Merge pull request #17587 from thomasvl/patch_objc_to_28

CherryPick the newer work for Objective-C over the the 28 branch.

conformance/failure_list_objc.txt

@@ -1,7 +1,2 @@
 # JSON input or output tests are skipped (in conformance_objc.m) as mobile
 # platforms don't support JSON wire format to avoid code bloat.
-
-Required.Editions_Proto2.ProtobufInput.UnknownOrdering.ProtobufOutput
-Required.Editions_Proto3.ProtobufInput.UnknownOrdering.ProtobufOutput
-Required.Proto2.ProtobufInput.UnknownOrdering.ProtobufOutput
-Required.Proto3.ProtobufInput.UnknownOrdering.ProtobufOutput

objectivec/GPBCodedInputStream.m

@@ -13,6 +13,10 @@
 #import "GPBUtilities_PackagePrivate.h"
 #import "GPBWireFormat.h"
 
+// TODO: Consider using on other functions to reduce bloat when
+// some compiler optimizations are enabled.
+#define GPB_NOINLINE __attribute__((noinline))
+
 NSString *const GPBCodedInputStreamException = GPBNSStringifySymbol(GPBCodedInputStreamException);
 
 NSString *const GPBCodedInputStreamUnderlyingErrorKey =
@@ -28,7 +32,8 @@ NSString *const GPBCodedInputStreamErrorDomain =
 //  int CodedInputStream::default_recursion_limit_ = 100;
 static const NSUInteger kDefaultRecursionLimit = 100;
 
-static void RaiseException(NSInteger code, NSString *reason) {
+GPB_NOINLINE
+void GPBRaiseStreamError(NSInteger code, NSString *reason) {
   NSDictionary *errorInfo = nil;
   if ([reason length]) {
     errorInfo = @{GPBErrorReasonKey : reason};
@@ -44,7 +49,7 @@ static void RaiseException(NSInteger code, NSString *reason) {
 
 GPB_INLINE void CheckRecursionLimit(GPBCodedInputStreamState *state) {
   if (state->recursionDepth >= kDefaultRecursionLimit) {
-    RaiseException(GPBCodedInputStreamErrorRecursionDepthExceeded, nil);
+    GPBRaiseStreamError(GPBCodedInputStreamErrorRecursionDepthExceeded, nil);
   }
 }
 
@@ -56,19 +61,19 @@ GPB_INLINE void CheckFieldSize(uint64_t size) {
   if (size > 0x7fffffff) {
     // TODO: Maybe a different error code for this, but adding one is a breaking
     // change so reuse an existing one.
-    RaiseException(GPBCodedInputStreamErrorInvalidSize, nil);
+    GPBRaiseStreamError(GPBCodedInputStreamErrorInvalidSize, nil);
   }
 }
 
 static void CheckSize(GPBCodedInputStreamState *state, size_t size) {
   size_t newSize = state->bufferPos + size;
   if (newSize > state->bufferSize) {
-    RaiseException(GPBCodedInputStreamErrorInvalidSize, nil);
+    GPBRaiseStreamError(GPBCodedInputStreamErrorInvalidSize, nil);
   }
   if (newSize > state->currentLimit) {
     // Fast forward to end of currentLimit;
     state->bufferPos = state->currentLimit;
-    RaiseException(GPBCodedInputStreamErrorSubsectionLimitReached, nil);
+    GPBRaiseStreamError(GPBCodedInputStreamErrorSubsectionLimitReached, nil);
   }
 }
 
@@ -110,7 +115,7 @@ static int64_t ReadRawVarint64(GPBCodedInputStreamState *state) {
     }
     shift += 7;
   }
-  RaiseException(GPBCodedInputStreamErrorInvalidVarInt, @"Invalid VarInt64");
+  GPBRaiseStreamError(GPBCodedInputStreamErrorInvalidVarInt, @"Invalid VarInt64");
   return 0;
 }
 
@@ -201,11 +206,11 @@ int32_t GPBCodedInputStreamReadTag(GPBCodedInputStreamState *state) {
   state->lastTag = ReadRawVarint32(state);
   // Tags have to include a valid wireformat.
   if (!GPBWireFormatIsValidTag(state->lastTag)) {
-    RaiseException(GPBCodedInputStreamErrorInvalidTag, @"Invalid wireformat in tag.");
+    GPBRaiseStreamError(GPBCodedInputStreamErrorInvalidTag, @"Invalid wireformat in tag.");
   }
   // Zero is not a valid field number.
   if (GPBWireFormatGetTagFieldNumber(state->lastTag) == 0) {
-    RaiseException(GPBCodedInputStreamErrorInvalidTag,
+    GPBRaiseStreamError(GPBCodedInputStreamErrorInvalidTag,
                         @"A zero field number on the wire is invalid.");
   }
   return state->lastTag;
@@ -231,7 +236,7 @@ NSString *GPBCodedInputStreamReadRetainedString(GPBCodedInputStreamState *state)
       NSLog(@"UTF-8 failure, is some field type 'string' when it should be "
             @"'bytes'?");
 #endif
-      RaiseException(GPBCodedInputStreamErrorInvalidUTF8, nil);
+      GPBRaiseStreamError(GPBCodedInputStreamErrorInvalidUTF8, nil);
     }
   }
   return result;
@@ -262,11 +267,65 @@ NSData *GPBCodedInputStreamReadRetainedBytesNoCopy(GPBCodedInputStreamState *sta
   return result;
 }
 
+static void SkipToEndGroupInternal(GPBCodedInputStreamState *state, uint32_t endGroupTag) {
+  CheckRecursionLimit(state);
+  ++state->recursionDepth;
+  while (YES) {
+    uint32_t tag = GPBCodedInputStreamReadTag(state);
+    if (tag == endGroupTag || tag == 0) {
+      GPBCodedInputStreamCheckLastTagWas(state, endGroupTag);  // Will fail for end of input.
+      --state->recursionDepth;
+      return;
+    }
+    switch (GPBWireFormatGetTagWireType(tag)) {
+      case GPBWireFormatVarint:
+        (void)ReadRawVarint64(state);
+        break;
+      case GPBWireFormatFixed64:
+        SkipRawData(state, sizeof(uint64_t));
+        break;
+      case GPBWireFormatLengthDelimited: {
+        uint64_t size = ReadRawVarint64(state);
+        CheckFieldSize(size);
+        size_t size2 = (size_t)size;  // Cast safe on 32bit because of CheckFieldSize() above.
+        SkipRawData(state, size2);
+        break;
+      }
+      case GPBWireFormatStartGroup:
+        SkipToEndGroupInternal(state, GPBWireFormatMakeTag(GPBWireFormatGetTagFieldNumber(tag),
+                                                           GPBWireFormatEndGroup));
+        break;
+      case GPBWireFormatEndGroup:
+        GPBRaiseStreamError(GPBCodedInputStreamErrorInvalidTag, @"Unmatched end group");
+        break;
+      case GPBWireFormatFixed32:
+        SkipRawData(state, sizeof(uint32_t));
+        break;
+    }
+  }
+}
+
+// This doesn't include the start group, but it collects all bytes until the end group including
+// the end group tag.
+NSData *GPBCodedInputStreamReadRetainedBytesToEndGroupNoCopy(GPBCodedInputStreamState *state,
+                                                             int32_t fieldNumber) {
+  // Better have just read the start of the group.
+  GPBCodedInputStreamCheckLastTagWas(state,
+                                     GPBWireFormatMakeTag(fieldNumber, GPBWireFormatStartGroup));
+  const uint8_t *start = state->bytes + state->bufferPos;
+  SkipToEndGroupInternal(state, GPBWireFormatMakeTag(fieldNumber, GPBWireFormatEndGroup));
+  // This will be after the end group tag.
+  const uint8_t *end = state->bytes + state->bufferPos;
+  return [[NSData alloc] initWithBytesNoCopy:(void *)start
+                                      length:(NSUInteger)(end - start)
+                                freeWhenDone:NO];
+}
+
 size_t GPBCodedInputStreamPushLimit(GPBCodedInputStreamState *state, size_t byteLimit) {
   byteLimit += state->bufferPos;
   size_t oldLimit = state->currentLimit;
   if (byteLimit > oldLimit) {
-    RaiseException(GPBCodedInputStreamErrorInvalidSubsectionLimit, nil);
+    GPBRaiseStreamError(GPBCodedInputStreamErrorInvalidSubsectionLimit, nil);
   }
   state->currentLimit = byteLimit;
   return oldLimit;
@@ -286,7 +345,7 @@ BOOL GPBCodedInputStreamIsAtEnd(GPBCodedInputStreamState *state) {
 
 void GPBCodedInputStreamCheckLastTagWas(GPBCodedInputStreamState *state, int32_t value) {
   if (state->lastTag != value) {
-    RaiseException(GPBCodedInputStreamErrorInvalidTag, @"Unexpected tag read");
+    GPBRaiseStreamError(GPBCodedInputStreamErrorInvalidTag, @"Unexpected tag read");
   }
 }
 
@@ -429,9 +488,9 @@ void GPBCodedInputStreamCheckLastTagWas(GPBCodedInputStreamState *state, int32_t
     extensionRegistry:(id<GPBExtensionRegistry>)extensionRegistry {
   CheckRecursionLimit(&state_);
   ++state_.recursionDepth;
-  [message mergeFromCodedInputStream:self extensionRegistry:extensionRegistry];
-  GPBCodedInputStreamCheckLastTagWas(&state_,
-                                     GPBWireFormatMakeTag(fieldNumber, GPBWireFormatEndGroup));
+  [message mergeFromCodedInputStream:self
+                   extensionRegistry:extensionRegistry
+                           endingTag:GPBWireFormatMakeTag(fieldNumber, GPBWireFormatEndGroup)];
   --state_.recursionDepth;
 }
 
@@ -452,8 +511,7 @@ void GPBCodedInputStreamCheckLastTagWas(GPBCodedInputStreamState *state, int32_t
   size_t length2 = (size_t)length;  // Cast safe on 32bit because of CheckFieldSize() above.
   size_t oldLimit = GPBCodedInputStreamPushLimit(&state_, length2);
   ++state_.recursionDepth;
-  [message mergeFromCodedInputStream:self extensionRegistry:extensionRegistry];
-  GPBCodedInputStreamCheckLastTagWas(&state_, 0);
+  [message mergeFromCodedInputStream:self extensionRegistry:extensionRegistry endingTag:0];
   --state_.recursionDepth;
   GPBCodedInputStreamPopLimit(&state_, oldLimit);
 }

objectivec/GPBCodedInputStream_PackagePrivate.h

@@ -53,6 +53,7 @@ typedef struct GPBCodedInputStreamState {
 
 CF_EXTERN_C_BEGIN
 
+void GPBRaiseStreamError(NSInteger code, NSString *reason);
 int32_t GPBCodedInputStreamReadTag(GPBCodedInputStreamState *state);
 
 double GPBCodedInputStreamReadDouble(GPBCodedInputStreamState *state);
@@ -75,6 +76,9 @@ NSData *GPBCodedInputStreamReadRetainedBytes(GPBCodedInputStreamState *state)
     __attribute((ns_returns_retained));
 NSData *GPBCodedInputStreamReadRetainedBytesNoCopy(GPBCodedInputStreamState *state)
     __attribute((ns_returns_retained));
+NSData *GPBCodedInputStreamReadRetainedBytesToEndGroupNoCopy(GPBCodedInputStreamState *state,
+                                                             int32_t fieldNumber)
+    __attribute((ns_returns_retained));
 
 size_t GPBCodedInputStreamPushLimit(GPBCodedInputStreamState *state, size_t byteLimit);
 void GPBCodedInputStreamPopLimit(GPBCodedInputStreamState *state, size_t oldLimit);

objectivec/GPBDescriptor.m

@@ -267,6 +267,10 @@ static NSArray *NewFieldsArrayForHasIndex(int hasIndex, NSArray *allMessageField
                        fields:(NSArray *)fields
                   storageSize:(uint32_t)storageSize
                    wireFormat:(BOOL)wireFormat {
+#if defined(DEBUG) && DEBUG && !defined(NS_BLOCK_ASSERTIONS)
+  // This is also checked by the generator.
+  NSAssert(!wireFormat || fields.count == 0, @"Internal error: MessageSets should not have fields");
+#endif
   if ((self = [super init])) {
     messageClass_ = messageClass;
     messageName_ = [messageName copy];
@@ -1139,8 +1143,18 @@ uint32_t GPBFieldAlternateTag(GPBFieldDescriptor *self) {
     GPBRuntimeMatchFailure();
   }
 
-#if defined(DEBUG) && DEBUG
+#if defined(DEBUG) && DEBUG && !defined(NS_BLOCK_ASSERTIONS)
   NSAssert(usesClassRefs, @"Internal error: all extensions should have class refs");
+
+  // These are also checked by the generator.
+  if ((desc->options & GPBExtensionSetWireFormat) != 0) {
+    NSAssert(desc->dataType == GPBDataTypeMessage,
+             @"Internal error: If a MessageSet extension is set, the data type must be a message.");
+    NSAssert((desc->options & GPBExtensionRepeated) == 0,
+             @"Internal Error: MessageSet extension can't be repeated.");
+    // NOTE: Could also check that the exteneded class is a MessageSet, but that would force the
+    // ObjC runtime to start up that class and that isn't desirable here.
+  }
 #endif
 
   if ((self = [super init])) {

objectivec/GPBMessage.m

@@ -5,9 +5,9 @@
 // license that can be found in the LICENSE file or at
 // https://developers.google.com/open-source/licenses/bsd
 
-#import <Foundation/Foundation.h>
-#import "GPBMessage_PackagePrivate.h"
+#import "GPBMessage.h"
 
+#import <Foundation/Foundation.h>
 #import <objc/message.h>
 #import <objc/runtime.h>
 #import <os/lock.h>
@@ -20,11 +20,18 @@
 #import "GPBDictionary_PackagePrivate.h"
 #import "GPBExtensionInternals.h"
 #import "GPBExtensionRegistry.h"
+#import "GPBMessage_PackagePrivate.h"
 #import "GPBRootObject_PackagePrivate.h"
+#import "GPBUnknownField.h"
+#import "GPBUnknownFieldSet.h"
 #import "GPBUnknownFieldSet_PackagePrivate.h"
 #import "GPBUnknownFields_PackagePrivate.h"
 #import "GPBUtilities_PackagePrivate.h"
 
+// TODO: Consider using on other functions to reduce bloat when
+// some compiler optimizations are enabled.
+#define GPB_NOINLINE __attribute__((noinline))
+
 // Returns a new instance that was automatically created by |autocreator| for
 // its field |field|.
 static GPBMessage *GPBCreateMessageWithAutocreator(Class msgClass, GPBMessage *autocreator,
@@ -64,7 +71,16 @@ NSString *const GPBMessageExceptionMessageTooLarge =
 
 @interface GPBMessage () {
  @package
+  // Only one of these two is ever set, GPBUnknownFieldSet is being deprecated and will
+  // eventually be removed, but because that api support mutation, once the property is
+  // fetch it must continue to be used so any mutations will be honored in future operations
+  // on the message.
+  // Read only operations that access these two/cause things to migration between them should
+  // be protected with an @synchronized(self) block (that way the code also doesn't have to
+  // worry about throws).
+  NSMutableData *unknownFieldData_;
   GPBUnknownFieldSet *unknownFields_;
+
   NSMutableDictionary *extensionMap_;
   // Readonly access to autocreatedExtensionMap_ is protected via readOnlyLock_.
   NSMutableDictionary *autocreatedExtensionMap_;
@@ -103,7 +119,6 @@ static id GetOrCreateMapIvarWithField(GPBMessage *self, GPBFieldDescriptor *fiel
 static id GetMapIvarWithField(GPBMessage *self, GPBFieldDescriptor *field);
 static NSMutableDictionary *CloneExtensionMap(NSDictionary *extensionMap, NSZone *zone)
     __attribute__((ns_returns_retained));
-static GPBUnknownFieldSet *GetOrMakeUnknownFields(GPBMessage *self);
 
 #if defined(DEBUG) && DEBUG
 static NSError *MessageError(NSInteger code, NSDictionary *userInfo) {
@@ -133,6 +148,178 @@ static NSError *ErrorFromException(NSException *exception) {
   return error;
 }
 
+// Helper to encode varints onto the mutable data, the max size need is 10 bytes.
+GPB_NOINLINE
+static uint8_t *EncodeVarintU64(uint64_t val, uint8_t *ptr) {
+  do {
+    uint8_t byte = val & 0x7fU;
+    val >>= 7;
+    if (val) byte |= 0x80U;
+    *(ptr++) = byte;
+  } while (val);
+  return ptr;
+}
+
+// Helper to encode varints onto the mutable data, the max size need is 5 bytes.
+GPB_NOINLINE
+static uint8_t *EncodeVarintU32(uint32_t val, uint8_t *ptr) {
+  do {
+    uint8_t byte = val & 0x7fU;
+    val >>= 7;
+    if (val) byte |= 0x80U;
+    *(ptr++) = byte;
+  } while (val);
+  return ptr;
+}
+
+// Helper to encode signed int32 values as varints onto the mutable data, the max size need is 10
+// bytes.
+GPB_NOINLINE
+static uint8_t *EncodeVarintS32(int32_t val, uint8_t *ptr) {
+  if (val >= 0) {
+    return EncodeVarintU32((uint32_t)val, ptr);
+  } else {
+    // Must sign-extend
+    int64_t extended = val;
+    return EncodeVarintU64((uint64_t)extended, ptr);
+  }
+}
+
+GPB_NOINLINE
+static void AddUnknownFieldVarint32(GPBMessage *self, uint32_t fieldNumber, int32_t value) {
+  if (self->unknownFields_) {
+    [self->unknownFields_ mergeVarintField:fieldNumber value:value];
+    return;
+  }
+  uint8_t buf[20];
+  uint8_t *ptr = buf;
+  ptr = EncodeVarintU32(GPBWireFormatMakeTag(fieldNumber, GPBWireFormatVarint), ptr);
+  ptr = EncodeVarintS32(value, ptr);
+
+  if (self->unknownFieldData_ == nil) {
+    self->unknownFieldData_ = [[NSMutableData alloc] initWithCapacity:ptr - buf];
+    GPBBecomeVisibleToAutocreator(self);
+  }
+  [self->unknownFieldData_ appendBytes:buf length:ptr - buf];
+}
+
+GPB_NOINLINE
+static void AddUnknownFieldLengthDelimited(GPBMessage *self, uint32_t fieldNumber, NSData *value) {
+  if (self->unknownFields_) {
+    [self->unknownFields_ mergeLengthDelimited:fieldNumber value:value];
+    return;
+  }
+  uint8_t buf[20];
+  uint8_t *ptr = buf;
+  ptr = EncodeVarintU32(GPBWireFormatMakeTag(fieldNumber, GPBWireFormatLengthDelimited), ptr);
+  ptr = EncodeVarintU64((uint64_t)value.length, ptr);
+
+  if (self->unknownFieldData_ == nil) {
+    self->unknownFieldData_ = [[NSMutableData alloc] initWithCapacity:(ptr - buf) + value.length];
+    GPBBecomeVisibleToAutocreator(self);
+  }
+  [self->unknownFieldData_ appendBytes:buf length:ptr - buf];
+  [self->unknownFieldData_ appendData:value];
+}
+
+GPB_NOINLINE
+static void AddUnknownMessageSetEntry(GPBMessage *self, uint32_t typeId, NSData *value) {
+  if (self->unknownFields_) {
+    // Legacy Set does this odd storage for MessageSet.
+    [self->unknownFields_ mergeLengthDelimited:typeId value:value];
+    return;
+  }
+
+  uint8_t buf[60];
+  uint8_t *ptr = buf;
+  ptr = EncodeVarintU32(GPBWireFormatMessageSetItemTag, ptr);
+  ptr = EncodeVarintU32(GPBWireFormatMessageSetTypeIdTag, ptr);
+  ptr = EncodeVarintU32(typeId, ptr);
+  ptr = EncodeVarintU32(GPBWireFormatMessageSetMessageTag, ptr);
+  ptr = EncodeVarintU64((uint64_t)value.length, ptr);
+  uint8_t *split = ptr;
+
+  ptr = EncodeVarintU32(GPBWireFormatMessageSetItemEndTag, ptr);
+  uint8_t *end = ptr;
+
+  if (self->unknownFieldData_ == nil) {
+    self->unknownFieldData_ = [[NSMutableData alloc] initWithCapacity:(end - buf) + value.length];
+    GPBBecomeVisibleToAutocreator(self);
+  }
+  [self->unknownFieldData_ appendBytes:buf length:split - buf];
+  [self->unknownFieldData_ appendData:value];
+  [self->unknownFieldData_ appendBytes:split length:end - split];
+}
+
+GPB_NOINLINE
+static void ParseUnknownField(GPBMessage *self, uint32_t tag, GPBCodedInputStream *input) {
+  if (self->unknownFields_) {
+    if (![self->unknownFields_ mergeFieldFrom:tag input:input]) {
+      GPBRaiseStreamError(GPBCodedInputStreamErrorInvalidTag, @"Unexpected end-group tag");
+    }
+    return;
+  }
+
+  uint8_t buf[20];
+  uint8_t *ptr = buf;
+  ptr = EncodeVarintU32(tag, ptr);  // All will need the tag
+  NSData *bytesToAppend = nil;
+
+  GPBCodedInputStreamState *state = &input->state_;
+
+  switch (GPBWireFormatGetTagWireType(tag)) {
+    case GPBWireFormatVarint: {
+      ptr = EncodeVarintU64(GPBCodedInputStreamReadUInt64(state), ptr);
+      break;
+    }
+    case GPBWireFormatFixed64: {
+      uint64_t value = GPBCodedInputStreamReadFixed64(state);
+      *(ptr++) = (uint8_t)(value) & 0xFF;
+      *(ptr++) = (uint8_t)(value >> 8) & 0xFF;
+      *(ptr++) = (uint8_t)(value >> 16) & 0xFF;
+      *(ptr++) = (uint8_t)(value >> 24) & 0xFF;
+      *(ptr++) = (uint8_t)(value >> 32) & 0xFF;
+      *(ptr++) = (uint8_t)(value >> 40) & 0xFF;
+      *(ptr++) = (uint8_t)(value >> 48) & 0xFF;
+      *(ptr++) = (uint8_t)(value >> 56) & 0xFF;
+      break;
+    }
+    case GPBWireFormatLengthDelimited: {
+      bytesToAppend = GPBCodedInputStreamReadRetainedBytes(state);
+      ptr = EncodeVarintU64((uint64_t)bytesToAppend.length, ptr);
+      break;
+    }
+    case GPBWireFormatStartGroup: {
+      bytesToAppend = GPBCodedInputStreamReadRetainedBytesToEndGroupNoCopy(
+          state, GPBWireFormatGetTagFieldNumber(tag));
+      break;
+    }
+    case GPBWireFormatEndGroup:
+      GPBRaiseStreamError(GPBCodedInputStreamErrorInvalidTag, @"Unexpected end-group tag");
+      break;
+    case GPBWireFormatFixed32: {
+      uint32_t value = GPBCodedInputStreamReadFixed32(state);
+      *(ptr++) = (uint8_t)(value) & 0xFF;
+      *(ptr++) = (uint8_t)(value >> 8) & 0xFF;
+      *(ptr++) = (uint8_t)(value >> 16) & 0xFF;
+      *(ptr++) = (uint8_t)(value >> 24) & 0xFF;
+      break;
+    }
+  }
+
+  if (self->unknownFieldData_ == nil) {
+    self->unknownFieldData_ =
+        [[NSMutableData alloc] initWithCapacity:(ptr - buf) + bytesToAppend.length];
+    GPBBecomeVisibleToAutocreator(self);
+  }
+
+  [self->unknownFieldData_ appendBytes:buf length:ptr - buf];
+  if (bytesToAppend) {
+    [self->unknownFieldData_ appendData:bytesToAppend];
+    [bytesToAppend release];
+  }
+}
+
 static void CheckExtension(GPBMessage *self, GPBExtensionDescriptor *extension) {
   if (![self isKindOfClass:extension.containingMessageClass]) {
     [NSException raise:NSInvalidArgumentException
@@ -729,8 +916,7 @@ static void DecodeSingleValueFromInputStream(GPBExtensionDescriptor *extension,
       if (!enumDescriptor.isClosed || enumDescriptor.enumVerifier(val)) {
         nsValue = [[NSNumber alloc] initWithInt:val];
       } else {
-        GPBUnknownFieldSet *unknownFields = GetOrMakeUnknownFields(messageToGetExtension);
-        [unknownFields mergeVarintField:extension->description_->fieldNumber value:val];
+        AddUnknownFieldVarint32(messageToGetExtension, extension->description_->fieldNumber, val);
         nsValue = nil;
       }
       break;
@@ -743,14 +929,12 @@ static void DecodeSingleValueFromInputStream(GPBExtensionDescriptor *extension,
             extensionRegistry:extensionRegistry];
       } else {
 // description->dataType == GPBDataTypeMessage
-        if (GPBExtensionIsWireFormat(description)) {
-          // For MessageSet fields the message length will have already been
-          // read.
-          [targetMessage mergeFromCodedInputStream:input extensionRegistry:extensionRegistry];
-        } else {
+#if defined(DEBUG) && DEBUG && !defined(NS_BLOCK_ASSERTIONS)
+        NSCAssert(!GPBExtensionIsWireFormat(description),
+                  @"Internal error: got a MessageSet extension when not expected.");
+#endif
         [input readMessage:targetMessage extensionRegistry:extensionRegistry];
       }
-      }
       // Nothing to add below since the caller provided the message (and added it).
       nsValue = nil;
       break;
@@ -922,12 +1106,50 @@ void GPBClearMessageAutocreator(GPBMessage *self) {
   self->autocreatorExtension_ = nil;
 }
 
-static GPBUnknownFieldSet *GetOrMakeUnknownFields(GPBMessage *self) {
-  if (!self->unknownFields_) {
-    self->unknownFields_ = [[GPBUnknownFieldSet alloc] init];
-    GPBBecomeVisibleToAutocreator(self);
+GPB_NOINLINE
+static void MergeUnknownFieldDataIntoFieldSet(GPBMessage *self, NSData *data,
+                                              GPBUnknownFieldSet *targetSet) {
+  GPBUnknownFieldSet *unknownFields = targetSet ? targetSet : self->unknownFields_;
+
+#if defined(DEBUG) && DEBUG
+  NSCAssert(unknownFields != nil, @"Internal error: unknown fields not initialized.");
+#endif
+
+  BOOL isMessageSet = self.descriptor.isWireFormat;
+  GPBUnknownFieldSet *decodeInto = isMessageSet ? [[GPBUnknownFieldSet alloc] init] : unknownFields;
+
+  GPBCodedInputStream *input = [[GPBCodedInputStream alloc] initWithData:data];
+  @try {
+    [decodeInto mergeFromCodedInputStream:input];
+  } @catch (NSException *exception) {
+#if defined(DEBUG) && DEBUG
+    NSLog(@"%@: Internal exception while parsing the unknown fields into a Set: %@", [self class],
+          exception);
+#endif
+  }
+  [input release];
+
+  if (isMessageSet) {
+    // Need to transform the groups back into how Message feeds the data into a MessageSet when
+    // doing a full MessageSet based decode.
+    GPBUnknownField *groupField = [decodeInto getField:GPBWireFormatMessageSetItem];
+    for (GPBUnknownFieldSet *group in groupField.groupList) {
+      GPBUnknownField *typeIdField = [group getField:GPBWireFormatMessageSetTypeId];
+      GPBUnknownField *messageField = [group getField:GPBWireFormatMessageSetMessage];
+      if (typeIdField.varintList.count != 1 || messageField.lengthDelimitedList.count != 1) {
+#if defined(DEBUG) && DEBUG
+        NSCAssert(NO, @"Internal error: MessageSet group missing typeId or message.");
+#endif
+        continue;
+      }
+      int32_t fieldNumber = (int32_t)[typeIdField.varintList valueAtIndex:0];
+      GPBUnknownField *messageSetField = [[GPBUnknownField alloc] initWithNumber:fieldNumber];
+      [messageSetField addLengthDelimited:messageField.lengthDelimitedList[0]];
+      [unknownFields addField:messageSetField];
+      [messageSetField release];
+    }
+    [decodeInto release];
   }
-  return self->unknownFields_;
 }
 
 @implementation GPBMessage
@@ -1027,7 +1249,7 @@ static GPBUnknownFieldSet *GetOrMakeUnknownFields(GPBMessage *self) {
                                    error:(NSError **)errorPtr {
   if ((self = [self init])) {
     @try {
-      [self mergeFromCodedInputStream:input extensionRegistry:extensionRegistry];
+      [self mergeFromCodedInputStream:input extensionRegistry:extensionRegistry endingTag:0];
       if (errorPtr) {
         *errorPtr = nil;
       }
@@ -1164,8 +1386,12 @@ static GPBUnknownFieldSet *GetOrMakeUnknownFields(GPBMessage *self) {
   GPBMessage *result = [[descriptor.messageClass allocWithZone:zone] init];
 
   [self copyFieldsInto:result zone:zone descriptor:descriptor];
-  // Make immutable copies of the extra bits.
+
+  @synchronized(self) {
     result->unknownFields_ = [unknownFields_ copyWithZone:zone];
+    result->unknownFieldData_ = [unknownFieldData_ mutableCopyWithZone:zone];
+  }
+
   result->extensionMap_ = CloneExtensionMap(extensionMap_, zone);
   return result;
 }
@@ -1239,6 +1465,8 @@ static GPBUnknownFieldSet *GetOrMakeUnknownFields(GPBMessage *self) {
 
   [extensionMap_ release];
   extensionMap_ = nil;
+  [unknownFieldData_ release];
+  unknownFieldData_ = nil;
   [unknownFields_ release];
   unknownFields_ = nil;
 
@@ -1442,12 +1670,21 @@ static GPBUnknownFieldSet *GetOrMakeUnknownFields(GPBMessage *self) {
                               sortedExtensions:sortedExtensions];
     }
   }
+  @synchronized(self) {
+    if (unknownFieldData_) {
+#if defined(DEBUG) && DEBUG
+      NSAssert(unknownFields_ == nil, @"Internal error both unknown states were set");
+#endif
+      [output writeRawData:unknownFieldData_];
+    } else {
       if (descriptor.isWireFormat) {
         [unknownFields_ writeAsMessageSetTo:output];
       } else {
         [unknownFields_ writeToCodedOutputStream:output];
       }
     }
+  }  // @synchronized(self)
+}
 
 - (void)writeDelimitedToOutputStream:(NSOutputStream *)output {
   GPBCodedOutputStream *codedOutput = [[GPBCodedOutputStream alloc] initWithOutputStream:output];
@@ -2078,8 +2315,7 @@ static GPBUnknownFieldSet *GetOrMakeUnknownFields(GPBMessage *self) {
 - (void)mergeFromData:(NSData *)data extensionRegistry:(id<GPBExtensionRegistry>)extensionRegistry {
   GPBCodedInputStream *input = [[GPBCodedInputStream alloc] initWithData:data];
   @try {
-    [self mergeFromCodedInputStream:input extensionRegistry:extensionRegistry];
-    [input checkLastTagWas:0];
+    [self mergeFromCodedInputStream:input extensionRegistry:extensionRegistry endingTag:0];
   } @finally {
     [input release];
   }
@@ -2088,9 +2324,10 @@ static GPBUnknownFieldSet *GetOrMakeUnknownFields(GPBMessage *self) {
 - (BOOL)mergeFromData:(NSData *)data
     extensionRegistry:(nullable id<GPBExtensionRegistry>)extensionRegistry
                 error:(NSError **)errorPtr {
+  GPBBecomeVisibleToAutocreator(self);
   GPBCodedInputStream *input = [[GPBCodedInputStream alloc] initWithData:data];
   @try {
-    [self mergeFromCodedInputStream:input extensionRegistry:extensionRegistry];
+    [self mergeFromCodedInputStream:input extensionRegistry:extensionRegistry endingTag:0];
     [input checkLastTagWas:0];
     if (errorPtr) {
       *errorPtr = nil;
@@ -2167,11 +2404,25 @@ static GPBUnknownFieldSet *GetOrMakeUnknownFields(GPBMessage *self) {
 #pragma mark - Unknown Field Support
 
 - (GPBUnknownFieldSet *)unknownFields {
+  @synchronized(self) {
+    if (unknownFieldData_) {
+#if defined(DEBUG) && DEBUG
+      NSAssert(unknownFields_ == nil, @"Internal error both unknown states were set");
+#endif
+      unknownFields_ = [[GPBUnknownFieldSet alloc] init];
+      MergeUnknownFieldDataIntoFieldSet(self, unknownFieldData_, nil);
+      [unknownFieldData_ release];
+      unknownFieldData_ = nil;
+    }
     return unknownFields_;
+  }  // @synchronized(self)
 }
 
 - (void)setUnknownFields:(GPBUnknownFieldSet *)unknownFields {
-  if (unknownFields != unknownFields_) {
+  if (unknownFields != unknownFields_ || unknownFieldData_ != nil) {
+    // Changing sets or clearing.
+    [unknownFieldData_ release];
+    unknownFieldData_ = nil;
     [unknownFields_ release];
     unknownFields_ = [unknownFields copy];
     GPBBecomeVisibleToAutocreator(self);
@@ -2182,89 +2433,90 @@ static GPBUnknownFieldSet *GetOrMakeUnknownFields(GPBMessage *self) {
       extensionRegistry:(id<GPBExtensionRegistry>)extensionRegistry {
   uint32_t typeId = 0;
   NSData *rawBytes = nil;
-  GPBExtensionDescriptor *extension = nil;
   GPBCodedInputStreamState *state = &input->state_;
+  BOOL gotType = NO;
+  BOOL gotBytes = NO;
   while (true) {
     uint32_t tag = GPBCodedInputStreamReadTag(state);
-    if (tag == 0) {
+    if (tag == GPBWireFormatMessageSetItemEndTag || tag == 0) {
       break;
     }
 
     if (tag == GPBWireFormatMessageSetTypeIdTag) {
-      typeId = GPBCodedInputStreamReadUInt32(state);
-      if (typeId != 0) {
-        extension = [extensionRegistry extensionForDescriptor:[self descriptor] fieldNumber:typeId];
+      uint32_t tmp = GPBCodedInputStreamReadUInt32(state);
+      // Spec says only use the first value.
+      if (!gotType) {
+        gotType = YES;
+        typeId = tmp;
       }
     } else if (tag == GPBWireFormatMessageSetMessageTag) {
+      if (gotBytes) {
+        // Skip over the payload instead of collecting it.
+        [input skipField:tag];
+      } else {
         rawBytes = [GPBCodedInputStreamReadRetainedBytesNoCopy(state) autorelease];
+        gotBytes = YES;
+      }
     } else {
+      // Don't capture unknowns within the message set impl group.
       if (![input skipField:tag]) {
         break;
       }
     }
   }
 
-  [input checkLastTagWas:GPBWireFormatMessageSetItemEndTag];
+  // If we get here because of end of input (tag zero) or the wrong end tag (within the skipField:),
+  // this will error.
+  GPBCodedInputStreamCheckLastTagWas(state, GPBWireFormatMessageSetItemEndTag);
+
+  if (!gotType || !gotBytes) {
+    // upb_Decoder_DecodeMessageSetItem does't keep this partial as an unknown field, it just drops
+    // it, so do the same thing.
+    return;
+  }
 
-  if (rawBytes != nil && typeId != 0) {
-    if (extension != nil) {
+  GPBExtensionDescriptor *extension = [extensionRegistry extensionForDescriptor:[self descriptor]
+                                                                    fieldNumber:typeId];
+  if (extension) {
+#if defined(DEBUG) && DEBUG && !defined(NS_BLOCK_ASSERTIONS)
+    NSAssert(extension.dataType == GPBDataTypeMessage,
+             @"Internal Error: MessageSet extension must be a message field.");
+    NSAssert(GPBExtensionIsWireFormat(extension->description_),
+             @"Internal Error: MessageSet extension must have message_set_wire_format set.");
+    NSAssert(!GPBExtensionIsRepeated(extension->description_),
+             @"Internal Error: MessageSet extension can't be repeated.");
+#endif
+    // Look up the existing one to merge to or create a new one.
+    GPBMessage *targetMessage = [self getExistingExtension:extension];
+    if (!targetMessage) {
+      GPBDescriptor *descriptor = [extension.msgClass descriptor];
+      targetMessage = [[descriptor.messageClass alloc] init];
+      [self setExtension:extension value:targetMessage];
+      [targetMessage release];
+    }
     GPBCodedInputStream *newInput = [[GPBCodedInputStream alloc] initWithData:rawBytes];
     @try {
-        ExtensionMergeFromInputStream(extension, extension.packable, newInput, extensionRegistry,
-                                      self);
+      [targetMessage mergeFromCodedInputStream:newInput
+                             extensionRegistry:extensionRegistry
+                                     endingTag:0];
     } @finally {
       [newInput release];
     }
   } else {
-      GPBUnknownFieldSet *unknownFields = GetOrMakeUnknownFields(self);
+    // The extension isn't in the registry, but it was well formed, so the whole group structure
+    // get preserved as an unknown field.
+
     // rawBytes was created via a NoCopy, so it can be reusing a
     // subrange of another NSData that might go out of scope as things
     // unwind, so a copy is needed to ensure what is saved in the
     // unknown fields stays valid.
     NSData *cloned = [NSData dataWithData:rawBytes];
-      [unknownFields mergeMessageSetMessage:typeId data:cloned];
-    }
-  }
-}
-
-- (BOOL)parseUnknownField:(GPBCodedInputStream *)input
-        extensionRegistry:(id<GPBExtensionRegistry>)extensionRegistry
-                      tag:(uint32_t)tag {
-  GPBWireFormat wireType = GPBWireFormatGetTagWireType(tag);
-  int32_t fieldNumber = GPBWireFormatGetTagFieldNumber(tag);
-
-  GPBDescriptor *descriptor = [self descriptor];
-  GPBExtensionDescriptor *extension = [extensionRegistry extensionForDescriptor:descriptor
-                                                                    fieldNumber:fieldNumber];
-  if (extension == nil) {
-    if (descriptor.wireFormat && GPBWireFormatMessageSetItemTag == tag) {
-      [self parseMessageSet:input extensionRegistry:extensionRegistry];
-      return YES;
-    }
-  } else {
-    if (extension.wireType == wireType) {
-      ExtensionMergeFromInputStream(extension, extension.packable, input, extensionRegistry, self);
-      return YES;
-    }
-    // Primitive, repeated types can be packed on unpacked on the wire, and are
-    // parsed either way.
-    if ([extension isRepeated] && !GPBDataTypeIsObject(extension->description_->dataType) &&
-        (extension.alternateWireType == wireType)) {
-      ExtensionMergeFromInputStream(extension, !extension.packable, input, extensionRegistry, self);
-      return YES;
-    }
-  }
-  if ([GPBUnknownFieldSet isFieldTag:tag]) {
-    GPBUnknownFieldSet *unknownFields = GetOrMakeUnknownFields(self);
-    return [unknownFields mergeFieldFrom:tag input:input];
-  } else {
-    return NO;
+    AddUnknownMessageSetEntry(self, typeId, cloned);
   }
 }
 
 - (void)addUnknownMapEntry:(int32_t)fieldNum value:(NSData *)data {
-  GPBUnknownFieldSet *unknownFields = GetOrMakeUnknownFields(self);
-  [unknownFields addUnknownMapEntry:fieldNum value:data];
+  AddUnknownFieldLengthDelimited(self, fieldNum, data);
 }
 
 #pragma mark - MergeFromCodedInputStream Support
@@ -2337,8 +2589,7 @@ static void MergeSingleFieldFromCodedInputStream(GPBMessage *self, GPBFieldDescr
       if (!GPBFieldIsClosedEnum(field) || [field isValidEnumValue:val]) {
         GPBSetInt32IvarWithFieldPrivate(self, field, val);
       } else {
-        GPBUnknownFieldSet *unknownFields = GetOrMakeUnknownFields(self);
-        [unknownFields mergeVarintField:GPBFieldNumber(field) value:val];
+        AddUnknownFieldVarint32(self, GPBFieldNumber(field), val);
       }
     }
   }  // switch
@@ -2387,8 +2638,7 @@ static void MergeRepeatedPackedFieldFromCodedInputStream(GPBMessage *self,
         if (!GPBFieldIsClosedEnum(field) || [field isValidEnumValue:val]) {
           [(GPBEnumArray *)genericArray addRawValue:val];
         } else {
-          GPBUnknownFieldSet *unknownFields = GetOrMakeUnknownFields(self);
-          [unknownFields mergeVarintField:GPBFieldNumber(field) value:val];
+          AddUnknownFieldVarint32(self, GPBFieldNumber(field), val);
         }
         break;
       }
@@ -2456,8 +2706,7 @@ static void MergeRepeatedNotPackedFieldFromCodedInputStream(
       if (!GPBFieldIsClosedEnum(field) || [field isValidEnumValue:val]) {
         [(GPBEnumArray *)genericArray addRawValue:val];
       } else {
-        GPBUnknownFieldSet *unknownFields = GetOrMakeUnknownFields(self);
-        [unknownFields mergeVarintField:GPBFieldNumber(field) value:val];
+        AddUnknownFieldVarint32(self, GPBFieldNumber(field), val);
       }
       break;
     }
@@ -2465,18 +2714,27 @@ static void MergeRepeatedNotPackedFieldFromCodedInputStream(
 }
 
 - (void)mergeFromCodedInputStream:(GPBCodedInputStream *)input
-                extensionRegistry:(id<GPBExtensionRegistry>)extensionRegistry {
+                extensionRegistry:(id<GPBExtensionRegistry>)extensionRegistry
+                        endingTag:(uint32_t)endingTag {
+#if defined(DEBUG) && DEBUG
+  NSAssert(endingTag == 0 || GPBWireFormatGetTagWireType(endingTag) == GPBWireFormatEndGroup,
+           @"endingTag should have been an endGroup tag");
+#endif  // DEBUG
   GPBDescriptor *descriptor = [self descriptor];
   GPBCodedInputStreamState *state = &input->state_;
   uint32_t tag = 0;
   NSUInteger startingIndex = 0;
   NSArray *fields = descriptor->fields_;
+  BOOL isMessageSetWireFormat = descriptor.isWireFormat;
   NSUInteger numFields = fields.count;
   while (YES) {
     BOOL merged = NO;
     tag = GPBCodedInputStreamReadTag(state);
-    if (tag == 0) {
-      break;  // Reached end.
+    if (tag == endingTag || tag == 0) {
+      // If we got to the end (tag zero), when we were expecting the end group, this will
+      // raise the error.
+      GPBCodedInputStreamCheckLastTagWas(state, endingTag);
+      return;
     }
     for (NSUInteger i = 0; i < numFields; ++i) {
       if (startingIndex >= numFields) startingIndex = 0;
@@ -2514,8 +2772,9 @@ static void MergeRepeatedNotPackedFieldFromCodedInputStream(
       }
     }  // for(i < numFields)
 
-    if (!merged && (tag != 0)) {
-      // Primitive, repeated types can be packed on unpacked on the wire, and
+    if (merged) continue;  // On to the next tag
+
+    // Primitive, repeated types can be packed or unpacked on the wire, and
     // are parsed either way.  The above loop covered tag in the preferred
     // for, so this need to check the alternate form.
     for (NSUInteger i = 0; i < numFields; ++i) {
@@ -2540,20 +2799,39 @@ static void MergeRepeatedNotPackedFieldFromCodedInputStream(
         startingIndex += 1;
       }
     }
-    }
 
-    if (!merged) {
-      if (tag == 0) {
-        // zero signals EOF / limit reached
-        return;
+    if (merged) continue;  // On to the next tag
+
+    if (isMessageSetWireFormat) {
+      if (GPBWireFormatMessageSetItemTag == tag) {
+        [self parseMessageSet:input extensionRegistry:extensionRegistry];
+        continue;  // On to the next tag
+      }
     } else {
-        if (![self parseUnknownField:input extensionRegistry:extensionRegistry tag:tag]) {
-          // it's an endgroup tag
-          return;
+      // ObjC Runtime currently doesn't track if a message supported extensions, so the check is
+      // always done.
+      GPBExtensionDescriptor *extension =
+          [extensionRegistry extensionForDescriptor:descriptor
+                                        fieldNumber:GPBWireFormatGetTagFieldNumber(tag)];
+      if (extension) {
+        GPBWireFormat wireType = GPBWireFormatGetTagWireType(tag);
+        if (extension.wireType == wireType) {
+          ExtensionMergeFromInputStream(extension, extension.packable, input, extensionRegistry,
+                                        self);
+          continue;  // On to the next tag
+        }
+        // Primitive, repeated types can be packed on unpacked on the wire, and are
+        // parsed either way.
+        if ([extension isRepeated] && !GPBDataTypeIsObject(extension->description_->dataType) &&
+            (extension.alternateWireType == wireType)) {
+          ExtensionMergeFromInputStream(extension, !extension.packable, input, extensionRegistry,
+                                        self);
+          continue;  // On to the next tag
+        }
       }
     }
-    }  // if(!merged)
 
+    ParseUnknownField(self, tag, input);
   }  // while(YES)
 }
 
@@ -2675,10 +2953,29 @@ static void MergeRepeatedNotPackedFieldFromCodedInputStream(
   }  // for(fields)
 
   // Unknown fields.
-  if (!unknownFields_) {
-    [self setUnknownFields:other.unknownFields];
+  if (unknownFields_) {
+#if defined(DEBUG) && DEBUG
+    NSAssert(unknownFieldData_ == nil, @"Internal error both unknown states were set");
+#endif
+    @synchronized(other) {
+      if (other->unknownFields_) {
+#if defined(DEBUG) && DEBUG
+        NSAssert(other->unknownFieldData_ == nil, @"Internal error both unknown states were set");
+#endif
+        [unknownFields_ mergeUnknownFields:other->unknownFields_];
+      } else if (other->unknownFieldData_) {
+        MergeUnknownFieldDataIntoFieldSet(self, other->unknownFieldData_, nil);
+      }
+    }  // @synchronized(other)
+  } else {
+    NSData *otherData = GPBMessageUnknownFieldsData(other);
+    if (otherData) {
+      if (unknownFieldData_) {
+        [unknownFieldData_ appendData:otherData];
       } else {
-    [unknownFields_ mergeUnknownFields:other.unknownFields];
+        unknownFieldData_ = [otherData mutableCopy];
+      }
+    }
   }
 
   // Extensions
@@ -2850,15 +3147,72 @@ static void MergeRepeatedNotPackedFieldFromCodedInputStream(
     }
   }
 
-  // nil and empty are equal
-  GPBUnknownFieldSet *otherUnknowns = otherMsg->unknownFields_;
-  if ([unknownFields_ countOfFields] != 0 || [otherUnknowns countOfFields] != 0) {
-    if (![unknownFields_ isEqual:otherUnknowns]) {
-      return NO;
+  // Mutation while another thread is doing read only access is invalid, so the only thing we
+  // need to guard against is concurrent r/o access, so we can grab the values (and retain them)
+  // so we have a version to compare against safely incase the second access causes the transform
+  // between internal states.
+  GPBUnknownFieldSet *selfUnknownFields;
+  NSData *selfUnknownFieldData;
+  @synchronized(self) {
+    selfUnknownFields = [unknownFields_ retain];
+    selfUnknownFieldData = [unknownFieldData_ retain];
+  }
+  GPBUnknownFieldSet *otherUnknownFields;
+  NSData *otherUnknownFieldData;
+  @synchronized(otherMsg) {
+    otherUnknownFields = [otherMsg->unknownFields_ retain];
+    otherUnknownFieldData = [otherMsg->unknownFieldData_ retain];
+  }
+#if defined(DEBUG) && DEBUG && !defined(NS_BLOCK_ASSERTIONS)
+  if (selfUnknownFields) {
+    NSAssert(selfUnknownFieldData == nil, @"Internal error both unknown states were set");
+  }
+  if (otherUnknownFields) {
+    NSAssert(otherUnknownFieldData == nil, @"Internal error both unknown states were set");
+  }
+#endif
+  // Since a developer can set the legacy unknownFieldSet, treat nil and empty as the same.
+  if (selfUnknownFields && selfUnknownFields.countOfFields == 0) {
+    [selfUnknownFields release];
+    selfUnknownFields = nil;
+  }
+  if (otherUnknownFields && otherUnknownFields.countOfFields == 0) {
+    [otherUnknownFields release];
+    otherUnknownFields = nil;
+  }
+
+  BOOL result = YES;
+
+  if (selfUnknownFieldData && otherUnknownFieldData) {
+    // Both had data, compare it.
+    result = [selfUnknownFieldData isEqual:otherUnknownFieldData];
+  } else if (selfUnknownFields && otherUnknownFields) {
+    // Both had fields set, compare them.
+    result = [selfUnknownFields isEqual:otherUnknownFields];
+  } else {
+    // At this point, we're done to one have a set/nothing, and the other having data/nothing.
+    GPBUnknownFieldSet *theSet = selfUnknownFields ? selfUnknownFields : otherUnknownFields;
+    NSData *theData = selfUnknownFieldData ? selfUnknownFieldData : otherUnknownFieldData;
+    if (theSet) {
+      if (theData) {
+        GPBUnknownFieldSet *tempSet = [[GPBUnknownFieldSet alloc] init];
+        MergeUnknownFieldDataIntoFieldSet(self, theData, tempSet);
+        result = [tempSet isEqual:theSet];
+        [tempSet release];
+      } else {
+        result = NO;
+      }
+    } else {
+      // It was a data/nothing and nothing, so they equal if the other didn't have data.
+      result = theData == nil;
     }
   }
 
-  return YES;
+  [selfUnknownFields release];
+  [selfUnknownFieldData release];
+  [otherUnknownFields release];
+  [otherUnknownFieldData release];
+  return result;
 }
 
 // It is very difficult to implement a generic hash for ProtoBuf messages that
@@ -3121,11 +3475,20 @@ static void MergeRepeatedNotPackedFieldFromCodedInputStream(
   }  // for(fields)
 
   // Add any unknown fields.
+  @synchronized(self) {
+    if (unknownFieldData_) {
+#if defined(DEBUG) && DEBUG
+      NSAssert(unknownFields_ == nil, @"Internal error both unknown states were set");
+#endif
+      result += [unknownFieldData_ length];
+    } else {
       if (descriptor.wireFormat) {
         result += [unknownFields_ serializedSizeAsMessageSet];
       } else {
         result += [unknownFields_ serializedSize];
       }
+    }
+  }  // @synchronized(self)
 
   // Add any extensions.
   for (GPBExtensionDescriptor *extension in extensionMap_) {
@@ -3554,8 +3917,7 @@ GPB_INLINE BOOL GPBIsCaseOfSelForOneOf(const char *selName, size_t selNameLength
     if (data.length) {
       GPBCodedInputStream *input = [[GPBCodedInputStream alloc] initWithData:data];
       @try {
-        [self mergeFromCodedInputStream:input extensionRegistry:nil];
-        [input checkLastTagWas:0];
+        [self mergeFromCodedInputStream:input extensionRegistry:nil endingTag:0];
       } @finally {
         [input release];
       }
@@ -3652,4 +4014,32 @@ id GPBGetObjectIvarWithField(GPBMessage *self, GPBFieldDescriptor *field) {
   return expected;
 }
 
+NSData *GPBMessageUnknownFieldsData(GPBMessage *self) {
+  NSData *result = nil;
+  @synchronized(self) {
+    GPBUnknownFieldSet *unknownFields = self->unknownFields_;
+    if (unknownFields) {
+#if defined(DEBUG) && DEBUG
+      NSCAssert(self->unknownFieldData_ == nil, @"Internal error both unknown states were set");
+#endif
+      if (self.descriptor.isWireFormat) {
+        NSMutableData *mutableData =
+            [NSMutableData dataWithLength:unknownFields.serializedSizeAsMessageSet];
+        GPBCodedOutputStream *output = [[GPBCodedOutputStream alloc] initWithData:mutableData];
+        [unknownFields writeAsMessageSetTo:output];
+        [output flush];
+        [output release];
+        result = mutableData;
+      } else {
+        result = [unknownFields data];
+      }
+    } else {
+      // Internally we can borrow it without a copy since this is immediately used by callers
+      // and multithreaded access with any mutation is not allow on messages.
+      result = self->unknownFieldData_;
+    }
+  }  // @synchronized(self)
+  return result;
+}
+
 #pragma clang diagnostic pop

objectivec/GPBMessage_PackagePrivate.h

@@ -47,15 +47,15 @@ typedef struct GPBMessage_Storage *GPBMessage_StoragePtr;
 // Parses a message of this type from the input and merges it with this
 // message.
 //
+// `endingTag` should be zero if expected to consume to the end of input, but if
+// the input is supposed to be a Group, it should be the endgroup tag to look for.
+//
 // Warning:  This does not verify that all required fields are present in
 // the input message.
-// Note:  The caller should call
-// -[CodedInputStream checkLastTagWas:] after calling this to
-// verify that the last tag seen was the appropriate end-group tag,
-// or zero for EOF.
 // NOTE: This will throw if there is an error while parsing.
 - (void)mergeFromCodedInputStream:(GPBCodedInputStream *)input
-                extensionRegistry:(id<GPBExtensionRegistry>)extensionRegistry;
+                extensionRegistry:(id<GPBExtensionRegistry>)extensionRegistry
+                        endingTag:(uint32_t)endingTag;
 
 - (void)addUnknownMapEntry:(int32_t)fieldNum value:(NSData *)data;
 
@@ -81,4 +81,7 @@ void GPBAutocreatedDictionaryModified(GPBMessage *self, id dictionary);
 // autocreated reference to this message.
 void GPBClearMessageAutocreator(GPBMessage *self);
 
+// The data (or null) from the unknown fields of a message;
+NSData *GPBMessageUnknownFieldsData(GPBMessage *self);
+
 CF_EXTERN_C_END

objectivec/GPBUnknownFieldSet.m

@@ -154,7 +154,10 @@ static void CopyWorker(__unused const void *key, const void *value, void *contex
 - (NSString *)description {
   NSMutableString *description =
       [NSMutableString stringWithFormat:@"<%@ %p>: TextFormat: {\n", [self class], self];
+#pragma clang diagnostic push
+#pragma clang diagnostic ignored "-Wdeprecated-declarations"
   NSString *textFormat = GPBTextFormatForUnknownFieldSet(self, @"  ");
+#pragma clang diagnostic pop
   [description appendString:textFormat];
   [description appendString:@"}"];
   return description;
@@ -212,10 +215,6 @@ static void GPBUnknownFieldSetSerializedSizeAsMessageSet(__unused const void *ke
   return data;
 }
 
-+ (BOOL)isFieldTag:(int32_t)tag {
-  return GPBWireFormatGetTagWireType(tag) != GPBWireFormatEndGroup;
-}
-
 - (void)addField:(GPBUnknownField *)field {
   int32_t number = [field number];
   checkNumber(number);
@@ -272,6 +271,11 @@ static void GPBUnknownFieldSetMergeUnknownFields(__unused const void *key, const
   [[self mutableFieldForNumber:number create:YES] addVarint:value];
 }
 
+- (void)mergeLengthDelimited:(int32_t)fieldNum value:(NSData *)value {
+  checkNumber(fieldNum);
+  [[self mutableFieldForNumber:fieldNum create:YES] addLengthDelimited:value];
+}
+
 - (BOOL)mergeFieldFrom:(int32_t)tag input:(GPBCodedInputStream *)input {
   NSAssert(GPBWireFormatIsValidTag(tag), @"Got passed an invalid tag");
   int32_t number = GPBWireFormatGetTagFieldNumber(tag);
@@ -314,15 +318,6 @@ static void GPBUnknownFieldSetMergeUnknownFields(__unused const void *key, const
   }
 }
 
-- (void)mergeMessageSetMessage:(int32_t)number data:(NSData *)messageData {
-  [[self mutableFieldForNumber:number create:YES] addLengthDelimited:messageData];
-}
-
-- (void)addUnknownMapEntry:(int32_t)fieldNum value:(NSData *)data {
-  GPBUnknownField *field = [self mutableFieldForNumber:fieldNum create:YES];
-  [field addLengthDelimited:data];
-}
-
 - (void)mergeFromCodedInputStream:(GPBCodedInputStream *)input {
   while (YES) {
     int32_t tag = GPBCodedInputStreamReadTag(&input->state_);

objectivec/GPBUnknownFieldSet_PackagePrivate.h

@@ -14,8 +14,6 @@
 
 @interface GPBUnknownFieldSet ()
 
-+ (BOOL)isFieldTag:(int32_t)tag;
-
 - (NSData *)data;
 
 - (size_t)serializedSize;
@@ -29,9 +27,7 @@
 - (void)mergeFromCodedInputStream:(GPBCodedInputStream *)input;
 
 - (void)mergeVarintField:(int32_t)number value:(int32_t)value;
+- (void)mergeLengthDelimited:(int32_t)number value:(NSData *)value;
 - (BOOL)mergeFieldFrom:(int32_t)tag input:(GPBCodedInputStream *)input;
-- (void)mergeMessageSetMessage:(int32_t)number data:(NSData *)messageData;
-
-- (void)addUnknownMapEntry:(int32_t)fieldNum value:(NSData *)data;
 
 @end

objectivec/GPBUnknownFields.m

@@ -12,7 +12,9 @@
 #import "GPBCodedInputStream_PackagePrivate.h"
 #import "GPBCodedOutputStream.h"
 #import "GPBCodedOutputStream_PackagePrivate.h"
+#import "GPBDescriptor.h"
 #import "GPBMessage.h"
+#import "GPBMessage_PackagePrivate.h"
 #import "GPBUnknownField.h"
 #import "GPBUnknownFieldSet_PackagePrivate.h"
 #import "GPBUnknownField_PackagePrivate.h"
@@ -195,12 +197,9 @@ static BOOL MergeFromInputStream(GPBUnknownFields *self, GPBCodedInputStream *in
   self = [super init];
   if (self) {
     fields_ = [[NSMutableArray alloc] init];
-    // TODO: b/349146447 - Move off the legacy class and directly to the data once Message is
-    // updated.
-    GPBUnknownFieldSet *legacyUnknownFields = [message unknownFields];
-    if (legacyUnknownFields) {
-      GPBCodedInputStream *input =
-          [[GPBCodedInputStream alloc] initWithData:[legacyUnknownFields data]];
+    NSData *data = GPBMessageUnknownFieldsData(message);
+    if (data) {
+      GPBCodedInputStream *input = [[GPBCodedInputStream alloc] initWithData:data];
       // Parse until the end of the data (tag will be zero).
       if (!MergeFromInputStream(self, input, 0)) {
         [input release];

objectivec/GPBUtilities.h

@@ -40,7 +40,9 @@ NSString *GPBTextFormatForMessage(GPBMessage *message, NSString *__nullable line
  * @return An NSString with the TextFormat of the unknown field set.
  **/
 NSString *GPBTextFormatForUnknownFieldSet(GPBUnknownFieldSet *__nullable unknownSet,
-                                          NSString *__nullable lineIndent);
+                                          NSString *__nullable lineIndent)
+    __attribute__((deprecated("Use GPBTextFormatForMessage will include the unknown fields, and "
+                              "GPBUnknownFieldSet it going away.")));
 
 /**
  * Checks if the given field number is set on a message.

objectivec/GPBUtilities.m

@@ -5,7 +5,7 @@
 // license that can be found in the LICENSE file or at
 // https://developers.google.com/open-source/licenses/bsd
 
-#import "GPBUtilities_PackagePrivate.h"
+#import "GPBUtilities.h"
 
 #import <objc/runtime.h>
 
@@ -15,6 +15,9 @@
 #import "GPBMessage_PackagePrivate.h"
 #import "GPBUnknownField.h"
 #import "GPBUnknownFieldSet.h"
+#import "GPBUnknownField_PackagePrivate.h"
+#import "GPBUnknownFields.h"
+#import "GPBUtilities_PackagePrivate.h"
 
 // Direct access is use for speed, to avoid even internally declaring things
 // read/write, etc. The warning is enabled in the project to ensure code calling
@@ -1929,6 +1932,78 @@ static void AppendTextFormatForMessageExtensionRange(GPBMessage *message, NSArra
   }  // for..in(activeExtensions)
 }
 
+static void AppendTextFormatForUnknownFields(GPBUnknownFields *ufs, NSMutableString *toStr,
+                                             NSString *lineIndent) {
+#if defined(DEBUG) && DEBUG
+  NSCAssert(!ufs.empty, @"Internal Error: No unknown fields to format.");
+#endif
+  // Extract the fields and sort them by field number. Use a stable sort and sort just by the field
+  // numbers, that way the output will still show the order the different types were added as well
+  // as maintaining the order within a give number/type pair (i.e. - repeated fields say in order).
+  NSMutableArray *sortedFields = [[NSMutableArray alloc] initWithCapacity:ufs.count];
+  for (GPBUnknownField *field in ufs) {
+    [sortedFields addObject:field];
+  }
+  [sortedFields
+      sortWithOptions:NSSortStable
+      usingComparator:^NSComparisonResult(GPBUnknownField *field1, GPBUnknownField *field2) {
+        int32_t fieldNumber1 = field1->number_;
+        int32_t fieldNumber2 = field2->number_;
+        if (fieldNumber1 < fieldNumber2) {
+          return NSOrderedAscending;
+        } else if (fieldNumber1 > fieldNumber2) {
+          return NSOrderedDescending;
+        } else {
+          return NSOrderedSame;
+        }
+      }];
+
+  NSString *subIndent = nil;
+
+  for (GPBUnknownField *field in sortedFields) {
+    int32_t fieldNumber = field->number_;
+    switch (field->type_) {
+      case GPBUnknownFieldTypeVarint:
+        [toStr appendFormat:@"%@%d: %llu\n", lineIndent, fieldNumber, field->storage_.intValue];
+        break;
+      case GPBUnknownFieldTypeFixed32:
+        [toStr appendFormat:@"%@%d: 0x%X\n", lineIndent, fieldNumber,
+                            (uint32_t)field->storage_.intValue];
+        break;
+      case GPBUnknownFieldTypeFixed64:
+        [toStr appendFormat:@"%@%d: 0x%llX\n", lineIndent, fieldNumber, field->storage_.intValue];
+        break;
+      case GPBUnknownFieldTypeLengthDelimited:
+        [toStr appendFormat:@"%@%d: ", lineIndent, fieldNumber];
+        AppendBufferAsString(field->storage_.lengthDelimited, toStr);
+        [toStr appendString:@"\n"];
+        break;
+      case GPBUnknownFieldTypeGroup: {
+        GPBUnknownFields *group = field->storage_.group;
+        if (group.empty) {
+          [toStr appendFormat:@"%@%d: {}\n", lineIndent, fieldNumber];
+        } else {
+          [toStr appendFormat:@"%@%d: {\n", lineIndent, fieldNumber];
+          if (subIndent == nil) {
+            subIndent = [lineIndent stringByAppendingString:@"  "];
+          }
+          AppendTextFormatForUnknownFields(group, toStr, subIndent);
+          [toStr appendFormat:@"%@}\n", lineIndent];
+        }
+      } break;
+      case GPBUnknownFieldTypeLegacy:
+#if defined(DEBUG) && DEBUG
+        NSCAssert(
+            NO,
+            @"Internal error: Shouldn't have gotten a legacy field type in the unknown fields.");
+#endif
+        break;
+    }
+  }
+  [subIndent release];
+  [sortedFields release];
+}
+
 static void AppendTextFormatForMessage(GPBMessage *message, NSMutableString *toStr,
                                        NSString *lineIndent) {
   GPBDescriptor *descriptor = [message descriptor];
@@ -1951,11 +2026,12 @@ static void AppendTextFormatForMessage(GPBMessage *message, NSMutableString *toS
     }
   }
 
-  NSString *unknownFieldsStr = GPBTextFormatForUnknownFieldSet(message.unknownFields, lineIndent);
-  if ([unknownFieldsStr length] > 0) {
+  GPBUnknownFields *ufs = [[GPBUnknownFields alloc] initFromMessage:message];
+  if (ufs.count > 0) {
     [toStr appendFormat:@"%@# --- Unknown fields ---\n", lineIndent];
-    [toStr appendString:unknownFieldsStr];
+    AppendTextFormatForUnknownFields(ufs, toStr, lineIndent);
   }
+  [ufs release];
 }
 
 NSString *GPBTextFormatForMessage(GPBMessage *message, NSString *lineIndent) {

objectivec/Tests/GPBCodedInputStreamTests.m

@@ -6,12 +6,16 @@
 // https://developers.google.com/open-source/licenses/bsd
 
 #import <Foundation/Foundation.h>
-#import "GPBTestUtilities.h"
 
 #import "GPBCodedInputStream.h"
+#import "GPBCodedInputStream_PackagePrivate.h"
 #import "GPBCodedOutputStream.h"
+#import "GPBTestUtilities.h"
+#import "GPBUnknownField.h"
 #import "GPBUnknownFieldSet_PackagePrivate.h"
+#import "GPBUnknownFields.h"
 #import "GPBUtilities_PackagePrivate.h"
+#import "GPBWireFormat.h"
 #import "objectivec/Tests/Unittest.pbobjc.h"
 
 @interface CodedInputStreamTests : GPBTestCase
@@ -266,20 +270,55 @@
   TestAllTypes* message = [self allSetRepeatedCount:kGPBDefaultRepeatCount];
   NSData* rawBytes = message.data;
 
-  // Create two parallel inputs.  Parse one as unknown fields while using
-  // skipField() to skip each field on the other.  Expect the same tags.
+  TestEmptyMessage* empty = [TestEmptyMessage parseFromData:rawBytes error:NULL];
+  XCTAssertNotNil(empty);
+  GPBUnknownFields* ufs = [[[GPBUnknownFields alloc] initFromMessage:empty] autorelease];
+  NSMutableArray<NSNumber*>* fieldNumbers = [NSMutableArray arrayWithCapacity:ufs.count];
+  for (GPBUnknownField* field in ufs) {
+    GPBWireFormat wireFormat;
+    switch (field.type) {
+      case GPBUnknownFieldTypeFixed32:
+        wireFormat = GPBWireFormatFixed32;
+        break;
+      case GPBUnknownFieldTypeFixed64:
+        wireFormat = GPBWireFormatFixed64;
+        break;
+      case GPBUnknownFieldTypeVarint:
+        wireFormat = GPBWireFormatVarint;
+        break;
+      case GPBUnknownFieldTypeLengthDelimited:
+        wireFormat = GPBWireFormatLengthDelimited;
+        break;
+      case GPBUnknownFieldTypeGroup:
+        wireFormat = GPBWireFormatStartGroup;
+        break;
+      case GPBUnknownFieldTypeLegacy:
+        XCTFail(@"Legacy field type not expected");
+        wireFormat = GPBWireFormatVarint;
+        break;
+    }
+    uint32_t tag = GPBWireFormatMakeTag(field.number, wireFormat);
+    [fieldNumbers addObject:@(tag)];
+  }
+
+  // Check the tags compared to what's in the UnknownFields to confirm the stream is
+  // skipping as expected (this covers the tags within a group also).
   GPBCodedInputStream* input1 = [GPBCodedInputStream streamWithData:rawBytes];
   GPBCodedInputStream* input2 = [GPBCodedInputStream streamWithData:rawBytes];
   GPBUnknownFieldSet* unknownFields = [[[GPBUnknownFieldSet alloc] init] autorelease];
 
+  NSUInteger idx = 0;
   while (YES) {
     int32_t tag = [input1 readTag];
     XCTAssertEqual(tag, [input2 readTag]);
     if (tag == 0) {
+      XCTAssertEqual(idx, fieldNumbers.count);
       break;
     }
+    XCTAssertEqual(tag, [fieldNumbers[idx] intValue]);
     [unknownFields mergeFieldFrom:tag input:input1];
     [input2 skipField:tag];
+    ++idx;
   }
 }
 
@@ -444,4 +483,102 @@
   }
 }
 
+- (void)assertReadByteToEndGroupFails:(NSData*)data {
+  GPBCodedInputStream* input = [GPBCodedInputStream streamWithData:data];
+  uint32_t tag = [input readTag];
+  XCTAssertThrows(GPBCodedInputStreamReadRetainedBytesToEndGroupNoCopy(
+      &input->state_, GPBWireFormatGetTagFieldNumber(tag)));
+}
+
+- (void)assertReadByteToEndGroup:(NSData*)data value:(NSData*)value {
+  GPBCodedInputStream* input = [GPBCodedInputStream streamWithData:data];
+  uint32_t tag = [input readTag];
+  NSData* readValue = GPBCodedInputStreamReadRetainedBytesToEndGroupNoCopy(
+      &input->state_, GPBWireFormatGetTagFieldNumber(tag));
+  XCTAssertNotNil(readValue);
+  XCTAssertEqualObjects(readValue, value);
+  [readValue release];
+}
+
+static NSData* DataForGroupsOfDepth(NSUInteger depth) {
+  NSMutableData* data = [NSMutableData dataWithCapacity:0];
+
+  uint32_t byte = 35;  // 35 = 0b100011 -> field 4/start group
+  for (NSUInteger i = 0; i < depth; ++i) {
+    [data appendBytes:&byte length:1];
+  }
+
+  byte = 8;  // 8 = 0b1000, -> field 1/varint
+  [data appendBytes:&byte length:1];
+  byte = 1;  // 1 -> varint value of 1
+  [data appendBytes:&byte length:1];
+
+  byte = 36;  // 36 = 0b100100 -> field 4/end group
+  for (NSUInteger i = 0; i < depth; ++i) {
+    [data appendBytes:&byte length:1];
+  }
+  return data;
+}
+
+- (void)testBytesToEndGroup {
+  // 35 = 0b100011 -> field 4/start group
+  // 36 = 0b100100 -> field 4/end group
+  // 43 = 0b101011 -> field 5/end group
+  // 44 = 0b101100 -> field 5/end group
+  // 8 = 0b1000, 1 -> field 1/varint, value of 1
+  // 21 = 0b10101, 0x78, 0x56, 0x34, 0x12 -> field 2/fixed32, value of 0x12345678
+  // 25 = 0b11001, 0xf0, 0xde, 0xbc, 0x9a, 0x78, 0x56, 0x34, 0x12 -> field 3/fixed64,
+  //                                                                 value of 0x123456789abcdef0LL
+  // 50 = 0b110010, 0x0 -> field 6/length delimited, length 0
+  // 50 = 0b110010, 0x1, 42 -> field 6/length delimited, length 1, byte 42
+  // 0 -> field 0 which is invalid/varint
+  // 15 = 0b1111 -> field 1, wire type 7 which is invalid
+
+  [self assertReadByteToEndGroup:bytes(35, 36) value:bytes(36)];              // empty group
+  [self assertReadByteToEndGroup:bytes(35, 8, 1, 36) value:bytes(8, 1, 36)];  // varint
+  [self assertReadByteToEndGroup:bytes(35, 21, 0x78, 0x56, 0x34, 0x12, 36)    // fixed32
+                           value:bytes(21, 0x78, 0x56, 0x34, 0x12, 36)];
+  [self assertReadByteToEndGroup:bytes(35, 25, 0xf0, 0xde, 0xbc, 0x9a, 0x78, 0x56, 0x34, 0x12,
+                                       36)  // fixed64
+                           value:bytes(25, 0xf0, 0xde, 0xbc, 0x9a, 0x78, 0x56, 0x34, 0x12, 36)];
+  [self assertReadByteToEndGroup:bytes(35, 50, 0, 36)
+                           value:bytes(50, 0, 36)];  // length delimited, length 0
+  [self assertReadByteToEndGroup:bytes(35, 50, 1, 42, 36)
+                           value:bytes(50, 1, 42, 36)];  // length delimited, length 1, byte 42
+  [self assertReadByteToEndGroup:bytes(35, 43, 44, 36) value:bytes(43, 44, 36)];  // Sub group
+  [self assertReadByteToEndGroup:bytes(35, 8, 1, 43, 8, 1, 44,
+                                       36)  // varint and sub group with varint
+                           value:bytes(8, 1, 43, 8, 1, 44, 36)];
+
+  [self assertReadByteToEndGroupFails:bytes(35, 0, 36)];                 // Invalid field number
+  [self assertReadByteToEndGroupFails:bytes(35, 15, 36)];                // Invalid wire type
+  [self assertReadByteToEndGroupFails:bytes(35, 21, 0x78, 0x56, 0x34)];  // truncated fixed32
+  [self assertReadByteToEndGroupFails:bytes(35, 25, 0xf0, 0xde, 0xbc, 0x9a, 0x78, 0x56,
+                                            0x34)];  // truncated fixed64
+
+  // Mising end group
+  [self assertReadByteToEndGroupFails:bytes(35)];
+  [self assertReadByteToEndGroupFails:bytes(35, 8, 1)];
+  [self assertReadByteToEndGroupFails:bytes(35, 43)];
+  [self assertReadByteToEndGroupFails:bytes(35, 43, 8, 1)];
+
+  // Wrong end group
+  [self assertReadByteToEndGroupFails:bytes(35, 44)];
+  [self assertReadByteToEndGroupFails:bytes(35, 8, 1, 44)];
+  [self assertReadByteToEndGroupFails:bytes(35, 43, 36)];
+  [self assertReadByteToEndGroupFails:bytes(35, 43, 8, 1, 36)];
+  [self assertReadByteToEndGroupFails:bytes(35, 43, 44, 44)];
+  [self assertReadByteToEndGroupFails:bytes(35, 43, 8, 1, 44, 44)];
+
+  // This is the same limit as within GPBCodedInputStream.
+  const NSUInteger kDefaultRecursionLimit = 100;
+  // That depth parses.
+  NSData* testData = DataForGroupsOfDepth(kDefaultRecursionLimit);
+  [self assertReadByteToEndGroup:testData
+                           value:[testData subdataWithRange:NSMakeRange(1, testData.length - 1)]];
+  // One more level deep fails.
+  testData = DataForGroupsOfDepth(kDefaultRecursionLimit + 1);
+  [self assertReadByteToEndGroupFails:testData];
+}
+
 @end

objectivec/Tests/GPBMessageTests+Serialization.m

@@ -419,6 +419,17 @@
 
   // All the values should be in unknown fields.
 
+  GPBUnknownFields *ufs = [[GPBUnknownFields alloc] initFromMessage:msg];
+  XCTAssertEqual(ufs.count, 3U);
+  uint64_t varint;
+  XCTAssertTrue([ufs getFirst:Message2_FieldNumber_OptionalEnum varint:&varint]);
+  XCTAssertEqual(varint, (uint64_t)Message3_Enum_Extra3);
+  XCTAssertTrue([ufs getFirst:Message2_FieldNumber_RepeatedEnumArray varint:&varint]);
+  XCTAssertEqual(varint, (uint64_t)Message3_Enum_Extra3);
+  XCTAssertTrue([ufs getFirst:Message2_FieldNumber_OneofEnum varint:&varint]);
+  XCTAssertEqual(varint, (uint64_t)Message3_Enum_Extra3);
+  [ufs release];
+
   GPBUnknownFieldSet *unknownFields = msg.unknownFields;
 
   XCTAssertEqual([unknownFields countOfFields], 3U);
@@ -1398,6 +1409,9 @@
   int32_t val = -1;
   XCTAssertTrue([msg1.knownMapField getEnum:&val forKey:0]);
   XCTAssertEqual(val, Proto2MapEnum_Proto2MapEnumFoo);
+  GPBUnknownFields *ufs = [[GPBUnknownFields alloc] initFromMessage:msg1];
+  XCTAssertEqual(ufs.count, 1U);
+  [ufs release];
   XCTAssertEqual(msg1.unknownFields.countOfFields, 1U);
 
   data = [msg1 data];
@@ -1410,6 +1424,9 @@
   XCTAssertEqual(msg2.unknownMapField.count, 1U);
   XCTAssertTrue([msg2.unknownMapField getEnum:&val forKey:0]);
   XCTAssertEqual(val, Proto2MapEnumPlusExtra_EProto2MapEnumExtra);
+  ufs = [[GPBUnknownFields alloc] initFromMessage:msg2];
+  XCTAssertTrue(ufs.empty);
+  [ufs release];
   XCTAssertEqual(msg2.unknownFields.countOfFields, 0U);
 
   XCTAssertEqualObjects(orig, msg2);

objectivec/Tests/GPBMessageTests.m

@@ -5,16 +5,18 @@
 // license that can be found in the LICENSE file or at
 // https://developers.google.com/open-source/licenses/bsd
 
-#import "GPBTestUtilities.h"
-
 #import <objc/runtime.h>
 
+#import "GPBArray.h"
 #import "GPBArray_PackagePrivate.h"
 #import "GPBDescriptor.h"
 #import "GPBDictionary_PackagePrivate.h"
 #import "GPBMessage_PackagePrivate.h"
+#import "GPBTestUtilities.h"
+#import "GPBUnknownField.h"
 #import "GPBUnknownFieldSet_PackagePrivate.h"
 #import "GPBUnknownField_PackagePrivate.h"
+#import "GPBUnknownFields.h"
 #import "objectivec/Tests/Unittest.pbobjc.h"
 #import "objectivec/Tests/UnittestImport.pbobjc.h"
 #import "objectivec/Tests/UnittestObjc.pbobjc.h"
@@ -501,6 +503,11 @@
 
   [message setUnknownFields:unknownFields];
 
+  GPBUnknownFields *ufs = [[[GPBUnknownFields alloc] init] autorelease];
+  [ufs addFieldNumber:1234 fixed32:1234];
+  [ufs addFieldNumber:2345 varint:54321];
+  [message mergeUnknownFields:ufs extensionRegistry:nil];
+
   NSString *description = [message description];
   XCTAssertGreaterThan([description length], 0U);
 
@@ -985,6 +992,19 @@
   XCTAssertFalse([message hasOptionalNestedMessage]);
   [message.optionalNestedMessage setUnknownFields:unknownFields];
   XCTAssertTrue([message hasOptionalNestedMessage]);
+
+  message.optionalNestedMessage = nil;
+  XCTAssertFalse([message hasOptionalNestedMessage]);
+  GPBUnknownFields *ufs = [[[GPBUnknownFields alloc] init] autorelease];
+  [ufs addFieldNumber:1 varint:1];
+  [message.optionalNestedMessage mergeUnknownFields:ufs extensionRegistry:nil];
+  XCTAssertTrue([message hasOptionalNestedMessage]);
+
+  message.optionalNestedMessage = nil;
+  XCTAssertFalse([message hasOptionalNestedMessage]);
+  [ufs clear];  // Also make sure merging zero length forces it to become visible.
+  [message.optionalNestedMessage mergeUnknownFields:ufs extensionRegistry:nil];
+  XCTAssertTrue([message hasOptionalNestedMessage]);
 }
 
 - (void)testSetAutocreatedSubmessageToSelf {
@@ -1481,6 +1501,19 @@
   XCTAssertFalse([msg hasExtension:[UnittestRoot repeatedNestedEnumExtension]]);
   XCTAssertFalse([msg hasExtension:[UnittestRoot repeatedForeignEnumExtension]]);
 
+  GPBUnknownFields *ufs = [[[GPBUnknownFields alloc] initFromMessage:msg] autorelease];
+  XCTAssertEqual(ufs.count, 3);
+  uint64_t varint;
+  XCTAssertTrue([ufs getFirst:[UnittestRoot optionalNestedEnumExtension].fieldNumber
+                       varint:&varint]);
+  XCTAssertEqual(varint, 10);
+  XCTAssertTrue([ufs getFirst:[UnittestRoot repeatedNestedEnumExtension].fieldNumber
+                       varint:&varint]);
+  XCTAssertEqual(varint, 11);
+  XCTAssertTrue([ufs getFirst:[UnittestRoot repeatedForeignEnumExtension].fieldNumber
+                       varint:&varint]);
+  XCTAssertEqual(varint, 12);
+
   GPBUnknownFieldSet *unknownFields = msg.unknownFields;
   GPBUnknownField *field =
       [unknownFields getField:[UnittestRoot optionalNestedEnumExtension].fieldNumber];
@@ -1523,6 +1556,18 @@
   expected = @[ @4, @6 ];
   XCTAssertEqualObjects([msg getExtension:[UnittestRoot repeatedForeignEnumExtension]], expected);
 
+  ufs = [[[GPBUnknownFields alloc] initFromMessage:msg] autorelease];
+  XCTAssertEqual(ufs.count, 3);
+  XCTAssertTrue([ufs getFirst:[UnittestRoot optionalNestedEnumExtension].fieldNumber
+                       varint:&varint]);
+  XCTAssertEqual(varint, 10);
+  XCTAssertTrue([ufs getFirst:[UnittestRoot repeatedNestedEnumExtension].fieldNumber
+                       varint:&varint]);
+  XCTAssertEqual(varint, 11);
+  XCTAssertTrue([ufs getFirst:[UnittestRoot repeatedForeignEnumExtension].fieldNumber
+                       varint:&varint]);
+  XCTAssertEqual(varint, 12);
+
   unknownFields = msg.unknownFields;
   field = [unknownFields getField:[UnittestRoot optionalNestedEnumExtension].fieldNumber];
   XCTAssertNotNil(field);
@@ -1840,6 +1885,9 @@
   [unknowns mergeVarintField:123 value:456];
   GPBMessage *message = [GPBMessage message];
   [message setUnknownFields:unknowns];
+  GPBUnknownFields *ufs = [[[GPBUnknownFields alloc] init] autorelease];
+  [ufs addFieldNumber:1234 varint:5678];
+  [message mergeUnknownFields:ufs extensionRegistry:nil];
   NSData *data = [message data];
   GPBMessage *message2 = [GPBMessage parseFromData:data extensionRegistry:nil error:NULL];
   XCTAssertEqualObjects(message, message2);
@@ -1850,12 +1898,19 @@
   [unknowns1 mergeVarintField:123 value:456];
   GPBMessage *message1 = [GPBMessage message];
   [message1 setUnknownFields:unknowns1];
+  GPBUnknownFields *ufs1 = [[[GPBUnknownFields alloc] init] autorelease];
+  [ufs1 addFieldNumber:1234 varint:5678];
+  [message1 mergeUnknownFields:ufs1 extensionRegistry:nil];
 
   GPBUnknownFieldSet *unknowns2 = [[[GPBUnknownFieldSet alloc] init] autorelease];
   [unknowns2 mergeVarintField:789 value:987];
   [unknowns2 mergeVarintField:654 value:321];
   GPBMessage *message2 = [GPBMessage message];
   [message2 setUnknownFields:unknowns2];
+  GPBUnknownFields *ufs2 = [[[GPBUnknownFields alloc] init] autorelease];
+  [ufs2 addFieldNumber:2345 fixed32:6789];
+  [ufs2 addFieldNumber:3456 fixed32:7890];
+  [message2 mergeUnknownFields:ufs2 extensionRegistry:nil];
 
   NSMutableData *delimitedData = [NSMutableData data];
   [delimitedData appendData:[message1 delimitedData]];
@@ -2022,6 +2077,54 @@
   XCTAssertEqual([msgPrime.mumbleArray valueAtIndex:4], EnumTestMsg_MyEnum_NegTwo);
 }
 
+- (void)testCloseEnumsValuesOutOfRange {
+  // The unknown values should all make it into the unknown fields.
+  EnumTestMsg *msg1 = [EnumTestMsg message];
+  msg1.bar = EnumTestMsg_MyEnum_NegTwo;
+  msg1.baz = EnumTestMsg_MyEnum_Two;
+  [msg1.mumbleArray addValue:EnumTestMsg_MyEnum_Two];
+  [msg1.mumbleArray addValue:EnumTestMsg_MyEnum_NegTwo];
+
+  NSData *data = [msg1 data];
+  XCTAssertNotNil(data);
+
+  EnumTestMsgPrime *msg2 = [EnumTestMsgPrime parseFromData:data error:NULL];
+  XCTAssertNotNil(msg2);
+  XCTAssertEqualObjects(data, [msg2 data]);
+  XCTAssertFalse(msg2.hasBar);
+  XCTAssertFalse(msg2.hasBaz);
+  XCTAssertEqual(msg2.mumbleArray_Count, 0U);
+
+  GPBUnknownFields *ufs = [[[GPBUnknownFields alloc] initFromMessage:msg2] autorelease];
+  XCTAssertEqual(ufs.count, 4U);
+  uint64_t varint;
+  XCTAssertTrue([ufs getFirst:EnumTestMsg_FieldNumber_Bar varint:&varint]);
+  XCTAssertEqual(varint, (uint64_t)EnumTestMsg_MyEnum_NegTwo);
+  XCTAssertTrue([ufs getFirst:EnumTestMsg_FieldNumber_Baz varint:&varint]);
+  XCTAssertEqual(varint, (uint64_t)EnumTestMsg_MyEnum_Two);
+  NSArray<GPBUnknownField *> *fields = [ufs fields:EnumTestMsg_FieldNumber_MumbleArray];
+  XCTAssertEqual(fields.count, 2U);
+  XCTAssertEqual(fields[0].varint, (uint64_t)EnumTestMsg_MyEnum_Two);
+  XCTAssertEqual(fields[1].varint, (uint64_t)EnumTestMsg_MyEnum_NegTwo);
+
+  GPBUnknownFieldSet *unknownFields = msg2.unknownFields;
+  XCTAssertNotNil(unknownFields);
+  XCTAssertEqual(unknownFields.countOfFields, 3U);
+  XCTAssertTrue([unknownFields hasField:EnumTestMsg_FieldNumber_Bar]);
+  XCTAssertTrue([unknownFields hasField:EnumTestMsg_FieldNumber_Baz]);
+  XCTAssertTrue([unknownFields hasField:EnumTestMsg_FieldNumber_MumbleArray]);
+  GPBUnknownField *field = [unknownFields getField:EnumTestMsg_FieldNumber_Bar];
+  XCTAssertEqual(field.varintList.count, 1U);
+  XCTAssertEqual([field.varintList valueAtIndex:0], (uint64_t)EnumTestMsg_MyEnum_NegTwo);
+  field = [unknownFields getField:EnumTestMsg_FieldNumber_Baz];
+  XCTAssertEqual(field.varintList.count, 1U);
+  XCTAssertEqual([field.varintList valueAtIndex:0], (uint64_t)EnumTestMsg_MyEnum_Two);
+  field = [unknownFields getField:EnumTestMsg_FieldNumber_MumbleArray];
+  XCTAssertEqual(field.varintList.count, 2U);
+  XCTAssertEqual([field.varintList valueAtIndex:0], (uint64_t)EnumTestMsg_MyEnum_Two);
+  XCTAssertEqual([field.varintList valueAtIndex:1], (uint64_t)EnumTestMsg_MyEnum_NegTwo);
+}
+
 - (void)testReservedWordNaming {
   // names.cc has some special handing to make sure that some "reserved" objc
   // names get renamed in a way so they don't conflict.

objectivec/Tests/GPBUnknownFieldSetTest.m

@@ -6,6 +6,7 @@
 // https://developers.google.com/open-source/licenses/bsd
 
 #import "GPBTestUtilities.h"
+#import "GPBUnknownFieldSet.h"
 
 #import "GPBUnknownFieldSet_PackagePrivate.h"
 #import "GPBUnknownField_PackagePrivate.h"
@@ -281,6 +282,10 @@
   TestEmptyMessage* destination2 = [TestEmptyMessage message];
   [destination2 mergeFrom:source3];
 
+  XCTAssertEqualObjects(destination1.unknownFields, destination2.unknownFields);
+  XCTAssertEqualObjects(destination1.unknownFields, source3.unknownFields);
+  XCTAssertEqualObjects(destination2.unknownFields, source3.unknownFields);
+
   XCTAssertEqualObjects(destination1.data, destination2.data);
   XCTAssertEqualObjects(destination1.data, source3.data);
   XCTAssertEqualObjects(destination2.data, source3.data);
@@ -376,6 +381,180 @@
   XCTAssertEqual(0x7FFFFFFFFFFFFFFFULL, [field2.varintList valueAtIndex:0]);
 }
 
+static NSData* DataForGroupsOfDepth(NSUInteger depth) {
+  NSMutableData* data = [NSMutableData dataWithCapacity:0];
+
+  uint32_t byte = 35;  // 35 = 0b100011 -> field 4/start group
+  for (NSUInteger i = 0; i < depth; ++i) {
+    [data appendBytes:&byte length:1];
+  }
+
+  byte = 8;  // 8 = 0b1000, -> field 1/varint
+  [data appendBytes:&byte length:1];
+  byte = 1;  // 1 -> varint value of 1
+  [data appendBytes:&byte length:1];
+
+  byte = 36;  // 36 = 0b100100 -> field 4/end group
+  for (NSUInteger i = 0; i < depth; ++i) {
+    [data appendBytes:&byte length:1];
+  }
+  return data;
+}
+
+- (void)testParsingNestingGroupData {
+  // 35 = 0b100011 -> field 4/start group
+  // 36 = 0b100100 -> field 4/end group
+  // 43 = 0b101011 -> field 5/end group
+  // 44 = 0b101100 -> field 5/end group
+  // 8 = 0b1000, 1 -> field 1/varint, value of 1
+  // 21 = 0b10101, 0x78, 0x56, 0x34, 0x12 -> field 2/fixed32, value of 0x12345678
+  // 25 = 0b11001, 0xf0, 0xde, 0xbc, 0x9a, 0x78, 0x56, 0x34, 0x12 -> field 3/fixed64,
+  //                                                                 value of 0x123456789abcdef0LL
+  // 50 = 0b110010, 0x0 -> field 6/length delimited, length 0
+  // 50 = 0b110010, 0x1, 42 -> field 6/length delimited, length 1, byte 42
+  // 0 -> field 0 which is invalid/varint
+  // 15 = 0b1111 -> field 1, wire type 7 which is invalid
+
+  TestEmptyMessage* m = [TestEmptyMessage parseFromData:DataFromBytes(35, 36)
+                                                  error:NULL];  // empty group
+  XCTAssertEqual(m.unknownFields.countOfFields, (NSUInteger)1);
+  GPBUnknownField* field = [m.unknownFields getField:4];
+  XCTAssertEqual(field.groupList.count, (NSUInteger)1);
+  GPBUnknownFieldSet* group = field.groupList[0];
+  XCTAssertEqual(group.countOfFields, (NSUInteger)0);
+
+  m = [TestEmptyMessage parseFromData:DataFromBytes(35, 8, 1, 36) error:NULL];  // varint
+  XCTAssertEqual(m.unknownFields.countOfFields, (NSUInteger)1);
+  field = [m.unknownFields getField:4];
+  XCTAssertEqual(field.groupList.count, (NSUInteger)1);
+  group = field.groupList[0];
+  field = [group getField:1];
+  XCTAssertEqual(field.varintList.count, (NSUInteger)1);
+  XCTAssertEqual([field.varintList valueAtIndex:0], 1);
+
+  m = [TestEmptyMessage parseFromData:DataFromBytes(35, 21, 0x78, 0x56, 0x34, 0x12, 36)
+                                error:NULL];  // fixed32
+  XCTAssertEqual(m.unknownFields.countOfFields, (NSUInteger)1);
+  field = [m.unknownFields getField:4];
+  XCTAssertEqual(field.groupList.count, (NSUInteger)1);
+  group = field.groupList[0];
+  field = [group getField:2];
+  XCTAssertEqual(field.fixed32List.count, (NSUInteger)1);
+  XCTAssertEqual([field.fixed32List valueAtIndex:0], 0x12345678);
+
+  m = [TestEmptyMessage
+      parseFromData:DataFromBytes(35, 25, 0xf0, 0xde, 0xbc, 0x9a, 0x78, 0x56, 0x34, 0x12,
+                                  36)
+              error:NULL];  // fixed64
+  XCTAssertEqual(m.unknownFields.countOfFields, (NSUInteger)1);
+  field = [m.unknownFields getField:4];
+  XCTAssertEqual(field.groupList.count, (NSUInteger)1);
+  group = field.groupList[0];
+  field = [group getField:3];
+  XCTAssertEqual(field.fixed64List.count, (NSUInteger)1);
+  XCTAssertEqual([field.fixed64List valueAtIndex:0], 0x123456789abcdef0LL);
+
+  m = [TestEmptyMessage parseFromData:DataFromBytes(35, 50, 0, 36)
+                                error:NULL];  // length delimited, length 0
+  XCTAssertEqual(m.unknownFields.countOfFields, (NSUInteger)1);
+  field = [m.unknownFields getField:4];
+  XCTAssertEqual(field.groupList.count, (NSUInteger)1);
+  group = field.groupList[0];
+  field = [group getField:6];
+  XCTAssertEqual(field.lengthDelimitedList.count, (NSUInteger)1);
+  XCTAssertEqualObjects(field.lengthDelimitedList[0], [NSData data]);
+
+  m = [TestEmptyMessage parseFromData:DataFromBytes(35, 50, 1, 42, 36)
+                                error:NULL];  // length delimited, length 1, byte 42
+  XCTAssertEqual(m.unknownFields.countOfFields, (NSUInteger)1);
+  field = [m.unknownFields getField:4];
+  XCTAssertEqual(field.groupList.count, (NSUInteger)1);
+  group = field.groupList[0];
+  field = [group getField:6];
+  XCTAssertEqual(field.lengthDelimitedList.count, (NSUInteger)1);
+  XCTAssertEqualObjects(field.lengthDelimitedList[0], DataFromBytes(42));
+
+  m = [TestEmptyMessage parseFromData:DataFromBytes(35, 43, 44, 36) error:NULL];  // Sub group
+  field = [m.unknownFields getField:4];
+  XCTAssertEqual(field.groupList.count, (NSUInteger)1);
+  group = field.groupList[0];
+  XCTAssertEqual(group.countOfFields, (NSUInteger)1);
+  field = [group getField:5];
+  XCTAssertEqual(field.groupList.count, (NSUInteger)1);
+  group = field.groupList[0];
+  XCTAssertEqual(group.countOfFields, (NSUInteger)0);
+
+  m = [TestEmptyMessage parseFromData:DataFromBytes(35, 8, 1, 43, 8, 2, 44, 36)
+                                error:NULL];  // varint and sub group with varint
+  XCTAssertEqual(m.unknownFields.countOfFields, (NSUInteger)1);
+  field = [m.unknownFields getField:4];
+  XCTAssertEqual(field.groupList.count, (NSUInteger)1);
+  group = field.groupList[0];
+  XCTAssertEqual(group.countOfFields, (NSUInteger)2);
+  field = [group getField:1];
+  XCTAssertEqual(field.varintList.count, (NSUInteger)1);
+  XCTAssertEqual([field.varintList valueAtIndex:0], 1);
+  field = [group getField:5];
+  XCTAssertEqual(field.groupList.count, (NSUInteger)1);
+  group = field.groupList[0];
+  field = [group getField:1];
+  XCTAssertEqual(field.varintList.count, (NSUInteger)1);
+  XCTAssertEqual([field.varintList valueAtIndex:0], 2);
+
+  XCTAssertNil([TestEmptyMessage parseFromData:DataFromBytes(35, 0, 36)
+                                         error:NULL]);  // Invalid field number
+  XCTAssertNil([TestEmptyMessage parseFromData:DataFromBytes(35, 15, 36)
+                                         error:NULL]);  // Invalid wire type
+  XCTAssertNil([TestEmptyMessage parseFromData:DataFromBytes(35, 21, 0x78, 0x56, 0x34)
+                                         error:NULL]);  // truncated fixed32
+  XCTAssertNil([TestEmptyMessage
+      parseFromData:DataFromBytes(35, 25, 0xf0, 0xde, 0xbc, 0x9a, 0x78, 0x56,
+                                  0x34)
+              error:NULL]);  // truncated fixed64
+
+  // Mising end group
+  XCTAssertNil([TestEmptyMessage parseFromData:DataFromBytes(35) error:NULL]);
+  XCTAssertNil([TestEmptyMessage parseFromData:DataFromBytes(35, 8, 1) error:NULL]);
+  XCTAssertNil([TestEmptyMessage parseFromData:DataFromBytes(35, 43) error:NULL]);
+  XCTAssertNil([TestEmptyMessage parseFromData:DataFromBytes(35, 43, 8, 1) error:NULL]);
+
+  // Wrong end group
+  XCTAssertNil([TestEmptyMessage parseFromData:DataFromBytes(35, 44) error:NULL]);
+  XCTAssertNil([TestEmptyMessage parseFromData:DataFromBytes(35, 8, 1, 44) error:NULL]);
+  XCTAssertNil([TestEmptyMessage parseFromData:DataFromBytes(35, 43, 36) error:NULL]);
+  XCTAssertNil([TestEmptyMessage parseFromData:DataFromBytes(35, 43, 8, 1, 36) error:NULL]);
+  XCTAssertNil([TestEmptyMessage parseFromData:DataFromBytes(35, 43, 44, 44) error:NULL]);
+  XCTAssertNil([TestEmptyMessage parseFromData:DataFromBytes(35, 43, 8, 1, 44, 44) error:NULL]);
+
+  // This is the same limit as within GPBCodedInputStream.
+  const NSUInteger kDefaultRecursionLimit = 100;
+  // That depth parses.
+  NSData* testData = DataForGroupsOfDepth(kDefaultRecursionLimit);
+  m = [TestEmptyMessage parseFromData:testData error:NULL];
+  XCTAssertEqual(m.unknownFields.countOfFields, (NSUInteger)1);
+  field = [m.unknownFields getField:4];
+  for (NSUInteger i = 0; i < kDefaultRecursionLimit; ++i) {
+    XCTAssertEqual(field.varintList.count, (NSUInteger)0);
+    XCTAssertEqual(field.fixed32List.count, (NSUInteger)0);
+    XCTAssertEqual(field.fixed64List.count, (NSUInteger)0);
+    XCTAssertEqual(field.lengthDelimitedList.count, (NSUInteger)0);
+    XCTAssertEqual(field.groupList.count, (NSUInteger)1);
+    group = field.groupList[0];
+    XCTAssertEqual(group.countOfFields, (NSUInteger)1);
+    field = [group getField:(i < (kDefaultRecursionLimit - 1) ? 4 : 1)];
+  }
+  // field is of the inner most group
+  XCTAssertEqual(field.varintList.count, (NSUInteger)1);
+  XCTAssertEqual([field.varintList valueAtIndex:0], (NSUInteger)1);
+  XCTAssertEqual(field.fixed32List.count, (NSUInteger)0);
+  XCTAssertEqual(field.fixed64List.count, (NSUInteger)0);
+  XCTAssertEqual(field.lengthDelimitedList.count, (NSUInteger)0);
+  XCTAssertEqual(field.groupList.count, (NSUInteger)0);
+  // One more level deep fails.
+  testData = DataForGroupsOfDepth(kDefaultRecursionLimit + 1);
+  XCTAssertNil([TestEmptyMessage parseFromData:testData error:NULL]);
+}
+
 #pragma mark - Field tests
 // Some tests directly on fields since the dictionary in FieldSet can gate
 // testing some of these.

objectivec/Tests/GPBUnknownFieldsTest.m

@@ -847,4 +847,163 @@
   XCTAssertEqual(varint, 0x7FFFFFFFFFFFFFFFL);
 }
 
+static NSData* DataForGroupsOfDepth(NSUInteger depth) {
+  NSMutableData* data = [NSMutableData dataWithCapacity:0];
+
+  uint32_t byte = 35;  // 35 = 0b100011 -> field 4/start group
+  for (NSUInteger i = 0; i < depth; ++i) {
+    [data appendBytes:&byte length:1];
+  }
+
+  byte = 8;  // 8 = 0b1000, -> field 1/varint
+  [data appendBytes:&byte length:1];
+  byte = 1;  // 1 -> varint value of 1
+  [data appendBytes:&byte length:1];
+
+  byte = 36;  // 36 = 0b100100 -> field 4/end group
+  for (NSUInteger i = 0; i < depth; ++i) {
+    [data appendBytes:&byte length:1];
+  }
+  return data;
+}
+
+- (void)testParsingNestingGroupData {
+  // 35 = 0b100011 -> field 4/start group
+  // 36 = 0b100100 -> field 4/end group
+  // 43 = 0b101011 -> field 5/end group
+  // 44 = 0b101100 -> field 5/end group
+  // 8 = 0b1000, 1 -> field 1/varint, value of 1
+  // 21 = 0b10101, 0x78, 0x56, 0x34, 0x12 -> field 2/fixed32, value of 0x12345678
+  // 25 = 0b11001, 0xf0, 0xde, 0xbc, 0x9a, 0x78, 0x56, 0x34, 0x12 -> field 3/fixed64,
+  //                                                                 value of 0x123456789abcdef0LL
+  // 50 = 0b110010, 0x0 -> field 6/length delimited, length 0
+  // 50 = 0b110010, 0x1, 42 -> field 6/length delimited, length 1, byte 42
+  // 0 -> field 0 which is invalid/varint
+  // 15 = 0b1111 -> field 1, wire type 7 which is invalid
+
+  TestEmptyMessage* m = [TestEmptyMessage parseFromData:DataFromBytes(35, 36)
+                                                  error:NULL];  // empty group
+  GPBUnknownFields* ufs = [[[GPBUnknownFields alloc] initFromMessage:m] autorelease];
+  XCTAssertEqual(ufs.count, (NSUInteger)1);
+  GPBUnknownFields* group = [ufs firstGroup:4];
+  XCTAssertTrue(group.empty);
+
+  m = [TestEmptyMessage parseFromData:DataFromBytes(35, 8, 1, 36) error:NULL];  // varint
+  ufs = [[[GPBUnknownFields alloc] initFromMessage:m] autorelease];
+  XCTAssertEqual(ufs.count, (NSUInteger)1);
+  group = [ufs firstGroup:4];
+  XCTAssertEqual(group.count, (NSUInteger)1);
+  uint64_t varint = 0;
+  XCTAssertTrue([group getFirst:1 varint:&varint]);
+  XCTAssertEqual(varint, 1);
+
+  m = [TestEmptyMessage parseFromData:DataFromBytes(35, 21, 0x78, 0x56, 0x34, 0x12, 36)
+                                error:NULL];  // fixed32
+  ufs = [[[GPBUnknownFields alloc] initFromMessage:m] autorelease];
+  XCTAssertEqual(ufs.count, (NSUInteger)1);
+  group = [ufs firstGroup:4];
+  XCTAssertEqual(group.count, (NSUInteger)1);
+  uint32_t fixed32 = 0;
+  XCTAssertTrue([group getFirst:2 fixed32:&fixed32]);
+  XCTAssertEqual(fixed32, 0x12345678);
+
+  m = [TestEmptyMessage
+      parseFromData:DataFromBytes(35, 25, 0xf0, 0xde, 0xbc, 0x9a, 0x78, 0x56, 0x34, 0x12,
+                                  36)
+              error:NULL];  // fixed64
+  ufs = [[[GPBUnknownFields alloc] initFromMessage:m] autorelease];
+  XCTAssertEqual(ufs.count, (NSUInteger)1);
+  group = [ufs firstGroup:4];
+  XCTAssertEqual(group.count, (NSUInteger)1);
+  uint64_t fixed64 = 0;
+  XCTAssertTrue([group getFirst:3 fixed64:&fixed64]);
+  XCTAssertEqual(fixed64, 0x123456789abcdef0LL);
+
+  m = [TestEmptyMessage parseFromData:DataFromBytes(35, 50, 0, 36)
+                                error:NULL];  // length delimited, length 0
+  ufs = [[[GPBUnknownFields alloc] initFromMessage:m] autorelease];
+  XCTAssertEqual(ufs.count, (NSUInteger)1);
+  group = [ufs firstGroup:4];
+  XCTAssertEqual(group.count, (NSUInteger)1);
+  NSData* lengthDelimited = [group firstLengthDelimited:6];
+  XCTAssertEqualObjects(lengthDelimited, [NSData data]);
+
+  m = [TestEmptyMessage parseFromData:DataFromBytes(35, 50, 1, 42, 36)
+                                error:NULL];  // length delimited, length 1, byte 42
+  ufs = [[[GPBUnknownFields alloc] initFromMessage:m] autorelease];
+  XCTAssertEqual(ufs.count, (NSUInteger)1);
+  group = [ufs firstGroup:4];
+  XCTAssertEqual(group.count, (NSUInteger)1);
+  lengthDelimited = [group firstLengthDelimited:6];
+  XCTAssertEqualObjects(lengthDelimited, DataFromBytes(42));
+
+  m = [TestEmptyMessage parseFromData:DataFromBytes(35, 43, 44, 36) error:NULL];  // Sub group
+  ufs = [[[GPBUnknownFields alloc] initFromMessage:m] autorelease];
+  XCTAssertEqual(ufs.count, (NSUInteger)1);
+  group = [ufs firstGroup:4];
+  XCTAssertEqual(group.count, (NSUInteger)1);
+  group = [group firstGroup:5];
+  XCTAssertTrue(group.empty);
+
+  m = [TestEmptyMessage parseFromData:DataFromBytes(35, 8, 1, 43, 8, 2, 44, 36)
+                                error:NULL];  // varint and sub group with varint
+  ufs = [[[GPBUnknownFields alloc] initFromMessage:m] autorelease];
+  XCTAssertEqual(ufs.count, (NSUInteger)1);
+  group = [ufs firstGroup:4];
+  XCTAssertEqual(group.count, (NSUInteger)2);
+  varint = 0;
+  XCTAssertTrue([group getFirst:1 varint:&varint]);
+  XCTAssertEqual(varint, 1);
+  group = [group firstGroup:5];
+  XCTAssertEqual(group.count, (NSUInteger)1);
+  XCTAssertTrue([group getFirst:1 varint:&varint]);
+  XCTAssertEqual(varint, 2);
+
+  XCTAssertNil([TestEmptyMessage parseFromData:DataFromBytes(35, 0, 36)
+                                         error:NULL]);  // Invalid field number
+  XCTAssertNil([TestEmptyMessage parseFromData:DataFromBytes(35, 15, 36)
+                                         error:NULL]);  // Invalid wire type
+  XCTAssertNil([TestEmptyMessage parseFromData:DataFromBytes(35, 21, 0x78, 0x56, 0x34)
+                                         error:NULL]);  // truncated fixed32
+  XCTAssertNil([TestEmptyMessage
+      parseFromData:DataFromBytes(35, 25, 0xf0, 0xde, 0xbc, 0x9a, 0x78, 0x56,
+                                  0x34)
+              error:NULL]);  // truncated fixed64
+
+  // Mising end group
+  XCTAssertNil([TestEmptyMessage parseFromData:DataFromBytes(35) error:NULL]);
+  XCTAssertNil([TestEmptyMessage parseFromData:DataFromBytes(35, 8, 1) error:NULL]);
+  XCTAssertNil([TestEmptyMessage parseFromData:DataFromBytes(35, 43) error:NULL]);
+  XCTAssertNil([TestEmptyMessage parseFromData:DataFromBytes(35, 43, 8, 1) error:NULL]);
+
+  // Wrong end group
+  XCTAssertNil([TestEmptyMessage parseFromData:DataFromBytes(35, 44) error:NULL]);
+  XCTAssertNil([TestEmptyMessage parseFromData:DataFromBytes(35, 8, 1, 44) error:NULL]);
+  XCTAssertNil([TestEmptyMessage parseFromData:DataFromBytes(35, 43, 36) error:NULL]);
+  XCTAssertNil([TestEmptyMessage parseFromData:DataFromBytes(35, 43, 8, 1, 36) error:NULL]);
+  XCTAssertNil([TestEmptyMessage parseFromData:DataFromBytes(35, 43, 44, 44) error:NULL]);
+  XCTAssertNil([TestEmptyMessage parseFromData:DataFromBytes(35, 43, 8, 1, 44, 44) error:NULL]);
+
+  // This is the same limit as within GPBCodedInputStream.
+  const NSUInteger kDefaultRecursionLimit = 100;
+  // That depth parses.
+  NSData* testData = DataForGroupsOfDepth(kDefaultRecursionLimit);
+  m = [TestEmptyMessage parseFromData:testData error:NULL];
+  ufs = [[[GPBUnknownFields alloc] initFromMessage:m] autorelease];
+  XCTAssertEqual(ufs.count, (NSUInteger)1);
+  group = [ufs firstGroup:4];
+  for (NSUInteger i = 1; i < kDefaultRecursionLimit; ++i) {
+    XCTAssertEqual(group.count, (NSUInteger)1);
+    group = [group firstGroup:4];
+  }
+  // group is now the inner most group.
+  XCTAssertEqual(group.count, (NSUInteger)1);
+  varint = 0;
+  XCTAssertTrue([group getFirst:1 varint:&varint]);
+  XCTAssertEqual(varint, 1);
+  // One more level deep fails.
+  testData = DataForGroupsOfDepth(kDefaultRecursionLimit + 1);
+  XCTAssertNil([TestEmptyMessage parseFromData:testData error:NULL]);
+}
+
 @end

objectivec/Tests/GPBUtilitiesTests.m

@@ -6,17 +6,16 @@
 // https://developers.google.com/open-source/licenses/bsd
 
 #import <XCTest/XCTest.h>
-
-#import "GPBUtilities_PackagePrivate.h"
-
 #import <objc/runtime.h>
 
-#import "GPBTestUtilities.h"
-
 #import "GPBDescriptor.h"
 #import "GPBDescriptor_PackagePrivate.h"
 #import "GPBMessage.h"
+#import "GPBTestUtilities.h"
+#import "GPBUnknownField.h"
 #import "GPBUnknownField_PackagePrivate.h"
+#import "GPBUtilities.h"
+#import "GPBUtilities_PackagePrivate.h"
 
 #import "objectivec/Tests/MapUnittest.pbobjc.h"
 #import "objectivec/Tests/Unittest.pbobjc.h"
@@ -168,6 +167,54 @@
   [expected release];
 }
 
+- (void)testTextFormatUnknownFields {
+  GPBUnknownFields *ufs = [[[GPBUnknownFields alloc] init] autorelease];
+  [ufs addFieldNumber:100 varint:5];
+  [ufs addFieldNumber:100 varint:4];
+  [ufs addFieldNumber:10 varint:1];
+  [ufs addFieldNumber:300 fixed32:0x50];
+  [ufs addFieldNumber:300 fixed32:0x40];
+  [ufs addFieldNumber:10 fixed32:0x10];
+  [ufs addFieldNumber:200 fixed64:0x5000];
+  [ufs addFieldNumber:200 fixed64:0x4000];
+  [ufs addFieldNumber:10 fixed64:0x1000];
+  [ufs addFieldNumber:10 lengthDelimited:DataFromCStr("foo")];
+  [ufs addFieldNumber:10 lengthDelimited:DataFromCStr("bar")];
+  GPBUnknownFields *group = [ufs addGroupWithFieldNumber:150];
+  [group addFieldNumber:2 varint:2];
+  [group addFieldNumber:1 varint:1];
+  group = [ufs addGroupWithFieldNumber:150];
+  [group addFieldNumber:1 varint:1];
+  [group addFieldNumber:3 fixed32:0x3];
+  [group addFieldNumber:2 fixed64:0x2];
+  TestEmptyMessage *message = [TestEmptyMessage message];
+  [message mergeUnknownFields:ufs extensionRegistry:nil];
+
+  NSString *expected = @"# --- Unknown fields ---\n"
+                       @"10: 1\n"
+                       @"10: 0x10\n"
+                       @"10: 0x1000\n"
+                       @"10: \"foo\"\n"
+                       @"10: \"bar\"\n"
+                       @"100: 5\n"
+                       @"100: 4\n"
+                       @"150: {\n"
+                       @"  1: 1\n"
+                       @"  2: 2\n"
+                       @"}\n"
+                       @"150: {\n"
+                       @"  1: 1\n"
+                       @"  2: 0x2\n"
+                       @"  3: 0x3\n"
+                       @"}\n"
+                       @"200: 0x5000\n"
+                       @"200: 0x4000\n"
+                       @"300: 0x50\n"
+                       @"300: 0x40\n";
+  NSString *result = GPBTextFormatForMessage(message, nil);
+  XCTAssertEqualObjects(expected, result);
+}
+
 - (void)testSetRepeatedFields {
   TestAllTypes *message = [TestAllTypes message];
 
@@ -204,11 +251,27 @@ static GPBUnknownFieldSet *UnknownFieldsSetHelper(int num) {
   return result;
 }
 
+// Helper to add an unknown field data to messages.
+static void AddUnknownFields(GPBMessage *message, int num) {
+  GPBUnknownFields *ufs = [[GPBUnknownFields alloc] init];
+  [ufs addFieldNumber:num varint:num];
+  [message mergeUnknownFields:ufs extensionRegistry:nil];
+  [ufs release];
+}
+
+static BOOL HasUnknownFields(GPBMessage *message) {
+  GPBUnknownFields *ufs = [[GPBUnknownFields alloc] initFromMessage:message];
+  BOOL result = !ufs.empty;
+  [ufs release];
+  return result;
+}
+
 - (void)testDropMessageUnknownFieldsRecursively {
   TestAllExtensions *message = [TestAllExtensions message];
 
   // Give it unknownFields.
   message.unknownFields = UnknownFieldsSetHelper(777);
+  AddUnknownFields(message, 1777);
 
   // Given it extensions that include a message with unknown fields of its own.
   {
@@ -219,18 +282,21 @@ static GPBUnknownFieldSet *UnknownFieldsSetHelper(int num) {
     OptionalGroup_extension *optionalGroup = [OptionalGroup_extension message];
     optionalGroup.a = 123;
     optionalGroup.unknownFields = UnknownFieldsSetHelper(779);
+    AddUnknownFields(optionalGroup, 1779);
     [message setExtension:[UnittestRoot optionalGroupExtension] value:optionalGroup];
 
     // Message
     TestAllTypes_NestedMessage *nestedMessage = [TestAllTypes_NestedMessage message];
     nestedMessage.bb = 456;
     nestedMessage.unknownFields = UnknownFieldsSetHelper(778);
+    AddUnknownFields(nestedMessage, 1778);
     [message setExtension:[UnittestRoot optionalNestedMessageExtension] value:nestedMessage];
 
     // Repeated Group
     RepeatedGroup_extension *repeatedGroup = [[RepeatedGroup_extension alloc] init];
     repeatedGroup.a = 567;
     repeatedGroup.unknownFields = UnknownFieldsSetHelper(780);
+    AddUnknownFields(repeatedGroup, 1780);
     [message addExtension:[UnittestRoot repeatedGroupExtension] value:repeatedGroup];
     [repeatedGroup release];
 
@@ -238,6 +304,7 @@ static GPBUnknownFieldSet *UnknownFieldsSetHelper(int num) {
     nestedMessage = [[TestAllTypes_NestedMessage alloc] init];
     nestedMessage.bb = 678;
     nestedMessage.unknownFields = UnknownFieldsSetHelper(781);
+    AddUnknownFields(nestedMessage, 1781);
     [message addExtension:[UnittestRoot repeatedNestedMessageExtension] value:nestedMessage];
     [nestedMessage release];
   }
@@ -245,6 +312,7 @@ static GPBUnknownFieldSet *UnknownFieldsSetHelper(int num) {
   // Confirm everything is there.
 
   XCTAssertNotNil(message);
+  XCTAssertTrue(HasUnknownFields(message));
   XCTAssertNotNil(message.unknownFields);
   XCTAssertTrue([message hasExtension:[UnittestRoot optionalInt32Extension]]);
 
@@ -254,6 +322,7 @@ static GPBUnknownFieldSet *UnknownFieldsSetHelper(int num) {
         [message getExtension:[UnittestRoot optionalGroupExtension]];
     XCTAssertNotNil(optionalGroup);
     XCTAssertEqual(optionalGroup.a, 123);
+    XCTAssertTrue(HasUnknownFields(optionalGroup));
     XCTAssertNotNil(optionalGroup.unknownFields);
   }
 
@@ -263,6 +332,7 @@ static GPBUnknownFieldSet *UnknownFieldsSetHelper(int num) {
         [message getExtension:[UnittestRoot optionalNestedMessageExtension]];
     XCTAssertNotNil(nestedMessage);
     XCTAssertEqual(nestedMessage.bb, 456);
+    XCTAssertTrue(HasUnknownFields(nestedMessage));
     XCTAssertNotNil(nestedMessage.unknownFields);
   }
 
@@ -273,6 +343,7 @@ static GPBUnknownFieldSet *UnknownFieldsSetHelper(int num) {
     RepeatedGroup_extension *repeatedGroup = repeatedGroups.firstObject;
     XCTAssertNotNil(repeatedGroup);
     XCTAssertEqual(repeatedGroup.a, 567);
+    XCTAssertTrue(HasUnknownFields(repeatedGroup));
     XCTAssertNotNil(repeatedGroup.unknownFields);
   }
 
@@ -284,6 +355,7 @@ static GPBUnknownFieldSet *UnknownFieldsSetHelper(int num) {
     TestAllTypes_NestedMessage *repeatedNestedMessage = repeatedNestedMessages.firstObject;
     XCTAssertNotNil(repeatedNestedMessage);
     XCTAssertEqual(repeatedNestedMessage.bb, 678);
+    XCTAssertTrue(HasUnknownFields(repeatedNestedMessage));
     XCTAssertNotNil(repeatedNestedMessage.unknownFields);
   }
 
@@ -293,6 +365,7 @@ static GPBUnknownFieldSet *UnknownFieldsSetHelper(int num) {
   // Confirm unknowns are gone from within the messages.
 
   XCTAssertNotNil(message);
+  XCTAssertFalse(HasUnknownFields(message));
   XCTAssertNil(message.unknownFields);
   XCTAssertTrue([message hasExtension:[UnittestRoot optionalInt32Extension]]);
 
@@ -302,6 +375,7 @@ static GPBUnknownFieldSet *UnknownFieldsSetHelper(int num) {
         [message getExtension:[UnittestRoot optionalGroupExtension]];
     XCTAssertNotNil(optionalGroup);
     XCTAssertEqual(optionalGroup.a, 123);
+    XCTAssertFalse(HasUnknownFields(optionalGroup));
     XCTAssertNil(optionalGroup.unknownFields);
   }
 
@@ -311,6 +385,7 @@ static GPBUnknownFieldSet *UnknownFieldsSetHelper(int num) {
         [message getExtension:[UnittestRoot optionalNestedMessageExtension]];
     XCTAssertNotNil(nestedMessage);
     XCTAssertEqual(nestedMessage.bb, 456);
+    XCTAssertFalse(HasUnknownFields(nestedMessage));
     XCTAssertNil(nestedMessage.unknownFields);
   }
 
@@ -321,6 +396,7 @@ static GPBUnknownFieldSet *UnknownFieldsSetHelper(int num) {
     RepeatedGroup_extension *repeatedGroup = repeatedGroups.firstObject;
     XCTAssertNotNil(repeatedGroup);
     XCTAssertEqual(repeatedGroup.a, 567);
+    XCTAssertFalse(HasUnknownFields(repeatedGroup));
     XCTAssertNil(repeatedGroup.unknownFields);
   }
 
@@ -332,6 +408,7 @@ static GPBUnknownFieldSet *UnknownFieldsSetHelper(int num) {
     TestAllTypes_NestedMessage *repeatedNestedMessage = repeatedNestedMessages.firstObject;
     XCTAssertNotNil(repeatedNestedMessage);
     XCTAssertEqual(repeatedNestedMessage.bb, 678);
+    XCTAssertFalse(HasUnknownFields(repeatedNestedMessage));
     XCTAssertNil(repeatedNestedMessage.unknownFields);
   }
 }
@@ -342,10 +419,12 @@ static GPBUnknownFieldSet *UnknownFieldsSetHelper(int num) {
   {
     ForeignMessage *foreignMessage = [ForeignMessage message];
     foreignMessage.unknownFields = UnknownFieldsSetHelper(100);
+    AddUnknownFields(foreignMessage, 1000);
     [message.mapInt32ForeignMessage setObject:foreignMessage forKey:100];
 
     foreignMessage = [ForeignMessage message];
     foreignMessage.unknownFields = UnknownFieldsSetHelper(101);
+    AddUnknownFields(foreignMessage, 1001);
     [message.mapStringForeignMessage setObject:foreignMessage forKey:@"101"];
   }
 
@@ -356,12 +435,14 @@ static GPBUnknownFieldSet *UnknownFieldsSetHelper(int num) {
   {
     ForeignMessage *foreignMessage = [message.mapInt32ForeignMessage objectForKey:100];
     XCTAssertNotNil(foreignMessage);
+    XCTAssertTrue(HasUnknownFields(foreignMessage));
     XCTAssertNotNil(foreignMessage.unknownFields);
   }
 
   {
     ForeignMessage *foreignMessage = [message.mapStringForeignMessage objectForKey:@"101"];
     XCTAssertNotNil(foreignMessage);
+    XCTAssertTrue(HasUnknownFields(foreignMessage));
     XCTAssertNotNil(foreignMessage.unknownFields);
   }
 
@@ -374,12 +455,14 @@ static GPBUnknownFieldSet *UnknownFieldsSetHelper(int num) {
   {
     ForeignMessage *foreignMessage = [message.mapInt32ForeignMessage objectForKey:100];
     XCTAssertNotNil(foreignMessage);
+    XCTAssertFalse(HasUnknownFields(foreignMessage));
     XCTAssertNil(foreignMessage.unknownFields);
   }
 
   {
     ForeignMessage *foreignMessage = [message.mapStringForeignMessage objectForKey:@"101"];
     XCTAssertNotNil(foreignMessage);
+    XCTAssertFalse(HasUnknownFields(foreignMessage));
     XCTAssertNil(foreignMessage.unknownFields);
   }
 }

objectivec/Tests/GPBWireFormatTests.m

@@ -5,11 +5,13 @@
 // license that can be found in the LICENSE file or at
 // https://developers.google.com/open-source/licenses/bsd
 
-#import "GPBTestUtilities.h"
-
 #import "GPBCodedInputStream.h"
 #import "GPBMessage_PackagePrivate.h"
+#import "GPBTestUtilities.h"
+#import "GPBUnknownField.h"
 #import "GPBUnknownField_PackagePrivate.h"
+#import "GPBUnknownFields.h"
+#import "GPBWireFormat.h"
 #import "objectivec/Tests/Unittest.pbobjc.h"
 #import "objectivec/Tests/UnittestMset.pbobjc.h"
 
@@ -110,32 +112,43 @@
 }
 
 const int kUnknownTypeId = 1550055;
+const int kUnknownTypeId2 = 1550056;
 
 - (void)testSerializeMessageSet {
   // Set up a MSetMessage with two known messages and an unknown one.
   MSetMessage* message_set = [MSetMessage message];
   [[message_set getExtension:[MSetMessageExtension1 messageSetExtension]] setI:123];
   [[message_set getExtension:[MSetMessageExtension2 messageSetExtension]] setStr:@"foo"];
+
   GPBUnknownField* unknownField =
       [[[GPBUnknownField alloc] initWithNumber:kUnknownTypeId] autorelease];
-  [unknownField addLengthDelimited:[NSData dataWithBytes:"bar" length:3]];
+  [unknownField addLengthDelimited:DataFromCStr("bar")];
   GPBUnknownFieldSet* unknownFieldSet = [[[GPBUnknownFieldSet alloc] init] autorelease];
   [unknownFieldSet addField:unknownField];
   [message_set setUnknownFields:unknownFieldSet];
 
+  GPBUnknownFields* ufs = [[[GPBUnknownFields alloc] init] autorelease];
+  GPBUnknownFields* group = [ufs addGroupWithFieldNumber:GPBWireFormatMessageSetItem];
+  [group addFieldNumber:GPBWireFormatMessageSetTypeId varint:kUnknownTypeId2];
+  [group addFieldNumber:GPBWireFormatMessageSetMessage lengthDelimited:DataFromCStr("baz")];
+  [message_set mergeUnknownFields:ufs extensionRegistry:[MSetUnittestMsetRoot extensionRegistry]];
+
   NSData* data = [message_set data];
 
   // Parse back using MSetRawMessageSet and check the contents.
   MSetRawMessageSet* raw = [MSetRawMessageSet parseFromData:data error:NULL];
 
+  GPBUnknownFields* ufs2 = [[[GPBUnknownFields alloc] initFromMessage:raw] autorelease];
+  XCTAssertTrue(ufs2.empty);
   XCTAssertEqual([raw.unknownFields countOfFields], (NSUInteger)0);
 
-  XCTAssertEqual(raw.itemArray.count, (NSUInteger)3);
+  XCTAssertEqual(raw.itemArray.count, (NSUInteger)4);
   XCTAssertEqual((uint32_t)[raw.itemArray[0] typeId],
                  [MSetMessageExtension1 messageSetExtension].fieldNumber);
   XCTAssertEqual((uint32_t)[raw.itemArray[1] typeId],
                  [MSetMessageExtension2 messageSetExtension].fieldNumber);
   XCTAssertEqual([raw.itemArray[2] typeId], kUnknownTypeId);
+  XCTAssertEqual([raw.itemArray[3] typeId], kUnknownTypeId2);
 
   MSetMessageExtension1* message1 =
       [MSetMessageExtension1 parseFromData:[((MSetRawMessageSet_Item*)raw.itemArray[0]) message]
@@ -147,7 +160,8 @@ const int kUnknownTypeId = 1550055;
                                      error:NULL];
   XCTAssertEqualObjects(message2.str, @"foo");
 
-  XCTAssertEqualObjects([raw.itemArray[2] message], [NSData dataWithBytes:"bar" length:3]);
+  XCTAssertEqualObjects([raw.itemArray[2] message], DataFromCStr("bar"));
+  XCTAssertEqualObjects([raw.itemArray[3] message], DataFromCStr("baz"));
 }
 
 - (void)testParseMessageSet {
@@ -175,7 +189,7 @@ const int kUnknownTypeId = 1550055;
   {
     MSetRawMessageSet_Item* item = [MSetRawMessageSet_Item message];
     item.typeId = kUnknownTypeId;
-    item.message = [NSData dataWithBytes:"bar" length:3];
+    item.message = DataFromCStr("bar");
     [raw.itemArray addObject:item];
   }
 
@@ -190,11 +204,100 @@ const int kUnknownTypeId = 1550055;
   XCTAssertEqualObjects([[messageSet getExtension:[MSetMessageExtension2 messageSetExtension]] str],
                         @"foo");
 
+  GPBUnknownFields* ufs = [[[GPBUnknownFields alloc] initFromMessage:messageSet] autorelease];
+  XCTAssertEqual(ufs.count, (NSUInteger)1);
+  GPBUnknownFields* group = [ufs firstGroup:GPBWireFormatMessageSetItem];
+  XCTAssertNotNil(group);
+  XCTAssertEqual(group.count, (NSUInteger)2);
+  uint64_t varint = 0;
+  XCTAssertTrue([group getFirst:GPBWireFormatMessageSetTypeId varint:&varint]);
+  XCTAssertEqual(varint, kUnknownTypeId);
+  XCTAssertEqualObjects([group firstLengthDelimited:GPBWireFormatMessageSetMessage],
+                        DataFromCStr("bar"));
+
   XCTAssertEqual([messageSet.unknownFields countOfFields], (NSUInteger)1);
   GPBUnknownField* unknownField = [messageSet.unknownFields getField:kUnknownTypeId];
   XCTAssertNotNil(unknownField);
   XCTAssertEqual(unknownField.lengthDelimitedList.count, (NSUInteger)1);
-  XCTAssertEqualObjects(unknownField.lengthDelimitedList[0], [NSData dataWithBytes:"bar" length:3]);
+  XCTAssertEqualObjects(unknownField.lengthDelimitedList[0], DataFromCStr("bar"));
+}
+
+- (void)testParseMessageSet_FirstValueSticks {
+  MSetRawBreakableMessageSet* raw = [MSetRawBreakableMessageSet message];
+
+  {
+    MSetRawBreakableMessageSet_Item* item = [MSetRawBreakableMessageSet_Item message];
+
+    [item.typeIdArray addValue:[MSetMessageExtension1 messageSetExtension].fieldNumber];
+    MSetMessageExtension1* message1 = [MSetMessageExtension1 message];
+    message1.i = 123;
+    NSData* itemData = [message1 data];
+    [item.messageArray addObject:itemData];
+
+    [item.typeIdArray addValue:[MSetMessageExtension2 messageSetExtension].fieldNumber];
+    MSetMessageExtension2* message2 = [MSetMessageExtension2 message];
+    message2.str = @"foo";
+    itemData = [message2 data];
+    [item.messageArray addObject:itemData];
+
+    [raw.itemArray addObject:item];
+  }
+
+  NSData* data = [raw data];
+
+  // Parse as a MSetMessage and check the contents.
+  NSError* err = nil;
+  MSetMessage* messageSet = [MSetMessage parseFromData:data
+                                     extensionRegistry:[MSetUnittestMsetRoot extensionRegistry]
+                                                 error:&err];
+  XCTAssertNotNil(messageSet);
+  XCTAssertNil(err);
+  XCTAssertTrue([messageSet hasExtension:[MSetMessageExtension1 messageSetExtension]]);
+  XCTAssertEqual([[messageSet getExtension:[MSetMessageExtension1 messageSetExtension]] i], 123);
+  XCTAssertFalse([messageSet hasExtension:[MSetMessageExtension2 messageSetExtension]]);
+  GPBUnknownFields* ufs = [[[GPBUnknownFields alloc] initFromMessage:messageSet] autorelease];
+  XCTAssertTrue(ufs.empty);
+}
+
+- (void)testParseMessageSet_PartialValuesDropped {
+  MSetRawBreakableMessageSet* raw = [MSetRawBreakableMessageSet message];
+
+  {
+    MSetRawBreakableMessageSet_Item* item = [MSetRawBreakableMessageSet_Item message];
+    [item.typeIdArray addValue:[MSetMessageExtension1 messageSetExtension].fieldNumber];
+    // No payload.
+    [raw.itemArray addObject:item];
+  }
+
+  {
+    MSetRawBreakableMessageSet_Item* item = [MSetRawBreakableMessageSet_Item message];
+    // No type ID.
+    MSetMessageExtension2* message = [MSetMessageExtension2 message];
+    message.str = @"foo";
+    NSData* itemData = [message data];
+    [item.messageArray addObject:itemData];
+    [raw.itemArray addObject:item];
+  }
+
+  {
+    MSetRawBreakableMessageSet_Item* item = [MSetRawBreakableMessageSet_Item message];
+    // Neither type ID nor payload.
+    [raw.itemArray addObject:item];
+  }
+
+  NSData* data = [raw data];
+
+  // Parse as a MSetMessage and check the contents.
+  NSError* err = nil;
+  MSetMessage* messageSet = [MSetMessage parseFromData:data
+                                     extensionRegistry:[MSetUnittestMsetRoot extensionRegistry]
+                                                 error:&err];
+  XCTAssertNotNil(messageSet);
+  XCTAssertNil(err);
+  XCTAssertEqual([messageSet extensionsCurrentlySet].count,
+                 (NSUInteger)0);  // None because they were all partial and dropped.
+  GPBUnknownFields* ufs = [[[GPBUnknownFields alloc] initFromMessage:messageSet] autorelease];
+  XCTAssertTrue(ufs.empty);
 }
 
 - (void)assertFieldsInOrder:(NSData*)data {

objectivec/Tests/unittest_mset.proto

@@ -14,6 +14,7 @@ option objc_class_prefix = "MSet";
 // A message with message_set_wire_format.
 message Message {
   option message_set_wire_format = true;
+
   extensions 4 to max;
 }
 
@@ -40,3 +41,12 @@ message RawMessageSet {
     required bytes message = 3;
   }
 }
+
+// MessageSet wire format is equivalent to this but since the fields
+// are repeated they can be left off or over present to testing.
+message RawBreakableMessageSet {
+  repeated group Item = 1 {
+    repeated int32 type_id = 2;
+    repeated bytes message = 3;
+  }
+}

objectivec/Tests/unittest_objc.proto

@@ -7,11 +7,11 @@
 
 syntax = "proto2";
 
+package objc.protobuf.tests;
+
 import "google/protobuf/any.proto";
 import "objectivec/Tests/unittest.proto";
 
-package objc.protobuf.tests;
-
 // Explicit empty prefix, tests some validations code paths also.
 option objc_class_prefix = "";
 
@@ -167,7 +167,6 @@ message self {
   }
 
   extensions 3000 to 3999;
-
 }
 
 enum retain {
@@ -700,13 +699,17 @@ message JustToScopeExtensions {
 
     repeated string mutableCopy_val_lower_complex_repeated = 2711;
     repeated string mutableCopy_Val_upper_complex_repeated = 2712;
-    repeated string mutableCopyvalue_lower_no_underscore_complex_repeated = 2713;
-    repeated string mutableCopyValue_upper_no_underscore_complex_repeated = 2714;
+    repeated string mutableCopyvalue_lower_no_underscore_complex_repeated =
+        2713;
+    repeated string mutableCopyValue_upper_no_underscore_complex_repeated =
+        2714;
 
     repeated int32 mutableCopy_val_lower_primitive_repeated = 2715;
     repeated int32 mutableCopy_Val_upper_primitive_repeated = 2716;
-    repeated int32 mutableCopyvalue_lower_no_underscore_primitive_repeated = 2717;
-    repeated int32 mutableCopyValue_upper_no_underscore_primitive_repeated = 2718;
+    repeated int32 mutableCopyvalue_lower_no_underscore_primitive_repeated =
+        2717;
+    repeated int32 mutableCopyValue_upper_no_underscore_primitive_repeated =
+        2718;
 
     repeated self mutableCopy_val_lower_message_repeated = 2719;
     repeated self mutableCopy_Val_upper_message_repeated = 2720;
@@ -794,6 +797,20 @@ message EnumTestMsg {
   repeated MyEnum mumble = 4;
 }
 
+message EnumTestMsgPrime {
+  enum MyEnumPrime {
+    ZERO = 0;
+    ONE = 1;
+    NEG_ONE = -1;
+    // Lacks 2, -2.
+  }
+  optional MyEnumPrime foo = 1;
+  optional MyEnumPrime bar = 2 [default = ONE];
+  optional MyEnumPrime baz = 3 [default = NEG_ONE];
+
+  repeated MyEnumPrime mumble = 4;
+}
+
 // Test case for https://github.com/protocolbuffers/protobuf/issues/1453
 // Message with no explicit defaults, but a non zero default for an enum.
 message MessageWithOneBasedEnum {

src/google/protobuf/compiler/objectivec/extension.cc

@@ -39,6 +39,14 @@ ExtensionGenerator::ExtensionGenerator(
   ABSL_CHECK(!descriptor->is_map())
       << "error: Extension is a map<>!"
       << " That used to be blocked by the compiler.";
+  if (descriptor->containing_type()->options().message_set_wire_format()) {
+    ABSL_CHECK(descriptor->type() == FieldDescriptor::TYPE_MESSAGE)
+        << "error: Extension to a message_set_wire_format message and the type "
+           "wasn't a message!";
+    ABSL_CHECK(!descriptor->is_repeated())
+        << "error: Extension to a message_set_wire_format message should not "
+           "be repeated!";
+  }
 }
 
 void ExtensionGenerator::GenerateMembersHeader(io::Printer* printer) const {

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

@@ -202,8 +202,12 @@ MessageGenerator::MessageGenerator(const std::string& file_description_name,
       class_name_(ClassName(descriptor_)),
       deprecated_attribute_(
           GetOptionalDeprecatedAttribute(descriptor, descriptor->file())) {
-  ABSL_DCHECK(!descriptor->options().map_entry())
+  ABSL_CHECK(!descriptor->options().map_entry())
       << "error: MessageGenerator create of a map<>!";
+  ABSL_CHECK(!descriptor->options().message_set_wire_format() ||
+             descriptor->field_count() == 0)
+      << "error: MessageGenerator message_set_wire_format should never have "
+         "fields!";
   for (int i = 0; i < descriptor_->real_oneof_decl_count(); i++) {
     oneof_generators_.push_back(std::make_unique<OneofGenerator>(
         descriptor_->real_oneof_decl(i), generation_options));