortype == kUpb_FieldType_Message ||
-             field->descriptortype == kUpb_FieldType_Group);
-  if (sub->ext & kUpb_ExtMode_IsMapEntry) {
-    UPB_ASSERT(field->descriptortype == kUpb_FieldType_Message);
-    field->mode = (field->mode & ~kUpb_FieldMode_Mask) | kUpb_FieldMode_Map;
+  UPB_ASSERT(sub);
+
+  const bool sub_is_map = sub->ext & kUpb_ExtMode_IsMapEntry;
+
+  switch (field->descriptortype) {
+    case kUpb_FieldType_Message:
+      if (sub_is_map) {
+        const bool table_is_map = table->ext & kUpb_ExtMode_IsMapEntry;
+        if (UPB_UNLIKELY(table_is_map)) return false;
+
+        field->mode = (field->mode & ~kUpb_FieldMode_Mask) | kUpb_FieldMode_Map;
+      }
+      break;
+
+    case kUpb_FieldType_Group:
+      if (UPB_UNLIKELY(sub_is_map)) return false;
+      break;
+
+    default:
+      return false;
   }
+
   upb_MiniTableSub* table_sub = (void*)&table->subs[field->submsg_index];
   table_sub->submsg = sub;
   return true;
@@ -6770,6 +6843,8 @@ bool upb_MiniTable_SetSubEnum(upb_MiniTable* table, upb_MiniTableField* field,
   UPB_ASSERT((uintptr_t)table->fields <= (uintptr_t)field &&
              (uintptr_t)field <
                  (uintptr_t)(table->fields + table->field_count));
+  UPB_ASSERT(sub);
+
   upb_MiniTableSub* table_sub = (void*)&table->subs[field->submsg_index];
   table_sub->subenum = sub;
   return true;
@@ -7152,9 +7227,27 @@ void _upb_DefBuilder_OomErr(upb_DefBuilder* ctx) {
   _upb_DefBuilder_FailJmp(ctx);
 }
 
+// Verify a relative identifier string. The loop is branchless for speed.
+static void _upb_DefBuilder_CheckIdentNotFull(upb_DefBuilder* ctx,
+                                              upb_StringView name) {
+  bool good = name.size > 0;
+
+  for (size_t i = 0; i < name.size; i++) {
+    const char c = name.data[i];
+    const char d = c | 0x20;  // force lowercase
+    const bool is_alpha = (('a' <= d) & (d <= 'z')) | (c == '_');
+    const bool is_numer = ('0' <= c) & (c <= '9') & (i != 0);
+
+    good &= is_alpha | is_numer;
+  }
+
+  if (!good) _upb_DefBuilder_CheckIdentSlow(ctx, name, false);
+}
+
 const char* _upb_DefBuilder_MakeFullName(upb_DefBuilder* ctx,
                                          const char* prefix,
                                          upb_StringView name) {
+  _upb_DefBuilder_CheckIdentNotFull(ctx, name);
   if (prefix) {
     // ret = prefix + '.' + name;
     size_t n = strlen(prefix);
@@ -7270,7 +7363,7 @@ static bool TryGetChar(const char** src, const char* end, char* ch) {
   return true;
 }
 
-static char TryGetHexDigit(const char** src, const char* end) {
+static int TryGetHexDigit(const char** src, const char* end) {
   char ch;
   if (!TryGetChar(src, end, &ch)) return -1;
   if ('0' <= ch && ch <= '9') {
@@ -7287,10 +7380,10 @@ static char TryGetHexDigit(const char** src, const char* end) {
 static char upb_DefBuilder_ParseHexEscape(upb_DefBuilder* ctx,
                                           const upb_FieldDef* f,
                                           const char** src, const char* end) {
-  char hex_digit = TryGetHexDigit(src, end);
+  int hex_digit = TryGetHexDigit(src, end);
   if (hex_digit < 0) {
     _upb_DefBuilder_Errf(
-        ctx, "\\x cannot be followed by non-hex digit in field '%s' default",
+        ctx, "\\x must be followed by at least one hex digit (field='%s')",
         upb_FieldDef_FullName(f));
     return 0;
   }
@@ -7466,7 +7559,7 @@ err:
 }
 
 bool _upb_DefPool_InsertExt(upb_DefPool* s, const upb_MiniTableExtension* ext,
-                            upb_FieldDef* f) {
+                            const upb_FieldDef* f) {
   return upb_inttable_insert(&s->exts, (uintptr_t)ext, upb_value_constptr(f),
                              s->arena);
 }
@@ -7698,12 +7791,6 @@ static const upb_FileDef* _upb_DefPool_AddFile(
     const upb_MiniTableFile* layout, upb_Status* status) {
   const upb_StringView name = UPB_DESC(FileDescriptorProto_name)(file_proto);
 
-  if (name.size == 0) {
-    upb_Status_SetErrorFormat(status,
-                              "missing name in google_protobuf_FileDescriptorProto");
-    return NULL;
-  }
-
   // Determine whether we already know about this file.
   {
     upb_value v;
@@ -7912,6 +7999,7 @@ struct upb_EnumDef {
   int res_range_count;
   int res_name_count;
   int32_t defaultval;
+  bool is_closed;
   bool is_sorted;  // Whether all of the values are defined in ascending order.
 };
 
@@ -7919,15 +8007,6 @@ upb_EnumDef* _upb_EnumDef_At(const upb_EnumDef* e, int i) {
   return (upb_EnumDef*)&e[i];
 }
 
-// TODO: Maybe implement this on top of a ZCOS instead?
-void _upb_EnumDef_Debug(const upb_EnumDef* e) {
-  fprintf(stderr, "enum %s (%p) {\n", e->full_name, e);
-  fprintf(stderr, " value_count: %d\n", e->value_count);
-  fprintf(stderr, " default:     %d\n", e->defaultval);
-  fprintf(stderr, " is_sorted:   %d\n", e->is_sorted);
-  fprintf(stderr, "}\n");
-}
-
 const upb_MiniTableEnum* _upb_EnumDef_MiniTable(const upb_EnumDef* e) {
   return e->layout;
 }
@@ -8023,10 +8102,7 @@ const upb_EnumValueDef* upb_EnumDef_Value(const upb_EnumDef* e, int i) {
   return _upb_EnumValueDef_At(e->values, i);
 }
 
-bool upb_EnumDef_IsClosed(const upb_EnumDef* e) {
-  if (UPB_TREAT_PROTO2_ENUMS_LIKE_PROTO3) return false;
-  return upb_FileDef_Syntax(e->file) == kUpb_Syntax_Proto2;
-}
+bool upb_EnumDef_IsClosed(const upb_EnumDef* e) { return e->is_closed; }
 
 bool upb_EnumDef_MiniDescriptorEncode(const upb_EnumDef* e, upb_Arena* a,
                                       upb_StringView* out) {
@@ -8106,12 +8182,14 @@ static void create_enumdef(upb_DefBuilder* ctx, const char* prefix,
   e->file = _upb_DefBuilder_File(ctx);
 
   name = UPB_DESC(EnumDescriptorProto_name)(enum_proto);
-  _upb_DefBuilder_CheckIdentNotFull(ctx, name);
 
   e->full_name = _upb_DefBuilder_MakeFullName(ctx, prefix, name);
   _upb_DefBuilder_Add(ctx, e->full_name,
                       _upb_DefType_Pack(e, UPB_DEFTYPE_ENUM));
 
+  e->is_closed = (!UPB_TREAT_PROTO2_ENUMS_LIKE_PROTO3) &&
+                 (upb_FileDef_Syntax(e->file) == kUpb_Syntax_Proto2);
+
   values = UPB_DESC(EnumDescriptorProto_value)(enum_proto, &n_value);
 
   bool ok = upb_strtable_init(&e->ntoi, n_value, ctx->arena);
@@ -8144,7 +8222,7 @@ static void create_enumdef(upb_DefBuilder* ctx, const char* prefix,
 
   upb_inttable_compact(&e->iton, ctx->arena);
 
-  if (upb_FileDef_Syntax(e->file) == kUpb_Syntax_Proto2) {
+  if (e->is_closed) {
     if (ctx->layout) {
       UPB_ASSERT(ctx->enum_count < ctx->layout->enum_count);
       e->layout = ctx->layout->enums[ctx->enum_count++];
@@ -8443,10 +8521,11 @@ struct upb_FieldDef {
   uint16_t index_;
   uint16_t layout_index;  // Index into msgdef->layout->fields or file->exts
   bool has_default;
-  bool is_extension_;
-  bool is_packed_;
-  bool proto3_optional_;
-  bool has_json_name_;
+  bool has_json_name;
+  bool has_presence;
+  bool is_extension;
+  bool is_packed;
+  bool is_proto3_optional;
   upb_FieldType type_;
   upb_Label label_;
 #if UINTPTR_MAX == 0xffffffff
@@ -8513,11 +8592,9 @@ upb_Label upb_FieldDef_Label(const upb_FieldDef* f) { return f->label_; }
 
 uint32_t upb_FieldDef_Number(const upb_FieldDef* f) { return f->number_; }
 
-bool upb_FieldDef_IsExtension(const upb_FieldDef* f) {
-  return f->is_extension_;
-}
+bool upb_FieldDef_IsExtension(const upb_FieldDef* f) { return f->is_extension; }
 
-bool upb_FieldDef_IsPacked(const upb_FieldDef* f) { return f->is_packed_; }
+bool upb_FieldDef_IsPacked(const upb_FieldDef* f) { return f->is_packed; }
 
 const char* upb_FieldDef_Name(const upb_FieldDef* f) {
   return _upb_DefBuilder_FullToShort(f->full_name);
@@ -8528,7 +8605,7 @@ const char* upb_FieldDef_JsonName(const upb_FieldDef* f) {
 }
 
 bool upb_FieldDef_HasJsonName(const upb_FieldDef* f) {
-  return f->has_json_name_;
+  return f->has_json_name;
 }
 
 const upb_FileDef* upb_FieldDef_File(const upb_FieldDef* f) { return f->file; }
@@ -8538,11 +8615,11 @@ const upb_MessageDef* upb_FieldDef_ContainingType(const upb_FieldDef* f) {
 }
 
 const upb_MessageDef* upb_FieldDef_ExtensionScope(const upb_FieldDef* f) {
-  return f->is_extension_ ? f->scope.extension_scope : NULL;
+  return f->is_extension ? f->scope.extension_scope : NULL;
 }
 
 const upb_OneofDef* upb_FieldDef_ContainingOneof(const upb_FieldDef* f) {
-  return f->is_extension_ ? NULL : f->scope.oneof;
+  return f->is_extension ? NULL : f->scope.oneof;
 }
 
 const upb_OneofDef* upb_FieldDef_RealContainingOneof(const upb_FieldDef* f) {
@@ -8619,22 +8696,18 @@ const upb_MiniTableExtension* _upb_FieldDef_ExtensionMiniTable(
 }
 
 bool _upb_FieldDef_IsClosedEnum(const upb_FieldDef* f) {
-  if (UPB_TREAT_PROTO2_ENUMS_LIKE_PROTO3) return false;
   if (f->type_ != kUpb_FieldType_Enum) return false;
-
-  // TODO: Maybe make is_proto2 a bool at creation?
-  const upb_FileDef* file = upb_EnumDef_File(f->sub.enumdef);
-  return upb_FileDef_Syntax(file) == kUpb_Syntax_Proto2;
+  return upb_EnumDef_IsClosed(f->sub.enumdef);
 }
 
 bool _upb_FieldDef_IsProto3Optional(const upb_FieldDef* f) {
-  return f->proto3_optional_;
+  return f->is_proto3_optional;
 }
 
 int _upb_FieldDef_LayoutIndex(const upb_FieldDef* f) { return f->layout_index; }
 
 uint64_t _upb_FieldDef_Modifiers(const upb_FieldDef* f) {
-  uint64_t out = f->is_packed_ ? kUpb_FieldModifier_IsPacked : 0;
+  uint64_t out = f->is_packed ? kUpb_FieldModifier_IsPacked : 0;
 
   switch (f->label_) {
     case kUpb_Label_Optional:
@@ -8657,13 +8730,7 @@ uint64_t _upb_FieldDef_Modifiers(const upb_FieldDef* f) {
 }
 
 bool upb_FieldDef_HasDefault(const upb_FieldDef* f) { return f->has_default; }
-
-bool upb_FieldDef_HasPresence(const upb_FieldDef* f) {
-  if (upb_FieldDef_IsRepeated(f)) return false;
-  const upb_FileDef* file = upb_FieldDef_File(f);
-  return upb_FieldDef_IsSubMessage(f) || upb_FieldDef_ContainingOneof(f) ||
-         upb_FileDef_Syntax(file) == kUpb_Syntax_Proto2;
-}
+bool upb_FieldDef_HasPresence(const upb_FieldDef* f) { return f->has_presence; }
 
 bool upb_FieldDef_HasSubDef(const upb_FieldDef* f) {
   return upb_FieldDef_IsSubMessage(f) ||
@@ -8924,10 +8991,17 @@ static void _upb_FieldDef_Create(upb_DefBuilder* ctx, const char* prefix,
   }
 
   const upb_StringView name = UPB_DESC(FieldDescriptorProto_name)(field_proto);
-  _upb_DefBuilder_CheckIdentNotFull(ctx, name);
 
-  f->has_json_name_ = UPB_DESC(FieldDescriptorProto_has_json_name)(field_proto);
-  if (f->has_json_name_) {
+  f->full_name = _upb_DefBuilder_MakeFullName(ctx, prefix, name);
+  f->label_ = (int)UPB_DESC(FieldDescriptorProto_label)(field_proto);
+  f->number_ = UPB_DESC(FieldDescriptorProto_number)(field_proto);
+  f->is_proto3_optional =
+      UPB_DESC(FieldDescriptorProto_proto3_optional)(field_proto);
+  f->msgdef = m;
+  f->scope.oneof = NULL;
+
+  f->has_json_name = UPB_DESC(FieldDescriptorProto_has_json_name)(field_proto);
+  if (f->has_json_name) {
     const upb_StringView sv =
         UPB_DESC(FieldDescriptorProto_json_name)(field_proto);
     f->json_name = upb_strdup2(sv.data, sv.size, ctx->arena);
@@ -8936,14 +9010,6 @@ static void _upb_FieldDef_Create(upb_DefBuilder* ctx, const char* prefix,
   }
   if (!f->json_name) _upb_DefBuilder_OomErr(ctx);
 
-  f->full_name = _upb_DefBuilder_MakeFullName(ctx, prefix, name);
-  f->label_ = (int)UPB_DESC(FieldDescriptorProto_label)(field_proto);
-  f->number_ = UPB_DESC(FieldDescriptorProto_number)(field_proto);
-  f->proto3_optional_ =
-      UPB_DESC(FieldDescriptorProto_proto3_optional)(field_proto);
-  f->msgdef = m;
-  f->scope.oneof = NULL;
-
   const bool has_type = UPB_DESC(FieldDescriptorProto_has_type)(field_proto);
   const bool has_type_name =
       UPB_DESC(FieldDescriptorProto_has_type_name)(field_proto);
@@ -9021,21 +9087,26 @@ static void _upb_FieldDef_Create(upb_DefBuilder* ctx, const char* prefix,
   UPB_DEF_SET_OPTIONS(f->opts, FieldDescriptorProto, FieldOptions, field_proto);
 
   if (UPB_DESC(FieldOptions_has_packed)(f->opts)) {
-    f->is_packed_ = UPB_DESC(FieldOptions_packed)(f->opts);
+    f->is_packed = UPB_DESC(FieldOptions_packed)(f->opts);
   } else {
     // Repeated fields default to packed for proto3 only.
-    f->is_packed_ = upb_FieldDef_IsPrimitive(f) &&
-                    f->label_ == kUpb_Label_Repeated &&
-                    upb_FileDef_Syntax(f->file) == kUpb_Syntax_Proto3;
+    f->is_packed = upb_FieldDef_IsPrimitive(f) &&
+                   f->label_ == kUpb_Label_Repeated &&
+                   upb_FileDef_Syntax(f->file) == kUpb_Syntax_Proto3;
   }
+
+  f->has_presence =
+      (!upb_FieldDef_IsRepeated(f)) &&
+      (upb_FieldDef_IsSubMessage(f) || upb_FieldDef_ContainingOneof(f) ||
+       (upb_FileDef_Syntax(f->file) == kUpb_Syntax_Proto2));
 }
 
 static void _upb_FieldDef_CreateExt(upb_DefBuilder* ctx, const char* prefix,
                                     const UPB_DESC(FieldDescriptorProto) *
                                         field_proto,
                                     upb_MessageDef* m, upb_FieldDef* f) {
+  f->is_extension = true;
   _upb_FieldDef_Create(ctx, prefix, field_proto, m, f);
-  f->is_extension_ = true;
 
   if (UPB_DESC(FieldDescriptorProto_has_oneof_index)(field_proto)) {
     _upb_DefBuilder_Errf(ctx, "oneof_index provided for extension field (%s)",
@@ -9055,11 +9126,11 @@ static void _upb_FieldDef_CreateNotExt(upb_DefBuilder* ctx, const char* prefix,
                                        const UPB_DESC(FieldDescriptorProto) *
                                            field_proto,
                                        upb_MessageDef* m, upb_FieldDef* f) {
+  f->is_extension = false;
   _upb_FieldDef_Create(ctx, prefix, field_proto, m, f);
-  f->is_extension_ = false;
 
   if (!UPB_DESC(FieldDescriptorProto_has_oneof_index)(field_proto)) {
-    if (f->proto3_optional_) {
+    if (f->is_proto3_optional) {
       _upb_DefBuilder_Errf(
           ctx,
           "non-extension field (%s) with proto3_optional was not in a oneof",
@@ -9068,19 +9139,24 @@ static void _upb_FieldDef_CreateNotExt(upb_DefBuilder* ctx, const char* prefix,
   }
 
   _upb_MessageDef_InsertField(ctx, m, f);
+}
 
-  if (!ctx->layout) return;
+upb_FieldDef* _upb_Extensions_New(
+    upb_DefBuilder* ctx, int n,
+    const UPB_DESC(FieldDescriptorProto) * const* protos, const char* prefix,
+    upb_MessageDef* m) {
+  _upb_DefType_CheckPadding(sizeof(upb_FieldDef));
+  upb_FieldDef* defs =
+      (upb_FieldDef*)_upb_DefBuilder_Alloc(ctx, sizeof(upb_FieldDef) * n);
 
-  const upb_MiniTable* mt = upb_MessageDef_MiniTable(m);
-  const upb_MiniTableField* fields = mt->fields;
-  for (int i = 0; i < mt->field_count; i++) {
-    if (fields[i].number == f->number_) {
-      f->layout_index = i;
-      return;
-    }
+  for (int i = 0; i < n; i++) {
+    upb_FieldDef* f = &defs[i];
+
+    _upb_FieldDef_CreateExt(ctx, prefix, protos[i], m, f);
+    f->index_ = i;
   }
 
-  UPB_ASSERT(false);  // It should be impossible to reach this point.
+  return defs;
 }
 
 upb_FieldDef* _upb_FieldDefs_New(
@@ -9091,28 +9167,23 @@ upb_FieldDef* _upb_FieldDefs_New(
   upb_FieldDef* defs =
       (upb_FieldDef*)_upb_DefBuilder_Alloc(ctx, sizeof(upb_FieldDef) * n);
 
-  // If we are creating extensions then is_sorted will be NULL.
-  // If we are not creating extensions then is_sorted will be non-NULL.
-  if (is_sorted) {
-    uint32_t previous = 0;
-    for (int i = 0; i < n; i++) {
-      upb_FieldDef* f = &defs[i];
-
-      _upb_FieldDef_CreateNotExt(ctx, prefix, protos[i], m, f);
-      f->index_ = i;
-      if (!ctx->layout) f->layout_index = i;
-
-      const uint32_t current = f->number_;
-      if (previous > current) *is_sorted = false;
-      previous = current;
+  uint32_t previous = 0;
+  for (int i = 0; i < n; i++) {
+    upb_FieldDef* f = &defs[i];
+
+    _upb_FieldDef_CreateNotExt(ctx, prefix, protos[i], m, f);
+    f->index_ = i;
+    if (!ctx->layout) {
+      // Speculate that the def fields are sorted.  We will always sort the
+      // MiniTable fields, so if defs are sorted then indices will match.
+      //
+      // If this is incorrect, we will overwrite later.
+      f->layout_index = i;
     }
-  } else {
-    for (int i = 0; i < n; i++) {
-      upb_FieldDef* f = &defs[i];
 
-      _upb_FieldDef_CreateExt(ctx, prefix, protos[i], m, f);
-      f->index_ = i;
-    }
+    const uint32_t current = f->number_;
+    if (previous > current) *is_sorted = false;
+    previous = current;
   }
 
   return defs;
@@ -9168,9 +9239,12 @@ static int _upb_FieldDef_Compare(const void* p1, const void* p2) {
   return (v1 < v2) ? -1 : (v1 > v2);
 }
 
+// _upb_FieldDefs_Sorted() is mostly a pure function of its inputs, but has one
+// critical side effect that we depend on: it sets layout_index appropriately
+// for non-sorted lists of fields.
 const upb_FieldDef** _upb_FieldDefs_Sorted(const upb_FieldDef* f, int n,
                                            upb_Arena* a) {
-  // TODO: Try to replace this arena alloc with a persistent scratch buffer.
+  // TODO(salo): Replace this arena alloc with a persistent scratch buffer.
   upb_FieldDef** out = (upb_FieldDef**)upb_Arena_Malloc(a, n * sizeof(void*));
   if (!out) return NULL;
 
@@ -9187,7 +9261,7 @@ const upb_FieldDef** _upb_FieldDefs_Sorted(const upb_FieldDef* f, int n,
 
 bool upb_FieldDef_MiniDescriptorEncode(const upb_FieldDef* f, upb_Arena* a,
                                        upb_StringView* out) {
-  UPB_ASSERT(f->is_extension_);
+  UPB_ASSERT(f->is_extension);
 
   upb_DescState s;
   _upb_DescState_Init(&s);
@@ -9225,7 +9299,10 @@ static void resolve_extension(upb_DefBuilder* ctx, const char* prefix,
         "field number %u in extension %s has no extension range in message %s",
         (unsigned)f->number_, f->full_name, upb_MessageDef_FullName(m));
   }
+}
 
+void _upb_FieldDef_BuildMiniTableExtension(upb_DefBuilder* ctx,
+                                           const upb_FieldDef* f) {
   const upb_MiniTableExtension* ext = _upb_FieldDef_ExtensionMiniTable(f);
 
   if (ctx->layout) {
@@ -9244,8 +9321,8 @@ static void resolve_extension(upb_DefBuilder* ctx, const char* prefix,
       sub.subenum = _upb_EnumDef_MiniTable(f->sub.enumdef);
     }
     bool ok2 = upb_MiniTableExtension_Build(desc.data, desc.size, mut_ext,
-                                            upb_MessageDef_MiniTable(m), sub,
-                                            ctx->status);
+                                            upb_MessageDef_MiniTable(f->msgdef),
+                                            sub, ctx->status);
     if (!ok2) _upb_DefBuilder_Errf(ctx, "Could not build extension mini table");
   }
 
@@ -9290,7 +9367,7 @@ void _upb_FieldDef_Resolve(upb_DefBuilder* ctx, const char* prefix,
   resolve_subdef(ctx, prefix, f);
   resolve_default(ctx, f, field_proto);
 
-  if (f->is_extension_) {
+  if (f->is_extension) {
     resolve_extension(ctx, prefix, f, field_proto);
   }
 }
@@ -9302,6 +9379,7 @@ struct upb_FileDef {
   const UPB_DESC(FileOptions) * opts;
   const char* name;
   const char* package;
+  const char* edition;
 
   const upb_FileDef** deps;
   const int32_t* public_deps;
@@ -9338,6 +9416,10 @@ const char* upb_FileDef_Package(const upb_FileDef* f) {
   return f->package ? f->package : "";
 }
 
+const char* upb_FileDef_Edition(const upb_FileDef* f) {
+  return f->edition ? f->edition : "";
+}
+
 const char* _upb_FileDef_RawPackage(const upb_FileDef* f) { return f->package; }
 
 upb_Syntax upb_FileDef_Syntax(const upb_FileDef* f) { return f->syntax; }
@@ -9485,13 +9567,14 @@ void _upb_FileDef_Create(upb_DefBuilder* ctx,
     }
   }
 
-  if (!UPB_DESC(FileDescriptorProto_has_name)(file_proto)) {
-    _upb_DefBuilder_Errf(ctx, "File has no name");
+  upb_StringView name = UPB_DESC(FileDescriptorProto_name)(file_proto);
+  file->name = strviewdup(ctx, name);
+  if (strlen(file->name) != name.size) {
+    _upb_DefBuilder_Errf(ctx, "File name contained embedded NULL");
   }
 
-  file->name = strviewdup(ctx, UPB_DESC(FileDescriptorProto_name)(file_proto));
-
   upb_StringView package = UPB_DESC(FileDescriptorProto_package)(file_proto);
+
   if (package.size) {
     _upb_DefBuilder_CheckIdentFull(ctx, package);
     file->package = strviewdup(ctx, package);
@@ -9499,6 +9582,18 @@ void _upb_FileDef_Create(upb_DefBuilder* ctx,
     file->package = NULL;
   }
 
+  upb_StringView edition = UPB_DESC(FileDescriptorProto_edition)(file_proto);
+
+  if (edition.size == 0) {
+    file->edition = NULL;
+  } else {
+    // TODO(b/267770604): How should we validate this?
+    file->edition = strviewdup(ctx, edition);
+    if (strlen(file->edition) != edition.size) {
+      _upb_DefBuilder_Errf(ctx, "Edition name contained embedded NULL");
+    }
+  }
+
   if (UPB_DESC(FileDescriptorProto_has_syntax)(file_proto)) {
     upb_StringView syntax = UPB_DESC(FileDescriptorProto_syntax)(file_proto);
 
@@ -9567,8 +9662,7 @@ void _upb_FileDef_Create(upb_DefBuilder* ctx,
   // Create extensions.
   exts = UPB_DESC(FileDescriptorProto_extension)(file_proto, &n);
   file->top_lvl_ext_count = n;
-  file->top_lvl_exts =
-      _upb_FieldDefs_New(ctx, n, exts, file->package, NULL, NULL);
+  file->top_lvl_exts = _upb_Extensions_New(ctx, n, exts, file->package, NULL);
 
   // Create messages.
   msgs = UPB_DESC(FileDescriptorProto_message_type)(file_proto, &n);
@@ -9592,11 +9686,19 @@ void _upb_FileDef_Create(upb_DefBuilder* ctx,
     _upb_FieldDef_Resolve(ctx, file->package, f);
   }
 
-  if (!ctx->layout) {
-    for (int i = 0; i < file->top_lvl_msg_count; i++) {
-      upb_MessageDef* m = (upb_MessageDef*)upb_FileDef_TopLevelMessage(file, i);
-      _upb_MessageDef_LinkMiniTable(ctx, m);
-    }
+  for (int i = 0; i < file->top_lvl_msg_count; i++) {
+    upb_MessageDef* m = (upb_MessageDef*)upb_FileDef_TopLevelMessage(file, i);
+    _upb_MessageDef_CreateMiniTable(ctx, (upb_MessageDef*)m);
+  }
+
+  for (int i = 0; i < file->top_lvl_ext_count; i++) {
+    upb_FieldDef* f = (upb_FieldDef*)upb_FileDef_TopLevelExtension(file, i);
+    _upb_FieldDef_BuildMiniTableExtension(ctx, f);
+  }
+
+  for (int i = 0; i < file->top_lvl_msg_count; i++) {
+    upb_MessageDef* m = (upb_MessageDef*)upb_FileDef_TopLevelMessage(file, i);
+    _upb_MessageDef_LinkMiniTable(ctx, m);
   }
 
   if (file->ext_count) {
@@ -10109,6 +10211,8 @@ bool upb_MessageDef_IsMessageSet(const upb_MessageDef* m) {
 static upb_MiniTable* _upb_MessageDef_MakeMiniTable(upb_DefBuilder* ctx,
                                                     const upb_MessageDef* m) {
   upb_StringView desc;
+  // Note: this will assign layout_index for fields, so upb_FieldDef_MiniTable()
+  // is safe to call only after this call.
   bool ok = upb_MessageDef_MiniDescriptorEncode(m, ctx->tmp_arena, &desc);
   if (!ok) _upb_DefBuilder_OomErr(ctx);
 
@@ -10128,23 +10232,6 @@ void _upb_MessageDef_Resolve(upb_DefBuilder* ctx, upb_MessageDef* m) {
     _upb_FieldDef_Resolve(ctx, m->full_name, f);
   }
 
-  if (!ctx->layout) {
-    m->layout = _upb_MessageDef_MakeMiniTable(ctx, m);
-    if (!m->layout) _upb_DefBuilder_OomErr(ctx);
-  }
-
-#ifndef NDEBUG
-  for (int i = 0; i < m->field_count; i++) {
-    const upb_FieldDef* f = upb_MessageDef_Field(m, i);
-    const int layout_index = _upb_FieldDef_LayoutIndex(f);
-    UPB_ASSERT(layout_index < m->layout->field_count);
-    const upb_MiniTableField* mt_f = &m->layout->fields[layout_index];
-    UPB_ASSERT(upb_FieldDef_Type(f) == upb_MiniTableField_Type(mt_f));
-    UPB_ASSERT(upb_FieldDef_HasPresence(f) ==
-               upb_MiniTableField_HasPresence(mt_f));
-  }
-#endif
-
   m->in_message_set = false;
   for (int i = 0; i < upb_MessageDef_NestedExtensionCount(m); i++) {
     upb_FieldDef* ext = (upb_FieldDef*)upb_MessageDef_NestedExtension(m, i);
@@ -10207,8 +10294,39 @@ void _upb_MessageDef_InsertField(upb_DefBuilder* ctx, upb_MessageDef* m,
   if (!ok) _upb_DefBuilder_OomErr(ctx);
 }
 
+void _upb_MessageDef_CreateMiniTable(upb_DefBuilder* ctx, upb_MessageDef* m) {
+  if (ctx->layout == NULL) {
+    m->layout = _upb_MessageDef_MakeMiniTable(ctx, m);
+  } else {
+    UPB_ASSERT(ctx->msg_count < ctx->layout->msg_count);
+    m->layout = ctx->layout->msgs[ctx->msg_count++];
+    UPB_ASSERT(m->field_count == m->layout->field_count);
+
+    // We don't need the result of this call, but it will assign layout_index
+    // for all the fields in O(n lg n) time.
+    _upb_FieldDefs_Sorted(m->fields, m->field_count, ctx->tmp_arena);
+  }
+
+  for (int i = 0; i < m->nested_msg_count; i++) {
+    upb_MessageDef* nested =
+        (upb_MessageDef*)upb_MessageDef_NestedMessage(m, i);
+    _upb_MessageDef_CreateMiniTable(ctx, nested);
+  }
+}
+
 void _upb_MessageDef_LinkMiniTable(upb_DefBuilder* ctx,
                                    const upb_MessageDef* m) {
+  for (int i = 0; i < upb_MessageDef_NestedExtensionCount(m); i++) {
+    const upb_FieldDef* ext = upb_MessageDef_NestedExtension(m, i);
+    _upb_FieldDef_BuildMiniTableExtension(ctx, ext);
+  }
+
+  for (int i = 0; i < m->nested_msg_count; i++) {
+    _upb_MessageDef_LinkMiniTable(ctx, upb_MessageDef_NestedMessage(m, i));
+  }
+
+  if (ctx->layout) return;
+
   for (int i = 0; i < m->field_count; i++) {
     const upb_FieldDef* f = upb_MessageDef_Field(m, i);
     const upb_MessageDef* sub_m = upb_FieldDef_MessageSubDef(f);
@@ -10236,9 +10354,17 @@ void _upb_MessageDef_LinkMiniTable(upb_DefBuilder* ctx,
     }
   }
 
-  for (int i = 0; i < m->nested_msg_count; i++) {
-    _upb_MessageDef_LinkMiniTable(ctx, upb_MessageDef_NestedMessage(m, i));
+#ifndef NDEBUG
+  for (int i = 0; i < m->field_count; i++) {
+    const upb_FieldDef* f = upb_MessageDef_Field(m, i);
+    const int layout_index = _upb_FieldDef_LayoutIndex(f);
+    UPB_ASSERT(layout_index < m->layout->field_count);
+    const upb_MiniTableField* mt_f = &m->layout->fields[layout_index];
+    UPB_ASSERT(upb_FieldDef_Type(f) == upb_MiniTableField_Type(mt_f));
+    UPB_ASSERT(upb_FieldDef_HasPresence(f) ==
+               upb_MiniTableField_HasPresence(mt_f));
   }
+#endif
 }
 
 static uint64_t _upb_MessageDef_Modifiers(const upb_MessageDef* m) {
@@ -10371,7 +10497,6 @@ static void create_msgdef(upb_DefBuilder* ctx, const char* prefix,
   m->is_sorted = true;
 
   name = UPB_DESC(DescriptorProto_name)(msg_proto);
-  _upb_DefBuilder_CheckIdentNotFull(ctx, name);
 
   m->full_name = _upb_DefBuilder_MakeFullName(ctx, prefix, name);
   _upb_DefBuilder_Add(ctx, m->full_name, _upb_DefType_Pack(m, UPB_DEFTYPE_MSG));
@@ -10390,17 +10515,6 @@ static void create_msgdef(upb_DefBuilder* ctx, const char* prefix,
   ok = upb_strtable_init(&m->ntof, n_oneof + n_field, ctx->arena);
   if (!ok) _upb_DefBuilder_OomErr(ctx);
 
-  if (ctx->layout) {
-    /* create_fielddef() below depends on this being set. */
-    UPB_ASSERT(ctx->msg_count < ctx->layout->msg_count);
-    m->layout = ctx->layout->msgs[ctx->msg_count++];
-    UPB_ASSERT(n_field == m->layout->field_count);
-  } else {
-    /* Allocate now (to allow cross-linking), populate later. */
-    m->layout = _upb_DefBuilder_Alloc(
-        ctx, sizeof(*m->layout) + sizeof(_upb_FastTable_Entry));
-  }
-
   UPB_DEF_SET_OPTIONS(m->opts, DescriptorProto, MessageOptions, msg_proto);
 
   m->oneof_count = n_oneof;
@@ -10441,7 +10555,7 @@ static void create_msgdef(upb_DefBuilder* ctx, const char* prefix,
   const UPB_DESC(FieldDescriptorProto)* const* exts =
       UPB_DESC(DescriptorProto_extension)(msg_proto, &n_ext);
   m->nested_ext_count = n_ext;
-  m->nested_exts = _upb_FieldDefs_New(ctx, n_ext, exts, m->full_name, m, NULL);
+  m->nested_exts = _upb_Extensions_New(ctx, n_ext, exts, m->full_name, m);
 
   const UPB_DESC(DescriptorProto)* const* msgs =
       UPB_DESC(DescriptorProto_nested_type)(msg_proto, &n_msg);
@@ -10848,7 +10962,6 @@ static void create_service(upb_DefBuilder* ctx,
   s->file = _upb_DefBuilder_File(ctx);
 
   name = UPB_DESC(ServiceDescriptorProto_name)(svc_proto);
-  _upb_DefBuilder_CheckIdentNotFull(ctx, name);
   const char* package = _upb_FileDef_RawPackage(s->file);
   s->full_name = _upb_DefBuilder_MakeFullName(ctx, package, name);
   _upb_DefBuilder_Add(ctx, s->full_name,
@@ -11420,33 +11533,48 @@ static const char* _upb_Decoder_DecodeToMap(upb_Decoder* d, const char* ptr,
   upb_Map** map_p = UPB_PTR_AT(msg, field->offset, upb_Map*);
   upb_Map* map = *map_p;
   upb_MapEntry ent;
+  UPB_ASSERT(upb_MiniTableField_Type(field) == kUpb_FieldType_Message);
   const upb_MiniTable* entry = subs[field->submsg_index].submsg;
 
+  UPB_ASSERT(entry->field_count == 2);
+  UPB_ASSERT(!upb_IsRepeatedOrMap(&entry->fields[0]));
+  UPB_ASSERT(!upb_IsRepeatedOrMap(&entry->fields[1]));
+
   if (!map) {
     map = _upb_Decoder_CreateMap(d, entry);
     *map_p = map;
   }
 
-  /* Parse map entry. */
+  // Parse map entry.
   memset(&ent, 0, sizeof(ent));
 
   if (entry->fields[1].descriptortype == kUpb_FieldType_Message ||
       entry->fields[1].descriptortype == kUpb_FieldType_Group) {
-    /* Create proactively to handle the case where it doesn't appear. */
-    ent.data.v.val =
-        upb_value_ptr(_upb_Message_New(entry->subs[0].submsg, &d->arena));
+    const upb_MiniTable* submsg_table = entry->subs[0].submsg;
+    // Any sub-message entry must be linked.  We do not allow dynamic tree
+    // shaking in this case.
+    UPB_ASSERT(submsg_table);
+
+    // Create proactively to handle the case where it doesn't appear. */
+    ent.data.v.val = upb_value_ptr(_upb_Message_New(submsg_table, &d->arena));
   }
 
-  const char* start = ptr;
   ptr =
       _upb_Decoder_DecodeSubMessage(d, ptr, &ent.data, subs, field, val->size);
   // check if ent had any unknown fields
   size_t size;
   upb_Message_GetUnknown(&ent.data, &size);
   if (size != 0) {
+    char* buf;
+    size_t size;
     uint32_t tag = ((uint32_t)field->number << 3) | kUpb_WireType_Delimited;
-    _upb_Decoder_AddUnknownVarints(d, msg, tag, (uint32_t)(ptr - start));
-    if (!_upb_Message_AddUnknown(msg, start, ptr - start, &d->arena)) {
+    upb_EncodeStatus status =
+        upb_Encode(&ent.data, entry, 0, &d->arena, &buf, &size);
+    if (status != kUpb_EncodeStatus_Ok) {
+      _upb_Decoder_ErrorJmp(d, kUpb_DecodeStatus_OutOfMemory);
+    }
+    _upb_Decoder_AddUnknownVarints(d, msg, tag, size);
+    if (!_upb_Message_AddUnknown(msg, buf, size, &d->arena)) {
       _upb_Decoder_ErrorJmp(d, kUpb_DecodeStatus_OutOfMemory);
     }
   } else {
@@ -13601,6 +13729,15 @@ static void encode_msgset_item(upb_encstate* e,
   encode_tag(e, kUpb_MsgSet_Item, kUpb_WireType_StartGroup);
 }
 
+static void encode_ext(upb_encstate* e, const upb_Message_Extension* ext,
+                       bool is_message_set) {
+  if (UPB_UNLIKELY(is_message_set)) {
+    encode_msgset_item(e, ext);
+  } else {
+    encode_field(e, &ext->data, &ext->ext->sub, &ext->ext->field);
+  }
+}
+
 static void encode_message(upb_encstate* e, const upb_Message* msg,
                            const upb_MiniTable* m, size_t* size) {
   size_t pre_len = e->limit - e->ptr;
@@ -13630,12 +13767,17 @@ static void encode_message(upb_encstate* e, const upb_Message* msg,
     size_t ext_count;
     const upb_Message_Extension* ext = _upb_Message_Getexts(msg, &ext_count);
     if (ext_count) {
-      const upb_Message_Extension* end = ext + ext_count;
-      for (; ext != end; ext++) {
-        if (UPB_UNLIKELY(m->ext == kUpb_ExtMode_IsMessageSet)) {
-          encode_msgset_item(e, ext);
-        } else {
-          encode_field(e, &ext->data, &ext->ext->sub, &ext->ext->field);
+      if (e->options & kUpb_EncodeOption_Deterministic) {
+        _upb_sortedmap sorted;
+        _upb_mapsorter_pushexts(&e->sorter, ext, ext_count, &sorted);
+        while (_upb_sortedmap_nextext(&e->sorter, &sorted, &ext)) {
+          encode_ext(e, ext, m->ext == kUpb_ExtMode_IsMessageSet);
+        }
+        _upb_mapsorter_popmap(&e->sorter, &sorted);
+      } else {
+        const upb_Message_Extension* end = ext + ext_count;
+        for (; ext != end; ext++) {
+          encode_ext(e, ext, m->ext == kUpb_ExtMode_IsMessageSet);
         }
       }
     }