This ChangeSet must be null-merged to 5.1. Applied innodb-5.0-ss982, -ss998, -ss1003

Fixes:
- Bug #15815: Very poor performance with multiple queries running concurrently
- Bug #22868: 'Thread thrashing' with > 50 concurrent conns under an upd-intensive workloadw
- Bug #23769: Debug assertion failure with innodb_locks_unsafe_for_binlog
- Bug #24089: Race condition in fil_flush_file_spaces()


innobase/buf/buf0buf.c:
  Applied innodb-5.0-ss982, -ss998, -ss1003
  
  Revision r1000:
  branches/5.0: Merge r999 from trunk:
  
  Reduce buffer pool mutex contention under >= 4 big concurrent
  CPU-bound SELECT queries.  (Bug #22868)
  
  Fix: replace the mutex by one mutex protecting the 'flush list'
  (and the free list) and several mutexes protecting portions of the
  buffer pool, where we keep several indivudual LRU lists of pages.
  
  This patch is from Sunny Bains and Heikki Tuuri.
innobase/buf/buf0flu.c:
  Applied innodb-5.0-ss982, -ss998, -ss1003
  
  Revision r1000:
  branches/5.0: Merge r999 from trunk:
  
  Reduce buffer pool mutex contention under >= 4 big concurrent
  CPU-bound SELECT queries.  (Bug #22868)
  
  Fix: replace the mutex by one mutex protecting the 'flush list'
  (and the free list) and several mutexes protecting portions of the
  buffer pool, where we keep several indivudual LRU lists of pages.
  
  This patch is from Sunny Bains and Heikki Tuuri.
innobase/buf/buf0lru.c:
  Applied innodb-5.0-ss982, -ss998, -ss1003
  
  Revision r1000:
  branches/5.0: Merge r999 from trunk:
  
  Reduce buffer pool mutex contention under >= 4 big concurrent
  CPU-bound SELECT queries.  (Bug #22868)
  
  Fix: replace the mutex by one mutex protecting the 'flush list'
  (and the free list) and several mutexes protecting portions of the
  buffer pool, where we keep several indivudual LRU lists of pages.
  
  This patch is from Sunny Bains and Heikki Tuuri.
innobase/dict/dict0crea.c:
  Applied innodb-5.0-ss982, -ss998, -ss1003
  
  Revision r974:
  branches/5.0: Port r973 from trunk.
  
  Do not break the latching order in TRUNCATE TABLE.
  
  dict_truncate_index_tree(): Replace parameter rec_t* rec with
  btr_pcur_t* pcur.  Reposition pcur before calling btr_create().
  
  sync_thread_add_level(): Remove the relaxation of the assertion added in r968.
innobase/fil/fil0fil.c:
  Applied innodb-5.0-ss982, -ss998, -ss1003
  
  Revision r1003:
  branches/5.0: Merge r1002 from trunk:
  
  fil_flush_file_spaces(): Copy the system->unflushed_spaces list to an
  array while holding the mutex.  This removes the crash-triggering
  race condition that was introduced when fixing Bug 15653.  (Bug #24089)
innobase/include/buf0buf.h:
  Applied innodb-5.0-ss982, -ss998, -ss1003
  
  Revision r1000:
  branches/5.0: Merge r999 from trunk:
  
  Reduce buffer pool mutex contention under >= 4 big concurrent
  CPU-bound SELECT queries.  (Bug #22868)
  
  Fix: replace the mutex by one mutex protecting the 'flush list'
  (and the free list) and several mutexes protecting portions of the
  buffer pool, where we keep several indivudual LRU lists of pages.
  
  This patch is from Sunny Bains and Heikki Tuuri.
innobase/include/buf0buf.ic:
  Applied innodb-5.0-ss982, -ss998, -ss1003
  
  Revision r1000:
  branches/5.0: Merge r999 from trunk:
  
  Reduce buffer pool mutex contention under >= 4 big concurrent
  CPU-bound SELECT queries.  (Bug #22868)
  
  Fix: replace the mutex by one mutex protecting the 'flush list'
  (and the free list) and several mutexes protecting portions of the
  buffer pool, where we keep several indivudual LRU lists of pages.
  
  This patch is from Sunny Bains and Heikki Tuuri.
innobase/include/dict0crea.h:
  Applied innodb-5.0-ss982, -ss998, -ss1003
  
  Revision r974:
  branches/5.0: Port r973 from trunk.
  
  Do not break the latching order in TRUNCATE TABLE.
  
  dict_truncate_index_tree(): Replace parameter rec_t* rec with
  btr_pcur_t* pcur.  Reposition pcur before calling btr_create().
  
  sync_thread_add_level(): Remove the relaxation of the assertion added in r968.
innobase/include/sync0arr.h:
  Applied innodb-5.0-ss982, -ss998, -ss1003
  
  Revision r1001:
  branches/5.0: Reduce locking contention:
  
  Bug #15815: 'Thread thrashing' with > 50 concurrent connections under
  an update-intensive workload.
  Fix: Introduce one event per InnoDB semaphore.
  
  This patch is from Sunny Bains and Heikki Tuuri.
  This patch will not be merged to trunk (MySQL/InnoDB 5.1) yet,
  because it tries to address the problem in a different way.
innobase/include/sync0rw.h:
  Applied innodb-5.0-ss982, -ss998, -ss1003
  
  Revision r1001:
  branches/5.0: Reduce locking contention:
  
  Bug #15815: 'Thread thrashing' with > 50 concurrent connections under
  an update-intensive workload.
  Fix: Introduce one event per InnoDB semaphore.
  
  This patch is from Sunny Bains and Heikki Tuuri.
  This patch will not be merged to trunk (MySQL/InnoDB 5.1) yet,
  because it tries to address the problem in a different way.
innobase/include/sync0rw.ic:
  Applied innodb-5.0-ss982, -ss998, -ss1003
  
  Revision r1001:
  branches/5.0: Reduce locking contention:
  
  Bug #15815: 'Thread thrashing' with > 50 concurrent connections under
  an update-intensive workload.
  Fix: Introduce one event per InnoDB semaphore.
  
  This patch is from Sunny Bains and Heikki Tuuri.
  This patch will not be merged to trunk (MySQL/InnoDB 5.1) yet,
  because it tries to address the problem in a different way.
innobase/include/sync0sync.h:
  Applied innodb-5.0-ss982, -ss998, -ss1003
  
  Revision r1001:
  branches/5.0: Reduce locking contention:
  
  Bug #15815: 'Thread thrashing' with > 50 concurrent connections under
  an update-intensive workload.
  Fix: Introduce one event per InnoDB semaphore.
  
  This patch is from Sunny Bains and Heikki Tuuri.
  This patch will not be merged to trunk (MySQL/InnoDB 5.1) yet,
  because it tries to address the problem in a different way.
innobase/os/os0sync.c:
  Applied innodb-5.0-ss982, -ss998, -ss1003
  
  Revision r1001:
  branches/5.0: Reduce locking contention:
  
  Bug #15815: 'Thread thrashing' with > 50 concurrent connections under
  an update-intensive workload.
  Fix: Introduce one event per InnoDB semaphore.
  
  This patch is from Sunny Bains and Heikki Tuuri.
  This patch will not be merged to trunk (MySQL/InnoDB 5.1) yet,
  because it tries to address the problem in a different way.
innobase/row/row0mysql.c:
  Applied innodb-5.0-ss982, -ss998, -ss1003
  
  Revision r974:
  branches/5.0: Port r973 from trunk.
  
  Do not break the latching order in TRUNCATE TABLE.
  
  dict_truncate_index_tree(): Replace parameter rec_t* rec with
  btr_pcur_t* pcur.  Reposition pcur before calling btr_create().
  
  sync_thread_add_level(): Remove the relaxation of the assertion added in r968.
innobase/row/row0sel.c:
  Applied innodb-5.0-ss982, -ss998, -ss1003
  
  Revision r982:
  branches/5.0: row_sel(): Do not try to acquire a LOCK_REC_NOT_GAP lock
  on the supremum record.  Instead, skip to the next record.  (Bug #23769)
  This fix was backported from r623 in the 5.1 tree.
innobase/srv/srv0start.c:
  Applied innodb-5.0-ss982, -ss998, -ss1003
  
  Revision r926:
  Refer to bug: 22268. Since no one tries to run 5.0 on Windows 95/ME it was
  decided to raise the limit of srv_max_n_threads to 10000 on Windows.
innobase/sync/sync0arr.c:
  Applied innodb-5.0-ss982, -ss998, -ss1003
  
  Revision r1001:
  branches/5.0: Reduce locking contention:
  
  Bug #15815: 'Thread thrashing' with > 50 concurrent connections under
  an update-intensive workload.
  Fix: Introduce one event per InnoDB semaphore.
  
  This patch is from Sunny Bains and Heikki Tuuri.
  This patch will not be merged to trunk (MySQL/InnoDB 5.1) yet,
  because it tries to address the problem in a different way.
innobase/sync/sync0rw.c:
  Applied innodb-5.0-ss982, -ss998, -ss1003
  
  Revision r1001:
  branches/5.0: Reduce locking contention:
  
  Bug #15815: 'Thread thrashing' with > 50 concurrent connections under
  an update-intensive workload.
  Fix: Introduce one event per InnoDB semaphore.
  
  This patch is from Sunny Bains and Heikki Tuuri.
  This patch will not be merged to trunk (MySQL/InnoDB 5.1) yet,
  because it tries to address the problem in a different way.
innobase/sync/sync0sync.c:
  Applied innodb-5.0-ss982, -ss998, -ss1003
  
  Revision r969:
  branches/5.0: Port r968 from trunk:
  
  sync_thread_add_level(): When level == SYNC_TREE_NODE, allow the latching
  order to be violated if the thread holds dict_operation_lock, whose level is
  SYNC_DICT_OPERATION.  This removes the assertion failure of TRUNCATE TABLE
  #ifdef UNIV_SYNC_DEBUG.
  
  
  Revision r974:
  branches/5.0: Port r973 from trunk.
  
  Do not break the latching order in TRUNCATE TABLE.
  
  dict_truncate_index_tree(): Replace parameter rec_t* rec with
  btr_pcur_t* pcur.  Reposition pcur before calling btr_create().
  
  sync_thread_add_level(): Remove the relaxation of the assertion added in r968.
  
  
  Revision r1001:
  branches/5.0: Reduce locking contention:
  
  Bug #15815: 'Thread thrashing' with > 50 concurrent connections under
  an update-intensive workload.
  Fix: Introduce one event per InnoDB semaphore.
  
  This patch is from Sunny Bains and Heikki Tuuri.
  This patch will not be merged to trunk (MySQL/InnoDB 5.1) yet,
  because it tries to address the problem in a different way.

innobase/buf/buf0buf.c

@@ -221,6 +221,9 @@ in the free list to the frames.
 5) When we have AWE enabled, we disable adaptive hash indexes.
 */
 
+/* Value in microseconds */
+static const int WAIT_FOR_READ	= 20000;
+
 buf_pool_t*	buf_pool = NULL; /* The buffer buf_pool of the database */
 
 #ifdef UNIV_DEBUG
@@ -488,6 +491,9 @@ buf_block_init(
 
 	block->n_pointers = 0;
 
+	mutex_create(&block->mutex);
+	mutex_set_level(&block->mutex, SYNC_BUF_BLOCK);
+
 	rw_lock_create(&(block->lock));
 	ut_ad(rw_lock_validate(&(block->lock)));
 
@@ -756,8 +762,15 @@ buf_awe_map_page_to_frame(
 	bck = UT_LIST_GET_LAST(buf_pool->awe_LRU_free_mapped);
 
 	while (bck) {	
-		if (bck->state == BUF_BLOCK_FILE_PAGE
-	    	    && (bck->buf_fix_count != 0 || bck->io_fix != 0)) {
+		ibool skip;
+
+		mutex_enter(&bck->mutex);
+
+		skip = (bck->state == BUF_BLOCK_FILE_PAGE
+			&& (bck->buf_fix_count != 0 || bck->io_fix != 0));
+
+		if (skip) {
+			mutex_exit(&bck->mutex);
 
 			/* We have to skip this */
 			bck = UT_LIST_GET_PREV(awe_LRU_free_mapped, bck);
@@ -790,6 +803,8 @@ buf_awe_map_page_to_frame(
 
 			buf_pool->n_pages_awe_remapped++;
 			
+			mutex_exit(&bck->mutex);
+
 			return;
 		}
 	}
@@ -828,13 +843,22 @@ buf_block_make_young(
 /*=================*/
 	buf_block_t*	block)	/* in: block to make younger */
 {
+#ifdef UNIV_SYNC_DEBUG
+	ut_ad(!mutex_own(&(buf_pool->mutex)));
+#endif /* UNIV_SYNC_DEBUG */
+
+	/* Note that we read freed_page_clock's without holding any mutex:
+	this is allowed since the result is used only in heuristics */
+
 	if (buf_pool->freed_page_clock >= block->freed_page_clock 
-				+ 1 + (buf_pool->curr_size / 1024)) {
+				+ 1 + (buf_pool->curr_size / 4)) {
 
+		mutex_enter(&buf_pool->mutex);
 		/* There has been freeing activity in the LRU list:
 		best to move to the head of the LRU list */
 
 		buf_LRU_make_block_young(block);
+		mutex_exit(&buf_pool->mutex);
 	}
 }
 
@@ -869,12 +893,16 @@ buf_block_free(
 /*===========*/
 	buf_block_t*	block)	/* in, own: block to be freed */
 {
-	ut_a(block->state != BUF_BLOCK_FILE_PAGE);
-
 	mutex_enter(&(buf_pool->mutex));
 
+	mutex_enter(&block->mutex);
+
+	ut_a(block->state != BUF_BLOCK_FILE_PAGE);
+
 	buf_LRU_block_free_non_file_page(block);
 
+	mutex_exit(&block->mutex);
+
 	mutex_exit(&(buf_pool->mutex));
 }
 
@@ -1093,9 +1121,8 @@ buf_page_get_gen(
 #endif
 	buf_pool->n_page_gets++;
 loop:
-	mutex_enter_fast(&(buf_pool->mutex));
-
 	block = NULL;
+	mutex_enter_fast(&(buf_pool->mutex));
 	
 	if (guess) {
 		block = buf_block_align(guess);
@@ -1133,6 +1160,8 @@ loop:
 		goto loop;
 	}
 
+	mutex_enter(&block->mutex);
+
 	ut_a(block->state == BUF_BLOCK_FILE_PAGE);
 
 	must_read = FALSE;
@@ -1142,9 +1171,9 @@ loop:
 		must_read = TRUE;
 
 		if (mode == BUF_GET_IF_IN_POOL) {
-
 			/* The page is only being read to buffer */
-			mutex_exit(&(buf_pool->mutex));
+			mutex_exit(&buf_pool->mutex);
+			mutex_exit(&block->mutex);
 
 			return(NULL);
 		}
@@ -1168,7 +1197,7 @@ loop:
 #else
 	buf_block_buf_fix_inc(block);
 #endif
-	buf_block_make_young(block);
+	mutex_exit(&buf_pool->mutex);
 
 	/* Check if this is the first access to the page */
 
@@ -1176,10 +1205,13 @@ loop:
 
 	block->accessed = TRUE;
 
+	mutex_exit(&block->mutex);
+
+	buf_block_make_young(block);
+
 #ifdef UNIV_DEBUG_FILE_ACCESSES
 	ut_a(block->file_page_was_freed == FALSE);
 #endif	
-	mutex_exit(&(buf_pool->mutex));
 
 #ifdef UNIV_DEBUG
 	buf_dbg_counter++;
@@ -1204,13 +1236,14 @@ loop:
 		}
 
 		if (!success) {
-			mutex_enter(&(buf_pool->mutex));
+			mutex_enter(&block->mutex);
 
 			block->buf_fix_count--;
+
+			mutex_exit(&block->mutex);
 #ifdef UNIV_SYNC_DEBUG
 			rw_lock_s_unlock(&(block->debug_latch));
 #endif			
-			mutex_exit(&(buf_pool->mutex));
 
 			return(NULL);
 		}
@@ -1221,18 +1254,16 @@ loop:
 			completes */
 
 		        for (;;) {
-			        mutex_enter(&(buf_pool->mutex));
+				mutex_enter(&block->mutex);
 
 		                if (block->io_fix == BUF_IO_READ) {
 
-				        mutex_exit(&(buf_pool->mutex));
+					mutex_exit(&block->mutex);
 				  
-				        /* Sleep 20 milliseconds */
-
-				        os_thread_sleep(20000);
+					os_thread_sleep(WAIT_FOR_READ);
 				} else {
 				  
-				       mutex_exit(&(buf_pool->mutex));
+					mutex_exit(&block->mutex);
 
 				       break;
 				}
@@ -1290,15 +1321,15 @@ buf_page_optimistic_get_func(
 
 	ut_ad(mtr && block);
 	ut_ad((rw_latch == RW_S_LATCH) || (rw_latch == RW_X_LATCH));
-	
-	mutex_enter(&(buf_pool->mutex));
 
 	/* If AWE is used, block may have a different frame now, e.g., NULL */
-	
+
+	mutex_enter(&block->mutex);
+
 	if (UNIV_UNLIKELY(block->state != BUF_BLOCK_FILE_PAGE)
-			|| UNIV_UNLIKELY(block->frame != guess)) {
-	exit_func:
-		mutex_exit(&(buf_pool->mutex));
+	    || UNIV_UNLIKELY(block->frame != guess)) {
+
+		mutex_exit(&block->mutex);
 
 		return(FALSE);
 	}
@@ -1308,15 +1339,14 @@ buf_page_optimistic_get_func(
 #else
 	buf_block_buf_fix_inc(block);
 #endif
-	buf_block_make_young(block);
-
-	/* Check if this is the first access to the page */
-
 	accessed = block->accessed;
-
 	block->accessed = TRUE;
 
-	mutex_exit(&(buf_pool->mutex));
+	mutex_exit(&block->mutex);
+
+	buf_block_make_young(block);
+
+	/* Check if this is the first access to the page */
 
 	ut_ad(!ibuf_inside() || ibuf_page(block->space, block->offset));
 
@@ -1331,13 +1361,16 @@ buf_page_optimistic_get_func(
 	}
 
 	if (UNIV_UNLIKELY(!success)) {
-		mutex_enter(&(buf_pool->mutex));
+		mutex_enter(&block->mutex);
 		
 		block->buf_fix_count--;
+
+		mutex_exit(&block->mutex);
+
 #ifdef UNIV_SYNC_DEBUG
 		rw_lock_s_unlock(&(block->debug_latch));
 #endif
-		goto exit_func;
+		return(FALSE);
 	}
 
 	if (UNIV_UNLIKELY(!UT_DULINT_EQ(modify_clock, block->modify_clock))) {
@@ -1350,13 +1383,16 @@ buf_page_optimistic_get_func(
 			rw_lock_x_unlock(&(block->lock));
 		}
 
-		mutex_enter(&(buf_pool->mutex));
+		mutex_enter(&block->mutex);
 		
 		block->buf_fix_count--;
+
+		mutex_exit(&block->mutex);
+
 #ifdef UNIV_SYNC_DEBUG
 		rw_lock_s_unlock(&(block->debug_latch));
 #endif
-		goto exit_func;
+		return(FALSE);
 	}
 
 	mtr_memo_push(mtr, block, fix_type);
@@ -1413,10 +1449,10 @@ buf_page_get_known_nowait(
 	ut_ad(mtr);
 	ut_ad((rw_latch == RW_S_LATCH) || (rw_latch == RW_X_LATCH));
 	
-	mutex_enter(&(buf_pool->mutex));
-
 	block = buf_block_align(guess);
 
+	mutex_enter(&block->mutex);
+
 	if (block->state == BUF_BLOCK_REMOVE_HASH) {
 	        /* Another thread is just freeing the block from the LRU list
 	        of the buffer pool: do not try to access this page; this
@@ -1425,7 +1461,7 @@ buf_page_get_known_nowait(
 		we have already removed it from the page address hash table
 		of the buffer pool. */
 
-	        mutex_exit(&(buf_pool->mutex));
+		mutex_exit(&block->mutex);
 
 		return(FALSE);
 	}
@@ -1437,12 +1473,12 @@ buf_page_get_known_nowait(
 #else
 	buf_block_buf_fix_inc(block);
 #endif
+	mutex_exit(&block->mutex);
+
 	if (mode == BUF_MAKE_YOUNG) {
 		buf_block_make_young(block);
 	}
 
-	mutex_exit(&(buf_pool->mutex));
-
 	ut_ad(!ibuf_inside() || (mode == BUF_KEEP_OLD));
 
 	if (rw_latch == RW_S_LATCH) {
@@ -1456,13 +1492,15 @@ buf_page_get_known_nowait(
 	}
 	
 	if (!success) {
-		mutex_enter(&(buf_pool->mutex));
+		mutex_enter(&block->mutex);
 		
 		block->buf_fix_count--;
+
+		mutex_exit(&block->mutex);
+
 #ifdef UNIV_SYNC_DEBUG
 		rw_lock_s_unlock(&(block->debug_latch));
 #endif		
-		mutex_exit(&(buf_pool->mutex));
 
 		return(FALSE);
 	}
@@ -1510,8 +1548,7 @@ buf_page_init_for_backup_restore(
 	block->offset 		= offset;
 
 	block->lock_hash_val	= 0;
-	block->lock_mutex	= NULL;
-	
+
 	block->freed_page_clock = 0;
 
 	block->newest_modification = ut_dulint_zero;
@@ -1543,6 +1580,7 @@ buf_page_init(
 {
 #ifdef UNIV_SYNC_DEBUG
 	ut_ad(mutex_own(&(buf_pool->mutex)));
+	ut_ad(mutex_own(&(block->mutex)));
 #endif /* UNIV_SYNC_DEBUG */
 	ut_a(block->state != BUF_BLOCK_FILE_PAGE);
 
@@ -1557,8 +1595,7 @@ buf_page_init(
 	block->index		= NULL;
 	
 	block->lock_hash_val	= lock_rec_hash(space, offset);
-	block->lock_mutex	= NULL;
-	
+
 	/* Insert into the hash table of file pages */
 
         if (buf_page_hash_get(space, offset)) {
@@ -1650,6 +1687,7 @@ buf_page_init_for_read(
 	ut_a(block);
 
 	mutex_enter(&(buf_pool->mutex));
+	mutex_enter(&block->mutex);
 
 	if (fil_tablespace_deleted_or_being_deleted_in_mem(space,
 							tablespace_version)) {
@@ -1662,7 +1700,9 @@ buf_page_init_for_read(
 		/* The page belongs to a space which has been deleted or is
 		being deleted, or the page is already in buf_pool, return */
 
+		mutex_exit(&block->mutex);
 		mutex_exit(&(buf_pool->mutex));
+
 		buf_block_free(block);
 
 		if (mode == BUF_READ_IBUF_PAGES_ONLY) {
@@ -1682,6 +1722,7 @@ buf_page_init_for_read(
 	buf_LRU_add_block(block, TRUE); 	/* TRUE == to old blocks */
 	
 	block->io_fix = BUF_IO_READ;
+
 	buf_pool->n_pend_reads++;
 	
 	/* We set a pass-type x-lock on the frame because then the same
@@ -1693,6 +1734,7 @@ buf_page_init_for_read(
 	
 	rw_lock_x_lock_gen(&(block->lock), BUF_IO_READ);
 	
+	mutex_exit(&block->mutex);
  	mutex_exit(&(buf_pool->mutex));
 
 	if (mode == BUF_READ_IBUF_PAGES_ONLY) {
@@ -1757,6 +1799,8 @@ buf_page_create(
 
 	block = free_block;
 	
+	mutex_enter(&block->mutex);
+
 	buf_page_init(space, offset, block);
 
 	/* The block must be put to the LRU list */
@@ -1767,13 +1811,15 @@ buf_page_create(
 #else
 	buf_block_buf_fix_inc(block);
 #endif
+	buf_pool->n_pages_created++;
+
+	mutex_exit(&(buf_pool->mutex));
+
 	mtr_memo_push(mtr, block, MTR_MEMO_BUF_FIX);
 
 	block->accessed = TRUE;
 	
-	buf_pool->n_pages_created++;
-
-	mutex_exit(&(buf_pool->mutex));
+	mutex_exit(&block->mutex);
 
 	/* Delete possible entries for the page from the insert buffer:
 	such can exist if the page belonged to an index which was dropped */
@@ -1822,6 +1868,12 @@ buf_page_io_complete(
 
 	ut_a(block->state == BUF_BLOCK_FILE_PAGE);
 
+	/* We do not need protect block->io_fix here by block->mutex to read
+	it because this is the only function where we can change the value
+	from BUF_IO_READ or BUF_IO_WRITE to some other value, and our code
+	ensures that this is the only thread that handles the i/o for this
+	block. */
+
 	io_type = block->io_fix;
 
 	if (io_type == BUF_IO_READ) {
@@ -1890,11 +1942,12 @@ buf_page_io_complete(
 		}
 	}
 	
+	mutex_enter(&(buf_pool->mutex));
+	mutex_enter(&block->mutex);
+
 #ifdef UNIV_IBUF_DEBUG
 	ut_a(ibuf_count_get(block->space, block->offset) == 0);
 #endif
-	mutex_enter(&(buf_pool->mutex));
-	
 	/* Because this thread which does the unlocking is not the same that
 	did the locking, we use a pass value != 0 in unlock, which simply
 	removes the newest lock debug record, without checking the thread
@@ -1937,6 +1990,7 @@ buf_page_io_complete(
 #endif /* UNIV_DEBUG */
 	}
 	
+	mutex_exit(&block->mutex);
 	mutex_exit(&(buf_pool->mutex));
 
 #ifdef UNIV_DEBUG
@@ -1999,6 +2053,8 @@ buf_validate(void)
 
 		block = buf_pool_get_nth_block(buf_pool, i);
 
+		mutex_enter(&block->mutex);
+
 		if (block->state == BUF_BLOCK_FILE_PAGE) {
 
 			ut_a(buf_page_hash_get(block->space,
@@ -2042,6 +2098,8 @@ buf_validate(void)
 		} else if (block->state == BUF_BLOCK_NOT_USED) {
 			n_free++;
 		}
+
+		mutex_exit(&block->mutex);
  	}
 
 	if (n_lru + n_free > buf_pool->curr_size) {
@@ -2185,11 +2243,17 @@ buf_get_latched_pages_number(void)
 
         for (i = 0; i < buf_pool->curr_size; i++) {
 
-               block = buf_pool_get_nth_block(buf_pool, i);
+		block = buf_pool_get_nth_block(buf_pool, i);
+
+		if (block->magic_n == BUF_BLOCK_MAGIC_N) {
+			mutex_enter(&block->mutex);
 
-               if (((block->buf_fix_count != 0) || (block->io_fix != 0)) &&
-                    block->magic_n == BUF_BLOCK_MAGIC_N )
-                       fixed_pages_number++;
+			if (block->buf_fix_count != 0 || block->io_fix != 0) {
+				fixed_pages_number++;
+			}
+
+			mutex_exit(&block->mutex);
+		}
         }
 
         mutex_exit(&(buf_pool->mutex));
@@ -2354,16 +2418,21 @@ buf_all_freed(void)
 
 		block = buf_pool_get_nth_block(buf_pool, i);
 
+		mutex_enter(&block->mutex);
+
 		if (block->state == BUF_BLOCK_FILE_PAGE) {
 
 			if (!buf_flush_ready_for_replace(block)) {
 
 				fprintf(stderr,
 					"Page %lu %lu still fixed or dirty\n",
-					(ulong) block->space, (ulong) block->offset);
+					(ulong) block->space,
+					(ulong) block->offset);
 			    	ut_error;
 			}
 		}
+
+		mutex_exit(&block->mutex);
  	}
 
 	mutex_exit(&(buf_pool->mutex));

innobase/buf/buf0flu.c

@@ -114,6 +114,7 @@ buf_flush_ready_for_replace(
 {
 #ifdef UNIV_SYNC_DEBUG
 	ut_ad(mutex_own(&(buf_pool->mutex)));
+	ut_ad(mutex_own(&block->mutex));
 #endif /* UNIV_SYNC_DEBUG */
 	if (block->state != BUF_BLOCK_FILE_PAGE) {
 		ut_print_timestamp(stderr);
@@ -148,6 +149,7 @@ buf_flush_ready_for_flush(
 {
 #ifdef UNIV_SYNC_DEBUG
 	ut_ad(mutex_own(&(buf_pool->mutex)));
+	ut_ad(mutex_own(&(block->mutex)));
 #endif /* UNIV_SYNC_DEBUG */
 	ut_a(block->state == BUF_BLOCK_FILE_PAGE);
 
@@ -539,8 +541,15 @@ buf_flush_try_page(
 
 	ut_a(!block || block->state == BUF_BLOCK_FILE_PAGE);
 
+	if (!block) {
+		mutex_exit(&(buf_pool->mutex));
+		return(0);
+	}
+
+	mutex_enter(&block->mutex);
+
 	if (flush_type == BUF_FLUSH_LIST
-	    && block && buf_flush_ready_for_flush(block, flush_type)) {
+	    && buf_flush_ready_for_flush(block, flush_type)) {
 	
 		block->io_fix = BUF_IO_WRITE;
 
@@ -578,6 +587,7 @@ buf_flush_try_page(
 			locked = TRUE;
 		}
 
+		mutex_exit(&block->mutex);
 		mutex_exit(&(buf_pool->mutex));
 
 		if (!locked) {
@@ -598,8 +608,8 @@ buf_flush_try_page(
 		
 		return(1);
 
-	} else if (flush_type == BUF_FLUSH_LRU && block
-			&& buf_flush_ready_for_flush(block, flush_type)) {
+	} else if (flush_type == BUF_FLUSH_LRU
+		   && buf_flush_ready_for_flush(block, flush_type)) {
 
 		/* VERY IMPORTANT:
 		Because any thread may call the LRU flush, even when owning
@@ -639,14 +649,15 @@ buf_flush_try_page(
 		buf_pool mutex: this ensures that the latch is acquired
 		immediately. */
 		
+		mutex_exit(&block->mutex);
 		mutex_exit(&(buf_pool->mutex));
 
 		buf_flush_write_block_low(block);
 
 		return(1);
 
-	} else if (flush_type == BUF_FLUSH_SINGLE_PAGE && block
-			&& buf_flush_ready_for_flush(block, flush_type)) {
+	} else if (flush_type == BUF_FLUSH_SINGLE_PAGE
+		   && buf_flush_ready_for_flush(block, flush_type)) {
 	
 		block->io_fix = BUF_IO_WRITE;
 
@@ -672,6 +683,7 @@ buf_flush_try_page(
 
 		(buf_pool->n_flush[flush_type])++;
 
+		mutex_exit(&block->mutex);
 		mutex_exit(&(buf_pool->mutex));
 
 		rw_lock_s_lock_gen(&(block->lock), BUF_IO_WRITE);
@@ -688,11 +700,12 @@ buf_flush_try_page(
 		buf_flush_write_block_low(block);
 		
 		return(1);
-	} else {
-		mutex_exit(&(buf_pool->mutex));
+	}
 
-		return(0);
-	}		
+	mutex_exit(&block->mutex);
+	mutex_exit(&(buf_pool->mutex));
+
+	return(0);
 }
 
 /***************************************************************
@@ -737,34 +750,48 @@ buf_flush_try_neighbors(
 		block = buf_page_hash_get(space, i);
 		ut_a(!block || block->state == BUF_BLOCK_FILE_PAGE);
 
-		if (block && flush_type == BUF_FLUSH_LRU && i != offset
-		    && !block->old) {
+		if (!block) {
+
+			continue;
+
+		} else if (flush_type == BUF_FLUSH_LRU && i != offset
+			   && !block->old) {
 
 		        /* We avoid flushing 'non-old' blocks in an LRU flush,
 		        because the flushed blocks are soon freed */
 
 		        continue;
-		}
+		} else {
+
+			mutex_enter(&block->mutex);
+
+			if (buf_flush_ready_for_flush(block, flush_type)
+			    && (i == offset || block->buf_fix_count == 0)) {
+				/* We only try to flush those
+				neighbors != offset where the buf fix count is
+				zero, as we then know that we probably can
+				latch the page without a semaphore wait.
+				Semaphore waits are expensive because we must
+				flush the doublewrite buffer before we start
+				waiting. */
 
-		if (block && buf_flush_ready_for_flush(block, flush_type)
-		   && (i == offset || block->buf_fix_count == 0)) {
-			/* We only try to flush those neighbors != offset
-			where the buf fix count is zero, as we then know that
-			we probably can latch the page without a semaphore
-			wait. Semaphore waits are expensive because we must
-			flush the doublewrite buffer before we start
-			waiting. */
+				mutex_exit(&block->mutex);
 
-			mutex_exit(&(buf_pool->mutex));
+				mutex_exit(&(buf_pool->mutex));
 
-			/* Note: as we release the buf_pool mutex above, in
-			buf_flush_try_page we cannot be sure the page is still
-			in a flushable state: therefore we check it again
-			inside that function. */
+				/* Note: as we release the buf_pool mutex
+				above, in buf_flush_try_page we cannot be sure
+				the page is still in a flushable state:
+				therefore we check it again inside that
+				function. */
 
-			count += buf_flush_try_page(space, i, flush_type);
+				count += buf_flush_try_page(space, i,
+							    flush_type);
 
-			mutex_enter(&(buf_pool->mutex));
+				mutex_enter(&(buf_pool->mutex));
+			} else {
+				mutex_exit(&block->mutex);
+			}
 		}
 	}
 				
@@ -858,12 +885,15 @@ buf_flush_batch(
 	    	while ((block != NULL) && !found) {
 			ut_a(block->state == BUF_BLOCK_FILE_PAGE);
 
+			mutex_enter(&block->mutex);
+
 			if (buf_flush_ready_for_flush(block, flush_type)) {
 
 				found = TRUE;
 				space = block->space;
 				offset = block->offset;
 	    
+				mutex_exit(&block->mutex);
 				mutex_exit(&(buf_pool->mutex));
 
 				old_page_count = page_count;
@@ -881,10 +911,14 @@ buf_flush_batch(
 
 			} else if (flush_type == BUF_FLUSH_LRU) {
 
+				mutex_exit(&block->mutex);
+
 				block = UT_LIST_GET_PREV(LRU, block);
 			} else {
 				ut_ad(flush_type == BUF_FLUSH_LIST);
 
+				mutex_exit(&block->mutex);
+
 				block = UT_LIST_GET_PREV(flush_list, block);
 			}
 	    	}
@@ -966,10 +1000,14 @@ buf_flush_LRU_recommendation(void)
 	       				+ BUF_FLUSH_EXTRA_MARGIN)
 	       && (distance < BUF_LRU_FREE_SEARCH_LEN)) {
 
+		mutex_enter(&block->mutex);
+
 		if (buf_flush_ready_for_replace(block)) {
 			n_replaceable++;
 		}
 
+		mutex_exit(&block->mutex);
+
 		distance++;
 			
 		block = UT_LIST_GET_PREV(LRU, block);

innobase/buf/buf0lru.c

@@ -86,6 +86,9 @@ scan_again:
 	block = UT_LIST_GET_LAST(buf_pool->LRU);
 
 	while (block != NULL) {
+
+		mutex_enter(&block->mutex);
+
 	        ut_a(block->state == BUF_BLOCK_FILE_PAGE);
 
 		if (block->space == id
@@ -112,6 +115,8 @@ scan_again:
 			if (block->is_hashed) {
 				page_no = block->offset;
 			
+				mutex_exit(&block->mutex);
+
 				mutex_exit(&(buf_pool->mutex));
 
 				/* Note that the following call will acquire
@@ -138,6 +143,7 @@ scan_again:
 			buf_LRU_block_free_hashed_page(block);
 		}
 next_page:
+		mutex_exit(&block->mutex);
 		block = UT_LIST_GET_PREV(LRU, block);
 	}
 
@@ -211,6 +217,9 @@ buf_LRU_search_and_free_block(
 
 	while (block != NULL) {
 	        ut_a(block->in_LRU_list);
+
+		mutex_enter(&block->mutex);
+
 		if (buf_flush_ready_for_replace(block)) {
 
 #ifdef UNIV_DEBUG
@@ -225,6 +234,7 @@ buf_LRU_search_and_free_block(
 			buf_LRU_block_remove_hashed_page(block);
 
 			mutex_exit(&(buf_pool->mutex));
+			mutex_exit(&block->mutex);
 
 			/* Remove possible adaptive hash index built on the
 			page; in the case of AWE the block may not have a
@@ -233,15 +243,21 @@ buf_LRU_search_and_free_block(
 			if (block->frame) {
 				btr_search_drop_page_hash_index(block->frame);
 			}
-			mutex_enter(&(buf_pool->mutex));
 
 			ut_a(block->buf_fix_count == 0);
 
+			mutex_enter(&(buf_pool->mutex));
+			mutex_enter(&block->mutex);
+
 			buf_LRU_block_free_hashed_page(block);
 			freed = TRUE;
+			mutex_exit(&block->mutex);
 
 			break;
 		}
+
+		mutex_exit(&block->mutex);
+
 		block = UT_LIST_GET_PREV(LRU, block);
 		distance++;
 
@@ -415,8 +431,12 @@ loop:
 			}
 		}
 		
+		mutex_enter(&block->mutex);
+
 		block->state = BUF_BLOCK_READY_FOR_USE;
 
+		mutex_exit(&block->mutex);
+
 		mutex_exit(&(buf_pool->mutex));
 
 		if (started_monitor) {
@@ -818,6 +838,7 @@ buf_LRU_block_free_non_file_page(
 {
 #ifdef UNIV_SYNC_DEBUG
 	ut_ad(mutex_own(&(buf_pool->mutex)));
+	ut_ad(mutex_own(&block->mutex));
 #endif /* UNIV_SYNC_DEBUG */
 	ut_ad(block);
 	
@@ -857,6 +878,7 @@ buf_LRU_block_remove_hashed_page(
 {
 #ifdef UNIV_SYNC_DEBUG
 	ut_ad(mutex_own(&(buf_pool->mutex)));
+	ut_ad(mutex_own(&block->mutex));
 #endif /* UNIV_SYNC_DEBUG */
 	ut_ad(block);
 	
@@ -914,6 +936,7 @@ buf_LRU_block_free_hashed_page(
 {
 #ifdef UNIV_SYNC_DEBUG
 	ut_ad(mutex_own(&(buf_pool->mutex)));
+	ut_ad(mutex_own(&block->mutex));
 #endif /* UNIV_SYNC_DEBUG */
 	ut_a(block->state == BUF_BLOCK_REMOVE_HASH);
 

innobase/dict/dict0crea.c

@@ -724,8 +724,10 @@ dict_truncate_index_tree(
 				/* out: new root page number, or
 				FIL_NULL on failure */
 	dict_table_t*	table,	/* in: the table the index belongs to */
-	rec_t*		rec,	/* in: record in the clustered index of
-				SYS_INDEXES table */
+	btr_pcur_t*	pcur,	/* in/out: persistent cursor pointing to
+				record in the clustered index of
+				SYS_INDEXES table. The cursor may be
+				repositioned in this call. */
 	mtr_t*		mtr)	/* in: mtr having the latch
 				on the record page. The mtr may be
 				committed and restarted in this call. */
@@ -734,6 +736,7 @@ dict_truncate_index_tree(
 	ulint		space;
 	ulint		type;
 	dulint		index_id;
+	rec_t*		rec;
 	byte*		ptr;
 	ulint		len;
 	ulint		comp;
@@ -744,6 +747,7 @@ dict_truncate_index_tree(
 #endif /* UNIV_SYNC_DEBUG */
 
 	ut_a(!dict_sys->sys_indexes->comp);
+	rec = btr_pcur_get_rec(pcur);
 	ptr = rec_get_nth_field_old(rec, DICT_SYS_INDEXES_PAGE_NO_FIELD, &len);
 
 	ut_ad(len == 4);
@@ -809,10 +813,11 @@ dict_truncate_index_tree(
 	/* We will need to commit the mini-transaction in order to avoid
 	deadlocks in the btr_create() call, because otherwise we would
 	be freeing and allocating pages in the same mini-transaction. */
+	btr_pcur_store_position(pcur, mtr);
 	mtr_commit(mtr);
-	/* mtr_commit() will invalidate rec. */
-	rec = NULL;
+
 	mtr_start(mtr);
+	btr_pcur_restore_position(BTR_MODIFY_LEAF, pcur, mtr);
 
 	/* Find the index corresponding to this SYS_INDEXES record. */
 	for (index = UT_LIST_GET_FIRST(table->indexes);

innobase/fil/fil0fil.c

@@ -4285,29 +4285,47 @@ fil_flush_file_spaces(
 {
 	fil_system_t*	system	= fil_system;
 	fil_space_t*	space;
+	ulint*		space_ids;
+	ulint		n_space_ids;
+	ulint		i;
 
 	mutex_enter(&(system->mutex));
 
-	space = UT_LIST_GET_FIRST(system->unflushed_spaces);
+	n_space_ids = UT_LIST_GET_LEN(system->unflushed_spaces);
+	if (n_space_ids == 0) {
 
-	while (space) {
-		if (space->purpose == purpose && !space->is_being_deleted) {
+		mutex_exit(&system->mutex);
+		return;
+	}
 
-			space->n_pending_flushes++; /* prevent dropping of the
-						    space while we are
-						    flushing */
-			mutex_exit(&(system->mutex));
+	/* Assemble a list of space ids to flush.  Previously, we
+	traversed system->unflushed_spaces and called UT_LIST_GET_NEXT()
+	on a space that was just removed from the list by fil_flush().
+	Thus, the space could be dropped and the memory overwritten. */
+	space_ids = mem_alloc(n_space_ids * sizeof *space_ids);
 
-			fil_flush(space->id);
+	n_space_ids = 0;
 
-			mutex_enter(&(system->mutex));
+	for (space = UT_LIST_GET_FIRST(system->unflushed_spaces);
+	     space;
+	     space = UT_LIST_GET_NEXT(unflushed_spaces, space)) {
+
+		if (space->purpose == purpose && !space->is_being_deleted) {
 
-			space->n_pending_flushes--;
+			space_ids[n_space_ids++] = space->id;
 		}
-		space = UT_LIST_GET_NEXT(unflushed_spaces, space);
 	}
-	
-	mutex_exit(&(system->mutex));
+
+	mutex_exit(&system->mutex);
+
+	/* Flush the spaces.  It will not hurt to call fil_flush() on
+	a non-existing space id. */
+	for (i = 0; i < n_space_ids; i++) {
+
+		fil_flush(space_ids[i]);
+	}
+
+	mem_free(space_ids);
 }
 
 /**********************************************************************

innobase/include/buf0buf.h

@@ -461,8 +461,8 @@ Gets the mutex number protecting the page record lock hash chain in the lock
 table. */
 UNIV_INLINE
 mutex_t*
-buf_frame_get_lock_mutex(
-/*=====================*/
+buf_frame_get_mutex(
+/*================*/
 			/* out: mutex */
 	byte*	ptr);	/* in: pointer to within a buffer frame */
 /***********************************************************************
@@ -713,7 +713,10 @@ struct buf_block_struct{
 
 	ulint		magic_n;	/* magic number to check */
 	ulint		state;		/* state of the control block:
-					BUF_BLOCK_NOT_USED, ... */
+					BUF_BLOCK_NOT_USED, ...; changing
+					this is only allowed when a thread
+					has BOTH the buffer pool mutex AND
+					block->mutex locked */
 	byte*		frame;		/* pointer to buffer frame which
 					is of size UNIV_PAGE_SIZE, and
 					aligned to an address divisible by
@@ -731,8 +734,12 @@ struct buf_block_struct{
 	ulint		offset;		/* page number within the space */
 	ulint		lock_hash_val;	/* hashed value of the page address
 					in the record lock hash table */
-	mutex_t*	lock_mutex;	/* mutex protecting the chain in the
-					record lock hash table */
+	mutex_t		mutex;		/* mutex protecting this block:
+					state (also protected by the buffer
+					pool mutex), io_fix, buf_fix_count,
+					and accessed; we introduce this new
+					mutex in InnoDB-5.1 to relieve
+					contention on the buffer pool mutex */
 	rw_lock_t	lock;		/* read-write lock of the buffer
 					frame */
 	buf_block_t*	hash;		/* node used in chaining to the page
@@ -788,20 +795,27 @@ struct buf_block_struct{
 					in heuristic algorithms, because of
 					the possibility of a wrap-around! */
 	ulint		freed_page_clock;/* the value of freed_page_clock
-					buffer pool when this block was
-					last time put to the head of the
-					LRU list */
+					of the buffer pool when this block was
+					the last time put to the head of the
+					LRU list; a thread is allowed to
+					read this for heuristic purposes
+					without holding any mutex or latch */
 	ibool		old;		/* TRUE if the block is in the old
 					blocks in the LRU list */
 	ibool		accessed;	/* TRUE if the page has been accessed
 					while in the buffer pool: read-ahead
 					may read in pages which have not been
-					accessed yet */
+					accessed yet; this is protected by
+					block->mutex; a thread is allowed to
+					read this for heuristic purposes
+					without holding any mutex or latch */
 	ulint		buf_fix_count;	/* count of how manyfold this block
-					is currently bufferfixed */
+					is currently bufferfixed; this is
+					protected by block->mutex */
 	ulint		io_fix;		/* if a read is pending to the frame,
 					io_fix is BUF_IO_READ, in the case
-					of a write BUF_IO_WRITE, otherwise 0 */
+					of a write BUF_IO_WRITE, otherwise 0;
+					this is protected by block->mutex */
 	/* 4. Optimistic search field */
 
 	dulint		modify_clock;	/* this clock is incremented every
@@ -962,7 +976,9 @@ struct buf_pool_struct{
 					number of buffer blocks removed from
 					the end of the LRU list; NOTE that
 					this counter may wrap around at 4
-					billion! */
+					billion! A thread is allowed to
+					read this for heuristic purposes
+					without holding any mutex or latch */
 	ulint		LRU_flush_ended;/* when an LRU flush ends for a page,
 					this is incremented by one; this is
 					set to zero when a buffer block is

innobase/include/buf0buf.ic

@@ -330,8 +330,8 @@ Gets the mutex number protecting the page record lock hash chain in the lock
 table. */
 UNIV_INLINE
 mutex_t*
-buf_frame_get_lock_mutex(
-/*=====================*/
+buf_frame_get_mutex(
+/*================*/
 			/* out: mutex */
 	byte*	ptr)	/* in: pointer to within a buffer frame */
 {
@@ -339,7 +339,7 @@ buf_frame_get_lock_mutex(
 
 	block = buf_block_align(ptr);
 
-	return(block->lock_mutex);
+	return(&block->mutex);
 }
 
 /*************************************************************************
@@ -512,6 +512,7 @@ buf_block_buf_fix_inc_debug(
 	ret = rw_lock_s_lock_func_nowait(&(block->debug_latch), file, line);
 
 	ut_ad(ret == TRUE);
+	ut_ad(mutex_own(&block->mutex));
 #endif
 	block->buf_fix_count++;
 }
@@ -524,6 +525,9 @@ buf_block_buf_fix_inc(
 /*==================*/
 	buf_block_t*	block)	/* in: block to bufferfix */
 {
+#ifdef UNIV_SYNC_DEBUG
+	ut_ad(mutex_own(&block->mutex));
+#endif
 	block->buf_fix_count++;
 }
 #endif /* UNIV_SYNC_DEBUG */
@@ -618,23 +622,24 @@ buf_page_release(
 	
 	ut_ad(block);
 
-	mutex_enter_fast(&(buf_pool->mutex));
-
 	ut_a(block->state == BUF_BLOCK_FILE_PAGE);
 	ut_a(block->buf_fix_count > 0);
 
 	if (rw_latch == RW_X_LATCH && mtr->modifications) {
-
+		mutex_enter(&buf_pool->mutex);
 		buf_flush_note_modification(block, mtr);
+		mutex_exit(&buf_pool->mutex);
 	}
 
+	mutex_enter(&block->mutex);
+
 #ifdef UNIV_SYNC_DEBUG
 	rw_lock_s_unlock(&(block->debug_latch));
 #endif
 	buf_fix_count = block->buf_fix_count;
 	block->buf_fix_count = buf_fix_count - 1;
 
-	mutex_exit(&(buf_pool->mutex));
+	mutex_exit(&block->mutex);
 
 	if (rw_latch == RW_S_LATCH) {
 		rw_lock_s_unlock(&(block->lock));

innobase/include/dict0crea.h

@@ -62,8 +62,10 @@ dict_truncate_index_tree(
 				/* out: new root page number, or
 				FIL_NULL on failure */
 	dict_table_t*	table,	/* in: the table the index belongs to */
-	rec_t*		rec,	/* in: record in the clustered index of
-				SYS_INDEXES table */
+	btr_pcur_t*	pcur,	/* in/out: persistent cursor pointing to
+				record in the clustered index of
+				SYS_INDEXES table. The cursor may be
+				repositioned in this call. */
 	mtr_t*		mtr);	/* in: mtr having the latch
 				on the record page. The mtr may be
 				committed and restarted in this call. */

innobase/include/sync0arr.h

@@ -75,17 +75,12 @@ sync_array_free_cell(
 	sync_array_t*	arr,	/* in: wait array */
         ulint    	index);  /* in: index of the cell in array */
 /**************************************************************************
-Looks for the cells in the wait array which refer
-to the wait object specified,
-and sets their corresponding events to the signaled state. In this
-way releases the threads waiting for the object to contend for the object.
-It is possible that no such cell is found, in which case does nothing. */
+Note that one of the wait objects was signalled. */
 
 void
-sync_array_signal_object(
-/*=====================*/
-	sync_array_t*	arr,	/* in: wait array */
-	void*		object);/* in: wait object */
+sync_array_object_signalled(
+/*========================*/
+	sync_array_t*	arr);	/* in: wait array */
 /**************************************************************************
 If the wakeup algorithm does not work perfectly at semaphore relases,
 this function will do the waking (see the comment in mutex_exit). This

innobase/include/sync0rw.h

@@ -411,6 +411,7 @@ blocked by readers, a writer may queue for the lock by setting the writer
 field. Then no new readers are allowed in. */
 
 struct rw_lock_struct {
+	os_event_t	event;	/* Used by sync0arr.c for thread queueing */
 	ulint	reader_count;	/* Number of readers who have locked this
 				lock in the shared mode */
 	ulint	writer; 	/* This field is set to RW_LOCK_EX if there

innobase/include/sync0rw.ic

@@ -382,7 +382,8 @@ rw_lock_s_unlock_func(
 	mutex_exit(mutex);
 
 	if (UNIV_UNLIKELY(sg)) {
-		sync_array_signal_object(sync_primary_wait_array, lock);
+		os_event_set(lock->event);
+		sync_array_object_signalled(sync_primary_wait_array);
 	}
 
         ut_ad(rw_lock_validate(lock));
@@ -462,7 +463,8 @@ rw_lock_x_unlock_func(
 	mutex_exit(&(lock->mutex));
 
 	if (UNIV_UNLIKELY(sg)) {
-		sync_array_signal_object(sync_primary_wait_array, lock);
+		os_event_set(lock->event);
+		sync_array_object_signalled(sync_primary_wait_array);
 	}
 
         ut_ad(rw_lock_validate(lock));

innobase/include/sync0sync.h

@@ -453,6 +453,7 @@ Do not use its fields directly! The structure used in the spin lock
 implementation of a mutual exclusion semaphore. */
 
 struct mutex_struct {
+	os_event_t	event;	/* Used by sync0arr.c for the wait queue */
 	ulint	lock_word;	/* This ulint is the target of the atomic
 				test-and-set instruction in Win32 */
 #if !defined(_WIN32) || !defined(UNIV_CAN_USE_X86_ASSEMBLER)

innobase/os/os0sync.c

@@ -21,6 +21,7 @@ Created 9/6/1995 Heikki Tuuri
 
 /* Type definition for an operating system mutex struct */
 struct os_mutex_struct{ 
+	os_event_t	event;	/* Used by sync0arr.c for queing threads */
 	void*		handle;	/* OS handle to mutex */
 	ulint		count;	/* we use this counter to check
 				that the same thread does not
@@ -35,6 +36,7 @@ struct os_mutex_struct{
 /* Mutex protecting counts and the lists of OS mutexes and events */
 os_mutex_t	os_sync_mutex;
 ibool		os_sync_mutex_inited	= FALSE;
+ibool		os_sync_free_called	= FALSE;
 
 /* This is incremented by 1 in os_thread_create and decremented by 1 in
 os_thread_exit */
@@ -50,6 +52,10 @@ ulint	os_event_count		= 0;
 ulint	os_mutex_count		= 0;
 ulint	os_fast_mutex_count	= 0;
 
+/* Because a mutex is embedded inside an event and there is an
+event embedded inside a mutex, on free, this generates a recursive call.
+This version of the free event function doesn't acquire the global lock */
+static void os_event_free_internal(os_event_t	event);
 
 /*************************************************************
 Initializes global event and OS 'slow' mutex lists. */
@@ -76,6 +82,7 @@ os_sync_free(void)
 	os_event_t	event;
 	os_mutex_t	mutex;
 
+	os_sync_free_called = TRUE;
 	event = UT_LIST_GET_FIRST(os_event_list);
 
 	while (event) {
@@ -99,6 +106,7 @@ os_sync_free(void)
 
 	      mutex = UT_LIST_GET_FIRST(os_mutex_list);
 	}
+	os_sync_free_called = FALSE;
 }
 
 /*************************************************************
@@ -146,14 +154,21 @@ os_event_create(
 	event->signal_count = 0;
 #endif /* __WIN__ */
 
-        /* Put to the list of events */
-	os_mutex_enter(os_sync_mutex);
+	/* The os_sync_mutex can be NULL because during startup an event
+	can be created [ because it's embedded in the mutex/rwlock ] before
+	this module has been initialized */
+	if (os_sync_mutex != NULL) {
+		os_mutex_enter(os_sync_mutex);
+	}
 
+	/* Put to the list of events */
 	UT_LIST_ADD_FIRST(os_event_list, os_event_list, event);
 
 	os_event_count++;
 
-	os_mutex_exit(os_sync_mutex);
+	if (os_sync_mutex != NULL) {
+		os_mutex_exit(os_sync_mutex);
+	}
 
 	return(event);
 }
@@ -255,6 +270,35 @@ os_event_reset(
 #endif
 }
 
+/**************************************************************
+Frees an event object, without acquiring the global lock. */
+static
+void
+os_event_free_internal(
+/*===================*/
+	os_event_t	event)	/* in: event to free */
+{
+#ifdef __WIN__
+	ut_a(event);
+
+	ut_a(CloseHandle(event->handle));
+#else
+	ut_a(event);
+
+	/* This is to avoid freeing the mutex twice */
+	os_fast_mutex_free(&(event->os_mutex));
+
+	ut_a(0 == pthread_cond_destroy(&(event->cond_var)));
+#endif
+	/* Remove from the list of events */
+
+	UT_LIST_REMOVE(os_event_list, os_event_list, event);
+
+	os_event_count--;
+
+	ut_free(event);
+}
+
 /**************************************************************
 Frees an event object. */
 
@@ -456,6 +500,7 @@ os_mutex_create(
 
 	mutex_str->handle = mutex;
 	mutex_str->count = 0;
+	mutex_str->event = os_event_create(NULL);
 
 	if (os_sync_mutex_inited) {
 		/* When creating os_sync_mutex itself we cannot reserve it */
@@ -532,6 +577,10 @@ os_mutex_free(
 {
 	ut_a(mutex);
 
+	if (!os_sync_free_called) {
+		os_event_free_internal(mutex->event);
+	}
+
 	if (os_sync_mutex_inited) {
 		os_mutex_enter(os_sync_mutex);
 	}

innobase/row/row0mysql.c

@@ -2920,12 +2920,10 @@ do not allow the TRUNCATE. We also reserve the data dictionary latch. */
 			goto next_rec;
 		}
 
-		btr_pcur_store_position(&pcur, &mtr);
+		/* This call may commit and restart mtr
+		and reposition pcur. */
+		root_page_no = dict_truncate_index_tree(table, &pcur, &mtr);
 
-		/* This call may commit and restart mtr. */
-		root_page_no = dict_truncate_index_tree(table, rec, &mtr);
-
-		btr_pcur_restore_position(BTR_MODIFY_LEAF, &pcur, &mtr);
 		rec = btr_pcur_get_rec(&pcur);
 
 		if (root_page_no != FIL_NULL) {

innobase/row/row0sel.c

@@ -1323,6 +1323,12 @@ rec_loop:
 						ULINT_UNDEFINED, &heap);
 
 		if (srv_locks_unsafe_for_binlog) {
+
+			if (page_rec_is_supremum(rec)) {
+
+				goto next_rec;
+			}
+
 			lock_type = LOCK_REC_NOT_GAP;
 		} else {
 			lock_type = LOCK_ORDINARY;

innobase/srv/srv0start.c

@@ -1142,10 +1142,7 @@ innobase_start_or_create_for_mysql(void)
 
 #if defined(__NETWARE__)
 
-/* Create less event semaphores because Win 98/ME had difficulty creating
-40000 event semaphores.
-Comment from Novell, Inc.: also, these just take a lot of memory on
-NetWare. */
+	/* Comment from Novell, Inc.: These take a lot of memory on NetWare.*/
         srv_max_n_threads = 1000;
 #else
         if (srv_pool_size >= 1000 * 1024) {

innobase/sync/sync0arr.c

@@ -62,9 +62,6 @@ struct sync_cell_struct {
 	ibool		waiting;	/* TRUE if the thread has already
 					called sync_array_event_wait
 					on this cell */
-	ibool		event_set;	/* TRUE if the event is set */
-        os_event_t 	event;   	/* operating system event
-                                        semaphore handle */
 	time_t		reservation_time;/* time when the thread reserved
 					the wait cell */
 };
@@ -218,10 +215,7 @@ sync_array_create(
         for (i = 0; i < n_cells; i++) {
 		cell = sync_array_get_nth_cell(arr, i);        	
                 cell->wait_object = NULL;
-
-                /* Create an operating system event semaphore with no name */
-                cell->event = os_event_create(NULL);
-		cell->event_set = FALSE; /* it is created in reset state */
+		cell->waiting = FALSE;
 	}
 
 	return(arr);
@@ -235,19 +229,12 @@ sync_array_free(
 /*============*/
 	sync_array_t*	arr)	/* in, own: sync wait array */
 {
-        ulint           i;
-        sync_cell_t*   	cell;
 	ulint		protection;
 
         ut_a(arr->n_reserved == 0);
         
 	sync_array_validate(arr);
         
-        for (i = 0; i < arr->n_cells; i++) {
-		cell = sync_array_get_nth_cell(arr, i);        	
-		os_event_free(cell->event);
-        }
-
 	protection = arr->protection;
 
         /* Release the mutex protecting the wait array complex */
@@ -292,28 +279,20 @@ sync_array_validate(
         sync_array_exit(arr);
 }
 
-/***********************************************************************
-Puts the cell event in set state. */
-static
-void
-sync_cell_event_set(
-/*================*/
-	sync_cell_t*	cell)	/* in: array cell */
-{
-	os_event_set(cell->event);
-	cell->event_set = TRUE;
-}		
-
 /***********************************************************************
 Puts the cell event in reset state. */
 static
 void
 sync_cell_event_reset(
 /*==================*/
-	sync_cell_t*	cell)	/* in: array cell */
+	ulint		type,	/* in: lock type mutex/rw_lock */
+	void*		object) /* in: the rw_lock/mutex object */
 {
-	os_event_reset(cell->event);
-	cell->event_set = FALSE;
+	if (type == SYNC_MUTEX) {
+		os_event_reset(((mutex_t *) object)->event);
+	} else {
+		os_event_reset(((rw_lock_t *) object)->event);
+	}
 }		
 
 /**********************************************************************
@@ -346,14 +325,7 @@ sync_array_reserve_cell(
 
                 if (cell->wait_object == NULL) {
 
-                        /* Make sure the event is reset */
-			if (cell->event_set) {	
-                        	sync_cell_event_reset(cell);
-			}
-
-			cell->reservation_time = time(NULL);
-			cell->thread = os_thread_get_curr_id();
-
+			cell->waiting = FALSE;
 			cell->wait_object = object;
 
 			if (type == SYNC_MUTEX) {
@@ -363,7 +335,6 @@ sync_array_reserve_cell(
 			}
 				
 			cell->request_type = type;
-			cell->waiting = FALSE;
 			
 			cell->file = file;
 			cell->line = line;
@@ -373,6 +344,13 @@ sync_array_reserve_cell(
 			*index = i;
 
 			sync_array_exit(arr);
+
+			/* Make sure the event is reset */
+			sync_cell_event_reset(type, object);
+
+			cell->reservation_time = time(NULL);
+
+			cell->thread = os_thread_get_curr_id();
                         
                         return;
                 }
@@ -408,7 +386,12 @@ sync_array_wait_event(
 	ut_a(!cell->waiting);
 	ut_ad(os_thread_get_curr_id() == cell->thread);
 
-       	event = cell->event;
+	if (cell->request_type == SYNC_MUTEX) {
+		event = ((mutex_t*) cell->wait_object)->event;
+	} else {
+		event = ((rw_lock_t*) cell->wait_object)->event;
+	}
+
        	cell->waiting = TRUE;
 
 #ifdef UNIV_SYNC_DEBUG
@@ -510,10 +493,6 @@ sync_array_cell_print(
         if (!cell->waiting) {
 		fputs("wait has ended\n", file);
 	}
-
-        if (cell->event_set) {
-		fputs("wait is ending\n", file);
-	}
 }
 
 #ifdef UNIV_SYNC_DEBUG
@@ -623,7 +602,7 @@ sync_array_detect_deadlock(
 	
 	depth++;
 	
-	if (cell->event_set || !cell->waiting) {
+	if (!cell->waiting) {
 
 		return(FALSE); /* No deadlock here */
 	}
@@ -802,6 +781,7 @@ sync_array_free_cell(
 
         ut_a(cell->wait_object != NULL);
 
+	cell->waiting = FALSE;
 	cell->wait_object =  NULL;
 
 	ut_a(arr->n_reserved > 0);
@@ -811,44 +791,17 @@ sync_array_free_cell(
 }
 
 /**************************************************************************
-Looks for the cells in the wait array which refer to the wait object
-specified, and sets their corresponding events to the signaled state. In this
-way releases the threads waiting for the object to contend for the object.
-It is possible that no such cell is found, in which case does nothing. */
+Increments the signalled count. */
 
 void
-sync_array_signal_object(
-/*=====================*/
-	sync_array_t*	arr,	/* in: wait array */
-	void*		object)	/* in: wait object */
+sync_array_object_signalled(
+/*========================*/
+	sync_array_t*	arr)	/* in: wait array */
 {
-        sync_cell_t*   	cell;
-        ulint           count;
-        ulint           i;
-
         sync_array_enter(arr);
 
 	arr->sg_count++;
 
-	i = 0;
-	count = 0;
-
-        while (count < arr->n_reserved) {
-
-        	cell = sync_array_get_nth_cell(arr, i);
-
-                if (cell->wait_object != NULL) {
-
-                        count++;
-                        if (cell->wait_object == object) {
-
-                        	sync_cell_event_set(cell);
-                        }
-                }
-
-                i++;
-        }
-
         sync_array_exit(arr);
 }
 
@@ -881,7 +834,17 @@ sync_arr_wake_threads_if_sema_free(void)
 
                         if (sync_arr_cell_can_wake_up(cell)) {
 
-                        	sync_cell_event_set(cell);
+				if (cell->request_type == SYNC_MUTEX) {
+					mutex_t*	mutex;
+
+					mutex = cell->wait_object;
+					os_event_set(mutex->event);
+				} else {
+					rw_lock_t*	lock;
+
+					lock = cell->wait_object;
+					os_event_set(lock->event);
+				}
                         }
                 }
 
@@ -911,7 +874,7 @@ sync_array_print_long_waits(void)
 
         	cell = sync_array_get_nth_cell(sync_primary_wait_array, i);
 
-                if (cell->wait_object != NULL
+		if (cell->wait_object != NULL && cell->waiting
 		    && difftime(time(NULL), cell->reservation_time) > 240) {
 			fputs("InnoDB: Warning: a long semaphore wait:\n",
 				stderr);
@@ -919,7 +882,7 @@ sync_array_print_long_waits(void)
 			noticed = TRUE;
                 }
 
-                if (cell->wait_object != NULL
+		if (cell->wait_object != NULL && cell->waiting
 		    && difftime(time(NULL), cell->reservation_time)
 		    > fatal_timeout) {
 			fatal = TRUE;

innobase/sync/sync0rw.c

@@ -128,6 +128,7 @@ rw_lock_create_func(
 	lock->last_x_file_name = "not yet reserved";
 	lock->last_s_line = 0;
 	lock->last_x_line = 0;
+	lock->event = os_event_create(NULL);
 
 	mutex_enter(&rw_lock_list_mutex);
 	
@@ -163,6 +164,7 @@ rw_lock_free(
 	mutex_free(rw_lock_get_mutex(lock));
 
 	mutex_enter(&rw_lock_list_mutex);
+	os_event_free(lock->event);
 
 	if (UT_LIST_GET_PREV(list, lock)) {
 		ut_a(UT_LIST_GET_PREV(list, lock)->magic_n == RW_LOCK_MAGIC_N);

innobase/sync/sync0sync.c

@@ -212,6 +212,7 @@ mutex_create_func(
 	os_fast_mutex_init(&(mutex->os_fast_mutex));
 	mutex->lock_word = 0;
 #endif
+	mutex->event = os_event_create(NULL);
 	mutex_set_waiters(mutex, 0);
 	mutex->magic_n = MUTEX_MAGIC_N;
 #ifdef UNIV_SYNC_DEBUG
@@ -288,6 +289,8 @@ mutex_free(
 		mutex_exit(&mutex_list_mutex);
 	}
 
+	os_event_free(mutex->event);
+
 #if !defined(_WIN32) || !defined(UNIV_CAN_USE_X86_ASSEMBLER) 
 	os_fast_mutex_free(&(mutex->os_fast_mutex));
 #endif
@@ -564,8 +567,8 @@ mutex_signal_object(
 
 	/* The memory order of resetting the waiters field and
 	signaling the object is important. See LEMMA 1 above. */
-
-	sync_array_signal_object(sync_primary_wait_array, mutex);
+	os_event_set(mutex->event);
+	sync_array_object_signalled(sync_primary_wait_array);
 }
 
 #ifdef UNIV_SYNC_DEBUG
@@ -1114,6 +1117,7 @@ sync_thread_add_level(
 		ut_a(sync_thread_levels_g(array, SYNC_PURGE_SYS));
 	} else if (level == SYNC_TREE_NODE) {
 		ut_a(sync_thread_levels_contain(array, SYNC_INDEX_TREE)
+		     || sync_thread_levels_contain(array, SYNC_DICT_OPERATION)
 		     || sync_thread_levels_g(array, SYNC_TREE_NODE - 1));
 	} else if (level == SYNC_TREE_NODE_FROM_HASH) {
 		ut_a(1);