Merge branch '5.5-galera' into 10.0-galera

mysql-test/suite/galera/r/galera_concurrent_ctas.result

@@ -0,0 +1 @@
+# End of test

mysql-test/suite/galera/t/galera_concurrent_ctas.test

@@ -0,0 +1,57 @@
+--source include/galera_cluster.inc
+--source include/have_innodb.inc
+
+--write_file $MYSQLTEST_VARDIR/tmp/galera_concurrent.sql
+CREATE table t1 as SELECT SLEEP(0);
+DROP table t1;
+CREATE table t1 as SELECT SLEEP(0);
+DROP table t1;
+CREATE table t1 as SELECT SLEEP(0);
+DROP table t1;
+CREATE table t1 as SELECT SLEEP(0);
+DROP table t1;
+CREATE table t1 as SELECT SLEEP(0);
+DROP table t1;
+CREATE table t1 as SELECT SLEEP(0);
+DROP table t1;
+CREATE table t1 as SELECT SLEEP(0);
+DROP table t1;
+CREATE table t1 as SELECT SLEEP(0);
+DROP table t1;
+CREATE table t1 as SELECT SLEEP(0);
+CREATE table t2 as SELECT SLEEP(0);
+CREATE table t3 as SELECT SLEEP(0);
+CREATE table t4 as SELECT SLEEP(0);
+CREATE table t5 as SELECT SLEEP(0);
+CREATE table t6 as SELECT SLEEP(0);
+CREATE table t7 as SELECT SLEEP(0);
+CREATE table t8 as SELECT SLEEP(0);
+CREATE table t9 as SELECT SLEEP(0);
+DROP table t1;
+DROP table t2;
+DROP table t3;
+DROP table t4;
+DROP table t5;
+DROP table t6;
+DROP table t7;
+DROP table t8;
+DROP table t9;
+EOF
+
+let $run=10;
+
+while($run)
+{
+  --error 0,1
+  exec $MYSQL --user=root --host=127.0.0.1 --port=$NODE_MYPORT_1 test
+         < $MYSQLTEST_VARDIR/tmp/galera_concurrent.sql &
+       $MYSQL --user=root --host=127.0.0.1 --port=$NODE_MYPORT_2 test
+         < $MYSQLTEST_VARDIR/tmp/galera_concurrent.sql;
+  dec $run;
+}
+
+--remove_file $MYSQLTEST_VARDIR/tmp/galera_concurrent.sql
+
+--source include/galera_end.inc
+--echo # End of test
+

policy/apparmor/README

@@ -0,0 +1,5 @@
+Note: The included AppArmor profiles can be used for MariaDB Galera cluster.
+However, since these profiles had been tested for a limited set of scenarios,
+it is highly recommended to run them in "complain" mode and report any denials
+on mariadb.org/jira.
+

policy/apparmor/usr.sbin.mysqld

@@ -0,0 +1,150 @@
+# Last Modified: Fri Mar  1 18:55:47 2013
+# Based on usr.sbin.mysqld packaged in mysql-server in Ubuntu.
+# This AppArmor profile has been copied under BSD License from
+# Percona XtraDB Cluster, along with some additions.
+
+#include <tunables/global>
+
+/usr/sbin/mysqld flags=(complain) {
+  #include <abstractions/base>
+  #include <abstractions/mysql>
+  #include <abstractions/nameservice>
+  #include <abstractions/user-tmp>
+  #include <abstractions/winbind>
+
+  capability chown,
+  capability dac_override,
+  capability setgid,
+  capability setuid,
+  capability sys_rawio,
+  capability sys_resource,
+
+  network tcp,
+
+  /bin/dash rcx,
+  /dev/dm-0 r,
+  /etc/gai.conf r,
+  /etc/group r,
+  /etc/hosts.allow r,
+  /etc/hosts.deny r,
+  /etc/ld.so.cache r,
+  /etc/mtab r,
+  /etc/my.cnf r,
+  /etc/mysql/*.cnf r,
+  /etc/mysql/*.pem r,
+  /etc/mysql/conf.d/ r,
+  /etc/mysql/conf.d/* r,
+  /etc/nsswitch.conf r,
+  /etc/passwd r,
+  /etc/services r,
+  /run/mysqld/mysqld.pid w,
+  /run/mysqld/mysqld.sock w,
+  /sys/devices/system/cpu/ r,
+  owner /tmp/** lk,
+  /tmp/** rw,
+  /usr/lib/mysql/plugin/ r,
+  /usr/lib/mysql/plugin/*.so* mr,
+  /usr/sbin/mysqld mr,
+  /usr/share/mysql/** r,
+  /var/lib/mysql/ r,
+  /var/lib/mysql/** rwk,
+  /var/log/mysql.err rw,
+  /var/log/mysql.log rw,
+  /var/log/mysql/ r,
+  /var/log/mysql/* rw,
+  /var/run/mysqld/mysqld.pid w,
+  /var/run/mysqld/mysqld.sock w,
+
+
+  profile /bin/dash flags=(complain) {
+    #include <abstractions/base>
+    #include <abstractions/bash>
+    #include <abstractions/mysql>
+    #include <abstractions/nameservice>
+    #include <abstractions/perl>
+
+
+
+    /bin/cat rix,
+    /bin/dash rix,
+    /bin/date rix,
+    /bin/grep rix,
+    /bin/nc.openbsd rix,
+    /bin/netstat rix,
+    /bin/ps rix,
+    /bin/rm rix,
+    /bin/sed rix,
+    /bin/sleep rix,
+    /bin/tar rix,
+    /bin/which rix,
+    /dev/tty rw,
+    /etc/ld.so.cache r,
+    /etc/my.cnf r,
+    /proc/ r,
+    /proc/*/cmdline r,
+    /proc/*/fd/ r,
+    /proc/*/net/dev r,
+    /proc/*/net/if_inet6 r,
+    /proc/*/net/tcp r,
+    /proc/*/net/tcp6 r,
+    /proc/*/stat r,
+    /proc/*/status r,
+    /proc/sys/kernel/pid_max r,
+    /proc/tty/drivers r,
+    /proc/uptime r,
+    /proc/version r,
+    /sbin/ifconfig rix,
+    /sys/devices/system/cpu/ r,
+    /tmp/** rw,
+    /usr/bin/cut rix,
+    /usr/bin/dirname rix,
+    /usr/bin/gawk rix,
+    /usr/bin/innobackupex rix,
+    /usr/bin/mysql rix,
+    /usr/bin/perl rix,
+    /usr/bin/seq rix,
+    /usr/bin/wsrep_sst* rix,
+    /usr/bin/wsrep_sst_common r,
+    /usr/bin/xtrabackup* rix,
+    /var/lib/mysql/ r,
+    /var/lib/mysql/** rw,
+    /var/lib/mysql/*.log w,
+    /var/lib/mysql/*.err w,
+
+# MariaDB additions
+    ptrace peer=@{profile_name},
+
+    /bin/hostname rix,
+    /bin/ip rix,
+    /bin/mktemp rix,
+    /bin/ss rix,
+    /bin/sync rix,
+    /bin/touch rix,
+    /bin/uname rix,
+    /etc/mysql/*.cnf r,
+    /etc/mysql/conf.d/ r,
+    /etc/mysql/conf.d/* r,
+    /proc/*/attr/current r,
+    /proc/*/fdinfo/* r,
+    /proc/*/net/* r,
+    /proc/locks r,
+    /proc/sys/net/ipv4/ip_local_port_range r,
+    /run/mysqld/mysqld.sock rw,
+    /sbin/ip rix,
+    /usr/bin/basename rix,
+    /usr/bin/du rix,
+    /usr/bin/find rix,
+    /usr/bin/lsof rix,
+    /usr/bin/my_print_defaults rix,
+    /usr/bin/mysqldump rix,
+    /usr/bin/pv rix,
+    /usr/bin/rsync rix,
+    /usr/bin/socat rix,
+    /usr/bin/tail rix,
+    /usr/bin/timeout rix,
+    /usr/bin/xargs rix,
+    /usr/bin/xbstream rix,
+  }
+  # Site-specific additions and overrides. See local/README for details.
+  #include <local/usr.sbin.mysqld>
+}

policy/apparmor/usr.sbin.mysqld.local

@@ -0,0 +1,4 @@
+# Site-specific additions and overrides for usr.sbin.mysqld..
+# For more details, please see /etc/apparmor.d/local/README.
+# This AppArmor profile has been copied under BSD License from
+# Percona XtraDB Cluster, along with some additions.

policy/selinux/README

@@ -0,0 +1,20 @@
+Note: The included SELinux policy files can be used for MariaDB Galera cluster.
+However, since these policies had been tested for a limited set of scenarios,
+it is highly recommended that you run mysqld in "permissive" mode even with
+these policies installed and report any denials on mariadb.org/jira.
+
+
+How to generate and load the policy module of MariaDB Galera cluster ?
+  * Generate the SELinux policy module.
+    # cd <source>/policy/selinux/
+    # make -f /usr/share/selinux/devel/Makefile mariadb-server.pp
+
+  * Load the generated policy module.
+    # semodule -i /path/to/mariadb-server.pp
+
+  * Lastly, run the following command to allow tcp/4568 and udp/4567.
+    # semanage port -a -t mysqld_port_t -p tcp 4568
+    # semanage port -a -t mysqld_port_t -p udp 4567
+
+How to run mysqld in permissve mode ?
+  # semanage permissive -a mysqld_t

policy/selinux/mariadb-server.fc

@@ -0,0 +1,10 @@
+# This SELinux file contexts (.fc) file has been copied under BSD License from
+# Percona XtraDB Cluster.
+
+/etc/init\.d/rc\.d/mysql -- gen_context(system_u:object_r:mysqld_initrc_exec_t,s0)
+/var/lib/mysql/.*\.log --  gen_context(system_u:object_r:mysqld_log_t,s0)
+/var/lib/mysql/.*\.err --  gen_context(system_u:object_r:mysqld_log_t,s0)
+/var/lib/mysql/.*\.pid -- gen_context(system_u:object_r:mysqld_var_run_t,s0)
+/var/lib/mysql/.*\.cnf	--  gen_context(system_u:object_r:mysqld_etc_t,s0)
+/usr/bin/xtrabackup.* -- gen_context(system_u:object_r:mysqld_exec_t,s0)
+/usr/bin/wsrep.*  -- gen_context(system_u:object_r:mysqld_safe_exec_t,s0)

policy/selinux/mariadb-server.te

@@ -0,0 +1,99 @@
+# This SELinux type enforcement (.te) file has been copied under BSD License
+# from Percona XtraDB Cluster, along with some additions.
+
+module mariadb-server 1.0;
+
+require {
+	type user_tmp_t;
+	#type kerberos_master_port_t;
+	type mysqld_safe_t;
+	type tmp_t;
+	type tmpfs_t;
+	type hostname_exec_t;
+	type ifconfig_exec_t;
+	type sysctl_net_t;
+	type proc_net_t;
+	type port_t;
+	type mysqld_t;
+	type var_lib_t;
+	type rsync_exec_t;
+	type bin_t;
+	type shell_exec_t;
+	type anon_inodefs_t;
+	type fixed_disk_device_t;
+	class lnk_file read;
+	class process { getattr signull };
+	class unix_stream_socket connectto;
+	class capability { sys_resource sys_nice };
+	class tcp_socket { name_bind name_connect };
+	class file { execute setattr read create getattr execute_no_trans write ioctl open append unlink };
+	class sock_file { create unlink getattr };
+	class blk_file { read write open };
+	class dir { write search getattr add_name read remove_name open };
+
+# MariaDB additions
+	type kerberos_port_t;
+	type tram_port_t;
+	type mysqld_port_t;
+	class udp_socket name_bind;
+	class process setpgid;
+	class netlink_tcpdiag_socket { create nlmsg_read };
+}
+
+
+#============= mysqld_safe_t ==============
+allow mysqld_safe_t mysqld_t:process signull;
+allow mysqld_safe_t self:capability { sys_resource sys_nice };
+allow mysqld_safe_t tmp_t:file { create read write open getattr unlink ioctl setattr };
+allow mysqld_safe_t tmp_t:dir { write remove_name add_name };
+allow mysqld_safe_t tmp_t:sock_file { getattr unlink };
+allow mysqld_safe_t user_tmp_t:sock_file { getattr unlink };
+allow mysqld_safe_t var_lib_t:dir { write add_name };
+allow mysqld_safe_t var_lib_t:file { write ioctl setattr create open getattr append unlink };
+
+#============= mysqld_t ==============
+allow mysqld_t anon_inodefs_t:file write;
+allow mysqld_t tmp_t:sock_file { create unlink };
+allow mysqld_t tmpfs_t:dir { write search read remove_name open add_name };
+allow mysqld_t tmpfs_t:file { write getattr read create unlink open };
+allow mysqld_t fixed_disk_device_t:blk_file { read write open };
+allow mysqld_t ifconfig_exec_t:file { read execute open execute_no_trans getattr };
+
+#This rule allows connecting on 4444/4567/4568
+#allow mysqld_t kerberos_master_port_t:tcp_socket { name_bind name_connect };
+
+allow mysqld_t mysqld_safe_t:dir { getattr search };
+allow mysqld_t mysqld_safe_t:file { read open };
+allow mysqld_t self:unix_stream_socket connectto;
+allow mysqld_t port_t:tcp_socket { name_bind name_connect };
+allow mysqld_t proc_net_t:file { read getattr open };
+allow mysqld_t sysctl_net_t:dir search;
+allow mysqld_t var_lib_t:file { getattr open append };
+allow mysqld_t var_lib_t:sock_file { create unlink getattr };
+allow mysqld_t rsync_exec_t:file { read getattr open execute execute_no_trans };
+allow mysqld_t self:process getattr;
+allow mysqld_t hostname_exec_t:file { read getattr execute open execute_no_trans };
+allow mysqld_t user_tmp_t:dir { write add_name };
+allow mysqld_t user_tmp_t:file create;
+allow mysqld_t bin_t:lnk_file read;
+allow mysqld_t tmp_t:file { append create read write open getattr unlink setattr };
+
+# Allows too much leeway - the xtrabackup/wsrep rules in fc should fix it, but
+# keep for the moment.
+allow mysqld_t shell_exec_t:file { execute_no_trans getattr read execute open };
+allow mysqld_t bin_t:file { getattr read execute open execute_no_trans ioctl };
+
+# MariaDB additions
+allow mysqld_t self:process setpgid;
+# This rule allows port tcp/4444
+allow mysqld_t kerberos_port_t:tcp_socket { name_bind name_connect };
+# This rule allows port tcp/4567 (tram_port_t may not be available on
+# older versions)
+allow mysqld_t tram_port_t:tcp_socket name_bind;
+# This rule allows port udp/4567 (see README)
+allow mysqld_t mysqld_port_t:udp_socket name_bind;
+
+# Rules related to XtraBackup
+allow mysqld_t self:netlink_tcpdiag_socket { create nlmsg_read };
+allow mysqld_t sysctl_net_t:file { read getattr open };
+

sql/mdl.cc

@@ -1443,6 +1443,13 @@ MDL_wait::timed_wait(MDL_context_owner *owner, struct timespec *abs_timeout,
   while (!m_wait_status && !owner->is_killed() &&
          wait_result != ETIMEDOUT && wait_result != ETIME)
   {
+#ifdef WITH_WSREP
+    if (wsrep_thd_is_BF(owner->get_thd(), true))
+    {
+      wait_result= mysql_cond_wait(&m_COND_wait_status, &m_LOCK_wait_status);
+    }
+    else
+#endif
     wait_result= mysql_cond_timedwait(&m_COND_wait_status, &m_LOCK_wait_status,
                                       abs_timeout);
   }
@@ -1527,12 +1534,15 @@ void MDL_lock::Ticket_list::add_ticket(MDL_ticket *ticket)
         WSREP_DEBUG("MDL add_ticket inserted before: %lu %s",
                     wsrep_thd_thread_id(waiting->get_ctx()->wsrep_get_thd()),
                     wsrep_thd_query(waiting->get_ctx()->wsrep_get_thd()));
+        /* Insert the ticket before the first non-BF waiting thd. */
         m_list.insert_after(prev, ticket);
         added= true;
       }
       prev= waiting;
     }
-    if (!added)   m_list.push_back(ticket);
+
+    /* Otherwise, insert the ticket at the back of the waiting list. */
+    if (!added) m_list.push_back(ticket);
 
     while ((granted= itg++))
     {

sql/sql_class.h

@@ -4172,6 +4172,8 @@ class select_insert :public select_result_interceptor {
   virtual int send_data(List<Item> &items);
   virtual void store_values(List<Item> &values);
   virtual bool can_rollback_data() { return 0; }
+  bool prepare_eof();
+  bool send_ok_packet();
   bool send_eof();
   virtual void abort_result_set();
   /* not implemented: select_insert is never re-used in prepared statements */

sql/sql_insert.cc

@@ -3660,14 +3660,14 @@ void select_insert::store_values(List<Item> &values)
                                          TRG_EVENT_INSERT);
 }
 
-bool select_insert::send_eof()
+bool select_insert::prepare_eof()
 {
   int error;
   bool const trans_table= table->file->has_transactions();
-  ulonglong id, row_count;
   bool changed;
   killed_state killed_status= thd->killed;
-  DBUG_ENTER("select_insert::send_eof");
+
+  DBUG_ENTER("select_insert::prepare_eof");
   DBUG_PRINT("enter", ("trans_table=%d, table_type='%s'",
                        trans_table, table->file->table_type()));
 
@@ -3709,11 +3709,10 @@ bool select_insert::send_eof()
   */
 #ifdef WITH_WSREP
   if ((WSREP_EMULATE_BINLOG(thd) || mysql_bin_log.is_open()) &&
-      (!error || thd->transaction.stmt.modified_non_trans_table))
 #else
   if (mysql_bin_log.is_open() &&
-      (!error || thd->transaction.stmt.modified_non_trans_table))
 #endif
+      (!error || thd->transaction.stmt.modified_non_trans_table))
   {
     int errcode= 0;
     if (!error)
@@ -3725,7 +3724,7 @@ bool select_insert::send_eof()
                       trans_table, FALSE, FALSE, errcode))
     {
       table->file->ha_release_auto_increment();
-      DBUG_RETURN(1);
+      DBUG_RETURN(true);
     }
   }
   table->file->ha_release_auto_increment();
@@ -3733,27 +3732,49 @@ bool select_insert::send_eof()
   if (error)
   {
     table->file->print_error(error,MYF(0));
-    DBUG_RETURN(1);
+    DBUG_RETURN(true);
   }
-  char buff[160];
+
+  DBUG_RETURN(false);
+}
+
+bool select_insert::send_ok_packet() {
+  char  message[160];                           /* status message */
+  ulong row_count;                              /* rows affected */
+  ulong id;                                     /* last insert-id */
+
+  DBUG_ENTER("select_insert::send_ok_packet");
+
   if (info.ignore)
-    sprintf(buff, ER(ER_INSERT_INFO), (ulong) info.records,
+    my_snprintf(message, sizeof(message), ER(ER_INSERT_INFO),
-	    (ulong) (info.records - info.copied),
+                (ulong) info.records, (ulong) (info.records - info.copied),
-            (long) thd->get_stmt_da()->current_statement_warn_count());
+                (long) thd->get_stmt_da()->current_statement_warn_count());
   else
-    sprintf(buff, ER(ER_INSERT_INFO), (ulong) info.records,
+    my_snprintf(message, sizeof(message), ER(ER_INSERT_INFO),
-	    (ulong) (info.deleted+info.updated),
+                (ulong) info.records, (ulong) (info.deleted + info.updated),
-            (long) thd->get_stmt_da()->current_statement_warn_count());
+                (long) thd->get_stmt_da()->current_statement_warn_count());
+
   row_count= info.copied + info.deleted +
-             ((thd->client_capabilities & CLIENT_FOUND_ROWS) ?
+    ((thd->client_capabilities & CLIENT_FOUND_ROWS) ?
-              info.touched : info.updated);
+     info.touched : info.updated);
+
   id= (thd->first_successful_insert_id_in_cur_stmt > 0) ?
     thd->first_successful_insert_id_in_cur_stmt :
     (thd->arg_of_last_insert_id_function ?
      thd->first_successful_insert_id_in_prev_stmt :
      (info.copied ? autoinc_value_of_last_inserted_row : 0));
-  ::my_ok(thd, row_count, id, buff);
+
-  DBUG_RETURN(0);
+  ::my_ok(thd, row_count, id, message);
+
+  DBUG_RETURN(false);
+}
+
+bool select_insert::send_eof()
+{
+  bool res;
+  DBUG_ENTER("select_insert::send_eof");
+  res= (prepare_eof() || send_ok_packet());
+  DBUG_RETURN(res);
 }
 
 void select_insert::abort_result_set() {
@@ -4234,13 +4255,13 @@ void select_create::store_values(List<Item> &values)
 
 bool select_create::send_eof()
 {
-  if (select_insert::send_eof())
+  DBUG_ENTER("select_create::send_eof");
+  if (prepare_eof())
   {
     abort_result_set();
-    return 1;
+    DBUG_RETURN(true);
   }
 
-  exit_done= 1;                                 // Avoid double calls
   /*
     Do an implicit commit at end of statement for non-temporary
     tables.  This can fail, but we should unlock the table
@@ -4259,7 +4280,7 @@ bool select_create::send_eof()
                     thd->thread_id, thd->wsrep_conflict_state, thd->query());
         mysql_mutex_unlock(&thd->LOCK_wsrep_thd);
         abort_result_set();
-	return TRUE;
+        DBUG_RETURN(true);
       }
       mysql_mutex_unlock(&thd->LOCK_wsrep_thd);
 #endif /* WITH_WSREP */
@@ -4267,9 +4288,17 @@ bool select_create::send_eof()
   else if (!thd->is_current_stmt_binlog_format_row())
     table->s->table_creation_was_logged= 1;
 
+  /*
+    exit_done must only be set after last potential call to
+    abort_result_set().
+  */
+  exit_done= 1;                                 // Avoid double calls
+
   table->file->extra(HA_EXTRA_NO_IGNORE_DUP_KEY);
   table->file->extra(HA_EXTRA_WRITE_CANNOT_REPLACE);
 
+  send_ok_packet();
+
   if (m_plock)
   {
     MYSQL_LOCK *lock= *m_plock;
@@ -4290,12 +4319,12 @@ bool select_create::send_eof()
                                                 create_info->
                                                 pos_in_locked_tables,
                                                 table, lock))
-        return 0;                               // ok
+        DBUG_RETURN(false);                     // ok
       /* Fail. Continue without locking the table */
     }
     mysql_unlock_tables(thd, lock);
   }
-  return 0;
+  DBUG_RETURN(false);
 }
 
 

sql/wsrep_mysqld.cc

@@ -1461,10 +1461,21 @@ void wsrep_to_isolation_end(THD *thd)
       gra->wsrep_exec_mode, gra->wsrep_query_state, gra->wsrep_conflict_state, \
       gra->get_command(), gra->lex->sql_command, gra->query());
 
+/**
+  Check if request for the metadata lock should be granted to the requester.
+
+  @param  requestor_ctx        The MDL context of the requestor
+  @param  ticket               MDL ticket for the requested lock
+
+  @retval TRUE   Lock request can be granted
+  @retval FALSE  Lock request cannot be granted
+*/
+
 bool
 wsrep_grant_mdl_exception(MDL_context *requestor_ctx,
                           MDL_ticket *ticket
 ) {
+  /* Fallback to the non-wsrep behaviour */
   if (!WSREP_ON) return FALSE;
 
   THD *request_thd  = requestor_ctx->wsrep_get_thd();

sql/wsrep_thd.cc

@@ -579,6 +579,17 @@ int wsrep_abort_thd(void *bf_thd_ptr, void *victim_thd_ptr, my_bool signal)
            bf_thd->wsrep_exec_mode == TOTAL_ORDER) )                         &&
        victim_thd)
   {
+    if ((victim_thd->wsrep_conflict_state == MUST_ABORT) ||
+        (victim_thd->wsrep_conflict_state == ABORTED) ||
+        (victim_thd->wsrep_conflict_state == ABORTING))
+    {
+      WSREP_DEBUG("wsrep_abort_thd called by %llu with victim %llu already "
+                  "aborted. Ignoring.",
+                  (bf_thd) ? (long long)bf_thd->real_id : 0,
+                  (long long)victim_thd->real_id);
+      DBUG_RETURN(1);
+    }
+
     WSREP_DEBUG("wsrep_abort_thd, by: %llu, victim: %llu", (bf_thd) ?
                 (long long)bf_thd->real_id : 0, (long long)victim_thd->real_id);
     ha_wsrep_abort_transaction(bf_thd, victim_thd, signal);