If a transaction T1 needs to wait for a transaction T2, T2's commit will
skip the normal binlog_commit_wait_usec delay, in order not to needlessly
stall throughput.

This works by checking if T2 is already ready to commit. If so, it is woken
up. If not, we set a flag in T2 so that when it gets ready to commit, it
will do so immediately.

But there was a potential race due to insufficient locking, if T2 gets ready
to commit just at the point where T1 does the check. If the race hits, the
wakeup (and early commit) of T2 might be lost.

The race is only theoretical (from code inspection, no known test case), but
seems best to fix it anyway, by properly locking LOCK_prepare_ordered around
the check.

MDEV-8725: Assertion `!(thd->rgi_slave && thd-> rgi_slave->did_mark_start_commit)' failed in ha_rollback_trans

The assertion is there to catch cases where we rollback while
mark_start_commit() is active. This can allow following event groups
to be replicated too early, causing conflicts.

But in this case, we have an _explicit_ ROLLBACK event in the binlog,
which should not assert.

We fix this by delaying the mark_start_commit() in the explicit
ROLLBACK case. It seems safest to delay this in ROLLBACK case anyway,
and there should be no reason to try to optimise this corner case.

and vcol_supported_sql_funcs_innodb. Moving the test for using CHARSET(),
COLLATION(), COERCIBILITY() in virtual column from
vcol_supported_sql_funcs_xxx to vcol_blocked_sql_funcs_xxx,
as these functions are not supported in virtual columns any longer.
Discussed with Sanja on IRC.

- If run with valgrind, mysqltest will now wait longer when syncronizing slave with master
- Ensure that we wait with cleanup() until slave thread has stopped.
- Added signal_thd_deleted() to signal close_connections() that all THD's has been freed.
- Check in handle_fatal_signal() that we don't use variables that has been freed.
- Increased some timeouts when run with --valgrind

Other things:
- Fixed wrong test in one_thread_per_connection_end() if galera is used.
- Removed not needed calls to THD_CHECK_SENTRY() when we are calling 'delete thd'.

Part 5: Removing calls to current_thd in net_read calls, creating fields,
        query_cache, acl and some other places where thd was available

MDEV-8722 The patch for MDEV-8688 disabled equal field propagation for temporal column and BETWEEN and IN
Item::cmp_context was inconsistently used in combination with cmp_type()
and result_type() in different places of the code. Fixed to use cmp_type()
in all places where cmp_context is involved, to avoid unexpected results
for temporal data types (which have result_type()==STRING_RESULT and
cmp_type==TIME_RESULT).

SELECT..WHERE varchar_column=DATE'2001-01-01' AND varchar_column='2001-01-01'

The fix for MDEV-8688 fixed MDEV-8680 as well. Just adding a test case.

          SELECT..WHERE varchar_column IN (1,2,3) AND varchar_column=' 1';

Item_func_in::compare_collation() and Item_func_between::compare_collation(),
and adding Item_func_opt_neg::compare_collation() instead.

parent Item_func_opt_neg. A pre-requisite patch for a number of
upcoming equal field propagation related bug fixes.

Note, the patch for MDEV-8661 unintentionally fixed MDEV-8694 as well,
as a side effect. Adding a real clear fix: implementing
Item_func_like::propagate_equal_fields() with comments.

- Part 4: Removing calls to sql_alloc() and sql_calloc()

Other things:
- Added current_thd in some places to make it clear that it's called (easier to remove later)
- Move memory allocation from Item_func_case::fix_length_and_dec() to Item_func_case::fix_fields()
- Added mem_root to some new calls
- Fixed some wrong UNINIT_VAR() calls
- Fixed a bug in generate_partition_syntax() in case of errors
- Added mem_root to argument to new thread_info
- Simplified my_parse_error() call in sql_yacc.yy

- Part 3: Adding mem_root to push_back() and push_front()

Other things:
- Added THD as an argument to some partition functions.
- Added memory overflow checking for XML tag's in read_xml()

MDEV-8679 Equal field propagation is not used for VARCHAR when it safely could

Evict table metric

port of work by dveeden:
https://github.com/dveeden/mysql-server/commit/555ca34958a8ed191f34eeb2333e0d7c30e12fce

This is a port of dveeden's work:

https://github.com/dveeden/mysql-server/commit/e9d00e859e3854bf8f71873caa69c060e670350b

(Mostly changes in linenumbers, a bit of whitespace fix-up.)

- foreign_keys: adjusted according to code changes;
- type_spatial: adjusted according to code changes;
- type_spatial_indexes (for MyISAM): disabled till MDEV-8675 is fixed

Added MTR suppressions for database corruption that is intentionally
caused in test.

Restored self-initialization version of UNINIT_VAR() for all gcc versions.
Fixed UNINIT_VAR() usage: it is supposed to be used along with declaration.

- Added mem_root to all calls to new Item
- Added private method operator new(size_t size) to Item to ensure that
  we always use a mem_root when creating an item.

This saves use once call to current_thd per Item creation

Added mandatory thd parameter to Item (and all derivative classes) constructor.
Added thd parameter to all routines that may create items.
Also removed "current_thd" from Item::Item. This reduced number of
pthread_getspecific() calls from 290 to 177 per OLTP RO transaction.

- UNINIT_VAR() was required for 4.8.3 on openSUSE 13.2

Ensure that fields declared with NOT NULL doesn't have DEFAULT values if not specified and if not timestamp or auto_increment

In original code, sometimes one got an automatic DEFAULT value in some cases, in other cases not.

For example:
create table t1 (a int primary key)      - No default
create table t2 (a int, primary key(a))  - DEFAULT 0
create table t1 SELECT ....              - Default for all fields, even if they where defined as NOT NULL
ALTER TABLE ... MODIFY could sometimes add an unexpected DEFAULT value.

The patch is quite big because we had some many test cases that used
CREATE ... SELECT or CREATE ... (...PRIMARY KEY(xxx)) which doesn't have an automatic DEFAULT anymore.

Other things:
- Removed warnings from InnoDB when waiting from semaphore (got this when testing things with --big)

Issue was two fold (both in MyISAM and Aria)
- optimize and repair failed if there was an old .TMM file around. As optimized and repair are protected against multiple execution, I decided to change so that we just truncate the file if it exists.
- I had missed to check for error condition if creation of the temporary index file failed. This caused the strange behaviour that it looked as if optimized would have worked once.

MDEV-7649 wrong result when comparing utf8 column with an invalid literal
This is a preparatory patch for:
MDEV-8433 Make field<'broken-string' use indexes

This is a pre-requisite patch for:
- MDEV-8433 Make field<'broken-string' use indexes
- MDEV-8625 Bad result set with ignorable characters when using a prefix key
- MDEV-8626 Bad result set with contractions when using a prefix key

Analysis: Problem was that when a new tablespace is created a default
encryption info is also created and stored to the tablespace. Later a
new encryption information was created with correct key_id but that
does not affect on IV.

Fix: Push encryption mode and key_id to lower levels and create
correct encryption info when a new tablespace is created.

This fix does not contain test case because, currently incorrect
encryption key causes page corruption and a lot of error messages
to error log causing mtr to fail.