summaryrefslogtreecommitdiff
path: root/storage
diff options
context:
space:
mode:
authorSergei Golubchik <sergii@pisem.net>2012-03-28 01:04:46 +0200
committerSergei Golubchik <sergii@pisem.net>2012-03-28 01:04:46 +0200
commit20e706689df1eb87c696304797e9d6184c0a75bb (patch)
treeff3eab8fa3e060b34687c6e9819cb559ac67d5c5 /storage
parent3d0775e9af2fcf3fe92b7f19e299ea23068eca1e (diff)
parentbfaebe3f5e4b917c4498e234bad7a9d45d07ca62 (diff)
downloadmariadb-git-20e706689df1eb87c696304797e9d6184c0a75bb.tar.gz
mysql-5.5.22 merge
mysql-test/suite/innodb/t/group_commit_crash.test: remove autoincrement to avoid rbr being used for insert ... select mysql-test/suite/innodb/t/group_commit_crash_no_optimize_thread.test: remove autoincrement to avoid rbr being used for insert ... select mysys/my_addr_resolve.c: a pointer to a buffer is returned to the caller -> the buffer cannot be on the stack mysys/stacktrace.c: my_vsnprintf() is ok here, in 5.5
Diffstat (limited to 'storage')
-rw-r--r--storage/archive/ha_archive.cc9
-rw-r--r--storage/blackhole/ha_blackhole.cc2
-rw-r--r--storage/csv/ha_tina.cc3
-rw-r--r--storage/example/ha_example.cc3
-rw-r--r--storage/federated/ha_federated.cc2
-rw-r--r--storage/heap/hp_info.c2
-rw-r--r--storage/innobase/btr/btr0btr.c237
-rw-r--r--storage/innobase/btr/btr0cur.c251
-rw-r--r--storage/innobase/btr/btr0sea.c8
-rw-r--r--storage/innobase/buf/buf0buf.c28
-rw-r--r--storage/innobase/buf/buf0flu.c3
-rw-r--r--storage/innobase/fil/fil0fil.c31
-rw-r--r--storage/innobase/fsp/fsp0fsp.c403
-rw-r--r--storage/innobase/handler/ha_innodb.cc96
-rw-r--r--storage/innobase/ibuf/ibuf0ibuf.c58
-rw-r--r--storage/innobase/include/btr0btr.h29
-rw-r--r--storage/innobase/include/btr0btr.ic2
-rw-r--r--storage/innobase/include/btr0cur.h88
-rw-r--r--storage/innobase/include/btr0cur.ic29
-rw-r--r--storage/innobase/include/buf0buf.h44
-rw-r--r--storage/innobase/include/buf0buf.ic21
-rw-r--r--storage/innobase/include/fsp0fsp.h41
-rw-r--r--storage/innobase/include/log0log.h6
-rw-r--r--storage/innobase/include/mtr0mtr.h13
-rw-r--r--storage/innobase/include/mtr0mtr.ic7
-rw-r--r--storage/innobase/include/page0page.h43
-rw-r--r--storage/innobase/include/page0page.ic38
-rw-r--r--storage/innobase/include/sync0sync.h10
-rw-r--r--storage/innobase/include/trx0rec.ic7
-rw-r--r--storage/innobase/include/trx0undo.h13
-rw-r--r--storage/innobase/lock/lock0lock.c2
-rw-r--r--storage/innobase/log/log0log.c6
-rw-r--r--storage/innobase/mtr/mtr0mtr.c6
-rw-r--r--storage/innobase/page/page0cur.c23
-rw-r--r--storage/innobase/page/page0page.c68
-rw-r--r--storage/innobase/row/row0ins.c83
-rw-r--r--storage/innobase/row/row0merge.c35
-rw-r--r--storage/innobase/row/row0row.c29
-rw-r--r--storage/innobase/row/row0sel.c9
-rw-r--r--storage/innobase/row/row0upd.c64
-rw-r--r--storage/innobase/srv/srv0start.c5
-rw-r--r--storage/innobase/sync/sync0rw.c10
-rw-r--r--storage/innobase/sync/sync0sync.c16
-rw-r--r--storage/innobase/trx/trx0rec.c37
-rw-r--r--storage/innobase/trx/trx0sys.c34
-rw-r--r--storage/innobase/trx/trx0trx.c2
-rw-r--r--storage/innobase/trx/trx0undo.c43
-rw-r--r--storage/myisam/ft_boolean_search.c2
-rw-r--r--storage/myisam/ft_nlq_search.c3
-rw-r--r--storage/myisam/ha_myisam.cc6
-rw-r--r--storage/myisam/ha_myisam.h3
-rw-r--r--storage/myisam/mi_check.c21
-rw-r--r--storage/myisam/mi_packrec.c2
-rw-r--r--storage/myisam/mi_preload.c2
-rw-r--r--storage/myisam/myisamchk.c7
-rw-r--r--storage/myisam/myisamdef.h2
-rw-r--r--storage/myisam/sort.c18
-rw-r--r--storage/perfschema/ha_perfschema.cc2
-rw-r--r--storage/perfschema/pfs.cc2
-rw-r--r--storage/perfschema/table_ews_global_by_event_name.cc2
-rw-r--r--storage/perfschema/table_ews_global_by_event_name.h2
-rw-r--r--storage/xtradb/row/row0merge.c2
-rw-r--r--storage/xtradb/row/row0sel.c4
63 files changed, 1362 insertions, 717 deletions
diff --git a/storage/archive/ha_archive.cc b/storage/archive/ha_archive.cc
index 25812a7ad48..6b29c6d9d53 100644
--- a/storage/archive/ha_archive.cc
+++ b/storage/archive/ha_archive.cc
@@ -1,5 +1,5 @@
/*
- Copyright (c) 2004, 2011, Oracle and/or its affiliates
+ Copyright (c) 2004, 2012, Oracle and/or its affiliates
This program is free software; you can redistribute it and/or
modify it under the terms of the GNU General Public License
@@ -1399,6 +1399,8 @@ int ha_archive::optimize(THD* thd, HA_CHECK_OPT* check_opt)
char writer_filename[FN_REFLEN];
DBUG_ENTER("ha_archive::optimize");
+ mysql_mutex_lock(&share->mutex);
+
if (init_archive_reader())
DBUG_RETURN(errno);
@@ -1414,7 +1416,10 @@ int ha_archive::optimize(THD* thd, HA_CHECK_OPT* check_opt)
MY_REPLACE_EXT | MY_UNPACK_FILENAME);
if (!(azopen(&writer, writer_filename, O_CREAT|O_RDWR|O_BINARY)))
+ {
+ mysql_mutex_unlock(&share->mutex);
DBUG_RETURN(HA_ERR_CRASHED_ON_USAGE);
+ }
/*
Transfer the embedded FRM so that the file can be discoverable.
@@ -1500,10 +1505,12 @@ int ha_archive::optimize(THD* thd, HA_CHECK_OPT* check_opt)
rc= my_rename(writer_filename, share->data_file_name, MYF(0));
+ mysql_mutex_unlock(&share->mutex);
DBUG_RETURN(rc);
error:
DBUG_PRINT("ha_archive", ("Failed to recover, error was %d", rc));
azclose(&writer);
+ mysql_mutex_unlock(&share->mutex);
DBUG_RETURN(rc);
}
diff --git a/storage/blackhole/ha_blackhole.cc b/storage/blackhole/ha_blackhole.cc
index 7bdbb4349cd..036dd374cb0 100644
--- a/storage/blackhole/ha_blackhole.cc
+++ b/storage/blackhole/ha_blackhole.cc
@@ -1,4 +1,4 @@
-/* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
+/* Copyright (c) 2005, 2011, Oracle and/or its affiliates. All rights reserved.
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
diff --git a/storage/csv/ha_tina.cc b/storage/csv/ha_tina.cc
index 64fd7102e93..9091c101183 100644
--- a/storage/csv/ha_tina.cc
+++ b/storage/csv/ha_tina.cc
@@ -1,5 +1,4 @@
-/*
- Copyright (c) 2004, 2010, Oracle and/or its affiliates
+/* Copyright (c) 2004, 2011, Oracle and/or its affiliates.
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
diff --git a/storage/example/ha_example.cc b/storage/example/ha_example.cc
index dcbfb176c6a..17ac1e962aa 100644
--- a/storage/example/ha_example.cc
+++ b/storage/example/ha_example.cc
@@ -1,5 +1,4 @@
-/*
- Copyright (c) 2004, 2010, Oracle and/or its affiliates
+/* Copyright (c) 2004, 2011, Oracle and/or its affiliates.
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
diff --git a/storage/federated/ha_federated.cc b/storage/federated/ha_federated.cc
index 4294f4b094f..549001ab08b 100644
--- a/storage/federated/ha_federated.cc
+++ b/storage/federated/ha_federated.cc
@@ -1,4 +1,4 @@
-/* Copyright (c) 2004, 2010, Oracle and/or its affiliates
+/* Copyright (c) 2004, 2011, Oracle and/or its affiliates.
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
diff --git a/storage/heap/hp_info.c b/storage/heap/hp_info.c
index c72dbaf2065..17f8c3fe97a 100644
--- a/storage/heap/hp_info.c
+++ b/storage/heap/hp_info.c
@@ -1,4 +1,4 @@
-/* Copyright (C) 2000-2004 MySQL AB
+/* Copyright (c) 2000, 2011, Oracle and/or its affiliates. All rights reserved.
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
diff --git a/storage/innobase/btr/btr0btr.c b/storage/innobase/btr/btr0btr.c
index 88371208230..3bbf74f3423 100644
--- a/storage/innobase/btr/btr0btr.c
+++ b/storage/innobase/btr/btr0btr.c
@@ -11,8 +11,8 @@ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with
-this program; if not, write to the Free Software Foundation, Inc., 59 Temple
-Place, Suite 330, Boston, MA 02111-1307 USA
+this program; if not, write to the Free Software Foundation, Inc.,
+51 Franklin Street, Suite 500, Boston, MA 02110-1335 USA
*****************************************************************************/
@@ -56,11 +56,12 @@ btr_corruption_report(
(unsigned) buf_block_get_space(block),
(unsigned) buf_block_get_page_no(block),
index->name, index->table_name);
- buf_page_print(buf_block_get_frame(block), 0);
if (block->page.zip.data) {
buf_page_print(block->page.zip.data,
- buf_block_get_zip_size(block));
+ buf_block_get_zip_size(block),
+ BUF_PAGE_PRINT_NO_CRASH);
}
+ buf_page_print(buf_block_get_frame(block), 0, 0);
}
#ifdef UNIV_BLOB_DEBUG
@@ -926,28 +927,31 @@ btr_page_alloc_for_ibuf(
/**************************************************************//**
Allocates a new file page to be used in an index tree. NOTE: we assume
that the caller has made the reservation for free extents!
-@return new allocated block, x-latched; NULL if out of space */
-UNIV_INTERN
+@retval NULL if no page could be allocated
+@retval block, rw_lock_x_lock_count(&block->lock) == 1 if allocation succeeded
+(init_mtr == mtr, or the page was not previously freed in mtr)
+@retval block (not allocated or initialized) otherwise */
+static __attribute__((nonnull, warn_unused_result))
buf_block_t*
-btr_page_alloc(
-/*===========*/
+btr_page_alloc_low(
+/*===============*/
dict_index_t* index, /*!< in: index */
ulint hint_page_no, /*!< in: hint of a good page */
byte file_direction, /*!< in: direction where a possible
page split is made */
ulint level, /*!< in: level where the page is placed
in the tree */
- mtr_t* mtr) /*!< in: mtr */
+ mtr_t* mtr, /*!< in/out: mini-transaction
+ for the allocation */
+ mtr_t* init_mtr) /*!< in/out: mtr or another
+ mini-transaction in which the
+ page should be initialized.
+ If init_mtr!=mtr, but the page
+ is already X-latched in mtr, do
+ not initialize the page. */
{
fseg_header_t* seg_header;
page_t* root;
- buf_block_t* new_block;
- ulint new_page_no;
-
- if (dict_index_is_ibuf(index)) {
-
- return(btr_page_alloc_for_ibuf(index, mtr));
- }
root = btr_root_get(index, mtr);
@@ -961,17 +965,47 @@ btr_page_alloc(
reservation for free extents, and thus we know that a page can
be allocated: */
- new_page_no = fseg_alloc_free_page_general(seg_header, hint_page_no,
- file_direction, TRUE, mtr);
- if (new_page_no == FIL_NULL) {
+ return(fseg_alloc_free_page_general(
+ seg_header, hint_page_no, file_direction,
+ TRUE, mtr, init_mtr));
+}
- return(NULL);
+/**************************************************************//**
+Allocates a new file page to be used in an index tree. NOTE: we assume
+that the caller has made the reservation for free extents!
+@retval NULL if no page could be allocated
+@retval block, rw_lock_x_lock_count(&block->lock) == 1 if allocation succeeded
+(init_mtr == mtr, or the page was not previously freed in mtr)
+@retval block (not allocated or initialized) otherwise */
+UNIV_INTERN
+buf_block_t*
+btr_page_alloc(
+/*===========*/
+ dict_index_t* index, /*!< in: index */
+ ulint hint_page_no, /*!< in: hint of a good page */
+ byte file_direction, /*!< in: direction where a possible
+ page split is made */
+ ulint level, /*!< in: level where the page is placed
+ in the tree */
+ mtr_t* mtr, /*!< in/out: mini-transaction
+ for the allocation */
+ mtr_t* init_mtr) /*!< in/out: mini-transaction
+ for x-latching and initializing
+ the page */
+{
+ buf_block_t* new_block;
+
+ if (dict_index_is_ibuf(index)) {
+
+ return(btr_page_alloc_for_ibuf(index, mtr));
}
- new_block = buf_page_get(dict_index_get_space(index),
- dict_table_zip_size(index->table),
- new_page_no, RW_X_LATCH, mtr);
- buf_block_dbg_add_level(new_block, SYNC_TREE_NODE_NEW);
+ new_block = btr_page_alloc_low(
+ index, hint_page_no, file_direction, level, mtr, init_mtr);
+
+ if (new_block) {
+ buf_block_dbg_add_level(new_block, SYNC_TREE_NODE_NEW);
+ }
return(new_block);
}
@@ -1098,10 +1132,10 @@ btr_page_free(
buf_block_t* block, /*!< in: block to be freed, x-latched */
mtr_t* mtr) /*!< in: mtr */
{
- ulint level;
-
- level = btr_page_get_level(buf_block_get_frame(block), mtr);
+ const page_t* page = buf_block_get_frame(block);
+ ulint level = btr_page_get_level(page, mtr);
+ ut_ad(fil_page_get_type(block->frame) == FIL_PAGE_INDEX);
btr_page_free_low(index, block, level, mtr);
}
@@ -1215,9 +1249,11 @@ btr_page_get_father_node_ptr_func(
!= page_no)) {
rec_t* print_rec;
fputs("InnoDB: Dump of the child page:\n", stderr);
- buf_page_print(page_align(user_rec), 0);
+ buf_page_print(page_align(user_rec), 0,
+ BUF_PAGE_PRINT_NO_CRASH);
fputs("InnoDB: Dump of the parent page:\n", stderr);
- buf_page_print(page_align(node_ptr), 0);
+ buf_page_print(page_align(node_ptr), 0,
+ BUF_PAGE_PRINT_NO_CRASH);
fputs("InnoDB: Corruption of an index tree: table ", stderr);
ut_print_name(stderr, NULL, TRUE, index->table_name);
@@ -1341,16 +1377,12 @@ btr_create(
/* Allocate then the next page to the segment: it will be the
tree root page */
- page_no = fseg_alloc_free_page(buf_block_get_frame(
- ibuf_hdr_block)
- + IBUF_HEADER
- + IBUF_TREE_SEG_HEADER,
- IBUF_TREE_ROOT_PAGE_NO,
- FSP_UP, mtr);
- ut_ad(page_no == IBUF_TREE_ROOT_PAGE_NO);
-
- block = buf_page_get(space, zip_size, page_no,
- RW_X_LATCH, mtr);
+ block = fseg_alloc_free_page(
+ buf_block_get_frame(ibuf_hdr_block)
+ + IBUF_HEADER + IBUF_TREE_SEG_HEADER,
+ IBUF_TREE_ROOT_PAGE_NO,
+ FSP_UP, mtr);
+ ut_ad(buf_block_get_page_no(block) == IBUF_TREE_ROOT_PAGE_NO);
} else {
#ifdef UNIV_BLOB_DEBUG
if ((type & DICT_CLUSTERED) && !index->blobs) {
@@ -1654,8 +1686,8 @@ btr_page_reorganize_low(
if (UNIV_UNLIKELY(data_size1 != data_size2)
|| UNIV_UNLIKELY(max_ins_size1 != max_ins_size2)) {
- buf_page_print(page, 0);
- buf_page_print(temp_page, 0);
+ buf_page_print(page, 0, BUF_PAGE_PRINT_NO_CRASH);
+ buf_page_print(temp_page, 0, BUF_PAGE_PRINT_NO_CRASH);
fprintf(stderr,
"InnoDB: Error: page old data size %lu"
" new data size %lu\n"
@@ -1666,6 +1698,7 @@ btr_page_reorganize_low(
(unsigned long) data_size1, (unsigned long) data_size2,
(unsigned long) max_ins_size1,
(unsigned long) max_ins_size2);
+ ut_ad(0);
} else {
success = TRUE;
}
@@ -1829,7 +1862,7 @@ btr_root_raise_and_insert(
level = btr_page_get_level(root, mtr);
- new_block = btr_page_alloc(index, 0, FSP_NO_DIR, level, mtr);
+ new_block = btr_page_alloc(index, 0, FSP_NO_DIR, level, mtr, mtr);
new_page = buf_block_get_frame(new_block);
new_page_zip = buf_block_get_page_zip(new_block);
ut_a(!new_page_zip == !root_page_zip);
@@ -2565,7 +2598,7 @@ func_start:
/* 2. Allocate a new page to the index */
new_block = btr_page_alloc(cursor->index, hint_page_no, direction,
- btr_page_get_level(page, mtr), mtr);
+ btr_page_get_level(page, mtr), mtr, mtr);
new_page = buf_block_get_frame(new_block);
new_page_zip = buf_block_get_page_zip(new_block);
btr_page_create(new_block, new_page_zip, cursor->index,
@@ -3015,15 +3048,16 @@ btr_node_ptr_delete(
ut_a(err == DB_SUCCESS);
if (!compressed) {
- btr_cur_compress_if_useful(&cursor, mtr);
+ btr_cur_compress_if_useful(&cursor, FALSE, mtr);
}
}
/*************************************************************//**
If page is the only on its level, this function moves its records to the
-father page, thus reducing the tree height. */
+father page, thus reducing the tree height.
+@return father block */
static
-void
+buf_block_t*
btr_lift_page_up(
/*=============*/
dict_index_t* index, /*!< in: index tree */
@@ -3140,6 +3174,8 @@ btr_lift_page_up(
}
ut_ad(page_validate(father_page, index));
ut_ad(btr_check_node_ptr(index, father_block, mtr));
+
+ return(father_block);
}
/*************************************************************//**
@@ -3156,11 +3192,13 @@ UNIV_INTERN
ibool
btr_compress(
/*=========*/
- btr_cur_t* cursor, /*!< in: cursor on the page to merge or lift;
- the page must not be empty: in record delete
- use btr_discard_page if the page would become
- empty */
- mtr_t* mtr) /*!< in: mtr */
+ btr_cur_t* cursor, /*!< in/out: cursor on the page to merge
+ or lift; the page must not be empty:
+ when deleting records, use btr_discard_page()
+ if the page would become empty */
+ ibool adjust, /*!< in: TRUE if should adjust the
+ cursor position even if compression occurs */
+ mtr_t* mtr) /*!< in/out: mini-transaction */
{
dict_index_t* index;
ulint space;
@@ -3178,6 +3216,7 @@ btr_compress(
ulint* offsets;
ulint data_size;
ulint n_recs;
+ ulint nth_rec = 0; /* remove bogus warning */
ulint max_ins_size;
ulint max_ins_size_reorg;
@@ -3205,6 +3244,10 @@ btr_compress(
offsets = btr_page_get_father_block(NULL, heap, index, block, mtr,
&father_cursor);
+ if (adjust) {
+ nth_rec = page_rec_get_n_recs_before(btr_cur_get_rec(cursor));
+ }
+
/* Decide the page to which we try to merge and which will inherit
the locks */
@@ -3231,9 +3274,9 @@ btr_compress(
} else {
/* The page is the only one on the level, lift the records
to the father */
- btr_lift_page_up(index, block, mtr);
- mem_heap_free(heap);
- return(TRUE);
+
+ merge_block = btr_lift_page_up(index, block, mtr);
+ goto func_exit;
}
n_recs = page_get_n_recs(page);
@@ -3315,6 +3358,10 @@ err_exit:
btr_node_ptr_delete(index, block, mtr);
lock_update_merge_left(merge_block, orig_pred, block);
+
+ if (adjust) {
+ nth_rec += page_rec_get_n_recs_before(orig_pred);
+ }
} else {
rec_t* orig_succ;
#ifdef UNIV_BTR_DEBUG
@@ -3379,7 +3426,6 @@ err_exit:
}
btr_blob_dbg_remove(page, index, "btr_compress");
- mem_heap_free(heap);
if (!dict_index_is_clust(index) && page_is_leaf(merge_page)) {
/* Update the free bits of the B-tree page in the
@@ -3431,6 +3477,16 @@ err_exit:
btr_page_free(index, block, mtr);
ut_ad(btr_check_node_ptr(index, merge_block, mtr));
+func_exit:
+ mem_heap_free(heap);
+
+ if (adjust) {
+ btr_cur_position(
+ index,
+ page_rec_get_nth(merge_block->frame, nth_rec),
+ merge_block, cursor);
+ }
+
return(TRUE);
}
@@ -3867,7 +3923,7 @@ btr_index_rec_validate(
(ulong) rec_get_n_fields_old(rec), (ulong) n);
if (dump_on_error) {
- buf_page_print(page, 0);
+ buf_page_print(page, 0, BUF_PAGE_PRINT_NO_CRASH);
fputs("InnoDB: corrupt record ", stderr);
rec_print_old(stderr, rec);
@@ -3905,7 +3961,8 @@ btr_index_rec_validate(
(ulong) i, (ulong) len, (ulong) fixed_size);
if (dump_on_error) {
- buf_page_print(page, 0);
+ buf_page_print(page, 0,
+ BUF_PAGE_PRINT_NO_CRASH);
fputs("InnoDB: corrupt record ", stderr);
rec_print_new(stderr, rec, offsets);
@@ -4115,8 +4172,8 @@ loop:
btr_validate_report2(index, level, block, right_block);
fputs("InnoDB: broken FIL_PAGE_NEXT"
" or FIL_PAGE_PREV links\n", stderr);
- buf_page_print(page, 0);
- buf_page_print(right_page, 0);
+ buf_page_print(page, 0, BUF_PAGE_PRINT_NO_CRASH);
+ buf_page_print(right_page, 0, BUF_PAGE_PRINT_NO_CRASH);
ret = FALSE;
}
@@ -4125,8 +4182,8 @@ loop:
!= page_is_comp(page))) {
btr_validate_report2(index, level, block, right_block);
fputs("InnoDB: 'compact' flag mismatch\n", stderr);
- buf_page_print(page, 0);
- buf_page_print(right_page, 0);
+ buf_page_print(page, 0, BUF_PAGE_PRINT_NO_CRASH);
+ buf_page_print(right_page, 0, BUF_PAGE_PRINT_NO_CRASH);
ret = FALSE;
@@ -4149,8 +4206,8 @@ loop:
fputs("InnoDB: records in wrong order"
" on adjacent pages\n", stderr);
- buf_page_print(page, 0);
- buf_page_print(right_page, 0);
+ buf_page_print(page, 0, BUF_PAGE_PRINT_NO_CRASH);
+ buf_page_print(right_page, 0, BUF_PAGE_PRINT_NO_CRASH);
fputs("InnoDB: record ", stderr);
rec = page_rec_get_prev(page_get_supremum_rec(page));
@@ -4199,8 +4256,8 @@ loop:
fputs("InnoDB: node pointer to the page is wrong\n",
stderr);
- buf_page_print(father_page, 0);
- buf_page_print(page, 0);
+ buf_page_print(father_page, 0, BUF_PAGE_PRINT_NO_CRASH);
+ buf_page_print(page, 0, BUF_PAGE_PRINT_NO_CRASH);
fputs("InnoDB: node ptr ", stderr);
rec_print(stderr, node_ptr, index);
@@ -4232,8 +4289,10 @@ loop:
btr_validate_report1(index, level, block);
- buf_page_print(father_page, 0);
- buf_page_print(page, 0);
+ buf_page_print(father_page, 0,
+ BUF_PAGE_PRINT_NO_CRASH);
+ buf_page_print(page, 0,
+ BUF_PAGE_PRINT_NO_CRASH);
fputs("InnoDB: Error: node ptrs differ"
" on levels > 0\n"
@@ -4278,9 +4337,15 @@ loop:
btr_validate_report1(index, level,
block);
- buf_page_print(father_page, 0);
- buf_page_print(page, 0);
- buf_page_print(right_page, 0);
+ buf_page_print(
+ father_page, 0,
+ BUF_PAGE_PRINT_NO_CRASH);
+ buf_page_print(
+ page, 0,
+ BUF_PAGE_PRINT_NO_CRASH);
+ buf_page_print(
+ right_page, 0,
+ BUF_PAGE_PRINT_NO_CRASH);
}
} else {
page_t* right_father_page
@@ -4298,10 +4363,18 @@ loop:
btr_validate_report1(index, level,
block);
- buf_page_print(father_page, 0);
- buf_page_print(right_father_page, 0);
- buf_page_print(page, 0);
- buf_page_print(right_page, 0);
+ buf_page_print(
+ father_page, 0,
+ BUF_PAGE_PRINT_NO_CRASH);
+ buf_page_print(
+ right_father_page, 0,
+ BUF_PAGE_PRINT_NO_CRASH);
+ buf_page_print(
+ page, 0,
+ BUF_PAGE_PRINT_NO_CRASH);
+ buf_page_print(
+ right_page, 0,
+ BUF_PAGE_PRINT_NO_CRASH);
}
if (page_get_page_no(right_father_page)
@@ -4315,10 +4388,18 @@ loop:
btr_validate_report1(index, level,
block);
- buf_page_print(father_page, 0);
- buf_page_print(right_father_page, 0);
- buf_page_print(page, 0);
- buf_page_print(right_page, 0);
+ buf_page_print(
+ father_page, 0,
+ BUF_PAGE_PRINT_NO_CRASH);
+ buf_page_print(
+ right_father_page, 0,
+ BUF_PAGE_PRINT_NO_CRASH);
+ buf_page_print(
+ page, 0,
+ BUF_PAGE_PRINT_NO_CRASH);
+ buf_page_print(
+ right_page, 0,
+ BUF_PAGE_PRINT_NO_CRASH);
}
}
}
diff --git a/storage/innobase/btr/btr0cur.c b/storage/innobase/btr/btr0cur.c
index 1551aee5426..e227e47b52f 100644
--- a/storage/innobase/btr/btr0cur.c
+++ b/storage/innobase/btr/btr0cur.c
@@ -18,8 +18,8 @@ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with
-this program; if not, write to the Free Software Foundation, Inc., 59 Temple
-Place, Suite 330, Boston, MA 02111-1307 USA
+this program; if not, write to the Free Software Foundation, Inc.,
+51 Franklin Street, Suite 500, Boston, MA 02110-1335 USA
*****************************************************************************/
@@ -1860,7 +1860,7 @@ btr_cur_update_in_place(
page_zip = buf_block_get_page_zip(block);
/* Check that enough space is available on the compressed page. */
- if (UNIV_LIKELY_NULL(page_zip)
+ if (page_zip
&& !btr_cur_update_alloc_zip(page_zip, block, index,
rec_offs_size(offsets), FALSE, mtr)) {
return(DB_ZIP_OVERFLOW);
@@ -2051,7 +2051,7 @@ any_extern:
ut_a(!page_zip || page_zip_validate(page_zip, page));
#endif /* UNIV_ZIP_DEBUG */
- if (UNIV_LIKELY_NULL(page_zip)
+ if (page_zip
&& !btr_cur_update_alloc_zip(page_zip, block, index,
new_rec_size, TRUE, mtr)) {
err = DB_ZIP_OVERFLOW;
@@ -2209,7 +2209,9 @@ btr_cur_pessimistic_update(
/*=======================*/
ulint flags, /*!< in: undo logging, locking, and rollback
flags */
- btr_cur_t* cursor, /*!< in: cursor on the record to update */
+ btr_cur_t* cursor, /*!< in/out: cursor on the record to update;
+ cursor may become invalid if *big_rec == NULL
+ || !(flags & BTR_KEEP_POS_FLAG) */
mem_heap_t** heap, /*!< in/out: pointer to memory heap, or NULL */
big_rec_t** big_rec,/*!< out: big rec vector whose fields have to
be stored externally by the caller, or NULL */
@@ -2348,10 +2350,10 @@ btr_cur_pessimistic_update(
record to be inserted: we have to remember which fields were such */
ut_ad(!page_is_comp(page) || !rec_get_node_ptr_flag(rec));
- offsets = rec_get_offsets(rec, index, offsets, ULINT_UNDEFINED, heap);
+ ut_ad(rec_offs_validate(rec, index, offsets));
n_ext += btr_push_update_extern_fields(new_entry, update, *heap);
- if (UNIV_LIKELY_NULL(page_zip)) {
+ if (page_zip) {
ut_ad(page_is_comp(page));
if (page_zip_rec_needs_ext(
rec_get_converted_size(index, new_entry, n_ext),
@@ -2371,6 +2373,10 @@ make_external:
err = DB_TOO_BIG_RECORD;
goto return_after_reservations;
}
+
+ ut_ad(page_is_leaf(page));
+ ut_ad(dict_index_is_clust(index));
+ ut_ad(flags & BTR_KEEP_POS_FLAG);
}
/* Store state of explicit locks on rec on the page infimum record,
@@ -2398,6 +2404,8 @@ make_external:
rec = btr_cur_insert_if_possible(cursor, new_entry, n_ext, mtr);
if (rec) {
+ page_cursor->rec = rec;
+
lock_rec_restore_from_page_infimum(btr_cur_get_block(cursor),
rec, block);
@@ -2411,7 +2419,10 @@ make_external:
rec, index, offsets, mtr);
}
- btr_cur_compress_if_useful(cursor, mtr);
+ btr_cur_compress_if_useful(
+ cursor,
+ big_rec_vec != NULL && (flags & BTR_KEEP_POS_FLAG),
+ mtr);
if (page_zip && !dict_index_is_clust(index)
&& page_is_leaf(page)) {
@@ -2431,6 +2442,21 @@ make_external:
}
}
+ if (big_rec_vec) {
+ ut_ad(page_is_leaf(page));
+ ut_ad(dict_index_is_clust(index));
+ ut_ad(flags & BTR_KEEP_POS_FLAG);
+
+ /* btr_page_split_and_insert() in
+ btr_cur_pessimistic_insert() invokes
+ mtr_memo_release(mtr, index->lock, MTR_MEMO_X_LOCK).
+ We must keep the index->lock when we created a
+ big_rec, so that row_upd_clust_rec() can store the
+ big_rec in the same mini-transaction. */
+
+ mtr_x_lock(dict_index_get_lock(index), mtr);
+ }
+
/* Was the record to be updated positioned as the first user
record on its page? */
was_first = page_cur_is_before_first(page_cursor);
@@ -2446,6 +2472,7 @@ make_external:
ut_a(rec);
ut_a(err == DB_SUCCESS);
ut_a(dummy_big_rec == NULL);
+ page_cursor->rec = rec;
if (dict_index_is_sec_or_ibuf(index)) {
/* Update PAGE_MAX_TRX_ID in the index page header.
@@ -2880,10 +2907,12 @@ UNIV_INTERN
ibool
btr_cur_compress_if_useful(
/*=======================*/
- btr_cur_t* cursor, /*!< in: cursor on the page to compress;
- cursor does not stay valid if compression
- occurs */
- mtr_t* mtr) /*!< in: mtr */
+ btr_cur_t* cursor, /*!< in/out: cursor on the page to compress;
+ cursor does not stay valid if !adjust and
+ compression occurs */
+ ibool adjust, /*!< in: TRUE if should adjust the
+ cursor position even if compression occurs */
+ mtr_t* mtr) /*!< in/out: mini-transaction */
{
ut_ad(mtr_memo_contains(mtr,
dict_index_get_lock(btr_cur_get_index(cursor)),
@@ -2892,7 +2921,7 @@ btr_cur_compress_if_useful(
MTR_MEMO_PAGE_X_FIX));
return(btr_cur_compress_recommendation(cursor, mtr)
- && btr_compress(cursor, mtr));
+ && btr_compress(cursor, adjust, mtr));
}
/*******************************************************//**
@@ -3134,7 +3163,7 @@ return_after_reservations:
mem_heap_free(heap);
if (ret == FALSE) {
- ret = btr_cur_compress_if_useful(cursor, mtr);
+ ret = btr_cur_compress_if_useful(cursor, FALSE, mtr);
}
if (n_extents > 0) {
@@ -3870,10 +3899,10 @@ btr_cur_set_ownership_of_extern_field(
byte_val = byte_val | BTR_EXTERN_OWNER_FLAG;
}
- if (UNIV_LIKELY_NULL(page_zip)) {
+ if (page_zip) {
mach_write_to_1(data + local_len + BTR_EXTERN_LEN, byte_val);
page_zip_write_blob_ptr(page_zip, rec, index, offsets, i, mtr);
- } else if (UNIV_LIKELY(mtr != NULL)) {
+ } else if (mtr != NULL) {
mlog_write_ulint(data + local_len + BTR_EXTERN_LEN, byte_val,
MLOG_1BYTE, mtr);
@@ -4106,9 +4135,9 @@ The fields are stored on pages allocated from leaf node
file segment of the index tree.
@return DB_SUCCESS or DB_OUT_OF_FILE_SPACE */
UNIV_INTERN
-ulint
-btr_store_big_rec_extern_fields_func(
-/*=================================*/
+enum db_err
+btr_store_big_rec_extern_fields(
+/*============================*/
dict_index_t* index, /*!< in: index of rec; the index tree
MUST be X-latched */
buf_block_t* rec_block, /*!< in/out: block containing rec */
@@ -4117,38 +4146,37 @@ btr_store_big_rec_extern_fields_func(
the "external storage" flags in offsets
will not correspond to rec when
this function returns */
-#ifdef UNIV_DEBUG
- mtr_t* local_mtr, /*!< in: mtr containing the
- latch to rec and to the tree */
-#endif /* UNIV_DEBUG */
-#if defined UNIV_DEBUG || defined UNIV_BLOB_LIGHT_DEBUG
- ibool update_in_place,/*! in: TRUE if the record is updated
- in place (not delete+insert) */
-#endif /* UNIV_DEBUG || UNIV_BLOB_LIGHT_DEBUG */
- const big_rec_t*big_rec_vec) /*!< in: vector containing fields
+ const big_rec_t*big_rec_vec, /*!< in: vector containing fields
to be stored externally */
-
+ mtr_t* btr_mtr, /*!< in: mtr containing the
+ latches to the clustered index */
+ enum blob_op op) /*! in: operation code */
{
- ulint rec_page_no;
- byte* field_ref;
- ulint extern_len;
- ulint store_len;
- ulint page_no;
- ulint space_id;
- ulint zip_size;
- ulint prev_page_no;
- ulint hint_page_no;
- ulint i;
- mtr_t mtr;
- mem_heap_t* heap = NULL;
+ ulint rec_page_no;
+ byte* field_ref;
+ ulint extern_len;
+ ulint store_len;
+ ulint page_no;
+ ulint space_id;
+ ulint zip_size;
+ ulint prev_page_no;
+ ulint hint_page_no;
+ ulint i;
+ mtr_t mtr;
+ mtr_t* alloc_mtr;
+ mem_heap_t* heap = NULL;
page_zip_des_t* page_zip;
- z_stream c_stream;
+ z_stream c_stream;
+ buf_block_t** freed_pages = NULL;
+ ulint n_freed_pages = 0;
+ enum db_err error = DB_SUCCESS;
ut_ad(rec_offs_validate(rec, index, offsets));
ut_ad(rec_offs_any_extern(offsets));
- ut_ad(mtr_memo_contains(local_mtr, dict_index_get_lock(index),
+ ut_ad(btr_mtr);
+ ut_ad(mtr_memo_contains(btr_mtr, dict_index_get_lock(index),
MTR_MEMO_X_LOCK));
- ut_ad(mtr_memo_contains(local_mtr, rec_block, MTR_MEMO_PAGE_X_FIX));
+ ut_ad(mtr_memo_contains(btr_mtr, rec_block, MTR_MEMO_PAGE_X_FIX));
ut_ad(buf_block_get_frame(rec_block) == page_align(rec));
ut_a(dict_index_is_clust(index));
@@ -4161,7 +4189,7 @@ btr_store_big_rec_extern_fields_func(
rec_page_no = buf_block_get_page_no(rec_block);
ut_a(fil_page_get_type(page_align(rec)) == FIL_PAGE_INDEX);
- if (UNIV_LIKELY_NULL(page_zip)) {
+ if (page_zip) {
int err;
/* Zlib deflate needs 128 kilobytes for the default
@@ -4177,6 +4205,42 @@ btr_store_big_rec_extern_fields_func(
ut_a(err == Z_OK);
}
+ if (btr_blob_op_is_update(op)) {
+ /* Avoid reusing pages that have been previously freed
+ in btr_mtr. */
+ if (btr_mtr->n_freed_pages) {
+ if (heap == NULL) {
+ heap = mem_heap_create(
+ btr_mtr->n_freed_pages
+ * sizeof *freed_pages);
+ }
+
+ freed_pages = mem_heap_alloc(
+ heap,
+ btr_mtr->n_freed_pages
+ * sizeof *freed_pages);
+ n_freed_pages = 0;
+ }
+
+ /* Because btr_mtr will be committed after mtr, it is
+ possible that the tablespace has been extended when
+ the B-tree record was updated or inserted, or it will
+ be extended while allocating pages for big_rec.
+
+ TODO: In mtr (not btr_mtr), write a redo log record
+ about extending the tablespace to its current size,
+ and remember the current size. Whenever the tablespace
+ grows as pages are allocated, write further redo log
+ records to mtr. (Currently tablespace extension is not
+ covered by the redo log. If it were, the record would
+ only be written to btr_mtr, which is committed after
+ mtr.) */
+ alloc_mtr = btr_mtr;
+ } else {
+ /* Use the local mtr for allocations. */
+ alloc_mtr = &mtr;
+ }
+
#if defined UNIV_DEBUG || defined UNIV_BLOB_LIGHT_DEBUG
/* All pointers to externally stored columns in the record
must either be zero or they must be pointers to inherited
@@ -4191,7 +4255,7 @@ btr_store_big_rec_extern_fields_func(
/* Either this must be an update in place,
or the BLOB must be inherited, or the BLOB pointer
must be zero (will be written in this function). */
- ut_a(update_in_place
+ ut_a(op == BTR_STORE_UPDATE
|| (field_ref[BTR_EXTERN_LEN] & BTR_EXTERN_INHERITED_FLAG)
|| !memcmp(field_ref, field_ref_zero,
BTR_EXTERN_FIELD_REF_SIZE));
@@ -4216,7 +4280,7 @@ btr_store_big_rec_extern_fields_func(
prev_page_no = FIL_NULL;
- if (UNIV_LIKELY_NULL(page_zip)) {
+ if (page_zip) {
int err = deflateReset(&c_stream);
ut_a(err == Z_OK);
@@ -4236,18 +4300,24 @@ btr_store_big_rec_extern_fields_func(
hint_page_no = prev_page_no + 1;
}
+alloc_another:
block = btr_page_alloc(index, hint_page_no,
- FSP_NO_DIR, 0, &mtr);
+ FSP_NO_DIR, 0, alloc_mtr, &mtr);
if (UNIV_UNLIKELY(block == NULL)) {
-
mtr_commit(&mtr);
+ error = DB_OUT_OF_FILE_SPACE;
+ goto func_exit;
+ }
- if (UNIV_LIKELY_NULL(page_zip)) {
- deflateEnd(&c_stream);
- mem_heap_free(heap);
- }
-
- return(DB_OUT_OF_FILE_SPACE);
+ if (rw_lock_get_x_lock_count(&block->lock) > 1) {
+ /* This page must have been freed in
+ btr_mtr previously. Put it aside, and
+ allocate another page for the BLOB data. */
+ ut_ad(alloc_mtr == btr_mtr);
+ ut_ad(btr_blob_op_is_update(op));
+ ut_ad(n_freed_pages < btr_mtr->n_freed_pages);
+ freed_pages[n_freed_pages++] = block;
+ goto alloc_another;
}
page_no = buf_block_get_page_no(block);
@@ -4264,7 +4334,7 @@ btr_store_big_rec_extern_fields_func(
SYNC_EXTERN_STORAGE);
prev_page = buf_block_get_frame(prev_block);
- if (UNIV_LIKELY_NULL(page_zip)) {
+ if (page_zip) {
mlog_write_ulint(
prev_page + FIL_PAGE_NEXT,
page_no, MLOG_4BYTES, &mtr);
@@ -4281,7 +4351,7 @@ btr_store_big_rec_extern_fields_func(
}
- if (UNIV_LIKELY_NULL(page_zip)) {
+ if (page_zip) {
int err;
page_zip_des_t* blob_page_zip;
@@ -4364,11 +4434,15 @@ btr_store_big_rec_extern_fields_func(
goto next_zip_page;
}
- rec_block = buf_page_get(space_id, zip_size,
- rec_page_no,
- RW_X_LATCH, &mtr);
- buf_block_dbg_add_level(rec_block,
- SYNC_NO_ORDER_CHECK);
+ if (alloc_mtr == &mtr) {
+ rec_block = buf_page_get(
+ space_id, zip_size,
+ rec_page_no,
+ RW_X_LATCH, &mtr);
+ buf_block_dbg_add_level(
+ rec_block,
+ SYNC_NO_ORDER_CHECK);
+ }
if (err == Z_STREAM_END) {
mach_write_to_4(field_ref
@@ -4402,7 +4476,8 @@ btr_store_big_rec_extern_fields_func(
page_zip_write_blob_ptr(
page_zip, rec, index, offsets,
- big_rec_vec->fields[i].field_no, &mtr);
+ big_rec_vec->fields[i].field_no,
+ alloc_mtr);
next_zip_page:
prev_page_no = page_no;
@@ -4447,19 +4522,23 @@ next_zip_page:
extern_len -= store_len;
- rec_block = buf_page_get(space_id, zip_size,
- rec_page_no,
- RW_X_LATCH, &mtr);
- buf_block_dbg_add_level(rec_block,
- SYNC_NO_ORDER_CHECK);
+ if (alloc_mtr == &mtr) {
+ rec_block = buf_page_get(
+ space_id, zip_size,
+ rec_page_no,
+ RW_X_LATCH, &mtr);
+ buf_block_dbg_add_level(
+ rec_block,
+ SYNC_NO_ORDER_CHECK);
+ }
mlog_write_ulint(field_ref + BTR_EXTERN_LEN, 0,
- MLOG_4BYTES, &mtr);
+ MLOG_4BYTES, alloc_mtr);
mlog_write_ulint(field_ref
+ BTR_EXTERN_LEN + 4,
big_rec_vec->fields[i].len
- extern_len,
- MLOG_4BYTES, &mtr);
+ MLOG_4BYTES, alloc_mtr);
if (prev_page_no == FIL_NULL) {
btr_blob_dbg_add_blob(
@@ -4469,18 +4548,19 @@ next_zip_page:
mlog_write_ulint(field_ref
+ BTR_EXTERN_SPACE_ID,
- space_id,
- MLOG_4BYTES, &mtr);
+ space_id, MLOG_4BYTES,
+ alloc_mtr);
mlog_write_ulint(field_ref
+ BTR_EXTERN_PAGE_NO,
- page_no,
- MLOG_4BYTES, &mtr);
+ page_no, MLOG_4BYTES,
+ alloc_mtr);
mlog_write_ulint(field_ref
+ BTR_EXTERN_OFFSET,
FIL_PAGE_DATA,
- MLOG_4BYTES, &mtr);
+ MLOG_4BYTES,
+ alloc_mtr);
}
prev_page_no = page_no;
@@ -4494,8 +4574,23 @@ next_zip_page:
}
}
- if (UNIV_LIKELY_NULL(page_zip)) {
+func_exit:
+ if (page_zip) {
deflateEnd(&c_stream);
+ }
+
+ if (n_freed_pages) {
+ ulint i;
+
+ ut_ad(alloc_mtr == btr_mtr);
+ ut_ad(btr_blob_op_is_update(op));
+
+ for (i = 0; i < n_freed_pages; i++) {
+ btr_page_free_low(index, freed_pages[i], 0, alloc_mtr);
+ }
+ }
+
+ if (heap != NULL) {
mem_heap_free(heap);
}
@@ -4516,7 +4611,7 @@ next_zip_page:
ut_a(!(field_ref[BTR_EXTERN_LEN] & BTR_EXTERN_OWNER_FLAG));
}
#endif /* UNIV_DEBUG || UNIV_BLOB_LIGHT_DEBUG */
- return(DB_SUCCESS);
+ return(error);
}
/*******************************************************************//**
@@ -4719,7 +4814,7 @@ btr_free_externally_stored_field(
btr_page_free_low(index, ext_block, 0, &mtr);
- if (UNIV_LIKELY(page_zip != NULL)) {
+ if (page_zip) {
mach_write_to_4(field_ref + BTR_EXTERN_PAGE_NO,
next_page_no);
mach_write_to_4(field_ref + BTR_EXTERN_LEN + 4,
diff --git a/storage/innobase/btr/btr0sea.c b/storage/innobase/btr/btr0sea.c
index 1825544a265..53a0c0eb5d9 100644
--- a/storage/innobase/btr/btr0sea.c
+++ b/storage/innobase/btr/btr0sea.c
@@ -1,6 +1,6 @@
/*****************************************************************************
-Copyright (c) 1996, 2011, Oracle and/or its affiliates. All Rights Reserved.
+Copyright (c) 1996, 2012, Oracle and/or its affiliates. All Rights Reserved.
Copyright (c) 2008, Google Inc.
Portions of this file contain modifications contributed and copyrighted by
@@ -1176,7 +1176,7 @@ cleanup:
index->name, (ulong) block->n_pointers);
rw_lock_x_unlock(&btr_search_latch);
- btr_search_validate();
+ ut_ad(btr_search_validate());
} else {
rw_lock_x_unlock(&btr_search_latch);
}
@@ -1908,7 +1908,9 @@ btr_search_validate(void)
(ulong) block->curr_left_side);
if (n_page_dumps < 20) {
- buf_page_print(page, 0);
+ buf_page_print(
+ page, 0,
+ BUF_PAGE_PRINT_NO_CRASH);
n_page_dumps++;
}
}
diff --git a/storage/innobase/buf/buf0buf.c b/storage/innobase/buf/buf0buf.c
index 0721929abd3..7b13caf6216 100644
--- a/storage/innobase/buf/buf0buf.c
+++ b/storage/innobase/buf/buf0buf.c
@@ -605,8 +605,12 @@ void
buf_page_print(
/*===========*/
const byte* read_buf, /*!< in: a database page */
- ulint zip_size) /*!< in: compressed page size, or
- 0 for uncompressed pages */
+ ulint zip_size, /*!< in: compressed page size, or
+ 0 for uncompressed pages */
+ ulint flags) /*!< in: 0 or
+ BUF_PAGE_PRINT_NO_CRASH or
+ BUF_PAGE_PRINT_NO_FULL */
+
{
#ifndef UNIV_HOTBACKUP
dict_index_t* index;
@@ -615,17 +619,18 @@ buf_page_print(
ulint old_checksum;
ulint size = zip_size;
- ut_ad(0);
-
if (!size) {
size = UNIV_PAGE_SIZE;
}
- ut_print_timestamp(stderr);
- fprintf(stderr, " InnoDB: Page dump in ascii and hex (%lu bytes):\n",
- (ulong) size);
- ut_print_buf(stderr, read_buf, size);
- fputs("\nInnoDB: End of page dump\n", stderr);
+ if (!(flags & BUF_PAGE_PRINT_NO_FULL)) {
+ ut_print_timestamp(stderr);
+ fprintf(stderr,
+ " InnoDB: Page dump in ascii and hex (%lu bytes):\n",
+ (ulong) size);
+ ut_print_buf(stderr, read_buf, size);
+ fputs("\nInnoDB: End of page dump\n", stderr);
+ }
if (zip_size) {
/* Print compressed page. */
@@ -797,6 +802,8 @@ buf_page_print(
stderr);
break;
}
+
+ ut_ad(flags & BUF_PAGE_PRINT_NO_CRASH);
}
#ifndef UNIV_HOTBACKUP
@@ -3557,7 +3564,8 @@ corrupt:
"InnoDB: You may have to recover"
" from a backup.\n",
(ulong) bpage->offset);
- buf_page_print(frame, buf_page_get_zip_size(bpage));
+ buf_page_print(frame, buf_page_get_zip_size(bpage),
+ BUF_PAGE_PRINT_NO_CRASH);
fprintf(stderr,
"InnoDB: Database page corruption on disk"
" or a failed\n"
diff --git a/storage/innobase/buf/buf0flu.c b/storage/innobase/buf/buf0flu.c
index 06b3ec393c1..6e8e8fdda5a 100644
--- a/storage/innobase/buf/buf0flu.c
+++ b/storage/innobase/buf/buf0flu.c
@@ -757,7 +757,8 @@ buf_flush_buffered_writes(void)
if (UNIV_UNLIKELY
(!page_simple_validate_new(block->frame))) {
corrupted_page:
- buf_page_print(block->frame, 0);
+ buf_page_print(block->frame, 0,
+ BUF_PAGE_PRINT_NO_CRASH);
ut_print_timestamp(stderr);
fprintf(stderr,
diff --git a/storage/innobase/fil/fil0fil.c b/storage/innobase/fil/fil0fil.c
index 2e4c6aeeb60..091becca258 100644
--- a/storage/innobase/fil/fil0fil.c
+++ b/storage/innobase/fil/fil0fil.c
@@ -964,11 +964,6 @@ retry:
return;
}
- if (fil_system->n_open < fil_system->max_n_open) {
-
- return;
- }
-
space = fil_space_get_by_id(space_id);
if (space != NULL && space->stop_ios) {
@@ -985,6 +980,25 @@ retry:
mutex_exit(&fil_system->mutex);
+#ifndef UNIV_HOTBACKUP
+
+ /* Wake the i/o-handler threads to make sure pending
+ i/o's are performed */
+ os_aio_simulated_wake_handler_threads();
+
+ /* The sleep here is just to give IO helper threads a
+ bit of time to do some work. It is not required that
+ all IO related to the tablespace being renamed must
+ be flushed here as we do fil_flush() in
+ fil_rename_tablespace() as well. */
+ os_thread_sleep(20000);
+
+#endif /* UNIV_HOTBACKUP */
+
+ /* Flush tablespaces so that we can close modified
+ files in the LRU list */
+ fil_flush_file_spaces(FIL_TABLESPACE);
+
os_thread_sleep(20000);
count2++;
@@ -992,6 +1006,11 @@ retry:
goto retry;
}
+ if (fil_system->n_open < fil_system->max_n_open) {
+
+ return;
+ }
+
/* If the file is already open, no need to do anything; if the space
does not exist, we handle the situation in the function which called
this function */
@@ -2551,7 +2570,7 @@ fil_rename_tablespace(
retry:
count++;
- if (count > 1000) {
+ if (!(count % 1000)) {
ut_print_timestamp(stderr);
fputs(" InnoDB: Warning: problems renaming ", stderr);
ut_print_filename(stderr, old_name);
diff --git a/storage/innobase/fsp/fsp0fsp.c b/storage/innobase/fsp/fsp0fsp.c
index 2d626405c5c..6a3cd912141 100644
--- a/storage/innobase/fsp/fsp0fsp.c
+++ b/storage/innobase/fsp/fsp0fsp.c
@@ -1,6 +1,6 @@
/*****************************************************************************
-Copyright (c) 1995, 2011, Oracle and/or its affiliates. All Rights Reserved.
+Copyright (c) 1995, 2012, Oracle and/or its affiliates. All Rights Reserved.
This program is free software; you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free Software
@@ -250,28 +250,38 @@ fsp_fill_free_list(
descriptor page and ibuf bitmap page;
then we do not allocate more extents */
ulint space, /*!< in: space */
- fsp_header_t* header, /*!< in: space header */
- mtr_t* mtr); /*!< in: mtr */
+ fsp_header_t* header, /*!< in/out: space header */
+ mtr_t* mtr) /*!< in/out: mini-transaction */
+ __attribute__((nonnull));
/**********************************************************************//**
Allocates a single free page from a segment. This function implements
the intelligent allocation strategy which tries to minimize file space
fragmentation.
-@return the allocated page number, FIL_NULL if no page could be allocated */
+@retval NULL if no page could be allocated
+@retval block, rw_lock_x_lock_count(&block->lock) == 1 if allocation succeeded
+(init_mtr == mtr, or the page was not previously freed in mtr)
+@retval block (not allocated or initialized) otherwise */
static
-ulint
+buf_block_t*
fseg_alloc_free_page_low(
/*=====================*/
ulint space, /*!< in: space */
ulint zip_size,/*!< in: compressed page size in bytes
or 0 for uncompressed pages */
- fseg_inode_t* seg_inode, /*!< in: segment inode */
- ulint hint, /*!< in: hint of which page would be desirable */
+ fseg_inode_t* seg_inode, /*!< in/out: segment inode */
+ ulint hint, /*!< in: hint of which page would be
+ desirable */
byte direction, /*!< in: if the new page is needed because
of an index page split, and records are
inserted there in order, into which
direction they go alphabetically: FSP_DOWN,
FSP_UP, FSP_NO_DIR */
- mtr_t* mtr); /*!< in: mtr handle */
+ mtr_t* mtr, /*!< in/out: mini-transaction */
+ mtr_t* init_mtr)/*!< in/out: mtr or another mini-transaction
+ in which the page should be initialized.
+ If init_mtr!=mtr, but the page is already
+ latched in mtr, do not initialize the page. */
+ __attribute__((warn_unused_result, nonnull));
#endif /* !UNIV_HOTBACKUP */
/**********************************************************************//**
@@ -639,17 +649,18 @@ list, if not free limit == space size. This adding is necessary to make the
descriptor defined, as they are uninitialized above the free limit.
@return pointer to the extent descriptor, NULL if the page does not
exist in the space or if the offset exceeds the free limit */
-UNIV_INLINE
+UNIV_INLINE __attribute__((nonnull, warn_unused_result))
xdes_t*
xdes_get_descriptor_with_space_hdr(
/*===============================*/
- fsp_header_t* sp_header,/*!< in/out: space header, x-latched */
- ulint space, /*!< in: space id */
- ulint offset, /*!< in: page offset;
- if equal to the free limit,
- we try to add new extents to
- the space free list */
- mtr_t* mtr) /*!< in: mtr handle */
+ fsp_header_t* sp_header, /*!< in/out: space header, x-latched
+ in mtr */
+ ulint space, /*!< in: space id */
+ ulint offset, /*!< in: page offset; if equal
+ to the free limit, we try to
+ add new extents to the space
+ free list */
+ mtr_t* mtr) /*!< in/out: mini-transaction */
{
ulint limit;
ulint size;
@@ -657,11 +668,9 @@ xdes_get_descriptor_with_space_hdr(
ulint descr_page_no;
page_t* descr_page;
- ut_ad(mtr);
ut_ad(mtr_memo_contains(mtr, fil_space_get_latch(space, NULL),
MTR_MEMO_X_LOCK));
- ut_ad(mtr_memo_contains_page(mtr, sp_header, MTR_MEMO_PAGE_S_FIX)
- || mtr_memo_contains_page(mtr, sp_header, MTR_MEMO_PAGE_X_FIX));
+ ut_ad(mtr_memo_contains_page(mtr, sp_header, MTR_MEMO_PAGE_X_FIX));
ut_ad(page_offset(sp_header) == FSP_HEADER_OFFSET);
/* Read free limit and space size */
limit = mach_read_from_4(sp_header + FSP_FREE_LIMIT);
@@ -711,7 +720,7 @@ is necessary to make the descriptor defined, as they are uninitialized
above the free limit.
@return pointer to the extent descriptor, NULL if the page does not
exist in the space or if the offset exceeds the free limit */
-static
+static __attribute__((nonnull, warn_unused_result))
xdes_t*
xdes_get_descriptor(
/*================*/
@@ -720,7 +729,7 @@ xdes_get_descriptor(
or 0 for uncompressed pages */
ulint offset, /*!< in: page offset; if equal to the free limit,
we try to add new extents to the space free list */
- mtr_t* mtr) /*!< in: mtr handle */
+ mtr_t* mtr) /*!< in/out: mini-transaction */
{
buf_block_t* block;
fsp_header_t* sp_header;
@@ -1098,14 +1107,14 @@ fsp_header_get_tablespace_size(void)
Tries to extend a single-table tablespace so that a page would fit in the
data file.
@return TRUE if success */
-static
+static __attribute__((nonnull, warn_unused_result))
ibool
fsp_try_extend_data_file_with_pages(
/*================================*/
ulint space, /*!< in: space */
ulint page_no, /*!< in: page number */
- fsp_header_t* header, /*!< in: space header */
- mtr_t* mtr) /*!< in: mtr */
+ fsp_header_t* header, /*!< in/out: space header */
+ mtr_t* mtr) /*!< in/out: mini-transaction */
{
ibool success;
ulint actual_size;
@@ -1130,7 +1139,7 @@ fsp_try_extend_data_file_with_pages(
/***********************************************************************//**
Tries to extend the last data file of a tablespace if it is auto-extending.
@return FALSE if not auto-extending */
-static
+static __attribute__((nonnull))
ibool
fsp_try_extend_data_file(
/*=====================*/
@@ -1140,8 +1149,8 @@ fsp_try_extend_data_file(
the actual file size rounded down to
megabyte */
ulint space, /*!< in: space */
- fsp_header_t* header, /*!< in: space header */
- mtr_t* mtr) /*!< in: mtr */
+ fsp_header_t* header, /*!< in/out: space header */
+ mtr_t* mtr) /*!< in/out: mini-transaction */
{
ulint size;
ulint zip_size;
@@ -1277,7 +1286,7 @@ fsp_fill_free_list(
then we do not allocate more extents */
ulint space, /*!< in: space */
fsp_header_t* header, /*!< in/out: space header */
- mtr_t* mtr) /*!< in: mtr */
+ mtr_t* mtr) /*!< in/out: mini-transaction */
{
ulint limit;
ulint size;
@@ -1476,29 +1485,120 @@ fsp_alloc_free_extent(
}
/**********************************************************************//**
-Allocates a single free page from a space. The page is marked as used.
-@return the page offset, FIL_NULL if no page could be allocated */
+Allocates a single free page from a space. */
+static __attribute__((nonnull))
+void
+fsp_alloc_from_free_frag(
+/*=====================*/
+ fsp_header_t* header, /*!< in/out: tablespace header */
+ xdes_t* descr, /*!< in/out: extent descriptor */
+ ulint bit, /*!< in: slot to allocate in the extent */
+ mtr_t* mtr) /*!< in/out: mini-transaction */
+{
+ ulint frag_n_used;
+
+ ut_ad(xdes_get_state(descr, mtr) == XDES_FREE_FRAG);
+ ut_a(xdes_get_bit(descr, XDES_FREE_BIT, bit, mtr));
+ xdes_set_bit(descr, XDES_FREE_BIT, bit, FALSE, mtr);
+
+ /* Update the FRAG_N_USED field */
+ frag_n_used = mtr_read_ulint(header + FSP_FRAG_N_USED, MLOG_4BYTES,
+ mtr);
+ frag_n_used++;
+ mlog_write_ulint(header + FSP_FRAG_N_USED, frag_n_used, MLOG_4BYTES,
+ mtr);
+ if (xdes_is_full(descr, mtr)) {
+ /* The fragment is full: move it to another list */
+ flst_remove(header + FSP_FREE_FRAG, descr + XDES_FLST_NODE,
+ mtr);
+ xdes_set_state(descr, XDES_FULL_FRAG, mtr);
+
+ flst_add_last(header + FSP_FULL_FRAG, descr + XDES_FLST_NODE,
+ mtr);
+ mlog_write_ulint(header + FSP_FRAG_N_USED,
+ frag_n_used - FSP_EXTENT_SIZE, MLOG_4BYTES,
+ mtr);
+ }
+}
+
+/**********************************************************************//**
+Gets a buffer block for an allocated page.
+
+NOTE: If init_mtr != mtr, the block will only be initialized if it was
+not previously x-latched. It is assumed that the block has been
+x-latched only by mtr, and freed in mtr in that case.
+
+@return block, initialized if init_mtr==mtr
+or rw_lock_x_lock_count(&block->lock) == 1 */
static
-ulint
+buf_block_t*
+fsp_page_create(
+/*============*/
+ ulint space, /*!< in: space id of the allocated page */
+ ulint zip_size, /*!< in: compressed page size in bytes
+ or 0 for uncompressed pages */
+ ulint page_no, /*!< in: page number of the allocated page */
+ mtr_t* mtr, /*!< in: mini-transaction of the allocation */
+ mtr_t* init_mtr) /*!< in: mini-transaction for initializing
+ the page */
+{
+ buf_block_t* block
+ = buf_page_create(space, page_no, zip_size, init_mtr);
+#ifdef UNIV_SYNC_DEBUG
+ ut_ad(mtr_memo_contains(mtr, block, MTR_MEMO_PAGE_X_FIX)
+ == rw_lock_own(&block->lock, RW_LOCK_EX));
+#endif /* UNIV_SYNC_DEBUG */
+
+ /* Mimic buf_page_get(), but avoid the buf_pool->page_hash lookup. */
+ rw_lock_x_lock(&block->lock);
+ mutex_enter(&block->mutex);
+ buf_block_buf_fix_inc(block, __FILE__, __LINE__);
+ mutex_exit(&block->mutex);
+ mtr_memo_push(init_mtr, block, MTR_MEMO_PAGE_X_FIX);
+
+ if (init_mtr == mtr
+ || rw_lock_get_x_lock_count(&block->lock) == 1) {
+
+ /* Initialize the page, unless it was already
+ X-latched in mtr. (In this case, we would want to
+ allocate another page that has not been freed in mtr.) */
+ ut_ad(init_mtr == mtr
+ || !mtr_memo_contains(mtr, block, MTR_MEMO_PAGE_X_FIX));
+
+ fsp_init_file_page(block, init_mtr);
+ }
+
+ return(block);
+}
+
+/**********************************************************************//**
+Allocates a single free page from a space. The page is marked as used.
+@retval NULL if no page could be allocated
+@retval block, rw_lock_x_lock_count(&block->lock) == 1 if allocation succeeded
+(init_mtr == mtr, or the page was not previously freed in mtr)
+@retval block (not allocated or initialized) otherwise */
+static __attribute__((nonnull, warn_unused_result))
+buf_block_t*
fsp_alloc_free_page(
/*================*/
ulint space, /*!< in: space id */
ulint zip_size,/*!< in: compressed page size in bytes
or 0 for uncompressed pages */
ulint hint, /*!< in: hint of which page would be desirable */
- mtr_t* mtr) /*!< in: mtr handle */
+ mtr_t* mtr, /*!< in/out: mini-transaction */
+ mtr_t* init_mtr)/*!< in/out: mini-transaction in which the
+ page should be initialized
+ (may be the same as mtr) */
{
fsp_header_t* header;
fil_addr_t first;
xdes_t* descr;
- buf_block_t* block;
ulint free;
- ulint frag_n_used;
ulint page_no;
ulint space_size;
- ibool success;
ut_ad(mtr);
+ ut_ad(init_mtr);
header = fsp_get_space_header(space, zip_size, mtr);
@@ -1525,7 +1625,7 @@ fsp_alloc_free_page(
if (descr == NULL) {
/* No free space left */
- return(FIL_NULL);
+ return(NULL);
}
xdes_set_state(descr, XDES_FREE_FRAG, mtr);
@@ -1570,50 +1670,18 @@ fsp_alloc_free_page(
" space size %lu. Page no %lu.\n",
(ulong) space, (ulong) space_size,
(ulong) page_no);
- return(FIL_NULL);
+ return(NULL);
}
- success = fsp_try_extend_data_file_with_pages(space, page_no,
- header, mtr);
- if (!success) {
+ if (!fsp_try_extend_data_file_with_pages(space, page_no,
+ header, mtr)) {
/* No disk space left */
- return(FIL_NULL);
+ return(NULL);
}
}
- xdes_set_bit(descr, XDES_FREE_BIT, free, FALSE, mtr);
+ fsp_alloc_from_free_frag(header, descr, free, mtr);
- /* Update the FRAG_N_USED field */
- frag_n_used = mtr_read_ulint(header + FSP_FRAG_N_USED, MLOG_4BYTES,
- mtr);
- frag_n_used++;
- mlog_write_ulint(header + FSP_FRAG_N_USED, frag_n_used, MLOG_4BYTES,
- mtr);
- if (xdes_is_full(descr, mtr)) {
- /* The fragment is full: move it to another list */
- flst_remove(header + FSP_FREE_FRAG, descr + XDES_FLST_NODE,
- mtr);
- xdes_set_state(descr, XDES_FULL_FRAG, mtr);
-
- flst_add_last(header + FSP_FULL_FRAG, descr + XDES_FLST_NODE,
- mtr);
- mlog_write_ulint(header + FSP_FRAG_N_USED,
- frag_n_used - FSP_EXTENT_SIZE, MLOG_4BYTES,
- mtr);
- }
-
- /* Initialize the allocated page to the buffer pool, so that it can
- be obtained immediately with buf_page_get without need for a disk
- read. */
-
- buf_page_create(space, page_no, zip_size, mtr);
-
- block = buf_page_get(space, zip_size, page_no, RW_X_LATCH, mtr);
- buf_block_dbg_add_level(block, SYNC_FSP_PAGE);
-
- /* Prior contents of the page should be ignored */
- fsp_init_file_page(block, mtr);
-
- return(page_no);
+ return(fsp_page_create(space, zip_size, page_no, mtr, init_mtr));
}
/**********************************************************************//**
@@ -1652,6 +1720,9 @@ fsp_free_page(
fputs("InnoDB: Dump of descriptor: ", stderr);
ut_print_buf(stderr, ((byte*)descr) - 50, 200);
putc('\n', stderr);
+ /* Crash in debug version, so that we get a core dump
+ of this corruption. */
+ ut_ad(0);
if (state == XDES_FREE) {
/* We put here some fault tolerance: if the page
@@ -1670,6 +1741,9 @@ fsp_free_page(
"InnoDB: Dump of descriptor: ", (ulong) page);
ut_print_buf(stderr, ((byte*)descr) - 50, 200);
putc('\n', stderr);
+ /* Crash in debug version, so that we get a core dump
+ of this corruption. */
+ ut_ad(0);
/* We put here some fault tolerance: if the page
is already free, return without doing anything! */
@@ -1704,6 +1778,8 @@ fsp_free_page(
mtr);
fsp_free_extent(space, zip_size, page, mtr);
}
+
+ mtr->n_freed_pages++;
}
/**********************************************************************//**
@@ -1836,7 +1912,6 @@ fsp_alloc_seg_inode_page(
fseg_inode_t* inode;
buf_block_t* block;
page_t* page;
- ulint page_no;
ulint space;
ulint zip_size;
ulint i;
@@ -1847,15 +1922,15 @@ fsp_alloc_seg_inode_page(
zip_size = dict_table_flags_to_zip_size(
mach_read_from_4(FSP_SPACE_FLAGS + space_header));
- page_no = fsp_alloc_free_page(space, zip_size, 0, mtr);
+ block = fsp_alloc_free_page(space, zip_size, 0, mtr, mtr);
- if (page_no == FIL_NULL) {
+ if (block == NULL) {
return(FALSE);
}
- block = buf_page_get(space, zip_size, page_no, RW_X_LATCH, mtr);
buf_block_dbg_add_level(block, SYNC_FSP_PAGE);
+ ut_ad(rw_lock_get_x_lock_count(&block->lock) == 1);
block->check_index_page_at_flush = FALSE;
@@ -2258,19 +2333,20 @@ fseg_create_general(
}
if (page == 0) {
- page = fseg_alloc_free_page_low(space, zip_size,
- inode, 0, FSP_UP, mtr);
+ block = fseg_alloc_free_page_low(space, zip_size,
+ inode, 0, FSP_UP, mtr, mtr);
- if (page == FIL_NULL) {
+ if (block == NULL) {
fsp_free_seg_inode(space, zip_size, inode, mtr);
goto funct_exit;
}
- block = buf_page_get(space, zip_size, page, RW_X_LATCH, mtr);
+ ut_ad(rw_lock_get_x_lock_count(&block->lock) == 1);
+
header = byte_offset + buf_block_get_frame(block);
- mlog_write_ulint(header - byte_offset + FIL_PAGE_TYPE,
+ mlog_write_ulint(buf_block_get_frame(block) + FIL_PAGE_TYPE,
FIL_PAGE_TYPE_SYS, MLOG_2BYTES, mtr);
}
@@ -2447,8 +2523,10 @@ fseg_fill_free_list(
Allocates a free extent for the segment: looks first in the free list of the
segment, then tries to allocate from the space free list. NOTE that the extent
returned still resides in the segment free list, it is not yet taken off it!
-@return allocated extent, still placed in the segment free list, NULL
-if could not be allocated */
+@retval NULL if no page could be allocated
+@retval block, rw_lock_x_lock_count(&block->lock) == 1 if allocation succeeded
+(init_mtr == mtr, or the page was not previously freed in mtr)
+@retval block (not allocated or initialized) otherwise */
static
xdes_t*
fseg_alloc_free_extent(
@@ -2500,22 +2578,30 @@ fseg_alloc_free_extent(
Allocates a single free page from a segment. This function implements
the intelligent allocation strategy which tries to minimize file space
fragmentation.
-@return the allocated page number, FIL_NULL if no page could be allocated */
+@retval NULL if no page could be allocated
+@retval block, rw_lock_x_lock_count(&block->lock) == 1 if allocation succeeded
+(init_mtr == mtr, or the page was not previously freed in mtr)
+@retval block (not allocated or initialized) otherwise */
static
-ulint
+buf_block_t*
fseg_alloc_free_page_low(
/*=====================*/
ulint space, /*!< in: space */
ulint zip_size,/*!< in: compressed page size in bytes
or 0 for uncompressed pages */
- fseg_inode_t* seg_inode, /*!< in: segment inode */
- ulint hint, /*!< in: hint of which page would be desirable */
+ fseg_inode_t* seg_inode, /*!< in/out: segment inode */
+ ulint hint, /*!< in: hint of which page would be
+ desirable */
byte direction, /*!< in: if the new page is needed because
of an index page split, and records are
inserted there in order, into which
direction they go alphabetically: FSP_DOWN,
FSP_UP, FSP_NO_DIR */
- mtr_t* mtr) /*!< in: mtr handle */
+ mtr_t* mtr, /*!< in/out: mini-transaction */
+ mtr_t* init_mtr)/*!< in/out: mtr or another mini-transaction
+ in which the page should be initialized.
+ If init_mtr!=mtr, but the page is already
+ latched in mtr, do not initialize the page. */
{
fsp_header_t* space_header;
ulint space_size;
@@ -2526,7 +2612,6 @@ fseg_alloc_free_page_low(
ulint ret_page; /*!< the allocated page offset, FIL_NULL
if could not be allocated */
xdes_t* ret_descr; /*!< the extent of the allocated page */
- ibool frag_page_allocated = FALSE;
ibool success;
ulint n;
@@ -2548,6 +2633,7 @@ fseg_alloc_free_page_low(
if (descr == NULL) {
/* Hint outside space or too high above free limit: reset
hint */
+ /* The file space header page is always allocated. */
hint = 0;
descr = xdes_get_descriptor(space, zip_size, hint, mtr);
}
@@ -2558,15 +2644,19 @@ fseg_alloc_free_page_low(
&& mach_read_from_8(descr + XDES_ID) == seg_id
&& (xdes_get_bit(descr, XDES_FREE_BIT,
hint % FSP_EXTENT_SIZE, mtr) == TRUE)) {
-
+take_hinted_page:
/* 1. We can take the hinted page
=================================*/
ret_descr = descr;
ret_page = hint;
+ /* Skip the check for extending the tablespace. If the
+ page hint were not within the size of the tablespace,
+ we would have got (descr == NULL) above and reset the hint. */
+ goto got_hinted_page;
/*-----------------------------------------------------------*/
- } else if ((xdes_get_state(descr, mtr) == XDES_FREE)
- && ((reserved - used) < reserved / FSEG_FILLFACTOR)
- && (used >= FSEG_FRAG_LIMIT)) {
+ } else if (xdes_get_state(descr, mtr) == XDES_FREE
+ && reserved - used < reserved / FSEG_FILLFACTOR
+ && used >= FSEG_FRAG_LIMIT) {
/* 2. We allocate the free extent from space and can take
=========================================================
@@ -2584,7 +2674,7 @@ fseg_alloc_free_page_low(
/* Try to fill the segment free list */
fseg_fill_free_list(seg_inode, space, zip_size,
hint + FSP_EXTENT_SIZE, mtr);
- ret_page = hint;
+ goto take_hinted_page;
/*-----------------------------------------------------------*/
} else if ((direction != FSP_NO_DIR)
&& ((reserved - used) < reserved / FSEG_FILLFACTOR)
@@ -2632,7 +2722,7 @@ fseg_alloc_free_page_low(
first = flst_get_first(seg_inode + FSEG_FREE, mtr);
} else {
ut_error;
- return(FIL_NULL);
+ return(NULL);
}
ret_descr = xdes_lst_get_descriptor(space, zip_size,
@@ -2644,20 +2734,23 @@ fseg_alloc_free_page_low(
} else if (used < FSEG_FRAG_LIMIT) {
/* 6. We allocate an individual page from the space
===================================================*/
- ret_page = fsp_alloc_free_page(space, zip_size, hint, mtr);
- ret_descr = NULL;
+ buf_block_t* block = fsp_alloc_free_page(
+ space, zip_size, hint, mtr, init_mtr);
- frag_page_allocated = TRUE;
-
- if (ret_page != FIL_NULL) {
+ if (block != NULL) {
/* Put the page in the fragment page array of the
segment */
n = fseg_find_free_frag_page_slot(seg_inode, mtr);
- ut_a(n != FIL_NULL);
+ ut_a(n != ULINT_UNDEFINED);
- fseg_set_nth_frag_page_no(seg_inode, n, ret_page,
- mtr);
+ fseg_set_nth_frag_page_no(
+ seg_inode, n, buf_block_get_page_no(block),
+ mtr);
}
+
+ /* fsp_alloc_free_page() invoked fsp_init_file_page()
+ already. */
+ return(block);
/*-----------------------------------------------------------*/
} else {
/* 7. We allocate a new extent and take its first page
@@ -2675,7 +2768,7 @@ fseg_alloc_free_page_low(
if (ret_page == FIL_NULL) {
/* Page could not be allocated */
- return(FIL_NULL);
+ return(NULL);
}
if (space != 0) {
@@ -2693,38 +2786,22 @@ fseg_alloc_free_page_low(
" the space size %lu. Page no %lu.\n",
(ulong) space, (ulong) space_size,
(ulong) ret_page);
- return(FIL_NULL);
+ return(NULL);
}
success = fsp_try_extend_data_file_with_pages(
space, ret_page, space_header, mtr);
if (!success) {
/* No disk space left */
- return(FIL_NULL);
+ return(NULL);
}
}
}
- if (!frag_page_allocated) {
- /* Initialize the allocated page to buffer pool, so that it
- can be obtained immediately with buf_page_get without need
- for a disk read */
- buf_block_t* block;
- ulint zip_size = dict_table_flags_to_zip_size(
- mach_read_from_4(FSP_SPACE_FLAGS + space_header));
-
- block = buf_page_create(space, ret_page, zip_size, mtr);
- buf_block_dbg_add_level(block, SYNC_FSP_PAGE);
-
- if (UNIV_UNLIKELY(block != buf_page_get(space, zip_size,
- ret_page, RW_X_LATCH,
- mtr))) {
- ut_error;
- }
-
- /* The prior contents of the page should be ignored */
- fsp_init_file_page(block, mtr);
-
+got_hinted_page:
+ /* ret_descr == NULL if the block was allocated from free_frag
+ (XDES_FREE_FRAG) */
+ if (ret_descr != NULL) {
/* At this point we know the extent and the page offset.
The extent is still in the appropriate list (FSEG_NOT_FULL
or FSEG_FREE), and the page is not yet marked as used. */
@@ -2737,22 +2814,28 @@ fseg_alloc_free_page_low(
fseg_mark_page_used(seg_inode, space, zip_size, ret_page, mtr);
}
- buf_reset_check_index_page_at_flush(space, ret_page);
-
- return(ret_page);
+ return(fsp_page_create(
+ space, dict_table_flags_to_zip_size(
+ mach_read_from_4(FSP_SPACE_FLAGS
+ + space_header)),
+ ret_page, mtr, init_mtr));
}
/**********************************************************************//**
Allocates a single free page from a segment. This function implements
the intelligent allocation strategy which tries to minimize file space
fragmentation.
-@return allocated page offset, FIL_NULL if no page could be allocated */
+@retval NULL if no page could be allocated
+@retval block, rw_lock_x_lock_count(&block->lock) == 1 if allocation succeeded
+(init_mtr == mtr, or the page was not previously freed in mtr)
+@retval block (not allocated or initialized) otherwise */
UNIV_INTERN
-ulint
+buf_block_t*
fseg_alloc_free_page_general(
/*=========================*/
- fseg_header_t* seg_header,/*!< in: segment header */
- ulint hint, /*!< in: hint of which page would be desirable */
+ fseg_header_t* seg_header,/*!< in/out: segment header */
+ ulint hint, /*!< in: hint of which page would be
+ desirable */
byte direction,/*!< in: if the new page is needed because
of an index page split, and records are
inserted there in order, into which
@@ -2763,15 +2846,18 @@ fseg_alloc_free_page_general(
with fsp_reserve_free_extents, then there
is no need to do the check for this individual
page */
- mtr_t* mtr) /*!< in: mtr handle */
+ mtr_t* mtr, /*!< in/out: mini-transaction handle */
+ mtr_t* init_mtr)/*!< in/out: mtr or another mini-transaction
+ in which the page should be initialized.
+ If init_mtr!=mtr, but the page is already
+ latched in mtr, do not initialize the page. */
{
fseg_inode_t* inode;
ulint space;
ulint flags;
ulint zip_size;
rw_lock_t* latch;
- ibool success;
- ulint page_no;
+ buf_block_t* block;
ulint n_reserved;
space = page_get_space_id(page_align(seg_header));
@@ -2796,43 +2882,20 @@ fseg_alloc_free_page_general(
inode = fseg_inode_get(seg_header, space, zip_size, mtr);
- if (!has_done_reservation) {
- success = fsp_reserve_free_extents(&n_reserved, space, 2,
- FSP_NORMAL, mtr);
- if (!success) {
- return(FIL_NULL);
- }
+ if (!has_done_reservation
+ && !fsp_reserve_free_extents(&n_reserved, space, 2,
+ FSP_NORMAL, mtr)) {
+ return(NULL);
}
- page_no = fseg_alloc_free_page_low(space, zip_size,
- inode, hint, direction, mtr);
+ block = fseg_alloc_free_page_low(space, zip_size,
+ inode, hint, direction,
+ mtr, init_mtr);
if (!has_done_reservation) {
fil_space_release_free_extents(space, n_reserved);
}
- return(page_no);
-}
-
-/**********************************************************************//**
-Allocates a single free page from a segment. This function implements
-the intelligent allocation strategy which tries to minimize file space
-fragmentation.
-@return allocated page offset, FIL_NULL if no page could be allocated */
-UNIV_INTERN
-ulint
-fseg_alloc_free_page(
-/*=================*/
- fseg_header_t* seg_header,/*!< in: segment header */
- ulint hint, /*!< in: hint of which page would be desirable */
- byte direction,/*!< in: if the new page is needed because
- of an index page split, and records are
- inserted there in order, into which
- direction they go alphabetically: FSP_DOWN,
- FSP_UP, FSP_NO_DIR */
- mtr_t* mtr) /*!< in: mtr handle */
-{
- return(fseg_alloc_free_page_general(seg_header, hint, direction,
- FALSE, mtr));
+ return(block);
}
/**********************************************************************//**
@@ -3344,6 +3407,8 @@ crash:
descr + XDES_FLST_NODE, mtr);
fsp_free_extent(space, zip_size, page, mtr);
}
+
+ mtr->n_freed_pages++;
}
/**********************************************************************//**
diff --git a/storage/innobase/handler/ha_innodb.cc b/storage/innobase/handler/ha_innodb.cc
index 9fb7dd79f3d..c7b530f381a 100644
--- a/storage/innobase/handler/ha_innodb.cc
+++ b/storage/innobase/handler/ha_innodb.cc
@@ -26,8 +26,8 @@ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with
-this program; if not, write to the Free Software Foundation, Inc., 59 Temple
-Place, Suite 330, Boston, MA 02111-1307 USA
+this program; if not, write to the Free Software Foundation, Inc.,
+51 Franklin Street, Suite 500, Boston, MA 02110-1335 USA
*****************************************************************************/
@@ -3345,37 +3345,114 @@ normalize_table_name_low(
{
char* name_ptr;
char* db_ptr;
+ ulint db_len;
char* ptr;
/* Scan name from the end */
- ptr = strend(name)-1;
+ ptr = strend(name) - 1;
+ /* seek to the last path separator */
while (ptr >= name && *ptr != '\\' && *ptr != '/') {
ptr--;
}
name_ptr = ptr + 1;
- DBUG_ASSERT(ptr > name);
+ /* skip any number of path separators */
+ while (ptr >= name && (*ptr == '\\' || *ptr == '/')) {
+ ptr--;
+ }
- ptr--;
+ DBUG_ASSERT(ptr >= name);
+ /* seek to the last but one path separator or one char before
+ the beginning of name */
+ db_len = 0;
while (ptr >= name && *ptr != '\\' && *ptr != '/') {
ptr--;
+ db_len++;
}
db_ptr = ptr + 1;
- memcpy(norm_name, db_ptr, strlen(name) + 1 - (db_ptr - name));
+ memcpy(norm_name, db_ptr, db_len);
+
+ norm_name[db_len] = '/';
- norm_name[name_ptr - db_ptr - 1] = '/';
+ memcpy(norm_name + db_len + 1, name_ptr, strlen(name_ptr) + 1);
if (set_lower_case) {
innobase_casedn_str(norm_name);
}
}
+#if !defined(DBUG_OFF)
+/*********************************************************************
+Test normalize_table_name_low(). */
+static
+void
+test_normalize_table_name_low()
+/*===========================*/
+{
+ char norm_name[128];
+ const char* test_data[][2] = {
+ /* input, expected result */
+ {"./mysqltest/t1", "mysqltest/t1"},
+ {"./test/#sql-842b_2", "test/#sql-842b_2"},
+ {"./test/#sql-85a3_10", "test/#sql-85a3_10"},
+ {"./test/#sql2-842b-2", "test/#sql2-842b-2"},
+ {"./test/bug29807", "test/bug29807"},
+ {"./test/foo", "test/foo"},
+ {"./test/innodb_bug52663", "test/innodb_bug52663"},
+ {"./test/t", "test/t"},
+ {"./test/t1", "test/t1"},
+ {"./test/t10", "test/t10"},
+ {"/a/b/db/table", "db/table"},
+ {"/a/b/db///////table", "db/table"},
+ {"/a/b////db///////table", "db/table"},
+ {"/var/tmp/mysqld.1/#sql842b_2_10", "mysqld.1/#sql842b_2_10"},
+ {"db/table", "db/table"},
+ {"ddd/t", "ddd/t"},
+ {"d/ttt", "d/ttt"},
+ {"d/t", "d/t"},
+ {".\\mysqltest\\t1", "mysqltest/t1"},
+ {".\\test\\#sql-842b_2", "test/#sql-842b_2"},
+ {".\\test\\#sql-85a3_10", "test/#sql-85a3_10"},
+ {".\\test\\#sql2-842b-2", "test/#sql2-842b-2"},
+ {".\\test\\bug29807", "test/bug29807"},
+ {".\\test\\foo", "test/foo"},
+ {".\\test\\innodb_bug52663", "test/innodb_bug52663"},
+ {".\\test\\t", "test/t"},
+ {".\\test\\t1", "test/t1"},
+ {".\\test\\t10", "test/t10"},
+ {"C:\\a\\b\\db\\table", "db/table"},
+ {"C:\\a\\b\\db\\\\\\\\\\\\\\table", "db/table"},
+ {"C:\\a\\b\\\\\\\\db\\\\\\\\\\\\\\table", "db/table"},
+ {"C:\\var\\tmp\\mysqld.1\\#sql842b_2_10", "mysqld.1/#sql842b_2_10"},
+ {"db\\table", "db/table"},
+ {"ddd\\t", "ddd/t"},
+ {"d\\ttt", "d/ttt"},
+ {"d\\t", "d/t"},
+ };
+
+ for (size_t i = 0; i < UT_ARR_SIZE(test_data); i++) {
+ printf("test_normalize_table_name_low(): "
+ "testing \"%s\", expected \"%s\"... ",
+ test_data[i][0], test_data[i][1]);
+
+ normalize_table_name_low(norm_name, test_data[i][0], FALSE);
+
+ if (strcmp(norm_name, test_data[i][1]) == 0) {
+ printf("ok\n");
+ } else {
+ printf("got \"%s\"\n", norm_name);
+ ut_error;
+ }
+ }
+}
+#endif /* !DBUG_OFF */
+
/********************************************************************//**
Get the upper limit of the MySQL integral and floating-point type.
@return maximum allowed value for the field */
@@ -7371,6 +7448,11 @@ ha_innobase::delete_table(
DBUG_ENTER("ha_innobase::delete_table");
+ DBUG_EXECUTE_IF(
+ "test_normalize_table_name_low",
+ test_normalize_table_name_low();
+ );
+
/* Strangely, MySQL passes the table name without the '.frm'
extension, in contrast to ::create */
normalize_table_name(norm_name, name);
diff --git a/storage/innobase/ibuf/ibuf0ibuf.c b/storage/innobase/ibuf/ibuf0ibuf.c
index 96af9038be2..eea7bbb7eec 100644
--- a/storage/innobase/ibuf/ibuf0ibuf.c
+++ b/storage/innobase/ibuf/ibuf0ibuf.c
@@ -11,8 +11,8 @@ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with
-this program; if not, write to the Free Software Foundation, Inc., 59 Temple
-Place, Suite 330, Boston, MA 02111-1307 USA
+this program; if not, write to the Free Software Foundation, Inc.,
+51 Franklin Street, Suite 500, Boston, MA 02110-1335 USA
*****************************************************************************/
@@ -2180,14 +2180,14 @@ ibool
ibuf_add_free_page(void)
/*====================*/
{
- mtr_t mtr;
- page_t* header_page;
- ulint flags;
- ulint zip_size;
- ulint page_no;
- page_t* page;
- page_t* root;
- page_t* bitmap_page;
+ mtr_t mtr;
+ page_t* header_page;
+ ulint flags;
+ ulint zip_size;
+ buf_block_t* block;
+ page_t* page;
+ page_t* root;
+ page_t* bitmap_page;
mtr_start(&mtr);
@@ -2208,28 +2208,23 @@ ibuf_add_free_page(void)
of a deadlock. This is the reason why we created a special ibuf
header page apart from the ibuf tree. */
- page_no = fseg_alloc_free_page(
+ block = fseg_alloc_free_page(
header_page + IBUF_HEADER + IBUF_TREE_SEG_HEADER, 0, FSP_UP,
&mtr);
- if (UNIV_UNLIKELY(page_no == FIL_NULL)) {
+ if (block == NULL) {
mtr_commit(&mtr);
return(FALSE);
- } else {
- buf_block_t* block = buf_page_get(
- IBUF_SPACE_ID, 0, page_no, RW_X_LATCH, &mtr);
-
- ibuf_enter(&mtr);
-
- mutex_enter(&ibuf_mutex);
-
- root = ibuf_tree_root_get(&mtr);
+ }
- buf_block_dbg_add_level(block, SYNC_IBUF_TREE_NODE_NEW);
+ ut_ad(rw_lock_get_x_lock_count(&block->lock) == 1);
+ ibuf_enter(&mtr);
+ mutex_enter(&ibuf_mutex);
+ root = ibuf_tree_root_get(&mtr);
- page = buf_block_get_frame(block);
- }
+ buf_block_dbg_add_level(block, SYNC_IBUF_TREE_NODE_NEW);
+ page = buf_block_get_frame(block);
/* Add the page to the free list and update the ibuf size data */
@@ -2246,12 +2241,13 @@ ibuf_add_free_page(void)
(level 2 page) */
bitmap_page = ibuf_bitmap_get_map_page(
- IBUF_SPACE_ID, page_no, zip_size, &mtr);
+ IBUF_SPACE_ID, buf_block_get_page_no(block), zip_size, &mtr);
mutex_exit(&ibuf_mutex);
ibuf_bitmap_page_set_bits(
- bitmap_page, page_no, zip_size, IBUF_BITMAP_IBUF, TRUE, &mtr);
+ bitmap_page, buf_block_get_page_no(block), zip_size,
+ IBUF_BITMAP_IBUF, TRUE, &mtr);
ibuf_mtr_commit(&mtr);
@@ -3931,9 +3927,10 @@ ibuf_insert_to_index_page(
"InnoDB: but the number of fields does not match!\n",
stderr);
dump:
- buf_page_print(page, 0);
+ buf_page_print(page, 0, BUF_PAGE_PRINT_NO_CRASH);
dtuple_print(stderr, entry);
+ ut_ad(0);
fputs("InnoDB: The table where where"
" this index record belongs\n"
@@ -4506,12 +4503,14 @@ ibuf_merge_or_delete_for_page(
bitmap_page = ibuf_bitmap_get_map_page(space, page_no,
zip_size, &mtr);
- buf_page_print(bitmap_page, 0);
+ buf_page_print(bitmap_page, 0,
+ BUF_PAGE_PRINT_NO_CRASH);
ibuf_mtr_commit(&mtr);
fputs("\nInnoDB: Dump of the page:\n", stderr);
- buf_page_print(block->frame, 0);
+ buf_page_print(block->frame, 0,
+ BUF_PAGE_PRINT_NO_CRASH);
fprintf(stderr,
"InnoDB: Error: corruption in the tablespace."
@@ -4531,6 +4530,7 @@ ibuf_merge_or_delete_for_page(
(ulong) page_no,
(ulong)
fil_page_get_type(block->frame));
+ ut_ad(0);
}
}
diff --git a/storage/innobase/include/btr0btr.h b/storage/innobase/include/btr0btr.h
index 93b59fdfafb..6b997a2b083 100644
--- a/storage/innobase/include/btr0btr.h
+++ b/storage/innobase/include/btr0btr.h
@@ -11,8 +11,8 @@ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with
-this program; if not, write to the Free Software Foundation, Inc., 59 Temple
-Place, Suite 330, Boston, MA 02111-1307 USA
+this program; if not, write to the Free Software Foundation, Inc.,
+51 Franklin Street, Suite 500, Boston, MA 02110-1335 USA
*****************************************************************************/
@@ -519,11 +519,14 @@ UNIV_INTERN
ibool
btr_compress(
/*=========*/
- btr_cur_t* cursor, /*!< in: cursor on the page to merge or lift;
- the page must not be empty: in record delete
- use btr_discard_page if the page would become
- empty */
- mtr_t* mtr); /*!< in: mtr */
+ btr_cur_t* cursor, /*!< in/out: cursor on the page to merge
+ or lift; the page must not be empty:
+ when deleting records, use btr_discard_page()
+ if the page would become empty */
+ ibool adjust, /*!< in: TRUE if should adjust the
+ cursor position even if compression occurs */
+ mtr_t* mtr) /*!< in/out: mini-transaction */
+ __attribute__((nonnull));
/*************************************************************//**
Discards a page from a B-tree. This is used to remove the last record from
a B-tree page: the whole page must be removed at the same time. This cannot
@@ -574,7 +577,10 @@ btr_get_size(
/**************************************************************//**
Allocates a new file page to be used in an index tree. NOTE: we assume
that the caller has made the reservation for free extents!
-@return new allocated block, x-latched; NULL if out of space */
+@retval NULL if no page could be allocated
+@retval block, rw_lock_x_lock_count(&block->lock) == 1 if allocation succeeded
+(init_mtr == mtr, or the page was not previously freed in mtr)
+@retval block (not allocated or initialized) otherwise */
UNIV_INTERN
buf_block_t*
btr_page_alloc(
@@ -585,7 +591,12 @@ btr_page_alloc(
page split is made */
ulint level, /*!< in: level where the page is placed
in the tree */
- mtr_t* mtr); /*!< in: mtr */
+ mtr_t* mtr, /*!< in/out: mini-transaction
+ for the allocation */
+ mtr_t* init_mtr) /*!< in/out: mini-transaction
+ for x-latching and initializing
+ the page */
+ __attribute__((nonnull, warn_unused_result));
/**************************************************************//**
Frees a file page used in an index tree. NOTE: cannot free field external
storage pages because the page must contain info on its level. */
diff --git a/storage/innobase/include/btr0btr.ic b/storage/innobase/include/btr0btr.ic
index 1bd22a0ebc6..55bdb289b21 100644
--- a/storage/innobase/include/btr0btr.ic
+++ b/storage/innobase/include/btr0btr.ic
@@ -277,7 +277,7 @@ btr_node_ptr_get_child_page_no(
"InnoDB: a nonsensical page number 0"
" in a node ptr record at offset %lu\n",
(ulong) page_offset(rec));
- buf_page_print(page_align(rec), 0);
+ buf_page_print(page_align(rec), 0, 0);
}
return(page_no);
diff --git a/storage/innobase/include/btr0cur.h b/storage/innobase/include/btr0cur.h
index 4f33aacc48e..cbc6103c2ee 100644
--- a/storage/innobase/include/btr0cur.h
+++ b/storage/innobase/include/btr0cur.h
@@ -1,6 +1,6 @@
/*****************************************************************************
-Copyright (c) 1994, 2010, Innobase Oy. All Rights Reserved.
+Copyright (c) 1994, 2012, Oracle and/or its affiliates. All Rights Reserved.
This program is free software; you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free Software
@@ -11,8 +11,8 @@ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with
-this program; if not, write to the Free Software Foundation, Inc., 59 Temple
-Place, Suite 330, Boston, MA 02111-1307 USA
+this program; if not, write to the Free Software Foundation, Inc.,
+51 Franklin Street, Suite 500, Boston, MA 02110-1335 USA
*****************************************************************************/
@@ -36,6 +36,9 @@ Created 10/16/1994 Heikki Tuuri
#define BTR_NO_LOCKING_FLAG 2 /* do no record lock checking */
#define BTR_KEEP_SYS_FLAG 4 /* sys fields will be found from the
update vector or inserted entry */
+#define BTR_KEEP_POS_FLAG 8 /* btr_cur_pessimistic_update()
+ must keep cursor position when
+ moving columns to big_rec */
#ifndef UNIV_HOTBACKUP
#include "que0types.h"
@@ -310,7 +313,9 @@ btr_cur_pessimistic_update(
/*=======================*/
ulint flags, /*!< in: undo logging, locking, and rollback
flags */
- btr_cur_t* cursor, /*!< in: cursor on the record to update */
+ btr_cur_t* cursor, /*!< in/out: cursor on the record to update;
+ cursor may become invalid if *big_rec == NULL
+ || !(flags & BTR_KEEP_POS_FLAG) */
mem_heap_t** heap, /*!< in/out: pointer to memory heap, or NULL */
big_rec_t** big_rec,/*!< out: big rec vector whose fields have to
be stored externally by the caller, or NULL */
@@ -364,10 +369,13 @@ UNIV_INTERN
ibool
btr_cur_compress_if_useful(
/*=======================*/
- btr_cur_t* cursor, /*!< in: cursor on the page to compress;
+ btr_cur_t* cursor, /*!< in/out: cursor on the page to compress;
cursor does not stay valid if compression
occurs */
- mtr_t* mtr); /*!< in: mtr */
+ ibool adjust, /*!< in: TRUE if should adjust the
+ cursor position even if compression occurs */
+ mtr_t* mtr) /*!< in/out: mini-transaction */
+ __attribute__((nonnull));
/*******************************************************//**
Removes the record on which the tree cursor is positioned. It is assumed
that the mtr has an x-latch on the page where the cursor is positioned,
@@ -492,6 +500,27 @@ btr_cur_disown_inherited_fields(
const upd_t* update, /*!< in: update vector */
mtr_t* mtr) /*!< in/out: mini-transaction */
__attribute__((nonnull(2,3,4,5,6)));
+
+/** Operation code for btr_store_big_rec_extern_fields(). */
+enum blob_op {
+ /** Store off-page columns for a freshly inserted record */
+ BTR_STORE_INSERT = 0,
+ /** Store off-page columns for an insert by update */
+ BTR_STORE_INSERT_UPDATE,
+ /** Store off-page columns for an update */
+ BTR_STORE_UPDATE
+};
+
+/*******************************************************************//**
+Determine if an operation on off-page columns is an update.
+@return TRUE if op != BTR_STORE_INSERT */
+UNIV_INLINE
+ibool
+btr_blob_op_is_update(
+/*==================*/
+ enum blob_op op) /*!< in: operation */
+ __attribute__((warn_unused_result));
+
/*******************************************************************//**
Stores the fields in big_rec_vec to the tablespace and puts pointers to
them in rec. The extern flags in rec will have to be set beforehand.
@@ -499,52 +528,23 @@ The fields are stored on pages allocated from leaf node
file segment of the index tree.
@return DB_SUCCESS or DB_OUT_OF_FILE_SPACE */
UNIV_INTERN
-ulint
-btr_store_big_rec_extern_fields_func(
-/*=================================*/
+enum db_err
+btr_store_big_rec_extern_fields(
+/*============================*/
dict_index_t* index, /*!< in: index of rec; the index tree
MUST be X-latched */
buf_block_t* rec_block, /*!< in/out: block containing rec */
- rec_t* rec, /*!< in: record */
+ rec_t* rec, /*!< in/out: record */
const ulint* offsets, /*!< in: rec_get_offsets(rec, index);
the "external storage" flags in offsets
will not correspond to rec when
this function returns */
-#ifdef UNIV_DEBUG
- mtr_t* local_mtr, /*!< in: mtr containing the
- latch to rec and to the tree */
-#endif /* UNIV_DEBUG */
-#if defined UNIV_DEBUG || defined UNIV_BLOB_LIGHT_DEBUG
- ibool update_in_place,/*! in: TRUE if the record is updated
- in place (not delete+insert) */
-#endif /* UNIV_DEBUG || UNIV_BLOB_LIGHT_DEBUG */
- const big_rec_t*big_rec_vec) /*!< in: vector containing fields
+ const big_rec_t*big_rec_vec, /*!< in: vector containing fields
to be stored externally */
- __attribute__((nonnull));
-
-/** Stores the fields in big_rec_vec to the tablespace and puts pointers to
-them in rec. The extern flags in rec will have to be set beforehand.
-The fields are stored on pages allocated from leaf node
-file segment of the index tree.
-@param index in: clustered index; MUST be X-latched by mtr
-@param b in/out: block containing rec; MUST be X-latched by mtr
-@param rec in/out: clustered index record
-@param offsets in: rec_get_offsets(rec, index);
- the "external storage" flags in offsets will not be adjusted
-@param mtr in: mini-transaction that holds x-latch on index and b
-@param upd in: TRUE if the record is updated in place (not delete+insert)
-@param big in: vector containing fields to be stored externally
-@return DB_SUCCESS or DB_OUT_OF_FILE_SPACE */
-#ifdef UNIV_DEBUG
-# define btr_store_big_rec_extern_fields(index,b,rec,offsets,mtr,upd,big) \
- btr_store_big_rec_extern_fields_func(index,b,rec,offsets,mtr,upd,big)
-#elif defined UNIV_BLOB_LIGHT_DEBUG
-# define btr_store_big_rec_extern_fields(index,b,rec,offsets,mtr,upd,big) \
- btr_store_big_rec_extern_fields_func(index,b,rec,offsets,upd,big)
-#else
-# define btr_store_big_rec_extern_fields(index,b,rec,offsets,mtr,upd,big) \
- btr_store_big_rec_extern_fields_func(index,b,rec,offsets,big)
-#endif
+ mtr_t* btr_mtr, /*!< in: mtr containing the
+ latches to the clustered index */
+ enum blob_op op) /*! in: operation code */
+ __attribute__((nonnull, warn_unused_result));
/*******************************************************************//**
Frees the space in an externally stored field to the file space
diff --git a/storage/innobase/include/btr0cur.ic b/storage/innobase/include/btr0cur.ic
index 280583f6ccf..e31f77c77eb 100644
--- a/storage/innobase/include/btr0cur.ic
+++ b/storage/innobase/include/btr0cur.ic
@@ -1,6 +1,6 @@
/*****************************************************************************
-Copyright (c) 1994, 2009, Innobase Oy. All Rights Reserved.
+Copyright (c) 1994, 2012, Oracle and/or its affiliates. All Rights Reserved.
This program is free software; you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free Software
@@ -11,8 +11,8 @@ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with
-this program; if not, write to the Free Software Foundation, Inc., 59 Temple
-Place, Suite 330, Boston, MA 02111-1307 USA
+this program; if not, write to the Free Software Foundation, Inc.,
+51 Franklin Street, Suite 500, Boston, MA 02110-1335 USA
*****************************************************************************/
@@ -139,7 +139,7 @@ btr_cur_compress_recommendation(
btr_cur_t* cursor, /*!< in: btr cursor */
mtr_t* mtr) /*!< in: mtr */
{
- page_t* page;
+ const page_t* page;
ut_ad(mtr_memo_contains(mtr, btr_cur_get_block(cursor),
MTR_MEMO_PAGE_X_FIX));
@@ -197,4 +197,25 @@ btr_cur_can_delete_without_compress(
return(TRUE);
}
+
+/*******************************************************************//**
+Determine if an operation on off-page columns is an update.
+@return TRUE if op != BTR_STORE_INSERT */
+UNIV_INLINE
+ibool
+btr_blob_op_is_update(
+/*==================*/
+ enum blob_op op) /*!< in: operation */
+{
+ switch (op) {
+ case BTR_STORE_INSERT:
+ return(FALSE);
+ case BTR_STORE_INSERT_UPDATE:
+ case BTR_STORE_UPDATE:
+ return(TRUE);
+ }
+
+ ut_ad(0);
+ return(FALSE);
+}
#endif /* !UNIV_HOTBACKUP */
diff --git a/storage/innobase/include/buf0buf.h b/storage/innobase/include/buf0buf.h
index 456f077a13d..5054429156a 100644
--- a/storage/innobase/include/buf0buf.h
+++ b/storage/innobase/include/buf0buf.h
@@ -1,6 +1,6 @@
/*****************************************************************************
-Copyright (c) 1995, 2010, Innobase Oy. All Rights Reserved.
+Copyright (c) 1995, 2012, Oracle and/or its affiliates. All Rights Reserved.
This program is free software; you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free Software
@@ -11,8 +11,8 @@ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with
-this program; if not, write to the Free Software Foundation, Inc., 59 Temple
-Place, Suite 330, Boston, MA 02111-1307 USA
+this program; if not, write to the Free Software Foundation, Inc.,
+51 Franklin Street, Suite 500, Boston, MA 02110-1335 USA
*****************************************************************************/
@@ -596,6 +596,31 @@ buf_block_get_modify_clock(
#else /* !UNIV_HOTBACKUP */
# define buf_block_modify_clock_inc(block) ((void) 0)
#endif /* !UNIV_HOTBACKUP */
+/*******************************************************************//**
+Increments the bufferfix count. */
+UNIV_INLINE
+void
+buf_block_buf_fix_inc_func(
+/*=======================*/
+#ifdef UNIV_SYNC_DEBUG
+ const char* file, /*!< in: file name */
+ ulint line, /*!< in: line */
+#endif /* UNIV_SYNC_DEBUG */
+ buf_block_t* block) /*!< in/out: block to bufferfix */
+ __attribute__((nonnull));
+#ifdef UNIV_SYNC_DEBUG
+/** Increments the bufferfix count.
+@param b in/out: block to bufferfix
+@param f in: file name where requested
+@param l in: line number where requested */
+# define buf_block_buf_fix_inc(b,f,l) buf_block_buf_fix_inc_func(f,l,b)
+#else /* UNIV_SYNC_DEBUG */
+/** Increments the bufferfix count.
+@param b in/out: block to bufferfix
+@param f in: file name where requested
+@param l in: line number where requested */
+# define buf_block_buf_fix_inc(b,f,l) buf_block_buf_fix_inc_func(b)
+#endif /* UNIV_SYNC_DEBUG */
/********************************************************************//**
Calculates a page checksum which is stored to the page when it is written
to a file. Note that we must be careful to calculate the same value
@@ -680,6 +705,13 @@ buf_print(void);
/*============*/
#endif /* UNIV_DEBUG_PRINT || UNIV_DEBUG || UNIV_BUF_DEBUG */
#endif /* !UNIV_HOTBACKUP */
+enum buf_page_print_flags {
+ /** Do not crash at the end of buf_page_print(). */
+ BUF_PAGE_PRINT_NO_CRASH = 1,
+ /** Do not print the full page dump. */
+ BUF_PAGE_PRINT_NO_FULL = 2
+};
+
/********************************************************************//**
Prints a page to stderr. */
UNIV_INTERN
@@ -687,8 +719,12 @@ void
buf_page_print(
/*===========*/
const byte* read_buf, /*!< in: a database page */
- ulint zip_size); /*!< in: compressed page size, or
+ ulint zip_size, /*!< in: compressed page size, or
0 for uncompressed pages */
+ ulint flags) /*!< in: 0 or
+ BUF_PAGE_PRINT_NO_CRASH or
+ BUF_PAGE_PRINT_NO_FULL */
+ __attribute__((nonnull));
/********************************************************************//**
Decompress a block.
@return TRUE if successful */
diff --git a/storage/innobase/include/buf0buf.ic b/storage/innobase/include/buf0buf.ic
index 99e55df3312..917ee5dda84 100644
--- a/storage/innobase/include/buf0buf.ic
+++ b/storage/innobase/include/buf0buf.ic
@@ -1,6 +1,6 @@
/*****************************************************************************
-Copyright (c) 1995, 2010, Innobase Oy. All Rights Reserved.
+Copyright (c) 1995, 2012, Oracle and/or its affiliates. All Rights Reserved.
Copyright (c) 2008, Google Inc.
Portions of this file contain modifications contributed and copyrighted by
@@ -18,8 +18,8 @@ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with
-this program; if not, write to the Free Software Foundation, Inc., 59 Temple
-Place, Suite 330, Boston, MA 02111-1307 USA
+this program; if not, write to the Free Software Foundation, Inc.,
+51 Franklin Street, Suite 500, Boston, MA 02110-1335 USA
*****************************************************************************/
@@ -988,19 +988,6 @@ buf_block_buf_fix_inc_func(
block->page.buf_fix_count++;
}
-#ifdef UNIV_SYNC_DEBUG
-/** Increments the bufferfix count.
-@param b in/out: block to bufferfix
-@param f in: file name where requested
-@param l in: line number where requested */
-# define buf_block_buf_fix_inc(b,f,l) buf_block_buf_fix_inc_func(f,l,b)
-#else /* UNIV_SYNC_DEBUG */
-/** Increments the bufferfix count.
-@param b in/out: block to bufferfix
-@param f in: file name where requested
-@param l in: line number where requested */
-# define buf_block_buf_fix_inc(b,f,l) buf_block_buf_fix_inc_func(b)
-#endif /* UNIV_SYNC_DEBUG */
/*******************************************************************//**
Decrements the bufferfix count. */
@@ -1251,7 +1238,7 @@ buf_block_dbg_add_level(
where we have acquired latch */
ulint level) /*!< in: latching order level */
{
- sync_thread_add_level(&block->lock, level);
+ sync_thread_add_level(&block->lock, level, FALSE);
}
#endif /* UNIV_SYNC_DEBUG */
/********************************************************************//**
diff --git a/storage/innobase/include/fsp0fsp.h b/storage/innobase/include/fsp0fsp.h
index 4efabacb2cb..f07e3decc66 100644
--- a/storage/innobase/include/fsp0fsp.h
+++ b/storage/innobase/include/fsp0fsp.h
@@ -1,6 +1,6 @@
/*****************************************************************************
-Copyright (c) 1995, 2011, Oracle and/or its affiliates. All Rights Reserved.
+Copyright (c) 1995, 2012, Oracle and/or its affiliates. All Rights Reserved.
This program is free software; you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free Software
@@ -260,30 +260,33 @@ fseg_n_reserved_pages(
Allocates a single free page from a segment. This function implements
the intelligent allocation strategy which tries to minimize
file space fragmentation.
-@return the allocated page offset FIL_NULL if no page could be allocated */
-UNIV_INTERN
-ulint
-fseg_alloc_free_page(
-/*=================*/
- fseg_header_t* seg_header, /*!< in: segment header */
- ulint hint, /*!< in: hint of which page would be desirable */
- byte direction, /*!< in: if the new page is needed because
+@param[in/out] seg_header segment header
+@param[in] hint hint of which page would be desirable
+@param[in] direction if the new page is needed because
of an index page split, and records are
inserted there in order, into which
direction they go alphabetically: FSP_DOWN,
- FSP_UP, FSP_NO_DIR */
- mtr_t* mtr); /*!< in: mtr handle */
+ FSP_UP, FSP_NO_DIR
+@param[in/out] mtr mini-transaction
+@return X-latched block, or NULL if no page could be allocated */
+#define fseg_alloc_free_page(seg_header, hint, direction, mtr) \
+ fseg_alloc_free_page_general(seg_header, hint, direction, \
+ FALSE, mtr, mtr)
/**********************************************************************//**
Allocates a single free page from a segment. This function implements
the intelligent allocation strategy which tries to minimize file space
fragmentation.
-@return allocated page offset, FIL_NULL if no page could be allocated */
+@retval NULL if no page could be allocated
+@retval block, rw_lock_x_lock_count(&block->lock) == 1 if allocation succeeded
+(init_mtr == mtr, or the page was not previously freed in mtr)
+@retval block (not allocated or initialized) otherwise */
UNIV_INTERN
-ulint
+buf_block_t*
fseg_alloc_free_page_general(
/*=========================*/
- fseg_header_t* seg_header,/*!< in: segment header */
- ulint hint, /*!< in: hint of which page would be desirable */
+ fseg_header_t* seg_header,/*!< in/out: segment header */
+ ulint hint, /*!< in: hint of which page would be
+ desirable */
byte direction,/*!< in: if the new page is needed because
of an index page split, and records are
inserted there in order, into which
@@ -294,8 +297,12 @@ fseg_alloc_free_page_general(
with fsp_reserve_free_extents, then there
is no need to do the check for this individual
page */
- mtr_t* mtr) /*!< in/out: mini-transaction */
- __attribute__((warn_unused_result, nonnull(1,5)));
+ mtr_t* mtr, /*!< in/out: mini-transaction */
+ mtr_t* init_mtr)/*!< in/out: mtr or another mini-transaction
+ in which the page should be initialized.
+ If init_mtr!=mtr, but the page is already
+ latched in mtr, do not initialize the page. */
+ __attribute__((warn_unused_result, nonnull));
/**********************************************************************//**
Reserves free pages from a tablespace. All mini-transactions which may
use several pages from the tablespace should call this function beforehand
diff --git a/storage/innobase/include/log0log.h b/storage/innobase/include/log0log.h
index 1ae94a332e5..f2ab6a9898d 100644
--- a/storage/innobase/include/log0log.h
+++ b/storage/innobase/include/log0log.h
@@ -1,6 +1,6 @@
/*****************************************************************************
-Copyright (c) 1995, 2010, Innobase Oy. All Rights Reserved.
+Copyright (c) 1995, 2010, Oracle and/or its affiliates. All Rights Reserved.
Copyright (c) 2009, Google Inc.
Portions of this file contain modifications contributed and copyrighted by
@@ -18,8 +18,8 @@ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with
-this program; if not, write to the Free Software Foundation, Inc., 59 Temple
-Place, Suite 330, Boston, MA 02111-1307 USA
+this program; if not, write to the Free Software Foundation, Inc.,
+51 Franklin Street, Suite 500, Boston, MA 02110-1335 USA
*****************************************************************************/
diff --git a/storage/innobase/include/mtr0mtr.h b/storage/innobase/include/mtr0mtr.h
index 7f608546cc2..46f1ff9310c 100644
--- a/storage/innobase/include/mtr0mtr.h
+++ b/storage/innobase/include/mtr0mtr.h
@@ -1,6 +1,6 @@
/*****************************************************************************
-Copyright (c) 1995, 2009, Innobase Oy. All Rights Reserved.
+Copyright (c) 1995, 2012, Oracle and/or its affiliates. All Rights Reserved.
This program is free software; you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free Software
@@ -11,8 +11,8 @@ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with
-this program; if not, write to the Free Software Foundation, Inc., 59 Temple
-Place, Suite 330, Boston, MA 02111-1307 USA
+this program; if not, write to the Free Software Foundation, Inc.,
+51 Franklin Street, Suite 500, Boston, MA 02110-1335 USA
*****************************************************************************/
@@ -50,7 +50,9 @@ first 3 values must be RW_S_LATCH, RW_X_LATCH, RW_NO_LATCH */
#define MTR_MEMO_PAGE_S_FIX RW_S_LATCH
#define MTR_MEMO_PAGE_X_FIX RW_X_LATCH
#define MTR_MEMO_BUF_FIX RW_NO_LATCH
-#define MTR_MEMO_MODIFY 54
+#ifdef UNIV_DEBUG
+# define MTR_MEMO_MODIFY 54
+#endif /* UNIV_DEBUG */
#define MTR_MEMO_S_LOCK 55
#define MTR_MEMO_X_LOCK 56
@@ -376,6 +378,9 @@ struct mtr_struct{
ulint n_log_recs;
/* count of how many page initial log records
have been written to the mtr log */
+ ulint n_freed_pages;
+ /* number of pages that have been freed in
+ this mini-transaction */
ulint log_mode; /* specifies which operations should be
logged; default value MTR_LOG_ALL */
ib_uint64_t start_lsn;/* start lsn of the possible log entry for
diff --git a/storage/innobase/include/mtr0mtr.ic b/storage/innobase/include/mtr0mtr.ic
index 1db4a4bd735..755bd49057b 100644
--- a/storage/innobase/include/mtr0mtr.ic
+++ b/storage/innobase/include/mtr0mtr.ic
@@ -1,6 +1,6 @@
/*****************************************************************************
-Copyright (c) 1995, 2010, Innobase Oy. All Rights Reserved.
+Copyright (c) 1995, 2012, Oracle and/or its affiliates. All Rights Reserved.
This program is free software; you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free Software
@@ -11,8 +11,8 @@ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with
-this program; if not, write to the Free Software Foundation, Inc., 59 Temple
-Place, Suite 330, Boston, MA 02111-1307 USA
+this program; if not, write to the Free Software Foundation, Inc.,
+51 Franklin Street, Suite 500, Boston, MA 02110-1335 USA
*****************************************************************************/
@@ -46,6 +46,7 @@ mtr_start(
mtr->modifications = FALSE;
mtr->inside_ibuf = FALSE;
mtr->n_log_recs = 0;
+ mtr->n_freed_pages = 0;
ut_d(mtr->state = MTR_ACTIVE);
ut_d(mtr->magic_n = MTR_MAGIC_N);
diff --git a/storage/innobase/include/page0page.h b/storage/innobase/include/page0page.h
index 540ec7855bc..74e9ceca959 100644
--- a/storage/innobase/include/page0page.h
+++ b/storage/innobase/include/page0page.h
@@ -1,6 +1,6 @@
/*****************************************************************************
-Copyright (c) 1994, 2009, Innobase Oy. All Rights Reserved.
+Copyright (c) 1994, 2012, Oracle and/or its affiliates. All Rights Reserved.
This program is free software; you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free Software
@@ -11,8 +11,8 @@ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with
-this program; if not, write to the Free Software Foundation, Inc., 59 Temple
-Place, Suite 330, Boston, MA 02111-1307 USA
+this program; if not, write to the Free Software Foundation, Inc.,
+51 Franklin Street, Suite 500, Boston, MA 02110-1335 USA
*****************************************************************************/
@@ -281,16 +281,42 @@ page_get_supremum_offset(
const page_t* page); /*!< in: page which must have record(s) */
#define page_get_infimum_rec(page) ((page) + page_get_infimum_offset(page))
#define page_get_supremum_rec(page) ((page) + page_get_supremum_offset(page))
+
/************************************************************//**
-Returns the middle record of record list. If there are an even number
-of records in the list, returns the first record of upper half-list.
-@return middle record */
+Returns the nth record of the record list.
+This is the inverse function of page_rec_get_n_recs_before().
+@return nth record */
UNIV_INTERN
+const rec_t*
+page_rec_get_nth_const(
+/*===================*/
+ const page_t* page, /*!< in: page */
+ ulint nth) /*!< in: nth record */
+ __attribute__((nonnull, warn_unused_result));
+/************************************************************//**
+Returns the nth record of the record list.
+This is the inverse function of page_rec_get_n_recs_before().
+@return nth record */
+UNIV_INLINE
+rec_t*
+page_rec_get_nth(
+/*=============*/
+ page_t* page, /*< in: page */
+ ulint nth) /*!< in: nth record */
+ __attribute__((nonnull, warn_unused_result));
+
+#ifndef UNIV_HOTBACKUP
+/************************************************************//**
+Returns the middle record of the records on the page. If there is an
+even number of records in the list, returns the first record of the
+upper half-list.
+@return middle record */
+UNIV_INLINE
rec_t*
page_get_middle_rec(
/*================*/
- page_t* page); /*!< in: page */
-#ifndef UNIV_HOTBACKUP
+ page_t* page) /*!< in: page */
+ __attribute__((nonnull, warn_unused_result));
/*************************************************************//**
Compares a data tuple to a physical record. Differs from the function
cmp_dtuple_rec_with_match in the way that the record must reside on an
@@ -345,6 +371,7 @@ page_get_n_recs(
/***************************************************************//**
Returns the number of records before the given record in chain.
The number includes infimum and supremum records.
+This is the inverse function of page_rec_get_nth().
@return number of records */
UNIV_INTERN
ulint
diff --git a/storage/innobase/include/page0page.ic b/storage/innobase/include/page0page.ic
index 115cee64f8b..781ad029e87 100644
--- a/storage/innobase/include/page0page.ic
+++ b/storage/innobase/include/page0page.ic
@@ -1,6 +1,6 @@
/*****************************************************************************
-Copyright (c) 1994, 2009, Innobase Oy. All Rights Reserved.
+Copyright (c) 1994, 2012, Oracle and/or its affiliates. All Rights Reserved.
This program is free software; you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free Software
@@ -11,8 +11,8 @@ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with
-this program; if not, write to the Free Software Foundation, Inc., 59 Temple
-Place, Suite 330, Boston, MA 02111-1307 USA
+this program; if not, write to the Free Software Foundation, Inc.,
+51 Franklin Street, Suite 500, Boston, MA 02110-1335 USA
*****************************************************************************/
@@ -419,7 +419,37 @@ page_rec_is_infimum(
return(page_rec_is_infimum_low(page_offset(rec)));
}
+/************************************************************//**
+Returns the nth record of the record list.
+This is the inverse function of page_rec_get_n_recs_before().
+@return nth record */
+UNIV_INLINE
+rec_t*
+page_rec_get_nth(
+/*=============*/
+ page_t* page, /*!< in: page */
+ ulint nth) /*!< in: nth record */
+{
+ return((rec_t*) page_rec_get_nth_const(page, nth));
+}
+
#ifndef UNIV_HOTBACKUP
+/************************************************************//**
+Returns the middle record of the records on the page. If there is an
+even number of records in the list, returns the first record of the
+upper half-list.
+@return middle record */
+UNIV_INLINE
+rec_t*
+page_get_middle_rec(
+/*================*/
+ page_t* page) /*!< in: page */
+{
+ ulint middle = (page_get_n_recs(page) + PAGE_HEAP_NO_USER_LOW) / 2;
+
+ return(page_rec_get_nth(page, middle));
+}
+
/*************************************************************//**
Compares a data tuple to a physical record. Differs from the function
cmp_dtuple_rec_with_match in the way that the record must reside on an
@@ -710,7 +740,7 @@ page_rec_get_next_low(
(void*) rec,
(ulong) page_get_space_id(page),
(ulong) page_get_page_no(page));
- buf_page_print(page, 0);
+ buf_page_print(page, 0, 0);
ut_error;
}
diff --git a/storage/innobase/include/sync0sync.h b/storage/innobase/include/sync0sync.h
index b5bf30e758c..9b07c4758c9 100644
--- a/storage/innobase/include/sync0sync.h
+++ b/storage/innobase/include/sync0sync.h
@@ -1,6 +1,6 @@
/*****************************************************************************
-Copyright (c) 1995, 2011, Innobase Oy. All Rights Reserved.
+Copyright (c) 1995, 2012, Oracle and/or its affiliates. All Rights Reserved.
Copyright (c) 2008, Google Inc.
Portions of this file contain modifications contributed and copyrighted by
@@ -18,8 +18,8 @@ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with
-this program; if not, write to the Free Software Foundation, Inc., 59 Temple
-Place, Suite 330, Boston, MA 02111-1307 USA
+this program; if not, write to the Free Software Foundation, Inc.,
+51 Franklin Street, Suite 500, Boston, MA 02110-1335 USA
*****************************************************************************/
@@ -400,8 +400,10 @@ void
sync_thread_add_level(
/*==================*/
void* latch, /*!< in: pointer to a mutex or an rw-lock */
- ulint level); /*!< in: level in the latching order; if
+ ulint level, /*!< in: level in the latching order; if
SYNC_LEVEL_VARYING, nothing is done */
+ ibool relock) /*!< in: TRUE if re-entering an x-lock */
+ __attribute__((nonnull));
/******************************************************************//**
Removes a latch from the thread level array if it is found there.
@return TRUE if found in the array; it is no error if the latch is
diff --git a/storage/innobase/include/trx0rec.ic b/storage/innobase/include/trx0rec.ic
index f0b3276ed44..4fc5a7147f9 100644
--- a/storage/innobase/include/trx0rec.ic
+++ b/storage/innobase/include/trx0rec.ic
@@ -1,6 +1,6 @@
/*****************************************************************************
-Copyright (c) 1996, 2009, Innobase Oy. All Rights Reserved.
+Copyright (c) 1996, 2012, Oracle and/or its affiliates. All Rights Reserved.
This program is free software; you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free Software
@@ -11,8 +11,8 @@ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with
-this program; if not, write to the Free Software Foundation, Inc., 59 Temple
-Place, Suite 330, Boston, MA 02111-1307 USA
+this program; if not, write to the Free Software Foundation, Inc.,
+51 Franklin Street, Suite 500, Boston, MA 02110-1335 USA
*****************************************************************************/
@@ -107,6 +107,7 @@ trx_undo_rec_copy(
len = mach_read_from_2(undo_rec)
- ut_align_offset(undo_rec, UNIV_PAGE_SIZE);
+ ut_ad(len < UNIV_PAGE_SIZE);
return(mem_heap_dup(heap, undo_rec, len));
}
#endif /* !UNIV_HOTBACKUP */
diff --git a/storage/innobase/include/trx0undo.h b/storage/innobase/include/trx0undo.h
index 50aa6d0ac09..4a1e40af505 100644
--- a/storage/innobase/include/trx0undo.h
+++ b/storage/innobase/include/trx0undo.h
@@ -1,6 +1,6 @@
/*****************************************************************************
-Copyright (c) 1996, 2011, Oracle and/or its affiliates. All Rights Reserved.
+Copyright (c) 1996, 2012, Oracle and/or its affiliates. All Rights Reserved.
This program is free software; you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free Software
@@ -11,8 +11,8 @@ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with
-this program; if not, write to the Free Software Foundation, Inc., 59 Temple
-Place, Suite 330, Boston, MA 02111-1307 USA
+this program; if not, write to the Free Software Foundation, Inc.,
+51 Franklin Street, Suite 500, Boston, MA 02110-1335 USA
*****************************************************************************/
@@ -194,16 +194,17 @@ trx_undo_get_first_rec(
mtr_t* mtr); /*!< in: mtr */
/********************************************************************//**
Tries to add a page to the undo log segment where the undo log is placed.
-@return page number if success, else FIL_NULL */
+@return X-latched block if success, else NULL */
UNIV_INTERN
-ulint
+buf_block_t*
trx_undo_add_page(
/*==============*/
trx_t* trx, /*!< in: transaction */
trx_undo_t* undo, /*!< in: undo log memory object */
- mtr_t* mtr); /*!< in: mtr which does not have a latch to any
+ mtr_t* mtr) /*!< in: mtr which does not have a latch to any
undo log page; the caller must have reserved
the rollback segment mutex */
+ __attribute__((nonnull, warn_unused_result));
/********************************************************************//**
Frees the last undo log page.
The caller must hold the rollback segment mutex. */
diff --git a/storage/innobase/lock/lock0lock.c b/storage/innobase/lock/lock0lock.c
index 263516ab5fd..a516b5545de 100644
--- a/storage/innobase/lock/lock0lock.c
+++ b/storage/innobase/lock/lock0lock.c
@@ -1615,7 +1615,7 @@ lock_sec_rec_some_has_impl_off_kernel(
if (!lock_check_trx_id_sanity(page_get_max_trx_id(page),
rec, index, offsets, TRUE)) {
- buf_page_print(page, 0);
+ buf_page_print(page, 0, 0);
/* The page is corrupt: try to avoid a crash by returning
NULL */
diff --git a/storage/innobase/log/log0log.c b/storage/innobase/log/log0log.c
index fd2258945b6..c40292bcf1e 100644
--- a/storage/innobase/log/log0log.c
+++ b/storage/innobase/log/log0log.c
@@ -1,6 +1,6 @@
/*****************************************************************************
-Copyright (c) 1995, 2010, Innobase Oy. All Rights Reserved.
+Copyright (c) 1995, 2010, Oracle and/or its affiliates. All Rights Reserved.
Copyright (c) 2009, Google Inc.
Portions of this file contain modifications contributed and copyrighted by
@@ -18,8 +18,8 @@ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with
-this program; if not, write to the Free Software Foundation, Inc., 59 Temple
-Place, Suite 330, Boston, MA 02111-1307 USA
+this program; if not, write to the Free Software Foundation, Inc.,
+51 Franklin Street, Suite 500, Boston, MA 02110-1335 USA
*****************************************************************************/
diff --git a/storage/innobase/mtr/mtr0mtr.c b/storage/innobase/mtr/mtr0mtr.c
index 08234609ff0..d852ed6f496 100644
--- a/storage/innobase/mtr/mtr0mtr.c
+++ b/storage/innobase/mtr/mtr0mtr.c
@@ -1,6 +1,6 @@
/*****************************************************************************
-Copyright (c) 1995, 2009, Innobase Oy. All Rights Reserved.
+Copyright (c) 1995, 2012, Oracle and/or its affiliates. All Rights Reserved.
This program is free software; you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free Software
@@ -11,8 +11,8 @@ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with
-this program; if not, write to the Free Software Foundation, Inc., 59 Temple
-Place, Suite 330, Boston, MA 02111-1307 USA
+this program; if not, write to the Free Software Foundation, Inc.,
+51 Franklin Street, Suite 500, Boston, MA 02110-1335 USA
*****************************************************************************/
diff --git a/storage/innobase/page/page0cur.c b/storage/innobase/page/page0cur.c
index 936762b986a..d49b121afab 100644
--- a/storage/innobase/page/page0cur.c
+++ b/storage/innobase/page/page0cur.c
@@ -1,6 +1,6 @@
/*****************************************************************************
-Copyright (c) 1994, 2009, Innobase Oy. All Rights Reserved.
+Copyright (c) 1994, 2012, Oracle and/or its affiliates. All Rights Reserved.
This program is free software; you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free Software
@@ -11,8 +11,8 @@ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with
-this program; if not, write to the Free Software Foundation, Inc., 59 Temple
-Place, Suite 330, Boston, MA 02111-1307 USA
+this program; if not, write to the Free Software Foundation, Inc.,
+51 Franklin Street, Suite 500, Boston, MA 02110-1335 USA
*****************************************************************************/
@@ -900,7 +900,7 @@ page_cur_parse_insert_rec(
ut_print_buf(stderr, ptr2, 300);
putc('\n', stderr);
- buf_page_print(page, 0);
+ buf_page_print(page, 0, 0);
ut_error;
}
@@ -1180,14 +1180,15 @@ page_cur_insert_rec_zip_reorg(
/* Before trying to reorganize the page,
store the number of preceding records on the page. */
pos = page_rec_get_n_recs_before(rec);
+ ut_ad(pos > 0);
if (page_zip_reorganize(block, index, mtr)) {
/* The page was reorganized: Find rec by seeking to pos,
and update *current_rec. */
- rec = page + PAGE_NEW_INFIMUM;
-
- while (--pos) {
- rec = page + rec_get_next_offs(rec, TRUE);
+ if (pos > 1) {
+ rec = page_rec_get_nth(page, pos - 1);
+ } else {
+ rec = page + PAGE_NEW_INFIMUM;
}
*current_rec = rec;
@@ -1283,6 +1284,12 @@ page_cur_insert_rec_zip(
insert_rec = page_cur_insert_rec_zip_reorg(
current_rec, block, index, insert_rec,
page, page_zip, mtr);
+#ifdef UNIV_DEBUG
+ if (insert_rec) {
+ rec_offs_make_valid(
+ insert_rec, index, offsets);
+ }
+#endif /* UNIV_DEBUG */
}
return(insert_rec);
diff --git a/storage/innobase/page/page0page.c b/storage/innobase/page/page0page.c
index e250bb359fe..5f0380cb55f 100644
--- a/storage/innobase/page/page0page.c
+++ b/storage/innobase/page/page0page.c
@@ -11,8 +11,8 @@ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with
-this program; if not, write to the Free Software Foundation, Inc., 59 Temple
-Place, Suite 330, Boston, MA 02111-1307 USA
+this program; if not, write to the Free Software Foundation, Inc.,
+51 Franklin Street, Suite 500, Boston, MA 02110-1335 USA
*****************************************************************************/
@@ -148,7 +148,7 @@ page_dir_find_owner_slot(
fputs("\n"
"InnoDB: on that page!\n", stderr);
- buf_page_print(page, 0);
+ buf_page_print(page, 0, 0);
ut_error;
}
@@ -569,8 +569,10 @@ page_copy_rec_list_end_no_locks(
/* Track an assertion failure reported on the mailing
list on June 18th, 2003 */
- buf_page_print(new_page, 0);
- buf_page_print(page_align(rec), 0);
+ buf_page_print(new_page, 0,
+ BUF_PAGE_PRINT_NO_CRASH);
+ buf_page_print(page_align(rec), 0,
+ BUF_PAGE_PRINT_NO_CRASH);
ut_print_timestamp(stderr);
fprintf(stderr,
@@ -1452,55 +1454,54 @@ page_dir_balance_slot(
}
}
-#ifndef UNIV_HOTBACKUP
/************************************************************//**
-Returns the middle record of the record list. If there are an even number
-of records in the list, returns the first record of the upper half-list.
-@return middle record */
+Returns the nth record of the record list.
+This is the inverse function of page_rec_get_n_recs_before().
+@return nth record */
UNIV_INTERN
-rec_t*
-page_get_middle_rec(
-/*================*/
- page_t* page) /*!< in: page */
+const rec_t*
+page_rec_get_nth_const(
+/*===================*/
+ const page_t* page, /*!< in: page */
+ ulint nth) /*!< in: nth record */
{
- page_dir_slot_t* slot;
- ulint middle;
+ const page_dir_slot_t* slot;
ulint i;
ulint n_owned;
- ulint count;
- rec_t* rec;
-
- /* This many records we must leave behind */
- middle = (page_get_n_recs(page) + PAGE_HEAP_NO_USER_LOW) / 2;
+ const rec_t* rec;
- count = 0;
+ ut_ad(nth < UNIV_PAGE_SIZE / (REC_N_NEW_EXTRA_BYTES + 1));
for (i = 0;; i++) {
slot = page_dir_get_nth_slot(page, i);
n_owned = page_dir_slot_get_n_owned(slot);
- if (count + n_owned > middle) {
+ if (n_owned > nth) {
break;
} else {
- count += n_owned;
+ nth -= n_owned;
}
}
ut_ad(i > 0);
slot = page_dir_get_nth_slot(page, i - 1);
- rec = (rec_t*) page_dir_slot_get_rec(slot);
- rec = page_rec_get_next(rec);
-
- /* There are now count records behind rec */
+ rec = page_dir_slot_get_rec(slot);
- for (i = 0; i < middle - count; i++) {
- rec = page_rec_get_next(rec);
+ if (page_is_comp(page)) {
+ do {
+ rec = page_rec_get_next_low(rec, TRUE);
+ ut_ad(rec);
+ } while (nth--);
+ } else {
+ do {
+ rec = page_rec_get_next_low(rec, FALSE);
+ ut_ad(rec);
+ } while (nth--);
}
return(rec);
}
-#endif /* !UNIV_HOTBACKUP */
/***************************************************************//**
Returns the number of records before the given record in chain.
@@ -1562,6 +1563,7 @@ page_rec_get_n_recs_before(
n--;
ut_ad(n >= 0);
+ ut_ad(n < UNIV_PAGE_SIZE / (REC_N_NEW_EXTRA_BYTES + 1));
return((ulint) n);
}
@@ -1834,7 +1836,7 @@ page_check_dir(
fprintf(stderr,
"InnoDB: Page directory corruption:"
" infimum not pointed to\n");
- buf_page_print(page, 0);
+ buf_page_print(page, 0, 0);
}
if (UNIV_UNLIKELY(!page_rec_is_supremum_low(supremum_offs))) {
@@ -1842,7 +1844,7 @@ page_check_dir(
fprintf(stderr,
"InnoDB: Page directory corruption:"
" supremum not pointed to\n");
- buf_page_print(page, 0);
+ buf_page_print(page, 0, 0);
}
}
#endif /* !UNIV_HOTBACKUP */
@@ -2546,7 +2548,7 @@ func_exit2:
(ulong) page_get_space_id(page),
(ulong) page_get_page_no(page),
index->name);
- buf_page_print(page, 0);
+ buf_page_print(page, 0, 0);
}
return(ret);
diff --git a/storage/innobase/row/row0ins.c b/storage/innobase/row/row0ins.c
index 67846ab6f69..d02d0e986aa 100644
--- a/storage/innobase/row/row0ins.c
+++ b/storage/innobase/row/row0ins.c
@@ -11,8 +11,8 @@ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with
-this program; if not, write to the Free Software Foundation, Inc., 59 Temple
-Place, Suite 330, Boston, MA 02111-1307 USA
+this program; if not, write to the Free Software Foundation, Inc.,
+51 Franklin Street, Suite 500, Boston, MA 02110-1335 USA
*****************************************************************************/
@@ -23,6 +23,8 @@ Insert into a table
Created 4/20/1996 Heikki Tuuri
*******************************************************/
+#include "m_string.h" /* for my_sys.h */
+#include "my_sys.h" /* DEBUG_SYNC_C */
#include "row0ins.h"
#ifdef UNIV_NONINL
@@ -348,9 +350,9 @@ row_ins_clust_index_entry_by_modify(
return(DB_LOCK_TABLE_FULL);
}
- err = btr_cur_pessimistic_update(0, cursor,
- heap, big_rec, update,
- 0, thr, mtr);
+ err = btr_cur_pessimistic_update(
+ BTR_KEEP_POS_FLAG, cursor, heap, big_rec, update,
+ 0, thr, mtr);
}
return(err);
@@ -1979,6 +1981,7 @@ row_ins_index_entry_low(
ulint modify = 0; /* remove warning */
rec_t* insert_rec;
rec_t* rec;
+ ulint* offsets;
ulint err;
ulint n_unique;
big_rec_t* big_rec = NULL;
@@ -2086,6 +2089,64 @@ row_ins_index_entry_low(
err = row_ins_clust_index_entry_by_modify(
mode, &cursor, &heap, &big_rec, entry,
thr, &mtr);
+
+ if (big_rec) {
+ ut_a(err == DB_SUCCESS);
+ /* Write out the externally stored
+ columns while still x-latching
+ index->lock and block->lock. Allocate
+ pages for big_rec in the mtr that
+ modified the B-tree, but be sure to skip
+ any pages that were freed in mtr. We will
+ write out the big_rec pages before
+ committing the B-tree mini-transaction. If
+ the system crashes so that crash recovery
+ will not replay the mtr_commit(&mtr), the
+ big_rec pages will be left orphaned until
+ the pages are allocated for something else.
+
+ TODO: If the allocation extends the
+ tablespace, it will not be redo
+ logged, in either mini-transaction.
+ Tablespace extension should be
+ redo-logged in the big_rec
+ mini-transaction, so that recovery
+ will not fail when the big_rec was
+ written to the extended portion of the
+ file, in case the file was somehow
+ truncated in the crash. */
+
+ rec = btr_cur_get_rec(&cursor);
+ offsets = rec_get_offsets(
+ rec, index, NULL,
+ ULINT_UNDEFINED, &heap);
+
+ DEBUG_SYNC_C("before_row_ins_upd_extern");
+ err = btr_store_big_rec_extern_fields(
+ index, btr_cur_get_block(&cursor),
+ rec, offsets, big_rec, &mtr,
+ BTR_STORE_INSERT_UPDATE);
+ DEBUG_SYNC_C("after_row_ins_upd_extern");
+ /* If writing big_rec fails (for
+ example, because of DB_OUT_OF_FILE_SPACE),
+ the record will be corrupted. Even if
+ we did not update any externally
+ stored columns, our update could cause
+ the record to grow so that a
+ non-updated column was selected for
+ external storage. This non-update
+ would not have been written to the
+ undo log, and thus the record cannot
+ be rolled back.
+
+ However, because we have not executed
+ mtr_commit(mtr) yet, the update will
+ not be replayed in crash recovery, and
+ the following assertion failure will
+ effectively "roll back" the operation. */
+ ut_a(err == DB_SUCCESS);
+ goto stored_big_rec;
+ }
} else {
ut_ad(!n_ext);
err = row_ins_sec_index_entry_by_modify(
@@ -2114,10 +2175,13 @@ function_exit:
mtr_commit(&mtr);
if (UNIV_LIKELY_NULL(big_rec)) {
- rec_t* rec;
- ulint* offsets;
+ DBUG_EXECUTE_IF(
+ "row_ins_extern_checkpoint",
+ log_make_checkpoint_at(IB_ULONGLONG_MAX, TRUE););
+
mtr_start(&mtr);
+ DEBUG_SYNC_C("before_row_ins_extern_latch");
btr_cur_search_to_nth_level(index, 0, entry, PAGE_CUR_LE,
BTR_MODIFY_TREE, &cursor, 0,
__FILE__, __LINE__, &mtr);
@@ -2125,10 +2189,13 @@ function_exit:
offsets = rec_get_offsets(rec, index, NULL,
ULINT_UNDEFINED, &heap);
+ DEBUG_SYNC_C("before_row_ins_extern");
err = btr_store_big_rec_extern_fields(
index, btr_cur_get_block(&cursor),
- rec, offsets, &mtr, FALSE, big_rec);
+ rec, offsets, big_rec, &mtr, BTR_STORE_INSERT);
+ DEBUG_SYNC_C("after_row_ins_extern");
+stored_big_rec:
if (modify) {
dtuple_big_rec_free(big_rec);
} else {
diff --git a/storage/innobase/row/row0merge.c b/storage/innobase/row/row0merge.c
index d42f21241ca..82169e7efee 100644
--- a/storage/innobase/row/row0merge.c
+++ b/storage/innobase/row/row0merge.c
@@ -1,6 +1,6 @@
/*****************************************************************************
-Copyright (c) 2005, 2010, Innobase Oy. All Rights Reserved.
+Copyright (c) 1995, 2012, Oracle and/or its affiliates. All Rights Reserved.
This program is free software; you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free Software
@@ -582,7 +582,7 @@ row_merge_buf_write(
REC_STATUS_ORDINARY,
entry, n_fields,
&extra_size);
- ut_ad(size > extra_size);
+ ut_ad(size >= extra_size);
ut_ad(extra_size >= REC_N_NEW_EXTRA_BYTES);
extra_size -= REC_N_NEW_EXTRA_BYTES;
size -= REC_N_NEW_EXTRA_BYTES;
@@ -2010,7 +2010,7 @@ row_merge_drop_index(
tables in Innobase. Deleting a row from SYS_INDEXES table also
frees the file segments of the B-tree associated with the index. */
- static const char str1[] =
+ static const char sql[] =
"PROCEDURE DROP_INDEX_PROC () IS\n"
"BEGIN\n"
/* Rename the index, so that it will be dropped by
@@ -2034,9 +2034,19 @@ row_merge_drop_index(
ut_a(trx->dict_operation_lock_mode == RW_X_LATCH);
- err = que_eval_sql(info, str1, FALSE, trx);
+ err = que_eval_sql(info, sql, FALSE, trx);
- ut_a(err == DB_SUCCESS);
+
+ if (err != DB_SUCCESS) {
+ /* Even though we ensure that DDL transactions are WAIT
+ and DEADLOCK free, we could encounter other errors e.g.,
+ DB_TOO_MANY_TRANSACTIONS. */
+ trx->error_state = DB_SUCCESS;
+
+ ut_print_timestamp(stderr);
+ fprintf(stderr, " InnoDB: Error: row_merge_drop_index failed "
+ "with error code: %lu.\n", (ulint) err);
+ }
/* Replace this index with another equivalent index for all
foreign key constraints on this table where this index is used */
@@ -2336,7 +2346,7 @@ row_merge_rename_indexes(
/* We use the private SQL parser of Innobase to generate the
query graphs needed in renaming indexes. */
- static const char rename_indexes[] =
+ static const char sql[] =
"PROCEDURE RENAME_INDEXES_PROC () IS\n"
"BEGIN\n"
"UPDATE SYS_INDEXES SET NAME=SUBSTR(NAME,1,LENGTH(NAME)-1)\n"
@@ -2352,7 +2362,7 @@ row_merge_rename_indexes(
pars_info_add_ull_literal(info, "tableid", table->id);
- err = que_eval_sql(info, rename_indexes, FALSE, trx);
+ err = que_eval_sql(info, sql, FALSE, trx);
if (err == DB_SUCCESS) {
dict_index_t* index = dict_table_get_first_index(table);
@@ -2362,6 +2372,15 @@ row_merge_rename_indexes(
}
index = dict_table_get_next_index(index);
} while (index);
+ } else {
+ /* Even though we ensure that DDL transactions are WAIT
+ and DEADLOCK free, we could encounter other errors e.g.,
+ DB_TOO_MANY_TRANSACTIONS. */
+ trx->error_state = DB_SUCCESS;
+
+ ut_print_timestamp(stderr);
+ fprintf(stderr, " InnoDB: Error: row_merge_rename_indexes "
+ "failed with error code: %lu.\n", (ulint) err);
}
trx->op_info = "";
@@ -2399,7 +2418,7 @@ row_merge_rename_tables(
memcpy(old_name, old_table->name, strlen(old_table->name) + 1);
} else {
ut_print_timestamp(stderr);
- fprintf(stderr, "InnoDB: too long table name: '%s', "
+ fprintf(stderr, " InnoDB: too long table name: '%s', "
"max length is %d\n", old_table->name,
MAX_FULL_NAME_LEN);
ut_error;
diff --git a/storage/innobase/row/row0row.c b/storage/innobase/row/row0row.c
index 20fd475343e..85c46f26933 100644
--- a/storage/innobase/row/row0row.c
+++ b/storage/innobase/row/row0row.c
@@ -1,6 +1,6 @@
/*****************************************************************************
-Copyright (c) 1996, 2011, Oracle and/or its affiliates. All Rights Reserved.
+Copyright (c) 1996, 2012, Oracle and/or its affiliates. All Rights Reserved.
This program is free software; you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free Software
@@ -11,8 +11,8 @@ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with
-this program; if not, write to the Free Software Foundation, Inc., 59 Temple
-Place, Suite 330, Boston, MA 02111-1307 USA
+this program; if not, write to the Free Software Foundation, Inc.,
+51 Franklin Street, Suite 500, Boston, MA 02110-1335 USA
*****************************************************************************/
@@ -241,13 +241,22 @@ row_build(
ut_ad(rec_offs_validate(rec, index, offsets));
}
-#if 0 && defined UNIV_BLOB_NULL_DEBUG
- /* This one can fail in trx_rollback_active() if
- the server crashed during an insert before the
- btr_store_big_rec_extern_fields() did mtr_commit()
- all BLOB pointers to the clustered index record. */
- ut_a(!rec_offs_any_null_extern(rec, offsets));
-#endif /* 0 && UNIV_BLOB_NULL_DEBUG */
+#ifdef UNIV_BLOB_NULL_DEBUG
+ if (rec_offs_any_null_extern(rec, offsets)) {
+ /* This condition can occur during crash recovery
+ before trx_rollback_active() has completed execution.
+
+ This condition is possible if the server crashed
+ during an insert or update-by-delete-and-insert before
+ btr_store_big_rec_extern_fields() did mtr_commit() all
+ BLOB pointers to the freshly inserted clustered index
+ record. */
+ ut_a(trx_assert_recovered(
+ row_get_rec_trx_id(rec, index, offsets)));
+ ut_a(trx_undo_roll_ptr_is_insert(
+ row_get_rec_roll_ptr(rec, index, offsets)));
+ }
+#endif /* UNIV_BLOB_NULL_DEBUG */
if (type != ROW_COPY_POINTERS) {
/* Take a copy of rec to heap */
diff --git a/storage/innobase/row/row0sel.c b/storage/innobase/row/row0sel.c
index 0d1d3aad1c6..7bec0a26225 100644
--- a/storage/innobase/row/row0sel.c
+++ b/storage/innobase/row/row0sel.c
@@ -3998,7 +3998,8 @@ rec_loop:
wrong_offs:
if (srv_force_recovery == 0 || moves_up == FALSE) {
ut_print_timestamp(stderr);
- buf_page_print(page_align(rec), 0);
+ buf_page_print(page_align(rec), 0,
+ BUF_PAGE_PRINT_NO_CRASH);
fprintf(stderr,
"\nInnoDB: rec address %p,"
" buf block fix count %lu\n",
@@ -4017,7 +4018,7 @@ wrong_offs:
"InnoDB: restore from a backup, or"
" dump + drop + reimport the table.\n",
stderr);
-
+ ut_ad(0);
err = DB_CORRUPTION;
goto lock_wait_or_error;
@@ -4369,7 +4370,9 @@ no_gap_lock:
applicable to unique secondary indexes. Current behaviour is
to widen the scope of a lock on an already delete marked record
if the same record is deleted twice by the same transaction */
- if (index == clust_index && unique_search) {
+ if (index == clust_index && unique_search
+ && !prebuilt->used_in_HANDLER) {
+
err = DB_RECORD_NOT_FOUND;
goto normal_return;
diff --git a/storage/innobase/row/row0upd.c b/storage/innobase/row/row0upd.c
index c57b829a1ca..4f5096a162b 100644
--- a/storage/innobase/row/row0upd.c
+++ b/storage/innobase/row/row0upd.c
@@ -11,8 +11,8 @@ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with
-this program; if not, write to the Free Software Foundation, Inc., 59 Temple
-Place, Suite 330, Boston, MA 02111-1307 USA
+this program; if not, write to the Free Software Foundation, Inc.,
+51 Franklin Street, Suite 500, Boston, MA 02110-1335 USA
*****************************************************************************/
@@ -23,6 +23,8 @@ Update of a row
Created 12/27/1996 Heikki Tuuri
*******************************************************/
+#include "m_string.h" /* for my_sys.h */
+#include "my_sys.h" /* DEBUG_SYNC_C */
#include "row0upd.h"
#ifdef UNIV_NONINL
@@ -2000,28 +2002,62 @@ row_upd_clust_rec(
ut_ad(!rec_get_deleted_flag(btr_pcur_get_rec(pcur),
dict_table_is_comp(index->table)));
- err = btr_cur_pessimistic_update(BTR_NO_LOCKING_FLAG, btr_cur,
- &heap, &big_rec, node->update,
- node->cmpl_info, thr, mtr);
- mtr_commit(mtr);
-
- if (err == DB_SUCCESS && big_rec) {
- ulint offsets_[REC_OFFS_NORMAL_SIZE];
- rec_t* rec;
+ err = btr_cur_pessimistic_update(
+ BTR_NO_LOCKING_FLAG | BTR_KEEP_POS_FLAG, btr_cur,
+ &heap, &big_rec, node->update, node->cmpl_info, thr, mtr);
+ if (big_rec) {
+ ulint offsets_[REC_OFFS_NORMAL_SIZE];
+ rec_t* rec;
rec_offs_init(offsets_);
- mtr_start(mtr);
+ ut_a(err == DB_SUCCESS);
+ /* Write out the externally stored
+ columns while still x-latching
+ index->lock and block->lock. Allocate
+ pages for big_rec in the mtr that
+ modified the B-tree, but be sure to skip
+ any pages that were freed in mtr. We will
+ write out the big_rec pages before
+ committing the B-tree mini-transaction. If
+ the system crashes so that crash recovery
+ will not replay the mtr_commit(&mtr), the
+ big_rec pages will be left orphaned until
+ the pages are allocated for something else.
+
+ TODO: If the allocation extends the tablespace, it
+ will not be redo logged, in either mini-transaction.
+ Tablespace extension should be redo-logged in the
+ big_rec mini-transaction, so that recovery will not
+ fail when the big_rec was written to the extended
+ portion of the file, in case the file was somehow
+ truncated in the crash. */
- ut_a(btr_pcur_restore_position(BTR_MODIFY_TREE, pcur, mtr));
rec = btr_cur_get_rec(btr_cur);
+ DEBUG_SYNC_C("before_row_upd_extern");
err = btr_store_big_rec_extern_fields(
index, btr_cur_get_block(btr_cur), rec,
rec_get_offsets(rec, index, offsets_,
ULINT_UNDEFINED, &heap),
- mtr, TRUE, big_rec);
- mtr_commit(mtr);
+ big_rec, mtr, BTR_STORE_UPDATE);
+ DEBUG_SYNC_C("after_row_upd_extern");
+ /* If writing big_rec fails (for example, because of
+ DB_OUT_OF_FILE_SPACE), the record will be corrupted.
+ Even if we did not update any externally stored
+ columns, our update could cause the record to grow so
+ that a non-updated column was selected for external
+ storage. This non-update would not have been written
+ to the undo log, and thus the record cannot be rolled
+ back.
+
+ However, because we have not executed mtr_commit(mtr)
+ yet, the update will not be replayed in crash
+ recovery, and the following assertion failure will
+ effectively "roll back" the operation. */
+ ut_a(err == DB_SUCCESS);
}
+ mtr_commit(mtr);
+
if (UNIV_LIKELY_NULL(heap)) {
mem_heap_free(heap);
}
diff --git a/storage/innobase/srv/srv0start.c b/storage/innobase/srv/srv0start.c
index edc655e93b9..343e41fe376 100644
--- a/storage/innobase/srv/srv0start.c
+++ b/storage/innobase/srv/srv0start.c
@@ -922,8 +922,9 @@ skip_size_check:
#endif /* UNIV_LOG_ARCHIVE */
min_flushed_lsn, max_flushed_lsn);
- if (UNIV_PAGE_SIZE
- != fsp_flags_get_page_size(flags)) {
+ if (!one_opened
+ && UNIV_PAGE_SIZE
+ != fsp_flags_get_page_size(flags)) {
ut_print_timestamp(stderr);
fprintf(stderr,
diff --git a/storage/innobase/sync/sync0rw.c b/storage/innobase/sync/sync0rw.c
index fc4f987fe65..8de9b40ef67 100644
--- a/storage/innobase/sync/sync0rw.c
+++ b/storage/innobase/sync/sync0rw.c
@@ -1,6 +1,6 @@
/*****************************************************************************
-Copyright (c) 1995, 2011, Oracle and/or its affiliates. All Rights Reserved.
+Copyright (c) 1995, 2012, Oracle and/or its affiliates. All Rights Reserved.
Copyright (c) 2008, Google Inc.
Portions of this file contain modifications contributed and copyrighted by
@@ -18,8 +18,8 @@ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with
-this program; if not, write to the Free Software Foundation, Inc., 59 Temple
-Place, Suite 330, Boston, MA 02111-1307 USA
+this program; if not, write to the Free Software Foundation, Inc.,
+51 Franklin Street, Suite 500, Boston, MA 02110-1335 USA
*****************************************************************************/
@@ -785,7 +785,9 @@ rw_lock_add_debug_info(
rw_lock_debug_mutex_exit();
if ((pass == 0) && (lock_type != RW_LOCK_WAIT_EX)) {
- sync_thread_add_level(lock, lock->level);
+ sync_thread_add_level(lock, lock->level,
+ lock_type == RW_LOCK_EX
+ && lock->lock_word < 0);
}
}
diff --git a/storage/innobase/sync/sync0sync.c b/storage/innobase/sync/sync0sync.c
index 81efe50d9bd..fb7101fdb8d 100644
--- a/storage/innobase/sync/sync0sync.c
+++ b/storage/innobase/sync/sync0sync.c
@@ -1,6 +1,6 @@
/*****************************************************************************
-Copyright (c) 1995, 2011, Innobase Oy. All Rights Reserved.
+Copyright (c) 1995, 2012, Oracle and/or its affiliates. All Rights Reserved.
Copyright (c) 2008, Google Inc.
Portions of this file contain modifications contributed and copyrighted by
@@ -18,8 +18,8 @@ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with
-this program; if not, write to the Free Software Foundation, Inc., 59 Temple
-Place, Suite 330, Boston, MA 02111-1307 USA
+this program; if not, write to the Free Software Foundation, Inc.,
+51 Franklin Street, Suite 500, Boston, MA 02110-1335 USA
*****************************************************************************/
@@ -693,7 +693,7 @@ mutex_set_debug_info(
ut_ad(mutex);
ut_ad(file_name);
- sync_thread_add_level(mutex, mutex->level);
+ sync_thread_add_level(mutex, mutex->level, FALSE);
mutex->file_name = file_name;
mutex->line = line;
@@ -1136,8 +1136,9 @@ void
sync_thread_add_level(
/*==================*/
void* latch, /*!< in: pointer to a mutex or an rw-lock */
- ulint level) /*!< in: level in the latching order; if
+ ulint level, /*!< in: level in the latching order; if
SYNC_LEVEL_VARYING, nothing is done */
+ ibool relock) /*!< in: TRUE if re-entering an x-lock */
{
ulint i;
sync_level_t* slot;
@@ -1188,6 +1189,10 @@ sync_thread_add_level(
array = thread_slot->levels;
+ if (relock) {
+ goto levels_ok;
+ }
+
/* NOTE that there is a problem with _NODE and _LEAF levels: if the
B-tree height changes, then a leaf can change to an internal node
or the other way around. We do not know at present if this can cause
@@ -1363,6 +1368,7 @@ sync_thread_add_level(
ut_error;
}
+levels_ok:
if (array->next_free == ULINT_UNDEFINED) {
ut_a(array->n_elems < array->max_elems);
diff --git a/storage/innobase/trx/trx0rec.c b/storage/innobase/trx/trx0rec.c
index 786c7be36a4..2f1389ae263 100644
--- a/storage/innobase/trx/trx0rec.c
+++ b/storage/innobase/trx/trx0rec.c
@@ -11,8 +11,8 @@ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with
-this program; if not, write to the Free Software Foundation, Inc., 59 Temple
-Place, Suite 330, Boston, MA 02111-1307 USA
+this program; if not, write to the Free Software Foundation, Inc.,
+51 Franklin Street, Suite 500, Boston, MA 02110-1335 USA
*****************************************************************************/
@@ -1198,6 +1198,7 @@ trx_undo_report_row_operation(
trx_t* trx;
trx_undo_t* undo;
ulint page_no;
+ buf_block_t* undo_block;
trx_rseg_t* rseg;
mtr_t mtr;
ulint err = DB_SUCCESS;
@@ -1240,10 +1241,13 @@ trx_undo_report_row_operation(
if (UNIV_UNLIKELY(!undo)) {
/* Did not succeed */
+ ut_ad(err != DB_SUCCESS);
mutex_exit(&(trx->undo_mutex));
return(err);
}
+
+ ut_ad(err == DB_SUCCESS);
} else {
ut_ad(op_type == TRX_UNDO_MODIFY_OP);
@@ -1257,30 +1261,30 @@ trx_undo_report_row_operation(
if (UNIV_UNLIKELY(!undo)) {
/* Did not succeed */
+ ut_ad(err != DB_SUCCESS);
mutex_exit(&(trx->undo_mutex));
return(err);
}
+ ut_ad(err == DB_SUCCESS);
offsets = rec_get_offsets(rec, index, offsets,
ULINT_UNDEFINED, &heap);
}
- page_no = undo->last_page_no;
-
mtr_start(&mtr);
+ page_no = undo->last_page_no;
+ undo_block = buf_page_get_gen(
+ undo->space, undo->zip_size, page_no, RW_X_LATCH,
+ undo->guess_block, BUF_GET, __FILE__, __LINE__, &mtr);
+ buf_block_dbg_add_level(undo_block, SYNC_TRX_UNDO_PAGE);
+
do {
- buf_block_t* undo_block;
page_t* undo_page;
ulint offset;
- undo_block = buf_page_get_gen(undo->space, undo->zip_size,
- page_no, RW_X_LATCH,
- undo->guess_block, BUF_GET,
- __FILE__, __LINE__, &mtr);
- buf_block_dbg_add_level(undo_block, SYNC_TRX_UNDO_PAGE);
-
undo_page = buf_block_get_frame(undo_block);
+ ut_ad(page_no == buf_block_get_page_no(undo_block));
if (op_type == TRX_UNDO_INSERT_OP) {
offset = trx_undo_page_report_insert(
@@ -1357,12 +1361,11 @@ trx_undo_report_row_operation(
a pessimistic insert in a B-tree, and we must reserve the
counterpart of the tree latch, which is the rseg mutex. */
- mutex_enter(&(rseg->mutex));
-
- page_no = trx_undo_add_page(trx, undo, &mtr);
-
- mutex_exit(&(rseg->mutex));
- } while (UNIV_LIKELY(page_no != FIL_NULL));
+ mutex_enter(&rseg->mutex);
+ undo_block = trx_undo_add_page(trx, undo, &mtr);
+ mutex_exit(&rseg->mutex);
+ page_no = undo->last_page_no;
+ } while (undo_block != NULL);
/* Did not succeed: out of space */
err = DB_OUT_OF_FILE_SPACE;
diff --git a/storage/innobase/trx/trx0sys.c b/storage/innobase/trx/trx0sys.c
index c0283448d10..90091c00228 100644
--- a/storage/innobase/trx/trx0sys.c
+++ b/storage/innobase/trx/trx0sys.c
@@ -1,6 +1,6 @@
/*****************************************************************************
-Copyright (c) 1996, 2011, Innobase Oy. All Rights Reserved.
+Copyright (c) 1996, 2012, Oracle and/or its affiliates. All Rights Reserved.
This program is free software; you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free Software
@@ -11,8 +11,8 @@ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with
-this program; if not, write to the Free Software Foundation, Inc., 59 Temple
-Place, Suite 330, Boston, MA 02111-1307 USA
+this program; if not, write to the Free Software Foundation, Inc.,
+51 Franklin Street, Suite 500, Boston, MA 02110-1335 USA
*****************************************************************************/
@@ -255,9 +255,7 @@ trx_sys_create_doublewrite_buf(void)
{
buf_block_t* block;
buf_block_t* block2;
-#ifdef UNIV_SYNC_DEBUG
buf_block_t* new_block;
-#endif /* UNIV_SYNC_DEBUG */
byte* doublewrite;
byte* fseg_header;
ulint page_no;
@@ -336,10 +334,9 @@ start_again:
for (i = 0; i < 2 * TRX_SYS_DOUBLEWRITE_BLOCK_SIZE
+ FSP_EXTENT_SIZE / 2; i++) {
- page_no = fseg_alloc_free_page(fseg_header,
- prev_page_no + 1,
- FSP_UP, &mtr);
- if (page_no == FIL_NULL) {
+ new_block = fseg_alloc_free_page(
+ fseg_header, prev_page_no + 1, FSP_UP, &mtr);
+ if (new_block == NULL) {
fprintf(stderr,
"InnoDB: Cannot create doublewrite"
" buffer: you must\n"
@@ -360,13 +357,8 @@ start_again:
the page position in the tablespace, then the page
has not been written to in doublewrite. */
-#ifdef UNIV_SYNC_DEBUG
- new_block =
-#endif /* UNIV_SYNC_DEBUG */
- buf_page_get(TRX_SYS_SPACE, 0, page_no,
- RW_X_LATCH, &mtr);
- buf_block_dbg_add_level(new_block,
- SYNC_NO_ORDER_CHECK);
+ ut_ad(rw_lock_get_x_lock_count(&new_block->lock) == 1);
+ page_no = buf_block_get_page_no(new_block);
if (i == FSP_EXTENT_SIZE / 2) {
ut_a(page_no == FSP_EXTENT_SIZE);
@@ -586,12 +578,16 @@ trx_sys_doublewrite_init_or_restore_pages(
if (buf_page_is_corrupted(page, zip_size)) {
fprintf(stderr,
"InnoDB: Dump of the page:\n");
- buf_page_print(read_buf, zip_size);
+ buf_page_print(
+ read_buf, zip_size,
+ BUF_PAGE_PRINT_NO_CRASH);
fprintf(stderr,
"InnoDB: Dump of"
" corresponding page"
" in doublewrite buffer:\n");
- buf_page_print(page, zip_size);
+ buf_page_print(
+ page, zip_size,
+ BUF_PAGE_PRINT_NO_CRASH);
fprintf(stderr,
"InnoDB: Also the page in the"
@@ -605,7 +601,7 @@ trx_sys_doublewrite_init_or_restore_pages(
"InnoDB: option:\n"
"InnoDB:"
" innodb_force_recovery=6\n");
- exit(1);
+ ut_error;
}
/* Write the good page from the
diff --git a/storage/innobase/trx/trx0trx.c b/storage/innobase/trx/trx0trx.c
index ab7677b5b35..85246aa6d1f 100644
--- a/storage/innobase/trx/trx0trx.c
+++ b/storage/innobase/trx/trx0trx.c
@@ -1065,6 +1065,8 @@ trx_commit_off_kernel(
ut_ad(UT_LIST_GET_LEN(trx->trx_locks) == 0);
UT_LIST_REMOVE(trx_list, trx_sys->trx_list, trx);
+
+ trx->error_state = DB_SUCCESS;
}
/****************************************************************//**
diff --git a/storage/innobase/trx/trx0undo.c b/storage/innobase/trx/trx0undo.c
index dae0637f72c..3d794c69c8b 100644
--- a/storage/innobase/trx/trx0undo.c
+++ b/storage/innobase/trx/trx0undo.c
@@ -1,6 +1,6 @@
/*****************************************************************************
-Copyright (c) 1996, 2009, Innobase Oy. All Rights Reserved.
+Copyright (c) 1996, 2012, Oracle and/or its affiliates. All Rights Reserved.
This program is free software; you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free Software
@@ -11,8 +11,8 @@ ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with
-this program; if not, write to the Free Software Foundation, Inc., 59 Temple
-Place, Suite 330, Boston, MA 02111-1307 USA
+this program; if not, write to the Free Software Foundation, Inc.,
+51 Franklin Street, Suite 500, Boston, MA 02110-1335 USA
*****************************************************************************/
@@ -876,9 +876,9 @@ trx_undo_discard_latest_update_undo(
#ifndef UNIV_HOTBACKUP
/********************************************************************//**
Tries to add a page to the undo log segment where the undo log is placed.
-@return page number if success, else FIL_NULL */
+@return X-latched block if success, else NULL */
UNIV_INTERN
-ulint
+buf_block_t*
trx_undo_add_page(
/*==============*/
trx_t* trx, /*!< in: transaction */
@@ -888,11 +888,10 @@ trx_undo_add_page(
the rollback segment mutex */
{
page_t* header_page;
+ buf_block_t* new_block;
page_t* new_page;
trx_rseg_t* rseg;
- ulint page_no;
ulint n_reserved;
- ibool success;
ut_ad(mutex_own(&(trx->undo_mutex)));
ut_ad(!mutex_own(&kernel_mutex));
@@ -902,37 +901,37 @@ trx_undo_add_page(
if (rseg->curr_size == rseg->max_size) {
- return(FIL_NULL);
+ return(NULL);
}
header_page = trx_undo_page_get(undo->space, undo->zip_size,
undo->hdr_page_no, mtr);
- success = fsp_reserve_free_extents(&n_reserved, undo->space, 1,
- FSP_UNDO, mtr);
- if (!success) {
+ if (!fsp_reserve_free_extents(&n_reserved, undo->space, 1,
+ FSP_UNDO, mtr)) {
- return(FIL_NULL);
+ return(NULL);
}
- page_no = fseg_alloc_free_page_general(header_page + TRX_UNDO_SEG_HDR
- + TRX_UNDO_FSEG_HEADER,
- undo->top_page_no + 1, FSP_UP,
- TRUE, mtr);
+ new_block = fseg_alloc_free_page_general(
+ TRX_UNDO_SEG_HDR + TRX_UNDO_FSEG_HEADER
+ + header_page,
+ undo->top_page_no + 1, FSP_UP, TRUE, mtr, mtr);
fil_space_release_free_extents(undo->space, n_reserved);
- if (page_no == FIL_NULL) {
+ if (new_block == NULL) {
/* No space left */
- return(FIL_NULL);
+ return(NULL);
}
- undo->last_page_no = page_no;
+ ut_ad(rw_lock_get_x_lock_count(&new_block->lock) == 1);
+ buf_block_dbg_add_level(new_block, SYNC_TRX_UNDO_PAGE);
+ undo->last_page_no = buf_block_get_page_no(new_block);
- new_page = trx_undo_page_get(undo->space, undo->zip_size,
- page_no, mtr);
+ new_page = buf_block_get_frame(new_block);
trx_undo_page_init(new_page, undo->type, mtr);
@@ -941,7 +940,7 @@ trx_undo_add_page(
undo->size++;
rseg->curr_size++;
- return(page_no);
+ return(new_block);
}
/********************************************************************//**
diff --git a/storage/myisam/ft_boolean_search.c b/storage/myisam/ft_boolean_search.c
index 56d0149a748..f419ac3d689 100644
--- a/storage/myisam/ft_boolean_search.c
+++ b/storage/myisam/ft_boolean_search.c
@@ -1,4 +1,4 @@
-/* Copyright (c) 2001, 2010, Oracle and/or its affiliates. All rights reserved.
+/* Copyright (c) 2001, 2011, Oracle and/or its affiliates. All rights reserved.
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
diff --git a/storage/myisam/ft_nlq_search.c b/storage/myisam/ft_nlq_search.c
index 6b36891ccf6..e31075e55d1 100644
--- a/storage/myisam/ft_nlq_search.c
+++ b/storage/myisam/ft_nlq_search.c
@@ -1,5 +1,4 @@
-/*
- Copyright (c) 2001, 2010, Oracle and/or its affiliates
+/* Copyright (c) 2001, 2011, Oracle and/or its affiliates.
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
diff --git a/storage/myisam/ha_myisam.cc b/storage/myisam/ha_myisam.cc
index 87fccb6f449..3f99dfc2186 100644
--- a/storage/myisam/ha_myisam.cc
+++ b/storage/myisam/ha_myisam.cc
@@ -1,5 +1,5 @@
/*
- Copyright (c) 2000, 2011, Oracle and/or its affiliates
+ Copyright (c) 2000, 2012, Oracle and/or its affiliates.
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -74,10 +74,10 @@ static MYSQL_THDVAR_ULONG(repair_threads, PLUGIN_VAR_RQCMDARG,
"disables parallel repair", NULL, NULL,
1, 1, ULONG_MAX, 1);
-static MYSQL_THDVAR_ULONG(sort_buffer_size, PLUGIN_VAR_RQCMDARG,
+static MYSQL_THDVAR_ULONGLONG(sort_buffer_size, PLUGIN_VAR_RQCMDARG,
"The buffer that is allocated when sorting the index when doing "
"a REPAIR or when creating indexes with CREATE INDEX or ALTER TABLE", NULL, NULL,
- 8192*1024, (long) (MIN_SORT_BUFFER + MALLOC_OVERHEAD), ULONG_MAX, 1);
+ 8192 * 1024, (long) (MIN_SORT_BUFFER + MALLOC_OVERHEAD), SIZE_T_MAX, 1);
static MYSQL_SYSVAR_BOOL(use_mmap, opt_myisam_use_mmap, PLUGIN_VAR_NOCMDARG,
"Use memory mapping for reading and writing MyISAM tables", NULL, NULL, FALSE);
diff --git a/storage/myisam/ha_myisam.h b/storage/myisam/ha_myisam.h
index 7439ea290c6..aa00398e7e7 100644
--- a/storage/myisam/ha_myisam.h
+++ b/storage/myisam/ha_myisam.h
@@ -1,5 +1,5 @@
/*
- Copyright (c) 2000, 2011, Oracle and/or its affiliates
+ Copyright (c) 2000, 2012, Oracle and/or its affiliates.
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -36,7 +36,6 @@ typedef struct st_ha_create_information HA_CREATE_INFO;
#define HA_RECOVER_FULL_BACKUP 16 /* Make a copy of index file too */
#define HA_RECOVER_OFF 32 /* No automatic recover */
-extern ulong myisam_sort_buffer_size;
extern TYPELIB myisam_recover_typelib;
extern const char *myisam_recover_names[];
extern ulonglong myisam_recover_options;
diff --git a/storage/myisam/mi_check.c b/storage/myisam/mi_check.c
index 04d2263ab5b..7e97a751376 100644
--- a/storage/myisam/mi_check.c
+++ b/storage/myisam/mi_check.c
@@ -1,5 +1,5 @@
/*
- Copyright (c) 2000, 2011, Oracle and/or its affiliates
+ Copyright (c) 2000, 2012, Oracle and/or its affiliates.
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -2406,7 +2406,7 @@ int mi_repair_by_sort(HA_CHECK *param, register MI_INFO *info,
if (_create_index_by_sort(&sort_param,
(my_bool) (!(param->testflag & T_VERBOSE)),
- (uint) param->sort_buffer_length))
+ param->sort_buffer_length))
{
param->retry_repair= 1;
if (! param->error_printed)
@@ -4312,14 +4312,6 @@ int recreate_table(HA_CHECK *param, MI_INFO **org_info, char *filename)
u_ptr->seg=keyseg;
keyseg+=u_ptr->keysegs+1;
}
- if (share.options & HA_OPTION_COMPRESS_RECORD)
- share.base.records=max_records=info.state->records;
- else if (share.base.min_pack_length)
- max_records=(ha_rows) (mysql_file_seek(info.dfile, 0L, MY_SEEK_END,
- MYF(0)) /
- (ulong) share.base.min_pack_length);
- else
- max_records=0;
unpack= (share.options & HA_OPTION_COMPRESS_RECORD) &&
(param->testflag & T_UNPACK);
share.options&= ~HA_OPTION_TEMP_COMPRESS_RECORD;
@@ -4329,10 +4321,17 @@ int recreate_table(HA_CHECK *param, MI_INFO **org_info, char *filename)
set_if_bigger(file_length,param->max_data_file_length);
set_if_bigger(file_length,tmp_length);
set_if_bigger(file_length,(ulonglong) share.base.max_data_file_length);
+
+ if (share.options & HA_OPTION_COMPRESS_RECORD)
+ share.base.records= max_records= info.state->records;
+ else if (!(share.options & HA_OPTION_PACK_RECORD))
+ max_records= (ha_rows) (file_length / share.base.pack_reclength);
+ else
+ max_records= 0;
(void) mi_close(*org_info);
bzero((char*) &create_info,sizeof(create_info));
- create_info.max_rows=max(max_records,share.base.records);
+ create_info.max_rows= max_records;
create_info.reloc_rows=share.base.reloc;
create_info.old_options=(share.options |
(unpack ? HA_OPTION_TEMP_COMPRESS_RECORD : 0));
diff --git a/storage/myisam/mi_packrec.c b/storage/myisam/mi_packrec.c
index c154ec3b817..7e2403b64c9 100644
--- a/storage/myisam/mi_packrec.c
+++ b/storage/myisam/mi_packrec.c
@@ -1,4 +1,4 @@
-/* Copyright (c) 2000, 2010, Oracle and/or its affiliates. All rights reserved.
+/* Copyright (c) 2000, 2011, Oracle and/or its affiliates. All rights reserved.
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
diff --git a/storage/myisam/mi_preload.c b/storage/myisam/mi_preload.c
index 767858a6ea0..1a2d5aac94f 100644
--- a/storage/myisam/mi_preload.c
+++ b/storage/myisam/mi_preload.c
@@ -1,4 +1,4 @@
-/* Copyright (c) 2003, 2010, Oracle and/or its affiliates. All rights reserved.
+/* Copyright (c) 2003, 2011, Oracle and/or its affiliates. All rights reserved.
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
diff --git a/storage/myisam/myisamchk.c b/storage/myisam/myisamchk.c
index 32d39e5a09d..fa8cc679997 100644
--- a/storage/myisam/myisamchk.c
+++ b/storage/myisam/myisamchk.c
@@ -1,5 +1,4 @@
-/*
- Copyright (c) 2000, 2011, Oracle and/or its affiliates
+/* Copyright (c) 2000, 2012, Oracle and/or its affiliates.
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -297,9 +296,9 @@ static struct my_option my_long_options[] =
INT_MAX32, (long) MALLOC_OVERHEAD, (long) 1L, 0},
{ "sort_buffer_size", OPT_SORT_BUFFER_SIZE, "",
&check_param.sort_buffer_length,
- &check_param.sort_buffer_length, 0, GET_ULONG, REQUIRED_ARG,
+ &check_param.sort_buffer_length, 0, GET_ULL, REQUIRED_ARG,
(long) SORT_BUFFER_INIT, (long) (MIN_SORT_BUFFER + MALLOC_OVERHEAD),
- ULONG_MAX, (long) MALLOC_OVERHEAD, (long) 1L, 0},
+ SIZE_T_MAX, (long) MALLOC_OVERHEAD, (long) 1L, 0},
{ "sort_key_blocks", OPT_SORT_KEY_BLOCKS, "",
&check_param.sort_key_blocks,
&check_param.sort_key_blocks, 0, GET_ULONG, REQUIRED_ARG,
diff --git a/storage/myisam/myisamdef.h b/storage/myisam/myisamdef.h
index a72b329a4c0..e3c1647cd59 100644
--- a/storage/myisam/myisamdef.h
+++ b/storage/myisam/myisamdef.h
@@ -1,5 +1,5 @@
/*
- Copyright (c) 2000, 2011, Oracle and/or its affiliates
+ Copyright (c) 2000, 2012, Oracle and/or its affiliates.
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
diff --git a/storage/myisam/sort.c b/storage/myisam/sort.c
index 6da2c5c7816..2c4639b3143 100644
--- a/storage/myisam/sort.c
+++ b/storage/myisam/sort.c
@@ -1,5 +1,4 @@
-/*
- Copyright (c) 2000, 2011, Oracle and/or its affiliates
+/* Copyright (c) 2000, 2012, Oracle and/or its affiliates.
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
@@ -100,10 +99,11 @@ my_var_write(MI_SORT_PARAM *info, IO_CACHE *to_file, uchar *bufs);
*/
int _create_index_by_sort(MI_SORT_PARAM *info,my_bool no_messages,
- ulong sortbuff_size)
+ ulonglong sortbuff_size)
{
int error,maxbuffer,skr;
- uint memavl,old_memavl,keys,sort_length;
+ uint sort_length, keys;
+ ulonglong memavl, old_memavl;
DYNAMIC_ARRAY buffpek;
ha_rows records;
uchar **sort_keys;
@@ -135,6 +135,9 @@ int _create_index_by_sort(MI_SORT_PARAM *info,my_bool no_messages,
sort_length= info->key_length;
LINT_INIT(keys);
+ if ((memavl - sizeof(BUFFPEK)) / (sort_length + sizeof(char *)) > UINT_MAX32)
+ memavl= sizeof(BUFFPEK) + UINT_MAX32 * (sort_length + sizeof(char *));
+
while (memavl >= MIN_SORT_BUFFER)
{
if ((records < UINT_MAX32) &&
@@ -311,7 +314,8 @@ pthread_handler_t thr_find_all_keys(void *arg)
{
MI_SORT_PARAM *sort_param= (MI_SORT_PARAM*) arg;
int error;
- uint memavl,old_memavl,keys,sort_length;
+ ulonglong memavl, old_memavl;
+ uint keys, sort_length;
uint idx, maxbuffer;
uchar **sort_keys=0;
@@ -352,6 +356,10 @@ pthread_handler_t thr_find_all_keys(void *arg)
sort_length= sort_param->key_length;
maxbuffer= 1;
+ if ((memavl - sizeof(BUFFPEK)) / (sort_length +
+ sizeof(char *)) > UINT_MAX32)
+ memavl= sizeof(BUFFPEK) + UINT_MAX32 * (sort_length + sizeof(char *));
+
while (memavl >= MIN_SORT_BUFFER)
{
if ((my_off_t) (idx+1)*(sort_length+sizeof(char*)) <=
diff --git a/storage/perfschema/ha_perfschema.cc b/storage/perfschema/ha_perfschema.cc
index d7771de7fb8..0fb86cfe5cd 100644
--- a/storage/perfschema/ha_perfschema.cc
+++ b/storage/perfschema/ha_perfschema.cc
@@ -1,4 +1,4 @@
-/* Copyright (c) 2008, 2010, Oracle and/or its affiliates. All rights reserved.
+/* Copyright (c) 2008, 2011, Oracle and/or its affiliates. All rights reserved.
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
diff --git a/storage/perfschema/pfs.cc b/storage/perfschema/pfs.cc
index 27350c59604..9a6ada2f814 100644
--- a/storage/perfschema/pfs.cc
+++ b/storage/perfschema/pfs.cc
@@ -1,4 +1,4 @@
-/* Copyright (c) 2008, 2010, Oracle and/or its affiliates. All rights reserved.
+/* Copyright (c) 2008, 2011, Oracle and/or its affiliates. All rights reserved.
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
diff --git a/storage/perfschema/table_ews_global_by_event_name.cc b/storage/perfschema/table_ews_global_by_event_name.cc
index 01d9bc5678f..3177584231d 100644
--- a/storage/perfschema/table_ews_global_by_event_name.cc
+++ b/storage/perfschema/table_ews_global_by_event_name.cc
@@ -1,4 +1,4 @@
-/* Copyright (c) 2010, 2010, Oracle and/or its affiliates. All rights reserved.
+/* Copyright (c) 2010, Oracle and/or its affiliates. All rights reserved.
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
diff --git a/storage/perfschema/table_ews_global_by_event_name.h b/storage/perfschema/table_ews_global_by_event_name.h
index 9b2b5cba5a8..7e66448e96c 100644
--- a/storage/perfschema/table_ews_global_by_event_name.h
+++ b/storage/perfschema/table_ews_global_by_event_name.h
@@ -1,4 +1,4 @@
-/* Copyright (c) 2010, 2010, Oracle and/or its affiliates. All rights reserved.
+/* Copyright (c) 2010, Oracle and/or its affiliates. All rights reserved.
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
diff --git a/storage/xtradb/row/row0merge.c b/storage/xtradb/row/row0merge.c
index 30cfa7c23af..9390946ad88 100644
--- a/storage/xtradb/row/row0merge.c
+++ b/storage/xtradb/row/row0merge.c
@@ -583,7 +583,7 @@ row_merge_buf_write(
REC_STATUS_ORDINARY,
entry, n_fields,
&extra_size);
- ut_ad(size > extra_size);
+ ut_ad(size >= extra_size);
ut_ad(extra_size >= REC_N_NEW_EXTRA_BYTES);
extra_size -= REC_N_NEW_EXTRA_BYTES;
size -= REC_N_NEW_EXTRA_BYTES;
diff --git a/storage/xtradb/row/row0sel.c b/storage/xtradb/row/row0sel.c
index f4a91e68024..8e6b01f2a6d 100644
--- a/storage/xtradb/row/row0sel.c
+++ b/storage/xtradb/row/row0sel.c
@@ -4604,7 +4604,9 @@ no_gap_lock:
applicable to unique secondary indexes. Current behaviour is
to widen the scope of a lock on an already delete marked record
if the same record is deleted twice by the same transaction */
- if (index == clust_index && unique_search) {
+ if (index == clust_index && unique_search
+ && !prebuilt->used_in_HANDLER) {
+
err = DB_RECORD_NOT_FOUND;
goto normal_return;