[MDEV-5726] Assertion `status_var.memory_used == 0' fails on disconnect after SELECT with materialization+semijoin Created: 2014-02-25  Updated: 2014-03-09  Resolved: 2014-03-09

Status: Closed
Project: MariaDB Server
Component/s: None
Affects Version/s: 10.0.8
Fix Version/s: 10.0.9

Type: Bug Priority: Major
Reporter: Elena Stepanova Assignee: Unassigned
Resolution: Fixed Votes: 0
Labels: None

Issue Links:
Relates

 Description   

The test case is not great, it uses a sleep, but more civilized ways (like count_sessions and such) didn't do the trick reliably – the problem happens anyway, but the server goes down before the assertion failure fully blossoms. The idea of the rough test is to show that the failure actually happens on normal session disconnect, rather than on server shutdown or KILL.

--connect(con1,localhost,root,,)
 
 
 
SET optimizer_switch = 'materialization=on,semijoin=on';
 
 
 
CREATE TABLE t1 (f1 VARCHAR(3));
 
CREATE TABLE t2 (f2 VARCHAR(3), INDEX(f2));
 
INSERT INTO t1 VALUES ('foo'),('bar');
 
INSERT INTO t2 VALUES ('baz'),('qux');
 
 
 
SELECT * FROM t1 WHERE ( 'abc', 'xyz' ) IN ( 
  SELECT f1, f1 FROM t1 WHERE f1 IN ( SELECT MAX( f2) FROM t2 )
 
);
 
 
 
--disconnect con1
 
--sleep 2
 
 
 
--connection default
 
DROP TABLE t1, t2;
 

10.0/sql/sql_class.cc:1636: virtual THD::~THD(): Assertion `status_var.memory_used == 0' failed.
 
140225 12:45:33 [ERROR] mysqld got signal 6 ;
 

#6  0x00007f4fca099621 in *__GI___assert_fail (assertion=0xee0df7 "status_var.memory_used == 0", file=<optimized out>, line=1636, function=0xee3430 "virtual THD::~THD()") at assert.c:81
 
#7  0x0000000000636516 in THD::~THD (this=0x7f4fc2484070, __in_chrg=<optimized out>) at 10.0/sql/sql_class.cc:1636
 
#8  0x000000000063669e in THD::~THD (this=0x7f4fc2484070, __in_chrg=<optimized out>) at 10.0/sql/sql_class.cc:1641
 
#9  0x00000000005b7c2f in unlink_thd (thd=0x7f4fc2484070) at 10.0/sql/mysqld.cc:2722
 
#10 0x00000000005b7f8f in one_thread_per_connection_end (thd=0x7f4fc2484070, put_in_cache=true) at 10.0/sql/mysqld.cc:2833
 
#11 0x0000000000787867 in do_handle_one_connection (thd_arg=0x7f4fc2484070) at 10.0/sql/sql_connect.cc:1390
 
#12 0x0000000000787514 in handle_one_connection (arg=0x7f4fc2484070) at 10.0/sql/sql_connect.cc:1293
 
#13 0x0000000000a2a921 in pfs_spawn_thread (arg=0x7f4fc367cd50) at 10.0/storage/perfschema/pfs.cc:1853
 
#14 0x00007f4fcbc3db50 in start_thread (arg=<optimized out>) at pthread_create.c:304
 
#15 0x00007f4fca148a7d in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:112
 
#16 0x0000000000000000 in ?? ()
 

Trying to get some variables.
 
Some pointers may be invalid and cause the dump to abort.
 
Query (0x0): 
Connection ID (thread ID): 4
 
Status: KILL_CONNECTION

Stack trace from:

revision-id: monty@askmonty.org-20140222011156-i1pa6zm2iltb4usv
 
revno: 4007
 
branch-nick: 10.0
 
Build with: 'cmake . -DCMAKE_BUILD_TYPE=Debug && make'



 Comments   
Comment by Elena Stepanova [ 2014-03-04 ]

Doesn't fail on revno 4017, need to find out why, or at least when it stopped failing.

Comment by Elena Stepanova [ 2014-03-09 ]

The bug was gone along with the following merge:

revno: 4015 [merge]
 
revision-id: sergii@pisem.net-20140228090031-f1iw9irc6a8m7fbv
 
parent: bar@mnogosearch.org-20140227154400-3rs1m2c47lqum7js
 
parent: sergii@pisem.net-20140227214342-gpyum52oiuxm4tgs
 
committer: Sergei Golubchik <sergii@pisem.net>
 
branch nick: 10.0
 
timestamp: Fri 2014-02-28 10:00:31 +0100
 
message:
 
  merge

or, more precisely, with the following nested merge:

    revno: 4007.1.1 [merge]
 
    revision-id: sergii@pisem.net-20140226142807-snbx56wm1am1t90o
 
    parent: monty@askmonty.org-20140222011156-i1pa6zm2iltb4usv
 
    parent: sergii@pisem.net-20140225150435-rpxxfpnsij95boxo
 
    committer: Sergei Golubchik <sergii@pisem.net>
 
    branch nick: 10.0
 
    timestamp: Wed 2014-02-26 15:28:07 +0100
 
    message:
 
      10.0-base merge
 
        ------------------------------------------------------------
 
        revno: 3427.43.5 [merge]
 
        revision-id: sergii@pisem.net-20140225150435-rpxxfpnsij95boxo
 
        parent: psergey@askmonty.org-20140225090157-keyn5zcp8c94u20w
 
        parent: sergii@pisem.net-20140222215120-1thda2f7s8luzgaq
 
        committer: Sergei Golubchik <sergii@pisem.net>
 
        branch nick: 10.0-base
 
        timestamp: Tue 2014-02-25 16:04:35 +0100
 
        message:
 
          5.5 merge

Unfortunately, digging deeper is impossible because the assertion does not exist on 5.5, so it was not reproducible on 5.5 code base either before or after the fix.

Generated at Thu Feb 08 07:06:36 UTC 2024 using Jira 8.20.16#820016-sha1:9d11dbea5f4be3d4cc21f03a88dd11d8c8687422.