Merge serg.mylan:/usr/home/serg/Abk/mysql-5.0

into serg.mylan:/usr/home/serg/Abk/mysql-5.0-xa


configure.in:
  Auto merged
mysql-test/r/drop_temp_table.result:
  Auto merged
mysql-test/r/innodb.result:
  Auto merged
mysql-test/t/innodb.test:
  Auto merged
sql/ha_innodb.cc:
  Auto merged
sql/item_func.h:
  Auto merged
sql/log.cc:
  Auto merged
sql/mysql_priv.h:
  Auto merged
sql/mysqld.cc:
  Auto merged
sql/set_var.h:
  Auto merged
sql/sql_base.cc:
  Auto merged
sql/sql_delete.cc:
  Auto merged
sql/sql_insert.cc:
  Auto merged
sql/sql_lex.cc:
  Auto merged
sql/sql_parse.cc:
  Auto merged

client/mysqlbinlog.cc

@@ -1100,7 +1100,7 @@ at offset %lu ; this could be a log format error or read error",
           /* EOF can't be hit here normally, so it's a real error */
           die("Could not read a Rotate_log_event event \
 at offset %lu ; this could be a log format error or read error",
-              tmp_pos); 
+              tmp_pos);
       }
       else
         break;
@@ -1169,9 +1169,15 @@ static int dump_local_log_entries(const char* logname)
     Log_event* ev = Log_event::read_log_event(file, description_event);
     if (!ev)
     {
-      if (file->error)
+      /*
+        if binlog wasn't closed properly ("in use" flag is set) don't complain
+        about a corruption, but treat it as EOF and move to the next binlog.
+      */
+      if (description_event->flags & LOG_EVENT_BINLOG_IN_USE_F)
+        file->error= 0;
+      else if (file->error)
       {
-	fprintf(stderr,
+        fprintf(stderr,
                 "Could not read entry at offset %s:"
                 "Error in log format or read error\n",
                 llstr(old_off,llbuff));

configure.in

@@ -1887,7 +1887,7 @@ AC_CHECK_FUNCS(alarm bcmp bfill bmove bzero chsize cuserid fchmod fcntl \
   getcwd gethostbyaddr_r gethostbyname_r getpass getpassphrase getpwnam \
   getpwuid getrlimit getrusage getwd gmtime_r index initgroups isnan \
   localtime_r locking longjmp lrand48 madvise mallinfo memcpy memmove \
-  mkstemp mlockall perror poll pread pthread_attr_create mmap \
+  mkstemp mlockall perror poll pread pthread_attr_create mmap getpagesize \
   pthread_attr_getstacksize pthread_attr_setprio pthread_attr_setschedparam \
   pthread_attr_setstacksize pthread_condattr_create pthread_getsequence_np \
   pthread_key_delete pthread_rwlock_rdlock pthread_setprio \

include/my_global.h

@@ -761,7 +761,7 @@ typedef long		int32;
 #endif
 typedef unsigned long	uint32; /* Short for unsigned integer >= 32 bits */
 #else
-error "Neither int or long is of 4 bytes width"
+#error "Neither int or long is of 4 bytes width"
 #endif
 
 #if !defined(HAVE_ULONG) && !defined(HAVE_LINUXTHREADS) && !defined(__USE_MISC)

include/my_pthread.h

@@ -689,12 +689,12 @@ extern pthread_t shutdown_th, main_th, signal_th;
 #define thread_safe_add(V,C,L) (pthread_mutex_lock((L)), (V)+=(C), pthread_mutex_unlock((L)))
 #define thread_safe_sub(V,C,L) \
         (pthread_mutex_lock((L)), (V)-=(C), pthread_mutex_unlock((L)))
-#if defined (__GNUC__) || defined (__cplusplus)
+#ifdef __cplusplus
-static inline bool thread_safe_dec_and_test(ulong V, pthread_mutex_t *L)
+static inline bool thread_safe_dec_and_test(ulong &V, pthread_mutex_t *L)
 {
   ulong res;
   pthread_mutex_lock(L);
-  res=V--;
+  res=--V;
   pthread_mutex_unlock(L);
   return res==0;
 }

innobase/include/trx0sys.ic

@@ -315,7 +315,7 @@ trx_is_active(
 	}
 
 	if (ut_dulint_cmp(trx_id, trx_sys->max_trx_id) >= 0) {
-		
+
 		/* There must be corruption: we return TRUE because this
 		function is only called by lock_clust_rec_some_has_impl()
 		and row_vers_impl_x_locked_off_kernel() and they have
@@ -325,8 +325,9 @@ trx_is_active(
 	}
 
 	trx = trx_get_on_id(trx_id);
-	if (trx && (trx->conc_state == TRX_ACTIVE)) {
+        if (trx && (trx->conc_state == TRX_ACTIVE ||
-		
+                        trx->conc_state == TRX_PREPARED)) {
+
 		return(TRUE);
 	}
 

innobase/read/read0read.c

@@ -153,10 +153,11 @@ read_view_open_now(
 	/* No active transaction should be visible, except cr_trx */
 
 	while (trx) {
-		if (trx != cr_trx && trx->conc_state == TRX_ACTIVE) {
+                if (trx != cr_trx && (trx->conc_state == TRX_ACTIVE ||
+                                        trx->conc_state == TRX_PREPARED)) {
 
 			read_view_set_nth_trx_id(view, n, trx->id);
-		
+
 			n++;
 
 			/* NOTE that a transaction whose trx number is <
@@ -164,7 +165,7 @@ read_view_open_now(
 			in the middle of its commit! Note that when a
 			transaction starts, we initialize trx->no to
 			ut_dulint_max. */
-		
+
 			if (ut_dulint_cmp(view->low_limit_no, trx->no) > 0) {
 
 				view->low_limit_no = trx->no;

innobase/trx/trx0trx.c

@@ -1617,20 +1617,21 @@ trx_print(
 		(ulong) ut_dulint_get_high(trx->id),
 		 (ulong) ut_dulint_get_low(trx->id));
 
-  	switch (trx->conc_state) {
+        switch (trx->conc_state) {
 		case TRX_NOT_STARTED:
 			fputs(", not started", f);
 			break;
 		case TRX_ACTIVE:
+		case TRX_PREPARED:
 			fprintf(f, ", ACTIVE %lu sec",
 				(ulong)difftime(time(NULL), trx->start_time));
-									break;
+                        break;
 		case TRX_COMMITTED_IN_MEMORY:
 			fputs(", COMMITTED IN MEMORY", f);
 			break;
 		default:
 			fprintf(f, " state %lu", (ulong) trx->conc_state);
-  	}
+        }
 
 #ifdef UNIV_LINUX
 	fprintf(f, ", process no %lu", trx->mysql_process_no);

mysql-test/r/ctype_ucs.result

@@ -523,11 +523,12 @@ create table t2 (c char(30)) charset=ucs2;
 set @v=convert('abc' using ucs2);
 reset master;
 insert into t2 values (@v);
-show binlog events from 95;
+show binlog events from 96;
 Log_name	Pos	Event_type	Server_id	End_log_pos	Info
-master-bin.000001	95	User var	1	135	@`v`=_ucs2 0x006100620063 COLLATE ucs2_general_ci
+master-bin.000001	96	User var	1	136	@`v`=_ucs2 0x006100620063 COLLATE ucs2_general_ci
-master-bin.000001	135	Query	1	225	use `test`; insert into t2 values (@v)
+master-bin.000001	136	Query	1	226	use `test`; insert into t2 values (@v)
 /*!40019 SET @@session.max_insert_delayed_threads=0*/;
+ROLLBACK;
 SET @`v`:=_ucs2 0x006100620063 COLLATE `ucs2_general_ci`;
 use test;
 SET TIMESTAMP=10000;

mysql-test/r/innodb.result

@@ -271,7 +271,7 @@ n
 6
 7
 rollback to savepoint `my_savepoint`;
-ERROR HY000: Got error 153 during ROLLBACK
+ERROR 42000: SAVEPOINT my_savepoint does not exist
 set autocommit=1;
 rollback;
 drop table t1;
@@ -1633,14 +1633,14 @@ t2	CREATE TABLE `t2` (
 drop table t2, t1;
 show status like "binlog_cache_use";
 Variable_name	Value
-Binlog_cache_use	25
+Binlog_cache_use	152
 show status like "binlog_cache_disk_use";
 Variable_name	Value
 Binlog_cache_disk_use	0
 create table t1 (a int) engine=innodb;
 show status like "binlog_cache_use";
 Variable_name	Value
-Binlog_cache_use	26
+Binlog_cache_use	153
 show status like "binlog_cache_disk_use";
 Variable_name	Value
 Binlog_cache_disk_use	1
@@ -1649,7 +1649,7 @@ delete from t1;
 commit;
 show status like "binlog_cache_use";
 Variable_name	Value
-Binlog_cache_use	27
+Binlog_cache_use	154
 show status like "binlog_cache_disk_use";
 Variable_name	Value
 Binlog_cache_disk_use	1

mysql-test/r/insert_select.result

@@ -74,8 +74,8 @@ insert into t1 select * from t2;
 ERROR 23000: Duplicate entry '2' for key 1
 show binlog events;
 Log_name	Pos	Event_type	Server_id	End_log_pos	Info
-master-bin.000001	4	Format_desc	1	95	Server ver: VERSION, Binlog ver: 4
+master-bin.000001	4	Format_desc	1	96	Server ver: VERSION, Binlog ver: 4
-master-bin.000001	95	Query	1	190	use `test`; insert into t1 select * from t2
+master-bin.000001	96	Query	1	191	use `test`; insert into t1 select * from t2
 select * from t1;
 a
 1
@@ -88,7 +88,7 @@ create table t2(unique(a)) select a from t1;
 ERROR 23000: Duplicate entry '1' for key 1
 show binlog events;
 Log_name	Pos	Event_type	Server_id	End_log_pos	Info
-master-bin.000001	4	Format_desc	1	95	Server ver: VERSION, Binlog ver: 4
+master-bin.000001	4	Format_desc	1	96	Server ver: VERSION, Binlog ver: 4
 drop table t1;
 create table t1 (a int not null);
 create table t2 (a int not null);

mysql-test/r/mix_innodb_myisam_binlog.result

@@ -6,12 +6,12 @@ begin;
 insert into t1 values(1);
 insert into t2 select * from t1;
 commit;
-show binlog events from 95;
+show binlog events from 96;
 Log_name	Pos	Event_type	Server_id	End_log_pos	Info
-master-bin.000001	95	Query	1	#	use `test`; BEGIN
+master-bin.000001	96	Query	1	#	use `test`; BEGIN
-master-bin.000001	164	Query	1	#	use `test`; insert into t1 values(1)
+master-bin.000001	165	Query	1	#	use `test`; insert into t1 values(1)
-master-bin.000001	252	Query	1	#	use `test`; insert into t2 select * from t1
+master-bin.000001	253	Query	1	#	use `test`; insert into t2 select * from t1
-master-bin.000001	347	Query	1	#	use `test`; COMMIT
+master-bin.000001	348	Xid	1	#	COMMIT /* xid=7 */
 delete from t1;
 delete from t2;
 reset master;
@@ -21,12 +21,12 @@ insert into t2 select * from t1;
 rollback;
 Warnings:
 Warning	1196	Some non-transactional changed tables couldn't be rolled back
-show binlog events from 95;
+show binlog events from 96;
 Log_name	Pos	Event_type	Server_id	End_log_pos	Info
-master-bin.000001	95	Query	1	#	use `test`; BEGIN
+master-bin.000001	96	Query	1	#	use `test`; BEGIN
-master-bin.000001	164	Query	1	#	use `test`; insert into t1 values(2)
+master-bin.000001	165	Query	1	#	use `test`; insert into t1 values(2)
-master-bin.000001	252	Query	1	#	use `test`; insert into t2 select * from t1
+master-bin.000001	253	Query	1	#	use `test`; insert into t2 select * from t1
-master-bin.000001	347	Query	1	#	use `test`; ROLLBACK
+master-bin.000001	348	Query	1	#	use `test`; ROLLBACK
 delete from t1;
 delete from t2;
 reset master;
@@ -39,15 +39,15 @@ rollback to savepoint my_savepoint;
 Warnings:
 Warning	1196	Some non-transactional changed tables couldn't be rolled back
 commit;
-show binlog events from 95;
+show binlog events from 96;
 Log_name	Pos	Event_type	Server_id	End_log_pos	Info
-master-bin.000001	95	Query	1	#	use `test`; BEGIN
+master-bin.000001	96	Query	1	#	use `test`; BEGIN
-master-bin.000001	164	Query	1	#	use `test`; insert into t1 values(3)
+master-bin.000001	165	Query	1	#	use `test`; insert into t1 values(3)
-master-bin.000001	252	Query	1	#	use `test`; savepoint my_savepoint
+master-bin.000001	253	Query	1	#	use `test`; savepoint my_savepoint
-master-bin.000001	338	Query	1	#	use `test`; insert into t1 values(4)
+master-bin.000001	339	Query	1	#	use `test`; insert into t1 values(4)
-master-bin.000001	426	Query	1	#	use `test`; insert into t2 select * from t1
+master-bin.000001	427	Query	1	#	use `test`; insert into t2 select * from t1
-master-bin.000001	521	Query	1	#	use `test`; rollback to savepoint my_savepoint
+master-bin.000001	522	Query	1	#	use `test`; rollback to savepoint my_savepoint
-master-bin.000001	619	Query	1	#	use `test`; COMMIT
+master-bin.000001	620	Xid	1	#	COMMIT /* xid=24 */
 delete from t1;
 delete from t2;
 reset master;
@@ -65,16 +65,16 @@ select a from t1 order by a;
 a
 5
 7
-show binlog events from 95;
+show binlog events from 96;
 Log_name	Pos	Event_type	Server_id	End_log_pos	Info
-master-bin.000001	95	Query	1	#	use `test`; BEGIN
+master-bin.000001	96	Query	1	#	use `test`; BEGIN
-master-bin.000001	164	Query	1	#	use `test`; insert into t1 values(5)
+master-bin.000001	165	Query	1	#	use `test`; insert into t1 values(5)
-master-bin.000001	252	Query	1	#	use `test`; savepoint my_savepoint
+master-bin.000001	253	Query	1	#	use `test`; savepoint my_savepoint
-master-bin.000001	338	Query	1	#	use `test`; insert into t1 values(6)
+master-bin.000001	339	Query	1	#	use `test`; insert into t1 values(6)
-master-bin.000001	426	Query	1	#	use `test`; insert into t2 select * from t1
+master-bin.000001	427	Query	1	#	use `test`; insert into t2 select * from t1
-master-bin.000001	521	Query	1	#	use `test`; rollback to savepoint my_savepoint
+master-bin.000001	522	Query	1	#	use `test`; rollback to savepoint my_savepoint
-master-bin.000001	619	Query	1	#	use `test`; insert into t1 values(7)
+master-bin.000001	620	Query	1	#	use `test`; insert into t1 values(7)
-master-bin.000001	707	Query	1	#	use `test`; COMMIT
+master-bin.000001	708	Xid	1	#	COMMIT /* xid=36 */
 delete from t1;
 delete from t2;
 reset master;
@@ -87,40 +87,47 @@ insert into t2 select * from t1;
 select get_lock("a",10);
 get_lock("a",10)
 1
-show binlog events from 95;
+show binlog events from 96;
 Log_name	Pos	Event_type	Server_id	End_log_pos	Info
-master-bin.000001	95	Query	1	#	use `test`; BEGIN
+master-bin.000001	96	Query	1	#	use `test`; BEGIN
-master-bin.000001	164	Query	1	#	use `test`; insert into t1 values(8)
+master-bin.000001	165	Query	1	#	use `test`; insert into t1 values(8)
-master-bin.000001	252	Query	1	#	use `test`; insert into t2 select * from t1
+master-bin.000001	253	Query	1	#	use `test`; insert into t2 select * from t1
-master-bin.000001	347	Query	1	#	use `test`; ROLLBACK
+master-bin.000001	348	Query	1	#	use `test`; ROLLBACK
+master-bin.000001	420	Query	1	#	use `test`; DO RELEASE_LOCK("a")
 delete from t1;
 delete from t2;
 reset master;
 insert into t1 values(9);
 insert into t2 select * from t1;
-show binlog events from 95;
+show binlog events from 96;
 Log_name	Pos	Event_type	Server_id	End_log_pos	Info
-master-bin.000001	95	Query	1	#	use `test`; insert into t1 values(9)
+master-bin.000001	96	Query	1	#	use `test`; BEGIN
-master-bin.000001	183	Query	1	#	use `test`; insert into t2 select * from t1
+master-bin.000001	165	Query	1	#	use `test`; insert into t1 values(9)
+master-bin.000001	253	Xid	1	#	COMMIT /* xid=59 */
+master-bin.000001	280	Query	1	#	use `test`; insert into t2 select * from t1
 delete from t1;
 delete from t2;
 reset master;
 insert into t1 values(10);
 begin;
 insert into t2 select * from t1;
-show binlog events from 95;
+show binlog events from 96;
 Log_name	Pos	Event_type	Server_id	End_log_pos	Info
-master-bin.000001	95	Query	1	#	use `test`; insert into t1 values(10)
+master-bin.000001	96	Query	1	#	use `test`; BEGIN
-master-bin.000001	184	Query	1	#	use `test`; insert into t2 select * from t1
+master-bin.000001	165	Query	1	#	use `test`; insert into t1 values(10)
+master-bin.000001	254	Xid	1	#	COMMIT /* xid=65 */
+master-bin.000001	281	Query	1	#	use `test`; insert into t2 select * from t1
 insert into t1 values(11);
 commit;
-show binlog events from 95;
+show binlog events from 96;
 Log_name	Pos	Event_type	Server_id	End_log_pos	Info
-master-bin.000001	95	Query	1	#	use `test`; insert into t1 values(10)
+master-bin.000001	96	Query	1	#	use `test`; BEGIN
-master-bin.000001	184	Query	1	#	use `test`; insert into t2 select * from t1
+master-bin.000001	165	Query	1	#	use `test`; insert into t1 values(10)
-master-bin.000001	279	Query	1	#	use `test`; BEGIN
+master-bin.000001	254	Xid	1	#	COMMIT /* xid=65 */
-master-bin.000001	348	Query	1	#	use `test`; insert into t1 values(11)
+master-bin.000001	281	Query	1	#	use `test`; insert into t2 select * from t1
-master-bin.000001	437	Query	1	#	use `test`; COMMIT
+master-bin.000001	376	Query	1	#	use `test`; BEGIN
+master-bin.000001	445	Query	1	#	use `test`; insert into t1 values(11)
+master-bin.000001	534	Xid	1	#	COMMIT /* xid=67 */
 alter table t2 engine=INNODB;
 delete from t1;
 delete from t2;
@@ -129,12 +136,12 @@ begin;
 insert into t1 values(12);
 insert into t2 select * from t1;
 commit;
-show binlog events from 95;
+show binlog events from 96;
 Log_name	Pos	Event_type	Server_id	End_log_pos	Info
-master-bin.000001	95	Query	1	#	use `test`; BEGIN
+master-bin.000001	96	Query	1	#	use `test`; BEGIN
-master-bin.000001	164	Query	1	#	use `test`; insert into t1 values(12)
+master-bin.000001	165	Query	1	#	use `test`; insert into t1 values(12)
-master-bin.000001	253	Query	1	#	use `test`; insert into t2 select * from t1
+master-bin.000001	254	Query	1	#	use `test`; insert into t2 select * from t1
-master-bin.000001	348	Query	1	#	use `test`; COMMIT
+master-bin.000001	349	Xid	1	#	COMMIT /* xid=77 */
 delete from t1;
 delete from t2;
 reset master;
@@ -142,7 +149,7 @@ begin;
 insert into t1 values(13);
 insert into t2 select * from t1;
 rollback;
-show binlog events from 95;
+show binlog events from 96;
 Log_name	Pos	Event_type	Server_id	End_log_pos	Info
 delete from t1;
 delete from t2;
@@ -154,11 +161,11 @@ insert into t1 values(15);
 insert into t2 select * from t1;
 rollback to savepoint my_savepoint;
 commit;
-show binlog events from 95;
+show binlog events from 96;
 Log_name	Pos	Event_type	Server_id	End_log_pos	Info
-master-bin.000001	95	Query	1	#	use `test`; BEGIN
+master-bin.000001	96	Query	1	#	use `test`; BEGIN
-master-bin.000001	164	Query	1	#	use `test`; insert into t1 values(14)
+master-bin.000001	165	Query	1	#	use `test`; insert into t1 values(14)
-master-bin.000001	253	Query	1	#	use `test`; COMMIT
+master-bin.000001	254	Xid	1	#	COMMIT /* xid=93 */
 delete from t1;
 delete from t2;
 reset master;
@@ -174,12 +181,12 @@ select a from t1 order by a;
 a
 16
 18
-show binlog events from 95;
+show binlog events from 96;
 Log_name	Pos	Event_type	Server_id	End_log_pos	Info
-master-bin.000001	95	Query	1	#	use `test`; BEGIN
+master-bin.000001	96	Query	1	#	use `test`; BEGIN
-master-bin.000001	164	Query	1	#	use `test`; insert into t1 values(16)
+master-bin.000001	165	Query	1	#	use `test`; insert into t1 values(16)
-master-bin.000001	253	Query	1	#	use `test`; insert into t1 values(18)
+master-bin.000001	254	Query	1	#	use `test`; insert into t1 values(18)
-master-bin.000001	342	Query	1	#	use `test`; COMMIT
+master-bin.000001	343	Xid	1	#	COMMIT /* xid=104 */
 delete from t1;
 delete from t2;
 alter table t2 type=MyISAM;

mysql-test/r/mysqlbinlog.result

@@ -15,6 +15,7 @@ flush logs;
 
 --- Local --
 /*!40019 SET @@session.max_insert_delayed_threads=0*/;
+ROLLBACK;
 use test;
 SET TIMESTAMP=1000000000;
 SET @@session.foreign_key_checks=1, @@session.sql_auto_is_null=1, @@session.unique_checks=1;
@@ -47,6 +48,7 @@ insert into t1 values ("Alas");
 
 --- --database --
 /*!40019 SET @@session.max_insert_delayed_threads=0*/;
+ROLLBACK;
 SET INSERT_ID=1;
 
 --- --position --
@@ -60,6 +62,7 @@ insert into t1 values ("Alas");
 
 --- Remote --
 /*!40019 SET @@session.max_insert_delayed_threads=0*/;
+ROLLBACK;
 use test;
 SET TIMESTAMP=1000000000;
 SET @@session.foreign_key_checks=1, @@session.sql_auto_is_null=1, @@session.unique_checks=1;
@@ -92,6 +95,7 @@ insert into t1 values ("Alas");
 
 --- --database --
 /*!40019 SET @@session.max_insert_delayed_threads=0*/;
+ROLLBACK;
 SET INSERT_ID=1;
 
 --- --position --
@@ -105,6 +109,7 @@ insert into t1 values ("Alas");
 
 --- reading stdin --
 /*!40019 SET @@session.max_insert_delayed_threads=0*/;
+ROLLBACK;
 use test;
 SET TIMESTAMP=1065204671;
 BEGIN;

mysql-test/r/mysqlbinlog2.result

@@ -16,6 +16,7 @@ insert into t1 values(null, "f");
 
 --- Local --
 /*!40019 SET @@session.max_insert_delayed_threads=0*/;
+ROLLBACK;
 use test;
 SET TIMESTAMP=1579609942;
 SET @@session.foreign_key_checks=1, @@session.sql_auto_is_null=1, @@session.unique_checks=1;
@@ -40,6 +41,7 @@ insert into t1 values(null, "e");
 
 --- offset --
 /*!40019 SET @@session.max_insert_delayed_threads=0*/;
+ROLLBACK;
 SET INSERT_ID=1;
 use test;
 SET TIMESTAMP=1579609942;
@@ -75,6 +77,7 @@ insert into t1 values(null, "e");
 
 --- stop-position --
 /*!40019 SET @@session.max_insert_delayed_threads=0*/;
+ROLLBACK;
 use test;
 SET TIMESTAMP=1579609942;
 SET @@session.foreign_key_checks=1, @@session.sql_auto_is_null=1, @@session.unique_checks=1;
@@ -93,6 +96,7 @@ insert into t1 values(null, "c");
 
 --- start-datetime --
 /*!40019 SET @@session.max_insert_delayed_threads=0*/;
+ROLLBACK;
 SET INSERT_ID=3;
 use test;
 SET TIMESTAMP=1579609944;
@@ -109,6 +113,7 @@ insert into t1 values(null, "e");
 
 --- stop-datetime --
 /*!40019 SET @@session.max_insert_delayed_threads=0*/;
+ROLLBACK;
 use test;
 SET TIMESTAMP=1579609942;
 SET @@session.foreign_key_checks=1, @@session.sql_auto_is_null=1, @@session.unique_checks=1;
@@ -124,6 +129,7 @@ insert into t1 values(null, "b");
 
 --- Local with 2 binlogs on command line --
 /*!40019 SET @@session.max_insert_delayed_threads=0*/;
+ROLLBACK;
 use test;
 SET TIMESTAMP=1579609942;
 SET @@session.foreign_key_checks=1, @@session.sql_auto_is_null=1, @@session.unique_checks=1;
@@ -155,6 +161,7 @@ insert into t1 values(null, "f");
 
 --- offset --
 /*!40019 SET @@session.max_insert_delayed_threads=0*/;
+ROLLBACK;
 SET INSERT_ID=1;
 use test;
 SET TIMESTAMP=1579609942;
@@ -204,6 +211,7 @@ insert into t1 values(null, "f");
 
 --- stop-position --
 /*!40019 SET @@session.max_insert_delayed_threads=0*/;
+ROLLBACK;
 use test;
 SET TIMESTAMP=1579609942;
 SET @@session.foreign_key_checks=1, @@session.sql_auto_is_null=1, @@session.unique_checks=1;
@@ -229,6 +237,7 @@ SET INSERT_ID=6;
 
 --- start-datetime --
 /*!40019 SET @@session.max_insert_delayed_threads=0*/;
+ROLLBACK;
 SET INSERT_ID=3;
 use test;
 SET TIMESTAMP=1579609944;
@@ -252,6 +261,7 @@ insert into t1 values(null, "f");
 
 --- stop-datetime --
 /*!40019 SET @@session.max_insert_delayed_threads=0*/;
+ROLLBACK;
 use test;
 SET TIMESTAMP=1579609942;
 SET @@session.foreign_key_checks=1, @@session.sql_auto_is_null=1, @@session.unique_checks=1;
@@ -267,6 +277,7 @@ insert into t1 values(null, "b");
 
 --- Remote --
 /*!40019 SET @@session.max_insert_delayed_threads=0*/;
+ROLLBACK;
 use test;
 SET TIMESTAMP=1579609942;
 SET @@session.foreign_key_checks=1, @@session.sql_auto_is_null=1, @@session.unique_checks=1;
@@ -291,6 +302,7 @@ insert into t1 values(null, "e");
 
 --- offset --
 /*!40019 SET @@session.max_insert_delayed_threads=0*/;
+ROLLBACK;
 SET INSERT_ID=1;
 use test;
 SET TIMESTAMP=1579609942;
@@ -326,6 +338,7 @@ insert into t1 values(null, "e");
 
 --- stop-position --
 /*!40019 SET @@session.max_insert_delayed_threads=0*/;
+ROLLBACK;
 use test;
 SET TIMESTAMP=1579609942;
 SET @@session.foreign_key_checks=1, @@session.sql_auto_is_null=1, @@session.unique_checks=1;
@@ -344,6 +357,7 @@ insert into t1 values(null, "c");
 
 --- start-datetime --
 /*!40019 SET @@session.max_insert_delayed_threads=0*/;
+ROLLBACK;
 SET INSERT_ID=3;
 use test;
 SET TIMESTAMP=1579609944;
@@ -360,6 +374,7 @@ insert into t1 values(null, "e");
 
 --- stop-datetime --
 /*!40019 SET @@session.max_insert_delayed_threads=0*/;
+ROLLBACK;
 use test;
 SET TIMESTAMP=1579609942;
 SET @@session.foreign_key_checks=1, @@session.sql_auto_is_null=1, @@session.unique_checks=1;
@@ -375,6 +390,7 @@ insert into t1 values(null, "b");
 
 --- Remote with 2 binlogs on command line --
 /*!40019 SET @@session.max_insert_delayed_threads=0*/;
+ROLLBACK;
 use test;
 SET TIMESTAMP=1579609942;
 SET @@session.foreign_key_checks=1, @@session.sql_auto_is_null=1, @@session.unique_checks=1;
@@ -406,6 +422,7 @@ insert into t1 values(null, "f");
 
 --- offset --
 /*!40019 SET @@session.max_insert_delayed_threads=0*/;
+ROLLBACK;
 SET INSERT_ID=1;
 use test;
 SET TIMESTAMP=1579609942;
@@ -455,6 +472,7 @@ insert into t1 values(null, "f");
 
 --- stop-position --
 /*!40019 SET @@session.max_insert_delayed_threads=0*/;
+ROLLBACK;
 use test;
 SET TIMESTAMP=1579609942;
 SET @@session.foreign_key_checks=1, @@session.sql_auto_is_null=1, @@session.unique_checks=1;
@@ -480,6 +498,7 @@ SET INSERT_ID=6;
 
 --- start-datetime --
 /*!40019 SET @@session.max_insert_delayed_threads=0*/;
+ROLLBACK;
 SET INSERT_ID=3;
 use test;
 SET TIMESTAMP=1579609944;
@@ -503,6 +522,7 @@ insert into t1 values(null, "f");
 
 --- stop-datetime --
 /*!40019 SET @@session.max_insert_delayed_threads=0*/;
+ROLLBACK;
 use test;
 SET TIMESTAMP=1579609942;
 SET @@session.foreign_key_checks=1, @@session.sql_auto_is_null=1, @@session.unique_checks=1;
@@ -518,6 +538,7 @@ insert into t1 values(null, "b");
 
 --- to-last-log --
 /*!40019 SET @@session.max_insert_delayed_threads=0*/;
+ROLLBACK;
 use test;
 SET TIMESTAMP=1579609942;
 SET @@session.foreign_key_checks=1, @@session.sql_auto_is_null=1, @@session.unique_checks=1;

mysql-test/r/rpl000015.result

@@ -1,7 +1,7 @@
 reset master;
 show master status;
 File	Position	Binlog_Do_DB	Binlog_Ignore_DB
-master-bin.000001	95		
+master-bin.000001	96		
 reset slave;
 show slave status;
 Slave_IO_State	Master_Host	Master_User	Master_Port	Connect_Retry	Master_Log_File	Read_Master_Log_Pos	Relay_Log_File	Relay_Log_Pos	Relay_Master_Log_File	Slave_IO_Running	Slave_SQL_Running	Replicate_Do_DB	Replicate_Ignore_DB	Replicate_Do_Table	Replicate_Ignore_Table	Replicate_Wild_Do_Table	Replicate_Wild_Ignore_Table	Last_Errno	Last_Error	Skip_Counter	Exec_Master_Log_Pos	Relay_Log_Space	Until_Condition	Until_Log_File	Until_Log_Pos	Master_SSL_Allowed	Master_SSL_CA_File	Master_SSL_CA_Path	Master_SSL_Cert	Master_SSL_Cipher	Master_SSL_Key	Seconds_Behind_Master
@@ -17,7 +17,7 @@ Slave_IO_State	Master_Host	Master_User	Master_Port	Connect_Retry	Master_Log_File
 start slave;
 show slave status;
 Slave_IO_State	Master_Host	Master_User	Master_Port	Connect_Retry	Master_Log_File	Read_Master_Log_Pos	Relay_Log_File	Relay_Log_Pos	Relay_Master_Log_File	Slave_IO_Running	Slave_SQL_Running	Replicate_Do_DB	Replicate_Ignore_DB	Replicate_Do_Table	Replicate_Ignore_Table	Replicate_Wild_Do_Table	Replicate_Wild_Ignore_Table	Last_Errno	Last_Error	Skip_Counter	Exec_Master_Log_Pos	Relay_Log_Space	Until_Condition	Until_Log_File	Until_Log_Pos	Master_SSL_Allowed	Master_SSL_CA_File	Master_SSL_CA_Path	Master_SSL_Cert	Master_SSL_Cipher	Master_SSL_Key	Seconds_Behind_Master
-#	127.0.0.1	root	MASTER_PORT	7	master-bin.000001	95	#	#	master-bin.000001	Yes	Yes							0		0	95	#	None		0	No						#
+#	127.0.0.1	root	MASTER_PORT	7	master-bin.000001	96	#	#	master-bin.000001	Yes	Yes							0		0	96	#	None		0	No						#
 drop table if exists t1;
 create table t1 (n int);
 insert into t1 values (10),(45),(90);

mysql-test/r/rpl_change_master.result

@@ -16,11 +16,11 @@ n
 1
 show slave status;
 Slave_IO_State	Master_Host	Master_User	Master_Port	Connect_Retry	Master_Log_File	Read_Master_Log_Pos	Relay_Log_File	Relay_Log_Pos	Relay_Master_Log_File	Slave_IO_Running	Slave_SQL_Running	Replicate_Do_DB	Replicate_Ignore_DB	Replicate_Do_Table	Replicate_Ignore_Table	Replicate_Wild_Do_Table	Replicate_Wild_Ignore_Table	Last_Errno	Last_Error	Skip_Counter	Exec_Master_Log_Pos	Relay_Log_Space	Until_Condition	Until_Log_File	Until_Log_Pos	Master_SSL_Allowed	Master_SSL_CA_File	Master_SSL_CA_Path	Master_SSL_Cert	Master_SSL_Cipher	Master_SSL_Key	Seconds_Behind_Master
-#	127.0.0.1	root	MASTER_MYPORT	1	master-bin.000001	376	#	#	master-bin.000001	No	No							0		0	288	#	None		0	No						#
+#	127.0.0.1	root	MASTER_MYPORT	1	master-bin.000001	377	#	#	master-bin.000001	No	No							0		0	289	#	None		0	No						#
 change master to master_user='root';
 show slave status;
 Slave_IO_State	Master_Host	Master_User	Master_Port	Connect_Retry	Master_Log_File	Read_Master_Log_Pos	Relay_Log_File	Relay_Log_Pos	Relay_Master_Log_File	Slave_IO_Running	Slave_SQL_Running	Replicate_Do_DB	Replicate_Ignore_DB	Replicate_Do_Table	Replicate_Ignore_Table	Replicate_Wild_Do_Table	Replicate_Wild_Ignore_Table	Last_Errno	Last_Error	Skip_Counter	Exec_Master_Log_Pos	Relay_Log_Space	Until_Condition	Until_Log_File	Until_Log_Pos	Master_SSL_Allowed	Master_SSL_CA_File	Master_SSL_CA_Path	Master_SSL_Cert	Master_SSL_Cipher	Master_SSL_Key	Seconds_Behind_Master
-#	127.0.0.1	root	MASTER_MYPORT	1	master-bin.000001	288	#	#	master-bin.000001	No	No							0		0	288	#	None		0	No						#
+#	127.0.0.1	root	MASTER_MYPORT	1	master-bin.000001	289	#	#	master-bin.000001	No	No							0		0	289	#	None		0	No						#
 select release_lock("a");
 release_lock("a")
 1

mysql-test/r/rpl_charset.result

@@ -103,7 +103,7 @@ a	b
 1	cp850_general_ci
 drop database mysqltest2;
 drop database mysqltest3;
-show binlog events from 95;
+show binlog events from 96;
 Log_name	Pos	Event_type	Server_id	End_log_pos	Info
 master-bin.000001	#	Query	1	#	drop database if exists mysqltest2
 master-bin.000001	#	Query	1	#	drop database if exists mysqltest3
@@ -175,6 +175,7 @@ select hex(c1), hex(c2) from t1;
 hex(c1)	hex(c2)
 CDF32C20E7E020F0FBE1E0EBEAF3	CDF32C20E7E020F0FBE1E0EBEAF3
 /*!40019 SET @@session.max_insert_delayed_threads=0*/;
+ROLLBACK;
 SET TIMESTAMP=1000000000;
 SET @@session.foreign_key_checks=1, @@session.sql_auto_is_null=1, @@session.unique_checks=1;
 SET @@session.sql_mode=0;

mysql-test/r/rpl_error_ignored_table.result

@@ -9,7 +9,7 @@ insert into t1 values (1),(1);
 ERROR 23000: Duplicate entry '1' for key 1
 show slave status;
 Slave_IO_State	Master_Host	Master_User	Master_Port	Connect_Retry	Master_Log_File	Read_Master_Log_Pos	Relay_Log_File	Relay_Log_Pos	Relay_Master_Log_File	Slave_IO_Running	Slave_SQL_Running	Replicate_Do_DB	Replicate_Ignore_DB	Replicate_Do_Table	Replicate_Ignore_Table	Replicate_Wild_Do_Table	Replicate_Wild_Ignore_Table	Last_Errno	Last_Error	Skip_Counter	Exec_Master_Log_Pos	Relay_Log_Space	Until_Condition	Until_Log_File	Until_Log_Pos	Master_SSL_Allowed	Master_SSL_CA_File	Master_SSL_CA_Path	Master_SSL_Cert	Master_SSL_Cipher	Master_SSL_Key	Seconds_Behind_Master
-#	127.0.0.1	root	MASTER_PORT	1	master-bin.000001	287	#	#	master-bin.000001	Yes	Yes				test.t3,test.t1,test.t2			0		0	287	#	None		0	No						#
+#	127.0.0.1	root	MASTER_PORT	1	master-bin.000001	288	#	#	master-bin.000001	Yes	Yes				test.t3,test.t1,test.t2			0		0	288	#	None		0	No						#
 show tables like 't1';
 Tables_in_test (t1)
 drop table t1;
@@ -26,14 +26,14 @@ select (@id := id) - id from t3;
 0
 kill @id;
 drop table t2,t3;
-show binlog events from 95;
+show binlog events from 96;
 Log_name	Pos	Event_type	Server_id	End_log_pos	Info
-master-bin.000001	95	Query	1	194	use `test`; create table t1 (a int primary key)
+master-bin.000001	96	Query	1	195	use `test`; create table t1 (a int primary key)
-master-bin.000001	194	Query	1	287	use `test`; insert into t1 values (1),(1)
+master-bin.000001	195	Query	1	288	use `test`; insert into t1 values (1),(1)
-master-bin.000001	287	Query	1	364	use `test`; drop table t1
+master-bin.000001	288	Query	1	365	use `test`; drop table t1
-master-bin.000001	364	Query	1	463	use `test`; create table t2 (a int primary key)
+master-bin.000001	365	Query	1	464	use `test`; create table t2 (a int primary key)
-master-bin.000001	463	Query	1	551	use `test`; insert into t2 values(1)
+master-bin.000001	464	Query	1	552	use `test`; insert into t2 values(1)
-master-bin.000001	551	Query	1	639	use `test`; create table t3 (id int)
+master-bin.000001	552	Query	1	640	use `test`; create table t3 (id int)
-master-bin.000001	639	Query	1	741	use `test`; insert into t3 values(connection_id())
+master-bin.000001	640	Query	1	742	use `test`; insert into t3 values(connection_id())
-master-bin.000001	741	Query	1	861	use `test`; update t2 set a = a + 1 + get_lock('crash_lock%20C', 10)
+master-bin.000001	742	Query	1	862	use `test`; update t2 set a = a + 1 + get_lock('crash_lock%20C', 10)
-master-bin.000001	861	Query	1	941	use `test`; drop table t2,t3
+master-bin.000001	862	Query	1	942	use `test`; drop table t2,t3

mysql-test/r/rpl_flush_log_loop.result

@@ -14,4 +14,4 @@ start slave;
 flush logs;
 show slave status;
 Slave_IO_State	Master_Host	Master_User	Master_Port	Connect_Retry	Master_Log_File	Read_Master_Log_Pos	Relay_Log_File	Relay_Log_Pos	Relay_Master_Log_File	Slave_IO_Running	Slave_SQL_Running	Replicate_Do_DB	Replicate_Ignore_DB	Replicate_Do_Table	Replicate_Ignore_Table	Replicate_Wild_Do_Table	Replicate_Wild_Ignore_Table	Last_Errno	Last_Error	Skip_Counter	Exec_Master_Log_Pos	Relay_Log_Space	Until_Condition	Until_Log_File	Until_Log_Pos	Master_SSL_Allowed	Master_SSL_CA_File	Master_SSL_CA_Path	Master_SSL_Cert	Master_SSL_Cipher	Master_SSL_Key	Seconds_Behind_Master
-#	127.0.0.1	root	SLAVE_PORT	60	slave-bin.000001	206	#	#	slave-bin.000001	Yes	Yes							0		0	206	#	None		0	No						#
+#	127.0.0.1	root	SLAVE_PORT	60	slave-bin.000001	207	#	#	slave-bin.000001	Yes	Yes							0		0	207	#	None		0	No						#

mysql-test/r/rpl_flush_tables.result

@@ -14,27 +14,27 @@ rename table t1 to t5, t2 to t1;
 flush no_write_to_binlog tables;
 show binlog events;
 Log_name	Pos	Event_type	Server_id	End_log_pos	Info
-master-bin.000001	4	Format_desc	1	95	Server ver: SERVER_VERSION, Binlog ver: 4
+master-bin.000001	4	Format_desc	1	96	Server ver: SERVER_VERSION, Binlog ver: 4
-master-bin.000001	95	Query	1	182	use `test`; create table t1 (a int)
+master-bin.000001	96	Query	1	183	use `test`; create table t1 (a int)
-master-bin.000001	182	Query	1	272	use `test`; insert into t1 values (10)
+master-bin.000001	183	Query	1	273	use `test`; insert into t1 values (10)
-master-bin.000001	272	Query	1	359	use `test`; create table t2 (a int)
+master-bin.000001	273	Query	1	360	use `test`; create table t2 (a int)
-master-bin.000001	359	Query	1	469	use `test`; create table t3 (a int) engine=merge union(t1)
+master-bin.000001	360	Query	1	470	use `test`; create table t3 (a int) engine=merge union(t1)
-master-bin.000001	469	Query	1	556	use `test`; create table t4 (a int)
+master-bin.000001	470	Query	1	557	use `test`; create table t4 (a int)
-master-bin.000001	556	Query	1	651	use `test`; insert into t4 select * from t3
+master-bin.000001	557	Query	1	652	use `test`; insert into t4 select * from t3
-master-bin.000001	651	Query	1	746	use `test`; rename table t1 to t5, t2 to t1
+master-bin.000001	652	Query	1	747	use `test`; rename table t1 to t5, t2 to t1
 select * from t3;
 a
 flush tables;
 show binlog events;
 Log_name	Pos	Event_type	Server_id	End_log_pos	Info
-master-bin.000001	4	Format_desc	1	95	Server ver: SERVER_VERSION, Binlog ver: 4
+master-bin.000001	4	Format_desc	1	96	Server ver: SERVER_VERSION, Binlog ver: 4
-master-bin.000001	95	Query	1	182	use `test`; create table t1 (a int)
+master-bin.000001	96	Query	1	183	use `test`; create table t1 (a int)
-master-bin.000001	182	Query	1	272	use `test`; insert into t1 values (10)
+master-bin.000001	183	Query	1	273	use `test`; insert into t1 values (10)
-master-bin.000001	272	Query	1	359	use `test`; create table t2 (a int)
+master-bin.000001	273	Query	1	360	use `test`; create table t2 (a int)
-master-bin.000001	359	Query	1	469	use `test`; create table t3 (a int) engine=merge union(t1)
+master-bin.000001	360	Query	1	470	use `test`; create table t3 (a int) engine=merge union(t1)
-master-bin.000001	469	Query	1	556	use `test`; create table t4 (a int)
+master-bin.000001	470	Query	1	557	use `test`; create table t4 (a int)
-master-bin.000001	556	Query	1	651	use `test`; insert into t4 select * from t3
+master-bin.000001	557	Query	1	652	use `test`; insert into t4 select * from t3
-master-bin.000001	651	Query	1	746	use `test`; rename table t1 to t5, t2 to t1
+master-bin.000001	652	Query	1	747	use `test`; rename table t1 to t5, t2 to t1
-master-bin.000001	746	Query	1	822	use `test`; flush tables
+master-bin.000001	747	Query	1	823	use `test`; flush tables
 select * from t3;
 a

mysql-test/r/rpl_loaddata.result

@@ -22,7 +22,7 @@ day	id	category	name
 2003-03-22	2416	a	bbbbb
 show master status;
 File	Position	Binlog_Do_DB	Binlog_Ignore_DB
-slave-bin.000001	1096		
+slave-bin.000001	1097		
 drop table t1;
 drop table t2;
 drop table t3;
@@ -33,7 +33,7 @@ set global sql_slave_skip_counter=1;
 start slave;
 show slave status;
 Slave_IO_State	Master_Host	Master_User	Master_Port	Connect_Retry	Master_Log_File	Read_Master_Log_Pos	Relay_Log_File	Relay_Log_Pos	Relay_Master_Log_File	Slave_IO_Running	Slave_SQL_Running	Replicate_Do_DB	Replicate_Ignore_DB	Replicate_Do_Table	Replicate_Ignore_Table	Replicate_Wild_Do_Table	Replicate_Wild_Ignore_Table	Last_Errno	Last_Error	Skip_Counter	Exec_Master_Log_Pos	Relay_Log_Space	Until_Condition	Until_Log_File	Until_Log_Pos	Master_SSL_Allowed	Master_SSL_CA_File	Master_SSL_CA_Path	Master_SSL_Cert	Master_SSL_Cipher	Master_SSL_Key	Seconds_Behind_Master
-#	127.0.0.1	root	MASTER_PORT	1	master-bin.000001	1559	#	#	master-bin.000001	Yes	Yes							0		0	1559	#	None		0	No						#
+#	127.0.0.1	root	MASTER_PORT	1	master-bin.000001	1560	#	#	master-bin.000001	Yes	Yes							0		0	1560	#	None		0	No						#
 set sql_log_bin=0;
 delete from t1;
 set sql_log_bin=1;
@@ -43,7 +43,7 @@ change master to master_user='test';
 change master to master_user='root';
 show slave status;
 Slave_IO_State	Master_Host	Master_User	Master_Port	Connect_Retry	Master_Log_File	Read_Master_Log_Pos	Relay_Log_File	Relay_Log_Pos	Relay_Master_Log_File	Slave_IO_Running	Slave_SQL_Running	Replicate_Do_DB	Replicate_Ignore_DB	Replicate_Do_Table	Replicate_Ignore_Table	Replicate_Wild_Do_Table	Replicate_Wild_Ignore_Table	Last_Errno	Last_Error	Skip_Counter	Exec_Master_Log_Pos	Relay_Log_Space	Until_Condition	Until_Log_File	Until_Log_Pos	Master_SSL_Allowed	Master_SSL_CA_File	Master_SSL_CA_Path	Master_SSL_Cert	Master_SSL_Cipher	Master_SSL_Key	Seconds_Behind_Master
-#	127.0.0.1	root	MASTER_PORT	1	master-bin.000001	1667	#	#	master-bin.000001	No	No							0		0	1667	#	None		0	No						#
+#	127.0.0.1	root	MASTER_PORT	1	master-bin.000001	1668	#	#	master-bin.000001	No	No							0		0	1668	#	None		0	No						#
 set global sql_slave_skip_counter=1;
 start slave;
 set sql_log_bin=0;
@@ -64,5 +64,5 @@ terminated by ',' optionally enclosed by '%' escaped by '@' lines terminated by
 ERROR 23000: Duplicate entry '2003-03-22' for key 1
 show master status;
 File	Position	Binlog_Do_DB	Binlog_Ignore_DB
-master-bin.000001	536		
+master-bin.000001	537		
 drop table t2;

mysql-test/r/rpl_loaddata_rule_m.result

@@ -10,8 +10,8 @@ create database mysqltest;
 create table t1(a int, b int, unique(b));
 use mysqltest;
 load data infile '../../std_data/rpl_loaddata.dat' into table test.t1;
-show binlog events from 95;
+show binlog events from 96;
 Log_name	Pos	Event_type	Server_id	End_log_pos	Info
-master-bin.000001	95	Query	1	197	drop database if exists mysqltest
+master-bin.000001	96	Query	1	198	drop database if exists mysqltest
-master-bin.000001	197	Query	1	291	create database mysqltest
+master-bin.000001	198	Query	1	292	create database mysqltest
 drop database mysqltest;

mysql-test/r/rpl_loaddata_rule_s.result

@@ -10,5 +10,5 @@ load data infile '../../std_data/rpl_loaddata.dat' into table test.t1;
 select count(*) from t1;
 count(*)
 2
-show binlog events from 95;
+show binlog events from 96;
 Log_name	Pos	Event_type	Server_id	End_log_pos	Info

mysql-test/r/rpl_log.result

@@ -19,25 +19,25 @@ count(*)
 drop table t1;
 show binlog events;
 Log_name	Pos	Event_type	Server_id	End_log_pos	Info
-master-bin.000001	4	Format_desc	1	95	Server ver: VERSION, Binlog ver: 4
+master-bin.000001	4	Format_desc	1	96	Server ver: VERSION, Binlog ver: 4
-master-bin.000001	95	Query	1	217	use `test`; create table t1(n int not null auto_increment primary key)
+master-bin.000001	96	Query	1	218	use `test`; create table t1(n int not null auto_increment primary key)
-master-bin.000001	217	Intvar	1	245	INSERT_ID=1
+master-bin.000001	218	Intvar	1	246	INSERT_ID=1
-master-bin.000001	245	Query	1	337	use `test`; insert into t1 values (NULL)
+master-bin.000001	246	Query	1	338	use `test`; insert into t1 values (NULL)
-master-bin.000001	337	Query	1	414	use `test`; drop table t1
+master-bin.000001	338	Query	1	415	use `test`; drop table t1
-master-bin.000001	414	Query	1	518	use `test`; create table t1 (word char(20) not null)
+master-bin.000001	415	Query	1	519	use `test`; create table t1 (word char(20) not null)
-master-bin.000001	518	Create_file	1	1188	db=test;table=t1;file_id=1;block_len=581
+master-bin.000001	519	Create_file	1	1189	db=test;table=t1;file_id=1;block_len=581
-master-bin.000001	1188	Exec_load	1	1211	;file_id=1
+master-bin.000001	1189	Exec_load	1	1212	;file_id=1
-master-bin.000001	1211	Query	1	1288	use `test`; drop table t1
+master-bin.000001	1212	Query	1	1289	use `test`; drop table t1
-show binlog events from 95 limit 1;
+show binlog events from 96 limit 1;
 Log_name	Pos	Event_type	Server_id	End_log_pos	Info
-master-bin.000001	95	Query	1	217	use `test`; create table t1(n int not null auto_increment primary key)
+master-bin.000001	96	Query	1	218	use `test`; create table t1(n int not null auto_increment primary key)
-show binlog events from 95 limit 2;
+show binlog events from 96 limit 2;
 Log_name	Pos	Event_type	Server_id	End_log_pos	Info
-master-bin.000001	95	Query	1	217	use `test`; create table t1(n int not null auto_increment primary key)
+master-bin.000001	96	Query	1	218	use `test`; create table t1(n int not null auto_increment primary key)
-master-bin.000001	217	Intvar	1	245	INSERT_ID=1
+master-bin.000001	218	Intvar	1	246	INSERT_ID=1
-show binlog events from 95 limit 2,1;
+show binlog events from 96 limit 2,1;
 Log_name	Pos	Event_type	Server_id	End_log_pos	Info
-master-bin.000001	245	Query	1	337	use `test`; insert into t1 values (NULL)
+master-bin.000001	246	Query	1	338	use `test`; insert into t1 values (NULL)
 flush logs;
 create table t5 (a int);
 drop table t5;
@@ -49,24 +49,24 @@ insert into t1 values (1);
 drop table t1;
 show binlog events;
 Log_name	Pos	Event_type	Server_id	End_log_pos	Info
-master-bin.000001	4	Format_desc	1	95	Server ver: VERSION, Binlog ver: 4
+master-bin.000001	4	Format_desc	1	96	Server ver: VERSION, Binlog ver: 4
-master-bin.000001	95	Query	1	217	use `test`; create table t1(n int not null auto_increment primary key)
+master-bin.000001	96	Query	1	218	use `test`; create table t1(n int not null auto_increment primary key)
-master-bin.000001	217	Intvar	1	245	INSERT_ID=1
+master-bin.000001	218	Intvar	1	246	INSERT_ID=1
-master-bin.000001	245	Query	1	337	use `test`; insert into t1 values (NULL)
+master-bin.000001	246	Query	1	338	use `test`; insert into t1 values (NULL)
-master-bin.000001	337	Query	1	414	use `test`; drop table t1
+master-bin.000001	338	Query	1	415	use `test`; drop table t1
-master-bin.000001	414	Query	1	518	use `test`; create table t1 (word char(20) not null)
+master-bin.000001	415	Query	1	519	use `test`; create table t1 (word char(20) not null)
-master-bin.000001	518	Create_file	1	1188	db=test;table=t1;file_id=1;block_len=581
+master-bin.000001	519	Create_file	1	1189	db=test;table=t1;file_id=1;block_len=581
-master-bin.000001	1188	Exec_load	1	1211	;file_id=1
+master-bin.000001	1189	Exec_load	1	1212	;file_id=1
-master-bin.000001	1211	Query	1	1288	use `test`; drop table t1
+master-bin.000001	1212	Query	1	1289	use `test`; drop table t1
-master-bin.000001	1288	Rotate	1	1332	master-bin.000002;pos=4
+master-bin.000001	1289	Rotate	1	1333	master-bin.000002;pos=4
 show binlog events in 'master-bin.000002';
 Log_name	Pos	Event_type	Server_id	End_log_pos	Info
-master-bin.000002	4	Format_desc	1	95	Server ver: VERSION, Binlog ver: 4
+master-bin.000002	4	Format_desc	1	96	Server ver: VERSION, Binlog ver: 4
-master-bin.000002	95	Query	1	182	use `test`; create table t5 (a int)
+master-bin.000002	96	Query	1	183	use `test`; create table t5 (a int)
-master-bin.000002	182	Query	1	259	use `test`; drop table t5
+master-bin.000002	183	Query	1	260	use `test`; drop table t5
-master-bin.000002	259	Query	1	346	use `test`; create table t1 (n int)
+master-bin.000002	260	Query	1	347	use `test`; create table t1 (n int)
-master-bin.000002	346	Query	1	435	use `test`; insert into t1 values (1)
+master-bin.000002	347	Query	1	436	use `test`; insert into t1 values (1)
-master-bin.000002	435	Query	1	512	use `test`; drop table t1
+master-bin.000002	436	Query	1	513	use `test`; drop table t1
 show binary logs;
 Log_name
 master-bin.000001
@@ -78,26 +78,26 @@ slave-bin.000001
 slave-bin.000002
 show binlog events in 'slave-bin.000001' from 4;
 Log_name	Pos	Event_type	Server_id	End_log_pos	Info
-slave-bin.000001	4	Format_desc	2	95	Server ver: VERSION, Binlog ver: 4
+slave-bin.000001	4	Format_desc	2	96	Server ver: VERSION, Binlog ver: 4
-slave-bin.000001	95	Query	1	217	use `test`; create table t1(n int not null auto_increment primary key)
+slave-bin.000001	96	Query	1	218	use `test`; create table t1(n int not null auto_increment primary key)
-slave-bin.000001	217	Intvar	1	245	INSERT_ID=1
+slave-bin.000001	218	Intvar	1	246	INSERT_ID=1
-slave-bin.000001	245	Query	1	337	use `test`; insert into t1 values (NULL)
+slave-bin.000001	246	Query	1	338	use `test`; insert into t1 values (NULL)
-slave-bin.000001	337	Query	1	414	use `test`; drop table t1
+slave-bin.000001	338	Query	1	415	use `test`; drop table t1
-slave-bin.000001	414	Query	1	518	use `test`; create table t1 (word char(20) not null)
+slave-bin.000001	415	Query	1	519	use `test`; create table t1 (word char(20) not null)
-slave-bin.000001	518	Create_file	1	1197	db=test;table=t1;file_id=1;block_len=581
+slave-bin.000001	519	Create_file	1	1198	db=test;table=t1;file_id=1;block_len=581
-slave-bin.000001	1197	Exec_load	1	1220	;file_id=1
+slave-bin.000001	1198	Exec_load	1	1221	;file_id=1
-slave-bin.000001	1220	Query	1	1297	use `test`; drop table t1
+slave-bin.000001	1221	Query	1	1298	use `test`; drop table t1
-slave-bin.000001	1297	Query	1	1384	use `test`; create table t5 (a int)
+slave-bin.000001	1298	Query	1	1385	use `test`; create table t5 (a int)
-slave-bin.000001	1384	Query	1	1461	use `test`; drop table t5
+slave-bin.000001	1385	Query	1	1462	use `test`; drop table t5
-slave-bin.000001	1461	Rotate	2	1504	slave-bin.000002;pos=4
+slave-bin.000001	1462	Rotate	2	1505	slave-bin.000002;pos=4
 show binlog events in 'slave-bin.000002' from 4;
 Log_name	Pos	Event_type	Server_id	End_log_pos	Info
-slave-bin.000002	4	Format_desc	2	95	Server ver: VERSION, Binlog ver: 4
+slave-bin.000002	4	Format_desc	2	96	Server ver: VERSION, Binlog ver: 4
-slave-bin.000002	95	Query	1	182	use `test`; create table t1 (n int)
+slave-bin.000002	96	Query	1	183	use `test`; create table t1 (n int)
-slave-bin.000002	182	Query	1	271	use `test`; insert into t1 values (1)
+slave-bin.000002	183	Query	1	272	use `test`; insert into t1 values (1)
-slave-bin.000002	271	Query	1	348	use `test`; drop table t1
+slave-bin.000002	272	Query	1	349	use `test`; drop table t1
 show slave status;
 Slave_IO_State	Master_Host	Master_User	Master_Port	Connect_Retry	Master_Log_File	Read_Master_Log_Pos	Relay_Log_File	Relay_Log_Pos	Relay_Master_Log_File	Slave_IO_Running	Slave_SQL_Running	Replicate_Do_DB	Replicate_Ignore_DB	Replicate_Do_Table	Replicate_Ignore_Table	Replicate_Wild_Do_Table	Replicate_Wild_Ignore_Table	Last_Errno	Last_Error	Skip_Counter	Exec_Master_Log_Pos	Relay_Log_Space	Until_Condition	Until_Log_File	Until_Log_Pos	Master_SSL_Allowed	Master_SSL_CA_File	Master_SSL_CA_Path	Master_SSL_Cert	Master_SSL_Cipher	Master_SSL_Key	Seconds_Behind_Master
-#	127.0.0.1	root	MASTER_PORT	1	master-bin.000002	512	#	#	master-bin.000002	Yes	Yes							0		0	512	#	None		0	No						#
+#	127.0.0.1	root	MASTER_PORT	1	master-bin.000002	513	#	#	master-bin.000002	Yes	Yes							0		0	513	#	None		0	No						#
 show binlog events in 'slave-bin.000005' from 4;
 ERROR HY000: Error when executing command SHOW BINLOG EVENTS: Could not find target log

mysql-test/r/rpl_log_pos.result

@@ -6,10 +6,10 @@ drop table if exists t1,t2,t3,t4,t5,t6,t7,t8,t9;
 start slave;
 show master status;
 File	Position	Binlog_Do_DB	Binlog_Ignore_DB
-master-bin.000001	95		
+master-bin.000001	96		
 show slave status;
 Slave_IO_State	Master_Host	Master_User	Master_Port	Connect_Retry	Master_Log_File	Read_Master_Log_Pos	Relay_Log_File	Relay_Log_Pos	Relay_Master_Log_File	Slave_IO_Running	Slave_SQL_Running	Replicate_Do_DB	Replicate_Ignore_DB	Replicate_Do_Table	Replicate_Ignore_Table	Replicate_Wild_Do_Table	Replicate_Wild_Ignore_Table	Last_Errno	Last_Error	Skip_Counter	Exec_Master_Log_Pos	Relay_Log_Space	Until_Condition	Until_Log_File	Until_Log_Pos	Master_SSL_Allowed	Master_SSL_CA_File	Master_SSL_CA_Path	Master_SSL_Cert	Master_SSL_Cipher	Master_SSL_Key	Seconds_Behind_Master
-#	127.0.0.1	root	MASTER_PORT	1	master-bin.000001	95	#	#	master-bin.000001	Yes	Yes							0		0	95	#	None		0	No						#
+#	127.0.0.1	root	MASTER_PORT	1	master-bin.000001	96	#	#	master-bin.000001	Yes	Yes							0		0	96	#	None		0	No						#
 stop slave;
 change master to master_log_pos=73;
 start slave;
@@ -30,13 +30,13 @@ Slave_IO_State	Master_Host	Master_User	Master_Port	Connect_Retry	Master_Log_File
 #	127.0.0.1	root	MASTER_PORT	1	master-bin.000001	173	#	#	master-bin.000001	No	Yes							0		0	173	#	None		0	No						#
 show master status;
 File	Position	Binlog_Do_DB	Binlog_Ignore_DB
-master-bin.000001	95		
+master-bin.000001	96		
 create table if not exists t1 (n int);
 drop table if exists t1;
 create table t1 (n int);
 insert into t1 values (1),(2),(3);
 stop slave;
-change master to master_log_pos=95;
+change master to master_log_pos=96;
 start slave;
 select * from t1;
 n

mysql-test/r/rpl_max_relay_size.result

@@ -16,7 +16,7 @@ select @@global.max_relay_log_size;
 start slave;
 show slave status;
 Slave_IO_State	Master_Host	Master_User	Master_Port	Connect_Retry	Master_Log_File	Read_Master_Log_Pos	Relay_Log_File	Relay_Log_Pos	Relay_Master_Log_File	Slave_IO_Running	Slave_SQL_Running	Replicate_Do_DB	Replicate_Ignore_DB	Replicate_Do_Table	Replicate_Ignore_Table	Replicate_Wild_Do_Table	Replicate_Wild_Ignore_Table	Last_Errno	Last_Error	Skip_Counter	Exec_Master_Log_Pos	Relay_Log_Space	Until_Condition	Until_Log_File	Until_Log_Pos	Master_SSL_Allowed	Master_SSL_CA_File	Master_SSL_CA_Path	Master_SSL_Cert	Master_SSL_Cipher	Master_SSL_Key	Seconds_Behind_Master
-#	127.0.0.1	root	MASTER_PORT	1	master-bin.000001	73751	#	#	master-bin.000001	Yes	Yes							0		0	73751	#	None		0	No						#
+#	127.0.0.1	root	MASTER_PORT	1	master-bin.000001	73752	#	#	master-bin.000001	Yes	Yes							0		0	73752	#	None		0	No						#
 stop slave;
 reset slave;
 set global max_relay_log_size=(5*4096);
@@ -26,7 +26,7 @@ select @@global.max_relay_log_size;
 start slave;
 show slave status;
 Slave_IO_State	Master_Host	Master_User	Master_Port	Connect_Retry	Master_Log_File	Read_Master_Log_Pos	Relay_Log_File	Relay_Log_Pos	Relay_Master_Log_File	Slave_IO_Running	Slave_SQL_Running	Replicate_Do_DB	Replicate_Ignore_DB	Replicate_Do_Table	Replicate_Ignore_Table	Replicate_Wild_Do_Table	Replicate_Wild_Ignore_Table	Last_Errno	Last_Error	Skip_Counter	Exec_Master_Log_Pos	Relay_Log_Space	Until_Condition	Until_Log_File	Until_Log_Pos	Master_SSL_Allowed	Master_SSL_CA_File	Master_SSL_CA_Path	Master_SSL_Cert	Master_SSL_Cipher	Master_SSL_Key	Seconds_Behind_Master
-#	127.0.0.1	root	MASTER_PORT	1	master-bin.000001	73751	#	#	master-bin.000001	Yes	Yes							0		0	73751	#	None		0	No						#
+#	127.0.0.1	root	MASTER_PORT	1	master-bin.000001	73752	#	#	master-bin.000001	Yes	Yes							0		0	73752	#	None		0	No						#
 stop slave;
 reset slave;
 set global max_relay_log_size=0;
@@ -36,7 +36,7 @@ select @@global.max_relay_log_size;
 start slave;
 show slave status;
 Slave_IO_State	Master_Host	Master_User	Master_Port	Connect_Retry	Master_Log_File	Read_Master_Log_Pos	Relay_Log_File	Relay_Log_Pos	Relay_Master_Log_File	Slave_IO_Running	Slave_SQL_Running	Replicate_Do_DB	Replicate_Ignore_DB	Replicate_Do_Table	Replicate_Ignore_Table	Replicate_Wild_Do_Table	Replicate_Wild_Ignore_Table	Last_Errno	Last_Error	Skip_Counter	Exec_Master_Log_Pos	Relay_Log_Space	Until_Condition	Until_Log_File	Until_Log_Pos	Master_SSL_Allowed	Master_SSL_CA_File	Master_SSL_CA_Path	Master_SSL_Cert	Master_SSL_Cipher	Master_SSL_Key	Seconds_Behind_Master
-#	127.0.0.1	root	MASTER_PORT	1	master-bin.000001	73751	#	#	master-bin.000001	Yes	Yes							0		0	73751	#	None		0	No						#
+#	127.0.0.1	root	MASTER_PORT	1	master-bin.000001	73752	#	#	master-bin.000001	Yes	Yes							0		0	73752	#	None		0	No						#
 stop slave;
 reset slave;
 flush logs;
@@ -49,13 +49,13 @@ flush logs;
 create table t1 (a int);
 show slave status;
 Slave_IO_State	Master_Host	Master_User	Master_Port	Connect_Retry	Master_Log_File	Read_Master_Log_Pos	Relay_Log_File	Relay_Log_Pos	Relay_Master_Log_File	Slave_IO_Running	Slave_SQL_Running	Replicate_Do_DB	Replicate_Ignore_DB	Replicate_Do_Table	Replicate_Ignore_Table	Replicate_Wild_Do_Table	Replicate_Wild_Ignore_Table	Last_Errno	Last_Error	Skip_Counter	Exec_Master_Log_Pos	Relay_Log_Space	Until_Condition	Until_Log_File	Until_Log_Pos	Master_SSL_Allowed	Master_SSL_CA_File	Master_SSL_CA_Path	Master_SSL_Cert	Master_SSL_Cipher	Master_SSL_Key	Seconds_Behind_Master
-#	127.0.0.1	root	MASTER_PORT	1	master-bin.000001	73838	#	#	master-bin.000001	Yes	Yes							0		0	73838	#	None		0	No						#
+#	127.0.0.1	root	MASTER_PORT	1	master-bin.000001	73839	#	#	master-bin.000001	Yes	Yes							0		0	73839	#	None		0	No						#
 flush logs;
 drop table t1;
 show slave status;
 Slave_IO_State	Master_Host	Master_User	Master_Port	Connect_Retry	Master_Log_File	Read_Master_Log_Pos	Relay_Log_File	Relay_Log_Pos	Relay_Master_Log_File	Slave_IO_Running	Slave_SQL_Running	Replicate_Do_DB	Replicate_Ignore_DB	Replicate_Do_Table	Replicate_Ignore_Table	Replicate_Wild_Do_Table	Replicate_Wild_Ignore_Table	Last_Errno	Last_Error	Skip_Counter	Exec_Master_Log_Pos	Relay_Log_Space	Until_Condition	Until_Log_File	Until_Log_Pos	Master_SSL_Allowed	Master_SSL_CA_File	Master_SSL_CA_Path	Master_SSL_Cert	Master_SSL_Cipher	Master_SSL_Key	Seconds_Behind_Master
-#	127.0.0.1	root	MASTER_PORT	1	master-bin.000001	73915	#	#	master-bin.000001	Yes	Yes							0		0	73915	#	None		0	No						#
+#	127.0.0.1	root	MASTER_PORT	1	master-bin.000001	73916	#	#	master-bin.000001	Yes	Yes							0		0	73916	#	None		0	No						#
 flush logs;
 show master status;
 File	Position	Binlog_Do_DB	Binlog_Ignore_DB
-master-bin.000002	95		
+master-bin.000002	96		

mysql-test/r/rpl_relayrotate.result

@@ -18,5 +18,5 @@ max(a)
 8000
 show slave status;
 Slave_IO_State	Master_Host	Master_User	Master_Port	Connect_Retry	Master_Log_File	Read_Master_Log_Pos	Relay_Log_File	Relay_Log_Pos	Relay_Master_Log_File	Slave_IO_Running	Slave_SQL_Running	Replicate_Do_DB	Replicate_Ignore_DB	Replicate_Do_Table	Replicate_Ignore_Table	Replicate_Wild_Do_Table	Replicate_Wild_Ignore_Table	Last_Errno	Last_Error	Skip_Counter	Exec_Master_Log_Pos	Relay_Log_Space	Until_Condition	Until_Log_File	Until_Log_Pos	Master_SSL_Allowed	Master_SSL_CA_File	Master_SSL_CA_Path	Master_SSL_Cert	Master_SSL_Cipher	Master_SSL_Key	Seconds_Behind_Master
-#	127.0.0.1	root	MASTER_MYPORT	1	master-bin.000001	743228	#	#	master-bin.000001	Yes	Yes							0		0	743228	#	None		0	No						#
+#	127.0.0.1	root	MASTER_MYPORT	1	master-bin.000001	743186	#	#	master-bin.000001	Yes	Yes							0		0	743186	#	None		0	No						#
 drop table t1;

mysql-test/r/rpl_replicate_do.result

@@ -28,4 +28,4 @@ ERROR 42S02: Table 'test.t11' doesn't exist
 drop table if exists t1,t2,t11;
 show slave status;
 Slave_IO_State	Master_Host	Master_User	Master_Port	Connect_Retry	Master_Log_File	Read_Master_Log_Pos	Relay_Log_File	Relay_Log_Pos	Relay_Master_Log_File	Slave_IO_Running	Slave_SQL_Running	Replicate_Do_DB	Replicate_Ignore_DB	Replicate_Do_Table	Replicate_Ignore_Table	Replicate_Wild_Do_Table	Replicate_Wild_Ignore_Table	Last_Errno	Last_Error	Skip_Counter	Exec_Master_Log_Pos	Relay_Log_Space	Until_Condition	Until_Log_File	Until_Log_Pos	Master_SSL_Allowed	Master_SSL_CA_File	Master_SSL_CA_Path	Master_SSL_Cert	Master_SSL_Cipher	Master_SSL_Key	Seconds_Behind_Master
-#	127.0.0.1	root	MASTER_PORT	1	master-bin.000001	1617	#	#	master-bin.000001	Yes	Yes			test.t1				0		0	1617	#	None		0	No						#
+#	127.0.0.1	root	MASTER_PORT	1	master-bin.000001	1618	#	#	master-bin.000001	Yes	Yes			test.t1				0		0	1618	#	None		0	No						#

mysql-test/r/rpl_reset_slave.result

@@ -6,12 +6,12 @@ drop table if exists t1,t2,t3,t4,t5,t6,t7,t8,t9;
 start slave;
 show slave status;
 Slave_IO_State	Master_Host	Master_User	Master_Port	Connect_Retry	Master_Log_File	Read_Master_Log_Pos	Relay_Log_File	Relay_Log_Pos	Relay_Master_Log_File	Slave_IO_Running	Slave_SQL_Running	Replicate_Do_DB	Replicate_Ignore_DB	Replicate_Do_Table	Replicate_Ignore_Table	Replicate_Wild_Do_Table	Replicate_Wild_Ignore_Table	Last_Errno	Last_Error	Skip_Counter	Exec_Master_Log_Pos	Relay_Log_Space	Until_Condition	Until_Log_File	Until_Log_Pos	Master_SSL_Allowed	Master_SSL_CA_File	Master_SSL_CA_Path	Master_SSL_Cert	Master_SSL_Cipher	Master_SSL_Key	Seconds_Behind_Master
-#	127.0.0.1	root	MASTER_PORT	1	master-bin.000001	95	#	#	master-bin.000001	Yes	Yes							0		0	95	#	None		0	No						#
+#	127.0.0.1	root	MASTER_PORT	1	master-bin.000001	96	#	#	master-bin.000001	Yes	Yes							0		0	96	#	None		0	No						#
 stop slave;
 change master to master_user='test';
 show slave status;
 Slave_IO_State	Master_Host	Master_User	Master_Port	Connect_Retry	Master_Log_File	Read_Master_Log_Pos	Relay_Log_File	Relay_Log_Pos	Relay_Master_Log_File	Slave_IO_Running	Slave_SQL_Running	Replicate_Do_DB	Replicate_Ignore_DB	Replicate_Do_Table	Replicate_Ignore_Table	Replicate_Wild_Do_Table	Replicate_Wild_Ignore_Table	Last_Errno	Last_Error	Skip_Counter	Exec_Master_Log_Pos	Relay_Log_Space	Until_Condition	Until_Log_File	Until_Log_Pos	Master_SSL_Allowed	Master_SSL_CA_File	Master_SSL_CA_Path	Master_SSL_Cert	Master_SSL_Cipher	Master_SSL_Key	Seconds_Behind_Master
-#	127.0.0.1	test	MASTER_PORT	1	master-bin.000001	95	#	#	master-bin.000001	No	No							0		0	95	#	None		0	No						#
+#	127.0.0.1	test	MASTER_PORT	1	master-bin.000001	96	#	#	master-bin.000001	No	No							0		0	96	#	None		0	No						#
 reset slave;
 show slave status;
 Slave_IO_State	Master_Host	Master_User	Master_Port	Connect_Retry	Master_Log_File	Read_Master_Log_Pos	Relay_Log_File	Relay_Log_Pos	Relay_Master_Log_File	Slave_IO_Running	Slave_SQL_Running	Replicate_Do_DB	Replicate_Ignore_DB	Replicate_Do_Table	Replicate_Ignore_Table	Replicate_Wild_Do_Table	Replicate_Wild_Ignore_Table	Last_Errno	Last_Error	Skip_Counter	Exec_Master_Log_Pos	Relay_Log_Space	Until_Condition	Until_Log_File	Until_Log_Pos	Master_SSL_Allowed	Master_SSL_CA_File	Master_SSL_CA_Path	Master_SSL_Cert	Master_SSL_Cipher	Master_SSL_Key	Seconds_Behind_Master
@@ -19,7 +19,7 @@ Slave_IO_State	Master_Host	Master_User	Master_Port	Connect_Retry	Master_Log_File
 start slave;
 show slave status;
 Slave_IO_State	Master_Host	Master_User	Master_Port	Connect_Retry	Master_Log_File	Read_Master_Log_Pos	Relay_Log_File	Relay_Log_Pos	Relay_Master_Log_File	Slave_IO_Running	Slave_SQL_Running	Replicate_Do_DB	Replicate_Ignore_DB	Replicate_Do_Table	Replicate_Ignore_Table	Replicate_Wild_Do_Table	Replicate_Wild_Ignore_Table	Last_Errno	Last_Error	Skip_Counter	Exec_Master_Log_Pos	Relay_Log_Space	Until_Condition	Until_Log_File	Until_Log_Pos	Master_SSL_Allowed	Master_SSL_CA_File	Master_SSL_CA_Path	Master_SSL_Cert	Master_SSL_Cipher	Master_SSL_Key	Seconds_Behind_Master
-#	127.0.0.1	root	MASTER_PORT	1	master-bin.000001	95	#	#	master-bin.000001	Yes	Yes							0		0	95	#	None		0	No						#
+#	127.0.0.1	root	MASTER_PORT	1	master-bin.000001	96	#	#	master-bin.000001	Yes	Yes							0		0	96	#	None		0	No						#
 stop slave;
 reset slave;
 start slave;

mysql-test/r/rpl_rotate_logs.result

@@ -16,7 +16,7 @@ create table t1 (s text);
 insert into t1 values('Could not break slave'),('Tried hard');
 show slave status;
 Slave_IO_State	Master_Host	Master_User	Master_Port	Connect_Retry	Master_Log_File	Read_Master_Log_Pos	Relay_Log_File	Relay_Log_Pos	Relay_Master_Log_File	Slave_IO_Running	Slave_SQL_Running	Replicate_Do_DB	Replicate_Ignore_DB	Replicate_Do_Table	Replicate_Ignore_Table	Replicate_Wild_Do_Table	Replicate_Wild_Ignore_Table	Last_Errno	Last_Error	Skip_Counter	Exec_Master_Log_Pos	Relay_Log_Space	Until_Condition	Until_Log_File	Until_Log_Pos	Master_SSL_Allowed	Master_SSL_CA_File	Master_SSL_CA_Path	Master_SSL_Cert	Master_SSL_Cipher	Master_SSL_Key	Seconds_Behind_Master
-#	127.0.0.1	root	MASTER_PORT	60	master-bin.000001	549	#	#	master-bin.000001	Yes	Yes							0		0	549	#	None		0	No						#
+#	127.0.0.1	root	MASTER_PORT	60	master-bin.000001	550	#	#	master-bin.000001	Yes	Yes							0		0	550	#	None		0	No						#
 select * from t1;
 s
 Could not break slave
@@ -57,7 +57,7 @@ master-bin.000003
 insert into t2 values (65);
 show slave status;
 Slave_IO_State	Master_Host	Master_User	Master_Port	Connect_Retry	Master_Log_File	Read_Master_Log_Pos	Relay_Log_File	Relay_Log_Pos	Relay_Master_Log_File	Slave_IO_Running	Slave_SQL_Running	Replicate_Do_DB	Replicate_Ignore_DB	Replicate_Do_Table	Replicate_Ignore_Table	Replicate_Wild_Do_Table	Replicate_Wild_Ignore_Table	Last_Errno	Last_Error	Skip_Counter	Exec_Master_Log_Pos	Relay_Log_Space	Until_Condition	Until_Log_File	Until_Log_Pos	Master_SSL_Allowed	Master_SSL_CA_File	Master_SSL_CA_Path	Master_SSL_Cert	Master_SSL_Cipher	Master_SSL_Key	Seconds_Behind_Master
-#	127.0.0.1	root	MASTER_PORT	60	master-bin.000003	497	#	#	master-bin.000003	Yes	Yes							0		0	497	#	None		0	No						#
+#	127.0.0.1	root	MASTER_PORT	60	master-bin.000003	498	#	#	master-bin.000003	Yes	Yes							0		0	498	#	None		0	No						#
 select * from t2;
 m
 34
@@ -79,13 +79,13 @@ master-bin.000004
 master-bin.000005
 show master status;
 File	Position	Binlog_Do_DB	Binlog_Ignore_DB
-master-bin.000005	2050		
+master-bin.000005	2051		
 select * from t4;
 a
 testing temporary tables part 2
 show slave status;
 Slave_IO_State	Master_Host	Master_User	Master_Port	Connect_Retry	Master_Log_File	Read_Master_Log_Pos	Relay_Log_File	Relay_Log_Pos	Relay_Master_Log_File	Slave_IO_Running	Slave_SQL_Running	Replicate_Do_DB	Replicate_Ignore_DB	Replicate_Do_Table	Replicate_Ignore_Table	Replicate_Wild_Do_Table	Replicate_Wild_Ignore_Table	Last_Errno	Last_Error	Skip_Counter	Exec_Master_Log_Pos	Relay_Log_Space	Until_Condition	Until_Log_File	Until_Log_Pos	Master_SSL_Allowed	Master_SSL_CA_File	Master_SSL_CA_Path	Master_SSL_Cert	Master_SSL_Cipher	Master_SSL_Key	Seconds_Behind_Master
-#	127.0.0.1	root	MASTER_PORT	60	master-bin.000005	2050	#	#	master-bin.000005	Yes	Yes							0		0	2050	#	None		0	No						#
+#	127.0.0.1	root	MASTER_PORT	60	master-bin.000005	2051	#	#	master-bin.000005	Yes	Yes							0		0	2051	#	None		0	No						#
 lock tables t3 read;
 select count(*) from t3 where n >= 4;
 count(*)

mysql-test/r/rpl_server_id1.result

@@ -10,7 +10,7 @@ stop slave;
 change master to master_port=SLAVE_PORT;
 show slave status;
 Slave_IO_State	Master_Host	Master_User	Master_Port	Connect_Retry	Master_Log_File	Read_Master_Log_Pos	Relay_Log_File	Relay_Log_Pos	Relay_Master_Log_File	Slave_IO_Running	Slave_SQL_Running	Replicate_Do_DB	Replicate_Ignore_DB	Replicate_Do_Table	Replicate_Ignore_Table	Replicate_Wild_Do_Table	Replicate_Wild_Ignore_Table	Last_Errno	Last_Error	Skip_Counter	Exec_Master_Log_Pos	Relay_Log_Space	Until_Condition	Until_Log_File	Until_Log_Pos	Master_SSL_Allowed	Master_SSL_CA_File	Master_SSL_CA_Path	Master_SSL_Cert	Master_SSL_Cipher	Master_SSL_Key	Seconds_Behind_Master
-	127.0.0.1	root	SLAVE_PORT	1		4	slave-relay-bin.000001	4		No	No						#	0		0	0	95	None		0	No						NULL
+	127.0.0.1	root	SLAVE_PORT	1		4	slave-relay-bin.000001	4		No	No						#	0		0	0	96	None		0	No						NULL
 start slave;
 insert into t1 values (1);
 show status like "slave_running";

mysql-test/r/rpl_server_id2.result

@@ -10,7 +10,7 @@ stop slave;
 change master to master_port=SLAVE_PORT;
 show slave status;
 Slave_IO_State	Master_Host	Master_User	Master_Port	Connect_Retry	Master_Log_File	Read_Master_Log_Pos	Relay_Log_File	Relay_Log_Pos	Relay_Master_Log_File	Slave_IO_Running	Slave_SQL_Running	Replicate_Do_DB	Replicate_Ignore_DB	Replicate_Do_Table	Replicate_Ignore_Table	Replicate_Wild_Do_Table	Replicate_Wild_Ignore_Table	Last_Errno	Last_Error	Skip_Counter	Exec_Master_Log_Pos	Relay_Log_Space	Until_Condition	Until_Log_File	Until_Log_Pos	Master_SSL_Allowed	Master_SSL_CA_File	Master_SSL_CA_Path	Master_SSL_Cert	Master_SSL_Cipher	Master_SSL_Key	Seconds_Behind_Master
-	127.0.0.1	root	SLAVE_PORT	1		4	slave-relay-bin.000001	4		No	No						#	0		0	0	95	None		0	No						NULL
+	127.0.0.1	root	SLAVE_PORT	1		4	slave-relay-bin.000001	4		No	No						#	0		0	0	96	None		0	No						NULL
 start slave;
 insert into t1 values (1);
 select * from t1;

mysql-test/r/rpl_temporary.result

@@ -38,19 +38,19 @@ f
 7
 show binlog events;
 Log_name	Pos	Event_type	Server_id	End_log_pos	Info
-master-bin.000001	4	Format_desc	1	95	Server ver: VERSION, Binlog ver: 4
+master-bin.000001	4	Format_desc	1	96	Server ver: VERSION, Binlog ver: 4
-master-bin.000001	95	Query	1	185	use `test`; drop table if exists t1,t2
+master-bin.000001	96	Query	1	186	use `test`; drop table if exists t1,t2
-master-bin.000001	185	Query	1	271	use `test`; create table t1(f int)
+master-bin.000001	186	Query	1	272	use `test`; create table t1(f int)
-master-bin.000001	271	Query	1	357	use `test`; create table t2(f int)
+master-bin.000001	272	Query	1	358	use `test`; create table t2(f int)
-master-bin.000001	357	Query	1	483	use `test`; insert into t1 values (1),(2),(3),(4),(5),(6),(7),(8),(9),(10)
+master-bin.000001	358	Query	1	484	use `test`; insert into t1 values (1),(2),(3),(4),(5),(6),(7),(8),(9),(10)
-master-bin.000001	483	Query	1	579	use `test`; create temporary table t3(f int)
+master-bin.000001	484	Query	1	580	use `test`; create temporary table t3(f int)
-master-bin.000001	579	Query	1	684	use `test`; insert into t3 select * from t1 where f<6
+master-bin.000001	580	Query	1	685	use `test`; insert into t3 select * from t1 where f<6
-master-bin.000001	684	Query	1	780	use `test`; create temporary table t3(f int)
+master-bin.000001	685	Query	1	781	use `test`; create temporary table t3(f int)
-master-bin.000001	780	Query	1	882	use `test`; insert into t2 select count(*) from t3
+master-bin.000001	781	Query	1	883	use `test`; insert into t2 select count(*) from t3
-master-bin.000001	882	Query	1	988	use `test`; insert into t3 select * from t1 where f>=4
+master-bin.000001	883	Query	1	989	use `test`; insert into t3 select * from t1 where f>=4
-master-bin.000001	988	Query	1	1075	use `test`; drop temporary table t3
+master-bin.000001	989	Query	1	1076	use `test`; drop temporary table t3
-master-bin.000001	1075	Query	1	1177	use `test`; insert into t2 select count(*) from t3
+master-bin.000001	1076	Query	1	1178	use `test`; insert into t2 select count(*) from t3
-master-bin.000001	1177	Query	1	1264	use `test`; drop temporary table t3
+master-bin.000001	1178	Query	1	1265	use `test`; drop temporary table t3
 drop table t1, t2;
 use test;
 SET TIMESTAMP=1040323920;

mysql-test/r/rpl_timezone.result

@@ -32,13 +32,13 @@ t
 2004-06-11 09:39:02
 show binlog events;
 Log_name	Pos	Event_type	Server_id	End_log_pos	Info
-master-bin.000001	4	Format_desc	1	95	Server ver: VERSION, Binlog ver: 4
+master-bin.000001	4	Format_desc	1	96	Server ver: VERSION, Binlog ver: 4
-master-bin.000001	95	Query	1	188	use `test`; create table t1 (t timestamp)
+master-bin.000001	96	Query	1	189	use `test`; create table t1 (t timestamp)
-master-bin.000001	188	Query	1	280	use `test`; create table t2 (t char(32))
+master-bin.000001	189	Query	1	281	use `test`; create table t2 (t char(32))
-master-bin.000001	280	Query	1	372	use `test`; SET ONE_SHOT TIME_ZONE='UTC'
+master-bin.000001	281	Query	1	373	use `test`; SET ONE_SHOT TIME_ZONE='UTC'
-master-bin.000001	372	Query	1	496	use `test`; insert into t1 values ('20040101000000'), ('20040611093902')
+master-bin.000001	373	Query	1	497	use `test`; insert into t1 values ('20040101000000'), ('20040611093902')
-master-bin.000001	496	Query	1	574	use `test`; delete from t1
+master-bin.000001	497	Query	1	575	use `test`; delete from t1
-master-bin.000001	574	Query	1	698	use `test`; insert into t1 values ('20040101000000'), ('20040611093902')
+master-bin.000001	575	Query	1	699	use `test`; insert into t1 values ('20040101000000'), ('20040611093902')
 set time_zone='MET';
 insert into t2 (select t from t1);
 select * from t1;

mysql-test/r/rpl_until.result

@@ -14,14 +14,14 @@ insert into t2 values (3),(4);
 drop table t2;
 show binlog events;
 Log_name	Pos	Event_type	Server_id	End_log_pos	Info
-master-bin.000001	4	Format_desc	1	95	Server ver: VERSION, Binlog ver: 4
+master-bin.000001	4	Format_desc	1	96	Server ver: VERSION, Binlog ver: 4
-master-bin.000001	95	Query	1	217	use `test`; create table t1(n int not null auto_increment primary key)
+master-bin.000001	96	Query	1	218	use `test`; create table t1(n int not null auto_increment primary key)
-master-bin.000001	217	Query	1	318	use `test`; insert into t1 values (1),(2),(3),(4)
+master-bin.000001	218	Query	1	319	use `test`; insert into t1 values (1),(2),(3),(4)
-master-bin.000001	318	Query	1	395	use `test`; drop table t1
+master-bin.000001	319	Query	1	396	use `test`; drop table t1
-master-bin.000001	395	Query	1	517	use `test`; create table t2(n int not null auto_increment primary key)
+master-bin.000001	396	Query	1	518	use `test`; create table t2(n int not null auto_increment primary key)
-master-bin.000001	517	Query	1	610	use `test`; insert into t2 values (1),(2)
+master-bin.000001	518	Query	1	611	use `test`; insert into t2 values (1),(2)
-master-bin.000001	610	Query	1	703	use `test`; insert into t2 values (3),(4)
+master-bin.000001	611	Query	1	704	use `test`; insert into t2 values (3),(4)
-master-bin.000001	703	Query	1	780	use `test`; drop table t2
+master-bin.000001	704	Query	1	781	use `test`; drop table t2
 start slave until master_log_file='master-bin.000001', master_log_pos=304;
 select * from t1;
 n
@@ -31,7 +31,7 @@ n
 4
 show slave status;
 Slave_IO_State	Master_Host	Master_User	Master_Port	Connect_Retry	Master_Log_File	Read_Master_Log_Pos	Relay_Log_File	Relay_Log_Pos	Relay_Master_Log_File	Slave_IO_Running	Slave_SQL_Running	Replicate_Do_DB	Replicate_Ignore_DB	Replicate_Do_Table	Replicate_Ignore_Table	Replicate_Wild_Do_Table	Replicate_Wild_Ignore_Table	Last_Errno	Last_Error	Skip_Counter	Exec_Master_Log_Pos	Relay_Log_Space	Until_Condition	Until_Log_File	Until_Log_Pos	Master_SSL_Allowed	Master_SSL_CA_File	Master_SSL_CA_Path	Master_SSL_Cert	Master_SSL_Cipher	Master_SSL_Key	Seconds_Behind_Master
-#	127.0.0.1	root	MASTER_MYPORT	1	master-bin.000001	780	slave-relay-bin.000004	#	master-bin.000001	Yes	No							0		0	318	#	Master	master-bin.000001	304	No						#
+#	127.0.0.1	root	MASTER_MYPORT	1	master-bin.000001	781	slave-relay-bin.000004	#	master-bin.000001	Yes	No							0		0	319	#	Master	master-bin.000001	304	No						#
 start slave until master_log_file='master-no-such-bin.000001', master_log_pos=291;
 select * from t1;
 n
@@ -41,7 +41,7 @@ n
 4
 show slave status;
 Slave_IO_State	Master_Host	Master_User	Master_Port	Connect_Retry	Master_Log_File	Read_Master_Log_Pos	Relay_Log_File	Relay_Log_Pos	Relay_Master_Log_File	Slave_IO_Running	Slave_SQL_Running	Replicate_Do_DB	Replicate_Ignore_DB	Replicate_Do_Table	Replicate_Ignore_Table	Replicate_Wild_Do_Table	Replicate_Wild_Ignore_Table	Last_Errno	Last_Error	Skip_Counter	Exec_Master_Log_Pos	Relay_Log_Space	Until_Condition	Until_Log_File	Until_Log_Pos	Master_SSL_Allowed	Master_SSL_CA_File	Master_SSL_CA_Path	Master_SSL_Cert	Master_SSL_Cipher	Master_SSL_Key	Seconds_Behind_Master
-#	127.0.0.1	root	MASTER_MYPORT	1	master-bin.000001	780	slave-relay-bin.000004	#	master-bin.000001	Yes	No							0		0	318	#	Master	master-no-such-bin.000001	291	No						#
+#	127.0.0.1	root	MASTER_MYPORT	1	master-bin.000001	781	slave-relay-bin.000004	#	master-bin.000001	Yes	No							0		0	319	#	Master	master-no-such-bin.000001	291	No						#
 start slave until relay_log_file='slave-relay-bin.000004', relay_log_pos=710;
 select * from t2;
 n
@@ -49,13 +49,13 @@ n
 2
 show slave status;
 Slave_IO_State	Master_Host	Master_User	Master_Port	Connect_Retry	Master_Log_File	Read_Master_Log_Pos	Relay_Log_File	Relay_Log_Pos	Relay_Master_Log_File	Slave_IO_Running	Slave_SQL_Running	Replicate_Do_DB	Replicate_Ignore_DB	Replicate_Do_Table	Replicate_Ignore_Table	Replicate_Wild_Do_Table	Replicate_Wild_Ignore_Table	Last_Errno	Last_Error	Skip_Counter	Exec_Master_Log_Pos	Relay_Log_Space	Until_Condition	Until_Log_File	Until_Log_Pos	Master_SSL_Allowed	Master_SSL_CA_File	Master_SSL_CA_Path	Master_SSL_Cert	Master_SSL_Cipher	Master_SSL_Key	Seconds_Behind_Master
-#	127.0.0.1	root	MASTER_MYPORT	1	master-bin.000001	780	slave-relay-bin.000004	#	master-bin.000001	Yes	No							0		0	610	#	Relay	slave-relay-bin.000004	710	No						#
+#	127.0.0.1	root	MASTER_MYPORT	1	master-bin.000001	781	slave-relay-bin.000004	#	master-bin.000001	Yes	No							0		0	611	#	Relay	slave-relay-bin.000004	710	No						#
 start slave;
 stop slave;
 start slave until master_log_file='master-bin.000001', master_log_pos=710;
 show slave status;
 Slave_IO_State	Master_Host	Master_User	Master_Port	Connect_Retry	Master_Log_File	Read_Master_Log_Pos	Relay_Log_File	Relay_Log_Pos	Relay_Master_Log_File	Slave_IO_Running	Slave_SQL_Running	Replicate_Do_DB	Replicate_Ignore_DB	Replicate_Do_Table	Replicate_Ignore_Table	Replicate_Wild_Do_Table	Replicate_Wild_Ignore_Table	Last_Errno	Last_Error	Skip_Counter	Exec_Master_Log_Pos	Relay_Log_Space	Until_Condition	Until_Log_File	Until_Log_Pos	Master_SSL_Allowed	Master_SSL_CA_File	Master_SSL_CA_Path	Master_SSL_Cert	Master_SSL_Cipher	Master_SSL_Key	Seconds_Behind_Master
-#	127.0.0.1	root	MASTER_MYPORT	1	master-bin.000001	780	slave-relay-bin.000004	#	master-bin.000001	Yes	No							0		0	780	#	Master	master-bin.000001	710	No						#
+#	127.0.0.1	root	MASTER_MYPORT	1	master-bin.000001	781	slave-relay-bin.000004	#	master-bin.000001	Yes	No							0		0	781	#	Master	master-bin.000001	710	No						#
 start slave until master_log_file='master-bin', master_log_pos=561;
 ERROR HY000: Incorrect parameter or combination of parameters for START SLAVE UNTIL
 start slave until master_log_file='master-bin.000001', master_log_pos=561, relay_log_pos=12;

mysql-test/r/rpl_user_variables.result

@@ -76,35 +76,35 @@ abcn1n2
 NULL
 NULL
 NULL
-show binlog events from 95;
+show binlog events from 96;
 Log_name	Pos	Event_type	Server_id	End_log_pos	Info
-slave-bin.000001	95	Query	1	186	use `test`; create table t1(n char(30))
+slave-bin.000001	96	Query	1	187	use `test`; create table t1(n char(30))
-slave-bin.000001	186	User var	2	229	@`i1`=12345678901234
+slave-bin.000001	187	User var	2	230	@`i1`=12345678901234
-slave-bin.000001	229	User var	2	272	@`i2`=-12345678901234
+slave-bin.000001	230	User var	2	273	@`i2`=-12345678901234
-slave-bin.000001	272	User var	2	315	@`i3`=0
+slave-bin.000001	273	User var	2	316	@`i3`=0
-slave-bin.000001	315	User var	2	358	@`i4`=-1
+slave-bin.000001	316	User var	2	359	@`i4`=-1
-slave-bin.000001	358	Query	1	470	use `test`; insert into t1 values (@i1), (@i2), (@i3), (@i4)
+slave-bin.000001	359	Query	1	471	use `test`; insert into t1 values (@i1), (@i2), (@i3), (@i4)
-slave-bin.000001	470	User var	2	509	@`r1`=12.5
+slave-bin.000001	471	User var	2	510	@`r1`=12.5
-slave-bin.000001	509	User var	2	548	@`r2`=-12.5
+slave-bin.000001	510	User var	2	549	@`r2`=-12.5
-slave-bin.000001	548	Query	1	646	use `test`; insert into t1 values (@r1), (@r2)
+slave-bin.000001	549	Query	1	647	use `test`; insert into t1 values (@r1), (@r2)
-slave-bin.000001	646	User var	2	695	@`s1`=_latin1 0x5468697320697320612074657374 COLLATE latin1_swedish_ci
+slave-bin.000001	647	User var	2	696	@`s1`=_latin1 0x5468697320697320612074657374 COLLATE latin1_swedish_ci
-slave-bin.000001	695	User var	2	730	@`s2`=_latin1 "" COLLATE latin1_swedish_ci
+slave-bin.000001	696	User var	2	731	@`s2`=_latin1 "" COLLATE latin1_swedish_ci
-slave-bin.000001	730	User var	2	772	@`s3`=_latin1 0x61626327646566 COLLATE latin1_swedish_ci
+slave-bin.000001	731	User var	2	773	@`s3`=_latin1 0x61626327646566 COLLATE latin1_swedish_ci
-slave-bin.000001	772	User var	2	814	@`s4`=_latin1 0x6162635C646566 COLLATE latin1_swedish_ci
+slave-bin.000001	773	User var	2	815	@`s4`=_latin1 0x6162635C646566 COLLATE latin1_swedish_ci
-slave-bin.000001	814	User var	2	856	@`s5`=_latin1 0x61626327646566 COLLATE latin1_swedish_ci
+slave-bin.000001	815	User var	2	857	@`s5`=_latin1 0x61626327646566 COLLATE latin1_swedish_ci
-slave-bin.000001	856	Query	1	975	use `test`; insert into t1 values (@s1), (@s2), (@s3), (@s4), (@s5)
+slave-bin.000001	857	Query	1	976	use `test`; insert into t1 values (@s1), (@s2), (@s3), (@s4), (@s5)
-slave-bin.000001	975	User var	2	1001	@`n1`=NULL
+slave-bin.000001	976	User var	2	1002	@`n1`=NULL
-slave-bin.000001	1001	Query	1	1092	use `test`; insert into t1 values (@n1)
+slave-bin.000001	1002	Query	1	1093	use `test`; insert into t1 values (@n1)
-slave-bin.000001	1092	User var	2	1118	@`n2`=NULL
+slave-bin.000001	1093	User var	2	1119	@`n2`=NULL
-slave-bin.000001	1118	Query	1	1209	use `test`; insert into t1 values (@n2)
+slave-bin.000001	1119	Query	1	1210	use `test`; insert into t1 values (@n2)
-slave-bin.000001	1209	Query	1	1326	use `test`; insert into t1 values (@a:=0), (@a:=@a+1), (@a:=@a+1)
+slave-bin.000001	1210	Query	1	1327	use `test`; insert into t1 values (@a:=0), (@a:=@a+1), (@a:=@a+1)
-slave-bin.000001	1326	User var	2	1368	@`a`=2
+slave-bin.000001	1327	User var	2	1369	@`a`=2
-slave-bin.000001	1368	Query	1	1469	use `test`; insert into t1 values (@a+(@b:=@a+1))
+slave-bin.000001	1369	Query	1	1470	use `test`; insert into t1 values (@a+(@b:=@a+1))
-slave-bin.000001	1469	User var	2	1506	@`q`=_latin1 0x616263 COLLATE latin1_swedish_ci
+slave-bin.000001	1470	User var	2	1507	@`q`=_latin1 0x616263 COLLATE latin1_swedish_ci
-slave-bin.000001	1506	Query	1	1639	use `test`; insert t1 values (@q), (@q:=concat(@q, 'n1')), (@q:=concat(@q, 'n2'))
+slave-bin.000001	1507	Query	1	1640	use `test`; insert t1 values (@q), (@q:=concat(@q, 'n1')), (@q:=concat(@q, 'n2'))
-slave-bin.000001	1639	User var	2	1681	@`a`=5
+slave-bin.000001	1640	User var	2	1682	@`a`=5
-slave-bin.000001	1681	Query	1	1776	use `test`; insert into t1 values (@a),(@a)
+slave-bin.000001	1682	Query	1	1777	use `test`; insert into t1 values (@a),(@a)
-slave-bin.000001	1776	User var	2	1801	@`a`=NULL
+slave-bin.000001	1777	User var	2	1802	@`a`=NULL
-slave-bin.000001	1801	Query	1	1903	use `test`; insert into t1 values (@a),(@a),(@a*5)
+slave-bin.000001	1802	Query	1	1904	use `test`; insert into t1 values (@a),(@a),(@a*5)
 drop table t1;
 stop slave;

mysql-test/r/user_var.result

@@ -174,14 +174,15 @@ INSERT INTO t1 VALUES(@`a b`);
 set @var1= "';aaa";
 SET @var2=char(ascii('a'));
 insert into t1 values (@var1),(@var2);
-show binlog events from 95;
+show binlog events from 96;
 Log_name	Pos	Event_type	Server_id	End_log_pos	Info
-master-bin.000001	95	User var	1	136	@`a b`=_latin1 0x68656C6C6F COLLATE latin1_swedish_ci
+master-bin.000001	96	User var	1	137	@`a b`=_latin1 0x68656C6C6F COLLATE latin1_swedish_ci
-master-bin.000001	136	Query	1	229	use `test`; INSERT INTO t1 VALUES(@`a b`)
+master-bin.000001	137	Query	1	230	use `test`; INSERT INTO t1 VALUES(@`a b`)
-master-bin.000001	229	User var	1	271	@`var1`=_latin1 0x273B616161 COLLATE latin1_swedish_ci
+master-bin.000001	230	User var	1	272	@`var1`=_latin1 0x273B616161 COLLATE latin1_swedish_ci
-master-bin.000001	271	User var	1	309	@`var2`=_latin1 0x61 COLLATE latin1_swedish_ci
+master-bin.000001	272	User var	1	310	@`var2`=_latin1 0x61 COLLATE latin1_swedish_ci
-master-bin.000001	309	Query	1	410	use `test`; insert into t1 values (@var1),(@var2)
+master-bin.000001	310	Query	1	411	use `test`; insert into t1 values (@var1),(@var2)
 /*!40019 SET @@session.max_insert_delayed_threads=0*/;
+ROLLBACK;
 SET @`a b`:=_latin1 0x68656C6C6F COLLATE `latin1_swedish_ci`;
 use test;
 SET TIMESTAMP=10000;

mysql-test/t/ctype_ucs.test

@@ -338,7 +338,7 @@ create table t2 (c char(30)) charset=ucs2;
 set @v=convert('abc' using ucs2);
 reset master;
 insert into t2 values (@v);
-show binlog events from 95;
+show binlog events from 96;
 # more important than SHOW BINLOG EVENTS, mysqlbinlog (where we
 # absolutely need variables names to be quoted and strings to be
 # escaped).

mysql-test/t/innodb.test

@@ -139,7 +139,7 @@ select n from t1;
 rollback to savepoint `savept2`;
 release savepoint `my_savepoint`;
 select n from t1;
--- error 1181
+-- error 1305
 rollback to savepoint `my_savepoint`;
 set autocommit=1;
 # nop

mysql-test/t/mix_innodb_myisam_binlog.test

@@ -26,7 +26,8 @@ insert into t2 select * from t1;
 commit;
 
 --replace_column 5 #
-show binlog events from 95;
+--replace_result "xid=12" "xid=7"
+show binlog events from 96;
 
 delete from t1;
 delete from t2;
@@ -39,7 +40,7 @@ insert into t2 select * from t1;
 rollback;
 
 --replace_column 5 #
-show binlog events from 95;
+show binlog events from 96;
 
 delete from t1;
 delete from t2;
@@ -54,7 +55,8 @@ rollback to savepoint my_savepoint;
 commit;
 
 --replace_column 5 #
-show binlog events from 95;
+--replace_result "xid=45" "xid=24"
+show binlog events from 96;
 
 delete from t1;
 delete from t2;
@@ -71,7 +73,8 @@ commit;
 select a from t1 order by a; # check that savepoints work :)
 
 --replace_column 5 #
-show binlog events from 95;
+--replace_result "xid=67" "xid=36"
+show binlog events from 96;
 
 # and when ROLLBACK is not explicit?
 delete from t1;
@@ -92,7 +95,7 @@ connection con2;
 # logging has been done, we use a user lock.
 select get_lock("a",10);
 --replace_column 5 #
-show binlog events from 95;
+show binlog events from 96;
 
 # and when not in a transact1on?
 delete from t1;
@@ -103,7 +106,8 @@ insert into t1 values(9);
 insert into t2 select * from t1;
 
 --replace_column 5 #
-show binlog events from 95;
+--replace_result "xid=116" "xid=59"
+show binlog events from 96;
 
 # Check that when the query updat1ng the MyISAM table is the first in the
 # transaction, we log it immediately.
@@ -115,12 +119,14 @@ insert into t1 values(10); # first make t1 non-empty
 begin;
 insert into t2 select * from t1;
 --replace_column 5 #
-show binlog events from 95;
+--replace_result "xid=130" "xid=65"
+show binlog events from 96;
 insert into t1 values(11);
 commit;
 
 --replace_column 5 #
-show binlog events from 95;
+--replace_result "xid=130" "xid=65" "xid=133" "xid=67"
+show binlog events from 96;
 
 
 # Check that things work like before this BEGIN/ROLLBACK code was added,
@@ -138,7 +144,8 @@ insert into t2 select * from t1;
 commit;
 
 --replace_column 5 #
-show binlog events from 95;
+--replace_result "xid=152" "xid=77"
+show binlog events from 96;
 
 delete from t1;
 delete from t2;
@@ -150,7 +157,7 @@ insert into t2 select * from t1;
 rollback;
 
 --replace_column 5 #
-show binlog events from 95;
+show binlog events from 96;
 
 delete from t1;
 delete from t2;
@@ -165,7 +172,8 @@ rollback to savepoint my_savepoint;
 commit;
 
 --replace_column 5 #
-show binlog events from 95;
+--replace_result "xid=184" "xid=93"
+show binlog events from 96;
 
 delete from t1;
 delete from t2;
@@ -182,7 +190,8 @@ commit;
 select a from t1 order by a; # check that savepoints work :)
 
 --replace_column 5 #
-show binlog events from 95;
+--replace_result "xid=205" "xid=104"
+show binlog events from 96;
 
 # Test for BUG#5714, where a MyISAM update in the transaction used to
 # release row-level locks in InnoDB

mysql-test/t/mysqlbinlog.test

@@ -61,7 +61,7 @@ select "--- --database --" as "";
 select "--- --position --" as "";
 --enable_query_log
 --replace_result $MYSQL_TEST_DIR MYSQL_TEST_DIR
---exec $MYSQL_BINLOG --short-form --local-load=$MYSQL_TEST_DIR/var/tmp/ --position=118 $MYSQL_TEST_DIR/var/log/master-bin.000002
+--exec $MYSQL_BINLOG --short-form --local-load=$MYSQL_TEST_DIR/var/tmp/ --position=119 $MYSQL_TEST_DIR/var/log/master-bin.000002
 
 # These are tests for remote binlog.
 # They should return the same as previous test.
@@ -93,7 +93,7 @@ select "--- --database --" as "";
 select "--- --position --" as "";
 --enable_query_log
 --replace_result $MYSQL_TEST_DIR MYSQL_TEST_DIR
---exec $MYSQL_BINLOG --short-form --local-load=$MYSQL_TEST_DIR/var/tmp/ --read-from-remote-server --position=118 --user=root --host=127.0.0.1 --port=$MASTER_MYPORT master-bin.000002
+--exec $MYSQL_BINLOG --short-form --local-load=$MYSQL_TEST_DIR/var/tmp/ --read-from-remote-server --position=119 --user=root --host=127.0.0.1 --port=$MASTER_MYPORT master-bin.000002
 
 # Bug#7853 (mysqlbinlog does not accept input from stdin)
 --disable_query_log
@@ -107,3 +107,4 @@ select "--- reading stdin --" as "";
 
 # clean up
 drop table t1, t2;
+

mysql-test/t/mysqlbinlog2.test

@@ -46,11 +46,11 @@ select "--- offset --" as "";
 --disable_query_log
 select "--- start-position --" as "";
 --enable_query_log
---exec $MYSQL_BINLOG --short-form --start-position=601 $MYSQL_TEST_DIR/var/log/master-bin.000001 
+--exec $MYSQL_BINLOG --short-form --start-position=602 $MYSQL_TEST_DIR/var/log/master-bin.000001 
 --disable_query_log
 select "--- stop-position --" as "";
 --enable_query_log
---exec $MYSQL_BINLOG --short-form --stop-position=601 $MYSQL_TEST_DIR/var/log/master-bin.000001 
+--exec $MYSQL_BINLOG --short-form --stop-position=602 $MYSQL_TEST_DIR/var/log/master-bin.000001 
 --disable_query_log
 select "--- start-datetime --" as "";
 --enable_query_log
@@ -75,11 +75,11 @@ select "--- offset --" as "";
 --disable_query_log
 select "--- start-position --" as "";
 --enable_query_log
---exec $MYSQL_BINLOG --short-form --start-position=601 $MYSQL_TEST_DIR/var/log/master-bin.000001 $MYSQL_TEST_DIR/var/log/master-bin.000002
+--exec $MYSQL_BINLOG --short-form --start-position=602 $MYSQL_TEST_DIR/var/log/master-bin.000001 $MYSQL_TEST_DIR/var/log/master-bin.000002
 --disable_query_log
 select "--- stop-position --" as "";
 --enable_query_log
---exec $MYSQL_BINLOG --short-form --stop-position=123 $MYSQL_TEST_DIR/var/log/master-bin.000001 $MYSQL_TEST_DIR/var/log/master-bin.000002
+--exec $MYSQL_BINLOG --short-form --stop-position=124 $MYSQL_TEST_DIR/var/log/master-bin.000001 $MYSQL_TEST_DIR/var/log/master-bin.000002
 --disable_query_log
 select "--- start-datetime --" as "";
 --enable_query_log
@@ -102,11 +102,11 @@ select "--- offset --" as "";
 --disable_query_log
 select "--- start-position --" as "";
 --enable_query_log
---exec $MYSQL_BINLOG --short-form --start-position=601 --read-from-remote-server --user=root --host=127.0.0.1 --port=$MASTER_MYPORT master-bin.000001 
+--exec $MYSQL_BINLOG --short-form --start-position=602 --read-from-remote-server --user=root --host=127.0.0.1 --port=$MASTER_MYPORT master-bin.000001 
 --disable_query_log
 select "--- stop-position --" as "";
 --enable_query_log
---exec $MYSQL_BINLOG --short-form --stop-position=601 --read-from-remote-server --user=root --host=127.0.0.1 --port=$MASTER_MYPORT master-bin.000001 
+--exec $MYSQL_BINLOG --short-form --stop-position=602 --read-from-remote-server --user=root --host=127.0.0.1 --port=$MASTER_MYPORT master-bin.000001 
 --disable_query_log
 select "--- start-datetime --" as "";
 --enable_query_log
@@ -129,11 +129,11 @@ select "--- offset --" as "";
 --disable_query_log
 select "--- start-position --" as "";
 --enable_query_log
---exec $MYSQL_BINLOG --short-form --start-position=601 --read-from-remote-server --user=root --host=127.0.0.1 --port=$MASTER_MYPORT master-bin.000001  master-bin.000002
+--exec $MYSQL_BINLOG --short-form --start-position=602 --read-from-remote-server --user=root --host=127.0.0.1 --port=$MASTER_MYPORT master-bin.000001  master-bin.000002
 --disable_query_log
 select "--- stop-position --" as "";
 --enable_query_log
---exec $MYSQL_BINLOG --short-form --stop-position=123 --read-from-remote-server --user=root --host=127.0.0.1 --port=$MASTER_MYPORT master-bin.000001  master-bin.000002
+--exec $MYSQL_BINLOG --short-form --stop-position=124 --read-from-remote-server --user=root --host=127.0.0.1 --port=$MASTER_MYPORT master-bin.000001  master-bin.000002
 --disable_query_log
 select "--- start-datetime --" as "";
 --enable_query_log

mysql-test/t/rpl_charset.test

@@ -107,7 +107,7 @@ connection master;
 drop database mysqltest2;
 drop database mysqltest3;
 --replace_column 2 # 5 #
-show binlog events from 95;
+show binlog events from 96;
 sync_slave_with_master;
 
 # Check that we can change global.collation_server (since 5.0.3)
@@ -156,3 +156,4 @@ connection master;
 --exec $MYSQL_BINLOG --short-form $MYSQL_TEST_DIR/var/log/master-bin.000001 
 drop table t1;
 sync_slave_with_master;
+

mysql-test/t/rpl_error_ignored_table.test

@@ -48,7 +48,7 @@ connection master;
 --error 0,1053;
 reap;
 connection master1;
-show binlog events from 95;
+show binlog events from 96;
 save_master_pos;
 connection slave;
 # SQL slave thread should not have stopped (because table of the killed

mysql-test/t/rpl_loaddata_rule_m.test

@@ -19,5 +19,5 @@ create database mysqltest;
 create table t1(a int, b int, unique(b));
 use mysqltest;
 load data infile '../../std_data/rpl_loaddata.dat' into table test.t1;
-show binlog events from 95; # should be nothing
+show binlog events from 96; # should be nothing
 drop database mysqltest;

mysql-test/t/rpl_loaddata_rule_s.test

@@ -17,4 +17,4 @@ save_master_pos;
 connection slave;
 sync_with_master;
 select count(*) from t1; # check that LOAD was replicated
-show binlog events from 95; # should be nothing
+show binlog events from 96; # should be nothing

mysql-test/t/rpl_log.test

@@ -38,9 +38,9 @@ select count(*) from t1;
 drop table t1;
 --replace_result $VERSION VERSION
 show binlog events;
-show binlog events from 95 limit 1;
+show binlog events from 96 limit 1;
-show binlog events from 95 limit 2;
+show binlog events from 96 limit 2;
-show binlog events from 95 limit 2,1;
+show binlog events from 96 limit 2,1;
 flush logs;
 
 # We need an extra update before doing save_master_pos.

mysql-test/t/rpl_log_pos.test

@@ -38,7 +38,7 @@ insert into t1 values (1),(2),(3);
 save_master_pos;
 connection slave;
 stop slave;
-change master to master_log_pos=95;
+change master to master_log_pos=96;
 start slave;
 sync_with_master;
 select * from t1;

mysql-test/t/rpl_user_variables.test

@@ -46,7 +46,7 @@ save_master_pos;
 connection slave;
 sync_with_master;
 select * from t1;
-show binlog events from 95;
+show binlog events from 96;
 connection master;
 drop table t1;
 save_master_pos;

mysql-test/t/strict.test

@@ -866,12 +866,12 @@ DROP TABLE t1;
 
 CREATE TABLE t1 (col1 CHAR(5), col2 VARCHAR(6));
 INSERT INTO t1 VALUES ('hello', 'hello'),('he', 'he'),('hello   ', 'hello ');
---error 1400
+--error 1406
 INSERT INTO t1 (col1) VALUES ('hellobob');
 --error 1265
 INSERT INTO t1 (col2) VALUES ('hellobob');
 INSERT INTO t1 (col2) VALUES ('hello  ');
---error 1400
+--error 1406
 UPDATE t1 SET col1 ='hellobob' WHERE col1 ='he';
 --error 1265
 UPDATE t1 SET col2 ='hellobob' WHERE col2 ='he';

mysql-test/t/user_var.test

@@ -110,7 +110,7 @@ INSERT INTO t1 VALUES(@`a b`);
 set @var1= "';aaa";
 SET @var2=char(ascii('a'));
 insert into t1 values (@var1),(@var2);
-show binlog events from 95;
+show binlog events from 96;
 # more important than SHOW BINLOG EVENTS, mysqlbinlog (where we
 # absolutely need variables names to be quoted and strings to be
 # escaped).

mysys/hash.c

@@ -103,7 +103,7 @@ static inline void hash_free_elements(HASH *hash)
     hash_free()
     hash   the hash to delete elements of
 
-  NOTES: Hash can't be reused wuthing calling hash_init again.
+  NOTES: Hash can't be reused without calling hash_init again.
 */
 
 void hash_free(HASH *hash)

sql/examples/ha_archive.cc

@@ -135,6 +135,23 @@ static HASH archive_open_tables;
 #define DATA_BUFFER_SIZE 2       // Size of the data used in the data file
 #define ARCHIVE_CHECK_HEADER 254 // The number we use to determine corruption
 
+/* dummy handlerton - only to have something to return from archive_db_init */
+static handlerton archive_hton = {
+  0,       /* slot */
+  0,       /* savepoint size. */
+  0,       /* close_connection */
+  0,       /* savepoint */
+  0,       /* rollback to savepoint */
+  0,       /* releas savepoint */
+  0,       /* commit */
+  0,       /* rollback */
+  0,       /* prepare */
+  0,       /* recover */
+  0,       /* commit_by_xid */
+  0        /* rollback_by_xid */
+};
+
+
 /*
   Used for hash table that tracks open tables.
 */
@@ -154,19 +171,20 @@ static byte* archive_get_key(ARCHIVE_SHARE *share,uint *length,
     void
 
   RETURN
-    FALSE       OK
+    &archive_hton OK
-    TRUE        Error
+    0             Error
 */
 
-bool archive_db_init()
+handlerton *archive_db_init()
 {
   archive_inited= 1;
   VOID(pthread_mutex_init(&archive_mutex, MY_MUTEX_INIT_FAST));
-  return (hash_init(&archive_open_tables, system_charset_info, 32, 0, 0,
+  if (hash_init(&archive_open_tables, system_charset_info, 32, 0, 0,
-                    (hash_get_key) archive_get_key, 0, 0));
+                (hash_get_key) archive_get_key, 0, 0))
+    return 0;
+  return &archive_hton;
 }
 
-
 /*
   Release the archive handler.
 
@@ -734,7 +752,7 @@ int ha_archive::rnd_next(byte *buf)
 }
 
 
-/* 
+/*
   Thanks to the table flag HA_REC_NOT_IN_SEQ this will be called after
   each call to ha_archive::rnd_next() if an ordering of the rows is
   needed.
@@ -743,7 +761,7 @@ int ha_archive::rnd_next(byte *buf)
 void ha_archive::position(const byte *record)
 {
   DBUG_ENTER("ha_archive::position");
-  ha_store_ptr(ref, ref_length, current_position);
+  my_store_ptr(ref, ref_length, current_position);
   DBUG_VOID_RETURN;
 }
 
@@ -760,7 +778,7 @@ int ha_archive::rnd_pos(byte * buf, byte *pos)
   DBUG_ENTER("ha_archive::rnd_pos");
   statistic_increment(table->in_use->status_var.ha_read_rnd_next_count,
 		      &LOCK_status);
-  current_position= ha_get_ptr(pos, ref_length);
+  current_position= my_get_ptr(pos, ref_length);
   z_off_t seek= gzseek(archive, current_position, SEEK_SET);
 
   DBUG_RETURN(get_row(archive, buf));

sql/examples/ha_archive.h

@@ -106,6 +106,6 @@ public:
                              enum thr_lock_type lock_type);
 };
 
-bool archive_db_init(void);
+handlerton *archive_db_init(void);
 bool archive_db_end(void);
 

sql/ha_berkeley.cc

@@ -102,11 +102,34 @@ static int write_status(DB *status_block, char *buff, uint length);
 static void update_status(BDB_SHARE *share, TABLE *table);
 static void berkeley_noticecall(DB_ENV *db_env, db_notices notice);
 
+static int berkeley_close_connection(THD *thd);
+static int berkeley_commit(THD *thd, bool all);
+static int berkeley_rollback(THD *thd, bool all);
 
+static handlerton berkeley_hton = {
+  0, /* slot */
+  0, /* savepoint size */
+  berkeley_close_connection,
+  NULL, /* savepoint_set */
+  NULL, /* savepoint_rollback */
+  NULL, /* savepoint_release */
+  berkeley_commit,
+  berkeley_rollback,
+  NULL, /* prepare */
+  NULL, /* recover */
+  NULL, /* commit_by_xid */
+  NULL /* rollback_by_xid */
+};
+
+typedef struct st_berkeley_trx_data {
+  DB_TXN *all;
+  DB_TXN *stmt;
+  uint bdb_lock_count;
+} berkeley_trx_data;
 
 /* General functions */
 
-bool berkeley_init(void)
+handlerton *berkeley_init(void)
 {
   DBUG_ENTER("berkeley_init");
 
@@ -135,7 +158,7 @@ bool berkeley_init(void)
   berkeley_log_file_size= max(berkeley_log_file_size, 10*1024*1024L);
 
   if (db_env_create(&db_env,0))
-    DBUG_RETURN(1); /* purecov: inspected */
+    DBUG_RETURN(0);
   db_env->set_errcall(db_env,berkeley_print_error);
   db_env->set_errpfx(db_env,"bdb");
   db_env->set_noticecall(db_env, berkeley_noticecall);
@@ -163,16 +186,15 @@ bool berkeley_init(void)
 		   DB_INIT_LOG | DB_INIT_MPOOL | DB_INIT_TXN |
 		   DB_CREATE | DB_THREAD, 0666))
   {
-    db_env->close(db_env,0); /* purecov: inspected */
+    db_env->close(db_env,0);
-    db_env=0; /* purecov: inspected */
+    db_env=0;
-    goto err;
+    DBUG_RETURN(0);
   }
 
   (void) hash_init(&bdb_open_tables,system_charset_info,32,0,0,
 		   (hash_get_key) bdb_get_key,0,0);
   pthread_mutex_init(&bdb_mutex,MY_MUTEX_INIT_FAST);
-err:
+  DBUG_RETURN(&berkeley_hton);
-  DBUG_RETURN(db_env == 0);
 }
 
 
@@ -190,6 +212,11 @@ bool berkeley_end(void)
   DBUG_RETURN(error != 0);
 }
 
+static int berkeley_close_connection(THD *thd)
+{
+  my_free((gptr)thd->ha_data[berkeley_hton.slot], MYF(0));
+}
+
 bool berkeley_flush_logs()
 {
   int error;
@@ -208,26 +235,29 @@ bool berkeley_flush_logs()
   DBUG_RETURN(result);
 }
 
-
+static int berkeley_commit(THD *thd, bool all)
-int berkeley_commit(THD *thd, void *trans)
 {
   DBUG_ENTER("berkeley_commit");
-  DBUG_PRINT("trans",("ending transaction %s",
+  DBUG_PRINT("trans",("ending transaction %s", all ? "all" : "stmt"));
-		      trans == thd->transaction.stmt.bdb_tid ? "stmt" : "all"));
+  berkeley_trx_data *trx=(berkeley_trx_data *)thd->ha_data[berkeley_hton.slot];
-  int error=txn_commit((DB_TXN*) trans,0);
+  DB_TXN **txn= all ? &trx->all : &trx->stmt;
+  int error=txn_commit(*txn,0);
+  *txn=0;
 #ifndef DBUG_OFF
   if (error)
-    DBUG_PRINT("error",("error: %d",error)); /* purecov: inspected */
+    DBUG_PRINT("error",("error: %d",error));
 #endif
   DBUG_RETURN(error);
 }
 
-int berkeley_rollback(THD *thd, void *trans)
+static int berkeley_rollback(THD *thd, bool all)
 {
   DBUG_ENTER("berkeley_rollback");
-  DBUG_PRINT("trans",("aborting transaction %s",
+  DBUG_PRINT("trans",("aborting transaction %s", all ? "all" : "stmt"));
-		      trans == thd->transaction.stmt.bdb_tid ? "stmt" : "all"));
+  berkeley_trx_data *trx=(berkeley_trx_data *)thd->ha_data[berkeley_hton.slot];
-  int error=txn_abort((DB_TXN*) trans);
+  DB_TXN **txn= all ? &trx->all : &trx->stmt;
+  int error=txn_abort(*txn);
+  *txn=0;
   DBUG_RETURN(error);
 }
 
@@ -1842,62 +1872,65 @@ int ha_berkeley::reset(void)
 int ha_berkeley::external_lock(THD *thd, int lock_type)
 {
   int error=0;
+  berkeley_trx_data *trx=(berkeley_trx_data *)thd->ha_data[berkeley_hton.slot];
   DBUG_ENTER("ha_berkeley::external_lock");
+  if (!trx)
+  {
+    thd->ha_data[berkeley_hton.slot]= trx= (berkeley_trx_data *)
+      my_malloc(sizeof(*trx), MYF(MY_ZEROFILL));
+    if (!trx)
+      DBUG_RETURN(1);
+  }
   if (lock_type != F_UNLCK)
   {
-    if (!thd->transaction.bdb_lock_count++)
+    if (!trx->bdb_lock_count++)
     {
-      DBUG_ASSERT(thd->transaction.stmt.bdb_tid == 0);
+      DBUG_ASSERT(trx->stmt == 0);
       transaction=0;				// Safety
       /* First table lock, start transaction */
       if ((thd->options & (OPTION_NOT_AUTOCOMMIT | OPTION_BEGIN |
-			   OPTION_TABLE_LOCK)) &&
+                           OPTION_TABLE_LOCK)) && !trx->all)
-	  !thd->transaction.all.bdb_tid)
       {
 	/* We have to start a master transaction */
 	DBUG_PRINT("trans",("starting transaction all:  options: 0x%lx",
                             (ulong) thd->options));
-	if ((error=txn_begin(db_env, 0,
+        if ((error=txn_begin(db_env, 0, &trx->all, 0)))
-			     (DB_TXN**) &thd->transaction.all.bdb_tid,
-			     0)))
 	{
-	  thd->transaction.bdb_lock_count--;	// We didn't get the lock /* purecov: inspected */
+          trx->bdb_lock_count--;        // We didn't get the lock
-	  DBUG_RETURN(error); /* purecov: inspected */
+          DBUG_RETURN(error);
 	}
+        trans_register_ha(thd, TRUE, &berkeley_hton);
 	if (thd->in_lock_tables)
 	  DBUG_RETURN(0);			// Don't create stmt trans
       }
       DBUG_PRINT("trans",("starting transaction stmt"));
-      if ((error=txn_begin(db_env,
+      if ((error=txn_begin(db_env, trx->all, &trx->stmt, 0)))
-			   (DB_TXN*) thd->transaction.all.bdb_tid,
-			   (DB_TXN**) &thd->transaction.stmt.bdb_tid,
-			   0)))
       {
 	/* We leave the possible master transaction open */
-	thd->transaction.bdb_lock_count--;	// We didn't get the lock /* purecov: inspected */
+        trx->bdb_lock_count--;                  // We didn't get the lock
-	DBUG_RETURN(error); /* purecov: inspected */
+        DBUG_RETURN(error);
       }
+      trans_register_ha(thd, FALSE, &berkeley_hton);
     }
-    transaction= (DB_TXN*) thd->transaction.stmt.bdb_tid;
+    transaction= trx->stmt;
   }
   else
   {
     lock.type=TL_UNLOCK;			// Unlocked
     thread_safe_add(share->rows, changed_rows, &share->mutex);
     changed_rows=0;
-    if (!--thd->transaction.bdb_lock_count)
+    if (!--trx->bdb_lock_count)
     {
-      if (thd->transaction.stmt.bdb_tid)
+      if (trx->stmt)
       {
 	/*
-	   F_UNLOCK is done without a transaction commit / rollback.
+	   F_UNLCK is done without a transaction commit / rollback.
 	   This happens if the thread didn't update any rows
 	   We must in this case commit the work to keep the row locks
 	*/
 	DBUG_PRINT("trans",("commiting non-updating transaction"));
-	error=txn_commit((DB_TXN*) thd->transaction.stmt.bdb_tid,0);
+        error= txn_commit(trx->stmt,0);
-	thd->transaction.stmt.bdb_tid=0;
+        trx->stmt= transaction= 0;
-	transaction=0;
       }
     }
   }
@@ -1915,14 +1948,20 @@ int ha_berkeley::start_stmt(THD *thd)
 {
   int error=0;
   DBUG_ENTER("ha_berkeley::start_stmt");
-  if (!thd->transaction.stmt.bdb_tid)
+  berkeley_trx_data *trx=(berkeley_trx_data *)thd->ha_data[berkeley_hton.slot];
+  DBUG_ASSERT(trx);
+  /*
+    note that trx->stmt may have been already initialized as start_stmt()
+    is called for *each table* not for each storage engine,
+    and there could be many bdb tables referenced in the query
+  */
+  if (!trx->stmt)
   {
     DBUG_PRINT("trans",("starting transaction stmt"));
-    error=txn_begin(db_env, (DB_TXN*) thd->transaction.all.bdb_tid,
+    error=txn_begin(db_env, trx->all, &trx->stmt, 0);
-		    (DB_TXN**) &thd->transaction.stmt.bdb_tid,
+    trans_register_ha(thd, FALSE, &berkeley_hton);
-		    0);
   }
-  transaction= (DB_TXN*) thd->transaction.stmt.bdb_tid;
+  transaction= trx->stmt;
   DBUG_RETURN(error);
 }
 
@@ -2258,6 +2297,8 @@ int ha_berkeley::analyze(THD* thd, HA_CHECK_OPT* check_opt)
   uint i;
   DB_BTREE_STAT *stat=0;
   DB_TXN_STAT *txn_stat_ptr= 0;
+  berkeley_trx_data *trx=(berkeley_trx_data *)thd->ha_data[berkeley_hton.slot];
+  DBUG_ASSERT(trx);
 
   /*
    Original bdb documentation says:
@@ -2272,13 +2313,10 @@ int ha_berkeley::analyze(THD* thd, HA_CHECK_OPT* check_opt)
       txn_stat_ptr && txn_stat_ptr->st_nactive>=2)
   {
     DB_TXN_ACTIVE *atxn_stmt= 0, *atxn_all= 0;
-    
+
-    DB_TXN *txn_all= (DB_TXN*) thd->transaction.all.bdb_tid;
+    u_int32_t all_id= trx->all->id(trx->all);
-    u_int32_t all_id= txn_all->id(txn_all);
+    u_int32_t stmt_id= trx->stmt->id(trx->stmt);
-    
+
-    DB_TXN *txn_stmt= (DB_TXN*) thd->transaction.stmt.bdb_tid;
-    u_int32_t stmt_id= txn_stmt->id(txn_stmt);
-    
     DB_TXN_ACTIVE *cur= txn_stat_ptr->st_txnarray;
     DB_TXN_ACTIVE *end= cur + txn_stat_ptr->st_nactive;
     for (; cur!=end && (!atxn_stmt || !atxn_all); cur++)
@@ -2286,7 +2324,7 @@ int ha_berkeley::analyze(THD* thd, HA_CHECK_OPT* check_opt)
       if (cur->txnid==all_id) atxn_all= cur;
       if (cur->txnid==stmt_id) atxn_stmt= cur;
     }
-    
+
     if (atxn_stmt && atxn_all &&
 	log_compare(&atxn_stmt->lsn,&atxn_all->lsn))
     {

sql/ha_berkeley.h

@@ -168,9 +168,7 @@ extern char *berkeley_home, *berkeley_tmpdir, *berkeley_logdir;
 extern long berkeley_lock_scan_time;
 extern TYPELIB berkeley_lock_typelib;
 
-bool berkeley_init(void);
+handlerton *berkeley_init(void);
 bool berkeley_end(void);
 bool berkeley_flush_logs(void);
-int berkeley_commit(THD *thd, void *trans);
-int berkeley_rollback(THD *thd, void *trans);
 int berkeley_show_logs(Protocol *protocol);

sql/ha_innodb.cc

@@ -14,14 +14,12 @@
    along with this program; if not, write to the Free Software
    Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA */
 
-/* This file defines the InnoDB handler: the interface between MySQL and
+/* This file defines the InnoDB handler: the interface between MySQL and InnoDB
-InnoDB
 NOTE: You can only use noninlined InnoDB functions in this file, because we
 have disables the InnoDB inlining in this file. */
 
-/* TODO list for the InnoDB handler in 4.1:
+/* TODO list for the InnoDB handler in 5.0:
-  - Remove the flag innodb_active_trans from thd and replace it with a
+  - Remove the flag trx->active_trans and look at the InnoDB
-    function call innodb_active_trans(thd), which looks at the InnoDB
     trx struct state field
   - Find out what kind of problems the OS X case-insensitivity causes to
     table and database names; should we 'normalize' the names like we do
@@ -141,8 +139,6 @@ ulong	innobase_active_counter	= 0;
 
 char*	innobase_home 	= NULL;
 
-char    innodb_dummy_stmt_trx_handle = 'D';
-
 static HASH 	innobase_open_tables;
 
 #ifdef __NETWARE__  	/* some special cleanup for NetWare */
@@ -153,6 +149,27 @@ static mysql_byte* innobase_get_key(INNOBASE_SHARE *share,uint *length,
 			      my_bool not_used __attribute__((unused)));
 static INNOBASE_SHARE *get_share(const char *table_name);
 static void free_share(INNOBASE_SHARE *share);
+static int innobase_close_connection(THD* thd);
+static int innobase_commit(THD* thd, bool all);
+static int innobase_rollback(THD* thd, bool all);
+static int innobase_rollback_to_savepoint(THD* thd, void *savepoint);
+static int innobase_savepoint(THD* thd, void *savepoint);
+static int innobase_release_savepoint(THD* thd, void *savepoint);
+
+static handlerton innobase_hton = {
+  0,				/* slot */
+  sizeof(trx_named_savept_t),	/* savepoint size. TODO: use it */
+  innobase_close_connection,
+  innobase_savepoint,
+  innobase_rollback_to_savepoint,
+  innobase_release_savepoint,
+  innobase_commit,		/* commit */
+  innobase_rollback,		/* rollback */
+  innobase_xa_prepare,		/* prepare */
+  innobase_xa_recover,		/* recover */
+  innobase_commit_by_xid,	/* commit_by_xid */
+  innobase_rollback_by_xid	/* rollback_by_xid */
+};
 
 /*********************************************************************
 Commits a transaction in an InnoDB database. */
@@ -250,7 +267,7 @@ struct show_var_st innodb_status_variables[]= {
   {"rows_updated",
   (char*) &export_vars.innodb_rows_updated,               SHOW_LONG},
   {NullS, NullS, SHOW_LONG}};
-      
+
 /* General functions */
 
 /**********************************************************************
@@ -317,9 +334,11 @@ documentation, see handler.cc. */
 void
 innobase_release_temporary_latches(
 /*===============================*/
-        void*   innobase_tid)
+        THD *thd)
 {
-        innobase_release_stat_resources((trx_t*)innobase_tid);
+  trx_t *trx= (trx_t*) thd->ha_data[innobase_hton.slot];
+  if (trx)
+        innobase_release_stat_resources(trx);
 }
 
 /************************************************************************
@@ -652,25 +671,17 @@ check_trx_exists(
 
 	ut_ad(thd == current_thd);
 
-	trx = (trx_t*) thd->transaction.all.innobase_tid;
+        trx = (trx_t*) thd->ha_data[innobase_hton.slot];
 
 	if (trx == NULL) {
 	        DBUG_ASSERT(thd != NULL);
 		trx = trx_allocate_for_mysql();
 
 		trx->mysql_thd = thd;
-		trx->mysql_query_str = &((*thd).query);
+		trx->mysql_query_str = &(thd->query);
-		
+                trx->active_trans = 0;
-		thd->transaction.all.innobase_tid = trx;
 
-		/* The execution of a single SQL statement is denoted by
+                thd->ha_data[innobase_hton.slot] = trx;
-		a 'transaction' handle which is a dummy pointer: InnoDB
-		remembers internally where the latest SQL statement
-		started, and if error handling requires rolling back the
-		latest statement, InnoDB does a rollback to a savepoint. */
-
-		thd->transaction.stmt.innobase_tid =
-		                  (void*)&innodb_dummy_stmt_trx_handle;
 	} else {
 		if (trx->magic_n != TRX_MAGIC_N) {
 			mem_analyze_corruption((byte*)trx);
@@ -707,7 +718,7 @@ ha_innobase::update_thd(
 {
 	row_prebuilt_t*	prebuilt = (row_prebuilt_t*) innobase_prebuilt;
 	trx_t*		trx;
-	
+
 	trx = check_trx_exists(thd);
 
 	if (prebuilt->trx != trx) {
@@ -720,6 +731,24 @@ ha_innobase::update_thd(
 	return(0);
 }
 
+/*************************************************************************
+Registers the InnoDB transaction in MySQL, to receive commit/rollback
+events. This function must be called every time InnoDB starts a
+transaction internally. */
+static
+void
+register_trans(
+/*============*/
+	THD*	thd)	/* in: thd to use the handle */
+{
+        /* register the start of the statement */
+        trans_register_ha(thd, FALSE, &innobase_hton);
+        if (thd->options & (OPTION_NOT_AUTOCOMMIT | OPTION_BEGIN)) {
+
+              /* no autocommit mode, register for a transaction */
+              trans_register_ha(thd, TRUE, &innobase_hton);
+        }
+}
 
 /*   BACKGROUND INFO: HOW THE MYSQL QUERY CACHE WORKS WITH INNODB
      ------------------------------------------------------------
@@ -813,12 +842,7 @@ innobase_query_caching_of_table_permitted(
 		return((my_bool)FALSE);
 	}
 
-	trx = (trx_t*) thd->transaction.all.innobase_tid;
+        trx = check_trx_exists(thd);
-
-	if (trx == NULL) {
-		trx = check_trx_exists(thd);
-	}
-
 	if (trx->has_search_latch) {
 		ut_print_timestamp(stderr);
 		fprintf(stderr,
@@ -871,7 +895,11 @@ innobase_query_caching_of_table_permitted(
 	/* The call of row_search_.. will start a new transaction if it is
 	not yet started */
 
-	thd->transaction.all.innodb_active_trans = 1;
+        if (trx->active_trans == 0) {
+
+                register_trans(thd);
+                trx->active_trans = 1;
+        }
 
 	if (row_search_check_if_query_cache_permitted(trx, norm_name)) {
 
@@ -983,7 +1011,12 @@ ha_innobase::init_table_handle_for_HANDLER(void)
 
 	/* Set the MySQL flag to mark that there is an active transaction */
 
-	current_thd->transaction.all.innodb_active_trans = 1;
+        if (prebuilt->trx->active_trans == 0) {
+
+                register_trans(current_thd);
+
+                prebuilt->trx->active_trans = 1;
+        }
 
         /* We did the necessary inits in this function, no need to repeat them
         in row_search_for_mysql */
@@ -1013,7 +1046,7 @@ ha_innobase::init_table_handle_for_HANDLER(void)
 /*************************************************************************
 Opens an InnoDB database. */
 
-bool
+handlerton *
 innobase_init(void)
 /*===============*/
 			/* out: TRUE if error */
@@ -1090,7 +1123,7 @@ innobase_init(void)
 			"InnoDB: syntax error in innodb_data_file_path");
 	  	my_free(internal_innobase_data_file_path,
 						MYF(MY_ALLOW_ZERO_PTR));
-	  	DBUG_RETURN(TRUE);
+	  	DBUG_RETURN(0);
 	}
 
 	/* -------------- Log files ---------------------------*/
@@ -1122,7 +1155,7 @@ innobase_init(void)
 
 	  	my_free(internal_innobase_data_file_path,
 						MYF(MY_ALLOW_ZERO_PTR));
-		DBUG_RETURN(TRUE);
+		DBUG_RETURN(0);
 	}
 
 	/* --------------------------------------------------*/
@@ -1212,7 +1245,7 @@ innobase_init(void)
 	if (err != DB_SUCCESS) {
 	  	my_free(internal_innobase_data_file_path,
 						MYF(MY_ALLOW_ZERO_PTR));
-		DBUG_RETURN(1);
+		DBUG_RETURN(0);
 	}
 
 	(void) hash_init(&innobase_open_tables,system_charset_info, 32, 0, 0,
@@ -1235,7 +1268,7 @@ innobase_init(void)
 		glob_mi.pos = trx_sys_mysql_master_log_pos;
 	}
 */
-  	DBUG_RETURN(0);
+	DBUG_RETURN(&innobase_hton);
 }
 
 /***********************************************************************
@@ -1359,7 +1392,12 @@ innobase_start_trx_and_assign_read_view(
 
 	/* Set the MySQL flag to mark that there is an active transaction */
 
-	current_thd->transaction.all.innodb_active_trans = 1;
+        if (trx->active_trans == 0) {
+
+                register_trans(current_thd);
+
+                trx->active_trans = 1;
+        }
 
 	DBUG_RETURN(0);
 }
@@ -1368,15 +1406,14 @@ innobase_start_trx_and_assign_read_view(
 Commits a transaction in an InnoDB database or marks an SQL statement
 ended. */
 
-int
+static int
 innobase_commit(
 /*============*/
 			/* out: 0 */
 	THD*	thd,	/* in: MySQL thread handle of the user for whom
 			the transaction should be committed */
-	void*	trx_handle)/* in: InnoDB trx handle or
+        bool    all)    /* in: TRUE - commit transaction
-			&innodb_dummy_stmt_trx_handle: the latter means
+                               FALSE - the current SQL statement ended */
-			that the current SQL statement ended */
 {
 	trx_t*		trx;
 
@@ -1391,7 +1428,7 @@ innobase_commit(
 
 	innobase_release_stat_resources(trx);
 
-	/* The flag thd->transaction.all.innodb_active_trans is set to 1 in
+        /* The flag trx->active_trans is set to 1 in
 
 	1. ::external_lock(),
 	2. ::start_stmt(),
@@ -1406,23 +1443,22 @@ innobase_commit(
 	For the time being, we play safe and do the cleanup though there should
 	be nothing to clean up. */
 
-	if (thd->transaction.all.innodb_active_trans == 0
+        if (trx->active_trans == 0
 	    && trx->conc_state != TRX_NOT_STARTED) {
 	    
 	        fprintf(stderr,
-"InnoDB: Error: thd->transaction.all.innodb_active_trans == 0\n"
+"InnoDB: Error: trx->active_trans == 0\n"
 "InnoDB: but trx->conc_state != TRX_NOT_STARTED\n");
 	}
 
-	if (trx_handle != (void*)&innodb_dummy_stmt_trx_handle
+        if (all || (!(thd->options & (OPTION_NOT_AUTOCOMMIT | OPTION_BEGIN)))) {
-	    || (!(thd->options & (OPTION_NOT_AUTOCOMMIT | OPTION_BEGIN)))) {
 	        
  		/* We were instructed to commit the whole transaction, or
 		this is an SQL statement end and autocommit is on */
 
 		innobase_commit_low(trx);
 
-		thd->transaction.all.innodb_active_trans = 0;
+                trx->active_trans = 0;
 	} else {
 	        /* We just mark the SQL statement ended and do not do a
 		transaction commit */
@@ -1448,6 +1484,11 @@ innobase_commit(
 	DBUG_RETURN(0);
 }
 
+/*
+  don't delete it - it may be re-enabled later
+  as an optimization for the most common case InnoDB+binlog
+*/
+#if 0
 /*********************************************************************
 This is called when MySQL writes the binlog entry for the current
 transaction. Writes to the InnoDB tablespace info which tells where the
@@ -1473,18 +1514,51 @@ innobase_report_binlog_offset_and_commit(
 
 	ut_a(trx != NULL);
 
-	trx->mysql_log_file_name = log_file_name;  	
+	trx->mysql_log_file_name = log_file_name;
 	trx->mysql_log_offset = (ib_longlong)end_offset;
-	
+
 	trx->flush_log_later = TRUE;
 
-  	innobase_commit(thd, trx_handle);
+	innobase_commit(thd, trx_handle);
 
 	trx->flush_log_later = FALSE;
 
 	return(0);
 }
 
+/***********************************************************************
+This function stores the binlog offset and flushes logs. */
+
+void
+innobase_store_binlog_offset_and_flush_log(
+/*=======================================*/
+    char *binlog_name,          /* in: binlog name */
+    longlong	offset)		/* in: binlog offset */
+{
+	mtr_t mtr;
+
+	assert(binlog_name != NULL);
+
+	/* Start a mini-transaction */
+        mtr_start_noninline(&mtr);
+
+	/* Update the latest MySQL binlog name and offset info
+        in trx sys header */
+
+        trx_sys_update_mysql_binlog_offset(
+            binlog_name,
+            offset,
+            TRX_SYS_MYSQL_LOG_INFO, &mtr);
+
+        /* Commits the mini-transaction */
+        mtr_commit(&mtr);
+
+	/* Syncronous flush of the log buffer to disk */
+	log_buffer_flush_to_disk();
+}
+
+#endif
+
 /*********************************************************************
 This is called after MySQL has written the binlog entry for the current
 transaction. Flushes the InnoDB log files to disk if required. */
@@ -1493,20 +1567,23 @@ int
 innobase_commit_complete(
 /*=====================*/
                                 /* out: 0 */
-        void*   trx_handle)     /* in: InnoDB trx handle */
+        THD*    thd)            /* in: user thread */
 {
 	trx_t*	trx;
 
-	if (srv_flush_log_at_trx_commit == 0) {
+        trx = (trx_t*) thd->ha_data[innobase_hton.slot];
 
-	        return(0);
+        if (trx && trx->active_trans) {
-	}
 
-	trx = (trx_t*)trx_handle;
+                trx->active_trans = 0;
 
-	ut_a(trx != NULL);
+                if (srv_flush_log_at_trx_commit == 0) {
 
-  	trx_commit_complete_for_mysql(trx);
+                        return(0);
+                }
+
+                trx_commit_complete_for_mysql(trx);
+        }
 
 	return(0);
 }
@@ -1514,15 +1591,14 @@ innobase_commit_complete(
 /*********************************************************************
 Rolls back a transaction or the latest SQL statement. */
 
-int
+static int
 innobase_rollback(
 /*==============*/
 			/* out: 0 or error number */
 	THD*	thd,	/* in: handle to the MySQL thread of the user
 			whose transaction should be rolled back */
-	void*	trx_handle)/* in: InnoDB trx handle or a dummy stmt handle;
+        bool    all)    /* in: TRUE - commit transaction
-			the latter means we roll back the latest SQL
+                               FALSE - the current SQL statement ended */
-			statement */
 {
 	int	error = 0;
 	trx_t*	trx;
@@ -1546,11 +1622,10 @@ innobase_rollback(
 		row_unlock_table_autoinc_for_mysql(trx);
 	}
 
-	if (trx_handle != (void*)&innodb_dummy_stmt_trx_handle
+        if (all || (!(thd->options & (OPTION_NOT_AUTOCOMMIT | OPTION_BEGIN)))) {
-	    || (!(thd->options & (OPTION_NOT_AUTOCOMMIT | OPTION_BEGIN)))) {
 
 		error = trx_rollback_for_mysql(trx);
-		thd->transaction.all.innodb_active_trans = 0;
+                trx->active_trans = 0;
 	} else {
 		error = trx_rollback_last_sql_stat_for_mysql(trx);
 	}
@@ -1594,17 +1669,14 @@ innobase_rollback_trx(
 /*********************************************************************
 Rolls back a transaction to a savepoint. */
 
-int
+static int
 innobase_rollback_to_savepoint(
 /*===========================*/
 				/* out: 0 if success, HA_ERR_NO_SAVEPOINT if
 				no savepoint with the given name */
 	THD*	thd,		/* in: handle to the MySQL thread of the user
 				whose transaction should be rolled back */
-	char*	savepoint_name,	/* in: savepoint name */
+        void *savepoint)        /* in: savepoint data */
-	my_off_t* binlog_cache_pos)/* out: position which corresponds to the
-				savepoint in the binlog cache of this
-				transaction, not defined if error */
 {
 	ib_longlong mysql_binlog_cache_pos;
 	int	    error = 0;
@@ -1620,34 +1692,35 @@ innobase_rollback_to_savepoint(
 
 	innobase_release_stat_resources(trx);
 
-	error = trx_rollback_to_savepoint_for_mysql(trx, savepoint_name,
+        /* TODO: use provided savepoint data area to store savepoint data */
+        char name[16]; sprintf(name, "s_%08lx", savepoint);
+        error = trx_rollback_to_savepoint_for_mysql(trx, name,
 						&mysql_binlog_cache_pos);
-	*binlog_cache_pos = (my_off_t)mysql_binlog_cache_pos;
-
 	DBUG_RETURN(convert_error_code_to_mysql(error, NULL));
 }
 
 /*********************************************************************
 Release transaction savepoint name. */
 
-int
+static int
-innobase_release_savepoint_name(
+innobase_release_savepoint(
 /*===========================*/
 				/* out: 0 if success, HA_ERR_NO_SAVEPOINT if
 				no savepoint with the given name */
 	THD*	thd,		/* in: handle to the MySQL thread of the user
 				whose transaction should be rolled back */
-	char*	savepoint_name)	/* in: savepoint name */
+        void *savepoint)        /* in: savepoint data */
 {
-	ib_longlong mysql_binlog_cache_pos;
 	int	    error = 0;
 	trx_t*	    trx;
 
-	DBUG_ENTER("innobase_release_savepoint_name");
+	DBUG_ENTER("innobase_release_savepoint");
 
 	trx = check_trx_exists(thd);
 
-	error = trx_release_savepoint_for_mysql(trx, savepoint_name);
+        /* TODO: use provided savepoint data area to store savepoint data */
+        char name[16]; sprintf(name, "s_%08lx", savepoint);
+	error = trx_release_savepoint_for_mysql(trx, name);
 
 	DBUG_RETURN(convert_error_code_to_mysql(error, NULL));
 }
@@ -1655,17 +1728,12 @@ innobase_release_savepoint_name(
 /*********************************************************************
 Sets a transaction savepoint. */
 
-int
+static int
 innobase_savepoint(
 /*===============*/
 				/* out: always 0, that is, always succeeds */
 	THD*	thd,		/* in: handle to the MySQL thread */
-	char*	savepoint_name,	/* in: savepoint name */
+        void *savepoint)        /* in: savepoint data */
-	my_off_t binlog_cache_pos)/* in: offset up to which the current
-				transaction has cached log entries to its
-				binlog cache, not defined if no transaction
-				active, or we are in the autocommit state, or
-				binlogging is not switched on */
 {
 	int	error = 0;
 	trx_t*	trx;
@@ -1686,14 +1754,12 @@ innobase_savepoint(
 
 	innobase_release_stat_resources(trx);
 
-	/* Setting a savepoint starts a transaction inside InnoDB since
+        /* cannot happen outside of transaction */
-	it allocates resources for it (memory to store the savepoint name,
+        DBUG_ASSERT(trx->active_trans);
-	for example) */
 
-	thd->transaction.all.innodb_active_trans = 1;
+        /* TODO: use provided savepoint data area to store savepoint data */
-
+        char name[16]; sprintf(name, "s_%08lx", savepoint);
-	error = trx_savepoint_for_mysql(trx, savepoint_name,
+        error = trx_savepoint_for_mysql(trx, name, (ib_longlong)0);
-					     (ib_longlong)binlog_cache_pos);
 
 	DBUG_RETURN(convert_error_code_to_mysql(error, NULL));
 }
@@ -1701,25 +1767,14 @@ innobase_savepoint(
 /*********************************************************************
 Frees a possible InnoDB trx object associated with the current THD. */
 
-int
+static int
 innobase_close_connection(
 /*======================*/
 			/* out: 0 or error number */
 	THD*	thd)	/* in: handle to the MySQL thread of the user
-			whose transaction should be rolled back */
+			whose resources should be free'd */
 {
-	trx_t*	trx;
+        trx_free_for_mysql((trx_t*)thd->ha_data[innobase_hton.slot]);
-
-	trx = (trx_t*)thd->transaction.all.innobase_tid;
-
-	if (NULL != trx) {
-	        innobase_rollback(thd, (void*)trx);
-
-		trx_free_for_mysql(trx);
-
-		thd->transaction.all.innobase_tid = NULL;
-	}
-
 	return(0);
 }
 
@@ -2535,19 +2590,19 @@ ha_innobase::write_row(
   	DBUG_ENTER("ha_innobase::write_row");
 
 	if (prebuilt->trx !=
-			(trx_t*) current_thd->transaction.all.innobase_tid) {
+                        (trx_t*) current_thd->ha_data[innobase_hton.slot]) {
 		fprintf(stderr,
 "InnoDB: Error: the transaction object for the table handle is at\n"
 "InnoDB: %p, but for the current thread it is at %p\n",
 			prebuilt->trx,
-			current_thd->transaction.all.innobase_tid);
+                        (trx_t*) current_thd->ha_data[innobase_hton.slot]);
 		fputs("InnoDB: Dump of 200 bytes around prebuilt: ", stderr);
 		ut_print_buf(stderr, ((const byte*)prebuilt) - 100, 200);
 		fputs("\n"
 			"InnoDB: Dump of 200 bytes around transaction.all: ",
 			stderr);
 		ut_print_buf(stderr,
-			((byte*)(&(current_thd->transaction.all))) - 100, 200);
+                        ((byte*)(&(current_thd->ha_data[innobase_hton.slot]))) - 100, 200);
 		putc('\n', stderr);
 		ut_error;
 	}
@@ -2599,7 +2654,7 @@ ha_innobase::write_row(
 			/* Altering to InnoDB format */
 			innobase_commit(user_thd, prebuilt->trx);
 			/* Note that this transaction is still active. */
-			user_thd->transaction.all.innodb_active_trans = 1;
+			prebuilt->trx->active_trans = 1;
 			/* We will need an IX lock on the destination table. */
 		        prebuilt->sql_stat_start = TRUE;
 		} else {
@@ -2614,7 +2669,7 @@ ha_innobase::write_row(
 			locks, so they have to be acquired again. */
 			innobase_commit(user_thd, prebuilt->trx);
 			/* Note that this transaction is still active. */
-			user_thd->transaction.all.innodb_active_trans = 1;
+			prebuilt->trx->active_trans = 1;
 			/* Re-acquire the table lock on the source table. */
 			row_lock_table_for_mysql(prebuilt, src_table, mode);
 			/* We will need an IX lock on the destination table. */
@@ -2902,7 +2957,7 @@ ha_innobase::update_row(
 	DBUG_ENTER("ha_innobase::update_row");
 
 	ut_ad(prebuilt->trx ==
-		(trx_t*) current_thd->transaction.all.innobase_tid);
+                (trx_t*) current_thd->ha_data[innobase_hton.slot]);
 
         if (table->timestamp_field_type & TIMESTAMP_AUTO_SET_ON_UPDATE)
                 table->timestamp_field->set_time();
@@ -2963,7 +3018,7 @@ ha_innobase::delete_row(
 	DBUG_ENTER("ha_innobase::delete_row");
 
 	ut_ad(prebuilt->trx ==
-		(trx_t*) current_thd->transaction.all.innobase_tid);
+                (trx_t*) current_thd->ha_data[innobase_hton.slot]);
 
 	if (last_query_id != user_thd->query_id) {
 	        prebuilt->sql_stat_start = TRUE;
@@ -3173,7 +3228,7 @@ ha_innobase::index_read(
   	DBUG_ENTER("index_read");
 
 	ut_ad(prebuilt->trx ==
-		(trx_t*) current_thd->transaction.all.innobase_tid);
+                (trx_t*) current_thd->ha_data[innobase_hton.slot]);
 
   	statistic_increment(current_thd->status_var.ha_read_key_count,
 			    &LOCK_status);
@@ -3288,7 +3343,7 @@ ha_innobase::change_active_index(
 
 	ut_ad(user_thd == current_thd);
 	ut_ad(prebuilt->trx ==
-	     (trx_t*) current_thd->transaction.all.innobase_tid);
+             (trx_t*) current_thd->ha_data[innobase_hton.slot]);
 
 	active_index = keynr;
 
@@ -3376,7 +3431,7 @@ ha_innobase::general_fetch(
 	DBUG_ENTER("general_fetch");
 
 	ut_ad(prebuilt->trx ==
-	     (trx_t*) current_thd->transaction.all.innobase_tid);
+             (trx_t*) current_thd->ha_data[innobase_hton.slot]);
 
 	innodb_srv_conc_enter_innodb(prebuilt->trx);
 
@@ -3602,7 +3657,7 @@ ha_innobase::rnd_pos(
 			    &LOCK_status);
 
 	ut_ad(prebuilt->trx ==
-		(trx_t*) current_thd->transaction.all.innobase_tid);
+                (trx_t*) current_thd->ha_data[innobase_hton.slot]);
 
 	if (prebuilt->clust_index_was_generated) {
 		/* No primary key was defined for the table and we
@@ -3651,7 +3706,7 @@ ha_innobase::position(
 	uint		len;
 
 	ut_ad(prebuilt->trx ==
-		(trx_t*) current_thd->transaction.all.innobase_tid);
+                (trx_t*) current_thd->ha_data[innobase_hton.slot]);
 
 	if (prebuilt->clust_index_was_generated) {
 		/* No primary key was defined for the table and we
@@ -4141,7 +4196,7 @@ ha_innobase::discard_or_import_tablespace(
 
 	ut_a(prebuilt->trx && prebuilt->trx->magic_n == TRX_MAGIC_N);
 	ut_a(prebuilt->trx ==
-		(trx_t*) current_thd->transaction.all.innobase_tid);
+                (trx_t*) current_thd->ha_data[innobase_hton.slot]);
 
 	dict_table = prebuilt->table;
 	trx = prebuilt->trx;
@@ -4847,7 +4902,7 @@ ha_innobase::check(
 
 	ut_a(prebuilt->trx && prebuilt->trx->magic_n == TRX_MAGIC_N);
 	ut_a(prebuilt->trx ==
-		(trx_t*) current_thd->transaction.all.innobase_tid);
+                (trx_t*) current_thd->ha_data[innobase_hton.slot]);
 
 	if (prebuilt->mysql_template == NULL) {
 		/* Build the template; we will use a dummy template
@@ -5285,7 +5340,11 @@ ha_innobase::start_stmt(
 	}
 
 	/* Set the MySQL flag to mark that there is an active transaction */
-	thd->transaction.all.innodb_active_trans = 1;
+        if (trx->active_trans == 0) {
+
+                register_trans(thd);
+                trx->active_trans = 1;
+        }
 
 	return(0);
 }
@@ -5353,7 +5412,11 @@ ha_innobase::external_lock(
 
 		/* Set the MySQL flag to mark that there is an active
 		transaction */
-		thd->transaction.all.innodb_active_trans = 1;
+                if (trx->active_trans == 0) {
+
+                        register_trans(thd);
+                        trx->active_trans = 1;
+                }
 
 		trx->n_mysql_tables_in_use++;
 		prebuilt->mysql_has_locked = TRUE;
@@ -5433,8 +5496,8 @@ ha_innobase::external_lock(
 		innobase_release_stat_resources(trx);
 
 		if (!(thd->options & (OPTION_NOT_AUTOCOMMIT | OPTION_BEGIN))) {
-			if (thd->transaction.all.innodb_active_trans != 0) {
+                        if (trx->active_trans != 0) {
-		    	        innobase_commit(thd, trx);
+                                innobase_commit(thd, TRUE);
 			}
 		} else {
 			if (trx->isolation_level <= TRX_ISO_READ_COMMITTED
@@ -5513,12 +5576,16 @@ ha_innobase::transactional_table_lock(
 	/* MySQL is setting a new transactional table lock */
 
 	/* Set the MySQL flag to mark that there is an active transaction */
-	thd->transaction.all.innodb_active_trans = 1;
+        if (trx->active_trans == 0) {
+
+                register_trans(thd);
+                trx->active_trans = 1;
+        }
 
 	if (thd->in_lock_tables && thd->variables.innodb_table_locks) {
 		ulint	error = DB_SUCCESS;
 
-		error = row_lock_table_for_mysql(prebuilt,NULL, 
+		error = row_lock_table_for_mysql(prebuilt,NULL,
 						LOCK_TABLE_TRANSACTIONAL);
 
 		if (error != DB_SUCCESS) {
@@ -5917,7 +5984,7 @@ ha_innobase::innobase_read_and_init_auto_inc(
 
   	ut_a(prebuilt);
 	ut_a(prebuilt->trx ==
-		(trx_t*) current_thd->transaction.all.innobase_tid);
+                (trx_t*) current_thd->ha_data[innobase_hton.slot]);
 	ut_a(prebuilt->table);
 	
 	/* In case MySQL calls this in the middle of a SELECT query, release
@@ -6027,37 +6094,6 @@ ha_innobase::get_auto_increment()
 	return((ulonglong) nr);
 }
 
-/***********************************************************************
-This function stores the binlog offset and flushes logs. */
-
-void 
-innobase_store_binlog_offset_and_flush_log(
-/*=======================================*/
-    char *binlog_name,          /* in: binlog name */
-    longlong 	offset)		/* in: binlog offset */
-{
-	mtr_t mtr;
-	
-	assert(binlog_name != NULL);
-
-	/* Start a mini-transaction */
-        mtr_start_noninline(&mtr); 
-
-	/* Update the latest MySQL binlog name and offset info
-        in trx sys header */
-
-        trx_sys_update_mysql_binlog_offset(
-            binlog_name,
-            offset,
-            TRX_SYS_MYSQL_LOG_INFO, &mtr);
-
-        /* Commits the mini-transaction */
-        mtr_commit(&mtr);
-        
-	/* Syncronous flush of the log buffer to disk */
-	log_buffer_flush_to_disk();
-}
-
 
 int
 ha_innobase::cmp_ref(
@@ -6252,9 +6288,7 @@ innobase_xa_prepare(
 
         trx = check_trx_exists(thd);
 
-	/* TODO: Get X/Open XA Transaction Identification from MySQL*/
+        trx->xid=thd->transaction.xid;
-	memset(&trx->xid, 0, sizeof(trx->xid));
-	trx->xid.formatID = -1;
 
 	/* Release a possible FIFO ticket and search latch. Since we will
 	reserve the kernel mutex, we have to release the search system latch
@@ -6265,14 +6299,14 @@ innobase_xa_prepare(
 	if (trx->active_trans == 0 && trx->conc_state != TRX_NOT_STARTED) {
 
 		fprintf(stderr,
-"InnoDB: Error: thd->transaction.all.innodb_active_trans == 0\n"
+"InnoDB: Error: trx->active_trans == 0\n"
 "InnoDB: but trx->conc_state != TRX_NOT_STARTED\n");
 	}
 
 	if (all || (!(thd->options & (OPTION_NOT_AUTOCOMMIT | OPTION_BEGIN)))) {
-	        
+
- 		/* We were instructed to prepare the whole transaction, or
+                /* We were instructed to prepare the whole transaction, or
-		this is an SQL statement end and autocommit is on */
+                this is an SQL statement end and autocommit is on */
 
 		error = trx_prepare_for_mysql(trx);
 	} else {

sql/ha_innodb.h

@@ -245,38 +245,31 @@ extern ulong srv_thread_concurrency;
 
 extern TYPELIB innobase_lock_typelib;
 
-bool innobase_init(void);
+handlerton *innobase_init(void);
 bool innobase_end(void);
 bool innobase_flush_logs(void);
 uint innobase_get_free_space(void);
 
-int innobase_commit(THD *thd, void* trx_handle);
+/*
+  don't delete it - it may be re-enabled later
+  as an optimization for the most common case InnoDB+binlog
+*/
+#if 0
 int innobase_report_binlog_offset_and_commit(
         THD*    thd,
 	void*	trx_handle,
         char*   log_file_name,
         my_off_t end_offset);
-int innobase_commit_complete(
+int innobase_commit_complete(void* trx_handle);
-        void*   trx_handle);
+void innobase_store_binlog_offset_and_flush_log(char *binlog_name,longlong offset);
-int innobase_rollback(THD *thd, void* trx_handle);
+#endif
-int innobase_rollback_to_savepoint(
+
-	THD*	thd,
-	char*	savepoint_name,
-	my_off_t* binlog_cache_pos);
-int innobase_savepoint(
-	THD*	thd,
-	char*	savepoint_name,
-	my_off_t binlog_cache_pos);
-int innobase_release_savepoint_name(
-	THD*	thd,
-	char*	savepoint_name);
-int innobase_close_connection(THD *thd);
 int innobase_drop_database(char *path);
 bool innodb_show_status(THD* thd);
 bool innodb_mutex_show_status(THD* thd);
 void innodb_export_status(void);
 
-void innobase_release_temporary_latches(void* innobase_tid);
+void innobase_release_temporary_latches(THD *thd);
 
 void innobase_store_binlog_offset_and_flush_log(char *binlog_name,longlong offset);
 

sql/ha_ndbcluster.cc

@@ -12,7 +12,7 @@
 
   You should have received a copy of the GNU General Public License
   along with this program; if not, write to the Free Software
-  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA 
+  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
 */
 
 /*
@@ -45,6 +45,23 @@ static const int max_transactions= 256;
 
 static const char *ha_ndb_ext=".ndb";
 
+static int ndbcluster_close_connection(THD *thd);
+static int ndbcluster_commit(THD *thd, bool all);
+static int ndbcluster_rollback(THD *thd, bool all);
+
+static handlerton ndbcluster_hton = {
+  0, /* slot */
+  0, /* savepoint size */
+  ndbcluster_close_connection,
+  NULL, /* savepoint_set */
+  NULL, /* savepoint_rollback */
+  NULL, /* savepoint_release */
+  ndbcluster_commit,
+  ndbcluster_rollback,
+  NULL, /* prepare */
+  NULL  /* recover */
+};
+
 #define NDB_HIDDEN_PRIMARY_KEY_LENGTH 8
 
 
@@ -258,6 +275,8 @@ Thd_ndb::Thd_ndb()
   ndb= new Ndb(g_ndb_cluster_connection, "");
   lock_count= 0;
   count= 0;
+  all= NULL;
+  stmt= NULL;
   error= 0;
 }
 
@@ -268,10 +287,18 @@ Thd_ndb::~Thd_ndb()
   ndb= 0;
 }
 
+inline
+Thd_ndb *
+get_thd_ndb(THD *thd) { return (Thd_ndb *) thd->ha_data[ndbcluster_hton.slot]; }
+
+inline
+void
+set_thd_ndb(THD *thd, Thd_ndb *thd_ndb) { thd->ha_data[ndbcluster_hton.slot]= thd_ndb; }
+
 inline
 Ndb *ha_ndbcluster::get_ndb()
 {
-  return ((Thd_ndb*)current_thd->transaction.thd_ndb)->ndb;
+  return get_thd_ndb(table->in_use)->ndb;
 }
 
 /*
@@ -315,7 +342,7 @@ void ha_ndbcluster::records_update()
   }
   {
     THD *thd= current_thd;
-    if (((Thd_ndb*)(thd->transaction.thd_ndb))->error)
+    if (get_thd_ndb(thd)->error)
       info->no_uncommitted_rows_count= 0;
   }
   records= info->records+ info->no_uncommitted_rows_count;
@@ -327,8 +354,7 @@ void ha_ndbcluster::no_uncommitted_rows_execute_failure()
   if (m_ha_not_exact_count)
     return;
   DBUG_ENTER("ha_ndbcluster::no_uncommitted_rows_execute_failure");
-  THD *thd= current_thd;
+  get_thd_ndb(table->in_use)->error= 1;
-  ((Thd_ndb*)(thd->transaction.thd_ndb))->error= 1;
   DBUG_VOID_RETURN;
 }
 
@@ -338,7 +364,7 @@ void ha_ndbcluster::no_uncommitted_rows_init(THD *thd)
     return;
   DBUG_ENTER("ha_ndbcluster::no_uncommitted_rows_init");
   struct Ndb_table_local_info *info= (struct Ndb_table_local_info *)m_table_info;
-  Thd_ndb *thd_ndb= (Thd_ndb *)thd->transaction.thd_ndb;
+  Thd_ndb *thd_ndb= get_thd_ndb(thd);
   if (info->last_count != thd_ndb->count)
   {
     info->last_count= thd_ndb->count;
@@ -370,14 +396,15 @@ void ha_ndbcluster::no_uncommitted_rows_reset(THD *thd)
   if (m_ha_not_exact_count)
     return;
   DBUG_ENTER("ha_ndbcluster::no_uncommitted_rows_reset");
-  ((Thd_ndb*)(thd->transaction.thd_ndb))->count++;
+  Thd_ndb *thd_ndb= get_thd_ndb(thd);
-  ((Thd_ndb*)(thd->transaction.thd_ndb))->error= 0;
+  thd_ndb->count++;
+  thd_ndb->error= 0;
   DBUG_VOID_RETURN;
 }
 
 /*
   Take care of the error that occured in NDB
-  
+
   RETURN
     0	No error
     #   The mapped error code
@@ -3136,9 +3163,9 @@ THR_LOCK_DATA **ha_ndbcluster::store_lock(THD *thd,
   As MySQL will execute an external lock for every new table it uses
   we can use this to start the transactions.
   If we are in auto_commit mode we just need to start a transaction
-  for the statement, this will be stored in transaction.stmt.
+  for the statement, this will be stored in thd_ndb.stmt.
   If not, we have to start a master transaction if there doesn't exist
-  one from before, this will be stored in transaction.all
+  one from before, this will be stored in thd_ndb.all
  
   When a table lock is held one transaction will be started which holds
   the table lock and for each statement a hupp transaction will be started  
@@ -3157,7 +3184,7 @@ int ha_ndbcluster::external_lock(THD *thd, int lock_type)
   if (check_ndb_connection())
     DBUG_RETURN(1);
  
-  Thd_ndb *thd_ndb= (Thd_ndb*)thd->transaction.thd_ndb;
+  Thd_ndb *thd_ndb= get_thd_ndb(thd);
   Ndb *ndb= thd_ndb->ndb;
 
   DBUG_PRINT("enter", ("transaction.thd_ndb->lock_count: %d", 
@@ -3173,18 +3200,19 @@ int ha_ndbcluster::external_lock(THD *thd, int lock_type)
       if (!(thd->options & (OPTION_NOT_AUTOCOMMIT | OPTION_BEGIN | OPTION_TABLE_LOCK))) 
       {
         // Autocommit transaction
-        DBUG_ASSERT(!thd->transaction.stmt.ndb_tid);
+        DBUG_ASSERT(!thd_ndb->stmt);
         DBUG_PRINT("trans",("Starting transaction stmt"));      
 
         trans= ndb->startTransaction();
         if (trans == NULL)
           ERR_RETURN(ndb->getNdbError());
 	no_uncommitted_rows_reset(thd);
-        thd->transaction.stmt.ndb_tid= trans;
+        thd_ndb->stmt= trans;
+        trans_register_ha(thd, FALSE, &ndbcluster_hton);
       } 
       else 
       { 
-        if (!thd->transaction.all.ndb_tid)
+        if (!thd_ndb->all)
 	{
           // Not autocommit transaction
           // A "master" transaction ha not been started yet
@@ -3194,6 +3222,8 @@ int ha_ndbcluster::external_lock(THD *thd, int lock_type)
           if (trans == NULL)
             ERR_RETURN(ndb->getNdbError());
 	  no_uncommitted_rows_reset(thd);
+          thd_ndb->all= trans; 
+          trans_register_ha(thd, TRUE, &ndbcluster_hton);
 
           /*
             If this is the start of a LOCK TABLE, a table look 
@@ -3207,7 +3237,6 @@ int ha_ndbcluster::external_lock(THD *thd, int lock_type)
             DBUG_PRINT("info", ("Locking the table..." ));
           }
 
-          thd->transaction.all.ndb_tid= trans; 
         }
       }
     }
@@ -3232,9 +3261,7 @@ int ha_ndbcluster::external_lock(THD *thd, int lock_type)
     else
       m_transaction_on= thd->variables.ndb_use_transactions;
 
-    m_active_trans= thd->transaction.all.ndb_tid ? 
+    m_active_trans= thd_ndb->all ? thd_ndb->all : thd_ndb->stmt;
-      (NdbTransaction*)thd->transaction.all.ndb_tid:
-      (NdbTransaction*)thd->transaction.stmt.ndb_tid;
     DBUG_ASSERT(m_active_trans);
     // Start of transaction
     m_retrieve_all_fields= FALSE;
@@ -3260,7 +3287,7 @@ int ha_ndbcluster::external_lock(THD *thd, int lock_type)
       DBUG_PRINT("trans", ("Last external_lock"));
       PRINT_OPTION_FLAGS(thd);
 
-      if (thd->transaction.stmt.ndb_tid)
+      if (thd_ndb->stmt)
       {
         /*
           Unlock is done without a transaction commit / rollback.
@@ -3269,7 +3296,7 @@ int ha_ndbcluster::external_lock(THD *thd, int lock_type)
         */
         DBUG_PRINT("trans",("ending non-updating transaction"));
         ndb->closeTransaction(m_active_trans);
-        thd->transaction.stmt.ndb_tid= 0;
+        thd_ndb->stmt= NULL;
       }
     }
     m_table= NULL;
@@ -3316,14 +3343,14 @@ int ha_ndbcluster::start_stmt(THD *thd)
   DBUG_ENTER("start_stmt");
   PRINT_OPTION_FLAGS(thd);
 
-  NdbTransaction *trans= (NdbTransaction*)thd->transaction.stmt.ndb_tid;
+  Thd_ndb *thd_ndb= get_thd_ndb(thd);
+  NdbTransaction *trans= thd_ndb->stmt;
   if (!trans){
-    Ndb *ndb= ((Thd_ndb*)thd->transaction.thd_ndb)->ndb;
+    Ndb *ndb= thd_ndb->ndb;
     DBUG_PRINT("trans",("Starting transaction stmt"));  
 
 #if 0    
-    NdbTransaction *tablock_trans= 
+    NdbTransaction *tablock_trans= thd_ndb->all;
-      (NdbTransaction*)thd->transaction.all.ndb_tid;
     DBUG_PRINT("info", ("tablock_trans: %x", (UintPtr)tablock_trans));
     DBUG_ASSERT(tablock_trans);
 //    trans= ndb->hupp(tablock_trans);
@@ -3332,7 +3359,8 @@ int ha_ndbcluster::start_stmt(THD *thd)
     if (trans == NULL)
       ERR_RETURN(ndb->getNdbError());
     no_uncommitted_rows_reset(thd);
-    thd->transaction.stmt.ndb_tid= trans;
+    thd_ndb->stmt= trans;
+    trans_register_ha(thd, FALSE, &ndbcluster_hton);
   }
   m_active_trans= trans;
 
@@ -3349,15 +3377,16 @@ int ha_ndbcluster::start_stmt(THD *thd)
   Commit a transaction started in NDB 
  */
 
-int ndbcluster_commit(THD *thd, void *ndb_transaction)
+int ndbcluster_commit(THD *thd, bool all)
 {
   int res= 0;
-  Ndb *ndb= ((Thd_ndb*)thd->transaction.thd_ndb)->ndb;
+  Thd_ndb *thd_ndb= get_thd_ndb(thd);
-  NdbTransaction *trans= (NdbTransaction*)ndb_transaction;
+  Ndb *ndb= thd_ndb->ndb;
+  NdbTransaction *trans= all ? thd_ndb->all : thd_ndb->stmt;
 
   DBUG_ENTER("ndbcluster_commit");
   DBUG_PRINT("transaction",("%s",
-                            trans == thd->transaction.stmt.ndb_tid ? 
+                            trans == thd_ndb->stmt ? 
                             "stmt" : "all"));
   DBUG_ASSERT(ndb && trans);
 
@@ -3371,6 +3400,7 @@ int ndbcluster_commit(THD *thd, void *ndb_transaction)
       ndbcluster_print_error(res, error_op);
   }
   ndb->closeTransaction(trans);
+  thd_ndb->all= thd_ndb->stmt= NULL;
   DBUG_RETURN(res);
 }
 
@@ -3379,15 +3409,16 @@ int ndbcluster_commit(THD *thd, void *ndb_transaction)
   Rollback a transaction started in NDB
  */
 
-int ndbcluster_rollback(THD *thd, void *ndb_transaction)
+int ndbcluster_rollback(THD *thd, bool all)
 {
   int res= 0;
-  Ndb *ndb= ((Thd_ndb*)thd->transaction.thd_ndb)->ndb;
+  Thd_ndb *thd_ndb= get_thd_ndb(thd);
-  NdbTransaction *trans= (NdbTransaction*)ndb_transaction;
+  Ndb *ndb= thd_ndb->ndb;
+  NdbTransaction *trans= all ? thd_ndb->all : thd_ndb->stmt;
 
   DBUG_ENTER("ndbcluster_rollback");
   DBUG_PRINT("transaction",("%s",
-                            trans == thd->transaction.stmt.ndb_tid ? 
+                            trans == thd_ndb->stmt ? 
                             "stmt" : "all"));
   DBUG_ASSERT(ndb && trans);
 
@@ -3401,7 +3432,8 @@ int ndbcluster_rollback(THD *thd, void *ndb_transaction)
       ndbcluster_print_error(res, error_op);
   }
   ndb->closeTransaction(trans);
-  DBUG_RETURN(0);
+  thd_ndb->all= thd_ndb->stmt= NULL;
+  DBUG_RETURN(res);
 }
 
 
@@ -4257,12 +4289,12 @@ void ha_ndbcluster::release_thd_ndb(Thd_ndb* thd_ndb)
 
 Ndb* check_ndb_in_thd(THD* thd)
 {
-  Thd_ndb *thd_ndb= (Thd_ndb*)thd->transaction.thd_ndb;  
+  Thd_ndb *thd_ndb= get_thd_ndb(thd);
   if (!thd_ndb)
   {
     if (!(thd_ndb= ha_ndbcluster::seize_thd_ndb()))
       return NULL;
-    thd->transaction.thd_ndb= thd_ndb;
+    set_thd_ndb(thd, thd_ndb);
   }
   return thd_ndb->ndb;
 }
@@ -4282,16 +4314,16 @@ int ha_ndbcluster::check_ndb_connection()
 }
 
 
-void ndbcluster_close_connection(THD *thd)
+int ndbcluster_close_connection(THD *thd)
 {
-  Thd_ndb *thd_ndb= (Thd_ndb*)thd->transaction.thd_ndb;
+  Thd_ndb *thd_ndb= get_thd_ndb(thd);
   DBUG_ENTER("ndbcluster_close_connection");
   if (thd_ndb)
   {
     ha_ndbcluster::release_thd_ndb(thd_ndb);
-    thd->transaction.thd_ndb= NULL;
+    set_thd_ndb(thd, NULL); // not strictly required but does not hurt either
   }
-  DBUG_VOID_RETURN;
+  DBUG_RETURN(0);
 }
 
 
@@ -4554,7 +4586,8 @@ static int connect_callback()
   return 0;
 }
 
-bool ndbcluster_init()
+handlerton *
+ndbcluster_init()
 {
   int res;
   DBUG_ENTER("ndbcluster_init");
@@ -4630,11 +4663,11 @@ bool ndbcluster_init()
   }
   
   ndbcluster_inited= 1;
-  DBUG_RETURN(FALSE);
+  DBUG_RETURN(&ndbcluster_hton);
 
  ndbcluster_init_error:
   ndbcluster_end();
-  DBUG_RETURN(TRUE);
+  DBUG_RETURN(NULL);
 }
 
 

sql/ha_ndbcluster.h

@@ -74,6 +74,8 @@ class Thd_ndb {
   Ndb *ndb;
   ulong count;
   uint lock_count;
+  NdbTransaction *all;
+  NdbTransaction *stmt;
   int error;
 };
 
@@ -286,14 +288,9 @@ private:
 
 extern struct show_var_st ndb_status_variables[];
 
-bool ndbcluster_init(void);
+handlerton *ndbcluster_init(void);
 bool ndbcluster_end(void);
 
-int ndbcluster_commit(THD *thd, void* ndb_transaction);
-int ndbcluster_rollback(THD *thd, void* ndb_transaction);
-
-void ndbcluster_close_connection(THD *thd);
-
 int ndbcluster_discover(THD* thd, const char* dbname, const char* name,
 			const void** frmblob, uint* frmlen);
 int ndbcluster_find_files(THD *thd,const char *db,const char *path,

sql/handler.h

@@ -93,6 +93,11 @@
 #define HA_KEY_SWITCH_NONUNIQ_SAVE 2
 #define HA_KEY_SWITCH_ALL_SAVE     3
 
+/*
+  Note: the following includes binlog and closing 0.
+  so: innodb+bdb+ndb+binlog+0
+*/
+#define MAX_HA 5
 
 /*
   Bits in index_ddl_flags(KEY *wanted_index)
@@ -192,16 +197,12 @@ enum row_type { ROW_TYPE_NOT_USED=-1, ROW_TYPE_DEFAULT, ROW_TYPE_FIXED,
 #define HA_CREATE_USED_COMMENT          (1L << 16)
 #define HA_CREATE_USED_PASSWORD         (1L << 17)
 
-typedef struct st_thd_trans {
+typedef ulonglong my_xid;
-  void *bdb_tid;
+#define MYSQL_XID_PREFIX "MySQLXid"
-  void *innobase_tid;
+#define MYSQL_XID_PREFIX_LEN 8 // must be a multiple of 8
-  bool innodb_active_trans;
+#define MYSQL_XID_OFFSET (MYSQL_XID_PREFIX_LEN+sizeof(server_id))
-  void *ndb_tid;
+#define MYSQL_XID_GTRID_LEN (MYSQL_XID_OFFSET+sizeof(my_xid))
-} THD_TRANS;
 
-#ifndef XIDDATASIZE  /* no xa.h included */
-
-/* XXX - may be we should disable xa completely in this case ? */
 #define XIDDATASIZE 128
 #define MAXGTRIDSIZE 64
 #define MAXBQUALSIZE 64
@@ -210,22 +211,106 @@ struct xid_t {
   long formatID;
   long gtrid_length;
   long bqual_length;
-  char data[XIDDATASIZE];
+  char data[XIDDATASIZE];  // not \0-terminated !
-};
 
+  bool eq(LEX_STRING *l) { return eq(l->length, 0, l->str); }
+  bool eq(long g, long b, const char *d)
+  { return g == gtrid_length && b == bqual_length && !memcmp(d, data, g+b); }
+  void set(LEX_STRING *l) { set(l->length, 0, l->str); }
+  void set(ulonglong l)
+  {
+    set(MYSQL_XID_PREFIX_LEN, 0, MYSQL_XID_PREFIX);
+    *(ulong*)(data+MYSQL_XID_PREFIX_LEN)=server_id;
+    *(my_xid*)(data+MYSQL_XID_OFFSET)=l;
+    gtrid_length=MYSQL_XID_GTRID_LEN;
+  }
+  void set(long g, long b, const char *d)
+  {
+    formatID=1;
+    gtrid_length= g;
+    bqual_length= b;
+    memcpy(data, d, g+b);
+  }
+  bool is_null() { return formatID == -1; }
+  void null() { formatID= -1; }
+  my_xid quick_get_my_xid()
+  {
+    return *(my_xid*)(data+MYSQL_XID_OFFSET);
+  }
+  my_xid get_my_xid()
+  {
+    return gtrid_length == MYSQL_XID_GTRID_LEN && bqual_length == 0 &&
+           *(ulong*)(data+MYSQL_XID_PREFIX_LEN) == server_id &&
+           !memcmp(data, MYSQL_XID_PREFIX, MYSQL_XID_PREFIX_LEN) ?
+           quick_get_my_xid() : 0;
+  }
+};
 typedef struct xid_t XID;
 
+/* for recover() handlerton call */
+#define MIN_XID_LIST_SIZE  128
+#define MAX_XID_LIST_SIZE  (1024*128)
 
-#endif
+/*
+  handlerton is a singleton structure - one instance per storage engine -
+  to provide access to storage engine functionality that works on
+  "global" level (unlike handler class that works on per-table basis)
 
+  usually handlerton instance is defined statically in ha_xxx.cc as
+
+  static handlerton { ... } xxx_hton;
+
+  savepoint_*, prepare, recover, and *_by_xid pointers can be 0.
+*/
 typedef struct
 {
-   byte slot;
+  /*
+    each storage engine has it's own memory area (actually a pointer)
+    in the thd, for storing per-connection information.
+    It is accessed as
+
+      thd->ha_data[xxx_hton.slot]
+
+   slot number is initialized by MySQL after xxx_init() is called.
+   */
+   uint slot;
+   /*
+     to store per-savepoint data storage engine is provided with an area
+     of a requested size (0 is ok here).
+     savepoint_offset must be initialized statically to the size of
+     the needed memory to store per-savepoint information.
+     After xxx_init it is changed to be an offset to savepoint storage
+     area and need not be used by storage engine.
+     see binlog_hton and binlog_savepoint_set/rollback for an example.
+   */
    uint savepoint_offset;
+   /*
+     handlerton methods:
+
+     close_connection is only called if
+     thd->ha_data[xxx_hton.slot] is non-zero, so even if you don't need
+     this storage area - set it to something, so that MySQL would know
+     this storage engine was accessed in this connection
+   */
    int  (*close_connection)(THD *thd);
+   /*
+     sv points to an uninitialized storage area of requested size
+     (see savepoint_offset description)
+   */
    int  (*savepoint_set)(THD *thd, void *sv);
+   /*
+     sv points to a storage area, that was earlier passed
+     to the savepoint_set call
+   */
    int  (*savepoint_rollback)(THD *thd, void *sv);
    int  (*savepoint_release)(THD *thd, void *sv);
+   /*
+     'all' is true if it's a real commit, that makes persistent changes
+     'all' is false if it's not in fact a commit but an end of the
+     statement that is part of the transaction.
+     NOTE 'all' is also false in auto-commit mode where 'end of statement'
+     and 'real commit' mean the same event.
+   */
    int  (*commit)(THD *thd, bool all);
    int  (*rollback)(THD *thd, bool all);
    int  (*prepare)(THD *thd, bool all);
@@ -234,6 +319,16 @@ typedef struct
    int  (*rollback_by_xid)(XID *xid);
 } handlerton;
 
+typedef struct st_thd_trans
+{
+  /* number of entries in the ht[] */
+  uint        nht;
+  /* true is not all entries in the ht[] support 2pc */
+  bool        no_2pc;
+  /* storage engines that registered themselves for this transaction */
+  handlerton *ht[MAX_HA];
+} THD_TRANS;
+
 enum enum_tx_isolation { ISO_READ_UNCOMMITTED, ISO_READ_COMMITTED,
 			 ISO_REPEATABLE_READ, ISO_SERIALIZABLE};
 
@@ -268,6 +363,9 @@ typedef struct st_table TABLE;
 struct st_foreign_key_info;
 typedef struct st_foreign_key_info FOREIGN_KEY_INFO;
 
+typedef struct st_savepoint SAVEPOINT;
+extern ulong savepoint_alloc_size;
+
 typedef struct st_ha_check_opt
 {
   ulong sort_buffer_size;
@@ -626,57 +724,72 @@ public:
 extern struct show_table_type_st sys_table_types[];
 extern const char *ha_row_type[];
 extern TYPELIB tx_isolation_typelib;
+extern handlerton *handlertons[MAX_HA];
+extern ulong total_ha, total_ha_2pc;
 
 	/* Wrapper functions */
-#define ha_commit_stmt(thd) (ha_commit_trans((thd), &((thd)->transaction.stmt)))
+#define ha_commit_stmt(thd) (ha_commit_trans((thd), FALSE))
-#define ha_rollback_stmt(thd) (ha_rollback_trans((thd), &((thd)->transaction.stmt)))
+#define ha_rollback_stmt(thd) (ha_rollback_trans((thd), FALSE))
-#define ha_commit(thd) (ha_commit_trans((thd), &((thd)->transaction.all)))
+#define ha_commit(thd) (ha_commit_trans((thd), TRUE))
-#define ha_rollback(thd) (ha_rollback_trans((thd), &((thd)->transaction.all)))
+#define ha_rollback(thd) (ha_rollback_trans((thd), TRUE))
 
 #define ha_supports_generate(T) (T != DB_TYPE_INNODB && \
                                  T != DB_TYPE_BERKELEY_DB && \
                                  T != DB_TYPE_NDBCLUSTER)
 
+/* lookups */
 enum db_type ha_resolve_by_name(const char *name, uint namelen);
 const char *ha_get_storage_engine(enum db_type db_type);
 handler *get_new_handler(TABLE *table, enum db_type db_type);
-my_off_t ha_get_ptr(byte *ptr, uint pack_length);
-void ha_store_ptr(byte *buff, uint pack_length, my_off_t pos);
-int ha_init(void);
-int ha_panic(enum ha_panic_function flag);
-void ha_close_connection(THD* thd);
 enum db_type ha_checktype(enum db_type database_type);
+
+/* basic stuff */
+int ha_init(void);
+TYPELIB *ha_known_exts(void);
+int ha_panic(enum ha_panic_function flag);
+int ha_update_statistics();
+void ha_close_connection(THD* thd);
+bool ha_flush_logs(void);
+void ha_drop_database(char* path);
 int ha_create_table(const char *name, HA_CREATE_INFO *create_info,
 		    bool update_create_info);
+int ha_delete_table(enum db_type db_type, const char *path);
+
+/* discovery */
 int ha_create_table_from_engine(THD* thd, const char *db, const char *name,
 				bool create_if_found);
-int ha_delete_table(enum db_type db_type, const char *path);
+int ha_discover(THD* thd, const char* dbname, const char* name,
-void ha_drop_database(char* path);
+                const void** frmblob, uint* frmlen);
+int ha_find_files(THD *thd,const char *db,const char *path,
+                  const char *wild, bool dir,List<char>* files);
+int ha_table_exists(THD* thd, const char* db, const char* name);
+
+/* key cache */
 int ha_init_key_cache(const char *name, KEY_CACHE *key_cache);
 int ha_resize_key_cache(KEY_CACHE *key_cache);
 int ha_change_key_cache_param(KEY_CACHE *key_cache);
+int ha_change_key_cache(KEY_CACHE *old_key_cache, KEY_CACHE *new_key_cache);
 int ha_end_key_cache(KEY_CACHE *key_cache);
-int ha_start_stmt(THD *thd);
+
-int ha_report_binlog_offset_and_commit(THD *thd, char *log_file_name,
+/* weird stuff */
-				       my_off_t end_offset);
-int ha_commit_complete(THD *thd);
 int ha_release_temporary_latches(THD *thd);
-int ha_update_statistics();
+
-int ha_commit_trans(THD *thd, THD_TRANS *trans);
+/* transactions: interface to handlerton functions */
-int ha_rollback_trans(THD *thd, THD_TRANS *trans);
-int ha_rollback_to_savepoint(THD *thd, char *savepoint_name);
-int ha_savepoint(THD *thd, char *savepoint_name);
-int ha_release_savepoint_name(THD *thd, char *savepoint_name);
-int ha_autocommit_or_rollback(THD *thd, int error);
-void ha_set_spin_retries(uint retries);
-bool ha_flush_logs(void);
-int ha_enable_transaction(THD *thd, bool on);
-int ha_change_key_cache(KEY_CACHE *old_key_cache,
-			KEY_CACHE *new_key_cache);
-int ha_discover(THD* thd, const char* dbname, const char* name,
-		const void** frmblob, uint* frmlen);
-int ha_find_files(THD *thd,const char *db,const char *path,
-		  const char *wild, bool dir,List<char>* files);
-int ha_table_exists(THD* thd, const char* db, const char* name);
-TYPELIB *ha_known_exts(void);
 int ha_start_consistent_snapshot(THD *thd);
+int ha_commit_or_rollback_by_xid(LEX_STRING *ident, bool commit);
+int ha_commit_one_phase(THD *thd, bool all);
+int ha_rollback_trans(THD *thd, bool all);
+int ha_prepare(THD *thd);
+int ha_recover(HASH *commit_list);
+
+/* transactions: these functions never call handlerton functions directly */
+int ha_commit_trans(THD *thd, bool all);
+int ha_autocommit_or_rollback(THD *thd, int error);
+int ha_enable_transaction(THD *thd, bool on);
+void trans_register_ha(THD *thd, bool all, handlerton *ht);
+
+/* savepoints */
+int ha_rollback_to_savepoint(THD *thd, SAVEPOINT *sv);
+int ha_savepoint(THD *thd, SAVEPOINT *sv);
+int ha_release_savepoint(THD *thd, SAVEPOINT *sv);
+

sql/item_func.cc

@@ -2823,7 +2823,7 @@ void item_user_lock_release(User_level_lock *ull)
     tmp.copy(command, strlen(command), tmp.charset());
     tmp.append(ull->key,ull->key_length);
     tmp.append("\")", 2);
-    Query_log_event qev(current_thd, tmp.ptr(), tmp.length(),1, FALSE);
+    Query_log_event qev(current_thd, tmp.ptr(), tmp.length(), 0, FALSE);
     qev.error_code=0; // this query is always safe to run on slave
     mysql_bin_log.write(&qev);
   }

sql/lex.h

@@ -19,11 +19,11 @@
 
 #include "lex_symbol.h"
 
-/* We don't want to include sql_yacc.h into gen_lex_hash */
 SYM_GROUP sym_group_common= {"", ""};
 SYM_GROUP sym_group_geom= {"Spatial extentions", "HAVE_SPATIAL"};
 SYM_GROUP sym_group_rtree= {"RTree keys", "HAVE_RTREE_KEYS"};
 
+/* We don't want to include sql_yacc.h into gen_lex_hash */
 #ifdef NO_YACC_SYMBOLS
 #define SYM_OR_NULL(A) 0
 #else
@@ -314,6 +314,7 @@ static SYMBOL symbols[] = {
   { "MERGE",		SYM(MERGE_SYM)},
   { "MICROSECOND",	SYM(MICROSECOND_SYM)},
   { "MIDDLEINT",	SYM(MEDIUMINT)},	/* For powerbuilder */
+  { "MIGRATE",          SYM(MIGRATE_SYM)},
   { "MINUTE",		SYM(MINUTE_SYM)},
   { "MINUTE_MICROSECOND", SYM(MINUTE_MICROSECOND_SYM)},
   { "MINUTE_SECOND",	SYM(MINUTE_SECOND_SYM)},
@@ -346,6 +347,7 @@ static SYMBOL symbols[] = {
   { "OFFSET",		SYM(OFFSET_SYM)},
   { "OLD_PASSWORD",	SYM(OLD_PASSWORD)},
   { "ON",		SYM(ON)},
+  { "ONE",              SYM(ONE_SYM)},
   { "ONE_SHOT",		SYM(ONE_SHOT_SYM)},
   { "OPEN",		SYM(OPEN_SYM)},
   { "OPTIMIZE",		SYM(OPTIMIZE)},
@@ -359,6 +361,7 @@ static SYMBOL symbols[] = {
   { "PACK_KEYS",	SYM(PACK_KEYS_SYM)},
   { "PARTIAL",		SYM(PARTIAL)},
   { "PASSWORD",		SYM(PASSWORD)},
+  { "PHASE",            SYM(PHASE_SYM)},
   { "POINT",		SYM(POINT_SYM)},
   { "POLYGON",		SYM(POLYGON)},
   { "PRECISION",	SYM(PRECISION)},
@@ -380,6 +383,7 @@ static SYMBOL symbols[] = {
   { "READ",		SYM(READ_SYM)},
   { "READS",		SYM(READS_SYM)},
   { "REAL",		SYM(REAL)},
+  { "RECOVER",          SYM(RECOVER_SYM)},
   { "REDUNDANT",	SYM(REDUNDANT_SYM)},
   { "REFERENCES",	SYM(REFERENCES)},
   { "REGEXP",		SYM(REGEXP)},
@@ -398,6 +402,7 @@ static SYMBOL symbols[] = {
   { "RESET",		SYM(RESET_SYM)},
   { "RESTORE",		SYM(RESTORE_SYM)},
   { "RESTRICT",		SYM(RESTRICT)},
+  { "RESUME",           SYM(RESUME_SYM)},
   { "RETURN",           SYM(RETURN_SYM)},
   { "RETURNS",		SYM(RETURNS_SYM)},
   { "REVOKE",		SYM(REVOKE)},
@@ -467,6 +472,7 @@ static SYMBOL symbols[] = {
   { "STRIPED",		SYM(RAID_STRIPED_SYM)},
   { "SUBJECT",		SYM(SUBJECT_SYM)},
   { "SUPER",		SYM(SUPER_SYM)},
+  { "SUSPEND",          SYM(SUSPEND_SYM)},
   { "TABLE",		SYM(TABLE_SYM)},
   { "TABLES",		SYM(TABLES)},
   { "TABLESPACE",	SYM(TABLESPACE)},
@@ -528,6 +534,7 @@ static SYMBOL symbols[] = {
   { "WRITE",		SYM(WRITE_SYM)},
   { "X509",		SYM(X509_SYM)},
   { "XOR",		SYM(XOR)},
+  { "XA",               SYM(XA_SYM)},
   { "YEAR",		SYM(YEAR_SYM)},
   { "YEAR_MONTH",	SYM(YEAR_MONTH_SYM)},
   { "ZEROFILL",		SYM(ZEROFILL)},

sql/log_event.cc

@@ -269,9 +269,10 @@ const char* Log_event::get_type_str()
   case DELETE_FILE_EVENT: return "Delete_file";
   case EXEC_LOAD_EVENT: return "Exec_load";
   case RAND_EVENT: return "RAND";
+  case XID_EVENT: return "Xid";
   case USER_VAR_EVENT: return "User var";
   case FORMAT_DESCRIPTION_EVENT: return "Format_desc";
-  default: return "Unknown";				/* impossible */ 
+  default: return "Unknown";				/* impossible */
   }
 }
 
@@ -286,16 +287,15 @@ Log_event::Log_event(THD* thd_arg, uint16 flags_arg, bool using_trans)
 {
   server_id=	thd->server_id;
   when=		thd->start_time;
-  cache_stmt=	(using_trans &&
+  cache_stmt=	using_trans;
-		 (thd->options & (OPTION_NOT_AUTOCOMMIT | OPTION_BEGIN)));
 }
 
 
 /*
-  This minimal constructor is for when you are not even sure that there is a
+  This minimal constructor is for when you are not even sure that there
-  valid THD. For example in the server when we are shutting down or flushing
+  is a valid THD. For example in the server when we are shutting down or
-  logs after receiving a SIGHUP (then we must write a Rotate to the binlog but
+  flushing logs after receiving a SIGHUP (then we must write a Rotate to
-  we have no THD, so we need this minimal constructor).
+  the binlog but we have no THD, so we need this minimal constructor).
 */
 
 Log_event::Log_event()
@@ -314,12 +314,12 @@ Log_event::Log_event()
 */
 
 Log_event::Log_event(const char* buf,
-                     const Format_description_log_event* description_event) 
+                     const Format_description_log_event* description_event)
   :temp_buf(0), cache_stmt(0)
 {
 #ifndef MYSQL_CLIENT
   thd = 0;
-#endif  
+#endif
   when = uint4korr(buf);
   server_id = uint4korr(buf + SERVER_ID_OFFSET);
   if (description_event->binlog_version==1)
@@ -331,14 +331,14 @@ Log_event::Log_event(const char* buf,
   /* 4.0 or newer */
   log_pos= uint4korr(buf + LOG_POS_OFFSET);
   /*
-    If the log is 4.0 (so here it can only be a 4.0 relay log read by the SQL
+    If the log is 4.0 (so here it can only be a 4.0 relay log read by
-    thread or a 4.0 master binlog read by the I/O thread), log_pos is the
+    the SQL thread or a 4.0 master binlog read by the I/O thread),
-    beginning of the event: we transform it into the end of the event, which is
+    log_pos is the beginning of the event: we transform it into the end
-    more useful.
+    of the event, which is more useful.
-    But how do you know that the log is 4.0: you know it if description_event
+    But how do you know that the log is 4.0: you know it if
-    is version 3 *and* you are not reading a Format_desc (remember that
+    description_event is version 3 *and* you are not reading a
-    mysqlbinlog starts by assuming that 5.0 logs are in 4.0 format, until it
+    Format_desc (remember that mysqlbinlog starts by assuming that 5.0
-    finds a Format_desc).
+    logs are in 4.0 format, until it finds a Format_desc).
   */
   if (description_event->binlog_version==3 &&
       buf[EVENT_TYPE_OFFSET]<FORMAT_DESCRIPTION_EVENT && log_pos)
@@ -346,13 +346,13 @@ Log_event::Log_event(const char* buf,
       /*
         If log_pos=0, don't change it. log_pos==0 is a marker to mean
         "don't change rli->group_master_log_pos" (see
-        inc_group_relay_log_pos()). As it is unreal log_pos, adding the event
+        inc_group_relay_log_pos()). As it is unreal log_pos, adding the
-        len's is nonsense. For example, a fake Rotate event should 
+        event len's is nonsense. For example, a fake Rotate event should
         not have its log_pos (which is 0) changed or it will modify
-        Exec_master_log_pos in SHOW SLAVE STATUS, displaying a nonsense value
+        Exec_master_log_pos in SHOW SLAVE STATUS, displaying a nonsense
-        of (a non-zero offset which does not exist in the master's binlog, so
+        value of (a non-zero offset which does not exist in the master's
-        which will cause problems if the user uses this value in
+        binlog, so which will cause problems if the user uses this value
-        CHANGE MASTER).
+        in CHANGE MASTER).
       */
     log_pos+= uint4korr(buf + EVENT_LEN_OFFSET);
   }
@@ -363,16 +363,17 @@ Log_event::Log_event(const char* buf,
       (buf[EVENT_TYPE_OFFSET] == ROTATE_EVENT))
   {
     /*
-      These events always have a header which stops here (i.e. their header is
+      These events always have a header which stops here (i.e. their
-      FROZEN).
+      header is FROZEN).
     */
     /*
-      Initialization to zero of all other Log_event members as they're not
+      Initialization to zero of all other Log_event members as they're
-      specified. Currently there are no such members; in the future there will
+      not specified. Currently there are no such members; in the future
-      be an event UID (but Format_description and Rotate don't need this UID,
+      there will be an event UID (but Format_description and Rotate
-      as they are not propagated through --log-slave-updates (remember the UID
+      don't need this UID, as they are not propagated through
-      is used to not play a query twice when you have two masters which are
+      --log-slave-updates (remember the UID is used to not play a query
-      slaves of a 3rd master). Then we are done.
+      twice when you have two masters which are slaves of a 3rd master).
+      Then we are done.
     */
     return;
   }
@@ -402,31 +403,32 @@ int Log_event::exec_event(struct st_relay_log_info* rli)
     only in the case discussed above, 'if (rli)' is useless here.
     But as we are not 100% sure, keep it for now.
   */
-  if (rli)  
+  if (rli)
   {
     /*
-      If in a transaction, and if the slave supports transactions,
+      If in a transaction, and if the slave supports transactions, just
-      just inc_event_relay_log_pos(). We only have to check for OPTION_BEGIN
+      inc_event_relay_log_pos(). We only have to check for OPTION_BEGIN
-      (not OPTION_NOT_AUTOCOMMIT) as transactions are logged
+      (not OPTION_NOT_AUTOCOMMIT) as transactions are logged with
-      with BEGIN/COMMIT, not with SET AUTOCOMMIT= .
+      BEGIN/COMMIT, not with SET AUTOCOMMIT= .
-      
+
       CAUTION: opt_using_transactions means
-      innodb || bdb ; suppose the master supports InnoDB and BDB, 
+      innodb || bdb ; suppose the master supports InnoDB and BDB,
       but the slave supports only BDB, problems
-      will arise: 
+      will arise:
       - suppose an InnoDB table is created on the master,
       - then it will be MyISAM on the slave
-      - but as opt_using_transactions is true, the slave will believe he is
+      - but as opt_using_transactions is true, the slave will believe he
-      transactional with the MyISAM table. And problems will come when one
+      is transactional with the MyISAM table. And problems will come
-      does START SLAVE; STOP SLAVE; START SLAVE; (the slave will resume at
+      when one does START SLAVE; STOP SLAVE; START SLAVE; (the slave
-      BEGIN whereas there has not been any rollback). This is the problem of
+      will resume at BEGIN whereas there has not been any rollback).
-      using opt_using_transactions instead of a finer
+      This is the problem of using opt_using_transactions instead of a
-      "does the slave support _the_transactional_handler_used_on_the_master_".
+      finer "does the slave support
-      
+      _the_transactional_handler_used_on_the_master_".
-      More generally, we'll have problems when a query mixes a transactional
+
-      handler and MyISAM and STOP SLAVE is issued in the middle of the
+      More generally, we'll have problems when a query mixes a
-      "transaction". START SLAVE will resume at BEGIN while the MyISAM table
+      transactional handler and MyISAM and STOP SLAVE is issued in the
-      has already been updated.
+      middle of the "transaction". START SLAVE will resume at BEGIN
+      while the MyISAM table has already been updated.
     */
     if ((thd->options & OPTION_BEGIN) && opt_using_transactions)
       rli->inc_event_relay_log_pos();
@@ -435,8 +437,8 @@ int Log_event::exec_event(struct st_relay_log_info* rli)
       rli->inc_group_relay_log_pos(log_pos);
       flush_relay_log_info(rli);
       /* 
-         Note that Rotate_log_event::exec_event() does not call this function,
+         Note that Rotate_log_event::exec_event() does not call this
-         so there is no chance that a fake rotate event resets
+         function, so there is no chance that a fake rotate event resets
          last_master_timestamp.
          Note that we update without mutex (probably ok - except in some very
          rare cases, only consequence is that value may take some time to
@@ -649,11 +651,9 @@ end:
 #ifndef MYSQL_CLIENT
 #define UNLOCK_MUTEX if (log_lock) pthread_mutex_unlock(log_lock);
 #define LOCK_MUTEX if (log_lock) pthread_mutex_lock(log_lock);
-#define max_allowed_packet current_thd->variables.max_allowed_packet
 #else
 #define UNLOCK_MUTEX
 #define LOCK_MUTEX
-#define max_allowed_packet (*mysql_get_parameters()->p_max_allowed_packet)
 #endif
 
 /*
@@ -670,16 +670,17 @@ Log_event* Log_event::read_log_event(IO_CACHE* file,
 #else
 Log_event* Log_event::read_log_event(IO_CACHE* file,
                                      const Format_description_log_event *description_event)
-#endif  
+#endif
 {
+  DBUG_ENTER("Log_event::read_log_event(IO_CACHE *, Format_description_log_event *");
   DBUG_ASSERT(description_event);
   char head[LOG_EVENT_MINIMAL_HEADER_LEN];
   /*
     First we only want to read at most LOG_EVENT_MINIMAL_HEADER_LEN, just to
     check the event for sanity and to know its length; no need to really parse
     it. We say "at most" because this could be a 3.23 master, which has header
-    of 13 bytes, whereas LOG_EVENT_MINIMAL_HEADER_LEN is 19 bytes (it's "minimal"
+    of 13 bytes, whereas LOG_EVENT_MINIMAL_HEADER_LEN is 19 bytes (it's
-    over the set {MySQL >=4.0}).
+    "minimal" over the set {MySQL >=4.0}).
   */
   uint header_size= min(description_event->common_header_len,
                         LOG_EVENT_MINIMAL_HEADER_LEN);
@@ -692,17 +693,21 @@ Log_event* Log_event::read_log_event(IO_CACHE* file,
 failed my_b_read"));
     UNLOCK_MUTEX;
     /*
-      No error here; it could be that we are at the file's end. However if the
+      No error here; it could be that we are at the file's end. However
-      next my_b_read() fails (below), it will be an error as we were able to
+      if the next my_b_read() fails (below), it will be an error as we
-      read the first bytes.
+      were able to read the first bytes.
     */
-    return 0;
+    DBUG_RETURN(0);
   }
 
   uint data_len = uint4korr(head + EVENT_LEN_OFFSET);
   char *buf= 0;
   const char *error= 0;
   Log_event *res=  0;
+#ifndef max_allowed_packet
+  THD *thd=current_thd;
+  uint max_allowed_packet= thd ? thd->variables.max_allowed_packet : ~0;
+#endif
 
   if (data_len > max_allowed_packet)
   {
@@ -729,16 +734,16 @@ failed my_b_read"));
     error = "read error";
     goto err;
   }
-  if ((res= read_log_event(buf, data_len, &error,
+  if ((res= read_log_event(buf, data_len, &error, description_event)))
-                           description_event))) 
     res->register_temp_buf(buf);
 
 err:
   UNLOCK_MUTEX;
-  if (error)
+  if (!res)
   {
-    sql_print_error("\
+    DBUG_ASSERT(error);
-Error in Log_event::read_log_event(): '%s', data_len: %d, event_type: %d",
+    sql_print_error("Error in Log_event::read_log_event(): "
+                    "'%s', data_len: %d, event_type: %d",
 		    error,data_len,head[EVENT_TYPE_OFFSET]);
     my_free(buf, MYF(MY_ALLOW_ZERO_PTR));
     /*
@@ -751,7 +756,7 @@ Error in Log_event::read_log_event(): '%s', data_len: %d, event_type: %d",
     */
     file->error= -1;
   }
-  return res;
+  DBUG_RETURN(res);
 }
 
 
@@ -775,7 +780,7 @@ Log_event* Log_event::read_log_event(const char* buf, uint event_len,
     *error="Sanity check failed";		// Needed to free buffer
     DBUG_RETURN(NULL); // general sanity check - will fail on a partial read
   }
-  
+
   switch(buf[EVENT_TYPE_OFFSET]) {
   case QUERY_EVENT:
     ev  = new Query_log_event(buf, event_len, description_event);
@@ -809,14 +814,15 @@ Log_event* Log_event::read_log_event(const char* buf, uint event_len,
   case START_EVENT_V3: /* this is sent only by MySQL <=4.x */
     ev = new Start_log_event_v3(buf, description_event);
     break;
-#ifdef HAVE_REPLICATION
   case STOP_EVENT:
     ev = new Stop_log_event(buf, description_event);
     break;
-#endif /* HAVE_REPLICATION */
   case INTVAR_EVENT:
     ev = new Intvar_log_event(buf, description_event);
     break;
+  case XID_EVENT:
+    ev = new Xid_log_event(buf, description_event);
+    break;
   case RAND_EVENT:
     ev = new Rand_log_event(buf, description_event);
     break;
@@ -831,14 +837,15 @@ Log_event* Log_event::read_log_event(const char* buf, uint event_len,
     ev= NULL;
     break;
   }
+
   /*
-    is_valid() are small event-specific sanity tests which are important; for
+    is_valid() are small event-specific sanity tests which are
-    example there are some my_malloc() in constructors
+    important; for example there are some my_malloc() in constructors
-    (e.g. Query_log_event::Query_log_event(char*...)); when these my_malloc()
+    (e.g. Query_log_event::Query_log_event(char*...)); when these
-    fail we can't return an error out of the constructor (because constructor
+    my_malloc() fail we can't return an error out of the constructor
-    is "void") ; so instead we leave the pointer we wanted to allocate
+    (because constructor is "void") ; so instead we leave the pointer we
-    (e.g. 'query') to 0 and we test it in is_valid(). Same for
+    wanted to allocate (e.g. 'query') to 0 and we test it in is_valid().
-    Format_description_log_event, member 'post_header_len'. 
+    Same for Format_description_log_event, member 'post_header_len'.
   */
   if (!ev || !ev->is_valid())
   {
@@ -923,8 +930,8 @@ void Query_log_event::pack_info(Protocol *protocol)
   char *buf, *pos;
   if (!(buf= my_malloc(9 + db_len + q_len, MYF(MY_WME))))
     return;
-  pos= buf;    
+  pos= buf;
-  if (!(flags & LOG_EVENT_SUPPRESS_USE_F) 
+  if (!(flags & LOG_EVENT_SUPPRESS_USE_F)
       && db && db_len)
   {
     pos= strmov(buf, "use `");
@@ -1082,8 +1089,8 @@ bool Query_log_event::write(IO_CACHE* file)
 
   return (write_header(file, event_length) ||
           my_b_safe_write(file, (byte*) buf, (uint) (start-buf)) ||
-  	  my_b_safe_write(file, (db) ? (byte*) db : (byte*)"", db_len + 1) ||
+          my_b_safe_write(file, (db) ? (byte*) db : (byte*)"", db_len + 1) ||
-  	  my_b_safe_write(file, (byte*) query, q_len)) ? 1 : 0;
+          my_b_safe_write(file, (byte*) query, q_len)) ? 1 : 0;
 }
 
 
@@ -1095,7 +1102,7 @@ bool Query_log_event::write(IO_CACHE* file)
 Query_log_event::Query_log_event(THD* thd_arg, const char* query_arg,
 				 ulong query_length, bool using_trans,
 				 bool suppress_use)
-  :Log_event(thd_arg, 
+  :Log_event(thd_arg,
 	     ((thd_arg->tmp_table_used ? LOG_EVENT_THREAD_SPECIFIC_F : 0)
 	      | (suppress_use          ? LOG_EVENT_SUPPRESS_USE_F    : 0)),
 	     using_trans),
@@ -1303,18 +1310,12 @@ void Query_log_event::print(FILE* file, bool short_form,
   my_fwrite(file, (byte*) buff, (uint) (end-buff),MYF(MY_NABP | MY_WME));
   if (flags & LOG_EVENT_THREAD_SPECIFIC_F)
     fprintf(file,"SET @@session.pseudo_thread_id=%lu;\n",(ulong)thread_id);
-  /*
-    Now the session variables;
-    it's more efficient to pass SQL_MODE as a number instead of a
-    comma-separated list.
-    FOREIGN_KEY_CHECKS, SQL_AUTO_IS_NULL, UNIQUE_CHECKS are session-only
-    variables (they have no global version; they're not listed in sql_class.h),
-    The tests below work for pure binlogs or pure relay logs. Won't work for
-    mixed relay logs but we don't create mixed relay logs (that is, there is no
-    relay log with a format change except within the 3 first events, which
-    mysqlbinlog handles gracefully). So this code should always be good.
-  */
 
+  /*
+    If flags2_inited==0, this is an event from 3.23 or 4.0; nothing to
+    print (remember we don't produce mixed relay logs so there cannot be
+    5.0 events before that one so there is nothing to reset).
+  */
   if (likely(flags2_inited)) /* likely as this will mainly read 5.0 logs */
   {
     /* tmp is a bitmask of bits which have changed. */
@@ -1343,9 +1344,16 @@ void Query_log_event::print(FILE* file, bool short_form,
   }
 
   /*
-    If flags2_inited==0, this is an event from 3.23 or 4.0; nothing to print
+    Now the session variables;
-    (remember we don't produce mixed relay logs so there cannot be 5.0 events
+    it's more efficient to pass SQL_MODE as a number instead of a
-    before that one so there is nothing to reset).
+    comma-separated list.
+    FOREIGN_KEY_CHECKS, SQL_AUTO_IS_NULL, UNIQUE_CHECKS are session-only
+    variables (they have no global version; they're not listed in
+    sql_class.h), The tests below work for pure binlogs or pure relay
+    logs. Won't work for mixed relay logs but we don't create mixed
+    relay logs (that is, there is no relay log with a format change
+    except within the 3 first events, which mysqlbinlog handles
+    gracefully). So this code should always be good.
   */
 
   if (likely(sql_mode_inited))
@@ -1439,7 +1447,7 @@ int Query_log_event::exec_event(struct st_relay_log_info* rli)
     thd->query_length= q_len;
     thd->query = (char*)query;
     VOID(pthread_mutex_lock(&LOCK_thread_count));
-    thd->query_id = query_id++;
+    thd->query_id = next_query_id();
     VOID(pthread_mutex_unlock(&LOCK_thread_count));
     thd->variables.pseudo_thread_id= thread_id;		// for temp tables
     mysql_log.write(thd,COM_QUERY,"%s",thd->query);
@@ -1587,7 +1595,7 @@ Default database: '%s'. Query: '%s'",
       We may also want an option to tell the slave to ignore "affected"
       mismatch. This mismatch could be implemented with a new ER_ code, and
       to ignore it you would use --slave-skip-errors...
-        
+
       To do the comparison we need to know the value of "affected" which the
       above mysql_parse() computed. And we need to know the value of
       "affected" in the master's binlog. Both will be implemented later. The
@@ -1674,14 +1682,20 @@ void Start_log_event_v3::print(FILE* file, bool short_form, LAST_EVENT_INFO* las
       fprintf(file," at startup");
     fputc('\n', file);
   }
+  if (!artificial_event && created)
+  {
 #ifdef WHEN_WE_HAVE_THE_RESET_CONNECTION_SQL_COMMAND
-  /*
+    /*
-    This is for mysqlbinlog: like in replication, we want to delete the stale
+      This is for mysqlbinlog: like in replication, we want to delete the stale
-    tmp files left by an unclean shutdown of mysqld (temporary tables). Probably
+      tmp files left by an unclean shutdown of mysqld (temporary tables)
-    this can be done with RESET CONNECTION (syntax to be defined).
+      and rollback unfinished transaction.
-  */
+      Probably this can be done with RESET CONNECTION (syntax to be defined).
-  fprintf(file,"RESET CONNECTION;\n");
+    */
+    fprintf(file,"RESET CONNECTION;\n");
+#else
+    fprintf(file,"ROLLBACK;\n");
 #endif
+  }
   fflush(file);
 }
 #endif /* MYSQL_CLIENT */
@@ -1757,34 +1771,14 @@ int Start_log_event_v3::exec_event(struct st_relay_log_info* rli)
       close_temporary_tables(thd);
       cleanup_load_tmpdir();
     }
-    /*
-      As a transaction NEVER spans on 2 or more binlogs:
-      if we have an active transaction at this point, the master died while
-      writing the transaction to the binary log, i.e. while flushing the binlog
-      cache to the binlog. As the write was started, the transaction had been
-      committed on the master, so we lack of information to replay this
-      transaction on the slave; all we can do is stop with error.
-      Note: this event could be sent by the master to inform us of the format
-      of its binlog; in other words maybe it is not at its original place when
-      it comes to us; we'll know this by checking log_pos ("artificial" events
-      have log_pos == 0).
-    */
-    if (!artificial_event && (thd->options & OPTION_BEGIN))
-    {
-      slave_print_error(rli, 0, "\
-Rolling back unfinished transaction (no COMMIT or ROLLBACK) from relay log. \
-A probable cause is that the master died while writing the transaction to its \
-binary log.");
-      return(1);
-    }
     break;
 
-    /* 
+    /*
        Now the older formats; in that case load_tmpdir is cleaned up by the I/O
        thread.
     */
   case 1:
-    if (strncmp(rli->relay_log.description_event_for_exec->server_version, 
+    if (strncmp(rli->relay_log.description_event_for_exec->server_version,
                 "3.23.57",7) >= 0 && created)
     {
       /*
@@ -1816,7 +1810,7 @@ binary log.");
 
   SYNOPSIS
     Format_description_log_event::Format_description_log_event
-      binlog_version          	  the binlog version for which we want to build
+      binlog_version              the binlog version for which we want to build
                                   an event. Can be 1 (=MySQL 3.23), 3 (=4.0.x
                                   x>=2 and 4.1) or 4 (MySQL 5.0). Note that the
                                   old 4.0 (binlog version 2) is not supported;
@@ -1836,8 +1830,7 @@ binary log.");
 */
 
 Format_description_log_event::
-Format_description_log_event(uint8 binlog_ver,
+Format_description_log_event(uint8 binlog_ver, const char* server_ver)
-                             const char* server_ver) 
   :Start_log_event_v3()
 {
   created= when;
@@ -1849,7 +1842,7 @@ Format_description_log_event(uint8 binlog_ver,
     number_of_event_types= LOG_EVENT_TYPES;
     /* we'll catch my_malloc() error in is_valid() */
     post_header_len=(uint8*) my_malloc(number_of_event_types*sizeof(uint8),
-                                       MYF(0)); 
+                                       MYF(MY_ZEROFILL));
     /*
       This long list of assignments is not beautiful, but I see no way to
       make it nicer, as the right members are #defines, not array members, so
@@ -1859,18 +1852,13 @@ Format_description_log_event(uint8 binlog_ver,
     {
       post_header_len[START_EVENT_V3-1]= START_V3_HEADER_LEN;
       post_header_len[QUERY_EVENT-1]= QUERY_HEADER_LEN;
-      post_header_len[STOP_EVENT-1]= 0;
       post_header_len[ROTATE_EVENT-1]= ROTATE_HEADER_LEN;
-      post_header_len[INTVAR_EVENT-1]= 0;
       post_header_len[LOAD_EVENT-1]= LOAD_HEADER_LEN;
-      post_header_len[SLAVE_EVENT-1]= 0;
       post_header_len[CREATE_FILE_EVENT-1]= CREATE_FILE_HEADER_LEN;
       post_header_len[APPEND_BLOCK_EVENT-1]= APPEND_BLOCK_HEADER_LEN;
       post_header_len[EXEC_LOAD_EVENT-1]= EXEC_LOAD_HEADER_LEN;
       post_header_len[DELETE_FILE_EVENT-1]= DELETE_FILE_HEADER_LEN;
       post_header_len[NEW_LOAD_EVENT-1]= post_header_len[LOAD_EVENT-1];
-      post_header_len[RAND_EVENT-1]= 0;
-      post_header_len[USER_VAR_EVENT-1]= 0;
       post_header_len[FORMAT_DESCRIPTION_EVENT-1]= FORMAT_DESCRIPTION_HEADER_LEN;
     }
     break;
@@ -1886,7 +1874,7 @@ Format_description_log_event(uint8 binlog_ver,
     else
       strmov(server_version, server_ver ? server_ver : "4.0");
     common_header_len= binlog_ver==1 ? OLD_HEADER_LEN :
-      LOG_EVENT_MINIMAL_HEADER_LEN;  
+      LOG_EVENT_MINIMAL_HEADER_LEN;
     /*
       The first new event in binlog version 4 is Format_desc. So any event type
       after that does not exist in older versions. We use the events known by
@@ -1896,7 +1884,7 @@ Format_description_log_event(uint8 binlog_ver,
     */
     number_of_event_types= FORMAT_DESCRIPTION_EVENT - 1;
     post_header_len=(uint8*) my_malloc(number_of_event_types*sizeof(uint8),
-                                       MYF(0)); 
+                                       MYF(0));
     if (post_header_len)
     {
       post_header_len[START_EVENT_V3-1]= START_V3_HEADER_LEN;
@@ -1927,7 +1915,7 @@ Format_description_log_event(uint8 binlog_ver,
   length different from this version, but we don't know this length as we
   have not read the Format_description_log_event which says it, yet. This
   length is in the post-header of the event, but we don't know where the
-  post-header starts. 
+  post-header starts.
   So this type of event HAS to:
   - either have the header's length at the beginning (in the header, at a
   fixed position which will never be changed), not in the post-header. That
@@ -1943,7 +1931,7 @@ Format_description_log_event(const char* buf,
                              uint event_len,
                              const
                              Format_description_log_event*
-                             description_event) 
+                             description_event)
   :Start_log_event_v3(buf, description_event)
 {
   DBUG_ENTER("Format_description_log_event::Format_description_log_event(char*,...)");
@@ -1953,12 +1941,11 @@ Format_description_log_event(const char* buf,
   number_of_event_types=
     event_len-(LOG_EVENT_MINIMAL_HEADER_LEN+ST_COMMON_HEADER_LEN_OFFSET+1);
   DBUG_PRINT("info", ("common_header_len=%d number_of_event_types=%d",
-                      common_header_len, number_of_event_types)); 
+                      common_header_len, number_of_event_types));
   /* If alloc fails, we'll detect it in is_valid() */
   post_header_len= (uint8*) my_memdup((byte*)buf+ST_COMMON_HEADER_LEN_OFFSET+1,
                                       number_of_event_types*
-                                      sizeof(*post_header_len),
+                                      sizeof(*post_header_len), MYF(0));
-                                      MYF(0));  
   DBUG_VOID_RETURN;
 }
 
@@ -1973,17 +1960,17 @@ bool Format_description_log_event::write(IO_CACHE* file)
   int2store(buff + ST_BINLOG_VER_OFFSET,binlog_version);
   memcpy((char*) buff + ST_SERVER_VER_OFFSET,server_version,ST_SERVER_VER_LEN);
   int4store(buff + ST_CREATED_OFFSET,created);
-  buff[ST_COMMON_HEADER_LEN_OFFSET]= LOG_EVENT_HEADER_LEN;    
+  buff[ST_COMMON_HEADER_LEN_OFFSET]= LOG_EVENT_HEADER_LEN;
   memcpy((char*) buff+ST_COMMON_HEADER_LEN_OFFSET+1, (byte*) post_header_len,
          LOG_EVENT_TYPES);
   return (write_header(file, sizeof(buff)) ||
           my_b_safe_write(file, buff, sizeof(buff)));
 }
-  
+
 /*
   SYNOPSIS
     Format_description_log_event::exec_event()
-  
+
   IMPLEMENTATION
     Save the information which describes the binlog's format, to be able to
     read all coming events.
@@ -1994,11 +1981,32 @@ bool Format_description_log_event::write(IO_CACHE* file)
 int Format_description_log_event::exec_event(struct st_relay_log_info* rli)
 {
   DBUG_ENTER("Format_description_log_event::exec_event");
-  
+
-  /* save the information describing this binlog */  
+  /* save the information describing this binlog */
   delete rli->relay_log.description_event_for_exec;
   rli->relay_log.description_event_for_exec= this;
 
+  /*
+    As a transaction NEVER spans on 2 or more binlogs:
+    if we have an active transaction at this point, the master died
+    while writing the transaction to the binary log, i.e. while
+    flushing the binlog cache to the binlog. As the write was started,
+    the transaction had been committed on the master, so we lack of
+    information to replay this transaction on the slave; all we can do
+    is stop with error.
+    Note: this event could be sent by the master to inform us of the
+    format of its binlog; in other words maybe it is not at its
+    original place when it comes to us; we'll know this by checking
+    log_pos ("artificial" events have log_pos == 0).
+  */
+  if (!artificial_event && created && thd->transaction.all.nht)
+  {
+    slave_print_error(rli, 0, "Rolling back unfinished transaction (no "
+                      "COMMIT or ROLLBACK) from relay log. A probable cause "
+                      "is that the master died while writing the transaction "
+                      "to its binary log.");
+    end_trans(thd, ROLLBACK);
+  }
   /*
     If this event comes from ourselves, there is no cleaning task to perform,
     we don't call Start_log_event_v3::exec_event() (this was just to update the
@@ -2027,18 +2035,18 @@ int Format_description_log_event::exec_event(struct st_relay_log_info* rli)
     If the event was not requested by the slave i.e. the master sent it while
     the slave asked for a position >4, the event will make
     rli->group_master_log_pos advance. Say that the slave asked for position
-    1000, and the Format_desc event's end is 95. Then in the beginning of
+    1000, and the Format_desc event's end is 96. Then in the beginning of
-    replication rli->group_master_log_pos will be 0, then 95, then jump to first
+    replication rli->group_master_log_pos will be 0, then 96, then jump to
-    really asked event (which is >95). So this is ok.
+    first really asked event (which is >96). So this is ok.
   */
   DBUG_RETURN(Start_log_event_v3::exec_event(rli));
 }
 #endif
 
   /**************************************************************************
-  	Load_log_event methods
+        Load_log_event methods
    General note about Load_log_event: the binlogging of LOAD DATA INFILE is
-   going to be changed in 5.0 (or maybe in 4.1; not decided yet).
+   going to be changed in 5.0 (or maybe in 5.1; not decided yet).
    However, the 5.0 slave could still have to read such events (from a 4.x
    master), convert them (which just means maybe expand the header, when 5.0
    servers have a UID in events) (remember that whatever is after the header
@@ -2062,7 +2070,7 @@ void Load_log_event::pack_info(Protocol *protocol)
   char *buf, *pos;
   uint buf_len;
 
-  buf_len= 
+  buf_len=
     5 + db_len + 3 +                        // "use DB; "
     18 + fname_len + 2 +                    // "LOAD DATA INFILE 'file''"
     7 +					    // LOCAL
@@ -2072,8 +2080,8 @@ void Load_log_event::pack_info(Protocol *protocol)
     23 + sql_ex.enclosed_len*4 + 2 +        // " OPTIONALLY ENCLOSED BY 'str'"
     12 + sql_ex.escaped_len*4 + 2 +         // " ESCAPED BY 'str'"
     21 + sql_ex.line_term_len*4 + 2 +       // " FIELDS TERMINATED BY 'str'"
-    19 + sql_ex.line_start_len*4 + 2 +      // " LINES STARTING BY 'str'" 
+    19 + sql_ex.line_start_len*4 + 2 +      // " LINES STARTING BY 'str'"
-    15 + 22 +                               // " IGNORE xxx  LINES" 
+    15 + 22 +                               // " IGNORE xxx  LINES"
     3 + (num_fields-1)*2 + field_block_len; // " (field1, field2, ...)"
 
   if (!(buf= my_malloc(buf_len, MYF(MY_WME))))
@@ -2553,7 +2561,7 @@ int Load_log_event::exec_event(NET* net, struct st_relay_log_info* rli,
   {
     thd->set_time((time_t)when);
     VOID(pthread_mutex_lock(&LOCK_thread_count));
-    thd->query_id = query_id++;
+    thd->query_id = next_query_id();
     VOID(pthread_mutex_unlock(&LOCK_thread_count));
     /*
       Initing thd->row_count is not necessary in theory as this variable has no
@@ -3068,6 +3076,72 @@ int Rand_log_event::exec_event(struct st_relay_log_info* rli)
 #endif /* !MYSQL_CLIENT */
 
 
+/**************************************************************************
+  Xid_log_event methods
+**************************************************************************/
+
+#if defined(HAVE_REPLICATION) && !defined(MYSQL_CLIENT)
+void Xid_log_event::pack_info(Protocol *protocol)
+{
+  char buf[128], *pos;
+  pos= strmov(buf, "COMMIT /* xid=");
+  pos= longlong10_to_str(xid, pos, 10);
+  pos= strmov(pos, " */");
+  protocol->store(buf, (uint) (pos-buf), &my_charset_bin);
+}
+#endif
+
+/*
+  NOTE it's ok not to use int8store here,
+  as long as xid_t::set(ulonglong) and
+  xid_t::get_my_xid doesn't do it either
+
+  we don't care about actual values of xids as long as
+  identical numbers compare identically
+*/
+Xid_log_event::Xid_log_event(const char* buf,
+                             const Format_description_log_event* description_event)
+  :Log_event(buf, description_event)
+{
+  buf+= description_event->common_header_len;
+  xid=*((my_xid *)buf);
+}
+
+
+bool Xid_log_event::write(IO_CACHE* file)
+{
+  return write_header(file, sizeof(xid)) ||
+         my_b_safe_write(file, (byte*) &xid, sizeof(xid));
+}
+
+
+#ifdef MYSQL_CLIENT
+void Xid_log_event::print(FILE* file, bool short_form, LAST_EVENT_INFO* last_event_info)
+{
+  if (!short_form)
+  {
+    char buf[64];
+    longlong10_to_str(xid, buf, 10);
+
+    print_header(file);
+    fprintf(file, "\tXid = %s\n", buf);
+    fflush(file);
+  }
+  fprintf(file, "COMMIT;\n");
+}
+#endif /* MYSQL_CLIENT */
+
+
+#if defined(HAVE_REPLICATION) && !defined(MYSQL_CLIENT)
+int Xid_log_event::exec_event(struct st_relay_log_info* rli)
+{
+  rli->inc_event_relay_log_pos();
+  /* For a slave Xid_log_event is COMMIT */
+  return end_trans(thd, COMMIT);
+}
+#endif /* !MYSQL_CLIENT */
+
+
 /**************************************************************************
   User_var_log_event methods
 **************************************************************************/
@@ -3443,7 +3517,7 @@ Slave_log_event::Slave_log_event(THD* thd_arg,
   DBUG_ENTER("Slave_log_event");
   if (!rli->inited)				// QQ When can this happen ?
     DBUG_VOID_RETURN;
-  
+
   MASTER_INFO* mi = rli->mi;
   // TODO: re-write this better without holding both locks at the same time
   pthread_mutex_lock(&mi->data_lock);
@@ -3577,7 +3651,7 @@ void Stop_log_event::print(FILE* file, bool short_form, LAST_EVENT_INFO* last_ev
 /*
   Stop_log_event::exec_event()
 
-  The master stopped. 
+  The master stopped.
   We used to clean up all temporary tables but this is useless as, as the
   master has shut down properly, it has written all DROP TEMPORARY TABLE and DO
   RELEASE_LOCK (prepared statements' deletion is TODO).

sql/log_event.h

@@ -302,6 +302,31 @@ struct sql_ex_info
 #define LOG_EVENT_TIME_F            0x1
 #define LOG_EVENT_FORCED_ROTATE_F   0x2
 #endif
+
+/*
+   This flag only makes sense for Format_description_log_event. It is set
+   when the event is written, and *reset* when a binlog file is
+   closed (yes, it's the only case when MySQL modifies already written
+   part of binlog).  Thus it is a reliable indicator that binlog was
+   closed correctly.  (Stop_log_event is not enough, there's always a
+   small chance that mysqld crashes in the middle of insert and end of
+   the binlog would look like a Stop_log_event).
+
+   This flag is used to detect a restart after a crash, and to provide
+   "unbreakable" binlog. The problem is that on a crash storage engines
+   rollback automatically, while binlog does not.  To solve this we use this
+   flag and automatically append ROLLBACK to every non-closed binlog (append
+   virtually, on reading, file itself is not changed). If this flag is found,
+   mysqlbinlog simply prints "ROLLBACK" Replication master does not abort on
+   binlog corruption, but takes it as EOF, and replication slave forces a
+   rollback in this case.
+
+   Note, that old binlogs does not have this flag set, so we get a
+   a backward-compatible behaviour.
+*/
+
+#define LOG_EVENT_BINLOG_IN_USE_F       0x1
+
 /*
    If the query depends on the thread (for example: TEMPORARY TABLE).
    Currently this is used by mysqlbinlog to know it must print
@@ -310,26 +335,6 @@ struct sql_ex_info
 */
 #define LOG_EVENT_THREAD_SPECIFIC_F 0x4
 
-/*
-   OPTIONS_WRITTEN_TO_BIN_LOG are the bits of thd->options which must be written
-   to the binlog. OPTIONS_WRITTEN_TO_BINLOG could be written into the
-   Format_description_log_event, so that if later we don't want to replicate a
-   variable we did replicate, or the contrary, it's doable. But it should not be
-   too hard to decide once for all of what we replicate and what we don't, among
-   the fixed 32 bits of thd->options.
-   I (Guilhem) have read through every option's usage, and it looks like
-   OPTION_AUTO_IS_NULL and OPTION_NO_FOREIGN_KEYS are the only ones which alter
-   how the query modifies the table. It's good to replicate
-   OPTION_RELAXED_UNIQUE_CHECKS too because otherwise, the slave may insert data
-   slower than the master, in InnoDB.
-   OPTION_BIG_SELECTS is not needed (the slave thread runs with
-   max_join_size=HA_POS_ERROR) and OPTION_BIG_TABLES is not needed either, as
-   the manual says (because a too big in-memory temp table is automatically
-   written to disk).
-*/
-#define OPTIONS_WRITTEN_TO_BIN_LOG (OPTION_AUTO_IS_NULL | \
-OPTION_NO_FOREIGN_KEY_CHECKS | OPTION_RELAXED_UNIQUE_CHECKS)
-
 /*
   Suppress the generation of 'USE' statements before the actual
   statement. This flag should be set for any events that does not need
@@ -343,24 +348,51 @@ OPTION_NO_FOREIGN_KEY_CHECKS | OPTION_RELAXED_UNIQUE_CHECKS)
  */
 #define LOG_EVENT_SUPPRESS_USE_F    0x8
 
+/*
+   OPTIONS_WRITTEN_TO_BIN_LOG are the bits of thd->options which must be
+   written to the binlog. OPTIONS_WRITTEN_TO_BINLOG could be written
+   into the Format_description_log_event, so that if later we don't want
+   to replicate a variable we did replicate, or the contrary, it's
+   doable. But it should not be too hard to decide once for all of what
+   we replicate and what we don't, among the fixed 32 bits of
+   thd->options.
+   I (Guilhem) have read through every option's usage, and it looks like
+   OPTION_AUTO_IS_NULL and OPTION_NO_FOREIGN_KEYS are the only ones
+   which alter how the query modifies the table. It's good to replicate
+   OPTION_RELAXED_UNIQUE_CHECKS too because otherwise, the slave may
+   insert data slower than the master, in InnoDB.
+   OPTION_BIG_SELECTS is not needed (the slave thread runs with
+   max_join_size=HA_POS_ERROR) and OPTION_BIG_TABLES is not needed
+   either, as the manual says (because a too big in-memory temp table is
+   automatically written to disk).
+*/
+#define OPTIONS_WRITTEN_TO_BIN_LOG (OPTION_AUTO_IS_NULL | \
+OPTION_NO_FOREIGN_KEY_CHECKS | OPTION_RELAXED_UNIQUE_CHECKS)
+
 enum Log_event_type
 {
   /*
     Every time you update this enum (when you add a type), you have to
-    update the code of Format_description_log_event::Format_description_log_event().
+    fix Format_description_log_event::Format_description_log_event().
-    Make sure you always insert new types ***BEFORE*** ENUM_END_EVENT.
   */
   UNKNOWN_EVENT= 0, START_EVENT_V3, QUERY_EVENT, STOP_EVENT, ROTATE_EVENT,
   INTVAR_EVENT, LOAD_EVENT, SLAVE_EVENT, CREATE_FILE_EVENT,
   APPEND_BLOCK_EVENT, EXEC_LOAD_EVENT, DELETE_FILE_EVENT,
   /*
-    NEW_LOAD_EVENT is like LOAD_EVENT except that it has a longer sql_ex,
+    NEW_LOAD_EVENT is like LOAD_EVENT except that it has a longer
-    allowing multibyte TERMINATED BY etc; both types share the same class
+    sql_ex, allowing multibyte TERMINATED BY etc; both types share the
-    (Load_log_event)
+    same class (Load_log_event)
   */
   NEW_LOAD_EVENT,
   RAND_EVENT, USER_VAR_EVENT,
   FORMAT_DESCRIPTION_EVENT,
+  XID_EVENT,
+
+  /*
+    add new events here - right above this comment!
+    existing events should never change their numbers
+  */
+
   ENUM_END_EVENT /* end marker */
 };
 
@@ -461,16 +493,15 @@ public:
   ulong data_written;
 
   /*
-    The master's server id (is preserved in the relay log; used to prevent from
+    The master's server id (is preserved in the relay log; used to
-    infinite loops in circular replication).
+    prevent from infinite loops in circular replication).
   */
   uint32 server_id;
 
   /*
-    Some 16 flags. Only one is really used now; look above for
+    Some 16 flags. Look above for LOG_EVENT_TIME_F,
-    LOG_EVENT_TIME_F, LOG_EVENT_FORCED_ROTATE_F,
+    LOG_EVENT_FORCED_ROTATE_F, LOG_EVENT_THREAD_SPECIFIC_F, and
-    LOG_EVENT_THREAD_SPECIFIC_F, and LOG_EVENT_SUPPRESS_USE_F for
+    LOG_EVENT_SUPPRESS_USE_F for notes.
-    notes.
   */
   uint16 flags;
 
@@ -566,11 +597,11 @@ public:
       temp_buf = 0;
     }
   }
-  virtual int get_data_size() { return 0;}
   /*
     Get event length for simple events. For complicated events the length
     is calculated during write()
   */
+  virtual int get_data_size() { return 0;}
   static Log_event* read_log_event(const char* buf, uint event_len,
 				   const char **error,
                                    const Format_description_log_event
@@ -862,8 +893,8 @@ public:
     binary log) was created.  In the other case (i.e. this event is at
     the start of a binary log created by FLUSH LOGS or automatic
     rotation), 'created' should be 0.  This "trick" is used by MySQL
-    >=4.0.14 slaves to know if they must drop the stale temporary
+    >=4.0.14 slaves to know whether they must drop stale temporary
-    tables or not.
+    tables and whether they should abort unfinished transaction.
 
     Note that when 'created'!=0, it is always equal to the event's
     timestamp; indeed Start_log_event is written only in log.cc where
@@ -1037,6 +1068,40 @@ class Rand_log_event: public Log_event
   bool is_valid() const { return 1; }
 };
 
+/*****************************************************************************
+
+  Xid Log Event class
+
+  Logs xid of the transaction-to-be-committed in the 2pc protocol.
+  Has no meaning in replication, slaves ignore it.
+
+ ****************************************************************************/
+#ifdef MYSQL_CLIENT
+typedef ulong my_xid;
+#endif
+
+class Xid_log_event: public Log_event
+{
+ public:
+   my_xid xid;
+
+#ifndef MYSQL_CLIENT
+  Xid_log_event(THD* thd_arg, my_xid x): Log_event(thd_arg,0,0), xid(x) {}
+#ifdef HAVE_REPLICATION
+  void pack_info(Protocol* protocol);
+  int exec_event(struct st_relay_log_info* rli);
+#endif /* HAVE_REPLICATION */
+#else
+  void print(FILE* file, bool short_form = 0, LAST_EVENT_INFO* last_event_info= 0);
+#endif
+
+  Xid_log_event(const char* buf, const Format_description_log_event* description_event);
+  ~Xid_log_event() {}
+  Log_event_type get_type_code() { return XID_EVENT;}
+  int get_data_size() { return sizeof(xid); }
+  bool write(IO_CACHE* file);
+  bool is_valid() const { return 1; }
+};
 
 /*****************************************************************************
 
@@ -1086,8 +1151,6 @@ public:
   Stop Log Event class
 
  ****************************************************************************/
-#ifdef HAVE_REPLICATION
-
 class Stop_log_event: public Log_event
 {
 public:
@@ -1107,14 +1170,11 @@ public:
   bool is_valid() const { return 1; }
 };
 
-#endif /* HAVE_REPLICATION */
-
-
 /*****************************************************************************
 
   Rotate Log Event class
 
-  This will be depricated when we move to using sequence ids.
+  This will be deprecated when we move to using sequence ids.
 
  ****************************************************************************/
 

sql/mysql_priv.h

@@ -409,9 +409,16 @@ inline THD *_current_thd(void)
 }
 #define current_thd _current_thd()
 
+/*
+  External variables
+*/
+extern ulong server_id, concurrency;
+
+
 typedef my_bool (*qc_engine_callback)(THD *thd, char *table_key,
                                       uint key_length,
                                       ulonglong *engine_data);
+
 #include "sql_string.h"
 #include "sql_list.h"
 #include "sql_map.h"
@@ -449,6 +456,14 @@ bool delete_precheck(THD *thd, TABLE_LIST *tables);
 bool insert_precheck(THD *thd, TABLE_LIST *tables);
 bool create_table_precheck(THD *thd, TABLE_LIST *tables,
                            TABLE_LIST *create_table);
+
+enum enum_mysql_completiontype {
+  ROLLBACK_RELEASE=-2, ROLLBACK=1,  ROLLBACK_AND_CHAIN=7,
+  COMMIT_RELEASE=-1,   COMMIT=0,    COMMIT_AND_CHAIN=6
+};
+
+int end_trans(THD *thd, enum enum_mysql_completiontype completion);
+
 Item *negate_expression(THD *thd, Item *expr);
 #include "sql_class.h"
 #include "sql_acl.h"
@@ -572,6 +587,8 @@ bool mysql_preload_keys(THD* thd, TABLE_LIST* table_list);
 int reassign_keycache_tables(THD* thd, KEY_CACHE *src_cache,
                              KEY_CACHE *dst_cache);
 
+bool mysql_xa_recover(THD *thd);
+
 bool check_simple_select();
 
 SORT_FIELD * make_unireg_sortorder(ORDER *order, uint *length);
@@ -944,11 +961,9 @@ void sql_print_information(const char *format, ...);
 
 bool fn_format_relative_to_data_home(my_string to, const char *name,
 				     const char *dir, const char *extension);
-bool open_log(MYSQL_LOG *log, const char *hostname,
+File open_binlog(IO_CACHE *log, const char *log_file_name,
-	      const char *opt_name, const char *extension,
+                 const char **errmsg);
-	      const char *index_file_name,
+handlerton *binlog_init();
-	      enum_log_type type, bool read_append,
-	      bool no_auto_events, ulong max_size);
 
 /* mysqld.cc */
 extern void yyerror(const char*);
@@ -1003,7 +1018,7 @@ extern double last_query_cost;
 extern double log_10[32];
 extern ulonglong log_10_int[20];
 extern ulonglong keybuff_size;
-extern ulong refresh_version,flush_version, thread_id,query_id;
+extern ulong refresh_version,flush_version, thread_id;
 extern ulong binlog_cache_use, binlog_cache_disk_use;
 extern ulong aborted_threads,aborted_connects;
 extern ulong delayed_insert_timeout;
@@ -1013,8 +1028,6 @@ extern ulong delayed_rows_in_use,delayed_insert_errors;
 extern ulong slave_open_temp_tables;
 extern ulong query_cache_size, query_cache_min_res_unit;
 extern ulong thd_startup_options, slow_launch_threads, slow_launch_time;
-extern ulong server_id, concurrency;
-extern ulong ha_read_count, ha_discover_count;
 extern ulong table_cache_size;
 extern ulong max_connections,max_connect_errors, connect_timeout;
 extern ulong slave_net_timeout;
@@ -1056,6 +1069,7 @@ extern uint opt_large_page_size;
 
 extern MYSQL_LOG mysql_log,mysql_slow_log,mysql_bin_log;
 extern FILE *bootstrap_file;
+extern int bootstrap_error;
 extern pthread_key(MEM_ROOT**,THR_MALLOC);
 extern pthread_mutex_t LOCK_mysql_create_db,LOCK_Acl,LOCK_open,
        LOCK_thread_count,LOCK_mapped_file,LOCK_user_locks, LOCK_status,
@@ -1259,6 +1273,14 @@ SQL_CRYPT *get_crypt_for_frm(void);
 
 #include "sql_view.h"
 
+/* query_id */
+
+typedef ulonglong query_id_t;
+extern query_id_t query_id;
+
+/* increment query_id and return it.  */
+inline query_id_t next_query_id() { return query_id++; }
+
 /* Some inline functions for more speed */
 
 inline bool add_item_to_list(THD *thd, Item *item)

sql/mysqld.cc

@@ -54,7 +54,8 @@
 #endif
 #ifdef HAVE_NDBCLUSTER_DB
 #define OPT_NDBCLUSTER_DEFAULT 0
-#if defined(NOT_ENOUGH_TESTED) && defined(NDB_SHM_TRANSPORTER) && MYSQL_VERSION_ID >= 50000
+#if defined(NOT_ENOUGH_TESTED) \
+  && defined(NDB_SHM_TRANSPORTER) && MYSQL_VERSION_ID >= 50000
 #define OPT_NDB_SHM_DEFAULT 1
 #else
 #define OPT_NDB_SHM_DEFAULT 0
@@ -147,11 +148,11 @@ int deny_severity = LOG_WARNING;
 #include <sys/mman.h>
 #endif
 
+#ifdef __NETWARE__
 #define zVOLSTATE_ACTIVE 6
 #define zVOLSTATE_DEACTIVE 2
 #define zVOLSTATE_MAINTENANCE 3
 
-#ifdef __NETWARE__
 #include <nks/vm.h>
 #include <library.h>
 #include <monitor.h>
@@ -245,6 +246,10 @@ const char *sql_mode_names[] =
 };
 TYPELIB sql_mode_typelib= { array_elements(sql_mode_names)-1,"",
 			    sql_mode_names, NULL };
+const char *tc_heuristic_recover_names[]= { "COMMIT", "ROLLBACK", NullS };
+TYPELIB tc_heuristic_recover_typelib=
+  { array_elements(tc_heuristic_recover_names)-1,"",
+    tc_heuristic_recover_names, NULL };
 const char *first_keyword= "first", *binary_keyword= "BINARY";
 const char *my_localhost= "localhost", *delayed_user= "DELAYED";
 #if SIZEOF_OFF_T > 4 && defined(BIG_TABLES)
@@ -300,14 +305,13 @@ my_bool opt_secure_auth= 0;
 my_bool opt_short_log_format= 0;
 my_bool opt_log_queries_not_using_indexes= 0;
 my_bool lower_case_file_system= 0;
-my_bool opt_innodb_safe_binlog= 0;
 my_bool opt_large_pages= 0;
-uint   opt_large_page_size= 0;
+uint    opt_large_page_size= 0;
 my_bool opt_old_style_user_limits= 0;
 /*
-  True if there is at least one per-hour limit for some user, so we should check
+  True if there is at least one per-hour limit for some user, so we should
-  them before each query (and possibly reset counters when hour is changed).
+  check them before each query (and possibly reset counters when hour is
-  False otherwise.
+  changed). False otherwise.
 */
 volatile bool mqh_used = 0;
 my_bool sp_automatic_privileges= 1;
@@ -315,7 +319,7 @@ my_bool sp_automatic_privileges= 1;
 uint mysqld_port, test_flags, select_errors, dropping_tables, ha_open_options;
 uint delay_key_write_options, protocol_version;
 uint lower_case_table_names;
-uint opt_crash_binlog_innodb;
+uint tc_heuristic_recover= 0;
 uint volatile thread_count, thread_running, kill_cached_threads, wake_thread;
 
 ulong back_log, connect_timeout, concurrency;
@@ -327,7 +331,7 @@ ulong slave_net_timeout;
 ulong thread_cache_size=0, binlog_cache_size=0, max_binlog_cache_size=0;
 ulong query_cache_size=0;
 ulong refresh_version, flush_version;	/* Increments on each reload */
-ulong query_id;
+query_id_t query_id;
 ulong aborted_threads, killed_threads, aborted_connects;
 ulong delayed_insert_timeout, delayed_insert_limit, delayed_queue_size;
 ulong delayed_insert_threads, delayed_insert_writes, delayed_rows_in_use;
@@ -384,6 +388,7 @@ Le_creator le_creator;
 
 
 FILE *bootstrap_file;
+int bootstrap_error;
 
 I_List<i_string_pair> replicate_rewrite_db;
 I_List<i_string> replicate_do_db, replicate_ignore_db;
@@ -445,7 +450,7 @@ static my_bool opt_do_pstack, opt_noacl, opt_bootstrap, opt_myisam_log;
 static int cleanup_done;
 static ulong opt_specialflag, opt_myisam_block_size;
 static char *opt_logname, *opt_update_logname, *opt_binlog_index_name;
-static char *opt_slow_logname;
+static char *opt_slow_logname, *opt_tc_log_file, *opt_tc_heuristic_recover;
 static char *mysql_home_ptr, *pidfile_name_ptr;
 static char **defaults_argv;
 static char *opt_bin_logname;
@@ -527,7 +532,7 @@ static char *get_relative_path(const char *path);
 static void fix_paths(void);
 extern "C" pthread_handler_decl(handle_connections_sockets,arg);
 extern "C" pthread_handler_decl(kill_server_thread,arg);
-static int bootstrap(FILE *file);
+static void bootstrap(FILE *file);
 static void close_server_sock();
 static bool read_init_file(char *file_name);
 #ifdef __NT__
@@ -969,6 +974,8 @@ void clean_up(bool print_message)
     udf_free();
 #endif
   (void) ha_panic(HA_PANIC_CLOSE);	/* close all tables and logs */
+  if (tc_log)
+    tc_log->close();
   delete_elements(&key_caches, (void (*)(const char*, gptr)) free_key_cache);
   multi_keycache_free();
   end_thr_alarm(1);			/* Free allocated memory */
@@ -1576,11 +1583,11 @@ void mysql_down_server_cb(void *, void *)
 
 // destroy callback resources
 void mysql_cb_destroy(void *)
-{  
+{
-  UnRegisterEventNotification(eh);  // cleanup down event notification    	  
+  UnRegisterEventNotification(eh);  // cleanup down event notification
   NX_UNWRAP_INTERFACE(ref);
-  /* Deregister NSS volume deactivation event */  
+  /* Deregister NSS volume deactivation event */
-  NX_UNWRAP_INTERFACE(refneb);  	
+  NX_UNWRAP_INTERFACE(refneb);
   if (neb_consumer_id)
     UnRegisterConsumer(neb_consumer_id, NULL);
 }
@@ -1703,7 +1710,6 @@ ulong neb_event_callback(struct EventBlock *eblock)
       nw_panic = TRUE;
       event_flag= TRUE;
       kill_server(0);
- 
     }
   }
   return 0;
@@ -1761,7 +1767,7 @@ static void getvolumeID(BYTE *volumeName)
   datavolid.clockSeqLow= info.vol.volumeID.clockSeqLow;
   /* This is guranteed to be 6-byte length (but sizeof() would be better) */
   memcpy(datavolid.node, info.vol.volumeID.node, (unsigned int) 6);
-		
+
 exit:
   if (rootKey)
     zClose(rootKey);
@@ -2174,8 +2180,8 @@ static void check_data_home(const char *path)
 
 
 /*
-  All global error messages are sent here where the first one is stored for
+  All global error messages are sent here where the first one is stored
-  the client
+  for the client
 */
 
 
@@ -2211,7 +2217,7 @@ extern "C" int my_message_sql(uint error, const char *str, myf MyFlags)
                            (thd->lex->current_select ?
                             thd->lex->current_select->no_error : 0),
                            (int) thd->is_fatal_error));
-                           
+
       push_warning(thd, MYSQL_ERROR::WARN_LEVEL_ERROR, error, str);
     }
     else
@@ -2301,45 +2307,14 @@ extern "C" pthread_handler_decl(handle_shutdown,arg)
 #endif
 
 
-const char *load_default_groups[]= { 
+const char *load_default_groups[]= {
 #ifdef HAVE_NDBCLUSTER_DB
 "mysql_cluster",
 #endif
-"mysqld","server",MYSQL_BASE_VERSION,0,0};
+"mysqld","server", MYSQL_BASE_VERSION, 0, 0};
 static const int load_default_groups_sz=
 sizeof(load_default_groups)/sizeof(load_default_groups[0]);
 
-bool open_log(MYSQL_LOG *log, const char *hostname,
-	      const char *opt_name, const char *extension,
-	      const char *index_file_name,
-	      enum_log_type type, bool read_append,
-	      bool no_auto_events, ulong max_size)
-{
-  char tmp[FN_REFLEN];
-  if (!opt_name || !opt_name[0])
-  {
-    /*
-      TODO: The following should be using fn_format();  We just need to
-      first change fn_format() to cut the file name if it's too long.
-    */
-    strmake(tmp,hostname,FN_REFLEN-5);
-    strmov(fn_ext(tmp),extension);
-    opt_name=tmp;
-  }
-  // get rid of extension if the log is binary to avoid problems
-  if (type == LOG_BIN)
-  {
-    char *p = fn_ext(opt_name);
-    uint length=(uint) (p-opt_name);
-    strmake(tmp,opt_name,min(length,FN_REFLEN));
-    opt_name=tmp;
-  }
-  return log->open(opt_name, type, 0, index_file_name,
-		   (read_append) ? SEQ_READ_APPEND : WRITE_CACHE,
-		   no_auto_events, max_size, 0);
-}
-
-
 /*
   Initialize one of the global date/time format variables
 
@@ -2347,7 +2322,7 @@ bool open_log(MYSQL_LOG *log, const char *hostname,
     init_global_datetime_format()
     format_type		What kind of format should be supported
     var_ptr		Pointer to variable that should be updated
-  
+
   NOTES
     The default value is taken from either opt_date_time_formats[] or
     the ISO format (ANSI SQL)
@@ -2630,8 +2605,7 @@ static int init_server_components()
 #endif
   /* Setup log files */
   if (opt_log)
-    open_log(&mysql_log, glob_hostname, opt_logname, ".log", NullS,
+    mysql_log.open_query_log(opt_logname);
-	     LOG_NORMAL, 0, 0, 0);
   if (opt_update_log)
   {
     /*
@@ -2684,57 +2658,15 @@ version 5.0 and above. It is replaced by the binary log. Now starting MySQL \
 with --log-bin instead.");
     }
   }
-  if (opt_slow_log)
+  if (opt_log_slave_updates && !opt_bin_log)
-    open_log(&mysql_slow_log, glob_hostname, opt_slow_logname, "-slow.log",
+  {
- 	     NullS, LOG_NORMAL, 0, 0, 0);
+      sql_print_warning("You need to use --log-bin to make "
+                        "--log-slave-updates work.");
+      unireg_abort(1);
+  }
 
-  if (opt_bin_log)
+  if (opt_slow_log)
-  {
+    mysql_slow_log.open_slow_log(opt_slow_logname);
-    /* If we fail to open binlog, it's going to hinder our recovery, so die */
-    if (open_log(&mysql_bin_log, glob_hostname, opt_bin_logname, "-bin",
-		 opt_binlog_index_name, LOG_BIN, 0, 0, max_binlog_size))
-      unireg_abort(1);
-    using_update_log=1;
-#ifdef HAVE_REPLICATION
-    if (expire_logs_days)
-    {
-      long purge_time= time(0) - expire_logs_days*24*60*60;
-      if (purge_time >= 0)
-	mysql_bin_log.purge_logs_before_date(purge_time);
-    }
-#endif
-    if (!opt_bin_logname && !opt_binlog_index_name)
-    {
-      /*
-        User didn't give us info to name the binlog index file.
-        Picking `hostname`-bin.index like did in 4.x, causes replication to
-        fail if the hostname is changed later. So, we would like to instead
-        require a name. But as we don't want to break many existing setups, we
-        only give warning, not error.
-      */
-      sql_print_warning("\
-No argument was provided to --log-bin, and --log-bin-index was not used; \
-so replication may break when this MySQL server acts as a master and \
-has his hostname changed!! Please use '--log-bin=%s' to avoid \
-this problem.",
-                        mysql_bin_log.get_name());
-    }
-  }
-  else
-  {
-    if (opt_log_slave_updates)
-    {
-      sql_print_error("\
-You need to use --log-bin=# to make --log-slave-updates work.");
-      unireg_abort(1);
-    }
-    if (opt_binlog_index_name)
-    {
-      sql_print_error("\
-You need to use --log-bin=# to make --log-bin-index work.");
-      unireg_abort(1);
-    }
-  }
 
 #ifdef HAVE_REPLICATION
   if (opt_log_slave_updates && replicate_same_server_id)
@@ -2766,39 +2698,33 @@ server.");
     }
   }
 
-  if (opt_innodb_safe_binlog)
+  if (opt_bin_log)
   {
-    if (have_innodb != SHOW_OPTION_YES)
+    char buf[FN_REFLEN];
-      sql_print_warning("--innodb-safe-binlog is meaningful only if "
+    const char *ln;
-                      "the InnoDB storage engine is enabled in the server.");
+    ln= mysql_bin_log.generate_name(opt_bin_logname, "-bin", 1, buf);
-#ifdef HAVE_INNOBASE_DB
+    if (!opt_bin_logname && !opt_binlog_index_name)
-    if (innobase_flush_log_at_trx_commit != 1)
-    {
-      sql_print_warning("--innodb-safe-binlog is meaningful only if "
-                      "innodb_flush_log_at_trx_commit is 1; now setting it "
-                      "to 1.");
-      innobase_flush_log_at_trx_commit= 1;
-    }
-    if (innobase_unix_file_flush_method)
     {
       /*
-        This option has so many values that it's hard to know which value is
+        User didn't give us info to name the binlog index file.
-        good (especially "littlesync", and on Windows... see
+        Picking `hostname`-bin.index like did in 4.x, causes replication to
-        srv/srv0start.c).
+        fail if the hostname is changed later. So, we would like to instead
+        require a name. But as we don't want to break many existing setups, we
+        only give warning, not error.
       */
-      sql_print_warning("--innodb-safe-binlog requires that "
+      sql_print_warning("No argument was provided to --log-bin, and "
-                      "the innodb_flush_method actually synchronizes the "
+                        "--log-bin-index was not used; so replication "
-                      "InnoDB log to disk; it is your responsibility "
+                        "may break when this MySQL server acts as a "
-                      "to verify that the method you chose does it.");
+                        "master and has his hostname changed!! Please "
+                        "use '--log-bin=%s' to avoid this problem.", ln);
     }
-    if (sync_binlog_period != 1)
+    if (ln == buf)
     {
-      sql_print_warning("--innodb-safe-binlog is meaningful only if "
+      my_free(opt_bin_logname, MYF(MY_ALLOW_ZERO_PTR));
-                      "the global sync_binlog variable is 1; now setting it "
+      opt_bin_logname=my_strdup(buf, MYF(0));
-                      "to 1.");
-      sync_binlog_period= 1;
     }
-#endif
+    mysql_bin_log.open_index_file(opt_binlog_index_name, ln);
+    using_update_log=1;
   }
 
   if (ha_init())
@@ -2806,21 +2732,33 @@ server.");
     sql_print_error("Can't init databases");
     unireg_abort(1);
   }
+  tc_log= total_ha_2pc > 1 ? opt_bin_log  ?
+                 (TC_LOG *)&mysql_bin_log :
+                 (TC_LOG *)&tc_log_mmap   :
+                 (TC_LOG *)&tc_log_dummy;
+
+  if (tc_log->open(opt_tc_log_file))
+  {
+    sql_print_error("Can't init tc log");
+    unireg_abort(1);
+  }
+
+  if (opt_bin_log && mysql_bin_log.open(opt_bin_logname, LOG_BIN, 0,
+                                        WRITE_CACHE, 0, max_binlog_size, 0))
+      unireg_abort(1);
+
+#ifdef HAVE_REPLICATION
+  if (opt_bin_log && expire_logs_days)
+  {
+    long purge_time= time(0) - expire_logs_days*24*60*60;
+    if (purge_time >= 0)
+      mysql_bin_log.purge_logs_before_date(purge_time);
+  }
+#endif
+
   if (opt_myisam_log)
     (void) mi_log(1);
 
-  /*
-    Now that InnoDB is initialized, we can know the last good binlog position
-    and cut the binlog if needed. This function does nothing if there was no
-    crash recovery by InnoDB.
-  */
-  if (opt_innodb_safe_binlog)
-  {
-    /* not fatal if fails (but print errors) */
-    mysql_bin_log.cut_spurious_tail();
-  }
-  mysql_bin_log.report_pos_in_innodb();
-
   /* call ha_init_key_cache() on all key caches to init them */
   process_key_caches(&ha_init_key_cache);
 
@@ -3137,9 +3075,9 @@ we force server id to 2, but this MySQL server will not act as a slave.");
 
   if (opt_bootstrap)
   {
-    int error=bootstrap(stdin);
+    bootstrap(stdin);
     end_thr_alarm(1);				// Don't allow alarms
-    unireg_abort(error ? 1 : 0);
+    unireg_abort(bootstrap_error ? 1 : 0);
   }
   if (opt_init_file)
   {
@@ -3327,10 +3265,10 @@ int main(int argc, char **argv)
   /* When several instances are running on the same machine, we
      need to have an  unique  named  hEventShudown  through the
      application PID e.g.: MySQLShutdown1890; MySQLShutdown2342
-  */ 
+  */
   int10_to_str((int) GetCurrentProcessId(),strmov(shutdown_event_name,
                                                   "MySQLShutdown"), 10);
-  
+
   /* Must be initialized early for comparison of service name */
   system_charset_info= &my_charset_utf8_general_ci;
 
@@ -3429,7 +3367,7 @@ int main(int argc, char **argv)
   create MySQL privilege tables without having to start a full MySQL server.
 */
 
-static int bootstrap(FILE *file)
+static void bootstrap(FILE *file)
 {
   int error= 0;
   DBUG_ENTER("bootstrap");
@@ -3449,7 +3387,8 @@ static int bootstrap(FILE *file)
 		     (void*) thd))
   {
     sql_print_warning("Can't create thread to handle bootstrap");
-    DBUG_RETURN(-1);
+    bootstrap_error=-1;
+    DBUG_VOID_RETURN;
   }
   /* Wait for thread to die */
   (void) pthread_mutex_lock(&LOCK_thread_count);
@@ -3464,13 +3403,7 @@ static int bootstrap(FILE *file)
   handle_bootstrap((void *)thd);
 #endif
 
-  error= thd->is_fatal_error;
+  DBUG_VOID_RETURN;
-#ifndef EMBEDDED_LIBRARY
-  net_end(&thd->net);
-#endif
-  thd->cleanup();
-  delete thd;
-  DBUG_RETURN(error);
 }
 
 
@@ -3481,7 +3414,7 @@ static bool read_init_file(char *file_name)
   DBUG_PRINT("enter",("name: %s",file_name));
   if (!(file=my_fopen(file_name,O_RDONLY,MYF(MY_WME))))
     return(1);
-  bootstrap(file);				/* Ignore errors from this */
+  bootstrap(file);
   (void) my_fclose(file,MYF(MY_WME));
   return 0;
 }
@@ -3960,7 +3893,7 @@ pthread_handler_decl(handle_connections_shared_memory,arg)
     /*
        it can be after shutdown command
     */
-    if (abort_loop) 
+    if (abort_loop)
       goto error;
 
     HANDLE handle_client_file_map= 0;
@@ -4140,7 +4073,7 @@ enum options_mysqld
   OPT_MASTER_HOST,             OPT_MASTER_USER,
   OPT_MASTER_PASSWORD,         OPT_MASTER_PORT,
   OPT_MASTER_INFO_FILE,        OPT_MASTER_CONNECT_RETRY,
-  OPT_MASTER_RETRY_COUNT,
+  OPT_MASTER_RETRY_COUNT,      OPT_LOG_TC, OPT_LOG_TC_SIZE,
   OPT_MASTER_SSL,              OPT_MASTER_SSL_KEY,
   OPT_MASTER_SSL_CERT,         OPT_MASTER_SSL_CAPATH,
   OPT_MASTER_SSL_CIPHER,       OPT_MASTER_SSL_CA,
@@ -4154,7 +4087,7 @@ enum options_mysqld
   OPT_SAFEMALLOC_MEM_LIMIT,    OPT_REPLICATE_DO_TABLE,
   OPT_REPLICATE_IGNORE_TABLE,  OPT_REPLICATE_WILD_DO_TABLE,
   OPT_REPLICATE_WILD_IGNORE_TABLE, OPT_REPLICATE_SAME_SERVER_ID,
-  OPT_DISCONNECT_SLAVE_EVENT_COUNT,
+  OPT_DISCONNECT_SLAVE_EVENT_COUNT, OPT_TC_HEURISTIC_RECOVER,
   OPT_ABORT_SLAVE_EVENT_COUNT,
   OPT_INNODB_DATA_HOME_DIR,
   OPT_INNODB_DATA_FILE_PATH,
@@ -4577,6 +4510,14 @@ Disable with --skip-isam.",
    "Log slow queries to this log file. Defaults logging to hostname-slow.log file.",
    (gptr*) &opt_slow_logname, (gptr*) &opt_slow_logname, 0, GET_STR, OPT_ARG,
    0, 0, 0, 0, 0, 0},
+  {"log-tc", OPT_LOG_TC,
+   "Path to transaction coordinator log (used for transactions that affect "
+   "more than one storage engine, when binary log is disabled)",
+   (gptr*) &opt_tc_log_file, (gptr*) &opt_tc_log_file, 0, GET_STR,
+   REQUIRED_ARG, 0, 0, 0, 0, 0, 0},
+  {"log-tc-size", OPT_LOG_TC_SIZE, "Size of transaction coordinator log.",
+   (gptr*) &opt_tc_log_size, (gptr*) &opt_tc_log_size, 0, GET_ULONG,
+   REQUIRED_ARG, TC_LOG_MIN_SIZE, TC_LOG_MIN_SIZE, ~0, 0, TC_LOG_PAGE_SIZE, 0},
   {"log-update", OPT_UPDATE_LOG,
    "The update log is deprecated since version 5.0, is replaced by the binary \
 log and this option justs turns on --log-bin instead.",
@@ -4912,6 +4853,10 @@ log and this option does nothing anymore.",
   {"symbolic-links", 's', "Enable symbolic link support.",
    (gptr*) &my_use_symdir, (gptr*) &my_use_symdir, 0, GET_BOOL, NO_ARG,
    IF_PURIFY(0,1), 0, 0, 0, 0, 0},
+  {"tc-heuristic-recover", OPT_TC_HEURISTIC_RECOVER,
+   "Decision to use in heuristic recover process. Possible values are COMMIT or ROLLBACK",
+   (gptr*) &opt_tc_heuristic_recover, (gptr*) &opt_tc_heuristic_recover,
+   0, GET_STR, REQUIRED_ARG, 0, 0, 0, 0, 0, 0},
   {"temp-pool", OPT_TEMP_POOL,
    "Using this option will cause most temporary files created to use a small set of names, rather than a unique name for each new file.",
    (gptr*) &use_temp_pool, (gptr*) &use_temp_pool, 0, GET_BOOL, NO_ARG, 1,
@@ -4982,12 +4927,6 @@ log and this option does nothing anymore.",
    "The number of seconds the mysqld server is waiting for a connect packet before responding with 'Bad handshake'.",
     (gptr*) &connect_timeout, (gptr*) &connect_timeout,
    0, GET_ULONG, REQUIRED_ARG, CONNECT_TIMEOUT, 2, LONG_TIMEOUT, 0, 1, 0 },
-#ifdef HAVE_REPLICATION
-  {"crash_binlog_innodb", OPT_CRASH_BINLOG_INNODB,
-   "Used only for testing, to crash when writing Nth event to binlog.",
-   (gptr*) &opt_crash_binlog_innodb, (gptr*) &opt_crash_binlog_innodb,
-   0, GET_UINT, REQUIRED_ARG, 0, 0, ~(uint)0, 0, 1, 0},
-#endif
   { "date_format", OPT_DATE_FORMAT,
     "The DATE format (For future).",
     (gptr*) &opt_date_time_formats[MYSQL_TIMESTAMP_DATE],
@@ -5108,26 +5047,6 @@ log and this option does nothing anymore.",
    "How many files at the maximum InnoDB keeps open at the same time.",
    (gptr*) &innobase_open_files, (gptr*) &innobase_open_files, 0,
    GET_LONG, REQUIRED_ARG, 300L, 10L, ~0L, 0, 1L, 0},
-#ifdef HAVE_REPLICATION
-  /*
-    Disabled for the 4.1.3 release. Disabling just this paragraph of code is
-    enough, as then user can't set it to 1 so it will always be ignored in the
-    rest of code.
-  */
-#if MYSQL_VERSION_ID >= 40103
-  /*
-    innodb_safe_binlog is not a variable, just an option. Does not make
-    sense to make it a variable, as it is only used at startup (and so the
-    value would be lost at next startup, so setting it on the fly would have no
-    effect).
-  */
-  {"innodb_safe_binlog", OPT_INNODB_SAFE_BINLOG,
-   "After a crash recovery by InnoDB, truncate the binary log after the last "
-   "not-rolled-back statement/transaction.",
-   (gptr*) &opt_innodb_safe_binlog, (gptr*) &opt_innodb_safe_binlog,
-   0, GET_BOOL, NO_ARG, 0, 0, 1, 0, 1, 0},
-#endif
-#endif
   {"innodb_sync_spin_loops", OPT_INNODB_SYNC_SPIN_LOOPS,
    "Count of spin-loop rounds in InnoDB mutexes",
    (gptr*) &srv_n_spin_wait_rounds,
@@ -5529,12 +5448,9 @@ struct show_var_st status_vars[]= {
   {"Aborted_connects",         (char*) &aborted_connects,       SHOW_LONG},
   {"Binlog_cache_disk_use",    (char*) &binlog_cache_disk_use,  SHOW_LONG},
   {"Binlog_cache_use",         (char*) &binlog_cache_use,       SHOW_LONG},
-  {"Bytes_received",           (char*) offsetof(STATUS_VAR, bytes_received),
+  {"Bytes_received",           (char*) offsetof(STATUS_VAR, bytes_received), SHOW_LONG_STATUS},
-   SHOW_LONG_STATUS},
+  {"Bytes_sent",               (char*) offsetof(STATUS_VAR, bytes_sent), SHOW_LONG_STATUS},
-  {"Bytes_sent",               (char*) offsetof(STATUS_VAR, bytes_sent),
+  {"Com_admin_commands",       (char*) offsetof(STATUS_VAR, com_other), SHOW_LONG_STATUS},
-   SHOW_LONG_STATUS},
-  {"Com_admin_commands",       (char*) offsetof(STATUS_VAR, com_other),
-   SHOW_LONG_STATUS},
   {"Com_alter_db",	       (char*) offsetof(STATUS_VAR, com_stat[(uint) SQLCOM_ALTER_DB]), SHOW_LONG_STATUS},
   {"Com_alter_table",	       (char*) offsetof(STATUS_VAR, com_stat[(uint) SQLCOM_ALTER_TABLE]), SHOW_LONG_STATUS},
   {"Com_analyze",	       (char*) offsetof(STATUS_VAR, com_stat[(uint) SQLCOM_ANALYZE]), SHOW_LONG_STATUS},
@@ -5621,62 +5537,45 @@ struct show_var_st status_vars[]= {
   {"Com_unlock_tables",	       (char*) offsetof(STATUS_VAR, com_stat[(uint) SQLCOM_UNLOCK_TABLES]), SHOW_LONG_STATUS},
   {"Com_update",	       (char*) offsetof(STATUS_VAR, com_stat[(uint) SQLCOM_UPDATE]), SHOW_LONG_STATUS},
   {"Com_update_multi",	       (char*) offsetof(STATUS_VAR, com_stat[(uint) SQLCOM_UPDATE_MULTI]), SHOW_LONG_STATUS},
+  {"Com_xa_commit",            (char*) offsetof(STATUS_VAR, com_stat[(uint) SQLCOM_XA_COMMIT]),SHOW_LONG_STATUS},
+  {"Com_xa_end",               (char*) offsetof(STATUS_VAR, com_stat[(uint) SQLCOM_XA_END]),SHOW_LONG_STATUS},
+  {"Com_xa_prepare",           (char*) offsetof(STATUS_VAR, com_stat[(uint) SQLCOM_XA_PREPARE]),SHOW_LONG_STATUS},
+  {"Com_xa_recover",           (char*) offsetof(STATUS_VAR, com_stat[(uint) SQLCOM_XA_RECOVER]),SHOW_LONG_STATUS},
+  {"Com_xa_rollback",          (char*) offsetof(STATUS_VAR, com_stat[(uint) SQLCOM_XA_ROLLBACK]),SHOW_LONG_STATUS},
+  {"Com_xa_start",             (char*) offsetof(STATUS_VAR, com_stat[(uint) SQLCOM_XA_START]),SHOW_LONG_STATUS},
   {"Connections",              (char*) &thread_id,              SHOW_LONG_CONST},
-  {"Created_tmp_disk_tables",  (char*) offsetof(STATUS_VAR,
+  {"Created_tmp_disk_tables",  (char*) offsetof(STATUS_VAR, created_tmp_disk_tables), SHOW_LONG_STATUS},
-						created_tmp_disk_tables),
-   SHOW_LONG_STATUS},
   {"Created_tmp_files",	       (char*) &my_tmp_file_created,	SHOW_LONG},
-  {"Created_tmp_tables",       (char*) offsetof(STATUS_VAR,
+  {"Created_tmp_tables",       (char*) offsetof(STATUS_VAR, created_tmp_tables), SHOW_LONG_STATUS},
-						created_tmp_tables),
-   SHOW_LONG_STATUS},
   {"Delayed_errors",           (char*) &delayed_insert_errors,  SHOW_LONG},
   {"Delayed_insert_threads",   (char*) &delayed_insert_threads, SHOW_LONG_CONST},
   {"Delayed_writes",           (char*) &delayed_insert_writes,  SHOW_LONG},
   {"Flush_commands",           (char*) &refresh_version,        SHOW_LONG_CONST},
-  {"Handler_commit",           (char*) offsetof(STATUS_VAR, ha_commit_count),
+  {"Handler_commit",           (char*) offsetof(STATUS_VAR, ha_commit_count), SHOW_LONG_STATUS},
-   SHOW_LONG_STATUS},
+  {"Handler_delete",           (char*) offsetof(STATUS_VAR, ha_delete_count), SHOW_LONG_STATUS},
-  {"Handler_delete",           (char*) offsetof(STATUS_VAR, ha_delete_count),
+  {"Handler_discover",         (char*) offsetof(STATUS_VAR, ha_discover_count), SHOW_LONG_STATUS},
-   SHOW_LONG_STATUS},
+  {"Handler_prepare",          (char*) offsetof(STATUS_VAR, ha_prepare_count),  SHOW_LONG_STATUS},
-  {"Handler_discover",         (char*) &ha_discover_count,      SHOW_LONG},
+  {"Handler_read_first",       (char*) offsetof(STATUS_VAR, ha_read_first_count), SHOW_LONG_STATUS},
-  {"Handler_read_first",       (char*) offsetof(STATUS_VAR,
+  {"Handler_read_key",         (char*) offsetof(STATUS_VAR, ha_read_key_count), SHOW_LONG_STATUS},
-						ha_read_first_count),
+  {"Handler_read_next",        (char*) offsetof(STATUS_VAR, ha_read_next_count), SHOW_LONG_STATUS},
-   SHOW_LONG_STATUS},
+  {"Handler_read_prev",        (char*) offsetof(STATUS_VAR, ha_read_prev_count), SHOW_LONG_STATUS},
-  {"Handler_read_key",         (char*) offsetof(STATUS_VAR, ha_read_key_count),
+  {"Handler_read_rnd",         (char*) offsetof(STATUS_VAR, ha_read_rnd_count), SHOW_LONG_STATUS},
-   SHOW_LONG_STATUS},
+  {"Handler_read_rnd_next",    (char*) offsetof(STATUS_VAR, ha_read_rnd_next_count), SHOW_LONG_STATUS},
-  {"Handler_read_next",        (char*) offsetof(STATUS_VAR,
+  {"Handler_rollback",         (char*) offsetof(STATUS_VAR, ha_rollback_count), SHOW_LONG_STATUS},
-						ha_read_next_count),
+  {"Handler_savepoint",        (char*) offsetof(STATUS_VAR, ha_savepoint_count), SHOW_LONG_STATUS},
-   SHOW_LONG_STATUS},
+  {"Handler_savepoint_rollback",(char*) offsetof(STATUS_VAR, ha_savepoint_rollback_count), SHOW_LONG_STATUS},
-  {"Handler_read_prev",        (char*) offsetof(STATUS_VAR,
+  {"Handler_update",           (char*) offsetof(STATUS_VAR, ha_update_count), SHOW_LONG_STATUS},
-						ha_read_prev_count),
+  {"Handler_write",            (char*) offsetof(STATUS_VAR, ha_write_count), SHOW_LONG_STATUS},
-   SHOW_LONG_STATUS},
-  {"Handler_read_rnd",         (char*) offsetof(STATUS_VAR, ha_read_rnd_count),
-   SHOW_LONG_STATUS},
-  {"Handler_read_rnd_next",    (char*) offsetof(STATUS_VAR,
-						ha_read_rnd_next_count),
-   SHOW_LONG_STATUS},
-  {"Handler_rollback",         (char*) offsetof(STATUS_VAR, ha_rollback_count),
-   SHOW_LONG_STATUS},
-  {"Handler_update",           (char*) offsetof(STATUS_VAR, ha_update_count),
-   SHOW_LONG_STATUS},
-  {"Handler_write",            (char*) offsetof(STATUS_VAR, ha_write_count),
-   SHOW_LONG_STATUS},
 #ifdef HAVE_INNOBASE_DB
   {"Innodb_",                  (char*) &innodb_status_variables, SHOW_VARS},
 #endif /*HAVE_INNOBASE_DB*/
-  {"Key_blocks_not_flushed",   (char*) &dflt_key_cache_var.global_blocks_changed,
+  {"Key_blocks_not_flushed",   (char*) &dflt_key_cache_var.global_blocks_changed, SHOW_KEY_CACHE_LONG},
-   SHOW_KEY_CACHE_LONG},
+  {"Key_blocks_unused",        (char*) &dflt_key_cache_var.blocks_unused, SHOW_KEY_CACHE_CONST_LONG},
-  {"Key_blocks_unused",        (char*) &dflt_key_cache_var.blocks_unused,
+  {"Key_blocks_used",          (char*) &dflt_key_cache_var.blocks_used, SHOW_KEY_CACHE_CONST_LONG},
-   SHOW_KEY_CACHE_CONST_LONG},
+  {"Key_read_requests",        (char*) &dflt_key_cache_var.global_cache_r_requests, SHOW_KEY_CACHE_LONG},
-  {"Key_blocks_used",          (char*) &dflt_key_cache_var.blocks_used,
+  {"Key_reads",                (char*) &dflt_key_cache_var.global_cache_read, SHOW_KEY_CACHE_LONG},
-   SHOW_KEY_CACHE_CONST_LONG},
+  {"Key_write_requests",       (char*) &dflt_key_cache_var.global_cache_w_requests, SHOW_KEY_CACHE_LONG},
-  {"Key_read_requests",        (char*) &dflt_key_cache_var.global_cache_r_requests,
+  {"Key_writes",               (char*) &dflt_key_cache_var.global_cache_write, SHOW_KEY_CACHE_LONG},
-   SHOW_KEY_CACHE_LONG},
-  {"Key_reads",                (char*) &dflt_key_cache_var.global_cache_read,
-   SHOW_KEY_CACHE_LONG},
-  {"Key_write_requests",       (char*) &dflt_key_cache_var.global_cache_w_requests,
-   SHOW_KEY_CACHE_LONG},
-  {"Key_writes",               (char*) &dflt_key_cache_var.global_cache_write,
-   SHOW_KEY_CACHE_LONG},
   {"Last_query_cost",          (char*) &last_query_cost,        SHOW_DOUBLE},
   {"Max_used_connections",     (char*) &max_used_connections,  SHOW_LONG},
 #ifdef HAVE_NDBCLUSTER_DB
@@ -5686,53 +5585,32 @@ struct show_var_st status_vars[]= {
   {"Open_files",               (char*) &my_file_opened,         SHOW_LONG_CONST},
   {"Open_streams",             (char*) &my_stream_opened,       SHOW_LONG_CONST},
   {"Open_tables",              (char*) 0,                       SHOW_OPENTABLES},
-  {"Opened_tables",            (char*) offsetof(STATUS_VAR, opened_tables),
+  {"Opened_tables",            (char*) offsetof(STATUS_VAR, opened_tables), SHOW_LONG_STATUS},
-   SHOW_LONG_STATUS},
 #ifdef HAVE_QUERY_CACHE
-  {"Qcache_free_blocks",       (char*) &query_cache.free_memory_blocks,
+  {"Qcache_free_blocks",       (char*) &query_cache.free_memory_blocks, SHOW_LONG_CONST},
-   SHOW_LONG_CONST},
+  {"Qcache_free_memory",       (char*) &query_cache.free_memory, SHOW_LONG_CONST},
-  {"Qcache_free_memory",       (char*) &query_cache.free_memory,
-   SHOW_LONG_CONST},
   {"Qcache_hits",              (char*) &query_cache.hits,       SHOW_LONG},
   {"Qcache_inserts",           (char*) &query_cache.inserts,    SHOW_LONG},
   {"Qcache_lowmem_prunes",     (char*) &query_cache.lowmem_prunes, SHOW_LONG},
   {"Qcache_not_cached",        (char*) &query_cache.refused,    SHOW_LONG},
   {"Qcache_queries_in_cache",  (char*) &query_cache.queries_in_cache, SHOW_LONG_CONST},
-  {"Qcache_total_blocks",      (char*) &query_cache.total_blocks,
+  {"Qcache_total_blocks",      (char*) &query_cache.total_blocks, SHOW_LONG_CONST},
-   SHOW_LONG_CONST},
 #endif /*HAVE_QUERY_CACHE*/
   {"Questions",                (char*) 0,                       SHOW_QUESTION},
   {"Rpl_status",               (char*) 0,                 SHOW_RPL_STATUS},
-  {"Select_full_join",         (char*) offsetof(STATUS_VAR,
+  {"Select_full_join",         (char*) offsetof(STATUS_VAR, select_full_join_count), SHOW_LONG_STATUS},
-						select_full_join_count),
+  {"Select_full_range_join",   (char*) offsetof(STATUS_VAR, select_full_range_join_count), SHOW_LONG_STATUS},
-   SHOW_LONG_STATUS},
+  {"Select_range",             (char*) offsetof(STATUS_VAR, select_range_count), SHOW_LONG_STATUS},
-  {"Select_full_range_join",   (char*) offsetof(STATUS_VAR,
+  {"Select_range_check",       (char*) offsetof(STATUS_VAR, select_range_check_count), SHOW_LONG_STATUS},
-						select_full_range_join_count),
+  {"Select_scan",	       (char*) offsetof(STATUS_VAR, select_scan_count), SHOW_LONG_STATUS},
-   SHOW_LONG_STATUS},
-  {"Select_range",             (char*) offsetof(STATUS_VAR,
-						select_range_count),
-   SHOW_LONG_STATUS},
-  {"Select_range_check",       (char*) offsetof(STATUS_VAR,
-						select_range_check_count),
-   SHOW_LONG_STATUS},
-  {"Select_scan",	       (char*) offsetof(STATUS_VAR, select_scan_count),
-   SHOW_LONG_STATUS},
   {"Slave_open_temp_tables",   (char*) &slave_open_temp_tables, SHOW_LONG},
   {"Slave_running",            (char*) 0, SHOW_SLAVE_RUNNING},
   {"Slow_launch_threads",      (char*) &slow_launch_threads,    SHOW_LONG},
-  {"Slow_queries",             (char*) offsetof(STATUS_VAR, long_query_count),
+  {"Slow_queries",             (char*) offsetof(STATUS_VAR, long_query_count), SHOW_LONG_STATUS},
-   SHOW_LONG_STATUS},
+  {"Sort_merge_passes",	       (char*) offsetof(STATUS_VAR, filesort_merge_passes), SHOW_LONG_STATUS},
-  {"Sort_merge_passes",	       (char*) offsetof(STATUS_VAR,
+  {"Sort_range",	       (char*) offsetof(STATUS_VAR, filesort_range_count), SHOW_LONG_STATUS},
-						filesort_merge_passes),
+  {"Sort_rows",		       (char*) offsetof(STATUS_VAR, filesort_rows), SHOW_LONG_STATUS},
-   SHOW_LONG_STATUS},
+  {"Sort_scan",		       (char*) offsetof(STATUS_VAR, filesort_scan_count), SHOW_LONG_STATUS},
-  {"Sort_range",	       (char*) offsetof(STATUS_VAR,
-						filesort_range_count),
-   SHOW_LONG_STATUS},
-  {"Sort_rows",		       (char*) offsetof(STATUS_VAR, filesort_rows),
-   SHOW_LONG_STATUS},
-  {"Sort_scan",		       (char*) offsetof(STATUS_VAR,
-						filesort_scan_count),
-   SHOW_LONG_STATUS},
 #ifdef HAVE_OPENSSL
   {"Ssl_accept_renegotiates",  (char*) 0, 	SHOW_SSL_CTX_SESS_ACCEPT_RENEGOTIATE},
   {"Ssl_accepts",              (char*) 0,  	SHOW_SSL_CTX_SESS_ACCEPT},
@@ -5760,6 +5638,9 @@ struct show_var_st status_vars[]= {
 #endif /* HAVE_OPENSSL */
   {"Table_locks_immediate",    (char*) &locks_immediate,        SHOW_LONG},
   {"Table_locks_waited",       (char*) &locks_waited,           SHOW_LONG},
+  {"Tc_log_max_pages_used",    (char*) &tc_log_max_pages_used,  SHOW_LONG},
+  {"Tc_log_page_size",         (char*) &tc_log_page_size,       SHOW_LONG},
+  {"Tc_log_page_waits",        (char*) &tc_log_page_waits,      SHOW_LONG},
   {"Threads_cached",           (char*) &cached_thread_count,    SHOW_LONG_CONST},
   {"Threads_connected",        (char*) &thread_count,           SHOW_INT_CONST},
   {"Threads_created",	       (char*) &thread_created,		SHOW_LONG_CONST},
@@ -5852,7 +5733,8 @@ static void mysql_init_variables(void)
   mysql_home[0]= pidfile_name[0]= log_error_file[0]= 0;
   opt_log= opt_update_log= opt_bin_log= opt_slow_log= 0;
   opt_disable_networking= opt_skip_show_db=0;
-  opt_logname= opt_update_logname= opt_binlog_index_name= opt_slow_logname=0;
+  opt_logname= opt_update_logname= opt_binlog_index_name= opt_slow_logname= 0;
+  opt_tc_log_file= "tc.log";      // no hostname in tc_log file name !
   opt_secure_auth= 0;
   opt_bootstrap= opt_myisam_log= 0;
   mqh_used= 0;
@@ -6545,6 +6427,16 @@ get_one_option(int optid, const struct my_option *opt __attribute__((unused)),
     ha_open_options|=HA_OPEN_ABORT_IF_CRASHED;
     break;
   }
+  case OPT_TC_HEURISTIC_RECOVER:
+  {
+    if ((tc_heuristic_recover=find_type(argument,
+                                        &tc_heuristic_recover_typelib, 2)) <=0)
+    {
+      fprintf(stderr, "Unknown option to tc-heuristic-recover: %s\n",argument);
+      exit(1);
+    }
+    break;
+  }
   case OPT_SQL_MODE:
   {
     sql_mode_str= argument;

sql/set_var.h

@@ -779,7 +779,7 @@ public:
   } save_result;
   LEX_STRING base;			/* for structs */
 
-  set_var(enum_var_type type_arg, sys_var *var_arg, LEX_STRING *base_name_arg,
+  set_var(enum_var_type type_arg, sys_var *var_arg, const LEX_STRING *base_name_arg,
 	  Item *value_arg)
     :var(var_arg), type(type_arg), base(*base_name_arg)
   {

sql/share/errmsg.txt

@@ -5306,6 +5306,18 @@ ER_CANNOT_USER
 	eng "Operation %s failed for %.256s"
 	ger "Das Kommando %s scheiterte für %.256s"
 	norwegian-ny "Operation %s failed for '%.256s'"
+ER_XAER_NOTA XAE04
+        eng "XAER_NOTA: Unknown XID"
+ER_XAER_INVAL XAE05
+        eng "XAER_INVAL: Invalid arguments (or unsupported command)"
+ER_XAER_RMFAIL XAE07
+        eng "XAER_RMFAIL: The command cannot be executed in the %.64s state"
+ER_XAER_OUTSIDE XAE09
+        eng "XAER_OUTSIDE: Some work is done outside global transaction"
+ER_XAER_RMERR XAE03
+        eng "XAER_RMERR: Fatal error occurred in the transaction branch - check your data for consistency"
+ER_XA_RBROLLBACK XA100
+        eng "XA_RBROLLBACK: Transaction branch was rolled back"
 ER_NONEXISTING_PROC_GRANT 42000 
 	eng "There is no such grant defined for user '%-.32s' on host '%-.64s' on routine '%-.64s'"
 ER_PROC_AUTO_GRANT_FAIL

sql/slave.cc

@@ -1,15 +1,15 @@
 /* Copyright (C) 2000-2003 MySQL AB
-   
+
    This program is free software; you can redistribute it and/or modify
    it under the terms of the GNU General Public License as published by
    the Free Software Foundation; either version 2 of the License, or
    (at your option) any later version.
-   
+
    This program is distributed in the hope that it will be useful,
    but WITHOUT ANY WARRANTY; without even the implied warranty of
    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
    GNU General Public License for more details.
-   
+
    You should have received a copy of the GNU General Public License
    along with this program; if not, write to the Free Software
    Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA */
@@ -158,7 +158,7 @@ int init_slave()
     sql_print_error("Failed to allocate memory for the master info structure");
     goto err;
   }
-    
+
   if (init_master_info(active_mi,master_info_file,relay_log_info_file,
 		       !master_host, (SLAVE_IO | SLAVE_SQL)))
   {
@@ -551,9 +551,9 @@ int purge_relay_logs(RELAY_LOG_INFO* rli, THD *thd, bool just_reset,
   /*
     Even if rli->inited==0, we still try to empty rli->master_log_* variables.
     Indeed, rli->inited==0 does not imply that they already are empty.
-    It could be that slave's info initialization partly succeeded : 
+    It could be that slave's info initialization partly succeeded :
     for example if relay-log.info existed but *relay-bin*.*
-    have been manually removed, init_relay_log_info reads the old 
+    have been manually removed, init_relay_log_info reads the old
     relay-log.info and fills rli->master_log_*, then init_relay_log_info
     checks for the existence of the relay log, this fails and
     init_relay_log_info leaves rli->inited to 0.
@@ -562,7 +562,7 @@ int purge_relay_logs(RELAY_LOG_INFO* rli, THD *thd, bool just_reset,
     MASTER, the callers of purge_relay_logs, will delete bogus *.info files
     or replace them with correct files), however if the user does SHOW SLAVE
     STATUS before START SLAVE, he will see old, confusing rli->master_log_*.
-    In other words, we reinit rli->master_log_* for SHOW SLAVE STATUS 
+    In other words, we reinit rli->master_log_* for SHOW SLAVE STATUS
     to display fine in any case.
   */
 
@@ -671,11 +671,11 @@ int terminate_slave_thread(THD* thd, pthread_mutex_t* term_lock,
     }
   }
   DBUG_ASSERT(thd != 0);
+  THD_CHECK_SENTRY(thd);
   /*
-    Is is criticate to test if the slave is running. Otherwise, we might
+    Is is critical to test if the slave is running. Otherwise, we might
     be referening freed memory trying to kick it
   */
-  THD_CHECK_SENTRY(thd);
 
   while (*slave_running)			// Should always be true
   {
@@ -1674,7 +1674,8 @@ void end_master_info(MASTER_INFO* mi)
 }
 
 
-int init_relay_log_info(RELAY_LOG_INFO* rli, const char* info_fname)
+static int init_relay_log_info(RELAY_LOG_INFO* rli,
+                               const char* info_fname)
 {
   char fname[FN_REFLEN+128];
   int info_fd;
@@ -1682,7 +1683,7 @@ int init_relay_log_info(RELAY_LOG_INFO* rli, const char* info_fname)
   int error = 0;
   DBUG_ENTER("init_relay_log_info");
 
-  if (rli->inited)				// Set if this function called
+  if (rli->inited)                       // Set if this function called
     DBUG_RETURN(0);
   fn_format(fname, info_fname, mysql_data_home, "", 4+32);
   pthread_mutex_lock(&rli->data_lock);
@@ -1693,23 +1694,10 @@ int init_relay_log_info(RELAY_LOG_INFO* rli, const char* info_fname)
   rli->log_space_limit= relay_log_space_limit;
   rli->log_space_total= 0;
 
-  // TODO: make this work with multi-master
-  if (!opt_relay_logname)
-  {
-    char tmp[FN_REFLEN];
-    /*
-      TODO: The following should be using fn_format();  We just need to
-      first change fn_format() to cut the file name if it's too long.
-    */
-    strmake(tmp,glob_hostname,FN_REFLEN-5);
-    strmov(strcend(tmp,'.'),"-relay-bin");
-    opt_relay_logname=my_strdup(tmp,MYF(MY_WME));
-  }
-
   /*
     The relay log will now be opened, as a SEQ_READ_APPEND IO_CACHE.
-    Note that the I/O thread flushes it to disk after writing every event, in
+    Note that the I/O thread flushes it to disk after writing every
-    flush_master_info(mi, 1).
+    event, in flush_master_info(mi, 1).
   */
 
   /*
@@ -1721,16 +1709,25 @@ int init_relay_log_info(RELAY_LOG_INFO* rli, const char* info_fname)
     switch to using max_binlog_size for the relay log) and update
     rli->relay_log.max_size (and mysql_bin_log.max_size).
   */
-
-  if (open_log(&rli->relay_log, glob_hostname, opt_relay_logname,
-	       "-relay-bin", opt_relaylog_index_name,
-	       LOG_BIN, 1 /* read_append cache */,
-	       0 /* starting from 5.0 we want relay logs to have auto events */,
-               max_relay_log_size ? max_relay_log_size : max_binlog_size))
   {
-    pthread_mutex_unlock(&rli->data_lock);
+    char buf[FN_REFLEN];
-    sql_print_error("Failed in open_log() called from init_relay_log_info()");
+    const char *ln;
-    DBUG_RETURN(1);
+    ln= rli->relay_log.generate_name(opt_relay_logname, "-relay-bin",
+                                     1, buf);
+
+    /*
+      note, that if open() fails, we'll still have index file open
+      but a destructor will take care of that
+    */
+    if (rli->relay_log.open_index_file(opt_relaylog_index_name, ln) ||
+        rli->relay_log.open(ln, LOG_BIN, 0, SEQ_READ_APPEND, 0,
+                            (max_relay_log_size ? max_relay_log_size :
+                            max_binlog_size), 1))
+    {
+      pthread_mutex_unlock(&rli->data_lock);
+      sql_print_error("Failed in open_log() called from init_relay_log_info()");
+      DBUG_RETURN(1);
+    }
   }
 
   /* if file does not exist */
@@ -1993,9 +1990,9 @@ void clear_until_condition(RELAY_LOG_INFO* rli)
 
 
 int init_master_info(MASTER_INFO* mi, const char* master_info_fname,
-		     const char* slave_info_fname,
+                     const char* slave_info_fname,
-		     bool abort_if_no_master_info_file,
+                     bool abort_if_no_master_info_file,
-		     int thread_mask)
+                     int thread_mask)
 {
   int fd,error;
   char fname[FN_REFLEN+128];
@@ -2009,7 +2006,7 @@ int init_master_info(MASTER_INFO* mi, const char* master_info_fname,
       last time. If this case pos_in_file would be set and we would
       get a crash when trying to read the signature for the binary
       relay log.
-      
+
       We only rewind the read position if we are starting the SQL
       thread. The handle_slave_sql thread assumes that the read
       position is at the beginning of the file, and will read the
@@ -2035,7 +2032,7 @@ int init_master_info(MASTER_INFO* mi, const char* master_info_fname,
   fd = mi->fd;
 
   /* does master.info exist ? */
-  
+
   if (access(fname,F_OK))
   {
     if (abort_if_no_master_info_file)
@@ -2071,7 +2068,7 @@ file '%s')", fname);
   {
     if (fd >= 0)
       reinit_io_cache(&mi->file, READ_CACHE, 0L,0,0);
-    else 
+    else
     {
       if ((fd = my_open(fname, O_RDWR|O_BINARY, MYF(MY_WME))) < 0 )
       {
@@ -2091,52 +2088,52 @@ file '%s')", fname);
     mi->fd = fd;
     int port, connect_retry, master_log_pos, ssl= 0, lines;
     char *first_non_digit;
-    
+
     /*
        Starting from 4.1.x master.info has new format. Now its
-       first line contains number of lines in file. By reading this 
+       first line contains number of lines in file. By reading this
-       number we will be always distinguish to which version our 
+       number we will be always distinguish to which version our
-       master.info corresponds to. We can't simply count lines in 
+       master.info corresponds to. We can't simply count lines in
        file since versions before 4.1.x could generate files with more
        lines than needed.
-       If first line doesn't contain a number or contain number less than 
+       If first line doesn't contain a number or contain number less than
        14 then such file is treated like file from pre 4.1.1 version.
-       There is no ambiguity when reading an old master.info, as before 
+       There is no ambiguity when reading an old master.info, as before
        4.1.1, the first line contained the binlog's name, which is either
-       empty or has an extension (contains a '.'), so can't be confused 
+       empty or has an extension (contains a '.'), so can't be confused
        with an integer.
 
-       So we're just reading first line and trying to figure which version 
+       So we're just reading first line and trying to figure which version
        is this.
     */
-    
+
-    /* 
+    /*
-       The first row is temporarily stored in mi->master_log_name, 
+       The first row is temporarily stored in mi->master_log_name,
-       if it is line count and not binlog name (new format) it will be 
+       if it is line count and not binlog name (new format) it will be
        overwritten by the second row later.
     */
     if (init_strvar_from_file(mi->master_log_name,
 			      sizeof(mi->master_log_name), &mi->file,
 			      ""))
       goto errwithmsg;
-    
+
     lines= strtoul(mi->master_log_name, &first_non_digit, 10);
 
-    if (mi->master_log_name[0]!='\0' && 
+    if (mi->master_log_name[0]!='\0' &&
         *first_non_digit=='\0' && lines >= LINES_IN_MASTER_INFO_WITH_SSL)
     {                                          // Seems to be new format
-      if (init_strvar_from_file(mi->master_log_name,     
+      if (init_strvar_from_file(mi->master_log_name,
             sizeof(mi->master_log_name), &mi->file, ""))
         goto errwithmsg;
     }
     else
       lines= 7;
-    
+
     if (init_intvar_from_file(&master_log_pos, &mi->file, 4) ||
 	init_strvar_from_file(mi->host, sizeof(mi->host), &mi->file,
 			      master_host) ||
 	init_strvar_from_file(mi->user, sizeof(mi->user), &mi->file,
-			      master_user) || 
+			      master_user) ||
         init_strvar_from_file(mi->password, SCRAMBLED_PASSWORD_CHAR_LENGTH+1,
                               &mi->file, master_password) ||
 	init_intvar_from_file(&port, &mi->file, master_port) ||
@@ -2144,17 +2141,17 @@ file '%s')", fname);
 			      master_connect_retry))
       goto errwithmsg;
 
-    /* 
+    /*
-       If file has ssl part use it even if we have server without 
+       If file has ssl part use it even if we have server without
-       SSL support. But these option will be ignored later when 
+       SSL support. But these option will be ignored later when
-       slave will try connect to master, so in this case warning 
+       slave will try connect to master, so in this case warning
        is printed.
      */
-    if (lines >= LINES_IN_MASTER_INFO_WITH_SSL && 
+    if (lines >= LINES_IN_MASTER_INFO_WITH_SSL &&
         (init_intvar_from_file(&ssl, &mi->file, master_ssl) ||
-         init_strvar_from_file(mi->ssl_ca, sizeof(mi->ssl_ca), 
+         init_strvar_from_file(mi->ssl_ca, sizeof(mi->ssl_ca),
                                &mi->file, master_ssl_ca) ||
-         init_strvar_from_file(mi->ssl_capath, sizeof(mi->ssl_capath), 
+         init_strvar_from_file(mi->ssl_capath, sizeof(mi->ssl_capath),
                                &mi->file, master_ssl_capath) ||
          init_strvar_from_file(mi->ssl_cert, sizeof(mi->ssl_cert),
                                &mi->file, master_ssl_cert) ||
@@ -2169,7 +2166,7 @@ file '%s')", fname);
                       "('%s') are ignored because this MySQL slave was compiled "
                       "without SSL support.", fname);
 #endif /* HAVE_OPENSSL */
-    
+
     /*
       This has to be handled here as init_intvar_from_file can't handle
       my_off_t types
@@ -2189,15 +2186,15 @@ file '%s')", fname);
 
   mi->inited = 1;
   // now change cache READ -> WRITE - must do this before flush_master_info
-  reinit_io_cache(&mi->file, WRITE_CACHE,0L,0,1);
+  reinit_io_cache(&mi->file, WRITE_CACHE, 0L, 0, 1);
   if ((error=test(flush_master_info(mi, 1))))
     sql_print_error("Failed to flush master info file");
   pthread_mutex_unlock(&mi->data_lock);
   DBUG_RETURN(error);
-  
+
 errwithmsg:
   sql_print_error("Error reading master configuration");
-  
+
 err:
   if (fd >= 0)
   {
@@ -2968,8 +2965,7 @@ static ulong read_event(MYSQL* mysql, MASTER_INFO *mi, bool* suppress_warnings)
       *suppress_warnings= TRUE;
     }
     else
-      sql_print_error("Error reading packet from server: %s (\
+      sql_print_error("Error reading packet from server: %s ( server_errno=%d)",
-server_errno=%d)",
 		      mysql_error(mysql), mysql_errno(mysql));
     return packet_error;
   }
@@ -3115,17 +3111,17 @@ static int exec_relay_log_event(THD* thd, RELAY_LOG_INFO* rli)
 {
   /*
      We acquire this mutex since we need it for all operations except
-     event execution. But we will release it in places where we will 
+     event execution. But we will release it in places where we will
      wait for something for example inside of next_event().
    */
   pthread_mutex_lock(&rli->data_lock);
-  
+
-  if (rli->until_condition!=RELAY_LOG_INFO::UNTIL_NONE && 
+  if (rli->until_condition!=RELAY_LOG_INFO::UNTIL_NONE &&
-      rli->is_until_satisfied()) 
+      rli->is_until_satisfied())
   {
     sql_print_error("Slave SQL thread stopped because it reached its"
                     " UNTIL position %ld", (long) rli->until_pos());
-    /* 
+    /*
       Setting abort_slave flag because we do not want additional message about
       error in query execution to be printed.
     */
@@ -3133,11 +3129,11 @@ static int exec_relay_log_event(THD* thd, RELAY_LOG_INFO* rli)
     pthread_mutex_unlock(&rli->data_lock);
     return 1;
   }
-  
+
   Log_event * ev = next_event(rli);
-  
+
   DBUG_ASSERT(rli->sql_thd==thd);
-  
+
   if (sql_slave_killed(thd,rli))
   {
     pthread_mutex_unlock(&rli->data_lock);
@@ -3162,13 +3158,13 @@ static int exec_relay_log_event(THD* thd, RELAY_LOG_INFO* rli)
       events created by the creation/rotation of the relay log (remember that
       now the relay log starts with its Format_desc, has a Rotate etc).
     */
-    
+
     DBUG_PRINT("info",("type_code=%d, server_id=%d",type_code,ev->server_id));
-    
+
     if ((ev->server_id == (uint32) ::server_id &&
          !replicate_same_server_id &&
          type_code != FORMAT_DESCRIPTION_EVENT) ||
- 	(rli->slave_skip_counter && 
+        (rli->slave_skip_counter &&
          type_code != ROTATE_EVENT && type_code != STOP_EVENT &&
          type_code != START_EVENT_V3 && type_code!= FORMAT_DESCRIPTION_EVENT))
     {
@@ -3177,24 +3173,24 @@ static int exec_relay_log_event(THD* thd, RELAY_LOG_INFO* rli)
         rli->inc_event_relay_log_pos();
       else
       {
-        rli->inc_group_relay_log_pos((type_code == ROTATE_EVENT || 
+        rli->inc_group_relay_log_pos((type_code == ROTATE_EVENT ||
                                       type_code == STOP_EVENT ||
                                       type_code == FORMAT_DESCRIPTION_EVENT) ?
                                      LL(0) : ev->log_pos,
                                      1/* skip lock*/);
         flush_relay_log_info(rli);
       }
-      
+
       /*
- 	Protect against common user error of setting the counter to 1
+        Protect against common user error of setting the counter to 1
- 	instead of 2 while recovering from an insert which used auto_increment,
+        instead of 2 while recovering from an insert which used auto_increment,
- 	rand or user var.
+        rand or user var.
       */
-      if (rli->slave_skip_counter && 
+      if (rli->slave_skip_counter &&
- 	  !((type_code == INTVAR_EVENT ||
+          !((type_code == INTVAR_EVENT ||
-             type_code == RAND_EVENT || 
+             type_code == RAND_EVENT ||
              type_code == USER_VAR_EVENT) &&
- 	    rli->slave_skip_counter == 1) &&
+            rli->slave_skip_counter == 1) &&
           /*
             The events from ourselves which have something to do with the relay
             log itself must be skipped, true, but they mustn't decrement
@@ -3209,11 +3205,11 @@ static int exec_relay_log_event(THD* thd, RELAY_LOG_INFO* rli)
              type_code == START_EVENT_V3 || type_code == FORMAT_DESCRIPTION_EVENT)))
         --rli->slave_skip_counter;
       pthread_mutex_unlock(&rli->data_lock);
-      delete ev;     
+      delete ev;
-      return 0;					// avoid infinite update loops
+      return 0;                                 // avoid infinite update loops
-    } 
+    }
     pthread_mutex_unlock(&rli->data_lock);
-  
+
     thd->server_id = ev->server_id; // use the original server id for logging
     thd->set_time();				// time the query
     thd->lex->current_select= 0;
@@ -3222,7 +3218,7 @@ static int exec_relay_log_event(THD* thd, RELAY_LOG_INFO* rli)
     ev->thd = thd;
     exec_res = ev->exec_event(rli);
     DBUG_ASSERT(rli->sql_thd==thd);
-    /* 
+    /*
        Format_description_log_event should not be deleted because it will be
        used to read info about the relay log's format; it will be deleted when
        the SQL thread does not need it, i.e. when this thread terminates.
@@ -3257,17 +3253,17 @@ extern "C" pthread_handler_decl(handle_slave_io,arg)
 {
   THD *thd; // needs to be first for thread_stack
   MYSQL *mysql;
-  MASTER_INFO *mi = (MASTER_INFO*)arg; 
+  MASTER_INFO *mi = (MASTER_INFO*)arg;
   char llbuff[22];
   uint retry_count;
-  
+
   // needs to call my_thread_init(), otherwise we get a coredump in DBUG_ stuff
   my_thread_init();
   DBUG_ENTER("handle_slave_io");
 
 #ifndef DBUG_OFF
 slave_begin:
-#endif  
+#endif
   DBUG_ASSERT(mi->inited);
   mysql= NULL ;
   retry_count= 0;
@@ -3276,10 +3272,10 @@ slave_begin:
   /* Inform waiting threads that slave has started */
   mi->slave_run_id++;
 
-#ifndef DBUG_OFF  
+#ifndef DBUG_OFF
   mi->events_till_abort = abort_slave_event_count;
-#endif  
+#endif
-  
+
   thd= new THD; // note that contructor of THD uses DBUG_ !
   THD_CHECK_SENTRY(thd);
 
@@ -3300,17 +3296,16 @@ slave_begin:
   mi->abort_slave = 0;
   pthread_mutex_unlock(&mi->run_lock);
   pthread_cond_broadcast(&mi->start_cond);
-  
+
   DBUG_PRINT("master_info",("log_file_name: '%s'  position: %s",
 			    mi->master_log_name,
 			    llstr(mi->master_log_pos,llbuff)));
-  
+
   if (!(mi->mysql = mysql = mysql_init(NULL)))
   {
     sql_print_error("Slave I/O thread: error in mysql_init()");
     goto err;
   }
-  
 
   thd->proc_info = "Connecting to master";
   // we can get killed during safe_connect
@@ -3346,11 +3341,11 @@ connected:
     if (register_slave_on_master(mysql) ||  update_slave_list(mysql, mi))
       goto err;
   }
-  
+
   DBUG_PRINT("info",("Starting reading binary log from master"));
   while (!io_slave_killed(thd,mi))
   {
-    bool suppress_warnings= 0;    
+    bool suppress_warnings= 0;
     thd->proc_info = "Requesting binlog dump";
     if (request_dump(mysql, mi, &suppress_warnings))
     {
@@ -3361,7 +3356,7 @@ connected:
 dump");
 	goto err;
       }
-	  
+
       mi->slave_running= MYSQL_SLAVE_RUN_NOT_CONNECT;
       thd->proc_info= "Waiting to reconnect after a failed binlog dump request";
 #ifdef SIGNAL_WITH_VIO_CLOSE
@@ -3405,12 +3400,12 @@ after reconnect");
 
     while (!io_slave_killed(thd,mi))
     {
-      bool suppress_warnings= 0;    
+      bool suppress_warnings= 0;
-      /* 
+      /*
          We say "waiting" because read_event() will wait if there's nothing to
-         read. But if there's something to read, it will not wait. The important
+         read. But if there's something to read, it will not wait. The
-         thing is to not confuse users by saying "reading" whereas we're in fact
+         important thing is to not confuse users by saying "reading" whereas
-         receiving nothing.
+         we're in fact receiving nothing.
       */
       thd->proc_info = "Waiting for master to send event";
       ulong event_len = read_event(mysql, mi, &suppress_warnings);
@@ -3420,7 +3415,7 @@ after reconnect");
 	  sql_print_information("Slave I/O thread killed while reading event");
 	goto err;
       }
-	  	  
+
       if (event_len == packet_error)
       {
 	uint mysql_error_number= mysql_errno(mysql);
@@ -3451,7 +3446,7 @@ max_allowed_packet",
 	    goto err;				// Don't retry forever
 	  safe_sleep(thd,mi->connect_retry,(CHECK_KILLED_FUNC)io_slave_killed,
 		     (void*) mi);
-	}	    
+	}
 	if (io_slave_killed(thd,mi))
 	{
 	  if (global_system_variables.log_warnings)
@@ -3474,7 +3469,7 @@ reconnect done to recover from failed read");
 	}
 	goto connected;
       } // if (event_len == packet_error)
-	  
+
       retry_count=0;			// ok event, reset retry counter
       thd->proc_info = "Queueing master event to the relay log";
       if (queue_event(mi,(const char*)mysql->net.read_pos + 1,
@@ -3935,6 +3930,7 @@ static int process_io_rotate(MASTER_INFO *mi, Rotate_log_event *rev)
   if (disconnect_slave_event_count)
     events_till_disconnect++;
 #endif
+
   /*
     If description_event_for_queue is format <4, there is conversion in the
     relay log to the slave's format (4). And Rotate can mean upgrade or
@@ -3958,8 +3954,8 @@ static int process_io_rotate(MASTER_INFO *mi, Rotate_log_event *rev)
 }
 
 /*
-  Reads a 3.23 event and converts it to the slave's format. This code was copied
+  Reads a 3.23 event and converts it to the slave's format. This code was
-  from MySQL 4.0.
+  copied from MySQL 4.0.
 */
 static int queue_binlog_ver_1_event(MASTER_INFO *mi, const char *buf,
 			   ulong event_len)
@@ -4222,9 +4218,9 @@ int queue_event(MASTER_INFO* mi,const char* buf, ulong event_len)
       to write this event again).
     */
     /*
-      We are the only thread which reads/writes description_event_for_queue. The
+      We are the only thread which reads/writes description_event_for_queue.
-      relay_log struct does not move (though some members of it can change), so
+      The relay_log struct does not move (though some members of it can
-      we needn't any lock (no rli->data_lock, no log lock).
+      change), so we needn't any lock (no rli->data_lock, no log lock).
     */
     Format_description_log_event* tmp;
     const char* errmsg;
@@ -4615,7 +4611,7 @@ Log_event* next_event(RELAY_LOG_INFO* rli)
       /* This is an assertion which sometimes fails, let's try to track it */
       char llbuf1[22], llbuf2[22];
       DBUG_PRINT("info", ("my_b_tell(cur_log)=%s rli->event_relay_log_pos=%s",
-                          llstr(my_b_tell(cur_log),llbuf1), 
+                          llstr(my_b_tell(cur_log),llbuf1),
                           llstr(rli->event_relay_log_pos,llbuf2)));
       DBUG_ASSERT(my_b_tell(cur_log) >= BIN_LOG_HEADER_SIZE);
       DBUG_ASSERT(my_b_tell(cur_log) == rli->event_relay_log_pos);
@@ -4635,7 +4631,7 @@ Log_event* next_event(RELAY_LOG_INFO* rli)
     */
     if ((ev=Log_event::read_log_event(cur_log,0,
                                       rli->relay_log.description_event_for_exec)))
-      
+
     {
       DBUG_ASSERT(thd==rli->sql_thd);
       /*

sql/slave.h

@@ -1,15 +1,15 @@
 /* Copyright (C) 2000-2003 MySQL AB
-   
+
    This program is free software; you can redistribute it and/or modify
    it under the terms of the GNU General Public License as published by
    the Free Software Foundation; either version 2 of the License, or
    (at your option) any later version.
-   
+
    This program is distributed in the hope that it will be useful,
    but WITHOUT ANY WARRANTY; without even the implied warranty of
    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
    GNU General Public License for more details.
-   
+
    You should have received a copy of the GNU General Public License
    along with this program; if not, write to the Free Software
    Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA */
@@ -392,11 +392,11 @@ typedef struct st_master_info
   my_bool ssl; // enables use of SSL connection if true
   char ssl_ca[FN_REFLEN], ssl_capath[FN_REFLEN], ssl_cert[FN_REFLEN];
   char ssl_cipher[FN_REFLEN], ssl_key[FN_REFLEN];
-  
+
   my_off_t master_log_pos;
   File fd; // we keep the file open, so we need to remember the file pointer
   IO_CACHE file;
-  
+
   pthread_mutex_t data_lock,run_lock;
   pthread_cond_t data_cond,start_cond,stop_cond;
   THD *io_thd;
@@ -412,7 +412,7 @@ typedef struct st_master_info
   volatile bool abort_slave;
   volatile uint slave_running;
   volatile ulong slave_run_id;
-  /* 
+  /*
      The difference in seconds between the clock of the master and the clock of
      the slave (second - first). It must be signed as it may be <0 or >0.
      clock_diff_with_master is computed when the I/O thread starts; for this the
@@ -421,8 +421,8 @@ typedef struct st_master_info
      clock_of_slave - last_timestamp_executed_by_SQL_thread - clock_diff_with_master
 
   */
-  long clock_diff_with_master; 
+  long clock_diff_with_master;
-  
+
   st_master_info()
     :ssl(0), fd(-1),  io_thd(0), inited(0),
      abort_slave(0),slave_running(0), slave_run_id(0)
@@ -430,7 +430,7 @@ typedef struct st_master_info
     host[0] = 0; user[0] = 0; password[0] = 0;
     ssl_ca[0]= 0; ssl_capath[0]= 0; ssl_cert[0]= 0;
     ssl_cipher[0]= 0; ssl_key[0]= 0;
-    
+
     bzero((char*) &file, sizeof(file));
     pthread_mutex_init(&run_lock, MY_MUTEX_INIT_FAST);
     pthread_mutex_init(&data_lock, MY_MUTEX_INIT_FAST);
@@ -550,7 +550,6 @@ int init_master_info(MASTER_INFO* mi, const char* master_info_fname,
 		     bool abort_if_no_master_info_file,
 		     int thread_mask);
 void end_master_info(MASTER_INFO* mi);
-int init_relay_log_info(RELAY_LOG_INFO* rli, const char* info_fname);
 void end_relay_log_info(RELAY_LOG_INFO* rli);
 void lock_slave_threads(MASTER_INFO* mi);
 void unlock_slave_threads(MASTER_INFO* mi);

sql/sp_head.cc

@@ -1243,7 +1243,7 @@ sp_instr_stmt::exec_stmt(THD *thd, LEX *lex)
   thd->free_list= NULL;
 
   VOID(pthread_mutex_lock(&LOCK_thread_count));
-  thd->query_id= query_id++;
+  thd->query_id= next_query_id();
   VOID(pthread_mutex_unlock(&LOCK_thread_count));
 
   reset_stmt_for_execute(thd, lex);

sql/sql_base.cc

@@ -384,8 +384,8 @@ void close_thread_tables(THD *thd, bool lock_in_use, bool skip_derived,
   {
     TABLE *table, *next;
     /*
-      Close all derived tables generated from questions like
+      Close all derived tables generated in queries like
-      SELECT * from (select * from t1))
+      SELECT * FROM (SELECT * FROM t1)
     */
     for (table= thd->derived_tables ; table ; table= next)
     {
@@ -405,6 +405,18 @@ void close_thread_tables(THD *thd, bool lock_in_use, bool skip_derived,
     mysql_unlock_tables(thd, thd->lock);
     thd->lock=0;
   }
+  /*
+    assume handlers auto-commit (if some doesn't - transaction handling
+    in MySQL should be redesigned to support it; it's a big change,
+    and it's not worth it - better to commit explicitly only writing
+    transactions, read-only ones should better take care of themselves.
+    saves some work in 2pc too)
+    see also sql_parse.cc - dispatch_command()
+  */
+  bzero(&thd->transaction.stmt, sizeof(thd->transaction.stmt));
+  if (!thd->active_transaction())
+    thd->transaction.xid.null();
+
   /* VOID(pthread_sigmask(SIG_SETMASK,&thd->block_signals,NULL)); */
   if (!lock_in_use)
     VOID(pthread_mutex_lock(&LOCK_open));
@@ -1824,7 +1836,7 @@ TABLE *open_ltable(THD *thd, TABLE_LIST *table_list, thr_lock_type lock_type)
     -1 - error
 
   NOTE
-    The lock will automaticly be freed by close_thread_tables()
+    The lock will automaticaly be freed by close_thread_tables()
 */
 
 int simple_open_n_lock_tables(THD *thd, TABLE_LIST *tables)
@@ -1851,7 +1863,7 @@ int simple_open_n_lock_tables(THD *thd, TABLE_LIST *tables)
     TRUE  - error
 
   NOTE
-    The lock will automaticly be freed by close_thread_tables()
+    The lock will automaticaly be freed by close_thread_tables()
 */
 
 bool open_and_lock_tables(THD *thd, TABLE_LIST *tables)

sql/sql_class.cc

@@ -194,6 +194,7 @@ THD::THD()
   file_id = 0;
   warn_id= 0;
   db_charset= global_system_variables.collation_database;
+  bzero(ha_data, sizeof(ha_data));
   mysys_var=0;
 #ifndef DBUG_OFF
   dbug_sentry=THD_SENTRY_MAGIC;
@@ -205,7 +206,6 @@ THD::THD()
   ull=0;
   system_thread= cleanup_done= abort_on_warning= 0;
   peer_port= 0;					// For SHOW PROCESSLIST
-  transaction.changed_tables = 0;
 #ifdef	__WIN__
   real_id = 0;
 #endif
@@ -240,9 +240,7 @@ THD::THD()
   /* For user vars replication*/
   if (opt_bin_log)
     my_init_dynamic_array(&user_var_events,
-			  sizeof(BINLOG_USER_VAR_EVENT *),
+			  sizeof(BINLOG_USER_VAR_EVENT *), 16, 16);
-			  16,
-			  16);
   else
     bzero((char*) &user_var_events, sizeof(user_var_events));
 
@@ -252,26 +250,8 @@ THD::THD()
   protocol_prep.init(this);
 
   tablespace_op=FALSE;
-#ifdef USING_TRANSACTIONS
-  bzero((char*) &transaction,sizeof(transaction));
-  /*
-    Binlog is always open (if needed) before a THD is created (including
-    bootstrap).
-  */
-  if (opt_using_transactions && mysql_bin_log.is_open())
-  {
-    if (open_cached_file(&transaction.trans_log,
-			 mysql_tmpdir, LOG_PREFIX, binlog_cache_size,
-			 MYF(MY_WME)))
-      killed= KILL_CONNECTION;
-    transaction.trans_log.end_of_file= max_binlog_cache_size;
-  }
-#endif
-  init_sql_alloc(&transaction.mem_root, ALLOC_ROOT_MIN_BLOCK_SIZE, 0);
-  {
     ulong tmp=sql_rnd_with_mutex();
     randominit(&rand, tmp + (ulong) &rand, tmp + (ulong) ::query_id);
-  }
 }
 
 
@@ -320,9 +300,12 @@ void THD::init_for_queries()
 
   reset_root_defaults(mem_root, variables.query_alloc_block_size,
                       variables.query_prealloc_size);
+#ifdef USING_TRANSACTIONS
   reset_root_defaults(&transaction.mem_root,
                       variables.trans_alloc_block_size,
                       variables.trans_prealloc_size);
+#endif
+  transaction.xid.null();
 }
 
 
@@ -407,13 +390,8 @@ THD::~THD()
 #endif
   if (!cleanup_done)
     cleanup();
-#ifdef USING_TRANSACTIONS
+
-  if (opt_using_transactions)
-  {
-    close_cached_file(&transaction.trans_log);
     ha_close_connection(this);
-  }
-#endif
 
   sp_cache_clear(&sp_proc_cache);
   sp_cache_clear(&sp_func_cache);
@@ -426,7 +404,9 @@ THD::~THD()
   safeFree(ip);
   safeFree(db);
   free_root(&warn_root,MYF(0));
+#ifdef USING_TRANSACTIONS
   free_root(&transaction.mem_root,MYF(0));
+#endif
   mysys_var=0;					// Safety (shouldn't be needed)
   pthread_mutex_destroy(&LOCK_delete);
 #ifndef DBUG_OFF
@@ -868,7 +848,6 @@ bool select_send::send_data(List<Item> &items)
     InnoDB adaptive hash S-latch to avoid thread deadlocks if it was reserved
     by thd
   */
-  if (thd->transaction.all.innobase_tid)
     ha_release_temporary_latches(thd);
 #endif
 
@@ -903,7 +882,6 @@ bool select_send::send_eof()
   /* We may be passing the control from mysqld to the client: release the
      InnoDB adaptive hash S-latch to avoid thread deadlocks if it was reserved
      by thd */
-  if (thd->transaction.all.innobase_tid)
     ha_release_temporary_latches(thd);
 #endif
 

sql/sql_class.h

@@ -43,6 +43,96 @@ enum enum_check_fields { CHECK_FIELD_IGNORE, CHECK_FIELD_WARN,
 extern char internal_table_name[2];
 extern const char **errmesg;
 
+#define TC_LOG_PAGE_SIZE   8192
+#define TC_LOG_MIN_SIZE    (3*TC_LOG_PAGE_SIZE)
+extern uint opt_tc_log_size;
+extern uint tc_log_max_pages_used;
+extern uint tc_log_page_size;
+extern uint tc_log_page_waits;
+
+#define TC_HEURISTIC_RECOVER_COMMIT   1
+#define TC_HEURISTIC_RECOVER_ROLLBACK 2
+extern uint tc_heuristic_recover;
+
+/*
+  Transaction Coordinator log - a base abstract class
+  for two different implementations
+*/
+class TC_LOG
+{
+  public:
+  int using_heuristic_recover();
+  TC_LOG() {}
+  virtual ~TC_LOG() {}
+
+  virtual int open(const char *opt_name)=0;
+  virtual void close()=0;
+  virtual int log(THD *thd, my_xid xid)=0;
+  virtual void unlog(ulong cookie, my_xid xid)=0;
+};
+
+class TC_LOG_DUMMY: public TC_LOG // use it to disable the logging
+{
+  public:
+  int open(const char *opt_name)        { return 0; }
+  void close()                          { }
+  int log(THD *thd, my_xid xid)         { return 1; }
+  void unlog(ulong cookie, my_xid xid)  { }
+};
+
+class TC_LOG_MMAP: public TC_LOG
+{
+  private:
+
+  typedef enum {
+    POOL,                 // page is in pool
+    ERROR,                // last sync failed
+    DIRTY                 // new xids added since last sync
+  } PAGE_STATE;
+
+  typedef struct st_page {
+    struct st_page *next; // page a linked in a fifo queue
+    my_xid *start, *end;  // usable area of a page
+    my_xid *ptr;          // next xid will be written here
+    int size, free;       // max and current number of free xid slots on the page
+    int waiters;          // number of waiters on condition
+    PAGE_STATE state;     // see above
+    pthread_mutex_t lock; // to access page data or control structure
+    pthread_cond_t  cond; // to wait for a sync
+  } PAGE;
+
+  char logname[FN_REFLEN];
+  File fd;
+  uint file_length, npages, inited;
+  uchar *data;
+  struct st_page *pages, *syncing, *active, *pool, *pool_last;
+  /*
+    note that, e.g. LOCK_active is only used to protect
+    'active' pointer, to protect the content of the active page
+    one has to use active->lock.
+    Same for LOCK_pool and LOCK_sync
+  */
+  pthread_mutex_t LOCK_active, LOCK_pool, LOCK_sync;
+  pthread_cond_t COND_pool, COND_active;
+
+  public:
+  TC_LOG_MMAP(): inited(0) {}
+  int open(const char *opt_name);
+  void close();
+  int log(THD *thd, my_xid xid);
+  void unlog(ulong cookie, my_xid xid);
+  int recover();
+
+  private:
+  void get_active_from_pool();
+  int sync();
+  int overflow();
+};
+
+extern TC_LOG *tc_log;
+extern TC_LOG_MMAP tc_log_mmap;
+extern TC_LOG_DUMMY tc_log_dummy;
+
 /* log info errors */
 #define LOG_INFO_EOF -1
 #define LOG_INFO_IO  -2
@@ -81,8 +171,18 @@ typedef struct st_user_var_events
 
 class Log_event;
 
-class MYSQL_LOG
+/*
- {
+  TODO split MYSQL_LOG into base MYSQL_LOG and
+  MYSQL_QUERY_LOG, MYSQL_SLOW_LOG, MYSQL_BIN_LOG
+  most of the code from MYSQL_LOG should be in the MYSQL_BIN_LOG
+  only (TC_LOG included)
+
+  TODO use mmap instead of IO_CACHE for binlog
+  (mmap+fsync is two times faster than write+fsync)
+*/
+
+class MYSQL_LOG: public TC_LOG
+{
  private:
   /* LOCK_log and LOCK_index are inited by init_pthread_objects() */
   pthread_mutex_t LOCK_log, LOCK_index;
@@ -108,8 +208,8 @@ class MYSQL_LOG
     etc. So in 4.x this is 1 for relay logs, 0 for binlogs.
     In 5.0 it's 0 for relay logs too!
   */
-  bool no_auto_events;			     
+  bool no_auto_events;
-  /* 
+  /*
      The max size before rotation (usable only if log_type == LOG_BIN: binary
      logs and relay logs).
      For a binlog, max_size should be max_binlog_size.
@@ -117,16 +217,26 @@ class MYSQL_LOG
      max_binlog_size otherwise.
      max_size is set in init(), and dynamically changed (when one does SET
      GLOBAL MAX_BINLOG_SIZE|MAX_RELAY_LOG_SIZE) by fix_max_binlog_size and
-     fix_max_relay_log_size). 
+     fix_max_relay_log_size).
   */
   ulong max_size;
+
+  ulong prepared_xids; /* for tc log - number of xids to remember */
+  pthread_mutex_t LOCK_prep_xids;
+  pthread_cond_t  COND_prep_xids;
   friend class Log_event;
 
 public:
   MYSQL_LOG();
   ~MYSQL_LOG();
 
-  /* 
+  int open(const char *opt_name);
+  void close();
+  int log(THD *thd, my_xid xid);
+  void unlog(ulong cookie, my_xid xid);
+  int recover(IO_CACHE *log, Format_description_log_event *fdle);
+
+  /*
      These describe the log's format. This is used only for relay logs.
      _for_exec is used by the SQL thread, _for_queue by the I/O thread. It's
      necessary to have 2 distinct objects, because the I/O thread may be reading
@@ -145,7 +255,7 @@ public:
   {
 #ifndef DBUG_OFF
     char buf1[22],buf2[22];
-#endif	
+#endif
     DBUG_ENTER("harvest_bytes_written");
     (*counter)+=bytes_written;
     DBUG_PRINT("info",("counter: %s  bytes_written: %s", llstr(*counter,buf1),
@@ -162,18 +272,36 @@ public:
 	    bool no_auto_events_arg, ulong max_size);
   void init_pthread_objects();
   void cleanup();
-  bool open(const char *log_name,enum_log_type log_type,
+  bool open(const char *log_name,
-	    const char *new_name, const char *index_file_name_arg,
+            enum_log_type log_type,
+            const char *new_name,
 	    enum cache_type io_cache_type_arg,
 	    bool no_auto_events_arg, ulong max_size,
             bool null_created);
+  const char *generate_name(const char *log_name, const char *suffix,
+                            bool strip_ext, char *buff);
+  /* simplified open_xxx wrappers for the gigantic open above */
+  bool open_query_log(const char *log_name)
+  {
+    char buf[FN_REFLEN];
+    return open(generate_name(log_name, ".log", 0, buf),
+                LOG_NORMAL, 0, WRITE_CACHE, 0, 0, 0);
+  }
+  bool open_slow_log(const char *log_name)
+  {
+    char buf[FN_REFLEN];
+    return open(generate_name(log_name, "-slow.log", 0, buf),
+                LOG_NORMAL, 0, WRITE_CACHE, 0, 0, 0);
+  }
+  bool open_index_file(const char *index_file_name_arg,
+                       const char *log_name);
   void new_file(bool need_lock= 1);
   bool write(THD *thd, enum enum_server_command command,
 	     const char *format,...);
   bool write(THD *thd, const char *query, uint query_length,
 	     time_t query_start=0);
   bool write(Log_event* event_info); // binary log write
-  bool write(THD *thd, IO_CACHE *cache, bool commit_or_rollback);
+  bool write(THD *thd, IO_CACHE *cache);
 
   /*
     v stands for vector
@@ -181,20 +309,18 @@ public:
   */
   bool appendv(const char* buf,uint len,...);
   bool append(Log_event* ev);
-  
+
   int generate_new_name(char *new_name,const char *old_name);
   void make_log_name(char* buf, const char* log_ident);
   bool is_active(const char* log_file_name);
   int update_log_index(LOG_INFO* linfo, bool need_update_threads);
-  int purge_logs(const char *to_log, bool included, 
+  int purge_logs(const char *to_log, bool included,
                  bool need_mutex, bool need_update_threads,
                  ulonglong *decrease_log_space);
   int purge_logs_before_date(time_t purge_time);
-  int purge_first_log(struct st_relay_log_info* rli, bool included); 
+  int purge_first_log(struct st_relay_log_info* rli, bool included);
   bool reset_logs(THD* thd);
   void close(uint exiting);
-  bool cut_spurious_tail();
-  void report_pos_in_innodb();
 
   // iterating through the log index file
   int find_log_pos(LOG_INFO* linfo, const char* log_name,
@@ -487,6 +613,10 @@ typedef struct system_status_var
   ulong ha_rollback_count;
   ulong ha_update_count;
   ulong ha_write_count;
+  ulong ha_prepare_count;
+  ulong ha_discover_count;
+  ulong ha_savepoint_count;
+  ulong ha_savepoint_rollback_count;
 
   /* KEY_CACHE parts. These are copies of the original */
   ulong key_blocks_changed;
@@ -765,6 +895,14 @@ private:
   Statement *last_found_statement;
 };
 
+struct st_savepoint {
+  struct st_savepoint *prev;
+  char                *name;
+  uint                 length, nht;
+};
+
+enum xa_states {XA_NOTR=0, XA_ACTIVE, XA_IDLE, XA_PREPARED};
+extern const char *xa_state_names[];
 
 /*
   A registry for item tree transformations performed during
@@ -907,15 +1045,15 @@ public:
   thr_lock_type update_lock_default;
   delayed_insert *di;
   my_bool    tablespace_op;	/* This is TRUE in DISCARD/IMPORT TABLESPACE */
+  /* container for handler's private per-connection data */
+  void *ha_data[MAX_HA];
   struct st_transactions {
-    IO_CACHE trans_log;                 // Inited ONLY if binlog is open !
+    SAVEPOINT *savepoints;
     THD_TRANS all;			// Trans since BEGIN WORK
     THD_TRANS stmt;			// Trans for current statement
-    uint bdb_lock_count;
-#ifdef HAVE_NDBCLUSTER_DB
-    void* thd_ndb;
-#endif
     bool on;
+    XID  xid;
+    enum xa_states xa_state;
     /*
        Tables changed in transaction (that must be invalidated in query cache).
        List contain only transactional tables, that not invalidated in query
@@ -926,8 +1064,18 @@ public:
     void cleanup()
     {
       changed_tables = 0;
+#ifdef USING_TRANSACTIONS
       free_root(&mem_root,MYF(MY_KEEP_PREALLOC));
+#endif
     }
+#ifdef USING_TRANSACTIONS
+    st_transactions()
+    {
+      bzero((char*)this, sizeof(*this));
+      xid.null();
+      init_sql_alloc(&mem_root, ALLOC_ROOT_MIN_BLOCK_SIZE, 0);
+    }
+#endif
   } transaction;
   Field      *dupp_field;
 #ifndef __WIN__
@@ -1126,7 +1274,7 @@ public:
   inline ulonglong insert_id(void)
   {
     if (!last_insert_id_used)
-    {      
+    {
       last_insert_id_used=1;
       current_insert_id=last_insert_id;
     }
@@ -1135,13 +1283,11 @@ public:
   inline ulonglong found_rows(void)
   {
     return limit_found_rows;
-  }                                                                         
+  }
   inline bool active_transaction()
   {
-#ifdef USING_TRANSACTIONS    
+#ifdef USING_TRANSACTIONS
-    return (transaction.all.bdb_tid != 0 ||
+    return server_status & SERVER_STATUS_IN_TRANS;
-	    transaction.all.innodb_active_trans != 0 ||
-	    transaction.all.ndb_tid != 0);
 #else
     return 0;
 #endif
@@ -1665,7 +1811,7 @@ class multi_delete :public select_result_interceptor
   ha_rows deleted, found;
   uint num_of_tables;
   int error;
-  bool do_delete, transactional_tables, log_delayed, normal_tables;
+  bool do_delete, transactional_tables, normal_tables;
 public:
   multi_delete(THD *thd, TABLE_LIST *dt, uint num_of_tables);
   ~multi_delete();
@@ -1692,7 +1838,7 @@ class multi_update :public select_result_interceptor
   uint table_count;
   Copy_field *copy_field;
   enum enum_duplicates handle_duplicates;
-  bool do_update, trans_safe, transactional_tables, log_delayed, ignore;
+  bool do_update, trans_safe, transactional_tables, ignore;
 
 public:
   multi_update(THD *thd_arg, TABLE_LIST *ut, TABLE_LIST *leaves_list,

sql/sql_delete.cc

@@ -36,8 +36,8 @@ bool mysql_delete(THD *thd, TABLE_LIST *table_list, COND *conds,
   TABLE		*table;
   SQL_SELECT	*select=0;
   READ_RECORD	info;
-  bool 		using_limit=limit != HA_POS_ERROR;
+  bool          using_limit=limit != HA_POS_ERROR;
-  bool		transactional_table, log_delayed, safe_update, const_cond; 
+  bool		transactional_table, safe_update, const_cond;
   ha_rows	deleted;
   SELECT_LEX   *select_lex= &thd->lex->select_lex;
   DBUG_ENTER("mysql_delete");
@@ -233,7 +233,6 @@ cleanup:
 
   delete select;
   transactional_table= table->file->has_transactions();
-  log_delayed= (transactional_table || table->s->tmp_table);
   /*
     We write to the binary log even if we deleted no row, because maybe the
     user is using this command to ensure that a table is clean on master *and
@@ -249,11 +248,11 @@ cleanup:
       if (error <= 0)
         thd->clear_error();
       Query_log_event qinfo(thd, thd->query, thd->query_length,
-			    log_delayed, FALSE);
+			    transactional_table, FALSE);
       if (mysql_bin_log.write(&qinfo) && transactional_table)
 	error=1;
     }
-    if (!log_delayed)
+    if (!transactional_table)
       thd->options|=OPTION_STATUS_NO_TRANS_UPDATE;
   }
   if (transactional_table)
@@ -398,7 +397,7 @@ multi_delete::multi_delete(THD *thd_arg, TABLE_LIST *dt,
 			   uint num_of_tables_arg)
   : delete_tables(dt), thd(thd_arg), deleted(0), found(0),
     num_of_tables(num_of_tables_arg), error(0),
-    do_delete(0), transactional_tables(0), log_delayed(0), normal_tables(0)
+    do_delete(0), transactional_tables(0), normal_tables(0)
 {
   tempfiles = (Unique **) sql_calloc(sizeof(Unique *) * (num_of_tables-1));
 }
@@ -445,9 +444,7 @@ multi_delete::initialize_tables(JOIN *join)
       tbl->no_cache= 1;
       tbl->used_keys.clear_all();
       if (tbl->file->has_transactions())
-	log_delayed= transactional_tables= 1;
+	transactional_tables= 1;
-      else if (tbl->s->tmp_table != NO_TMP_TABLE)
-	log_delayed= 1;
       else
 	normal_tables= 1;
     }
@@ -670,14 +667,14 @@ bool multi_delete::send_eof()
       if (error <= 0)
         thd->clear_error();
       Query_log_event qinfo(thd, thd->query, thd->query_length,
-			    log_delayed, FALSE);
+			    transactional_tables, FALSE);
       if (mysql_bin_log.write(&qinfo) && !normal_tables)
 	local_error=1;  // Log write failed: roll back the SQL statement
     }
-    if (!log_delayed)
+    if (!transactional_tables)
       thd->options|=OPTION_STATUS_NO_TRANS_UPDATE;
   }
-  /* Commit or rollback the current SQL statement */ 
+  /* Commit or rollback the current SQL statement */
   if (transactional_tables)
     if (ha_autocommit_or_rollback(thd,local_error > 0))
       local_error=1;
@@ -736,7 +733,7 @@ bool mysql_truncate(THD *thd, TABLE_LIST *table_list, bool dont_send_ok)
       If we return here we will not have logged the truncation to the bin log
       and we will not send_ok() to the client.
     */
-    goto end; 
+    goto end;
   }
 
   (void) sprintf(path,"%s/%s/%s%s",mysql_data_home,table_list->db,
@@ -768,7 +765,7 @@ bool mysql_truncate(THD *thd, TABLE_LIST *table_list, bool dont_send_ok)
 
   *fn_ext(path)=0;				// Remove the .frm extension
   error= ha_create_table(path,&create_info,1);
-  query_cache_invalidate3(thd, table_list, 0); 
+  query_cache_invalidate3(thd, table_list, 0);
 
 end:
   if (!dont_send_ok)
@@ -779,7 +776,7 @@ end:
       {
         thd->clear_error();
 	Query_log_event qinfo(thd, thd->query, thd->query_length,
-			      thd->tmp_table, FALSE);
+			      0, FALSE);
 	mysql_bin_log.write(&qinfo);
       }
       send_ok(thd);		// This should return record count

sql/sql_insert.cc

@@ -168,7 +168,7 @@ bool mysql_insert(THD *thd,TABLE_LIST *table_list,
     runs without --log-update or --log-bin).
   */
   bool log_on= (thd->options & OPTION_BIN_LOG) || (!(thd->master_access & SUPER_ACL));
-  bool transactional_table, log_delayed;
+  bool transactional_table;
   uint value_count;
   ulong counter = 1;
   ulonglong id;
@@ -433,7 +433,7 @@ bool mysql_insert(THD *thd,TABLE_LIST *table_list,
     /*
       Invalidate the table in the query cache if something changed.
       For the transactional algorithm to work the invalidation must be
-      before binlog writing and ha_autocommit_...
+      before binlog writing and ha_autocommit_or_rollback
     */
     if (info.copied || info.deleted || info.updated)
     {
@@ -442,7 +442,6 @@ bool mysql_insert(THD *thd,TABLE_LIST *table_list,
 
     transactional_table= table->file->has_transactions();
 
-    log_delayed= (transactional_table || table->s->tmp_table);
     if ((info.copied || info.deleted || info.updated) &&
 	(error <= 0 || !transactional_table))
     {
@@ -451,11 +450,11 @@ bool mysql_insert(THD *thd,TABLE_LIST *table_list,
         if (error <= 0)
           thd->clear_error();
 	Query_log_event qinfo(thd, thd->query, thd->query_length,
-			      log_delayed, FALSE);
+			      transactional_table, FALSE);
 	if (mysql_bin_log.write(&qinfo) && transactional_table)
 	  error=1;
       }
-      if (!log_delayed)
+      if (!transactional_table)
 	thd->options|=OPTION_STATUS_NO_TRANS_UPDATE;
     }
     if (transactional_table)
@@ -1974,7 +1973,7 @@ bool select_insert::send_eof()
 
   /*
     We must invalidate the table in the query cache before binlog writing
-    and ha_autocommit_...
+    and ha_autocommit_or_rollback
   */
 
   if (info.copied || info.deleted || info.updated)

sql/sql_lex.cc

@@ -141,7 +141,7 @@ void lex_start(THD *thd, uchar *buf,uint length)
   lex->view_prepare_mode= FALSE;
   lex->derived_tables= 0;
   lex->lock_option= TL_READ;
-  lex->found_colon= 0;
+  lex->found_semicolon= 0;
   lex->safe_to_cache_query= 1;
   lex->time_zone_tables_used= 0;
   lex->leaf_tables_insert= lex->proc_table= lex->query_tables= 0;
@@ -970,7 +970,7 @@ int yylex(void *arg, void *yythd)
             (thd->command != COM_PREPARE))
         {
 	  lex->safe_to_cache_query= 0;
-          lex->found_colon=    (char*) lex->ptr;
+          lex->found_semicolon=(char*) lex->ptr;
           thd->server_status|= SERVER_MORE_RESULTS_EXISTS;
           lex->next_state=     MY_LEX_END;
           return (END_OF_INPUT);

sql/sql_lex.h

@@ -87,6 +87,8 @@ enum enum_sql_command {
   SQLCOM_PREPARE, SQLCOM_EXECUTE, SQLCOM_DEALLOCATE_PREPARE,
   SQLCOM_CREATE_VIEW, SQLCOM_DROP_VIEW,
   SQLCOM_CREATE_TRIGGER, SQLCOM_DROP_TRIGGER,
+  SQLCOM_XA_START, SQLCOM_XA_END, SQLCOM_XA_PREPARE,
+  SQLCOM_XA_COMMIT, SQLCOM_XA_ROLLBACK, SQLCOM_XA_RECOVER,
   /* This should be the last !!! */
 
   SQLCOM_END
@@ -635,6 +637,9 @@ struct st_trg_chistics
 
 extern sys_var_long_ptr trg_new_row_fake_var;
 
+enum xa_option_words {XA_NONE, XA_JOIN, XA_RESUME, XA_ONE_PHASE,
+                      XA_SUSPEND, XA_FOR_MIGRATE};
+
 /* The state of the lex parsing. This is saved in the THD struct */
 
 typedef struct st_lex
@@ -654,12 +659,12 @@ typedef struct st_lex
   char *backup_dir;				/* For RESTORE/BACKUP */
   char* to_log;                                 /* For PURGE MASTER LOGS TO */
   char* x509_subject,*x509_issuer,*ssl_cipher;
-  char* found_colon;                            /* For multi queries - next query */
+  char* found_semicolon;                        /* For multi queries - next query */
   String *wild;
   sql_exchange *exchange;
   select_result *result;
   Item *default_value, *on_update_value;
-  LEX_STRING *comment, name_and_length;
+  LEX_STRING comment, ident;
   LEX_USER *grant_user;
   gptr yacc_yyss,yacc_yyvs;
   THD *thd;
@@ -689,7 +694,6 @@ typedef struct st_lex
   List<LEX_STRING>    view_list; // view list (list of field names in view)
   SQL_LIST	      proc_list, auxilliary_table_list, save_list;
   create_field	      *last_field;
-  char		      *savepoint_name;		// Transaction savepoint id
   udf_func udf;
   HA_CHECK_OPT   check_opt;			// check/repair options
   HA_CREATE_INFO create_info;
@@ -703,7 +707,10 @@ typedef struct st_lex
   enum enum_duplicates duplicates;
   enum enum_tx_isolation tx_isolation;
   enum enum_ha_read_modes ha_read_mode;
+  union {
   enum ha_rkey_function ha_rkey_mode;
+    enum xa_option_words xa_opt;
+  };
   enum enum_var_type option_type;
   enum enum_view_create_mode create_view_mode;
   enum enum_drop_mode drop_mode;

sql/sql_load.cc

@@ -105,7 +105,7 @@ bool mysql_load(THD *thd,sql_exchange *ex,TABLE_LIST *table_list,
   char *tdb= thd->db ? thd->db : db;		// Result is never null
   ulong skip_lines= ex->skip_lines;
   int res;
-  bool transactional_table, log_delayed;
+  bool transactional_table;
   DBUG_ENTER("mysql_load");
 
 #ifdef EMBEDDED_LIBRARY
@@ -133,7 +133,6 @@ bool mysql_load(THD *thd,sql_exchange *ex,TABLE_LIST *table_list,
   }
   table= table_list->table;
   transactional_table= table->file->has_transactions();
-  log_delayed= (transactional_table || table->s->tmp_table);
 
   if (!fields.elements)
   {
@@ -263,7 +262,7 @@ bool mysql_load(THD *thd,sql_exchange *ex,TABLE_LIST *table_list,
     lf_info.handle_dup = handle_duplicates;
     lf_info.wrote_create_file = 0;
     lf_info.last_pos_in_file = HA_POS_ERROR;
-    lf_info.log_delayed= log_delayed;
+    lf_info.log_delayed= transactional_table;
     read_info.set_io_cache_arg((void*) &lf_info);
   }
 #endif /*!EMBEDDED_LIBRARY*/
@@ -365,7 +364,7 @@ bool mysql_load(THD *thd,sql_exchange *ex,TABLE_LIST *table_list,
       /* If the file was not empty, wrote_create_file is true */
       if (lf_info.wrote_create_file)
       {
-        Delete_file_log_event d(thd, db, log_delayed);
+        Delete_file_log_event d(thd, db, transactional_table);
         mysql_bin_log.write(&d);
       }
     }
@@ -377,7 +376,7 @@ bool mysql_load(THD *thd,sql_exchange *ex,TABLE_LIST *table_list,
 	  (ulong) (info.records - info.copied), (ulong) thd->cuted_fields);
   send_ok(thd,info.copied+info.deleted,0L,name);
 
-  if (!log_delayed)
+  if (!transactional_table)
     thd->options|=OPTION_STATUS_NO_TRANS_UPDATE;
 #ifndef EMBEDDED_LIBRARY
   if (mysql_bin_log.is_open())
@@ -387,16 +386,16 @@ bool mysql_load(THD *thd,sql_exchange *ex,TABLE_LIST *table_list,
       block will be logged only after Execute_load_log_event (which is wrong),
       when read_info is destroyed.
     */
-    read_info.end_io_cache(); 
+    read_info.end_io_cache();
     if (lf_info.wrote_create_file)
     {
-      Execute_load_log_event e(thd, db, log_delayed);
+      Execute_load_log_event e(thd, db, transactional_table);
       mysql_bin_log.write(&e);
     }
   }
 #endif /*!EMBEDDED_LIBRARY*/
   if (transactional_table)
-    error=ha_autocommit_or_rollback(thd,error); 
+    error=ha_autocommit_or_rollback(thd,error);
 
 err:
   if (thd->lock)
@@ -404,7 +403,7 @@ err:
     mysql_unlock_tables(thd, thd->lock);
     thd->lock=0;
   }
-  thd->abort_on_warning= 0;  
+  thd->abort_on_warning= 0;
   DBUG_RETURN(error);
 }
 
@@ -732,12 +731,11 @@ READ_INFO::READ_INFO(File file_par, uint tot_length, CHARSET_INFO *cs,
       my_free((gptr) buffer,MYF(0)); /* purecov: inspected */
       error=1;
     }
-    else 
+    else
     {
       /*
 	init_io_cache() will not initialize read_function member
-	if the cache is READ_NET. The reason is explained in
+	if the cache is READ_NET. So we work around the problem with a
-	mysys/mf_iocache.c. So we work around the problem with a
 	manual assignment
       */
       need_end_io_cache = 1;

sql/sql_parse.cc

@@ -83,6 +83,10 @@ const char *command_name[]={
   "Error"					// Last command number
 };
 
+const char *xa_state_names[]={
+  "NON-EXISTING", "ACTIVE", "IDLE", "PREPARED"
+};
+
 static char empty_c_string[1]= {0};		// Used for not defined 'db'
 
 #ifdef __WIN__
@@ -152,7 +156,7 @@ static bool begin_trans(THD *thd)
 		   OPTION_BEGIN);
     thd->server_status|= SERVER_STATUS_IN_TRANS;
     if (lex->start_transaction_opt & MYSQL_START_TRANS_OPT_WITH_CONS_SNAPSHOT)
-      error= ha_start_consistent_snapshot(thd);    
+      error= ha_start_consistent_snapshot(thd);
   }
   return error;
 }
@@ -555,6 +559,12 @@ void free_max_user_conn(void)
 
   sql_command is actually set to SQLCOM_END sometimes
   so we need the +1 to include it in the array.
+
+  numbers are:
+     0  - read-only query
+  != 0  - query that may change a table
+     2  - query that returns meaningful ROW_COUNT() -
+          a number of modified rows
 */
 
 char  uc_update_queries[SQLCOM_END+1];
@@ -566,23 +576,23 @@ void init_update_queries(void)
   uc_update_queries[SQLCOM_CREATE_TABLE]=1;
   uc_update_queries[SQLCOM_CREATE_INDEX]=1;
   uc_update_queries[SQLCOM_ALTER_TABLE]=1;
-  uc_update_queries[SQLCOM_UPDATE]=1;
+  uc_update_queries[SQLCOM_UPDATE]=2;
-  uc_update_queries[SQLCOM_INSERT]=1;
+  uc_update_queries[SQLCOM_UPDATE_MULTI]=2;
-  uc_update_queries[SQLCOM_INSERT_SELECT]=1;
+  uc_update_queries[SQLCOM_INSERT]=2;
-  uc_update_queries[SQLCOM_DELETE]=1;
+  uc_update_queries[SQLCOM_INSERT_SELECT]=2;
+  uc_update_queries[SQLCOM_DELETE]=2;
+  uc_update_queries[SQLCOM_DELETE_MULTI]=2;
   uc_update_queries[SQLCOM_TRUNCATE]=1;
   uc_update_queries[SQLCOM_DROP_TABLE]=1;
   uc_update_queries[SQLCOM_LOAD]=1;
   uc_update_queries[SQLCOM_CREATE_DB]=1;
   uc_update_queries[SQLCOM_DROP_DB]=1;
-  uc_update_queries[SQLCOM_REPLACE]=1;
+  uc_update_queries[SQLCOM_REPLACE]=2;
-  uc_update_queries[SQLCOM_REPLACE_SELECT]=1;
+  uc_update_queries[SQLCOM_REPLACE_SELECT]=2;
   uc_update_queries[SQLCOM_RENAME_TABLE]=1;
   uc_update_queries[SQLCOM_BACKUP_TABLE]=1;
   uc_update_queries[SQLCOM_RESTORE_TABLE]=1;
-  uc_update_queries[SQLCOM_DELETE_MULTI]=1;
   uc_update_queries[SQLCOM_DROP_INDEX]=1;
-  uc_update_queries[SQLCOM_UPDATE_MULTI]=1;
   uc_update_queries[SQLCOM_CREATE_VIEW]=1;
   uc_update_queries[SQLCOM_DROP_VIEW]=1;
 }
@@ -824,17 +834,6 @@ static int check_connection(THD *thd)
     return(ER_OUT_OF_RESOURCES);
 
   thd->client_capabilities=uint2korr(net->read_pos);
-#ifdef TO_BE_REMOVED_IN_4_1_RELEASE
-  /*
-    This is just a safety check against any client that would use the old
-    CLIENT_CHANGE_USER flag
-  */
-  if ((thd->client_capabilities & CLIENT_PROTOCOL_41) &&
-      !(thd->client_capabilities & (CLIENT_RESERVED |
-				    CLIENT_SECURE_CONNECTION |
-				    CLIENT_MULTI_RESULTS)))
-    thd->client_capabilities&= ~CLIENT_PROTOCOL_41;
-#endif
   if (thd->client_capabilities & CLIENT_PROTOCOL_41)
   {
     thd->client_capabilities|= ((ulong) uint2korr(net->read_pos+2)) << 16;
@@ -1190,24 +1189,25 @@ extern "C" pthread_handler_decl(handle_bootstrap,arg)
       We don't need to obtain LOCK_thread_count here because in bootstrap
       mode we have only one thread.
     */
-    thd->query_id=query_id++;
+    thd->query_id=next_query_id();
-    if (mqh_used && thd->user_connect && check_mqh(thd, SQLCOM_END))
-    {
-      thd->net.error = 0;
-      close_thread_tables(thd);			// Free tables
-      free_root(thd->mem_root,MYF(MY_KEEP_PREALLOC));
-      break;
-    }
     mysql_parse(thd,thd->query,length);
     close_thread_tables(thd);			// Free tables
     if (thd->is_fatal_error)
       break;
     free_root(thd->mem_root,MYF(MY_KEEP_PREALLOC));
+#ifdef USING_TRANSACTIONS
     free_root(&thd->transaction.mem_root,MYF(MY_KEEP_PREALLOC));
+#endif
   }
 
   /* thd->fatal_error should be set in case something went wrong */
 end:
+  bootstrap_error= thd->is_fatal_error;
+
+  net_end(&thd->net);
+  thd->cleanup();
+  delete thd;
+
 #ifndef EMBEDDED_LIBRARY
   (void) pthread_mutex_lock(&LOCK_thread_count);
   thread_count--;
@@ -1216,7 +1216,7 @@ end:
   my_thread_end();
   pthread_exit(0);
 #endif
-  DBUG_RETURN(0);				// Never reached
+  DBUG_RETURN(0);
 }
 
     /* This works because items are allocated with sql_alloc() */
@@ -1282,45 +1282,27 @@ int mysql_table_dump(THD* thd, char* db, char* tbl_name, int fd)
     my_error(ER_GET_ERRNO, MYF(0), error);
 
 err:
-  close_thread_tables(thd);
   DBUG_RETURN(error);
 }
 
 /*
   Ends the current transaction and (maybe) begin the next
-  First uint4 in packet is completion type
-  Remainder is savepoint name (if required)
 
   SYNOPSIS
-    mysql_endtrans()
+    end_trans()
       thd            Current thread
       completion     Completion type
-      savepoint_name Savepoint when doing ROLLBACK_SAVEPOINT_NAME 
-                     or RELEASE_SAVEPOINT_NAME
-      release        (OUT) indicator for release operation
 
   RETURN
     0 - OK
 */
 
-enum enum_mysql_completiontype {
+int end_trans(THD *thd, enum enum_mysql_completiontype completion)
-  ROLLBACK_RELEASE=-2,
-  COMMIT_RELEASE=-1,
-  COMMIT=0,
-  ROLLBACK=1,
-  SAVEPOINT_NAME_ROLLBACK=2,
-  SAVEPOINT_NAME_RELEASE=4,
-  COMMIT_AND_CHAIN=6,
-  ROLLBACK_AND_CHAIN=7
-};
-
-int mysql_endtrans(THD *thd, enum enum_mysql_completiontype completion, 
-		    char *savepoint_name)
 {
   bool do_release= 0;
   int res= 0;
   LEX *lex= thd->lex;
-  DBUG_ENTER("mysql_endtrans");
+  DBUG_ENTER("end_trans");
 
   switch (completion) {
   case COMMIT:
@@ -1331,80 +1313,39 @@ int mysql_endtrans(THD *thd, enum enum_mysql_completiontype completion,
     */
     thd->options&= ~(ulong) (OPTION_BEGIN | OPTION_STATUS_NO_TRANS_UPDATE);
     thd->server_status&= ~SERVER_STATUS_IN_TRANS;
-    if (!(res= ha_commit(thd)))
+    res= ha_commit(thd);
-      send_ok(thd);
     break;
   case COMMIT_RELEASE:
-    do_release= 1;
+    do_release= 1; /* fall through */
   case COMMIT_AND_CHAIN:
     res= end_active_trans(thd);
     if (!res && completion == COMMIT_AND_CHAIN)
       res= begin_trans(thd);
-    if (!res)
-      send_ok(thd);
     break;
   case ROLLBACK_RELEASE:
-    do_release= 1;
+    do_release= 1; /* fall through */
   case ROLLBACK:
   case ROLLBACK_AND_CHAIN:
   {
-    bool warn= 0;
     thd->server_status&= ~SERVER_STATUS_IN_TRANS;
-    if (!ha_rollback(thd))
+    if (ha_rollback(thd))
-    {
-      /*
-        If a non-transactional table was updated, warn; don't warn if this is a
-        slave thread (because when a slave thread executes a ROLLBACK, it has
-        been read from the binary log, so it's 100% sure and normal to produce
-        error ER_WARNING_NOT_COMPLETE_ROLLBACK. If we sent the warning to the
-        slave SQL thread, it would not stop the thread but just be printed in
-        the error log; but we don't want users to wonder why they have this
-        message in the error log, so we don't send it.
-      */
-      warn= (thd->options & OPTION_STATUS_NO_TRANS_UPDATE) && 
-      	    !thd->slave_thread;
-    }
-    else
       res= -1;
     thd->options&= ~(ulong) (OPTION_BEGIN | OPTION_STATUS_NO_TRANS_UPDATE);
     if (!res && (completion == ROLLBACK_AND_CHAIN))
       res= begin_trans(thd);
-
-    if (!res) 
-    {
-      if (warn)
-        push_warning(thd, MYSQL_ERROR::WARN_LEVEL_WARN,
-                     ER_WARNING_NOT_COMPLETE_ROLLBACK,
-                     ER(ER_WARNING_NOT_COMPLETE_ROLLBACK));
-      send_ok(thd);
-    }
     break;
   }
-  case SAVEPOINT_NAME_ROLLBACK:
-    if (!(res=ha_rollback_to_savepoint(thd, savepoint_name)))
-    {
-      if ((thd->options & OPTION_STATUS_NO_TRANS_UPDATE) && !thd->slave_thread)
-	push_warning(thd, MYSQL_ERROR::WARN_LEVEL_WARN,
-		     ER_WARNING_NOT_COMPLETE_ROLLBACK,
-		     ER(ER_WARNING_NOT_COMPLETE_ROLLBACK));
-      send_ok(thd);
-    }
-    break;
-  case SAVEPOINT_NAME_RELEASE:
-    if (!(res=ha_release_savepoint_name(thd, savepoint_name)))
-      send_ok(thd);
-    break;
   default:
     res= -1;
     my_error(ER_UNKNOWN_COM_ERROR, MYF(0));
     DBUG_RETURN(-1);
   }
-  
+
   if (res < 0)
     my_error(thd->killed_errno(), MYF(0));
   else if ((res == 0) && do_release)
-    thd->killed= THD::KILL_CONNECTION; 
+    thd->killed= THD::KILL_CONNECTION;
-  
+
   DBUG_RETURN(res);
 }
 
@@ -1519,7 +1460,7 @@ bool dispatch_command(enum enum_server_command command, THD *thd,
   VOID(pthread_mutex_lock(&LOCK_thread_count));
   thd->query_id=query_id;
   if (command != COM_STATISTICS && command != COM_PING)
-    query_id++;
+    next_query_id();
   thread_running++;
   /* TODO: set thd->lex->sql_command to SQLCOM_END here */
   VOID(pthread_mutex_unlock(&LOCK_thread_count));
@@ -1676,9 +1617,9 @@ bool dispatch_command(enum enum_server_command command, THD *thd,
     DBUG_PRINT("query",("%-.4096s",thd->query));
     mysql_parse(thd,thd->query, thd->query_length);
 
-    while (!thd->killed && thd->lex->found_colon && !thd->net.report_error)
+    while (!thd->killed && thd->lex->found_semicolon && !thd->net.report_error)
     {
-      char *packet= thd->lex->found_colon;
+      char *packet= thd->lex->found_semicolon;
       /*
         Multiple queries exits, execute them individually
 	in embedded server - just store them to be executed later 
@@ -1698,7 +1639,7 @@ bool dispatch_command(enum enum_server_command command, THD *thd,
       VOID(pthread_mutex_lock(&LOCK_thread_count));
       thd->query_length= length;
       thd->query= packet;
-      thd->query_id= query_id++;
+      thd->query_id= next_query_id();
       /* TODO: set thd->lex->sql_command to SQLCOM_END here */
       VOID(pthread_mutex_unlock(&LOCK_thread_count));
 #ifndef EMBEDDED_LIBRARY
@@ -2010,6 +1951,17 @@ bool dispatch_command(enum enum_server_command command, THD *thd,
     thd->proc_info="closing tables";
     close_thread_tables(thd);			/* Free tables */
   }
+  /*
+    assume handlers auto-commit (if some doesn't - transaction handling
+    in MySQL should be redesigned to support it; it's a big change,
+    and it's not worth it - better to commit explicitly only writing
+    transactions, read-only ones should better take care of themselves.
+    saves some work in 2pc too)
+    see also sql_base.cc - close_thread_tables()
+  */
+  bzero(&thd->transaction.stmt, sizeof(thd->transaction.stmt));
+  if (!thd->active_transaction())
+    thd->transaction.xid.null();
 
   /* report error issued during command execution */
   if (thd->killed_errno() && !thd->net.report_error)
@@ -2411,7 +2363,7 @@ mysql_execute_command(THD *thd)
   */
   if (opt_readonly &&
       !(thd->slave_thread || (thd->master_access & SUPER_ACL)) &&
-      (uc_update_queries[lex->sql_command] > 0))
+      uc_update_queries[lex->sql_command])
   {
     my_error(ER_OPTION_PREVENTS_STATEMENT, MYF(0), "--read-only");
     DBUG_RETURN(-1);
@@ -2677,7 +2629,7 @@ mysql_execute_command(THD *thd)
   {
     if (check_global_access(thd, REPL_SLAVE_ACL))
       goto error;
-    res = show_binlog_events(thd);
+    res = mysql_show_binlog_events(thd);
     break;
   }
 #endif
@@ -2712,7 +2664,7 @@ mysql_execute_command(THD *thd)
         check_access(thd, INDEX_ACL, first_table->db,
                      &first_table->grant.privilege, 0, 0))
       goto error;
-    res= mysql_assign_to_keycache(thd, first_table, &lex->name_and_length);
+    res= mysql_assign_to_keycache(thd, first_table, &lex->ident);
     break;
   }
   case SQLCOM_PRELOAD_KEYS:
@@ -3333,7 +3285,7 @@ unsent_create_error:
                   first_table->ancestor && first_table->ancestor->next_local);
       my_error(ER_VIEW_DELETE_MERGE_VIEW, MYF(0),
                first_table->view_db.str, first_table->view_name.str);
-      res= -1;
+      res= FALSE;
       break;
     }
 
@@ -3358,7 +3310,6 @@ unsent_create_error:
     }
     else
       res= TRUE;
-    close_thread_tables(thd);
     break;
   }
   case SQLCOM_DROP_TABLE:
@@ -3881,7 +3832,7 @@ unsent_create_error:
     */
     if (check_db_used(thd, all_tables))
       goto error;
-    res= mysql_ha_read(thd, first_table, lex->ha_read_mode, lex->backup_dir,
+    res= mysql_ha_read(thd, first_table, lex->ha_read_mode, lex->ident.str,
                        lex->insert_list, lex->ha_rkey_mode, select_lex->where,
                        select_lex->select_limit, select_lex->offset_limit);
     break;
@@ -3889,32 +3840,121 @@ unsent_create_error:
   case SQLCOM_BEGIN:
     if (begin_trans(thd))
       goto error;
-    else
+    send_ok(thd);
-      send_ok(thd);
     break;
   case SQLCOM_COMMIT:
-    if (mysql_endtrans(thd, lex->tx_release ? COMMIT_RELEASE :
+    if (end_trans(thd, lex->tx_release ? COMMIT_RELEASE :
-    		       lex->tx_chain ? COMMIT_AND_CHAIN : COMMIT, 0))
+                              lex->tx_chain ? COMMIT_AND_CHAIN : COMMIT))
       goto error;
+    send_ok(thd);
     break;
   case SQLCOM_ROLLBACK:
-    if (mysql_endtrans(thd, lex->tx_release ? ROLLBACK_RELEASE :
+    if (end_trans(thd, lex->tx_release ? ROLLBACK_RELEASE :
-    		       lex->tx_chain ? ROLLBACK_AND_CHAIN : ROLLBACK, 0))
+                              lex->tx_chain ? ROLLBACK_AND_CHAIN : ROLLBACK))
-      goto error;
-    break;
-  case SQLCOM_ROLLBACK_TO_SAVEPOINT:
-    if (mysql_endtrans(thd, SAVEPOINT_NAME_ROLLBACK, lex->savepoint_name))
-      goto error;
-    break;
-  case SQLCOM_SAVEPOINT:
-    if (!ha_savepoint(thd, lex->savepoint_name))
-      send_ok(thd);
-    else
       goto error;
+    send_ok(thd);
     break;
   case SQLCOM_RELEASE_SAVEPOINT:
-    if (mysql_endtrans(thd, SAVEPOINT_NAME_RELEASE, lex->savepoint_name))
+  {
-      goto error;
+    SAVEPOINT **sv;
+    for (sv=&thd->transaction.savepoints; *sv; sv=&(*sv)->prev)
+    {
+      if (my_strnncoll(system_charset_info,
+                       (uchar *)lex->ident.str, lex->ident.length,
+                       (uchar *)(*sv)->name, (*sv)->length) == 0)
+        break;
+    }
+    if (*sv)
+    {
+      if (ha_release_savepoint(thd, *sv))
+        res= TRUE; // cannot happen
+      else
+        send_ok(thd);
+      *sv=(*sv)->prev;
+    }
+    else
+    {
+      my_error(ER_SP_DOES_NOT_EXIST, MYF(0), "SAVEPOINT", lex->ident.str);
+      res= TRUE;
+    }
+    break;
+  }
+  case SQLCOM_ROLLBACK_TO_SAVEPOINT:
+  {
+    SAVEPOINT **sv;
+    for (sv=&thd->transaction.savepoints; *sv; sv=&(*sv)->prev)
+    {
+      if (my_strnncoll(system_charset_info,
+                       (uchar *)lex->ident.str, lex->ident.length,
+                       (uchar *)(*sv)->name, (*sv)->length) == 0)
+        break;
+    }
+    if (*sv)
+    {
+      if (ha_rollback_to_savepoint(thd, *sv))
+        res= TRUE; // cannot happen
+      else
+      {
+        if ((thd->options & OPTION_STATUS_NO_TRANS_UPDATE) &&
+            !thd->slave_thread)
+          push_warning(thd, MYSQL_ERROR::WARN_LEVEL_WARN,
+                       ER_WARNING_NOT_COMPLETE_ROLLBACK,
+                       ER(ER_WARNING_NOT_COMPLETE_ROLLBACK));
+        send_ok(thd);
+      }
+      *sv=(*sv)->prev;
+    }
+    else
+    {
+      my_error(ER_SP_DOES_NOT_EXIST, MYF(0), "SAVEPOINT", lex->ident.str);
+      res= TRUE;
+    }
+    break;
+  }
+  case SQLCOM_SAVEPOINT:
+    if (!(thd->options & (OPTION_NOT_AUTOCOMMIT | OPTION_BEGIN)) ||
+        !opt_using_transactions)
+      send_ok(thd);
+    else
+    {
+      SAVEPOINT **sv, *newsv;
+      for (sv=&thd->transaction.savepoints; *sv; sv=&(*sv)->prev)
+      {
+        if (my_strnncoll(system_charset_info,
+                         (uchar *)lex->ident.str, lex->ident.length,
+                         (uchar *)(*sv)->name, (*sv)->length) == 0)
+          break;
+      }
+      if (*sv) /* old savepoint of the same name exists */
+      {
+        newsv=*sv;
+        ha_release_savepoint(thd, *sv); // it cannot fail
+        *sv=(*sv)->prev;
+      }
+      else if ((newsv=(SAVEPOINT *) alloc_root(&thd->transaction.mem_root,
+                                               savepoint_alloc_size)) == 0)
+      {
+        my_error(ER_OUT_OF_RESOURCES, MYF(0));
+        res= TRUE;
+        break;
+      }
+      newsv->name=strmake_root(&thd->transaction.mem_root,
+                               lex->ident.str, lex->ident.length);
+      newsv->length=lex->ident.length;
+      /*
+        if we'll get an error here, don't add new savepoint to the list.
+        we'll lose a little bit of memory in transaction mem_root, but it'll
+        be free'd when transaction ends anyway
+      */
+      if (ha_savepoint(thd, newsv))
+        res= TRUE;
+      else
+      {
+        newsv->prev=thd->transaction.savepoints;
+        thd->transaction.savepoints=newsv;
+        send_ok(thd);
+      }
+    }
     break;
   case SQLCOM_CREATE_PROCEDURE:
   case SQLCOM_CREATE_SPFUNCTION:
@@ -3931,12 +3971,12 @@ unsent_create_error:
       lex->sphead= 0;
       goto error;
     }
-    
+
     if (!lex->sphead->m_db.str || !lex->sphead->m_db.str[0])
     {
       lex->sphead->m_db.length= strlen(thd->db);
-      lex->sphead->m_db.str= strmake_root(thd->mem_root, thd->db, 
+      lex->sphead->m_db.str= strmake_root(thd->mem_root, thd->db,
-      					   lex->sphead->m_db.length);
+                                           lex->sphead->m_db.length);
     }
 
     name= lex->sphead->name(&namelen);
@@ -4291,7 +4331,157 @@ unsent_create_error:
     res= mysql_create_or_drop_trigger(thd, all_tables, 0);
     break;
   }
-  default:					/* Impossible */
+  case SQLCOM_XA_START:
+    if (thd->transaction.xa_state == XA_IDLE && thd->lex->xa_opt == XA_RESUME)
+    {
+      if (! thd->transaction.xid.eq(&thd->lex->ident))
+      {
+        my_error(ER_XAER_NOTA, MYF(0));
+        break;
+      }
+      thd->transaction.xa_state=XA_ACTIVE;
+      send_ok(thd);
+      res=TRUE;
+      break;
+    }
+    if (thd->lex->ident.length > MAXGTRIDSIZE || thd->lex->xa_opt != XA_NONE)
+    { // JOIN is not supported yet. TODO
+      my_error(ER_XAER_INVAL, MYF(0));
+      break;
+    }
+    if (thd->transaction.xa_state != XA_NOTR)
+    {
+      my_error(ER_XAER_RMFAIL, MYF(0),
+               xa_state_names[thd->transaction.xa_state]);
+      break;
+    }
+    if (thd->active_transaction() || thd->locked_tables)
+    {
+      my_error(ER_XAER_OUTSIDE, MYF(0));
+      break;
+    }
+    DBUG_ASSERT(thd->transaction.xid.is_null());
+    thd->transaction.xa_state=XA_ACTIVE;
+    thd->transaction.xid.set(&thd->lex->ident);
+    thd->options= ((thd->options & (ulong) ~(OPTION_STATUS_NO_TRANS_UPDATE)) |
+                   OPTION_BEGIN);
+    thd->server_status|= SERVER_STATUS_IN_TRANS;
+    send_ok(thd);
+    res=TRUE;
+    break;
+  case SQLCOM_XA_END:
+    /* fake it */
+    if (thd->lex->xa_opt != XA_NONE)
+    { // SUSPEND and FOR MIGRATE are not supported yet. TODO
+      my_error(ER_XAER_INVAL, MYF(0));
+      break;
+    }
+    if (thd->transaction.xa_state != XA_ACTIVE)
+    {
+      my_error(ER_XAER_RMFAIL, MYF(0),
+               xa_state_names[thd->transaction.xa_state]);
+      break;
+    }
+    if (!thd->transaction.xid.eq(&thd->lex->ident))
+    {
+      my_error(ER_XAER_NOTA, MYF(0));
+      break;
+    }
+    thd->transaction.xa_state=XA_IDLE;
+    send_ok(thd);
+    res=TRUE;
+    break;
+  case SQLCOM_XA_PREPARE:
+    if (thd->transaction.xa_state != XA_IDLE)
+    {
+      my_error(ER_XAER_RMFAIL, MYF(0),
+               xa_state_names[thd->transaction.xa_state]);
+      break;
+    }
+    if (!thd->transaction.xid.eq(&thd->lex->ident))
+    {
+      my_error(ER_XAER_NOTA, MYF(0));
+      break;
+    }
+    if (ha_prepare(thd))
+    {
+      my_error(ER_XA_RBROLLBACK, MYF(0));
+      thd->transaction.xa_state=XA_NOTR;
+      break;
+    }
+    res=TRUE;
+    thd->transaction.xa_state=XA_PREPARED;
+    send_ok(thd);
+    break;
+  case SQLCOM_XA_COMMIT:
+    if (!thd->transaction.xid.eq(&thd->lex->ident))
+    {
+      if (!(res= !ha_commit_or_rollback_by_xid(&thd->lex->ident, 1)))
+        my_error(ER_XAER_NOTA, MYF(0));
+      break;
+    }
+    if (thd->transaction.xa_state == XA_IDLE && thd->lex->xa_opt == XA_ONE_PHASE)
+    {
+      int r;
+      if ((r= ha_commit(thd)))
+        my_error(r == 1 ? ER_XA_RBROLLBACK : ER_XAER_RMERR, MYF(0));
+      else
+      {
+        send_ok(thd);
+        res= TRUE;
+      }
+    }
+    else
+    if (thd->transaction.xa_state == XA_PREPARED && thd->lex->xa_opt == XA_NONE)
+    {
+      if (ha_commit_one_phase(thd, 1))
+        my_error(ER_XAER_RMERR, MYF(0));
+      else
+      {
+        send_ok(thd);
+        res= TRUE;
+      }
+    }
+    else
+    {
+      my_error(ER_XAER_RMFAIL, MYF(0),
+               xa_state_names[thd->transaction.xa_state]);
+      break;
+    }
+    thd->options&= ~(ulong) (OPTION_BEGIN | OPTION_STATUS_NO_TRANS_UPDATE);
+    thd->server_status&= ~SERVER_STATUS_IN_TRANS;
+    thd->transaction.xa_state=XA_NOTR;
+    break;
+  case SQLCOM_XA_ROLLBACK:
+    if (!thd->transaction.xid.eq(&thd->lex->ident))
+    {
+      if (!(res= !ha_commit_or_rollback_by_xid(&thd->lex->ident, 0)))
+        my_error(ER_XAER_NOTA, MYF(0));
+      break;
+    }
+    if (thd->transaction.xa_state != XA_IDLE &&
+        thd->transaction.xa_state != XA_PREPARED)
+    {
+      my_error(ER_XAER_RMFAIL, MYF(0),
+               xa_state_names[thd->transaction.xa_state]);
+      break;
+    }
+    if (ha_rollback(thd))
+      my_error(ER_XAER_RMERR, MYF(0));
+    else
+    {
+      send_ok(thd);
+      res= TRUE;
+    }
+    thd->options&= ~(ulong) (OPTION_BEGIN | OPTION_STATUS_NO_TRANS_UPDATE);
+    thd->server_status&= ~SERVER_STATUS_IN_TRANS;
+    thd->transaction.xa_state=XA_NOTR;
+    break;
+  case SQLCOM_XA_RECOVER:
+    res= !mysql_xa_recover(thd);
+    break;
+  default:
+    DBUG_ASSERT(0);                             /* Impossible */
     send_ok(thd);
     break;
   }
@@ -4326,22 +4516,10 @@ unsent_create_error:
     the statement is not DELETE, INSERT or UPDATE (or a CALL executing
     such a statement), but -1 is what JDBC and ODBC wants.
    */
-  switch (lex->sql_command) {
+  if (lex->sql_command != SQLCOM_CALL && uc_update_queries[lex->sql_command]<2)
-  case SQLCOM_UPDATE:
-  case SQLCOM_UPDATE_MULTI:
-  case SQLCOM_REPLACE:
-  case SQLCOM_INSERT:
-  case SQLCOM_REPLACE_SELECT:
-  case SQLCOM_INSERT_SELECT:
-  case SQLCOM_DELETE:
-  case SQLCOM_DELETE_MULTI:
-  case SQLCOM_CALL:
-    break;
-  default:
     thd->row_count_func= -1;
-  }
   goto cleanup;
-  
+
 error:
   res= 1;
 
@@ -5133,17 +5311,7 @@ bool add_field_to_list(THD *thd, char *field_name, enum_field_types type,
   new_field->charset=cs;
   new_field->geom_type= (Field::geometry_type) uint_geom_type;
 
-  if (!comment)
+  new_field->comment=*comment;
-  {
-    new_field->comment.str=0;
-    new_field->comment.length=0;
-  }
-  else
-  {
-    /* In this case comment is always of type Item_string */
-    new_field->comment.str=   (char*) comment->str;
-    new_field->comment.length=comment->length;
-  }
   /*
     Set flag if this field doesn't have a default value
     Enum values has always the first value as a default (set in

sql/sql_repl.cc

@@ -25,49 +25,34 @@ int max_binlog_dump_events = 0; // unlimited
 my_bool opt_sporadic_binlog_dump_fail = 0;
 static int binlog_dump_count = 0;
 
-int check_binlog_magic(IO_CACHE* log, const char** errmsg)
+/*
-{
-  char magic[4];
-  DBUG_ASSERT(my_b_tell(log) == 0);
-
-  if (my_b_read(log, (byte*) magic, sizeof(magic)))
-  {
-    *errmsg = "I/O error reading the header from the binary log";
-    sql_print_error("%s, errno=%d, io cache code=%d", *errmsg, my_errno,
-		    log->error);
-    return 1;
-  }
-  if (memcmp(magic, BINLOG_MAGIC, sizeof(magic)))
-  {
-    *errmsg = "Binlog has bad magic number;  It's not a binary log file that can be used by this version of MySQL";
-    return 1;
-  }
-  return 0;
-}
-
- /* 
     fake_rotate_event() builds a fake (=which does not exist physically in any
     binlog) Rotate event, which contains the name of the binlog we are going to
     send to the slave (because the slave may not know it if it just asked for
     MASTER_LOG_FILE='', MASTER_LOG_POS=4).
-    < 4.0.14, fake_rotate_event() was called only if the requested pos was
+    < 4.0.14, fake_rotate_event() was called only if the requested pos was 4.
-    4. After this version we always call it, so that a 3.23.58 slave can rely on
+    After this version we always call it, so that a 3.23.58 slave can rely on
     it to detect if the master is 4.0 (and stop) (the _fake_ Rotate event has
     zeros in the good positions which, by chance, make it possible for the 3.23
     slave to detect that this event is unexpected) (this is luck which happens
     because the master and slave disagree on the size of the header of
     Log_event).
- 
+
-    Relying on the event length of the Rotate event instead of these well-placed
+    Relying on the event length of the Rotate event instead of these
-    zeros was not possible as Rotate events have a variable-length part. 
+    well-placed zeros was not possible as Rotate events have a variable-length
+    part.
 */
 
 static int fake_rotate_event(NET* net, String* packet, char* log_file_name,
                              ulonglong position, const char** errmsg)
 {
   DBUG_ENTER("fake_rotate_event");
-  char header[LOG_EVENT_HEADER_LEN], buf[ROTATE_HEADER_LEN];
+  char header[LOG_EVENT_HEADER_LEN], buf[ROTATE_HEADER_LEN+100];
-  memset(header, 0, 4); // 'when' (the timestamp) does not matter, is set to 0
+  /*
+    'when' (the timestamp) is set to 0 so that slave could distinguish between
+    real and fake Rotate events (if necessary)
+  */
+  memset(header, 0, 4);
   header[EVENT_TYPE_OFFSET] = ROTATE_EVENT;
 
   char* p = log_file_name+dirname_length(log_file_name);
@@ -76,10 +61,10 @@ static int fake_rotate_event(NET* net, String* packet, char* log_file_name,
   int4store(header + SERVER_ID_OFFSET, server_id);
   int4store(header + EVENT_LEN_OFFSET, event_len);
   int2store(header + FLAGS_OFFSET, 0);
-  
+
   // TODO: check what problems this may cause and fix them
   int4store(header + LOG_POS_OFFSET, 0);
-  
+
   packet->append(header, sizeof(header));
   int8store(buf+R_POS_OFFSET,position);
   packet->append(buf, ROTATE_HEADER_LEN);
@@ -164,41 +149,6 @@ static int send_file(THD *thd)
 }
 
 
-File open_binlog(IO_CACHE *log, const char *log_file_name,
-		 const char **errmsg)
-{
-  File file;
-  DBUG_ENTER("open_binlog");
-
-  if ((file = my_open(log_file_name, O_RDONLY | O_BINARY, MYF(MY_WME))) < 0)
-  {
-    sql_print_error("Failed to open log (\
-file '%s', errno %d)", log_file_name, my_errno);
-    *errmsg = "Could not open log file";	// This will not be sent
-    goto err;
-  }
-  if (init_io_cache(log, file, IO_SIZE*2, READ_CACHE, 0, 0,
-		    MYF(MY_WME | MY_DONT_CHECK_FILESIZE)))
-  {
-    sql_print_error("Failed to create a cache on log (\
-file '%s')", log_file_name);
-    *errmsg = "Could not open log file";	// This will not be sent
-    goto err;
-  }
-  if (check_binlog_magic(log,errmsg))
-    goto err;
-  DBUG_RETURN(file);
-
-err:
-  if (file >= 0)
-  {
-    my_close(file,MYF(0));
-    end_io_cache(log);
-  }
-  DBUG_RETURN(-1);
-}
-
-
 /*
   Adjust the position pointer in the binary log file for all running slaves
 
@@ -330,7 +280,7 @@ bool purge_master_logs_before_date(THD* thd, time_t purge_time)
 
 int test_for_non_eof_log_read_errors(int error, const char **errmsg)
 {
-  if (error == LOG_READ_EOF) 
+  if (error == LOG_READ_EOF)
     return 0;
   my_errno= ER_MASTER_FATAL_ERROR_READING_BINLOG;
   switch (error) {
@@ -375,6 +325,7 @@ void mysql_binlog_send(THD* thd, char* log_ident, my_off_t pos,
   const char *errmsg = "Unknown error";
   NET* net = &thd->net;
   pthread_mutex_t *log_lock;
+  bool binlog_can_be_corrupted= FALSE;
 #ifndef DBUG_OFF
   int left_events = max_binlog_dump_events;
 #endif
@@ -442,37 +393,38 @@ impossible position";
 
   /*
     Tell the client about the log name with a fake Rotate event;
-    this is needed even if we also send a Format_description_log_event just
+    this is needed even if we also send a Format_description_log_event
-    after, because that event does not contain the binlog's name.
+    just after, because that event does not contain the binlog's name.
-    Note that as this Rotate event is sent before Format_description_log_event,
+    Note that as this Rotate event is sent before
-    the slave cannot have any info to understand this event's format, so the
+    Format_description_log_event, the slave cannot have any info to
-    header len of Rotate_log_event is FROZEN
+    understand this event's format, so the header len of
-    (so in 5.0 it will have a header shorter than other events except
+    Rotate_log_event is FROZEN (so in 5.0 it will have a header shorter
-    FORMAT_DESCRIPTION_EVENT). 
+    than other events except FORMAT_DESCRIPTION_EVENT).
-    Before 4.0.14 we called fake_rotate_event below only if 
+    Before 4.0.14 we called fake_rotate_event below only if (pos ==
-    (pos == BIN_LOG_HEADER_SIZE), because if this is false then the slave
+    BIN_LOG_HEADER_SIZE), because if this is false then the slave
     already knows the binlog's name.
-    Since, we always call fake_rotate_event; if the slave already knew the log's
+    Since, we always call fake_rotate_event; if the slave already knew
-    name (ex: CHANGE MASTER TO MASTER_LOG_FILE=...) this is useless but does
+    the log's name (ex: CHANGE MASTER TO MASTER_LOG_FILE=...) this is
-    not harm much. It is nice for 3.23 (>=.58) slaves which test Rotate events
+    useless but does not harm much. It is nice for 3.23 (>=.58) slaves
-    to see if the master is 4.0 (then they choose to stop because they can't
+    which test Rotate events to see if the master is 4.0 (then they
-    replicate 4.0); by always calling fake_rotate_event we are sure that
+    choose to stop because they can't replicate 4.0); by always calling
-    3.23.58 and newer will detect the problem as soon as replication starts
+    fake_rotate_event we are sure that 3.23.58 and newer will detect the
-    (BUG#198).
+    problem as soon as replication starts (BUG#198).
     Always calling fake_rotate_event makes sending of normal
-    (=from-binlog) Rotate events a priori unneeded, but it is not so simple:
+    (=from-binlog) Rotate events a priori unneeded, but it is not so
-    the 2 Rotate events are not equivalent, the normal one is before the Stop
+    simple: the 2 Rotate events are not equivalent, the normal one is
-    event, the fake one is after. If we don't send the normal one, then the
+    before the Stop event, the fake one is after. If we don't send the
-    Stop event will be interpreted (by existing 4.0 slaves) as "the master
+    normal one, then the Stop event will be interpreted (by existing 4.0
-    stopped", which is wrong. So for safety, given that we want minimum
+    slaves) as "the master stopped", which is wrong. So for safety,
-    modification of 4.0, we send the normal and fake Rotates.
+    given that we want minimum modification of 4.0, we send the normal
+    and fake Rotates.
   */
   if (fake_rotate_event(net, packet, log_file_name, pos, &errmsg))
   {
-    /* 
+    /*
-       This error code is not perfect, as fake_rotate_event() does not read
+       This error code is not perfect, as fake_rotate_event() does not
-       anything from the binlog; if it fails it's because of an error in
+       read anything from the binlog; if it fails it's because of an
-       my_net_write(), fortunately it will say it in errmsg. 
+       error in my_net_write(), fortunately it will say so in errmsg.
     */
     my_errno= ER_MASTER_FATAL_ERROR_READING_BINLOG;
     goto err;
@@ -480,30 +432,35 @@ impossible position";
   packet->set("\0", 1, &my_charset_bin);
 
   /*
-    We can set log_lock now, it does not move (it's a member of mysql_bin_log,
+    We can set log_lock now, it does not move (it's a member of
-    and it's already inited, and it will be destroyed only at shutdown).
+    mysql_bin_log, and it's already inited, and it will be destroyed
+    only at shutdown).
   */
-  log_lock = mysql_bin_log.get_log_lock();  
+  log_lock = mysql_bin_log.get_log_lock();
   if (pos > BIN_LOG_HEADER_SIZE)
-   {
+  {
-     /* Try to find a Format_description_log_event at the beginning of the binlog */
+     /*
+       Try to find a Format_description_log_event at the beginning of
+       the binlog
+     */
      if (!(error = Log_event::read_log_event(&log, packet, log_lock)))
      {
        /*
-         The packet has offsets equal to the normal offsets in a binlog event
+         The packet has offsets equal to the normal offsets in a binlog
-         +1 (the first character is \0).
+         event +1 (the first character is \0).
        */
        DBUG_PRINT("info",
                   ("Looked for a Format_description_log_event, found event type %d",
                    (*packet)[EVENT_TYPE_OFFSET+1]));
        if ((*packet)[EVENT_TYPE_OFFSET+1] == FORMAT_DESCRIPTION_EVENT)
        {
+         binlog_can_be_corrupted= (*packet)[FLAGS_OFFSET+1] & LOG_EVENT_BINLOG_IN_USE_F;
          /*
            mark that this event with "log_pos=0", so the slave
            should not increment master's binlog position
            (rli->group_master_log_pos)
          */
-         int4store(packet->c_ptr() +LOG_POS_OFFSET+1,0);
+         int4store(packet->c_ptr()+LOG_POS_OFFSET+1, 0);
          /* send it */
          if (my_net_write(net, (char*)packet->ptr(), packet->length()))
          {
@@ -512,24 +469,25 @@ impossible position";
            goto err;
          }
          /*
-           No need to save this event. We are only doing simple reads (no real
+           No need to save this event. We are only doing simple reads
-           parsing of the events) so we don't need it. And so we don't need the
+           (no real parsing of the events) so we don't need it. And so
-           artificial Format_description_log_event of 3.23&4.x.
+           we don't need the artificial Format_description_log_event of
+           3.23&4.x.
          */
        }
      }
      else
        if (test_for_non_eof_log_read_errors(error, &errmsg))
          goto err;
-     /* 
+     /*
         else: it's EOF, nothing to do, go on reading next events, the
         Format_description_log_event will be found naturally if it is written.
      */
      /* reset the packet as we wrote to it in any case */
      packet->set("\0", 1, &my_charset_bin);
-   } /* end of if (pos > BIN_LOG_HEADER_SIZE); if false, the Format_description_log_event
+   } /* end of if (pos > BIN_LOG_HEADER_SIZE); if false, the
-        event will be found naturally. */
+        Format_description_log_event event will be found naturally. */
-  
+
   /* seek to the requested position, to start the requested dump */
   my_b_seek(&log, pos);			// Seek will done on next read
 
@@ -546,6 +504,12 @@ impossible position";
 	goto err;
       }
 #endif
+
+      if ((*packet)[EVENT_TYPE_OFFSET+1] == FORMAT_DESCRIPTION_EVENT)
+        binlog_can_be_corrupted= (*packet)[FLAGS_OFFSET+1] & LOG_EVENT_BINLOG_IN_USE_F;
+      else if ((*packet)[EVENT_TYPE_OFFSET+1] == STOP_EVENT)
+        binlog_can_be_corrupted= FALSE;
+
       if (my_net_write(net, (char*)packet->ptr(), packet->length()))
       {
 	errmsg = "Failed on my_net_write()";
@@ -565,19 +529,25 @@ impossible position";
       }
       packet->set("\0", 1, &my_charset_bin);
     }
+
+    /*
+      here we were reading binlog that was not closed properly (as a result
+      of a crash ?). treat any corruption as EOF
+    */
+    if (binlog_can_be_corrupted && error != LOG_READ_MEM)
+      error=LOG_READ_EOF;
     /*
       TODO: now that we are logging the offset, check to make sure
       the recorded offset and the actual match.
-      Guilhem 2003-06: this is not true if this master is a slave <4.0.15
+      Guilhem 2003-06: this is not true if this master is a slave
-      running with --log-slave-updates, because then log_pos may be the offset
+      <4.0.15 running with --log-slave-updates, because then log_pos may
-      in the-master-of-this-master's binlog.
+      be the offset in the-master-of-this-master's binlog.
     */
-
     if (test_for_non_eof_log_read_errors(error, &errmsg))
       goto err;
 
     if (!(flags & BINLOG_DUMP_NON_BLOCK) &&
-       mysql_bin_log.is_active(log_file_name))
+        mysql_bin_log.is_active(log_file_name))
     {
       /*
 	Block until there is more data in the log
@@ -613,9 +583,9 @@ impossible position";
 	  now, but we'll be quick and just read one record
 
 	  TODO:
-	  Add an counter that is incremented for each time we update
+          Add an counter that is incremented for each time we update the
-	  the binary log.  We can avoid the following read if the counter
+          binary log.  We can avoid the following read if the counter
-	  has not been updated since last read.
+          has not been updated since last read.
 	*/
 
 	pthread_mutex_lock(log_lock);
@@ -708,16 +678,17 @@ impossible position";
       (void) my_close(file, MYF(MY_WME));
 
       /*
-        Call fake_rotate_event() in case the previous log (the one which we have
+        Call fake_rotate_event() in case the previous log (the one which
-        just finished reading) did not contain a Rotate event (for example (I
+        we have just finished reading) did not contain a Rotate event
-        don't know any other example) the previous log was the last one before
+        (for example (I don't know any other example) the previous log
-        the master was shutdown & restarted).
+        was the last one before the master was shutdown & restarted).
-        This way we tell the slave about the new log's name and position.
+        This way we tell the slave about the new log's name and
-        If the binlog is 5.0, the next event we are going to read and send is
+        position.  If the binlog is 5.0, the next event we are going to
-        Format_description_log_event.
+        read and send is Format_description_log_event.
       */
       if ((file=open_binlog(&log, log_file_name, &errmsg)) < 0 ||
-	  fake_rotate_event(net, packet, log_file_name, BIN_LOG_HEADER_SIZE, &errmsg))
+	  fake_rotate_event(net, packet, log_file_name, BIN_LOG_HEADER_SIZE,
+                            &errmsg))
       {
 	my_errno= ER_MASTER_FATAL_ERROR_READING_BINLOG;
 	goto err;
@@ -762,17 +733,17 @@ int start_slave(THD* thd , MASTER_INFO* mi,  bool net_report)
   int slave_errno= 0;
   int thread_mask;
   DBUG_ENTER("start_slave");
-  
+
   if (check_access(thd, SUPER_ACL, any_db,0,0,0))
     DBUG_RETURN(1);
   lock_slave_threads(mi);  // this allows us to cleanly read slave_running
   // Get a mask of _stopped_ threads
   init_thread_mask(&thread_mask,mi,1 /* inverse */);
   /*
-    Below we will start all stopped threads.
+    Below we will start all stopped threads.  But if the user wants to
-    But if the user wants to start only one thread, do as if the other thread
+    start only one thread, do as if the other thread was running (as we
-    was running (as we don't wan't to touch the other thread), so set the
+    don't wan't to touch the other thread), so set the bit to 0 for the
-    bit to 0 for the other thread
+    other thread
   */
   if (thd->lex->slave_thd_opt)
     thread_mask&= thd->lex->slave_thd_opt;
@@ -783,9 +754,9 @@ int start_slave(THD* thd , MASTER_INFO* mi,  bool net_report)
       slave_errno=ER_MASTER_INFO;
     else if (server_id_supplied && *mi->host)
     {
-      /* 
+      /*
-         If we will start SQL thread we will care about UNTIL options 
+         If we will start SQL thread we will care about UNTIL options If
-         If not and they are specified we will ignore them and warn user 
+         not and they are specified we will ignore them and warn user
          about this fact.
       */
       if (thread_mask & SLAVE_SQL)
@@ -796,13 +767,13 @@ int start_slave(THD* thd , MASTER_INFO* mi,  bool net_report)
         {
           mi->rli.until_condition= RELAY_LOG_INFO::UNTIL_MASTER_POS;
           mi->rli.until_log_pos= thd->lex->mi.pos;
-          /* 
+          /*
-             We don't check thd->lex->mi.log_file_name for NULL here 
+             We don't check thd->lex->mi.log_file_name for NULL here
              since it is checked in sql_yacc.yy
           */
           strmake(mi->rli.until_log_name, thd->lex->mi.log_file_name,
               sizeof(mi->rli.until_log_name)-1);
-        } 
+        }
         else if (thd->lex->mi.relay_log_pos)
         {
           mi->rli.until_condition= RELAY_LOG_INFO::UNTIL_RELAY_POS;
@@ -826,15 +797,15 @@ int start_slave(THD* thd , MASTER_INFO* mi,  bool net_report)
               p_end points to the first invalid character. If it equals
               to p, no digits were found, error. If it contains '\0' it
               means  conversion went ok.
-            */ 
+            */
             if (p_end==p || *p_end)
               slave_errno=ER_BAD_SLAVE_UNTIL_COND;
           }
           else
             slave_errno=ER_BAD_SLAVE_UNTIL_COND;
-          
+
           /* mark the cached result of the UNTIL comparison as "undefined" */
-          mi->rli.until_log_names_cmp_result= 
+          mi->rli.until_log_names_cmp_result=
             RELAY_LOG_INFO::UNTIL_LOG_NAMES_CMP_UNKNOWN;
 
           /* Issuing warning then started without --skip-slave-start */
@@ -842,14 +813,13 @@ int start_slave(THD* thd , MASTER_INFO* mi,  bool net_report)
             push_warning(thd, MYSQL_ERROR::WARN_LEVEL_NOTE, ER_MISSING_SKIP_SLAVE, 
                          ER(ER_MISSING_SKIP_SLAVE));
         }
-        
+
         pthread_mutex_unlock(&mi->rli.data_lock);
       }
       else if (thd->lex->mi.pos || thd->lex->mi.relay_log_pos)
         push_warning(thd, MYSQL_ERROR::WARN_LEVEL_NOTE, ER_UNTIL_COND_IGNORED,
             ER(ER_UNTIL_COND_IGNORED));
-        
+
-      
       if (!slave_errno)
         slave_errno = start_slave_threads(0 /*no mutex */,
 					1 /* wait for start */,
@@ -864,9 +834,9 @@ int start_slave(THD* thd , MASTER_INFO* mi,  bool net_report)
     //no error if all threads are already started, only a warning
     push_warning(thd, MYSQL_ERROR::WARN_LEVEL_NOTE, ER_SLAVE_WAS_RUNNING,
                  ER(ER_SLAVE_WAS_RUNNING));
-  
+
   unlock_slave_threads(mi);
-  
+
   if (slave_errno)
   {
     if (net_report)
@@ -966,7 +936,7 @@ int reset_slave(THD *thd, MASTER_INFO* mi)
 			       1 /* just reset */,
 			       &errmsg)))
     goto err;
-  
+
   /*
     Clear master's log coordinates and reset host/user/etc to the values
     specified in mysqld's options (only for good display of SHOW SLAVE STATUS;
@@ -975,13 +945,13 @@ int reset_slave(THD *thd, MASTER_INFO* mi)
     STATUS; before doing START SLAVE;
   */
   init_master_info_with_options(mi);
-  /* 
+  /*
      Reset errors (the idea is that we forget about the
      old master).
   */
   clear_slave_error(&mi->rli);
   clear_until_condition(&mi->rli);
-  
+
   // close master_info_file, relay_log_info_file, set mi->inited=rli->inited=0
   end_master_info(mi);
   // and delete these two files
@@ -1288,7 +1258,7 @@ int cmp_master_pos(const char* log_file_name1, ulonglong log_pos1,
 }
 
 
-bool show_binlog_events(THD* thd)
+bool mysql_show_binlog_events(THD* thd)
 {
   Protocol *protocol= thd->protocol;
   DBUG_ENTER("show_binlog_events");
@@ -1297,7 +1267,7 @@ bool show_binlog_events(THD* thd)
   IO_CACHE log;
   File file = -1;
   Format_description_log_event *description_event= new
-    Format_description_log_event(3); /* MySQL 4.0 by default */ 
+    Format_description_log_event(3); /* MySQL 4.0 by default */
 
   Log_event::init_show_field_list(&field_list);
   if (protocol->send_fields(&field_list,
@@ -1314,7 +1284,7 @@ bool show_binlog_events(THD* thd)
     pthread_mutex_t *log_lock = mysql_bin_log.get_log_lock();
     LOG_INFO linfo;
     Log_event* ev;
-  
+
     limit_start= thd->lex->current_select->offset_limit;
     limit_end= thd->lex->current_select->select_limit + limit_start;
 
@@ -1338,15 +1308,15 @@ bool show_binlog_events(THD* thd)
 
     pthread_mutex_lock(log_lock);
 
-    /* 
+    /*
        open_binlog() sought to position 4.
-       Read the first event in case it's a Format_description_log_event, to know the
+       Read the first event in case it's a Format_description_log_event, to
-       format. If there's no such event, we are 3.23 or 4.x. This code, like
+       know the format. If there's no such event, we are 3.23 or 4.x. This
-       before, can't read 3.23 binlogs.
+       code, like before, can't read 3.23 binlogs.
        This code will fail on a mixed relay log (one which has Format_desc then
        Rotate then Format_desc).
     */
-    
+
     ev = Log_event::read_log_event(&log,(pthread_mutex_t*)0,description_event);
     if (ev)
     {
@@ -1366,7 +1336,7 @@ bool show_binlog_events(THD* thd)
       errmsg="Invalid Format_description event; could be out of memory";
       goto err;
     }
-    
+
     for (event_count = 0;
 	 (ev = Log_event::read_log_event(&log,(pthread_mutex_t*)0,description_event)); )
     {

sql/sql_repl.h

@@ -36,15 +36,16 @@ extern I_List<i_string> binlog_do_db, binlog_ignore_db;
 extern int max_binlog_dump_events;
 extern my_bool opt_sporadic_binlog_dump_fail;
 
-#define KICK_SLAVE(thd) { pthread_mutex_lock(&(thd)->LOCK_delete); (thd)->awake(THD::NOT_KILLED); pthread_mutex_unlock(&(thd)->LOCK_delete); }
+#define KICK_SLAVE(thd) do {                                            \
-
+                          pthread_mutex_lock(&(thd)->LOCK_delete);      \
-File open_binlog(IO_CACHE *log, const char *log_file_name,
+                          (thd)->awake(THD::NOT_KILLED);                \
-		 const char **errmsg);
+                          pthread_mutex_unlock(&(thd)->LOCK_delete);    \
+                        } while(0)
 
 int start_slave(THD* thd, MASTER_INFO* mi, bool net_report);
 int stop_slave(THD* thd, MASTER_INFO* mi, bool net_report);
 bool change_master(THD* thd, MASTER_INFO* mi);
-bool show_binlog_events(THD* thd);
+bool mysql_show_binlog_events(THD* thd);
 int cmp_master_pos(const char* log_file_name1, ulonglong log_pos1,
 		   const char* log_file_name2, ulonglong log_pos2);
 int reset_slave(THD *thd, MASTER_INFO* mi);

sql/sql_table.cc

@@ -14,7 +14,6 @@
    along with this program; if not, write to the Free Software
    Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA */
 
-
 /* drop and alter of tables */
 
 #include "mysql_priv.h"
@@ -275,9 +274,7 @@ int mysql_rm_table_part2(THD *thd, TABLE_LIST *tables, bool if_exists,
     {
       if (!error)
         thd->clear_error();
-      Query_log_event qinfo(thd, thd->query, thd->query_length,
+      Query_log_event qinfo(thd, thd->query, thd->query_length, FALSE, FALSE);
-                            tmp_table_deleted && !some_tables_deleted,
-			    FALSE);
       mysql_bin_log.write(&qinfo);
     }
   }
@@ -1297,7 +1294,7 @@ int mysql_prepare_table(THD *thd, HA_CREATE_INFO *create_info,
     create_info		Create information (like MAX_ROWS)
     fields		List of fields to create
     keys		List of keys to create
-    tmp_table		Set to 1 if this is an internal temporary table
+    internal_tmp_table  Set to 1 if this is an internal temporary table
 			(From ALTER TABLE)
 
   DESCRIPTION
@@ -1316,7 +1313,7 @@ int mysql_prepare_table(THD *thd, HA_CREATE_INFO *create_info,
 bool mysql_create_table(THD *thd,const char *db, const char *table_name,
                         HA_CREATE_INFO *create_info,
                         List<create_field> &fields,
-                        List<Key> &keys,bool tmp_table,
+                        List<Key> &keys,bool internal_tmp_table,
                         uint select_field_count)
 {
   char		path[FN_REFLEN];
@@ -1385,7 +1382,7 @@ bool mysql_create_table(THD *thd,const char *db, const char *table_name,
   }
 
   if (mysql_prepare_table(thd, create_info, fields,
-			  keys, tmp_table, db_options, file,
+			  keys, internal_tmp_table, db_options, file,
 			  key_info_buffer, &key_count,
 			  select_field_count))
     DBUG_RETURN(TRUE);
@@ -1419,7 +1416,7 @@ bool mysql_create_table(THD *thd,const char *db, const char *table_name,
   if (wait_if_global_read_lock(thd, 0, 1))
     DBUG_RETURN(error);
   VOID(pthread_mutex_lock(&LOCK_open));
-  if (!tmp_table && !(create_info->options & HA_LEX_CREATE_TMP_TABLE))
+  if (!internal_tmp_table && !(create_info->options & HA_LEX_CREATE_TMP_TABLE))
   {
     if (!access(path,F_OK))
     {
@@ -1486,13 +1483,10 @@ bool mysql_create_table(THD *thd,const char *db, const char *table_name,
     }
     thd->tmp_table_used= 1;
   }
-  if (!tmp_table && mysql_bin_log.is_open())
+  if (!internal_tmp_table && mysql_bin_log.is_open())
   {
     thd->clear_error();
-    Query_log_event qinfo(thd, thd->query, thd->query_length,
+    Query_log_event qinfo(thd, thd->query, thd->query_length, FALSE, FALSE);
-                          test(create_info->options &
-                               HA_LEX_CREATE_TMP_TABLE),
-			  FALSE);
     mysql_bin_log.write(&qinfo);
   }
   error= FALSE;
@@ -2473,10 +2467,7 @@ bool mysql_create_like_table(THD* thd, TABLE_LIST* table,
   if (mysql_bin_log.is_open())
   {
     thd->clear_error();
-    Query_log_event qinfo(thd, thd->query, thd->query_length,
+    Query_log_event qinfo(thd, thd->query, thd->query_length, FALSE, FALSE);
-			  test(create_info->options &
-			       HA_LEX_CREATE_TMP_TABLE), 
-			  FALSE);
     mysql_bin_log.write(&qinfo);
   }
   res= FALSE;
@@ -2586,7 +2577,7 @@ mysql_discard_or_import_tablespace(THD *thd,
     goto err;
   if (mysql_bin_log.is_open())
   {
-    Query_log_event qinfo(thd, thd->query, thd->query_length, 0, FALSE);
+    Query_log_event qinfo(thd, thd->query, thd->query_length, FALSE, FALSE);
     mysql_bin_log.write(&qinfo);
   }
 err:
@@ -2979,7 +2970,7 @@ bool mysql_alter_table(THD *thd,char *new_db, char *new_name,
       if (mysql_bin_log.is_open())
       {
 	thd->clear_error();
-	Query_log_event qinfo(thd, thd->query, thd->query_length, 0, FALSE);
+	Query_log_event qinfo(thd, thd->query, thd->query_length, FALSE, FALSE);
 	mysql_bin_log.write(&qinfo);
       }
       if (do_send_ok)
@@ -3396,7 +3387,7 @@ bool mysql_alter_table(THD *thd,char *new_db, char *new_name,
     if (mysql_bin_log.is_open())
     {
       thd->clear_error();
-      Query_log_event qinfo(thd, thd->query, thd->query_length, 0, FALSE);
+      Query_log_event qinfo(thd, thd->query, thd->query_length, FALSE, FALSE);
       mysql_bin_log.write(&qinfo);
     }
     goto end_temporary;
@@ -3530,7 +3521,7 @@ bool mysql_alter_table(THD *thd,char *new_db, char *new_name,
   if (mysql_bin_log.is_open())
   {
     thd->clear_error();
-    Query_log_event qinfo(thd, thd->query, thd->query_length, 0, FALSE);
+    Query_log_event qinfo(thd, thd->query, thd->query_length, FALSE, FALSE);
     mysql_bin_log.write(&qinfo);
   }
   VOID(pthread_cond_broadcast(&COND_refresh));

sql/sql_trigger.cc

@@ -147,7 +147,7 @@ bool Table_triggers_list::create_trigger(THD *thd, TABLE_LIST *tables)
   /* Let us check if trigger with the same name exists */
   while ((name= it++))
   {
-    if (my_strcasecmp(system_charset_info, lex->name_and_length.str,
+    if (my_strcasecmp(system_charset_info, lex->ident.str,
                       name->str) == 0)
     {
       my_message(ER_TRG_ALREADY_EXISTS, ER(ER_TRG_ALREADY_EXISTS), MYF(0));
@@ -238,7 +238,7 @@ bool Table_triggers_list::drop_trigger(THD *thd, TABLE_LIST *tables)
   {
     it_def++;
 
-    if (my_strcasecmp(system_charset_info, lex->name_and_length.str,
+    if (my_strcasecmp(system_charset_info, lex->ident.str,
                       name->str) == 0)
     {
       /*
@@ -428,15 +428,15 @@ bool Table_triggers_list::check_n_load(THD *thd, const char *db,
 
         if (!(trg_name_buff= alloc_root(&table->mem_root,
                                         sizeof(LEX_STRING) +
-                                        lex.name_and_length.length + 1)))
+                                        lex.ident.length + 1)))
           goto err_with_lex_cleanup;
 
         trg_name_str= (LEX_STRING *)trg_name_buff;
         trg_name_buff+= sizeof(LEX_STRING);
-        memcpy(trg_name_buff, lex.name_and_length.str,
+        memcpy(trg_name_buff, lex.ident.str,
-               lex.name_and_length.length + 1);
+               lex.ident.length + 1);
         trg_name_str->str= trg_name_buff;
-        trg_name_str->length= lex.name_and_length.length;
+        trg_name_str->length= lex.ident.length;
 
         if (triggers->names_list.push_back(trg_name_str, &table->mem_root))
           goto err_with_lex_cleanup;

sql/sql_update.cc

@@ -117,7 +117,7 @@ int mysql_update(THD *thd,
 {
   bool		using_limit= limit != HA_POS_ERROR;
   bool		safe_update= thd->options & OPTION_SAFE_UPDATES;
-  bool		used_key_is_modified, transactional_table, log_delayed;
+  bool		used_key_is_modified, transactional_table;
   int           res;
   int		error=0;
   uint		used_index;
@@ -469,7 +469,6 @@ int mysql_update(THD *thd,
     query_cache_invalidate3(thd, table_list, 1);
   }
 
-  log_delayed= (transactional_table || table->s->tmp_table);
   if ((updated || (error < 0)) && (error <= 0 || !transactional_table))
   {
     if (mysql_bin_log.is_open())
@@ -477,11 +476,11 @@ int mysql_update(THD *thd,
       if (error <= 0)
         thd->clear_error();
       Query_log_event qinfo(thd, thd->query, thd->query_length,
-			    log_delayed, FALSE);
+			    transactional_table, FALSE);
       if (mysql_bin_log.write(&qinfo) && transactional_table)
 	error=1;				// Rollback update
     }
-    if (!log_delayed)
+    if (!transactional_table)
       thd->options|=OPTION_STATUS_NO_TRANS_UPDATE;
   }
   if (transactional_table)
@@ -1000,7 +999,6 @@ multi_update::initialize_tables(JOIN *join)
     DBUG_RETURN(1);
   main_table=join->join_tab->table;
   trans_safe= transactional_tables= main_table->file->has_transactions();
-  log_delayed= trans_safe || main_table->s->tmp_table != NO_TMP_TABLE;
   table_to_update= 0;
 
   /* Create a temporary table for keys to all tables, except main table */
@@ -1335,17 +1333,13 @@ int multi_update::do_updates(bool from_send_error)
 	    goto err;
 	}
 	updated++;
-	if (table->s->tmp_table != NO_TMP_TABLE)
-	  log_delayed= 1;
       }
     }
 
     if (updated != org_updated)
     {
-      if (table->s->tmp_table != NO_TMP_TABLE)
-	log_delayed= 1;				// Tmp tables forces delay log
       if (table->file->has_transactions())
-	log_delayed= transactional_tables= 1;
+	transactional_tables= 1;
       else
 	trans_safe= 0;				// Can't do safe rollback
     }
@@ -1366,10 +1360,8 @@ err:
 
   if (updated != org_updated)
   {
-    if (table->s->tmp_table != NO_TMP_TABLE)
-      log_delayed= 1;
     if (table->file->has_transactions())
-      log_delayed= transactional_tables= 1;
+      transactional_tables= 1;
     else
       trans_safe= 0;
   }
@@ -1411,11 +1403,11 @@ bool multi_update::send_eof()
       if (local_error <= 0)
         thd->clear_error();
       Query_log_event qinfo(thd, thd->query, thd->query_length,
-			    log_delayed, FALSE);
+			    transactional_tables, FALSE);
       if (mysql_bin_log.write(&qinfo) && trans_safe)
 	local_error= 1;				// Rollback update
     }
-    if (!log_delayed)
+    if (!transactional_tables)
       thd->options|=OPTION_STATUS_NO_TRANS_UPDATE;
   }