Merge innodb-zip-ss6900 to mysql-5.1-bugteam.

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

@@ -441,6 +441,7 @@ t3	CREATE TABLE `t3` (
   KEY `c` (`c`)
 ) ENGINE=InnoDB DEFAULT CHARSET=latin1
 alter table t2 drop index b, add index (b);
+ERROR 42000: Incorrect index name 'b'
 show create table t2;
 Table	Create Table
 t2	CREATE TABLE `t2` (
@@ -451,8 +452,8 @@ t2	CREATE TABLE `t2` (
   `e` int(11) DEFAULT NULL,
   PRIMARY KEY (`a`),
   UNIQUE KEY `dc` (`d`,`c`),
-  KEY `c` (`c`),
   KEY `b` (`b`),
+  KEY `c` (`c`),
   CONSTRAINT `t2_ibfk_1` FOREIGN KEY (`b`) REFERENCES `t1` (`b`) ON DELETE CASCADE,
   CONSTRAINT `t2_ibfk_2` FOREIGN KEY (`c`) REFERENCES `t3` (`c`),
   CONSTRAINT `t2_ibfk_3` FOREIGN KEY (`d`) REFERENCES `t4` (`d`)

mysql-test/suite/innodb/r/innodb_bug44571.result

@@ -2,8 +2,7 @@ CREATE TABLE bug44571 (foo INT) ENGINE=InnoDB;
 ALTER TABLE bug44571 CHANGE foo bar INT;
 ALTER TABLE bug44571 ADD INDEX bug44571b (foo);
 ERROR 42000: Key column 'foo' doesn't exist in table
-ALTER TABLE bug44571 ADD INDEX bug44571b (bar);
-ERROR HY000: Incorrect key file for table 'bug44571'; try to repair it
-CREATE INDEX bug44571b ON bug44571 (bar);
-ERROR HY000: Incorrect key file for table 'bug44571'; try to repair it
+ALTER TABLE bug44571 ADD INDEX bug44571c (bar);
+DROP INDEX bug44571c ON bug44571;
+CREATE INDEX bug44571c ON bug44571 (bar);
 DROP TABLE bug44571;

mysql-test/suite/innodb/r/innodb_bug47622.result

+CREATE TABLE bug47622(
+`rule_key` int(11) NOT NULL DEFAULT '0',
+`seq` smallint(6) NOT NULL DEFAULT '0',
+`action` smallint(6) NOT NULL DEFAULT '0',
+`arg_id` smallint(6) DEFAULT NULL,
+`else_ind` TINYINT NOT NULL,
+KEY IDX_A (`arg_id`)
+) ENGINE=InnoDB;
+ALTER TABLE bug47622 ADD UNIQUE IDX_B (rule_key,else_ind,seq,action,arg_id);
+drop index IDX_B on bug47622;
+create index idx on bug47622(seq, arg_id);
+ALTER TABLE bug47622 ADD UNIQUE IDX_X (rule_key,else_ind,seq,action);
+drop table bug47622;
+CREATE TABLE bug47622 (
+`a` int(11) NOT NULL,
+`b` int(11) DEFAULT NULL,
+`c` char(10) DEFAULT NULL,
+`d` varchar(20) DEFAULT NULL,
+PRIMARY KEY (`a`),
+KEY `b` (`b`)
+) ENGINE=InnoDB;
+alter table bug47622 add unique index (c), add index (d);
+drop table bug47622;

mysql-test/suite/innodb/r/innodb_bug51378.result

+create table bug51378 (
+col1 int not null,
+col2 blob not null,
+col3 time not null) engine = innodb;
+create unique index idx on bug51378(col1, col2(31));
+alter table bug51378 add unique index idx2(col1, col2(31));
+create unique index idx3 on bug51378(col1, col3);
+SHOW CREATE TABLE bug51378;
+Table	Create Table
+bug51378	CREATE TABLE `bug51378` (
+  `col1` int(11) NOT NULL,
+  `col2` blob NOT NULL,
+  `col3` time NOT NULL,
+  UNIQUE KEY `idx3` (`col1`,`col3`),
+  UNIQUE KEY `idx` (`col1`,`col2`(31)),
+  UNIQUE KEY `idx2` (`col1`,`col2`(31))
+) ENGINE=InnoDB DEFAULT CHARSET=latin1
+drop index idx3 on bug51378;
+SHOW CREATE TABLE bug51378;
+Table	Create Table
+bug51378	CREATE TABLE `bug51378` (
+  `col1` int(11) NOT NULL,
+  `col2` blob NOT NULL,
+  `col3` time NOT NULL,
+  UNIQUE KEY `idx` (`col1`,`col2`(31)),
+  UNIQUE KEY `idx2` (`col1`,`col2`(31))
+) ENGINE=InnoDB DEFAULT CHARSET=latin1
+alter table bug51378 add primary key idx3(col1, col2(31));
+SHOW CREATE TABLE bug51378;
+Table	Create Table
+bug51378	CREATE TABLE `bug51378` (
+  `col1` int(11) NOT NULL,
+  `col2` blob NOT NULL,
+  `col3` time NOT NULL,
+  PRIMARY KEY (`col1`,`col2`(31)),
+  UNIQUE KEY `idx` (`col1`,`col2`(31)),
+  UNIQUE KEY `idx2` (`col1`,`col2`(31))
+) ENGINE=InnoDB DEFAULT CHARSET=latin1
+drop table bug51378;
+create table bug51378 (
+col1 int not null,
+col2 blob not null,
+col3 time not null, primary key(col1, col2(31))) engine = innodb;
+create unique index idx on bug51378(col1, col2(31));
+SHOW CREATE TABLE bug51378;
+Table	Create Table
+bug51378	CREATE TABLE `bug51378` (
+  `col1` int(11) NOT NULL,
+  `col2` blob NOT NULL,
+  `col3` time NOT NULL,
+  PRIMARY KEY (`col1`,`col2`(31)),
+  UNIQUE KEY `idx` (`col1`,`col2`(31))
+) ENGINE=InnoDB DEFAULT CHARSET=latin1
+drop table bug51378;
+create table bug51378 (
+col1 int not null,
+col2 int ) engine = innodb;
+create unique index idx on bug51378(col1, col2);
+SHOW CREATE TABLE bug51378;
+Table	Create Table
+bug51378	CREATE TABLE `bug51378` (
+  `col1` int(11) NOT NULL,
+  `col2` int(11) DEFAULT NULL,
+  UNIQUE KEY `idx` (`col1`,`col2`)
+) ENGINE=InnoDB DEFAULT CHARSET=latin1
+drop table bug51378;

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

@@ -26,7 +26,7 @@ replace into t1 select * from t2;
 connection b;
 set session transaction isolation level read committed;
 set autocommit=0;
-# should not cuase a lock wait.
+# should not cause a lock wait.
 delete from t2 where a=5;
 commit;
 delete from t2;
@@ -42,7 +42,7 @@ insert into t1 select * from t2;
 connection b;
 set session transaction isolation level read committed;
 set autocommit=0;
-# should not cuase a lock wait.
+# should not cause a lock wait.
 delete from t2 where a=5;
 commit;
 delete from t2;

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

 -- source include/have_innodb.inc
 
+let $MYSQLD_DATADIR= `select @@datadir`;
+
 let $innodb_file_format_check_orig=`select @@innodb_file_format_check`;
 
 create table t1(a int not null, b int, c char(10) not null, d varchar(20)) engine = innodb;
@@ -139,6 +141,8 @@ show create table t4;
 --error ER_CANT_CREATE_TABLE
 alter table t3 add constraint dc foreign key (a) references t1(a);
 show create table t3;
+# this should be fixed by MySQL (see Bug #51451)
+--error ER_WRONG_NAME_FOR_INDEX
 alter table t2 drop index b, add index (b);
 show create table t2;
 --error ER_ROW_IS_REFERENCED_2
@@ -146,7 +150,9 @@ delete from t1;
 --error ER_CANT_DROP_FIELD_OR_KEY
 drop index dc on t4;
 # there is no foreign key dc on t3
---replace_regex /'\.\/test\/#sql2-[0-9a-f-]*'/'#sql2-temporary'/
+--replace_regex /'[^']*test\/#sql2-[0-9a-f-]*'/'#sql2-temporary'/
+# Embedded server doesn't chdir to data directory
+--replace_result $MYSQLD_DATADIR ./ master-data/ ''
 --error ER_ERROR_ON_RENAME
 alter table t3 drop foreign key dc;
 alter table t4 drop foreign key dc;

mysql-test/suite/innodb/t/innodb_bug44571.test

 #
 # Bug#44571 InnoDB Plugin crashes on ADD INDEX
 # http://bugs.mysql.com/44571
+# Please also refer to related fix in
+# http://bugs.mysql.com/47621
 #
 -- source include/have_innodb.inc
 -- source suite/innodb/include/have_innodb_plugin.inc
 
 CREATE TABLE bug44571 (foo INT) ENGINE=InnoDB;
 ALTER TABLE bug44571 CHANGE foo bar INT;
+# Create index with the old column name will fail,
+# because the CHANGE foo bar is successful. And
+# the column name change would communicate to
+# InnoDB with the fix from bug #47621
 -- error ER_KEY_COLUMN_DOES_NOT_EXITS
 ALTER TABLE bug44571 ADD INDEX bug44571b (foo);
-# The following will fail, because the CHANGE foo bar was
-# not communicated to InnoDB.
---error ER_NOT_KEYFILE
-ALTER TABLE bug44571 ADD INDEX bug44571b (bar);
---error ER_NOT_KEYFILE
-CREATE INDEX bug44571b ON bug44571 (bar);
+# The following create indexes should succeed,
+# indirectly confirm the CHANGE foo bar is successful.
+ALTER TABLE bug44571 ADD INDEX bug44571c (bar);
+DROP INDEX bug44571c ON bug44571;
+CREATE INDEX bug44571c ON bug44571 (bar);
 DROP TABLE bug44571;

mysql-test/suite/innodb/t/innodb_bug47622.test

+# This is the test for bug 47622. There could be index
+# metadata sequence mismatch between MySQL and Innodb
+# after creating index through FIC interfaces.
+# We resolve the problem by sync the index sequence
+# up when opening the table.
+
+--source include/have_innodb.inc
+
+connect (a,localhost,root,,);
+connect (b,localhost,root,,);
+
+# Create a table with a non-unique index
+CREATE TABLE bug47622(
+	`rule_key` int(11) NOT NULL DEFAULT '0',
+	`seq` smallint(6) NOT NULL DEFAULT '0',
+	`action` smallint(6) NOT NULL DEFAULT '0',
+	`arg_id` smallint(6) DEFAULT NULL,
+	`else_ind` TINYINT NOT NULL,
+	KEY IDX_A (`arg_id`)
+) ENGINE=InnoDB;
+
+connection a;
+
+# A subsequent creating unique index should not trigger
+# any error message. Unique index would be ranked ahead
+# of regular index.
+ALTER TABLE bug47622 ADD UNIQUE IDX_B (rule_key,else_ind,seq,action,arg_id);
+
+drop index IDX_B on bug47622;
+
+# In another connection, create additional set of normal
+# index and unique index. Again, unique index would be ranked
+# ahead of regular index.
+connection b;
+create index idx on bug47622(seq, arg_id);
+
+ALTER TABLE bug47622 ADD UNIQUE IDX_X (rule_key,else_ind,seq,action);
+
+drop table bug47622;
+
+# Create a table with one Primary key and a non-unique key
+CREATE TABLE bug47622 (
+  `a` int(11) NOT NULL,
+  `b` int(11) DEFAULT NULL,
+  `c` char(10) DEFAULT NULL,
+  `d` varchar(20) DEFAULT NULL,
+  PRIMARY KEY (`a`),
+  KEY `b` (`b`)
+) ENGINE=InnoDB;
+
+# Add two index with one unique and one non-unique.
+# Index sequence is "PRIMARY", "c", "b" and "d"
+alter table bug47622 add unique index (c), add index (d);
+
+drop table bug47622;

mysql-test/suite/innodb/t/innodb_bug51378.test

+# This is the test for bug 51378. Unique index created
+# through "create index" and "alter table add unique index"
+# interfaces should not be treated as primary index if indexed
+# columns contain one or more column prefix(es) (only prefix/part of
+# the column is indexed)
+# On the other hand, if there is a unique index covers all
+# columns of a table, and they are non-null columns, and
+# full length of the column are indexed, then this index
+# will be created as primary index
+# Following queries test various scenario, no mismatch
+# error message should be printed.
+--source include/have_innodb.inc
+
+# Create a table contains a BLOB column
+create table bug51378 (
+	col1 int not null,
+	col2 blob not null,
+	col3 time not null) engine = innodb;
+
+# Create following unique indexes on 'col1' and 'col2(31)'
+# of the table, the index should not be treated as primary
+# key because it indexes only first 31 bytes of col2.
+# Thus it contains "column prefix", and will not be
+# upgraded to primary index.
+# There should not be mismatch message printed in the
+# errorlog
+create unique index idx on bug51378(col1, col2(31));
+
+alter table bug51378 add unique index idx2(col1, col2(31));
+
+# Unique index on 'col1' and 'col3' will be created as primary index,
+# since the index does not contain column prefix
+create unique index idx3 on bug51378(col1, col3);
+
+# Show create table would show idx3 created as unique index, internally,
+# idx3 is treated as primary index both by MySQL and Innodb
+SHOW CREATE TABLE bug51378;
+
+# "GEN_CLUST_INDEX" will be re-created as default primary index
+# after idx3 is dropped
+drop index idx3 on bug51378;
+
+SHOW CREATE TABLE bug51378;
+
+# Or we can add the primary key through alter table interfaces
+alter table bug51378 add primary key idx3(col1, col2(31));
+
+SHOW CREATE TABLE bug51378;
+
+drop table bug51378;
+
+# Or we can create such primary key through create table interfaces
+create table bug51378 (
+        col1 int not null,
+        col2 blob not null,
+        col3 time not null, primary key(col1, col2(31))) engine = innodb;
+
+# Unique index on one or more column prefix(es) will be created
+# as non-cluster index
+create unique index idx on bug51378(col1, col2(31));
+
+SHOW CREATE TABLE bug51378;
+
+drop table bug51378;
+
+# If a table has a NULLABLE column, unique index on it will not
+# be treated as primary index.
+create table bug51378 (
+	col1 int not null,
+        col2 int ) engine = innodb;
+
+# This will be created as non-cluster index since col2 is nullable
+create unique index idx on bug51378(col1, col2);
+
+SHOW CREATE TABLE bug51378;
+
+drop table bug51378;

storage/innodb_plugin/CMakeLists.txt

@@ -78,7 +78,11 @@ SET(INNODB_PLUGIN_SOURCES	btr/btr0btr.c btr/btr0cur.c btr/btr0pcur.c btr/btr0sea
 			trx/trx0i_s.c trx/trx0purge.c trx/trx0rec.c trx/trx0roll.c trx/trx0rseg.c
 			trx/trx0sys.c trx/trx0trx.c trx/trx0undo.c
 			usr/usr0sess.c
-			ut/ut0byte.c ut/ut0dbg.c ut/ut0mem.c ut/ut0rnd.c ut/ut0ut.c ut/ut0vec.c
+			ut/ut0byte.c ut/ut0dbg.c ut/ut0mem.c ut/ut0rbt.c ut/ut0rnd.c ut/ut0ut.c ut/ut0vec.c
 			ut/ut0list.c ut/ut0wqueue.c)
-ADD_DEFINITIONS(-DHAVE_WINDOWS_ATOMICS -DIB_HAVE_PAUSE_INSTRUCTION)
+# Windows atomics do not perform well. Disable Windows atomics by default.
+# See bug#52102 for details.
+#ADD_DEFINITIONS(-DHAVE_WINDOWS_ATOMICS -DINNODB_RW_LOCKS_USE_ATOMICS -DHAVE_IB_PAUSE_INSTRUCTION)
+ADD_DEFINITIONS(-DHAVE_IB_PAUSE_INSTRUCTION)
+
 MYSQL_STORAGE_ENGINE(INNODB_PLUGIN)

storage/innodb_plugin/ChangeLog

+2010-03-18	The InnoDB Team
+
+	* CMakeLists.txt:
+	Fix Bug#52102 InnoDB Plugin shows performance drop compared to
+	InnoDB (Windows)
+
+2010-03-18	The InnoDB Team
+
+	* buf0buf.ic:
+	When comparing the time of the first access to a block against
+	innodb_old_blocks_time, use 32-bit arithmetics. The comparison was
+	incorrect on 64-bit systems.
+
+2010-03-11	The InnoDB Team
+
+	* buf0buf.h, buf0buf.ic:
+	Fix and clarify the latching of some buf_block_t members.
+	Note that check_index_page_at_flush is not protected by any mutex.
+	Note and assert that lock_hash_val is protected by the rw-latch.
+
+2010-03-10	The InnoDB Team
+
+	* trx/trx0sys.c:
+	Fix Bug#51653 outdated reference to set-variable
+
+2010-03-10	The InnoDB Team
+
+	* handler/ha_innodb.cc, mysql-test/innodb_bug21704.result,
+	mysql-test/innodb_bug47621.result, mysql-test/innodb_bug47621.test:
+	Fix Bug#47621 MySQL and InnoDB data dictionaries will become out of
+	sync when renaming columns
+
+2010-03-10	The InnoDB Team
+
+	* handler/ha_innodb.cc:
+	Fix Bug#51356 Many Valgrind errors in error messages
+	with concurrent DDL
+
+2010-03-10	The InnoDB Team
+
+	* handler/ha_innodb.cc, handler/handler0alter.cc,
+	mysql-test/innodb_bug51378.result, mysql-test/innodb_bug51378.test:
+	Fix Bug#51378 Init 'ref_length' to correct value, in case an out
+	of bound MySQL primary_key
+
+2010-03-10	The InnoDB Team
+
+	* log/log0recv.c:
+	Remove a bogus assertion about page numbers exceeding 0x90000000
+	in the redo log. Abort when encountering a corrupted redo log
+	record, unless innodb_force_recovery is set.
+
+2010-03-09	The InnoDB Team
+
+	* handler/ha_innodb.cc:
+	Make SHOW ENGINE INNODB MUTEX STATUS display SUM(os_waits)
+	for the buffer pool block mutexes and locks.
+
+2010-03-08	The InnoDB Team
+
+	* fil/fil0fil.c:
+	Fix ALTER TABLE ... IMPORT TABLESPACE of compressed tables.
+
+2010-03-03	The InnoDB Team
+
+	* handler/handler0alter.cc, innodb-index.result, innodb-index.test,
+	innodb.result, innodb.test:
+	Disallow a duplicate index name when creating an index.
+
+2010-02-11	The InnoDB Team
+
+	* include/mem0mem.h, include/mem0mem.ic, mem/mem0mem.c:
+	Fix Bug#49535 Available memory check slows down crash
+	recovery tens of times
+
+2010-02-09	The InnoDB Team
+
+	* buf/buf0buf.c:
+	Fix Bug#38901 InnoDB logs error repeatedly when trying to load
+	page into buffer pool
+
+2010-02-09	The InnoDB Team
+
+	* srv/srv0srv.c:
+	Let the master thread sleep if the amount of work to be done is
+	calibrated as taking less than a second.
+
+2010-02-04	The InnoDB Team
+
+	* btr/btr0btr.c, btr/btr0cur.c, btr/btr0pcur.c, buf/buf0buf.c,
+	include/btr0btr.h, include/btr0cur.h, include/btr0pcur.h,
+	include/btr0pcur.ic, include/buf0buf.h, row/row0ins.c, row/row0sel.c:
+	Pass the file name and line number of the caller of the
+	b-tree cursor functions to the buffer pool requests, in order
+	to make the latch diagnostics more accurate.
+
+2010-02-03	The InnoDB Team
+
+	* lock/lock0lock.c:
+	Fix Bug#49001 SHOW INNODB STATUS deadlock info incorrect
+	when deadlock detection aborts
+
+2010-02-03	The InnoDB Team
+
+	* buf/buf0lru.c:
+	Fix Bug#35077 Very slow DROP TABLE (ALTER TABLE, OPTIMIZE TABLE)
+	on compressed tables
+
+2010-02-03	The InnoDB Team
+
+	* handler/ha_innodb.cc, include/row0mysql.h, row/row0mysql.c:
+	Clean up CHECK TABLE error handling.
+
+2010-02-01	The InnoDB Team
+
+	* handler/ha_innodb.cc, mysql-test/innodb-autoinc.test,
+	mysql-test/innodb-autoinc.result,
+	mysql-test/innodb-autoinc-44030.test,
+	mysql-test/innodb-autoinc-44030.result:
+	Fix Bug#49497 Error 1467 (ER_AUTOINC_READ_FAILED) on inserting
+	a negative value
+
+2010-01-27	The InnoDB Team
+
+	* include/row0mysql.h, log/log0recv.c, row/row0mysql.c:
+	Drop temporary tables at startup.
+	This addresses the third aspect of
+	Bug#41609 Crash recovery does not work for InnoDB temporary tables.
+
+2010-01-21	The InnoDB Team
+
+	* buf/buf0buf.c:
+	Do not merge buffered inserts to compressed pages before
+	the redo log has been applied in crash recovery.
+
+2010-01-13	The InnoDB Team
+
+	* row/row0sel.c:
+	On the READ UNCOMMITTED isolation level, do not attempt to access
+	a clustered index record that has been marked for deletion. The
+	built-in InnoDB in MySQL 5.1 and earlier would attempt to retrieve
+	a previous version of the record in this case.
+
+2010-01-13	The InnoDB Team
+
+	* buf/buf0buf.c:
+	When disabling the adaptive hash index, check the block state
+	before checking block->is_hashed, because the latter may be
+	uninitialized right after server startup.
+
+2010-01-12	The InnoDB Team
+
+	* handler/ha_innodb.cc, handler/ha_innodb.h:
+	Fix Bug#46193 crash when accessing tables after enabling
+	innodb_force_recovery option
+
+2010-01-12	The InnoDB Team
+
+	* row/row0mysql.c:
+	Fix Bug#49238 Creating/Dropping a temporary table while at 1023
+	transactions will cause assert.
+
+2009-12-02	The InnoDB Team
+
+	* srv/srv0start.c:
+	Display the zlib version number at startup.
+	InnoDB compressed tables use zlib, and the implementation depends
+	on the zlib function compressBound(), whose definition was slightly
+	changed in zlib version 1.2.3.1 in 2006. MySQL bundles zlib 1.2.3
+	from 2005, but some installations use a more recent zlib.
+
+2009-11-30	The InnoDB Team
+
+	* dict/dict0crea.c, dict/dict0mem.c, dict/dict0load.c,
+	dict/dict0boot.c, fil/fil0fil.c, handler/ha_innodb.cc,
+	include/dict0mem.h, row/row0mysql.c:
+	Fix the bogus warning messages for non-existing temporary
+	tables that were reported in
+	Bug#41609 Crash recovery does not work for InnoDB temporary tables.
+	The actual crash recovery bug was corrected on 2009-04-29.
+
+2009-11-27	The InnoDB Team
+
+	InnoDB Plugin 1.0.6 released
+
 2009-11-20	The InnoDB Team
 
 	* handler/ha_innodb.cc:
@@ -79,8 +264,8 @@
 	sync/sync0arr.c, sync/sync0sync.c, thr/thr0loc.c, trx/trx0i_s.c,
 	trx/trx0purge.c, trx/trx0rseg.c, trx/trx0sys.c, trx/trx0undo.c,
 	usr/usr0sess.c, ut/ut0mem.c:
-	Fix Bug #45992 innodb memory not freed after shutdown
-	Fix Bug #46656 InnoDB plugin: memory leaks (Valgrind)
+	Fix Bug#45992 innodb memory not freed after shutdown
+	Fix Bug#46656 InnoDB plugin: memory leaks (Valgrind)
 
 2009-10-29	The InnoDB Team
 
@@ -422,7 +607,7 @@
 	* dict/dict0dict.c:
 	When an index column cannot be found in the table during index
 	creation, display additional diagnostic before an assertion failure.
-	This does NOT fix Bug #44571 InnoDB Plugin crashes on ADD INDEX,
+	This does NOT fix Bug#44571 InnoDB Plugin crashes on ADD INDEX,
 	but it helps understand the reason of the crash.
 
 2009-06-17	The InnoDB Team
@@ -535,6 +720,12 @@
 	Fix Bug#44320 InnoDB: missing DB_ROLL_PTR in Table Monitor COLUMNS
 	output
 
+2009-04-29	The InnoDB Team
+
+	* fil/fil0fil.c, include/fil0fil.h, include/mtr0mtr.h,
+	log/log0recv.c:
+	Fix Bug#41609 Crash recovery does not work for InnoDB temporary tables
+
 2009-04-23	The InnoDB Team
 
 	* row/row0mysql.c:

storage/innodb_plugin/Makefile.am

@@ -217,6 +217,7 @@ noinst_HEADERS=		\
 			include/ut0lst.h	\
 			include/ut0mem.h	\
 			include/ut0mem.ic	\
+			include/ut0rbt.h	\
 			include/ut0rnd.h	\
 			include/ut0rnd.ic	\
 			include/ut0sort.h	\
@@ -318,6 +319,7 @@ libinnobase_a_SOURCES=	\
 			ut/ut0dbg.c			\
 			ut/ut0list.c			\
 			ut/ut0mem.c			\
+			ut/ut0rbt.c			\
 			ut/ut0rnd.c			\
 			ut/ut0ut.c			\
 			ut/ut0vec.c			\

storage/innodb_plugin/btr/btr0btr.c

 /*****************************************************************************
 
-Copyright (c) 1994, 2009, Innobase Oy. All Rights Reserved.
+Copyright (c) 1994, 2010, 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
@@ -592,13 +592,15 @@ an x-latch on the tree.
 @return	rec_get_offsets() of the node pointer record */
 static
 ulint*
-btr_page_get_father_node_ptr(
-/*=========================*/
+btr_page_get_father_node_ptr_func(
+/*==============================*/
 	ulint*		offsets,/*!< in: work area for the return value */
 	mem_heap_t*	heap,	/*!< in: memory heap to use */
 	btr_cur_t*	cursor,	/*!< in: cursor pointing to user record,
 				out: cursor on node pointer record,
 				its page x-latched */
+	const char*	file,	/*!< in: file name */
+	ulint		line,	/*!< in: line where called */
 	mtr_t*		mtr)	/*!< in: mtr */
 {
 	dtuple_t*	tuple;
@@ -622,7 +624,8 @@ btr_page_get_father_node_ptr(
 	tuple = dict_index_build_node_ptr(index, user_rec, 0, heap, level);
 
 	btr_cur_search_to_nth_level(index, level + 1, tuple, PAGE_CUR_LE,
-				    BTR_CONT_MODIFY_TREE, cursor, 0, mtr);
+				    BTR_CONT_MODIFY_TREE, cursor, 0,
+				    file, line, mtr);
 
 	node_ptr = btr_cur_get_rec(cursor);
 	ut_ad(!page_rec_is_comp(node_ptr)
@@ -670,6 +673,9 @@ btr_page_get_father_node_ptr(
 	return(offsets);
 }
 
+#define btr_page_get_father_node_ptr(of,heap,cur,mtr)			\
+	btr_page_get_father_node_ptr_func(of,heap,cur,__FILE__,__LINE__,mtr)
+
 /************************************************************//**
 Returns the upper level node pointer to a page. It is assumed that mtr holds
 an x-latch on the tree.
@@ -1662,11 +1668,13 @@ Inserts a data tuple to a tree on a non-leaf level. It is assumed
 that mtr holds an x-latch on the tree. */
 UNIV_INTERN
 void
-btr_insert_on_non_leaf_level(
-/*=========================*/
+btr_insert_on_non_leaf_level_func(
+/*==============================*/
 	dict_index_t*	index,	/*!< in: index */
 	ulint		level,	/*!< in: level, must be > 0 */
 	dtuple_t*	tuple,	/*!< in: the record to be inserted */
+	const char*	file,	/*!< in: file name */
+	ulint		line,	/*!< in: line where called */
 	mtr_t*		mtr)	/*!< in: mtr */
 {
 	big_rec_t*	dummy_big_rec;
@@ -1678,7 +1686,7 @@ btr_insert_on_non_leaf_level(
 
 	btr_cur_search_to_nth_level(index, level, tuple, PAGE_CUR_LE,
 				    BTR_CONT_MODIFY_TREE,
-				    &cursor, 0, mtr);
+				    &cursor, 0, file, line, mtr);
 
 	err = btr_cur_pessimistic_insert(BTR_NO_LOCKING_FLAG
 					 | BTR_KEEP_SYS_FLAG

storage/innodb_plugin/btr/btr0cur.c

 /*****************************************************************************
 
-Copyright (c) 1994, 2009, Innobase Oy. All Rights Reserved.
+Copyright (c) 1994, 2010, Innobase Oy. All Rights Reserved.
 Copyright (c) 2008, Google Inc.
 
 Portions of this file contain modifications contributed and copyrighted by
@@ -342,6 +342,8 @@ btr_cur_search_to_nth_level(
 	ulint		has_search_latch,/*!< in: info on the latch mode the
 				caller currently has on btr_search_latch:
 				RW_S_LATCH, or 0 */
+	const char*	file,	/*!< in: file name */
+	ulint		line,	/*!< in: line where called */
 	mtr_t*		mtr)	/*!< in: mtr */
 {
 	page_cur_t*	page_cursor;
@@ -520,7 +522,7 @@ btr_cur_search_to_nth_level(
 retry_page_get:
 		block = buf_page_get_gen(space, zip_size, page_no,
 					 rw_latch, guess, buf_mode,
-					 __FILE__, __LINE__, mtr);
+					 file, line, mtr);
 		if (block == NULL) {
 			/* This must be a search to perform an insert;
 			try insert to the insert buffer */
@@ -677,13 +679,15 @@ btr_cur_search_to_nth_level(
 Opens a cursor at either end of an index. */
 UNIV_INTERN
 void
-btr_cur_open_at_index_side(
-/*=======================*/
+btr_cur_open_at_index_side_func(
+/*============================*/
 	ibool		from_left,	/*!< in: TRUE if open to the low end,
 					FALSE if to the high end */
 	dict_index_t*	index,		/*!< in: index */
 	ulint		latch_mode,	/*!< in: latch mode */
 	btr_cur_t*	cursor,		/*!< in: cursor */
+	const char*	file,		/*!< in: file name */
+	ulint		line,		/*!< in: line where called */
 	mtr_t*		mtr)		/*!< in: mtr */
 {
 	page_cur_t*	page_cursor;
@@ -728,7 +732,7 @@ btr_cur_open_at_index_side(
 		page_t*		page;
 		block = buf_page_get_gen(space, zip_size, page_no,
 					 RW_NO_LATCH, NULL, BUF_GET,
-					 __FILE__, __LINE__, mtr);
+					 file, line, mtr);
 		page = buf_block_get_frame(block);
 		ut_ad(0 == ut_dulint_cmp(index->id,
 					 btr_page_get_index_id(page)));
@@ -808,11 +812,13 @@ btr_cur_open_at_index_side(
 Positions a cursor at a randomly chosen position within a B-tree. */
 UNIV_INTERN
 void
-btr_cur_open_at_rnd_pos(
-/*====================*/
+btr_cur_open_at_rnd_pos_func(
+/*=========================*/
 	dict_index_t*	index,		/*!< in: index */
 	ulint		latch_mode,	/*!< in: BTR_SEARCH_LEAF, ... */
 	btr_cur_t*	cursor,		/*!< in/out: B-tree cursor */
+	const char*	file,		/*!< in: file name */
+	ulint		line,		/*!< in: line where called */
 	mtr_t*		mtr)		/*!< in: mtr */
 {
 	page_cur_t*	page_cursor;
@@ -847,7 +853,7 @@ btr_cur_open_at_rnd_pos(
 
 		block = buf_page_get_gen(space, zip_size, page_no,
 					 RW_NO_LATCH, NULL, BUF_GET,
-					 __FILE__, __LINE__, mtr);
+					 file, line, mtr);
 		page = buf_block_get_frame(block);
 		ut_ad(0 == ut_dulint_cmp(index->id,
 					 btr_page_get_index_id(page)));
@@ -3100,7 +3106,8 @@ btr_estimate_n_rows_in_range(
 
 		btr_cur_search_to_nth_level(index, 0, tuple1, mode1,
 					    BTR_SEARCH_LEAF | BTR_ESTIMATE,
-					    &cursor, 0, &mtr);
+					    &cursor, 0,
+					    __FILE__, __LINE__, &mtr);
 	} else {
 		btr_cur_open_at_index_side(TRUE, index,
 					   BTR_SEARCH_LEAF | BTR_ESTIMATE,
@@ -3117,7 +3124,8 @@ btr_estimate_n_rows_in_range(
 
 		btr_cur_search_to_nth_level(index, 0, tuple2, mode2,
 					    BTR_SEARCH_LEAF | BTR_ESTIMATE,
-					    &cursor, 0, &mtr);
+					    &cursor, 0,
+					    __FILE__, __LINE__, &mtr);
 	} else {
 		btr_cur_open_at_index_side(FALSE, index,
 					   BTR_SEARCH_LEAF | BTR_ESTIMATE,
@@ -4252,7 +4260,7 @@ btr_free_externally_stored_field(
 		/* In the rollback of uncommitted transactions, we may
 		encounter a clustered index record whose BLOBs have
 		not been written.  There is nothing to free then. */
-		ut_a(rb_ctx == RB_RECOVERY);
+		ut_a(rb_ctx == RB_RECOVERY || rb_ctx == RB_RECOVERY_PURGE_REC);
 		return;
 	}
 
@@ -4298,7 +4306,7 @@ btr_free_externally_stored_field(
 		    || (mach_read_from_1(field_ref + BTR_EXTERN_LEN)
 			& BTR_EXTERN_OWNER_FLAG)
 		    /* Rollback and inherited field */
-		    || (rb_ctx != RB_NONE
+		    || ((rb_ctx == RB_NORMAL || rb_ctx == RB_RECOVERY)
 			&& (mach_read_from_1(field_ref + BTR_EXTERN_LEN)
 			    & BTR_EXTERN_INHERITED_FLAG))) {
 

storage/innodb_plugin/btr/btr0pcur.c

 /*****************************************************************************
 
-Copyright (c) 1996, 2009, Innobase Oy. All Rights Reserved.
+Copyright (c) 1996, 2010, 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
@@ -205,10 +205,12 @@ record and it can be restored on a user record whose ordering fields
 are identical to the ones of the original user record */
 UNIV_INTERN
 ibool
-btr_pcur_restore_position(
-/*======================*/
+btr_pcur_restore_position_func(
+/*===========================*/
 	ulint		latch_mode,	/*!< in: BTR_SEARCH_LEAF, ... */
 	btr_pcur_t*	cursor,		/*!< in: detached persistent cursor */
+	const char*	file,		/*!< in: file name */
+	ulint		line,		/*!< in: line where called */
 	mtr_t*		mtr)		/*!< in: mtr */
 {
 	dict_index_t*	index;
@@ -217,6 +219,9 @@ btr_pcur_restore_position(
 	ulint		old_mode;
 	mem_heap_t*	heap;
 
+	ut_ad(mtr);
+	ut_ad(mtr->state == MTR_ACTIVE);
+
 	index = btr_cur_get_index(btr_pcur_get_btr_cur(cursor));
 
 	if (UNIV_UNLIKELY(cursor->old_stored != BTR_PCUR_OLD_STORED)
@@ -257,7 +262,8 @@ btr_pcur_restore_position(
 		if (UNIV_LIKELY(buf_page_optimistic_get(
 					latch_mode,
 					cursor->block_when_stored,
-					cursor->modify_clock, mtr))) {
+					cursor->modify_clock,
+					file, line, mtr))) {
 			cursor->pos_state = BTR_PCUR_IS_POSITIONED;
 
 			buf_block_dbg_add_level(btr_pcur_get_block(cursor),
@@ -312,8 +318,8 @@ btr_pcur_restore_position(
 		mode = PAGE_CUR_L;
 	}
 
-	btr_pcur_open_with_no_init(index, tuple, mode, latch_mode,
-				   cursor, 0, mtr);
+	btr_pcur_open_with_no_init_func(index, tuple, mode, latch_mode,
+					cursor, 0, file, line, mtr);
 
 	/* Restore the old search mode */
 	cursor->search_mode = old_mode;
@@ -553,8 +559,8 @@ before first in tree. The latching mode must be BTR_SEARCH_LEAF or
 BTR_MODIFY_LEAF. */
 UNIV_INTERN
 void
-btr_pcur_open_on_user_rec(
-/*======================*/
+btr_pcur_open_on_user_rec_func(
+/*===========================*/
 	dict_index_t*	index,		/*!< in: index */
 	const dtuple_t*	tuple,		/*!< in: tuple on which search done */
 	ulint		mode,		/*!< in: PAGE_CUR_L, ... */
@@ -562,9 +568,12 @@ btr_pcur_open_on_user_rec(
 					BTR_MODIFY_LEAF */
 	btr_pcur_t*	cursor,		/*!< in: memory buffer for persistent
 					cursor */
+	const char*	file,		/*!< in: file name */
+	ulint		line,		/*!< in: line where called */
 	mtr_t*		mtr)		/*!< in: mtr */
 {
-	btr_pcur_open(index, tuple, mode, latch_mode, cursor, mtr);
+	btr_pcur_open_func(index, tuple, mode, latch_mode, cursor,
+			   file, line, mtr);
 
 	if ((mode == PAGE_CUR_GE) || (mode == PAGE_CUR_G)) {
 

storage/innodb_plugin/buf/buf0buddy.c

 /*****************************************************************************
 
-Copyright (c) 2006, 2009, Innobase Oy. All Rights Reserved.
+Copyright (c) 2006, 2010, 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
@@ -391,6 +391,8 @@ buf_buddy_relocate_block(
 		UT_LIST_ADD_FIRST(list, buf_pool->zip_clean, dpage);
 	}
 
+	UNIV_MEM_INVALID(bpage, sizeof *bpage);
+
 	mutex_exit(&buf_pool_zip_mutex);
 	return(TRUE);
 }

storage/innodb_plugin/buf/buf0buf.c

 /*****************************************************************************
 
-Copyright (c) 1995, 2009, Innobase Oy. All Rights Reserved.
+Copyright (c) 1995, 2010, Innobase Oy. All Rights Reserved.
 Copyright (c) 2008, Google Inc.
 
 Portions of this file contain modifications contributed and copyrighted by
@@ -242,6 +242,8 @@ the read requests for the whole area.
 #ifndef UNIV_HOTBACKUP
 /** Value in microseconds */
 static const int WAIT_FOR_READ	= 5000;
+/** Number of attemtps made to read in a page in the buffer pool */
+static const ulint BUF_PAGE_READ_MAX_RETRIES = 100;
 
 /** The buffer buf_pool of the database */
 UNIV_INTERN buf_pool_t*	buf_pool = NULL;
@@ -1058,7 +1060,9 @@ buf_pool_drop_hash_index(void)
 				when we have an x-latch on btr_search_latch;
 				see the comment in buf0buf.h */
 
-				if (!block->is_hashed) {
+				if (buf_block_get_state(block)
+				    != BUF_BLOCK_FILE_PAGE
+				    || !block->is_hashed) {
 					continue;
 				}
 
@@ -1187,8 +1191,6 @@ buf_relocate(
 
 	HASH_DELETE(buf_page_t, hash, buf_pool->page_hash, fold, bpage);
 	HASH_INSERT(buf_page_t, hash, buf_pool->page_hash, fold, dpage);
-
-	UNIV_MEM_INVALID(bpage, sizeof *bpage);
 }
 
 /********************************************************************//**
@@ -2034,8 +2036,10 @@ buf_page_get_gen(
 	unsigned	access_time;
 	ulint		fix_type;
 	ibool		must_read;
+	ulint		retries = 0;
 
 	ut_ad(mtr);
+	ut_ad(mtr->state == MTR_ACTIVE);
 	ut_ad((rw_latch == RW_S_LATCH)
 	      || (rw_latch == RW_X_LATCH)
 	      || (rw_latch == RW_NO_LATCH));
@@ -2088,7 +2092,29 @@ buf_page_get_gen(
 			return(NULL);
 		}
 
-		buf_read_page(space, zip_size, offset);
+		if (buf_read_page(space, zip_size, offset)) {
+			retries = 0;
+		} else if (retries < BUF_PAGE_READ_MAX_RETRIES) {
+			++retries;
+		} else {
+			fprintf(stderr, "InnoDB: Error: Unable"
+				" to read tablespace %lu page no"
+				" %lu into the buffer pool after"
+				" %lu attempts\n"
+				"InnoDB: The most probable cause"
+				" of this error may be that the"
+				" table has been corrupted.\n"
+				"InnoDB: You can try to fix this"
+				" problem by using"
+				" innodb_force_recovery.\n"
+				"InnoDB: Please see reference manual"
+				" for more details.\n"
+				"InnoDB: Aborting...\n",
+				space, offset,
+				BUF_PAGE_READ_MAX_RETRIES);
+
+			ut_error;
+		}
 
 #if defined UNIV_DEBUG || defined UNIV_BUF_DEBUG
 		ut_a(++buf_dbg_counter % 37 || buf_validate());
@@ -2196,22 +2222,8 @@ buf_page_get_gen(
 			ut_ad(!block->page.in_flush_list);
 		} else {
 			/* Relocate buf_pool->flush_list. */
-			buf_page_t*	b;
-
-			b = UT_LIST_GET_PREV(list, &block->page);
-			ut_ad(block->page.in_flush_list);
-			UT_LIST_REMOVE(list, buf_pool->flush_list,
-				       &block->page);
-
-			if (b) {
-				UT_LIST_INSERT_AFTER(
-					list, buf_pool->flush_list, b,
-					&block->page);
-			} else {
-				UT_LIST_ADD_FIRST(
-					list, buf_pool->flush_list,
-					&block->page);
-			}
+			buf_flush_relocate_on_flush_list(bpage,
+							 &block->page);
 		}
 
 		/* Buffer-fix, I/O-fix, and X-latch the block
@@ -2225,6 +2237,9 @@ buf_page_get_gen(
 		block->page.buf_fix_count = 1;
 		buf_block_set_io_fix(block, BUF_IO_READ);
 		rw_lock_x_lock(&block->lock);
+
+		UNIV_MEM_INVALID(bpage, sizeof *bpage);
+
 		mutex_exit(&block->mutex);
 		mutex_exit(&buf_pool_zip_mutex);
 		buf_pool->n_pend_unzip++;
@@ -2237,7 +2252,7 @@ buf_page_get_gen(
 		while not holding buf_pool_mutex or block->mutex. */
 		success = buf_zip_decompress(block, srv_use_checksums);
 
-		if (UNIV_LIKELY(success)) {
+		if (UNIV_LIKELY(success && !recv_no_ibuf_operations)) {
 			ibuf_merge_or_delete_for_page(block, space, offset,
 						      zip_size, TRUE);
 		}
@@ -2356,8 +2371,8 @@ page.
 @return	TRUE if success */
 UNIV_INTERN
 ibool
-buf_page_optimistic_get_func(
-/*=========================*/
+buf_page_optimistic_get(
+/*====================*/
 	ulint		rw_latch,/*!< in: RW_S_LATCH, RW_X_LATCH */
 	buf_block_t*	block,	/*!< in: guessed buffer block */
 	ib_uint64_t	modify_clock,/*!< in: modify clock value if mode is
@@ -2370,7 +2385,9 @@ buf_page_optimistic_get_func(
 	ibool		success;
 	ulint		fix_type;
 
-	ut_ad(mtr && block);
+	ut_ad(block);
+	ut_ad(mtr);
+	ut_ad(mtr->state == MTR_ACTIVE);
 	ut_ad((rw_latch == RW_S_LATCH) || (rw_latch == RW_X_LATCH));
 
 	mutex_enter(&block->mutex);
@@ -2482,6 +2499,7 @@ buf_page_get_known_nowait(
 	ulint		fix_type;
 
 	ut_ad(mtr);
+	ut_ad(mtr->state == MTR_ACTIVE);
 	ut_ad((rw_latch == RW_S_LATCH) || (rw_latch == RW_X_LATCH));
 
 	mutex_enter(&block->mutex);
@@ -2581,6 +2599,9 @@ buf_page_try_get_func(
 	ibool		success;
 	ulint		fix_type;
 
+	ut_ad(mtr);
+	ut_ad(mtr->state == MTR_ACTIVE);
+
 	buf_pool_mutex_enter();
 	block = buf_block_hash_get(space_id, page_no);
 
@@ -2954,6 +2975,7 @@ buf_page_create(
 	ulint		time_ms		= ut_time_ms();
 
 	ut_ad(mtr);
+	ut_ad(mtr->state == MTR_ACTIVE);
 	ut_ad(space || !zip_size);
 
 	free_block = buf_LRU_get_free_block(0);

storage/innodb_plugin/buf/buf0flu.c

 /*****************************************************************************
 
-Copyright (c) 1995, 2009, Innobase Oy. All Rights Reserved.
+Copyright (c) 1995, 2010, 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
@@ -87,6 +87,138 @@ buf_flush_validate_low(void);
 /*========================*/
 #endif /* UNIV_DEBUG || UNIV_BUF_DEBUG */
 
+/********************************************************************//**
+Insert a block in the flush_rbt and returns a pointer to its
+predecessor or NULL if no predecessor. The ordering is maintained
+on the basis of the <oldest_modification, space, offset> key.
+@return pointer to the predecessor or NULL if no predecessor. */
+static
+buf_page_t*
+buf_flush_insert_in_flush_rbt(
+/*==========================*/
+	buf_page_t*	bpage)		/*!< in: bpage to be inserted. */
+{
+	buf_page_t*		prev = NULL;
+	const ib_rbt_node_t*	c_node;
+	const ib_rbt_node_t*	p_node;
+
+	ut_ad(buf_pool_mutex_own());
+
+	/* Insert this buffer into the rbt. */
+	c_node = rbt_insert(buf_pool->flush_rbt, &bpage, &bpage);
+	ut_a(c_node != NULL);
+
+	/* Get the predecessor. */
+	p_node = rbt_prev(buf_pool->flush_rbt, c_node);
+
+	if (p_node != NULL) {
+		prev = *rbt_value(buf_page_t*, p_node);
+		ut_a(prev != NULL);
+	}
+
+	return(prev);
+}
+
+/********************************************************************//**
+Delete a bpage from the flush_rbt. */
+static
+void
+buf_flush_delete_from_flush_rbt(
+/*============================*/
+	buf_page_t*	bpage)		/*!< in: bpage to be removed. */
+{
+
+	ibool	ret = FALSE;
+
+	ut_ad(buf_pool_mutex_own());
+	ret = rbt_delete(buf_pool->flush_rbt, &bpage);
+	ut_ad(ret);
+}
+
+/********************************************************************//**
+Compare two modified blocks in the buffer pool. The key for comparison
+is:
+key = <oldest_modification, space, offset>
+This comparison is used to maintian ordering of blocks in the
+buf_pool->flush_rbt.
+Note that for the purpose of flush_rbt, we only need to order blocks
+on the oldest_modification. The other two fields are used to uniquely
+identify the blocks.
+@return < 0 if b2 < b1, 0 if b2 == b1, > 0 if b2 > b1 */
+static
+int
+buf_flush_block_cmp(
+/*================*/
+	const void*	p1,		/*!< in: block1 */
+	const void*	p2)		/*!< in: block2 */
+{
+	int		ret;
+
+	ut_ad(p1 != NULL);
+	ut_ad(p2 != NULL);
+
+	const buf_page_t* b1 = *(const buf_page_t**) p1;
+	const buf_page_t* b2 = *(const buf_page_t**) p2;
+
+	ut_ad(b1 != NULL);
+	ut_ad(b2 != NULL);
+
+	ut_ad(b1->in_flush_list);
+	ut_ad(b2->in_flush_list);
+
+	if (b2->oldest_modification
+	    > b1->oldest_modification) {
+		return(1);
+	}
+
+	if (b2->oldest_modification
+	    < b1->oldest_modification) {
+		return(-1);
+	}
+
+	/* If oldest_modification is same then decide on the space. */
+	ret = (int)(b2->space - b1->space);
+
+	/* Or else decide ordering on the offset field. */
+	return(ret ? ret : (int)(b2->offset - b1->offset));
+}
+
+/********************************************************************//**
+Initialize the red-black tree to speed up insertions into the flush_list
+during recovery process. Should be called at the start of recovery
+process before any page has been read/written. */
+UNIV_INTERN
+void
+buf_flush_init_flush_rbt(void)
+/*==========================*/
+{
+	buf_pool_mutex_enter();
+
+	/* Create red black tree for speedy insertions in flush list. */
+	buf_pool->flush_rbt = rbt_create(sizeof(buf_page_t*),
+					 buf_flush_block_cmp);
+	buf_pool_mutex_exit();
+}
+
+/********************************************************************//**
+Frees up the red-black tree. */
+UNIV_INTERN
+void
+buf_flush_free_flush_rbt(void)
+/*==========================*/
+{
+	buf_pool_mutex_enter();
+
+#if defined UNIV_DEBUG || defined UNIV_BUF_DEBUG
+	ut_a(buf_flush_validate_low());
+#endif /* UNIV_DEBUG || UNIV_BUF_DEBUG */
+
+	rbt_free(buf_pool->flush_rbt);
+	buf_pool->flush_rbt = NULL;
+
+	buf_pool_mutex_exit();
+}
+
 /********************************************************************//**
 Inserts a modified block into the flush list. */
 UNIV_INTERN
@@ -100,6 +232,13 @@ buf_flush_insert_into_flush_list(
 	      || (UT_LIST_GET_FIRST(buf_pool->flush_list)->oldest_modification
 		  <= block->page.oldest_modification));
 
+	/* If we are in the recovery then we need to update the flush
+	red-black tree as well. */
+	if (UNIV_LIKELY_NULL(buf_pool->flush_rbt)) {
+		buf_flush_insert_sorted_into_flush_list(block);
+		return;
+	}
+
 	ut_ad(buf_block_get_state(block) == BUF_BLOCK_FILE_PAGE);
 	ut_ad(block->page.in_LRU_list);
 	ut_ad(block->page.in_page_hash);
@@ -136,12 +275,27 @@ buf_flush_insert_sorted_into_flush_list(
 	ut_d(block->page.in_flush_list = TRUE);
 
 	prev_b = NULL;
-	b = UT_LIST_GET_FIRST(buf_pool->flush_list);
 
-	while (b && b->oldest_modification > block->page.oldest_modification) {
-		ut_ad(b->in_flush_list);
-		prev_b = b;
-		b = UT_LIST_GET_NEXT(list, b);
+	/* For the most part when this function is called the flush_rbt
+	should not be NULL. In a very rare boundary case it is possible
+	that the flush_rbt has already been freed by the recovery thread
+	before the last page was hooked up in the flush_list by the
+	io-handler thread. In that case we'll  just do a simple
+	linear search in the else block. */
+	if (buf_pool->flush_rbt) {
+
+		prev_b = buf_flush_insert_in_flush_rbt(&block->page);
+
+	} else {
+
+		b = UT_LIST_GET_FIRST(buf_pool->flush_list);
+
+		while (b && b->oldest_modification
+		       > block->page.oldest_modification) {
+			ut_ad(b->in_flush_list);
+			prev_b = b;
+			b = UT_LIST_GET_NEXT(list, b);
+		}
 	}
 
 	if (prev_b == NULL) {
@@ -237,7 +391,6 @@ buf_flush_remove(
 	ut_ad(buf_pool_mutex_own());
 	ut_ad(mutex_own(buf_page_get_mutex(bpage)));
 	ut_ad(bpage->in_flush_list);
-	ut_d(bpage->in_flush_list = FALSE);
 
 	switch (buf_page_get_state(bpage)) {
 	case BUF_BLOCK_ZIP_PAGE:
@@ -259,12 +412,78 @@ buf_flush_remove(
 		break;
 	}
 
+	/* If the flush_rbt is active then delete from it as well. */
+	if (UNIV_LIKELY_NULL(buf_pool->flush_rbt)) {
+		buf_flush_delete_from_flush_rbt(bpage);
+	}
+
+	/* Must be done after we have removed it from the flush_rbt
+	because we assert on in_flush_list in comparison function. */
+	ut_d(bpage->in_flush_list = FALSE);
+
 	bpage->oldest_modification = 0;
 
 	ut_d(UT_LIST_VALIDATE(list, buf_page_t, buf_pool->flush_list,
 			      ut_ad(ut_list_node_313->in_flush_list)));
 }
 
+/********************************************************************//**
+Relocates a buffer control block on the flush_list.
+Note that it is assumed that the contents of bpage has already been
+copied to dpage. */
+UNIV_INTERN
+void
+buf_flush_relocate_on_flush_list(
+/*=============================*/
+	buf_page_t*	bpage,	/*!< in/out: control block being moved */
+	buf_page_t*	dpage)	/*!< in/out: destination block */
+{
+	buf_page_t* prev;
+	buf_page_t* prev_b = NULL;
+
+	ut_ad(buf_pool_mutex_own());
+
+	ut_ad(mutex_own(buf_page_get_mutex(bpage)));
+
+	ut_ad(bpage->in_flush_list);
+	ut_ad(dpage->in_flush_list);
+
+	/* If recovery is active we must swap the control blocks in
+	the flush_rbt as well. */
+	if (UNIV_LIKELY_NULL(buf_pool->flush_rbt)) {
+		buf_flush_delete_from_flush_rbt(bpage);
+		prev_b = buf_flush_insert_in_flush_rbt(dpage);
+	}
+
+	/* Must be done after we have removed it from the flush_rbt
+	because we assert on in_flush_list in comparison function. */
+	ut_d(bpage->in_flush_list = FALSE);
+
+	prev = UT_LIST_GET_PREV(list, bpage);
+	UT_LIST_REMOVE(list, buf_pool->flush_list, bpage);
+
+	if (prev) {
+		ut_ad(prev->in_flush_list);
+		UT_LIST_INSERT_AFTER(
+			list,
+			buf_pool->flush_list,
+			prev, dpage);
+	} else {
+		UT_LIST_ADD_FIRST(
+			list,
+			buf_pool->flush_list,
+			dpage);
+	}
+
+	/* Just an extra check. Previous in flush_list
+	should be the same control block as in flush_rbt. */
+	ut_a(!buf_pool->flush_rbt || prev_b == prev);
+
+#if defined UNIV_DEBUG || defined UNIV_BUF_DEBUG
+	ut_a(buf_flush_validate_low());
+#endif /* UNIV_DEBUG || UNIV_BUF_DEBUG */
+}
+
 /********************************************************************//**
 Updates the flush system data structures when a write is completed. */
 UNIV_INTERN
@@ -1367,24 +1586,45 @@ ibool
 buf_flush_validate_low(void)
 /*========================*/
 {
-	buf_page_t*	bpage;
+	buf_page_t*		bpage;
+	const ib_rbt_node_t*	rnode = NULL;
 
 	UT_LIST_VALIDATE(list, buf_page_t, buf_pool->flush_list,
 			 ut_ad(ut_list_node_313->in_flush_list));
 
 	bpage = UT_LIST_GET_FIRST(buf_pool->flush_list);
 
+	/* If we are in recovery mode i.e.: flush_rbt != NULL
+	then each block in the flush_list must also be present
+	in the flush_rbt. */
+	if (UNIV_LIKELY_NULL(buf_pool->flush_rbt)) {
+		rnode = rbt_first(buf_pool->flush_rbt);
+	}
+
 	while (bpage != NULL) {
 		const ib_uint64_t om = bpage->oldest_modification;
 		ut_ad(bpage->in_flush_list);
 		ut_a(buf_page_in_file(bpage));
 		ut_a(om > 0);
 
+		if (UNIV_LIKELY_NULL(buf_pool->flush_rbt)) {
+			ut_a(rnode);
+			buf_page_t* rpage = *rbt_value(buf_page_t*,
+						       rnode);
+			ut_a(rpage);
+			ut_a(rpage == bpage);
+			rnode = rbt_next(buf_pool->flush_rbt, rnode);
+		}
+
 		bpage = UT_LIST_GET_NEXT(list, bpage);
 
 		ut_a(!bpage || om >= bpage->oldest_modification);
 	}
 
+	/* By this time we must have exhausted the traversal of
+	flush_rbt (if active) as well. */
+	ut_a(rnode == NULL);
+
 	return(TRUE);
 }
 

storage/innodb_plugin/buf/buf0lru.c

 /*****************************************************************************
 
-Copyright (c) 1995, 2009, Innobase Oy. All Rights Reserved.
+Copyright (c) 1995, 2010, 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
@@ -350,17 +350,31 @@ buf_LRU_invalidate_tablespace(
 	bpage = UT_LIST_GET_LAST(buf_pool->LRU);
 
 	while (bpage != NULL) {
-		mutex_t*	block_mutex = buf_page_get_mutex(bpage);
 		buf_page_t*	prev_bpage;
+		ibool		prev_bpage_buf_fix = FALSE;
 
 		ut_a(buf_page_in_file(bpage));
 
-		mutex_enter(block_mutex);
 		prev_bpage = UT_LIST_GET_PREV(LRU, bpage);
 
-		if (buf_page_get_space(bpage) == id) {
-			if (bpage->buf_fix_count > 0
-			    || buf_page_get_io_fix(bpage) != BUF_IO_NONE) {
+		/* 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. */
+
+		if (buf_page_get_space(bpage) != id) {
+			/* Skip this block, as it does not belong to
+			the space that is being invalidated. */
+		} else if (buf_page_get_io_fix(bpage) != BUF_IO_NONE) {
+			/* We cannot remove this page during this scan
+			yet; maybe the system is currently reading it
+			in, or flushing the modifications to the file */
+
+			all_freed = FALSE;
+		} else {
+			mutex_t* block_mutex = buf_page_get_mutex(bpage);
+			mutex_enter(block_mutex);
+
+			if (bpage->buf_fix_count > 0) {
 
 				/* We cannot remove this page during
 				this scan yet; maybe the system is
@@ -380,8 +394,40 @@ buf_LRU_invalidate_tablespace(
 					(ulong) buf_page_get_page_no(bpage));
 			}
 #endif
-			if (buf_page_get_state(bpage) == BUF_BLOCK_FILE_PAGE
-			    && ((buf_block_t*) bpage)->is_hashed) {
+			if (buf_page_get_state(bpage) != BUF_BLOCK_FILE_PAGE) {
+				/* This is a compressed-only block
+				descriptor.  Ensure that prev_bpage
+				cannot be relocated when bpage is freed. */
+				if (UNIV_LIKELY(prev_bpage != NULL)) {
+					switch (buf_page_get_state(
+							prev_bpage)) {
+					case BUF_BLOCK_FILE_PAGE:
+						/* Descriptors of uncompressed
+						blocks will not be relocated,
+						because we are holding the
+						buf_pool_mutex. */
+						break;
+					case BUF_BLOCK_ZIP_PAGE:
+					case BUF_BLOCK_ZIP_DIRTY:
+						/* Descriptors of compressed-
+						only blocks can be relocated,
+						unless they are buffer-fixed.
+						Because both bpage and
+						prev_bpage are protected by
+						buf_pool_zip_mutex, it is
+						not necessary to acquire
+						further mutexes. */
+						ut_ad(&buf_pool_zip_mutex
+						      == block_mutex);
+						ut_ad(mutex_own(block_mutex));
+						prev_bpage_buf_fix = TRUE;
+						prev_bpage->buf_fix_count++;
+						break;
+					default:
+						ut_error;
+					}
+				}
+			} else if (((buf_block_t*) bpage)->is_hashed) {
 				ulint	page_no;
 				ulint	zip_size;
 
@@ -405,7 +451,8 @@ buf_LRU_invalidate_tablespace(
 				buf_flush_remove(bpage);
 			}
 
-			/* Remove from the LRU list */
+			/* Remove from the LRU list. */
+
 			if (buf_LRU_block_remove_hashed_page(bpage, TRUE)
 			    != BUF_BLOCK_ZIP_FREE) {
 				buf_LRU_block_free_hashed_page((buf_block_t*)
@@ -417,18 +464,27 @@ buf_LRU_invalidate_tablespace(
 				ut_ad(block_mutex == &buf_pool_zip_mutex);
 				ut_ad(!mutex_own(block_mutex));
 
-				/* The compressed block descriptor
-				(bpage) has been deallocated and
-				block_mutex released.  Also,
-				buf_buddy_free() may have relocated
-				prev_bpage.  Rescan the LRU list. */
+				if (prev_bpage_buf_fix) {
+					/* We temporarily buffer-fixed
+					prev_bpage, so that
+					buf_buddy_free() could not
+					relocate it, in case it was a
+					compressed-only block
+					descriptor. */
+
+					mutex_enter(block_mutex);
+					ut_ad(prev_bpage->buf_fix_count > 0);
+					prev_bpage->buf_fix_count--;
+					mutex_exit(block_mutex);
+				}
 
-				bpage = UT_LIST_GET_LAST(buf_pool->LRU);
-				continue;
+				goto next_page_no_mutex;
 			}
-		}
 next_page:
-		mutex_exit(block_mutex);
+			mutex_exit(block_mutex);
+		}
+
+next_page_no_mutex:
 		bpage = prev_bpage;
 	}
 
@@ -1474,26 +1530,8 @@ buf_LRU_free_block(
 			if (b->state == BUF_BLOCK_ZIP_PAGE) {
 				buf_LRU_insert_zip_clean(b);
 			} else {
-				buf_page_t* prev;
-
-				ut_ad(b->in_flush_list);
-				ut_d(bpage->in_flush_list = FALSE);
-
-				prev = UT_LIST_GET_PREV(list, b);
-				UT_LIST_REMOVE(list, buf_pool->flush_list, b);
-
-				if (prev) {
-					ut_ad(prev->in_flush_list);
-					UT_LIST_INSERT_AFTER(
-						list,
-						buf_pool->flush_list,
-						prev, b);
-				} else {
-					UT_LIST_ADD_FIRST(
-						list,
-						buf_pool->flush_list,
-						b);
-				}
+				/* Relocate on buf_pool->flush_list. */
+				buf_flush_relocate_on_flush_list(bpage, b);
 			}
 
 			bpage->zip.data = NULL;

storage/innodb_plugin/buf/buf0rea.c

 /*****************************************************************************
 
-Copyright (c) 1995, 2009, Innobase Oy. All Rights Reserved.
+Copyright (c) 1995, 2010, 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
@@ -608,14 +608,14 @@ buf_read_recv_pages(
 		while (buf_pool->n_pend_reads >= recv_n_pool_free_frames / 2) {
 
 			os_aio_simulated_wake_handler_threads();
-			os_thread_sleep(500000);
+			os_thread_sleep(10000);
 
 			count++;
 
-			if (count > 100) {
+			if (count > 1000) {
 				fprintf(stderr,
 					"InnoDB: Error: InnoDB has waited for"
-					" 50 seconds for pending\n"
+					" 10 seconds for pending\n"
 					"InnoDB: reads to the buffer pool to"
 					" be finished.\n"
 					"InnoDB: Number of pending reads %lu,"

storage/innodb_plugin/dict/dict0boot.c

 /*****************************************************************************
 
-Copyright (c) 1996, 2009, Innobase Oy. All Rights Reserved.
+Copyright (c) 1996, 2010, 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
@@ -274,6 +274,9 @@ dict_boot(void)
 	and (TYPE & DICT_TF_FORMAT_MASK) are nonzero and TYPE = table->flags */
 	dict_mem_table_add_col(table, heap, "TYPE", DATA_INT, 0, 4);
 	dict_mem_table_add_col(table, heap, "MIX_ID", DATA_BINARY, 0, 0);
+	/* MIX_LEN may contain additional table flags when
+	ROW_FORMAT!=REDUNDANT.  Currently, these flags include
+	DICT_TF2_TEMPORARY. */
 	dict_mem_table_add_col(table, heap, "MIX_LEN", DATA_INT, 0, 4);
 	dict_mem_table_add_col(table, heap, "CLUSTER_NAME", DATA_BINARY, 0, 0);
 	dict_mem_table_add_col(table, heap, "SPACE", DATA_INT, 0, 4);
@@ -355,7 +358,7 @@ dict_boot(void)
 	dict_mem_table_add_col(table, heap, "SPACE", DATA_INT, 0, 4);
 	dict_mem_table_add_col(table, heap, "PAGE_NO", DATA_INT, 0, 4);
 
-	/* The '+ 2' below comes from the 2 system fields */
+	/* The '+ 2' below comes from the fields DB_TRX_ID, DB_ROLL_PTR */
 #if DICT_SYS_INDEXES_PAGE_NO_FIELD != 6 + 2
 #error "DICT_SYS_INDEXES_PAGE_NO_FIELD != 6 + 2"
 #endif
@@ -364,6 +367,9 @@ dict_boot(void)
 #endif
 #if DICT_SYS_INDEXES_TYPE_FIELD != 4 + 2
 #error "DICT_SYS_INDEXES_TYPE_FIELD != 4 + 2"
+#endif
+#if DICT_SYS_INDEXES_NAME_FIELD != 1 + 2
+#error "DICT_SYS_INDEXES_NAME_FIELD != 1 + 2"
 #endif
 
 	table->id = DICT_INDEXES_ID;

storage/innodb_plugin/dict/dict0dict.c

 /*****************************************************************************
 
-Copyright (c) 1996, 2009, Innobase Oy. All Rights Reserved.
+Copyright (c) 1996, 2010, 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
@@ -140,7 +140,7 @@ static
 void
 dict_field_print_low(
 /*=================*/
-	dict_field_t*	field);	/*!< in: field */
+	const dict_field_t*	field);	/*!< in: field */
 /*********************************************************************//**
 Frees a foreign key struct. */
 static
@@ -1460,6 +1460,7 @@ dict_index_add_to_cache(
 
 	if (!dict_index_find_cols(table, index)) {
 
+		dict_mem_index_free(index);
 		return(DB_CORRUPTION);
 	}
 
@@ -4402,7 +4403,7 @@ static
 void
 dict_field_print_low(
 /*=================*/
-	dict_field_t*	field)	/*!< in: field */
+	const dict_field_t*	field)	/*!< in: field */
 {
 	ut_ad(mutex_own(&(dict_sys->mutex)));
 
@@ -4775,6 +4776,8 @@ dict_table_check_for_dup_indexes(
 	const dict_index_t*	index1;
 	const dict_index_t*	index2;
 
+	ut_ad(mutex_own(&dict_sys->mutex));
+
 	/* The primary index _must_ exist */
 	ut_a(UT_LIST_GET_LEN(table->indexes) > 0);
 

storage/innodb_plugin/dict/dict0load.c

 /*****************************************************************************
 
-Copyright (c) 1996, 2009, Innobase Oy. All Rights Reserved.
+Copyright (c) 1996, 2010, 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
@@ -260,7 +260,7 @@ dict_sys_tables_get_flags(
 		return(0);
 	}
 
-	field = rec_get_nth_field_old(rec, 4, &len);
+	field = rec_get_nth_field_old(rec, 4/*N_COLS*/, &len);
 	n_cols = mach_read_from_4(field);
 
 	if (UNIV_UNLIKELY(!(n_cols & 0x80000000UL))) {
@@ -390,15 +390,35 @@ dict_check_tablespaces_and_store_max_id(
 
 		mtr_commit(&mtr);
 
-		if (space_id != 0 && in_crash_recovery) {
+		if (space_id == 0) {
+			/* The system tablespace always exists. */
+		} else if (in_crash_recovery) {
 			/* Check that the tablespace (the .ibd file) really
-			exists; print a warning to the .err log if not */
-
-			fil_space_for_table_exists_in_mem(space_id, name,
-							  FALSE, TRUE, TRUE);
-		}
+			exists; print a warning to the .err log if not.
+			Do not print warnings for temporary tables. */
+			ibool	is_temp;
+
+			field = rec_get_nth_field_old(rec, 4, &len);
+			if (0x80000000UL &  mach_read_from_4(field)) {
+				/* ROW_FORMAT=COMPACT: read the is_temp
+				flag from SYS_TABLES.MIX_LEN. */
+				field = rec_get_nth_field_old(rec, 7, &len);
+				is_temp = mach_read_from_4(field)
+					& DICT_TF2_TEMPORARY;
+			} else {
+				/* For tables created with old versions
+				of InnoDB, SYS_TABLES.MIX_LEN may contain
+				garbage.  Such tables would always be
+				in ROW_FORMAT=REDUNDANT.  Pretend that
+				all such tables are non-temporary.  That is,
+				do not suppress error printouts about
+				temporary tables not being found. */
+				is_temp = FALSE;
+			}
 
-		if (space_id != 0 && !in_crash_recovery) {
+			fil_space_for_table_exists_in_mem(
+				space_id, name, is_temp, TRUE, !is_temp);
+		} else {
 			/* It is a normal database startup: create the space
 			object and check that the .ibd file exists. */
 
@@ -894,43 +914,72 @@ dict_load_table(
 				(ulong) flags);
 			goto err_exit;
 		}
+	} else {
+		flags = 0;
+	}
 
-		if (fil_space_for_table_exists_in_mem(space, name, FALSE,
-						      FALSE, FALSE)) {
-			/* Ok; (if we did a crash recovery then the tablespace
-			can already be in the memory cache) */
-		} else {
-			/* In >= 4.1.9, InnoDB scans the data dictionary also
-			at a normal mysqld startup. It is an error if the
-			space object does not exist in memory. */
+	ut_a(name_of_col_is(sys_tables, sys_index, 4, "N_COLS"));
 
+	field = rec_get_nth_field_old(rec, 4, &len);
+	n_cols = mach_read_from_4(field);
+
+	/* The high-order bit of N_COLS is the "compact format" flag.
+	For tables in that format, MIX_LEN may hold additional flags. */
+	if (n_cols & 0x80000000UL) {
+		ulint	flags2;
+
+		flags |= DICT_TF_COMPACT;
+
+		ut_a(name_of_col_is(sys_tables, sys_index, 7, "MIX_LEN"));
+		field = rec_get_nth_field_old(rec, 7, &len);
+
+		flags2 = mach_read_from_4(field);
+
+		if (flags2 & (~0 << (DICT_TF2_BITS - DICT_TF2_SHIFT))) {
+			ut_print_timestamp(stderr);
+			fputs("  InnoDB: Warning: table ", stderr);
+			ut_print_filename(stderr, name);
+			fprintf(stderr, "\n"
+				"InnoDB: in InnoDB data dictionary"
+				" has unknown flags %lx.\n",
+				(ulong) flags2);
+
+			flags2 &= ~(~0 << (DICT_TF2_BITS - DICT_TF2_SHIFT));
+		}
+
+		flags |= flags2 << DICT_TF2_SHIFT;
+	}
+
+	/* See if the tablespace is available. */
+	if (space == 0) {
+		/* The system tablespace is always available. */
+	} else if (!fil_space_for_table_exists_in_mem(
+			   space, name,
+			   (flags >> DICT_TF2_SHIFT) & DICT_TF2_TEMPORARY,
+			   FALSE, FALSE)) {
+
+		if ((flags >> DICT_TF2_SHIFT) & DICT_TF2_TEMPORARY) {
+			/* Do not bother to retry opening temporary tables. */
+			ibd_file_missing = TRUE;
+		} else {
 			ut_print_timestamp(stderr);
 			fprintf(stderr,
-				"  InnoDB: error: space object of table %s,\n"
+				"  InnoDB: error: space object of table");
+			ut_print_filename(stderr, name);
+			fprintf(stderr, ",\n"
 				"InnoDB: space id %lu did not exist in memory."
 				" Retrying an open.\n",
-				name, (ulong)space);
+				(ulong) space);
 			/* Try to open the tablespace */
 			if (!fil_open_single_table_tablespace(
-				    TRUE, space, flags, name)) {
-				/* We failed to find a sensible tablespace
-				file */
+				    TRUE, space,
+				    flags & ~(~0 << DICT_TF_BITS), name)) {
+				/* We failed to find a sensible
+				tablespace file */
 
 				ibd_file_missing = TRUE;
 			}
 		}
-	} else {
-		flags = 0;
-	}
-
-	ut_a(name_of_col_is(sys_tables, sys_index, 4, "N_COLS"));
-
-	field = rec_get_nth_field_old(rec, 4, &len);
-	n_cols = mach_read_from_4(field);
-
-	/* The high-order bit of N_COLS is the "compact format" flag. */
-	if (n_cols & 0x80000000UL) {
-		flags |= DICT_TF_COMPACT;
 	}
 
 	table = dict_mem_table_create(name, space, n_cols & ~0x80000000UL,

storage/innodb_plugin/dict/dict0mem.c

@@ -59,7 +59,7 @@ dict_mem_table_create(
 	mem_heap_t*	heap;
 
 	ut_ad(name);
-	ut_a(!(flags & (~0 << DICT_TF_BITS)));
+	ut_a(!(flags & (~0 << DICT_TF2_BITS)));
 
 	heap = mem_heap_create(DICT_HEAP_SIZE);
 

storage/innodb_plugin/fil/fil0fil.c

 /*****************************************************************************
 
-Copyright (c) 1995, 2009, Innobase Oy. All Rights Reserved.
+Copyright (c) 1995, 2010, 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
@@ -38,6 +38,7 @@ Created 10/25/1995 Heikki Tuuri
 #include "mtr0mtr.h"
 #include "mtr0log.h"
 #include "dict0dict.h"
+#include "page0page.h"
 #include "page0zip.h"
 #ifndef UNIV_HOTBACKUP
 # include "buf0lru.h"
@@ -1097,10 +1098,13 @@ fil_space_create(
 	fil_space_t*	space;
 
 	/* The tablespace flags (FSP_SPACE_FLAGS) should be 0 for
-	ROW_FORMAT=COMPACT (table->flags == DICT_TF_COMPACT) and
+	ROW_FORMAT=COMPACT
+	((table->flags & ~(~0 << DICT_TF_BITS)) == DICT_TF_COMPACT) and
 	ROW_FORMAT=REDUNDANT (table->flags == 0).  For any other
-	format, the tablespace flags should equal table->flags. */
+	format, the tablespace flags should equal
+	(table->flags & ~(~0 << DICT_TF_BITS)). */
 	ut_a(flags != DICT_TF_COMPACT);
+	ut_a(!(flags & (~0UL << DICT_TF_BITS)));
 
 try_again:
 	/*printf(
@@ -2582,10 +2586,13 @@ fil_create_new_single_table_tablespace(
 
 	ut_a(size >= FIL_IBD_FILE_INITIAL_SIZE);
 	/* The tablespace flags (FSP_SPACE_FLAGS) should be 0 for
-	ROW_FORMAT=COMPACT (table->flags == DICT_TF_COMPACT) and
+	ROW_FORMAT=COMPACT
+	((table->flags & ~(~0 << DICT_TF_BITS)) == DICT_TF_COMPACT) and
 	ROW_FORMAT=REDUNDANT (table->flags == 0).  For any other
-	format, the tablespace flags should equal table->flags. */
+	format, the tablespace flags should equal
+	(table->flags & ~(~0 << DICT_TF_BITS)). */
 	ut_a(flags != DICT_TF_COMPACT);
+	ut_a(!(flags & (~0UL << DICT_TF_BITS)));
 
 	path = fil_make_ibd_name(tablename, is_temp);
 
@@ -2786,6 +2793,7 @@ fil_reset_too_high_lsns(
 	ib_int64_t	offset;
 	ulint		zip_size;
 	ibool		success;
+	page_zip_des_t	page_zip;
 
 	filepath = fil_make_ibd_name(name, FALSE);
 
@@ -2833,6 +2841,12 @@ fil_reset_too_high_lsns(
 	space_id = fsp_header_get_space_id(page);
 	zip_size = fsp_header_get_zip_size(page);
 
+	page_zip_des_init(&page_zip);
+	page_zip_set_size(&page_zip, zip_size);
+	if (zip_size) {
+		page_zip.data = page + UNIV_PAGE_SIZE;
+	}
+
 	ut_print_timestamp(stderr);
 	fprintf(stderr,
 		"  InnoDB: Flush lsn in the tablespace file %lu"
@@ -2867,20 +2881,23 @@ fil_reset_too_high_lsns(
 			/* We have to reset the lsn */
 
 			if (zip_size) {
-				memcpy(page + UNIV_PAGE_SIZE, page, zip_size);
+				memcpy(page_zip.data, page, zip_size);
 				buf_flush_init_for_writing(
-					page, page + UNIV_PAGE_SIZE,
-					current_lsn);
+					page, &page_zip, current_lsn);
+				success = os_file_write(
+					filepath, file, page_zip.data,
+					(ulint) offset & 0xFFFFFFFFUL,
+					(ulint) (offset >> 32), zip_size);
 			} else {
 				buf_flush_init_for_writing(
 					page, NULL, current_lsn);
+				success = os_file_write(
+					filepath, file, page,
+					(ulint)(offset & 0xFFFFFFFFUL),
+					(ulint)(offset >> 32),
+					UNIV_PAGE_SIZE);
 			}
-			success = os_file_write(filepath, file, page,
-						(ulint)(offset & 0xFFFFFFFFUL),
-						(ulint)(offset >> 32),
-						zip_size
-						? zip_size
-						: UNIV_PAGE_SIZE);
+
 			if (!success) {
 
 				goto func_exit;
@@ -2956,10 +2973,13 @@ fil_open_single_table_tablespace(
 	filepath = fil_make_ibd_name(name, FALSE);
 
 	/* The tablespace flags (FSP_SPACE_FLAGS) should be 0 for
-	ROW_FORMAT=COMPACT (table->flags == DICT_TF_COMPACT) and
+	ROW_FORMAT=COMPACT
+	((table->flags & ~(~0 << DICT_TF_BITS)) == DICT_TF_COMPACT) and
 	ROW_FORMAT=REDUNDANT (table->flags == 0).  For any other
-	format, the tablespace flags should equal table->flags. */
+	format, the tablespace flags should equal
+	(table->flags & ~(~0 << DICT_TF_BITS)). */
 	ut_a(flags != DICT_TF_COMPACT);
+	ut_a(!(flags & (~0UL << DICT_TF_BITS)));
 
 	file = os_file_create_simple_no_error_handling(
 		filepath, OS_FILE_OPEN, OS_FILE_READ_ONLY, &success);
@@ -3011,7 +3031,8 @@ fil_open_single_table_tablespace(
 
 	ut_free(buf2);
 
-	if (UNIV_UNLIKELY(space_id != id || space_flags != flags)) {
+	if (UNIV_UNLIKELY(space_id != id
+			  || space_flags != (flags & ~(~0 << DICT_TF_BITS)))) {
 		ut_print_timestamp(stderr);
 
 		fputs("  InnoDB: Error: tablespace id and flags in file ",
@@ -4781,8 +4802,10 @@ void
 fil_close(void)
 /*===========*/
 {
+#ifndef UNIV_HOTBACKUP
 	/* The mutex should already have been freed. */
 	ut_ad(fil_system->mutex.magic_n == 0);
+#endif /* !UNIV_HOTBACKUP */
 
 	hash_table_free(fil_system->spaces);
 

storage/innodb_plugin/fsp/fsp0fsp.c

 /*****************************************************************************
 
-Copyright (c) 1995, 2009, Innobase Oy. All Rights Reserved.
+Copyright (c) 1995, 2010, 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
@@ -386,11 +386,11 @@ UNIV_INLINE
 ibool
 xdes_get_bit(
 /*=========*/
-	xdes_t*	descr,	/*!< in: descriptor */
-	ulint	bit,	/*!< in: XDES_FREE_BIT or XDES_CLEAN_BIT */
-	ulint	offset,	/*!< in: page offset within extent:
-			0 ... FSP_EXTENT_SIZE - 1 */
-	mtr_t*	mtr)	/*!< in: mtr */
+	const xdes_t*	descr,	/*!< in: descriptor */
+	ulint		bit,	/*!< in: XDES_FREE_BIT or XDES_CLEAN_BIT */
+	ulint		offset,	/*!< in: page offset within extent:
+				0 ... FSP_EXTENT_SIZE - 1 */
+	mtr_t*		mtr)	/*!< in: mtr */
 {
 	ulint	index;
 	ulint	byte_index;
@@ -527,8 +527,8 @@ UNIV_INLINE
 ulint
 xdes_get_n_used(
 /*============*/
-	xdes_t*	descr,	/*!< in: descriptor */
-	mtr_t*	mtr)	/*!< in: mtr */
+	const xdes_t*	descr,	/*!< in: descriptor */
+	mtr_t*		mtr)	/*!< in: mtr */
 {
 	ulint	i;
 	ulint	count	= 0;
@@ -551,8 +551,8 @@ UNIV_INLINE
 ibool
 xdes_is_free(
 /*=========*/
-	xdes_t*	descr,	/*!< in: descriptor */
-	mtr_t*	mtr)	/*!< in: mtr */
+	const xdes_t*	descr,	/*!< in: descriptor */
+	mtr_t*		mtr)	/*!< in: mtr */
 {
 	if (0 == xdes_get_n_used(descr, mtr)) {
 
@@ -569,8 +569,8 @@ UNIV_INLINE
 ibool
 xdes_is_full(
 /*=========*/
-	xdes_t*	descr,	/*!< in: descriptor */
-	mtr_t*	mtr)	/*!< in: mtr */
+	const xdes_t*	descr,	/*!< in: descriptor */
+	mtr_t*		mtr)	/*!< in: mtr */
 {
 	if (FSP_EXTENT_SIZE == xdes_get_n_used(descr, mtr)) {
 
@@ -586,7 +586,7 @@ UNIV_INLINE
 void
 xdes_set_state(
 /*===========*/
-	xdes_t*	descr,	/*!< in: descriptor */
+	xdes_t*	descr,	/*!< in/out: descriptor */
 	ulint	state,	/*!< in: state to set */
 	mtr_t*	mtr)	/*!< in: mtr handle */
 {
@@ -605,8 +605,8 @@ UNIV_INLINE
 ulint
 xdes_get_state(
 /*===========*/
-	xdes_t*	descr,	/*!< in: descriptor */
-	mtr_t*	mtr)	/*!< in: mtr handle */
+	const xdes_t*	descr,	/*!< in: descriptor */
+	mtr_t*		mtr)	/*!< in: mtr handle */
 {
 	ulint	state;
 
@@ -705,7 +705,7 @@ UNIV_INLINE
 xdes_t*
 xdes_get_descriptor_with_space_hdr(
 /*===============================*/
-	fsp_header_t*	sp_header,/*!< in: space header, x-latched */
+	fsp_header_t*	sp_header,/*!< in/out: space header, x-latched */
 	ulint		space,	/*!< in: space id */
 	ulint		offset,	/*!< in: page offset;
 				if equal to the free limit,
@@ -869,9 +869,7 @@ fsp_init_file_page_low(
 		return;
 	}
 
-#ifdef UNIV_BASIC_LOG_DEBUG
-	memset(page, 0xff, UNIV_PAGE_SIZE);
-#endif
+	UNIV_MEM_INVALID(page, UNIV_PAGE_SIZE);
 	mach_write_to_4(page + FIL_PAGE_OFFSET, buf_block_get_page_no(block));
 	memset(page + FIL_PAGE_LSN, 0, 8);
 	mach_write_to_4(page + FIL_PAGE_ARCH_LOG_NO_OR_SPACE_ID,
@@ -1342,7 +1340,7 @@ fsp_fill_free_list(
 					descriptor page and ibuf bitmap page;
 					then we do not allocate more extents */
 	ulint		space,		/*!< in: space */
-	fsp_header_t*	header,		/*!< in: space header */
+	fsp_header_t*	header,		/*!< in/out: space header */
 	mtr_t*		mtr)		/*!< in: mtr */
 {
 	ulint	limit;

storage/innodb_plugin/ha/ha0ha.c

@@ -101,6 +101,8 @@ ha_clear(
 	ulint	i;
 	ulint	n;
 
+	ut_ad(table);
+	ut_ad(table->magic_n == HASH_TABLE_MAGIC_N);
 #ifdef UNIV_SYNC_DEBUG
 	ut_ad(rw_lock_own(&btr_search_latch, RW_LOCK_EXCLUSIVE));
 #endif /* UNIV_SYNC_DEBUG */
@@ -146,7 +148,9 @@ ha_insert_for_fold_func(
 	ha_node_t*	prev_node;
 	ulint		hash;
 
-	ut_ad(table && data);
+	ut_ad(data);
+	ut_ad(table);
+	ut_ad(table->magic_n == HASH_TABLE_MAGIC_N);
 #if defined UNIV_AHI_DEBUG || defined UNIV_DEBUG
 	ut_a(block->frame == page_align(data));
 #endif /* UNIV_AHI_DEBUG || UNIV_DEBUG */
@@ -237,6 +241,8 @@ ha_delete_hash_node(
 	hash_table_t*	table,		/*!< in: hash table */
 	ha_node_t*	del_node)	/*!< in: node to be deleted */
 {
+	ut_ad(table);
+	ut_ad(table->magic_n == HASH_TABLE_MAGIC_N);
 #if defined UNIV_AHI_DEBUG || defined UNIV_DEBUG
 # ifndef UNIV_HOTBACKUP
 	if (table->adaptive) {
@@ -267,6 +273,8 @@ ha_search_and_update_if_found_func(
 {
 	ha_node_t*	node;
 
+	ut_ad(table);
+	ut_ad(table->magic_n == HASH_TABLE_MAGIC_N);
 	ASSERT_HASH_MUTEX_OWN(table, fold);
 #if defined UNIV_AHI_DEBUG || defined UNIV_DEBUG
 	ut_a(new_block->frame == page_align(new_data));
@@ -304,6 +312,8 @@ ha_remove_all_nodes_to_page(
 {
 	ha_node_t*	node;
 
+	ut_ad(table);
+	ut_ad(table->magic_n == HASH_TABLE_MAGIC_N);
 	ASSERT_HASH_MUTEX_OWN(table, fold);
 
 	node = ha_chain_get_first(table, fold);
@@ -353,6 +363,8 @@ ha_validate(
 	ibool		ok	= TRUE;
 	ulint		i;
 
+	ut_ad(table);
+	ut_ad(table->magic_n == HASH_TABLE_MAGIC_N);
 	ut_a(start_index <= end_index);
 	ut_a(start_index < hash_get_n_cells(table));
 	ut_a(end_index < hash_get_n_cells(table));
@@ -391,6 +403,8 @@ ha_print_info(
 	FILE*		file,	/*!< in: file where to print */
 	hash_table_t*	table)	/*!< in: hash table */
 {
+	ut_ad(table);
+	ut_ad(table->magic_n == HASH_TABLE_MAGIC_N);
 #ifdef UNIV_DEBUG
 /* Some of the code here is disabled for performance reasons in production
 builds, see http://bugs.mysql.com/36941 */

storage/innodb_plugin/ha/hash0hash.c

@@ -119,7 +119,7 @@ hash_create(
 	table->heaps = NULL;
 #endif /* !UNIV_HOTBACKUP */
 	table->heap = NULL;
-	table->magic_n = HASH_TABLE_MAGIC_N;
+	ut_d(table->magic_n = HASH_TABLE_MAGIC_N);
 
 	/* Initialize the cell array */
 	hash_table_clear(table);
@@ -135,6 +135,8 @@ hash_table_free(
 /*============*/
 	hash_table_t*	table)	/*!< in, own: hash table */
 {
+	ut_ad(table);
+	ut_ad(table->magic_n == HASH_TABLE_MAGIC_N);
 #ifndef UNIV_HOTBACKUP
 	ut_a(table->mutexes == NULL);
 #endif /* !UNIV_HOTBACKUP */
@@ -160,6 +162,8 @@ hash_create_mutexes_func(
 {
 	ulint	i;
 
+	ut_ad(table);
+	ut_ad(table->magic_n == HASH_TABLE_MAGIC_N);
 	ut_a(n_mutexes > 0);
 	ut_a(ut_is_2pow(n_mutexes));
 

storage/innodb_plugin/handler/ha_innodb.h

 /*****************************************************************************
 
-Copyright (c) 2000, 2009, MySQL AB & Innobase Oy. All Rights Reserved.
+Copyright (c) 2000, 2010, MySQL AB & 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
@@ -27,15 +27,31 @@ Place, Suite 330, Boston, MA 02111-1307 USA
 #pragma interface			/* gcc class implementation */
 #endif
 
+/* Structure defines translation table between mysql index and innodb
+index structures */
+typedef struct innodb_idx_translate_struct {
+	ulint		index_count;	/*!< number of valid index entries
+					in the index_mapping array */
+	ulint		array_size;	/*!< array size of index_mapping */
+	dict_index_t**	index_mapping;	/*!< index pointer array directly
+					maps to index in Innodb from MySQL
+					array index */
+} innodb_idx_translate_t;
+
+
 /** InnoDB table share */
 typedef struct st_innobase_share {
-	THR_LOCK	lock;		/*!< MySQL lock protecting
-					this structure */
-	const char*	table_name;	/*!< InnoDB table name */
-	uint		use_count;	/*!< reference count,
-					incremented in get_share()
-					and decremented in free_share() */
-	void*		table_name_hash;/*!< hash table chain node */
+	THR_LOCK		lock;		/*!< MySQL lock protecting
+						this structure */
+	const char*		table_name;	/*!< InnoDB table name */
+	uint			use_count;	/*!< reference count,
+						incremented in get_share()
+						and decremented in
+						free_share() */
+	void*			table_name_hash;/*!< hash table chain node */
+	innodb_idx_translate_t	idx_trans_tbl;	/*!< index translation
+						table between MySQL and
+						Innodb */
 } INNOBASE_SHARE;
 
 
@@ -91,9 +107,8 @@ class ha_innobase: public handler
 	ulint innobase_reset_autoinc(ulonglong auto_inc);
 	ulint innobase_get_autoinc(ulonglong* value);
 	ulint innobase_update_autoinc(ulonglong	auto_inc);
-	ulint innobase_initialize_autoinc();
+	void innobase_initialize_autoinc();
 	dict_index_t* innobase_get_index(uint keynr);
- 	ulonglong innobase_get_int_col_max_value(const Field* field);
 
 	/* Init values for the class: */
  public:

storage/innodb_plugin/handler/handler0alter.cc

 /*****************************************************************************
 
-Copyright (c) 2005, 2009, Innobase Oy. All Rights Reserved.
+Copyright (c) 2005, 2010, 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
@@ -229,9 +229,11 @@ static
 int
 innobase_check_index_keys(
 /*======================*/
-	const KEY*	key_info,	/*!< in: Indexes to be created */
-	ulint		num_of_keys)	/*!< in: Number of indexes to
-					be created */
+	const KEY*		key_info,	/*!< in: Indexes to be
+						created */
+	ulint			num_of_keys,	/*!< in: Number of
+						indexes to be created */
+	const dict_table_t*	table)		/*!< in: Existing indexes */
 {
 	ulint		key_num;
 
@@ -248,9 +250,22 @@ innobase_check_index_keys(
 			const KEY&	key2 = key_info[i];
 
 			if (0 == strcmp(key.name, key2.name)) {
-				sql_print_error("InnoDB: key name `%s` appears"
-						" twice in CREATE INDEX\n",
-						key.name);
+				my_error(ER_WRONG_NAME_FOR_INDEX, MYF(0),
+					 key.name);
+
+				return(ER_WRONG_NAME_FOR_INDEX);
+			}
+		}
+
+		/* Check that the same index name does not already exist. */
+
+		for (const dict_index_t* index
+			     = dict_table_get_first_index(table);
+		     index; index = dict_table_get_next_index(index)) {
+
+			if (0 == strcmp(key.name, index->name)) {
+				my_error(ER_WRONG_NAME_FOR_INDEX, MYF(0),
+					 key.name);
 
 				return(ER_WRONG_NAME_FOR_INDEX);
 			}
@@ -258,7 +273,7 @@ innobase_check_index_keys(
 
 		/* Check that MySQL does not try to create a column
 		prefix index field on an inappropriate data type and
-		that the same colum does not appear twice in the index. */
+		that the same column does not appear twice in the index. */
 
 		for (ulint i = 0; i < key.key_parts; i++) {
 			const KEY_PART_INFO&	key_part1
@@ -289,14 +304,8 @@ innobase_check_index_keys(
 					}
 				}
 
-				sql_print_error("InnoDB: MySQL is trying to"
-						" create a column prefix"
-						" index field on an"
-						" inappropriate data type."
-						" column `%s`,"
-						" index `%s`.\n",
-						field->field_name,
-						key.name);
+				my_error(ER_WRONG_KEY_COLUMN, MYF(0),
+					 field->field_name);
 				return(ER_WRONG_KEY_COLUMN);
 			}
 
@@ -309,11 +318,8 @@ innobase_check_index_keys(
 					continue;
 				}
 
-				sql_print_error("InnoDB: column `%s`"
-						" is not allowed to occur"
-						" twice in index `%s`.\n",
-						key_part1.field->field_name,
-						key.name);
+				my_error(ER_WRONG_KEY_COLUMN, MYF(0),
+					 key_part1.field->field_name);
 				return(ER_WRONG_KEY_COLUMN);
 			}
 		}
@@ -522,12 +528,14 @@ innobase_create_key_def(
 				     key_info->name, "PRIMARY");
 
 	/* If there is a UNIQUE INDEX consisting entirely of NOT NULL
-	columns, MySQL will treat it as a PRIMARY KEY unless the
-	table already has one. */
+	columns and if the index does not contain column prefix(es)
+	(only prefix/part of the column is indexed), MySQL will treat the
+	index as a PRIMARY KEY unless the table already has one. */
 
 	if (!new_primary && (key_info->flags & HA_NOSAME)
+	    && (!(key_info->flags & HA_KEY_HAS_PART_KEY_SEG))
 	    && row_table_got_default_clust_index(table)) {
-		uint	key_part = key_info->key_parts;
+		uint    key_part = key_info->key_parts;
 
 		new_primary = TRUE;
 
@@ -656,12 +664,18 @@ ha_innobase::add_index(
 	innodb_table = indexed_table
 		= dict_table_get(prebuilt->table->name, FALSE);
 
+	if (UNIV_UNLIKELY(!innodb_table)) {
+		error = HA_ERR_NO_SUCH_TABLE;
+		goto err_exit;
+	}
+
 	/* Check if the index name is reserved. */
 	if (innobase_index_name_is_reserved(trx, key_info, num_of_keys)) {
 		error = -1;
 	} else {
 		/* Check that index keys are sensible */
-		error = innobase_check_index_keys(key_info, num_of_keys);
+		error = innobase_check_index_keys(key_info, num_of_keys,
+						  innodb_table);
 	}
 
 	if (UNIV_UNLIKELY(error)) {
@@ -764,6 +778,10 @@ ha_innobase::add_index(
 
 	ut_ad(error == DB_SUCCESS);
 
+	/* We will need to rebuild index translation table. Set
+	valid index entry count in the translation table to zero */
+	share->idx_trans_tbl.index_count = 0;
+
 	/* Commit the data dictionary transaction in order to release
 	the table locks on the system tables.  This means that if
 	MySQL crashes while creating a new primary key inside
@@ -799,18 +817,6 @@ ha_innobase::add_index(
 					index, num_of_idx, table);
 
 error_handling:
-#ifdef UNIV_DEBUG
-	/* TODO: At the moment we can't handle the following statement
-	in our debugging code below:
-
-	alter table t drop index b, add index (b);
-
-	The fix will have to parse the SQL and note that the index
-	being added has the same name as the one being dropped and
-	ignore that in the dup index check.*/
-	//dict_table_check_for_dup_indexes(prebuilt->table);
-#endif
-
 	/* After an error, remove all those index definitions from the
 	dictionary which were defined. */
 
@@ -822,6 +828,8 @@ ha_innobase::add_index(
 		row_mysql_lock_data_dictionary(trx);
 		dict_locked = TRUE;
 
+		ut_d(dict_table_check_for_dup_indexes(prebuilt->table));
+
 		if (!new_primary) {
 			error = row_merge_rename_indexes(trx, indexed_table);
 
@@ -1198,9 +1206,11 @@ ha_innobase::final_drop_index(
 		ut_a(!index->to_be_dropped);
 	}
 
-#ifdef UNIV_DEBUG
-	dict_table_check_for_dup_indexes(prebuilt->table);
-#endif
+	/* We will need to rebuild index translation table. Set
+	valid index entry count in the translation table to zero */
+	share->idx_trans_tbl.index_count = 0;
+
+	ut_d(dict_table_check_for_dup_indexes(prebuilt->table));
 
 func_exit:
 	trx_commit_for_mysql(trx);

storage/innodb_plugin/ibuf/ibuf0ibuf.c

@@ -730,24 +730,41 @@ page containing the descriptor bits for the file page; the bitmap page
 is x-latched */
 static
 page_t*
-ibuf_bitmap_get_map_page(
-/*=====================*/
-	ulint	space,	/*!< in: space id of the file page */
-	ulint	page_no,/*!< in: page number of the file page */
-	ulint	zip_size,/*!< in: compressed page size in bytes;
-			0 for uncompressed pages */
-	mtr_t*	mtr)	/*!< in: mtr */
+ibuf_bitmap_get_map_page_func(
+/*==========================*/
+	ulint		space,	/*!< in: space id of the file page */
+	ulint		page_no,/*!< in: page number of the file page */
+	ulint		zip_size,/*!< in: compressed page size in bytes;
+				0 for uncompressed pages */
+	const char*	file,	/*!< in: file name */
+	ulint		line,	/*!< in: line where called */
+	mtr_t*		mtr)	/*!< in: mtr */
 {
 	buf_block_t*	block;
 
-	block = buf_page_get(space, zip_size,
-			     ibuf_bitmap_page_no_calc(zip_size, page_no),
-			     RW_X_LATCH, mtr);
+	block = buf_page_get_gen(space, zip_size,
+				 ibuf_bitmap_page_no_calc(zip_size, page_no),
+				 RW_X_LATCH, NULL, BUF_GET,
+				 file, line, mtr);
 	buf_block_dbg_add_level(block, SYNC_IBUF_BITMAP);
 
 	return(buf_block_get_frame(block));
 }
 
+/********************************************************************//**
+Gets the ibuf bitmap page where the bits describing a given file page are
+stored.
+@return bitmap page where the file page is mapped, that is, the bitmap
+page containing the descriptor bits for the file page; the bitmap page
+is x-latched
+@param space	in: space id of the file page
+@param page_no	in: page number of the file page
+@param zip_size	in: compressed page size in bytes; 0 for uncompressed pages
+@param mtr	in: mini-transaction */
+#define ibuf_bitmap_get_map_page(space, page_no, zip_size, mtr)		\
+	ibuf_bitmap_get_map_page_func(space, page_no, zip_size,		\
+				      __FILE__, __LINE__, mtr)
+
 /************************************************************************//**
 Sets the free bits of the page in the ibuf bitmap. This is done in a separate
 mini-transaction, hence this operation does not restrict further work to only

storage/innodb_plugin/include/btr0btr.h

 /*****************************************************************************
 
-Copyright (c) 1994, 2009, Innobase Oy. All Rights Reserved.
+Copyright (c) 1994, 2010, 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
@@ -193,6 +193,10 @@ btr_leaf_page_release(
 	mtr_t*		mtr);		/*!< in: mtr */
 /**************************************************************//**
 Gets the child node file address in a node pointer.
+NOTE: the offsets array must contain all offsets for the record since
+we read the last field according to offsets and assume that it contains
+the child page number. In other words offsets must have been retrieved
+with rec_get_offsets(n_fields=ULINT_UNDEFINED).
 @return	child node address */
 UNIV_INLINE
 ulint
@@ -317,12 +321,16 @@ Inserts a data tuple to a tree on a non-leaf level. It is assumed
 that mtr holds an x-latch on the tree. */
 UNIV_INTERN
 void
-btr_insert_on_non_leaf_level(
-/*=========================*/
+btr_insert_on_non_leaf_level_func(
+/*==============================*/
 	dict_index_t*	index,	/*!< in: index */
 	ulint		level,	/*!< in: level, must be > 0 */
 	dtuple_t*	tuple,	/*!< in: the record to be inserted */
+	const char*	file,	/*!< in: file name */
+	ulint		line,	/*!< in: line where called */
 	mtr_t*		mtr);	/*!< in: mtr */
+# define btr_insert_on_non_leaf_level(i,l,t,m)				\
+	btr_insert_on_non_leaf_level_func(i,l,t,__FILE__,__LINE__,m)
 #endif /* !UNIV_HOTBACKUP */
 /****************************************************************//**
 Sets a record as the predefined minimum record. */

storage/innodb_plugin/include/btr0btr.ic

 /*****************************************************************************
 
-Copyright (c) 1994, 2009, Innobase Oy. All Rights Reserved.
+Copyright (c) 1994, 2010, 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
@@ -255,6 +255,10 @@ btr_page_set_prev(
 
 /**************************************************************//**
 Gets the child node file address in a node pointer.
+NOTE: the offsets array must contain all offsets for the record since
+we read the last field according to offsets and assume that it contains
+the child page number. In other words offsets must have been retrieved
+with rec_get_offsets(n_fields=ULINT_UNDEFINED).
 @return	child node address */
 UNIV_INLINE
 ulint

storage/innodb_plugin/include/btr0cur.h

 /*****************************************************************************
 
-Copyright (c) 1994, 2009, Innobase Oy. All Rights Reserved.
+Copyright (c) 1994, 2010, 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
@@ -152,29 +152,39 @@ btr_cur_search_to_nth_level(
 	ulint		has_search_latch,/*!< in: latch mode the caller
 				currently has on btr_search_latch:
 				RW_S_LATCH, or 0 */
+	const char*	file,	/*!< in: file name */
+	ulint		line,	/*!< in: line where called */
 	mtr_t*		mtr);	/*!< in: mtr */
 /*****************************************************************//**
 Opens a cursor at either end of an index. */
 UNIV_INTERN
 void
-btr_cur_open_at_index_side(
-/*=======================*/
+btr_cur_open_at_index_side_func(
+/*============================*/
 	ibool		from_left,	/*!< in: TRUE if open to the low end,
 					FALSE if to the high end */
 	dict_index_t*	index,		/*!< in: index */
 	ulint		latch_mode,	/*!< in: latch mode */
 	btr_cur_t*	cursor,		/*!< in: cursor */
+	const char*	file,		/*!< in: file name */
+	ulint		line,		/*!< in: line where called */
 	mtr_t*		mtr);		/*!< in: mtr */
+#define btr_cur_open_at_index_side(f,i,l,c,m)				\
+	btr_cur_open_at_index_side_func(f,i,l,c,__FILE__,__LINE__,m)
 /**********************************************************************//**
 Positions a cursor at a randomly chosen position within a B-tree. */
 UNIV_INTERN
 void
-btr_cur_open_at_rnd_pos(
-/*====================*/
+btr_cur_open_at_rnd_pos_func(
+/*=========================*/
 	dict_index_t*	index,		/*!< in: index */
 	ulint		latch_mode,	/*!< in: BTR_SEARCH_LEAF, ... */
 	btr_cur_t*	cursor,		/*!< in/out: B-tree cursor */
+	const char*	file,		/*!< in: file name */
+	ulint		line,		/*!< in: line where called */
 	mtr_t*		mtr);		/*!< in: mtr */
+#define btr_cur_open_at_rnd_pos(i,l,c,m)				\
+	btr_cur_open_at_rnd_pos_func(i,l,c,__FILE__,__LINE__,m)
 /*************************************************************//**
 Tries to perform an insert to a page in an index tree, next to cursor.
 It is assumed that mtr holds an x-latch on the page. The operation does

storage/innodb_plugin/include/btr0pcur.h

 /*****************************************************************************
 
-Copyright (c) 1996, 2009, Innobase Oy. All Rights Reserved.
+Copyright (c) 1996, 2010, 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
@@ -82,8 +82,8 @@ Initializes and opens a persistent cursor to an index tree. It should be
 closed with btr_pcur_close. */
 UNIV_INLINE
 void
-btr_pcur_open(
-/*==========*/
+btr_pcur_open_func(
+/*===============*/
 	dict_index_t*	index,	/*!< in: index */
 	const dtuple_t*	tuple,	/*!< in: tuple on which search done */
 	ulint		mode,	/*!< in: PAGE_CUR_L, ...;
@@ -94,14 +94,18 @@ btr_pcur_open(
 				record! */
 	ulint		latch_mode,/*!< in: BTR_SEARCH_LEAF, ... */
 	btr_pcur_t*	cursor, /*!< in: memory buffer for persistent cursor */
+	const char*	file,	/*!< in: file name */
+	ulint		line,	/*!< in: line where called */
 	mtr_t*		mtr);	/*!< in: mtr */
+#define btr_pcur_open(i,t,md,l,c,m)				\
+	btr_pcur_open_func(i,t,md,l,c,__FILE__,__LINE__,m)
 /**************************************************************//**
 Opens an persistent cursor to an index tree without initializing the
 cursor. */
 UNIV_INLINE
 void
-btr_pcur_open_with_no_init(
-/*=======================*/
+btr_pcur_open_with_no_init_func(
+/*============================*/
 	dict_index_t*	index,	/*!< in: index */
 	const dtuple_t*	tuple,	/*!< in: tuple on which search done */
 	ulint		mode,	/*!< in: PAGE_CUR_L, ...;
@@ -119,7 +123,12 @@ btr_pcur_open_with_no_init(
 	ulint		has_search_latch,/*!< in: latch mode the caller
 				currently has on btr_search_latch:
 				RW_S_LATCH, or 0 */
+	const char*	file,	/*!< in: file name */
+	ulint		line,	/*!< in: line where called */
 	mtr_t*		mtr);	/*!< in: mtr */
+#define btr_pcur_open_with_no_init(ix,t,md,l,cur,has,m)			\
+	btr_pcur_open_with_no_init_func(ix,t,md,l,cur,has,__FILE__,__LINE__,m)
+
 /*****************************************************************//**
 Opens a persistent cursor at either end of an index. */
 UNIV_INLINE
@@ -160,8 +169,8 @@ before first in tree. The latching mode must be BTR_SEARCH_LEAF or
 BTR_MODIFY_LEAF. */
 UNIV_INTERN
 void
-btr_pcur_open_on_user_rec(
-/*======================*/
+btr_pcur_open_on_user_rec_func(
+/*===========================*/
 	dict_index_t*	index,		/*!< in: index */
 	const dtuple_t*	tuple,		/*!< in: tuple on which search done */
 	ulint		mode,		/*!< in: PAGE_CUR_L, ... */
@@ -169,17 +178,25 @@ btr_pcur_open_on_user_rec(
 					BTR_MODIFY_LEAF */
 	btr_pcur_t*	cursor,		/*!< in: memory buffer for persistent
 					cursor */
+	const char*	file,		/*!< in: file name */
+	ulint		line,		/*!< in: line where called */
 	mtr_t*		mtr);		/*!< in: mtr */
+#define btr_pcur_open_on_user_rec(i,t,md,l,c,m)				\
+	btr_pcur_open_on_user_rec_func(i,t,md,l,c,__FILE__,__LINE__,m)
 /**********************************************************************//**
 Positions a cursor at a randomly chosen position within a B-tree. */
 UNIV_INLINE
 void
-btr_pcur_open_at_rnd_pos(
-/*=====================*/
+btr_pcur_open_at_rnd_pos_func(
+/*==========================*/
 	dict_index_t*	index,		/*!< in: index */
 	ulint		latch_mode,	/*!< in: BTR_SEARCH_LEAF, ... */
 	btr_pcur_t*	cursor,		/*!< in/out: B-tree pcur */
+	const char*	file,		/*!< in: file name */
+	ulint		line,		/*!< in: line where called */
 	mtr_t*		mtr);		/*!< in: mtr */
+#define btr_pcur_open_at_rnd_pos(i,l,c,m)				\
+	btr_pcur_open_at_rnd_pos_func(i,l,c,__FILE__,__LINE__,m)
 /**************************************************************//**
 Frees the possible old_rec_buf buffer of a persistent cursor and sets the
 latch mode of the persistent cursor to BTR_NO_LATCHES. */
@@ -218,11 +235,15 @@ record and it can be restored on a user record whose ordering fields
 are identical to the ones of the original user record */
 UNIV_INTERN
 ibool
-btr_pcur_restore_position(
-/*======================*/
+btr_pcur_restore_position_func(
+/*===========================*/
 	ulint		latch_mode,	/*!< in: BTR_SEARCH_LEAF, ... */
 	btr_pcur_t*	cursor,		/*!< in: detached persistent cursor */
+	const char*	file,		/*!< in: file name */
+	ulint		line,		/*!< in: line where called */
 	mtr_t*		mtr);		/*!< in: mtr */
+#define btr_pcur_restore_position(l,cur,mtr)				\
+	btr_pcur_restore_position_func(l,cur,__FILE__,__LINE__,mtr)
 /**************************************************************//**
 If the latch mode of the cursor is BTR_LEAF_SEARCH or BTR_LEAF_MODIFY,
 releases the page latch and bufferfix reserved by the cursor.
@@ -260,20 +281,13 @@ btr_pcur_get_mtr(
 /*=============*/
 	btr_pcur_t*	cursor);	/*!< in: persistent cursor */
 /**************************************************************//**
-Commits the pcur mtr and sets the pcur latch mode to BTR_NO_LATCHES,
+Commits the mtr and sets the pcur latch mode to BTR_NO_LATCHES,
 that is, the cursor becomes detached. If there have been modifications
 to the page where pcur is positioned, this can be used instead of
 btr_pcur_release_leaf. Function btr_pcur_store_position should be used
 before calling this, if restoration of cursor is wanted later. */
 UNIV_INLINE
 void
-btr_pcur_commit(
-/*============*/
-	btr_pcur_t*	pcur);	/*!< in: persistent cursor */
-/**************************************************************//**
-Differs from btr_pcur_commit in that we can specify the mtr to commit. */
-UNIV_INLINE
-void
 btr_pcur_commit_specify_mtr(
 /*========================*/
 	btr_pcur_t*	pcur,	/*!< in: persistent cursor */

storage/innodb_plugin/include/btr0pcur.ic

 /*****************************************************************************
 
-Copyright (c) 1996, 2009, Innobase Oy. All Rights Reserved.
+Copyright (c) 1996, 2010, 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
@@ -395,30 +395,13 @@ btr_pcur_move_to_next(
 }
 
 /**************************************************************//**
-Commits the pcur mtr and sets the pcur latch mode to BTR_NO_LATCHES,
+Commits the mtr and sets the pcur latch mode to BTR_NO_LATCHES,
 that is, the cursor becomes detached. If there have been modifications
 to the page where pcur is positioned, this can be used instead of
 btr_pcur_release_leaf. Function btr_pcur_store_position should be used
 before calling this, if restoration of cursor is wanted later. */
 UNIV_INLINE
 void
-btr_pcur_commit(
-/*============*/
-	btr_pcur_t*	pcur)	/*!< in: persistent cursor */
-{
-	ut_a(pcur->pos_state == BTR_PCUR_IS_POSITIONED);
-
-	pcur->latch_mode = BTR_NO_LATCHES;
-
-	mtr_commit(pcur->mtr);
-
-	pcur->pos_state = BTR_PCUR_WAS_POSITIONED;
-}
-
-/**************************************************************//**
-Differs from btr_pcur_commit in that we can specify the mtr to commit. */
-UNIV_INLINE
-void
 btr_pcur_commit_specify_mtr(
 /*========================*/
 	btr_pcur_t*	pcur,	/*!< in: persistent cursor */
@@ -483,8 +466,8 @@ Initializes and opens a persistent cursor to an index tree. It should be
 closed with btr_pcur_close. */
 UNIV_INLINE
 void
-btr_pcur_open(
-/*==========*/
+btr_pcur_open_func(
+/*===============*/
 	dict_index_t*	index,	/*!< in: index */
 	const dtuple_t*	tuple,	/*!< in: tuple on which search done */
 	ulint		mode,	/*!< in: PAGE_CUR_L, ...;
@@ -495,6 +478,8 @@ btr_pcur_open(
 				record! */
 	ulint		latch_mode,/*!< in: BTR_SEARCH_LEAF, ... */
 	btr_pcur_t*	cursor, /*!< in: memory buffer for persistent cursor */
+	const char*	file,	/*!< in: file name */
+	ulint		line,	/*!< in: line where called */
 	mtr_t*		mtr)	/*!< in: mtr */
 {
 	btr_cur_t*	btr_cursor;
@@ -511,7 +496,7 @@ btr_pcur_open(
 	btr_cursor = btr_pcur_get_btr_cur(cursor);
 
 	btr_cur_search_to_nth_level(index, 0, tuple, mode, latch_mode,
-				    btr_cursor, 0, mtr);
+				    btr_cursor, 0, file, line, mtr);
 	cursor->pos_state = BTR_PCUR_IS_POSITIONED;
 
 	cursor->trx_if_known = NULL;
@@ -522,8 +507,8 @@ Opens an persistent cursor to an index tree without initializing the
 cursor. */
 UNIV_INLINE
 void
-btr_pcur_open_with_no_init(
-/*=======================*/
+btr_pcur_open_with_no_init_func(
+/*============================*/
 	dict_index_t*	index,	/*!< in: index */
 	const dtuple_t*	tuple,	/*!< in: tuple on which search done */
 	ulint		mode,	/*!< in: PAGE_CUR_L, ...;
@@ -541,6 +526,8 @@ btr_pcur_open_with_no_init(
 	ulint		has_search_latch,/*!< in: latch mode the caller
 				currently has on btr_search_latch:
 				RW_S_LATCH, or 0 */
+	const char*	file,	/*!< in: file name */
+	ulint		line,	/*!< in: line where called */
 	mtr_t*		mtr)	/*!< in: mtr */
 {
 	btr_cur_t*	btr_cursor;
@@ -553,7 +540,8 @@ btr_pcur_open_with_no_init(
 	btr_cursor = btr_pcur_get_btr_cur(cursor);
 
 	btr_cur_search_to_nth_level(index, 0, tuple, mode, latch_mode,
-				    btr_cursor, has_search_latch, mtr);
+				    btr_cursor, has_search_latch,
+				    file, line, mtr);
 	cursor->pos_state = BTR_PCUR_IS_POSITIONED;
 
 	cursor->old_stored = BTR_PCUR_OLD_NOT_STORED;
@@ -600,11 +588,13 @@ btr_pcur_open_at_index_side(
 Positions a cursor at a randomly chosen position within a B-tree. */
 UNIV_INLINE
 void
-btr_pcur_open_at_rnd_pos(
-/*=====================*/
+btr_pcur_open_at_rnd_pos_func(
+/*==========================*/
 	dict_index_t*	index,		/*!< in: index */
 	ulint		latch_mode,	/*!< in: BTR_SEARCH_LEAF, ... */
 	btr_pcur_t*	cursor,		/*!< in/out: B-tree pcur */
+	const char*	file,		/*!< in: file name */
+	ulint		line,		/*!< in: line where called */
 	mtr_t*		mtr)		/*!< in: mtr */
 {
 	/* Initialize the cursor */
@@ -614,8 +604,9 @@ btr_pcur_open_at_rnd_pos(
 
 	btr_pcur_init(cursor);
 
-	btr_cur_open_at_rnd_pos(index, latch_mode,
-				btr_pcur_get_btr_cur(cursor), mtr);
+	btr_cur_open_at_rnd_pos_func(index, latch_mode,
+				     btr_pcur_get_btr_cur(cursor),
+				     file, line, mtr);
 	cursor->pos_state = BTR_PCUR_IS_POSITIONED;
 	cursor->old_stored = BTR_PCUR_OLD_NOT_STORED;
 

storage/innodb_plugin/include/buf0buf.h

 /*****************************************************************************
 
-Copyright (c) 1995, 2009, Innobase Oy. All Rights Reserved.
+Copyright (c) 1995, 2010, 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
@@ -33,6 +33,7 @@ Created 11/5/1995 Heikki Tuuri
 #include "hash0hash.h"
 #include "ut0byte.h"
 #include "page0types.h"
+#include "ut0rbt.h"
 #ifndef UNIV_HOTBACKUP
 #include "os0proc.h"
 
@@ -202,20 +203,14 @@ with care. */
 #define buf_page_get_with_no_latch(SP, ZS, OF, MTR)	   buf_page_get_gen(\
 				SP, ZS, OF, RW_NO_LATCH, NULL,\
 				BUF_GET_NO_LATCH, __FILE__, __LINE__, MTR)
-/**************************************************************//**
-NOTE! The following macros should be used instead of
-buf_page_optimistic_get_func, to improve debugging. Only values RW_S_LATCH and
-RW_X_LATCH are allowed as LA! */
-#define buf_page_optimistic_get(LA, BL, MC, MTR)			     \
-	buf_page_optimistic_get_func(LA, BL, MC, __FILE__, __LINE__, MTR)
 /********************************************************************//**
 This is the general function used to get optimistic access to a database
 page.
 @return	TRUE if success */
 UNIV_INTERN
 ibool
-buf_page_optimistic_get_func(
-/*=========================*/
+buf_page_optimistic_get(
+/*====================*/
 	ulint		rw_latch,/*!< in: RW_S_LATCH, RW_X_LATCH */
 	buf_block_t*	block,	/*!< in: guessed block */
 	ib_uint64_t	modify_clock,/*!< in: modify clock value if mode is
@@ -1185,15 +1180,21 @@ struct buf_block_struct{
 	rw_lock_t	lock;		/*!< read-write lock of the buffer
 					frame */
 	unsigned	lock_hash_val:32;/*!< hashed value of the page address
-					in the record lock hash table */
-	unsigned	check_index_page_at_flush:1;
+					in the record lock hash table;
+					protected by buf_block_t::lock
+					(or buf_block_t::mutex, buf_pool_mutex
+				        in buf_page_get_gen(),
+					buf_page_init_for_read()
+					and buf_page_create()) */
+	ibool		check_index_page_at_flush;
 					/*!< TRUE if we know that this is
 					an index page, and want the database
 					to check its consistency before flush;
 					note that there may be pages in the
 					buffer pool which are index pages,
 					but this flag is not set because
-					we do not keep track of all pages */
+					we do not keep track of all pages;
+					NOT protected by any mutex */
 	/* @} */
 	/** @name Optimistic search field */
 	/* @{ */
@@ -1359,6 +1360,19 @@ struct buf_pool_struct{
 					/*!< this is in the set state
 					when there is no flush batch
 					of the given type running */
+	ib_rbt_t*	flush_rbt;	/* !< a red-black tree is used
+					exclusively during recovery to
+					speed up insertions in the
+					flush_list. This tree contains
+					blocks in order of
+					oldest_modification LSN and is
+					kept in sync with the
+					flush_list.
+					Each member of the tree MUST
+					also be on the flush_list.
+					This tree is relevant only in
+					recovery and is set to NULL
+					once the recovery is over. */
 	ulint		freed_page_clock;/*!< a sequence number used
 					to count the number of buffer
 					blocks removed from the end of

storage/innodb_plugin/include/buf0buf.ic

 /*****************************************************************************
 
-Copyright (c) 1995, 2009, Innobase Oy. All Rights Reserved.
+Copyright (c) 1995, 2010, Innobase Oy. All Rights Reserved.
 Copyright (c) 2008, Google Inc.
 
 Portions of this file contain modifications contributed and copyrighted by
@@ -81,7 +81,7 @@ buf_page_peek_if_too_old(
 		unsigned	access_time = buf_page_is_accessed(bpage);
 
 		if (access_time > 0
-		    && (ut_time_ms() - access_time)
+		    && ((ib_uint32_t) (ut_time_ms() - access_time))
 		    >= buf_LRU_old_threshold_ms) {
 			return(TRUE);
 		}
@@ -705,6 +705,12 @@ buf_block_get_lock_hash_val(
 /*========================*/
 	const buf_block_t*	block)	/*!< in: block */
 {
+	ut_ad(block);
+	ut_ad(buf_page_in_file(&block->page));
+#ifdef UNIV_SYNC_DEBUG
+	ut_ad(rw_lock_own(&(((buf_block_t*) block)->lock), RW_LOCK_EXCLUSIVE)
+	      || rw_lock_own(&(((buf_block_t*) block)->lock), RW_LOCK_SHARED));
+#endif /* UNIV_SYNC_DEBUG */
 	return(block->lock_hash_val);
 }
 

storage/innodb_plugin/include/buf0flu.h

 /*****************************************************************************
 
-Copyright (c) 1995, 2009, Innobase Oy. All Rights Reserved.
+Copyright (c) 1995, 2010, 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
@@ -40,6 +40,16 @@ buf_flush_remove(
 /*=============*/
 	buf_page_t*	bpage);	/*!< in: pointer to the block in question */
 /********************************************************************//**
+Relocates a buffer control block on the flush_list.
+Note that it is assumed that the contents of bpage has already been
+copied to dpage. */
+UNIV_INTERN
+void
+buf_flush_relocate_on_flush_list(
+/*=============================*/
+	buf_page_t*	bpage,	/*!< in/out: control block being moved */
+	buf_page_t*	dpage);	/*!< in/out: destination block */
+/********************************************************************//**
 Updates the flush system data structures when a write is completed. */
 UNIV_INTERN
 void
@@ -139,8 +149,8 @@ how much redo the workload is generating and at what rate. */
 
 struct buf_flush_stat_struct
 {
-	ib_uint64_t	redo;		/**< amount of redo generated. */
-	ulint		n_flushed;	/**< number of pages flushed. */
+	ib_uint64_t	redo;		/*!< amount of redo generated. */
+	ulint		n_flushed;	/*!< number of pages flushed. */
 };
 
 /** Statistics for selecting flush rate of dirty pages. */
@@ -175,6 +185,22 @@ buf_flush_validate(void);
 /*====================*/
 #endif /* UNIV_DEBUG || UNIV_BUF_DEBUG */
 
+/******************************************************************//**
+Initialize the red-black tree to speed up insertions into the flush_list
+during recovery process. Should be called at the start of recovery
+process before any page has been read/written. */
+UNIV_INTERN
+void
+buf_flush_init_flush_rbt(void);
+/*==========================*/
+
+/******************************************************************//**
+Frees up the red-black tree. */
+UNIV_INTERN
+void
+buf_flush_free_flush_rbt(void);
+/*==========================*/
+
 /** When buf_flush_free_margin is called, it tries to make this many blocks
 available to replacement in the free list and at the end of the LRU list (to
 make sure that a read-ahead batch can be read efficiently in a single

storage/innodb_plugin/include/data0type.ic

 /*****************************************************************************
 
-Copyright (c) 1996, 2009, Innobase Oy. All Rights Reserved.
+Copyright (c) 1996, 2010, 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
@@ -285,6 +285,10 @@ dtype_new_store_for_order_and_null_size(
 #endif
 	ulint	len;
 
+	ut_ad(type);
+	ut_ad(type->mtype >= DATA_VARCHAR);
+	ut_ad(type->mtype <= DATA_MYSQL);
+
 	buf[0] = (byte)(type->mtype & 0xFFUL);
 
 	if (type->prtype & DATA_BINARY_TYPE) {

storage/innodb_plugin/include/dict0boot.h

 /*****************************************************************************
 
-Copyright (c) 1996, 2009, Innobase Oy. All Rights Reserved.
+Copyright (c) 1996, 2010, 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
@@ -137,6 +137,7 @@ clustered index */
 #define DICT_SYS_INDEXES_PAGE_NO_FIELD	 8
 #define DICT_SYS_INDEXES_SPACE_NO_FIELD	 7
 #define DICT_SYS_INDEXES_TYPE_FIELD	 6
+#define DICT_SYS_INDEXES_NAME_FIELD	 3
 
 /* When a row id which is zero modulo this number (which must be a power of
 two) is assigned, the field DICT_HDR_ROW_ID on the dictionary header page is

storage/innodb_plugin/include/dict0mem.h

@@ -80,21 +80,39 @@ combination of types */
 /** File format */
 /* @{ */
 #define DICT_TF_FORMAT_SHIFT		5	/* file format */
-#define DICT_TF_FORMAT_MASK		(127 << DICT_TF_FORMAT_SHIFT)
+#define DICT_TF_FORMAT_MASK		\
+((~(~0 << (DICT_TF_BITS - DICT_TF_FORMAT_SHIFT))) << DICT_TF_FORMAT_SHIFT)
 #define DICT_TF_FORMAT_51		0	/*!< InnoDB/MySQL up to 5.1 */
 #define DICT_TF_FORMAT_ZIP		1	/*!< InnoDB plugin for 5.1:
 						compressed tables,
 						new BLOB treatment */
 /** Maximum supported file format */
 #define DICT_TF_FORMAT_MAX		DICT_TF_FORMAT_ZIP
-
+/* @} */
 #define DICT_TF_BITS			6	/*!< number of flag bits */
 #if (1 << (DICT_TF_BITS - DICT_TF_FORMAT_SHIFT)) <= DICT_TF_FORMAT_MAX
 # error "DICT_TF_BITS is insufficient for DICT_TF_FORMAT_MAX"
 #endif
 /* @} */
+
+/** @brief Additional table flags.
+
+These flags will be stored in SYS_TABLES.MIX_LEN.  All unused flags
+will be written as 0.  The column may contain garbage for tables
+created with old versions of InnoDB that only implemented
+ROW_FORMAT=REDUNDANT. */
+/* @{ */
+#define DICT_TF2_SHIFT			DICT_TF_BITS
+						/*!< Shift value for
+						table->flags. */
+#define DICT_TF2_TEMPORARY		1	/*!< TRUE for tables from
+						CREATE TEMPORARY TABLE. */
+#define DICT_TF2_BITS			(DICT_TF2_SHIFT + 1)
+						/*!< Total number of bits
+						in table->flags. */
 /* @} */
 
+
 /**********************************************************************//**
 Creates a table memory object.
 @return	own: table object */
@@ -374,7 +392,7 @@ struct dict_table_struct{
 	unsigned	space:32;
 				/*!< space where the clustered index of the
 				table is placed */
-	unsigned	flags:DICT_TF_BITS;/*!< DICT_TF_COMPACT, ... */
+	unsigned	flags:DICT_TF2_BITS;/*!< DICT_TF_COMPACT, ... */
 	unsigned	ibd_file_missing:1;
 				/*!< TRUE if this is in a single-table
 				tablespace and the .ibd file is missing; then

storage/innodb_plugin/include/fil0fil.h

 /*****************************************************************************
 
-Copyright (c) 1995, 2009, Innobase Oy. All Rights Reserved.
+Copyright (c) 1995, 2010, 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
@@ -110,9 +110,10 @@ extern fil_addr_t	fil_addr_null;
 					contents of this field is valid
 					for all uncompressed pages. */
 #define FIL_PAGE_FILE_FLUSH_LSN	26	/*!< this is only defined for the
-					first page in a data file: the file
-					has been flushed to disk at least up
-					to this lsn */
+					first page in a system tablespace
+					data file (ibdata*, not *.ibd):
+					the file has been flushed to disk
+					at least up to this lsn */
 #define FIL_PAGE_ARCH_LOG_NO_OR_SPACE_ID  34 /*!< starting from 4.1.x this
 					contains the space id of the page */
 #define FIL_PAGE_DATA		38	/*!< start of the data on the page */

storage/innodb_plugin/include/hash0hash.h

@@ -434,11 +434,12 @@ struct hash_table_struct {
 				these heaps */
 #endif /* !UNIV_HOTBACKUP */
 	mem_heap_t*	heap;
+#ifdef UNIV_DEBUG
 	ulint		magic_n;
+# define HASH_TABLE_MAGIC_N	76561114
+#endif /* UNIV_DEBUG */
 };
 
-#define HASH_TABLE_MAGIC_N	76561114
-
 #ifndef UNIV_NONINL
 #include "hash0hash.ic"
 #endif

storage/innodb_plugin/include/hash0hash.ic

@@ -35,6 +35,8 @@ hash_get_nth_cell(
 	hash_table_t*	table,	/*!< in: hash table */
 	ulint		n)	/*!< in: cell index */
 {
+	ut_ad(table);
+	ut_ad(table->magic_n == HASH_TABLE_MAGIC_N);
 	ut_ad(n < table->n_cells);
 
 	return(table->array + n);
@@ -48,6 +50,8 @@ hash_table_clear(
 /*=============*/
 	hash_table_t*	table)	/*!< in/out: hash table */
 {
+	ut_ad(table);
+	ut_ad(table->magic_n == HASH_TABLE_MAGIC_N);
 	memset(table->array, 0x0,
 	       table->n_cells * sizeof(*table->array));
 }
@@ -61,6 +65,8 @@ hash_get_n_cells(
 /*=============*/
 	hash_table_t*	table)	/*!< in: table */
 {
+	ut_ad(table);
+	ut_ad(table->magic_n == HASH_TABLE_MAGIC_N);
 	return(table->n_cells);
 }
 
@@ -74,6 +80,8 @@ hash_calc_hash(
 	ulint		fold,	/*!< in: folded value */
 	hash_table_t*	table)	/*!< in: hash table */
 {
+	ut_ad(table);
+	ut_ad(table->magic_n == HASH_TABLE_MAGIC_N);
 	return(ut_hash_ulint(fold, table->n_cells));
 }
 
@@ -88,6 +96,8 @@ hash_get_mutex_no(
 	hash_table_t*	table,	/*!< in: hash table */
 	ulint		fold)	/*!< in: fold */
 {
+	ut_ad(table);
+	ut_ad(table->magic_n == HASH_TABLE_MAGIC_N);
 	ut_ad(ut_is_2pow(table->n_mutexes));
 	return(ut_2pow_remainder(hash_calc_hash(fold, table),
 				 table->n_mutexes));
@@ -103,6 +113,8 @@ hash_get_nth_heap(
 	hash_table_t*	table,	/*!< in: hash table */
 	ulint		i)	/*!< in: index of the heap */
 {
+	ut_ad(table);
+	ut_ad(table->magic_n == HASH_TABLE_MAGIC_N);
 	ut_ad(i < table->n_mutexes);
 
 	return(table->heaps[i]);
@@ -120,6 +132,9 @@ hash_get_heap(
 {
 	ulint	i;
 
+	ut_ad(table);
+	ut_ad(table->magic_n == HASH_TABLE_MAGIC_N);
+
 	if (table->heap) {
 		return(table->heap);
 	}
@@ -139,6 +154,8 @@ hash_get_nth_mutex(
 	hash_table_t*	table,	/*!< in: hash table */
 	ulint		i)	/*!< in: index of the mutex */
 {
+	ut_ad(table);
+	ut_ad(table->magic_n == HASH_TABLE_MAGIC_N);
 	ut_ad(i < table->n_mutexes);
 
 	return(table->mutexes + i);
@@ -156,6 +173,9 @@ hash_get_mutex(
 {
 	ulint	i;
 
+	ut_ad(table);
+	ut_ad(table->magic_n == HASH_TABLE_MAGIC_N);
+
 	i = hash_get_mutex_no(table, fold);
 
 	return(hash_get_nth_mutex(table, i));

storage/innodb_plugin/include/lock0lock.h

@@ -613,13 +613,16 @@ lock_rec_print(
 	FILE*		file,	/*!< in: file where to print */
 	const lock_t*	lock);	/*!< in: record type lock */
 /*********************************************************************//**
-Prints info of locks for all transactions. */
+Prints info of locks for all transactions.
+@return FALSE if not able to obtain kernel mutex
+and exits without printing info */
 UNIV_INTERN
-void
+ibool
 lock_print_info_summary(
 /*====================*/
-	FILE*	file);	/*!< in: file where to print */
-/*********************************************************************//**
+	FILE*	file,	/*!< in: file where to print */
+	ibool   nowait);/*!< in: whether to wait for the kernel mutex */
+/*************************************************************************
 Prints info of locks for each transaction. */
 UNIV_INTERN
 void

storage/innodb_plugin/include/log0log.h

 /*****************************************************************************
 
-Copyright (c) 1995, 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
-
-*****************************************************************************/
-/*****************************************************************************
-
-Copyright (c) 1995, 2009, Innobase Oy. All Rights Reserved.
+Copyright (c) 1995, 2010, Innobase Oy. All Rights Reserved.
 Copyright (c) 2009, Google Inc.
 
 Portions of this file contain modifications contributed and copyrighted by
@@ -825,7 +808,17 @@ struct log_struct{
 					written to some log group; for this to
 					be advanced, it is enough that the
 					write i/o has been completed for all
-					log groups */
+					log groups.
+					Note that since InnoDB currently
+					has only one log group therefore
+					this value is redundant. Also it
+					is possible that this value
+					falls behind the
+					flushed_to_disk_lsn transiently.
+					It is appropriate to use either
+					flushed_to_disk_lsn or
+					write_lsn which are always
+					up-to-date and accurate. */
 	ib_uint64_t	write_lsn;	/*!< end lsn for the current running
 					write */
 	ulint		write_end_offset;/*!< the data in buffer has

storage/innodb_plugin/include/log0log.ic

 /*****************************************************************************
 
-Copyright (c) 1995, 2009, Innobase Oy. All Rights Reserved.
+Copyright (c) 1995, 2010, 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
@@ -314,12 +314,15 @@ log_reserve_and_write_fast(
 	ulint		data_len;
 #ifdef UNIV_LOG_LSN_DEBUG
 	/* length of the LSN pseudo-record */
-	ulint		lsn_len = 1
-		+ mach_get_compressed_size(log_sys->lsn >> 32)
-		+ mach_get_compressed_size(log_sys->lsn & 0xFFFFFFFFUL);
+	ulint		lsn_len;
 #endif /* UNIV_LOG_LSN_DEBUG */
 
 	mutex_enter(&log_sys->mutex);
+#ifdef UNIV_LOG_LSN_DEBUG
+	lsn_len = 1
+		+ mach_get_compressed_size(log_sys->lsn >> 32)
+		+ mach_get_compressed_size(log_sys->lsn & 0xFFFFFFFFUL);
+#endif /* UNIV_LOG_LSN_DEBUG */
 
 	data_len = len
 #ifdef UNIV_LOG_LSN_DEBUG

storage/innodb_plugin/include/log0recv.h

 /*****************************************************************************
 
-Copyright (c) 1997, 2009, Innobase Oy. All Rights Reserved.
+Copyright (c) 1997, 2010, 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
@@ -176,6 +176,12 @@ UNIV_INTERN
 void
 recv_recovery_from_checkpoint_finish(void);
 /*======================================*/
+/********************************************************//**
+Initiates the rollback of active transactions. */
+UNIV_INTERN
+void
+recv_recovery_rollback_active(void);
+/*===============================*/
 /*******************************************************//**
 Scans log from a buffer and stores new log data to the parsing buffer.
 Parses and hashes the log records if new data found.  Unless
@@ -258,12 +264,14 @@ void
 recv_sys_init(
 /*==========*/
 	ulint	available_memory);	/*!< in: available memory in bytes */
+#ifndef UNIV_HOTBACKUP
 /********************************************************//**
 Reset the state of the recovery system variables. */
 UNIV_INTERN
 void
 recv_sys_var_init(void);
 /*===================*/
+#endif /* !UNIV_HOTBACKUP */
 /*******************************************************************//**
 Empties the hash table of stored log records, applying them to appropriate
 pages. */

storage/innodb_plugin/include/mem0dbg.h

 /*****************************************************************************
 
-Copyright (c) 1994, 2009, Innobase Oy. All Rights Reserved.
+Copyright (c) 1994, 2010, 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
@@ -28,6 +28,13 @@ Created 6/9/1994 Heikki Tuuri
 check fields whose sizes are given below */
 
 #ifdef UNIV_MEM_DEBUG
+# ifndef UNIV_HOTBACKUP
+/* The mutex which protects in the debug version the hash table
+containing the list of live memory heaps, and also the global
+variables in mem0dbg.c. */
+extern mutex_t	mem_hash_mutex;
+# endif /* !UNIV_HOTBACKUP */
+
 #define MEM_FIELD_HEADER_SIZE	ut_calc_align(2 * sizeof(ulint),\
 						UNIV_MEM_ALIGNMENT)
 #define MEM_FIELD_TRAILER_SIZE	sizeof(ulint)

storage/innodb_plugin/include/mem0dbg.ic

 /*****************************************************************************
 
-Copyright (c) 1994, 2009, Innobase Oy. All Rights Reserved.
+Copyright (c) 1994, 2010, 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
@@ -25,9 +25,6 @@ Created 6/8/1994 Heikki Tuuri
 *************************************************************************/
 
 #ifdef UNIV_MEM_DEBUG
-# ifndef UNIV_HOTBACKUP
-extern mutex_t	mem_hash_mutex;
-# endif /* !UNIV_HOTBACKUP */
 extern ulint	mem_current_allocated_memory;
 
 /******************************************************************//**

storage/innodb_plugin/include/mem0mem.h

 /*****************************************************************************
 
-Copyright (c) 1994, 2009, Innobase Oy. All Rights Reserved.
+Copyright (c) 1994, 2010, 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
@@ -359,6 +359,9 @@ struct mem_block_info_struct {
 			to the heap is also the first block in this list,
 			though it also contains the base node of the list. */
 	ulint	len;	/*!< physical length of this block in bytes */
+	ulint	total_size; /* physical length in bytes of all blocks
+			in the heap. This is defined only in the base
+			node and is set to ULINT_UNDEFINED in others. */
 	ulint	type;	/*!< type of heap: MEM_HEAP_DYNAMIC, or
 			MEM_HEAP_BUF possibly ORed to MEM_HEAP_BTR_SEARCH */
 	ulint	free;	/*!< offset in bytes of the first free position for

storage/innodb_plugin/include/mem0mem.ic

 /*****************************************************************************
 
-Copyright (c) 1994, 2009, Innobase Oy. All Rights Reserved.
+Copyright (c) 1994, 2010, 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
@@ -579,18 +579,12 @@ mem_heap_get_size(
 /*==============*/
 	mem_heap_t*	heap)	/*!< in: heap */
 {
-	mem_block_t*	block;
 	ulint		size	= 0;
 
 	ut_ad(mem_heap_check(heap));
 
-	block = heap;
-
-	while (block != NULL) {
+	size = heap->total_size;
 
-		size += mem_block_get_len(block);
-		block = UT_LIST_GET_NEXT(list, block);
-	}
 #ifndef UNIV_HOTBACKUP
 	if (heap->free_block) {
 		size += UNIV_PAGE_SIZE;

storage/innodb_plugin/include/mtr0mtr.ic

 /*****************************************************************************
 
-Copyright (c) 1995, 2009, Innobase Oy. All Rights Reserved.
+Copyright (c) 1995, 2010, 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
@@ -70,6 +70,7 @@ mtr_memo_push(
 	ut_ad(type <= MTR_MEMO_X_LOCK);
 	ut_ad(mtr);
 	ut_ad(mtr->magic_n == MTR_MAGIC_N);
+	ut_ad(mtr->state == MTR_ACTIVE);
 
 	memo = &(mtr->memo);
 
@@ -92,6 +93,7 @@ mtr_set_savepoint(
 
 	ut_ad(mtr);
 	ut_ad(mtr->magic_n == MTR_MAGIC_N);
+	ut_ad(mtr->state == MTR_ACTIVE);
 
 	memo = &(mtr->memo);
 
@@ -149,6 +151,7 @@ mtr_memo_contains(
 
 	ut_ad(mtr);
 	ut_ad(mtr->magic_n == MTR_MAGIC_N);
+	ut_ad(mtr->state == MTR_ACTIVE || mtr->state == MTR_COMMITTING);
 
 	memo = &(mtr->memo);
 

storage/innodb_plugin/include/os0file.h

-/*****************************************************************************
-
-Copyright (c) 1995, 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
-
-*****************************************************************************/
 /***********************************************************************
 
-Copyright (c) 1995, 2009, Innobase Oy. All Rights Reserved.
+Copyright (c) 1995, 2010, Innobase Oy. All Rights Reserved.
 Copyright (c) 2009, Percona Inc.
 
 Portions of this file contain modifications contributed and copyrighted

storage/innodb_plugin/include/que0que.ic

 /*****************************************************************************
 
-Copyright (c) 1996, 2009, Innobase Oy. All Rights Reserved.
+Copyright (c) 1996, 2010, 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
@@ -38,6 +38,20 @@ thr_get_trx(
 	return(thr->graph->trx);
 }
 
+/*******************************************************************//**
+Determines if this thread is rolling back an incomplete transaction
+in crash recovery.
+@return TRUE if thr is rolling back an incomplete transaction in crash
+recovery */
+UNIV_INLINE
+ibool
+thr_is_recv(
+/*========*/
+	const que_thr_t*	thr)	/*!< in: query thread */
+{
+	return(trx_is_recv(thr->graph->trx));
+}
+
 /***********************************************************************//**
 Gets the first thr in a fork. */
 UNIV_INLINE