| Bug #16805 | Events: crash if procedure, function, and time | ||
|---|---|---|---|
| Submitted: | 26 Jan 2006 14:41 | Modified: | 26 Jun 2006 18:35 |
| Reporter: | Peter Gulutzan | Email Updates: | |
| Status: | Can't repeat | Impact on me: | |
| Category: | MySQL Server: Stored Routines | Severity: | S3 (Non-critical) |
| Version: | 5.1.6-alpha-debug | OS: | Linux (SUSE 10.0) |
| Assigned to: | Andrey Hristov | CPU Architecture: | Any |
[26 Jan 2006 17:29]
MySQL Verification Team
Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 1121926064 (LWP 24665)]
0x081fdd14 in Item::delete_self (this=0x8f8f8f8f) at item.h:779
779 cleanup();
(gdb) bt full
#0 0x081fdd14 in Item::delete_self (this=0x8f8f8f8f) at item.h:779
No locals.
#1 0x081f74b9 in Query_arena::free_items (this=0x9241468) at sql_class.cc:1605
_db_func_ = 0x0
_db_file_ = 0x0
next = (class Item *) 0x0
_db_level_ = 143343420
_db_framep_ = (char **) 0x924171c
#2 0x0836c4af in sp_head::destroy (this=0x9241468) at sp_head.cc:639
i = (sp_instr *) 0x0
lex = (LEX *) 0x91f73f4
_db_func_ = 0x42df33a8 "è3ßBAG8\bh\024$\t\031[s\b÷\003"
[10 Mar 2006 15:56]
Andrey Hristov
I get this spurious crash too. No idea what is happening it. Maybe because the event's SP is firstly compiled in the connection that creates it, though the mem_root of thd is substituted
[10 Mar 2006 16:10]
Andrey Hristov
Good backtrace from Miguel, 8F means trashed memory, so it's a double free :(
[10 Mar 2006 17:48]
Andrey Hristov
Can't track down the problem because the server crashes with the scenario event without an event . See bug #18140
[25 May 2006 18:53]
MySQL Verification Team
I was unable to repeat anymore this issue running the test case about 3 hours. miguel@hegel:~/dbs/mysql-5.1-new> bk changes | head ChangeSet@1.2188, 2006-05-25 17:58:52+05:00, gluh@eagle.intranet.mysql.r18.ru test fix ChangeSet@1.2186.1.1, 2006-05-25 13:49:25+02:00, pekka@mysql.com Merge mysql.com:/space/pekka/ndb/version/my51 into mysql.com:/space/pekka/ndb/version/my51-bug14509 ChangeSet@1.2187, 2006-05-25 16:06:43+05:00, gluh@mysql.com after merge fix
[7 Jun 2006 22:15]
Peter Gulutzan
The crash does not repeat. However, mysqld is repeatedly displays: 060607 16:12:10 [Warning] SCHEDULER: [`root`@`localhost`][`db78`.`ex`] Out of range value for column 'NODEGROUP' at row -6510615555426900571] Perhaps this is expected behaviour, but I believe it will make people wonder if something is wrong.
[8 Jun 2006 12:59]
Andrey Hristov
This is a partitioning problem. Reported as bug#20340.
[26 Jun 2006 18:35]
Andrey Hristov
Hello, the actual cause of the second problem you have found has been found in the partitioning codebase and therefore a new bug report has been created, namely bug#20340 (Paritioning does not obey to column type). Please subscribe to it to keep track of the issue. Thank you for your reports!
Description: I create an event which invokes a procedure which invokes a function. I leave it running for 2 to 12 hours. Crash. Stack trace: 0x819951a handle_segfault + 494 0xffffe420 _end + -140167152 (nil) 0x83094e7 _ZN7sp_head7executeEP3THD + 1119 0x830a7ec _ZN7sp_head17execute_procedureEP3THDP4ListI4ItemE + 1464 0x8322b12 _ZN11event_timed7executeEP3THDP11st_mem_root + 338 0x831c792 event_executor_worker + 814 0x40050297 _end + 933910151 0x4018b37e _end + 935200622 How to repeat: DELIMITER // CREATE TABLE t9 (s1 int)// CREATE FUNCTION `partition_exists`(p_p_name varchar(64)) RETURNS int(11) return (select count(*) from information_schema.partitions where partition_name = p_p_name) // CREATE PROCEDURE p9 () begin declare continue handler for sqlexception insert into t9 values (partiton_exists('p2')); insert into t9 values (partition_exists('p1')); end// create event ex on schedule every 1 second do call p9()// set global event_scheduler = 1// Nothing is being inserted into t9; that's expected behaviour. I'm not sure how long it ran before the crash. At least two hours, but perhaps many more.