Details
-
Bug
-
Status: Closed (View Workflow)
-
Major
-
Resolution: Duplicate
-
10.1(EOL), 10.2(EOL)
-
None
-
10.1.24, 10.2.10, 10.2.11
Description
Might be related to MDEV-5518 / https://bugs.mysql.com/bug.php?id=73819
innodb_fts.fulltext_misc 'innodb_plugin' w1 [ fail ] Found warnings/errors in server log file!
|
Test ended at 2017-04-27 13:05:34
|
line
|
==1037== Thread 22:
|
==1037== Syscall param write(buf) points to uninitialised byte(s)
|
==1037== at 0x4E42B1D: ??? (in /lib64/libpthread-2.18.so)
|
==1037== by 0xF1EB0B: my_write (my_write.c:49)
|
==1037== by 0xEF31ED: inline_mysql_file_write (mysql_file.h:1131)
|
==1037== by 0xEF5E79: _my_b_cache_write (mf_iocache.c:1643)
|
==1037== by 0xEF66D4: my_b_flush_io_cache (mf_iocache.c:1843)
|
==1037== by 0xEF68C2: end_io_cache (mf_iocache.c:1902)
|
==1037== by 0xA8A5A5: maria_extra (ma_extra.c:164)
|
==1037== by 0xA4547A: ha_maria::extra(ha_extra_function) (ha_maria.cc:2583)
|
==1037== by 0x6F5367: select_union::flush() (sql_union.cc:111)
|
==1037== by 0x6F7C07: st_select_lex_unit::exec() (sql_union.cc:849)
|
==1037== by 0x6F5045: mysql_union(THD*, LEX*, select_result*, st_select_lex_unit*, unsigned long) (sql_union.cc:39)
|
==1037== by 0x658035: handle_select(THD*, LEX*, select_result*, unsigned long) (sql_select.cc:362)
|
==1037== Address 0x183197eb is 11 bytes inside a block of size 131,080 alloc'd
|
==1037== at 0x4C280F3: malloc (vg_replace_malloc.c:299)
|
==1037== by 0xF178DA: my_malloc (my_malloc.c:101)
|
==1037== by 0xEF3B76: init_io_cache (mf_iocache.c:272)
|
==1037== by 0xA8A512: maria_extra (ma_extra.c:145)
|
==1037== by 0xA4547A: ha_maria::extra(ha_extra_function) (ha_maria.cc:2583)
|
==1037== by 0x6F5590: select_union::create_result_table(THD*, List<Item>*, bool, unsigned long long, char const*, bool, bool, bool) (sql_union.cc:168)
|
==1037== by 0x6F6AFE: st_select_lex_unit::prepare(THD*, select_result*, unsigned long) (sql_union.cc:585)
|
==1037== by 0x6F502B: mysql_union(THD*, LEX*, select_result*, st_select_lex_unit*, unsigned long) (sql_union.cc:37)
|
==1037== by 0x658035: handle_select(THD*, LEX*, select_result*, unsigned long) (sql_select.cc:362)
|
==1037== by 0x627253: execute_sqlcom_select(THD*, TABLE_LIST*) (sql_parse.cc:5920)
|
==1037== by 0x61D721: mysql_execute_command(THD*) (sql_parse.cc:2979)
|
==1037== by 0x62A952: mysql_parse(THD*, char*, unsigned int, Parser_state*) (sql_parse.cc:7343)
|
Attachments
Issue Links
- relates to
-
-
- Closed
-
-
-
- Closed
-
Activity
it appeared that boolean mode allow absence of index.
internal temporary table table is aria, returned value of match is the same independent of original type of the table, but bug reproduceble only with innodb as t1 table.
It is 2 SELECTs with number 3, probably again problem of reset counter of SELECTs because of the error
The problem is fixed. the fact that bisect can't find normally the commit which fixed it it came form merge.
Also query in one of the test proposed here start return incorrect usage of FT and also the fact that it was not repeatable without FT and binary patch in particular makes me think that problem was there.
So I will just uncomment commented out test in fulltext_misc
OK, bisect was done incorrectly.
here is patch which fixed the problem:
HEAD ist jetzt bei b20f821e072 Fix Innodb ASAN error on init
b20f821e0723c27d6ce121aafb2eb6c6d89bcd57 is the first bad commit
commit b20f821e0723c27d6ce121aafb2eb6c6d89bcd57
Author: Vicențiu Ciorbaru <vicentiu@mariadb.org>
Date: Wed Jan 24 15:18:36 2018 +0200
Fix Innodb ASAN error on init
Backport 7c03edf2fe66855a8ce8f2575c3aaf66af975377 from xtradb to innodb
If remove "binary mode" it return error that can not find fulltext index and it looks correct, why it do not need index in binary mode.