MDEV-7487 Semi-join optimization for single-table update/delete statements

This patch allows to use semi-join optimization at the top level of
single-table update and delete statements.
The problem of supporting such optimization became easy to resolve after
processing a single-table update/delete statement started using JOIN
structure. This allowed to use JOIN::prepare() not only for multi-table
updates/deletes but for single-table ones as well. This was done in the
patch for mdev-28883:
Re-design the upper level of handling UPDATE and DELETE statements.

Note that JOIN::prepare() detects all subqueries that can be considered
as candidates for semi-join optimization. The code added by this patch
looks for such candidates at the top level and if such candidates are found
in the processed single-table update/delete the statement is handled in
the same way as a multi-table update/delete.

    Approved by Oleksandr Byelkin <sanja@mariadb.com>

mysql-test/main/log_state.result

@@ -243,7 +243,7 @@ rows_examined	sql_text
 4	UPDATE t1 SET a=a+sleep(.02) WHERE a>2
 8	UPDATE t1 SET a=a+sleep(.02) ORDER BY a DESC
 1	UPDATE t2 set b=b+sleep(.02) limit 1
-4	UPDATE t1 SET a=a+sleep(.02) WHERE a in (SELECT b from t2)
+10	UPDATE t1 SET a=a+sleep(.02) WHERE a in (SELECT b from t2)
 6	DELETE FROM t1 WHERE a=a+sleep(.02) ORDER BY a LIMIT 2
 disconnect con2;
 connection default;

mysql-test/main/myisam_explain_non_select_all.result

@@ -218,14 +218,16 @@ INSERT INTO t2 VALUES (1), (2), (3);
 #
 EXPLAIN UPDATE        t1 SET a = 10 WHERE 1 IN (SELECT 1 FROM t2 WHERE t2.b < 3);
 id	select_type	table	type	possible_keys	key	key_len	ref	rows	Extra
-1	PRIMARY	t1	ALL	NULL	NULL	NULL	NULL	3	Using where
-2	SUBQUERY	t2	ALL	NULL	NULL	NULL	NULL	3	Using where
+1	PRIMARY	<subquery2>	eq_ref	distinct_key	distinct_key	4	func	1	
+1	PRIMARY	t1	ALL	NULL	NULL	NULL	NULL	3	
+2	MATERIALIZED	t2	ALL	NULL	NULL	NULL	NULL	3	Using where
 FLUSH STATUS;
 FLUSH TABLES;
 EXPLAIN EXTENDED UPDATE        t1 SET a = 10 WHERE 1 IN (SELECT 1 FROM t2 WHERE t2.b < 3);
 id	select_type	table	type	possible_keys	key	key_len	ref	rows	filtered	Extra
-1	PRIMARY	t1	ALL	NULL	NULL	NULL	NULL	3	100.00	Using where
-2	SUBQUERY	t2	ALL	NULL	NULL	NULL	NULL	3	100.00	Using where
+1	PRIMARY	<subquery2>	eq_ref	distinct_key	distinct_key	4	func	1	100.00	
+1	PRIMARY	t1	ALL	NULL	NULL	NULL	NULL	3	100.00	
+2	MATERIALIZED	t2	ALL	NULL	NULL	NULL	NULL	3	100.00	Using where
 # Status of EXPLAIN EXTENDED query
 Variable_name	Value
 Handler_read_key	4
@@ -246,8 +248,9 @@ Handler_read_key	4
 Handler_read_rnd_next	5
 # Status of testing query execution:
 Variable_name	Value
-Handler_read_key	4
-Handler_read_rnd_next	5
+Handler_read_key	5
+Handler_read_rnd	3
+Handler_read_rnd_next	12
 Handler_update	3
 
 DROP TABLE t1, t2;
@@ -263,13 +266,13 @@ INSERT INTO t2 VALUES (1), (2), (3);
 EXPLAIN UPDATE        t1 SET a = 10 WHERE a IN (SELECT b FROM t2 WHERE t1.a < 3);
 id	select_type	table	type	possible_keys	key	key_len	ref	rows	Extra
 1	PRIMARY	t1	ALL	NULL	NULL	NULL	NULL	3	Using where
-2	DEPENDENT SUBQUERY	t2	ALL	NULL	NULL	NULL	NULL	3	Using where
+1	PRIMARY	t2	ALL	NULL	NULL	NULL	NULL	3	Using where; FirstMatch(t1)
 FLUSH STATUS;
 FLUSH TABLES;
 EXPLAIN EXTENDED UPDATE        t1 SET a = 10 WHERE a IN (SELECT b FROM t2 WHERE t1.a < 3);
 id	select_type	table	type	possible_keys	key	key_len	ref	rows	filtered	Extra
 1	PRIMARY	t1	ALL	NULL	NULL	NULL	NULL	3	100.00	Using where
-2	DEPENDENT SUBQUERY	t2	ALL	NULL	NULL	NULL	NULL	3	100.00	Using where
+1	PRIMARY	t2	ALL	NULL	NULL	NULL	NULL	3	100.00	Using where; FirstMatch(t1)
 Warnings:
 Note	1276	Field or reference 'test.t1.a' of SELECT #2 was resolved in SELECT #1
 # Status of EXPLAIN EXTENDED query
@@ -900,14 +903,16 @@ INSERT INTO t2 VALUES (1), (2), (3), (1000);
 #
 EXPLAIN UPDATE t1 SET a = 10 WHERE a IN (SELECT a FROM t2);
 id	select_type	table	type	possible_keys	key	key_len	ref	rows	Extra
-1	PRIMARY	t1	ALL	NULL	NULL	NULL	NULL	3	Using where
-2	DEPENDENT SUBQUERY	t2	ALL	NULL	NULL	NULL	NULL	4	Using where
+1	PRIMARY	t1	ALL	NULL	NULL	NULL	NULL	3	
+1	PRIMARY	<subquery2>	eq_ref	distinct_key	distinct_key	4	func	1	
+2	MATERIALIZED	t2	ALL	NULL	NULL	NULL	NULL	3	
 FLUSH STATUS;
 FLUSH TABLES;
 EXPLAIN EXTENDED UPDATE t1 SET a = 10 WHERE a IN (SELECT a FROM t2);
 id	select_type	table	type	possible_keys	key	key_len	ref	rows	filtered	Extra
-1	PRIMARY	t1	ALL	NULL	NULL	NULL	NULL	3	100.00	Using where
-2	DEPENDENT SUBQUERY	t2	ALL	NULL	NULL	NULL	NULL	4	100.00	Using where
+1	PRIMARY	t1	ALL	NULL	NULL	NULL	NULL	3	100.00	
+1	PRIMARY	<subquery2>	eq_ref	distinct_key	distinct_key	4	func	1	100.00	
+2	MATERIALIZED	t2	ALL	NULL	NULL	NULL	NULL	3	100.00	
 # Status of EXPLAIN EXTENDED query
 Variable_name	Value
 Handler_read_key	4
@@ -928,8 +933,8 @@ Handler_read_key	4
 Handler_read_rnd_next	9
 # Status of testing query execution:
 Variable_name	Value
-Handler_read_key	4
-Handler_read_rnd_next	10
+Handler_read_key	7
+Handler_read_rnd_next	8
 Handler_update	3
 
 DROP TABLE t1, t2;
@@ -986,14 +991,14 @@ INSERT INTO t1 VALUES (1), (2), (3), (4), (5);
 #
 EXPLAIN DELETE   FROM t1 WHERE a1 IN (SELECT a2 FROM t2 WHERE a2 > 2);
 id	select_type	table	type	possible_keys	key	key_len	ref	rows	Extra
-1	PRIMARY	t1	ALL	NULL	NULL	NULL	NULL	5	Using where
-2	DEPENDENT SUBQUERY	t2	ALL	NULL	NULL	NULL	NULL	5	Using where
+1	PRIMARY	t1	ALL	NULL	NULL	NULL	NULL	5	
+1	PRIMARY	t2	ALL	NULL	NULL	NULL	NULL	5	Using where; FirstMatch(t1)
 FLUSH STATUS;
 FLUSH TABLES;
 EXPLAIN EXTENDED DELETE   FROM t1 WHERE a1 IN (SELECT a2 FROM t2 WHERE a2 > 2);
 id	select_type	table	type	possible_keys	key	key_len	ref	rows	filtered	Extra
-1	PRIMARY	t1	ALL	NULL	NULL	NULL	NULL	5	100.00	Using where
-2	DEPENDENT SUBQUERY	t2	ALL	NULL	NULL	NULL	NULL	5	100.00	Using where
+1	PRIMARY	t1	ALL	NULL	NULL	NULL	NULL	5	100.00	
+1	PRIMARY	t2	ALL	NULL	NULL	NULL	NULL	5	100.00	Using where; FirstMatch(t1)
 # Status of EXPLAIN EXTENDED query
 Variable_name	Value
 Handler_read_key	4
@@ -2823,14 +2828,14 @@ INSERT INTO t2 VALUES (1), (2), (3);
 EXPLAIN UPDATE        t1 SET a = 10 WHERE a IN (SELECT * FROM (SELECT b FROM t2 ORDER BY b LIMIT 2,2) x);
 id	select_type	table	type	possible_keys	key	key_len	ref	rows	Extra
 1	PRIMARY	t1	ALL	NULL	NULL	NULL	NULL	3	Using where
-2	DEPENDENT SUBQUERY	<derived3>	unique_subquery	distinct_key	distinct_key	5	func	1	
+1	PRIMARY	<derived3>	ref	key0	key0	5	test.t1.a	2	FirstMatch(t1)
 3	DERIVED	t2	ALL	NULL	NULL	NULL	NULL	3	Using filesort
 FLUSH STATUS;
 FLUSH TABLES;
 EXPLAIN EXTENDED UPDATE        t1 SET a = 10 WHERE a IN (SELECT * FROM (SELECT b FROM t2 ORDER BY b LIMIT 2,2) x);
 id	select_type	table	type	possible_keys	key	key_len	ref	rows	filtered	Extra
 1	PRIMARY	t1	ALL	NULL	NULL	NULL	NULL	3	100.00	Using where
-2	DEPENDENT SUBQUERY	<derived3>	unique_subquery	distinct_key	distinct_key	5	func	1	100.00	
+1	PRIMARY	<derived3>	ref	key0	key0	5	test.t1.a	2	100.00	FirstMatch(t1)
 3	DERIVED	t2	ALL	NULL	NULL	NULL	NULL	3	100.00	Using filesort
 # Status of EXPLAIN EXTENDED query
 Variable_name	Value

mysql-test/main/opt_trace.result

@@ -4466,6 +4466,16 @@ explain delete t0,t1 from t0, t1 where t0.a=t1.a and t1.a<3	{
                   }
                 }
               },
+              {
+                "table": "t0",
+                "rowid_filters": [
+                  {
+                    "key": "a",
+                    "build_cost": 0.174715752,
+                    "rows": 3
+                  }
+                ]
+              },
               {
                 "selectivity_for_indexes": [
                   {
@@ -4531,6 +4541,16 @@ explain delete t0,t1 from t0, t1 where t0.a=t1.a and t1.a<3	{
                   }
                 }
               },
+              {
+                "table": "t1",
+                "rowid_filters": [
+                  {
+                    "key": "a",
+                    "build_cost": 0.174715752,
+                    "rows": 3
+                  }
+                ]
+              },
               {
                 "selectivity_for_indexes": [
                   {

mysql-test/main/opt_trace_security.test

@@ -20,9 +20,9 @@ delimiter ;|
 
 --change_user foo
 set optimizer_trace="enabled=on";
-# --error 1142
-# select * from db1.t1;
-# select * from information_schema.OPTIMIZER_TRACE;
+--error 1142
+select * from db1.t1;
+select * from information_schema.OPTIMIZER_TRACE;
 set optimizer_trace="enabled=off";
 
 --change_user root

sql/opt_subselect.cc

@@ -567,11 +567,11 @@ bool SELECT_LEX::is_sj_conversion_prohibited(THD *thd)
   switch (thd->lex->sql_command) {
   case SQLCOM_UPDATE:
     return
-      !((Sql_cmd_update *) cmd)->is_multitable() ||
+      !((Sql_cmd_update *) cmd)->is_multitable() &&
       ((Sql_cmd_update *) cmd)->processing_as_multitable_update_prohibited(thd);
   case SQLCOM_DELETE:
     return
-      !((Sql_cmd_delete *) cmd)->is_multitable() ||
+      !((Sql_cmd_delete *) cmd)->is_multitable() &&
       ((Sql_cmd_delete *) cmd)->processing_as_multitable_delete_prohibited(thd);
   default:
     return false;

sql/opt_trace.cc

@@ -491,8 +491,7 @@ void Opt_trace_start::init(THD *thd,
       !list_has_optimizer_trace_table(tbl) &&
       !sets_var_optimizer_trace(sql_command, set_vars) &&
       !thd->system_thread &&
-      !ctx->disable_tracing_if_required() &&
-      !(thd->lex->context_analysis_only & CONTEXT_ANALYSIS_ONLY_PREPARE))
+      !ctx->disable_tracing_if_required())
   {
     ctx->start(thd, tbl, sql_command, query, query_length, query_charset,
                thd->variables.optimizer_trace_max_mem_size);

sql/sql_base.cc

@@ -1214,7 +1214,7 @@ TABLE_LIST* find_dup_table(THD *thd, TABLE_LIST *table, TABLE_LIST *table_list,
         }
       }
       else if (thd->lex->sql_command == SQLCOM_DELETE)
-      {
+     {
         Sql_cmd_delete *cmd= (Sql_cmd_delete *) (thd->lex->m_sql_cmd);
         if (cmd->is_multitable() || derived->derived->outer_select())
           materialize= false;

sql/sql_delete.cc

@@ -348,7 +348,6 @@ bool Sql_cmd_delete::delete_from_single_table(THD *thd)
   query_plan.using_filesort= FALSE;
 
   THD_STAGE_INFO(thd, stage_init_update);
-  create_explain_query(thd->lex, thd->mem_root);
 
   const bool delete_history= table_list->vers_conditions.delete_history;
   DBUG_ASSERT(!(delete_history && table_list->period_conditions.is_set()));
@@ -1668,6 +1667,10 @@ bool Sql_cmd_delete::prepare_inner(THD *thd)
     {
       goto err;
     }
+
+    if (!multitable &&
+        select_lex->sj_subselects.elements)
+      multitable= true;
   }
 
   if (multitable)

sql/sql_select.cc

@@ -5972,8 +5972,7 @@ make_join_statistics(JOIN *join, List<TABLE_LIST> &tables_list,
           s->needed_reg=select->needed_reg;
           select->quick=0;
           impossible_range= records == 0 && s->table->reginfo.impossible_range;
-          if (join->thd->lex->sql_command == SQLCOM_SELECT &&
-              optimizer_flag(join->thd, OPTIMIZER_SWITCH_USE_ROWID_FILTER))
+          if (optimizer_flag(join->thd, OPTIMIZER_SWITCH_USE_ROWID_FILTER))
             s->table->init_cost_info_for_usable_range_rowid_filters(join->thd);
         }
         if (!impossible_range)

sql/sql_update.cc

@@ -382,7 +382,6 @@ bool Sql_cmd_update::update_single_table(THD *thd)
   DBUG_ENTER("Sql_cmd_update::update_single_table");
 
   THD_STAGE_INFO(thd, stage_init_update);
-  create_explain_query(thd->lex, thd->mem_root);
 
   thd->table_map_for_update= 0;
 
@@ -2479,6 +2478,8 @@ int multi_update::do_updates()
     table = cur_table->table;
     if (table == table_to_update)
       continue;					// Already updated
+    if (table->file->pushed_rowid_filter)
+      table->file->disable_pushed_rowid_filter();
     org_updated= updated;
     tmp_table= tmp_tables[cur_table->shared];
     tmp_table->file->extra(HA_EXTRA_CACHE);	// Change to read cache
@@ -2673,7 +2674,8 @@ int multi_update::do_updates()
     check_opt_it.rewind();
     while (TABLE *tbl= check_opt_it++)
         tbl->file->ha_rnd_end();
-
+    if (table->file->save_pushed_rowid_filter)
+      table->file->enable_pushed_rowid_filter();
   }
   DBUG_RETURN(0);
 
@@ -2684,6 +2686,8 @@ int multi_update::do_updates()
   }
 
 err2:
+  if (table->file->save_pushed_rowid_filter)
+    table->file->enable_pushed_rowid_filter();
   if (table->file->inited)
     (void) table->file->ha_rnd_end();
   if (tmp_table->file->inited)
@@ -2987,7 +2991,9 @@ bool Sql_cmd_update::prepare_inner(THD *thd)
     {
       goto err;
     }
-
+    if (!multitable &&
+        select_lex->sj_subselects.elements)
+      multitable= true;
   }
 
   if (table_list->has_period())