database/mariadb
1
0
Fork 0
mirror of https://github.com/MariaDB/server.git synced 2026-10-23 18:49:29 +03:00
Code Activity
Files
c37016b0e5881a0cfa17ab4c640dad44c4c63f9d
mariadb/sql/sql_delete.cc
Arcadiy Ivanov f1a72e2cca MDEV-40376 Protocol::end_statement assertion when window function fills HEAP tmp table 
`save_window_function_values()` did not take into account that one can
get `HA_ERR_RECORD_FILE_FULL` on `ha_update_row()`. With blob support in
heap, it can now happen more easily (expressions wider than 512
characters are promoted to TEXT in tmp tables, and each update of a blob
result column allocates a continuation record). However, it could also
happen with Aria tables, which is apparently not tested. The error was
returned as a plain `true` with an empty diagnostics area: debug builds
hit `Assertion '0'` in `Protocol::end_statement()`, release builds send
a bare OK packet after the result set metadata, which the client
mis-parses (appears as a hang / lost connection).

Errors of the computation are now exposed:
`save_window_function_values()` calls `handler::print_error()` for any
failure it does not recover from, the previously ignored `ha_rnd_pos()`
return values are checked and reported, and `compute_window_func()`
distinguishes a read error from EOF and stops the row scan as soon as
the per-row loop fails.

The overflow itself is recovered from by converting the tmp table to
the disk-based tmp engine in place: the rows keep their identity and
their contents, only their positions change, and those are translated
in the rowid sequence that the computation is built on.

The filesort result of a window sort is a sequence of row positions
that covers every row of the table exactly once (the sort is set up
without a limit) and is the only place where the positions are kept:
the row scan and all frame cursors read the rows through it. The
conversion (`Window_rowid_remapper`) therefore copies the rows in the
order of that sequence, which makes the new position of a row known as
soon as the row has been written, and stores it back into the very slot
the old position was read from.

For the new position to always fit into the slot, a window sort stores
the row positions in slots of a fixed 8 bytes
(`Filesort::min_ref_length`, set to the new `TMP_TABLE_MAX_REF_LENGTH`),
which holds the position of any engine an internal tmp table can use: a
pointer into memory or a data file offset. `make_sortkey()` zero-pads a
shorter position to its slot, and `SORT_INFO::ref_length` carries the
slot width to the readers, so `init_read_record()` no longer re-derives
it from `handler::ref_length`, which changes when the table is
converted. Sorts that do not set a minimum, and engines whose positions
are wider than 8 bytes, are unchanged byte for byte. A sequence held in
memory is thereby rewritten in place, keeping its layout, so the frame
cursors' places in it stay valid.

A sequence held in a temporary file is written and read through an
`IO_CACHE`, which encrypts the file when tmp file encryption is
enabled, so the file can not be rewritten in place: the old sequence is
streamed out of its cache and the translated sequence into a new cached
temporary file, which then replaces the old one under the cache. The
frame cursors' slave caches stay linked to the master through
`next_file_user` across the replacement, and are re-created from the
new master afterwards (`Frame_cursor::rowids_rewritten()`). Should
re-creating one fail, the cursor forgets its already-released cache, so
that it is not released a second time when the cursor is destroyed, and
reports the failure.

The row whose update did not fit is written from `record[0]`, which
holds its new image, instead of being read from the table, so the
conversion applies the update that failed. Blob values of `record[0]`
that were read from the HEAP table can point into the handler's shared
blob reassembly buffer (`hp_read_blobs()`, blob values that span
multiple allocation blocks), which reading the other rows overwrites,
so they are first given memory of their own.

When the window sort was set up with a deferred filter (a HAVING clause
deferred to the final ORDER BY sort), the sequence omits the rows the
filter rejected and the conversion drops them: every later reader of
the table applies the same filter, either directly or by reading the
table through a sort that does, so such rows can never reach the
result.

`create_internal_tmp_table_from_heap()` gains a `Tmp_table_row_copier`
hook for the copy from the HEAP table to the Aria or MyISAM table that
replaces it. The plain copy loop and the append of the pending
`record[0]` become `Tmp_table_default_copier`, used when the caller
passes no copier, so both cases of the conversion run through the same
code, defined together with `Window_rowid_remapper` just before the
function. `ha_end_bulk_insert()` is called also when the copy fails, so
the new handler does not keep bulk insert state when its table is
dropped; the pending row is thereby written while the bulk insert is
still active, which does not delay its duplicate detection, as bulk
insert does not cache unique keys (`maria_init_bulk_insert()` skips
`HA_NOSAME` keys). An ignored duplicate of the pending row is passed to
the caller in `Tmp_table_row_copier::duplicate_row_error`, which sets
`*is_duplicate`.

Carrying the running computation over the conversion, instead of
re-running it, has two visible consequences:

 - Compound expressions containing window functions (`items_to_copy`)
   are evaluated exactly once per row, as they are when the table does
   not overflow, so the conversion does not have to be refused when
   such an expression is non-deterministic (`RAND_TABLE_BIT`: a user
   variable assignment, a non-deterministic stored function); those
   statements produce their result.
 - The statement sorts once, so sort related aggregate warnings, e.g.
   the `max_sort_length` truncation counts, are reported for one pass.

The rows are copied in the order of the sequence, so the converted
table does not hold them in the order they were inserted in. This is
safe even when that order is what the table was built for
(`TABLE::keep_row_order`, set for `ROWNUM()` with GROUP BY or ORDER
BY): before HEAP supported blobs, such a tmp table was created in the
disk engine from the start and these statements simply ran, so refusing
here would be a regression against that behavior.

1. `ROWNUM()` values are materialized into the tmp table rows during
   the fill, before the window computation begins; the copy moves them
   verbatim, so their pairing with the rows can not change.
2. The rowid sequence is translated in place, so the running window
   computation continues over exactly the same row order, and
   tie-sensitive window function values (`ROW_NUMBER`, frames over tied
   keys) keep the values the interrupted pass had already produced.
3. Consumers that scan the converted table afterwards see the rows in
   the copy order. Among rows with equal sort keys that order differs
   from the insertion order, but such tie order is unspecified and
   already differs between the memory and disk tmp engines.

The converted table keeps `keep_row_order= true` so the copy order is
also the order the disk engine preserves from then on.

Tests, in `heap.blob_window_overflow` unless noted: a single window
over all rows and a partitioned window overflow mid-computation and
must transparently convert; the side-effecting window expression test
verifies that the user variable is assigned exactly once per row,
against the same query that does not overflow; a fourth test covers the
sequence being an array in memory instead of a merge file; a fifth test
covers the `keep_row_order` conversion, verifying that every `ROWNUM()`
value keeps its insertion-order pairing with its row across the
conversion; a sixth test covers blob values larger than a HEAP
allocation block surviving the conversion of the pending row;
`heap.blob_window_overflow_encrypt` runs the conversion with encrypted
temporary files; `heap.blob_window_overflow_debug` covers the failure
to re-create a slave cache of the sequence
(`simulate_window_seq_slave_oom`), which previously hung the server in
the cursor destructor.
2026-10-11 18:42:16 +03:00

2258 lines
67 KiB
C++
Raw Blame History

/*
Copyright (c) 2000, 2019, Oracle and/or its affiliates.
Copyright (c) 2010, 2022, MariaDB
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 St, Fifth Floor, Boston, MA 02110-1335 USA */
/*
Delete of records tables.
Multi-table deletes were introduced by Monty and Sinisa
*/
#include "mariadb.h"
#include "sql_priv.h"
#include "unireg.h"
#include "sql_delete.h"
#include "sql_cache.h" // query_cache_*
#include "sql_base.h" // open_temprary_table
#include "lock.h" // unlock_table_name
#include "sql_view.h" // check_key_in_view, mysql_frm_type
#include "sql_parse.h" // mysql_init_select
#include "filesort.h" // filesort
#include "sql_handler.h" // mysql_ha_rm_tables
#include "sql_select.h"
#include "sp_head.h"
#include "sql_trigger.h"
#include "sql_statistics.h"
#include "transaction.h"
#include "records.h" // init_read_record,
#include "filesort.h"
#include "uniques.h"
#include "sql_derived.h" // mysql_handle_derived
#include "key.h"
// end_read_record
#include "sql_insert.h" // fix_rownum_pointers
#include "sql_partition.h" // make_used_partitions_str
#ifdef WITH_WSREP
#include "wsrep_mysqld.h"
#endif
#define MEM_STRIP_BUF_SIZE ((size_t) thd->variables.sortbuff_size)
/*
@brief
Print query plan of a single-table DELETE command
@detail
This function is used by EXPLAIN DELETE and by SHOW EXPLAIN when it is
invoked on a running DELETE statement.
*/
Explain_delete* Delete_plan::save_explain_delete_data(THD *thd, MEM_ROOT *mem_root)
{
Explain_query *query= thd->lex->explain;
Explain_delete *explain=
new (mem_root) Explain_delete(mem_root, thd->lex->analyze_stmt);
if (!explain)
return 0;
if (deleting_all_rows)
{
explain->deleting_all_rows= true;
explain->select_type= "SIMPLE";
explain->rows= scanned_rows;
}
else
{
explain->deleting_all_rows= false;
if (Update_plan::save_explain_data_intern(thd, mem_root, explain,
thd->lex->analyze_stmt))
return 0;
}
query->add_upd_del_plan(explain);
return explain;
}
Explain_update*
Update_plan::save_explain_update_data(THD *thd, MEM_ROOT *mem_root)
{
Explain_query *query= thd->lex->explain;
Explain_update* explain=
new (mem_root) Explain_update(mem_root, thd->lex->analyze_stmt);
if (!explain)
return 0;
if (save_explain_data_intern(thd, mem_root, explain, thd->lex->analyze_stmt))
return 0;
query->add_upd_del_plan(explain);
return explain;
}
bool Update_plan::save_explain_data_intern(THD *thd,
MEM_ROOT *mem_root,
Explain_update *explain,
bool is_analyze)
{
explain->select_type= "SIMPLE";
explain->table_name.append(table->alias);
explain->impossible_where= false;
explain->no_partitions= false;
if (impossible_where)
{
explain->impossible_where= true;
return 0;
}
if (no_partitions)
{
explain->no_partitions= true;
return 0;
}
if (is_analyze ||
(thd->variables.log_slow_verbosity &
LOG_SLOW_VERBOSITY_ENGINE))
{
explain->table_tracker.set_gap_tracker(&explain->extra_time_tracker);
table->file->set_time_tracker(&explain->table_tracker);
if (table->file->handler_stats &&
table->s->tmp_table != RESULT_TMP_TABLE &&
table->s->tmp_table != INTERNAL_TMP_TABLE)
explain->handler_for_stats= table->file;
}
select_lex->set_explain_type(TRUE);
explain->select_type= select_lex->type;
/* Partitions */
{
#ifdef WITH_PARTITION_STORAGE_ENGINE
partition_info *part_info;
if ((part_info= table->part_info))
{
make_used_partitions_str(mem_root, part_info, &explain->used_partitions,
explain->used_partitions_list);
explain->used_partitions_set= true;
}
else
explain->used_partitions_set= false;
#else
/* just produce empty column if partitioning is not compiled in */
explain->used_partitions_set= false;
#endif
}
/* Set jtype */
if (select && select->quick)
{
if (is_index_merge(select->quick->get_type()))
explain->jtype= JT_INDEX_MERGE;
else
explain->jtype= JT_RANGE;
}
else
{
if (index == MAX_KEY)
explain->jtype= JT_ALL;
else
explain->jtype= JT_NEXT;
}
explain->using_where= MY_TEST(select && select->cond);
explain->where_cond= select? select->cond: NULL;
if (using_filesort)
if (!(explain->filesort_tracker= new (mem_root) Filesort_tracker(is_analyze)))
return 1;
explain->using_io_buffer= using_io_buffer;
append_possible_keys(mem_root, explain->possible_keys, table,
possible_keys);
explain->quick_info= NULL;
/* Calculate key_len */
if (select && select->quick)
{
explain->quick_info= select->quick->get_explain(mem_root);
}
else
{
if (index != MAX_KEY)
{
explain->key.set(mem_root, &table->key_info[index],
table->key_info[index].key_length);
}
}
explain->rows= scanned_rows;
if (select && select->quick &&
select->quick->get_type() == QUICK_SELECT_I::QS_TYPE_RANGE)
{
explain_append_mrr_info((QUICK_RANGE_SELECT*)select->quick,
&explain->mrr_type);
}
/* Save subquery children */
for (SELECT_LEX_UNIT *unit= select_lex->first_inner_unit();
unit;
unit= unit->next_unit())
{
if (unit->explainable())
explain->add_child(unit->first_select()->select_number);
}
return 0;
}
static bool record_should_be_deleted(THD *thd, TABLE *table, SQL_SELECT *sel,
Explain_delete *explain, bool truncate_history)
{
explain->tracker.on_record_read();
thd->inc_examined_row_count();
if (table->vfield)
(void) table->update_virtual_fields(table->file, VCOL_UPDATE_FOR_DELETE);
if (!sel || sel->skip_record(thd) > 0)
{
explain->tracker.on_record_after_where();
return true;
}
return false;
}
static
int update_portion_of_time(THD *thd, TABLE *table,
const vers_select_conds_t &period_conds,
bool *inside_period)
{
bool lcond= period_conds.field_start->val_datetime_packed(thd)
< period_conds.start.item->val_datetime_packed(thd);
bool rcond= period_conds.field_end->val_datetime_packed(thd)
> period_conds.end.item->val_datetime_packed(thd);
*inside_period= !lcond && !rcond;
if (*inside_period)
return 0;
DBUG_ASSERT(!table->triggers
|| !table->triggers->has_triggers(TRG_EVENT_INSERT,
TRG_ACTION_BEFORE));
int res= 0;
Item *src= lcond ? period_conds.start.item : period_conds.end.item;
uint dst_fieldno= lcond ? table->s->period.end_fieldno
: table->s->period.start_fieldno;
ulonglong prev_insert_id= table->file->next_insert_id;
store_record(table, record[1]);
if (likely(!res))
res= src->save_in_field(table->field[dst_fieldno], true);
if (likely(!res))
{
table->period_prepare_autoinc();
res= table->update_generated_fields();
}
if(likely(!res))
res= table->file->ha_update_row(table->record[1], table->record[0]);
if (likely(!res) && table->triggers)
res= table->triggers->process_triggers(thd, TRG_EVENT_INSERT,
TRG_ACTION_AFTER, true, nullptr);
restore_record(table, record[1]);
if (res)
table->file->restore_auto_increment(prev_insert_id);
if (likely(!res) && lcond && rcond)
res= table->period_make_insert(period_conds.end.item,
table->field[table->s->period.start_fieldno]);
return res;
}
/**
Delete a record stored in:
replace= true: record[0]
replace= false: record[1]
with regard to the treat_versioned flag, which can be false for a versioned
table in case of versioned->versioned replication.
For a versioned case, we detect a few conditions, under which we should delete
a row instead of updating it to a history row.
This includes:
* History deletion by user;
* History collision, in case of REPLACE or very fast sequence of dmls
so that timestamp doesn't change;
* History collision in the parent table
A normal delete is processed here as well.
*/
template <bool replace>
int TABLE::delete_row(bool treat_versioned)
{
int err= 0;
bool remembered_pos= false;
uchar *del_buf= record[replace ? 1 : 0];
bool delete_row= !treat_versioned
|| in_use->lex->vers_conditions.delete_history
|| versioned(VERS_TRX_ID)
|| !vers_end_field()->is_max(
vers_end_field()->ptr_in_record(del_buf));
if (!delete_row)
{
if ((err= file->extra(HA_EXTRA_REMEMBER_POS)))
return err;
remembered_pos= true;
if (replace)
{
store_record(this, record[2]);
restore_record(this, record[1]);
}
else
{
store_record(this, record[1]);
}
vers_update_end();
Field *row_start= vers_start_field();
Field *row_end= vers_end_field();
// Don't make history row with negative lifetime
delete_row= row_start->cmp(row_start->ptr, row_end->ptr) > 0;
if (likely(!delete_row))
err= file->ha_update_row(record[1], record[0]);
if (unlikely(err))
{
/*
MDEV-23644: we get HA_ERR_FOREIGN_DUPLICATE_KEY iff we already got
history row with same trx_id which is the result of foreign key
action, so we don't need one more history row.
Additionally, delete the row if versioned record already exists.
This happens on replace, a very fast sequence of inserts and deletes,
or if timestamp is frozen.
*/
delete_row= err == HA_ERR_FOUND_DUPP_KEY
|| err == HA_ERR_FOUND_DUPP_UNIQUE
|| err == HA_ERR_FOREIGN_DUPLICATE_KEY;
if (!delete_row)
return err;
}
if (delete_row)
del_buf= record[1];
if (replace)
restore_record(this, record[2]);
}
if (delete_row)
err= file->ha_delete_row(del_buf);
if (remembered_pos)
(void) file->extra(HA_EXTRA_RESTORE_POS);
return err;
}
template int TABLE::delete_row<true>(bool treat_versioned);
template int TABLE::delete_row<false>(bool treat_versioned);
/**
@brief Special handling of single-table deletes after prepare phase
@param thd global context the processed statement
@returns false on success, true on error
*/
bool Sql_cmd_delete::delete_from_single_table(THD *thd)
{
int error;
int loc_error;
bool transactional_table;
bool const_cond;
bool safe_update;
bool const_cond_result;
bool return_error= 0;
bool binlogged= 0;
TABLE *table;
SQL_SELECT *select= 0;
SORT_INFO *file_sort= 0;
READ_RECORD info;
bool reverse= FALSE;
bool binlog_is_row;
killed_state killed_status= NOT_KILLED;
THD::enum_binlog_query_type query_type= THD::ROW_QUERY_TYPE;
bool will_batch= FALSE;
bool has_triggers= false;
SELECT_LEX_UNIT *unit = &lex->unit;
SELECT_LEX *select_lex= unit->first_select();
SELECT_LEX *returning= thd->lex->has_returning() ? thd->lex->returning() : 0;
TABLE_LIST *const table_list = select_lex->get_table_list();
ulonglong options= select_lex->options;
ORDER *order= select_lex->order_list.first;
COND *conds= select_lex->join->conds;
ha_rows limit= unit->lim.get_select_limit();
bool using_limit= limit != HA_POS_ERROR;
Delete_plan query_plan(thd->mem_root);
Explain_delete *explain;
Unique * deltempfile= NULL;
bool delete_record= false;
bool delete_while_scanning= table_list->delete_while_scanning;
bool portion_of_time_through_update;
/*
TRUE if we are after the call to
select_lex->optimize_unflattened_subqueries(true) and before the
call to select_lex->optimize_unflattened_subqueries(false), to
ensure a call to
select_lex->optimize_unflattened_subqueries(false) happens which
avoid 2nd ps mem leaks when e.g. the first execution produces
empty result and the second execution produces a non-empty set
*/
bool optimize_subqueries= FALSE;
DBUG_ENTER("Sql_cmd_delete::delete_single_table");
query_plan.index= MAX_KEY;
query_plan.using_filesort= FALSE;
THD_STAGE_INFO(thd, stage_init_update);
const bool delete_history= table_list->vers_conditions.delete_history;
DBUG_ASSERT(!(delete_history && table_list->period_conditions.is_set()));
if (table_list->handle_derived(thd->lex, DT_MERGE_FOR_INSERT))
DBUG_RETURN(1);
if (table_list->handle_derived(thd->lex, DT_PREPARE))
DBUG_RETURN(1);
table= table_list->table;
if (!table_list->single_table_updatable())
{
my_error(ER_NON_UPDATABLE_TABLE, MYF(0), table_list->alias.str, "DELETE");
DBUG_RETURN(TRUE);
}
if (!table || !table->is_created())
{
my_error(ER_VIEW_DELETE_MERGE_VIEW, MYF(0),
table_list->view_db.str, table_list->view_name.str);
DBUG_RETURN(TRUE);
}
query_plan.select_lex= thd->lex->first_select_lex();
query_plan.table= table;
thd->lex->promote_select_describe_flag_if_needed();
/*
Apply the IN=>EXISTS transformation to all constant subqueries
and optimize them.
Constant subqueries are treated in a special way here: they can be
evaluated even in EXPLAIN statement, so their query plan must be
fully initialized for computation.
*/
if (select_lex->optimize_constant_subqueries())
DBUG_RETURN(TRUE);
optimize_subqueries= TRUE;
const_cond= (!conds || conds->const_item());
safe_update= (thd->variables.option_bits & OPTION_SAFE_UPDATES) &&
!thd->lex->describe;
if (safe_update && const_cond)
{
my_message(ER_UPDATE_WITHOUT_KEY_IN_SAFE_MODE,
ER_THD(thd, ER_UPDATE_WITHOUT_KEY_IN_SAFE_MODE), MYF(0));
DBUG_RETURN(TRUE);
}
const_cond_result= const_cond && (!conds || conds->val_bool());
if (unlikely(thd->is_error()))
{
/* Error evaluating val_bool(). */
DBUG_RETURN(TRUE);
}
/*
Test if the user wants to delete all rows and deletion doesn't have
any side-effects (because of triggers), so we can use optimized
handler::delete_all_rows() method.
We can use delete_all_rows() if and only if:
- We allow new functions (not using option --skip-new), and are
not in safe mode (not using option --safe-mode)
- There is no limit clause
- The condition is constant
- If there is a condition, then it produces a non-zero value
- If the current command is DELETE FROM with no where clause, then:
- We should not be binlogging this statement in row-based, and
- there should be no delete triggers associated with the table.
*/
has_triggers= table->triggers && table->triggers->has_delete_triggers();
transactional_table= table->file->has_transactions_and_rollback();
deleted= 0;
if (!returning && !using_limit && const_cond_result &&
!thd->is_current_stmt_binlog_format_row() && !has_triggers &&
!table->versioned(VERS_TIMESTAMP) && !table_list->has_period())
{
/* Update the table->file->stats.records number */
table->file->info(HA_STATUS_VARIABLE | HA_STATUS_NO_LOCK);
set_statistics_for_table(thd, table);
ha_rows const maybe_deleted= table->file->stats.records;
DBUG_PRINT("debug", ("Trying to use delete_all_rows()"));
query_plan.set_delete_all_rows(maybe_deleted);
if (thd->lex->describe)
goto produce_explain_and_leave;
table->file->prepare_for_modify(false, false);
if (likely(!(error=table->file->ha_delete_all_rows())))
{
/*
If delete_all_rows() is used, it is not possible to log the
query in row format, so we have to log it in statement format.
*/
query_type= THD::STMT_QUERY_TYPE;
error= -1;
deleted= maybe_deleted;
if (!query_plan.save_explain_delete_data(thd, thd->mem_root))
error= 1;
goto cleanup;
}
if (error != HA_ERR_WRONG_COMMAND)
{
table->file->print_error(error,MYF(0));
error=0;
goto cleanup;
}
/* Handler didn't support fast delete; Delete rows one by one */
query_plan.cancel_delete_all_rows();
}
if (conds)
{
Item::cond_result result;
conds= conds->remove_eq_conds(thd, &result, true);
if (result == Item::COND_FALSE) // Impossible where
{
limit= 0;
query_plan.set_impossible_where();
if (thd->lex->describe || thd->lex->analyze_stmt)
goto produce_explain_and_leave;
}
}
if (conds && thd->lex->are_date_funcs_used())
{
/* Rewrite datetime comparison conditions into sargable */
conds= conds->top_level_transform(thd, &Item::date_conds_transformer,
(uchar *) 0);
}
if (conds && optimizer_flag(thd, OPTIMIZER_SWITCH_SARGABLE_CASEFOLD))
{
conds= conds->top_level_transform(thd, &Item::varchar_upper_cmp_transformer,
(uchar *) 0);
}
if ((conds || order) && substitute_indexed_vcols_for_table(table, conds,
order, select_lex))
DBUG_RETURN(1); // Fatal error
#ifdef WITH_PARTITION_STORAGE_ENGINE
if (prune_partitions(thd, table, conds))
{
if (optimize_subqueries && select_lex->optimize_unflattened_subqueries(false))
DBUG_RETURN(TRUE);
optimize_subqueries= FALSE;
free_underlaid_joins(thd, select_lex);
query_plan.set_no_partitions();
if (thd->lex->describe || thd->lex->analyze_stmt)
goto produce_explain_and_leave;
if (thd->binlog_for_noop_dml(transactional_table))
DBUG_RETURN(1);
if (!thd->lex->current_select->leaf_tables_saved)
{
thd->lex->current_select->save_leaf_tables(thd);
thd->lex->current_select->leaf_tables_saved= true;
thd->lex->current_select->first_cond_optimization= 0;
}
my_ok(thd, 0);
DBUG_RETURN(0);
}
#endif
/* Update the table->file->stats.records number */
table->file->info(HA_STATUS_VARIABLE | HA_STATUS_NO_LOCK);
set_statistics_for_table(thd, table);
table->covering_keys.clear_all();
table->opt_range_keys.clear_all();
select=make_select(table, 0, 0, conds, (SORT_INFO*) 0, 0, &error);
if (unlikely(error))
DBUG_RETURN(TRUE);
if ((select && select->check_quick(thd, safe_update, limit,
Item_func::BITMAP_ALL)) || !limit ||
table->stat_records() == 0)
{
query_plan.set_impossible_where();
if (thd->lex->describe || thd->lex->analyze_stmt)
goto produce_explain_and_leave;
delete select;
if (select_lex->optimize_unflattened_subqueries(false))
DBUG_RETURN(TRUE);
optimize_subqueries= FALSE;
free_underlaid_joins(thd, select_lex);
/*
Error was already created by quick select evaluation (check_quick()).
TODO: Add error code output parameter to Item::val_xxx() methods.
Currently they rely on the user checking DA for
errors when unwinding the stack after calling Item::val_xxx().
*/
if (unlikely(thd->is_error()))
DBUG_RETURN(TRUE);
if (thd->binlog_for_noop_dml(transactional_table))
DBUG_RETURN(1);
if (!thd->lex->current_select->leaf_tables_saved)
{
thd->lex->current_select->save_leaf_tables(thd);
thd->lex->current_select->leaf_tables_saved= true;
thd->lex->current_select->first_cond_optimization= 0;
}
my_ok(thd, 0);
DBUG_RETURN(0); // Nothing to delete
}
/* If running in safe sql mode, don't allow updates without keys */
if (!select || !select->quick)
{
thd->set_status_no_index_used();
if (safe_update && !using_limit)
{
delete select;
if (optimize_subqueries && select_lex->optimize_unflattened_subqueries(false))
DBUG_RETURN(TRUE);
optimize_subqueries= FALSE;
free_underlaid_joins(thd, select_lex);
my_message(ER_UPDATE_WITHOUT_KEY_IN_SAFE_MODE,
ER_THD(thd, ER_UPDATE_WITHOUT_KEY_IN_SAFE_MODE), MYF(0));
DBUG_RETURN(TRUE);
}
}
if (options & OPTION_QUICK)
(void) table->file->extra(HA_EXTRA_QUICK);
/*
Estimate the number of scanned rows and have it accessible in
JOIN::choose_subquery_plan() from the outer join through
JOIN::sql_cmd_dml
*/
scanned_rows= query_plan.scanned_rows= select ?
select->records : table->file->stats.records;
select_lex->join->sql_cmd_dml= this;
DBUG_ASSERT(optimize_subqueries);
if (select_lex->optimize_unflattened_subqueries(false))
DBUG_RETURN(TRUE);
optimize_subqueries= FALSE;
if (order)
{
table->update_const_key_parts(conds);
order= simple_remove_const(order, conds);
if (select && select->quick && select->quick->unique_key_range())
{ // Single row select (always "ordered")
query_plan.using_filesort= FALSE;
query_plan.index= MAX_KEY;
}
else
{
ha_rows scanned_limit= query_plan.scanned_rows;
table->no_keyread= 1;
query_plan.index= get_index_for_order(order, table, select, limit,
&scanned_limit,
&query_plan.using_filesort,
&reverse);
table->no_keyread= 0;
if (!query_plan.using_filesort)
query_plan.scanned_rows= scanned_limit;
}
}
query_plan.select= select;
query_plan.possible_keys= select? select->possible_keys: key_map(0);
/*
Ok, we have generated a query plan for the DELETE.
- if we're running EXPLAIN DELETE, goto produce explain output
- otherwise, execute the query plan
*/
if (thd->lex->describe)
goto produce_explain_and_leave;
if (!(explain= query_plan.save_explain_delete_data(thd, thd->mem_root)))
goto got_error;
ANALYZE_START_TRACKING(thd, &explain->command_tracker);
DBUG_EXECUTE_IF("show_explain_probe_delete_exec_start",
dbug_serve_apcs(thd, 1););
if (!(select && select->quick))
status_var_increment(thd->status_var.delete_scan_count);
binlog_is_row= thd->is_current_stmt_binlog_format_row();
DBUG_PRINT("info", ("binlog_is_row: %s", binlog_is_row ? "TRUE" : "FALSE"));
/*
We can use direct delete (delete that is done silently in the handler)
if none of the following conditions are true:
- There are triggers
- There is binary logging
- There is a virtual not stored column in the WHERE clause
- ORDER BY or LIMIT
- As this requires the rows to be deleted in a specific order
- Note that Spider can handle ORDER BY and LIMIT in a cluster with
one data node. These conditions are therefore checked in
direct_delete_rows_init().
Direct delete does not require a WHERE clause
Later we also ensure that we are only using one table (no sub queries)
*/
if ((table->file->ha_table_flags() & HA_CAN_DIRECT_UPDATE_AND_DELETE) &&
!has_triggers && !binlog_is_row && !returning &&
!table_list->has_period())
{
table->mark_columns_needed_for_delete();
if (!table->check_virtual_columns_marked_for_read())
{
DBUG_PRINT("info", ("Trying direct delete"));
bool use_direct_delete= !select || !select->cond;
if (!use_direct_delete &&
(select->cond->used_tables() & ~RAND_TABLE_BIT) == table->map)
{
DBUG_ASSERT(!table->file->pushed_cond);
if (!table->file->cond_push(select->cond))
{
use_direct_delete= TRUE;
table->file->pushed_cond= select->cond;
}
}
if (use_direct_delete && !table->file->direct_delete_rows_init())
{
/* Direct deleting is supported */
DBUG_PRINT("info", ("Using direct delete"));
THD_STAGE_INFO(thd, stage_updating);
if (!(error= table->file->ha_direct_delete_rows(&deleted)))
error= -1;
goto terminate_delete;
}
}
}
if (query_plan.using_filesort)
{
{
Filesort fsort(order, HA_POS_ERROR, true, select);
DBUG_ASSERT(query_plan.index == MAX_KEY);
Filesort_tracker *fs_tracker=
thd->lex->explain->get_upd_del_plan()->filesort_tracker;
if (!(file_sort= filesort(thd, table, &fsort, fs_tracker)))
goto got_error;
thd->ps_report_examined_row_count();
/*
Filesort has already found and selected the rows we want to delete,
so we don't need the where clause
*/
delete select;
/*
If we are not in DELETE ... RETURNING, we can free subqueries. (in
DELETE ... RETURNING we can't, because the RETURNING part may have
a subquery in it)
*/
if (!returning)
free_underlaid_joins(thd, select_lex);
select= 0;
}
}
/* If quick select is used, initialize it before retrieving rows. */
if (select && select->quick && select->quick->reset())
goto got_error;
init_table_full_scan_if_needed(table, conds, limit);
if (query_plan.index == MAX_KEY || (select && select->quick))
error= init_read_record(&info, thd, table, select, file_sort, 1, 1, FALSE);
else
error= init_read_record_idx(&info, thd, table, 1, query_plan.index,
reverse);
if (unlikely(error))
goto got_error;
if (unlikely(init_ftfuncs(thd, select_lex, 1)))
goto got_error;
if (table_list->has_period())
{
table->use_all_columns();
table->rpl_write_set= table->write_set;
// Initialize autoinc.
// We don't set next_number_field here, as it is handled manually.
if (table->found_next_number_field)
table->file->info(HA_STATUS_AUTO);
}
else
{
table->mark_columns_needed_for_delete();
}
if (!table->prepare_triggers_for_delete_stmt_or_event() &&
table->file->ha_table_flags() & HA_CAN_FORCE_BULK_DELETE)
will_batch= !table->file->start_bulk_delete();
/*
thd->get_stmt_da()->is_set() means first iteration of prepared statement
with array binding operation execution (non optimized so it is not
INSERT)
*/
if (returning && !thd->get_stmt_da()->is_set())
{
if (result->send_result_set_metadata(returning->returning_list,
Protocol::SEND_NUM_ROWS | Protocol::SEND_EOF))
goto cleanup;
}
explain= (Explain_delete*)thd->lex->explain->get_upd_del_plan();
explain->tracker.on_scan_init();
thd->get_stmt_da()->reset_current_row_for_warning(1);
if (!delete_while_scanning)
{
/*
The table we are going to delete appears in subqueries in the where
clause. Instead of deleting the rows, first mark them deleted.
*/
ha_rows tmplimit=limit;
deltempfile= new (thd->mem_root) Unique (refpos_order_cmp, table->file,
table->file->ref_length,
MEM_STRIP_BUF_SIZE);
THD_STAGE_INFO(thd, stage_searching_rows_for_update);
while (!(error=info.read_record()) && !thd->killed && !thd->is_error())
{
if (record_should_be_deleted(thd, table, select, explain, delete_history))
{
table->file->position(table->record[0]);
if ((error= deltempfile->unique_add((char*) table->file->ref)))
break;
if (!--tmplimit && using_limit)
break;
}
}
end_read_record(&info);
if (table->file->ha_index_or_rnd_end() || error > 0 ||
deltempfile->get(table) ||
init_read_record(&info, thd, table, 0, &deltempfile->sort, 0, 1, 0))
{
error= 1;
goto terminate_delete;
}
delete_record= true;
}
/*
From SQL2016, Part 2, 15.7 <Effect of deleting rows from base table>,
General Rules, 8), we can conclude that DELETE FOR PORTION OF time performs
0-2 INSERTS + DELETE. We can substitute INSERT+DELETE with one UPDATE, with
a condition of no side effects. The side effect is possible if there is a
BEFORE INSERT trigger, since it is the only one splitting DELETE and INSERT
operations.
Another possible side effect is related to tables of non-transactional
engines, since UPDATE is anyway atomic, and DELETE+INSERT is not.
This optimization is not possible for system-versioned table.
*/
portion_of_time_through_update=
!(table->triggers && table->triggers->has_triggers(TRG_EVENT_INSERT,
TRG_ACTION_BEFORE))
&& !table->versioned()
&& table->file->has_transactions();
table->file->prepare_for_modify(table->versioned(VERS_TIMESTAMP) ||
table_list->has_period(), true);
DBUG_ASSERT(table->file->inited != handler::NONE);
THD_STAGE_INFO(thd, stage_updating);
fix_rownum_pointers(thd, thd->lex->current_select, &deleted);
thd->get_stmt_da()->reset_current_row_for_warning(0);
while (likely(!(error=info.read_record())) && likely(!thd->killed) &&
likely(!thd->is_error()))
{
thd->get_stmt_da()->inc_current_row_for_warning();
if (delete_while_scanning)
delete_record= record_should_be_deleted(thd, table, select, explain,
delete_history);
if (delete_record)
{
bool trg_skip_row= false;
if (!delete_history && table->triggers &&
table->triggers->process_triggers(thd, TRG_EVENT_DELETE,
TRG_ACTION_BEFORE, FALSE,
&trg_skip_row))
{
error= 1;
break;
}
if (trg_skip_row)
continue;
// no LIMIT / OFFSET
if (returning && result->send_data(returning->returning_list) < 0)
{
error=1;
break;
}
if (table_list->has_period() && portion_of_time_through_update)
{
bool need_delete= true;
error= update_portion_of_time(thd, table, table_list->period_conditions,
&need_delete);
if (likely(!error) && need_delete)
error= table->delete_row();
}
else
{
error= table->delete_row();
ha_rows rows_inserted;
if (likely(!error) && table_list->has_period()
&& !portion_of_time_through_update)
error= table->insert_portion_of_time(thd, table_list->period_conditions,
&rows_inserted);
}
if (likely(!error))
{
deleted++;
if (!delete_history && table->triggers &&
table->triggers->process_triggers(thd, TRG_EVENT_DELETE,
TRG_ACTION_AFTER, false,
nullptr))
{
error= 1;
break;
}
if (!--limit && using_limit)
{
error= -1;
break;
}
}
else
{
table->file->print_error(error,
MYF(thd->lex->ignore ? ME_WARNING : 0));
if (thd->is_error())
{
error= 1;
break;
}
}
}
/*
Don't try unlocking the row if skip_record reported an error since in
this case the transaction might have been rolled back already.
*/
else if (likely(!thd->is_error()))
table->file->unlock_row(); // Row failed selection, release lock on it
else
break;
}
thd->get_stmt_da()->reset_current_row_for_warning(1);
terminate_delete:
killed_status= thd->killed;
if (unlikely(killed_status != NOT_KILLED || thd->is_error()))
error= 1; // Aborted
if (will_batch && unlikely((loc_error= table->file->end_bulk_delete())))
{
if (error != 1)
table->file->print_error(loc_error,MYF(0));
error=1;
}
THD_STAGE_INFO(thd, stage_end);
end_read_record(&info);
if (table_list->has_period())
table->file->ha_release_auto_increment();
if (options & OPTION_QUICK)
(void) table->file->extra(HA_EXTRA_NORMAL);
ANALYZE_STOP_TRACKING(thd, &explain->command_tracker);
cleanup:
/*
Invalidate the table in the query cache if something changed. This must
be before binlog writing and ha_autocommit_...
*/
if (deleted)
{
query_cache_invalidate3(thd, table_list, 1);
}
if (!thd->lex->current_select->leaf_tables_saved)
{
thd->lex->current_select->save_leaf_tables(thd);
thd->lex->current_select->leaf_tables_saved= true;
thd->lex->current_select->first_cond_optimization= false;
}
delete deltempfile;
deltempfile=NULL;
delete select;
select= NULL;
if (!transactional_table && deleted > 0)
thd->transaction->stmt.modified_non_trans_table=
thd->transaction->all.modified_non_trans_table= TRUE;
/* See similar binlogging code in sql_update.cc, for comments */
if (likely((error < 0) || thd->transaction->stmt.modified_non_trans_table ||
thd->log_current_statement()))
{
if ((WSREP_EMULATE_BINLOG(thd) || mysql_bin_log.is_open()) &&
table->s->using_binlog())
{
int errcode= 0;
if (error < 0)
thd->clear_error();
else
errcode= query_error_code(thd, killed_status == NOT_KILLED);
StatementBinlog stmt_binlog(thd, table->versioned(VERS_TRX_ID) ||
thd->binlog_need_stmt_format(transactional_table));
/*
[binlog]: If 'handler::delete_all_rows()' was called and the
storage engine does not inject the rows itself, we replicate
statement-based; otherwise, 'ha_delete_row()' was used to
delete specific rows which we might log row-based.
*/
int log_result= thd->binlog_query(query_type,
thd->query(), thd->query_length(),
transactional_table, FALSE, FALSE,
errcode);
if (log_result > 0)
{
error=1;
}
else
binlogged= 1;
}
}
if (!binlogged)
table->mark_as_not_binlogged();
DBUG_ASSERT(transactional_table || !deleted || thd->transaction->stmt.modified_non_trans_table);
if (likely(error < 0) ||
(thd->lex->ignore && !thd->is_error() && !thd->is_fatal_error))
{
if (thd->lex->analyze_stmt)
goto send_nothing_and_leave;
thd->collect_unit_results(0, deleted);
if (returning)
result->send_eof();
else
my_ok(thd, deleted);
DBUG_PRINT("info", ("%ld records deleted", (long) deleted));
}
delete file_sort;
if (optimize_subqueries && select_lex->optimize_unflattened_subqueries(false))
DBUG_RETURN(TRUE);
free_underlaid_joins(thd, select_lex);
if (table->file->pushed_cond)
table->file->cond_pop();
DBUG_RETURN(error >= 0 || thd->is_error());
/* Special exits */
produce_explain_and_leave:
/*
We come here for various "degenerate" query plans: impossible WHERE,
no-partitions-used, impossible-range, etc.
*/
if (!(query_plan.save_explain_delete_data(thd, thd->mem_root)))
goto got_error;
send_nothing_and_leave:
/*
ANALYZE DELETE jumps here. We can't send explain right here, because
we might be using ANALYZE DELETE ...RETURNING, in which case we have
Protocol_discard active.
*/
delete select;
delete file_sort;
if (!thd->is_error() && optimize_subqueries &&
select_lex->optimize_unflattened_subqueries(false))
DBUG_RETURN(TRUE);
free_underlaid_joins(thd, select_lex);
if (table->file->pushed_cond)
table->file->cond_pop();
DBUG_ASSERT(!return_error || thd->is_error() || thd->killed);
DBUG_RETURN((return_error || thd->is_error() || thd->killed) ? 1 : 0);
got_error:
return_error= 1;
goto send_nothing_and_leave;
}
/***************************************************************************
Delete multiple tables from join
***************************************************************************/
extern "C" int refpos_order_cmp(void *arg, const void *a, const void *b)
{
auto file= static_cast<handler *>(arg);
return file->cmp_ref(static_cast<const uchar *>(a),
static_cast<const uchar *>(b));
}
multi_delete::multi_delete(THD *thd_arg,
TABLE_LIST *dt,
uint num_of_tables_arg)
: select_result_interceptor(thd_arg),
delete_tables(dt),
deleted(0),
found(0),
table_count(num_of_tables_arg),
error(0),
do_delete(0),
transactional_tables(0),
normal_tables(0),
error_handled(0)
{
tmp_tables = thd->calloc<TABLE*>(table_count);
tmp_table_param = thd->calloc<TMP_TABLE_PARAM>(table_count);
}
int
multi_delete::prepare(List<Item> &values, SELECT_LEX_UNIT *u)
{
DBUG_ENTER("multi_delete::prepare");
unit= u;
do_delete= 1;
THD_STAGE_INFO(thd, stage_deleting_from_main_table);
DBUG_RETURN(0);
}
static TABLE *item_rowid_table(Item *item)
{
if (item->type() != Item::FUNC_ITEM)
return NULL;
Item_func *func= (Item_func *)item;
if (func->functype() != Item_func::TEMPTABLE_ROWID)
return NULL;
Item_temptable_rowid *itr= (Item_temptable_rowid *)func;
return itr->table;
}
/*
multi_delete stores a rowid and new field values for every updated row in a
temporary table (one temporary table per updated table). These rowids are
obtained via Item_temptable_rowid's by calling handler::position(). But if
the join is resolved via a temp table, rowids cannot be obtained from
handler::position() in the multi_update::send_data(). So, they're stored in
the join's temp table (JOIN::add_fields_for_current_rowid()) and here we
replace Item_temptable_rowid's (that would've done handler::position()) with
Item_field's (that will simply take the corresponding field value from the
temp table).
*/
int multi_delete::prepare2(JOIN *join)
{
if (!join->need_tmp || !join->tmp_table_keep_current_rowid)
return 0;
delete_while_scanning= false;
JOIN_TAB *tmptab= join->join_tab + join->exec_join_tab_cnt();
for (Item **it= tmptab->tmp_table_param->items_to_copy; *it ; it++)
{
TABLE *tbl= item_rowid_table(*it);
if (!tbl)
continue;
for (uint i= 0; i < table_count; i++)
{
for (Item **it2= tmp_table_param[i].items_to_copy; *it2; it2++)
{
if (item_rowid_table(*it2) != tbl)
continue;
Item_field *fld= new (thd->mem_root)
Item_field(thd, (*it)->get_tmp_table_field());
if (!fld)
return 1;
fld->result_field= (*it2)->get_tmp_table_field();
*it2= fld;
}
}
}
return 0;
}
void multi_delete::prepare_to_read_rows()
{
/* see multi_update::prepare_to_read_rows() */
for (TABLE_LIST *walk= delete_tables; walk; walk= walk->next_local)
{
TABLE_LIST *tbl= walk->correspondent_table->find_table_for_update();
tbl->table->mark_columns_needed_for_delete();
}
}
bool
multi_delete::initialize_tables(JOIN *join)
{
TABLE_LIST *walk;
DBUG_ENTER("initialize_tables");
if (unlikely((thd->variables.option_bits & OPTION_SAFE_UPDATES) &&
error_if_full_join(join)))
DBUG_RETURN(1);
main_table=join->join_tab->table;
table_map tables_to_delete_from= 0;
delete_while_scanning= true;
for (walk= delete_tables; walk; walk= walk->next_local)
{
TABLE_LIST *tbl= walk->correspondent_table->find_table_for_update();
tables_to_delete_from|= tbl->table->map;
/*
Ensure that filesort re-reads the row from the engine before
delete is called.
*/
join->map2table[tbl->table->tablenr]->keep_current_rowid= true;
if (delete_while_scanning &&
unique_table(thd, tbl, join->tables_list, 0))
{
/*
If the table we are going to delete from appears
in join, we need to defer delete. So the delete
doesn't interfere with the scanning of results.
*/
delete_while_scanning= false;
}
}
walk= delete_tables;
uint index= 0;
for (JOIN_TAB *tab= first_linear_tab(join, WITHOUT_BUSH_ROOTS,
WITH_CONST_TABLES);
tab;
tab= next_linear_tab(join, tab, WITHOUT_BUSH_ROOTS))
{
if (!tab->bush_children && tab->table->map & tables_to_delete_from)
{
/* We are going to delete from this table */
TABLE *tbl=walk->table=tab->table;
TABLE_LIST *prior= walk;
walk= walk->next_local;
/* Don't use KEYREAD optimization on this table */
tbl->no_keyread=1;
/* Don't use record cache */
tbl->no_cache= 1;
tbl->covering_keys.clear_all();
if (tbl->file->has_transactions())
transactional_tables= 1;
else
normal_tables= 1;
tbl->prepare_triggers_for_delete_stmt_or_event();
tbl->prepare_for_position();
tbl->file->prepare_for_modify(tbl->versioned(VERS_TIMESTAMP), true);
List<Item> temp_fields;
tbl->prepare_for_position();
join->map2table[tbl->tablenr]->keep_current_rowid= true;
Item_temptable_rowid *item=
new (thd->mem_root) Item_temptable_rowid(tbl);
if (!item)
DBUG_RETURN(1);
item->fix_fields(thd, 0);
if (temp_fields.push_back(item, thd->mem_root))
DBUG_RETURN(1);
/* Make an unique key over the first field to avoid duplicated updates */
ORDER *group= (ORDER*) alloc_root(thd->mem_root, sizeof(ORDER));
if (!group)
DBUG_RETURN(1);
bzero((char*) group, sizeof(*group));
group->direction= ORDER::ORDER_ASC;
group->item= (Item**) temp_fields.head_ref();
TMP_TABLE_PARAM *tmp_param;
prior->shared = index;
tmp_param= &tmp_table_param[prior->shared];
tmp_param->init();
tmp_param->tmp_name="update";
tmp_param->field_count= temp_fields.elements;
tmp_param->func_count= temp_fields.elements;
calc_group_buffer(tmp_param, group);
tmp_tables[index]=create_tmp_table(thd, tmp_param, temp_fields,
group, 0, 0,
TMP_TABLE_ALL_COLUMNS, HA_POS_ERROR, &empty_clex_str);
if (!tmp_tables[index])
DBUG_RETURN(1);
tmp_tables[index]->file->extra(HA_EXTRA_WRITE_CACHE);
++index;
}
else if ((tab->type != JT_SYSTEM && tab->type != JT_CONST) &&
walk == delete_tables)
{
/*
We are not deleting from the table we are scanning. In this
case send_data() shouldn't delete any rows a we may touch
the rows in the deleted table many times
*/
delete_while_scanning= false;
}
}
if (delete_while_scanning)
table_being_deleted= delete_tables;
if (init_ftfuncs(thd, thd->lex->current_select, 1))
DBUG_RETURN(true);
join->tmp_table_keep_current_rowid= TRUE;
DBUG_RETURN(thd->is_fatal_error);
}
multi_delete::~multi_delete()
{
for (TABLE_LIST *walk= delete_tables; walk; walk= walk->next_local)
{
TABLE *table= walk->table;
if (!table)
continue;
table->no_keyread=0;
table->no_cache= 0;
}
if (tmp_tables)
{
for (uint cnt = 0; cnt < table_count; cnt++)
{
if (tmp_tables[cnt])
{
free_tmp_table(thd, tmp_tables[cnt]);
tmp_table_param[cnt].cleanup();
}
}
}
}
int multi_delete::send_data(List<Item> &values)
{
int secure_counter= delete_while_scanning ? -1 : 0;
TABLE_LIST *del_table;
DBUG_ENTER("multi_delete::send_data");
bool ignore= thd->lex->ignore;
for (del_table= delete_tables;
del_table;
del_table= del_table->next_local, ++secure_counter)
{
TABLE *table= del_table->table;
// DELETE and TRUNCATE don't affect SEQUENCE, so bail early
if (table->file->ht->db_type == DB_TYPE_SEQUENCE)
continue;
/* Check if we are using outer join and we didn't find the row */
if (table->status & (STATUS_NULL_ROW | STATUS_DELETED))
continue;
table->file->position(table->record[0]);
++found;
if (secure_counter < 0)
{
bool trg_skip_row= false;
/* We are scanning the current table */
DBUG_ASSERT(del_table == table_being_deleted);
if (table->triggers &&
table->triggers->process_triggers(thd, TRG_EVENT_DELETE,
TRG_ACTION_BEFORE, false,
&trg_skip_row))
DBUG_RETURN(1);
if (trg_skip_row)
continue;
table->status|= STATUS_DELETED;
error= table->delete_row();
if (likely(!error))
{
deleted++;
if (!table->file->has_transactions())
thd->transaction->stmt.modified_non_trans_table= TRUE;
if (table->triggers &&
table->triggers->process_triggers(thd, TRG_EVENT_DELETE,
TRG_ACTION_AFTER, false,
nullptr))
DBUG_RETURN(1);
}
else if (!ignore)
{
/*
If the IGNORE option is used errors caused by ha_delete_row don't
have to stop the iteration.
*/
table->file->print_error(error,MYF(0));
DBUG_RETURN(1);
}
}
else
{
const uint offset= del_table->shared;
TABLE *tmp_table= tmp_tables[offset];
if (copy_funcs(tmp_table_param[offset].items_to_copy, thd))
DBUG_RETURN(1);
/* rowid field is NULL if join tmp table has null row from outer join */
if (tmp_table->field[0]->is_null())
continue;
error= tmp_table->file->ha_write_tmp_row(tmp_table->record[0]);
if (error)
{
--found;
if (error != HA_ERR_FOUND_DUPP_KEY &&
error != HA_ERR_FOUND_DUPP_UNIQUE)
{
bool is_duplicate;
if (create_internal_tmp_table_from_heap(thd, tmp_table,
tmp_table_param[offset].start_recinfo,
&tmp_table_param[offset].recinfo,
error, 1, &is_duplicate, NULL))
{
do_delete= 0;
DBUG_RETURN(1); // Not a table_is_full error
}
found++;
}
error= 0;
}
}
}
DBUG_RETURN(0);
}
void multi_delete::abort_result_set()
{
TABLE_LIST *cur_table;
DBUG_ENTER("multi_delete::abort_result_set");
/****************************************************************************
NOTE: if you change here be aware that almost the same code is in
multi_delete::send_eof().
***************************************************************************/
/* the error was handled or nothing deleted and no side effects return */
if (error_handled ||
(!thd->transaction->stmt.modified_non_trans_table && !deleted))
DBUG_VOID_RETURN;
/* Something already deleted so we have to invalidate cache */
if (deleted)
query_cache_invalidate3(thd, delete_tables, 1);
if (thd->transaction->stmt.modified_non_trans_table)
thd->transaction->all.modified_non_trans_table= TRUE;
thd->transaction->all.m_unsafe_rollback_flags|=
(thd->transaction->stmt.m_unsafe_rollback_flags & THD_TRANS::DID_WAIT);
/*
If rows from the first table only has been deleted and it is
transactional, just do rollback.
The same if all tables are transactional, regardless of where we are.
In all other cases do attempt deletes ...
*/
if (do_delete && normal_tables &&
(table_being_deleted != delete_tables ||
!table_being_deleted->table->file->has_transactions_and_rollback()))
{
/*
We have to execute the recorded do_deletes() and write info into the
error log
*/
error= 1;
send_eof();
DBUG_ASSERT(error_handled);
DBUG_VOID_RETURN;
}
if (thd->transaction->stmt.modified_non_trans_table ||
thd->log_current_statement())
{
/*
there is only side effects; to binlog with the error
*/
if (WSREP_EMULATE_BINLOG(thd) || mysql_bin_log.is_open())
{
StatementBinlog stmt_binlog(thd, thd->binlog_need_stmt_format(transactional_tables));
int errcode= query_error_code(thd, thd->killed == NOT_KILLED);
/* possible error of writing binary log is ignored deliberately */
(void) thd->binlog_query(THD::ROW_QUERY_TYPE,
thd->query(), thd->query_length(),
transactional_tables, FALSE, FALSE, errcode);
}
}
/*
Mark all temporay tables as not completely binlogged
All future usage of these tables will enforce row level logging, which
ensures that all future usage of them enforces row level logging.
*/
for (cur_table= delete_tables; cur_table; cur_table= cur_table->next_local)
cur_table->table->mark_as_not_binlogged();
DBUG_VOID_RETURN;
}
/**
Do delete from other tables.
@retval 0 ok
@retval 1 error
@todo Is there any reason not use the normal nested-loops join? If not, and
there is no documentation supporting it, this method and callee should be
removed and there should be hooks within normal execution.
*/
int multi_delete::do_deletes()
{
DBUG_ENTER("do_deletes");
DBUG_ASSERT(do_delete);
do_delete= 0; // Mark called
if (!found)
DBUG_RETURN(0);
table_being_deleted= (delete_while_scanning ? delete_tables->next_local :
delete_tables);
for (; table_being_deleted;
table_being_deleted= table_being_deleted->next_local)
{
TABLE *table = table_being_deleted->table;
// DELETE and TRUNCATE don't affect SEQUENCE, so bail early
if (table->file->ht->db_type == DB_TYPE_SEQUENCE)
continue;
int local_error= rowid_table_deletes(table, thd->lex->ignore);
if (unlikely(thd->killed) && likely(!local_error))
DBUG_RETURN(1);
if (unlikely(local_error == -1)) // End of file
local_error= 0;
if (unlikely(local_error))
DBUG_RETURN(local_error);
}
DBUG_RETURN(0);
}
/**
Implements the inner loop of nested-loops join within multi-DELETE
execution.
@param table The table from which to delete.
@param ignore If used, all non fatal errors will be translated
to warnings and we should not break the row-by-row iteration.
@return Status code
@retval 0 All ok.
@retval 1 Triggers or handler reported error.
@retval -1 End of file from handler.
*/
int multi_delete::rowid_table_deletes(TABLE *table, bool ignore)
{
int local_error= 0;
ha_rows last_deleted= deleted;
DBUG_ENTER("rowid_table_deletes");
TABLE *err_table= nullptr;
bool will_batch= !table->file->start_bulk_delete();
TABLE *tmp_table= tmp_tables[table_being_deleted->shared];
tmp_table->file->extra(HA_EXTRA_CACHE); // Change to read cache
if (unlikely((local_error= table->file->ha_rnd_init(0))))
{
err_table= table;
goto err;
}
table->file->extra(HA_EXTRA_NO_CACHE);
if (unlikely((local_error= tmp_table->file->ha_rnd_init(1))))
{
err_table= tmp_table;
goto err;
}
while (!thd->killed)
{
if (unlikely((local_error=
tmp_table->file->ha_rnd_next(tmp_table->record[0]))))
{
if (local_error == HA_ERR_END_OF_FILE)
{
local_error= 0;
break;
}
err_table= tmp_table;
goto err;
}
DBUG_ASSERT(!tmp_table->field[0]->is_null());
String rowid;
tmp_table->field[0]->val_str(&rowid);
if (unlikely((local_error= table->file->ha_rnd_pos(table->record[0],
(uchar*)rowid.ptr()))))
{
// Table aliased to itself had key deleted already
continue;
}
bool trg_skip_row= false;
if (table->triggers &&
unlikely(table->triggers->process_triggers(thd, TRG_EVENT_DELETE,
TRG_ACTION_BEFORE, FALSE,
&trg_skip_row)))
{
err_table= table;
local_error= 1;
break;
}
if (trg_skip_row)
continue;
local_error= table->delete_row();
if (unlikely(local_error) && !ignore)
{
table->file->print_error(local_error, MYF(0));
break;
}
/*
Increase the reported number of deleted rows only if no error occurred
during ha_delete_row.
Also, don't execute the AFTER trigger if the row operation failed.
*/
if (likely(!local_error))
{
deleted++;
if (table->triggers &&
table->triggers->process_triggers(thd, TRG_EVENT_DELETE,
TRG_ACTION_AFTER, FALSE,
nullptr))
{
err_table= table;
local_error= 1;
break;
}
}
}
if (will_batch)
{
int tmp_error= table->file->end_bulk_delete();
if (unlikely(tmp_error) && !local_error)
{
local_error= tmp_error;
table->file->print_error(local_error, MYF(0));
}
}
if (last_deleted != deleted && !table->file->has_transactions_and_rollback())
thd->transaction->stmt.modified_non_trans_table= TRUE;
err:
if (err_table)
err_table->file->print_error(local_error,MYF(ME_FATAL));
if (tmp_table->file->inited == handler::init_stat::RND)
tmp_table->file->ha_rnd_end();
if (table->file->inited == handler::init_stat::RND)
table->file->ha_rnd_end();
DBUG_RETURN(local_error);
}
/*
Send ok to the client
return: 0 success
1 error
*/
bool multi_delete::send_eof()
{
killed_state killed_status= NOT_KILLED;
THD_STAGE_INFO(thd, stage_deleting_from_reference_tables);
/* Does deletes for the last n - 1 tables, returns 0 if ok */
int local_error= do_deletes(); // returns 0 if success
/* compute a total error to know if something failed */
local_error= local_error || error;
killed_status= (local_error == 0)? NOT_KILLED : thd->killed;
/* reset used flags */
THD_STAGE_INFO(thd, stage_end);
/****************************************************************************
NOTE: if you change here be aware that almost the same code is in
multi_delete::abort_result_set().
***************************************************************************/
if (thd->transaction->stmt.modified_non_trans_table)
thd->transaction->all.modified_non_trans_table= TRUE;
thd->transaction->all.m_unsafe_rollback_flags|=
(thd->transaction->stmt.m_unsafe_rollback_flags & THD_TRANS::DID_WAIT);
/*
We must invalidate the query cache before binlog writing and
ha_autocommit_...
*/
if (deleted)
{
query_cache_invalidate3(thd, delete_tables, 1);
}
if (likely((local_error == 0) ||
thd->transaction->stmt.modified_non_trans_table) ||
thd->log_current_statement())
{
if(WSREP_EMULATE_BINLOG(thd) || mysql_bin_log.is_open())
{
int errcode= 0;
if (likely(local_error == 0))
thd->clear_error();
else
errcode= query_error_code(thd, killed_status == NOT_KILLED);
thd->used|= THD::THREAD_SPECIFIC_USED;
StatementBinlog stmt_binlog(thd, thd->binlog_need_stmt_format(transactional_tables));
if (unlikely(thd->binlog_query(THD::ROW_QUERY_TYPE,
thd->query(), thd->query_length(),
transactional_tables, FALSE, FALSE,
errcode) > 0) &&
!normal_tables)
{
local_error=1; // Log write failed: roll back the SQL statement
}
}
}
if (unlikely(local_error != 0))
{
error_handled= TRUE; // to force early leave from ::abort_result_set()
if (thd->killed == NOT_KILLED && !thd->get_stmt_da()->is_set())
{
/*
No error message was sent and query was not killed (in which case
mysql_execute_command() will send the error mesage).
*/
::my_ok(thd, deleted); // Ends the DELETE statement
}
return 1;
}
if (likely(!thd->lex->analyze_stmt))
{
::my_ok(thd, deleted);
}
return 0;
}
/**
@brief Remove ORDER BY from DELETE if it's used without limit clause
*/
void Sql_cmd_delete::remove_order_by_without_limit(THD *thd)
{
SELECT_LEX *const select_lex = thd->lex->first_select_lex();
if (select_lex->order_list.elements &&
!select_lex->limit_params.select_limit)
select_lex->order_list.empty();
}
/**
@brief Check whether processing to multi-table delete is prohibited
@param thd global context the processed statement
@returns true if processing as multitable is prohibited, false otherwise
@todo
Introduce handler level flag for storage engines that would prohibit
such conversion for any single-table delete.
*/
bool Sql_cmd_delete::processing_as_multitable_delete_prohibited(THD *thd)
{
return
(thd->lex->has_returning());
}
/**
@brief Perform precheck of table privileges for delete statements
@param thd global context the processed statement
@returns false on success, true on error
*/
bool Sql_cmd_delete::precheck(THD *thd)
{
if (!multitable)
{
if (delete_precheck(thd, lex->query_tables))
return true;
}
else
{
if (multi_delete_precheck(thd, lex->query_tables))
return true;
SELECT_LEX *select_lex= lex->first_select_lex();
/* condition will be TRUE on SP re-excuting */
if (select_lex->item_list.elements != 0)
select_lex->item_list.empty();
if (add_item_to_list(thd, new (thd->mem_root) Item_null(thd)))
return true;
}
#ifdef WITH_WSREP
WSREP_SYNC_WAIT(thd, WSREP_SYNC_WAIT_BEFORE_UPDATE_DELETE);
#endif
return false;
#ifdef WITH_WSREP
wsrep_error_label:
#endif
return true;
}
/**
@brief Perform context analysis for delete statements
@param thd global context the processed statement
@returns false on success, true on error
@note
The main bulk of the context analysis actions for a delete statement
is performed by a call of JOIN::prepare().
*/
bool Sql_cmd_delete::prepare_inner(THD *thd)
{
int err= 0;
TABLE_LIST *target_tbl;
JOIN *join;
SELECT_LEX *const select_lex = thd->lex->first_select_lex();
TABLE_LIST *const table_list = select_lex->get_table_list();
TABLE_LIST *aux_tables= thd->lex->auxiliary_table_list.first;
ulonglong select_options= select_lex->options;
bool free_join= 1;
SELECT_LEX *returning= thd->lex->has_returning() ? thd->lex->returning() : 0;
const bool delete_history= table_list->vers_conditions.delete_history;
DBUG_ASSERT(!(delete_history && table_list->period_conditions.is_set()));
DBUG_ENTER("Sql_cmd_delete::prepare_inner");
(void) read_statistics_for_tables_if_needed(thd, table_list);
THD_STAGE_INFO(thd, stage_init_update);
{
if (mysql_handle_derived(lex, DT_INIT))
DBUG_RETURN(TRUE);
if (mysql_handle_derived(lex, DT_MERGE_FOR_INSERT))
DBUG_RETURN(TRUE);
if (mysql_handle_derived(lex, DT_PREPARE))
DBUG_RETURN(TRUE);
}
if (!(result= new (thd->mem_root) multi_delete(thd, aux_tables,
lex->table_count_update)))
{
DBUG_RETURN(TRUE);
}
table_list->delete_while_scanning= true;
if (!multitable && !table_list->single_table_updatable())
{
my_error(ER_NON_UPDATABLE_TABLE, MYF(0), table_list->alias.str, "DELETE");
DBUG_RETURN(TRUE);
}
if (!multitable && (!table_list->table || !table_list->table->is_created()))
{
my_error(ER_VIEW_DELETE_MERGE_VIEW, MYF(0),
table_list->view_db.str, table_list->view_name.str);
DBUG_RETURN(TRUE);
}
if (setup_tables_and_check_access(thd, &select_lex->context,
&select_lex->top_join_list,
table_list, select_lex->leaf_tables,
false, DELETE_ACL, SELECT_ACL, true))
DBUG_RETURN(TRUE);
if (setup_tables(thd, &select_lex->context, &select_lex->top_join_list,
table_list, select_lex->leaf_tables, false, false))
DBUG_RETURN(TRUE);
if (!multitable)
{
if (select_lex->index_hints || table_list->index_hints)
{
if (!processing_as_multitable_delete_prohibited(thd))
{
multitable= true;
}
else
{
push_warning_printf(thd, Sql_condition::WARN_LEVEL_WARN,
WARN_INDEX_HINTS_IGNORED,
"%s", ER_THD(thd, WARN_INDEX_HINTS_IGNORED));
}
}
if (table_list->vers_conditions.is_set() && table_list->is_view_or_derived())
{
my_error(ER_IT_IS_A_VIEW, MYF(0), table_list->table_name.str);
DBUG_RETURN(true);
}
if (!multitable)
{
TABLE_LIST *update_source_table= 0;
if (((update_source_table=unique_table(thd, table_list,
table_list->next_global, 0)) ||
table_list->is_multitable()))
{
DBUG_ASSERT(update_source_table || table_list->view != 0);
if (!table_list->is_multitable() &&
!processing_as_multitable_delete_prohibited(thd))
{
multitable= true;
remove_order_by_without_limit(thd);
}
}
}
if (table_list->has_period())
{
if (table_list->is_view_or_derived())
{
my_error(ER_IT_IS_A_VIEW, MYF(0), table_list->table_name.str);
DBUG_RETURN(true);
}
if (select_lex->period_setup_conds(thd, table_list))
DBUG_RETURN(true);
}
if (select_lex->vers_setup_conds(thd, table_list))
DBUG_RETURN(TRUE);
/*
Application-time periods: if FOR PORTION OF ... syntax used, DELETE
statement could issue delete_row's mixed with write_row's. This causes
problems for myisam and corrupts table, if deleting while scanning.
*/
if (table_list->has_period()
|| unique_table(thd, table_list, table_list->next_global, 0))
table_list->delete_while_scanning= false;
}
{
if (thd->lex->describe)
select_options|= SELECT_DESCRIBE;
/*
When in EXPLAIN, delay deleting the joins so that they are still
available when we're producing EXPLAIN EXTENDED warning text.
*/
if (select_options & SELECT_DESCRIBE)
free_join= 0;
select_options|=
SELECT_NO_JOIN_CACHE | SELECT_NO_UNLOCK | OPTION_SETUP_TABLES_DONE;
if (!(join= new (thd->mem_root) JOIN(thd, empty_list,
select_options, result)))
DBUG_RETURN(TRUE);
THD_STAGE_INFO(thd, stage_init);
select_lex->join= join;
thd->lex->used_tables=0;
if ((err= join->prepare(table_list, select_lex->where,
select_lex->order_list.elements,
select_lex->order_list.first,
false, NULL, NULL, NULL,
select_lex, &lex->unit)))
{
goto err;
}
if (!multitable &&
select_lex->sj_subselects.elements)
multitable= true;
}
if (multitable)
{
/*
Multi-delete can't be constructed over-union => we always have
single SELECT on top and have to check underlying SELECTs of it
*/
lex->first_select_lex()->set_unique_exclude();
/* Fix tables-to-be-deleted-from list to point at opened tables */
for (target_tbl= (TABLE_LIST*) aux_tables;
target_tbl;
target_tbl= target_tbl->next_local)
{
target_tbl->table= target_tbl->correspondent_table->table;
if (target_tbl->correspondent_table->is_multitable())
{
my_error(ER_VIEW_DELETE_MERGE_VIEW, MYF(0),
target_tbl->correspondent_table->view_db.str,
target_tbl->correspondent_table->view_name.str);
DBUG_RETURN(TRUE);
}
if (!target_tbl->correspondent_table->single_table_updatable() ||
check_key_in_view(thd, target_tbl->correspondent_table))
{
my_error(ER_NON_UPDATABLE_TABLE, MYF(0),
target_tbl->table_name.str, "DELETE");
DBUG_RETURN(TRUE);
}
}
/*
Reset the exclude flag to false so it doesn't interfere
with further calls to unique_table
*/
lex->first_select_lex()->exclude_from_table_unique_test= FALSE;
}
if (!multitable && table_list->has_period())
{
if (!table_list->period_conditions.start.item->const_item()
|| !table_list->period_conditions.end.item->const_item())
{
my_error(ER_NOT_CONSTANT_EXPRESSION, MYF(0), "FOR PORTION OF");
DBUG_RETURN(true);
}
}
if (delete_history)
table_list->table->vers_write= false;
if (setup_returning_fields(thd, table_list) ||
setup_ftfuncs(select_lex))
goto err;
free_join= false;
if (returning)
(void) result->prepare(returning->returning_list, NULL);
err:
if (free_join)
{
THD_STAGE_INFO(thd, stage_end);
err|= (int)(select_lex->cleanup());
DBUG_RETURN(err || thd->is_error());
}
DBUG_RETURN(err);
}
/**
@brief Perform optimization and execution actions needed for deletes
@param thd global context the processed statement
@returns false on success, true on error
*/
bool Sql_cmd_delete::execute_inner(THD *thd)
{
Running_stmt_guard guard(thd, active_dml_stmt::DELETING_STMT);
if (!multitable)
{
if (lex->has_returning())
{
select_result *sel_result= NULL;
delete result;
/* This is DELETE ... RETURNING. It will return output to the client */
if (thd->lex->analyze_stmt)
{
/*
Actually, it is ANALYZE .. DELETE .. RETURNING. We need to produce
output and then discard it.
*/
sel_result= new (thd->mem_root) select_send_analyze(thd);
save_protocol= thd->protocol;
thd->protocol= new Protocol_discard(thd);
}
else
{
if (!lex->result && !(sel_result= new (thd->mem_root) select_send(thd)))
return true;
}
result= lex->result ? lex->result : sel_result;
}
}
bool res= multitable ? Sql_cmd_dml::execute_inner(thd)
: delete_from_single_table(thd);
res|= thd->is_error();
if (save_protocol)
{
delete thd->protocol;
thd->protocol= save_protocol;
}
{
if (unlikely(res))
{
if (multitable)
result->abort_result_set();
}
else
{
if (thd->lex->describe || thd->lex->analyze_stmt)
{
bool extended= thd->lex->describe & DESCRIBE_EXTENDED;
res= thd->lex->explain->send_explain(thd, extended);
}
}
}
if (result)
{
/* In single table case, this->deleted set by delete_from_single_table */
if (res && multitable)
deleted= ((multi_delete*)get_result())->num_deleted();
res= false;
delete result;
}
status_var_add(thd->status_var.rows_sent, thd->get_sent_row_count());
return res;
}
void Multidelete_prelocking_strategy::reset(THD *thd)
{
done= false;
}
/**
Call setup_tables() to populate lex->first_select_lex()->leaf_tables.
This is needed to properly cleanup the tables used in DELETE's top-level
select. See st_select_lex::cleanup().
*/
bool Multidelete_prelocking_strategy::handle_end(THD *thd)
{
DBUG_ENTER("Multidelete_prelocking_strategy::handle_end");
if (done)
DBUG_RETURN(0);
LEX *lex= thd->lex;
SELECT_LEX *select_lex= lex->first_select_lex();
TABLE_LIST *table_list= lex->query_tables;
done= true;
/*
This is done to resolve other base tables within derived tables in the
outermost select.
*/
if (mysql_handle_derived(lex, DT_INIT) ||
mysql_handle_derived(lex, DT_MERGE_FOR_INSERT) ||
mysql_handle_derived(lex, DT_PREPARE))
DBUG_RETURN(1);
if (setup_tables(thd, &select_lex->context, &select_lex->top_join_list,
table_list, select_lex->leaf_tables, false, true))
DBUG_RETURN(1);
DBUG_RETURN(0);
}
View Git Blame Copy Permalink