mariadb/storage/innobase/row/row0uins.cc
Marko Mäkelä a4948dafcd MDEV-11369 Instant ADD COLUMN for InnoDB
For InnoDB tables, adding, dropping and reordering columns has
required a rebuild of the table and all its indexes. Since MySQL 5.6
(and MariaDB 10.0) this has been supported online (LOCK=NONE), allowing
concurrent modification of the tables.

This work revises the InnoDB ROW_FORMAT=REDUNDANT, ROW_FORMAT=COMPACT
and ROW_FORMAT=DYNAMIC so that columns can be appended instantaneously,
with only minor changes performed to the table structure. The counter
innodb_instant_alter_column in INFORMATION_SCHEMA.GLOBAL_STATUS
is incremented whenever a table rebuild operation is converted into
an instant ADD COLUMN operation.

ROW_FORMAT=COMPRESSED tables will not support instant ADD COLUMN.

Some usability limitations will be addressed in subsequent work:

MDEV-13134 Introduce ALTER TABLE attributes ALGORITHM=NOCOPY
and ALGORITHM=INSTANT
MDEV-14016 Allow instant ADD COLUMN, ADD INDEX, LOCK=NONE

The format of the clustered index (PRIMARY KEY) is changed as follows:

(1) The FIL_PAGE_TYPE of the root page will be FIL_PAGE_TYPE_INSTANT,
and a new field PAGE_INSTANT will contain the original number of fields
in the clustered index ('core' fields).
If instant ADD COLUMN has not been used or the table becomes empty,
or the very first instant ADD COLUMN operation is rolled back,
the fields PAGE_INSTANT and FIL_PAGE_TYPE will be reset
to 0 and FIL_PAGE_INDEX.

(2) A special 'default row' record is inserted into the leftmost leaf,
between the page infimum and the first user record. This record is
distinguished by the REC_INFO_MIN_REC_FLAG, and it is otherwise in the
same format as records that contain values for the instantly added
columns. This 'default row' always has the same number of fields as
the clustered index according to the table definition. The values of
'core' fields are to be ignored. For other fields, the 'default row'
will contain the default values as they were during the ALTER TABLE
statement. (If the column default values are changed later, those
values will only be stored in the .frm file. The 'default row' will
contain the original evaluated values, which must be the same for
every row.) The 'default row' must be completely hidden from
higher-level access routines. Assertions have been added to ensure
that no 'default row' is ever present in the adaptive hash index
or in locked records. The 'default row' is never delete-marked.

(3) In clustered index leaf page records, the number of fields must
reside between the number of 'core' fields (dict_index_t::n_core_fields
introduced in this work) and dict_index_t::n_fields. If the number
of fields is less than dict_index_t::n_fields, the missing fields
are replaced with the column value of the 'default row'.
Note: The number of fields in the record may shrink if some of the
last instantly added columns are updated to the value that is
in the 'default row'. The function btr_cur_trim() implements this
'compression' on update and rollback; dtuple::trim() implements it
on insert.

(4) In ROW_FORMAT=COMPACT and ROW_FORMAT=DYNAMIC records, the new
status value REC_STATUS_COLUMNS_ADDED will indicate the presence of
a new record header that will encode n_fields-n_core_fields-1 in
1 or 2 bytes. (In ROW_FORMAT=REDUNDANT records, the record header
always explicitly encodes the number of fields.)

We introduce the undo log record type TRX_UNDO_INSERT_DEFAULT for
covering the insert of the 'default row' record when instant ADD COLUMN
is used for the first time. Subsequent instant ADD COLUMN can use
TRX_UNDO_UPD_EXIST_REC.

This is joint work with Vin Chen (陈福荣) from Tencent. The design
that was discussed in April 2017 would not have allowed import or
export of data files, because instead of the 'default row' it would
have introduced a data dictionary table. The test
rpl.rpl_alter_instant is exactly as contributed in pull request .
The test innodb.instant_alter is based on a contributed test.

The redo log record format changes for ROW_FORMAT=DYNAMIC and
ROW_FORMAT=COMPACT are as contributed. (With this change present,
crash recovery from MariaDB 10.3.1 will fail in spectacular ways!)
Also the semantics of higher-level redo log records that modify the
PAGE_INSTANT field is changed. The redo log format version identifier
was already changed to LOG_HEADER_FORMAT_CURRENT=103 in MariaDB 10.3.1.

Everything else has been rewritten by me. Thanks to Elena Stepanova,
the code has been tested extensively.

When rolling back an instant ADD COLUMN operation, we must empty the
PAGE_FREE list after deleting or shortening the 'default row' record,
by calling either btr_page_empty() or btr_page_reorganize(). We must
know the size of each entry in the PAGE_FREE list. If rollback left a
freed copy of the 'default row' in the PAGE_FREE list, we would be
unable to determine its size (if it is in ROW_FORMAT=COMPACT or
ROW_FORMAT=DYNAMIC) because it would contain more fields than the
rolled-back definition of the clustered index.

UNIV_SQL_DEFAULT: A new special constant that designates an instantly
added column that is not present in the clustered index record.

len_is_stored(): Check if a length is an actual length. There are
two magic length values: UNIV_SQL_DEFAULT, UNIV_SQL_NULL.

dict_col_t::def_val: The 'default row' value of the column.  If the
column is not added instantly, def_val.len will be UNIV_SQL_DEFAULT.

dict_col_t: Add the accessors is_virtual(), is_nullable(), is_instant(),
instant_value().

dict_col_t::remove_instant(): Remove the 'instant ADD' status of
a column.

dict_col_t::name(const dict_table_t& table): Replaces
dict_table_get_col_name().

dict_index_t::n_core_fields: The original number of fields.
For secondary indexes and if instant ADD COLUMN has not been used,
this will be equal to dict_index_t::n_fields.

dict_index_t::n_core_null_bytes: Number of bytes needed to
represent the null flags; usually equal to UT_BITS_IN_BYTES(n_nullable).

dict_index_t::NO_CORE_NULL_BYTES: Magic value signalling that
n_core_null_bytes was not initialized yet from the clustered index
root page.

dict_index_t: Add the accessors is_instant(), is_clust(),
get_n_nullable(), instant_field_value().

dict_index_t::instant_add_field(): Adjust clustered index metadata
for instant ADD COLUMN.

dict_index_t::remove_instant(): Remove the 'instant ADD' status
of a clustered index when the table becomes empty, or the very first
instant ADD COLUMN operation is rolled back.

dict_table_t: Add the accessors is_instant(), is_temporary(),
supports_instant().

dict_table_t::instant_add_column(): Adjust metadata for
instant ADD COLUMN.

dict_table_t::rollback_instant(): Adjust metadata on the rollback
of instant ADD COLUMN.

prepare_inplace_alter_table_dict(): First create the ctx->new_table,
and only then decide if the table really needs to be rebuilt.
We must split the creation of table or index metadata from the
creation of the dictionary table records and the creation of
the data. In this way, we can transform a table-rebuilding operation
into an instant ADD COLUMN operation. Dictionary objects will only
be added to cache when table rebuilding or index creation is needed.
The ctx->instant_table will never be added to cache.

dict_table_t::add_to_cache(): Modified and renamed from
dict_table_add_to_cache(). Do not modify the table metadata.
Let the callers invoke dict_table_add_system_columns() and if needed,
set can_be_evicted.

dict_create_sys_tables_tuple(), dict_create_table_step(): Omit the
system columns (which will now exist in the dict_table_t object
already at this point).

dict_create_table_step(): Expect the callers to invoke
dict_table_add_system_columns().

pars_create_table(): Before creating the table creation execution
graph, invoke dict_table_add_system_columns().

row_create_table_for_mysql(): Expect all callers to invoke
dict_table_add_system_columns().

create_index_dict(): Replaces row_merge_create_index_graph().

innodb_update_n_cols(): Renamed from innobase_update_n_virtual().
Call my_error() if an error occurs.

btr_cur_instant_init(), btr_cur_instant_init_low(),
btr_cur_instant_root_init():
Load additional metadata from the clustered index and set
dict_index_t::n_core_null_bytes. This is invoked
when table metadata is first loaded into the data dictionary.

dict_boot(): Initialize n_core_null_bytes for the four hard-coded
dictionary tables.

dict_create_index_step(): Initialize n_core_null_bytes. This is
executed as part of CREATE TABLE.

dict_index_build_internal_clust(): Initialize n_core_null_bytes to
NO_CORE_NULL_BYTES if table->supports_instant().

row_create_index_for_mysql(): Initialize n_core_null_bytes for
CREATE TEMPORARY TABLE.

commit_cache_norebuild(): Call the code to rename or enlarge columns
in the cache only if instant ADD COLUMN is not being used.
(Instant ADD COLUMN would copy all column metadata from
instant_table to old_table, including the names and lengths.)

PAGE_INSTANT: A new 13-bit field for storing dict_index_t::n_core_fields.
This is repurposing the 16-bit field PAGE_DIRECTION, of which only the
least significant 3 bits were used. The original byte containing
PAGE_DIRECTION will be accessible via the new constant PAGE_DIRECTION_B.

page_get_instant(), page_set_instant(): Accessors for the PAGE_INSTANT.

page_ptr_get_direction(), page_get_direction(),
page_ptr_set_direction(): Accessors for PAGE_DIRECTION.

page_direction_reset(): Reset PAGE_DIRECTION, PAGE_N_DIRECTION.

page_direction_increment(): Increment PAGE_N_DIRECTION
and set PAGE_DIRECTION.

rec_get_offsets(): Use the 'leaf' parameter for non-debug purposes,
and assume that heap_no is always set.
Initialize all dict_index_t::n_fields for ROW_FORMAT=REDUNDANT records,
even if the record contains fewer fields.

rec_offs_make_valid(): Add the parameter 'leaf'.

rec_copy_prefix_to_dtuple(): Assert that the tuple is only built
on the core fields. Instant ADD COLUMN only applies to the
clustered index, and we should never build a search key that has
more than the PRIMARY KEY and possibly DB_TRX_ID,DB_ROLL_PTR.
All these columns are always present.

dict_index_build_data_tuple(): Remove assertions that would be
duplicated in rec_copy_prefix_to_dtuple().

rec_init_offsets(): Support ROW_FORMAT=REDUNDANT records whose
number of fields is between n_core_fields and n_fields.

cmp_rec_rec_with_match(): Implement the comparison between two
MIN_REC_FLAG records.

trx_t::in_rollback: Make the field available in non-debug builds.

trx_start_for_ddl_low(): Remove dangerous error-tolerance.
A dictionary transaction must be flagged as such before it has generated
any undo log records. This is because trx_undo_assign_undo() will mark
the transaction as a dictionary transaction in the undo log header
right before the very first undo log record is being written.

btr_index_rec_validate(): Account for instant ADD COLUMN

row_undo_ins_remove_clust_rec(): On the rollback of an insert into
SYS_COLUMNS, revert instant ADD COLUMN in the cache by removing the
last column from the table and the clustered index.

row_search_on_row_ref(), row_undo_mod_parse_undo_rec(), row_undo_mod(),
trx_undo_update_rec_get_update(): Handle the 'default row'
as a special case.

dtuple_t::trim(index): Omit a redundant suffix of an index tuple right
before insert or update. After instant ADD COLUMN, if the last fields
of a clustered index tuple match the 'default row', there is no
need to store them. While trimming the entry, we must hold a page latch,
so that the table cannot be emptied and the 'default row' be deleted.

btr_cur_optimistic_update(), btr_cur_pessimistic_update(),
row_upd_clust_rec_by_insert(), row_ins_clust_index_entry_low():
Invoke dtuple_t::trim() if needed.

row_ins_clust_index_entry(): Restore dtuple_t::n_fields after calling
row_ins_clust_index_entry_low().

rec_get_converted_size(), rec_get_converted_size_comp(): Allow the number
of fields to be between n_core_fields and n_fields. Do not support
infimum,supremum. They are never supposed to be stored in dtuple_t,
because page creation nowadays uses a lower-level method for initializing
them.

rec_convert_dtuple_to_rec_comp(): Assign the status bits based on the
number of fields.

btr_cur_trim(): In an update, trim the index entry as needed. For the
'default row', handle rollback specially. For user records, omit
fields that match the 'default row'.

btr_cur_optimistic_delete_func(), btr_cur_pessimistic_delete():
Skip locking and adaptive hash index for the 'default row'.

row_log_table_apply_convert_mrec(): Replace 'default row' values if needed.
In the temporary file that is applied by row_log_table_apply(),
we must identify whether the records contain the extra header for
instantly added columns. For now, we will allocate an additional byte
for this for ROW_T_INSERT and ROW_T_UPDATE records when the source table
has been subject to instant ADD COLUMN. The ROW_T_DELETE records are
fine, as they will be converted and will only contain 'core' columns
(PRIMARY KEY and some system columns) that are converted from dtuple_t.

rec_get_converted_size_temp(), rec_init_offsets_temp(),
rec_convert_dtuple_to_temp(): Add the parameter 'status'.

REC_INFO_DEFAULT_ROW = REC_INFO_MIN_REC_FLAG | REC_STATUS_COLUMNS_ADDED:
An info_bits constant for distinguishing the 'default row' record.

rec_comp_status_t: An enum of the status bit values.

rec_leaf_format: An enum that replaces the bool parameter of
rec_init_offsets_comp_ordinary().
2017-10-06 09:50:10 +03:00

625 lines
17 KiB
C++

/*****************************************************************************
Copyright (c) 1997, 2017, Oracle and/or its affiliates. All Rights Reserved.
Copyright (c) 2017, MariaDB Corporation.
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
Foundation; version 2 of the License.
This program is distributed in the hope that it will be useful, but WITHOUT
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.,
51 Franklin Street, Suite 500, Boston, MA 02110-1335 USA
*****************************************************************************/
/**************************************************//**
@file row/row0uins.cc
Fresh insert undo
Created 2/25/1997 Heikki Tuuri
*******************************************************/
#include "row0uins.h"
#include "dict0dict.h"
#include "dict0stats.h"
#include "dict0boot.h"
#include "dict0crea.h"
#include "trx0undo.h"
#include "trx0roll.h"
#include "btr0btr.h"
#include "mach0data.h"
#include "row0undo.h"
#include "row0vers.h"
#include "row0log.h"
#include "trx0trx.h"
#include "trx0rec.h"
#include "row0row.h"
#include "row0upd.h"
#include "que0que.h"
#include "ibuf0ibuf.h"
#include "log0log.h"
#include "fil0fil.h"
/*************************************************************************
IMPORTANT NOTE: Any operation that generates redo MUST check that there
is enough space in the redo log before for that operation. This is
done by calling log_free_check(). The reason for checking the
availability of the redo log space before the start of the operation is
that we MUST not hold any synchonization objects when performing the
check.
If you make a change in this module make sure that no codepath is
introduced where a call to log_free_check() is bypassed. */
/***************************************************************//**
Removes a clustered index record. The pcur in node was positioned on the
record, now it is detached.
@return DB_SUCCESS or DB_OUT_OF_FILE_SPACE */
static MY_ATTRIBUTE((nonnull, warn_unused_result))
dberr_t
row_undo_ins_remove_clust_rec(
/*==========================*/
undo_node_t* node) /*!< in: undo node */
{
btr_cur_t* btr_cur;
ibool success;
dberr_t err;
ulint n_tries = 0;
mtr_t mtr;
dict_index_t* index = node->pcur.btr_cur.index;
bool online;
ut_ad(dict_index_is_clust(index));
ut_ad(node->trx->in_rollback);
mtr.start();
if (index->table->is_temporary()) {
ut_ad(node->rec_type == TRX_UNDO_INSERT_REC);
mtr.set_log_mode(MTR_LOG_NO_REDO);
} else {
mtr.set_named_space(index->space);
}
/* This is similar to row_undo_mod_clust(). The DDL thread may
already have copied this row from the log to the new table.
We must log the removal, so that the row will be correctly
purged. However, we can log the removal out of sync with the
B-tree modification. */
online = dict_index_is_online_ddl(index);
if (online) {
ut_ad(node->trx->dict_operation_lock_mode
!= RW_X_LATCH);
ut_ad(node->table->id != DICT_INDEXES_ID);
mtr_s_lock(dict_index_get_lock(index), &mtr);
}
success = btr_pcur_restore_position(
online
? BTR_MODIFY_LEAF | BTR_ALREADY_S_LATCHED
: BTR_MODIFY_LEAF, &node->pcur, &mtr);
ut_a(success);
btr_cur = btr_pcur_get_btr_cur(&node->pcur);
ut_ad(rec_get_trx_id(btr_cur_get_rec(btr_cur), btr_cur->index)
== node->trx->id);
ut_ad(!rec_get_deleted_flag(
btr_cur_get_rec(btr_cur),
dict_table_is_comp(btr_cur->index->table)));
if (online && dict_index_is_online_ddl(index)) {
const rec_t* rec = btr_cur_get_rec(btr_cur);
mem_heap_t* heap = NULL;
const ulint* offsets = rec_get_offsets(
rec, index, NULL, true, ULINT_UNDEFINED, &heap);
row_log_table_delete(rec, node->row, index, offsets, NULL);
mem_heap_free(heap);
}
switch (node->table->id) {
case DICT_INDEXES_ID:
ut_ad(!online);
ut_ad(node->trx->dict_operation_lock_mode == RW_X_LATCH);
ut_ad(node->rec_type == TRX_UNDO_INSERT_REC);
dict_drop_index_tree(
btr_pcur_get_rec(&node->pcur), &(node->pcur), &mtr);
mtr.commit();
mtr.start();
success = btr_pcur_restore_position(
BTR_MODIFY_LEAF, &node->pcur, &mtr);
ut_a(success);
break;
case DICT_COLUMNS_ID:
/* This is rolling back an INSERT into SYS_COLUMNS.
If it was part of an instant ADD COLUMN operation, we
must modify the table definition. At this point, any
corresponding operation to the 'default row' will have
been rolled back. */
ut_ad(!online);
ut_ad(node->trx->dict_operation_lock_mode == RW_X_LATCH);
ut_ad(node->rec_type == TRX_UNDO_INSERT_REC);
const rec_t* rec = btr_pcur_get_rec(&node->pcur);
if (rec_get_n_fields_old(rec)
!= DICT_NUM_FIELDS__SYS_COLUMNS) {
break;
}
ulint len;
const byte* data = rec_get_nth_field_old(
rec, DICT_FLD__SYS_COLUMNS__TABLE_ID, &len);
if (len != 8) {
break;
}
const table_id_t table_id = mach_read_from_8(data);
data = rec_get_nth_field_old(rec, DICT_FLD__SYS_COLUMNS__POS,
&len);
if (len != 4) {
break;
}
const unsigned pos = mach_read_from_4(data);
if (pos == 0 || pos >= (1U << 16)) {
break;
}
dict_table_t* table = dict_table_open_on_id(
table_id, true, DICT_TABLE_OP_OPEN_ONLY_IF_CACHED);
if (!table) {
break;
}
dict_index_t* index = dict_table_get_first_index(table);
if (index && index->is_instant()
&& DATA_N_SYS_COLS + 1 + pos == table->n_cols) {
/* This is the rollback of an instant ADD COLUMN.
Remove the column from the dictionary cache,
but keep the system columns. */
table->rollback_instant(pos);
}
dict_table_close(table, true, false);
}
if (btr_cur_optimistic_delete(btr_cur, 0, &mtr)) {
err = DB_SUCCESS;
goto func_exit;
}
btr_pcur_commit_specify_mtr(&node->pcur, &mtr);
retry:
/* If did not succeed, try pessimistic descent to tree */
mtr.start();
if (index->table->is_temporary()) {
mtr.set_log_mode(MTR_LOG_NO_REDO);
} else {
mtr.set_named_space(index->space);
}
success = btr_pcur_restore_position(
BTR_MODIFY_TREE | BTR_LATCH_FOR_DELETE,
&node->pcur, &mtr);
ut_a(success);
btr_cur_pessimistic_delete(&err, FALSE, btr_cur, 0, true, &mtr);
/* The delete operation may fail if we have little
file space left: TODO: easiest to crash the database
and restart with more file space */
if (err == DB_OUT_OF_FILE_SPACE
&& n_tries < BTR_CUR_RETRY_DELETE_N_TIMES) {
btr_pcur_commit_specify_mtr(&(node->pcur), &mtr);
n_tries++;
os_thread_sleep(BTR_CUR_RETRY_SLEEP_TIME);
goto retry;
}
func_exit:
btr_pcur_commit_specify_mtr(&node->pcur, &mtr);
if (err == DB_SUCCESS && node->rec_type == TRX_UNDO_INSERT_DEFAULT) {
/* When rolling back the very first instant ADD COLUMN
operation, reset the root page to the basic state. */
ut_ad(!index->table->is_temporary());
mtr.start();
if (page_t* root = btr_root_get(index, &mtr)) {
byte* page_type = root + FIL_PAGE_TYPE;
ut_ad(mach_read_from_2(page_type)
== FIL_PAGE_TYPE_INSTANT
|| mach_read_from_2(page_type)
== FIL_PAGE_INDEX);
mtr.set_named_space(index->space);
mlog_write_ulint(page_type, FIL_PAGE_INDEX,
MLOG_2BYTES, &mtr);
byte* instant = PAGE_INSTANT + PAGE_HEADER + root;
mlog_write_ulint(instant,
page_ptr_get_direction(instant + 1),
MLOG_2BYTES, &mtr);
}
mtr.commit();
}
return(err);
}
/***************************************************************//**
Removes a secondary index entry if found.
@return DB_SUCCESS, DB_FAIL, or DB_OUT_OF_FILE_SPACE */
static MY_ATTRIBUTE((nonnull, warn_unused_result))
dberr_t
row_undo_ins_remove_sec_low(
/*========================*/
ulint mode, /*!< in: BTR_MODIFY_LEAF or BTR_MODIFY_TREE,
depending on whether we wish optimistic or
pessimistic descent down the index tree */
dict_index_t* index, /*!< in: index */
dtuple_t* entry, /*!< in: index entry to remove */
que_thr_t* thr) /*!< in: query thread */
{
btr_pcur_t pcur;
btr_cur_t* btr_cur;
dberr_t err = DB_SUCCESS;
mtr_t mtr;
enum row_search_result search_result;
const bool modify_leaf = mode == BTR_MODIFY_LEAF;
memset(&pcur, 0, sizeof(pcur));
row_mtr_start(&mtr, index, !modify_leaf);
if (modify_leaf) {
mode = BTR_MODIFY_LEAF | BTR_ALREADY_S_LATCHED;
mtr_s_lock(dict_index_get_lock(index), &mtr);
} else {
ut_ad(mode == (BTR_MODIFY_TREE | BTR_LATCH_FOR_DELETE));
mtr_sx_lock(dict_index_get_lock(index), &mtr);
}
if (row_log_online_op_try(index, entry, 0)) {
goto func_exit_no_pcur;
}
if (dict_index_is_spatial(index)) {
if (modify_leaf) {
mode |= BTR_RTREE_DELETE_MARK;
}
btr_pcur_get_btr_cur(&pcur)->thr = thr;
mode |= BTR_RTREE_UNDO_INS;
}
search_result = row_search_index_entry(index, entry, mode,
&pcur, &mtr);
switch (search_result) {
case ROW_NOT_FOUND:
goto func_exit;
case ROW_FOUND:
if (dict_index_is_spatial(index)
&& rec_get_deleted_flag(
btr_pcur_get_rec(&pcur),
dict_table_is_comp(index->table))) {
ib::error() << "Record found in index " << index->name
<< " is deleted marked on insert rollback.";
ut_ad(0);
}
break;
case ROW_BUFFERED:
case ROW_NOT_DELETED_REF:
/* These are invalid outcomes, because the mode passed
to row_search_index_entry() did not include any of the
flags BTR_INSERT, BTR_DELETE, or BTR_DELETE_MARK. */
ut_error;
}
btr_cur = btr_pcur_get_btr_cur(&pcur);
if (modify_leaf) {
err = btr_cur_optimistic_delete(btr_cur, 0, &mtr)
? DB_SUCCESS : DB_FAIL;
} else {
/* Passing rollback=false here, because we are
deleting a secondary index record: the distinction
only matters when deleting a record that contains
externally stored columns. */
ut_ad(!dict_index_is_clust(index));
btr_cur_pessimistic_delete(&err, FALSE, btr_cur, 0,
false, &mtr);
}
func_exit:
btr_pcur_close(&pcur);
func_exit_no_pcur:
mtr_commit(&mtr);
return(err);
}
/***************************************************************//**
Removes a secondary index entry from the index if found. Tries first
optimistic, then pessimistic descent down the tree.
@return DB_SUCCESS or DB_OUT_OF_FILE_SPACE */
static MY_ATTRIBUTE((nonnull, warn_unused_result))
dberr_t
row_undo_ins_remove_sec(
/*====================*/
dict_index_t* index, /*!< in: index */
dtuple_t* entry, /*!< in: index entry to insert */
que_thr_t* thr) /*!< in: query thread */
{
dberr_t err;
ulint n_tries = 0;
/* Try first optimistic descent to the B-tree */
err = row_undo_ins_remove_sec_low(BTR_MODIFY_LEAF, index, entry, thr);
if (err == DB_SUCCESS) {
return(err);
}
/* Try then pessimistic descent to the B-tree */
retry:
err = row_undo_ins_remove_sec_low(
BTR_MODIFY_TREE | BTR_LATCH_FOR_DELETE,
index, entry, thr);
/* The delete operation may fail if we have little
file space left: TODO: easiest to crash the database
and restart with more file space */
if (err != DB_SUCCESS && n_tries < BTR_CUR_RETRY_DELETE_N_TIMES) {
n_tries++;
os_thread_sleep(BTR_CUR_RETRY_SLEEP_TIME);
goto retry;
}
return(err);
}
/***********************************************************//**
Parses the row reference and other info in a fresh insert undo record. */
static
void
row_undo_ins_parse_undo_rec(
/*========================*/
undo_node_t* node, /*!< in/out: row undo node */
ibool dict_locked) /*!< in: TRUE if own dict_sys->mutex */
{
dict_index_t* clust_index;
byte* ptr;
undo_no_t undo_no;
table_id_t table_id;
ulint type;
ulint dummy;
bool dummy_extern;
ut_ad(node);
ptr = trx_undo_rec_get_pars(node->undo_rec, &type, &dummy,
&dummy_extern, &undo_no, &table_id);
ut_ad(type == TRX_UNDO_INSERT_REC || type == TRX_UNDO_INSERT_DEFAULT);
node->rec_type = type;
node->update = NULL;
node->table = dict_table_open_on_id(
table_id, dict_locked, DICT_TABLE_OP_NORMAL);
/* Skip the UNDO if we can't find the table or the .ibd file. */
if (UNIV_UNLIKELY(node->table == NULL)) {
return;
}
if (UNIV_UNLIKELY(!fil_table_accessible(node->table))) {
close_table:
/* Normally, tables should not disappear or become
unaccessible during ROLLBACK, because they should be
protected by InnoDB table locks. TRUNCATE TABLE
or table corruption could be valid exceptions.
FIXME: When running out of temporary tablespace, it
would probably be better to just drop all temporary
tables (and temporary undo log records) of the current
connection, instead of doing this rollback. */
dict_table_close(node->table, dict_locked, FALSE);
node->table = NULL;
} else {
clust_index = dict_table_get_first_index(node->table);
if (clust_index != NULL) {
if (type == TRX_UNDO_INSERT_REC) {
ptr = trx_undo_rec_get_row_ref(
ptr, clust_index, &node->ref,
node->heap);
} else {
ut_ad(type == TRX_UNDO_INSERT_DEFAULT);
node->ref = &trx_undo_default_rec;
}
if (!row_undo_search_clust_to_pcur(node)) {
/* An error probably occurred during
an insert into the clustered index,
after we wrote the undo log record. */
goto close_table;
}
if (node->table->n_v_cols) {
trx_undo_read_v_cols(node->table, ptr,
node->row, false, NULL);
}
} else {
ib::warn() << "Table " << node->table->name
<< " has no indexes,"
" ignoring the table";
goto close_table;
}
}
}
/***************************************************************//**
Removes secondary index records.
@return DB_SUCCESS or DB_OUT_OF_FILE_SPACE */
static MY_ATTRIBUTE((nonnull, warn_unused_result))
dberr_t
row_undo_ins_remove_sec_rec(
/*========================*/
undo_node_t* node, /*!< in/out: row undo node */
que_thr_t* thr) /*!< in: query thread */
{
dberr_t err = DB_SUCCESS;
dict_index_t* index = node->index;
mem_heap_t* heap;
heap = mem_heap_create(1024);
while (index != NULL) {
dtuple_t* entry;
if (index->type & DICT_FTS) {
dict_table_next_uncorrupted_index(index);
continue;
}
/* An insert undo record TRX_UNDO_INSERT_REC will
always contain all fields of the index. It does not
matter if any indexes were created afterwards; all
index entries can be reconstructed from the row. */
entry = row_build_index_entry(
node->row, node->ext, index, heap);
if (UNIV_UNLIKELY(!entry)) {
/* The database must have crashed after
inserting a clustered index record but before
writing all the externally stored columns of
that record, or a statement is being rolled
back because an error occurred while storing
off-page columns.
Because secondary index entries are inserted
after the clustered index record, we may
assume that the secondary index record does
not exist. */
} else {
err = row_undo_ins_remove_sec(index, entry, thr);
if (UNIV_UNLIKELY(err != DB_SUCCESS)) {
goto func_exit;
}
}
mem_heap_empty(heap);
dict_table_next_uncorrupted_index(index);
}
func_exit:
node->index = index;
mem_heap_free(heap);
return(err);
}
/***********************************************************//**
Undoes a fresh insert of a row to a table. A fresh insert means that
the same clustered index unique key did not have any record, even delete
marked, at the time of the insert. InnoDB is eager in a rollback:
if it figures out that an index record will be removed in the purge
anyway, it will remove it in the rollback.
@return DB_SUCCESS or DB_OUT_OF_FILE_SPACE */
dberr_t
row_undo_ins(
/*=========*/
undo_node_t* node, /*!< in: row undo node */
que_thr_t* thr) /*!< in: query thread */
{
dberr_t err;
ibool dict_locked;
ut_ad(node->state == UNDO_NODE_INSERT);
ut_ad(node->trx->in_rollback);
ut_ad(trx_undo_roll_ptr_is_insert(node->roll_ptr));
dict_locked = node->trx->dict_operation_lock_mode == RW_X_LATCH;
row_undo_ins_parse_undo_rec(node, dict_locked);
if (node->table == NULL) {
return(DB_SUCCESS);
}
/* Iterate over all the indexes and undo the insert.*/
node->index = dict_table_get_first_index(node->table);
ut_ad(dict_index_is_clust(node->index));
switch (node->rec_type) {
default:
ut_ad(!"wrong undo record type");
case TRX_UNDO_INSERT_REC:
/* Skip the clustered index (the first index) */
node->index = dict_table_get_next_index(node->index);
dict_table_skip_corrupt_index(node->index);
err = row_undo_ins_remove_sec_rec(node, thr);
if (err != DB_SUCCESS) {
break;
}
/* fall through */
case TRX_UNDO_INSERT_DEFAULT:
log_free_check();
if (node->table->id == DICT_INDEXES_ID) {
ut_ad(node->rec_type == TRX_UNDO_INSERT_REC);
if (!dict_locked) {
mutex_enter(&dict_sys->mutex);
}
}
// FIXME: We need to update the dict_index_t::space and
// page number fields too.
err = row_undo_ins_remove_clust_rec(node);
if (node->table->id == DICT_INDEXES_ID
&& !dict_locked) {
mutex_exit(&dict_sys->mutex);
}
if (err == DB_SUCCESS && node->table->stat_initialized) {
/* Not protected by dict_table_stats_lock() for
performance reasons, we would rather get garbage
in stat_n_rows (which is just an estimate anyway)
than protecting the following code with a latch. */
dict_table_n_rows_dec(node->table);
/* Do not attempt to update statistics when
executing ROLLBACK in the InnoDB SQL
interpreter, because in that case we would
already be holding dict_sys->mutex, which
would be acquired when updating statistics. */
if (!dict_locked) {
dict_stats_update_if_needed(node->table);
}
}
}
dict_table_close(node->table, dict_locked, FALSE);
node->table = NULL;
return(err);
}