branches/innodb+: Merge revisions 2344:2454 from branches/zip

CMakeLists.txt

@@ -60,4 +60,5 @@ SET(INNOBASE_SOURCES  btr/btr0btr.c btr/btr0cur.c btr/btr0pcur.c btr/btr0sea.c
 
 IF(NOT SOURCE_SUBLIBS)
   ADD_LIBRARY(innobase ${INNOBASE_SOURCES})
+  ADD_DEPENDENCIES(innobase GenError)
 ENDIF(NOT SOURCE_SUBLIBS)

ChangeLog

+2008-05-06	The InnoDB Team
+
+	* handler/ha_innodb.cc, include/srv0srv.h, include/sync0sync.h,
+	include/trx0sys.h, mysql-test/innodb-zip.result,
+	mysql-test/innodb-zip.test, srv/srv0srv.c, srv/srv0start.c,
+	sync/sync0sync.c, trx/trx0sys.c:
+	Implement the system tablespace tagging
+
+	* handler/ha_innodb.cc, handler/i_s.cc, include/univ.i,
+	srv/srv0start.c:
+	Add InnoDB version in INFORMATION_SCHEMA.PLUGINS.PLUGIN_VERSION,
+	in the startup message and in a server variable innodb_version.
+
+	* sync/sync0sync.c:
+	Fix a bug in the sync debug code where a lock with level
+	SYNC_LEVEL_VARYING would cause an assertion failure when a thread
+	tried to release it.
+
+2008-04-30	The InnoDB Team
+
+	* Makefile.am:
+	Fix Bug#36434 ha_innodb.so is installed in the wrong directory
+
+	* handler/ha_innodb.cc:
+	Merge change from MySQL (Fix Bug#35406 5.1-opt crashes on select from
+	I_S.REFERENTIAL_CONSTRAINTS):
+	ChangeSet@1.2563, 2008-03-18 19:42:04+04:00, gluh@mysql.com +1 -0
+
+	* scripts/install_innodb_plugins.sql:
+	Added
+
+	* mysql-test/innodb.result:
+	Merge change from MySQL (this fixes the failing innodb test):
+	ChangeSet@1.1810.3601.4, 2008-02-07 02:33:21+04:00
+
+	* row/row0sel.c:
+	Fix Bug#35226 RBR event crashes slave
+
+	* handler/ha_innodb.cc:
+	Change the fix for Bug#32440 to show bytes instead of kilobytes in
+	INFORMATION_SCHEMA.TABLES.DATA_FREE
+
+	* handler/ha_innodb.cc, mysql-test/innodb.result,
+	mysql-test/innodb.test:
+	Fix Bug#29507 TRUNCATE shows to many rows effected
+
+	* handler/ha_innodb.cc, mysql-test/innodb.result,
+	mysql-test/innodb.test:
+	Fix Bug#35537 Innodb doesn't increment handler_update and
+	handler_delete
+
+2008-04-29	The InnoDB Team
+
+	* handler/i_s.cc, include/srv0start.h, srv/srv0start.c: 
+	Fix Bug#36310 InnoDB plugin crash
+
+2008-04-23	The InnoDB Team
+
+	* mysql-test/innodb_bug36169.result, mysql-test/innodb_bug36169.test,
+	row/row0mysql.c:
+	Fix Bug#36169 create innodb compressed table with too large row size
+	crashed
+
+	* (outside the source tree):
+	Fix Bug#36222 New InnoDB plugin 1.0 has wrong MKDIR_P defined in
+	Makefile.in
+
+2008-04-15	The InnoDB Team
+
+	InnoDB Plugin 1.0.0 released

Makefile.am

@@ -15,20 +15,21 @@
 
 # Process this file with automake to create Makefile.in
 
-MYSQLDATAdir =          $(localstatedir)
-MYSQLSHAREdir =         $(pkgdatadir)
-MYSQLBASEdir=           $(prefix)
-MYSQLLIBdir=            $(pkglibdir)
-INCLUDES =              -I$(top_srcdir)/include -I$(top_builddir)/include \
+MYSQLDATAdir=		$(localstatedir)
+MYSQLSHAREdir=		$(pkgdatadir)
+MYSQLBASEdir=		$(prefix)
+MYSQLLIBdir=		$(pkglibdir)
+pkgplugindir=		$(pkglibdir)/plugin
+INCLUDES=		-I$(top_srcdir)/include -I$(top_builddir)/include \
 			-I$(top_srcdir)/regex \
 			-I$(top_srcdir)/storage/innobase/include \
 			-I$(top_srcdir)/sql \
-                        -I$(srcdir)
+			-I$(srcdir)
 
-DEFS =			@DEFS@
+DEFS=			@DEFS@
 
 
-noinst_HEADERS = include/btr0btr.h include/btr0btr.ic			\
+noinst_HEADERS=		include/btr0btr.h include/btr0btr.ic		\
 			include/btr0cur.h include/btr0cur.ic		\
 			include/btr0pcur.h include/btr0pcur.ic		\
 			include/btr0sea.h include/btr0sea.ic		\
@@ -132,9 +133,9 @@ noinst_HEADERS = include/btr0btr.h include/btr0btr.ic			\
 			include/handler0alter.h				\
 			handler/i_s.h include/ut0rbt.h
 
-EXTRA_LIBRARIES =	libinnobase.a
-noinst_LIBRARIES =	@plugin_innobase_static_target@
-libinnobase_a_SOURCES =	btr/btr0btr.c btr/btr0cur.c btr/btr0pcur.c	\
+EXTRA_LIBRARIES=	libinnobase.a
+noinst_LIBRARIES=	@plugin_innobase_static_target@
+libinnobase_a_SOURCES=	btr/btr0btr.c btr/btr0cur.c btr/btr0pcur.c	\
 			btr/btr0sea.c buf/buf0buddy.c			\
 			buf/buf0buf.c buf/buf0flu.c			\
 			buf/buf0lru.c buf/buf0rea.c data/data0data.c	\
@@ -175,18 +176,18 @@ libinnobase_a_SOURCES =	btr/btr0btr.c btr/btr0cur.c btr/btr0pcur.c	\
 			handler/i_s.cc					\
 			handler/mysql_addons.cc ut/ut0rbt.c
 
-libinnobase_a_CXXFLAGS=	$(AM_CFLAGS) $(INNODB_CFLAGS)
-libinnobase_a_CFLAGS  =	$(AM_CFLAGS) $(INNODB_CFLAGS)
+libinnobase_a_CXXFLAGS=	$(AM_CFLAGS)
+libinnobase_a_CFLAGS=	$(AM_CFLAGS)
 
-EXTRA_LTLIBRARIES =	ha_innodb.la
-pkglib_LTLIBRARIES =	@plugin_innobase_shared_target@
+EXTRA_LTLIBRARIES=	ha_innodb.la
+pkgplugin_LTLIBRARIES=	@plugin_innobase_shared_target@
 
-ha_innodb_la_LDFLAGS =	-module -rpath $(MYSQLLIBdir)
-ha_innodb_la_CXXFLAGS=	$(AM_CFLAGS) $(INNODB_CFLAGS) $(INNODB_DYNAMIC_CFLAGS)
-ha_innodb_la_CFLAGS  =	$(AM_CFLAGS) $(INNODB_CFLAGS) $(INNODB_DYNAMIC_CFLAGS)
-ha_innodb_la_SOURCES =	$(libinnobase_a_SOURCES)
+ha_innodb_la_LDFLAGS=	-module -rpath $(pkgplugindir)
+ha_innodb_la_CXXFLAGS=	$(AM_CFLAGS) $(INNODB_DYNAMIC_CFLAGS)
+ha_innodb_la_CFLAGS=	$(AM_CFLAGS) $(INNODB_DYNAMIC_CFLAGS)
+ha_innodb_la_SOURCES=	$(libinnobase_a_SOURCES)
 
-EXTRA_DIST =		CMakeLists.txt plug.in \
+EXTRA_DIST=		CMakeLists.txt plug.in \
 			pars/make_bison.sh pars/make_flex.sh \
 			pars/pars0grm.y pars/pars0lex.l
 

btr/btr0btr.c

@@ -1807,7 +1807,35 @@ btr_page_split_and_insert(
 	} else {
 		direction = FSP_UP;
 		hint_page_no = page_no + 1;
-		split_rec = page_get_middle_rec(page);
+
+		if (page_get_n_recs(page) == 1) {
+			page_cur_t	pcur;
+
+			/* There is only one record in the index page
+			therefore we can't split the node in the middle
+			by default. We need to determine whether the
+			new record will be inserted to the left or right. */
+
+			/* Read the first (and only) record in the page. */
+			page_cur_set_before_first(block, &pcur);
+			page_cur_move_to_next(&pcur);
+			first_rec = page_cur_get_rec(&pcur);
+
+			offsets = rec_get_offsets(
+				first_rec, cursor->index, offsets,
+				n_uniq, &heap);
+
+			/* If the new record is less than the existing record
+			the the split in the middle will copy the existing
+			record to the new node. */
+			if (cmp_dtuple_rec(tuple, first_rec, offsets) < 0) {
+				split_rec = page_get_middle_rec(page);
+			} else {
+				split_rec = NULL;
+			}
+		} else {
+			split_rec = page_get_middle_rec(page);
+		}
 	}
 
 	/* 2. Allocate a new page to the index */
@@ -2945,9 +2973,9 @@ static
 void
 btr_index_rec_validate_report(
 /*==========================*/
-	const page_t*	page,	/* in: index page */
-	const rec_t*	rec,	/* in: index record */
-	dict_index_t*	index)	/* in: index */
+	const page_t*		page,	/* in: index page */
+	const rec_t*		rec,	/* in: index record */
+	const dict_index_t*	index)	/* in: index */
 {
 	fputs("InnoDB: Record in ", stderr);
 	dict_index_name_print(stderr, NULL, index);
@@ -2962,17 +2990,17 @@ UNIV_INTERN
 ibool
 btr_index_rec_validate(
 /*===================*/
-					/* out: TRUE if ok */
-	rec_t*		rec,		/* in: index record */
-	dict_index_t*	index,		/* in: index */
-	ibool		dump_on_error)	/* in: TRUE if the function
-					should print hex dump of record
-					and page on error */
+						/* out: TRUE if ok */
+	const rec_t*		rec,		/* in: index record */
+	const dict_index_t*	index,		/* in: index */
+	ibool			dump_on_error)	/* in: TRUE if the function
+						should print hex dump of record
+						and page on error */
 {
 	ulint		len;
 	ulint		n;
 	ulint		i;
-	page_t*		page;
+	const page_t*	page;
 	mem_heap_t*	heap	= NULL;
 	ulint		offsets_[REC_OFFS_NORMAL_SIZE];
 	ulint*		offsets	= offsets_;

btr/btr0cur.c

@@ -1984,7 +1984,7 @@ btr_cur_optimistic_update(
 	corresponding to new_entry is latched in mtr.
 	Thus the following call is safe. */
 	row_upd_index_replace_new_col_vals_index_pos(new_entry, index, update,
-						     FALSE, NULL, heap);
+						     FALSE, heap);
 	old_rec_size = rec_offs_size(offsets);
 	new_rec_size = rec_get_converted_size(index, new_entry, 0);
 
@@ -2261,7 +2261,7 @@ btr_cur_pessimistic_update(
 	purge would also have removed the clustered index record
 	itself.  Thus the following call is safe. */
 	row_upd_index_replace_new_col_vals_index_pos(new_entry, index, update,
-						     FALSE, *heap, *heap);
+						     FALSE, *heap);
 	if (!(flags & BTR_KEEP_SYS_FLAG)) {
 		row_upd_index_entry_sys_field(new_entry, index, DATA_ROLL_PTR,
 					      roll_ptr);
@@ -3132,7 +3132,7 @@ btr_cur_add_path_info(
 /***********************************************************************
 Estimates the number of rows in a given index range. */
 UNIV_INTERN
-ib_longlong
+ib_int64_t
 btr_estimate_n_rows_in_range(
 /*=========================*/
 				/* out: estimated number of rows */
@@ -3150,7 +3150,7 @@ btr_estimate_n_rows_in_range(
 	ibool		diverged;
 	ibool		diverged_lot;
 	ulint		divergence_level;
-	ib_longlong	n_rows;
+	ib_int64_t	n_rows;
 	ulint		i;
 	mtr_t		mtr;
 
@@ -3293,7 +3293,7 @@ btr_estimate_number_of_different_key_vals(
 	ulint		n_cols;
 	ulint		matched_fields;
 	ulint		matched_bytes;
-	ib_longlong*	n_diff;
+	ib_int64_t*	n_diff;
 	ulint		not_empty_flag	= 0;
 	ulint		total_external_size = 0;
 	ulint		i;
@@ -3310,7 +3310,7 @@ btr_estimate_number_of_different_key_vals(
 
 	n_cols = dict_index_get_n_unique(index);
 
-	n_diff = mem_zalloc((n_cols + 1) * sizeof(ib_longlong));
+	n_diff = mem_zalloc((n_cols + 1) * sizeof(ib_int64_t));
 
 	/* We sample some pages in the index to get an estimate */
 
@@ -3413,7 +3413,7 @@ btr_estimate_number_of_different_key_vals(
 	for (j = 0; j <= n_cols; j++) {
 		index->stat_n_diff_key_vals[j]
 			= ((n_diff[j]
-			    * (ib_longlong)index->stat_n_leaf_pages
+			    * (ib_int64_t)index->stat_n_leaf_pages
 			    + srv_stats_sample - 1
 			    + total_external_size
 			    + not_empty_flag)

buf/buf0buddy.c

@@ -24,15 +24,9 @@ Created December 2006 by Marko Makela
 Protected by buf_pool_mutex. */
 static ulint buf_buddy_n_frames;
 #endif /* UNIV_DEBUG */
-/** Counts of blocks allocated from the buddy system.
+/** Statistics of the buddy system, indexed by block size.
 Protected by buf_pool_mutex. */
-UNIV_INTERN ulint buf_buddy_used[BUF_BUDDY_SIZES + 1];
-/** Counts of blocks relocated by the buddy system.
-Protected by buf_pool_mutex. */
-UNIV_INTERN ib_uint64_t buf_buddy_relocated[BUF_BUDDY_SIZES + 1];
-/** Durations of block relocations.
-Protected by buf_pool_mutex. */
-UNIV_INTERN ullint buf_buddy_relocated_duration[BUF_BUDDY_SIZES + 1];
+UNIV_INTERN buf_buddy_stat_t buf_buddy_stat[BUF_BUDDY_SIZES + 1];
 
 /**************************************************************************
 Get the offset of the buddy of a compressed page frame. */
@@ -320,7 +314,7 @@ buf_buddy_alloc_low(
 	block = buf_buddy_alloc_from(block->frame, i, BUF_BUDDY_SIZES);
 
 func_exit:
-	buf_buddy_used[i]++;
+	buf_buddy_stat[i].used++;
 	return(block);
 }
 
@@ -463,9 +457,13 @@ buf_buddy_relocate(
 			mutex_exit(mutex);
 success:
 			UNIV_MEM_INVALID(src, size);
-			buf_buddy_relocated[i]++;
-			buf_buddy_relocated_duration[i]
-				+= ut_time_us(NULL) - usec;
+			{
+				buf_buddy_stat_t*	buddy_stat
+					= &buf_buddy_stat[i];
+				buddy_stat->relocated++;
+				buddy_stat->relocated_usec
+					+= ut_time_us(NULL) - usec;
+			}
 			return(TRUE);
 		}
 
@@ -498,9 +496,9 @@ buf_buddy_free_low(
 	ut_ad(buf_pool_mutex_own());
 	ut_ad(!mutex_own(&buf_pool_zip_mutex));
 	ut_ad(i <= BUF_BUDDY_SIZES);
-	ut_ad(buf_buddy_used[i] > 0);
+	ut_ad(buf_buddy_stat[i].used > 0);
 
-	buf_buddy_used[i]--;
+	buf_buddy_stat[i].used--;
 recombine:
 	UNIV_MEM_ASSERT_AND_ALLOC(buf, BUF_BUDDY_LOW << i);
 	ut_d(((buf_page_t*) buf)->state = BUF_BLOCK_ZIP_FREE);

buf/buf0buf.c

@@ -2646,7 +2646,7 @@ buf_page_init_for_read(
 	ulint		space,	/* in: space id */
 	ulint		zip_size,/* in: compressed page size, or 0 */
 	ibool		unzip,	/* in: TRUE=request uncompressed page */
-	ib_longlong	tablespace_version,/* in: prevents reading from a wrong
+	ib_int64_t	tablespace_version,/* in: prevents reading from a wrong
 				version of the tablespace in case we have done
 				DISCARD + IMPORT */
 	ulint		offset)	/* in: page number */
@@ -3608,7 +3608,7 @@ buf_get_latched_pages_number(void)
 	for (b = UT_LIST_GET_FIRST(buf_pool->zip_clean); b;
 	     b = UT_LIST_GET_NEXT(list, b)) {
 		ut_a(buf_page_get_state(b) == BUF_BLOCK_ZIP_PAGE);
-		ut_a(buf_page_get_io_fix(b) == BUF_IO_NONE);
+		ut_a(buf_page_get_io_fix(b) != BUF_IO_WRITE);
 
 		if (b->buf_fix_count != 0
 		    || buf_page_get_io_fix(b) != BUF_IO_NONE) {

buf/buf0flu.c

@@ -27,12 +27,6 @@ Created 11/11/1995 Heikki Tuuri
 #include "trx0sys.h"
 #include "srv0srv.h"
 
-/* When flushed, dirty blocks are searched in neighborhoods of this size, and
-flushed along with the original page. */
-
-#define BUF_FLUSH_AREA		ut_min(BUF_READ_AHEAD_AREA,\
-		buf_pool->curr_size / 16)
-
 #if defined UNIV_DEBUG || defined UNIV_BUF_DEBUG
 /**********************************************************************
 Validates the flush list. */
@@ -915,15 +909,21 @@ buf_flush_try_neighbors(
 
 	ut_ad(flush_type == BUF_FLUSH_LRU || flush_type == BUF_FLUSH_LIST);
 
-	low = (offset / BUF_FLUSH_AREA) * BUF_FLUSH_AREA;
-	high = (offset / BUF_FLUSH_AREA + 1) * BUF_FLUSH_AREA;
-
 	if (UT_LIST_GET_LEN(buf_pool->LRU) < BUF_LRU_OLD_MIN_LEN) {
 		/* If there is little space, it is better not to flush any
 		block except from the end of the LRU list */
 
 		low = offset;
 		high = offset + 1;
+	} else {
+		/* When flushed, dirty blocks are searched in neighborhoods of
+		this size, and flushed along with the original page. */
+
+		ulint	buf_flush_area	= ut_min(BUF_READ_AHEAD_AREA,
+						 buf_pool->curr_size / 16);
+
+		low = (offset / buf_flush_area) * buf_flush_area;
+		high = (offset / buf_flush_area + 1) * buf_flush_area;
 	}
 
 	/* fprintf(stderr, "Flush area: low %lu high %lu\n", low, high); */

buf/buf0rea.c

@@ -30,13 +30,13 @@ the accessed pages when deciding whether to read-ahead */
 
 /* 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)
+#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
 
 /* The linear read-ahead threshold */
-#define BUF_READ_AHEAD_LINEAR_THRESHOLD	(3 * BUF_READ_AHEAD_LINEAR_AREA / 8)
+#define LINEAR_AREA_THRESHOLD_COEF	5 / 8
 
 /* If there are buf_pool->curr_size per the number below pending reads, then
 read-ahead is not done: this is to prevent flooding the buffer pool with
@@ -67,7 +67,7 @@ buf_read_page_low(
 	ulint	space,	/* in: space id */
 	ulint	zip_size,/* in: compressed page size, or 0 */
 	ibool	unzip,	/* in: TRUE=request uncompressed page */
-	ib_longlong tablespace_version, /* in: if the space memory object has
+	ib_int64_t tablespace_version, /* in: if the space memory object has
 			this timestamp different from what we are giving here,
 			treat the tablespace as dropped; this is a timestamp we
 			use to stop dangling page reads from a tablespace
@@ -177,7 +177,7 @@ buf_read_ahead_random(
 	ulint	offset)	/* in: page number of a page which the current thread
 			wants to access */
 {
-	ib_longlong	tablespace_version;
+	ib_int64_t	tablespace_version;
 	ulint		recent_blocks	= 0;
 	ulint		count;
 	ulint		LRU_recent_limit;
@@ -185,6 +185,7 @@ buf_read_ahead_random(
 	ulint		low, high;
 	ulint		err;
 	ulint		i;
+	ulint		buf_read_ahead_random_area;
 
 	if (srv_startup_is_before_trx_rollback_phase) {
 		/* No read-ahead to avoid thread deadlocks */
@@ -207,10 +208,12 @@ buf_read_ahead_random(
 
 	tablespace_version = fil_space_get_version(space);
 
-	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;
+	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);
@@ -324,7 +327,7 @@ buf_read_page(
 	ulint	zip_size,/* in: compressed page size in bytes, or 0 */
 	ulint	offset)	/* in: page number */
 {
-	ib_longlong	tablespace_version;
+	ib_int64_t	tablespace_version;
 	ulint		count;
 	ulint		count2;
 	ulint		err;
@@ -392,7 +395,7 @@ buf_read_ahead_linear(
 	ulint	offset)	/* in: page number of a page; NOTE: the current thread
 			must want access to this page (see NOTE 3 above) */
 {
-	ib_longlong	tablespace_version;
+	ib_int64_t	tablespace_version;
 	buf_page_t*	bpage;
 	buf_frame_t*	frame;
 	buf_page_t*	pred_bpage	= NULL;
@@ -406,16 +409,18 @@ buf_read_ahead_linear(
 	ulint		low, high;
 	ulint		err;
 	ulint		i;
+	const ulint	buf_read_ahead_linear_area
+		= BUF_READ_AHEAD_LINEAR_AREA;
 
 	if (UNIV_UNLIKELY(srv_startup_is_before_trx_rollback_phase)) {
 		/* No read-ahead to avoid thread deadlocks */
 		return(0);
 	}
 
-	low  = (offset / BUF_READ_AHEAD_LINEAR_AREA)
-		* BUF_READ_AHEAD_LINEAR_AREA;
-	high = (offset / BUF_READ_AHEAD_LINEAR_AREA + 1)
-		* BUF_READ_AHEAD_LINEAR_AREA;
+	low  = (offset / buf_read_ahead_linear_area)
+		* buf_read_ahead_linear_area;
+	high = (offset / buf_read_ahead_linear_area + 1)
+		* buf_read_ahead_linear_area;
 
 	if ((offset != low) && (offset != high - 1)) {
 		/* This is not a border page of the area: return */
@@ -486,8 +491,8 @@ buf_read_ahead_linear(
 		}
 	}
 
-	if (fail_count > BUF_READ_AHEAD_LINEAR_AREA
-	    - BUF_READ_AHEAD_LINEAR_THRESHOLD) {
+	if (fail_count > buf_read_ahead_linear_area
+	    * LINEAR_AREA_THRESHOLD_COEF) {
 		/* Too many failures: return */
 
 		buf_pool_mutex_exit();
@@ -544,10 +549,10 @@ buf_read_ahead_linear(
 		return(0);
 	}
 
-	low  = (new_offset / BUF_READ_AHEAD_LINEAR_AREA)
-		* BUF_READ_AHEAD_LINEAR_AREA;
-	high = (new_offset / BUF_READ_AHEAD_LINEAR_AREA + 1)
-		* BUF_READ_AHEAD_LINEAR_AREA;
+	low  = (new_offset / buf_read_ahead_linear_area)
+		* buf_read_ahead_linear_area;
+	high = (new_offset / buf_read_ahead_linear_area + 1)
+		* buf_read_ahead_linear_area;
 
 	if ((new_offset != low) && (new_offset != high - 1)) {
 		/* This is not a border page of the area: return */
@@ -638,7 +643,7 @@ buf_read_ibuf_merge_pages(
 					to get read in, before this
 					function returns */
 	const ulint*	space_ids,	/* in: array of space ids */
-	const ib_longlong* space_versions,/* in: the spaces must have
+	const ib_int64_t* space_versions,/* in: the spaces must have
 					this version number
 					(timestamp), otherwise we
 					discard the read; we use this
@@ -723,7 +728,7 @@ buf_read_recv_pages(
 	ulint		n_stored)	/* in: number of page numbers
 					in the array */
 {
-	ib_longlong	tablespace_version;
+	ib_int64_t	tablespace_version;
 	ulint		count;
 	ulint		err;
 	ulint		i;

data/data0data.c

@@ -561,11 +561,21 @@ dtuple_convert_big_rec(
 	dict_field_t*	ifield;
 	ulint		size;
 	ulint		n_fields;
+	ulint		local_len;
+	ulint		local_prefix_len;
 
 	if (UNIV_UNLIKELY(!dict_index_is_clust(index))) {
 		return(NULL);
 	}
 
+	if (dict_table_get_format(index->table) < DICT_TF_FORMAT_ZIP) {
+		/* up to MySQL 5.1: store a 768-byte prefix locally */
+		local_len = BTR_EXTERN_FIELD_REF_SIZE + DICT_MAX_INDEX_COL_LEN;
+	} else {
+		/* new-format table: do not store any BLOB prefix locally */
+		local_len = BTR_EXTERN_FIELD_REF_SIZE;
+	}
+
 	ut_a(dtuple_check_typed_no_assert(entry));
 
 	size = rec_get_converted_size(index, entry, *n_ext);
@@ -598,9 +608,11 @@ dtuple_convert_big_rec(
 							     *n_ext),
 				      dict_table_is_comp(index->table),
 				      dict_table_zip_size(index->table))) {
-		ulint	i;
-		ulint	longest		= 0;
-		ulint	longest_i	= ULINT_MAX;
+		ulint			i;
+		ulint			longest		= 0;
+		ulint			longest_i	= ULINT_MAX;
+		byte*			data;
+		big_rec_field_t*	b;
 
 		for (i = dict_index_get_n_unique_in_tree(index);
 		     i < dtuple_get_n_fields(entry); i++) {
@@ -615,13 +627,13 @@ dtuple_convert_big_rec(
 			if (ifield->fixed_len
 			    || dfield_is_null(dfield)
 			    || dfield_is_ext(dfield)
+			    || dfield_get_len(dfield) <= local_len
 			    || dfield_get_len(dfield)
 			    <= BTR_EXTERN_FIELD_REF_SIZE * 2) {
 				goto skip_field;
 			}
 
-			savings = dfield_get_len(dfield)
-				- BTR_EXTERN_FIELD_REF_SIZE;
+			savings = dfield_get_len(dfield) - local_len;
 
 			/* Check that there would be savings */
 			if (longest >= savings) {
@@ -651,25 +663,32 @@ dtuple_convert_big_rec(
 
 		dfield = dtuple_get_nth_field(entry, longest_i);
 		ifield = dict_index_get_nth_field(index, longest_i);
-		vector->fields[n_fields].field_no = longest_i;
+		local_prefix_len = local_len - BTR_EXTERN_FIELD_REF_SIZE;
 
-		vector->fields[n_fields].len = dfield_get_len(dfield);
+		b = &vector->fields[n_fields];
+		b->field_no = longest_i;
+		b->len = dfield_get_len(dfield) - local_prefix_len;
+		b->data = (char*) dfield_get_data(dfield) + local_prefix_len;
 
-		vector->fields[n_fields].data = dfield_get_data(dfield);
+		/* Allocate the locally stored part of the column. */
+		data = mem_heap_alloc(heap, local_len);
 
-		/* Set the extern field reference in dfield to zero */
-		dfield_set_data(dfield,
-				mem_heap_zalloc(heap,
-						BTR_EXTERN_FIELD_REF_SIZE),
-				BTR_EXTERN_FIELD_REF_SIZE);
-		dfield_set_ext(dfield);
+		/* Copy the local prefix. */
+		memcpy(data, dfield_get_data(dfield), local_prefix_len);
+		/* Clear the extern field reference (BLOB pointer). */
+		memset(data + local_prefix_len, 0, BTR_EXTERN_FIELD_REF_SIZE);
 #if 0
 		/* The following would fail the Valgrind checks in
 		page_cur_insert_rec_low() and page_cur_insert_rec_zip().
 		The BLOB pointers in the record will be initialized after
 		the record and the BLOBs have been written. */
-		UNIV_MEM_ALLOC(dfield->data, BTR_EXTERN_FIELD_REF_SIZE);
+		UNIV_MEM_ALLOC(data + local_prefix_len,
+			       BTR_EXTERN_FIELD_REF_SIZE);
 #endif
+
+		dfield_set_data(dfield, data, local_len);
+		dfield_set_ext(dfield);
+
 		n_fields++;
 		(*n_ext)++;
 		ut_ad(n_fields < dtuple_get_n_fields(entry));
@@ -692,16 +711,26 @@ dtuple_convert_back_big_rec(
 	big_rec_t*	vector)	/* in, own: big rec vector; it is
 				freed in this function */
 {
-	dfield_t*	dfield;
-	ulint		i;
+	big_rec_field_t*		b	= vector->fields;
+	const big_rec_field_t* const	end	= b + vector->n_fields;
+
+	for (; b < end; b++) {
+		dfield_t*	dfield;
+		ulint		local_len;
 
-	for (i = 0; i < vector->n_fields; i++) {
+		dfield = dtuple_get_nth_field(entry, b->field_no);
+		local_len = dfield_get_len(dfield);
 
-		dfield = dtuple_get_nth_field(entry,
-					      vector->fields[i].field_no);
 		ut_ad(dfield_is_ext(dfield));
+		ut_ad(local_len >= BTR_EXTERN_FIELD_REF_SIZE);
+
+		local_len -= BTR_EXTERN_FIELD_REF_SIZE;
+
+		ut_ad(local_len <= DICT_MAX_INDEX_COL_LEN);
+
 		dfield_set_data(dfield,
-				vector->fields[i].data, vector->fields[i].len);
+				(char*) b->data - local_len,
+				b->len + local_len);
 	}
 
 	mem_heap_free(vector->heap);

dict/dict0boot.c

@@ -252,7 +252,10 @@ dict_boot(void)
 
 	dict_mem_table_add_col(table, heap, "NAME", DATA_BINARY, 0, 0);
 	dict_mem_table_add_col(table, heap, "ID", DATA_BINARY, 0, 0);
+	/* ROW_FORMAT = (N_COLS >> 31) ? COMPACT : REDUNDANT */
 	dict_mem_table_add_col(table, heap, "N_COLS", DATA_INT, 0, 4);
+	/* TYPE is either DICT_TABLE_ORDINARY, or (TYPE & DICT_TF_COMPACT)
+	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);
 	dict_mem_table_add_col(table, heap, "MIX_LEN", DATA_INT, 0, 4);

dict/dict0crea.c

@@ -77,11 +77,13 @@ dict_create_sys_tables_tuple(
 	dfield = dtuple_get_nth_field(entry, 3);
 
 	ptr = mem_heap_alloc(heap, 4);
-	if (table->flags & DICT_TF_COMPRESSED_MASK) {
+	if (table->flags & ~DICT_TF_COMPACT) {
 		ut_a(table->flags & DICT_TF_COMPACT);
-		mach_write_to_4(ptr, DICT_TABLE_COMPRESSED_BASE
-				+ ((table->flags & DICT_TF_COMPRESSED_MASK)
-				>> DICT_TF_COMPRESSED_SHIFT));
+		ut_a(dict_table_get_format(table) >= DICT_TF_FORMAT_ZIP);
+		ut_a((table->flags & DICT_TF_ZSSIZE_MASK)
+		     <= (DICT_TF_ZSSIZE_MAX << DICT_TF_ZSSIZE_SHIFT));
+		ut_a(!(table->flags & (~0 << DICT_TF_BITS)));
+		mach_write_to_4(ptr, table->flags);
 	} else {
 		mach_write_to_4(ptr, DICT_TABLE_ORDINARY);
 	}
@@ -255,9 +257,13 @@ dict_build_table_def_step(
 			is_path = FALSE;
 		}
 
+		ut_ad(dict_table_get_format(table) <= DICT_TF_FORMAT_MAX);
+		ut_ad(!dict_table_zip_size(table)
+		      || dict_table_get_format(table) >= DICT_TF_FORMAT_ZIP);
+
 		error = fil_create_new_single_table_tablespace(
 			&space, path_or_name, is_path,
-			dict_table_zip_size(table),
+			table->flags == DICT_TF_COMPACT ? 0 : table->flags,
 			FIL_IBD_FILE_INITIAL_SIZE);
 		table->space = (unsigned int) space;
 
@@ -272,10 +278,8 @@ dict_build_table_def_step(
 
 		mtr_commit(&mtr);
 	} else {
-		/* Create in the system tablespace: disallow compression */
-		if (table->flags & DICT_TF_COMPRESSED_MASK) {
-			return(DB_TABLE_ZIP_NO_IBD);
-		}
+		/* Create in the system tablespace: disallow new features */
+		table->flags &= DICT_TF_COMPACT;
 	}
 
 	row = dict_create_sys_tables_tuple(table, node->heap);

dict/dict0dict.c

@@ -342,7 +342,7 @@ void
 dict_table_autoinc_initialize(
 /*==========================*/
 	dict_table_t*	table,	/* in/out: table */
-	ib_longlong	value)	/* in: next value to assign to a row */
+	ib_uint64_t	value)	/* in: next value to assign to a row */
 {
 	ut_ad(mutex_own(&table->autoinc_mutex));
 
@@ -354,13 +354,13 @@ dict_table_autoinc_initialize(
 Reads the next autoinc value (== autoinc counter value), 0 if not yet
 initialized. */
 UNIV_INTERN
-ib_longlong
+ib_uint64_t
 dict_table_autoinc_read(
 /*====================*/
 					/* out: value for a new row, or 0 */
 	const dict_table_t*	table)	/* in: table */
 {
-	ib_longlong	value;
+	ib_int64_t	value;
 
 	ut_ad(mutex_own(&table->autoinc_mutex));
 
@@ -383,7 +383,7 @@ dict_table_autoinc_update(
 /*======================*/
 
 	dict_table_t*	table,	/* in/out: table */
-	ib_longlong	value)	/* in: value which was assigned to a row */
+	ib_uint64_t	value)	/* in: value which was assigned to a row */
 {
 	if (table->autoinc_inited && value > table->autoinc) {
 
@@ -1419,7 +1419,7 @@ dict_index_add_to_cache(
 		new_index->stat_n_diff_key_vals = mem_heap_alloc(
 			new_index->heap,
 			(1 + dict_index_get_n_unique(new_index))
-			* sizeof(ib_longlong));
+			* sizeof(ib_int64_t));
 		/* Give some sensible values to stat_n_... in case we do
 		not calculate statistics quickly enough */
 
@@ -3592,7 +3592,7 @@ dict_foreign_parse_drop_constraints(
 
 	ptr = dict_accept(cs, ptr, "FOREIGN", &success);
 
-	if (!success) {
+	if (!success || !my_isspace(cs, *ptr)) {
 
 		goto loop;
 	}
@@ -4087,12 +4087,13 @@ dict_table_print_low(
 
 	fprintf(stderr,
 		"--------------------------------------\n"
-		"TABLE: name %s, id %lu %lu, columns %lu, indexes %lu,"
-		" appr.rows %lu\n"
+		"TABLE: name %s, id %lu %lu, flags %lx, columns %lu,"
+		" indexes %lu, appr.rows %lu\n"
 		"  COLUMNS: ",
 		table->name,
 		(ulong) ut_dulint_get_high(table->id),
 		(ulong) ut_dulint_get_low(table->id),
+		(ulong) table->flags,
 		(ulong) table->n_cols,
 		(ulong) UT_LIST_GET_LEN(table->indexes),
 		(ulong) table->stat_n_rows);
@@ -4154,7 +4155,7 @@ dict_index_print_low(
 /*=================*/
 	dict_index_t*	index)	/* in: index */
 {
-	ib_longlong	n_vals;
+	ib_int64_t	n_vals;
 	ulint		i;
 	const char*	type_string;
 
@@ -4484,7 +4485,7 @@ dict_table_find_equivalent_index(
 		table, column_names, index->n_fields,
 		index, TRUE, FALSE);
 
-	mem_free(column_names);
+	mem_free((void*) column_names);
 
 	return(equiv_index);
 }

dict/dict0load.c

@@ -223,11 +223,11 @@ dict_print(void)
 }
 
 /************************************************************************
-Determine the compressed page size of a table described in SYS_TABLES. */
+Determine the flags of a table described in SYS_TABLES. */
 static
 ulint
-dict_sys_tables_get_zip_size(
-/*=========================*/
+dict_sys_tables_get_flags(
+/*======================*/
 				/* out: compressed page size in kilobytes;
 				or 0 if the tablespace is uncompressed,
 				ULINT_UNDEFINED on error */
@@ -236,29 +236,53 @@ dict_sys_tables_get_zip_size(
 	const byte*	field;
 	ulint		len;
 	ulint		n_cols;
-	ulint		table_type;
+	ulint		flags;
 
 	field = rec_get_nth_field_old(rec, 5, &len);
 	ut_a(len == 4);
 
-	table_type = mach_read_from_4(field);
+	flags = mach_read_from_4(field);
+
+	if (UNIV_LIKELY(flags == DICT_TABLE_ORDINARY)) {
+		return(0);
+	}
 
 	field = rec_get_nth_field_old(rec, 4, &len);
 	n_cols = mach_read_from_4(field);
 
-	if (UNIV_EXPECT(n_cols & 0x80000000UL, 0x80000000UL)
-			&& UNIV_LIKELY(table_type > DICT_TABLE_COMPRESSED_BASE)
-			&& UNIV_LIKELY(table_type
-					<= DICT_TABLE_COMPRESSED_BASE + 16)) {
+	if (UNIV_UNLIKELY(!(n_cols & 0x80000000UL))) {
+		/* New file formats require ROW_FORMAT=COMPACT. */
+		return(ULINT_UNDEFINED);
+	}
 
-		return(table_type - DICT_TABLE_COMPRESSED_BASE);
+	switch (flags & (DICT_TF_FORMAT_MASK | DICT_TF_COMPACT)) {
+	default:
+	case DICT_TF_FORMAT_51 << DICT_TF_FORMAT_SHIFT:
+	case DICT_TF_FORMAT_51 << DICT_TF_FORMAT_SHIFT | DICT_TF_COMPACT:
+		/* flags should be DICT_TABLE_ORDINARY,
+		or DICT_TF_FORMAT_MASK should be nonzero. */
+		return(ULINT_UNDEFINED);
+
+	case DICT_TF_FORMAT_ZIP << DICT_TF_FORMAT_SHIFT | DICT_TF_COMPACT:
+#if DICT_TF_FORMAT_MAX > DICT_TF_FORMAT_ZIP
+# error "missing case labels for DICT_TF_FORMAT_ZIP .. DICT_TF_FORMAT_MAX"
+#endif
+		/* We support this format. */
+		break;
 	}
 
-	if (UNIV_LIKELY(table_type == DICT_TABLE_ORDINARY)) {
-		return(0);
+	if (UNIV_UNLIKELY((flags & DICT_TF_ZSSIZE_MASK)
+			  > (DICT_TF_ZSSIZE_MAX << DICT_TF_ZSSIZE_SHIFT))) {
+		/* Unsupported compressed page size. */
+		return(ULINT_UNDEFINED);
 	}
 
-	return(ULINT_UNDEFINED);
+	if (UNIV_UNLIKELY(flags & (~0 << DICT_TF_BITS))) {
+		/* Some unused bits are set. */
+		return(ULINT_UNDEFINED);
+	}
+
+	return(flags);
 }
 
 /************************************************************************
@@ -321,14 +345,28 @@ dict_check_tablespaces_and_store_max_id(
 		const byte*	field;
 		ulint		len;
 		ulint		space_id;
-		ulint		zip_size_in_k;
+		ulint		flags;
 		char*		name;
 
 		field = rec_get_nth_field_old(rec, 0, &len);
 		name = mem_strdupl((char*) field, len);
 
-		zip_size_in_k = dict_sys_tables_get_zip_size(rec);
-		ut_a(zip_size_in_k != ULINT_UNDEFINED);
+		flags = dict_sys_tables_get_flags(rec);
+		if (UNIV_UNLIKELY(flags == ULINT_UNDEFINED)) {
+
+			field = rec_get_nth_field_old(rec, 5, &len);
+			flags = mach_read_from_4(field);
+
+			ut_print_timestamp(stderr);
+			fputs("  InnoDB: Error: table ", stderr);
+			ut_print_filename(stderr, name);
+			fprintf(stderr, "\n"
+				"InnoDB: in InnoDB data dictionary"
+				" has unknown type %lx.\n",
+				(ulong) flags);
+
+			goto loop;
+		}
 
 		field = rec_get_nth_field_old(rec, 9, &len);
 		ut_a(len == 4);
@@ -352,8 +390,7 @@ dict_check_tablespaces_and_store_max_id(
 			object and check that the .ibd file exists. */
 
 			fil_open_single_table_tablespace(FALSE, space_id,
-							 zip_size_in_k * 1024,
-							 name);
+							 flags, name);
 		}
 
 		mem_free(name);
@@ -784,7 +821,6 @@ dict_load_table(
 	ulint		n_cols;
 	ulint		flags;
 	ulint		err;
-	ulint		zip_size_in_k;
 	mtr_t		mtr;
 
 	ut_ad(mutex_own(&(dict_sys->mutex)));
@@ -833,8 +869,21 @@ dict_load_table(
 
 	/* Check if the tablespace exists and has the right name */
 	if (space != 0) {
-		zip_size_in_k = dict_sys_tables_get_zip_size(rec);
-		ut_a(zip_size_in_k != ULINT_UNDEFINED);
+		flags = dict_sys_tables_get_flags(rec);
+
+		if (UNIV_UNLIKELY(flags == ULINT_UNDEFINED)) {
+			field = rec_get_nth_field_old(rec, 5, &len);
+			flags = mach_read_from_4(field);
+
+			ut_print_timestamp(stderr);
+			fputs("  InnoDB: Error: table ", stderr);
+			ut_print_filename(stderr, name);
+			fprintf(stderr, "\n"
+				"InnoDB: in InnoDB data dictionary"
+				" has unknown type %lx.\n",
+				(ulong) flags);
+			goto err_exit;
+		}
 
 		if (fil_space_for_table_exists_in_mem(space, name, FALSE,
 						      FALSE, FALSE)) {
@@ -853,7 +902,7 @@ dict_load_table(
 				name, (ulong)space);
 			/* Try to open the tablespace */
 			if (!fil_open_single_table_tablespace(
-				    TRUE, space, zip_size_in_k << 10, name)) {
+				    TRUE, space, flags, name)) {
 				/* We failed to find a sensible tablespace
 				file */
 
@@ -861,7 +910,7 @@ dict_load_table(
 			}
 		}
 	} else {
-		zip_size_in_k = 0;
+		flags = 0;
 	}
 
 	ut_a(name_of_col_is(sys_tables, sys_index, 4, "N_COLS"));
@@ -869,8 +918,6 @@ dict_load_table(
 	field = rec_get_nth_field_old(rec, 4, &len);
 	n_cols = mach_read_from_4(field);
 
-	flags = zip_size_in_k << DICT_TF_COMPRESSED_SHIFT;
-
 	/* The high-order bit of N_COLS is the "compact format" flag. */
 	if (n_cols & 0x80000000UL) {
 		flags |= DICT_TF_COMPACT;
@@ -886,17 +933,6 @@ dict_load_table(
 	field = rec_get_nth_field_old(rec, 3, &len);
 	table->id = mach_read_from_8(field);
 
-	zip_size_in_k = dict_sys_tables_get_zip_size(rec);
-
-	if (UNIV_UNLIKELY(zip_size_in_k == ULINT_UNDEFINED)) {
-		field = rec_get_nth_field_old(rec, 5, &len);
-		ut_print_timestamp(stderr);
-		fprintf(stderr,
-			"  InnoDB: table %s: unknown table type %lu\n",
-			name, (ulong) mach_read_from_4(field));
-		goto err_exit;
-	}
-
 	btr_pcur_close(&pcur);
 	mtr_commit(&mtr);
 

dict/dict0mem.c

@@ -42,6 +42,7 @@ dict_mem_table_create(
 	mem_heap_t*	heap;
 
 	ut_ad(name);
+	ut_a(!(flags & (~0 << DICT_TF_BITS)));
 
 	heap = mem_heap_create(DICT_HEAP_SIZE);
 

ha/hash0hash.c

@@ -136,7 +136,8 @@ hash_create_mutexes_func(
 {
 	ulint	i;
 
-	ut_a(n_mutexes > 0 && ut_is_2pow(n_mutexes));
+	ut_a(n_mutexes > 0);
+	ut_a(ut_is_2pow(n_mutexes));
 
 	table->mutexes = mem_alloc(n_mutexes * sizeof(mutex_t));
 

handler/ha_innodb.h

@@ -73,7 +73,7 @@ class ha_innobase: public handler
 	void update_thd();
 	int change_active_index(uint keynr);
 	int general_fetch(uchar* buf, uint direction, uint match_mode);
-	int innobase_read_and_init_auto_inc(longlong* ret);
+	int innobase_read_and_init_auto_inc(ulonglong* ret);
 	ulong innobase_autoinc_lock();
 	ulong innobase_set_max_autoinc(ulonglong auto_inc);
 	ulong innobase_reset_autoinc(ulonglong auto_inc);
@@ -263,4 +263,5 @@ convert_error_code_to_mysql(
 /*========================*/
 				/* out: MySQL error code */
 	int		error,	/* in: InnoDB error code */
+	ulint		flags,	/* in: InnoDB table flags, or 0 */
 	MYSQL_THD	thd);	/* in: user thread handle or NULL */

handler/handler0alter.cc

@@ -720,7 +720,8 @@ ha_innobase::add_index(
 				break;
 			default:
 				error = convert_error_code_to_mysql(
-					trx->error_state, user_thd);
+					trx->error_state, innodb_table->flags,
+					user_thd);
 			}
 
 			row_mysql_unlock_data_dictionary(trx);
@@ -872,7 +873,9 @@ ha_innobase::add_index(
 		}
 
 convert_error:
-		error = convert_error_code_to_mysql(error, user_thd);
+		error = convert_error_code_to_mysql(error,
+						    innodb_table->flags,
+						    user_thd);
 	}
 
 	mem_heap_free(heap);
@@ -1095,7 +1098,7 @@ ha_innobase::final_drop_index(
 	transaction depends on an index that is being dropped. */
 	err = convert_error_code_to_mysql(
 		row_merge_lock_table(prebuilt->trx, prebuilt->table, LOCK_X),
-		user_thd);
+		prebuilt->table->flags, user_thd);
 
 	if (UNIV_UNLIKELY(err)) {
 

handler/i_s.h

@@ -12,7 +12,9 @@ Created July 18, 2007 Vasil Dimov
 extern struct st_mysql_plugin	i_s_innodb_trx;
 extern struct st_mysql_plugin	i_s_innodb_locks;
 extern struct st_mysql_plugin	i_s_innodb_lock_waits;
-extern struct st_mysql_plugin	i_s_innodb_zip;
-extern struct st_mysql_plugin	i_s_innodb_zip_reset;
+extern struct st_mysql_plugin	i_s_innodb_cmp;
+extern struct st_mysql_plugin	i_s_innodb_cmp_reset;
+extern struct st_mysql_plugin	i_s_innodb_cmpmem;
+extern struct st_mysql_plugin	i_s_innodb_cmpmem_reset;
 
 #endif /* i_s_h */

ibuf/ibuf0ibuf.c

@@ -1896,6 +1896,7 @@ ibuf_add_free_page(void)
 {
 	mtr_t	mtr;
 	page_t*	header_page;
+	ulint	flags;
 	ulint	zip_size;
 	ulint	page_no;
 	page_t*	page;
@@ -1906,7 +1907,8 @@ ibuf_add_free_page(void)
 
 	/* Acquire the fsp latch before the ibuf header, obeying the latching
 	order */
-	mtr_x_lock(fil_space_get_latch(IBUF_SPACE_ID, &zip_size), &mtr);
+	mtr_x_lock(fil_space_get_latch(IBUF_SPACE_ID, &flags), &mtr);
+	zip_size = dict_table_flags_to_zip_size(flags);
 
 	header_page = ibuf_header_page_get(&mtr);
 
@@ -1988,6 +1990,7 @@ ibuf_remove_free_page(void)
 	mtr_t	mtr;
 	mtr_t	mtr2;
 	page_t*	header_page;
+	ulint	flags;
 	ulint	zip_size;
 	ulint	page_no;
 	page_t*	page;
@@ -1998,7 +2001,8 @@ ibuf_remove_free_page(void)
 
 	/* Acquire the fsp latch before the ibuf header, obeying the latching
 	order */
-	mtr_x_lock(fil_space_get_latch(IBUF_SPACE_ID, &zip_size), &mtr);
+	mtr_x_lock(fil_space_get_latch(IBUF_SPACE_ID, &flags), &mtr);
+	zip_size = dict_table_flags_to_zip_size(flags);
 
 	header_page = ibuf_header_page_get(&mtr);
 
@@ -2169,7 +2173,7 @@ ibuf_get_merge_page_nos(
 	rec_t*		rec,	/* in: record from which we read up and down
 				in the chain of records */
 	ulint*		space_ids,/* in/out: space id's of the pages */
-	ib_longlong*	space_versions,/* in/out: tablespace version
+	ib_int64_t*	space_versions,/* in/out: tablespace version
 				timestamps; used to prevent reading in old
 				pages after DISCARD + IMPORT tablespace */
 	ulint*		page_nos,/* in/out: buffer for at least
@@ -2343,7 +2347,7 @@ ibuf_contract_ext(
 	btr_pcur_t	pcur;
 	ulint		page_nos[IBUF_MAX_N_PAGES_MERGED];
 	ulint		space_ids[IBUF_MAX_N_PAGES_MERGED];
-	ib_longlong	space_versions[IBUF_MAX_N_PAGES_MERGED];
+	ib_int64_t	space_versions[IBUF_MAX_N_PAGES_MERGED];
 	ulint		n_stored;
 	ulint		sum_sizes;
 	mtr_t		mtr;
@@ -2930,7 +2934,7 @@ ibuf_insert_low(
 	ulint		err;
 	ibool		do_merge;
 	ulint		space_ids[IBUF_MAX_N_PAGES_MERGED];
-	ib_longlong	space_versions[IBUF_MAX_N_PAGES_MERGED];
+	ib_int64_t	space_versions[IBUF_MAX_N_PAGES_MERGED];
 	ulint		page_nos[IBUF_MAX_N_PAGES_MERGED];
 	ulint		n_stored;
 	ulint		bits;

include/btr0btr.h

@@ -468,12 +468,12 @@ UNIV_INTERN
 ibool
 btr_index_rec_validate(
 /*===================*/
-					/* out: TRUE if ok */
-	rec_t*		rec,		/* in: index record */
-	dict_index_t*	index,		/* in: index */
-	ibool		dump_on_error);	/* in: TRUE if the function
-					should print hex dump of record
-					and page on error */
+						/* out: TRUE if ok */
+	const rec_t*		rec,		/* in: index record */
+	const dict_index_t*	index,		/* in: index */
+	ibool			dump_on_error);	/* in: TRUE if the function
+						should print hex dump of record
+						and page on error */
 /******************************************************************
 Checks the consistency of an index tree. */
 UNIV_INTERN

include/btr0cur.h

@@ -422,7 +422,7 @@ btr_cur_parse_del_mark_set_sec_rec(
 /***********************************************************************
 Estimates the number of rows in a given index range. */
 UNIV_INTERN
-ib_longlong
+ib_int64_t
 btr_estimate_n_rows_in_range(
 /*=========================*/
 				/* out: estimated number of rows */

include/buf0buddy.h

@@ -50,15 +50,21 @@ buf_buddy_free(
 	ulint	size)	/* in: block size, up to UNIV_PAGE_SIZE */
 	__attribute__((nonnull));
 
-/** Counts of blocks allocated from the buddy system.
-Protected by buf_pool_mutex. */
-extern ulint buf_buddy_used[BUF_BUDDY_SIZES + 1];
-/** Counts of blocks relocated by the buddy system.
-Protected by buf_pool_mutex. */
-extern ib_uint64_t buf_buddy_relocated[BUF_BUDDY_SIZES + 1];
-/** Durations of block relocations.
+/** Statistics of buddy blocks of a given size. */
+struct buf_buddy_stat_struct {
+	/** Number of blocks allocated from the buddy system. */
+	ulint		used;
+	/** Number of blocks relocated by the buddy system. */
+	ib_uint64_t	relocated;
+	/** Total duration of block relocations, in microseconds. */
+	ib_uint64_t	relocated_usec;
+};
+
+typedef struct buf_buddy_stat_struct buf_buddy_stat_t;
+
+/** Statistics of the buddy system, indexed by block size.
 Protected by buf_pool_mutex. */
-extern ullint buf_buddy_relocated_duration[BUF_BUDDY_SIZES + 1];
+extern buf_buddy_stat_t buf_buddy_stat[BUF_BUDDY_SIZES + 1];
 
 #ifndef UNIV_NONINL
 # include "buf0buddy.ic"

include/buf0buf.h

@@ -916,7 +916,7 @@ buf_page_init_for_read(
 	ulint		space,	/* in: space id */
 	ulint		zip_size,/* in: compressed page size, or 0 */
 	ibool		unzip,	/* in: TRUE=request uncompressed page */
-	ib_longlong	tablespace_version,/* in: prevents reading from a wrong
+	ib_int64_t	tablespace_version,/* in: prevents reading from a wrong
 				version of the tablespace in case we have done
 				DISCARD + IMPORT */
 	ulint		offset);/* in: page number */

include/buf0buf.ic

@@ -278,7 +278,7 @@ buf_page_get_mutex(
 	switch (buf_page_get_state(bpage)) {
 	case BUF_BLOCK_ZIP_FREE:
 		ut_error;
-		break;
+		return(NULL);
 	case BUF_BLOCK_ZIP_PAGE:
 	case BUF_BLOCK_ZIP_DIRTY:
 		return(&buf_pool_zip_mutex);

include/buf0rea.h

@@ -73,7 +73,7 @@ buf_read_ibuf_merge_pages(
 					to get read in, before this
 					function returns */
 	const ulint*	space_ids,	/* in: array of space ids */
-	const ib_longlong* space_versions,/* in: the spaces must have
+	const ib_int64_t* space_versions,/* in: the spaces must have
 					this version number
 					(timestamp), otherwise we
 					discard the read; we use this

include/db0err.h

@@ -74,8 +74,6 @@ enum db_err {
 
 	DB_PRIMARY_KEY_IS_NULL,		/* a column in the PRIMARY KEY
 					was found to be NULL */
-	DB_TABLE_ZIP_NO_IBD,		/* trying to create a compressed
-					table in the system tablespace */
 
 	/* The following are partial failure codes */
 	DB_FAIL = 1000,

include/dict0dict.h

@@ -179,12 +179,12 @@ void
 dict_table_autoinc_initialize(
 /*==========================*/
 	dict_table_t*	table,	/* in/out: table */
-	ib_longlong	value);	/* in: next value to assign to a row */
+	ib_uint64_t	value);	/* in: next value to assign to a row */
 /************************************************************************
 Reads the next autoinc value (== autoinc counter value), 0 if not yet
 initialized. */
 UNIV_INTERN
-ib_longlong
+ib_uint64_t
 dict_table_autoinc_read(
 /*====================*/
 					/* out: value for a new row, or 0 */
@@ -198,7 +198,7 @@ dict_table_autoinc_update(
 /*======================*/
 
 	dict_table_t*	table,	/* in/out: table */
-	ib_longlong	value);	/* in: value which was assigned to a row */
+	ib_uint64_t	value);	/* in: value which was assigned to a row */
 /************************************************************************
 Release the autoinc lock.*/
 UNIV_INTERN
@@ -631,6 +631,32 @@ dict_table_is_comp(
 					compact page format */
 	const dict_table_t*	table);	/* in: table */
 /************************************************************************
+Determine the file format of a table. */
+UNIV_INLINE
+ulint
+dict_table_get_format(
+/*==================*/
+					/* out: file format version */
+	const dict_table_t*	table);	/* in: table */
+/************************************************************************
+Set the file format of a table. */
+UNIV_INLINE
+void
+dict_table_set_format(
+/*==================*/
+	dict_table_t*	table,	/* in/out: table */
+	ulint		format);/* in: file format version */
+/************************************************************************
+Extract the compressed page size from table flags. */
+UNIV_INLINE
+ulint
+dict_table_flags_to_zip_size(
+/*=========================*/
+			/* out: compressed page size,
+			or 0 if not compressed */
+	ulint	flags)	/* in: flags */
+	__attribute__((const));
+/************************************************************************
 Check whether the table uses the compressed compact page format. */
 UNIV_INLINE
 ulint

include/dict0dict.ic

@@ -349,6 +349,57 @@ dict_table_is_comp(
 	return(UNIV_LIKELY(table->flags & DICT_TF_COMPACT));
 }
 
+/************************************************************************
+Determine the file format of a table. */
+UNIV_INLINE
+ulint
+dict_table_get_format(
+/*==================*/
+					/* out: file format version */
+	const dict_table_t*	table)	/* in: table */
+{
+	ut_ad(table);
+
+	return((table->flags & DICT_TF_FORMAT_MASK) >> DICT_TF_FORMAT_SHIFT);
+}
+
+/************************************************************************
+Determine the file format of a table. */
+UNIV_INLINE
+void
+dict_table_set_format(
+/*==================*/
+	dict_table_t*	table,	/* in/out: table */
+	ulint		format)	/* in: file format version */
+{
+	ut_ad(table);
+
+	table->flags = (table->flags & ~DICT_TF_FORMAT_MASK)
+		| (format << DICT_TF_FORMAT_SHIFT);
+}
+
+/************************************************************************
+Extract the compressed page size from table flags. */
+UNIV_INLINE
+ulint
+dict_table_flags_to_zip_size(
+/*=========================*/
+			/* out: compressed page size,
+			or 0 if not compressed */
+	ulint	flags)	/* in: flags */
+{
+	ulint	zip_size = flags & DICT_TF_ZSSIZE_MASK;
+
+	if (UNIV_UNLIKELY(zip_size)) {
+		zip_size = ((PAGE_ZIP_MIN_SIZE >> 1)
+			 << (zip_size >> DICT_TF_ZSSIZE_SHIFT));
+
+		ut_ad(zip_size <= UNIV_PAGE_SIZE);
+	}
+
+	return(zip_size);
+}
+
 /************************************************************************
 Check whether the table uses the compressed compact page format. */
 UNIV_INLINE
@@ -361,8 +412,7 @@ dict_table_zip_size(
 {
 	ut_ad(table);
 
-	return(UNIV_UNLIKELY((table->flags & DICT_TF_COMPRESSED_MASK)
-			     << (10 - DICT_TF_COMPRESSED_SHIFT)));
+	return(dict_table_flags_to_zip_size(table->flags));
 }
 
 /************************************************************************

include/dict0mem.h

@@ -40,12 +40,32 @@ combination of types */
 #define	DICT_TABLE_CLUSTER		3 /* this means that the table is
 					  really a cluster definition */
 #endif
-#define DICT_TABLE_COMPRESSED_BASE	0x8000	/* compressed tablespace */
 
-/* Table flags */
-#define DICT_TF_COMPACT			1	/* compact page format */
-#define DICT_TF_COMPRESSED_MASK		62	/* compressed page size, KiB */
-#define DICT_TF_COMPRESSED_SHIFT	1
+/* Table flags.  All unused bits must be 0. */
+#define DICT_TF_COMPACT			1	/* Compact page format.
+						This must be set for
+						new file formats
+						(later than
+						DICT_TF_FORMAT_51). */
+
+/* compressed page size (0=uncompressed, up to 15 compressed sizes) */
+#define DICT_TF_ZSSIZE_SHIFT		1
+#define DICT_TF_ZSSIZE_MASK		(15 << DICT_TF_ZSSIZE_SHIFT)
+#define DICT_TF_ZSSIZE_MAX (UNIV_PAGE_SIZE_SHIFT - PAGE_ZIP_MIN_SIZE_SHIFT + 1)
+
+
+#define DICT_TF_FORMAT_SHIFT		5	/* file format */
+#define DICT_TF_FORMAT_MASK		(127 << 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 */
+#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
 
 /**************************************************************************
 Creates a table memory object. */
@@ -226,7 +246,7 @@ struct dict_index_struct{
 			indexes;/* list of indexes of the table */
 	btr_search_t*	search_info; /* info used in optimistic searches */
 	/*----------------------*/
-	ib_longlong*	stat_n_diff_key_vals;
+	ib_int64_t*	stat_n_diff_key_vals;
 				/* approximate number of different key values
 				for this index, for each n-column prefix
 				where n <= dict_get_n_unique(index); we
@@ -314,6 +334,7 @@ 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	ibd_file_missing:1;
 				/* TRUE if this is in a single-table
 				tablespace and the .ibd file is missing; then
@@ -326,7 +347,6 @@ struct dict_table_struct{
 				TABLESPACE */
 	unsigned	cached:1;/* TRUE if the table object has been added
 				to the dictionary cache */
-	unsigned	flags:8;/* DICT_TF_COMPACT, ... */
 	unsigned	n_def:10;/* number of columns defined so far */
 	unsigned	n_cols:10;/* number of columns */
 	dict_col_t*	cols;	/* array of column descriptions */
@@ -397,7 +417,7 @@ struct dict_table_struct{
 	unsigned	stat_initialized:1; /* TRUE if statistics have
 				been calculated the first time
 				after database startup or table creation */
-	ib_longlong	stat_n_rows;
+	ib_int64_t	stat_n_rows;
 				/* approximate number of rows in the table;
 				we periodically calculate new estimates */
 	ulint		stat_clustered_index_size;
@@ -425,9 +445,9 @@ struct dict_table_struct{
 				/* TRUE if the autoinc counter has been
 				inited; MySQL gets the init value by executing
 				SELECT MAX(auto inc column) */
-	ib_longlong	autoinc;/* autoinc counter value to give to the
+	ib_uint64_t	autoinc;/* autoinc counter value to give to the
 				next inserted row */
-	ib_longlong	autoinc_increment;
+	ib_int64_t	autoinc_increment;
 				/* The increment step of the auto increment
 				column. Value must be greater than or equal
 				to 1 */

include/fil0fil.h

@@ -133,7 +133,7 @@ extern ulint	fil_n_pending_tablespace_flushes;
 /***********************************************************************
 Returns the version number of a tablespace, -1 if not found. */
 UNIV_INTERN
-ib_longlong
+ib_int64_t
 fil_space_get_version(
 /*==================*/
 			/* out: version number, -1 if the tablespace does not
@@ -214,6 +214,15 @@ fil_space_get_size(
 			/* out: space size, 0 if space not found */
 	ulint	id);	/* in: space id */
 /***********************************************************************
+Returns the flags of the space. The tablespace must be cached
+in the memory cache. */
+UNIV_INTERN
+ulint
+fil_space_get_flags(
+/*================*/
+			/* out: flags, ULINT_UNDEFINED if space not found */
+	ulint	id);	/* in: space id */
+/***********************************************************************
 Returns the compressed page size of the space, or 0 if the space
 is not compressed. The tablespace must be cached in the memory cache. */
 UNIV_INTERN
@@ -397,8 +406,7 @@ fil_create_new_single_table_tablespace(
 					table */
 	ibool		is_temp,	/* in: TRUE if a table created with
 					CREATE TEMPORARY TABLE */
-	ulint		zip_size,	/* in: compressed page size,
-					or 0 if uncompressed tablespace */
+	ulint		flags,		/* in: tablespace flags */
 	ulint		size);		/* in: the initial size of the
 					tablespace file in pages,
 					must be >= FIL_IBD_FILE_INITIAL_SIZE */
@@ -424,8 +432,7 @@ fil_open_single_table_tablespace(
 					faster (the OS caches them) than
 					accessing the first page of the file */
 	ulint		id,		/* in: space id */
-	ulint		zip_size,	/* in: compressed page size,
-					or 0 if uncompressed tablespace */
+	ulint		flags,		/* in: tablespace flags */
 	const char*	name);		/* in: table name in the
 					databasename/tablename format */
 /************************************************************************
@@ -479,7 +486,7 @@ fil_tablespace_deleted_or_being_deleted_in_mem(
 				/* out: TRUE if does not exist or is being\
 				deleted */
 	ulint		id,	/* in: space id */
-	ib_longlong	version);/* in: tablespace_version should be this; if
+	ib_int64_t	version);/* in: tablespace_version should be this; if
 				you pass -1 as the value of this, then this
 				parameter is ignored */
 /***********************************************************************

include/fsp0fsp.h

@@ -46,26 +46,25 @@ void
 fsp_init(void);
 /*==========*/
 /**************************************************************************
-Gets the current free limit of a tablespace. The free limit means the
-place of the first page which has never been put to the the free list
-for allocation. The space above that address is initialized to zero.
-Sets also the global variable log_fsp_current_free_limit. */
+Gets the current free limit of the system tablespace.  The free limit
+means the place of the first page which has never been put to the the
+free list for allocation.  The space above that address is initialized
+to zero.  Sets also the global variable log_fsp_current_free_limit. */
 UNIV_INTERN
 ulint
-fsp_header_get_free_limit(
-/*======================*/
+fsp_header_get_free_limit(void);
+/*===========================*/
 			/* out: free limit in megabytes */
-	ulint	space);	/* in: space id, must be 0 */
 /**************************************************************************
-Gets the size of the tablespace from the tablespace header. If we do not
-have an auto-extending data file, this should be equal to the size of the
-data files. If there is an auto-extending data file, this can be smaller. */
+Gets the size of the system tablespace from the tablespace header.  If
+we do not have an auto-extending data file, this should be equal to
+the size of the data files.  If there is an auto-extending data file,
+this can be smaller. */
 UNIV_INTERN
 ulint
-fsp_header_get_tablespace_size(
-/*===========================*/
+fsp_header_get_tablespace_size(void);
+/*================================*/
 			/* out: size in pages */
-	ulint	space);	/* in: space id, must be 0 */
 /**************************************************************************
 Reads the file space size stored in the header page. */
 UNIV_INTERN
@@ -83,6 +82,14 @@ fsp_header_get_space_id(
 				/* out: space id, ULINT UNDEFINED if error */
 	const page_t*	page);	/* in: first page of a tablespace */
 /**************************************************************************
+Reads the space flags from the first page of a tablespace. */
+UNIV_INTERN
+ulint
+fsp_header_get_flags(
+/*=================*/
+				/* out: flags */
+	const page_t*	page);	/* in: first page of a tablespace */
+/**************************************************************************
 Reads the compressed page size from the first page of a tablespace. */
 UNIV_INTERN
 ulint
@@ -101,8 +108,8 @@ fsp_header_init_fields(
 /*===================*/
 	page_t*	page,		/* in/out: first page in the space */
 	ulint	space_id,	/* in: space id */
-	ulint	zip_size);	/* in: compressed page size in bytes;
-				0 for uncompressed pages */
+	ulint	flags);		/* in: tablespace flags (FSP_SPACE_FLAGS):
+				0, or table->flags if newer than COMPACT */
 /**************************************************************************
 Initializes the space header of a new created space and creates also the
 insert buffer tree root if space == 0. */
@@ -357,6 +364,7 @@ fseg_validate(
 				/* out: TRUE if ok */
 	fseg_header_t*	header, /* in: segment header */
 	mtr_t*		mtr2);	/* in: mtr */
+#ifdef UNIV_BTR_PRINT
 /***********************************************************************
 Writes info of a segment. */
 UNIV_INTERN
@@ -365,6 +373,7 @@ fseg_print(
 /*=======*/
 	fseg_header_t*	header, /* in: segment header */
 	mtr_t*		mtr);	/* in: mtr */
+#endif /* UNIV_BTR_PRINT */
 
 /* Flags for fsp_reserve_free_extents */
 #define FSP_NORMAL	1000000

include/hash0hash.ic

@@ -70,6 +70,7 @@ hash_get_mutex_no(
 	hash_table_t*	table,	/* in: hash table */
 	ulint		fold)	/* in: fold */
 {
+	ut_ad(ut_is_2pow(table->n_mutexes));
 	return(ut_2pow_remainder(fold, table->n_mutexes));
 }
 

include/log0log.h

@@ -46,7 +46,7 @@ ulint
 log_calc_where_lsn_is(
 /*==================*/
 						/* out: log file number */
-	ib_longlong*	log_file_offset,	/* out: offset in that file
+	ib_int64_t*	log_file_offset,	/* out: offset in that file
 						(including the header) */
 	ib_uint64_t	first_header_lsn,	/* in: first log file start
 						lsn */
@@ -54,7 +54,7 @@ log_calc_where_lsn_is(
 						determine */
 	ulint		n_log_files,		/* in: total number of log
 						files */
-	ib_longlong	log_file_size);		/* in: log file size
+	ib_int64_t	log_file_size);		/* in: log file size
 						(including the header) */
 /****************************************************************
 Writes to the log the string given. The log must be released with

include/mach0data.ic

@@ -287,8 +287,8 @@ mach_write_ull(
 {
 	ut_ad(b);
 
-	mach_write_to_4(b, n >> 32);
-	mach_write_to_4(b + 4, n);
+	mach_write_to_4(b, (ulint) (n >> 32));
+	mach_write_to_4(b + 4, (ulint) n);
 }
 
 /************************************************************

include/mtr0mtr.h

@@ -130,8 +130,8 @@ flag value must give the length also! */
 						/* copy compact record list end
 						to a new created index page */
 #define MLOG_COMP_PAGE_REORGANIZE ((byte)46)	/* reorganize an index page */
-#define MLOG_ZIP_FILE_CREATE	((byte)47)	/* log record about creating a
-						compressed .ibd file */
+#define MLOG_FILE_CREATE2	((byte)47)	/* log record about creating
+						an .ibd file, with format */
 #define MLOG_ZIP_WRITE_NODE_PTR	((byte)48)	/* write the node pointer of
 						a record on a compressed
 						non-leaf B-tree page */

include/os0sync.h

@@ -40,7 +40,7 @@ struct os_event_struct {
 					in the signaled state, i.e., a thread
 					does not stop if it tries to wait for
 					this event */
-	ib_longlong	signal_count;	/* this is incremented each time
+	ib_int64_t	signal_count;	/* this is incremented each time
 					the event becomes signaled */
 	pthread_cond_t	cond_var;	/* condition variable is used in
 					waiting for the event */
@@ -118,7 +118,7 @@ that this thread should not wait in case of an intervening call to
 os_event_set() between this os_event_reset() and the
 os_event_wait_low() call. See comments for os_event_wait_low(). */
 UNIV_INTERN
-ib_longlong
+ib_int64_t
 os_event_reset(
 /*===========*/
 	os_event_t	event);	/* in: event to reset */
@@ -155,7 +155,7 @@ void
 os_event_wait_low(
 /*==============*/
 	os_event_t	event,		/* in: event to wait */
-	ib_longlong	reset_sig_count);/* in: zero or the value
+	ib_int64_t	reset_sig_count);/* in: zero or the value
 					returned by previous call of
 					os_event_reset(). */
 

mysql-test/innodb-zip-master.opt

---innodb_file_per_table
\ No newline at end of file