Slave_open_temp_tables fails to decrement on the slave w/ binlog off if master is killed

The fix [1] has been merged to PS 5.6 incorrectly: the part which decrements Slave_open_temp_tables on a slave with binlog off has been dropped. Diff of 5.6 trunk vs. PS 5.6 trunk in question is [2].

To repeat, add mysql-test/suite/rpl/t/rpl_create_drop_temp_table-slave.opt with "--skip-log-bin", and "--let $rpl_skip_reset_master_and_slave= 1" before --source include/master-slave.inc in the testcase script itself.

[1]

commit e1f5cb000cf42e337b7d9aa3e695b315950fcd03
Author: Venkatesh Duggirala <email address hidden>
Date: Tue Mar 18 18:40:47 2014 +0530

    Bug#18364070: Backporting Bug#18236612 to Mysql-5.6

    Problem: When Slave SQL thread detects that Master was restarted
    with the help of information sent by Master through 'FormatDescription'
    event, slave SQL drops all the opened temporary tables inorder to have
    proper cleanup. While slave SQL thread is dropping the temporary tables,
    it is not decrementing Slave_open_temp_tables count.

    Fix: Set slave_open_temp_tables=0 in close_temporary_tables(thd).

[2]

diff --git a/sql/sql_base.cc b/sql/sql_base.cc
index be8ed86..70dd46f 100644
--- a/sql/sql_base.cc
+++ b/sql/sql_base.cc
@@ -1572,8 +1581,7 @@ bool close_temporary_tables(THD *thd)
       close_temporary(t, 1, 1);
     }
     thd->temporary_tables= 0;
- if (thd->slave_thread)
- modify_slave_open_temp_tables(thd, -slave_open_temp_tables);
+ mysql_mutex_unlock(&thd->LOCK_temporary_tables);

     DBUG_RETURN(FALSE);
   }