Merge branch 'data-refactoring'. Ported benchmarks.

Conflicts:

	src/upb_data.c
	src/upb_data.h

5 files changed, 137 insertions, 105 deletions

diff --git a/src/upb.h b/src/upb.h
index 2809418..2c0a23d 100644
--- a/src/upb.h
+++ b/src/upb.h
@@ -125,11 +125,10 @@ struct upb_tag {
 
 // INTERNAL-ONLY: never refer to these types with a tag ("union", "struct").
 // Always use the typedefs.
-union _upb_array;
 struct _upb_msg;
 
-typedef union _upb_array upb_array;
 typedef struct _upb_msg upb_msg;
+
 typedef upb_atomic_refcount_t upb_data;
 
 typedef uint32_t upb_strlen_t;
@@ -143,6 +142,15 @@ typedef union {
   upb_data *base;
 } upb_strptr;
 
+typedef uint32_t upb_arraylen_t;
+
+typedef union {
+  // Must be first, for the UPB_STATIC_ARRAY_PTR_INIT() macro.
+  struct upb_norefcount_array *norefcount;
+  struct upb_refcounted_array *refcounted;
+  upb_data *base;
+} upb_arrayptr;
+
 // A single .proto value.  The owner must have an out-of-band way of knowing
 // the type, so that it knows which union member to use.
 union upb_value {
@@ -154,7 +162,7 @@ union upb_value {
   uint64_t uint64;
   bool _bool;
   upb_strptr str;
-  upb_array *arr;
+  upb_arrayptr arr;
   upb_msg *msg;
   upb_data *data;
 };
@@ -171,7 +179,7 @@ union upb_value_ptr {
   uint64_t *uint64;
   bool *_bool;
   upb_strptr *str;
-  upb_array **arr;
+  upb_arrayptr *arr;
   upb_msg **msg;
   upb_data **data;
   void *_void;
diff --git a/src/upb_data.c b/src/upb_data.c
index 1bb76aa..a0a588a 100644
--- a/src/upb_data.c
+++ b/src/upb_data.c
@@ -57,11 +57,11 @@ static void check_not_frozen(upb_data *d) {
 
 /* upb_string *******************************************************************/
 
-static char *_upb_string_setptr(upb_strptr s, char *ptr) {
+void _upb_string_setptr(upb_strptr s, char *ptr) {
   if(upb_data_hasflag(s.base, UPB_DATA_REFCOUNTED))
-    return s.refcounted->ptr = ptr;
+    s.refcounted->ptr = ptr;
   else
-    return s.norefcount->ptr = ptr;
+    s.norefcount->ptr = ptr;
 }
 
 static void _upb_string_set_bytelen(upb_strptr s, upb_strlen_t newlen) {
@@ -182,64 +182,79 @@ int upb_strcmp(upb_strptr s1, upb_strptr s2) {
 
 /* upb_array ******************************************************************/
 
-upb_array *upb_array_new() {
-  upb_array *a = malloc(sizeof(upb_refcounted_array));
-  data_init(&a->common.base, UPB_DATA_HEAPALLOCATED | UPB_DATA_REFCOUNTED);
-  a->refcounted.size = 0;
-  a->common.len = 0;
-  a->common.elements._void = NULL;
+static void _upb_array_setptr(upb_arrayptr a, void *ptr) {
+  if(upb_data_hasflag(a.base, UPB_DATA_REFCOUNTED))
+    a.refcounted->elements._void = ptr;
+  else
+    a.norefcount->elements._void = ptr;
+}
+
+static void _upb_array_setlen(upb_arrayptr a, upb_strlen_t newlen) {
+  if(upb_data_hasflag(a.base, UPB_DATA_REFCOUNTED)) {
+    a.refcounted->len = newlen;
+  } else {
+    a.norefcount->len = newlen;
+  }
+}
+
+upb_arrayptr upb_array_new() {
+  upb_arrayptr a;
+  a.refcounted = malloc(sizeof(struct upb_refcounted_array));
+  data_init(a.base, UPB_DATA_HEAPALLOCATED | UPB_DATA_REFCOUNTED);
+  a.refcounted->size = 0;
+  a.refcounted->len = 0;
+  a.refcounted->elements._void = NULL;
   return a;
 }
 
 // ONLY handles refcounted arrays for the moment.
-void _upb_array_free(upb_array *a, struct upb_fielddef *f)
+void _upb_array_free(upb_arrayptr a, struct upb_fielddef *f)
 {
   if(upb_elem_ismm(f)) {
-    for(upb_arraylen_t i = 0; i < a->refcounted.size; i++) {
+    for(upb_arraylen_t i = 0; i < a.refcounted->size; i++) {
       union upb_value_ptr p = _upb_array_getptr(a, f, i);
       if(!*p.data) continue;
       data_elem_unref(p, f);
     }
   }
-  if(a->refcounted.size != 0) free(a->common.elements._void);
-  free(a);
+  if(a.refcounted->size != 0) free(a.refcounted->elements._void);
+  free(a.refcounted);
 }
 
-static upb_arraylen_t array_get_size(upb_array *a) {
-  if(upb_data_hasflag(&a->common.base, UPB_DATA_REFCOUNTED)) {
-    return a->refcounted.size;
+static upb_arraylen_t array_get_size(upb_arrayptr a) {
+  if(upb_data_hasflag(a.base, UPB_DATA_REFCOUNTED)) {
+    return a.refcounted->size;
   } else {
-    return (a->norefcount.size_and_flags & 0xFFFFFFF8) >> 3;
+    return (a.norefcount->base.v & 0xFFFFFFF8) >> 3;
   }
 }
 
-static void array_set_size(upb_array *a, upb_arraylen_t newsize) {
-  if(upb_data_hasflag(&a->common.base, UPB_DATA_REFCOUNTED)) {
-    a->refcounted.size = newsize;
+static void array_set_size(upb_arrayptr a, upb_arraylen_t newsize) {
+  if(upb_data_hasflag(a.base, UPB_DATA_REFCOUNTED)) {
+    a.refcounted->size = newsize;
   } else {
-    a->norefcount.size_and_flags &= 0x7;
-    a->norefcount.size_and_flags |= (newsize << 3);
+    a.norefcount->base.v &= 0x7;
+    a.norefcount->base.v |= (newsize << 3);
   }
 }
 
-void upb_array_resize(upb_array *a, struct upb_fielddef *f, upb_strlen_t len) {
-  check_not_frozen(&a->common.base);
+void upb_array_resize(upb_arrayptr a, struct upb_fielddef *f, upb_strlen_t len) {
+  check_not_frozen(a.base);
   size_t type_size = upb_type_info[f->type].size;
   upb_arraylen_t old_size = array_get_size(a);
   if(old_size < len) {
     // Need to resize.
     size_t new_size = round_up_to_pow2(len);
-    a->common.elements._void = realloc(a->common.elements._void, new_size * type_size);
+    _upb_array_setptr(a, realloc(_upb_array_getptr_raw(a, 0, 0)._void, new_size * type_size));
     array_set_size(a, new_size);
-    memset(UPB_INDEX(a->common.elements._void, old_size, type_size),
+    memset(_upb_array_getptr_raw(a, old_size, type_size)._void,
            0,
            (new_size - old_size) * type_size);
   }
-  a->common.len = len;
+  _upb_array_setlen(a, len);
 }
 
 
-
 /* upb_msg ********************************************************************/
 
 static void upb_msg_sethas(upb_msg *msg, struct upb_fielddef *f) {
@@ -289,19 +304,15 @@ static union upb_value_ptr get_value_ptr(upb_msg *msg, struct upb_fielddef *f)
   union upb_value_ptr p = _upb_msg_getptr(msg, f);
   if(upb_isarray(f)) {
     if(!upb_msg_has(msg, f)) {
-      if(!*p.arr || !upb_data_only(*p.data)) {
-        printf("Initializing array field " UPB_STRFMT "\n", UPB_STRARG(f->name));
-        if(*p.arr)
+      if(upb_array_isnull(*p.arr) || !upb_data_only(*p.data)) {
+        if(!upb_array_isnull(*p.arr))
           upb_array_unref(*p.arr, f);
         *p.arr = upb_array_new();
-      } else {
-        printf("REUSING array field " UPB_STRFMT "\n", UPB_STRARG(f->name));
       }
       upb_array_truncate(*p.arr);
       upb_msg_sethas(msg, f);
     } else {
-      printf("APPENDING TO EXISTING array field " UPB_STRFMT "\n", UPB_STRARG(f->name));
-      assert(*p.arr);
+      assert(!upb_array_isnull(*p.arr));
     }
     upb_arraylen_t oldlen = upb_array_len(*p.arr);
     upb_array_resize(*p.arr, f, oldlen + 1);
diff --git a/src/upb_data.h b/src/upb_data.h
index 15e87c5..98243ce 100644
--- a/src/upb_data.h
+++ b/src/upb_data.h
@@ -213,9 +213,7 @@ upb_strptr upb_string_getref(upb_strptr s, int ref_flags);
 
 #define UPB_STRING_NULL_INITIALIZER {NULL}
 static const upb_strptr UPB_STRING_NULL = UPB_STRING_NULL_INITIALIZER;
-INLINE bool upb_string_isnull(upb_strptr s) {
-  return s.base == NULL;
-}
+INLINE bool upb_string_isnull(upb_strptr s) { return s.base == NULL; }
 
 // The caller releases a ref on src, which it must previously have owned a ref
 // on.
@@ -350,70 +348,77 @@ typedef struct upb_norefcount_string upb_static_string;
 
 /* upb_array ******************************************************************/
 
-typedef uint32_t upb_arraylen_t;
-
 // The comments attached to upb_string above also apply here.
-typedef struct {
-  upb_data base;
-  upb_arraylen_t len;
-  union upb_value_ptr elements;
-} upb_array_common;
-
-typedef struct {
-  uint32_t size_and_flags;
+struct upb_norefcount_array {
+  upb_data base;  // We co-opt the refcount for the size.
   upb_arraylen_t len;
   union upb_value_ptr elements;
-} upb_norefcount_array;
+};
 
-typedef struct {
+struct upb_refcounted_array {
   upb_data base;
   upb_arraylen_t len;
   union upb_value_ptr elements;
   upb_arraylen_t size;
-} upb_refcounted_array;
-
-union _upb_array {
-  upb_norefcount_array norefcount;
-  upb_array_common common;
-  upb_refcounted_array refcounted;
 };
 
-// This type can be used either to perform read-only access on an array,
-// or to statically define a non-reference-counted static array.
-#define UPB_DEFINE_MSG_ARRAY(type) \
-typedef struct type ## _array { \
-  upb_data base; \
-  upb_arraylen_t len;\
-  type **elements; \
-} type ## _array; \
+typedef struct upb_norefcount_array upb_static_array;
+#define UPB_STATIC_ARRAY_INIT(arr, len) {{0 | UPB_DATA_FROZEN}, len, {._void=arr}}
+#define UPB_STATIC_ARRAY_PTR_TYPED_INIT(static_arr) {{&static_arr}}
+
+#define UPB_ARRAY_NULL_INITIALIZER {NULL}
+static const upb_arrayptr UPB_ARRAY_NULL = UPB_ARRAY_NULL_INITIALIZER;
+INLINE bool upb_array_isnull(upb_arrayptr a) { return a.base == NULL; }
+INLINE bool upb_array_ptreql(upb_arrayptr a1, upb_arrayptr a2) {
+  return a1.base == a2.base;
+}
 
-#define UPB_MSG_ARRAY(type) struct type ## _array
+#define UPB_MSG_ARRAYPTR(type) type ## _array
+#define UPB_DEFINE_MSG_ARRAY(type) \
+typedef struct { upb_arrayptr ptr; } UPB_MSG_ARRAYPTR(type); \
+INLINE upb_arraylen_t type ## _array_len(UPB_MSG_ARRAYPTR(type) a) { \
+  return upb_array_len(a.ptr); \
+} \
+INLINE type* type ## _array_get(UPB_MSG_ARRAYPTR(type) a, upb_arraylen_t elem) { \
+  return *(type**)_upb_array_getptr_raw(a.ptr, elem, sizeof(void*))._void; \
+}
 
 // Constructs a newly-allocated, reference-counted array which starts out
 // empty.  Caller owns one ref on it.
-upb_array *upb_array_new(void);
+upb_arrayptr upb_array_new(void);
 
 // Returns the current number of elements in the array.
-INLINE size_t upb_array_len(upb_array *a) {
-  return a->common.len;
+INLINE size_t upb_array_len(upb_arrayptr a) {
+  if(upb_data_hasflag(a.base, UPB_DATA_REFCOUNTED))
+    return a.refcounted->len;
+  else
+    return a.norefcount->len;
 }
 
 // INTERNAL-ONLY:
 // Frees the given message and releases references on members.
-void _upb_array_free(upb_array *a, struct upb_fielddef *f);
+void _upb_array_free(upb_arrayptr a, struct upb_fielddef *f);
 
 // INTERNAL-ONLY:
 // Returns a pointer to the given elem.
-INLINE union upb_value_ptr _upb_array_getptr(upb_array *a,
-                                             struct upb_fielddef *f,
-                                             upb_arraylen_t elem) {
-  size_t type_size = upb_type_info[f->type].size;
+INLINE union upb_value_ptr _upb_array_getptr_raw(upb_arrayptr a,
+                                                 upb_arraylen_t elem,
+                                                 size_t type_size) {
   union upb_value_ptr p;
-  p._void = &a->common.elements.uint8[elem * type_size];
+  if(upb_data_hasflag(a.base, UPB_DATA_REFCOUNTED))
+    p._void = &a.refcounted->elements.uint8[elem * type_size];
+  else
+    p._void = &a.norefcount->elements.uint8[elem * type_size];
   return p;
 }
 
-INLINE union upb_value upb_array_get(upb_array *a, struct upb_fielddef *f,
+INLINE union upb_value_ptr _upb_array_getptr(upb_arrayptr a,
+                                             struct upb_fielddef *f,
+                                             upb_arraylen_t elem) {
+  return _upb_array_getptr_raw(a, elem, upb_type_info[f->type].size);
+}
+
+INLINE union upb_value upb_array_get(upb_arrayptr a, struct upb_fielddef *f,
                                      upb_arraylen_t elem) {
   assert(elem < upb_array_len(a));
   return upb_value_read(_upb_array_getptr(a, f, elem), f->type);
@@ -421,33 +426,36 @@ INLINE union upb_value upb_array_get(upb_array *a, struct upb_fielddef *f,
 
 // The caller releases a ref on the given array, which it must previously have
 // owned a ref on.
-INLINE void upb_array_unref(upb_array *a, struct upb_fielddef *f) {
-  if(_upb_data_unref(&a->common.base)) _upb_array_free(a, f);
+INLINE void upb_array_unref(upb_arrayptr a, struct upb_fielddef *f) {
+  if(_upb_data_unref(a.base)) _upb_array_free(a, f);
 }
 
 #if 0
 // Returns an array to which caller owns a ref, and contains the same contents
 // as src.  The returned value may be a copy of src, if the requested flags
 // were incompatible with src's.
-INLINE upb_array *upb_array_getref(upb_array *src, int ref_flags);
+INLINE upb_arrayptr upb_array_getref(upb_arrayptr src, int ref_flags);
 
 // Sets the given element in the array to val.  The current length of the array
 // must be greater than elem.  If the field type is dynamic, the array will
 // take a ref on val and release a ref on what was previously in the array.
-INLINE void upb_array_set(upb_array *a, struct upb_fielddef *f, int elem,
+INLINE void upb_array_set(upb_arrayptr a, struct upb_fielddef *f, int elem,
                           union upb_value val);
 
 
 // Note that array_append will attempt to take a reference on the given value,
 // so to avoid a copy use append_default and get.
-INLINE void upb_array_append(upb_array *a, struct upb_fielddef *f,
+INLINE void upb_array_append(upb_arrayptr a, struct upb_fielddef *f,
                              union upb_value val);
-INLINE void upb_array_append_default(upb_array *a, struct upb_fielddef *f,
+INLINE void upb_array_append_default(upb_arrayptr a, struct upb_fielddef *f,
                              union upb_value val);
 #endif
 
-INLINE void upb_array_truncate(upb_array *a) {
-  a->common.len = 0;
+INLINE void upb_array_truncate(upb_arrayptr a) {
+  if(upb_data_hasflag(a.base, UPB_DATA_REFCOUNTED))
+    a.refcounted->len = 0;
+  else
+    a.norefcount->len = 0;
 }
 
 
diff --git a/src/upb_def.c b/src/upb_def.c
index 4f57407..24a57a6 100644
--- a/src/upb_def.c
+++ b/src/upb_def.c
@@ -364,12 +364,14 @@ static struct upb_enumdef *enumdef_new(google_protobuf_EnumDescriptorProto *ed,
 {
   struct upb_enumdef *e = malloc(sizeof(*e));
   upb_def_init(&e->base, UPB_DEF_ENUM, fqname);
-  int num_values = ed->set_flags.has.value ? ed->value->len : 0;
+  int num_values = ed->set_flags.has.value ?
+      google_protobuf_EnumValueDescriptorProto_array_len(ed->value) : 0;
   upb_strtable_init(&e->ntoi, num_values, sizeof(struct ntoi_ent));
   upb_inttable_init(&e->iton, num_values, sizeof(struct iton_ent));
 
   for(int i = 0; i < num_values; i++) {
-    google_protobuf_EnumValueDescriptorProto *value = ed->value->elements[i];
+    google_protobuf_EnumValueDescriptorProto *value =
+        google_protobuf_EnumValueDescriptorProto_array_get(ed->value, i);
     struct ntoi_ent ntoi_ent = {{value->name, 0}, value->number};
     struct iton_ent iton_ent = {{value->number, 0}, value->name};
     upb_strtable_insert(&e->ntoi, &ntoi_ent.e);
@@ -521,20 +523,22 @@ static void insert_message(struct upb_strtable *t,
   upb_strptr fqname = try_define(t, base, name, status);
   if(upb_string_isnull(fqname)) return;
 
-  int num_fields = d->set_flags.has.field ? d->field->len : 0;
+  int num_fields = d->set_flags.has.field ?
+      google_protobuf_FieldDescriptorProto_array_len(d->field) : 0;
   struct symtab_ent e;
   e.e.key = fqname;
 
   // Gather our list of fields, sorting if necessary.
   struct upb_fielddef **fielddefs = malloc(sizeof(*fielddefs) * num_fields);
   for (int i = 0; i < num_fields; i++) {
-    google_protobuf_FieldDescriptorProto *fd = d->field->elements[i];
+    google_protobuf_FieldDescriptorProto *fd =
+        google_protobuf_FieldDescriptorProto_array_get(d->field, i);
     fielddefs[i] = fielddef_new(fd);
   }
-  if(sort) fielddef_sort(fielddefs, d->field->len);
+  if(sort) fielddef_sort(fielddefs, num_fields);
 
   // Create the msgdef with that list of fields.
-  e.def = UPB_UPCAST(msgdef_new(fielddefs, d->field->len, fqname, status));
+  e.def = UPB_UPCAST(msgdef_new(fielddefs, num_fields, fqname, status));
 
   // Cleanup.
   for (int i = 0; i < num_fields; i++) fielddef_free(fielddefs[i]);
@@ -546,12 +550,12 @@ static void insert_message(struct upb_strtable *t,
 
   /* Add nested messages and enums. */
   if(d->set_flags.has.nested_type)
-    for(unsigned int i = 0; i < d->nested_type->len; i++)
-      insert_message(t, d->nested_type->elements[i], fqname, sort, status);
+    for(unsigned int i = 0; i < google_protobuf_DescriptorProto_array_len(d->nested_type); i++)
+      insert_message(t, google_protobuf_DescriptorProto_array_get(d->nested_type, i), fqname, sort, status);
 
   if(d->set_flags.has.enum_type)
-    for(unsigned int i = 0; i < d->enum_type->len; i++)
-      insert_enum(t, d->enum_type->elements[i], fqname, status);
+    for(unsigned int i = 0; i < google_protobuf_EnumDescriptorProto_array_len(d->enum_type); i++)
+      insert_enum(t, google_protobuf_EnumDescriptorProto_array_get(d->enum_type, i), fqname, status);
 
 error:
   // Free the ref we got from try_define().
@@ -618,12 +622,12 @@ static void addfd(struct upb_strtable *addto, struct upb_strtable *existingdefs,
   }
 
   if(fd->set_flags.has.message_type)
-    for(unsigned int i = 0; i < fd->message_type->len; i++)
-      insert_message(addto, fd->message_type->elements[i], pkg, sort, status);
+    for(unsigned int i = 0; i < google_protobuf_DescriptorProto_array_len(fd->message_type); i++)
+      insert_message(addto, google_protobuf_DescriptorProto_array_get(fd->message_type, i), pkg, sort, status);
 
   if(fd->set_flags.has.enum_type)
-    for(unsigned int i = 0; i < fd->enum_type->len; i++)
-      insert_enum(addto, fd->enum_type->elements[i], pkg, status);
+    for(unsigned int i = 0; i < google_protobuf_EnumDescriptorProto_array_len(fd->enum_type); i++)
+      insert_enum(addto, google_protobuf_EnumDescriptorProto_array_get(fd->enum_type, i), pkg, status);
 
   upb_string_unref(pkg);
 
@@ -686,8 +690,9 @@ struct upb_symtab *upb_symtab_new()
   upb_strtable_init(&s->psymtab, 16, sizeof(struct symtab_ent));
 
   // Add descriptor.proto types to private symtable so we can parse descriptors.
+  // We know there is only 1.
   google_protobuf_FileDescriptorProto *fd =
-      upb_file_descriptor_set->file->elements[0];  // We know there is only 1.
+      google_protobuf_FileDescriptorProto_array_get(upb_file_descriptor_set->file, 0);
   struct upb_status status = UPB_STATUS_INIT;
   addfd(&s->psymtab, &s->symtab, fd, false, &status);
   if(!upb_ok(&status)) {
@@ -782,8 +787,8 @@ void upb_symtab_addfds(struct upb_symtab *s,
 
     {  // Read lock scope
       upb_rwlock_rdlock(&s->lock);
-      for(uint32_t i = 0; i < fds->file->len; i++) {
-        addfd(&tmp, &s->symtab, fds->file->elements[i], true, status);
+      for(uint32_t i = 0; i < google_protobuf_FileDescriptorProto_array_len(fds->file); i++) {
+        addfd(&tmp, &s->symtab, google_protobuf_FileDescriptorProto_array_get(fds->file, i), true, status);
         if(!upb_ok(status)) {
           free_symtab(&tmp);
           upb_rwlock_unlock(&s->lock);
diff --git a/src/upb_text.c b/src/upb_text.c
index 225b344..3981715 100644
--- a/src/upb_text.c
+++ b/src/upb_text.c
@@ -92,7 +92,7 @@ static void printmsg(struct upb_text_printer *printer,
     if(!upb_msg_has(msg, f)) continue;
     union upb_value v = upb_msg_get(msg, f);
     if(upb_isarray(f)) {
-      upb_array *arr = v.arr;
+      upb_arrayptr arr = v.arr;
       for(uint32_t j = 0; j < upb_array_len(arr); j++) {
         union upb_value elem = upb_array_get(arr, f, j);
         printval(printer, elem, f, stream);