Bug #6358 | SELECT COUNT(DISTINCT) crashes server | ||
---|---|---|---|
Submitted: | 1 Nov 2004 10:16 | Modified: | 10 Nov 2004 13:23 |
Reporter: | Marko Mäkelä | Email Updates: | |
Status: | Closed | Impact on me: | |
Category: | MySQL Server | Severity: | S1 (Critical) |
Version: | 5.0.2-bk | OS: | Linux (GNU/Linux) |
Assigned to: | CPU Architecture: | Any |
[1 Nov 2004 10:16]
Marko Mäkelä
[1 Nov 2004 11:18]
MySQL Verification Team
Back trace: (gdb) bt #0 0x081b13fc in free_tmp_table (thd=0x87a8bd8, entry=0x87aa9e0) at sql_select.cc:8314 #1 0x081092f4 in Item_sum_count_distinct::setup (this=0x87b5a70, thd=0x87a8bd8) at item_sum.cc:1368 #2 0x081b83c8 in JOIN::make_sum_func_list (this=0x87b5ed8, field_list=@0x87b6c30, send_fields=@0x87b6c60, before_group_by=true, recompute=true) at sql_select.cc:11782 #3 0x081a3860 in JOIN::exec (this=0x87b5ed8) at sql_select.cc:1475 #4 0x081a4b04 in mysql_select (thd=0x87a8bd8, rref_pointer_array=0x87a8ee8, tables=0x87b5cd0, wild_num=0, fields=@0x87a8e34, conds=0x0, og_num=1, order=0x0, group=0x87b5e90, having=0x0, proc_param=0x0, select_options=8669696, result=0x87b5ec8, unit=0x87a8c1c, select_lex=0x87a8dc8) at sql_select.cc:2029 #5 0x081a06aa in handle_select (thd=0x87a8bd8, lex=0x87a8c10, result=0x87b5ec8) at sql_select.cc:229 #6 0x08175e8d in mysql_execute_command (thd=0x87a8bd8) at sql_parse.cc:2069 #7 0x0817bc69 in mysql_parse (thd=0x87a8bd8, inBuf=0x87b58d0 "select a,count(distinct b)from crash group by a", length=47) at sql_parse.cc:4626 #8 0x08174b62 in dispatch_command (command=COM_QUERY, thd=0x87a8bd8, packet=0x87a04f9 "select a,count(distinct b)from crash group by a", packet_length=48) at sql_parse.cc:1518 #9 0x08174447 in do_command (thd=0x87a8bd8) at sql_parse.cc:1298 #10 0x081739ac in handle_one_connection (arg=0x87a8bd8) at sql_parse.cc:1034 #11 0xb7e4714b in pthread_start_thread () from /lib/libpthread.so.0 #12 0xb7e471df in pthread_start_thread_event () from /lib/libpthread.so.0 #13 0xb7d7a50a in clone () from /lib/libc.so.6
[1 Nov 2004 15:37]
Konstantin Osipov
To anyone who is going to assign this bug: I'm currently looking at replacing SUM(DISTINCT), COUNT(DISTINCT) and AVG(DISTINCT) with single implementation, which should solve this bug automatically.
[10 Nov 2004 13:23]
Marko Mäkelä
It appears that this bug has been fixed. I noticed two duplicates of this report that were not marked as duplicates, but closed, and I could not repeat the crash with a fresh 5.0 bk pull.