Proposed fix for bug#24491 "using alias from source table in insert ...

on duplicate key".

INSERT ... SELECT ... ON DUPLICATE KEY UPDATE which was used in
stored routine or as prepared statement and which in its ON DUPLICATE
KEY clause erroneously tried to assign value to a column mentioned only
in its SELECT part was properly emitting error on the first execution
but succeeded on the second and following executions.

Code which is responsible for name resolution of fields mentioned in
UPDATE clause (e.g. see select_insert::prepare()) modifies table list
and Name_resolution_context used in this process. It uses
Name_resolution_context_state::save_state/restore_state() to revert
these modifications. Unfortunately those two methods failed to revert
properly modifications to TABLE_LIST::next_name_resolution_table
and this broke name resolution process for successive executions.

This patch fixes Name_resolution_context_state::save_state/restore_state()
in such way that it properly handles TABLE_LIST::next_name_resolution_table.


mysql-test/r/ps.result:
  Added test case for bug#24491 "using alias from source table in insert ...
  on duplicate key"
mysql-test/r/sp-error.result:
  Added test case for bug#24491 "using alias from source table in insert ...
  on duplicate key"
mysql-test/t/ps.test:
  Added test case for bug#24491 "using alias from source table in insert ...
  on duplicate key"
mysql-test/t/sp-error.test:
  Added test case for bug#24491 "using alias from source table in insert ...
  on duplicate key"
sql/item.h:
  Name_resolution_context::save_state/restore_state():
    At the moment these methods are used only by code implementing
    INSERT and INSERT ... SELECT statements. This code doesn't modify
   'next_name_resolution_table' member of table list element
    corresponding to the first table of SELECT clause (pointed by
    'first_name_resolution_table'). But it modifies table list element
    corresponding to the target table of INSERT (pointed by 'table_list')
    So these methods were changed to reflect this.

1 files changed, 30 insertions, 0 deletions

diff --git a/mysql-test/t/ps.test b/mysql-test/t/ps.test
index 1c8d7b419a3..2059f2fb080 100644
--- a/mysql-test/t/ps.test
+++ b/mysql-test/t/ps.test
@@ -1540,4 +1540,34 @@ execute sq;
 deallocate prepare no_index;
 deallocate prepare sq;
 
+
+#
+# BUG#24491 "using alias from source table in insert ... on duplicate key"
+#
+--disable_warnings
+drop tables if exists t1;
+--enable_warnings
+create table t1 (id int primary key auto_increment, value varchar(10));
+insert into t1 (id, value) values (1, 'FIRST'), (2, 'SECOND'), (3, 'THIRD');
+# Let us prepare INSERT ... SELECT ... ON DUPLICATE KEY UPDATE statement
+# which in its ON DUPLICATE KEY clause erroneously tries to assign value
+# to a column which is mentioned only in SELECT part.
+prepare stmt from "insert into t1 (id, value) select * from (select 4 as i, 'FOURTH' as v) as y on duplicate key update v = 'DUP'";
+# Both first and second attempts to execute it should fail
+--error ER_BAD_FIELD_ERROR 
+execute stmt;
+--error ER_BAD_FIELD_ERROR
+execute stmt;
+deallocate prepare stmt;
+# And now the same test for more complex case which is more close
+# to the one that was reported originally.
+prepare stmt from "insert into t1 (id, value) select * from (select 4 as id, 'FOURTH' as value) as y on duplicate key update y.value = 'DUP'";
+--error ER_BAD_FIELD_ERROR 
+execute stmt;
+--error ER_BAD_FIELD_ERROR
+execute stmt;
+deallocate prepare stmt;
+drop tables t1;
+
+
 --echo End of 5.0 tests.