branches/zip: Minor cleanup and bug fixes

btr_page_reorganize_low(): Rename new_page to temp_page.

btr_store_big_rec_extern_fields(): FIL_PAGE_TYPE is 2 bytes, not 4.

buf_page_init(), buf_page_create(), buf_read_page_low(),
buf_page_init_for_read(): Add parameter zip_size.

buf_page_init_for_backup_restore(),
recv_apply_log_recs_for_backup(): Enclose in #ifdef UNIV_HOTBACKUP.

Enclose some debug code in #ifdef UNIV_LOG_REPLICATE.

page_zip_write_header_log(): Replace page_zip with a pointer to
the uncompressed page.

page_zip_write_rec(): Relax an assertion about blob_no + n_ext.

page_copy_rec_list_to_created_page_write_log(): Allow logging to be disabled.

btr/btr0btr.c

@@ -858,7 +858,7 @@ btr_page_reorganize_low(
 	dict_index_t*	index,	/* in: record descriptor */
 	mtr_t*		mtr)	/* in: mtr */
 {
-	page_t*	new_page;
+	page_t*	temp_page;
 	ulint	log_mode;
 	ulint	data_size1;
 	ulint	data_size2;
@@ -881,10 +881,10 @@ btr_page_reorganize_low(
 	/* Turn logging off */
 	log_mode = mtr_set_log_mode(mtr, MTR_LOG_NONE);
 
-	new_page = buf_frame_alloc();
+	temp_page = buf_frame_alloc();
 
 	/* Copy the old page to temporary space */
-	buf_frame_copy(new_page, page);
+	buf_frame_copy(temp_page, page);
 
 	if (UNIV_LIKELY(!recovery)) {
 		btr_search_drop_page_hash_index(page);
@@ -900,16 +900,16 @@ btr_page_reorganize_low(
 	do not copy the lock bits yet */
 
 	page_copy_rec_list_end_no_locks(page,
-				page_get_infimum_rec(new_page), index, mtr);
+				page_get_infimum_rec(temp_page), index, mtr);
 	/* Copy max trx id to recreated page */
-	page_set_max_trx_id(page, NULL, page_get_max_trx_id(new_page));
+	page_set_max_trx_id(page, NULL, page_get_max_trx_id(temp_page));
 
 	if (UNIV_LIKELY_NULL(page_zip)) {
 		if (UNIV_UNLIKELY(!page_zip_compress(
 				page_zip, page, index))) {
 
 			/* Restore the old page and exit. */
-			buf_frame_copy(page, new_page);
+			buf_frame_copy(page, temp_page);
 
 			goto func_exit;
 		}
@@ -917,7 +917,7 @@ btr_page_reorganize_low(
 
 	if (UNIV_LIKELY(!recovery)) {
 		/* Update the record lock bitmaps */
-		lock_move_reorganize_page(page, new_page);
+		lock_move_reorganize_page(page, temp_page);
 	}
 
 	data_size2 = page_get_data_size(page);
@@ -926,7 +926,7 @@ btr_page_reorganize_low(
 	if (UNIV_UNLIKELY(data_size1 != data_size2)
 			|| UNIV_UNLIKELY(max_ins_size1 != max_ins_size2)) {
 		buf_page_print(page);
-		buf_page_print(new_page);
+		buf_page_print(temp_page);
 		fprintf(stderr,
 "InnoDB: Error: page old data size %lu new data size %lu\n"
 "InnoDB: Error: page old max ins size %lu new max ins size %lu\n"
@@ -940,7 +940,7 @@ btr_page_reorganize_low(
 
 func_exit:
 	ut_ad(!page_zip || page_zip_validate(page_zip, page));
-	buf_frame_free(new_page);
+	buf_frame_free(temp_page);
 
 	/* Restore logging mode */
 	mtr_set_log_mode(mtr, log_mode);

btr/btr0cur.c

@@ -3605,7 +3605,7 @@ btr_store_big_rec_extern_fields(
 			if (UNIV_LIKELY_NULL(page_zip)) {
 				int	err;
 
-				mach_write_to_4(page + FIL_PAGE_TYPE,
+				mach_write_to_2(page + FIL_PAGE_TYPE,
 						FIL_PAGE_TYPE_ZBLOB);
 
 				c_stream.next_out = page

buf/buf0buf.c

@@ -1559,6 +1559,7 @@ buf_page_get_known_nowait(
 	return(TRUE);
 }
 
+#ifdef UNIV_HOTBACKUP
 /************************************************************************
 Inits a page to the buffer buf_pool, for use in ibbackup --restore. */
 
@@ -1594,9 +1595,12 @@ buf_page_init_for_backup_restore(
 	block->n_fields		= 1;
 	block->n_bytes		= 0;
 	block->side		= BTR_SEARCH_LEFT_SIDE;
+	page_zip_des_init(&block->page_zip);
+	/* TODO: allocate page_zip->data? */
 
 	block->file_page_was_freed = FALSE;
 }
+#endif /* UNIV_HOTBACKUP */
 
 /************************************************************************
 Inits a page to the buffer buf_pool. */
@@ -1607,12 +1611,15 @@ buf_page_init(
 	ulint		space,	/* in: space id */
 	ulint		offset,	/* in: offset of the page within space
 				in units of a page */
+	ulint		zip_size,/* in: compressed page size in bytes,
+				or 0 if uncompressed tablespace */
 	buf_block_t*	block)	/* in: block to init */
 {
 #ifdef UNIV_SYNC_DEBUG
 	ut_ad(mutex_own(&(buf_pool->mutex)));
 #endif /* UNIV_SYNC_DEBUG */
 	ut_a(block->state != BUF_BLOCK_FILE_PAGE);
+	ut_a(space || !zip_size);
 
 	/* Set the state of the block */
 	block->magic_n		= BUF_BLOCK_MAGIC_N;
@@ -1627,11 +1634,19 @@ buf_page_init(
 	block->lock_hash_val	= lock_rec_hash(space, offset);
 	block->lock_mutex	= NULL;
 
+	block->page_zip.size	= zip_size;
+	if (UNIV_UNLIKELY(zip_size)) {
+		/* TODO: allocate this from a separate pool */
+		block->page_zip.data = ut_malloc(zip_size);
+	} else {
+		block->page_zip.data = NULL;
+	}
+
 	/* Insert into the hash table of file pages */
 
-	if (buf_page_hash_get(space, offset)) {
+	if (UNIV_LIKELY_NULL(buf_page_hash_get(space, offset))) {
 		fprintf(stderr,
-"InnoDB: Error: page %lu %lu already found from the hash table\n",
+"InnoDB: Error: page %lu %lu already found in the hash table\n",
 			(ulong) space,
 			(ulong) offset);
 #ifdef UNIV_DEBUG
@@ -1640,7 +1655,7 @@ buf_page_init(
 		buf_validate();
 		buf_LRU_validate();
 #endif /* UNIV_DEBUG */
-		ut_a(0);
+		ut_error;
 	}
 
 	HASH_INSERT(buf_block_t, hash, buf_pool->page_hash,
@@ -1683,6 +1698,7 @@ buf_page_init_for_read(
 	ulint*		err,	/* out: DB_SUCCESS or DB_TABLESPACE_DELETED */
 	ulint		mode,	/* in: BUF_READ_IBUF_PAGES_ONLY, ... */
 	ulint		space,	/* in: space id */
+	ulint		zip_size,/* in: compressed page size, or 0 */
 	ib_longlong	tablespace_version,/* in: prevents reading from a wrong
 				version of the tablespace in case we have done
 				DISCARD + IMPORT */
@@ -1743,7 +1759,7 @@ buf_page_init_for_read(
 
 	ut_ad(block);
 
-	buf_page_init(space, offset, block);
+	buf_page_init(space, offset, zip_size, block);
 
 	/* The block must be put to the LRU list, to the old blocks */
 
@@ -1784,6 +1800,7 @@ buf_page_create(
 	ulint	space,	/* in: space id */
 	ulint	offset,	/* in: offset of the page within space in units of
 			a page */
+	ulint	zip_size,/* in: compressed page size, or 0 */
 	mtr_t*	mtr)	/* in: mini-transaction handle */
 {
 	buf_frame_t*	frame;
@@ -1791,6 +1808,7 @@ buf_page_create(
 	buf_block_t*	free_block	= NULL;
 
 	ut_ad(mtr);
+	ut_ad(space || !zip_size);
 
 	free_block = buf_LRU_get_free_block();
 
@@ -1825,7 +1843,7 @@ buf_page_create(
 
 	block = free_block;
 
-	buf_page_init(space, offset, block);
+	buf_page_init(space, offset, zip_size, block);
 
 	/* The block must be put to the LRU list */
 	buf_LRU_add_block(block, FALSE);
@@ -1914,7 +1932,8 @@ buf_page_io_complete(
 		/* From version 3.23.38 up we store the page checksum
 		to the 4 first bytes of the page end lsn field */
 
-		if (buf_page_is_corrupted(block->frame, 0/*TODO:zip_size*/)) {
+		if (buf_page_is_corrupted(block->frame/* TODO */,
+					block->page_zip.size)) {
 			fprintf(stderr,
 		"InnoDB: Database page corruption on disk or a failed\n"
 		"InnoDB: file read of page %lu.\n", (ulong) block->offset);

buf/buf0rea.c

@@ -65,6 +65,7 @@ buf_read_page_low(
 			ORed to OS_AIO_SIMULATED_WAKE_LATER (see below
 			at read-ahead functions) */
 	ulint	space,	/* in: space id */
+	ulint	zip_size,/* in: compressed page size, or 0 */
 	ib_longlong 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
@@ -122,8 +123,8 @@ buf_read_page_low(
 	or is being dropped; if we succeed in initing the page in the buffer
 	pool for read, then DISCARD cannot proceed until the read has
 	completed */
-	block = buf_page_init_for_read(err, mode, space, tablespace_version,
-								offset);
+	block = buf_page_init_for_read(err, mode, space, zip_size,
+					tablespace_version, offset);
 	if (block == NULL) {
 
 		return(0);
@@ -177,6 +178,7 @@ buf_read_ahead_random(
 			wants to access */
 {
 	ib_longlong	tablespace_version;
+	ulint		zip_size;
 	buf_block_t*	block;
 	ulint		recent_blocks	= 0;
 	ulint		count;
@@ -200,6 +202,8 @@ buf_read_ahead_random(
 		return(0);
 	}
 
+	zip_size = fil_space_get_zip_size(space);
+
 	/* Remember the tablespace version before we ask te tablespace size
 	below: if DISCARD + IMPORT changes the actual .ibd file meanwhile, we
 	do not try to read outside the bounds of the tablespace! */
@@ -269,7 +273,8 @@ buf_read_ahead_random(
 		if (!ibuf_bitmap_page(i)) {
 			count += buf_read_page_low(&err, FALSE, ibuf_mode
 					| OS_AIO_SIMULATED_WAKE_LATER,
-					space, tablespace_version, i);
+					space, zip_size,
+					tablespace_version, i);
 			if (err == DB_TABLESPACE_DELETED) {
 				ut_print_timestamp(stderr);
 				fprintf(stderr,
@@ -316,10 +321,12 @@ buf_read_page(
 	ulint	offset)	/* in: page number */
 {
 	ib_longlong	tablespace_version;
+	ulint		zip_size;
 	ulint		count;
 	ulint		count2;
 	ulint		err;
 
+	zip_size = fil_space_get_zip_size(space);
 	tablespace_version = fil_space_get_version(space);
 
 	count = buf_read_ahead_random(space, offset);
@@ -328,7 +335,7 @@ buf_read_page(
 	switches: hence TRUE */
 
 	count2 = buf_read_page_low(&err, TRUE, BUF_READ_ANY_PAGE, space,
-					tablespace_version, offset);
+					zip_size, tablespace_version, offset);
 	srv_buf_pool_reads+= count2;
 	if (err == DB_TABLESPACE_DELETED) {
 		ut_print_timestamp(stderr);
@@ -377,6 +384,7 @@ buf_read_ahead_linear(
 			must want access to this page (see NOTE 3 above) */
 {
 	ib_longlong	tablespace_version;
+	ulint		zip_size;
 	buf_block_t*	block;
 	buf_frame_t*	frame;
 	buf_block_t*	pred_block	= NULL;
@@ -416,6 +424,8 @@ buf_read_ahead_linear(
 		return(0);
 	}
 
+	zip_size = fil_space_get_zip_size(space);
+
 	/* Remember the tablespace version before we ask te tablespace size
 	below: if DISCARD + IMPORT changes the actual .ibd file meanwhile, we
 	do not try to read outside the bounds of the tablespace! */
@@ -556,7 +566,8 @@ buf_read_ahead_linear(
 		if (!ibuf_bitmap_page(i)) {
 			count += buf_read_page_low(&err, FALSE, ibuf_mode
 					| OS_AIO_SIMULATED_WAKE_LATER,
-					space,	tablespace_version, i);
+					space, zip_size,
+					tablespace_version, i);
 			if (err == DB_TABLESPACE_DELETED) {
 				ut_print_timestamp(stderr);
 				fprintf(stderr,
@@ -623,13 +634,9 @@ buf_read_ibuf_merge_pages(
 	}
 
 	for (i = 0; i < n_stored; i++) {
-		if ((i + 1 == n_stored) && sync) {
-			buf_read_page_low(&err, TRUE, BUF_READ_ANY_PAGE,
-				space_ids[i], space_versions[i], page_nos[i]);
-		} else {
-			buf_read_page_low(&err, FALSE, BUF_READ_ANY_PAGE,
-				space_ids[i], space_versions[i], page_nos[i]);
-		}
+		buf_read_page_low(&err, sync && (i + 1 == n_stored),
+				BUF_READ_ANY_PAGE, space_ids[i], 0,
+				space_versions[i], page_nos[i]);
 
 		if (err == DB_TABLESPACE_DELETED) {
 			/* We have deleted or are deleting the single-table
@@ -669,10 +676,12 @@ buf_read_recv_pages(
 	ulint	n_stored)	/* in: number of page numbers in the array */
 {
 	ib_longlong	tablespace_version;
+	ulint		zip_size;
 	ulint		count;
 	ulint		err;
 	ulint		i;
 
+	zip_size = fil_space_get_zip_size(space);
 	tablespace_version = fil_space_get_version(space);
 
 	for (i = 0; i < n_stored; i++) {
@@ -704,11 +713,11 @@ buf_read_recv_pages(
 
 		if ((i + 1 == n_stored) && sync) {
 			buf_read_page_low(&err, TRUE, BUF_READ_ANY_PAGE, space,
-				tablespace_version, page_nos[i]);
+				zip_size, tablespace_version, page_nos[i]);
 		} else {
 			buf_read_page_low(&err, FALSE, BUF_READ_ANY_PAGE
-				| OS_AIO_SIMULATED_WAKE_LATER,
-				space, tablespace_version, page_nos[i]);
+				| OS_AIO_SIMULATED_WAKE_LATER, space,
+				zip_size, tablespace_version, page_nos[i]);
 		}
 	}
 

fsp/fsp0fsp.c

@@ -890,7 +890,7 @@ fsp_header_init(
 
 	mtr_x_lock(fil_space_get_latch(space), mtr);
 
-	page = buf_page_create(space, 0, mtr);
+	page = buf_page_create(space, 0, 0/* TODO: zip_size!=16k? */, mtr);
 	buf_page_get(space, 0, RW_X_LATCH, mtr);
 #ifdef UNIV_SYNC_DEBUG
 	buf_page_dbg_add_level(page, SYNC_FSP_PAGE);
@@ -1272,7 +1272,8 @@ fsp_fill_free_list(
 			pages should be ignored. */
 
 			if (i > 0) {
-				descr_page = buf_page_create(space, i, mtr);
+				/* TODO: zip_size != 16384 */
+				descr_page = buf_page_create(space, i, 0, mtr);
 				buf_page_get(space, i, RW_X_LATCH, mtr);
 #ifdef UNIV_SYNC_DEBUG
 				buf_page_dbg_add_level(descr_page,
@@ -1290,8 +1291,10 @@ fsp_fill_free_list(
 
 			mtr_start(&ibuf_mtr);
 
+			/* TODO: no ibuf on compressed tablespaces */
 			ibuf_page = buf_page_create(space,
-					i + FSP_IBUF_BITMAP_OFFSET, &ibuf_mtr);
+					i + FSP_IBUF_BITMAP_OFFSET,
+					0, &ibuf_mtr);
 			buf_page_get(space, i + FSP_IBUF_BITMAP_OFFSET,
 							RW_X_LATCH, &ibuf_mtr);
 #ifdef UNIV_SYNC_DEBUG
@@ -1511,7 +1514,8 @@ fsp_alloc_free_page(
 	be obtained immediately with buf_page_get without need for a disk
 	read. */
 
-	buf_page_create(space, page_no, mtr);
+	buf_page_create(space, page_no,
+			mach_read_from_4(FSP_PAGE_ZIP_SIZE + header), mtr);
 
 	page = buf_page_get(space, page_no, RW_X_LATCH, mtr);
 #ifdef UNIV_SYNC_DEBUG
@@ -2544,9 +2548,14 @@ fseg_alloc_free_page_low(
 		can be obtained immediately with buf_page_get without need
 		for a disk read */
 
-		page = buf_page_create(space, ret_page, mtr);
+		page = buf_page_create(space, ret_page,
+				mach_read_from_4(FSP_PAGE_ZIP_SIZE
+				+ space_header), mtr);
 
-		ut_a(page == buf_page_get(space, ret_page, RW_X_LATCH, mtr));
+		if (UNIV_UNLIKELY(page != buf_page_get(
+				space, ret_page, RW_X_LATCH, mtr))) {
+			ut_error;
+		}
 
 #ifdef UNIV_SYNC_DEBUG
 		buf_page_dbg_add_level(page, SYNC_FSP_PAGE);

include/buf0buf.h

@@ -238,6 +238,7 @@ buf_page_create(
 	ulint	space,	/* in: space id */
 	ulint	offset,	/* in: offset of the page within space in units of
 			a page */
+	ulint	zip_size,/* in: compressed page size, or 0 */
 	mtr_t*	mtr);	/* in: mini-transaction handle */
 /************************************************************************
 Inits a page to the buffer buf_pool, for use in ibbackup --restore. */
@@ -679,6 +680,7 @@ buf_page_init_for_read(
 	ulint*		err,	/* out: DB_SUCCESS or DB_TABLESPACE_DELETED */
 	ulint		mode,	/* in: BUF_READ_IBUF_PAGES_ONLY, ... */
 	ulint		space,	/* in: space id */
+	ulint		zip_size,/* in: compressed page size, or 0 */
 	ib_longlong	tablespace_version,/* in: prevents reading from a wrong
 				version of the tablespace in case we have done
 				DISCARD + IMPORT */

include/log0recv.h

@@ -214,6 +214,7 @@ void
 recv_recovery_from_archive_finish(void);
 /*===================================*/
 #endif /* UNIV_LOG_ARCHIVE */
+#ifdef UNIV_LOG_REPLICATE
 /***********************************************************************
 Checks that a replica of a space is identical to the original space. */
 
@@ -235,6 +236,7 @@ recv_compare_spaces_low(
 	ulint	space1,	/* in: space id */
 	ulint	space2,	/* in: space id */
 	ulint	n_pages);/* in: number of pages */
+#endif /* UNIV_LOG_REPLICATE */
 
 /* Block of log record data */
 typedef struct recv_data_struct	recv_data_t;

include/page0zip.ic

@@ -272,10 +272,9 @@ Write a log record of writing to the uncompressed header portion of a page. */
 void
 page_zip_write_header_log(
 /*======================*/
-	const page_zip_des_t*	page_zip,/* in: compressed page */
-	ulint			offset,	/* in: offset to the data */
-	ulint			length,	/* in: length of the data */
-	mtr_t*			mtr);	/* in: mini-transaction */
+	const byte*	data,/* in: data on the uncompressed page */
+	ulint		length,	/* in: length of the data */
+	mtr_t*		mtr);	/* in: mini-transaction */
 
 /**************************************************************************
 Write data to the uncompressed header portion of a page.  The data must
@@ -307,7 +306,7 @@ page_zip_write_header(
 	/* ut_ad(page_zip_validate(page_zip, str - pos)); */
 
 	if (UNIV_LIKELY_NULL(mtr)) {
-		page_zip_write_header_log(page_zip, pos, length, mtr);
+		page_zip_write_header_log(str, length, mtr);
 	}
 }
 

log/log0recv.c

@@ -1490,9 +1490,11 @@ loop:
 	mutex_exit(&(recv_sys->mutex));
 }
 
+#ifdef UNIV_HOTBACKUP
 /* This page is allocated from the buffer pool and used in the function
 below */
 static page_t* recv_backup_application_page	= NULL;
+TODO: define recv_backup_application_page_zip;
 
 /***********************************************************************
 Applies log records in the hash table to a backup. */
@@ -1618,8 +1620,9 @@ skip_this_recv_addr:
 
 	recv_sys_empty_hash();
 }
+#endif /* UNIV_HOTBACKUP */
 
-#ifdef notdefined
+#ifdef UNIV_LOG_REPLICATE
 /***********************************************************************
 In the debug version, updates the replica of a file page, based on a log
 record. */

page/page0cur.c

@@ -1123,7 +1123,8 @@ byte*
 page_copy_rec_list_to_created_page_write_log(
 /*=========================================*/
 				/* out: 4-byte field where to
-				write the log data length */
+				write the log data length,
+				or NULL if logging is disabled */
 	page_t*		page,	/* in: index page */
 	dict_index_t*	index,	/* in: record descriptor */
 	mtr_t*		mtr)	/* in: mtr */
@@ -1136,8 +1137,9 @@ page_copy_rec_list_to_created_page_write_log(
 			page_is_comp(page)
 			? MLOG_COMP_LIST_END_COPY_CREATED
 			: MLOG_LIST_END_COPY_CREATED, 4);
-	ut_a(log_ptr);
-	mlog_close(mtr, log_ptr + 4);
+	if (UNIV_LIKELY(log_ptr != NULL)) {
+		mlog_close(mtr, log_ptr + 4);
+	}
 
 	return(log_ptr);
 }
@@ -1335,7 +1337,9 @@ page_copy_rec_list_end_to_created_page(
 
 	ut_a(log_data_len < 100 * UNIV_PAGE_SIZE);
 
-	mach_write_to_4(log_ptr, log_data_len);
+	if (UNIV_LIKELY(log_ptr != NULL)) {
+		mach_write_to_4(log_ptr, log_data_len);
+	}
 
 	if (page_is_comp(new_page)) {
 		rec_set_next_offs_new(insert_rec, PAGE_NEW_SUPREMUM);

page/page0zip.c

@@ -2045,7 +2045,7 @@ page_zip_write_rec(
 						externs - ext_end);
 			}
 
-			ut_a(blob_no + n_ext < page_zip->n_blobs);
+			ut_a(blob_no + n_ext <= page_zip->n_blobs);
 		}
 
 		/* Store separately trx_id, roll_ptr and
@@ -2829,12 +2829,13 @@ Write a log record of writing to the uncompressed header portion of a page. */
 void
 page_zip_write_header_log(
 /*======================*/
-	const page_zip_des_t*	page_zip,/* in: compressed page */
-	ulint			offset,	/* in: offset to the data */
-	ulint			length,	/* in: length of the data */
-	mtr_t*			mtr)	/* in: mini-transaction */
+	const byte*	data,	/* in: data on the uncompressed page */
+	ulint		length,	/* in: length of the data */
+	mtr_t*		mtr)	/* in: mini-transaction */
 {
-	byte*	log_ptr = mlog_open(mtr, 11 + 1 + 1);
+	byte*	log_ptr	= mlog_open(mtr, 11 + 1 + 1);
+	ulint	offset	= ut_align_offset(data, UNIV_PAGE_SIZE);
+
 	ut_ad(offset < PAGE_DATA);
 	ut_ad(offset + length < PAGE_DATA);
 #if PAGE_DATA > 255
@@ -2848,13 +2849,13 @@ page_zip_write_header_log(
 		return;
 	}
 
-	log_ptr = mlog_write_initial_log_record_fast(page_zip->data + offset,
+	log_ptr = mlog_write_initial_log_record_fast((byte*) data,
 				MLOG_ZIP_WRITE_HEADER, log_ptr, mtr);
 	*log_ptr++ = (byte) offset;
 	*log_ptr++ = (byte) length;
 	mlog_close(mtr, log_ptr);
 
-	mlog_catenate_string(mtr, page_zip->data + offset, length);
+	mlog_catenate_string(mtr, data, length);
 }
 
 /**************************************************************************