Bug #58703 crash when using --view-protocol
Submitted: 3 Dec 2010 15:55 Modified: 6 Dec 2010 17:09
Reporter: Matthias Leich Email Updates:
Status: Closed Impact on me:
None 
Category:MySQL Server Severity:S3 (Non-critical)
Version:5.6.1-m5-debug-log OS:Any
Assigned to: CPU Architecture:Any
Tags: crash, view-protocol

[3 Dec 2010 15:55] Matthias Leich
Description:
   ./mtr --view-protocol main.no-threads
gives a strange output.
I do not claim that the test must pass because the MTR option
"--view-protocol" does not fit to every test we have.
I apologize in advance if this is no bug
= the server abort is intended
  + the backtrace we see has the shape we
    really want to achieve

Result on mysql-trunk-stage revno: 3209 2010-11-30
--------------------------------------------------
...
main.no-threads                          [ fail ]
...
mysqltest: At line 4: Could not open connection 'util': 2013 Lost connection to MySQL server at 'waiting for initial communication packet', system error: 95
...
101203 18:47:29 - mysqld got signal 6 ;
...
Thread 1 (Thread 20904):
#0  0x00007fd5486c95ec in pthread_kill () from /lib64/libpthread.so.0
#1  0x0000000000b0b690 in my_write_core (sig=6) at mysys/stacktrace.c:330
#2  0x000000000058b622 in handle_segfault (sig=6) at sql/mysqld.cc:2511
#3  <signal handler called>
#4  0x00007fd5476d89e5 in raise () from /lib64/libc.so.6
#5  0x00007fd5476d9ee6 in abort () from /lib64/libc.so.6
#6  0x0000000000b1e20e in DbugExit (why=0x7fff10d53f40 "missing DBUG_RETURN or DBUG_VOID_RETURN macro in function \"native_password_authenticate\"\n")
                       at dbug/dbug.c:2127
    ---------------
    static void DbugExit(const char *why)
    {
        CODE_STATE *cs=code_state();
        (void) fprintf(stderr, ERR_ABORT, cs ? cs->process : "(null)", why);
        (void) fflush(stderr);
        DBUG_ABORT();
    }
    --------------
#7  0x0000000000b1cdb7 in _db_return_ (_line_=9120, _stack_frame_=0x7fff10d544d0) 
                       at dbug/dbug.c:1230
#8  0x00000000005bcc9a in acl_authenticate (thd=0x1ba6650, connect_errors=0, com_change_user_pkt_len=0) 
                       at sql/sql_acl.cc:9120
#9  0x00000000006f0ea0 in check_connection (thd=0x1ba6650) 
                       at sql_connect.cc:500
#10 0x00000000006f0fec in login_connection (thd=0x1ba6650) 
                       at sql/sql_connect.cc:558
#11 0x00000000006f1620 in do_handle_one_connection (thd_arg=0x1ba6650) 
                       at sql/sql_connect.cc:733
#12 0x000000000058f866 in handle_connection_in_main_thread (thd=0x1ba6650) 
                       at sql/mysqld.cc:5079
#13 0x000000000058fd9b in create_new_thread (thd=0x1ba6650) 
                       at sql/mysqld.cc:5194
#14 0x0000000000590293 in handle_connections_sockets () 
                       at sql/mysqld.cc:5453
#15 0x000000000058f35e in mysqld_main (argc=51, argv=0x134d0f8) 
                       at sql/mysqld.cc:4745
#16 0x00000000005875a4 in main (argc=9, argv=0x7fff10d54dd8) 
                       at sql/main.cc:24

How to repeat:
See above
[3 Dec 2010 16:01] Davi Arnaut
I think this was fixed recently, could you test with mysql-trunk-bugfixing? Thanks.
[3 Dec 2010 16:17] Valeriy Kravchuk
This is what I've got on current mysql-trunk tree:

macbook-pro:5.5-sec openxs$ cd ../trunk/mysql-test/
macbook-pro:mysql-test openxs$ ./mtr --view-protocol main.no-threads
Logging: ./mtr  --view-protocol main.no-threads
101203 18:13:47 [Warning] Setting lower_case_table_names=2 because file system for /var/folders/dX/dXCzvuSlHX4Op1g-o1jIWk+++TI/-Tmp-/9IgDVuAxe4/ is case insensitive
101203 18:13:47 [Note] Plugin 'FEDERATED' is disabled.
101203 18:13:47 [Note] Plugin 'ndbcluster' is disabled.
MySQL Version 5.6.1
Checking supported features...
 - skipping ndbcluster
 - SSL connections supported
 - binaries are debug compiled
Collecting tests...
vardir: /Users/openxs/dbs/trunk/mysql-test/var
Checking leftover processes...
Removing old var directory...
Creating var directory '/Users/openxs/dbs/trunk/mysql-test/var'...
Installing system database...
Using server port 58828

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

TEST                                      RESULT   TIME (ms)
------------------------------------------------------------

worker[1] Using MTR_BUILD_THREAD 300, with reserved ports 13000..13009
main.no-threads                          [ fail ]
        Test ended at 2010-12-03 18:16:07

CURRENT_TEST: main.no-threads
mysqltest: At line 4: Could not open connection 'util': 2013 Lost connection to MySQL server at 'waiting for initial communication packet', system error: 102
[3 Dec 2010 17:21] Matthias Leich
Result on mysql-trunk-bugfixing revno: 3411 2010-12-03
SELECT VERSION() "says" 5.6.1-m5-debug-log
-----------------------------------------------------
...
CURRENT_TEST: main.no-threads
mysqltest: At line 4: Could not open connection 'util': 2013 Lost connection to MySQL server at 'waiting for initial communication packet', system error: 95
...
and NO crash
So the crash seems to be fixed.
[6 Dec 2010 16:59] MySQL Verification Team
So if the crash not happen anymore could this bug closed or the Synopsis changed?. Thanks in advance.
[6 Dec 2010 17:09] Matthias Leich
I am closing this bug because it was fixed by
whatever unknown patch for another bug.
There is IMHO no need to document something because
- it is to be assumed that the other bug will get documented
- the current bug will only show up under special conditions
  which customers usually do not fulfill
  - use debug binaries
  - use mysqltest + the option "--view-protocol"