Bug #79145 Memory Leak in Ha_innopart_share::set_v_templ partitioned on virtual column

Description:
valgrind shows memory leak for subquery scan on partitioned table with virtual columns.

1,088 (176 direct, 912 indirect) bytes in 2 blocks are definitely lost in loss record 7 of 7
   at 0x4C2AB80: malloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
   by 0x18895CC: ut_allocator<unsigned char>::allocate(unsigned long, unsigned char const*, char const*, bool, bool) (ut0new.h:342)
   by 0x1894A17: Ha_innopart_share::set_v_templ(TABLE*, dict_table_t*, char const*) (ha_innopart.cc:237)
   by 0x1896911: ha_innopart::open(char const*, int, unsigned int) (ha_innopart.cc:1125)
   by 0xECBC72: handler::ha_open(TABLE*, char const*, int, int) (handler.cc:2629)
   by 0x15E320A: open_table_from_share(THD*, TABLE_SHARE*, char const*, unsigned int, unsigned int, unsigned int, TABLE*, bool) (table.cc:3300)
   by 0x1467E67: open_table(THD*, TABLE_LIST*, Open_table_context*) (sql_base.cc:3454)

304 bytes in 2 blocks are indirectly lost in loss record 6 of 7
   at 0x4C2AB80: malloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
   by 0x18895CC: ut_allocator<unsigned char>::allocate(unsigned long, unsigned char const*, char const*, bool, bool) (ut0new.h:342)
   by 0x186940B: innobase_build_v_templ(TABLE const*, dict_table_t const*, dict_vcol_templ_t*, dict_add_v_col_t const*, bool, char const*) (ha_innodb.cc:5157)
   by 0x1894B08: Ha_innopart_share::set_v_templ(TABLE*, dict_table_t*, char const*) (ha_innopart.cc:245)
   by 0x1896911: ha_innopart::open(char const*, int, unsigned int) (ha_innopart.cc:1125)
   by 0xECBC72: handler::ha_open(TABLE*, char const*, int, int) (handler.cc:2629)
   by 0x15E320A: open_table_from_share(THD*, TABLE_SHARE*, char const*, unsigned int, unsigned int, unsigned int, TABLE*, bool) (table.cc:3300)

How to repeat:
commit 4903997455887c434c6a06a6a6a8b4b24170d159
Author: Jimmy Yang <jimmy.yang@oracle.com>
Date:   Thu Nov 5 11:29:48 2015 +0800

create table t (
  a date,
  b blob,
  c tinyint,
  d longblob generated always as (c) virtual,
  e float,
  key i0001 (a),
  key i0000 (d(11))
) engine=innodb partition by key (c,a,e) partitions 2;
insert into t (a, b, c, e) values ('2015-10-13','a',-128,0);
insert into t (a, b, c, e) values ('2015-01-03','a',-128,0);
insert into t (a, b, c, e) values ('2015-09-14','a',0,0);
insert into t (a, b, c, e) values ('2015-07-19','a',0,0);
insert into t (a, b, c, e) values ('2014-10-30','a',-43,-32768);
insert into t (a, b, c, e) values ('2016-05-08','a',0,-1);
insert into t (a, b, c, e) values ('2015-05-04','a',43,0);
insert into t (a, b, c, e) values ('2015-06-03','a',1,1);
select d in(select d from t)from t group by d;

1. create a test file sample.test with above sql script.
2. ./mtr --valgrind sample

Documented fix in the MySQL 5.7.11 changelog as follows:

    Subquery scans on partitioned table with virtual columns could
    leak memory.

Closed.