Merge of XtraDB for 5.1.59

storage/xtradb/ChangeLog

+2011-08-08	The InnoDB Team
+
+	* row/row0sel.c:
+	Fix Bug#12835650 VARCHAR maximum length performance impact
+
+2011-08-08	The InnoDB Team
+
+	* handler/ha_innodb.cc:
+	Fix Bug#12770537 I_S.TABLES.DATA_LENGTH DOES NOT SHOW ON-DISK SIZE
+	FOR COMPRESSED INNODB
+
+2011-07-19	The InnoDB Team
+
+	* buf/buf0buf.c, buf/buf0rea.c, handler/ha_innodb.cc,
+	include/buf0buf.h, include/buf0buf.ic, include/srv0srv.h,
+	srv/srv0srv.c:
+	Fix Bug#12356373 by reintroducing random readahead
+
+2011-06-30	The InnoDB Team
+
+	* row/row0row.c:
+	Fix Bug#12637786 Wrong secondary index entries on CHAR and VARCHAR
+	columns in ROW_FORMAT=DYNAMIC and ROW_FORMAT=COMPRESSED
+
+2011-06-16	The InnoDB Team
+
+	* btr/btr0cur.c, buf/buf0buddy.c, buf/buf0buf.c, buf/buf0lru.c,
+	include/buf0buddy.h, include/buf0buddy.ic, include/buf0buf.h,
+	include/buf0buf.ic, include/buf0lru.h, include/buf0types.h:
+	Fix Bug#61188 DROP TABLE extremely slow
+
+2011-06-16	The InnoDB Team
+
+	* buf/buf0buddy.c, buf/buf0buf.c, buf/buf0flu.c, buf/buf0lru.c,
+	include/buf0buf.h, include/buf0lru.h:
+	Fix Bug#61341 buf_LRU_insert_zip_clean can be O(N) on LRU length
+
+2011-06-16	The InnoDB Team
+
+	* page/page0zip.c, rem/rem0rec.c:
+	Fix Bug#61191 question about page_zip_available()
+
+2011-06-16	The InnoDB Team
+
+	* btr/btr0btr.c, btr/btr0cur.c, include/btr0btr.h, include/btr0cur.h,
+	include/btr0cur.ic, include/buf0buf.h, include/buf0buf.ic,
+	include/page0cur.ic, include/page0page.h, include/page0page.ic,
+	include/sync0rw.ic, include/sync0sync.h, page/page0cur.c,
+	page/page0page.c, row/row0ins.c, row/row0upd.c,
+	sync/sync0rw.c, sync/sync0sync.c:
+	Fix Bug#12612184 Race condition after btr_cur_pessimistic_update()
+
+2011-06-09	The InnoDB Team
+	* btr/btr0cur.c, include/rem0rec.h, include/rem0rec.ic,
+	* row/row0row.c, row/row0vers.c, trx/trx0rec.c:
+	Instrumentation for Bug#12612184 Race condition in row_upd_clust_rec()
+
+2011-05-19	The InnoDB Team
+
+	* row/row0row.c:
+	Fix Bug#12429576 Assertion failure on purge of column prefix index
+
+2011-04-07	The InnoDB Team
+
+	* handler/ha_innodb.cc, handler/ha_innodb.h, handler/handler0alter.cc:
+	Fix Bug #52409 Assertion failure: long semaphore wait
+
+2011-04-07	The InnoDB Team
+
+	* handler/ha_innodb.cc, include/trx0trx.h, include/trx0undo.h,
+	log/log0log.c, trx/trx0sys.c, trx/trx0trx.c, trx/trx0undo.c:
+	Fix Bug #59641 Prepared XA transaction in system after hard crash
+	causes future shutdown hang
+
+2011-03-30	The InnoDB Team
+
+	* srv/srv0srv.c, sync/sync0arr.h, sync/sync0arr.c:
+	Fix Bug#11877216 InnoDB too eager to commit suicide on a busy server
+
+2011-03-15	The InnoDB Team
+
+	* btr/btr0cur.c, page/page0zip.c:
+	Fix Bug#11849231 inflateInit() invoked without initializing all memory
+
+2011-02-28	The InnoDB Team
+
+	* btr/btr0sea.c, buf/buf0buf.c, buf/buf0lru.c:
+	Fix Bug#58549 Race condition in buf_LRU_drop_page_hash_for_tablespace()
+	and compressed tables
+
+2011-02-15	The InnoDB Team
+
+	* sync/sync0rw.c, innodb_bug59307.test:
+	Bug#59307 Valgrind: uninitialized value in
+	rw_lock_set_writer_id_and_recursion_flag()
+
+2011-02-14	The InnoDB Team
+
+	* handler/handler0alter.cc:
+	Bug#59749 Enabling concurrent reads while creating non-primary
+	unique index gives failures
+
 2011-01-31	The InnoDB Team
 
 	* btr/btr0cur.c, include/row0upd.h,

storage/xtradb/btr/btr0pcur.c

 /*****************************************************************************
 
-Copyright (c) 1996, 2010, Innobase Oy. All Rights Reserved.
+Copyright (c) 1996, 2011, Oracle and/or its affiliates. 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
@@ -362,33 +362,6 @@ btr_pcur_restore_position_func(
 	return(FALSE);
 }
 
-/**************************************************************//**
-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.
-NOTE! In the case of BTR_LEAF_MODIFY, there should not exist changes
-made by the current mini-transaction to the data protected by the
-cursor latch, as then the latch must not be released until mtr_commit. */
-UNIV_INTERN
-void
-btr_pcur_release_leaf(
-/*==================*/
-	btr_pcur_t*	cursor, /*!< in: persistent cursor */
-	mtr_t*		mtr)	/*!< in: mtr */
-{
-	buf_block_t*	block;
-
-	ut_a(cursor->pos_state == BTR_PCUR_IS_POSITIONED);
-	ut_ad(cursor->latch_mode != BTR_NO_LATCHES);
-
-	block = btr_pcur_get_block(cursor);
-
-	btr_leaf_page_release(block, cursor->latch_mode, mtr);
-
-	cursor->latch_mode = BTR_NO_LATCHES;
-
-	cursor->pos_state = BTR_PCUR_WAS_POSITIONED;
-}
-
 /*********************************************************//**
 Moves the persistent cursor to the first record on the next page. Releases the
 latch on the current page, and bufferunfixes it. Note that there must not be

storage/xtradb/btr/btr0sea.c

@@ -1373,8 +1373,8 @@ btr_search_drop_page_hash_when_freed(
 	having to fear a deadlock. */
 
 	block = buf_page_get_gen(space, zip_size, page_no, RW_S_LATCH, NULL,
-				BUF_GET_IF_IN_POOL, __FILE__, __LINE__,
-				&mtr);
+				 BUF_PEEK_IF_IN_POOL, __FILE__, __LINE__,
+				 &mtr);
 	/* Because the buffer pool mutex was released by
 	buf_page_peek_if_search_hashed(), it is possible that the
 	block was removed from the buffer pool by another thread

storage/xtradb/buf/buf0flu.c

 /*****************************************************************************
 
-Copyright (c) 1995, 2010, Innobase Oy. All Rights Reserved.
+Copyright (c) 1995, 2011, Oracle and/or its affiliates. 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
@@ -459,7 +459,9 @@ buf_flush_remove(
 	case BUF_BLOCK_ZIP_DIRTY:
 		buf_page_set_state(bpage, BUF_BLOCK_ZIP_PAGE);
 		UT_LIST_REMOVE(flush_list, buf_pool->flush_list, bpage);
+#if defined UNIV_DEBUG || defined UNIV_BUF_DEBUG
 		buf_LRU_insert_zip_clean(bpage);
+#endif /* UNIV_DEBUG || UNIV_BUF_DEBUG */
 		break;
 	case BUF_BLOCK_FILE_PAGE:
 		UT_LIST_REMOVE(flush_list, buf_pool->flush_list, bpage);
@@ -769,7 +771,7 @@ corrupted_page:
 flush:
 	/* Now flush the doublewrite buffer data to disk */
 
-	fil_flush(srv_doublewrite_file ? TRX_DOUBLEWRITE_SPACE : TRX_SYS_SPACE);
+	fil_flush(srv_doublewrite_file ? TRX_DOUBLEWRITE_SPACE : TRX_SYS_SPACE, FALSE);
 
 	/* We know that the writes have been flushed to disk now
 	and in recovery we will find them in the doublewrite buffer
@@ -1084,7 +1086,7 @@ buf_flush_page_try(
 /*===============*/
 	buf_block_t*	block)		/*!< in/out: buffer control block */
 {
-	ut_ad(buf_pool_mutex_own());
+	//ut_ad(buf_pool_mutex_own());
 	ut_ad(buf_block_get_state(block) == BUF_BLOCK_FILE_PAGE);
 	ut_ad(mutex_own(&block->mutex));
 
@@ -1092,8 +1094,11 @@ buf_flush_page_try(
 		return(FALSE);
 	}
 
+	buf_pool_mutex_enter();
+
 	if (buf_pool->n_flush[BUF_FLUSH_LRU] > 0
 	    || buf_pool->init_flush[BUF_FLUSH_LRU]) {
+		buf_pool_mutex_exit();
 		/* There is already a flush batch of the same type running */
 		return(FALSE);
 	}

storage/xtradb/buf/buf0rea.c

@@ -38,6 +38,14 @@ Created 11/5/1995 Heikki Tuuri
 #include "srv0start.h"
 #include "srv0srv.h"
 
+/** The size in blocks of the area where the random read-ahead algorithm counts
+the accessed pages when deciding whether to read-ahead */
+#define	BUF_READ_AHEAD_RANDOM_AREA	BUF_READ_AHEAD_AREA
+
+/** There must be at least this many pages in buf_pool in the area to start
+a random read-ahead */
+#define BUF_READ_AHEAD_RANDOM_THRESHOLD	(5 + BUF_READ_AHEAD_RANDOM_AREA / 8)
+
 /** The linear read-ahead area size */
 #define	BUF_READ_AHEAD_LINEAR_AREA	BUF_READ_AHEAD_AREA
 
@@ -201,12 +209,178 @@ not_to_recover:
 	if (sync) {
 		/* The i/o is already completed when we arrive from
 		fil_read */
-		buf_page_io_complete(bpage, trx);
+		buf_page_io_complete(bpage);
 	}
 
 	return(1);
 }
 
+/********************************************************************//**
+Applies a random read-ahead in buf_pool if there are at least a threshold
+value of accessed pages from the random read-ahead area. Does not read any
+page, not even the one at the position (space, offset), if the read-ahead
+mechanism is not activated. NOTE 1: the calling thread may own latches on
+pages: to avoid deadlocks this function must be written such that it cannot
+end up waiting for these latches! NOTE 2: the calling thread must want
+access to the page given: this rule is set to prevent unintended read-aheads
+performed by ibuf routines, a situation which could result in a deadlock if
+the OS does not support asynchronous i/o.
+@return number of page read requests issued; NOTE that if we read ibuf
+pages, it may happen that the page at the given page number does not
+get read even if we return a positive value! */
+static
+ulint
+buf_read_ahead_random(
+/*==================*/
+	ulint	space,	/*!< in: space id */
+	ulint	zip_size,/*!< in: compressed page size in bytes, or 0 */
+	ulint	offset,	/*!< in: page number of a page which the current thread
+			wants to access */
+	trx_t*	trx)
+{
+	ib_int64_t	tablespace_version;
+	ulint		recent_blocks	= 0;
+	ulint		count;
+	ulint		ibuf_mode;
+	ulint		low, high;
+	ulint		err;
+	ulint		i;
+	ulint		buf_read_ahead_random_area;
+
+	if (!srv_random_read_ahead) {
+		/* Disabled by user */
+		return(0);
+	}
+
+	if (srv_startup_is_before_trx_rollback_phase) {
+		/* No read-ahead to avoid thread deadlocks */
+		return(0);
+	}
+
+	if (ibuf_bitmap_page(zip_size, offset)
+	    || trx_sys_hdr_page(space, offset)) {
+
+		/* If it is an ibuf bitmap page or trx sys hdr, we do
+		no read-ahead, as that could break the ibuf page access
+		order */
+
+		return(0);
+	}
+
+	/* Remember the tablespace version before we ask the tablespace size
+	below: if DISCARD + IMPORT changes the actual .ibd file meanwhile, we
+	do not try to read outside the bounds of the tablespace! */
+
+	tablespace_version = fil_space_get_version(space);
+
+	buf_read_ahead_random_area = BUF_READ_AHEAD_RANDOM_AREA;
+
+	low  = (offset / buf_read_ahead_random_area)
+		* buf_read_ahead_random_area;
+	high = (offset / buf_read_ahead_random_area + 1)
+		* buf_read_ahead_random_area;
+	if (high > fil_space_get_size(space)) {
+
+		high = fil_space_get_size(space);
+	}
+
+	//buf_pool_mutex_enter();
+	mutex_enter(&buf_pool_mutex);
+
+	if (buf_pool->n_pend_reads
+	    > buf_pool->curr_size / BUF_READ_AHEAD_PEND_LIMIT) {
+		//buf_pool_mutex_exit();
+		mutex_exit(&buf_pool_mutex);
+
+		return(0);
+	}
+	mutex_exit(&buf_pool_mutex);
+
+	/* Count how many blocks in the area have been recently accessed,
+	that is, reside near the start of the LRU list. */
+
+	rw_lock_s_lock(&page_hash_latch);
+	for (i = low; i < high; i++) {
+		const buf_page_t*	bpage = buf_page_hash_get(space, i);
+
+		if (bpage
+		    && buf_page_is_accessed(bpage)
+		    && buf_page_peek_if_young(bpage)) {
+
+			recent_blocks++;
+
+			if (recent_blocks >= BUF_READ_AHEAD_RANDOM_THRESHOLD) {
+
+				//buf_pool_mutex_exit();
+				rw_lock_s_unlock(&page_hash_latch);
+				goto read_ahead;
+			}
+		}
+	}
+
+	//buf_pool_mutex_exit();
+	rw_lock_s_unlock(&page_hash_latch);
+	/* Do nothing */
+	return(0);
+
+read_ahead:
+	/* Read all the suitable blocks within the area */
+
+	if (ibuf_inside()) {
+		ibuf_mode = BUF_READ_IBUF_PAGES_ONLY;
+	} else {
+		ibuf_mode = BUF_READ_ANY_PAGE;
+	}
+
+	count = 0;
+
+	for (i = low; i < high; i++) {
+		/* It is only sensible to do read-ahead in the non-sync aio
+		mode: hence FALSE as the first parameter */
+
+		if (!ibuf_bitmap_page(zip_size, i)) {
+			count += buf_read_page_low(
+				&err, FALSE,
+				ibuf_mode | OS_AIO_SIMULATED_WAKE_LATER,
+				space, zip_size, FALSE,
+				tablespace_version, i, trx);
+			if (err == DB_TABLESPACE_DELETED) {
+				ut_print_timestamp(stderr);
+				fprintf(stderr,
+					"  InnoDB: Warning: in random"
+					" readahead trying to access\n"
+					"InnoDB: tablespace %lu page %lu,\n"
+					"InnoDB: but the tablespace does not"
+					" exist or is just being dropped.\n",
+					(ulong) space, (ulong) i);
+			}
+		}
+	}
+
+	/* In simulated aio we wake the aio handler threads only after
+	queuing all aio requests, in native aio the following call does
+	nothing: */
+
+	os_aio_simulated_wake_handler_threads();
+
+#ifdef UNIV_DEBUG
+	if (buf_debug_prints && (count > 0)) {
+		fprintf(stderr,
+			"Random read-ahead space %lu offset %lu pages %lu\n",
+			(ulong) space, (ulong) offset,
+			(ulong) count);
+	}
+#endif /* UNIV_DEBUG */
+
+	/* Read ahead is considered one I/O operation for the purpose of
+	LRU policy decision. */
+	buf_LRU_stat_inc_io();
+
+	buf_pool->stat.n_ra_pages_read_rnd += count;
+	return(count);
+}
+
+
 /********************************************************************//**
 High-level function which reads a page asynchronously from a file to the
 buffer buf_pool if it is not already there. Sets the io_fix flag and sets
@@ -226,6 +400,9 @@ buf_read_page(
 	ulint		count;
 	ulint		err;
 
+	count = buf_read_ahead_random(space, zip_size, offset, trx);
+	srv_buf_pool_reads += count;
+
 	tablespace_version = fil_space_get_version(space);
 
 	/* We do the i/o in the synchronous aio mode to save thread

storage/xtradb/dict/dict0dict.c

@@ -4524,6 +4524,8 @@ dict_store_statistics(
 			break;
 		}
 
+		btr_pcur_store_position(&pcur, &mtr);
+
 		if (rec_get_deleted_flag(rec, 0)) {
 			/* don't count */
 			i--;
@@ -4564,6 +4566,10 @@ dict_store_statistics(
 		rests--;
 
 next_rec:
+		mtr_commit(&mtr);
+		mtr_start(&mtr);
+		btr_pcur_restore_position(BTR_MODIFY_LEAF, &pcur, &mtr);
+
 		btr_pcur_move_to_next_user_rec(&pcur, &mtr);
 	}
 	btr_pcur_close(&pcur);
@@ -4654,6 +4660,7 @@ dict_update_statistics(
 	do {
 		if (table->is_corrupt) {
 			ut_a(srv_pass_corrupt_table);
+			dict_table_stats_unlock(table, RW_X_LATCH);
 			return;
 		}
 

storage/xtradb/dict/dict0load.c

@@ -554,9 +554,10 @@ dict_load_columns(
 }
 
 /********************************************************************//**
-Loads definitions for index fields. */
+Loads definitions for index fields.
+@return DB_SUCCESS if ok, DB_CORRUPTION if failed */
 static
-void
+ulint
 dict_load_fields(
 /*=============*/
 	dict_index_t*	index,	/*!< in: index whose fields to load */
@@ -575,6 +576,7 @@ dict_load_fields(
 	byte*		buf;
 	ulint		i;
 	mtr_t		mtr;
+	ulint		error = DB_SUCCESS;
 
 	ut_ad(mutex_own(&(dict_sys->mutex)));
 
@@ -641,6 +643,26 @@ dict_load_fields(
 
 		field = rec_get_nth_field_old(rec, 4, &len);
 
+		if (prefix_len >= DICT_MAX_INDEX_COL_LEN) {
+			fprintf(stderr, "InnoDB: Error: load index"
+					" '%s' failed.\n"
+					"InnoDB: index field '%s' has a prefix"
+					" length of %lu bytes,\n"
+					"InnoDB: which exceeds the"
+					" maximum limit of %lu bytes.\n"
+					"InnoDB: Please use server that"
+					" supports long index prefix\n"
+					"InnoDB: or turn on"
+					" innodb_force_recovery to load"
+					" the table\n",
+				index->name, mem_heap_strdupl(
+						heap, (char*) field, len),
+		    		(ulong) prefix_len,
+				(ulong) (DICT_MAX_INDEX_COL_LEN - 1));
+			error = DB_CORRUPTION;
+			goto func_exit;
+		}
+
 		dict_mem_index_add_field(index,
 					 mem_heap_strdupl(heap,
 							  (char*) field, len),
@@ -650,8 +672,10 @@ next_rec:
 		btr_pcur_move_to_next_user_rec(&pcur, &mtr);
 	}
 
+func_exit:
 	btr_pcur_close(&pcur);
 	mtr_commit(&mtr);
+	return(error);
 }
 
 /********************************************************************//**
@@ -802,7 +826,25 @@ dict_load_indexes(
 						      space, type, n_fields);
 			index->id = id;
 
-			dict_load_fields(index, heap);
+			error = dict_load_fields(index, heap);
+
+			if (error != DB_SUCCESS) {
+				fprintf(stderr, "InnoDB: Error: load index '%s'"
+					" for table '%s' failed\n",
+					index->name, table->name);
+
+				/* If the force recovery flag is set, and
+				if the failed index is not the primary index, we
+				will continue and open other indexes */
+				if (srv_force_recovery
+				    && !(index->type & DICT_CLUSTERED)) {
+					error = DB_SUCCESS;
+					goto next_rec;
+				} else {
+					goto func_exit;
+				}
+			}
+
 			error = dict_index_add_to_cache(table, index, page_no,
 							FALSE);
 			/* The data dictionary tables should never contain
@@ -1028,9 +1070,18 @@ err_exit:
 		} else {
 			table->fk_max_recusive_level = 0;
 		}
-	} else if (!srv_force_recovery) {
-		dict_table_remove_from_cache(table);
-		table = NULL;
+	} else {
+		dict_index_t*	index;
+
+		/* Make sure that at least the clustered index was loaded.
+		Otherwise refuse to load the table */
+		index = dict_table_get_first_index(table);
+
+		if (!srv_force_recovery || !index
+		     || !(index->type & DICT_CLUSTERED)) {
+			dict_table_remove_from_cache(table);
+			table = NULL;
+		}
 	}
 #if 0
 	if (err != DB_SUCCESS && table != NULL) {

storage/xtradb/dict/dict0mem.c

@@ -36,6 +36,9 @@ Created 1/8/1996 Heikki Tuuri
 #ifndef UNIV_HOTBACKUP
 # include "lock0lock.h"
 #endif /* !UNIV_HOTBACKUP */
+#ifdef UNIV_BLOB_DEBUG
+# include "ut0rbt.h"
+#endif /* UNIV_BLOB_DEBUG */
 
 #define	DICT_HEAP_SIZE		100	/*!< initial memory heap size when
 					creating a table or index object */
@@ -318,6 +321,12 @@ dict_mem_index_free(
 {
 	ut_ad(index);
 	ut_ad(index->magic_n == DICT_INDEX_MAGIC_N);
+#ifdef UNIV_BLOB_DEBUG
+	if (index->blobs) {
+		mutex_free(&index->blobs_mutex);
+		rbt_free(index->blobs);
+	}
+#endif /* UNIV_BLOB_DEBUG */
 
 	mem_heap_free(index->heap);
 }

storage/xtradb/ha/hash0hash.c

@@ -127,70 +127,6 @@ hash_create(
 	return(table);
 }
 
-/*************************************************************//**
-*/
-UNIV_INTERN
-ulint
-hash_create_needed(
-/*===============*/
-	ulint	n)
-{
-	ulint	prime;
-	ulint	offset;
-
-	prime = ut_find_prime(n);
-
-	offset = (sizeof(hash_table_t) + 7) / 8;
-	offset *= 8;
-
-	return(offset + sizeof(hash_cell_t) * prime);
-}
-
-UNIV_INTERN
-void
-hash_create_init(
-/*=============*/
-	hash_table_t*	table,
-	ulint		n)
-{
-	ulint	prime;
-	ulint	offset;
-
-	prime = ut_find_prime(n);
-
-	offset = (sizeof(hash_table_t) + 7) / 8;
-	offset *= 8;
-
-	table->array = (hash_cell_t*)(((byte*)table) + offset);
-	table->n_cells = prime;
-# if defined UNIV_AHI_DEBUG || defined UNIV_DEBUG
-	table->adaptive = FALSE;
-# endif /* UNIV_AHI_DEBUG || UNIV_DEBUG */
-	table->n_mutexes = 0;
-	table->mutexes = NULL;
-	table->heaps = NULL;
-	table->heap = NULL;
-	ut_d(table->magic_n = HASH_TABLE_MAGIC_N);
-
-	/* Initialize the cell array */
-	hash_table_clear(table);
-}
-
-UNIV_INTERN
-void
-hash_create_reuse(
-/*==============*/
-	hash_table_t*	table)
-{
-	ulint	offset;
-
-	offset = (sizeof(hash_table_t) + 7) / 8;
-	offset *= 8;
-
-	table->array = (hash_cell_t*)(((byte*)table) + offset);
-	ut_ad(table->magic_n == HASH_TABLE_MAGIC_N);
-}
-
 /*************************************************************//**
 Frees a hash table. */
 UNIV_INTERN

storage/xtradb/handler/ha_innodb.h

@@ -319,15 +319,14 @@ innobase_trx_allocate(
 This function checks each index name for a table against reserved
 system default primary index name 'GEN_CLUST_INDEX'. If a name
 matches, this function pushes an warning message to the client,
-and returns true. */
+and returns true.
+@return true if the index name matches the reserved name */
 extern "C"
 bool
 innobase_index_name_is_reserved(
 /*============================*/
-					/* out: true if the index name
-					matches the reserved name */
-	const trx_t*	trx,		/* in: InnoDB transaction handle */
-	const KEY*	key_info,	/* in: Indexes to be created */
-	ulint		num_of_keys);	/* in: Number of indexes to
+	THD*		thd,		/*!< in/out: MySQL connection */
+	const KEY*	key_info,	/*!< in: Indexes to be created */
+	ulint		num_of_keys);	/*!< in: Number of indexes to
 					be created. */
 

storage/xtradb/handler/handler0alter.cc

@@ -649,44 +649,47 @@ ha_innobase::add_index(
 
 	update_thd();
 
-	heap = mem_heap_create(1024);
-
 	/* In case MySQL calls this in the middle of a SELECT query, release
 	possible adaptive hash latch to avoid deadlocks of threads. */
 	trx_search_latch_release_if_reserved(prebuilt->trx);
-	trx_start_if_not_started(prebuilt->trx);
+	if (prebuilt->trx->fake_changes) {
+		DBUG_RETURN(HA_ERR_WRONG_COMMAND);
+	}
 
-	/* Create a background transaction for the operations on
-	the data dictionary tables. */
-	trx = innobase_trx_allocate(user_thd);
-	trx_start_if_not_started(trx);
+	/* Check if the index name is reserved. */
+	if (innobase_index_name_is_reserved(user_thd, key_info, num_of_keys)) {
+		DBUG_RETURN(-1);
+	}
 
 	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;
+		DBUG_RETURN(HA_ERR_NO_SUCH_TABLE);
 	}
 
-	/* Check if the index name is reserved. */
-	if (innobase_index_name_is_reserved(trx, key_info, num_of_keys)) {
-                error = ER_WRONG_NAME_FOR_INDEX;
-	} else {
-		/* Check that index keys are sensible */
-		error = innobase_check_index_keys(key_info, num_of_keys,
-						  innodb_table);
-	}
+	/* Check that index keys are sensible */
+	error = innobase_check_index_keys(key_info, num_of_keys, innodb_table);
 
 	if (UNIV_UNLIKELY(error)) {
-err_exit:
+		DBUG_RETURN(error);
+	}
+
+	heap = mem_heap_create(1024);
+	trx_start_if_not_started(prebuilt->trx);
+
+	/* Create a background transaction for the operations on
+	the data dictionary tables. */
+	trx = innobase_trx_allocate(user_thd);
+	if (trx->fake_changes) {
 		mem_heap_free(heap);
 		trx_general_rollback_for_mysql(trx, NULL);
 		trx_free_for_mysql(trx);
-		trx_commit_for_mysql(prebuilt->trx);
-		DBUG_RETURN(error);
+		DBUG_RETURN(HA_ERR_WRONG_COMMAND);
 	}
 
+	trx_start_if_not_started(trx);
+
 	/* Create table containing all indexes to be built in this
 	alter table add index so that they are in the correct order
 	in the table. */
@@ -758,8 +761,12 @@ err_exit:
 
 			ut_d(dict_table_check_for_dup_indexes(innodb_table,
 							      FALSE));
+			mem_heap_free(heap);
+			trx_general_rollback_for_mysql(trx, NULL);
 			row_mysql_unlock_data_dictionary(trx);
-			goto err_exit;
+			trx_free_for_mysql(trx);
+			trx_commit_for_mysql(prebuilt->trx);
+			DBUG_RETURN(error);
 		}
 
 		trx->table_id = indexed_table->id;
@@ -782,10 +789,6 @@ err_exit:
 
 	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
@@ -911,6 +914,14 @@ error:
 		}
 
 convert_error:
+		if (error == DB_SUCCESS) {
+			/* Build index is successful. We will need to
+			rebuild index translation table.  Reset the
+			index entry count in the translation table
+			to zero, so that translation table will be rebuilt */
+			share->idx_trans_tbl.index_count = 0;
+		}
+
 		error = convert_error_code_to_mysql(error,
 						    innodb_table->flags,
 						    user_thd);
@@ -963,6 +974,10 @@ ha_innobase::prepare_drop_index(
 	trx_search_latch_release_if_reserved(prebuilt->trx);
 	trx = prebuilt->trx;
 
+	if (trx->fake_changes) {
+		DBUG_RETURN(HA_ERR_WRONG_COMMAND);
+	}
+
 	/* Test and mark all the indexes to be dropped */
 
 	row_mysql_lock_data_dictionary(trx);
@@ -1167,6 +1182,12 @@ ha_innobase::final_drop_index(
 	/* Create a background transaction for the operations on
 	the data dictionary tables. */
 	trx = innobase_trx_allocate(user_thd);
+	if (trx->fake_changes) {
+		trx_general_rollback_for_mysql(trx, NULL);
+		trx_free_for_mysql(trx);
+		DBUG_RETURN(HA_ERR_WRONG_COMMAND);
+	}
+
 	trx_start_if_not_started(trx);
 
 	/* Flag this transaction as a dictionary operation, so that

storage/xtradb/handler/innodb_patch_info.h

@@ -47,6 +47,5 @@ struct innodb_enhancement {
 {"innodb_fast_checksum","Using the checksum on 32bit-unit calculation","incompatible for unpatched ver.","http://www.percona.com/docs/wiki/percona-xtradb"},
 {"innodb_files_extend","allow >4GB transaction log files, and can vary universal page size of datafiles","incompatible for unpatched ver.","http://www.percona.com/docs/wiki/percona-xtradb"},
 {"innodb_sys_tables_sys_indexes","Expose InnoDB SYS_TABLES and SYS_INDEXES schema tables","","http://www.percona.com/docs/wiki/percona-xtradb"},
-{"innodb_buffer_pool_shm","Put buffer pool contents to shared memory segment and reuse it at clean restart [experimental]","","http://www.percona.com/docs/wiki/percona-xtradb"},
 {NULL, NULL, NULL, NULL}
 };

storage/xtradb/ibuf/ibuf0ibuf.c

@@ -2613,6 +2613,8 @@ ibuf_insert_low(
 
 	ut_a(trx_sys_multiple_tablespace_format);
 
+	ut_ad(!(thr_get_trx(thr)->fake_changes));
+
 	do_merge = FALSE;
 
 	mutex_enter(&ibuf_mutex);

storage/xtradb/include/fil0fil.h

@@ -670,8 +670,9 @@ UNIV_INTERN
 void
 fil_flush(
 /*======*/
-	ulint	space_id);	/*!< in: file space id (this can be a group of
+	ulint	space_id,	/*!< in: file space id (this can be a group of
 				log files or a tablespace of the database) */
+	ibool	metadata);
 /**********************************************************************//**
 Flushes to disk writes in file spaces of the given type possibly cached by
 the OS. */