Merge from mysql-5.5-innodb to mysql-5.5-bugteam

No conflicts

mysql-test/suite/innodb/r/innodb_bug56680.result

+SET GLOBAL tx_isolation='REPEATABLE-READ';
+SET GLOBAL innodb_file_format=Barracuda;
+SET GLOBAL innodb_file_per_table=on;
+CREATE TABLE bug56680(
+a INT AUTO_INCREMENT PRIMARY KEY,
+b CHAR(1),
+c INT,
+INDEX(b))
+ENGINE=InnoDB;
+INSERT INTO bug56680 VALUES(0,'x',1);
+BEGIN;
+SELECT b FROM bug56680;
+b
+x
+BEGIN;
+UPDATE bug56680 SET b='X';
+SELECT b FROM bug56680;
+b
+x
+SELECT * FROM bug56680;
+a	b	c
+1	x	1
+ROLLBACK;
+SELECT b FROM bug56680;
+b
+x
+SET GLOBAL tx_isolation='READ-UNCOMMITTED';
+INSERT INTO bug56680 SELECT 0,b,c FROM bug56680;
+INSERT INTO bug56680 SELECT 0,b,c FROM bug56680;
+INSERT INTO bug56680 SELECT 0,b,c FROM bug56680;
+INSERT INTO bug56680 SELECT 0,b,c FROM bug56680;
+INSERT INTO bug56680 SELECT 0,b,c FROM bug56680;
+INSERT INTO bug56680 SELECT 0,b,c FROM bug56680;
+INSERT INTO bug56680 SELECT 0,b,c FROM bug56680;
+INSERT INTO bug56680 SELECT 0,b,c FROM bug56680;
+INSERT INTO bug56680 SELECT 0,b,c FROM bug56680;
+INSERT INTO bug56680 SELECT 0,b,c FROM bug56680;
+INSERT INTO bug56680 SELECT 0,b,c FROM bug56680;
+BEGIN;
+SELECT b FROM bug56680 LIMIT 2;
+b
+x
+x
+BEGIN;
+DELETE FROM bug56680 WHERE a=1;
+INSERT INTO bug56680 VALUES(1,'X',1);
+SELECT b FROM bug56680 LIMIT 3;
+b
+X
+x
+x
+SELECT b FROM bug56680 LIMIT 2;
+b
+x
+x
+CHECK TABLE bug56680;
+Table	Op	Msg_type	Msg_text
+test.bug56680	check	status	OK
+ROLLBACK;
+SELECT b FROM bug56680 LIMIT 2;
+b
+x
+x
+CHECK TABLE bug56680;
+Table	Op	Msg_type	Msg_text
+test.bug56680	check	status	OK
+SELECT b FROM bug56680 LIMIT 2;
+b
+x
+x
+CREATE TABLE bug56680_2(
+a INT AUTO_INCREMENT PRIMARY KEY,
+b VARCHAR(2) CHARSET latin1 COLLATE latin1_german2_ci,
+c INT,
+INDEX(b))
+ENGINE=InnoDB;
+INSERT INTO bug56680_2 SELECT 0,_latin1 0xdf,c FROM bug56680;
+BEGIN;
+SELECT HEX(b) FROM bug56680_2 LIMIT 2;
+HEX(b)
+DF
+DF
+DELETE FROM bug56680_2 WHERE a=1;
+INSERT INTO bug56680_2 VALUES(1,'SS',1);
+SELECT HEX(b) FROM bug56680_2 LIMIT 3;
+HEX(b)
+5353
+DF
+DF
+CHECK TABLE bug56680_2;
+Table	Op	Msg_type	Msg_text
+test.bug56680_2	check	status	OK
+ALTER TABLE bug56680_2 ROW_FORMAT=COMPRESSED KEY_BLOCK_SIZE=1;
+SELECT HEX(b) FROM bug56680_2 LIMIT 2;
+HEX(b)
+5353
+DF
+DELETE FROM bug56680_2 WHERE a=1;
+INSERT INTO bug56680_2 VALUES(1,_latin1 0xdf,1);
+SELECT HEX(b) FROM bug56680_2 LIMIT 3;
+HEX(b)
+DF
+DF
+DF
+CHECK TABLE bug56680_2;
+Table	Op	Msg_type	Msg_text
+test.bug56680_2	check	status	OK
+DROP TABLE bug56680_2;
+DROP TABLE bug56680;

mysql-test/suite/innodb/r/innodb_bug57252.result

+cardinality
+10
+Table	Op	Msg_type	Msg_text
+test.bug57252	analyze	status	OK
+cardinality
+10

mysql-test/suite/innodb/t/innodb_bug56632.test

+#
+# Bug#56632: ALTER TABLE implicitly changes ROW_FORMAT to COMPRESSED
+# http://bugs.mysql.com/56632
+#
+# Innodb automatically uses compressed mode when the KEY_BLOCK_SIZE
+# parameter is used, except if the ROW_FORMAT is also specified, in
+# which case the KEY_BLOCK_SIZE is ignored and a warning is shown.
+# But Innodb was getting confused when neither of those parameters
+# was used on the ALTER statement after they were both used on the
+# CREATE.
+#
+# This will test the results of all 4 combinations of these two
+# parameters of the CREATE and ALTER.
+#
+#    Tests 1-5 use INNODB_STRICT_MODE=1 which returns an error
+#    if there is anything wrong with the statement.
+#
+# 1) CREATE with ROW_FORMAT=COMPACT & KEY_BLOCK_SIZE=1, ALTER with neither.
+#    Result; CREATE; fails with error ER_CANT_CREATE_TABLE
+# 2) CREATE with ROW_FORMAT=COMPACT, ALTER with KEY_BLOCK_SIZE=1
+#    Result; CREATE succeeds,
+#            ALTER quietly converts ROW_FORMAT to compressed.
+# 3) CREATE with KEY_BLOCK_SIZE=1, ALTER with ROW_FORMAT=COMPACT
+#    Result; CREATE quietly converts ROW_FORMAT to compressed,
+#            ALTER fails with error ER_CANT_CREATE_TABLE.
+# 4) CREATE with neither, ALTER with ROW_FORMAT=COMPACT & KEY_BLOCK_SIZE=1
+#    Result; CREATE succeeds,
+#            ALTER; fails with error ER_CANT_CREATE_TABLE
+# 5) CREATE with KEY_BLOCK_SIZE=3 (invalid), ALTER with neither.
+#    Result; CREATE; fails with error ER_CANT_CREATE_TABLE
+#
+#    Tests 6-11 use INNODB_STRICT_MODE=0 which automatically makes
+#    adjustments if the prameters are incompatible.
+#
+# 6) CREATE with ROW_FORMAT=COMPACT & KEY_BLOCK_SIZE=1, ALTER with neither.
+#    Result; CREATE succeeds, warns that KEY_BLOCK_SIZE is ignored.
+#            ALTER succeeds, no warnings.
+# 7) CREATE with ROW_FORMAT=COMPACT, ALTER with KEY_BLOCK_SIZE=1
+#    Result; CREATE succeeds,
+#            ALTER quietly converts ROW_FORMAT to compressed.
+# 8) CREATE with KEY_BLOCK_SIZE=1, ALTER with ROW_FORMAT=COMPACT
+#    Result; CREATE quietly converts ROW_FORMAT to compressed,
+#            ALTER succeeds, warns that KEY_BLOCK_SIZE is ignored.
+# 9) CREATE with neither, ALTER with ROW_FORMAT=COMPACT & KEY_BLOCK_SIZE=1
+#    Result; CREATE succeeds,
+#            ALTER succeeds, warns that KEY_BLOCK_SIZE is ignored.
+# 10) CREATE with KEY_BLOCK_SIZE=3 (invalid), ALTER with neither.
+#    Result; CREATE succeeds, warns that KEY_BLOCK_SIZE=3 is ignored.
+#            ALTER succeeds, warns that KEY_BLOCK_SIZE=3 is ignored.
+# 11) CREATE with KEY_BLOCK_SIZE=3 (invalid), ALTER with ROW_FORMAT=COMPACT.
+#    Result; CREATE succeeds, warns that KEY_BLOCK_SIZE=3 is ignored.
+#            ALTER succeeds, warns that KEY_BLOCK_SIZE=3 is ignored.
+# 12) CREATE with KEY_BLOCK_SIZE=3 (invalid), ALTER with KEY_BLOCK_SIZE=1.
+#    Result; CREATE succeeds, warns that KEY_BLOCK_SIZE=3 is ignored.
+#            ALTER succeeds, quietly converts ROW_FORMAT to compressed.
+
+-- source include/have_innodb.inc
+
+SET storage_engine=InnoDB;
+
+--disable_query_log
+# These values can change during the test
+LET $innodb_file_format_orig=`select @@innodb_file_format`;
+LET $innodb_file_format_max_orig=`select @@innodb_file_format_max`;
+LET $innodb_file_per_table_orig=`select @@innodb_file_per_table`;
+LET $innodb_strict_mode_orig=`select @@session.innodb_strict_mode`;
+--enable_query_log
+
+SET GLOBAL innodb_file_format=`Barracuda`;
+SET GLOBAL innodb_file_per_table=ON;
+
+# Innodb strict mode will cause an error on the CREATE or ALTER when;
+# 1. both ROW_FORMAT=COMPACT and KEY_BLOCK_SIZE=1,
+# 2. KEY_BLOCK_SIZE is not a valid number (0,1,2,4,8,16).
+# With innodb_strict_mode = OFF, These errors are corrected
+# and just a warning is returned.
+SET SESSION innodb_strict_mode = ON;
+
+--echo # Test 1) CREATE with ROW_FORMAT & KEY_BLOCK_SIZE, ALTER with neither
+DROP TABLE IF EXISTS bug56632;
+--error ER_CANT_CREATE_TABLE
+CREATE TABLE bug56632 ( i INT ) ROW_FORMAT=COMPACT KEY_BLOCK_SIZE=1;
+SHOW WARNINGS;
+
+--echo # Test 2) CREATE with ROW_FORMAT, ALTER with KEY_BLOCK_SIZE
+DROP TABLE IF EXISTS bug56632;
+CREATE TABLE bug56632 ( i INT ) ROW_FORMAT=COMPACT;
+SHOW WARNINGS;
+SHOW CREATE TABLE bug56632;
+SELECT TABLE_NAME,ROW_FORMAT,CREATE_OPTIONS FROM information_schema.tables WHERE TABLE_NAME = 'bug56632';
+ALTER TABLE bug56632 KEY_BLOCK_SIZE=1;
+SHOW WARNINGS;
+SHOW CREATE TABLE bug56632;
+SELECT TABLE_NAME,ROW_FORMAT,CREATE_OPTIONS FROM information_schema.tables WHERE TABLE_NAME = 'bug56632';
+
+--echo # Test 3) CREATE with KEY_BLOCK_SIZE, ALTER with ROW_FORMAT
+DROP TABLE IF EXISTS bug56632;
+CREATE TABLE bug56632 ( i INT ) KEY_BLOCK_SIZE=1;
+SHOW WARNINGS;
+SHOW CREATE TABLE bug56632;
+SELECT TABLE_NAME,ROW_FORMAT,CREATE_OPTIONS FROM information_schema.tables WHERE TABLE_NAME = 'bug56632';
+--disable_result_log
+--error ER_CANT_CREATE_TABLE
+ALTER TABLE bug56632 ROW_FORMAT=COMPACT;
+--enable_result_log
+SHOW CREATE TABLE bug56632;
+SELECT TABLE_NAME,ROW_FORMAT,CREATE_OPTIONS FROM information_schema.tables WHERE TABLE_NAME = 'bug56632';
+
+--echo # Test 4) CREATE with neither, ALTER with ROW_FORMAT & KEY_BLOCK_SIZE
+DROP TABLE IF EXISTS bug56632;
+CREATE TABLE bug56632 ( i INT );
+SHOW WARNINGS;
+SHOW CREATE TABLE bug56632;
+SELECT TABLE_NAME,ROW_FORMAT,CREATE_OPTIONS FROM information_schema.tables WHERE TABLE_NAME = 'bug56632';
+--disable_result_log
+--error ER_CANT_CREATE_TABLE
+ALTER TABLE bug56632 ROW_FORMAT=COMPACT KEY_BLOCK_SIZE=1;
+--enable_result_log
+SHOW CREATE TABLE bug56632;
+SELECT TABLE_NAME,ROW_FORMAT,CREATE_OPTIONS FROM information_schema.tables WHERE TABLE_NAME = 'bug56632';
+
+--echo # Test 5) CREATE with KEY_BLOCK_SIZE=3 (invalid).
+DROP TABLE IF EXISTS bug56632;
+--error ER_CANT_CREATE_TABLE
+CREATE TABLE bug56632 ( i INT ) KEY_BLOCK_SIZE=3;
+SHOW WARNINGS;
+
+SET SESSION innodb_strict_mode = OFF;
+
+--echo # Test 6) CREATE with ROW_FORMAT & KEY_BLOCK_SIZE, ALTER with neither
+DROP TABLE IF EXISTS bug56632;
+CREATE TABLE bug56632 ( i INT ) ROW_FORMAT=COMPACT KEY_BLOCK_SIZE=1;
+SHOW WARNINGS;
+SHOW CREATE TABLE bug56632;
+SELECT TABLE_NAME,ROW_FORMAT,CREATE_OPTIONS FROM information_schema.tables WHERE TABLE_NAME = 'bug56632';
+ALTER TABLE bug56632 ADD COLUMN f1 INT;
+SHOW WARNINGS;
+SHOW CREATE TABLE bug56632;
+SELECT TABLE_NAME,ROW_FORMAT,CREATE_OPTIONS FROM information_schema.tables WHERE TABLE_NAME = 'bug56632';
+
+--echo # Test 7) CREATE with ROW_FORMAT, ALTER with KEY_BLOCK_SIZE
+DROP TABLE IF EXISTS bug56632;
+CREATE TABLE bug56632 ( i INT ) ROW_FORMAT=COMPACT;
+SHOW WARNINGS;
+SHOW CREATE TABLE bug56632;
+SELECT TABLE_NAME,ROW_FORMAT,CREATE_OPTIONS FROM information_schema.tables WHERE TABLE_NAME = 'bug56632';
+ALTER TABLE bug56632 KEY_BLOCK_SIZE=1;
+SHOW WARNINGS;
+SHOW CREATE TABLE bug56632;
+SELECT TABLE_NAME,ROW_FORMAT,CREATE_OPTIONS FROM information_schema.tables WHERE TABLE_NAME = 'bug56632';
+
+--echo # Test 8) CREATE with KEY_BLOCK_SIZE, ALTER with ROW_FORMAT
+DROP TABLE IF EXISTS bug56632;
+CREATE TABLE bug56632 ( i INT ) KEY_BLOCK_SIZE=1;
+SHOW WARNINGS;
+SHOW CREATE TABLE bug56632;
+SELECT TABLE_NAME,ROW_FORMAT,CREATE_OPTIONS FROM information_schema.tables WHERE TABLE_NAME = 'bug56632';
+ALTER TABLE bug56632 ROW_FORMAT=COMPACT;
+SHOW WARNINGS;
+SHOW CREATE TABLE bug56632;
+SELECT TABLE_NAME,ROW_FORMAT,CREATE_OPTIONS FROM information_schema.tables WHERE TABLE_NAME = 'bug56632';
+
+--echo # Test 9) CREATE with neither, ALTER with ROW_FORMAT & KEY_BLOCK_SIZE
+DROP TABLE IF EXISTS bug56632;
+CREATE TABLE bug56632 ( i INT );
+SHOW WARNINGS;
+SHOW CREATE TABLE bug56632;
+SELECT TABLE_NAME,ROW_FORMAT,CREATE_OPTIONS FROM information_schema.tables WHERE TABLE_NAME = 'bug56632';
+ALTER TABLE bug56632 ROW_FORMAT=COMPACT KEY_BLOCK_SIZE=1;
+SHOW WARNINGS;
+SHOW CREATE TABLE bug56632;
+SELECT TABLE_NAME,ROW_FORMAT,CREATE_OPTIONS FROM information_schema.tables WHERE TABLE_NAME = 'bug56632';
+
+--echo # Test 10) CREATE with KEY_BLOCK_SIZE=3 (invalid), ALTER with neither.
+DROP TABLE IF EXISTS bug56632;
+CREATE TABLE bug56632 ( i INT ) KEY_BLOCK_SIZE=3;
+SHOW WARNINGS;
+SHOW CREATE TABLE bug56632;
+SELECT TABLE_NAME,ROW_FORMAT,CREATE_OPTIONS FROM information_schema.tables WHERE TABLE_NAME = 'bug56632';
+ALTER TABLE bug56632 ADD COLUMN f1 INT;
+SHOW WARNINGS;
+SHOW CREATE TABLE bug56632;
+SELECT TABLE_NAME,ROW_FORMAT,CREATE_OPTIONS FROM information_schema.tables WHERE TABLE_NAME = 'bug56632';
+
+--echo # Test 11) CREATE with KEY_BLOCK_SIZE=3 (invalid), ALTER with ROW_FORMAT=COMPACT.
+DROP TABLE IF EXISTS bug56632;
+CREATE TABLE bug56632 ( i INT ) KEY_BLOCK_SIZE=3;
+SHOW WARNINGS;
+SHOW CREATE TABLE bug56632;
+SELECT TABLE_NAME,ROW_FORMAT,CREATE_OPTIONS FROM information_schema.tables WHERE TABLE_NAME = 'bug56632';
+ALTER TABLE bug56632 ROW_FORMAT=COMPACT;
+SHOW WARNINGS;
+SHOW CREATE TABLE bug56632;
+SELECT TABLE_NAME,ROW_FORMAT,CREATE_OPTIONS FROM information_schema.tables WHERE TABLE_NAME = 'bug56632';
+
+--echo # Test 12) CREATE with KEY_BLOCK_SIZE=3 (invalid), ALTER with KEY_BLOCK_SIZE=1.
+DROP TABLE IF EXISTS bug56632;
+CREATE TABLE bug56632 ( i INT ) KEY_BLOCK_SIZE=3;
+SHOW WARNINGS;
+SHOW CREATE TABLE bug56632;
+SELECT TABLE_NAME,ROW_FORMAT,CREATE_OPTIONS FROM information_schema.tables WHERE TABLE_NAME = 'bug56632';
+ALTER TABLE bug56632 KEY_BLOCK_SIZE=1;
+SHOW WARNINGS;
+SHOW CREATE TABLE bug56632;
+SELECT TABLE_NAME,ROW_FORMAT,CREATE_OPTIONS FROM information_schema.tables WHERE TABLE_NAME = 'bug56632';
+
+--echo # Cleanup
+DROP TABLE IF EXISTS bug56632;
+
+--disable_query_log
+EVAL SET GLOBAL innodb_file_per_table=$innodb_file_per_table_orig;
+EVAL SET GLOBAL innodb_file_format_max=$innodb_file_format_max_orig;
+EVAL SET GLOBAL innodb_file_format=$innodb_file_format_orig;
+EVAL SET SESSION innodb_strict_mode=$innodb_strict_mode_orig;
+--enable_query_log
+

mysql-test/suite/innodb/t/innodb_bug56680.test

+#
+# Bug #56680 InnoDB may return wrong results from a case-insensitive index
+#
+-- source include/have_innodb.inc
+
+-- disable_query_log
+SET @tx_isolation_orig = @@tx_isolation;
+SET @innodb_file_per_table_orig = @@innodb_file_per_table;
+SET @innodb_file_format_orig = @@innodb_file_format;
+SET @innodb_file_format_max_orig = @@innodb_file_format_max;
+# The flag innodb_change_buffering_debug is only available in debug builds.
+# It instructs InnoDB to try to evict pages from the buffer pool when
+# change buffering is possible, so that the change buffer will be used
+# whenever possible.
+-- error 0,ER_UNKNOWN_SYSTEM_VARIABLE
+SET @innodb_change_buffering_debug_orig = @@innodb_change_buffering_debug;
+-- error 0,ER_UNKNOWN_SYSTEM_VARIABLE
+SET GLOBAL innodb_change_buffering_debug = 1;
+-- enable_query_log
+SET GLOBAL tx_isolation='REPEATABLE-READ';
+SET GLOBAL innodb_file_format=Barracuda;
+SET GLOBAL innodb_file_per_table=on;
+
+CREATE TABLE bug56680(
+       a INT AUTO_INCREMENT PRIMARY KEY,
+       b CHAR(1),
+       c INT,
+       INDEX(b))
+ENGINE=InnoDB;
+
+INSERT INTO bug56680 VALUES(0,'x',1);
+BEGIN;
+SELECT b FROM bug56680;
+
+connect (con1,localhost,root,,);
+connection con1;
+BEGIN;
+UPDATE bug56680 SET b='X';
+
+connection default;
+# This should return the last committed value 'x', but would return 'X'
+# due to a bug in row_search_for_mysql().
+SELECT b FROM bug56680;
+# This would always return the last committed value 'x'.
+SELECT * FROM bug56680;
+
+connection con1;
+ROLLBACK;
+disconnect con1;
+
+connection default;
+
+SELECT b FROM bug56680;
+
+# For the rest of this test, use the READ UNCOMMITTED isolation level
+# to see what exists in the secondary index.
+SET GLOBAL tx_isolation='READ-UNCOMMITTED';
+
+# Create enough rows for the table, so that the insert buffer will be
+# used for modifying the secondary index page. There must be multiple
+# index pages, because changes to the root page are never buffered.
+
+INSERT INTO bug56680 SELECT 0,b,c FROM bug56680;
+INSERT INTO bug56680 SELECT 0,b,c FROM bug56680;
+INSERT INTO bug56680 SELECT 0,b,c FROM bug56680;
+INSERT INTO bug56680 SELECT 0,b,c FROM bug56680;
+INSERT INTO bug56680 SELECT 0,b,c FROM bug56680;
+INSERT INTO bug56680 SELECT 0,b,c FROM bug56680;
+INSERT INTO bug56680 SELECT 0,b,c FROM bug56680;
+INSERT INTO bug56680 SELECT 0,b,c FROM bug56680;
+INSERT INTO bug56680 SELECT 0,b,c FROM bug56680;
+INSERT INTO bug56680 SELECT 0,b,c FROM bug56680;
+INSERT INTO bug56680 SELECT 0,b,c FROM bug56680;
+
+BEGIN;
+SELECT b FROM bug56680 LIMIT 2;
+
+connect (con1,localhost,root,,);
+connection con1;
+BEGIN;
+DELETE FROM bug56680 WHERE a=1;
+# This should be buffered, if innodb_change_buffering_debug = 1 is in effect.
+INSERT INTO bug56680 VALUES(1,'X',1);
+
+# This should force an insert buffer merge, and return 'X' in the first row.
+SELECT b FROM bug56680 LIMIT 3;
+
+connection default;
+SELECT b FROM bug56680 LIMIT 2;
+CHECK TABLE bug56680;
+
+connection con1;
+ROLLBACK;
+SELECT b FROM bug56680 LIMIT 2;
+CHECK TABLE bug56680;
+
+connection default;
+disconnect con1;
+
+SELECT b FROM bug56680 LIMIT 2;
+
+CREATE TABLE bug56680_2(
+       a INT AUTO_INCREMENT PRIMARY KEY,
+       b VARCHAR(2) CHARSET latin1 COLLATE latin1_german2_ci,
+       c INT,
+       INDEX(b))
+ENGINE=InnoDB;
+
+INSERT INTO bug56680_2 SELECT 0,_latin1 0xdf,c FROM bug56680;
+
+BEGIN;
+SELECT HEX(b) FROM bug56680_2 LIMIT 2;
+DELETE FROM bug56680_2 WHERE a=1;
+# This should be buffered, if innodb_change_buffering_debug = 1 is in effect.
+INSERT INTO bug56680_2 VALUES(1,'SS',1);
+
+# This should force an insert buffer merge, and return 'SS' in the first row.
+SELECT HEX(b) FROM bug56680_2 LIMIT 3;
+CHECK TABLE bug56680_2;
+
+# Test this with compressed tables.
+ALTER TABLE bug56680_2 ROW_FORMAT=COMPRESSED KEY_BLOCK_SIZE=1;
+
+SELECT HEX(b) FROM bug56680_2 LIMIT 2;
+DELETE FROM bug56680_2 WHERE a=1;
+# This should be buffered, if innodb_change_buffering_debug = 1 is in effect.
+INSERT INTO bug56680_2 VALUES(1,_latin1 0xdf,1);
+
+# This should force an insert buffer merge, and return 0xdf in the first row.
+SELECT HEX(b) FROM bug56680_2 LIMIT 3;
+CHECK TABLE bug56680_2;
+
+DROP TABLE bug56680_2;
+DROP TABLE bug56680;
+
+-- disable_query_log
+SET GLOBAL tx_isolation = @tx_isolation_orig;
+SET GLOBAL innodb_file_per_table = @innodb_file_per_table_orig;
+SET GLOBAL innodb_file_format = @innodb_file_format_orig;
+SET GLOBAL innodb_file_format_max = @innodb_file_format_max_orig;
+-- error 0, ER_UNKNOWN_SYSTEM_VARIABLE
+SET GLOBAL innodb_change_buffering_debug = @innodb_change_buffering_debug_orig;

mysql-test/suite/innodb/t/innodb_bug57252.test

+#
+# Bug#57252 disabling innobase_stats_on_metadata disables ANALYZE
+# http://bugs.mysql.com/57252
+#
+
+-- source include/have_innodb.inc
+
+-- disable_query_log
+-- disable_result_log
+
+SET @innodb_stats_on_metadata_orig = @@innodb_stats_on_metadata;
+
+CREATE TABLE bug57252 (a INT, KEY akey (a)) ENGINE=INNODB;
+
+BEGIN;
+let $i = 10;
+while ($i) {
+	eval INSERT INTO bug57252 VALUES ($i);
+	dec $i;
+}
+COMMIT;
+
+-- enable_result_log
+
+SET GLOBAL innodb_stats_on_metadata=0;
+
+# this calls ::info() without HA_STATUS_CONST and so
+# index->stat_n_diff_key_vals[] is not copied to the mysql-visible
+# rec_per_key
+SELECT cardinality FROM information_schema.statistics
+WHERE table_name='bug57252' AND index_name='akey';
+
+# this calls ::info() with HA_STATUS_CONST and so
+# index->stat_n_diff_key_vals[] is copied to the mysql-visible
+# rec_per_key at the end; when the bug is present dict_update_statistics()
+# is not called beforehand and so index->stat_n_diff_key_vals[] contains
+# an outdated data and thus we get an outdated data in the result when the
+# bug is present
+ANALYZE TABLE bug57252;
+
+SELECT cardinality FROM information_schema.statistics
+WHERE table_name='bug57252' AND index_name='akey';
+
+DROP TABLE bug57252;
+
+SET GLOBAL innodb_stats_on_metadata = @innodb_stats_on_metadata_orig;

mysql-test/suite/sys_vars/r/all_vars.result

@@ -3,6 +3,7 @@ create table t2 (variable_name text);
 load data infile "MYSQLTEST_VARDIR/tmp/sys_vars.all_vars.txt" into table t1;
 insert into t2 select variable_name from information_schema.global_variables;
 insert into t2 select variable_name from information_schema.session_variables;
+delete from t2 where variable_name='innodb_change_buffering_debug';
 update t2 set variable_name= replace(variable_name, "PERFORMANCE_SCHEMA_", "PFS_");
 select variable_name as `There should be *no* long test name listed below:` from t2
 where length(variable_name) > 50;

mysql-test/suite/sys_vars/t/all_vars.test

@@ -61,6 +61,9 @@ eval load data infile "$MYSQLTEST_VARDIR/tmp/sys_vars.all_vars.txt" into table t
 insert into t2 select variable_name from information_schema.global_variables;
 insert into t2 select variable_name from information_schema.session_variables;
 
+# This is only present in debug builds.
+delete from t2 where variable_name='innodb_change_buffering_debug';
+
 # Performance schema variables are too long for files named
 # 'mysql-test/suite/sys_vars/t/' ...
 # ... 'performance_schema_events_waits_history_long_size_basic-master.opt'

storage/innobase/CMakeLists.txt

@@ -125,9 +125,13 @@ IF(CMAKE_SYSTEM_NAME STREQUAL "SunOS")
   CHECK_FUNCTION_EXISTS(atomic_cas_ulong  HAVE_ATOMIC_CAS_ULONG)
   CHECK_FUNCTION_EXISTS(atomic_cas_32 HAVE_ATOMIC_CAS_32)
   CHECK_FUNCTION_EXISTS(atomic_cas_64 HAVE_ATOMIC_CAS_64)
-  CHECK_FUNCTION_EXISTS(atomic_add_long HAVE_ATOMIC_ADD_LONG)
-  IF(HAVE_ATOMIC_CAS_ULONG AND HAVE_ATOMIC_CAS_32 AND
-    HAVE_ATOMIC_CAS_64 AND HAVE_ATOMIC_ADD_LONG)
+  CHECK_FUNCTION_EXISTS(atomic_add_long_nv HAVE_ATOMIC_ADD_LONG_NV)
+  CHECK_FUNCTION_EXISTS(atomic_swap_uchar HAVE_ATOMIC_SWAP_UCHAR)
+  IF(HAVE_ATOMIC_CAS_ULONG AND
+     HAVE_ATOMIC_CAS_32 AND
+     HAVE_ATOMIC_CAS_64 AND
+     HAVE_ATOMIC_ADD_LONG_NV AND
+     HAVE_ATOMIC_SWAP_UCHAR)
     SET(HAVE_IB_SOLARIS_ATOMICS 1)
   ENDIF()
   
@@ -231,7 +235,7 @@ SET(INNOBASE_SOURCES	btr/btr0btr.c btr/btr0cur.c btr/btr0pcur.c btr/btr0sea.c
 			os/os0file.c os/os0proc.c os/os0sync.c os/os0thread.c
 			page/page0cur.c page/page0page.c page/page0zip.c
 			que/que0que.c
-			handler/ha_innodb.cc handler/handler0alter.cc handler/i_s.cc handler/mysql_addons.cc
+			handler/ha_innodb.cc handler/handler0alter.cc handler/i_s.cc
 			read/read0read.c
 			rem/rem0cmp.c rem/rem0rec.c
 			row/row0ext.c row/row0ins.c row/row0merge.c row/row0mysql.c row/row0purge.c row/row0row.c

storage/innobase/Makefile.am

@@ -115,7 +115,6 @@ noinst_HEADERS=		\
 			include/mtr0mtr.h	\
 			include/mtr0mtr.ic	\
 			include/mtr0types.h	\
-			include/mysql_addons.h	\
 			include/os0file.h	\
 			include/os0file.ic	\
 			include/os0proc.h	\
@@ -258,7 +257,6 @@ libinnobase_a_SOURCES=	\
 			handler/ha_innodb.cc		\
 			handler/handler0alter.cc	\
 			handler/i_s.cc			\
-			handler/mysql_addons.cc		\
 			ibuf/ibuf0ibuf.c		\
 			lock/lock0iter.c		\
 			lock/lock0lock.c		\

storage/innobase/btr/btr0cur.c

@@ -1744,7 +1744,7 @@ func_exit:
 See if there is enough place in the page modification log to log
 an update-in-place.
 @return	TRUE if enough place */
-static
+UNIV_INTERN
 ibool
 btr_cur_update_alloc_zip(
 /*=====================*/

storage/innobase/buf/buf0buddy.c

@@ -281,7 +281,7 @@ buf_buddy_alloc_from(
 
 /**********************************************************************//**
 Allocate a block.  The thread calling this function must hold
-buf_pool->mutex and must not hold buf_pool_zip_mutex or any block->mutex.
+buf_pool->mutex and must not hold buf_pool->zip_mutex or any block->mutex.
 The buf_pool->mutex may only be released and reacquired if lru != NULL.
 @return	allocated block, possibly NULL if lru==NULL */
 UNIV_INTERN

storage/innobase/buf/buf0buf.c

@@ -172,7 +172,7 @@ The chain of modified blocks (buf_pool->flush_list) contains the blocks
 holding file pages that have been modified in the memory
 but not written to disk yet. The block with the oldest modification
 which has not yet been written to disk is at the end of the chain.
-The access to this list is protected by flush_list_mutex.
+The access to this list is protected by buf_pool->flush_list_mutex.
 
 The chain of unmodified compressed blocks (buf_pool->zip_clean)
 contains the control blocks (buf_page_t) of those compressed pages
@@ -1882,8 +1882,8 @@ buf_pool_watch_set(
 			ut_ad(!bpage->in_page_hash);
 			ut_ad(bpage->buf_fix_count == 0);
 
-			/* bpage is pointing to buf_pool_watch[],
-			which is protected by buf_pool_mutex.
+			/* bpage is pointing to buf_pool->watch[],
+			which is protected by buf_pool->mutex.
 			Normally, buf_page_t objects are protected by
 			buf_block_t::mutex or buf_pool->zip_mutex or both. */
 
@@ -3008,6 +3008,46 @@ wait_until_unfixed:
 	bytes. */
 	UNIV_MEM_ASSERT_RW(&block->page, sizeof block->page);
 #endif
+#if defined UNIV_DEBUG || defined UNIV_IBUF_DEBUG
+	if ((mode == BUF_GET_IF_IN_POOL || mode == BUF_GET_IF_IN_POOL_OR_WATCH)
+	    && ibuf_debug) {
+		/* Try to evict the block from the buffer pool, to use the
+		insert buffer (change buffer) as much as possible. */
+
+		if (buf_LRU_free_block(&block->page, TRUE, NULL)
+		    == BUF_LRU_FREED) {
+			mutex_exit(&block->mutex);
+			if (mode == BUF_GET_IF_IN_POOL_OR_WATCH) {
+				/* Set the watch, as it would have
+				been set if the page were not in the
+				buffer pool in the first place. */
+				block = (buf_block_t*) buf_pool_watch_set(
+					space, offset, fold);
+
+				if (UNIV_LIKELY_NULL(block)) {
+
+					/* The page entered the buffer
+					pool for some reason. Try to
+					evict it again. */
+					goto got_block;
+				}
+			}
+			buf_pool_mutex_exit(buf_pool);
+			fprintf(stderr,
+				"innodb_change_buffering_debug evict %u %u\n",
+				(unsigned) space, (unsigned) offset);
+			return(NULL);
+		} else if (buf_flush_page_try(buf_pool, block)) {
+			fprintf(stderr,
+				"innodb_change_buffering_debug flush %u %u\n",
+				(unsigned) space, (unsigned) offset);
+			guess = block;
+			goto loop;
+		}
+
+		/* Failed to evict the page; change it directly */
+	}
+#endif /* UNIV_DEBUG || UNIV_IBUF_DEBUG */
 
 	buf_block_buf_fix_inc(block, file, line);
 

storage/innobase/buf/buf0flu.c

@@ -321,7 +321,7 @@ buf_flush_insert_sorted_into_flush_list(
 
 	buf_flush_list_mutex_enter(buf_pool);
 
-	/* The field in_LRU_list is protected by buf_pool_mutex, which
+	/* The field in_LRU_list is protected by buf_pool->mutex, which
 	we are not holding.  However, while a block is in the flush
 	list, it is dirty and cannot be discarded, not from the
 	page_hash or from the LRU list.  At most, the uncompressed
@@ -1061,7 +1061,7 @@ buf_flush_write_block_low(
 
 	ut_ad(buf_page_in_file(bpage));
 
-	/* We are not holding buf_pool_mutex or block_mutex here.
+	/* We are not holding buf_pool->mutex or block_mutex here.
 	Nevertheless, it is safe to access bpage, because it is
 	io_fixed and oldest_modification != 0.  Thus, it cannot be
 	relocated in the buffer pool or removed from flush_list or
@@ -1132,6 +1132,83 @@ buf_flush_write_block_low(
 	}
 }
 
+# if defined UNIV_DEBUG || defined UNIV_IBUF_DEBUG
+/********************************************************************//**
+Writes a flushable page asynchronously from the buffer pool to a file.
+NOTE: buf_pool->mutex and block->mutex must be held upon entering this
+function, and they will be released by this function after flushing.
+This is loosely based on buf_flush_batch() and buf_flush_page().
+@return TRUE if the page was flushed and the mutexes released */
+UNIV_INTERN
+ibool
+buf_flush_page_try(
+/*===============*/
+	buf_pool_t*	buf_pool,	/*!< in/out: buffer pool instance */
+	buf_block_t*	block)		/*!< in/out: buffer control block */
+{
+	ut_ad(buf_pool_mutex_own(buf_pool));
+	ut_ad(buf_block_get_state(block) == BUF_BLOCK_FILE_PAGE);
+	ut_ad(mutex_own(&block->mutex));
+
+	if (!buf_flush_ready_for_flush(&block->page, BUF_FLUSH_LRU)) {
+		return(FALSE);
+	}
+
+	if (buf_pool->n_flush[BUF_FLUSH_LRU] > 0
+	    || buf_pool->init_flush[BUF_FLUSH_LRU]) {
+		/* There is already a flush batch of the same type running */
+		return(FALSE);
+	}
+
+	buf_pool->init_flush[BUF_FLUSH_LRU] = TRUE;
+
+	buf_page_set_io_fix(&block->page, BUF_IO_WRITE);
+
+	buf_page_set_flush_type(&block->page, BUF_FLUSH_LRU);
+
+	if (buf_pool->n_flush[BUF_FLUSH_LRU]++ == 0) {
+
+		os_event_reset(buf_pool->no_flush[BUF_FLUSH_LRU]);
+	}
+
+	/* VERY IMPORTANT:
+	Because any thread may call the LRU flush, even when owning
+	locks on pages, to avoid deadlocks, we must make sure that the
+	s-lock is acquired on the page without waiting: this is
+	accomplished because buf_flush_ready_for_flush() must hold,
+	and that requires the page not to be bufferfixed. */
+
+	rw_lock_s_lock_gen(&block->lock, BUF_IO_WRITE);
+
+	/* Note that the s-latch is acquired before releasing the
+	buf_pool mutex: this ensures that the latch is acquired
+	immediately. */
+
+	mutex_exit(&block->mutex);
+	buf_pool_mutex_exit(buf_pool);
+
+	/* Even though block is not protected by any mutex at this
+	point, it is safe to access block, because it is io_fixed and
+	oldest_modification != 0.  Thus, it cannot be relocated in the
+	buffer pool or removed from flush_list or LRU_list. */
+
+	buf_flush_write_block_low(&block->page);
+
+	buf_pool_mutex_enter(buf_pool);
+	buf_pool->init_flush[BUF_FLUSH_LRU] = FALSE;
+
+	if (buf_pool->n_flush[BUF_FLUSH_LRU] == 0) {
+		/* The running flush batch has ended */
+		os_event_set(buf_pool->no_flush[BUF_FLUSH_LRU]);
+	}
+
+	buf_pool_mutex_exit(buf_pool);
+	buf_flush_buffered_writes();
+
+	return(TRUE);
+}
+# endif /* UNIV_DEBUG || UNIV_IBUF_DEBUG */
+
 /********************************************************************//**
 Writes a flushable page asynchronously from the buffer pool to a file.
 NOTE: in simulated aio we must call
@@ -2116,12 +2193,12 @@ buf_flush_validate_low(
 
 		ut_ad(bpage->in_flush_list);
 
-		/* A page in flush_list can be in BUF_BLOCK_REMOVE_HASH
-		state. This happens when a page is in the middle of
-		being relocated. In that case the original descriptor
-		can have this state and still be in the flush list
-		waiting to acquire the flush_list_mutex to complete
-		the relocation. */
+		/* A page in buf_pool->flush_list can be in
+		BUF_BLOCK_REMOVE_HASH state. This happens when a page
+		is in the middle of being relocated. In that case the
+		original descriptor can have this state and still be
+		in the flush list waiting to acquire the
+		buf_pool->flush_list_mutex to complete the relocation. */
 		ut_a(buf_page_in_file(bpage)
 		     || buf_page_get_state(bpage) == BUF_BLOCK_REMOVE_HASH);
 		ut_a(om > 0);

storage/innobase/buf/buf0lru.c

@@ -356,8 +356,8 @@ scan_again:
 		prev_bpage = UT_LIST_GET_PREV(LRU, bpage);
 
 		/* bpage->space and bpage->io_fix are protected by
-		buf_pool_mutex and block_mutex.  It is safe to check
-		them while holding buf_pool_mutex only. */
+		buf_pool->mutex and block_mutex.  It is safe to check
+		them while holding buf_pool->mutex only. */
 
 		if (buf_page_get_space(bpage) != id) {
 			/* Skip this block, as it does not belong to
@@ -403,7 +403,7 @@ scan_again:
 						/* Descriptors of uncompressed
 						blocks will not be relocated,
 						because we are holding the
-						buf_pool_mutex. */
+						buf_pool->mutex. */
 						break;
 					case BUF_BLOCK_ZIP_PAGE:
 					case BUF_BLOCK_ZIP_DIRTY:
@@ -1443,10 +1443,10 @@ Try to free a block.  If bpage is a descriptor of a compressed-only
 page, the descriptor object will be freed as well.
 
 NOTE: If this function returns BUF_LRU_FREED, it will temporarily
-release buf_pool_mutex.  Furthermore, the page frame will no longer be
+release buf_pool->mutex.  Furthermore, the page frame will no longer be
 accessible via bpage.
 
-The caller must hold buf_pool_mutex and buf_page_get_mutex(bpage) and
+The caller must hold buf_pool->mutex and buf_page_get_mutex(bpage) and
 release these two mutexes after the call.  No other
 buf_page_get_mutex() may be held when calling this function.
 @return BUF_LRU_FREED if freed, BUF_LRU_CANNOT_RELOCATE or

storage/innobase/handler/ha_innodb.h

@@ -109,6 +109,7 @@ class ha_innobase: public handler
 	ulint innobase_update_autoinc(ulonglong	auto_inc);
 	void innobase_initialize_autoinc();
 	dict_index_t* innobase_get_index(uint keynr);
+	int info_low(uint flag, bool called_from_analyze);
 
 	/* Init values for the class: */
  public:

storage/innobase/handler/mysql_addons.cc

-/*****************************************************************************
-
-Copyright (c) 2007, 2009, Innobase Oy. All Rights Reserved.
-
-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; version 2 of the License.
-
-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
-
-*****************************************************************************/
-
-/**************************************************//**
-@file handler/mysql_addons.cc
-This file contains functions that need to be added to
-MySQL code but have not been added yet.
-
-Whenever you add a function here submit a MySQL bug
-report (feature request) with the implementation. Then
-write the bug number in the comment before the
-function in this file.
-
-When MySQL commits the function it can be deleted from
-here. In a perfect world this file exists but is empty.
-
-Created November 07, 2007 Vasil Dimov
-*******************************************************/
-
-#ifndef MYSQL_SERVER
-#define MYSQL_SERVER
-#endif /* MYSQL_SERVER */
-
-#include <sql_priv.h>
-
-#include "mysql_addons.h"
-#include "univ.i"

storage/innobase/include/btr0cur.h

@@ -243,6 +243,22 @@ btr_cur_pessimistic_insert(
 	que_thr_t*	thr,	/*!< in: query thread or NULL */
 	mtr_t*		mtr);	/*!< in: mtr */
 /*************************************************************//**
+See if there is enough place in the page modification log to log
+an update-in-place.
+@return	TRUE if enough place */
+UNIV_INTERN
+ibool
+btr_cur_update_alloc_zip(
+/*=====================*/
+	page_zip_des_t*	page_zip,/*!< in/out: compressed page */
+	buf_block_t*	block,	/*!< in/out: buffer page */
+	dict_index_t*	index,	/*!< in: the index corresponding to the block */
+	ulint		length,	/*!< in: size needed */
+	ibool		create,	/*!< in: TRUE=delete-and-insert,
+				FALSE=update-in-place */
+	mtr_t*		mtr)	/*!< in: mini-transaction */
+	__attribute__((nonnull, warn_unused_result));
+/*************************************************************//**
 Updates a record when the update causes no size changes in its fields.
 @return	DB_SUCCESS or error number */
 UNIV_INTERN

storage/innobase/include/buf0buddy.h

@@ -36,7 +36,7 @@ Created December 2006 by Marko Makela
 
 /**********************************************************************//**
 Allocate a block.  The thread calling this function must hold
-buf_pool->mutex and must not hold buf_pool_zip_mutex or any
+buf_pool->mutex and must not hold buf_pool->zip_mutex or any
 block->mutex.  The buf_pool->mutex may only be released and reacquired
 if lru != NULL.  This function should only be used for allocating
 compressed page frames or control blocks (buf_page_t).  Allocated

storage/innobase/include/buf0buddy.ic

@@ -35,7 +35,7 @@ Created December 2006 by Marko Makela
 
 /**********************************************************************//**
 Allocate a block.  The thread calling this function must hold
-buf_pool->mutex and must not hold buf_pool_zip_mutex or any block->mutex.
+buf_pool->mutex and must not hold buf_pool->zip_mutex or any block->mutex.
 The buf_pool->mutex may only be released and reacquired if lru != NULL.
 @return	allocated block, possibly NULL if lru==NULL */
 UNIV_INTERN
@@ -86,7 +86,7 @@ buf_buddy_get_slot(
 
 /**********************************************************************//**
 Allocate a block.  The thread calling this function must hold
-buf_pool->mutex and must not hold buf_pool_zip_mutex or any
+buf_pool->mutex and must not hold buf_pool->zip_mutex or any
 block->mutex.  The buf_pool->mutex may only be released and reacquired
 if lru != NULL.  This function should only be used for allocating
 compressed page frames or control blocks (buf_page_t).  Allocated

storage/innobase/include/buf0buf.h

@@ -96,7 +96,7 @@ enum buf_page_state {
 	BUF_BLOCK_ZIP_FREE = 0,		/*!< contains a free
 					compressed page */
 	BUF_BLOCK_POOL_WATCH = 0,	/*!< a sentinel for the buffer pool
-					watch, element of buf_pool_watch[] */
+					watch, element of buf_pool->watch[] */
 	BUF_BLOCK_ZIP_PAGE,		/*!< contains a clean
 					compressed page */
 	BUF_BLOCK_ZIP_DIRTY,		/*!< contains a compressed
@@ -1210,10 +1210,10 @@ struct buf_page_struct{
 #endif /* !UNIV_HOTBACKUP */
 	page_zip_des_t	zip;		/*!< compressed page; zip.data
 					(but not the data it points to) is
-					also protected by buf_pool_mutex;
+					also protected by buf_pool->mutex;
 					state == BUF_BLOCK_ZIP_PAGE and
 					zip.data == NULL means an active
-					buf_pool_watch */
+					buf_pool->watch */
 #ifndef UNIV_HOTBACKUP
 	buf_page_t*	hash;		/*!< node used in chaining to
 					buf_pool->page_hash or
@@ -1224,15 +1224,16 @@ struct buf_page_struct{
 #endif /* UNIV_DEBUG */
 
 	/** @name Page flushing fields
-	All these are protected by buf_pool_mutex. */
+	All these are protected by buf_pool->mutex. */
 	/* @{ */
 
 	UT_LIST_NODE_T(buf_page_t) list;
 					/*!< based on state, this is a
 					list node, protected either by
-					buf_pool_mutex or by
-					flush_list_mutex, in one of the
-					following lists in buf_pool:
+					buf_pool->mutex or by
+					buf_pool->flush_list_mutex,
+					in one of the following lists in
+					buf_pool:
 
 					- BUF_BLOCK_NOT_USED:	free
 					- BUF_BLOCK_FILE_PAGE:	flush_list
@@ -1242,9 +1243,9 @@ struct buf_page_struct{
 
 					If bpage is part of flush_list
 					then the node pointers are
-					covered by flush_list_mutex.
+					covered by buf_pool->flush_list_mutex.
 					Otherwise these pointers are
-					protected by buf_pool_mutex.
+					protected by buf_pool->mutex.
 
 					The contents of the list node
 					is undefined if !in_flush_list
@@ -1256,17 +1257,18 @@ struct buf_page_struct{
 
 #ifdef UNIV_DEBUG
 	ibool		in_flush_list;	/*!< TRUE if in buf_pool->flush_list;
-					when flush_list_mutex is free, the
-					following should hold: in_flush_list
+					when buf_pool->flush_list_mutex is
+					free, the following should hold:
+					in_flush_list
 					== (state == BUF_BLOCK_FILE_PAGE
 					    || state == BUF_BLOCK_ZIP_DIRTY)
 					Writes to this field must be
 					covered by both block->mutex
-					and flush_list_mutex. Hence
+					and buf_pool->flush_list_mutex. Hence
 					reads can happen while holding
 					any one of the two mutexes */
 	ibool		in_free_list;	/*!< TRUE if in buf_pool->free; when
-					buf_pool_mutex is free, the following
+					buf_pool->mutex is free, the following
 					should hold: in_free_list
 					== (state == BUF_BLOCK_NOT_USED) */
 #endif /* UNIV_DEBUG */
@@ -1286,7 +1288,7 @@ struct buf_page_struct{
 					modifications are on disk.
 					Writes to this field must be
 					covered by both block->mutex
-					and flush_list_mutex. Hence
+					and buf_pool->flush_list_mutex. Hence
 					reads can happen while holding
 					any one of the two mutexes */
 	/* @} */
@@ -1661,20 +1663,13 @@ struct buf_pool_struct{
 	/* @} */
 };
 
-/** mutex protecting the buffer pool struct and control blocks, except the
-read-write lock in them */
-extern mutex_t	buf_pool_mutex;
-/** mutex protecting the control blocks of compressed-only pages
-(of type buf_page_t, not buf_block_t) */
-extern mutex_t	buf_pool_zip_mutex;
-
-/** @name Accessors for buf_pool_mutex.
-Use these instead of accessing buf_pool_mutex directly. */
+/** @name Accessors for buf_pool->mutex.
+Use these instead of accessing buf_pool->mutex directly. */
 /* @{ */
 
-/** Test if buf_pool_mutex is owned. */
+/** Test if a buffer pool mutex is owned. */
 #define buf_pool_mutex_own(b) mutex_own(&b->mutex)
-/** Acquire the buffer pool mutex. */
+/** Acquire a buffer pool mutex. */
 #define buf_pool_mutex_enter(b) do {		\
 	ut_ad(!mutex_own(&b->zip_mutex));	\
 	mutex_enter(&b->mutex);		\

storage/innobase/include/buf0flu.h

@@ -84,6 +84,21 @@ buf_flush_init_for_writing(
 	ib_uint64_t	newest_lsn);	/*!< in: newest modification lsn
 					to the page */
 #ifndef UNIV_HOTBACKUP
+# if defined UNIV_DEBUG || defined UNIV_IBUF_DEBUG
+/********************************************************************//**
+Writes a flushable page asynchronously from the buffer pool to a file.
+NOTE: buf_pool->mutex and block->mutex must be held upon entering this
+function, and they will be released by this function after flushing.
+This is loosely based on buf_flush_batch() and buf_flush_page().
+@return TRUE if the page was flushed and the mutexes released */
+UNIV_INTERN
+ibool
+buf_flush_page_try(
+/*===============*/
+	buf_pool_t*	buf_pool,	/*!< in/out: buffer pool instance */
+	buf_block_t*	block)		/*!< in/out: buffer control block */
+	__attribute__((nonnull, warn_unused_result));
+# endif /* UNIV_DEBUG || UNIV_IBUF_DEBUG */
 /*******************************************************************//**
 This utility flushes dirty blocks from the end of the LRU list.
 NOTE: The calling thread may own latches to pages: to avoid deadlocks,

storage/innobase/include/data0type.h

@@ -174,10 +174,13 @@ store the charset-collation number; one byte is left unused, though */
 /* Pack mbminlen, mbmaxlen to mbminmaxlen. */
 #define DATA_MBMINMAXLEN(mbminlen, mbmaxlen)	\
 	((mbmaxlen) * DATA_MBMAX + (mbminlen))
-/* Get mbminlen from mbminmaxlen. */
-#define DATA_MBMINLEN(mbminmaxlen) UNIV_EXPECT(((mbminmaxlen) % DATA_MBMAX), 1)
+/* Get mbminlen from mbminmaxlen. Cast the result of UNIV_EXPECT to ulint
+because in GCC it returns a long. */
+#define DATA_MBMINLEN(mbminmaxlen) ((ulint) \
+                                    UNIV_EXPECT(((mbminmaxlen) % DATA_MBMAX), \
+                                                1))
 /* Get mbmaxlen from mbminmaxlen. */
-#define DATA_MBMAXLEN(mbminmaxlen) ((mbminmaxlen) / DATA_MBMAX)
+#define DATA_MBMAXLEN(mbminmaxlen) ((ulint) ((mbminmaxlen) / DATA_MBMAX))
 
 #ifndef UNIV_HOTBACKUP
 /*********************************************************************//**

storage/innobase/include/ibuf0ibuf.h

@@ -63,6 +63,11 @@ typedef enum {
 /** Operations that can currently be buffered. */
 extern ibuf_use_t	ibuf_use;
 
+#if defined UNIV_DEBUG || defined UNIV_IBUF_DEBUG
+/** Flag to control insert buffer debugging. */
+extern uint		ibuf_debug;
+#endif /* UNIV_DEBUG || UNIV_IBUF_DEBUG */
+
 /** The insert buffer control structure */
 extern ibuf_t*		ibuf;
 

storage/innobase/include/mysql_addons.h

-/*****************************************************************************
-
-Copyright (c) 2007, 2009, Innobase Oy. All Rights Reserved.
-
-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; version 2 of the License.
-
-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
-
-*****************************************************************************/
-
-/**************************************************//**
-@file include/mysql_addons.h
-This file contains functions that need to be added to
-MySQL code but have not been added yet.
-
-Whenever you add a function here submit a MySQL bug
-report (feature request) with the implementation. Then
-write the bug number in the comment before the
-function in this file.
-
-When MySQL commits the function it can be deleted from
-here. In a perfect world this file exists but is empty.
-
-Created November 07, 2007 Vasil Dimov
-*******************************************************/

storage/innobase/include/os0sync.h

@@ -76,6 +76,12 @@ struct os_event_struct {
 					/*!< list of all created events */
 };
 
+/** Denotes an infinite delay for os_event_wait_time() */
+#define OS_SYNC_INFINITE_TIME   ULINT_UNDEFINED
+
+/** Return value of os_event_wait_time() when the time is exceeded */
+#define OS_SYNC_TIME_EXCEEDED   1
+
 /** Operating system mutex */
 typedef struct os_mutex_struct	os_mutex_str_t;
 /** Operating system mutex handle */
@@ -173,7 +179,23 @@ os_event_wait_low(
 					os_event_reset(). */
 
 #define os_event_wait(event) os_event_wait_low(event, 0)
+#define os_event_wait_time(e, t) os_event_wait_time_low(event, t, 0)
 
+/**********************************************************//**
+Waits for an event object until it is in the signaled state or
+a timeout is exceeded. In Unix the timeout is always infinite.
+@return	0 if success, OS_SYNC_TIME_EXCEEDED if timeout was exceeded */
+UNIV_INTERN
+ulint
+os_event_wait_time_low(
+/*===================*/
+	os_event_t	event,			/*!< in: event to wait */
+	ulint		time_in_usec,		/*!< in: timeout in
+						microseconds, or
+						OS_SYNC_INFINITE_TIME */
+	ib_int64_t	reset_sig_count);	/*!< in: zero or the value
+						returned by previous call of
+						os_event_reset(). */
 /*********************************************************//**
 Creates an operating system mutex semaphore. Because these are slow, the
 mutex semaphore of InnoDB itself (mutex_t) should be used where possible.

storage/innobase/include/rem0rec.h

@@ -803,7 +803,7 @@ rec_print(
 /*======*/
 	FILE*			file,	/*!< in: file where to print */
 	const rec_t*		rec,	/*!< in: physical record */
-	dict_index_t*	index);	/*!< in: record descriptor */
+	const dict_index_t*	index);	/*!< in: record descriptor */
 #endif /* UNIV_HOTBACKUP */
 
 /* Maximum lengths for the data in a physical record if the offsets

storage/innobase/include/row0mysql.h

@@ -538,6 +538,10 @@ struct mysql_row_templ_struct {
 					Innobase record in the current index;
 					not defined if template_type is
 					ROW_MYSQL_WHOLE_ROW */
+	ulint	clust_rec_field_no;	/*!< field number of the column in an
+					Innobase record in the clustered index;
+					not defined if template_type is
+					ROW_MYSQL_WHOLE_ROW */
 	ulint	mysql_col_offset;	/*!< offset of the column in the MySQL
 					row format */
 	ulint	mysql_col_len;		/*!< length of the column in the MySQL

storage/innobase/include/row0upd.h

@@ -167,8 +167,11 @@ row_upd_changes_field_size_or_external(
 	const upd_t*	update);/*!< in: update vector */
 #endif /* !UNIV_HOTBACKUP */
 /***********************************************************//**
-Replaces the new column values stored in the update vector to the record
-given. No field size changes are allowed. */
+Replaces the new column values stored in the update vector to the
+record given. No field size changes are allowed. This function is
+usually invoked on a clustered index. The only use case for a
+secondary index is row_ins_sec_index_entry_by_modify() or its
+counterpart in ibuf_insert_to_index_page(). */
 UNIV_INTERN
 void
 row_upd_rec_in_place(

storage/innobase/include/srv0srv.h

@@ -57,6 +57,15 @@ extern const char	srv_mysql50_table_name_prefix[9];
 thread starts running */
 extern os_event_t	srv_lock_timeout_thread_event;
 
+/* The monitor thread waits on this event. */
+extern os_event_t	srv_monitor_event;
+
+/* The lock timeout thread waits on this event. */
+extern os_event_t	srv_timeout_event;
+
+/* The error monitor thread waits on this event. */
+extern os_event_t	srv_error_event;
+
 /* If the last data file is auto-extended, we add this many pages to it
 at a time */
 #define SRV_AUTO_EXTEND_INCREMENT	\

storage/innobase/include/trx0trx.h

@@ -470,6 +470,20 @@ struct trx_struct{
 					of view of concurrency control:
 					TRX_ACTIVE, TRX_COMMITTED_IN_MEMORY,
 					... */
+	/*------------------------------*/
+	/* MySQL has a transaction coordinator to coordinate two phase
+       	commit between multiple storage engines and the binary log. When
+       	an engine participates in a transaction, it's responsible for
+       	registering itself using the trans_register_ha() API. */
+	unsigned	is_registered:1;/* This flag is set to 1 after the
+				       	transaction has been registered with
+				       	the coordinator using the XA API, and
+				       	is set to 0 after commit or rollback. */
+	unsigned	owns_prepare_mutex:1;/* 1 if owns prepare mutex, if
+					this is set to 1 then registered should
+					also be set to 1. This is used in the
+					XA code */
+	/*------------------------------*/
 	ulint		isolation_level;/* TRX_ISO_REPEATABLE_READ, ... */
 	ulint		check_foreigns;	/* normally TRUE, but if the user
 					wants to suppress foreign key checks,
@@ -500,9 +514,6 @@ struct trx_struct{
 					in that case we must flush the log
 					in trx_commit_complete_for_mysql() */
 	ulint		duplicates;	/*!< TRX_DUP_IGNORE | TRX_DUP_REPLACE */
-	ulint		active_trans;	/*!< 1 - if a transaction in MySQL
-					is active. 2 - if prepare_commit_mutex
-					was taken */
 	ulint		has_search_latch;
 					/* TRUE if this trx has latched the
 					search system latch in S-mode */

storage/innobase/lock/lock0lock.c

@@ -4859,11 +4859,11 @@ loop:
 			ut_a(rec);
 			offsets = rec_get_offsets(rec, index, offsets,
 						  ULINT_UNDEFINED, &heap);
-
+#if 0
 			fprintf(stderr,
 				"Validating %lu %lu\n",
 				(ulong) space, (ulong) page_no);
-
+#endif
 			lock_mutex_exit_kernel();
 
 			/* If this thread is holding the file space

storage/innobase/log/log0log.c

@@ -3098,10 +3098,15 @@ loop:
 
 	if (srv_fast_shutdown < 2
 	   && (srv_error_monitor_active
-	      || srv_lock_timeout_active || srv_monitor_active)) {
+	      || srv_lock_timeout_active
+	      || srv_monitor_active)) {
 
 		mutex_exit(&kernel_mutex);
 
+		os_event_set(srv_error_event);
+		os_event_set(srv_monitor_event);
+		os_event_set(srv_timeout_event);
+
 		goto loop;
 	}
 
@@ -3128,6 +3133,8 @@ loop:
 
 		log_buffer_flush_to_disk();
 
+		mutex_exit(&kernel_mutex);
+
 		return; /* We SKIP ALL THE REST !! */
 	}
 

storage/innobase/mysql-test/patches/README

-This directory contains patches that need to be applied to the MySQL
-source tree in order to get the mysql-test suite to succeed (when
-storage/innobase is replaced with this InnoDB branch). Things to keep
-in mind when adding new patches here:
-
-* The patch must be appliable from the mysql top-level source directory.
-
-* The patch filename must end in ".diff".
-
-* All patches here are expected to apply cleanly to the latest MySQL 5.1
-  tree when storage/innobase is replaced with this InnoDB branch. If
-  changes to either of those cause the patch to fail, then please check
-  whether the patch is still needed and, if yes, adjust it so it applies
-  cleanly.
-
-* If applicable, always submit the patch at http://bugs.mysql.com and
-  name the file here like bug%d.diff. Once the patch is committed to
-  MySQL remove the file from here.
-
-* If the patch cannot be proposed for inclusion in the MySQL source tree
-  (via http://bugs.mysql.com) then add a comment at the beginning of the
-  patch, explaining the problem it is solving, how it does solve it and
-  why it is not applicable for inclusion in the MySQL source tree.
-  Obviously this is a very bad situation and should be avoided at all
-  costs, especially for files that are in the MySQL source repository
-  (not in storage/innobase).
-
-* If you ever need to add a patch here that is not related to mysql-test
-  suite, then please move this directory from ./mysql-test/patches to
-  ./patches and remove this text.

storage/innobase/mysql-test/patches/index_merge_innodb-explain.diff

-InnoDB's estimate for the index cardinality depends on a pseudo random
-number generator (it picks up random pages to sample). After an
-optimization that was made in r2625 the following EXPLAINs started
-returning a different number of rows (3 instead of 4).
-
-This patch adjusts the result file.
-
-This patch cannot be proposed to MySQL because the failures occur only
-in this tree and do not occur in the standard InnoDB 5.1. Furthermore,
-the file index_merge2.inc is used by other engines too.
-
---- mysql-test/r/index_merge_innodb.result.orig	2008-09-30 18:32:13.000000000 +0300
-+++ mysql-test/r/index_merge_innodb.result	2008-09-30 18:33:01.000000000 +0300
-@@ -111,7 +111,7 @@
- explain select count(*) from t1 where
- key1a = 2 and key1b is null and  key2a = 2 and key2b is null;
- id	select_type	table	type	possible_keys	key	key_len	ref	rows	Extra
--1	SIMPLE	t1	index_merge	i1,i2	i1,i2	10,10	NULL	4	Using intersect(i1,i2); Using where; Using index
-+1	SIMPLE	t1	index_merge	i1,i2	i1,i2	10,10	NULL	3	Using intersect(i1,i2); Using where; Using index
- select count(*) from t1 where
- key1a = 2 and key1b is null and key2a = 2 and key2b is null;
- count(*)
-@@ -119,7 +119,7 @@
- explain select count(*) from t1 where
- key1a = 2 and key1b is null and key3a = 2 and key3b is null;
- id	select_type	table	type	possible_keys	key	key_len	ref	rows	Extra
--1	SIMPLE	t1	index_merge	i1,i3	i1,i3	10,10	NULL	4	Using intersect(i1,i3); Using where; Using index
-+1	SIMPLE	t1	index_merge	i1,i3	i1,i3	10,10	NULL	3	Using intersect(i1,i3); Using where; Using index
- select count(*) from t1 where
- key1a = 2 and key1b is null and key3a = 2 and key3b is null;
- count(*)

storage/innobase/mysql-test/patches/information_schema.diff

---- mysql-test/r/information_schema.result.orig	2009-01-31 03:38:50.000000000 +0200
-+++ mysql-test/r/information_schema.result	2009-01-31 07:51:58.000000000 +0200
-@@ -71,6 +71,13 @@
- TRIGGERS
- USER_PRIVILEGES
- VIEWS
-+INNODB_CMP_RESET
-+INNODB_TRX
-+INNODB_CMPMEM_RESET
-+INNODB_LOCK_WAITS
-+INNODB_CMPMEM
-+INNODB_CMP
-+INNODB_LOCKS
- columns_priv
- db
- event
-@@ -799,6 +806,8 @@
- TABLES	UPDATE_TIME	datetime
- TABLES	CHECK_TIME	datetime
- TRIGGERS	CREATED	datetime
-+INNODB_TRX	trx_started	datetime
-+INNODB_TRX	trx_wait_started	datetime
- event	execute_at	datetime
- event	last_executed	datetime
- event	starts	datetime
-@@ -852,7 +861,7 @@
- flush privileges;
- SELECT table_schema, count(*) FROM information_schema.TABLES WHERE table_schema IN ('mysql', 'INFORMATION_SCHEMA', 'test', 'mysqltest') AND table_name<>'ndb_binlog_index' AND table_name<>'ndb_apply_status' GROUP BY TABLE_SCHEMA;
- table_schema	count(*)
--information_schema	28
-+information_schema	35
- mysql	22
- create table t1 (i int, j int);
- create trigger trg1 before insert on t1 for each row
-@@ -1267,6 +1276,13 @@
- TRIGGERS	TRIGGER_SCHEMA
- USER_PRIVILEGES	GRANTEE
- VIEWS	TABLE_SCHEMA
-+INNODB_CMP_RESET	page_size
-+INNODB_TRX	trx_id
-+INNODB_CMPMEM_RESET	page_size
-+INNODB_LOCK_WAITS	requesting_trx_id
-+INNODB_CMPMEM	page_size
-+INNODB_CMP	page_size
-+INNODB_LOCKS	lock_id
- SELECT t.table_name, c1.column_name
- FROM information_schema.tables t
- INNER JOIN
-@@ -1310,6 +1326,13 @@
- TRIGGERS	TRIGGER_SCHEMA
- USER_PRIVILEGES	GRANTEE
- VIEWS	TABLE_SCHEMA
-+INNODB_CMP_RESET	page_size
-+INNODB_TRX	trx_id
-+INNODB_CMPMEM_RESET	page_size
-+INNODB_LOCK_WAITS	requesting_trx_id
-+INNODB_CMPMEM	page_size
-+INNODB_CMP	page_size
-+INNODB_LOCKS	lock_id
- SELECT MAX(table_name) FROM information_schema.tables WHERE table_schema IN ('mysql', 'INFORMATION_SCHEMA', 'test');
- MAX(table_name)
- VIEWS
-@@ -1386,6 +1409,13 @@
- FILES	information_schema.FILES	1
- GLOBAL_STATUS	information_schema.GLOBAL_STATUS	1
- GLOBAL_VARIABLES	information_schema.GLOBAL_VARIABLES	1
-+INNODB_CMP	information_schema.INNODB_CMP	1
-+INNODB_CMPMEM	information_schema.INNODB_CMPMEM	1
-+INNODB_CMPMEM_RESET	information_schema.INNODB_CMPMEM_RESET	1
-+INNODB_CMP_RESET	information_schema.INNODB_CMP_RESET	1
-+INNODB_LOCKS	information_schema.INNODB_LOCKS	1
-+INNODB_LOCK_WAITS	information_schema.INNODB_LOCK_WAITS	1
-+INNODB_TRX	information_schema.INNODB_TRX	1
- KEY_COLUMN_USAGE	information_schema.KEY_COLUMN_USAGE	1
- PARTITIONS	information_schema.PARTITIONS	1
- PLUGINS	information_schema.PLUGINS	1
-diff mysql-test/r/information_schema_db.result.orig mysql-test/r/information_schema_db.result
---- mysql-test/r/information_schema_db.result.orig	2008-08-04 09:27:49.000000000 +0300
-+++ mysql-test/r/information_schema_db.result	2008-10-07 12:26:31.000000000 +0300
-@@ -33,6 +33,13 @@
- TRIGGERS
- USER_PRIVILEGES
- VIEWS
-+INNODB_CMP_RESET
-+INNODB_TRX
-+INNODB_CMPMEM_RESET
-+INNODB_LOCK_WAITS
-+INNODB_CMPMEM
-+INNODB_CMP
-+INNODB_LOCKS
- show tables from INFORMATION_SCHEMA like 'T%';
- Tables_in_information_schema (T%)
- TABLES
-diff mysql-test/r/mysqlshow.result.orig mysql-test/r/mysqlshow.result
---- mysql-test/r/mysqlshow.result.orig	2008-08-04 09:27:51.000000000 +0300
-+++ mysql-test/r/mysqlshow.result	2008-10-07 12:35:39.000000000 +0300
-@@ -107,6 +107,13 @@
- | TRIGGERS                              |
- | USER_PRIVILEGES                       |
- | VIEWS                                 |
-+| INNODB_CMP_RESET                      |
-+| INNODB_TRX                            |
-+| INNODB_CMPMEM_RESET                   |
-+| INNODB_LOCK_WAITS                     |
-+| INNODB_CMPMEM                         |
-+| INNODB_CMP                            |
-+| INNODB_LOCKS                          |
- +---------------------------------------+
- Database: INFORMATION_SCHEMA
- +---------------------------------------+
-@@ -140,6 +147,13 @@
- | TRIGGERS                              |
- | USER_PRIVILEGES                       |
- | VIEWS                                 |
-+| INNODB_CMP_RESET                      |
-+| INNODB_TRX                            |
-+| INNODB_CMPMEM_RESET                   |
-+| INNODB_LOCK_WAITS                     |
-+| INNODB_CMPMEM                         |
-+| INNODB_CMP                            |
-+| INNODB_LOCKS                          |
- +---------------------------------------+
- Wildcard: inf_rmation_schema
- +--------------------+

storage/innobase/mysql-test/patches/innodb_change_buffering_basic.diff

---- mysql-test/suite/sys_vars/t/innodb_change_buffering_basic.test.orig	Mon Mar 15 16:15:22 2010
-+++ mysql-test/suite/sys_vars/t/innodb_change_buffering_basic.test	Fri Mar 19 01:19:09 2010
-@@ -11,8 +11,8 @@
- #
- # exists as global only
- #
----echo Valid values are 'inserts' and 'none' 
--select @@global.innodb_change_buffering in ('inserts', 'none');
-+--echo Valid values are 'inserts', 'deletes', 'changes', 'purges', 'all', and 'none' 
-+select @@global.innodb_change_buffering in ('inserts', 'deletes', 'changes', 'purges', 'all', 'none');
- select @@global.innodb_change_buffering;
- --error ER_INCORRECT_GLOBAL_LOCAL_VAR
- select @@session.innodb_change_buffering;
-
---- mysql-test/suite/sys_vars/r/innodb_change_buffering_basic.result.orig	Mon Mar 15 16:15:22 2010
-+++ mysql-test/suite/sys_vars/r/innodb_change_buffering_basic.result	Fri Mar 19 01:23:58 2010
-@@ -1,28 +1,28 @@
- SET @start_global_value = @@global.innodb_change_buffering;
- SELECT @start_global_value;
- @start_global_value
--inserts
--Valid values are 'inserts' and 'none' 
--select @@global.innodb_change_buffering in ('inserts', 'none');
--@@global.innodb_change_buffering in ('inserts', 'none')
-+all
-+Valid values are 'inserts', 'deletes', 'changes', 'purges', 'all', and 'none' 
-+select @@global.innodb_change_buffering in ('inserts', 'deletes', 'changes', 'purges', 'all', 'none');
-+@@global.innodb_change_buffering in ('inserts', 'deletes', 'changes', 'purges', 'all', 'none')
- 1
- select @@global.innodb_change_buffering;
- @@global.innodb_change_buffering
--inserts
-+all
- select @@session.innodb_change_buffering;
- ERROR HY000: Variable 'innodb_change_buffering' is a GLOBAL variable
- show global variables like 'innodb_change_buffering';
- Variable_name	Value
--innodb_change_buffering	inserts
-+innodb_change_buffering	all
- show session variables like 'innodb_change_buffering';
- Variable_name	Value
--innodb_change_buffering	inserts
-+innodb_change_buffering	all
- select * from information_schema.global_variables where variable_name='innodb_change_buffering';
- VARIABLE_NAME	VARIABLE_VALUE
--INNODB_CHANGE_BUFFERING	inserts
-+INNODB_CHANGE_BUFFERING	all
- select * from information_schema.session_variables where variable_name='innodb_change_buffering';
- VARIABLE_NAME	VARIABLE_VALUE
--INNODB_CHANGE_BUFFERING	inserts
-+INNODB_CHANGE_BUFFERING	all
- set global innodb_change_buffering='none';
- select @@global.innodb_change_buffering;
- @@global.innodb_change_buffering
-@@ -60,4 +60,4 @@
- SET @@global.innodb_change_buffering = @start_global_value;
- SELECT @@global.innodb_change_buffering;
- @@global.innodb_change_buffering
--inserts
-+all

storage/innobase/mysql-test/patches/innodb_file_per_table.diff

-diff mysql-test/suite/sys_vars/t/innodb_file_per_table_basic.test.orig mysql-test/suite/sys_vars/t/innodb_file_per_table_basic.test
---- mysql-test/suite/sys_vars/t/innodb_file_per_table_basic.test.orig	2008-10-07 11:32:30.000000000 +0300
-+++ mysql-test/suite/sys_vars/t/innodb_file_per_table_basic.test	2008-10-07 11:52:14.000000000 +0300
-@@ -37,10 +37,6 @@
- #   Check if Value can set                                         #
- ####################################################################
- 
----error ER_INCORRECT_GLOBAL_LOCAL_VAR
--SET @@GLOBAL.innodb_file_per_table=1;
----echo Expected error 'Read only variable'
--
- SELECT COUNT(@@GLOBAL.innodb_file_per_table);
- --echo 1 Expected
- 
-@@ -52,7 +48,7 @@
- # Check if the value in GLOBAL Table matches value in variable  #
- #################################################################
- 
--SELECT @@GLOBAL.innodb_file_per_table = VARIABLE_VALUE
-+SELECT IF(@@GLOBAL.innodb_file_per_table,'ON','OFF') = VARIABLE_VALUE
- FROM INFORMATION_SCHEMA.GLOBAL_VARIABLES
- WHERE VARIABLE_NAME='innodb_file_per_table';
- --echo 1 Expected
-diff mysql-test/suite/sys_vars/r/innodb_file_per_table_basic.result.orig mysql-test/suite/sys_vars/r/innodb_file_per_table_basic.result
---- mysql-test/suite/sys_vars/r/innodb_file_per_table_basic.result.orig	2008-10-07 11:32:02.000000000 +0300
-+++ mysql-test/suite/sys_vars/r/innodb_file_per_table_basic.result	2008-10-07 11:52:47.000000000 +0300
-@@ -4,18 +4,15 @@
- 1
- 1 Expected
- '#---------------------BS_STVARS_028_02----------------------#'
--SET @@GLOBAL.innodb_file_per_table=1;
--ERROR HY000: Variable 'innodb_file_per_table' is a read only variable
--Expected error 'Read only variable'
- SELECT COUNT(@@GLOBAL.innodb_file_per_table);
- COUNT(@@GLOBAL.innodb_file_per_table)
- 1
- 1 Expected
- '#---------------------BS_STVARS_028_03----------------------#'
--SELECT @@GLOBAL.innodb_file_per_table = VARIABLE_VALUE
-+SELECT IF(@@GLOBAL.innodb_file_per_table,'ON','OFF') = VARIABLE_VALUE
- FROM INFORMATION_SCHEMA.GLOBAL_VARIABLES
- WHERE VARIABLE_NAME='innodb_file_per_table';
--@@GLOBAL.innodb_file_per_table = VARIABLE_VALUE
-+IF(@@GLOBAL.innodb_file_per_table,'ON','OFF') = VARIABLE_VALUE
- 1
- 1 Expected
- SELECT COUNT(@@GLOBAL.innodb_file_per_table);

storage/innobase/mysql-test/patches/innodb_lock_wait_timeout.diff

---- mysql-test/suite/sys_vars/t/innodb_lock_wait_timeout_basic.test.orig	2008-08-04 09:28:16.000000000 +0300
-+++ mysql-test/suite/sys_vars/t/innodb_lock_wait_timeout_basic.test	2008-10-07 11:14:15.000000000 +0300
-@@ -37,10 +37,6 @@
- #   Check if Value can set                                         #
- ####################################################################
- 
----error ER_INCORRECT_GLOBAL_LOCAL_VAR
--SET @@GLOBAL.innodb_lock_wait_timeout=1;
----echo Expected error 'Read only variable'
--
- SELECT COUNT(@@GLOBAL.innodb_lock_wait_timeout);
- --echo 1 Expected
- 
-@@ -84,13 +80,9 @@
- SELECT COUNT(@@innodb_lock_wait_timeout);
- --echo 1 Expected
- 
----Error ER_INCORRECT_GLOBAL_LOCAL_VAR
- SELECT COUNT(@@local.innodb_lock_wait_timeout);
----echo Expected error 'Variable is a GLOBAL variable'
- 
----Error ER_INCORRECT_GLOBAL_LOCAL_VAR
- SELECT COUNT(@@SESSION.innodb_lock_wait_timeout);
----echo Expected error 'Variable is a GLOBAL variable'
- 
- SELECT COUNT(@@GLOBAL.innodb_lock_wait_timeout);
- --echo 1 Expected
---- mysql-test/suite/sys_vars/r/innodb_lock_wait_timeout_basic.result.orig	2008-08-04 09:27:50.000000000 +0300
-+++ mysql-test/suite/sys_vars/r/innodb_lock_wait_timeout_basic.result	2008-10-07 11:15:14.000000000 +0300
-@@ -4,9 +4,6 @@
- 1
- 1 Expected
- '#---------------------BS_STVARS_032_02----------------------#'
--SET @@GLOBAL.innodb_lock_wait_timeout=1;
--ERROR HY000: Variable 'innodb_lock_wait_timeout' is a read only variable
--Expected error 'Read only variable'
- SELECT COUNT(@@GLOBAL.innodb_lock_wait_timeout);
- COUNT(@@GLOBAL.innodb_lock_wait_timeout)
- 1
-@@ -39,11 +36,11 @@
- 1
- 1 Expected
- SELECT COUNT(@@local.innodb_lock_wait_timeout);
--ERROR HY000: Variable 'innodb_lock_wait_timeout' is a GLOBAL variable
--Expected error 'Variable is a GLOBAL variable'
-+COUNT(@@local.innodb_lock_wait_timeout)
-+1
- SELECT COUNT(@@SESSION.innodb_lock_wait_timeout);
--ERROR HY000: Variable 'innodb_lock_wait_timeout' is a GLOBAL variable
--Expected error 'Variable is a GLOBAL variable'
-+COUNT(@@SESSION.innodb_lock_wait_timeout)
-+1
- SELECT COUNT(@@GLOBAL.innodb_lock_wait_timeout);
- COUNT(@@GLOBAL.innodb_lock_wait_timeout)
- 1

storage/innobase/mysql-test/patches/innodb_thread_concurrency_basic.diff

---- mysql-test/suite/sys_vars/r/innodb_thread_concurrency_basic.result.orig	2008-12-04 18:45:52 -06:00
-+++ mysql-test/suite/sys_vars/r/innodb_thread_concurrency_basic.result	2009-02-12 02:05:48 -06:00
-@@ -1,19 +1,19 @@
- SET @global_start_value = @@global.innodb_thread_concurrency;
- SELECT @global_start_value;
- @global_start_value
--8
-+0
- '#--------------------FN_DYNVARS_046_01------------------------#'
- SET @@global.innodb_thread_concurrency = 0;
- SET @@global.innodb_thread_concurrency = DEFAULT;
- SELECT @@global.innodb_thread_concurrency;
- @@global.innodb_thread_concurrency
--8
-+0
- '#---------------------FN_DYNVARS_046_02-------------------------#'
- SET innodb_thread_concurrency = 1;
- ERROR HY000: Variable 'innodb_thread_concurrency' is a GLOBAL variable and should be set with SET GLOBAL
- SELECT @@innodb_thread_concurrency;
- @@innodb_thread_concurrency
--8
-+0
- SELECT local.innodb_thread_concurrency;
- ERROR 42S02: Unknown table 'local' in field list
- SET global innodb_thread_concurrency = 0;
-@@ -93,4 +93,4 @@
- SET @@global.innodb_thread_concurrency = @global_start_value;
- SELECT @@global.innodb_thread_concurrency;
- @@global.innodb_thread_concurrency
--8
-+0

storage/innobase/mysql-test/patches/partition_innodb.diff

-The partition_innodb test only fails if run immediately after innodb_trx_weight.
-The reason for this failure is that innodb_trx_weight creates deadlocks and
-leaves something like this in the SHOW ENGINE INNODB STATUS output:
-
-  ------------------------
-  LATEST DETECTED DEADLOCK
-  ------------------------
-  090213 10:26:25
-  *** (1) TRANSACTION:
-  TRANSACTION 313, ACTIVE 0 sec, OS thread id 13644672 inserting
-  mysql tables in use 1, locked 1
-  LOCK WAIT 4 lock struct(s), heap size 488, 3 row lock(s)
-  MySQL thread id 3, query id 36 localhost root update
-
-The regular expressions that partition_innodb is using are intended to extract
-the lock structs and row locks numbers from another part of the output:
-
-  ------------
-  TRANSACTIONS
-  ------------
-  Trx id counter 31D
-  Purge done for trx's n:o < 0 undo n:o < 0
-  History list length 4
-  LIST OF TRANSACTIONS FOR EACH SESSION:
-  ---TRANSACTION 0, not started, OS thread id 13645056
-  0 lock struct(s), heap size 488, 0 row lock(s)
-  MySQL thread id 8, query id 81 localhost root
-
-In the InnoDB Plugin a transaction id is not printed as 2 consecutive
-decimal integers (as it is in InnoDB 5.1) but rather as a single
-hexadecimal integer. Thus the regular expressions somehow pick the wrong
-part of the SHOW ENGINE INNODB STATUS output.
-
-So after the regular expressions are adjusted to the InnoDB Plugin's variant
-of trx_id prinout, then they pick the expected part of the output.
-
-This patch cannot be proposed to MySQL because the failures occur only
-in this tree and do not occur in the standard InnoDB 5.1.
-
---- mysql-test/t/partition_innodb.test	2008-11-14 22:51:17 +0000
-+++ mysql-test/t/partition_innodb.test	2009-02-13 07:36:07 +0000
-@@ -27,14 +27,14 @@
- 
- # grouping/referencing in replace_regex is very slow on long strings,
- # removing all before/after the interesting row before grouping/referencing
----replace_regex /.*---TRANSACTION [0-9]+ [0-9]+, .*, OS thread id [0-9]+// /MySQL thread id [0-9]+, query id [0-9]+ .*// /.*([0-9]+ lock struct\(s\)), heap size [0-9]+, ([0-9]+ row lock\(s\)).*/\1 \2/
-+--replace_regex /.*---TRANSACTION [0-9A-F]+, .*, OS thread id [0-9]+// /MySQL thread id [0-9]+, query id [0-9]+ .*// /.*([0-9]+ lock struct\(s\)), heap size [0-9]+, ([0-9]+ row lock\(s\)).*/\1 \2/
- SHOW ENGINE InnoDB STATUS;
- 
- UPDATE t1 SET data = data*2 WHERE data = 2;
- 
- # grouping/referencing in replace_regex is very slow on long strings,
- # removing all before/after the interesting row before grouping/referencing
----replace_regex /.*---TRANSACTION [0-9]+ [0-9]+, .*, OS thread id [0-9]+// /MySQL thread id [0-9]+, query id [0-9]+ .*// /.*([0-9]+ lock struct\(s\)), heap size [0-9]+, ([0-9]+ row lock\(s\)).*/\1 \2/
-+--replace_regex /.*---TRANSACTION [0-9A-F]+, .*, OS thread id [0-9]+// /MySQL thread id [0-9]+, query id [0-9]+ .*// /.*([0-9]+ lock struct\(s\)), heap size [0-9]+, ([0-9]+ row lock\(s\)).*/\1 \2/
- SHOW ENGINE InnoDB STATUS;
- 
- SET @@session.tx_isolation = @old_tx_isolation;
-

storage/innobase/os/os0sync.c

@@ -72,6 +72,9 @@ UNIV_INTERN ulint	os_event_count		= 0;
 UNIV_INTERN ulint	os_mutex_count		= 0;
 UNIV_INTERN ulint	os_fast_mutex_count	= 0;
 
+/* The number of microsecnds in a second. */
+static const ulint MICROSECS_IN_A_SECOND = 1000000;
+
 /* Because a mutex is embedded inside an event and there is an
 event embedded inside a mutex, on free, this generates a recursive call.
 This version of the free event function doesn't acquire the global lock */
@@ -121,6 +124,57 @@ os_cond_init(
 #endif
 }
 
+/*********************************************************//**
+Do a timed wait on condition variable.
+@return TRUE if timed out, FALSE otherwise */
+UNIV_INLINE
+ibool
+os_cond_wait_timed(
+/*===============*/
+	os_cond_t*		cond,		/*!< in: condition variable. */
+	os_fast_mutex_t*	mutex,		/*!< in: fast mutex */
+#ifndef __WIN__
+	const struct timespec*	abstime		/*!< in: timeout */
+#else
+	ulint			time_in_ms	/*!< in: timeout in
+						milliseconds*/
+#endif /* !__WIN__ */
+)
+{
+#ifdef __WIN__
+	BOOL			ret;
+
+	ut_a(sleep_condition_variable != NULL);
+
+	ret = sleep_condition_variable(cond, mutex, time_in_ms);
+
+	if (!ret && GetLastError() == WAIT_TIMEOUT) {
+		return(TRUE);
+	}
+
+	ut_a(ret);
+
+	return(FALSE);
+#else
+	int	ret;
+
+	ret = pthread_cond_timedwait(cond, mutex, abstime);
+
+	switch (ret) {
+	case 0:
+	case ETIMEDOUT:
+	       	break;
+
+	default:
+		fprintf(stderr, "  InnoDB: pthread_cond_timedwait() returned: "
+				"%d: abstime={%lu,%lu}\n",
+			       	ret, abstime->tv_sec, abstime->tv_nsec);
+		ut_error;
+	}
+
+	return(ret == ETIMEDOUT);
+#endif
+}
 /*********************************************************//**
 Wait on condition variable */
 UNIV_INLINE
@@ -572,6 +626,128 @@ os_event_wait_low(
 	}
 }
 
+/**********************************************************//**
+Waits for an event object until it is in the signaled state or
+a timeout is exceeded.
+@return	0 if success, OS_SYNC_TIME_EXCEEDED if timeout was exceeded */
+UNIV_INTERN
+ulint
+os_event_wait_time_low(
+/*===================*/
+	os_event_t	event,			/*!< in: event to wait */
+	ulint		time_in_usec,		/*!< in: timeout in
+						microseconds, or
+						OS_SYNC_INFINITE_TIME */
+	ib_int64_t	reset_sig_count)	/*!< in: zero or the value
+						returned by previous call of
+						os_event_reset(). */
+
+{
+	ibool		timed_out;
+	ib_int64_t	old_signal_count;
+
+#ifdef __WIN__
+	DWORD		time_in_ms;
+
+	if (!srv_use_native_conditions) {
+		DWORD	err;
+
+		ut_a(event);
+
+		if (time_in_usec != OS_SYNC_INFINITE_TIME) {
+			time_in_ms = time_in_ms / 1000;
+			err = WaitForSingleObject(event->handle, time_in_ms);
+		} else {
+			err = WaitForSingleObject(event->handle, INFINITE);
+		}
+
+		if (err == WAIT_OBJECT_0) {
+			return(0);
+		} else if (err == WAIT_TIMEOUT) {
+			return(OS_SYNC_TIME_EXCEEDED);
+		}
+
+		ut_error;
+		/* Dummy value to eliminate compiler warning. */
+		return(42);
+	} else {
+		ut_a(sleep_condition_variable != NULL);
+
+		if (time_in_usec != OS_SYNC_INFINITE_TIME) {
+			time_in_ms = time_in_usec / 1000;
+		} else {
+			time_in_ms = INFINITE;
+		}
+	}
+#else
+	struct timespec	abstime;
+
+	if (time_in_usec != OS_SYNC_INFINITE_TIME) {
+		struct timeval	tv;
+		int		ret;
+		ulint		sec;
+		ulint		usec;
+
+		ret = ut_usectime(&sec, &usec);
+		ut_a(ret == 0);
+
+		tv.tv_sec = sec;
+		tv.tv_usec = usec;
+
+		tv.tv_usec += time_in_usec;
+
+		if ((ulint) tv.tv_usec >= MICROSECS_IN_A_SECOND) {
+			tv.tv_sec += time_in_usec / MICROSECS_IN_A_SECOND;
+			tv.tv_usec %= MICROSECS_IN_A_SECOND;
+		}
+
+		abstime.tv_sec  = tv.tv_sec;
+		abstime.tv_nsec = tv.tv_usec * 1000;
+	} else {
+		abstime.tv_nsec = 999999999;
+		abstime.tv_sec = (time_t) ULINT_MAX;
+	}
+
+	ut_a(abstime.tv_nsec <= 999999999);
+
+#endif /* __WIN__ */
+
+	os_fast_mutex_lock(&event->os_mutex);
+
+	if (reset_sig_count) {
+		old_signal_count = reset_sig_count;
+	} else {
+		old_signal_count = event->signal_count;
+	}
+
+	do {
+		if (event->is_set == TRUE
+		    || event->signal_count != old_signal_count) {
+
+			break;
+		}
+
+		timed_out = os_cond_wait_timed(
+			&event->cond_var, &event->os_mutex,
+#ifndef __WIN__
+			&abstime
+#else
+			time_in_ms
+#endif /* !__WIN__ */
+		);
+
+	} while (!timed_out);
+
+	os_fast_mutex_unlock(&event->os_mutex);
+
+	if (srv_shutdown_state == SRV_SHUTDOWN_EXIT_THREADS) {
+
+		os_thread_exit(NULL);
+	}
+
+	return(timed_out ? OS_SYNC_TIME_EXCEEDED : 0);
+}
+
 /*********************************************************//**
 Creates an operating system mutex semaphore. Because these are slow, the
 mutex semaphore of InnoDB itself (mutex_t) should be used where possible.

storage/innobase/plug.in

@@ -146,10 +146,11 @@ MYSQL_PLUGIN_ACTIONS(innobase,  [
 
   AC_MSG_CHECKING(whether Solaris libc atomic functions are available)
   # either define HAVE_IB_SOLARIS_ATOMICS or not
-  AC_CHECK_FUNCS(atomic_add_long \
+  AC_CHECK_FUNCS(atomic_cas_ulong \
 		 atomic_cas_32 \
 		 atomic_cas_64 \
-		 atomic_cas_ulong,
+		 atomic_add_long_nv \
+		 atomic_swap_uchar,
 
 		 AC_DEFINE([HAVE_IB_SOLARIS_ATOMICS], [1],
 			   [Define to 1 if Solaris libc atomic functions \

storage/innobase/rem/rem0rec.c

@@ -1751,7 +1751,7 @@ rec_print(
 /*======*/
 	FILE*			file,	/*!< in: file where to print */
 	const rec_t*		rec,	/*!< in: physical record */
-	dict_index_t*	index)	/*!< in: record descriptor */
+	const dict_index_t*	index)	/*!< in: record descriptor */
 {
 	ut_ad(index);
 

storage/innobase/row/row0mysql.c

@@ -503,7 +503,7 @@ row_mysql_convert_row_to_innobase(
 					row is used, as row may contain
 					pointers to this record! */
 {
-	mysql_row_templ_t*	templ;
+	const mysql_row_templ_t*templ;
 	dfield_t*		dfield;
 	ulint			i;
 

storage/innobase/row/row0sel.c

@@ -2667,21 +2667,22 @@ row_sel_store_mysql_rec(
 	row_prebuilt_t*	prebuilt,	/*!< in: prebuilt struct */
 	const rec_t*	rec,		/*!< in: Innobase record in the index
 					which was described in prebuilt's
-					template; must be protected by
-					a page latch */
+					template, or in the clustered index;
+					must be protected by a page latch */
+	ibool		rec_clust,	/*!< in: TRUE if rec is in the
+					clustered index instead of
+					prebuilt->index */
 	const ulint*	offsets)	/*!< in: array returned by
-					rec_get_offsets() */
+					rec_get_offsets(rec) */
 {
-	mysql_row_templ_t*	templ;
 	mem_heap_t*	extern_field_heap	= NULL;
 	mem_heap_t*	heap;
-	const byte*		data;
-	ulint			len;
 	ulint		i;
 
 	ut_ad(prebuilt->mysql_template);
 	ut_ad(prebuilt->default_rec);
 	ut_ad(rec_offs_validate(rec, NULL, offsets));
+	ut_ad(!rec_get_deleted_flag(rec, rec_offs_comp(offsets)));
 
 	if (UNIV_LIKELY_NULL(prebuilt->blob_heap)) {
 		mem_heap_free(prebuilt->blob_heap);
@@ -2690,10 +2691,15 @@ row_sel_store_mysql_rec(
 
 	for (i = 0; i < prebuilt->n_template; i++) {
 
-		templ = prebuilt->mysql_template + i;
+		const mysql_row_templ_t*templ = prebuilt->mysql_template + i;
+		const byte*		data;
+		ulint			len;
+		ulint			field_no;
 
-		if (UNIV_UNLIKELY(rec_offs_nth_extern(offsets,
-						      templ->rec_field_no))) {
+		field_no = rec_clust
+			? templ->clust_rec_field_no : templ->rec_field_no;
+
+		if (UNIV_UNLIKELY(rec_offs_nth_extern(offsets, field_no))) {
 
 			/* Copy an externally stored field to the temporary
 			heap */
@@ -2721,7 +2727,7 @@ row_sel_store_mysql_rec(
 			data = btr_rec_copy_externally_stored_field(
 				rec, offsets,
 				dict_table_zip_size(prebuilt->table),
-				templ->rec_field_no, &len, heap);
+				field_no, &len, heap);
 
 			if (UNIV_UNLIKELY(!data)) {
 				/* The externally stored field
@@ -2742,8 +2748,7 @@ row_sel_store_mysql_rec(
 		} else {
 			/* Field is stored in the row. */
 
-			data = rec_get_nth_field(rec, offsets,
-						 templ->rec_field_no, &len);
+			data = rec_get_nth_field(rec, offsets, field_no, &len);
 
 			if (UNIV_UNLIKELY(templ->type == DATA_BLOB)
 			    && len != UNIV_SQL_NULL) {
@@ -3105,7 +3110,7 @@ row_sel_pop_cached_row_for_mysql(
 	row_prebuilt_t*	prebuilt)	/*!< in: prebuilt struct */
 {
 	ulint			i;
-	mysql_row_templ_t*	templ;
+	const mysql_row_templ_t*templ;
 	byte*			cached_rec;
 	ut_ad(prebuilt->n_fetch_cached > 0);
 	ut_ad(prebuilt->mysql_prefix_len <= prebuilt->mysql_row_len);
@@ -3162,15 +3167,21 @@ ibool
 row_sel_push_cache_row_for_mysql(
 /*=============================*/
 	row_prebuilt_t*	prebuilt,	/*!< in: prebuilt struct */
-	const rec_t*	rec,		/*!< in: record to push; must
-					be protected by a page latch */
-	const ulint*	offsets)	/*!< in: rec_get_offsets() */
+	const rec_t*	rec,		/*!< in: record to push, in the index
+					which was described in prebuilt's
+					template, or in the clustered index;
+					must be protected by a page latch */
+	ibool		rec_clust,	/*!< in: TRUE if rec is in the
+					clustered index instead of
+					prebuilt->index */
+	const ulint*	offsets)	/*!< in: rec_get_offsets(rec) */
 {
 	byte*	buf;
 	ulint	i;
 
 	ut_ad(prebuilt->n_fetch_cached < MYSQL_FETCH_CACHE_SIZE);
 	ut_ad(rec_offs_validate(rec, NULL, offsets));
+	ut_ad(!rec_get_deleted_flag(rec, rec_offs_comp(offsets)));
 	ut_a(!prebuilt->templ_contains_blob);
 
 	if (prebuilt->fetch_cache[0] == NULL) {
@@ -3199,7 +3210,7 @@ row_sel_push_cache_row_for_mysql(
 	if (UNIV_UNLIKELY(!row_sel_store_mysql_rec(
 				  prebuilt->fetch_cache[
 					  prebuilt->n_fetch_cached],
-				  prebuilt, rec, offsets))) {
+				  prebuilt, rec, rec_clust, offsets))) {
 		return(FALSE);
 	}
 
@@ -3600,7 +3611,8 @@ row_search_for_mysql(
 				ut_ad(!rec_get_deleted_flag(rec, comp));
 
 				if (!row_sel_store_mysql_rec(buf, prebuilt,
-							     rec, offsets)) {
+							     rec, FALSE,
+							     offsets)) {
 					/* Only fresh inserts may contain
 					incomplete externally stored
 					columns. Pretend that such
@@ -4234,7 +4246,6 @@ no_gap_lock:
 			is necessary, because we can only get the undo
 			information via the clustered index record. */
 
-			ut_ad(index != clust_index);
 			ut_ad(!dict_index_is_clust(index));
 
 			if (!lock_sec_rec_cons_read_sees(
@@ -4350,34 +4361,19 @@ requires_clust_rec:
 			goto next_rec;
 		}
 
-		if (prebuilt->need_to_access_clustered) {
-
 		result_rec = clust_rec;
-
-			ut_ad(rec_offs_validate(result_rec, clust_index,
-						offsets));
+		ut_ad(rec_offs_validate(result_rec, clust_index, offsets));
 	} else {
-			/* We used 'offsets' for the clust rec, recalculate
-			them for 'rec' */
-			offsets = rec_get_offsets(rec, index, offsets,
-						  ULINT_UNDEFINED, &heap);
 		result_rec = rec;
 	}
 
-		/* result_rec can legitimately be delete-marked
-		now that it has been established that it points to a
-		clustered index record that exists in the read view. */
-	} else {
-		result_rec = rec;
-		ut_ad(!rec_get_deleted_flag(rec, comp));
-	}
-
 	/* We found a qualifying record 'result_rec'. At this point,
 	'offsets' are associated with 'result_rec'. */
 
 	ut_ad(rec_offs_validate(result_rec,
 				result_rec != rec ? clust_index : index,
 				offsets));
+	ut_ad(!rec_get_deleted_flag(result_rec, comp));
 
 	/* At this point, the clustered index record is protected
 	by a page latch that was acquired when pcur was positioned.
@@ -4402,6 +4398,7 @@ requires_clust_rec:
 		cursor. */
 
 		if (!row_sel_push_cache_row_for_mysql(prebuilt, result_rec,
+						      result_rec != rec,
 						      offsets)) {
 			/* Only fresh inserts may contain incomplete
 			externally stored columns. Pretend that such
@@ -4419,15 +4416,31 @@ requires_clust_rec:
 
 		goto next_rec;
 	} else {
-		if (prebuilt->template_type == ROW_MYSQL_DUMMY_TEMPLATE) {
+		if (UNIV_UNLIKELY
+		    (prebuilt->template_type == ROW_MYSQL_DUMMY_TEMPLATE)) {
+			/* CHECK TABLE: fetch the row */
+
+			if (result_rec != rec
+			    && !prebuilt->need_to_access_clustered) {
+				/* We used 'offsets' for the clust
+				rec, recalculate them for 'rec' */
+				offsets = rec_get_offsets(rec, index, offsets,
+							  ULINT_UNDEFINED,
+							  &heap);
+				result_rec = rec;
+			}
+
 			memcpy(buf + 4, result_rec
 			       - rec_offs_extra_size(offsets),
 			       rec_offs_size(offsets));
 			mach_write_to_4(buf,
 					rec_offs_extra_size(offsets) + 4);
 		} else {
-			if (!row_sel_store_mysql_rec(buf, prebuilt,
-						     result_rec, offsets)) {
+			/* Returning a row to MySQL */
+
+			if (!row_sel_store_mysql_rec(buf, prebuilt, result_rec,
+						     result_rec != rec,
+						     offsets)) {
 				/* Only fresh inserts may contain
 				incomplete externally stored
 				columns. Pretend that such records do

storage/innobase/row/row0upd.c

@@ -466,8 +466,11 @@ row_upd_changes_field_size_or_external(
 #endif /* !UNIV_HOTBACKUP */
 
 /***********************************************************//**
-Replaces the new column values stored in the update vector to the record
-given. No field size changes are allowed. */
+Replaces the new column values stored in the update vector to the
+record given. No field size changes are allowed. This function is
+usually invoked on a clustered index. The only use case for a
+secondary index is row_ins_sec_index_entry_by_modify() or its
+counterpart in ibuf_insert_to_index_page(). */
 UNIV_INTERN
 void
 row_upd_rec_in_place(

storage/innobase/srv/srv0srv.c

@@ -695,6 +695,12 @@ struct srv_slot_struct{
 /* Table for MySQL threads where they will be suspended to wait for locks */
 UNIV_INTERN srv_slot_t*	srv_mysql_table = NULL;
 
+UNIV_INTERN os_event_t	srv_timeout_event;
+
+UNIV_INTERN os_event_t	srv_monitor_event;
+
+UNIV_INTERN os_event_t	srv_error_event;
+
 UNIV_INTERN os_event_t	srv_lock_timeout_thread_event;
 
 UNIV_INTERN srv_sys_t*	srv_sys	= NULL;
@@ -1012,6 +1018,12 @@ srv_init(void)
 		ut_a(slot->event);
 	}
 
+	srv_error_event = os_event_create(NULL);
+
+	srv_timeout_event = os_event_create(NULL);
+
+	srv_monitor_event = os_event_create(NULL);
+
 	srv_lock_timeout_thread_event = os_event_create(NULL);
 
 	for (i = 0; i < SRV_MASTER + 1; i++) {
@@ -2049,6 +2061,7 @@ srv_monitor_thread(
 			/*!< in: a dummy parameter required by
 			os_thread_create */
 {
+	ib_int64_t	sig_count;
 	double		time_elapsed;
 	time_t		current_time;
 	time_t		last_table_monitor_time;
@@ -2067,26 +2080,28 @@ srv_monitor_thread(
 #endif
 
 	UT_NOT_USED(arg);
-	srv_last_monitor_time = time(NULL);
-	last_table_monitor_time = time(NULL);
-	last_tablespace_monitor_time = time(NULL);
-	last_monitor_time = time(NULL);
+	srv_last_monitor_time = ut_time();
+	last_table_monitor_time = ut_time();
+	last_tablespace_monitor_time = ut_time();
+	last_monitor_time = ut_time();
 	mutex_skipped = 0;
 	last_srv_print_monitor = srv_print_innodb_monitor;
 loop:
 	srv_monitor_active = TRUE;
 
 	/* Wake up every 5 seconds to see if we need to print
-	monitor information. */
+	monitor information or if signalled at shutdown. */
 
-	os_thread_sleep(5000000);
+	sig_count = os_event_reset(srv_monitor_event);
 
-	current_time = time(NULL);
+	os_event_wait_time_low(srv_monitor_event, 5000000, sig_count);
+
+	current_time = ut_time();
 
 	time_elapsed = difftime(current_time, last_monitor_time);
 
 	if (time_elapsed > 15) {
-		last_monitor_time = time(NULL);
+		last_monitor_time = ut_time();
 
 		if (srv_print_innodb_monitor) {
 			/* Reset mutex_skipped counter everytime
@@ -2130,7 +2145,7 @@ loop:
 		if (srv_print_innodb_tablespace_monitor
 		    && difftime(current_time,
 				last_tablespace_monitor_time) > 60) {
-			last_tablespace_monitor_time = time(NULL);
+			last_tablespace_monitor_time = ut_time();
 
 			fputs("========================"
 			      "========================\n",
@@ -2156,7 +2171,7 @@ loop:
 		if (srv_print_innodb_table_monitor
 		    && difftime(current_time, last_table_monitor_time) > 60) {
 
-			last_table_monitor_time = time(NULL);
+			last_table_monitor_time = ut_time();
 
 			fputs("===========================================\n",
 			      stderr);
@@ -2216,16 +2231,20 @@ srv_lock_timeout_thread(
 	ibool		some_waits;
 	double		wait_time;
 	ulint		i;
+	ib_int64_t	sig_count;
 
 #ifdef UNIV_PFS_THREAD
 	pfs_register_thread(srv_lock_timeout_thread_key);
 #endif
 
 loop:
+
 	/* When someone is waiting for a lock, we wake up every second
 	and check if a timeout has passed for a lock wait */
 
-	os_thread_sleep(1000000);
+	sig_count = os_event_reset(srv_timeout_event);
+
+	os_event_wait_time_low(srv_timeout_event, 1000000, sig_count);
 
 	srv_lock_timeout_active = TRUE;
 
@@ -2320,6 +2339,7 @@ srv_error_monitor_thread(
 	ulint		fatal_cnt	= 0;
 	ib_uint64_t	old_lsn;
 	ib_uint64_t	new_lsn;
+	ib_int64_t	sig_count;
 
 	old_lsn = srv_start_lsn;
 
@@ -2395,7 +2415,9 @@ loop:
 
 	fflush(stderr);
 
-	os_thread_sleep(1000000);
+	sig_count = os_event_reset(srv_error_event);
+
+	os_event_wait_time_low(srv_error_event, 1000000, sig_count);
 
 	if (srv_shutdown_state < SRV_SHUTDOWN_CLEANUP) {
 
@@ -2646,6 +2668,21 @@ loop:
 	for (i = 0; i < 10; i++) {
 		ulint	cur_time = ut_time_ms();
 
+		/* ALTER TABLE in MySQL requires on Unix that the table handler
+		can drop tables lazily after there no longer are SELECT
+		queries to them. */
+
+		srv_main_thread_op_info = "doing background drop tables";
+
+		row_drop_tables_for_mysql_in_background();
+
+		srv_main_thread_op_info = "";
+
+		if (srv_fast_shutdown && srv_shutdown_state > 0) {
+
+			goto background_loop;
+		}
+
 		buf_get_total_stat(&buf_stat);
 
 		n_ios_old = log_sys->n_log_ios + buf_stat.n_pages_read
@@ -2654,7 +2691,8 @@ loop:
 		srv_main_thread_op_info = "sleeping";
 		srv_main_1_second_loops++;
 
-		if (next_itr_time > cur_time) {
+		if (next_itr_time > cur_time
+		    && srv_shutdown_state == SRV_SHUTDOWN_NONE) {
 
 			/* Get sleep interval in micro seconds. We use
 			ut_min() to avoid long sleep in case of
@@ -2668,21 +2706,6 @@ loop:
 		/* Each iteration should happen at 1 second interval. */
 		next_itr_time = ut_time_ms() + 1000;
 
-		/* ALTER TABLE in MySQL requires on Unix that the table handler
-		can drop tables lazily after there no longer are SELECT
-		queries to them. */
-
-		srv_main_thread_op_info = "doing background drop tables";
-
-		row_drop_tables_for_mysql_in_background();
-
-		srv_main_thread_op_info = "";
-
-		if (srv_fast_shutdown && srv_shutdown_state > 0) {
-
-			goto background_loop;
-		}
-
 		/* Flush logs if needed */
 		srv_sync_log_buffer_in_background();
 
@@ -2860,8 +2883,10 @@ background_loop:
 		MySQL tries to drop a table while there are still open handles
 		to it and we had to put it to the background drop queue.) */
 
+		if (srv_shutdown_state == SRV_SHUTDOWN_NONE) {
 			os_thread_sleep(100000);
 		}
+	}
 
 	if (srv_n_purge_threads == 0) {
 		srv_main_thread_op_info = "master purging";

storage/innobase/srv/srv0start.c

@@ -477,7 +477,6 @@ io_handler_thread(
 			the aio array */
 {
 	ulint	segment;
-	ulint	i;
 
 	segment = *((ulint*)arg);
 
@@ -490,7 +489,7 @@ io_handler_thread(
 	pfs_register_thread(io_handler_thread_key);
 #endif /* UNIV_PFS_THREAD */
 
-	for (i = 0;; i++) {
+	while (srv_shutdown_state != SRV_SHUTDOWN_EXIT_THREADS) {
 		fil_aio_wait(segment);
 
 		mutex_enter(&ios_mutex);
@@ -498,8 +497,6 @@ io_handler_thread(
 		mutex_exit(&ios_mutex);
 	}
 
-	thr_local_free(os_thread_get_curr_id());
-
 	/* We count the number of threads in os_thread_exit(). A created
 	thread should always use that to exit and not use return() to exit.
 	The thread actually never comes here because it is exited in an

storage/innobase/sync/sync0sync.c

@@ -1207,8 +1207,8 @@ sync_thread_add_level(
 
 	case SYNC_BUF_BLOCK:
 		/* Either the thread must own the buffer pool mutex
-		(buf_pool_mutex), or it is allowed to latch only ONE
-		buffer block (block->mutex or buf_pool_zip_mutex). */
+		(buf_pool->mutex), or it is allowed to latch only ONE
+		buffer block (block->mutex or buf_pool->zip_mutex). */
 		if (!sync_thread_levels_g(array, level, FALSE)) {
 			ut_a(sync_thread_levels_g(array, level - 1, TRUE));
 			ut_a(sync_thread_levels_contain(array, SYNC_BUF_POOL));

storage/innobase/trx/trx0i_s.c

@@ -38,8 +38,6 @@ Created July 17, 2007 Vasil Dimov
 
 #include <mysql/plugin.h>
 
-#include "mysql_addons.h"
-
 #include "buf0buf.h"
 #include "dict0dict.h"
 #include "ha0storage.h"

storage/innobase/trx/trx0trx.c

@@ -105,7 +105,11 @@ trx_create(
 	trx->is_purge = 0;
 	trx->is_recovered = 0;
 	trx->conc_state = TRX_NOT_STARTED;
-	trx->start_time = time(NULL);
+
+	trx->is_registered = 0;
+	trx->owns_prepare_mutex = 0;
+
+	trx->start_time = ut_time();
 
 	trx->isolation_level = TRX_ISO_REPEATABLE_READ;
 
@@ -124,7 +128,6 @@ trx_create(
 	trx->table_id = 0;
 
 	trx->mysql_thd = NULL;
-	trx->active_trans = 0;
 	trx->duplicates = 0;
 
 	trx->n_mysql_tables_in_use = 0;