[MDEV-23356] InnoDB: Failing assertion: field->col->mtype == type, crash or ASAN failures in row_sel_convert_mysql_key_to_innobase, InnoDB indexes are inconsistent after INDEX changes - Jira

Details

Type: Bug
Status: Closed (View Workflow)
Priority: Major
Resolution: Fixed
Affects Version/s: 10.4(EOL), 10.5
Fix Version/s: 10.4.16, 10.5.7
Component/s: Data Definition - Alter Table, Storage Engine - InnoDB
Labels:
- not-10.1
- not-10.2
- not-10.3
- regression

Description

--source include/have_innodb.inc

CREATE TABLE t1 (a INT, b INT, c CHAR(8), KEY ind1(c), KEY ind2(b)) ENGINE=InnoDB;

ALTER TABLE t1 RENAME INDEX ind2 TO ind3, RENAME INDEX ind1 TO ind2;

UPDATE t1 SET a = 1 WHERE c = 'foo';

# Cleanup

DROP TABLE t1;

10.5 c69520c9
2020-07-31 16:07:09 4 [ERROR] Found index ind2 whose column info does not match that of MariaDB.
2020-07-31 16:07:09 4 [ERROR] InnoDB indexes are inconsistent with what defined in .frm for table ./test/t1
2020-07-31 16:07:09 0x7ff7b8db1700 InnoDB: Assertion failure in file /data/src/10.5/storage/innobase/row/row0sel.cc line 2518
InnoDB: Failing assertion: field->col->mtype == type

#6 0x0000556da3e53f56 in ut_dbg_assertion_failed (expr=0x556da4546432 "field->col->mtype == type", file=0x556da4545d38 "/data/src/10.5/storage/innobase/row/row0sel.cc", line=2518) at /data/src/10.5/storage/innobase/ut/ut0dbg.cc:60
#7 0x0000556da3db5ac1 in row_sel_convert_mysql_key_to_innobase (tuple=0x7f147cb08d38, buf=0x7f147c0087fc "\217\217\217\217", buf_len=4, index=0x7f147c1f2098, key_ptr=0x7f147cb0d0ed " \245\217\217\217\217\217\217", key_len=9) at /data/src/10.5/storage/innobase/row/row0sel.cc:2518
#8 0x0000556da3bcd5de in ha_innobase::records_in_range (this=0x7f147c1f0f40, keynr=0, min_key=0x7f14ae07d570, max_key=0x7f14ae07d590, pages=0x7f14ae07d540) at /data/src/10.5/storage/innobase/handler/ha_innodb.cc:13979
#9 0x0000556da35feb92 in handler::multi_range_read_info_const (this=0x7f147c1f0f40, keyno=0, seq=0x7f14ae07d720, seq_init_param=0x7f14ae07d750, n_ranges_arg=0, bufsz=0x7f14ae07d624, flags=0x7f14ae07d620, cost=0x7f14ae07ddd0) at /data/src/10.5/sql/multi_range_read.cc:177
#10 0x0000556da3602880 in DsMrr_impl::dsmrr_info_const (this=0x7f147c1f1420, keyno=0, seq=0x7f14ae07d720, seq_init_param=0x7f14ae07d750, n_ranges=0, bufsz=0x7f14ae07dce8, flags=0x7f14ae07dce4, cost=0x7f14ae07ddd0) at /data/src/10.5/sql/multi_range_read.cc:1712
#11 0x0000556da3bd81d8 in ha_innobase::multi_range_read_info_const (this=0x7f147c1f0f40, keyno=0, seq=0x7f14ae07d720, seq_init_param=0x7f14ae07d750, n_ranges=0, bufsz=0x7f14ae07dce8, flags=0x7f14ae07dce4, cost=0x7f14ae07ddd0) at /data/src/10.5/storage/innobase/handler/ha_innodb.cc:20576
#12 0x0000556da390448b in check_quick_select (param=0x7f14ae07e060, idx=0, index_only=false, tree=0x7f147cb0d1a0, update_tbl_stats=true, mrr_flags=0x7f14ae07dce4, bufsize=0x7f14ae07dce8, cost=0x7f14ae07ddd0, is_ror_scan=0x7f14ae07dce2) at /data/src/10.5/sql/opt_range.cc:11159
#13 0x0000556da38fac99 in get_key_scans_params (param=0x7f14ae07e060, tree=0x7f147cb0d108, index_read_must_be_used=false, for_range_access=true, read_time=3.2000000000000002) at /data/src/10.5/sql/opt_range.cc:7443
#14 0x0000556da38ef459 in SQL_SELECT::test_quick_select (this=0x7f147c0149a8, thd=0x7f147c000b18, keys_to_use=..., prev_tables=0, limit=18446744073709551615, force_quick_range=false, ordered_output=false, remove_false_parts_of_where=false, only_single_index_range_scan=false) at /data/src/10.5/sql/opt_range.cc:2911
#15 0x0000556da352f972 in SQL_SELECT::check_quick (this=0x7f147c0149a8, thd=0x7f147c000b18, force_quick_range=false, limit=18446744073709551615) at /data/src/10.5/sql/opt_range.h:1655
#16 0x0000556da35269b9 in mysql_update (thd=0x7f147c000b18, table_list=0x7f147c0139c8, fields=..., values=..., conds=0x7f147c0144a8, order_num=0, order=0x0, limit=18446744073709551615, ignore=false, found_return=0x7f14ae07ef80, updated_return=0x7f14ae07f050) at /data/src/10.5/sql/sql_update.cc:555
#17 0x0000556da3405a77 in mysql_execute_command (thd=0x7f147c000b18) at /data/src/10.5/sql/sql_parse.cc:4400
#18 0x0000556da3411d4d in mysql_parse (thd=0x7f147c000b18, rawbuf=0x7f147c0138d0 "UPDATE t1 SET a = 1 WHERE c = 'foo'", length=35, parser_state=0x7f14ae07f520, is_com_multi=false, is_next_command=false) at /data/src/10.5/sql/sql_parse.cc:7993
#19 0x0000556da33fe046 in dispatch_command (command=COM_QUERY, thd=0x7f147c000b18, packet=0x7f147c1b0279 "UPDATE t1 SET a = 1 WHERE c = 'foo'", packet_length=35, is_com_multi=false, is_next_command=false) at /data/src/10.5/sql/sql_parse.cc:1867
#20 0x0000556da33fc77e in do_command (thd=0x7f147c000b18) at /data/src/10.5/sql/sql_parse.cc:1348
#21 0x0000556da35a207a in do_handle_one_connection (connect=0x556da6249ef8, put_in_cache=true) at /data/src/10.5/sql/sql_connect.cc:1410
#22 0x0000556da35a1de2 in handle_one_connection (arg=0x556da6249ef8) at /data/src/10.5/sql/sql_connect.cc:1312
#23 0x0000556da3ae0126 in pfs_spawn_thread (arg=0x556da6167db8) at /data/src/10.5/storage/perfschema/pfs.cc:2201
#24 0x00007f14baab44a4 in start_thread (arg=0x7f14ae080700) at pthread_create.c:456
#25 0x00007f14b8be8d0f in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:97

Reproducible the same way on debug and non-debug builds.
The ASAN build produces use-after-poison:

2020-07-31 16:05:56 4 [ERROR] Found index ind2 whose column info does not match that of MariaDB.

2020-07-31 16:05:56 4 [ERROR] InnoDB indexes are inconsistent with what defined in .frm for table ./test/t1

=================================================================

==30504==ERROR: AddressSanitizer: use-after-poison on address 0x61200005cb40 at pc 0x564333d8bcf6 bp 0x7ff894e65350 sp 0x7ff894e65348

READ of size 8 at 0x61200005cb40 thread T13

    #0 0x564333d8bcf5 in row_sel_convert_mysql_key_to_innobase(dtuple_t*, unsigned char*, unsigned long, dict_index_t*, unsigned char const*, unsigned long) /data/src/10.5/storage/innobase/row/row0sel.cc:2517

    #1 0x564333a3b878 in ha_innobase::records_in_range(unsigned int, st_key_range const*, st_key_range const*, st_page_range*) /data/src/10.5/storage/innobase/handler/ha_innodb.cc:13979

    #2 0x564332d10080 in handler::multi_range_read_info_const(unsigned int, st_range_seq_if*, void*, unsigned int, unsigned int*, unsigned int*, Cost_estimate*) /data/src/10.5/sql/multi_range_read.cc:177

    #3 0x564332d1b2ba in DsMrr_impl::dsmrr_info_const(unsigned int, st_range_seq_if*, void*, unsigned int, unsigned int*, unsigned int*, Cost_estimate*) /data/src/10.5/sql/multi_range_read.cc:1712

    #4 0x564333a51353 in ha_innobase::multi_range_read_info_const(unsigned int, st_range_seq_if*, void*, unsigned int, unsigned int*, unsigned int*, Cost_estimate*) /data/src/10.5/storage/innobase/handler/ha_innodb.cc:20576

    #5 0x5643333c7425 in check_quick_select /data/src/10.5/sql/opt_range.cc:11159

    #6 0x5643333b1899 in get_key_scans_params /data/src/10.5/sql/opt_range.cc:7443

    #7 0x564333397815 in SQL_SELECT::test_quick_select(THD*, Bitmap<64u>, unsigned long long, unsigned long long, bool, bool, bool, bool) /data/src/10.5/sql/opt_range.cc:2911

    #8 0x564332b66ee6 in SQL_SELECT::check_quick(THD*, bool, unsigned long long) /data/src/10.5/sql/opt_range.h:1655

    #9 0x564332b51db1 in mysql_update(THD*, TABLE_LIST*, List<Item>&, List<Item>&, Item*, unsigned int, st_order*, unsigned long long, bool, unsigned long long*, unsigned long long*) /data/src/10.5/sql/sql_update.cc:555

    #10 0x5643328b7372 in mysql_execute_command(THD*) /data/src/10.5/sql/sql_parse.cc:4400

    #11 0x5643328cdf97 in mysql_parse(THD*, char*, unsigned int, Parser_state*, bool, bool) /data/src/10.5/sql/sql_parse.cc:7993

    #12 0x5643328a87f7 in dispatch_command(enum_server_command, THD*, char*, unsigned int, bool, bool) /data/src/10.5/sql/sql_parse.cc:1867

    #13 0x5643328a567a in do_command(THD*) /data/src/10.5/sql/sql_parse.cc:1348

    #14 0x564332c62509 in do_handle_one_connection(CONNECT*, bool) /data/src/10.5/sql/sql_connect.cc:1410

    #15 0x564332c61f78 in handle_one_connection /data/src/10.5/sql/sql_connect.cc:1312

    #16 0x564333851220 in pfs_spawn_thread /data/src/10.5/storage/perfschema/pfs.cc:2201

    #17 0x7ff8a53c84a3 in start_thread (/lib/x86_64-linux-gnu/libpthread.so.0+0x74a3)

    #18 0x7ff8a34fcd0e in __clone (/lib/x86_64-linux-gnu/libc.so.6+0xe8d0e)

0x61200005cb40 is located 256 bytes inside of 312-byte region [0x61200005ca40,0x61200005cb78)

allocated by thread T13 here:

    #0 0x7ff8a569fd28 in malloc (/usr/lib/x86_64-linux-gnu/libasan.so.3+0xc1d28)

    #1 0x564333a6b70a in ut_allocator<unsigned char, true>::allocate(unsigned long, unsigned char const*, unsigned int, bool, bool) /data/src/10.5/storage/innobase/include/ut0new.h:374

    #2 0x564333be49a2 in mem_heap_create_block_func(mem_block_info_t*, unsigned long, char const*, unsigned int, unsigned long) /data/src/10.5/storage/innobase/mem/mem0mem.cc:277

    #3 0x5643339facc8 in mem_heap_create_func /data/src/10.5/storage/innobase/include/mem0mem.ic:375

    #4 0x564333a3b5a0 in ha_innobase::records_in_range(unsigned int, st_key_range const*, st_key_range const*, st_page_range*) /data/src/10.5/storage/innobase/handler/ha_innodb.cc:13965

    #5 0x564332d10080 in handler::multi_range_read_info_const(unsigned int, st_range_seq_if*, void*, unsigned int, unsigned int*, unsigned int*, Cost_estimate*) /data/src/10.5/sql/multi_range_read.cc:177

    #6 0x564332d1b2ba in DsMrr_impl::dsmrr_info_const(unsigned int, st_range_seq_if*, void*, unsigned int, unsigned int*, unsigned int*, Cost_estimate*) /data/src/10.5/sql/multi_range_read.cc:1712

    #7 0x564333a51353 in ha_innobase::multi_range_read_info_const(unsigned int, st_range_seq_if*, void*, unsigned int, unsigned int*, unsigned int*, Cost_estimate*) /data/src/10.5/storage/innobase/handler/ha_innodb.cc:20576

    #8 0x5643333c7425 in check_quick_select /data/src/10.5/sql/opt_range.cc:11159

    #9 0x5643333b1899 in get_key_scans_params /data/src/10.5/sql/opt_range.cc:7443

    #10 0x564333397815 in SQL_SELECT::test_quick_select(THD*, Bitmap<64u>, unsigned long long, unsigned long long, bool, bool, bool, bool) /data/src/10.5/sql/opt_range.cc:2911

    #11 0x564332b66ee6 in SQL_SELECT::check_quick(THD*, bool, unsigned long long) /data/src/10.5/sql/opt_range.h:1655

    #12 0x564332b51db1 in mysql_update(THD*, TABLE_LIST*, List<Item>&, List<Item>&, Item*, unsigned int, st_order*, unsigned long long, bool, unsigned long long*, unsigned long long*) /data/src/10.5/sql/sql_update.cc:555

    #13 0x5643328b7372 in mysql_execute_command(THD*) /data/src/10.5/sql/sql_parse.cc:4400

    #14 0x5643328cdf97 in mysql_parse(THD*, char*, unsigned int, Parser_state*, bool, bool) /data/src/10.5/sql/sql_parse.cc:7993

    #15 0x5643328a87f7 in dispatch_command(enum_server_command, THD*, char*, unsigned int, bool, bool) /data/src/10.5/sql/sql_parse.cc:1867

    #16 0x5643328a567a in do_command(THD*) /data/src/10.5/sql/sql_parse.cc:1348

    #17 0x564332c62509 in do_handle_one_connection(CONNECT*, bool) /data/src/10.5/sql/sql_connect.cc:1410

    #18 0x564332c61f78 in handle_one_connection /data/src/10.5/sql/sql_connect.cc:1312

    #19 0x564333851220 in pfs_spawn_thread /data/src/10.5/storage/perfschema/pfs.cc:2201

    #20 0x7ff8a53c84a3 in start_thread (/lib/x86_64-linux-gnu/libpthread.so.0+0x74a3)

Thread T13 created by T0 here:

    #0 0x7ff8a560ef59 in __interceptor_pthread_create (/usr/lib/x86_64-linux-gnu/libasan.so.3+0x30f59)

    #1 0x56433384c2e2 in my_thread_create /data/src/10.5/storage/perfschema/my_thread.h:38

    #2 0x56433385160f in pfs_spawn_thread_v1 /data/src/10.5/storage/perfschema/pfs.cc:2252

    #3 0x5643325edc67 in inline_mysql_thread_create /data/src/10.5/include/mysql/psi/mysql_thread.h:1321

    #4 0x564332600a79 in create_thread_to_handle_connection(CONNECT*) /data/src/10.5/sql/mysqld.cc:6020

    #5 0x564332600fee in create_new_thread(CONNECT*) /data/src/10.5/sql/mysqld.cc:6079

    #6 0x5643326011ab in handle_accepted_socket(st_mysql_socket, st_mysql_socket) /data/src/10.5/sql/mysqld.cc:6144

    #7 0x564332601b8e in handle_connections_sockets() /data/src/10.5/sql/mysqld.cc:6271

    #8 0x5643326003df in mysqld_main(int, char**) /data/src/10.5/sql/mysqld.cc:5666

    #9 0x5643325ec45f in main /data/src/10.5/sql/main.cc:25

    #10 0x7ff8a34342e0 in __libc_start_main (/lib/x86_64-linux-gnu/libc.so.6+0x202e0)

SUMMARY: AddressSanitizer: use-after-poison /data/src/10.5/storage/innobase/row/row0sel.cc:2517 in row_sel_convert_mysql_key_to_innobase(dtuple_t*, unsigned char*, unsigned long, dict_index_t*, unsigned char const*, unsigned long)

Shadow bytes around the buggy address:

  0x0c2480003910: fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa

  0x0c2480003920: fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa

  0x0c2480003930: fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa fa

  0x0c2480003940: fa fa fa fa fa fa fa fa 00 00 00 00 00 00 00 00

  0x0c2480003950: 00 00 00 00 00 00 00 00 00 00 00 f7 00 00 00 00

=>0x0c2480003960: 00 00 00 00 00 00 f7 f7[f7]f7 f7 f7 f7 f7 f7 fa

  0x0c2480003970: fa fa fa fa fa fa fa fa 00 00 00 00 00 00 00 00

  0x0c2480003980: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00

  0x0c2480003990: 00 00 00 00 00 00 00 00 00 00 00 04 fa fa fa fa

  0x0c24800039a0: fa fa fa fa fa fa fa fa 00 00 00 00 00 00 00 00

  0x0c24800039b0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00

Shadow byte legend (one shadow byte represents 8 application bytes):

  Addressable:           00

  Partially addressable: 01 02 03 04 05 06 07

  Heap left redzone:       fa

  Heap right redzone:      fb

  Freed heap region:       fd

  Stack left redzone:      f1

  Stack mid redzone:       f2

  Stack right redzone:     f3

  Stack partial redzone:   f4

  Stack after return:      f5

  Stack use after scope:   f8

  Global redzone:          f9

  Global init order:       f6

  Poisoned by user:        f7

  Container overflow:      fc

  Array cookie:            ac

  Intra object redzone:    bb

  ASan internal:           fe

  Left alloca redzone:     ca

  Right alloca redzone:    cb

==30504==ABORTING

The same test case with an explicit PK doesn't crash, but the error is still there:

--source include/have_innodb.inc

CREATE TABLE t1 (pk INT PRIMARY KEY, a INT, b INT, c CHAR(8), KEY ind1(c), KEY ind2(b)) ENGINE=InnoDB;

ALTER TABLE t1 RENAME INDEX ind2 TO ind3, RENAME INDEX ind1 TO ind2;

UPDATE t1 SET a = 1 WHERE c = 'foo';

# Cleanup

DROP TABLE t1;

bug3.mtype1 'innodb'                     [ fail ]  Found warnings/errors in server log file!

        Test ended at 2020-07-31 16:08:17

line

2020-07-31 16:08:17 4 [ERROR] Found index ind2 whose column info does not match that of MariaDB.

2020-07-31 16:08:17 4 [ERROR] InnoDB indexes are inconsistent with what defined in .frm for table ./test/t1

A similar test case with a longer CHAR causes a SIGSEGV:

--source include/have_innodb.inc

CREATE TABLE t1 (pk INT PRIMARY KEY, a INT, b INT, c CHAR(64), KEY ind1(c), KEY ind2(b)) ENGINE=InnoDB;

ALTER TABLE t1 RENAME INDEX ind2 TO ind3, RENAME INDEX ind1 TO ind2;

UPDATE t1 SET a = 1 WHERE c = 'foo';

# Cleanup

DROP TABLE t1;

#3  <signal handler called>

#4  0x000055d1efab2a90 in row_sel_convert_mysql_key_to_innobase (tuple=0x7f7664b09998, buf=0x7f76640094d0 "\001", buf_len=8, index=0x7f76641f2b48, key_ptr=0x7f7664b0e221 ' ' <repeats 56 times>, "\245\245\245\245\245\217\217", key_len=65) at /data/src/10.5/storage/innobase/row/row0sel.cc:2518

#5  0x000055d1ef8ca5de in ha_innobase::records_in_range (this=0x7f7664008640, keynr=1, min_key=0x7f769404e560, max_key=0x7f769404e580, pages=0x7f769404e530) at /data/src/10.5/storage/innobase/handler/ha_innodb.cc:13979

#6  0x000055d1ef2fbb92 in handler::multi_range_read_info_const (this=0x7f7664008640, keyno=1, seq=0x7f769404e710, seq_init_param=0x7f769404e740, n_ranges_arg=0, bufsz=0x7f769404e614, flags=0x7f769404e610, cost=0x7f769404edc0) at /data/src/10.5/sql/multi_range_read.cc:177

#7  0x000055d1ef2ff880 in DsMrr_impl::dsmrr_info_const (this=0x7f7664008b20, keyno=1, seq=0x7f769404e710, seq_init_param=0x7f769404e740, n_ranges=0, bufsz=0x7f769404ecd8, flags=0x7f769404ecd4, cost=0x7f769404edc0) at /data/src/10.5/sql/multi_range_read.cc:1712

#8  0x000055d1ef8d51d8 in ha_innobase::multi_range_read_info_const (this=0x7f7664008640, keyno=1, seq=0x7f769404e710, seq_init_param=0x7f769404e740, n_ranges=0, bufsz=0x7f769404ecd8, flags=0x7f769404ecd4, cost=0x7f769404edc0) at /data/src/10.5/storage/innobase/handler/ha_innodb.cc:20576

#9  0x000055d1ef60148b in check_quick_select (param=0x7f769404f060, idx=1, index_only=false, tree=0x7f7664b0e380, update_tbl_stats=true, mrr_flags=0x7f769404ecd4, bufsize=0x7f769404ecd8, cost=0x7f769404edc0, is_ror_scan=0x7f769404ecd2) at /data/src/10.5/sql/opt_range.cc:11159

#10 0x000055d1ef5f7c99 in get_key_scans_params (param=0x7f769404f060, tree=0x7f7664b0e2a8, index_read_must_be_used=false, for_range_access=true, read_time=3.2000000000000002) at /data/src/10.5/sql/opt_range.cc:7443

#11 0x000055d1ef5ec459 in SQL_SELECT::test_quick_select (this=0x7f76640149a8, thd=0x7f7664000b18, keys_to_use=..., prev_tables=0, limit=18446744073709551615, force_quick_range=false, ordered_output=false, remove_false_parts_of_where=false, only_single_index_range_scan=false) at /data/src/10.5/sql/opt_range.cc:2911

#12 0x000055d1ef22c972 in SQL_SELECT::check_quick (this=0x7f76640149a8, thd=0x7f7664000b18, force_quick_range=false, limit=18446744073709551615) at /data/src/10.5/sql/opt_range.h:1655

#13 0x000055d1ef2239b9 in mysql_update (thd=0x7f7664000b18, table_list=0x7f76640139c8, fields=..., values=..., conds=0x7f76640144a8, order_num=0, order=0x0, limit=18446744073709551615, ignore=false, found_return=0x7f769404ff80, updated_return=0x7f7694050050) at /data/src/10.5/sql/sql_update.cc:555

#14 0x000055d1ef102a77 in mysql_execute_command (thd=0x7f7664000b18) at /data/src/10.5/sql/sql_parse.cc:4400

#15 0x000055d1ef10ed4d in mysql_parse (thd=0x7f7664000b18, rawbuf=0x7f76640138d0 "UPDATE t1 SET a = 1 WHERE c = 'foo'", length=35, parser_state=0x7f7694050520, is_com_multi=false, is_next_command=false) at /data/src/10.5/sql/sql_parse.cc:7993

#16 0x000055d1ef0fb046 in dispatch_command (command=COM_QUERY, thd=0x7f7664000b18, packet=0x7f76641b0279 "UPDATE t1 SET a = 1 WHERE c = 'foo'", packet_length=35, is_com_multi=false, is_next_command=false) at /data/src/10.5/sql/sql_parse.cc:1867

#17 0x000055d1ef0f977e in do_command (thd=0x7f7664000b18) at /data/src/10.5/sql/sql_parse.cc:1348

#18 0x000055d1ef29f07a in do_handle_one_connection (connect=0x55d1f3781ef8, put_in_cache=true) at /data/src/10.5/sql/sql_connect.cc:1410

#19 0x000055d1ef29ede2 in handle_one_connection (arg=0x55d1f3781ef8) at /data/src/10.5/sql/sql_connect.cc:1312

#20 0x000055d1ef7dd126 in pfs_spawn_thread (arg=0x55d1f369fdb8) at /data/src/10.5/storage/perfschema/pfs.cc:2201

#21 0x00007f769a6b34a4 in start_thread (arg=0x7f7694051700) at pthread_create.c:456

#22 0x00007f76987e7d0f in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:97

The test case is not applicable to earlier versions due to RENAME INDEX.

Attachments

Issue Links

is caused by

MDEV-7318 RENAME INDEX

Closed

MDEV-13301 Optimize DROP INDEX, ADD INDEX into RENAME INDEX

Closed

Activity

Ascending order - Click to sort in descending order

Marko Mäkelä added a comment - 2020-07-31 13:34

I seem to remember that I pointed out the problem of multiple RENAME INDEX somewhere, but I cannot find that comment in ~~MDEV-7318~~.

I also seem to remember pointing out that in ALTER TABLE, we must distinguish two sets of column or index names: referring to the ‘old’ or the ‘altered’ table. ADD obviously refers to the ‘altered’ table, DROP refers to the ‘old’ table, and the ‘old’ and ‘altered’ names of RENAME must refer to the ‘old’ and ‘altered’ tables, respectively.

For example, the following two statement should be equivalent:

ALTER TABLE t RENAME INDEX i TO j, ADD INDEX(i), RENAME INDEX j TO k;

ALTER TABLE t RENAME INDEX j TO k, RENAME INDEX i TO j, ADD INDEX(i);

Both statements are only valid if the indexes i and j existed before the operation. The operations are executed as a set (with no specified order, contrary to what someone might expect).
The latter statement is easier to read ‘from left to right’.

The test innodb.innodb_rename_index in MySQL 5.7.1 is worth looking at. It includes many combinations. See especially the commit that introduced it and my extension of the test.

Marko Mäkelä added a comment - 2020-07-31 13:34 I seem to remember that I pointed out the problem of multiple RENAME INDEX somewhere, but I cannot find that comment in MDEV-7318 . I also seem to remember pointing out that in ALTER TABLE , we must distinguish two sets of column or index names: referring to the ‘old’ or the ‘altered’ table. ADD obviously refers to the ‘altered’ table, DROP refers to the ‘old’ table, and the ‘old’ and ‘altered’ names of RENAME must refer to the ‘old’ and ‘altered’ tables, respectively. For example, the following two statement should be equivalent: ALTER TABLE t RENAME INDEX i TO j, ADD INDEX (i), RENAME INDEX j TO k; ALTER TABLE t RENAME INDEX j TO k, RENAME INDEX i TO j, ADD INDEX (i); Both statements are only valid if the indexes i and j existed before the operation. The operations are executed as a set (with no specified order, contrary to what someone might expect). The latter statement is easier to read ‘from left to right’. The test innodb.innodb_rename_index in MySQL 5.7.1 is worth looking at. It includes many combinations. See especially the commit that introduced it and my extension of the test .

Elena Stepanova added a comment - 2020-07-31 14:02

As marko suspected, it can also be reproduced without RENAME, with ADD/DROP instead:

--source include/have_innodb.inc

CREATE TABLE t1 (a INT, b INT, c CHAR(8), KEY ind1(c), KEY ind2(b)) ENGINE=InnoDB;

ALTER TABLE t1 DROP INDEX ind2, ADD INDEX ind3(b), DROP INDEX ind1, ADD INDEX ind2(c);

UPDATE t1 SET a = 1 WHERE c = 'foo';

# Cleanup

DROP TABLE t1;

This way it is also reproducible on 10.4:

10.4 a874b6c4
2020-07-31 17:01:38 9 [ERROR] Found index ind3 whose column info does not match that of MariaDB.
2020-07-31 17:01:38 9 [ERROR] InnoDB indexes are inconsistent with what defined in .frm for table ./test/t1
2020-07-31 17:01:38 0x7fef656d2700 InnoDB: Assertion failure in file /data/src/10.4/storage/innobase/row/row0sel.cc line 2537
InnoDB: Failing assertion: field->col->mtype == type

#6 0x000055715bf7a49e in ut_dbg_assertion_failed (expr=0x55715c6f152a "field->col->mtype == type", file=0x55715c6f0e30 "/data/src/10.4/storage/innobase/row/row0sel.cc", line=2537) at /data/src/10.4/storage/innobase/ut/ut0dbg.cc:60
#7 0x000055715bedeb88 in row_sel_convert_mysql_key_to_innobase (tuple=0x7fef10069800, buf=0x7fef100468dc "", buf_len=4, index=0x7fef1015a1f0, key_ptr=0x7fef100687f5 " \245\217\217\217\217\217\217", key_len=9) at /data/src/10.4/storage/innobase/row/row0sel.cc:2537
#8 0x000055715bcfe138 in ha_innobase::records_in_range (this=0x7fef10147ef8, keynr=1, min_key=0x7fef656cf430, max_key=0x7fef656cf450) at /data/src/10.4/storage/innobase/handler/ha_innodb.cc:13726
#9 0x000055715b92272c in handler::multi_range_read_info_const (this=0x7fef10147ef8, keyno=1, seq=0x7fef656cf5f0, seq_init_param=0x7fef656cf620, n_ranges_arg=0, bufsz=0x7fef656cf4e4, flags=0x7fef656cf4e0, cost=0x7fef656cfca0) at /data/src/10.4/sql/multi_range_read.cc:126
#10 0x000055715b9262c2 in DsMrr_impl::dsmrr_info_const (this=0x7fef10148388, keyno=1, seq=0x7fef656cf5f0, seq_init_param=0x7fef656cf620, n_ranges=0, bufsz=0x7fef656cfbb8, flags=0x7fef656cfbb4, cost=0x7fef656cfca0) at /data/src/10.4/sql/multi_range_read.cc:1550
#11 0x000055715bd08c52 in ha_innobase::multi_range_read_info_const (this=0x7fef10147ef8, keyno=1, seq=0x7fef656cf5f0, seq_init_param=0x7fef656cf620, n_ranges=0, bufsz=0x7fef656cfbb8, flags=0x7fef656cfbb4, cost=0x7fef656cfca0) at /data/src/10.4/storage/innobase/handler/ha_innodb.cc:20370
#12 0x000055715bc37ffa in check_quick_select (param=0x7fef656cff10, idx=1, index_only=false, tree=0x7fef100688a8, update_tbl_stats=true, mrr_flags=0x7fef656cfbb4, bufsize=0x7fef656cfbb8, cost=0x7fef656cfca0, is_ror_scan=0x7fef656cfbb2) at /data/src/10.4/sql/opt_range.cc:11141
#13 0x000055715bc2e879 in get_key_scans_params (param=0x7fef656cff10, tree=0x7fef10068810, index_read_must_be_used=false, for_range_access=true, read_time=3.3000000000000003) at /data/src/10.4/sql/opt_range.cc:7429
#14 0x000055715bc232de in SQL_SELECT::test_quick_select (this=0x7fef100141d8, thd=0x7fef10000af0, keys_to_use=..., prev_tables=0, limit=18446744073709551615, force_quick_range=false, ordered_output=false, remove_false_parts_of_where=false, only_single_index_range_scan=false) at /data/src/10.4/sql/opt_range.cc:2925
#15 0x000055715b86290a in SQL_SELECT::check_quick (this=0x7fef100141d8, thd=0x7fef10000af0, force_quick_range=false, limit=18446744073709551615) at /data/src/10.4/sql/opt_range.h:1654
#16 0x000055715b859b45 in mysql_update (thd=0x7fef10000af0, table_list=0x7fef10013290, fields=..., values=..., conds=0x7fef10013cf0, order_num=0, order=0x0, limit=18446744073709551615, ignore=false, found_return=0x7fef656d0e40, updated_return=0x7fef656d0f00) at /data/src/10.4/sql/sql_update.cc:557
#17 0x000055715b74b8ad in mysql_execute_command (thd=0x7fef10000af0) at /data/src/10.4/sql/sql_parse.cc:4367
#18 0x000055715b757983 in mysql_parse (thd=0x7fef10000af0, rawbuf=0x7fef10013198 "UPDATE t1 SET a = 1 WHERE c = 'foo'", length=35, parser_state=0x7fef656d1570, is_com_multi=false, is_next_command=false) at /data/src/10.4/sql/sql_parse.cc:7896
#19 0x000055715b743eb8 in dispatch_command (command=COM_QUERY, thd=0x7fef10000af0, packet=0x7fef100083a1 "UPDATE t1 SET a = 1 WHERE c = 'foo'", packet_length=35, is_com_multi=false, is_next_command=false) at /data/src/10.4/sql/sql_parse.cc:1835
#20 0x000055715b74265a in do_command (thd=0x7fef10000af0) at /data/src/10.4/sql/sql_parse.cc:1353
#21 0x000055715b8cb8b4 in do_handle_one_connection (connect=0x55715f5708e0) at /data/src/10.4/sql/sql_connect.cc:1412
#22 0x000055715b8cb603 in handle_one_connection (arg=0x55715f5708e0) at /data/src/10.4/sql/sql_connect.cc:1316
#23 0x000055715c2cce03 in pfs_spawn_thread (arg=0x55715f4932a0) at /data/src/10.4/storage/perfschema/pfs.cc:1869
#24 0x00007fef6e2b14a4 in start_thread (arg=0x7fef656d2700) at pthread_create.c:456
#25 0x00007fef6c3e5d0f in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:97

Doesn't crash on 10.3

Elena Stepanova added a comment - 2020-07-31 14:02 As marko suspected, it can also be reproduced without RENAME, with ADD/DROP instead: --source include/have_innodb.inc CREATE TABLE t1 (a INT , b INT , c CHAR (8), KEY ind1(c), KEY ind2(b)) ENGINE=InnoDB; ALTER TABLE t1 DROP INDEX ind2, ADD INDEX ind3(b), DROP INDEX ind1, ADD INDEX ind2(c); UPDATE t1 SET a = 1 WHERE c = 'foo' ; # Cleanup DROP TABLE t1; This way it is also reproducible on 10.4: 10.4 a874b6c4 2020-07-31 17:01:38 9 [ERROR] Found index ind3 whose column info does not match that of MariaDB. 2020-07-31 17:01:38 9 [ERROR] InnoDB indexes are inconsistent with what defined in .frm for table ./test/t1 2020-07-31 17:01:38 0x7fef656d2700 InnoDB: Assertion failure in file /data/src/10.4/storage/innobase/row/row0sel.cc line 2537 InnoDB: Failing assertion: field->col->mtype == type #6 0x000055715bf7a49e in ut_dbg_assertion_failed (expr=0x55715c6f152a "field->col->mtype == type", file=0x55715c6f0e30 "/data/src/10.4/storage/innobase/row/row0sel.cc", line=2537) at /data/src/10.4/storage/innobase/ut/ut0dbg.cc:60 #7 0x000055715bedeb88 in row_sel_convert_mysql_key_to_innobase (tuple=0x7fef10069800, buf=0x7fef100468dc "", buf_len=4, index=0x7fef1015a1f0, key_ptr=0x7fef100687f5 " \245\217\217\217\217\217\217", key_len=9) at /data/src/10.4/storage/innobase/row/row0sel.cc:2537 #8 0x000055715bcfe138 in ha_innobase::records_in_range (this=0x7fef10147ef8, keynr=1, min_key=0x7fef656cf430, max_key=0x7fef656cf450) at /data/src/10.4/storage/innobase/handler/ha_innodb.cc:13726 #9 0x000055715b92272c in handler::multi_range_read_info_const (this=0x7fef10147ef8, keyno=1, seq=0x7fef656cf5f0, seq_init_param=0x7fef656cf620, n_ranges_arg=0, bufsz=0x7fef656cf4e4, flags=0x7fef656cf4e0, cost=0x7fef656cfca0) at /data/src/10.4/sql/multi_range_read.cc:126 #10 0x000055715b9262c2 in DsMrr_impl::dsmrr_info_const (this=0x7fef10148388, keyno=1, seq=0x7fef656cf5f0, seq_init_param=0x7fef656cf620, n_ranges=0, bufsz=0x7fef656cfbb8, flags=0x7fef656cfbb4, cost=0x7fef656cfca0) at /data/src/10.4/sql/multi_range_read.cc:1550 #11 0x000055715bd08c52 in ha_innobase::multi_range_read_info_const (this=0x7fef10147ef8, keyno=1, seq=0x7fef656cf5f0, seq_init_param=0x7fef656cf620, n_ranges=0, bufsz=0x7fef656cfbb8, flags=0x7fef656cfbb4, cost=0x7fef656cfca0) at /data/src/10.4/storage/innobase/handler/ha_innodb.cc:20370 #12 0x000055715bc37ffa in check_quick_select (param=0x7fef656cff10, idx=1, index_only=false, tree=0x7fef100688a8, update_tbl_stats=true, mrr_flags=0x7fef656cfbb4, bufsize=0x7fef656cfbb8, cost=0x7fef656cfca0, is_ror_scan=0x7fef656cfbb2) at /data/src/10.4/sql/opt_range.cc:11141 #13 0x000055715bc2e879 in get_key_scans_params (param=0x7fef656cff10, tree=0x7fef10068810, index_read_must_be_used=false, for_range_access=true, read_time=3.3000000000000003) at /data/src/10.4/sql/opt_range.cc:7429 #14 0x000055715bc232de in SQL_SELECT::test_quick_select (this=0x7fef100141d8, thd=0x7fef10000af0, keys_to_use=..., prev_tables=0, limit=18446744073709551615, force_quick_range=false, ordered_output=false, remove_false_parts_of_where=false, only_single_index_range_scan=false) at /data/src/10.4/sql/opt_range.cc:2925 #15 0x000055715b86290a in SQL_SELECT::check_quick (this=0x7fef100141d8, thd=0x7fef10000af0, force_quick_range=false, limit=18446744073709551615) at /data/src/10.4/sql/opt_range.h:1654 #16 0x000055715b859b45 in mysql_update (thd=0x7fef10000af0, table_list=0x7fef10013290, fields=..., values=..., conds=0x7fef10013cf0, order_num=0, order=0x0, limit=18446744073709551615, ignore=false, found_return=0x7fef656d0e40, updated_return=0x7fef656d0f00) at /data/src/10.4/sql/sql_update.cc:557 #17 0x000055715b74b8ad in mysql_execute_command (thd=0x7fef10000af0) at /data/src/10.4/sql/sql_parse.cc:4367 #18 0x000055715b757983 in mysql_parse (thd=0x7fef10000af0, rawbuf=0x7fef10013198 "UPDATE t1 SET a = 1 WHERE c = 'foo'", length=35, parser_state=0x7fef656d1570, is_com_multi=false, is_next_command=false) at /data/src/10.4/sql/sql_parse.cc:7896 #19 0x000055715b743eb8 in dispatch_command (command=COM_QUERY, thd=0x7fef10000af0, packet=0x7fef100083a1 "UPDATE t1 SET a = 1 WHERE c = 'foo'", packet_length=35, is_com_multi=false, is_next_command=false) at /data/src/10.4/sql/sql_parse.cc:1835 #20 0x000055715b74265a in do_command (thd=0x7fef10000af0) at /data/src/10.4/sql/sql_parse.cc:1353 #21 0x000055715b8cb8b4 in do_handle_one_connection (connect=0x55715f5708e0) at /data/src/10.4/sql/sql_connect.cc:1412 #22 0x000055715b8cb603 in handle_one_connection (arg=0x55715f5708e0) at /data/src/10.4/sql/sql_connect.cc:1316 #23 0x000055715c2cce03 in pfs_spawn_thread (arg=0x55715f4932a0) at /data/src/10.4/storage/perfschema/pfs.cc:1869 #24 0x00007fef6e2b14a4 in start_thread (arg=0x7fef656d2700) at pthread_create.c:456 #25 0x00007fef6c3e5d0f in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:97 Doesn't crash on 10.3

Marko Mäkelä added a comment - 2020-07-31 14:18

~~MDEV-13301~~ optimized DROP INDEX, ADD INDEX pairs internally in InnoDB into RENAME operations. There could be something wrong with that.

In addition to fixing the reported scenarios, I think that we must consider combinations of adding, dropping or renaming columns together with adding, dropping or renaming indexes.

We can only optimize DROP INDEX,ADD INDEX into RENAME INDEX if the underlying columns are identical (referring to the same columns that existed before the ALTER TABLE operation, by identity, not necessarily by name) and in the same order, and carrying the same ASC/DESC attribute (ok, that attribute might be unavailable until ~~MDEV-13756~~ is implemented).

Marko Mäkelä added a comment - 2020-07-31 14:18 MDEV-13301 optimized DROP INDEX , ADD INDEX pairs internally in InnoDB into RENAME operations. There could be something wrong with that. In addition to fixing the reported scenarios, I think that we must consider combinations of adding, dropping or renaming columns together with adding, dropping or renaming indexes. We can only optimize DROP INDEX , ADD INDEX into RENAME INDEX if the underlying columns are identical (referring to the same columns that existed before the ALTER TABLE operation, by identity, not necessarily by name) and in the same order, and carrying the same ASC / DESC attribute (ok, that attribute might be unavailable until MDEV-13756 is implemented).

People

Assignee:: Eugene Kosov (Inactive)

Reporter:: Elena Stepanova

Votes:: 0 Vote for this issue

Watchers:: 2 Start watching this issue

Dates

Created:: 2020-07-31 13:12

Updated:: 2020-11-11 22:43

Resolved:: 2020-10-26 14:40

Git Integration

Error rendering 'com.xiplink.jira.git.jira_git_plugin:git-issue-webpanel'. Please contact your Jira administrators.

MariaDB Server