Bug #46175: NULL read_view and consistent read assertion

The SE API requires mysql to notify the storage engine that
it's going to read certain tables at the beginning of the 
statement (by calling start_stmt(), store_lock() or
external_lock()).
These are typically called by the lock_tables(). 
However SHOW CREATE TABLE is not pre-locking the tables
because it's not expected to access the data at all.
But for some view definitions (that include comparing a
date/datetime/timestamp column to a string returning
scalar subquery) the JOIN::prepare may still access data
when materializing the scalar non-correlated subquery
in Arg_comparator::can_compare_as_dates().
Fixed by not materializing the subquery when the function
is called in a SHOW/EXPLAIN/CREATE VIEW

mysql-test/r/innodb_mysql.result

@@ -2251,4 +2251,26 @@ c >= '2009-10-09 00:00:00.001' AND c <= '2009-10-09 00:00:00.00';
 id	select_type	table	type	possible_keys	key	key_len	ref	rows	Extra
 1	SIMPLE	NULL	NULL	NULL	NULL	NULL	NULL	NULL	Impossible WHERE noticed after reading const tables
 DROP TABLE t1;
+#
+# Bug #46175: NULL read_view and consistent read assertion
+#
+CREATE TABLE t1(a CHAR(13),KEY(a)) ENGINE=innodb;
+CREATE TABLE t2(b DATETIME,KEY(b)) ENGINE=innodb;
+INSERT INTO t1 VALUES (),();
+INSERT INTO t2 VALUES (),();
+CREATE OR REPLACE VIEW v1 AS SELECT 1 FROM t2 
+WHERE b =(SELECT a FROM t1 LIMIT 1);
+CREATE PROCEDURE p1(num INT)
+BEGIN
+DECLARE i INT DEFAULT 0;
+REPEAT
+SHOW CREATE VIEW v1;
+SET i:=i+1;
+UNTIL i>num END REPEAT;
+END|
+# Should not crash
+# Should not crash
+DROP PROCEDURE p1;
+DROP VIEW v1;
+DROP TABLE t1,t2;
 End of 5.1 tests

mysql-test/t/innodb_mysql.test

@@ -489,5 +489,51 @@ EXPLAIN SELECT * FROM t1 WHERE a = 'TEST' AND
   c >= '2009-10-09 00:00:00.001' AND c <= '2009-10-09 00:00:00.00';
 DROP TABLE t1;
 
+--echo #
+--echo # Bug #46175: NULL read_view and consistent read assertion
+--echo #
+
+CREATE TABLE t1(a CHAR(13),KEY(a)) ENGINE=innodb;
+CREATE TABLE t2(b DATETIME,KEY(b)) ENGINE=innodb;
+INSERT INTO t1 VALUES (),();
+INSERT INTO t2 VALUES (),();
+CREATE OR REPLACE VIEW v1 AS SELECT 1 FROM t2 
+  WHERE b =(SELECT a FROM t1 LIMIT 1);
+
+--disable_query_log
+--disable_result_log
+CONNECT (con1, localhost, root,,);
+--enable_query_log
+--enable_result_log
+CONNECTION default;
+
+DELIMITER |;
+CREATE PROCEDURE p1(num INT)
+BEGIN
+  DECLARE i INT DEFAULT 0;
+  REPEAT
+    SHOW CREATE VIEW v1;
+    SET i:=i+1;
+  UNTIL i>num END REPEAT;
+END|
+DELIMITER ;|
+
+--echo # Should not crash
+--disable_query_log
+--disable_result_log
+--send CALL p1(1000)
+CONNECTION con1;
+--echo # Should not crash
+CALL p1(1000);
+
+CONNECTION default;
+--reap
+--enable_query_log
+--enable_result_log
+
+DISCONNECT con1;
+DROP PROCEDURE p1;
+DROP VIEW v1;
+DROP TABLE t1,t2;
 
 --echo End of 5.1 tests

sql/item_cmpfunc.cc

@@ -785,15 +785,21 @@ Arg_comparator::can_compare_as_dates(Item *a, Item *b, ulonglong *const_value)
 
   if (cmp_type != CMP_DATE_DFLT)
   {
+    THD *thd= current_thd;
     /*
       Do not cache GET_USER_VAR() function as its const_item() may return TRUE
       for the current thread but it still may change during the execution.
+      Don't use cache while in the context analysis mode only (i.e. for 
+      EXPLAIN/CREATE VIEW and similar queries). Cache is useless in such 
+      cases and can cause problems. For example evaluating subqueries can 
+      confuse storage engines since in context analysis mode tables 
+      aren't locked.
     */
-    if (cmp_type != CMP_DATE_WITH_DATE && str_arg->const_item() &&
+    if (!thd->is_context_analysis_only() &&
+        cmp_type != CMP_DATE_WITH_DATE && str_arg->const_item() &&
         (str_arg->type() != Item::FUNC_ITEM ||
         ((Item_func*)str_arg)->functype() != Item_func::GUSERVAR_FUNC))
     {
-      THD *thd= current_thd;
       ulonglong value;
       bool error;
       String tmp, *str_val= 0;