The bug was the result of the incomplete fix for bug lp bug 1008293.

the new file is fully synced to disk and binlog index. This fixes a window
where a crash would leave next server restart unable to detect that a crash
occured, causing recovery to fail.

MDEV-359: Fix another case where switch-off semisync could cause a race that ended with server crash.

This one was when the code releases and reaquires the lock with pthread_cond_wait() - and semisync is switched off meanwhile.

The semisync code does a fast-but-unsafe check for enabled or not without lock,
followed by a slow-but-safe check under lock. However, if the slow check failed,
the code still referenced not valid data (in an assert() expression), causing a
crash.

Fixed by not running the incorrect assert when semisync is disabled.

This is XtraDB bug lp:1015109, introduced by innodb_split_buf_pool_mutex.patch

Comment the offending assertion, until the fixed XtraDB is available

MySQL introduced a class Deferred_log_events. This class keeps a pointer
last_added. The code was keeping this pointer around even after the memory
pointed to was freed, and later comparing the bogus pointer against other
allocated memory. This is illegal, and can randomly produce false equal
comparisons depending on whatever the malloc() subsystem decides to return.

- put back the result encoding in func_in.result (messed up by kdiff3)
- update .result for other tests (checked)

Virtual columns of ENUM and SET data types were not supported properly
in the original patch that introduced virtual columns into MariaDB 5.2.
The problem was that for any  virtual column the patch used the 
interval_id field of the definition of the column in the frm file as
a reference to the virtual column expression.
The fix stores the optional interval_id of the virtual column in the
extended header of the virtual column expression. 

Attempt to make it easier to upgrade mysql->mariadb on Ubuntu precise.
It looks like we were missing conflicts: and replaces: on packages
mysql-server-5.5 and mysql-client-5.5.

make sure that find_date_time_item() is called before agg_arg_charsets_for_comparison().
optimize Item_func_conv_charset to avoid conversion if no string result is needed

On localized Windows versions, Windows uses localized time zone names and contain non-ASCII characters.  non-ASCII characters appear broken when displayed by clients
The fix is to declare system_time_zone variable to have UTF8 encoding and to convert tzname to UTF8.

mysql-test/suite/sphinx/sphinx.result:
  Removed sphinx_time, as it was depending on timing.
mysql-test/suite/sphinx/sphinx.test:
  Removed sphinx_time, as it was depending on timing.

mysqldump with --include-master-host-port putting quotes around port number
Patch from Stewart Smith

client/mysqldump.c:
  Remove quotes from MASTER_PORT

Fixed MDEV-306 / LP:1007967 - Assertion `table->file->stats.records > 0 || error' failed join_read_const_table on concurrent SELECT and DROP/ADD INDEX


sql/sql_table.cc:
  Added comment
storage/maria/ma_close.c:
  Don't store history if it's visible to all.
  This fixed the MDEV-306 bug
storage/maria/ma_delete_table.c:
  Removed old comment
  Delete history state for deleted tables
storage/maria/ma_info.c:
  More DBUG_PRINT
storage/maria/ma_open.c:
  More DBUG_PRINT

Analysis:
The fix for bug lp:985667 implements the method Item_subselect::no_rows_in_result()
for all main kinds of subqueries. The purpose of this method is to be called from
return_zero_rows() and set Items to some default value in the case when a query
returns no rows. Aggregates and subqueries require special treatment in this case.

Every implementation of Item_subselect::no_rows_in_result() called
Item_subselect::make_const() to set the subquery predicate to its default value
irrespective of where the predicate was located in the query. Once the predicate
was set to a constant it was never executed.

At the same time, the JOIN object of the fake select for UNIONs (the one used for
the final result of the UNION), was set after all subqueries in the union were
executed. Since we set the subquery as constant, it was never executed, and the
corresponding JOIN was never created.

In order to decide whether the result of NOT IN is NULL or FALSE, Item_in_optimizer
needs to check if the subquery result was empty or not. This is where we got the
crash, because subselect_union_engine::no_rows() checks for
unit->fake_select_lex->join->send_records, and the join object was NULL.

Solution:
If a subquery is in the HAVING clause it must be evaluated in order to know its
result, so that we can properly filter the result records. Once subqueries in the
HAVING clause are executed even in the case of no result rows, this specific
crash will be solved, because the UNION will be executed, and its JOIN will be
constructed. Therefore the fix for this crash is to narrow the fix for lp:985667,
and to apply Item_subselect::no_rows_in_result() only when the subquery predicate
is in the SELECT clause.