[MDEV-7892] server crash on updates with joins Created: 2015-03-31  Updated: 2015-06-09  Resolved: 2015-06-09

Status: Closed
Project: MariaDB Server
Component/s: Data Manipulation - Update
Affects Version/s: 10.0.17, 5.5
Fix Version/s: 5.5.44, 10.0.20

Type: Bug Priority: Critical
Reporter: Max Bubenick Assignee: Oleksandr Byelkin
Resolution: Duplicate Votes: 0
Labels: verified
Environment:

Tested only on centos6, any hardware vm and physical


Attachments: File bug_schema.sql.gz     File core.25994.gz    
Issue Links:
Blocks
is blocked by MDEV-8114 server crash on updates with joins st... Closed
Duplicate
duplicates MDEV-7613 MariaDB 5.5.40 server crash on update... Closed
is duplicated by MDEV-8106 MariaDB 10.0 and 10.1 crash Closed
Sprint: 5.5.44

 Description   

150331 14:49:17 [ERROR] mysqld got signal 11 ;
This could be because you hit a bug. It is also possible that this binary
or one of the libraries it was linked against is corrupt, improperly built,
or misconfigured. This error can also be caused by malfunctioning hardware.
 
To report this bug, see http://kb.askmonty.org/en/reporting-bugs
 
We will try our best to scrape up some info that will hopefully help
diagnose the problem, but since we have already crashed,
something is definitely wrong and this may fail.
 
Server version: 10.0.17-MariaDB-debug-log
key_buffer_size=134217728
read_buffer_size=131072
max_used_connections=10
max_threads=153
thread_count=2
It is possible that mysqld could use up to
key_buffer_size + (read_buffer_size + sort_buffer_size)*max_threads = 467212 K  bytes of memory
Hope that's ok; if not, decrease some variables in the equation.
 
Thread pointer: 0x0x314d9c0
Attempting backtrace. You can use the following information to find out
where mysqld died. If you see no messages after this, something went
terribly wrong...
stack_bottom = 0x7ffd98623e38 thread_stack 0x48000
mysys/stacktrace.c:246(my_print_stacktrace)[0xe52463]
sql/signal_handler.cc:155(handle_fatal_signal)[0x859d2d]
/lib64/libpthread.so.0[0x3aae20f710]
sql/opt_subselect.cc:3294(fix_semijoin_strategies_for_picked_join_order(JOIN*))[0x7cff3b]
sql/sql_select.cc:8446(get_best_combination(JOIN*))[0x6aeb38]
sql/sql_select.cc:4072(make_join_statistics)[0x6a4a07]
sql/sql_select.cc:1339(JOIN::optimize_inner())[0x69acb8]
sql/sql_select.cc:1024(JOIN::optimize())[0x699bc9]
sql/sql_lex.cc:3490(st_select_lex::optimize_unflattened_subqueries(bool))[0x658bf9]
sql/opt_subselect.cc:4950(JOIN::optimize_unflattened_subqueries())[0x7d44cc]
sql/sql_select.cc:1784(JOIN::optimize_inner())[0x69c742]
sql/sql_select.cc:1024(JOIN::optimize())[0x699bc9]
sql/sql_select.cc:3296(mysql_select(THD*, Item***, TABLE_LIST*, unsigned int, List<Item>&, Item*, unsigned int, st_order*, st_order*, Item*, st_order*, unsigned long long, select_result*, st_select_lex_unit*, st_select_lex*))[0x6a1b99]
sql/sql_update.cc:1567(mysql_multi_update(THD*, TABLE_LIST*, List<Item>*, List<Item>*, Item*, unsigned long long, enum_duplicates, bool, st_select_lex_unit*, st_select_lex*, multi_update**))[0x732668]
sql/sql_parse.cc:3366(mysql_execute_command(THD*))[0x665ea8]
sql/sql_parse.cc:6517(mysql_parse(THD*, char*, unsigned int, Parser_state*))[0x66ea01]
sql/sql_parse.cc:1302(dispatch_command(enum_server_command, THD*, char*, unsigned int))[0x660f9f]
sql/sql_parse.cc:996(do_command(THD*))[0x6602d5]
sql/sql_connect.cc:1375(do_handle_one_connection(THD*))[0x782480]
sql/sql_connect.cc:1290(handle_one_connection)[0x7821d3]
/lib64/libpthread.so.0[0x3aae2079d1]
/lib64/libc.so.6(clone+0x6d)[0x3aadee886d]
 
Trying to get some variables.
Some pointers may be invalid and cause the dump to abort.
Query (0x7ffd94053568): UPDATE table_a a JOIN table_b b ON a.c1 = b.c1 JOIN view_vw vw ON b.c2 = vw.c1 JOIN table_t del ON vw.c2 = del.c2 SET a.c2 = (         SELECT                 max(t.c1)         FROM                 table_t t,                 view_vw i         WHERE                 del.c2 = t.c2         AND vw.c3 = i.c3         AND t.c3 = 4 ) WHERE         a.c2 IS NULL OR a.c2 < '2011-05-01'
Connection ID (thread ID): 262
Status: NOT_KILLED
 
Optimizer switch: index_merge=on,index_merge_union=on,index_merge_sort_union=on,index_merge_intersection=on,index_merge_sort_intersection=off,engine_condition_pushdown=off,index_condition_pushdown=on,derived_merge=on,derived_with_keys=on,firstmatch=on,loosescan=on,materialization=on,in_to_exists=on,semijoin=on,partial_match_rowid_merge=on,partial_match_table_scan=on,subquery_cache=on,mrr=off,mrr_cost_based=off,mrr_sort_keys=off,outer_join_with_cache=on,semijoin_with_cache=on,join_cache_incremental=on,join_cache_hashed=on,join_cache_bka=on,optimize_join_buffer_size=off,table_elimination=on,extended_keys=on,exists_to_in=on

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

Please note that Ive also tried disabling semijoin and semijoin_with_cache with the same result, my.cnf is default. Attached are the core dump file and you can reproduce it loading the attached schema and running the update from the above trace.



 Comments   
Comment by Elena Stepanova [ 2015-04-01 ]

Thanks for the report and the test case.
It appears to be the same problem as MDEV-7613. The patch for that bug has not been pushed into the tree yet, but I checked locally that a patched build does not crash on the provided test case (and crashes without the patch).

MariaDB [bug2]> UPDATE table_a a JOIN table_b b ON a.c1 = b.c1 JOIN view_vw vw ON b.c2 = vw.c1 JOIN table_t del ON vw.c2 = del.c2 SET a.c2 = ( SELECT max(t.c1) FROM table_t t, view_vw i WHERE del.c2 = t.c2 AND vw.c3 = i.c3 AND t.c3 = 4 ) WHERE a.c2 IS NULL OR a.c2 < '2011-05-01';
Query OK, 0 rows affected (0.00 sec)
Rows matched: 0  Changed: 0  Warnings: 0

I will keep the bug report open until the patch for MDEV-7613 is pushed into the main tree, and will re-check it there before closing.

Comment by Elena Stepanova [ 2015-04-16 ]

I checked that the crash is not reproducible any longer on 5.5 tree after the bugfix for MDEV-7613 was pushed.

Comment by Elena Stepanova [ 2015-05-07 ]

The bug re-appeared with the following revision:

commit 8cbaafd22b145512cc91f7b512290320849e77bd
Author: Oleksandr Byelkin <sanja@mariadb.com>
Date:   Wed Apr 22 10:14:11 2015 +0200
 
    MDEV-8018: main.multi_update fails with --ps-protocol
    
    save_prep_leaf_tables() made recursive to work with underlying view
    
    Arena restoiring fixed in case of EOM.
 

So it's not really fixed in 5.5.43 and 10.0.18. See MDEV-8114 for further progress.

Comment by Elena Stepanova [ 2015-05-07 ]

Actually, I can safely re-open this bug because it didn't show anywhere in the git log or changelog.

Comment by Oleksandr Byelkin [ 2015-06-09 ]

Duplicate of MDEV-8114

Generated at Thu Feb 08 07:23:03 UTC 2024 using Jira 8.20.16#820016-sha1:9d11dbea5f4be3d4cc21f03a88dd11d8c8687422.