mirror/mariadb
1
0
Fork 0
mirror of https://github.com/MariaDB/server.git synced 2026-05-16 20:07:13 +02:00
Code Issues Projects Releases Packages Wiki Activity

Merge r3403..3414 from mysql-5.1-innodb:

Browse source 
------------------------------------------------------------
  revno: 3414
  revision-id: marko.makela@oracle.com-20100422093342-jf9ojlzdqsdebohn
  parent: marko.makela@oracle.com-20100421185359-8qaxoa2yyrpzwdd7
  committer: Marko M?kel? <marko.makela@oracle.com>
  branch nick: 5.1-innodb
  timestamp: Thu 2010-04-22 12:33:42 +0300
  message:
    Correct the definition of DICT_SYS_INDEXES_NAME_FIELD.
    When row_merge_drop_temp_indexes() was reworked to drop the indexes
    via the data dictionary cache, the code was broken because it would
    read the index name from the wrong field.
  modified:
    storage/innodb_plugin/ChangeLog 2425@16c675df-0fcb-4bc9-8058-dcc011a37293:branches%2Fzip%2FChangeLog
    storage/innodb_plugin/dict/dict0boot.c 2@16c675df-0fcb-4bc9-8058-dcc011a37293:trunk%2Fdict%2Fdict0boot.c
    storage/innodb_plugin/include/dict0boot.h 2@16c675df-0fcb-4bc9-8058-dcc011a37293:trunk%2Finclude%2Fdict0boot.h
  ------------------------------------------------------------
  revno: 3413
  revision-id: marko.makela@oracle.com-20100421185359-8qaxoa2yyrpzwdd7
  parent: marko.makela@oracle.com-20100421102723-0i80uezbyu0ekj5d
  committer: Marko M?kel? <marko.makela@oracle.com>
  branch nick: 5.1-innodb
  timestamp: Wed 2010-04-21 21:53:59 +0300
  message:
    btr_page_split_and_insert(): Avoid an infinite loop. (Bug #52964)
    
    btr_page_tuple_smaller(): New function, refactored from
    btr_page_split_and_insert().
    
    btr_page_get_split_rec(): Renamed from btr_page_get_sure_split_rec().
    Note that a NULL return may mean that the tuple is to be inserted into
    either the lower or upper page, to be determined by btr_page_tuple_smaller().
    
    btr_page_split_and_insert(): When btr_page_get_split_rec() returns NULL,
    invoke btr_page_tuple_smaller() to determine which half-page the tuple
    belongs to.
    
    Reviewed by Sunny Bains
  modified:
    storage/innodb_plugin/ChangeLog 2425@16c675df-0fcb-4bc9-8058-dcc011a37293:branches%2Fzip%2FChangeLog
    storage/innodb_plugin/btr/btr0btr.c 2@16c675df-0fcb-4bc9-8058-dcc011a37293:trunk%2Fbtr%2Fbtr0btr.c
  ------------------------------------------------------------
  revno: 3412
  revision-id: marko.makela@oracle.com-20100421102723-0i80uezbyu0ekj5d
  parent: marko.makela@oracle.com-20100421100029-mji6lmdgvuqh96qq
  committer: Marko M?kel? <marko.makela@oracle.com>
  branch nick: 5.1-innodb
  timestamp: Wed 2010-04-21 13:27:23 +0300
  message:
    dict_create_index_step(): Be strict about DYNAMIC and COMPRESSED tables.
    Bug #50495 is about REDUNDANT and COMPACT tables, after all.
  modified:
    mysql-test/suite/innodb_plugin/r/innodb-zip.result 2252@16c675df-0fcb-4bc9-8058-dcc011a37293:branches%2Fzip%2Fmysql-test%2Finnodb-zip.result
    mysql-test/suite/innodb_plugin/t/innodb-zip.test 2252@16c675df-0fcb-4bc9-8058-dcc011a37293:branches%2Fzip%2Fmysql-test%2Finnodb-zip.test
    storage/innodb_plugin/ChangeLog 2425@16c675df-0fcb-4bc9-8058-dcc011a37293:branches%2Fzip%2FChangeLog
    storage/innodb_plugin/dict/dict0crea.c 2@16c675df-0fcb-4bc9-8058-dcc011a37293:trunk%2Fdict%2Fdict0crea.c
  ------------------------------------------------------------
  revno: 3411
  revision-id: marko.makela@oracle.com-20100421100029-mji6lmdgvuqh96qq
  parent: marko.makela@oracle.com-20100421095033-0acvzxb8um8cms0a
  committer: Marko M?kel? <marko.makela@oracle.com>
  branch nick: 5.1-innodb
  timestamp: Wed 2010-04-21 13:00:29 +0300
  message:
    ha_innobase::add_index(): Only check for duplicate indexes
    when the data dictionary is locked.
    This fixes a UNIV_DEBUG assertion failure in innodb-index.test.
  modified:
    storage/innodb_plugin/handler/handler0alter.cc 1845@16c675df-0fcb-4bc9-8058-dcc011a37293:branches%2Fzip%2Fhandler%2Fhandler0alter.cc
  ------------------------------------------------------------
  revno: 3410
  revision-id: marko.makela@oracle.com-20100421095033-0acvzxb8um8cms0a
  parent: marko.makela@oracle.com-20100421094032-ir4glqk46qvg2ywn
  committer: Marko M?kel? <marko.makela@oracle.com>
  branch nick: 5.1-innodb
  timestamp: Wed 2010-04-21 12:50:33 +0300
  message:
    dtuple_convert_big_rec(): Store locally any fields whose maximum length
    is less than 256 bytes. (Bug #52745)
    Add related comments and debug assertions to the "offsets"
    functions in rem0rec.c.
    Approved by Sunny Bains
  modified:
    storage/innodb_plugin/ChangeLog 2425@16c675df-0fcb-4bc9-8058-dcc011a37293:branches%2Fzip%2FChangeLog
    storage/innodb_plugin/data/data0data.c 2@16c675df-0fcb-4bc9-8058-dcc011a37293:trunk%2Fdata%2Fdata0data.c
    storage/innodb_plugin/rem/rem0rec.c 2@16c675df-0fcb-4bc9-8058-dcc011a37293:trunk%2Frem%2Frem0rec.c
  ------------------------------------------------------------
  revno: 3409
  revision-id: marko.makela@oracle.com-20100421094032-ir4glqk46qvg2ywn
  parent: marko.makela@oracle.com-20100421091611-uu46iygmv5sizjru
  committer: Marko M?kel? <marko.makela@oracle.com>
  branch nick: 5.1-innodb
  timestamp: Wed 2010-04-21 12:40:32 +0300
  message:
    Adjust tests for the Bug #50495 fix.
  modified:
    mysql-test/suite/innodb_plugin/r/innodb-zip.result 2252@16c675df-0fcb-4bc9-8058-dcc011a37293:branches%2Fzip%2Fmysql-test%2Finnodb-zip.result
    mysql-test/suite/innodb_plugin/t/innodb-zip.test 2252@16c675df-0fcb-4bc9-8058-dcc011a37293:branches%2Fzip%2Fmysql-test%2Finnodb-zip.test
    mysql-test/suite/innodb_plugin/t/innodb_bug36169.test 2418@16c675df-0fcb-4bc9-8058-dcc011a37293:branches%2Fzip%2Fmysql-test%2Finnodb_bug36169.test
  ------------------------------------------------------------
  revno: 3408
  revision-id: marko.makela@oracle.com-20100421091611-uu46iygmv5sizjru
  parent: marko.makela@oracle.com-20100421063916-h3gwjea7jzjilyww
  committer: Marko M?kel? <marko.makela@oracle.com>
  branch nick: 5.1-innodb
  timestamp: Wed 2010-04-21 12:16:11 +0300
  message:
    rec_convert_dtuple_to_rec(): Correct the debug check.
    The "extern" accessor functions return zero or nonzero, not 0 or 1.
  modified:
    storage/innodb_plugin/rem/rem0rec.c 2@16c675df-0fcb-4bc9-8058-dcc011a37293:trunk%2Frem%2Frem0rec.c
  ------------------------------------------------------------
  revno: 3407
  revision-id: marko.makela@oracle.com-20100421063916-h3gwjea7jzjilyww
  parent: marko.makela@oracle.com-20100420201550-cax1xywvlcdshgfg
  committer: Marko M?kel? <marko.makela@oracle.com>
  branch nick: 5.1-innodb
  timestamp: Wed 2010-04-21 09:39:16 +0300
  message:
    rec_convert_dtuple_to_rec(): Add a debug check.
  modified:
    storage/innodb_plugin/rem/rem0rec.c 2@16c675df-0fcb-4bc9-8058-dcc011a37293:trunk%2Frem%2Frem0rec.c
  ------------------------------------------------------------
  revno: 3406
  revision-id: marko.makela@oracle.com-20100420201550-cax1xywvlcdshgfg
  parent: marko.makela@oracle.com-20100420114222-diq7h2hiom9ww6me
  committer: Marko Makela <marko.makela@oracle.com>
  branch nick: 5.1-innodb
  timestamp: Tue 2010-04-20 23:15:50 +0300
  message:
    btr_cur_optimistic_insert(): Remove unused variable "heap".
  modified:
    storage/innodb_plugin/btr/btr0cur.c 2@16c675df-0fcb-4bc9-8058-dcc011a37293:trunk%2Fbtr%2Fbtr0cur.c
  ------------------------------------------------------------
  revno: 3405
  revision-id: marko.makela@oracle.com-20100420114222-diq7h2hiom9ww6me
  parent: marko.makela@oracle.com-20100420082908-tpako8jd4imkh1xb
  committer: Marko Makela <marko.makela@oracle.com>
  branch nick: 5.1-innodb
  timestamp: Tue 2010-04-20 14:42:22 +0300
  message:
    dict_create_index_step(): Invoke dict_index_add_to_cache()
    in strict mode only if innodb_strict_mode is set. (Bug #50495)
    
    trx_is_strict(): New function, for checking innodb_strict_mode.
  modified:
    storage/innodb_plugin/ChangeLog 2425@16c675df-0fcb-4bc9-8058-dcc011a37293:branches%2Fzip%2FChangeLog
    storage/innodb_plugin/dict/dict0crea.c 2@16c675df-0fcb-4bc9-8058-dcc011a37293:trunk%2Fdict%2Fdict0crea.c
    storage/innodb_plugin/handler/ha_innodb.cc 2@16c675df-0fcb-4bc9-8058-dcc011a37293:trunk%2Fhandler%2Fha_innodb.cc
    storage/innodb_plugin/include/trx0trx.h 2@16c675df-0fcb-4bc9-8058-dcc011a37293:trunk%2Finclude%2Ftrx0trx.h
  ------------------------------------------------------------
  revno: 3404
  revision-id: marko.makela@oracle.com-20100420082908-tpako8jd4imkh1xb
  parent: marko.makela@oracle.com-20100419103603-u5pz4qc6hfhx4nua
  committer: Marko M?kel? <marko.makela@oracle.com>
  branch nick: 5.1-innodb
  timestamp: Tue 2010-04-20 11:29:08 +0300
  message:
    Implement UNIV_BTR_AVOID_COPY, an optimization of page splits.
  modified:
    storage/innodb_plugin/ChangeLog 2425@16c675df-0fcb-4bc9-8058-dcc011a37293:branches%2Fzip%2FChangeLog
    storage/innodb_plugin/btr/btr0btr.c 2@16c675df-0fcb-4bc9-8058-dcc011a37293:trunk%2Fbtr%2Fbtr0btr.c
    storage/innodb_plugin/include/univ.i 2@16c675df-0fcb-4bc9-8058-dcc011a37293:trunk%2Finclude%2Funiv.i
  ------------------------------------------------------------
  revno: 3403
  revision-id: marko.makela@oracle.com-20100419103603-u5pz4qc6hfhx4nua
  parent: marko.makela@oracle.com-20100419094405-fd3xwadullishv07
  committer: Marko =?ISO-8859-1?Q?M=E4kel=E4?= <marko.makela@oracle.com>
  branch nick: 5.1-innodb
  timestamp: Mon 2010-04-19 13:36:03 +0300
  message:
    Enable UNIV_DEBUG_VALGRIND when HAVE_purify is set.
  modified:
    storage/innobase/include/univ.i 2@cee13dc7-1704-0410-992b-c9b4543f1246:trunk%2Finclude%2Funiv.i
    storage/innodb_plugin/include/univ.i 2@16c675df-0fcb-4bc9-8058-dcc011a37293:trunk%2Finclude%2Funiv.i
This commit is contained in:
Vasil Dimov 2010-04-22 14:24:42 +03:00
commit 66b4a1e614
13 changed files with 200 additions and 57 deletions
Download patch file Download diff file Expand all files Collapse all files

111
storage/innobase/btr/btr0btr.c
View file

@ -1455,11 +1455,11 @@ Calculates a split record such that the tuple will certainly fit on
its half-page when the split is performed. We assume in this function
only that the cursor page has at least one user record.
@return split record, or NULL if tuple will be the first record on
upper half-page */
the lower or upper half-page (determined by btr_page_tuple_smaller()) */
static
rec_t*
btr_page_get_sure_split_rec(
/*========================*/
btr_page_get_split_rec(
/*===================*/
btr_cur_t* cursor, /*!< in: cursor at which insert should be made */
const dtuple_t* tuple, /*!< in: tuple to insert */
ulint n_ext) /*!< in: number of externally stored columns */
@ -1835,6 +1835,37 @@ btr_attach_half_pages(
btr_page_set_next(upper_page, upper_page_zip, next_page_no, mtr);
}
/*************************************************************//**
Determine if a tuple is smaller than any record on the page.
@return TRUE if smaller */
static
ibool
btr_page_tuple_smaller(
/*===================*/
btr_cur_t* cursor, /*!< in: b-tree cursor */
const dtuple_t* tuple, /*!< in: tuple to consider */
ulint* offsets,/*!< in/out: temporary storage */
ulint n_uniq, /*!< in: number of unique fields
in the index page records */
mem_heap_t** heap) /*!< in/out: heap for offsets */
{
buf_block_t* block;
const rec_t* first_rec;
page_cur_t pcur;
/* Read the first user record in the page. */
block = btr_cur_get_block(cursor);
page_cur_set_before_first(block, &pcur);
page_cur_move_to_next(&pcur);
first_rec = page_cur_get_rec(&pcur);
offsets = rec_get_offsets(
first_rec, cursor->index, offsets,
n_uniq, heap);
return(cmp_dtuple_rec(tuple, first_rec, offsets) < 0);
}
/*************************************************************//**
Splits an index page to halves and inserts the tuple. It is assumed
that mtr holds an x-latch to the index tree. NOTE: the tree x-latch is
@ -1909,49 +1940,45 @@ func_start:
if (n_iterations > 0) {
direction = FSP_UP;
hint_page_no = page_no + 1;
split_rec = btr_page_get_sure_split_rec(cursor, tuple, n_ext);
split_rec = btr_page_get_split_rec(cursor, tuple, n_ext);
if (UNIV_UNLIKELY(split_rec == NULL)) {
insert_left = btr_page_tuple_smaller(
cursor, tuple, offsets, n_uniq, &heap);
}
} else if (btr_page_get_split_rec_to_right(cursor, &split_rec)) {
direction = FSP_UP;
hint_page_no = page_no + 1;
insert_left = FALSE;
} else if (btr_page_get_split_rec_to_left(cursor, &split_rec)) {
direction = FSP_DOWN;
hint_page_no = page_no - 1;
ut_ad(split_rec);
} else {
direction = FSP_UP;
hint_page_no = page_no + 1;
if (page_get_n_recs(page) == 1) {
page_cur_t pcur;
/* If there is only one record in the index page, we
can't split the node in the middle by default. We need
to determine whether the new record will be inserted
to the left or right. */
/* There is only one record in the index page
therefore we can't split the node in the middle
by default. We need to determine whether the
new record will be inserted to the left or right. */
/* Read the first (and only) record in the page. */
page_cur_set_before_first(block, &pcur);
page_cur_move_to_next(&pcur);
first_rec = page_cur_get_rec(&pcur);
offsets = rec_get_offsets(
first_rec, cursor->index, offsets,
n_uniq, &heap);
/* If the new record is less than the existing record
the split in the middle will copy the existing
record to the new node. */
if (cmp_dtuple_rec(tuple, first_rec, offsets) < 0) {
split_rec = page_get_middle_rec(page);
} else {
split_rec = NULL;
}
} else {
if (page_get_n_recs(page) > 1) {
split_rec = page_get_middle_rec(page);
} else if (btr_page_tuple_smaller(cursor, tuple,
offsets, n_uniq, &heap)) {
split_rec = page_rec_get_next(
page_get_infimum_rec(page));
} else {
split_rec = NULL;
insert_left = FALSE;
}
}
/* At this point, insert_left is initialized if split_rec == NULL
and may be uninitialized otherwise. */
/* 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);
@ -1978,11 +2005,11 @@ func_start:
avoid further splits by inserting the record
to an empty page. */
split_rec = NULL;
goto insert_right;
goto insert_empty;
}
} else {
insert_right:
insert_left = FALSE;
insert_empty:
ut_ad(!split_rec);
buf = mem_alloc(rec_get_converted_size(cursor->index,
tuple, n_ext));
@ -2019,7 +2046,17 @@ insert_right:
}
/* 5. Move then the records to the new page */
if (direction == FSP_DOWN) {
if (direction == FSP_DOWN
#ifdef UNIV_BTR_AVOID_COPY
&& page_rec_is_supremum(move_limit)) {
/* Instead of moving all records, make the new page
the empty page. */
left_block = block;
right_block = new_block;
} else if (direction == FSP_DOWN
#endif /* UNIV_BTR_AVOID_COPY */
) {
/* fputs("Split left\n", stderr); */
if (0
@ -2062,6 +2099,14 @@ insert_right:
right_block = block;
lock_update_split_left(right_block, left_block);
#ifdef UNIV_BTR_AVOID_COPY
} else if (!split_rec) {
/* Instead of moving all records, make the new page
the empty page. */
left_block = new_block;
right_block = block;
#endif /* UNIV_BTR_AVOID_COPY */
} else {
/* fputs("Split right\n", stderr); */