[MDEV-22651] Assertion `dict_col_get_fixed_size(f.col, not_redundant()) <= 768' failed in dict_table_t::init_instant Created: 2020-05-21  Updated: 2020-07-24  Resolved: 2020-07-24

Status: Closed
Project: MariaDB Server
Component/s: Character Sets, Storage Engine - InnoDB
Affects Version/s: 10.5.2, 10.5.3, 10.4
Fix Version/s: 10.4.14, 10.5.5

Type: Bug Priority: Critical
Reporter: Roel Van de Paar Assignee: Marko Mäkelä
Resolution: Fixed Votes: 0
Labels: not-10.1, not-10.2, not-10.3, regression

Issue Links:
Problem/Incident
is caused by MDEV-15562 Instant DROP COLUMN or changing the o... Closed
Relates
relates to MDEV-22771 Assertion `fields[i].same(instant.fie... Closed

 Description    

USE test;
 
CREATE TABLE t (c INT);
 
ALTER TABLE t ADD d INT FIRST;
 
ALTER TABLE t ADD e CHAR(255) CHARACTER SET UTF32;

Leads to:

10.5.3 cfe5ee90c8e4b9dfa98a41fcd299197a59261be7

 
mysqld: /test/10.5_dbg/storage/innobase/handler/handler0alter.cc:173: void dict_table_t::init_instant(const dict_table_t&): Assertion `dict_col_get_fixed_size(f.col, not_redundant()) <= 768' failed.

10.5.3 cfe5ee90c8e4b9dfa98a41fcd299197a59261be7

 
Core was generated by `/test/MD110520-mariadb-10.5.3-linux-x86_64-dbg/bin/mysqld --no-defaults --core-'.
 
Program terminated with signal SIGABRT, Aborted.
 
#0  __pthread_kill (threadid=<optimized out>, signo=signo@entry=6)
 
    at ../sysdeps/unix/sysv/linux/pthread_kill.c:57
 
[Current thread is 1 (Thread 0x146a4b686700 (LWP 1681418))]
 
(gdb) bt
 
#0  __pthread_kill (threadid=<optimized out>, signo=signo@entry=6) at ../sysdeps/unix/sysv/linux/pthread_kill.c:57
 
#1  0x000055f30c918c11 in my_write_core (sig=sig@entry=6) at /test/10.5_dbg/mysys/stacktrace.c:518
 
#2  0x000055f30c0bdf8d in handle_fatal_signal (sig=6) at /test/10.5_dbg/sql/signal_handler.cc:329
 
#3  <signal handler called>
 
#4  __GI_raise (sig=sig@entry=6) at ../sysdeps/unix/sysv/linux/raise.c:51
 
#5  0x0000146a49dca801 in __GI_abort () at abort.c:79
 
#6  0x0000146a49dba39a in __assert_fail_base (fmt=0x146a49f417d8 "%s%s%s:%u: %s%sAssertion `%s' failed.\n%n", assertion=assertion@entry=0x55f30cd0c6b8 "dict_col_get_fixed_size(f.col, not_redundant()) <= 768", file=file@entry=0x55f30cd032e8 "/test/10.5_dbg/storage/innobase/handler/handler0alter.cc", line=line@entry=173, function=function@entry=0x55f30cd08420 <dict_table_t::init_instant(dict_table_t const&)::__PRETTY_FUNCTION__> "void dict_table_t::init_instant(const dict_table_t&)") at assert.c:92
 
#7  0x0000146a49dba412 in __GI___assert_fail (assertion=assertion@entry=0x55f30cd0c6b8 "dict_col_get_fixed_size(f.col, not_redundant()) <= 768", file=file@entry=0x55f30cd032e8 "/test/10.5_dbg/storage/innobase/handler/handler0alter.cc", line=line@entry=173, function=function@entry=0x55f30cd08420 <dict_table_t::init_instant(dict_table_t const&)::__PRETTY_FUNCTION__> "void dict_table_t::init_instant(const dict_table_t&)") at assert.c:101
 
#8  0x000055f30c4bbc9f in dict_table_t::init_instant (this=this@entry=0x146a28452598, table=...) at /test/10.5_dbg/storage/innobase/handler/handler0alter.cc:172
 
#9  0x000055f30c4c629c in dict_table_t::instant_column (this=0x146a28452598, table=..., col_map=<optimized out>) at /test/10.5_dbg/storage/innobase/handler/handler0alter.cc:630
 
#10 0x000055f30c4c6834 in ha_innobase_inplace_ctx::instant_column (this=this@entry=0x146a28475990) at /test/10.5_dbg/storage/innobase/handler/handler0alter.cc:1053
 
#11 0x000055f30c4c6a8a in innobase_instant_try (ha_alter_info=ha_alter_info@entry=0x146a4b6819d0, ctx=ctx@entry=0x146a28475990, altered_table=altered_table@entry=0x146a4b681a70, table=table@entry=0x146a284f0088, trx=trx@entry=0x146a370033b0) at /test/10.5_dbg/storage/innobase/handler/handler0alter.cc:5632
 
#12 0x000055f30c4ca27f in commit_try_norebuild (ha_alter_info=ha_alter_info@entry=0x146a4b6819d0, ctx=ctx@entry=0x146a28475990, altered_table=altered_table@entry=0x146a4b681a70, old_table=0x146a284f0088, trx=0x146a370033b0, table_name=<optimized out>) at /test/10.5_dbg/storage/innobase/handler/handler0alter.cc:10179
 
#13 0x000055f30c4b6959 in ha_innobase::commit_inplace_alter_table (this=0x146a284394a0, altered_table=0x146a4b681a70, ha_alter_info=0x146a4b6819d0, commit=true) at /test/10.5_dbg/storage/innobase/handler/handler0alter.cc:10899
 
#14 0x000055f30c0cc549 in handler::ha_commit_inplace_alter_table (this=0x146a284394a0, altered_table=altered_table@entry=0x146a4b681a70, ha_alter_info=ha_alter_info@entry=0x146a4b6819d0, commit=commit@entry=true) at /test/10.5_dbg/sql/handler.cc:4735
 
#15 0x000055f30bedeb4c in mysql_inplace_alter_table (thd=thd@entry=0x146a28415088, table_list=0x146a284741b0, table=table@entry=0x146a284f0088, altered_table=altered_table@entry=0x146a4b681a70, ha_alter_info=ha_alter_info@entry=0x146a4b6819d0, inplace_supported=inplace_supported@entry=HA_ALTER_INPLACE_INSTANT, target_mdl_request=0x146a4b682830, alter_ctx=0x146a4b6836c0) at /test/10.5_dbg/sql/sql_table.cc:7920
 
#16 0x000055f30beeaf17 in mysql_alter_table (thd=thd@entry=0x146a28415088, new_db=new_db@entry=0x146a28419940, new_name=new_name@entry=0x146a28419d48, create_info=create_info@entry=0x146a4b6842b0, table_list=<optimized out>, table_list@entry=0x146a284741b0, alter_info=alter_info@entry=0x146a4b6841e0, order_num=0, order=0x0, ignore=false, if_exists=false) at /test/10.5_dbg/sql/sql_table.cc:10442
 
#17 0x000055f30bf6e7d0 in Sql_cmd_alter_table::execute (this=<optimized out>, thd=0x146a28415088) at /test/10.5_dbg/sql/sql_alter.cc:532
 
#18 0x000055f30be19eda in mysql_execute_command (thd=thd@entry=0x146a28415088) at /test/10.5_dbg/sql/sql_parse.cc:5912
 
#19 0x000055f30be21804 in mysql_parse (thd=thd@entry=0x146a28415088, rawbuf=<optimized out>, length=<optimized out>, parser_state=parser_state@entry=0x146a4b6853e0, is_com_multi=is_com_multi@entry=false, is_next_command=is_next_command@entry=false) at /test/10.5_dbg/sql/sql_parse.cc:7957
 
#20 0x000055f30be0dffd in dispatch_command (command=command@entry=COM_QUERY, thd=thd@entry=0x146a28415088, packet=packet@entry=0x146a28467089 "ALTER TABLE t ADD e CHAR(255) CHARACTER SET UTF32", packet_length=packet_length@entry=49, is_com_multi=is_com_multi@entry=false, is_next_command=is_next_command@entry=false) at /test/10.5_dbg/sql/sql_parse.cc:1839
 
#21 0x000055f30be0c8cc in do_command (thd=0x146a28415088) at /test/10.5_dbg/sql/sql_parse.cc:1358
 
#22 0x000055f30bf6699d in do_handle_one_connection (connect=<optimized out>, connect@entry=0x146a2a4433a8, put_in_cache=put_in_cache@entry=true) at /test/10.5_dbg/sql/sql_connect.cc:1411
 
#23 0x000055f30bf670b9 in handle_one_connection (arg=arg@entry=0x146a2a4433a8) at /test/10.5_dbg/sql/sql_connect.cc:1313
 
#24 0x000055f30c3c510a in pfs_spawn_thread (arg=0x146a49045888) at /test/10.5_dbg/storage/perfschema/pfs.cc:2201
 
#25 0x0000146a4aaad6db in start_thread (arg=0x146a4b686700) at pthread_create.c:463
 
#26 0x0000146a49eab88f in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:95

Bug confirmed present in:
MariaDB: 10.4.13 (dbg), 10.5.2 (dbg), 10.5.3 (dbg)

Bug confirmed not present in:
MariaDB: 10.1.45 (dbg), 10.1.45 (opt), 10.2.32 (dbg), 10.2.32 (opt), 10.3.23 (dbg), 10.3.23 (opt), 10.4.13 (opt), 10.5.2 (opt), 10.5.3 (opt), 10.5.4 (dbg), 10.5.4 (opt)
MySQL: 5.5.62 (dbg), 5.5.62 (opt), 5.6.47 (dbg), 5.6.47 (opt), 5.7.29 (dbg), 5.7.29 (opt), 8.0.19 (dbg), 8.0.19 (opt)

Note that MariaDB 10.5.4 dbg does not crash.

MariaDB 10.3.23 (debug):

10.3.23 811e4409ce587bc4b693cc0262bccccacc0238fb

 
10.3.23>USE test;
 
Database changed
 
10.3.23>CREATE TABLE t (c INT);
 
Query OK, 0 rows affected (0.012 sec)
 
10.3.23>ALTER TABLE t ADD d INT FIRST;
 
Query OK, 0 rows affected (0.033 sec)
 
Records: 0  Duplicates: 0  Warnings: 0
 
10.3.23>ALTER TABLE t ADD e CHAR(255) CHARACTER SET UTF32;
 
Query OK, 0 rows affected (0.004 sec)
 
Records: 0  Duplicates: 0  Warnings: 0



 Comments   
Comment by Roel Van de Paar [ 2020-06-02 ]

Also see MDEV-22771. Issue may or may not be charset related. Set to critical until analyzed; feel free to adjust as needed.

Comment by Alexander Barkov [ 2020-06-11 ]

marko, kevg, can you please check this issue? Thanks.

Comment by Marko Mäkelä [ 2020-07-24 ]

This turned out to be a bogus debug assertion.

I can repeat this on 10.4 with an even simpler test case (basically any CREATE TABLE…ENGINE=InnoDB followed by the ALTER TABLE):

--source include/have_innodb.inc
 
CREATE TABLE t (c CHAR(255) CHARACTER SET UTF32) ENGINE=InnoDB;
 
ALTER TABLE t ADD e CHAR(255) CHARACTER SET UTF32 FIRST;
 
DROP TABLE t;

For the initially created column c the following will hold in handler::ha_prepare_inplace_alter_table():

10.4 f843e215f3f640de377d8fd60e85bb372fb138fd

 
(rr) p *((ha_innobase*)this)->m_prebuilt.table.indexes.start.fields[3].col
 
$17 = {prtype = 3932414, mtype = 13, len = 1020, mbminlen = 4, mbmaxlen = 4, 
  ind = 0, ord_part = 0, max_prefix = 0, def_val = {data = 0x0, 
    len = 4294967294}}

Note the len=1020 that corresponds to 4×255. This would also hold for f.col.len for the column e in dict_table_t::init_instant(), triggering the debug assertion failure.
I think that in MDEV-15562 I was trying to be consistent with dict_index_add_col() in that debug assertion, but I failed to notice that the logic is being applied to dict_index_t::fields[] and not dict_table_t::cols[]:

	/* Long fixed-length fields that need external storage are treated as
 
	variable-length fields, so that the extern flag can be embedded in
 
	the length word. */
 
 
 
	if (field->fixed_len > DICT_MAX_FIXED_COL_LEN) {
 
		field->fixed_len = 0;
 
	}

Comment by Marko Mäkelä [ 2020-07-24 ]

This bogus debug assertion was not present in release builds.

Generated at Thu Feb 08 09:16:24 UTC 2024 using Jira 8.20.16#820016-sha1:9d11dbea5f4be3d4cc21f03a88dd11d8c8687422.