| Bug #20437 | Events: crash with procedure altering partitions | ||
|---|---|---|---|
| Submitted: | 13 Jun 2006 20:23 | Modified: | 13 Jun 2006 22:47 |
| Reporter: | Peter Gulutzan | Email Updates: | |
| Status: | Can't repeat | Impact on me: | |
| Category: | MySQL Server: Partitions | Severity: | S3 (Non-critical) |
| Version: | 5.1.12-beta-debug | OS: | Linux (SUSE 10.0 / 64-bit) |
| Assigned to: | CPU Architecture: | Any | |
[13 Jun 2006 21:57]
Andrey Hristov
After 1-2 minutes I get the following stack. Seems quite unrelated to events but partitioning stuff.
Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 1118694320 (LWP 24559)]
0x082e8e4d in ha_myisam::external_lock (this=0x91fed20, thd=0x918c990, lock_type=2)
at ha_myisam.cc:1448
1448 return mi_lock_database(file, !table->s->tmp_table ?
(gdb) p this->table->s
$15 = (TABLE_SHARE *) 0x0
(gdb) f 0
#0 0x082e8e4d in ha_myisam::external_lock (this=0x91fed20, thd=0x918c990, lock_type=2)
at ha_myisam.cc:1448
1448 return mi_lock_database(file, !table->s->tmp_table ?
(gdb) l
1443 }
1444
1445
1446 int ha_myisam::external_lock(THD *thd, int lock_type)
1447 {
1448 return mi_lock_database(file, !table->s->tmp_table ?
1449 lock_type : ((lock_type == F_UNLCK) ?
1450 F_UNLCK : F_EXTRA_LCK));
1451 }
(gdb) bt
#0 0x082e8e4d in ha_myisam::external_lock (this=0x91fed20, thd=0x918c990, lock_type=2)
at ha_myisam.cc:1448
#1 0x082f0370 in ha_partition::external_lock (this=0x91feb60, thd=0x918c990, lock_type=2)
at ha_partition.cc:2374
#2 0x082e2c78 in handler::ha_external_lock (this=0x91feb60, thd=0x918c990, lock_type=2)
at handler.cc:3259
#3 0x081f5f73 in unlock_external (thd=0x918c990, table=0x91357e0, count=1) at lock.cc:641
#4 0x081f5389 in mysql_unlock_tables (thd=0x918c990, sql_lock=0x9135790) at lock.cc:271
#5 0x08241dc7 in close_thread_tables (thd=0x918c990, lock_in_use=false, skip_derived=false)
at sql_base.cc:1075
#6 0x0815cb94 in fast_end_partition (thd=0x918c990, copied=0, deleted=0, table=0x9195eb0,
table_list=0x91ee888, is_empty=false, lpt=0x42adafe0, written_bin_log=true)
at sql_partition.cc:3518
#7 0x08160d97 in fast_alter_partition_table (thd=0x918c990, table=0x9195eb0, alter_info=0x9230a8c,
create_info=0x9230930, table_list=0x91ee888, create_list=0x42adb4f0, key_list=0x42adb500,
db=0x91eea28 "db1", table_name=0x91ee860 "tmy", fast_alter_partition=1) at sql_partition.cc:5863
#8 0x0831bcb5 in mysql_alter_table (thd=0x918c990, new_db=0x91eea28 "db1", new_name=0x91ee860 "tmy",
create_info=0x9230930, table_list=0x91ee888, fields=@0x9230848, keys=@0x923083c, order_num=0,
order=0x0, handle_duplicates=DUP_ERROR, ignore=false, alter_info=0x9230a8c, do_send_ok=true)
at sql_table.cc:5577
#9 0x082176dc in mysql_execute_command (thd=0x918c990) at sql_parse.cc:3077
#10 0x08283f66 in Prepared_statement::execute (this=0x9230380, expanded_query=0x42add240,
open_cursor=false) at sql_prepare.cc:2930
#11 0x082827d6 in mysql_sql_stmt_execute (thd=0x918c990) at sql_prepare.cc:2324
#12 0x082164f4 in mysql_execute_command (thd=0x918c990) at sql_parse.cc:2555
#13 0x08379df3 in sp_instr_stmt::exec_core (this=0x91d8750, thd=0x918c990, nextp=0x42add950)
at sp_head.cc:2356
#14 0x08379a33 in sp_lex_keeper::reset_lex_and_exec_core (this=0x91d8778, thd=0x918c990,
nextp=0x42add950, open_tables=false, instr=0x91d8750) at sp_head.cc:2230
#15 0x08379c75 in sp_instr_stmt::execute (this=0x91d8750, thd=0x918c990, nextp=0x42add950)
at sp_head.cc:2307
#16 0x08376c44 in sp_head::execute (this=0x9234310, thd=0x918c990) at sp_head.cc:1084
#17 0x08377e1d in sp_head::execute_procedure (this=0x9234310, thd=0x918c990, args=0x922e04c)
at sp_head.cc:1559
#18 0x0821b705 in mysql_execute_command (thd=0x918c990) at sql_parse.cc:4596
#19 0x08379df3 in sp_instr_stmt::exec_core (this=0x9227ee8, thd=0x918c990, nextp=0x42ade170)
---Type <return> to continue, or q <return> to quit---
at sp_head.cc:2356
#20 0x08379a33 in sp_lex_keeper::reset_lex_and_exec_core (this=0x9227f10, thd=0x918c990,
nextp=0x42ade170, open_tables=false, instr=0x9227ee8) at sp_head.cc:2230
#21 0x08379c75 in sp_instr_stmt::execute (this=0x9227ee8, thd=0x918c990, nextp=0x42ade170)
at sp_head.cc:2307
#22 0x08376c44 in sp_head::execute (this=0x9227bb0, thd=0x918c990) at sp_head.cc:1084
#23 0x08377e1d in sp_head::execute_procedure (this=0x9227bb0, thd=0x918c990, args=0x42ade2e0)
at sp_head.cc:1559
#24 0x08393510 in Event_timed::execute (this=0x915b880, thd=0x918c990, mem_root=0x918c9b0)
at event_timed.cc:1413
#25 0x08388cf4 in event_worker_thread (arg=0x426eb300) at event_scheduler.cc:568
#26 0x40053aa7 in start_thread () from /lib/tls/libpthread.so.0
#27 0x40184c2e in clone () from /lib/tls/libc.so.6
[13 Jun 2006 22:47]
Peter Gulutzan
I have just pulled again from the MySQL-5.1 tree. Now I cannot repeat. And, although Andrey was able to repeat, he tells me that he was using an older version of the tree. Therefore I have changed my bug's status to "Can't Repeat". It was a verified crashing bug, but something in the last 24 hours has apparently fixed it.
Description: I create a procedure which creates tables, adds and drops partitions, inserts, updates, deletes. I create an event which calls this procedure. No problem. I create a second event which calls this same procedure. Crash. How to repeat: mysql> delimiter // mysql> set names utf8// Query OK, 0 rows affected (0.00 sec) mysql> create procedure pmy () -> begin -> declare v1 int; -> declare v2 int; -> declare v3 int default 0; -> declare continue handler for 1513 -> begin -> set @v = 'alter table tmy add partition (partition p'; -> set @v = concat(@v,v2); -> set @v = concat(@v,' values in ('); -> set @v = concat(@v,v2); -> set @v = concat(@v,'))'); -> prepare ペ from @v; -> execute ペ; -> end; -> drop table if exists tmy; -> create table tmy (s1 int, s2 int, s3 int, primary key (s1,s2,s3)) -> partition by list (s1) -> (partition p1 values in (1)); -> set v1 = 0; -> while v1 < 100 do -> set v2 = rand(v1) * 100; -> if mod(v1,1) = 0 then -> insert into tmy values (v2,v1,v3); -> set v3 = v3 + 1; -> end if; -> if mod(v1,2) = 0 then -> set v2 = rand(v1) * 100; -> update tmy set s1 = v2 where s3 = v3; -> set v3 = v3 + 1; -> end if; -> if mod(v1,3) = 0 then -> delete from tmy where s1 = v3; -> set v3 = v3 + 1; -> end if; -> if mod(v1,4) = 0 then -> set @ = 'alter table tmy drop partition p'; -> set @ = concat(@,v2); -> prepare ペ from @; -> execute ペ; -> end if; -> set v1 = v1 + 1; -> end while; -> end// Query OK, 0 rows affected (0.03 sec) mysql> create event emy1 on schedule every 1 second do call pmy()// Query OK, 1 row affected (0.00 sec) mysql> create event emy2 on schedule every 1 second do call pmy()// Query OK, 1 row affected (0.00 sec) mysql> set global event_scheduler = 1// Query OK, 0 rows affected (0.01 sec) ... at this point mysqld will crash. The same statements as above, condensed: delimiter // set names utf8// drop procedure pmy// create procedure pmy () begin declare v1 int; declare v2 int; declare v3 int default 0; declare continue handler for 1513 begin set @v = 'alter table tmy add partition (partition p'; set @v = concat(@v,v2); set @v = concat(@v,' values in ('); set @v = concat(@v,v2); set @v = concat(@v,'))'); prepare ペ from @v; execute ペ; end; drop table if exists tmy; create table tmy (s1 int, s2 int, s3 int, primary key (s1,s2,s3)) partition by list (s1) (partition p1 values in (1)); set v1 = 0; while v1 < 100 do set v2 = rand(v1) * 100; if mod(v1,1) = 0 then insert into tmy values (v2,v1,v3); set v3 = v3 + 1; end if; if mod(v1,2) = 0 then set v2 = rand(v1) * 100; update tmy set s1 = v2 where s3 = v3; set v3 = v3 + 1; end if; if mod(v1,3) = 0 then delete from tmy where s1 = v3; set v3 = v3 + 1; end if; if mod(v1,4) = 0 then set @ = 'alter table tmy drop partition p'; set @ = concat(@,v2); prepare ペ from @; execute ペ; end if; set v1 = v1 + 1; end while; end// create event emy1 on schedule every 1 second do call pmy()// create event emy2 on schedule every 1 second do call pmy()// set global event_scheduler = 1//