MySQL Lists are EOL. Please join:

List:Commits« Previous MessageNext Message »
From:Mattias Jonsson Date:March 30 2010 8:52pm
Subject:bzr commit into mysql-trunk-bugfixing branch (mattias.jonsson:3008)
Bug#51851
View as plain text  
#At file:///Users/mattiasj/clones/bzrroot/b51851-trunk-bugfixing/ based on revid:alik@stripped

 3008 Mattias Jonsson	2010-03-30
      Bug#51851: Server with SBR locks mutex twice on LOAD DATA
      into partitioned MyISAM table
      
      Problem was that the ha_data structure was introduced in 5.1
      and only used for partitioning first, but with the intention
      of be of use for others engines as well, and when used by other
      engines it would clash if it also was partitioned.
      
      Solution is to move the partitioning specific data to a separate
      structure, with its own mutex (which is used for auto_increment).
      
      Also did rename PARTITION_INFO to PARTITION_STATS since there
      already exist a class named partition_info, also cleaned up
      some related variables.
     @ mysql-test/r/partition_binlog_stmt.result
        Bug#51851: Server with SBR locks mutex twice on LOAD DATA
        into partitioned MyISAM table
        
        New result file
     @ mysql-test/t/partition_binlog_stmt.test
        Bug#51851: Server with SBR locks mutex twice on LOAD DATA
        into partitioned MyISAM table
        
        New result file
     @ sql/ha_ndbcluster.cc
        Bug#51851: Server with SBR locks mutex twice on LOAD DATA
        into partitioned MyISAM table
        
        Rename of PARTITION_INFO to PARTITION_STATS to better
        match the use (and there is also a class named
        partition_info...)
     @ sql/ha_ndbcluster.h
        Bug#51851: Server with SBR locks mutex twice on LOAD DATA
        into partitioned MyISAM table
        
        Rename of PARTITION_INFO to PARTITION_STATS to better
        match the use (and there is also a class named
        partition_info...)
     @ sql/ha_partition.cc
        Bug#51851: Server with SBR locks mutex twice on LOAD DATA
        into partitioned MyISAM table
        
        Removed the partitioning engines use of ha_data in
        TABLE_SHARE and added ha_part_data instead, since
        they collide if used in the same time.
        
        Rename of PARTITION_INFO to PARTITION_STATS to better
        match the use (and there is also a class named
        partition_info...)
        
        Removed some dead code.
     @ sql/ha_partition.h
        Bug#51851: Server with SBR locks mutex twice on LOAD DATA
        into partitioned MyISAM table
        
        Removed some dead code.
        
        Rename of PARTITION_INFO to PARTITION_STATS to better
        match the use (and there is also a class named
        partition_info...)
        
        Removed the partitioning engines use of ha_data in
        TABLE_SHARE and added ha_part_data instead, since
        they collide if used in the same time.
     @ sql/handler.cc
        Bug#51851: Server with SBR locks mutex twice on LOAD DATA
        into partitioned MyISAM table
        
        Rename of PARTITION_INFO to PARTITION_STATS to better
        match the use (and there is also a class named
        partition_info...)
     @ sql/handler.h
        Bug#51851: Server with SBR locks mutex twice on LOAD DATA
        into partitioned MyISAM table
        
        Rename of PARTITION_INFO to PARTITION_STATS to better
        match the use (and there is also a class named
        partition_info...)
     @ sql/mysql_priv.h
        Bug#51851: Server with SBR locks mutex twice on LOAD DATA
        into partitioned MyISAM table
        
        Removed the partitioning engines use of ha_data in
        TABLE_SHARE and added ha_part_data instead, since
        they collide if used in the same time.
        
        Added key_PARTITION_LOCK_auto_inc for instrumentation.
     @ sql/mysqld.cc
        Bug#51851: Server with SBR locks mutex twice on LOAD DATA
        into partitioned MyISAM table
        
        Removed the partitioning engines use of ha_data in
        TABLE_SHARE and added ha_part_data instead, since
        they collide if used in the same time.
        
        Added key_PARTITION_LOCK_auto_inc for instrumentation.
     @ sql/partition_info.h
        Bug#51851: Server with SBR locks mutex twice on LOAD DATA
        into partitioned MyISAM table
        
        Removed part_state* since it was not in use.
     @ sql/sql_partition.cc
        Bug#51851: Server with SBR locks mutex twice on LOAD DATA
        into partitioned MyISAM table
        
        Removed part_state* since it was not in use.
     @ sql/sql_partition.h
        Bug#51851: Server with SBR locks mutex twice on LOAD DATA
        into partitioned MyISAM table
        
        Cleaned up old commented out code.
        
        Removed part_state* since it was not in use.
     @ sql/sql_show.cc
        Bug#51851: Server with SBR locks mutex twice on LOAD DATA
        into partitioned MyISAM table
        
        Rename of PARTITION_INFO to PARTITION_STATS to better
        match the use (and there is also a class named
        partition_info...)
        
        Renamed partition_info to partition_info_str, since
        partition_info is a name of a class.
     @ sql/sql_table.cc
        Bug#51851: Server with SBR locks mutex twice on LOAD DATA
        into partitioned MyISAM table
        
        Renamed partition_info to partition_info_str, since
        partition_info is a name of a class.
     @ sql/table.cc
        Bug#51851: Server with SBR locks mutex twice on LOAD DATA
        into partitioned MyISAM table
        
        Removed the partitioning engines use of ha_data in
        TABLE_SHARE and added ha_part_data instead, since
        they collide if used in the same time.
        
        Renamed partition_info to partition_info_str, since
        partition_info is a name of a class.
        
        removed part_state* since it was not in use.
     @ sql/table.h
        Bug#51851: Server with SBR locks mutex twice on LOAD DATA
        into partitioned MyISAM table
        
        Removed the partitioning engines use of ha_data in
        TABLE_SHARE and added ha_part_data instead, since
        they collide if used in the same time.
        
        Renamed partition_info to partition_info_str, since
        partition_info is a name of a class.
        
        removed part_state* since it was not in use.

    added:
      mysql-test/r/partition_binlog_stmt.result
      mysql-test/t/partition_binlog_stmt.test
    modified:
      sql/ha_ndbcluster.cc
      sql/ha_ndbcluster.h
      sql/ha_partition.cc
      sql/ha_partition.h
      sql/handler.cc
      sql/handler.h
      sql/mysql_priv.h
      sql/mysqld.cc
      sql/partition_info.h
      sql/sql_partition.cc
      sql/sql_partition.h
      sql/sql_show.cc
      sql/sql_table.cc
      sql/table.cc
      sql/table.h
=== added file 'mysql-test/r/partition_binlog_stmt.result'
--- a/mysql-test/r/partition_binlog_stmt.result	1970-01-01 00:00:00 +0000
+++ b/mysql-test/r/partition_binlog_stmt.result	2010-03-30 20:52:45 +0000
@@ -0,0 +1,13 @@
+DROP TABLE IF EXISTS t1;
+#
+# Bug#51851: Server with SBR locks mutex twice on LOAD DATA into
+#            partitioned MyISAM table
+CREATE TABLE t1
+(id INT NOT NULL AUTO_INCREMENT PRIMARY KEY,
+name TINYBLOB NOT NULL,
+modified TIMESTAMP DEFAULT '0000-00-00 00:00:00',
+INDEX namelocs (name(255))) ENGINE = MyISAM
+PARTITION BY HASH(id) PARTITIONS 2;
+LOAD DATA LOCAL INFILE 'init_file.txt'
+INTO TABLE t1 (name);
+DROP TABLE t1;

=== added file 'mysql-test/t/partition_binlog_stmt.test'
--- a/mysql-test/t/partition_binlog_stmt.test	1970-01-01 00:00:00 +0000
+++ b/mysql-test/t/partition_binlog_stmt.test	2010-03-30 20:52:45 +0000
@@ -0,0 +1,28 @@
+--source include/have_partition.inc
+--source include/have_binlog_format_statement.inc
+
+--disable_warnings
+DROP TABLE IF EXISTS t1;
+--enable_warnings
+
+--echo #
+--echo # Bug#51851: Server with SBR locks mutex twice on LOAD DATA into
+--echo #            partitioned MyISAM table
+perl;
+open( INIT, ">init_file.txt");
+print INIT "abcd\n";
+close( INIT );
+EOF
+
+CREATE TABLE t1
+(id INT NOT NULL AUTO_INCREMENT PRIMARY KEY,
+ name TINYBLOB NOT NULL,
+ modified TIMESTAMP DEFAULT '0000-00-00 00:00:00',
+ INDEX namelocs (name(255))) ENGINE = MyISAM
+PARTITION BY HASH(id) PARTITIONS 2;
+
+LOAD DATA LOCAL INFILE 'init_file.txt'
+INTO TABLE t1 (name);
+
+--remove_file init_file.txt
+DROP TABLE t1;

=== modified file 'sql/ha_ndbcluster.cc'
--- a/sql/ha_ndbcluster.cc	2010-02-24 17:04:00 +0000
+++ b/sql/ha_ndbcluster.cc	2010-03-30 20:52:45 +0000
@@ -4238,7 +4238,7 @@ int ha_ndbcluster::info(uint flag)
 }
 
 
-void ha_ndbcluster::get_dynamic_partition_info(PARTITION_INFO *stat_info,
+void ha_ndbcluster::get_dynamic_partition_info(PARTITION_STATS *stat_info,
                                                uint part_id)
 {
   /* 
@@ -4246,7 +4246,7 @@ void ha_ndbcluster::get_dynamic_partitio
      implement ndb function which retrives the statistics
      about ndb partitions.
   */
-  bzero((char*) stat_info, sizeof(PARTITION_INFO));
+  bzero((char*) stat_info, sizeof(PARTITION_STATS));
   return;
 }
 

=== modified file 'sql/ha_ndbcluster.h'
--- a/sql/ha_ndbcluster.h	2010-01-07 05:42:07 +0000
+++ b/sql/ha_ndbcluster.h	2010-03-30 20:52:45 +0000
@@ -273,7 +273,7 @@ class ha_ndbcluster: public handler
   ha_rows estimate_rows_upper_bound()
     { return HA_POS_ERROR; }
   int info(uint);
-  void get_dynamic_partition_info(PARTITION_INFO *stat_info, uint part_id);
+  void get_dynamic_partition_info(PARTITION_STATS *stat_info, uint part_id);
   int extra(enum ha_extra_function operation);
   int extra_opt(enum ha_extra_function operation, ulong cache_size);
   int reset();

=== modified file 'sql/ha_partition.cc'
--- a/sql/ha_partition.cc	2010-03-03 14:58:12 +0000
+++ b/sql/ha_partition.cc	2010-03-30 20:52:45 +0000
@@ -2496,12 +2496,12 @@ err1:
   A destructor for partition-specific TABLE_SHARE data.
 */
 
-void ha_data_partition_destroy(void *ha_data)
+void ha_data_partition_destroy(void *ha_part_data)
 {
-  if (ha_data)
+  if (ha_part_data)
   {
-    HA_DATA_PARTITION *ha_data_partition= (HA_DATA_PARTITION*) ha_data;
-    pthread_mutex_destroy(&ha_data_partition->mutex);
+    HA_DATA_PARTITION *ha_data_partition= (HA_DATA_PARTITION*) ha_part_data;
+    mysql_mutex_destroy(&ha_data_partition->LOCK_auto_inc);
   }
 }
 
@@ -2641,28 +2641,29 @@ int ha_partition::open(const char *name,
     goto err_handler;
 
   /*
-    Use table_share->ha_data to share auto_increment_value among all handlers
-    for the same table.
+    Use table_share->ha_part_data to share auto_increment_value among
+    all handlers for the same table.
   */
   if (is_not_tmp_table)
     mysql_mutex_lock(&table_share->LOCK_ha_data);
-  if (!table_share->ha_data)
+  if (!table_share->ha_part_data)
   {
-    HA_DATA_PARTITION *ha_data;
+    HA_DATA_PARTITION *ha_part_data;
     /* currently only needed for auto_increment */
-    table_share->ha_data= ha_data= (HA_DATA_PARTITION*)
-                                   alloc_root(&table_share->mem_root,
-                                              sizeof(HA_DATA_PARTITION));
-    if (!ha_data)
+    table_share->ha_part_data= alloc_root(&table_share->mem_root,
+                                          sizeof(HA_DATA_PARTITION));
+    ha_part_data= (HA_DATA_PARTITION*) table_share->ha_part_data;
+    if (!ha_part_data)
     {
       if (is_not_tmp_table)
         mysql_mutex_unlock(&table_share->LOCK_ha_data);
       goto err_handler;
     }
-    DBUG_PRINT("info", ("table_share->ha_data 0x%p", ha_data));
-    bzero(ha_data, sizeof(HA_DATA_PARTITION));
-    table_share->ha_data_destroy= ha_data_partition_destroy;
-    pthread_mutex_init(&ha_data->mutex, MY_MUTEX_INIT_FAST);
+    DBUG_PRINT("info", ("table_share->ha_part_data 0x%p", ha_part_data));
+    bzero(ha_part_data, sizeof(HA_DATA_PARTITION));
+    table_share->ha_part_data_destroy= ha_data_partition_destroy;
+    mysql_mutex_init(key_PARTITION_LOCK_auto_inc, &ha_part_data->LOCK_auto_inc,
+                     MY_MUTEX_INIT_FAST);
   }
   if (is_not_tmp_table)
     mysql_mutex_unlock(&table_share->LOCK_ha_data);
@@ -3086,7 +3087,8 @@ int ha_partition::write_row(uchar * buf)
   longlong func_value;
   bool have_auto_increment= table->next_number_field && buf == table->record[0];
   my_bitmap_map *old_map;
-  HA_DATA_PARTITION *ha_data= (HA_DATA_PARTITION*) table_share->ha_data;
+  HA_DATA_PARTITION *ha_part_data= (HA_DATA_PARTITION*)
+                                            table_share->ha_part_data;
   THD *thd= ha_thd();
   timestamp_auto_set_type orig_timestamp_type= table->timestamp_field_type;
 #ifdef NOT_NEEDED
@@ -3106,7 +3108,7 @@ int ha_partition::write_row(uchar * buf)
   */
   if (have_auto_increment)
   {
-    if (!ha_data->auto_inc_initialized &&
+    if (!ha_part_data->auto_inc_initialized &&
         !table->s->next_number_keypart)
     {
       /*
@@ -3247,7 +3249,7 @@ int ha_partition::update_row(const uchar
 exit:
   /*
     if updating an auto_increment column, update
-    table_share->ha_data->next_auto_inc_val if needed.
+    table_share->ha_part_data->next_auto_inc_val if needed.
     (not to be used if auto_increment on secondary field in a multi-column
     index)
     mysql_update does not set table->next_number_field, so we use
@@ -3256,8 +3258,9 @@ exit:
   if (table->found_next_number_field && new_data == table->record[0] &&
       !table->s->next_number_keypart)
   {
-    HA_DATA_PARTITION *ha_data= (HA_DATA_PARTITION*) table_share->ha_data;
-    if (!ha_data->auto_inc_initialized)
+    HA_DATA_PARTITION *ha_part_data= (HA_DATA_PARTITION*)
+                                              table_share->ha_part_data;
+    if (!ha_part_data->auto_inc_initialized)
       info(HA_STATUS_AUTO);
     set_auto_increment_if_higher(table->found_next_number_field);
   }
@@ -3349,11 +3352,12 @@ int ha_partition::delete_all_rows()
   if (thd->lex->sql_command == SQLCOM_TRUNCATE)
   {
     Alter_info *alter_info= &thd->lex->alter_info;
-    HA_DATA_PARTITION *ha_data= (HA_DATA_PARTITION*) table_share->ha_data;
+    HA_DATA_PARTITION *ha_part_data= (HA_DATA_PARTITION*)
+                                              table_share->ha_part_data;
     /* TRUNCATE also means resetting auto_increment */
     lock_auto_increment();
-    ha_data->next_auto_inc_val= 0;
-    ha_data->auto_inc_initialized= FALSE;
+    ha_part_data->next_auto_inc_val= 0;
+    ha_part_data->auto_inc_initialized= FALSE;
     unlock_auto_increment();
     if (alter_info->flags & ALTER_ADMIN_PARTITION)
     {
@@ -5069,22 +5073,23 @@ int ha_partition::info(uint flag)
   if (flag & HA_STATUS_AUTO)
   {
     bool auto_inc_is_first_in_idx= (table_share->next_number_keypart == 0);
-    HA_DATA_PARTITION *ha_data= (HA_DATA_PARTITION*) table_share->ha_data;
+    HA_DATA_PARTITION *ha_part_data= (HA_DATA_PARTITION*)
+                                              table_share->ha_part_data;
     DBUG_PRINT("info", ("HA_STATUS_AUTO"));
     if (!table->found_next_number_field)
       stats.auto_increment_value= 0;
-    else if (ha_data->auto_inc_initialized)
+    else if (ha_part_data->auto_inc_initialized)
     {
       lock_auto_increment();
-      stats.auto_increment_value= ha_data->next_auto_inc_val;
+      stats.auto_increment_value= ha_part_data->next_auto_inc_val;
       unlock_auto_increment();
     }
     else
     {
       lock_auto_increment();
       /* to avoid two concurrent initializations, check again when locked */
-      if (ha_data->auto_inc_initialized)
-        stats.auto_increment_value= ha_data->next_auto_inc_val;
+      if (ha_part_data->auto_inc_initialized)
+        stats.auto_increment_value= ha_part_data->next_auto_inc_val;
       else
       {
         handler *file, **file_array;
@@ -5104,10 +5109,10 @@ int ha_partition::info(uint flag)
         stats.auto_increment_value= auto_increment_value;
         if (auto_inc_is_first_in_idx)
         {
-          set_if_bigger(ha_data->next_auto_inc_val, auto_increment_value);
-          ha_data->auto_inc_initialized= TRUE;
+          set_if_bigger(ha_part_data->next_auto_inc_val, auto_increment_value);
+          ha_part_data->auto_inc_initialized= TRUE;
           DBUG_PRINT("info", ("initializing next_auto_inc_val to %lu",
-                              (ulong) ha_data->next_auto_inc_val));
+                              (ulong) ha_part_data->next_auto_inc_val));
         }
       }
       unlock_auto_increment();
@@ -5281,7 +5286,7 @@ int ha_partition::info(uint flag)
 }
 
 
-void ha_partition::get_dynamic_partition_info(PARTITION_INFO *stat_info,
+void ha_partition::get_dynamic_partition_info(PARTITION_STATS *stat_info,
                                               uint part_id)
 {
   handler *file= m_file[part_id];
@@ -6496,11 +6501,12 @@ int ha_partition::reset_auto_increment(u
 {
   handler **file= m_file;
   int res;
-  HA_DATA_PARTITION *ha_data= (HA_DATA_PARTITION*) table_share->ha_data;
+  HA_DATA_PARTITION *ha_part_data= (HA_DATA_PARTITION*)
+                                            table_share->ha_part_data;
   DBUG_ENTER("ha_partition::reset_auto_increment");
   lock_auto_increment();
-  ha_data->auto_inc_initialized= FALSE;
-  ha_data->next_auto_inc_val= 0;
+  ha_part_data->auto_inc_initialized= FALSE;
+  ha_part_data->next_auto_inc_val= 0;
   do
   {
     if ((res= (*file)->ha_reset_auto_increment(value)) != 0)
@@ -6514,7 +6520,7 @@ int ha_partition::reset_auto_increment(u
 /**
   This method is called by update_auto_increment which in turn is called
   by the individual handlers as part of write_row. We use the
-  table_share->ha_data->next_auto_inc_val, or search all
+  table_share->ha_part_data->next_auto_inc_val, or search all
   partitions for the highest auto_increment_value if not initialized or
   if auto_increment field is a secondary part of a key, we must search
   every partition when holding a mutex to be sure of correctness.
@@ -6567,13 +6573,14 @@ void ha_partition::get_auto_increment(ul
   else
   {
     THD *thd= ha_thd();
-    HA_DATA_PARTITION *ha_data= (HA_DATA_PARTITION*) table_share->ha_data;
+    HA_DATA_PARTITION *ha_part_data= (HA_DATA_PARTITION*)
+                                              table_share->ha_part_data;
     /*
       This is initialized in the beginning of the first write_row call.
     */
-    DBUG_ASSERT(ha_data->auto_inc_initialized);
+    DBUG_ASSERT(ha_part_data->auto_inc_initialized);
     /*
-      Get a lock for handling the auto_increment in table_share->ha_data
+      Get a lock for handling the auto_increment in table_share->ha_part_data
       for avoiding two concurrent statements getting the same number.
     */ 
 
@@ -6600,8 +6607,8 @@ void ha_partition::get_auto_increment(ul
     }
 
     /* this gets corrected (for offset/increment) in update_auto_increment */
-    *first_value= ha_data->next_auto_inc_val;
-    ha_data->next_auto_inc_val+= nb_desired_values * increment;
+    *first_value= ha_part_data->next_auto_inc_val;
+    ha_part_data->next_auto_inc_val+= nb_desired_values * increment;
 
     unlock_auto_increment();
     DBUG_PRINT("info", ("*first_value: %lu", (ulong) *first_value));
@@ -6621,15 +6628,16 @@ void ha_partition::release_auto_incremen
   }
   else if (next_insert_id)
   {
-    HA_DATA_PARTITION *ha_data= (HA_DATA_PARTITION*) table_share->ha_data;
+    HA_DATA_PARTITION *ha_part_data= (HA_DATA_PARTITION*)
+                                              table_share->ha_part_data;
     ulonglong next_auto_inc_val;
     lock_auto_increment();
-    next_auto_inc_val= ha_data->next_auto_inc_val;
+    next_auto_inc_val= ha_part_data->next_auto_inc_val;
     if (next_insert_id < next_auto_inc_val &&
         auto_inc_interval_for_cur_row.maximum() >= next_auto_inc_val)
-      ha_data->next_auto_inc_val= next_insert_id;
-    DBUG_PRINT("info", ("ha_data->next_auto_inc_val: %lu",
-                        (ulong) ha_data->next_auto_inc_val));
+      ha_part_data->next_auto_inc_val= next_insert_id;
+    DBUG_PRINT("info", ("ha_part_data->next_auto_inc_val: %lu",
+                        (ulong) ha_part_data->next_auto_inc_val));
 
     /* Unlock the multi row statement lock taken in get_auto_increment */
     if (auto_increment_safe_stmt_log_lock)
@@ -6729,127 +6737,6 @@ int ha_partition::indexes_are_disabled(v
 }
 
 
-/****************************************************************************
-                MODULE Partition Share
-****************************************************************************/
-/*
-  Service routines for ... methods.
--------------------------------------------------------------------------
-  Variables for partition share methods. A hash used to track open tables.
-  A mutex for the hash table and an init variable to check if hash table
-  is initialized.
-  There is also a constant ending of the partition handler file name.
-*/
-
-#ifdef NOT_USED
-static HASH partition_open_tables;
-static mysql_mutex_t partition_mutex;
-static int partition_init= 0;
-
-
-/*
-  Function we use in the creation of our hash to get key.
-*/
-
-static uchar *partition_get_key(PARTITION_SHARE *share, size_t *length,
-			       my_bool not_used __attribute__ ((unused)))
-{
-  *length= share->table_name_length;
-  return (uchar *) share->table_name;
-}
-
-/*
-  Example of simple lock controls. The "share" it creates is structure we
-  will pass to each partition handler. Do you have to have one of these?
-  Well, you have pieces that are used for locking, and they are needed to
-  function.
-*/
-
-static PARTITION_SHARE *get_share(const char *table_name, TABLE *table)
-{
-  PARTITION_SHARE *share;
-  uint length;
-  char *tmp_name;
-
-  /*
-    So why does this exist? There is no way currently to init a storage
-    engine.
-    Innodb and BDB both have modifications to the server to allow them to
-    do this. Since you will not want to do this, this is probably the next
-    best method.
-  */
-  if (!partition_init)
-  {
-    /* Hijack a mutex for init'ing the storage engine */
-    mysql_mutex_lock(&LOCK_mysql_create_db);
-    if (!partition_init)
-    {
-      partition_init++;
-      mysql_mutex_init(INSTRUMENT_ME, &partition_mutex, MY_MUTEX_INIT_FAST);
-      (void) hash_init(&partition_open_tables, system_charset_info, 32, 0, 0,
-		       (hash_get_key) partition_get_key, 0, 0);
-    }
-    mysql_mutex_unlock(&LOCK_mysql_create_db);
-  }
-  mysql_mutex_lock(&partition_mutex);
-  length= (uint) strlen(table_name);
-
-  if (!(share= (PARTITION_SHARE *) hash_search(&partition_open_tables,
-					       (uchar *) table_name, length)))
-  {
-    if (!(share= (PARTITION_SHARE *)
-	  my_multi_malloc(MYF(MY_WME | MY_ZEROFILL),
-			  &share, (uint) sizeof(*share),
-			  &tmp_name, (uint) length + 1, NullS)))
-    {
-      mysql_mutex_unlock(&partition_mutex);
-      return NULL;
-    }
-
-    share->use_count= 0;
-    share->table_name_length= length;
-    share->table_name= tmp_name;
-    strmov(share->table_name, table_name);
-    if (my_hash_insert(&partition_open_tables, (uchar *) share))
-      goto error;
-    thr_lock_init(&share->lock);
-    mysql_mutex_init(INSTRUMENT_ME, &share->mutex, MY_MUTEX_INIT_FAST);
-  }
-  share->use_count++;
-  mysql_mutex_unlock(&partition_mutex);
-
-  return share;
-
-error:
-  mysql_mutex_unlock(&partition_mutex);
-  my_free((uchar*) share, MYF(0));
-
-  return NULL;
-}
-
-
-/*
-  Free lock controls. We call this whenever we close a table. If the table
-  had the last reference to the share then we free memory associated with
-  it.
-*/
-
-static int free_share(PARTITION_SHARE *share)
-{
-  mysql_mutex_lock(&partition_mutex);
-  if (!--share->use_count)
-  {
-    hash_delete(&partition_open_tables, (uchar *) share);
-    thr_lock_delete(&share->lock);
-    mysql_mutex_destroy(&share->mutex);
-    my_free((uchar*) share, MYF(0));
-  }
-  mysql_mutex_unlock(&partition_mutex);
-
-  return 0;
-}
-#endif /* NOT_USED */
-
 struct st_mysql_storage_engine partition_storage_engine=
 { MYSQL_HANDLERTON_INTERFACE_VERSION };
 

=== modified file 'sql/ha_partition.h'
--- a/sql/ha_partition.h	2010-02-06 10:28:06 +0000
+++ b/sql/ha_partition.h	2010-03-30 20:52:45 +0000
@@ -26,30 +26,15 @@ enum partition_keywords
   PKW_COLUMNS
 };
 
-/*
-  PARTITION_SHARE is a structure that will be shared amoung all open handlers
-  The partition implements the minimum of what you will probably need.
-*/
-
-#ifdef NOT_USED
-typedef struct st_partition_share
-{
-  char *table_name;
-  uint table_name_length, use_count;
-  mysql_mutex_t mutex;
-  THR_LOCK lock;
-} PARTITION_SHARE;
-#endif
 
 /**
   Partition specific ha_data struct.
-  @todo: move all partition specific data from TABLE_SHARE here.
 */
 typedef struct st_ha_data_partition
 {
-  ulonglong next_auto_inc_val;                 /**< first non reserved value */
   bool auto_inc_initialized;
-  pthread_mutex_t mutex;
+  mysql_mutex_t LOCK_auto_inc;                 /**< protecting auto_inc val */
+  ulonglong next_auto_inc_val;                 /**< first non reserved value */
 } HA_DATA_PARTITION;
 
 #define PARTITION_BYTES_IN_POS 2
@@ -519,7 +504,7 @@ public:
     -------------------------------------------------------------------------
   */
   virtual int info(uint);
-  void get_dynamic_partition_info(PARTITION_INFO *stat_info,
+  void get_dynamic_partition_info(PARTITION_STATS *stat_info,
                                   uint part_id);
   virtual int extra(enum ha_extra_function operation);
   virtual int extra_opt(enum ha_extra_function operation, ulong cachesize);
@@ -938,19 +923,23 @@ private:
   virtual int reset_auto_increment(ulonglong value);
   virtual void lock_auto_increment()
   {
+    HA_DATA_PARTITION *ha_part_data;
     /* lock already taken */
     if (auto_increment_safe_stmt_log_lock)
       return;
-    DBUG_ASSERT(table_share->ha_data && !auto_increment_lock);
+    ha_part_data= (HA_DATA_PARTITION*) table_share->ha_part_data;
+    DBUG_ASSERT(ha_part_data && !auto_increment_lock);
     if(table_share->tmp_table == NO_TMP_TABLE)
     {
       auto_increment_lock= TRUE;
-      mysql_mutex_lock(&table_share->LOCK_ha_data);
+      mysql_mutex_lock(&ha_part_data->LOCK_auto_inc);
     }
   }
   virtual void unlock_auto_increment()
   {
-    DBUG_ASSERT(table_share->ha_data);
+    HA_DATA_PARTITION *ha_part_data= (HA_DATA_PARTITION*)
+                                              table_share->ha_part_data;
+    DBUG_ASSERT(ha_part_data);
     /*
       If auto_increment_safe_stmt_log_lock is true, we have to keep the lock.
       It will be set to false and thus unlocked at the end of the statement by
@@ -958,20 +947,21 @@ private:
     */
     if(auto_increment_lock && !auto_increment_safe_stmt_log_lock)
     {
-      mysql_mutex_unlock(&table_share->LOCK_ha_data);
+      mysql_mutex_unlock(&ha_part_data->LOCK_auto_inc);
       auto_increment_lock= FALSE;
     }
   }
   virtual void set_auto_increment_if_higher(Field *field)
   {
-    HA_DATA_PARTITION *ha_data= (HA_DATA_PARTITION*) table_share->ha_data;
+    HA_DATA_PARTITION *ha_part_data= (HA_DATA_PARTITION*)
+                                              table_share->ha_part_data;
     ulonglong nr= (((Field_num*) field)->unsigned_flag ||
                    field->val_int() > 0) ? field->val_int() : 0;
     lock_auto_increment();
-    DBUG_ASSERT(ha_data->auto_inc_initialized == TRUE);
+    DBUG_ASSERT(ha_part_data->auto_inc_initialized == TRUE);
     /* must check when the mutex is taken */
-    if (nr >= ha_data->next_auto_inc_val)
-      ha_data->next_auto_inc_val= nr + 1;
+    if (nr >= ha_part_data->next_auto_inc_val)
+      ha_part_data->next_auto_inc_val= nr + 1;
     unlock_auto_increment();
   }
 

=== modified file 'sql/handler.cc'
--- a/sql/handler.cc	2010-02-16 22:19:47 +0000
+++ b/sql/handler.cc	2010-03-30 20:52:45 +0000
@@ -3517,7 +3517,7 @@ int handler::index_next_same(uchar *buf,
 }
 
 
-void handler::get_dynamic_partition_info(PARTITION_INFO *stat_info,
+void handler::get_dynamic_partition_info(PARTITION_STATS *stat_info,
                                          uint part_id)
 {
   info(HA_STATUS_CONST | HA_STATUS_TIME | HA_STATUS_VARIABLE |

=== modified file 'sql/handler.h'
--- a/sql/handler.h	2010-02-20 10:07:32 +0000
+++ b/sql/handler.h	2010-03-30 20:52:45 +0000
@@ -954,7 +954,7 @@ typedef struct {
   ulong check_time;
   ulong update_time;
   ulonglong check_sum;
-} PARTITION_INFO;
+} PARTITION_STATS;
 
 #define UNDEF_NODEGROUP 65535
 class Item;
@@ -1553,7 +1553,7 @@ public:
     { return (ha_rows) 10; }
   virtual void position(const uchar *record)=0;
   virtual int info(uint)=0; // see my_base.h for full description
-  virtual void get_dynamic_partition_info(PARTITION_INFO *stat_info,
+  virtual void get_dynamic_partition_info(PARTITION_STATS *stat_info,
                                           uint part_id);
   virtual int extra(enum ha_extra_function operation)
   { return 0; }

=== modified file 'sql/mysql_priv.h'
--- a/sql/mysql_priv.h	2010-03-15 21:20:20 +0000
+++ b/sql/mysql_priv.h	2010-03-30 20:52:45 +0000
@@ -2718,8 +2718,8 @@ extern PSI_mutex_key key_BINLOG_LOCK_ind
   key_master_info_data_lock, key_master_info_run_lock,
   key_mutex_slave_reporting_capability_err_lock, key_relay_log_info_data_lock,
   key_relay_log_info_log_space_lock, key_relay_log_info_run_lock,
-  key_structure_guard_mutex, key_TABLE_SHARE_LOCK_ha_data, key_LOCK_error_messages,
-  key_LOCK_thread_count;
+  key_structure_guard_mutex, key_TABLE_SHARE_LOCK_ha_data,
+  key_LOCK_error_messages, key_LOCK_thread_count, key_PARTITION_LOCK_auto_inc;
 
 extern PSI_rwlock_key key_rwlock_LOCK_grant, key_rwlock_LOCK_logger,
   key_rwlock_LOCK_sys_init_connect, key_rwlock_LOCK_sys_init_slave,

=== modified file 'sql/mysqld.cc'
--- a/sql/mysqld.cc	2010-03-05 13:08:21 +0000
+++ b/sql/mysqld.cc	2010-03-30 20:52:45 +0000
@@ -7872,8 +7872,9 @@ PSI_mutex_key key_BINLOG_LOCK_index, key
   key_master_info_data_lock, key_master_info_run_lock,
   key_mutex_slave_reporting_capability_err_lock, key_relay_log_info_data_lock,
   key_relay_log_info_log_space_lock, key_relay_log_info_run_lock,
-  key_structure_guard_mutex, key_TABLE_SHARE_LOCK_ha_data, key_LOCK_error_messages,
-  key_LOG_INFO_lock, key_LOCK_thread_count;
+  key_structure_guard_mutex, key_TABLE_SHARE_LOCK_ha_data,
+  key_LOCK_error_messages, key_LOG_INFO_lock, key_LOCK_thread_count,
+  key_PARTITION_LOCK_auto_inc;
 
 static PSI_mutex_info all_server_mutexes[]=
 {
@@ -7927,7 +7928,8 @@ static PSI_mutex_info all_server_mutexes
   { &key_TABLE_SHARE_LOCK_ha_data, "TABLE_SHARE::LOCK_ha_data", 0},
   { &key_LOCK_error_messages, "LOCK_error_messages", PSI_FLAG_GLOBAL},
   { &key_LOG_INFO_lock, "LOG_INFO::lock", 0},
-  { &key_LOCK_thread_count, "LOCK_thread_count", PSI_FLAG_GLOBAL}
+  { &key_LOCK_thread_count, "LOCK_thread_count", PSI_FLAG_GLOBAL},
+  { &key_PARTITION_LOCK_auto_inc, "HA_DATA_PARTITION::LOCK_auto_inc", 0}
 };
 
 PSI_rwlock_key key_rwlock_LOCK_grant, key_rwlock_LOCK_logger,

=== modified file 'sql/partition_info.h'
--- a/sql/partition_info.h	2009-12-17 17:39:10 +0000
+++ b/sql/partition_info.h	2010-03-30 20:52:45 +0000
@@ -151,8 +151,6 @@ public:
   char *part_func_string;
   char *subpart_func_string;
 
-  const char *part_state;
-
   partition_element *curr_part_elem;
   partition_element *current_partition;
   part_elem_value *curr_list_val;
@@ -173,7 +171,6 @@ public:
   partition_type subpart_type;
 
   uint part_info_len;
-  uint part_state_len;
   uint part_func_len;
   uint subpart_func_len;
 
@@ -226,13 +223,12 @@ public:
     list_array(NULL), err_value(0),
     part_info_string(NULL),
     part_func_string(NULL), subpart_func_string(NULL),
-    part_state(NULL),
     curr_part_elem(NULL), current_partition(NULL),
     curr_list_object(0), num_columns(0),
     default_engine_type(NULL),
     part_result_type(INT_RESULT),
     part_type(NOT_A_PARTITION), subpart_type(NOT_A_PARTITION),
-    part_info_len(0), part_state_len(0),
+    part_info_len(0),
     part_func_len(0), subpart_func_len(0),
     num_parts(0), num_subparts(0),
     count_curr_subparts(0), part_error_code(0),

=== modified file 'sql/sql_partition.cc'
--- a/sql/sql_partition.cc	2010-03-04 10:24:39 +0000
+++ b/sql/sql_partition.cc	2010-03-30 20:52:45 +0000
@@ -4165,7 +4165,6 @@ void get_partition_set(const TABLE *tabl
 
 bool mysql_unpack_partition(THD *thd,
                             const char *part_buf, uint part_info_len,
-                            const char *part_state, uint part_state_len,
                             TABLE* table, bool is_create_table_ind,
                             handlerton *default_db_type,
                             bool *work_part_info_used)
@@ -4201,8 +4200,6 @@ bool mysql_unpack_partition(THD *thd,
     goto end;
   }
   part_info= lex.part_info;
-  part_info->part_state= part_state;
-  part_info->part_state_len= part_state_len;
   DBUG_PRINT("info", ("Parse: %s", part_buf));
   if (parse_sql(thd, & parser_state, NULL) ||
       part_info->fix_parser_data(thd))

=== modified file 'sql/sql_partition.h'
--- a/sql/sql_partition.h	2009-12-16 13:47:07 +0000
+++ b/sql/sql_partition.h	2010-03-30 20:52:45 +0000
@@ -26,19 +26,6 @@
 #define HA_CAN_PARTITION_UNIQUE (1 << 2)
 #define HA_USE_AUTO_PARTITION (1 << 3)
 
-/*typedef struct {
-  ulonglong data_file_length;
-  ulonglong max_data_file_length;
-  ulonglong index_file_length;
-  ulonglong delete_length;
-  ha_rows records;
-  ulong mean_rec_length;
-  time_t create_time;
-  time_t check_time;
-  time_t update_time;
-  ulonglong check_sum;
-} PARTITION_INFO;
-*/
 typedef struct {
   longlong list_value;
   uint32 partition_id;
@@ -87,7 +74,6 @@ void get_full_part_id_from_key(const TAB
                                part_id_range *part_spec);
 bool mysql_unpack_partition(THD *thd, const char *part_buf,
                             uint part_info_len,
-                            const char *part_state, uint part_state_len,
                             TABLE *table, bool is_create_table_ind,
                             handlerton *default_db_type,
                             bool *work_part_info_used);

=== modified file 'sql/sql_show.cc'
--- a/sql/sql_show.cc	2010-03-15 21:20:20 +0000
+++ b/sql/sql_show.cc	2010-03-30 20:52:45 +0000
@@ -3726,7 +3726,7 @@ static int get_schema_tables_record(THD 
     }
 #ifdef WITH_PARTITION_STORAGE_ENGINE
     if (share->db_type() == partition_hton &&
-        share->partition_info_len)
+        share->partition_info_str_len)
     {
       tmp_db_type= share->default_part_db_type;
       is_partitioned= TRUE;
@@ -5289,7 +5289,7 @@ static void store_schema_partitions_reco
 {
   TABLE* table= schema_table;
   CHARSET_INFO *cs= system_charset_info;
-  PARTITION_INFO stat_info;
+  PARTITION_STATS stat_info;
   MYSQL_TIME time;
   file->get_dynamic_partition_info(&stat_info, part_id);
   table->field[0]->store(STRING_WITH_LEN("def"), cs);

=== modified file 'sql/sql_table.cc'
--- a/sql/sql_table.cc	2010-03-15 21:20:20 +0000
+++ b/sql/sql_table.cc	2010-03-30 20:52:45 +0000
@@ -1725,11 +1725,12 @@ bool mysql_write_frm(ALTER_PARTITION_PAR
           error= 1;
           goto err;
         }
-        share->partition_info= tmp_part_syntax_str;
+        share->partition_info_str= tmp_part_syntax_str;
       }
       else
-        memcpy((char*) share->partition_info, part_syntax_buf, syntax_len + 1);
-      share->partition_info_len= part_info->part_info_len= syntax_len;
+        memcpy((char*) share->partition_info_str, part_syntax_buf,
+               syntax_len + 1);
+      share->partition_info_str_len= part_info->part_info_len= syntax_len;
       part_info->part_info_string= part_syntax_buf;
     }
 #endif

=== modified file 'sql/table.cc'
--- a/sql/table.cc	2010-02-24 13:52:27 +0000
+++ b/sql/table.cc	2010-03-30 20:52:45 +0000
@@ -954,28 +954,28 @@ static int open_binary_frm(THD *thd, TAB
     }
     if (next_chunk + 5 < buff_end)
     {
-      uint32 partition_info_len = uint4korr(next_chunk);
+      uint32 partition_info_str_len = uint4korr(next_chunk);
 #ifdef WITH_PARTITION_STORAGE_ENGINE
       if ((share->partition_info_buffer_size=
-             share->partition_info_len= partition_info_len))
+             share->partition_info_str_len= partition_info_str_len))
       {
-        if (!(share->partition_info= (char*)
+        if (!(share->partition_info_str= (char*)
               memdup_root(&share->mem_root, next_chunk + 4,
-                          partition_info_len + 1)))
+                          partition_info_str_len + 1)))
         {
           my_free(buff, MYF(0));
           goto err;
         }
       }
 #else
-      if (partition_info_len)
+      if (partition_info_str_len)
       {
         DBUG_PRINT("info", ("WITH_PARTITION_STORAGE_ENGINE is not defined"));
         my_free(buff, MYF(0));
         goto err;
       }
 #endif
-      next_chunk+= 5 + partition_info_len;
+      next_chunk+= 5 + partition_info_str_len;
     }
 #if MYSQL_VERSION_ID < 50200
     if (share->mysql_version >= 50106 && share->mysql_version <= 50109)
@@ -1638,6 +1638,10 @@ static int open_binary_frm(THD *thd, TAB
   my_hash_free(&share->name_hash);
   if (share->ha_data_destroy)
     share->ha_data_destroy(share->ha_data);
+#ifdef WITH_PARTITION_STORAGE_ENGINE
+  if (share->ha_part_data_destroy)
+    share->ha_part_data_destroy(share->ha_part_data);
+#endif
 
   open_table_error(share, error, share->open_errno, errarg);
   DBUG_RETURN(error);
@@ -1829,7 +1833,7 @@ int open_table_from_share(THD *thd, TABL
   }
 
 #ifdef WITH_PARTITION_STORAGE_ENGINE
-  if (share->partition_info_len && outparam->file)
+  if (share->partition_info_str_len && outparam->file)
   {
   /*
     In this execution we must avoid calling thd->change_item_tree since
@@ -1850,10 +1854,8 @@ int open_table_from_share(THD *thd, TABL
     bool tmp;
     bool work_part_info_used;
 
-    tmp= mysql_unpack_partition(thd, share->partition_info,
-                                share->partition_info_len,
-                                share->part_state,
-                                share->part_state_len,
+    tmp= mysql_unpack_partition(thd, share->partition_info_str,
+                                share->partition_info_str_len,
                                 outparam, is_create_table,
                                 share->default_part_db_type,
                                 &work_part_info_used);

=== modified file 'sql/table.h'
--- a/sql/table.h	2010-02-24 13:52:27 +0000
+++ b/sql/table.h	2010-03-30 20:52:45 +0000
@@ -495,13 +495,11 @@ struct TABLE_SHARE
   int cached_row_logging_check;
 
 #ifdef WITH_PARTITION_STORAGE_ENGINE
-  /** @todo: Move into *ha_data for partitioning */
+  /* filled in when reading from frm */
   bool auto_partitioned;
-  const char *partition_info;
-  uint  partition_info_len;
+  const char *partition_info_str;
+  uint  partition_info_str_len;
   uint  partition_info_buffer_size;
-  const char *part_state;
-  uint part_state_len;
   handlerton *default_part_db_type;
 #endif
 
@@ -521,6 +519,14 @@ struct TABLE_SHARE
   void *ha_data;
   void (*ha_data_destroy)(void *); /* An optional destructor for ha_data */
 
+#ifdef WITH_PARTITION_STORAGE_ENGINE
+  /** place to store partition specific data, LOCK_ha_data hold while init. */
+  void *ha_part_data;
+  /* Destructor for ha_part_data */
+  void (*ha_part_data_destroy)(void *);
+#endif
+
+
   /** Instrumentation for this table share. */
   PSI_table_share *m_psi;
 


Attachment: [text/bzr-bundle]
Thread
bzr commit into mysql-trunk-bugfixing branch (mattias.jonsson:3008)Bug#51851Mattias Jonsson30 Mar