List:Commits« Previous MessageNext Message »
From:Jon Olav Hauglid Date:December 16 2009 11:32am
Subject:bzr commit into mysql-5.6-next-mr branch (jon.hauglid:3040) Bug#48541
View as plain text  
#At file:///export/home/z/mysql-next-4284-bug48541/ based on revid:kostja@stripped

 3040 Jon Olav Hauglid	2009-12-16
      Bug #48541 Deadlock between LOCK_open and LOCK_mdl
      
      The reason for the deadlock was an improper exit from
      MDL_context::wait_for_locks() which caused mysys_var->current_mutex to remain
      LOCK_mdl even though LOCK_mdl was no longer held by that connection. 
      
      This could for example lead to a deadlock in the following way:
      1) INSERT DELAYED tries to open a table but fails, and trying to recover it
      calls wait_for_locks().
      2) Due to a pending exclusive request, wait_for_locks() fails and exits without
      resetting mysys_var->current_mutex for the delayed insert handler thread. So it
      continues to point to LOCK_mdl.
      3) The handler thread manages to open a table.
      4) A different connection takes LOCK_open and tries to take LOCK_mdl.
      5) FLUSH TABLES from a third connection notices that the handler thread has a
      table open, and tries to kill it. This involves locking mysys_var->current_mutex
      while having LOCK_open locked. Since current_mutex mistakenly points to LOCK_mdl,
      we have a deadlock.
      
      This patch makes sure MDL_EXIT_COND() is called before exiting wait_for_locks().
      This clears mysys->current_mutex which resolves the issue. 
      
      An assert is added to recover_from_failed_open_table_attempt() after
      wait_for_locks() is called, to check that current_mutex is indeed reset.
      With this assert in place, existing tests in (e.g.) mdl_sync.test will fail
      without this patch.

    modified:
      sql/mdl.cc
      sql/sql_base.cc
=== modified file 'sql/mdl.cc'
--- a/sql/mdl.cc	2009-12-10 10:53:20 +0000
+++ b/sql/mdl.cc	2009-12-16 11:32:11 +0000
@@ -1254,7 +1254,8 @@ MDL_context::wait_for_locks(MDL_request_
     }
     if (!mdl_request)
     {
-      pthread_mutex_unlock(&LOCK_mdl);
+      /* As a side-effect MDL_EXIT_COND() unlocks LOCK_mdl. */
+      MDL_EXIT_COND(m_thd, mysys_var, old_msg);
       break;
     }
     pthread_cond_wait(&COND_mdl, &LOCK_mdl);

=== modified file 'sql/sql_base.cc'
--- a/sql/sql_base.cc	2009-12-16 08:33:54 +0000
+++ b/sql/sql_base.cc	2009-12-16 11:32:11 +0000
@@ -3693,6 +3693,7 @@ recover_from_failed_open_table_attempt(T
     case OT_WAIT:
       result= (thd->mdl_context.wait_for_locks(&m_mdl_requests) ||
                tdc_wait_for_old_versions(thd, &m_mdl_requests));
+      DBUG_ASSERT(thd->mysys_var->current_mutex == NULL);
       break;
     case OT_DISCOVER:
       {


Attachment: [text/bzr-bundle] bzr/jon.hauglid@sun.com-20091216113211-lysdhzll282ks668.bundle
Thread
bzr commit into mysql-5.6-next-mr branch (jon.hauglid:3040) Bug#48541Jon Olav Hauglid16 Dec