| Bug #68461 | potential deadlock that caused by concurrent DDL and 'SET GLOBAL ' command | ||
|---|---|---|---|
| Submitted: | 22 Feb 2013 8:07 | Modified: | 22 Feb 2013 13:35 |
| Reporter: | zhai weixiang (OCA) | Email Updates: | |
| Status: | Won't fix | Impact on me: | |
| Category: | MySQL Server | Severity: | S3 (Non-critical) |
| Version: | 5.1 | OS: | Any |
| Assigned to: | Jon Olav Hauglid | CPU Architecture: | Any |
[22 Feb 2013 8:37]
MySQL Verification Team
easy to repeat on 5.1.68
Attachment: bug68461_5.1.68_stack_traces.txt (text/plain), 9.40 KiB.
[22 Feb 2013 8:42]
MySQL Verification Team
Seems I repeated a slightly different deadlock, but still, none of this is repeatable on 5.5.30.
[22 Feb 2013 13:35]
Jon Olav Hauglid
Hello! Thank you for the bug report. This problem was fixed in 5.5.3 and we have no plans to backport the fix to 5.1. Therefore closing the bug as "Won't fix".
Description: interesting code path that may lead to deadlock( based on code review of MySQL 5.1.68) session 1 ( SET GLOBAL SLOW_QUERY_LOG = OFF; SET GLOBAL LOG_OUTPUT = 'TABLE'; SET GLOBAL SLOW_QUERY_LOG = ON; ) sys_var_log_state::update --->hold LOCK_global_system_variables LOGGER::activate_log_handler activate_log open_performance_schema_table open_ltable open_table --------->require LOCK_open session 2 ( a DDL request): mysql_alter_table mysql_create_table_no_lock ---->hold LOCK_open rea_create_table ha_create_table ha_innodb::create create_options_are_valid mysql_sys_var_char intern_sys_var_ptr ----->require LOCK_global_system_variables It seems 5.5 was not affected because LOCK_global_system_variables was released before activate_log_handler was called How to repeat: based on code review Suggested fix: Is it safe to release LOCK_global_system_variables before activate_log_handler was called in 5.1? I am not sure