Using hash value stored in ecma-string for lcache search; comparing string faster but with possible false negatives while performing lcache lookup.

This commit is contained in:
Ruben Ayrapetyan 2014-11-28 20:27:27 +03:00
parent fe7e34d93e
commit 3b29280b79
4 changed files with 139 additions and 95 deletions

View File

@ -743,7 +743,7 @@ typedef enum
/**
* ECMA string hash
*/
typedef uint16_t ecma_string_hash_t;
typedef uint8_t ecma_string_hash_t;
/**
* Number of string's last characters to use for hash calculation
@ -758,13 +758,13 @@ typedef struct
/** Reference counter for the string */
unsigned int refs : CONFIG_ECMA_REFERENCE_COUNTER_WIDTH;
/** Where the string's data is placed (ecma_string_container_t) */
unsigned int container : 3;
/** Flag indicating whether the string descriptor is placed
* in a stack variable (not in the heap) */
unsigned int is_stack_var : 1;
/** Where the string's data is placed (ecma_string_container_t) */
uint8_t container;
/** Hash of the string (calculated from two last characters of the string) */
ecma_string_hash_t hash;
@ -794,6 +794,9 @@ typedef struct
/** Identifier of magic string */
ecma_magic_string_id_t magic_string_id;
/** For zeroing and comparison in some cases */
uint32_t common_field;
} u;
} ecma_string_t;

View File

@ -308,6 +308,7 @@ ecma_init_ecma_string_from_lit_index (ecma_string_t *string_p, /**< descriptor t
string_p->container = ECMA_STRING_CONTAINER_LIT_TABLE;
string_p->hash = lit.data.lp.hash;
string_p->u.common_field = 0;
string_p->u.lit_index = lit_index;
} /* ecma_init_ecma_string_from_lit_index */
@ -331,6 +332,7 @@ ecma_init_ecma_string_from_magic_string_id (ecma_string_t *string_p, /**< descri
string_p->hash = ecma_chars_buffer_calc_hash_last_chars (ecma_get_magic_string_zt (magic_string_id),
ecma_magic_string_lengths [magic_string_id]);
string_p->u.common_field = 0;
string_p->u.magic_string_id = magic_string_id;
} /* ecma_init_ecma_string_from_magic_string_id */
@ -366,6 +368,7 @@ ecma_new_ecma_string (const ecma_char_t *string_p) /**< zero-terminated string *
string_desc_p->container = ECMA_STRING_CONTAINER_HEAP_CHUNKS;
string_desc_p->hash = ecma_chars_buffer_calc_hash_last_chars (string_p, length);
string_desc_p->u.common_field = 0;
ecma_collection_header_t *collection_p = ecma_new_chars_collection (string_p, length);
ECMA_SET_NON_NULL_POINTER (string_desc_p->u.collection_cp, collection_p);
@ -384,7 +387,6 @@ ecma_new_ecma_string_from_uint32 (uint32_t uint32_number) /**< UInt32-represente
string_desc_p->refs = 1;
string_desc_p->is_stack_var = false;
string_desc_p->container = ECMA_STRING_CONTAINER_UINT32_IN_DESC;
string_desc_p->u.uint32_number = uint32_number;
uint32_t last_two_digits = uint32_number % 100;
uint32_t digit_pl = last_two_digits / 10;
@ -414,6 +416,9 @@ ecma_new_ecma_string_from_uint32 (uint32_t uint32_number) /**< UInt32-represente
ecma_zt_string_length (char_buf)));
#endif /* !JERRY_NDEBUG */
string_desc_p->u.common_field = 0;
string_desc_p->u.uint32_number = uint32_number;
return string_desc_p;
} /* ecma_new_ecma_string_from_uint32 */
@ -442,6 +447,7 @@ ecma_new_ecma_string_from_number (ecma_number_t num) /**< ecma-number */
string_desc_p->container = ECMA_STRING_CONTAINER_HEAP_NUMBER;
string_desc_p->hash = ecma_chars_buffer_calc_hash_last_chars (str_buf, length);
string_desc_p->u.common_field = 0;
ecma_number_t *num_p = ecma_alloc_number ();
*num_p = num;
ECMA_SET_POINTER (string_desc_p->u.number_cp, num_p);
@ -529,15 +535,17 @@ ecma_concat_ecma_strings (ecma_string_t *string1_p, /**< first ecma-string */
int64_t length = (int64_t) str1_len + (int64_t) str2_len;
if (length > ECMA_STRING_MAX_CONCATENATION_LENGTH)
{
jerry_exit (ERR_OUT_OF_MEMORY);
}
ecma_string_t* string_desc_p = ecma_alloc_string ();
string_desc_p->refs = 1;
string_desc_p->is_stack_var = false;
string_desc_p->container = ECMA_STRING_CONTAINER_CONCATENATION;
if (length > ECMA_STRING_MAX_CONCATENATION_LENGTH)
{
jerry_exit (ERR_OUT_OF_MEMORY);
}
string_desc_p->u.common_field = 0;
string1_p = ecma_copy_or_ref_ecma_string (string1_p);
string2_p = ecma_copy_or_ref_ecma_string (string2_p);
@ -954,6 +962,28 @@ static bool __noinline
ecma_compare_ecma_strings_longpath (const ecma_string_t *string1_p, /* ecma-string */
const ecma_string_t *string2_p) /* ecma-string */
{
if (string1_p->container == string2_p->container)
{
if (string1_p->container == ECMA_STRING_CONTAINER_LIT_TABLE)
{
JERRY_ASSERT (string1_p->u.lit_index != string2_p->u.lit_index);
return false;
}
else if (string1_p->container == ECMA_STRING_CONTAINER_MAGIC_STRING)
{
JERRY_ASSERT (string1_p->u.magic_string_id != string2_p->u.magic_string_id);
return false;
}
else if (string1_p->container == ECMA_STRING_CONTAINER_UINT32_IN_DESC)
{
JERRY_ASSERT (string1_p->u.uint32_number != string2_p->u.uint32_number);
return false;
}
}
const int32_t string1_len = ecma_string_get_length (string1_p);
const int32_t string2_len = ecma_string_get_length (string2_p);
@ -1000,10 +1030,22 @@ ecma_compare_ecma_strings_longpath (const ecma_string_t *string1_p, /* ecma-stri
break;
}
case ECMA_STRING_CONTAINER_LIT_TABLE:
{
JERRY_ASSERT (string1_p->u.lit_index != string2_p->u.lit_index);
return false;
}
case ECMA_STRING_CONTAINER_MAGIC_STRING:
{
JERRY_ASSERT (string1_p->u.magic_string_id != string2_p->u.magic_string_id);
return false;
}
case ECMA_STRING_CONTAINER_UINT32_IN_DESC:
{
JERRY_UNREACHABLE ();
JERRY_ASSERT (string1_p->u.uint32_number != string2_p->u.uint32_number);
return false;
}
}
}
@ -1036,6 +1078,29 @@ ecma_compare_ecma_strings_longpath (const ecma_string_t *string1_p, /* ecma-stri
return is_equal;
} /* ecma_compare_ecma_strings_longpath */
/**
* Compare ecma-string to ecma-string if they're hashes are equal
*
* @return true - if strings are equal;
* false - may be.
*/
bool
ecma_compare_ecma_strings_equal_hashes (const ecma_string_t *string1_p, /* ecma-string */
const ecma_string_t *string2_p) /* ecma-string */
{
JERRY_ASSERT (string1_p->hash == string2_p->hash);
if (string1_p->container == string2_p->container
&& string1_p->u.common_field == string2_p->u.common_field)
{
return true;
}
else
{
return false;
}
} /* ecma_compare_ecma_strings_equal_hashes */
/**
* Compare ecma-string to ecma-string
*
@ -1048,28 +1113,23 @@ ecma_compare_ecma_strings (const ecma_string_t *string1_p, /* ecma-string */
{
JERRY_ASSERT (string1_p != NULL && string2_p != NULL);
if (string1_p->hash != string2_p->hash)
const bool is_equal_hashes = (string1_p->hash == string2_p->hash);
if (!is_equal_hashes)
{
return false;
}
const bool is_equal_containers = (string1_p->container == string2_p->container);
const bool is_equal_fields = (string1_p->u.common_field == string2_p->u.common_field);
if (string1_p->container == string2_p->container)
if (is_equal_containers && is_equal_fields)
{
if (likely (string1_p->container == ECMA_STRING_CONTAINER_LIT_TABLE))
{
return (string1_p->u.lit_index == string2_p->u.lit_index);
}
else if (likely (string1_p->container == ECMA_STRING_CONTAINER_MAGIC_STRING))
{
return (string1_p->u.magic_string_id == string2_p->u.magic_string_id);
}
else if (string1_p->container == ECMA_STRING_CONTAINER_UINT32_IN_DESC)
{
return (string1_p->u.uint32_number == string2_p->u.uint32_number);
}
return true;
}
else
{
return ecma_compare_ecma_strings_longpath (string1_p, string2_p);
}
return ecma_compare_ecma_strings_longpath (string1_p, string2_p);
} /* ecma_compare_ecma_strings */
/**
@ -1614,39 +1674,13 @@ ecma_is_string_magic (const ecma_string_t *string_p, /**< ecma-string */
/**
* Try to calculate hash of the ecma-string
*
* Note:
* Not all ecma-string containers provide ability to calculate hash.
* For now, only strings stored in literal table support calculating of hash.
*
* Warning:
* Anyway, if ecma-string is hashable, for given length of hash,
* the hash values should always be equal for equal strings
* irrespective of the strings' container type.
*
* If a ecma-string stored in a container of some type is hashable,
* then the ecma-string should always be hashable if stored in container
* of the same type.
*
* @return true - if hash was calculated,
* false - otherwise (the string is not hashable).
* @return calculated hash
*/
bool
ecma_string_try_hash (const ecma_string_t *string_p, /**< ecma-string to calculate hash for */
uint32_t hash_length_bits, /**< length of hash value, in bits */
uint32_t *out_hash_p) /**< out: hash value, if calculated */
ecma_string_hash_t
ecma_string_hash (const ecma_string_t *string_p) /**< ecma-string to calculate hash for */
{
JERRY_ASSERT (hash_length_bits < sizeof (uint32_t) * JERRY_BITSINBYTE);
uint32_t hash_mask = ((1u << hash_length_bits) - 1);
if (string_p->container == ECMA_STRING_CONTAINER_LIT_TABLE)
{
*out_hash_p = (string_p->u.lit_index) & hash_mask;
return true;
}
return false;
return (string_p->hash);
} /* ecma_string_try_hash */
/**
@ -1666,8 +1700,9 @@ ecma_chars_buffer_calc_hash_last_chars (const ecma_char_t *chars, /**< character
uint32_t t1 = (uint32_t) char1 + (uint32_t) char2;
uint32_t t2 = t1 * 0x24418b66;
uint32_t t3 = (t2 >> 16) ^ (t2 & 0xffffu);
uint32_t t4 = (t3 >> 8) ^ (t3 & 0xffu);
return (ecma_string_hash_t) t3;
return (ecma_string_hash_t) t4;
} /* ecma_chars_buffer_calc_hash_last_chars */
/**

View File

@ -137,6 +137,8 @@ extern ecma_number_t ecma_string_to_number (const ecma_string_t *str_p);
extern ssize_t ecma_string_to_zt_string (const ecma_string_t *string_desc_p,
ecma_char_t *buffer_p,
ssize_t buffer_size);
extern bool ecma_compare_ecma_strings_equal_hashes (const ecma_string_t *string1_p,
const ecma_string_t *string2_p);
extern bool ecma_compare_ecma_strings (const ecma_string_t *string1_p,
const ecma_string_t *string2_p);
extern bool ecma_compare_ecma_strings_relational (const ecma_string_t *string1_p,
@ -156,7 +158,7 @@ extern const ecma_char_t* ecma_get_magic_string_zt (ecma_magic_string_id_t id);
extern ecma_string_t* ecma_get_magic_string (ecma_magic_string_id_t id);
extern bool ecma_is_string_magic (const ecma_string_t *string_p, ecma_magic_string_id_t *out_id_p);
extern bool ecma_is_zt_string_magic (const ecma_char_t *zt_string_p, ecma_magic_string_id_t *out_id_p);
extern bool ecma_string_try_hash (const ecma_string_t *string_p, uint32_t hash_length_bits, uint32_t *out_hash_p);
extern ecma_string_hash_t ecma_string_hash (const ecma_string_t *string_p);
extern ecma_string_hash_t ecma_chars_buffer_calc_hash_last_chars (const ecma_char_t *chars, ecma_length_t length);
/* ecma-helpers-number.c */

View File

@ -44,7 +44,7 @@ typedef struct
/**
* LCache hash value length, in bits
*/
#define ECMA_LCACHE_HASH_BITS (8)
#define ECMA_LCACHE_HASH_BITS (sizeof (ecma_string_hash_t) * JERRY_BITSINBYTE)
/**
* Number of rows in LCache's hash table
@ -142,15 +142,28 @@ ecma_lcache_insert (ecma_object_t *object_p, /**< object */
JERRY_ASSERT (object_p != NULL);
JERRY_ASSERT (prop_name_p != NULL);
uint32_t hash_key;
if (!ecma_string_try_hash (prop_name_p, ECMA_LCACHE_HASH_BITS, &hash_key))
{
return;
}
ecma_string_hash_t hash_key = ecma_string_hash (prop_name_p);
if (prop_p != NULL)
{
if (unlikely (ecma_is_property_lcached (prop_p)))
{
uint16_t prop_cp;
ECMA_SET_NON_NULL_POINTER (prop_cp, prop_p);
int32_t entry_index;
for (entry_index = 0; entry_index < ECMA_LCACHE_HASH_ROW_LENGTH; entry_index++)
{
if (ecma_lcache_hash_table[hash_key][entry_index].prop_cp == prop_cp)
{
break;
}
}
JERRY_ASSERT (entry_index != ECMA_LCACHE_HASH_ROW_LENGTH);
ecma_lcache_invalidate_entry (&ecma_lcache_hash_table[hash_key][entry_index]);
}
JERRY_ASSERT (!ecma_is_property_lcached (prop_p));
ecma_set_property_lcached (prop_p, true);
}
@ -186,7 +199,7 @@ ecma_lcache_insert (ecma_object_t *object_p, /**< object */
* Lookup property in the LCache
*
* @return true - if (object, property name) pair is registered in LCache,
* false - otherwise.
* false - probably, not registered.
*/
bool
ecma_lcache_lookup (ecma_object_t *object_p, /**< object */
@ -198,28 +211,30 @@ ecma_lcache_lookup (ecma_object_t *object_p, /**< object */
* if return value is false,
* then the output parameter is not set */
{
uint32_t hash_key;
if (!ecma_string_try_hash (prop_name_p, ECMA_LCACHE_HASH_BITS, &hash_key))
{
return false;
}
ecma_string_hash_t hash_key = ecma_string_hash (prop_name_p);
unsigned int object_cp;
ECMA_SET_NON_NULL_POINTER (object_cp, object_p);
for (uint32_t i = 0; i < ECMA_LCACHE_HASH_ROW_LENGTH; i++)
{
if (ecma_lcache_hash_table[hash_key][i].object_cp == object_cp
&& ecma_compare_ecma_strings (prop_name_p,
ECMA_GET_NON_NULL_POINTER (ecma_lcache_hash_table[hash_key][i].prop_name_cp)))
if (ecma_lcache_hash_table[hash_key][i].object_cp == object_cp)
{
ecma_property_t *prop_p = ECMA_GET_POINTER (ecma_lcache_hash_table[hash_key][i].prop_cp);
JERRY_ASSERT (prop_p == NULL || ecma_is_property_lcached (prop_p));
ecma_string_t *entry_prop_name_p = ECMA_GET_NON_NULL_POINTER (ecma_lcache_hash_table[hash_key][i].prop_name_cp);
*prop_p_p = prop_p;
if (ecma_compare_ecma_strings_equal_hashes (prop_name_p, entry_prop_name_p))
{
ecma_property_t *prop_p = ECMA_GET_POINTER (ecma_lcache_hash_table[hash_key][i].prop_cp);
JERRY_ASSERT (prop_p == NULL || ecma_is_property_lcached (prop_p));
return true;
*prop_p_p = prop_p;
return true;
}
else
{
/* may be equal but it is long to compare it here */
}
}
}
@ -277,23 +292,12 @@ ecma_lcache_invalidate (ecma_object_t *object_p, /**< object */
ECMA_SET_NON_NULL_POINTER (object_cp, object_p);
ECMA_SET_POINTER (prop_cp, prop_p);
uint32_t hash_key;
ecma_string_hash_t hash_key = ecma_string_hash (prop_name_p);
if (!ecma_string_try_hash (prop_name_p, ECMA_LCACHE_HASH_BITS, &hash_key))
{
/* Property's name hash was not computed, so iterating the whole hash table */
for (uint32_t row_index = 0; row_index < ECMA_LCACHE_HASH_ROWS_COUNT; row_index++)
{
ecma_lcache_invalidate_row_for_object_property_pair (row_index, object_cp, prop_cp);
}
}
else
{
/* Property's name has was computed.
* Given (object, property name) pair should be in the row corresponding to computed hash.
*/
ecma_lcache_invalidate_row_for_object_property_pair (hash_key, object_cp, prop_cp);
}
/* Property's name has was computed.
* Given (object, property name) pair should be in the row corresponding to computed hash.
*/
ecma_lcache_invalidate_row_for_object_property_pair (hash_key, object_cp, prop_cp);
} /* ecma_lcache_invalidate */
/**