Making FLUSH TABLES WITH READ LOCK block COMMITs of existing transactions,

in a deadlock-free manner. This splits locking the global read lock in two steps.
This fixes a consequence of this bug, known as:
BUG#4953 'mysqldump --master-data may report incorrect binlog position if using InnoDB'
And a test.


sql/handler.cc:
  making COMMIT wait if FLUSH TABLES WITH READ LOCK happened.
sql/lock.cc:
  an additional stage so that FLUSH TABLES WITH READ LOCK blocks COMMIT:
  make_global_read_lock_block_commit():
  taking the global read lock is TWO steps (2nd step is optional; without
  it, COMMIT of existing transactions will be allowed):
  lock_global_read_lock() THEN make_global_read_lock_block_commit().
sql/mysql_priv.h:
  new argument to wait_if_global_read_lock()
sql/sql_class.h:
  THD::global_read_lock now an uint to reflect the 2 steps of global read lock (does not block COMMIT / does)
sql/sql_db.cc:
  update for new prototype
sql/sql_parse.cc:
  implementing the two steps of global read lock so that FLUSH TABLES WITH READ LOCK can block COMMIT without deadlocking with COMMITs.

mysql-test/r/flush_block_commit.result

+drop table if exists t1;
+create table t1 (a int) type=innodb;
+begin;
+insert into t1 values(1);
+flush tables with read lock;
+select * from t1;
+a
+ commit;
+select * from t1;
+a
+unlock tables;
+begin;
+select * from t1 for update;
+a
+1
+begin;
+ select * from t1 for update;
+ flush tables with read lock;
+commit;
+a
+1
+unlock tables;
+drop table t1;

mysql-test/t/flush_block_commit-master.opt

+--innodb_lock_wait_timeout=5

mysql-test/t/flush_block_commit.test

+# Let's see if FLUSH TABLES WITH READ LOCK blocks COMMIT of existing
+# transactions.
+# We verify that we did not introduce a deadlock.
+
+-- source include/have_innodb.inc
+
+connect (con1,localhost,root,,);
+connect (con2,localhost,root,,);
+connect (con3,localhost,root,,);
+connection con1;
+drop table if exists t1;
+create table t1 (a int) type=innodb;
+
+# blocks COMMIT ?
+
+begin;
+insert into t1 values(1);
+connection con2;
+flush tables with read lock;
+select * from t1;
+connection con1;
+send commit; # blocked by con2
+sleep 1;
+connection con2;
+select * from t1; # verify con1 was blocked and data did not move
+unlock tables;
+connection con1;
+reap;
+
+# No deadlock ?
+
+connection con1;
+begin;
+select * from t1 for update;
+connection con2;
+begin;
+send select * from t1 for update; # blocked by con1
+sleep 1;
+connection con3;
+send flush tables with read lock; # blocked by con2
+connection con1;
+commit; # should not be blocked by con3
+connection con2;
+reap;
+connection con3;
+reap;
+unlock tables;
+connection con1;
+drop table t1;

sql/handler.cc

@@ -342,18 +342,31 @@ int ha_commit_trans(THD *thd, THD_TRANS* trans)
 #ifdef USING_TRANSACTIONS
   if (opt_using_transactions)
   {
-    bool operation_done= 0;
+    bool operation_done= 0, need_start_waiters= 0;
     bool transaction_commited= 0;
-
-    /* Update the binary log if we have cached some queries */
-    if (trans == &thd->transaction.all && mysql_bin_log.is_open() &&
+    /* If transaction has done some updates to tables */
+    if (trans == &thd->transaction.all &&
 	my_b_tell(&thd->transaction.trans_log))
+    {
+      if (error= wait_if_global_read_lock(thd, 0, 0))
+      {
+        /*
+          Note that ROLLBACK [TO SAVEPOINT] does not have this test; it's
+          because ROLLBACK never updates data, so needn't wait on the lock.
+        */
+        my_error(ER_ERROR_DURING_COMMIT, MYF(0), error);
+        error= 1;
+      }
+      else
+        need_start_waiters= 1;
+      if (mysql_bin_log.is_open())
       {
         mysql_bin_log.write(thd, &thd->transaction.trans_log, 1);
         reinit_io_cache(&thd->transaction.trans_log,
                         WRITE_CACHE, (my_off_t) 0, 0, 1);
         thd->transaction.trans_log.end_of_file= max_binlog_cache_size;
       }
+    }
 #ifdef HAVE_BERKELEY_DB
     if (trans->bdb_tid)
     {
@@ -393,6 +406,8 @@ int ha_commit_trans(THD *thd, THD_TRANS* trans)
       statistic_increment(ha_commit_count,&LOCK_status);
       thd->transaction.cleanup();
     }
+    if (need_start_waiters)
+      start_waiting_global_read_lock(thd);
   }
 #endif // using transactions
   DBUG_RETURN(error);

sql/lock.cc

@@ -96,7 +96,7 @@ MYSQL_LOCK *mysql_lock_tables(THD *thd,TABLE **tables,uint count)
 	Someone has issued LOCK ALL TABLES FOR READ and we want a write lock
 	Wait until the lock is gone
       */
-      if (wait_if_global_read_lock(thd, 1))
+      if (wait_if_global_read_lock(thd, 1, 1))
       {
 	my_free((gptr) sql_lock,MYF(0));
 	sql_lock=0;
@@ -453,7 +453,7 @@ int lock_and_wait_for_table_name(THD *thd, TABLE_LIST *table_list)
   int error= -1;
   DBUG_ENTER("lock_and_wait_for_table_name");
 
-  if (wait_if_global_read_lock(thd,0))
+  if (wait_if_global_read_lock(thd, 0, 1))
     DBUG_RETURN(1);
   VOID(pthread_mutex_lock(&LOCK_open));
   if ((lock_retcode = lock_table_name(thd, table_list)) < 0)
@@ -667,14 +667,23 @@ static void print_lock_error(int error)
 
   The global locks are handled through the global variables:
   global_read_lock
+  global_read_lock_blocks_commit
   waiting_for_read_lock 
   protect_against_global_read_lock
+
+  Taking the global read lock is TWO steps (2nd step is optional; without
+  it, COMMIT of existing transactions will be allowed):
+  lock_global_read_lock() THEN make_global_read_lock_block_commit().
 ****************************************************************************/
 
 volatile uint global_read_lock=0;
+volatile uint global_read_lock_blocks_commit=0;
 static volatile uint protect_against_global_read_lock=0;
 static volatile uint waiting_for_read_lock=0;
 
+#define GOT_GLOBAL_READ_LOCK               1
+#define MADE_GLOBAL_READ_LOCK_BLOCK_COMMIT 2
+
 bool lock_global_read_lock(THD *thd)
 {
   DBUG_ENTER("lock_global_read_lock");
@@ -697,27 +706,40 @@ bool lock_global_read_lock(THD *thd)
       thd->exit_cond(old_message);
       DBUG_RETURN(1);
     }
-    thd->global_read_lock=1;
+    thd->global_read_lock= GOT_GLOBAL_READ_LOCK;
     global_read_lock++;
     thd->exit_cond(old_message);
   }
+  /*
+    We DON'T set global_read_lock_blocks_commit now, it will be set after
+    tables are flushed (as the present function serves for FLUSH TABLES WITH
+    READ LOCK only). Doing things in this order is necessary to avoid
+    deadlocks (we must allow COMMIT until all tables are closed; we should not
+    forbid it before, or we can have a 3-thread deadlock if 2 do SELECT FOR
+    UPDATE and one does FLUSH TABLES WITH READ LOCK).
+  */
   DBUG_RETURN(0);
 }
 
 void unlock_global_read_lock(THD *thd)
 {
   uint tmp;
-  thd->global_read_lock=0;
   pthread_mutex_lock(&LOCK_open);
   tmp= --global_read_lock;
+  if (thd->global_read_lock == MADE_GLOBAL_READ_LOCK_BLOCK_COMMIT)
+    --global_read_lock_blocks_commit;
   pthread_mutex_unlock(&LOCK_open);
   /* Send the signal outside the mutex to avoid a context switch */
   if (!tmp)
     pthread_cond_broadcast(&COND_refresh);
+  thd->global_read_lock= 0;
 }
 
+#define must_wait (global_read_lock &&                             \
+                   (is_not_commit ||                               \
+                    global_read_lock_blocks_commit))
 
-bool wait_if_global_read_lock(THD *thd, bool abort_on_refresh)
+bool wait_if_global_read_lock(THD *thd, bool abort_on_refresh, bool is_not_commit)
 {
   const char *old_message;
   bool result= 0, need_exit_cond;
@@ -725,7 +747,7 @@ bool wait_if_global_read_lock(THD *thd, bool abort_on_refresh)
 
   LINT_INIT(old_message);
   (void) pthread_mutex_lock(&LOCK_open);
-  if (need_exit_cond= (bool)global_read_lock)
+  if (need_exit_cond= must_wait)
   {
     if (thd->global_read_lock)		// This thread had the read locks
     {
@@ -735,7 +757,7 @@ bool wait_if_global_read_lock(THD *thd, bool abort_on_refresh)
     }
     old_message=thd->enter_cond(&COND_refresh, &LOCK_open,
 				"Waiting for release of readlock");
-    while (global_read_lock && ! thd->killed &&
+    while (must_wait && ! thd->killed &&
 	   (!abort_on_refresh || thd->version == refresh_version))
       (void) pthread_cond_wait(&COND_refresh,&LOCK_open);
     if (thd->killed)
@@ -762,3 +784,21 @@ void start_waiting_global_read_lock(THD *thd)
     pthread_cond_broadcast(&COND_refresh);
   DBUG_VOID_RETURN;
 }
+
+
+void make_global_read_lock_block_commit(THD *thd)
+{
+  /*
+    If we didn't succeed lock_global_read_lock(), or if we already suceeded
+    make_global_read_lock_block_commit(), do nothing.
+  */
+  if (thd->global_read_lock != GOT_GLOBAL_READ_LOCK)
+    return;
+  pthread_mutex_lock(&LOCK_open);
+  /* increment this BEFORE waiting on cond (otherwise race cond) */
+  global_read_lock_blocks_commit++;
+  while (protect_against_global_read_lock)
+    pthread_cond_wait(&COND_refresh, &LOCK_open);
+  pthread_mutex_unlock(&LOCK_open);
+  thd->global_read_lock= MADE_GLOBAL_READ_LOCK_BLOCK_COMMIT;
+}

sql/mysql_priv.h

@@ -762,8 +762,9 @@ void mysql_lock_abort_for_thread(THD *thd, TABLE *table);
 MYSQL_LOCK *mysql_lock_merge(MYSQL_LOCK *a,MYSQL_LOCK *b);
 bool lock_global_read_lock(THD *thd);
 void unlock_global_read_lock(THD *thd);
-bool wait_if_global_read_lock(THD *thd, bool abort_on_refresh);
+bool wait_if_global_read_lock(THD *thd, bool abort_on_refresh, bool is_not_commit);
 void start_waiting_global_read_lock(THD *thd);
+void make_global_read_lock_block_commit(THD *thd);
 
 /* Lock based on name */
 int lock_and_wait_for_table_name(THD *thd, TABLE_LIST *table_list);

sql/sql_class.h

@@ -478,7 +478,7 @@ class THD :public ilink
   ulong	     rand_saved_seed1, rand_saved_seed2;
   long	     dbug_thread_id;
   pthread_t  real_id;
-  uint	     current_tablenr,tmp_table,cond_count;
+  uint	     current_tablenr,tmp_table,cond_count,global_read_lock;
   uint	     server_status,open_options,system_thread;
   uint32     query_length;
   uint32     db_length;
@@ -489,7 +489,7 @@ class THD :public ilink
   bool	     set_query_id,locked,count_cuted_fields,some_tables_deleted;
   bool	     no_errors, allow_sum_func, password, fatal_error;
   bool	     query_start_used,last_insert_id_used,insert_id_used,rand_used;
-  bool	     in_lock_tables,global_read_lock;
+  bool	     in_lock_tables;
   bool       query_error, bootstrap, cleanup_done;
   bool	     safe_to_cache_query;
   bool	     volatile killed;

sql/sql_db.cc

@@ -42,7 +42,7 @@ int mysql_create_db(THD *thd, char *db, uint create_options, bool silent)
   VOID(pthread_mutex_lock(&LOCK_mysql_create_db));
 
   // do not create database if another thread is holding read lock
-  if (wait_if_global_read_lock(thd,0))
+  if (wait_if_global_read_lock(thd, 0, 1))
   {
     error= -1;
     goto exit2;
@@ -146,7 +146,7 @@ int mysql_rm_db(THD *thd,char *db,bool if_exists, bool silent)
   VOID(pthread_mutex_lock(&LOCK_mysql_create_db));
 
   // do not drop database if another thread is holding read lock
-  if (wait_if_global_read_lock(thd,0))
+  if (wait_if_global_read_lock(thd, 0, 1))
   {
     error= -1;
     goto exit2;

sql/sql_parse.cc

@@ -3708,7 +3708,11 @@ bool reload_acl_and_cache(THD *thd, ulong options, TABLE_LIST *tables)
     {
       if (lock_global_read_lock(thd))
 	return 1;
+      result=close_cached_tables(thd,(options & REFRESH_FAST) ? 0 : 1,
+                                 tables);
+      make_global_read_lock_block_commit(thd);
     }
+    else
       result=close_cached_tables(thd,(options & REFRESH_FAST) ? 0 : 1, tables);
   }
   if (options & REFRESH_HOSTS)