msgpack 1.5.4 → 1.8.1

data/ext/msgpack/unpacker.c

@@ -19,102 +19,147 @@
 #include "unpacker.h"
 #include "rmem.h"
 #include "extension_value_class.h"
+#include <assert.h>
+#include <limits.h>
 
-#if !defined(DISABLE_RMEM) && !defined(DISABLE_UNPACKER_STACK_RMEM) && \
-        MSGPACK_UNPACKER_STACK_CAPACITY * MSGPACK_UNPACKER_STACK_SIZE <= MSGPACK_RMEM_PAGE_SIZE
-#define UNPACKER_STACK_RMEM
+#if !defined(HAVE_RB_PROC_CALL_WITH_BLOCK)
+#define rb_proc_call_with_block(recv, argc, argv, block) rb_funcallv(recv, rb_intern("call"), argc, argv)
 #endif
 
+#ifndef HAVE_RB_GC_MARK_LOCATIONS
+// For TruffleRuby
+void rb_gc_mark_locations(const VALUE *start, const VALUE *end)
+{
+    VALUE *value = start;
+
+    while (value < end) {
+        rb_gc_mark(*value);
+        value++;
+    }
+}
+#endif
+
+struct protected_proc_call_args {
+    VALUE proc;
+    int argc;
+    VALUE *argv;
+};
+
+static VALUE protected_proc_call_safe(VALUE _args) {
+    struct protected_proc_call_args *args = (struct protected_proc_call_args *)_args;
+
+    return rb_proc_call_with_block(args->proc, args->argc, args->argv, Qnil);
+}
+
+static VALUE protected_proc_call(VALUE proc, int argc, VALUE *argv, int *raised) {
+    struct protected_proc_call_args args = {
+      .proc = proc,
+      .argc = argc,
+      .argv = argv,
+    };
+    return rb_protect(protected_proc_call_safe, (VALUE)&args, raised);
+}
+
 static int RAW_TYPE_STRING = 256;
 static int RAW_TYPE_BINARY = 257;
+static int16_t INITIAL_BUFFER_CAPACITY_MAX = SHRT_MAX;
 
-static ID s_call;
-
-#ifdef UNPACKER_STACK_RMEM
 static msgpack_rmem_t s_stack_rmem;
-#endif
 
 #if !defined(HAVE_RB_HASH_NEW_CAPA)
-static inline VALUE rb_hash_new_capa(long capa)
+static inline VALUE rb_hash_new_capa_inline(long capa)
 {
   return rb_hash_new();
 }
+#define rb_hash_new_capa rb_hash_new_capa_inline
 #endif
 
-void msgpack_unpacker_static_init()
+static inline int16_t initial_buffer_size(long size)
 {
-#ifdef UNPACKER_STACK_RMEM
-    msgpack_rmem_init(&s_stack_rmem);
-#endif
+    return (size > INITIAL_BUFFER_CAPACITY_MAX) ? INITIAL_BUFFER_CAPACITY_MAX : size;
+}
+
+void msgpack_unpacker_static_init(void)
+{
+    assert(sizeof(msgpack_unpacker_stack_entry_t) * MSGPACK_UNPACKER_STACK_CAPACITY <= MSGPACK_RMEM_PAGE_SIZE);
 
-    s_call = rb_intern("call");
+    msgpack_rmem_init(&s_stack_rmem);
 }
 
-void msgpack_unpacker_static_destroy()
+void msgpack_unpacker_static_destroy(void)
 {
-#ifdef UNPACKER_STACK_RMEM
     msgpack_rmem_destroy(&s_stack_rmem);
-#endif
 }
 
 #define HEAD_BYTE_REQUIRED 0xc1
 
-static inline msgpack_unpacker_stack_t* _msgpack_unpacker_new_stack(void) {
-#ifdef UNPACKER_STACK_RMEM
-    return msgpack_rmem_alloc(&s_stack_rmem);
-    /*memset(uk->stack, 0, MSGPACK_UNPACKER_STACK_CAPACITY);*/
-#else
-    /*uk->stack = calloc(MSGPACK_UNPACKER_STACK_CAPACITY, sizeof(msgpack_unpacker_stack_t));*/
-    return xmalloc(MSGPACK_UNPACKER_STACK_CAPACITY * sizeof(msgpack_unpacker_stack_t));
-#endif
+static inline bool _msgpack_unpacker_stack_init(msgpack_unpacker_stack_t *stack) {
+    if (!stack->data) {
+        stack->capacity = MSGPACK_UNPACKER_STACK_CAPACITY;
+        stack->data = msgpack_rmem_alloc(&s_stack_rmem);
+        stack->depth = 0;
+        return true;
+    }
+    return false;
 }
 
-msgpack_unpacker_t* _msgpack_unpacker_new(void)
-{
-    msgpack_unpacker_t* uk = ZALLOC_N(msgpack_unpacker_t, 1);
+static inline void _msgpack_unpacker_free_stack(msgpack_unpacker_stack_t* stack) {
+    if (stack->data) {
+        if (!msgpack_rmem_free(&s_stack_rmem, stack->data)) {
+            rb_bug("Failed to free an rmem pointer, memory leak?");
+        }
+        stack->data = NULL;
+        stack->depth = 0;
+    }
+}
 
+#define STACK_INIT(uk) bool stack_allocated = _msgpack_unpacker_stack_init(&uk->stack);
+#define STACK_FREE(uk) if (stack_allocated) { _msgpack_unpacker_free_stack(&uk->stack); }
+
+void _msgpack_unpacker_init(msgpack_unpacker_t* uk)
+{
     msgpack_buffer_init(UNPACKER_BUFFER_(uk));
 
     uk->head_byte = HEAD_BYTE_REQUIRED;
 
     uk->last_object = Qnil;
     uk->reading_raw = Qnil;
+}
 
-    uk->stack = _msgpack_unpacker_new_stack();
-    uk->stack_capacity = MSGPACK_UNPACKER_STACK_CAPACITY;
-
-    return uk;
+void _msgpack_unpacker_destroy(msgpack_unpacker_t* uk)
+{
+    _msgpack_unpacker_free_stack(&uk->stack);
+    msgpack_buffer_destroy(UNPACKER_BUFFER_(uk));
 }
 
-static inline void _msgpack_unpacker_free_stack(msgpack_unpacker_stack_t* stack) {
-    #ifdef UNPACKER_STACK_RMEM
-        msgpack_rmem_free(&s_stack_rmem, stack);
-    #else
-        xfree(stack);
-    #endif
+void msgpack_unpacker_mark_stack(msgpack_unpacker_stack_t* stack)
+{
+    if (stack->data) {
+        msgpack_unpacker_stack_entry_t* s = stack->data;
+        msgpack_unpacker_stack_entry_t* send = stack->data + stack->depth;
+        for(; s < send; s++) {
+            rb_gc_mark(s->object);
+            rb_gc_mark(s->key);
+        }
+    }
 }
 
-void _msgpack_unpacker_destroy(msgpack_unpacker_t* uk)
+void msgpack_unpacker_mark_key_cache(msgpack_key_cache_t *cache)
 {
-    _msgpack_unpacker_free_stack(uk->stack);
-    msgpack_buffer_destroy(UNPACKER_BUFFER_(uk));
+    const VALUE *entries = &cache->entries[0];
+    rb_gc_mark_locations(entries, entries + cache->length);
 }
 
 void msgpack_unpacker_mark(msgpack_unpacker_t* uk)
 {
     rb_gc_mark(uk->last_object);
     rb_gc_mark(uk->reading_raw);
-
-    msgpack_unpacker_stack_t* s = uk->stack;
-    msgpack_unpacker_stack_t* send = uk->stack + uk->stack_depth;
-    for(; s < send; s++) {
-        rb_gc_mark(s->object);
-        rb_gc_mark(s->key);
-    }
-
+    msgpack_unpacker_mark_stack(&uk->stack);
+    msgpack_unpacker_mark_key_cache(&uk->key_cache);
     /* See MessagePack_Buffer_wrap */
     /* msgpack_buffer_mark(UNPACKER_BUFFER_(uk)); */
     rb_gc_mark(uk->buffer_ref);
+    rb_gc_mark(uk->self);
 }
 
 void _msgpack_unpacker_reset(msgpack_unpacker_t* uk)
@@ -123,9 +168,8 @@ void _msgpack_unpacker_reset(msgpack_unpacker_t* uk)
 
     uk->head_byte = HEAD_BYTE_REQUIRED;
 
-    /*memset(uk->stack, 0, sizeof(msgpack_unpacker_t) * uk->stack_depth);*/
-    uk->stack_depth = 0;
-
+    /*memset(uk->stack, 0, sizeof(msgpack_unpacker_t) * uk->stack.depth);*/
+    uk->stack.depth = 0;
     uk->last_object = Qnil;
     uk->reading_raw = Qnil;
     uk->reading_raw_remaining = 0;
@@ -188,7 +232,13 @@ static inline int object_complete_ext(msgpack_unpacker_t* uk, int ext_type, VALU
 
     if(proc != Qnil) {
         VALUE obj;
-        obj = rb_funcall(proc, s_call, 1, str == Qnil ? rb_str_buf_new(0) : str);
+        VALUE arg = (str == Qnil ? rb_str_buf_new(0) : str);
+        int raised;
+        obj = protected_proc_call(proc, 1, &arg, &raised);
+        if (raised) {
+            uk->last_object = rb_errinfo();
+            return PRIMITIVE_RECURSIVE_RAISED;
+        }
         return object_complete(uk, obj);
     }
 
@@ -201,37 +251,37 @@ static inline int object_complete_ext(msgpack_unpacker_t* uk, int ext_type, VALU
 }
 
 /* stack funcs */
-static inline msgpack_unpacker_stack_t* _msgpack_unpacker_stack_top(msgpack_unpacker_t* uk)
+static inline msgpack_unpacker_stack_entry_t* _msgpack_unpacker_stack_entry_top(msgpack_unpacker_t* uk)
 {
-    return &uk->stack[uk->stack_depth-1];
+    return &uk->stack.data[uk->stack.depth-1];
 }
 
 static inline int _msgpack_unpacker_stack_push(msgpack_unpacker_t* uk, enum stack_type_t type, size_t count, VALUE object)
 {
     reset_head_byte(uk);
 
-    if(uk->stack_capacity - uk->stack_depth <= 0) {
+    if(uk->stack.capacity - uk->stack.depth <= 0) {
         return PRIMITIVE_STACK_TOO_DEEP;
     }
 
-    msgpack_unpacker_stack_t* next = &uk->stack[uk->stack_depth];
+    msgpack_unpacker_stack_entry_t* next = &uk->stack.data[uk->stack.depth];
     next->count = count;
     next->type = type;
     next->object = object;
     next->key = Qnil;
 
-    uk->stack_depth++;
+    uk->stack.depth++;
     return PRIMITIVE_CONTAINER_START;
 }
 
-static inline VALUE msgpack_unpacker_stack_pop(msgpack_unpacker_t* uk)
+static inline size_t msgpack_unpacker_stack_pop(msgpack_unpacker_t* uk)
 {
-    return --uk->stack_depth;
+    return --uk->stack.depth;
 }
 
 static inline bool msgpack_unpacker_stack_is_empty(msgpack_unpacker_t* uk)
 {
-    return uk->stack_depth == 0;
+    return uk->stack.depth == 0;
 }
 
 #ifdef USE_CASE_RANGE
@@ -250,17 +300,30 @@ static inline bool msgpack_unpacker_stack_is_empty(msgpack_unpacker_t* uk)
 
 #endif
 
+union msgpack_buffer_cast_block_t {
+    char buffer[8];
+    uint8_t u8;
+    uint16_t u16;
+    uint32_t u32;
+    uint64_t u64;
+    int8_t i8;
+    int16_t i16;
+    int32_t i32;
+    int64_t i64;
+    float f;
+    double d;
+};
 
 #define READ_CAST_BLOCK_OR_RETURN_EOF(cb, uk, n) \
-    union msgpack_buffer_cast_block_t* cb = msgpack_buffer_read_cast_block(UNPACKER_BUFFER_(uk), n); \
-    if(cb == NULL) { \
+    union msgpack_buffer_cast_block_t cb; \
+    if (!msgpack_buffer_read_all(UNPACKER_BUFFER_(uk), (char *)&cb.buffer, n)) { \
         return PRIMITIVE_EOF; \
     }
 
 static inline bool is_reading_map_key(msgpack_unpacker_t* uk)
 {
-    if(uk->stack_depth > 0) {
-        msgpack_unpacker_stack_t* top = _msgpack_unpacker_stack_top(uk);
+    if(uk->stack.depth > 0) {
+        msgpack_unpacker_stack_entry_t* top = _msgpack_unpacker_stack_entry_top(uk);
         if(top->type == STACK_TYPE_MAP_KEY) {
             return true;
         }
@@ -273,7 +336,8 @@ static int read_raw_body_cont(msgpack_unpacker_t* uk)
     size_t length = uk->reading_raw_remaining;
 
     if(uk->reading_raw == Qnil) {
-        uk->reading_raw = rb_str_buf_new(length);
+        size_t buffered_size = msgpack_buffer_all_readable_size(UNPACKER_BUFFER_(uk));
+        uk->reading_raw = rb_str_buf_new(length > buffered_size ? buffered_size : length);
     }
 
     do {
@@ -314,20 +378,15 @@ static inline int read_raw_body_begin(msgpack_unpacker_t* uk, int raw_type)
             reset_head_byte(uk);
             uk->reading_raw_remaining = 0;
 
-            msgpack_unpacker_stack_t* stack = uk->stack;
-            size_t stack_depth = uk->stack_depth;
-            size_t stack_capacity = uk->stack_capacity;
-
-            uk->stack = _msgpack_unpacker_new_stack();
-            uk->stack_depth = 0;
-            uk->stack_capacity = MSGPACK_UNPACKER_STACK_CAPACITY;
-
-            obj = rb_funcall(proc, s_call, 1, uk->buffer.owner);
+            _msgpack_unpacker_stack_push(uk, STACK_TYPE_RECURSIVE, 1, Qnil);
+            int raised;
+            obj = protected_proc_call(proc, 1, &uk->self, &raised);
+            msgpack_unpacker_stack_pop(uk);
 
-            _msgpack_unpacker_free_stack(uk->stack);
-            uk->stack = stack;
-            uk->stack_depth = stack_depth;
-            uk->stack_capacity = stack_capacity;
+            if (raised) {
+                uk->last_object = rb_errinfo();
+                return PRIMITIVE_RECURSIVE_RAISED;
+            }
 
             return object_complete(uk, obj);
         }
@@ -337,15 +396,32 @@ static inline int read_raw_body_begin(msgpack_unpacker_t* uk, int raw_type)
     size_t length = uk->reading_raw_remaining;
     if(length <= msgpack_buffer_top_readable_size(UNPACKER_BUFFER_(uk))) {
         int ret;
-        if ((uk->optimized_symbol_ext_type && uk->symbol_ext_type == raw_type) || (uk->symbolize_keys && is_reading_map_key(uk))) {
+        if ((uk->optimized_symbol_ext_type && uk->symbol_ext_type == raw_type)) {
             VALUE symbol = msgpack_buffer_read_top_as_symbol(UNPACKER_BUFFER_(uk), length, raw_type != RAW_TYPE_BINARY);
             ret = object_complete_symbol(uk, symbol);
+        } else if (is_reading_map_key(uk) && raw_type == RAW_TYPE_STRING) {
+           /* don't use zerocopy for hash keys but get a frozen string directly
+            * because rb_hash_aset freezes keys and it causes copying */
+            VALUE key;
+            if (uk->symbolize_keys) {
+                if (uk->use_key_cache) {
+                    key = msgpack_buffer_read_top_as_interned_symbol(UNPACKER_BUFFER_(uk), &uk->key_cache, length);
+                } else {
+                    key = msgpack_buffer_read_top_as_symbol(UNPACKER_BUFFER_(uk), length, true);
+                }
+                ret = object_complete_symbol(uk, key);
+            } else {
+                if (uk->use_key_cache) {
+                    key = msgpack_buffer_read_top_as_interned_string(UNPACKER_BUFFER_(uk), &uk->key_cache, length);
+                } else {
+                    key = msgpack_buffer_read_top_as_string(UNPACKER_BUFFER_(uk), length, true, true);
+                }
+
+                ret = object_complete(uk, key);
+            }
         } else {
             bool will_freeze = uk->freeze;
             if(raw_type == RAW_TYPE_STRING || raw_type == RAW_TYPE_BINARY) {
-               /* don't use zerocopy for hash keys but get a frozen string directly
-                * because rb_hash_aset freezes keys and it causes copying */
-                will_freeze = will_freeze || is_reading_map_key(uk);
                 VALUE string = msgpack_buffer_read_top_as_string(UNPACKER_BUFFER_(uk), length, will_freeze, raw_type == RAW_TYPE_STRING);
                 ret = object_complete(uk, string);
             } else {
@@ -390,14 +466,14 @@ static int read_primitive(msgpack_unpacker_t* uk)
         if(count == 0) {
             return object_complete(uk, rb_ary_new());
         }
-        return _msgpack_unpacker_stack_push(uk, STACK_TYPE_ARRAY, count, rb_ary_new2(count));
+        return _msgpack_unpacker_stack_push(uk, STACK_TYPE_ARRAY, count, rb_ary_new2(initial_buffer_size(count)));
 
     SWITCH_RANGE(b, 0x80, 0x8f)  // FixMap
         int count = b & 0x0f;
         if(count == 0) {
             return object_complete(uk, rb_hash_new());
         }
-        return _msgpack_unpacker_stack_push(uk, STACK_TYPE_MAP_KEY, count*2, rb_hash_new_capa(count));
+        return _msgpack_unpacker_stack_push(uk, STACK_TYPE_MAP_KEY, count*2, rb_hash_new_capa(initial_buffer_size(count)));
 
     SWITCH_RANGE(b, 0xc0, 0xdf)  // Variable
         switch(b) {
@@ -415,8 +491,8 @@ static int read_primitive(msgpack_unpacker_t* uk)
         case 0xc7: // ext 8
             {
                 READ_CAST_BLOCK_OR_RETURN_EOF(cb, uk, 2);
-                uint8_t length = cb->u8;
-                int ext_type = (signed char) cb->buffer[1];
+                uint8_t length = cb.u8;
+                int ext_type = (signed char) cb.buffer[1];
                 if(length == 0) {
                     return object_complete_ext(uk, ext_type, Qnil);
                 }
@@ -427,8 +503,8 @@ static int read_primitive(msgpack_unpacker_t* uk)
         case 0xc8: // ext 16
             {
                 READ_CAST_BLOCK_OR_RETURN_EOF(cb, uk, 3);
-                uint16_t length = _msgpack_be16(cb->u16);
-                int ext_type = (signed char) cb->buffer[2];
+                uint16_t length = _msgpack_be16(cb.u16);
+                int ext_type = (signed char) cb.buffer[2];
                 if(length == 0) {
                     return object_complete_ext(uk, ext_type, Qnil);
                 }
@@ -439,8 +515,8 @@ static int read_primitive(msgpack_unpacker_t* uk)
         case 0xc9: // ext 32
             {
                 READ_CAST_BLOCK_OR_RETURN_EOF(cb, uk, 5);
-                uint32_t length = _msgpack_be32(cb->u32);
-                int ext_type = (signed char) cb->buffer[4];
+                uint32_t length = _msgpack_be32(cb.u32);
+                int ext_type = (signed char) cb.buffer[4];
                 if(length == 0) {
                     return object_complete_ext(uk, ext_type, Qnil);
                 }
@@ -451,77 +527,77 @@ static int read_primitive(msgpack_unpacker_t* uk)
         case 0xca:  // float
             {
                 READ_CAST_BLOCK_OR_RETURN_EOF(cb, uk, 4);
-                cb->u32 = _msgpack_be_float(cb->u32);
-                return object_complete(uk, rb_float_new(cb->f));
+                cb.u32 = _msgpack_be_float(cb.u32);
+                return object_complete(uk, rb_float_new(cb.f));
             }
 
         case 0xcb:  // double
             {
                 READ_CAST_BLOCK_OR_RETURN_EOF(cb, uk, 8);
-                cb->u64 = _msgpack_be_double(cb->u64);
-                return object_complete(uk, rb_float_new(cb->d));
+                cb.u64 = _msgpack_be_double(cb.u64);
+                return object_complete(uk, rb_float_new(cb.d));
             }
 
         case 0xcc:  // unsigned int  8
             {
                 READ_CAST_BLOCK_OR_RETURN_EOF(cb, uk, 1);
-                uint8_t u8 = cb->u8;
+                uint8_t u8 = cb.u8;
                 return object_complete(uk, INT2NUM((int)u8));
             }
 
         case 0xcd:  // unsigned int 16
             {
                 READ_CAST_BLOCK_OR_RETURN_EOF(cb, uk, 2);
-                uint16_t u16 = _msgpack_be16(cb->u16);
+                uint16_t u16 = _msgpack_be16(cb.u16);
                 return object_complete(uk, INT2NUM((int)u16));
             }
 
         case 0xce:  // unsigned int 32
             {
                 READ_CAST_BLOCK_OR_RETURN_EOF(cb, uk, 4);
-                uint32_t u32 = _msgpack_be32(cb->u32);
+                uint32_t u32 = _msgpack_be32(cb.u32);
                 return object_complete(uk, ULONG2NUM(u32)); // long at least 32 bits
             }
 
         case 0xcf:  // unsigned int 64
             {
                 READ_CAST_BLOCK_OR_RETURN_EOF(cb, uk, 8);
-                uint64_t u64 = _msgpack_be64(cb->u64);
+                uint64_t u64 = _msgpack_be64(cb.u64);
                 return object_complete(uk, rb_ull2inum(u64));
             }
 
         case 0xd0:  // signed int  8
             {
                 READ_CAST_BLOCK_OR_RETURN_EOF(cb, uk, 1);
-                int8_t i8 = cb->i8;
+                int8_t i8 = cb.i8;
                 return object_complete(uk, INT2NUM((int)i8));
             }
 
         case 0xd1:  // signed int 16
             {
                 READ_CAST_BLOCK_OR_RETURN_EOF(cb, uk, 2);
-                int16_t i16 = _msgpack_be16(cb->i16);
+                int16_t i16 = _msgpack_be16(cb.i16);
                 return object_complete(uk, INT2NUM((int)i16));
             }
 
         case 0xd2:  // signed int 32
             {
                 READ_CAST_BLOCK_OR_RETURN_EOF(cb, uk, 4);
-                int32_t i32 = _msgpack_be32(cb->i32);
+                int32_t i32 = _msgpack_be32(cb.i32);
                 return object_complete(uk, LONG2NUM(i32)); // long at least 32 bits
             }
 
         case 0xd3:  // signed int 64
             {
                 READ_CAST_BLOCK_OR_RETURN_EOF(cb, uk, 8);
-                int64_t i64 = _msgpack_be64(cb->i64);
+                int64_t i64 = _msgpack_be64(cb.i64);
                 return object_complete(uk, rb_ll2inum(i64));
             }
 
         case 0xd4:  // fixext 1
             {
                 READ_CAST_BLOCK_OR_RETURN_EOF(cb, uk, 1);
-                int ext_type = cb->i8;
+                int ext_type = cb.i8;
                 uk->reading_raw_remaining = 1;
                 return read_raw_body_begin(uk, ext_type);
             }
@@ -529,7 +605,7 @@ static int read_primitive(msgpack_unpacker_t* uk)
         case 0xd5:  // fixext 2
             {
                 READ_CAST_BLOCK_OR_RETURN_EOF(cb, uk, 1);
-                int ext_type = cb->i8;
+                int ext_type = cb.i8;
                 uk->reading_raw_remaining = 2;
                 return read_raw_body_begin(uk, ext_type);
             }
@@ -537,7 +613,7 @@ static int read_primitive(msgpack_unpacker_t* uk)
         case 0xd6:  // fixext 4
             {
                 READ_CAST_BLOCK_OR_RETURN_EOF(cb, uk, 1);
-                int ext_type = cb->i8;
+                int ext_type = cb.i8;
                 uk->reading_raw_remaining = 4;
                 return read_raw_body_begin(uk, ext_type);
             }
@@ -545,7 +621,7 @@ static int read_primitive(msgpack_unpacker_t* uk)
         case 0xd7:  // fixext 8
             {
                 READ_CAST_BLOCK_OR_RETURN_EOF(cb, uk, 1);
-                int ext_type = cb->i8;
+                int ext_type = cb.i8;
                 uk->reading_raw_remaining = 8;
                 return read_raw_body_begin(uk, ext_type);
             }
@@ -553,7 +629,7 @@ static int read_primitive(msgpack_unpacker_t* uk)
         case 0xd8:  // fixext 16
             {
                 READ_CAST_BLOCK_OR_RETURN_EOF(cb, uk, 1);
-                int ext_type = cb->i8;
+                int ext_type = cb.i8;
                 uk->reading_raw_remaining = 16;
                 return read_raw_body_begin(uk, ext_type);
             }
@@ -562,7 +638,7 @@ static int read_primitive(msgpack_unpacker_t* uk)
         case 0xd9:  // raw 8 / str 8
             {
                 READ_CAST_BLOCK_OR_RETURN_EOF(cb, uk, 1);
-                uint8_t count = cb->u8;
+                uint8_t count = cb.u8;
                 /* read_raw_body_begin sets uk->reading_raw */
                 uk->reading_raw_remaining = count;
                 return read_raw_body_begin(uk, RAW_TYPE_STRING);
@@ -571,7 +647,7 @@ static int read_primitive(msgpack_unpacker_t* uk)
         case 0xda:  // raw 16 / str 16
             {
                 READ_CAST_BLOCK_OR_RETURN_EOF(cb, uk, 2);
-                uint16_t count = _msgpack_be16(cb->u16);
+                uint16_t count = _msgpack_be16(cb.u16);
                 /* read_raw_body_begin sets uk->reading_raw */
                 uk->reading_raw_remaining = count;
                 return read_raw_body_begin(uk, RAW_TYPE_STRING);
@@ -580,7 +656,7 @@ static int read_primitive(msgpack_unpacker_t* uk)
         case 0xdb:  // raw 32 / str 32
             {
                 READ_CAST_BLOCK_OR_RETURN_EOF(cb, uk, 4);
-                uint32_t count = _msgpack_be32(cb->u32);
+                uint32_t count = _msgpack_be32(cb.u32);
                 /* read_raw_body_begin sets uk->reading_raw */
                 uk->reading_raw_remaining = count;
                 return read_raw_body_begin(uk, RAW_TYPE_STRING);
@@ -589,7 +665,7 @@ static int read_primitive(msgpack_unpacker_t* uk)
         case 0xc4:  // bin 8
             {
                 READ_CAST_BLOCK_OR_RETURN_EOF(cb, uk, 1);
-                uint8_t count = cb->u8;
+                uint8_t count = cb.u8;
                 /* read_raw_body_begin sets uk->reading_raw */
                 uk->reading_raw_remaining = count;
                 return read_raw_body_begin(uk, RAW_TYPE_BINARY);
@@ -598,7 +674,7 @@ static int read_primitive(msgpack_unpacker_t* uk)
         case 0xc5:  // bin 16
             {
                 READ_CAST_BLOCK_OR_RETURN_EOF(cb, uk, 2);
-                uint16_t count = _msgpack_be16(cb->u16);
+                uint16_t count = _msgpack_be16(cb.u16);
                 /* read_raw_body_begin sets uk->reading_raw */
                 uk->reading_raw_remaining = count;
                 return read_raw_body_begin(uk, RAW_TYPE_BINARY);
@@ -607,7 +683,7 @@ static int read_primitive(msgpack_unpacker_t* uk)
         case 0xc6:  // bin 32
             {
                 READ_CAST_BLOCK_OR_RETURN_EOF(cb, uk, 4);
-                uint32_t count = _msgpack_be32(cb->u32);
+                uint32_t count = _msgpack_be32(cb.u32);
                 /* read_raw_body_begin sets uk->reading_raw */
                 uk->reading_raw_remaining = count;
                 return read_raw_body_begin(uk, RAW_TYPE_BINARY);
@@ -616,41 +692,41 @@ static int read_primitive(msgpack_unpacker_t* uk)
         case 0xdc:  // array 16
             {
                 READ_CAST_BLOCK_OR_RETURN_EOF(cb, uk, 2);
-                uint16_t count = _msgpack_be16(cb->u16);
+                uint16_t count = _msgpack_be16(cb.u16);
                 if(count == 0) {
                     return object_complete(uk, rb_ary_new());
                 }
-                return _msgpack_unpacker_stack_push(uk, STACK_TYPE_ARRAY, count, rb_ary_new2(count));
+                return _msgpack_unpacker_stack_push(uk, STACK_TYPE_ARRAY, count, rb_ary_new2(initial_buffer_size(count)));
             }
 
         case 0xdd:  // array 32
             {
                 READ_CAST_BLOCK_OR_RETURN_EOF(cb, uk, 4);
-                uint32_t count = _msgpack_be32(cb->u32);
+                uint32_t count = _msgpack_be32(cb.u32);
                 if(count == 0) {
                     return object_complete(uk, rb_ary_new());
                 }
-                return _msgpack_unpacker_stack_push(uk, STACK_TYPE_ARRAY, count, rb_ary_new2(count));
+                return _msgpack_unpacker_stack_push(uk, STACK_TYPE_ARRAY, count, rb_ary_new2(initial_buffer_size(count)));
             }
 
         case 0xde:  // map 16
             {
                 READ_CAST_BLOCK_OR_RETURN_EOF(cb, uk, 2);
-                uint16_t count = _msgpack_be16(cb->u16);
+                uint16_t count = _msgpack_be16(cb.u16);
                 if(count == 0) {
                     return object_complete(uk, rb_hash_new());
                 }
-                return _msgpack_unpacker_stack_push(uk, STACK_TYPE_MAP_KEY, count*2, rb_hash_new_capa(count));
+                return _msgpack_unpacker_stack_push(uk, STACK_TYPE_MAP_KEY, count*2, rb_hash_new_capa(initial_buffer_size(count)));
             }
 
         case 0xdf:  // map 32
             {
                 READ_CAST_BLOCK_OR_RETURN_EOF(cb, uk, 4);
-                uint32_t count = _msgpack_be32(cb->u32);
+                uint32_t count = _msgpack_be32(cb.u32);
                 if(count == 0) {
                     return object_complete(uk, rb_hash_new());
                 }
-                return _msgpack_unpacker_stack_push(uk, STACK_TYPE_MAP_KEY, count*2, rb_hash_new_capa(count));
+                return _msgpack_unpacker_stack_push(uk, STACK_TYPE_MAP_KEY, count*2, rb_hash_new_capa(initial_buffer_size(count)));
             }
 
         default:
@@ -676,12 +752,12 @@ int msgpack_unpacker_read_array_header(msgpack_unpacker_t* uk, uint32_t* result_
     } else if(b == 0xdc) {
         /* array 16 */
         READ_CAST_BLOCK_OR_RETURN_EOF(cb, uk, 2);
-        *result_size = _msgpack_be16(cb->u16);
+        *result_size = _msgpack_be16(cb.u16);
 
     } else if(b == 0xdd) {
         /* array 32 */
         READ_CAST_BLOCK_OR_RETURN_EOF(cb, uk, 4);
-        *result_size = _msgpack_be32(cb->u32);
+        *result_size = _msgpack_be32(cb.u32);
 
     } else {
         return PRIMITIVE_UNEXPECTED_TYPE;
@@ -704,12 +780,12 @@ int msgpack_unpacker_read_map_header(msgpack_unpacker_t* uk, uint32_t* result_si
     } else if(b == 0xde) {
         /* map 16 */
         READ_CAST_BLOCK_OR_RETURN_EOF(cb, uk, 2);
-        *result_size = _msgpack_be16(cb->u16);
+        *result_size = _msgpack_be16(cb.u16);
 
     } else if(b == 0xdf) {
         /* map 32 */
         READ_CAST_BLOCK_OR_RETURN_EOF(cb, uk, 4);
-        *result_size = _msgpack_be32(cb->u32);
+        *result_size = _msgpack_be32(cb.u32);
 
     } else {
         return PRIMITIVE_UNEXPECTED_TYPE;
@@ -721,9 +797,15 @@ int msgpack_unpacker_read_map_header(msgpack_unpacker_t* uk, uint32_t* result_si
 
 int msgpack_unpacker_read(msgpack_unpacker_t* uk, size_t target_stack_depth)
 {
+    STACK_INIT(uk);
+
     while(true) {
         int r = read_primitive(uk);
         if(r < 0) {
+            if (r != PRIMITIVE_EOF) {
+                // We keep the stack on EOF as the parsing may be resumed.
+                STACK_FREE(uk);
+            }
             return r;
         }
         if(r == PRIMITIVE_CONTAINER_START) {
@@ -732,12 +814,13 @@ int msgpack_unpacker_read(msgpack_unpacker_t* uk, size_t target_stack_depth)
         /* PRIMITIVE_OBJECT_COMPLETE */
 
         if(msgpack_unpacker_stack_is_empty(uk)) {
+            STACK_FREE(uk);
             return PRIMITIVE_OBJECT_COMPLETE;
         }
 
         container_completed:
         {
-            msgpack_unpacker_stack_t* top = _msgpack_unpacker_stack_top(uk);
+            msgpack_unpacker_stack_entry_t* top = _msgpack_unpacker_stack_entry_top(uk);
             switch(top->type) {
             case STACK_TYPE_ARRAY:
                 rb_ary_push(top->object, uk->last_object);
@@ -755,12 +838,16 @@ int msgpack_unpacker_read(msgpack_unpacker_t* uk, size_t target_stack_depth)
                 }
                 top->type = STACK_TYPE_MAP_KEY;
                 break;
+            case STACK_TYPE_RECURSIVE:
+                STACK_FREE(uk);
+                return PRIMITIVE_OBJECT_COMPLETE;
             }
             size_t count = --top->count;
 
             if(count == 0) {
                 object_complete(uk, top->object);
                 if(msgpack_unpacker_stack_pop(uk) <= target_stack_depth) {
+                    STACK_FREE(uk);
                     return PRIMITIVE_OBJECT_COMPLETE;
                 }
                 goto container_completed;
@@ -771,9 +858,12 @@ int msgpack_unpacker_read(msgpack_unpacker_t* uk, size_t target_stack_depth)
 
 int msgpack_unpacker_skip(msgpack_unpacker_t* uk, size_t target_stack_depth)
 {
+    STACK_INIT(uk);
+
     while(true) {
         int r = read_primitive(uk);
         if(r < 0) {
+            STACK_FREE(uk);
             return r;
         }
         if(r == PRIMITIVE_CONTAINER_START) {
@@ -782,12 +872,13 @@ int msgpack_unpacker_skip(msgpack_unpacker_t* uk, size_t target_stack_depth)
         /* PRIMITIVE_OBJECT_COMPLETE */
 
         if(msgpack_unpacker_stack_is_empty(uk)) {
+            STACK_FREE(uk);
             return PRIMITIVE_OBJECT_COMPLETE;
         }
 
         container_completed:
         {
-            msgpack_unpacker_stack_t* top = _msgpack_unpacker_stack_top(uk);
+            msgpack_unpacker_stack_entry_t* top = _msgpack_unpacker_stack_entry_top(uk);
 
             /* this section optimized out */
             // TODO object_complete still creates objects which should be optimized out
@@ -797,6 +888,7 @@ int msgpack_unpacker_skip(msgpack_unpacker_t* uk, size_t target_stack_depth)
             if(count == 0) {
                 object_complete(uk, Qnil);
                 if(msgpack_unpacker_stack_pop(uk) <= target_stack_depth) {
+                    STACK_FREE(uk);
                     return PRIMITIVE_OBJECT_COMPLETE;
                 }
                 goto container_completed;