From afffa9eaeb58efedbe695b6f985ea7c810871af2 Mon Sep 17 00:00:00 2001
From: Joshua Haberman <jhaberman@gmail.com>
Date: Thu, 10 Mar 2022 15:21:55 -0800
Subject: [PATCH] Two Ruby changes to unblock the release

---
 bazel/amalgamate.py  |  3 +++
 upb/def.c            |  7 ++++---
 upb/port_def.inc     |  8 ++++++++
 upb/port_undef.inc   |  1 +
 upb/table.c          | 10 +++++-----
 upb/table_internal.h |  2 ++
 6 files changed, 23 insertions(+), 8 deletions(-)

diff --git a/bazel/amalgamate.py b/bazel/amalgamate.py
index 103b11c351..70399085ca 100755
--- a/bazel/amalgamate.py
+++ b/bazel/amalgamate.py
@@ -44,6 +44,9 @@ class Amalgamator:
 
     self.output_c.write("/* Amalgamated source file */\n")
     self.output_c.write('#include "%supb.h"\n' % (prefix))
+    if prefix == "ruby-":
+      self.output_h.write("// Ruby is still using proto3 enum semantics for proto2\n")
+      self.output_h.write("#define UPB_DISABLE_PROTO2_ENUM_CHECKING\n")
     self.output_c.write(open("upb/port_def.inc").read())
 
     self.output_h.write("/* Amalgamated source file */\n")
diff --git a/upb/def.c b/upb/def.c
index 451aecdaf3..140465aab5 100644
--- a/upb/def.c
+++ b/upb/def.c
@@ -1414,6 +1414,7 @@ static uint8_t map_descriptortype(const upb_FieldDef* f) {
     return kUpb_FieldType_Bytes;
   } else if (type == kUpb_FieldType_Enum &&
              (f->sub.enumdef->file->syntax == kUpb_Syntax_Proto3 ||
+              UPB_TREAT_PROTO2_ENUMS_LIKE_PROTO3 ||
               // TODO(https://github.com/protocolbuffers/upb/issues/541):
               // fix map enum values to check for unknown enum values and put
               // them in the unknown field set.
@@ -1584,11 +1585,11 @@ static void make_layout(symtab_addctx* ctx, const upb_MessageDef* m) {
 
     fill_fieldlayout(field, f);
 
-    if (upb_FieldDef_IsSubMessage(f)) {
+    if (field->descriptortype == kUpb_FieldType_Message ||
+        field->descriptortype == kUpb_FieldType_Group) {
       field->submsg_index = sublayout_count++;
       subs[field->submsg_index].submsg = upb_FieldDef_MessageSubDef(f)->layout;
-    } else if (upb_FieldDef_CType(f) == kUpb_CType_Enum &&
-               f->sub.enumdef->file->syntax == kUpb_Syntax_Proto2) {
+    } else if (field->descriptortype == kUpb_FieldType_Enum) {
       field->submsg_index = sublayout_count++;
       subs[field->submsg_index].subenum = upb_FieldDef_EnumSubDef(f)->layout;
       UPB_ASSERT(subs[field->submsg_index].subenum);
diff --git a/upb/port_def.inc b/upb/port_def.inc
index 75d416d1ce..2b240ff6e8 100644
--- a/upb/port_def.inc
+++ b/upb/port_def.inc
@@ -251,3 +251,11 @@ void __asan_unpoison_memory_region(void const volatile *addr, size_t size);
 #define UPB_UNPOISON_MEMORY_REGION(addr, size) \
   ((void)(addr), (void)(size))
 #endif
+
+/* Disable proto2 arena behavior (TEMPORARY) **********************************/
+
+#ifdef UPB_DISABLE_PROTO2_ENUM_CHECKING
+#define UPB_TREAT_PROTO2_ENUMS_LIKE_PROTO3 1
+#else
+#define UPB_TREAT_PROTO2_ENUMS_LIKE_PROTO3 0
+#endif
diff --git a/upb/port_undef.inc b/upb/port_undef.inc
index 70956df25d..3ef2d0dbcd 100644
--- a/upb/port_undef.inc
+++ b/upb/port_undef.inc
@@ -59,3 +59,4 @@
 #undef UPB_POISON_MEMORY_REGION
 #undef UPB_UNPOISON_MEMORY_REGION
 #undef UPB_ASAN
+#undef UPB_TREAT_PROTO2_ENUMS_LIKE_PROTO3
diff --git a/upb/table.c b/upb/table.c
index 33b3a9dc6e..63fe3cc960 100644
--- a/upb/table.c
+++ b/upb/table.c
@@ -433,14 +433,14 @@ const uint64_t kWyhashSalt[5] = {
     0x082EFA98EC4E6C89ULL, 0x452821E638D01377ULL,
 };
 
-static uint32_t table_hash(const char* p, size_t n) {
+uint32_t _upb_Hash(const char* p, size_t n) {
   return Wyhash(p, n, 0, kWyhashSalt);
 }
 
 static uint32_t strhash(upb_tabkey key) {
   uint32_t len;
   char* str = upb_tabstr(key, &len);
-  return table_hash(str, len);
+  return _upb_Hash(str, len);
 }
 
 static bool streql(upb_tabkey k1, lookupkey_t k2) {
@@ -496,20 +496,20 @@ bool upb_strtable_insert(upb_strtable* t, const char* k, size_t len,
   tabkey = strcopy(key, a);
   if (tabkey == 0) return false;
 
-  hash = table_hash(key.str.str, key.str.len);
+  hash = _upb_Hash(key.str.str, key.str.len);
   insert(&t->t, key, tabkey, v, hash, &strhash, &streql);
   return true;
 }
 
 bool upb_strtable_lookup2(const upb_strtable* t, const char* key, size_t len,
                           upb_value* v) {
-  uint32_t hash = table_hash(key, len);
+  uint32_t hash = _upb_Hash(key, len);
   return lookup(&t->t, strkey2(key, len), v, hash, &streql);
 }
 
 bool upb_strtable_remove2(upb_strtable* t, const char* key, size_t len,
                           upb_value* val) {
-  uint32_t hash = table_hash(key, len);
+  uint32_t hash = _upb_Hash(key, len);
   upb_tabkey tabkey;
   return rm(&t->t, strkey2(key, len), val, &tabkey, hash, &streql);
 }
diff --git a/upb/table_internal.h b/upb/table_internal.h
index 9fb2d238cd..74076551de 100644
--- a/upb/table_internal.h
+++ b/upb/table_internal.h
@@ -374,6 +374,8 @@ void upb_inttable_iter_setdone(upb_inttable_iter* i);
 bool upb_inttable_iter_isequal(const upb_inttable_iter* i1,
                                const upb_inttable_iter* i2);
 
+uint32_t _upb_Hash(const char* p, size_t n);
+
 #ifdef __cplusplus
 } /* extern "C" */
 #endif