Commit 99cd1a61 authored by Vasil Dimov's avatar Vasil Dimov

Merge mysql-5.1-innodb -> mysql-5.5-innodb

parents f45f5902 a491492f
CREATE TABLE bug53046_1 (c1 INT PRIMARY KEY) ENGINE=INNODB;
CREATE TABLE bug53046_2 (c2 INT PRIMARY KEY,
FOREIGN KEY (c2) REFERENCES bug53046_1(c1)
ON UPDATE CASCADE ON DELETE CASCADE) ENGINE=INNODB;
INSERT INTO bug53046_1 VALUES (1);
INSERT INTO bug53046_1 SELECT c1+(SELECT MAX(c1) FROM bug53046_1)
FROM bug53046_1;
INSERT INTO bug53046_1 SELECT c1+(SELECT MAX(c1) FROM bug53046_1)
FROM bug53046_1;
INSERT INTO bug53046_1 SELECT c1+(SELECT MAX(c1) FROM bug53046_1)
FROM bug53046_1;
INSERT INTO bug53046_1 SELECT c1+(SELECT MAX(c1) FROM bug53046_1)
FROM bug53046_1;
INSERT INTO bug53046_1 SELECT c1+(SELECT MAX(c1) FROM bug53046_1)
FROM bug53046_1;
INSERT INTO bug53046_2 VALUES (1), (2);
ANALYZE TABLE bug53046_1;
Table Op Msg_type Msg_text
test.bug53046_1 analyze status OK
SHOW TABLE STATUS LIKE 'bug53046_1';
UPDATE bug53046_1 SET c1 = c1 - 1;
DELETE FROM bug53046_1;
INSERT INTO bug53046_1 VALUES (1);
INSERT INTO bug53046_2 VALUES (1);
TRUNCATE TABLE bug53046_2;
DROP TABLE bug53046_2;
DROP TABLE bug53046_1;
#
# http://bugs.mysql.com/53046
# dict_update_statistics_low can still be run concurrently on same table
#
# This is a symbolic test, it would not fail if the bug is present.
# Rather those SQL commands have been used during manual testing under
# UNIV_DEBUG & UNIV_SYNC_DEBUG to test all changed codepaths for locking
# correctness.
#
-- source include/have_innodb.inc
CREATE TABLE bug53046_1 (c1 INT PRIMARY KEY) ENGINE=INNODB;
CREATE TABLE bug53046_2 (c2 INT PRIMARY KEY,
FOREIGN KEY (c2) REFERENCES bug53046_1(c1)
ON UPDATE CASCADE ON DELETE CASCADE) ENGINE=INNODB;
INSERT INTO bug53046_1 VALUES (1);
let $i = 5;
while ($i) {
eval INSERT INTO bug53046_1 SELECT c1+(SELECT MAX(c1) FROM bug53046_1)
FROM bug53046_1;
dec $i;
}
INSERT INTO bug53046_2 VALUES (1), (2);
# CREATE TABLE innodb_table_monitor (a int) ENGINE=INNODB;
# wait more than 1 minute and observe the mysqld output
# DROP TABLE innodb_table_monitor;
ANALYZE TABLE bug53046_1;
# this prints create time and other nondeterministic data
-- disable_result_log
SHOW TABLE STATUS LIKE 'bug53046_1';
-- enable_result_log
UPDATE bug53046_1 SET c1 = c1 - 1;
DELETE FROM bug53046_1;
INSERT INTO bug53046_1 VALUES (1);
INSERT INTO bug53046_2 VALUES (1);
TRUNCATE TABLE bug53046_2;
DROP TABLE bug53046_2;
DROP TABLE bug53046_1;
......@@ -3633,8 +3633,6 @@ btr_estimate_number_of_different_key_vals(
also the pages used for external storage of fields (those pages are
included in index->stat_n_leaf_pages) */
dict_index_stat_mutex_enter(index);
for (j = 0; j <= n_cols; j++) {
index->stat_n_diff_key_vals[j]
= ((n_diff[j]
......@@ -3664,8 +3662,6 @@ btr_estimate_number_of_different_key_vals(
index->stat_n_diff_key_vals[j] += add_on;
}
dict_index_stat_mutex_exit(index);
mem_free(n_diff);
if (UNIV_LIKELY_NULL(heap)) {
mem_heap_free(heap);
......
......@@ -91,9 +91,18 @@ UNIV_INTERN mysql_pfs_key_t dict_foreign_err_mutex_key;
/** Identifies generated InnoDB foreign key names */
static char dict_ibfk[] = "_ibfk_";
/** array of mutexes protecting dict_index_t::stat_n_diff_key_vals[] */
#define DICT_INDEX_STAT_MUTEX_SIZE 32
mutex_t dict_index_stat_mutex[DICT_INDEX_STAT_MUTEX_SIZE];
/** array of rw locks protecting
dict_table_t::stat_initialized
dict_table_t::stat_n_rows (*)
dict_table_t::stat_clustered_index_size
dict_table_t::stat_sum_of_other_index_sizes
dict_table_t::stat_modified_counter (*)
dict_table_t::indexes*::stat_n_diff_key_vals[]
dict_table_t::indexes*::stat_index_size
dict_table_t::indexes*::stat_n_leaf_pages
(*) those are not always protected for performance reasons */
#define DICT_TABLE_STATS_LATCHES_SIZE 64
static rw_lock_t dict_table_stats_latches[DICT_TABLE_STATS_LATCHES_SIZE];
/*******************************************************************//**
Tries to find column names for the index and sets the col field of the
......@@ -254,43 +263,65 @@ dict_mutex_exit_for_mysql(void)
mutex_exit(&(dict_sys->mutex));
}
/** Get the mutex that protects index->stat_n_diff_key_vals[] */
#define GET_INDEX_STAT_MUTEX(index) \
(&dict_index_stat_mutex[ut_fold_ull(index->id) \
% DICT_INDEX_STAT_MUTEX_SIZE])
/** Get the latch that protects the stats of a given table */
#define GET_TABLE_STATS_LATCH(table) \
(&dict_table_stats_latches[ut_fold_ull(table->id) \
% DICT_TABLE_STATS_LATCHES_SIZE])
/**********************************************************************//**
Lock the appropriate mutex to protect index->stat_n_diff_key_vals[].
index->id is used to pick the right mutex and it should not change
before dict_index_stat_mutex_exit() is called on this index. */
Lock the appropriate latch to protect a given table's statistics.
table->id is used to pick the corresponding latch from a global array of
latches. */
UNIV_INTERN
void
dict_index_stat_mutex_enter(
/*========================*/
const dict_index_t* index) /*!< in: index */
dict_table_stats_lock(
/*==================*/
const dict_table_t* table, /*!< in: table */
ulint latch_mode) /*!< in: RW_S_LATCH or
RW_X_LATCH */
{
ut_ad(index != NULL);
ut_ad(index->magic_n == DICT_INDEX_MAGIC_N);
ut_ad(index->cached);
ut_ad(!index->to_be_dropped);
ut_ad(table != NULL);
ut_ad(table->magic_n == DICT_TABLE_MAGIC_N);
mutex_enter(GET_INDEX_STAT_MUTEX(index));
switch (latch_mode) {
case RW_S_LATCH:
rw_lock_s_lock(GET_TABLE_STATS_LATCH(table));
break;
case RW_X_LATCH:
rw_lock_x_lock(GET_TABLE_STATS_LATCH(table));
break;
case RW_NO_LATCH:
/* fall through */
default:
ut_error;
}
}
/**********************************************************************//**
Unlock the appropriate mutex that protects index->stat_n_diff_key_vals[]. */
Unlock the latch that has been locked by dict_table_stats_lock() */
UNIV_INTERN
void
dict_index_stat_mutex_exit(
/*=======================*/
const dict_index_t* index) /*!< in: index */
dict_table_stats_unlock(
/*====================*/
const dict_table_t* table, /*!< in: table */
ulint latch_mode) /*!< in: RW_S_LATCH or
RW_X_LATCH */
{
ut_ad(index != NULL);
ut_ad(index->magic_n == DICT_INDEX_MAGIC_N);
ut_ad(index->cached);
ut_ad(!index->to_be_dropped);
ut_ad(table != NULL);
ut_ad(table->magic_n == DICT_TABLE_MAGIC_N);
mutex_exit(GET_INDEX_STAT_MUTEX(index));
switch (latch_mode) {
case RW_S_LATCH:
rw_lock_s_unlock(GET_TABLE_STATS_LATCH(table));
break;
case RW_X_LATCH:
rw_lock_x_unlock(GET_TABLE_STATS_LATCH(table));
break;
case RW_NO_LATCH:
/* fall through */
default:
ut_error;
}
}
/********************************************************************//**
......@@ -682,9 +713,9 @@ dict_init(void)
mutex_create(dict_foreign_err_mutex_key,
&dict_foreign_err_mutex, SYNC_ANY_LATCH);
for (i = 0; i < DICT_INDEX_STAT_MUTEX_SIZE; i++) {
mutex_create(PFS_NOT_INSTRUMENTED,
&dict_index_stat_mutex[i], SYNC_INDEX_TREE);
for (i = 0; i < DICT_TABLE_STATS_LATCHES_SIZE; i++) {
rw_lock_create(PFS_NOT_INSTRUMENTED,
&dict_table_stats_latches[i], SYNC_INDEX_TREE);
}
}
......@@ -715,12 +746,11 @@ dict_table_get(
mutex_exit(&(dict_sys->mutex));
if (table != NULL) {
if (!table->stat_initialized) {
/* If table->ibd_file_missing == TRUE, this will
print an error message and return without doing
anything. */
dict_update_statistics(table);
}
/* If table->ibd_file_missing == TRUE, this will
print an error message and return without doing
anything. */
dict_update_statistics(table, TRUE /* only update stats
if they have not been initialized */);
}
return(table);
......@@ -4201,6 +4231,10 @@ void
dict_update_statistics_low(
/*=======================*/
dict_table_t* table, /*!< in/out: table */
ibool only_calc_if_missing_stats,/*!< in: only
update/recalc the stats if they have
not been initialized yet, otherwise
do nothing */
ibool has_dict_mutex __attribute__((unused)))
/*!< in: TRUE if the caller has the
dictionary mutex */
......@@ -4231,6 +4265,12 @@ dict_update_statistics_low(
return;
}
dict_table_stats_lock(table, RW_X_LATCH);
if (only_calc_if_missing_stats && table->stat_initialized) {
dict_table_stats_unlock(table, RW_X_LATCH);
return;
}
do {
if (UNIV_LIKELY
......@@ -4276,13 +4316,9 @@ dict_update_statistics_low(
index = dict_table_get_first_index(table);
dict_index_stat_mutex_enter(index);
table->stat_n_rows = index->stat_n_diff_key_vals[
dict_index_get_n_unique(index)];
dict_index_stat_mutex_exit(index);
table->stat_clustered_index_size = index->stat_index_size;
table->stat_sum_of_other_index_sizes = sum_of_index_sizes
......@@ -4291,6 +4327,8 @@ dict_update_statistics_low(
table->stat_initialized = TRUE;
table->stat_modified_counter = 0;
dict_table_stats_unlock(table, RW_X_LATCH);
}
/*********************************************************************//**
......@@ -4300,9 +4338,13 @@ UNIV_INTERN
void
dict_update_statistics(
/*===================*/
dict_table_t* table) /*!< in/out: table */
dict_table_t* table, /*!< in/out: table */
ibool only_calc_if_missing_stats)/*!< in: only
update/recalc the stats if they have
not been initialized yet, otherwise
do nothing */
{
dict_update_statistics_low(table, FALSE);
dict_update_statistics_low(table, only_calc_if_missing_stats, FALSE);
}
/**********************************************************************//**
......@@ -4382,7 +4424,11 @@ dict_table_print_low(
ut_ad(mutex_own(&(dict_sys->mutex)));
dict_update_statistics_low(table, TRUE);
dict_update_statistics_low(table,
FALSE /* update even if initialized */,
TRUE /* we have the dict mutex */);
dict_table_stats_lock(table, RW_S_LATCH);
fprintf(stderr,
"--------------------------------------\n"
......@@ -4410,6 +4456,8 @@ dict_table_print_low(
index = UT_LIST_GET_NEXT(indexes, index);
}
dict_table_stats_unlock(table, RW_S_LATCH);
foreign = UT_LIST_GET_FIRST(table->foreign_list);
while (foreign != NULL) {
......@@ -4458,8 +4506,6 @@ dict_index_print_low(
ut_ad(mutex_own(&(dict_sys->mutex)));
dict_index_stat_mutex_enter(index);
if (index->n_user_defined_cols > 0) {
n_vals = index->stat_n_diff_key_vals[
index->n_user_defined_cols];
......@@ -4467,8 +4513,6 @@ dict_index_print_low(
n_vals = index->stat_n_diff_key_vals[1];
}
dict_index_stat_mutex_exit(index);
fprintf(stderr,
" INDEX: name %s, id %llu, fields %lu/%lu,"
" uniq %lu, type %lu\n"
......@@ -4955,8 +4999,8 @@ dict_close(void)
mem_free(dict_sys);
dict_sys = NULL;
for (i = 0; i < DICT_INDEX_STAT_MUTEX_SIZE; i++) {
mutex_free(&dict_index_stat_mutex[i]);
for (i = 0; i < DICT_TABLE_STATS_LATCHES_SIZE; i++) {
rw_lock_free(&dict_table_stats_latches[i]);
}
}
#endif /* !UNIV_HOTBACKUP */
......@@ -346,7 +346,11 @@ dict_process_sys_tables_rec(
/* Update statistics if DICT_TABLE_UPDATE_STATS
is set */
dict_update_statistics_low(*table, TRUE);
dict_update_statistics_low(*table,
FALSE /* update even if
initialized */,
TRUE /* we have the dict
mutex */);
}
return(NULL);
......
......@@ -7534,6 +7534,7 @@ ha_innobase::estimate_rows_upper_bound(void)
dict_index_t* index;
ulonglong estimate;
ulonglong local_data_file_length;
ulint stat_n_leaf_pages;
DBUG_ENTER("estimate_rows_upper_bound");
......@@ -7553,10 +7554,12 @@ ha_innobase::estimate_rows_upper_bound(void)
index = dict_table_get_first_index(prebuilt->table);
ut_a(index->stat_n_leaf_pages > 0);
stat_n_leaf_pages = index->stat_n_leaf_pages;
ut_a(stat_n_leaf_pages > 0);
local_data_file_length =
((ulonglong) index->stat_n_leaf_pages) * UNIV_PAGE_SIZE;
((ulonglong) stat_n_leaf_pages) * UNIV_PAGE_SIZE;
/* Calculate a minimum length for a clustered index record and from
......@@ -7752,7 +7755,9 @@ ha_innobase::info_low(
prebuilt->trx->op_info = "updating table statistics";
dict_update_statistics(ib_table);
dict_update_statistics(ib_table,
FALSE /* update even if stats
are initialized */);
prebuilt->trx->op_info = "returning various info to MySQL";
}
......@@ -7771,6 +7776,9 @@ ha_innobase::info_low(
}
if (flag & HA_STATUS_VARIABLE) {
dict_table_stats_lock(ib_table, RW_S_LATCH);
n_rows = ib_table->stat_n_rows;
/* Because we do not protect stat_n_rows by any mutex in a
......@@ -7820,6 +7828,8 @@ ha_innobase::info_low(
ib_table->stat_sum_of_other_index_sizes)
* UNIV_PAGE_SIZE;
dict_table_stats_unlock(ib_table, RW_S_LATCH);
/* Since fsp_get_available_space_in_free_extents() is
acquiring latches inside InnoDB, we do not call it if we
are asked by MySQL to avoid locking. Another reason to
......@@ -7896,6 +7906,8 @@ ha_innobase::info_low(
table->s->keys);
}
dict_table_stats_lock(ib_table, RW_S_LATCH);
for (i = 0; i < table->s->keys; i++) {
ulong j;
/* We could get index quickly through internal
......@@ -7933,8 +7945,6 @@ ha_innobase::info_low(
break;
}
dict_index_stat_mutex_enter(index);
if (index->stat_n_diff_key_vals[j + 1] == 0) {
rec_per_key = stats.records;
......@@ -7943,8 +7953,6 @@ ha_innobase::info_low(
index->stat_n_diff_key_vals[j + 1]);
}
dict_index_stat_mutex_exit(index);
/* Since MySQL seems to favor table scans
too much over index searches, we pretend
index selectivity is 2 times better than
......@@ -7961,6 +7969,8 @@ ha_innobase::info_low(
(ulong) rec_per_key;
}
}
dict_table_stats_unlock(ib_table, RW_S_LATCH);
}
if (srv_force_recovery >= SRV_FORCE_NO_IBUF_MERGE) {
......
......@@ -1083,6 +1083,10 @@ void
dict_update_statistics_low(
/*=======================*/
dict_table_t* table, /*!< in/out: table */
ibool only_calc_if_missing_stats,/*!< in: only
update/recalc the stats if they have
not been initialized yet, otherwise
do nothing */
ibool has_dict_mutex);/*!< in: TRUE if the caller has the
dictionary mutex */
/*********************************************************************//**
......@@ -1092,7 +1096,11 @@ UNIV_INTERN
void
dict_update_statistics(
/*===================*/
dict_table_t* table); /*!< in/out: table */
dict_table_t* table, /*!< in/out: table */
ibool only_calc_if_missing_stats);/*!< in: only
update/recalc the stats if they have
not been initialized yet, otherwise
do nothing */
/********************************************************************//**
Reserves the dictionary system mutex for MySQL. */
UNIV_INTERN
......@@ -1106,21 +1114,25 @@ void
dict_mutex_exit_for_mysql(void);
/*===========================*/
/**********************************************************************//**
Lock the appropriate mutex to protect index->stat_n_diff_key_vals[].
index->id is used to pick the right mutex and it should not change
before dict_index_stat_mutex_exit() is called on this index. */
Lock the appropriate latch to protect a given table's statistics.
table->id is used to pick the corresponding latch from a global array of
latches. */
UNIV_INTERN
void
dict_index_stat_mutex_enter(
/*========================*/
const dict_index_t* index); /*!< in: index */
dict_table_stats_lock(
/*==================*/
const dict_table_t* table, /*!< in: table */
ulint latch_mode); /*!< in: RW_S_LATCH or
RW_X_LATCH */
/**********************************************************************//**
Unlock the appropriate mutex that protects index->stat_n_diff_key_vals[]. */
Unlock the latch that has been locked by dict_table_stats_lock() */
UNIV_INTERN
void
dict_index_stat_mutex_exit(
/*=======================*/
const dict_index_t* index); /*!< in: index */
dict_table_stats_unlock(
/*====================*/
const dict_table_t* table, /*!< in: table */
ulint latch_mode); /*!< in: RW_S_LATCH or
RW_X_LATCH */
/********************************************************************//**
Checks if the database name in two table names is the same.
@return TRUE if same db name */
......
......@@ -930,7 +930,8 @@ row_update_statistics_if_needed(
if (counter > 2000000000
|| ((ib_int64_t)counter > 16 + table->stat_n_rows / 16)) {
dict_update_statistics(table);
dict_update_statistics(table, FALSE /* update even if stats
are initialized */);
}
}
......@@ -3020,7 +3021,8 @@ row_truncate_table_for_mysql(
dict_table_autoinc_lock(table);
dict_table_autoinc_initialize(table, 1);
dict_table_autoinc_unlock(table);
dict_update_statistics(table);
dict_update_statistics(table, FALSE /* update even if stats are
initialized */);
trx_commit_for_mysql(trx);
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment