MDEV-7565: Server crash with Signal 6 (part 2)

followup test suite and its fix.

Problem was in rewriting left expression which had 2 references on it. Solved with making subselect reference main.

Item_in_optimized can have not Item_in_subselect reference in left part so type casting with no check is dangerous.

Item::cols() should be checked after Item::fix_fields().

Preparation of subselect moved earlier (before checks which needs it prepared).

MDEV-8845: Table disappear after modifying FK

Added test case.

Fixed incorrect access to freed memory.

Made no_rows_in_result()/restore_to_before_no_rows_in_result() not looking
annecessary deep with walk() method.

The patch was pushed into 10.0, but it needs to be applied to 5.5 as well

The test restarts the server and expects that the feedback plugin
will send a report on shutdown, and will write about it in the error
log. But the server is only given 10 sec to shut down properly,
which is not always enough.
Added a parameter to restart_mysqld.inc, and set it to a bigger
value in feedback_plugin_send

The culprit is the feedback_plugin_load test, which is run both separately
and from inside feedback_plugin_send.test. The test queries I_S.FEEDBACK,
and every time it does, 'FEEDBACK used' value is increased.
Fixed by checking that the value is increased instead of recording the actual
value in the result file.

The test would take really long if it was run without --mem.
Fixed by adding --innodb-flush-log-at-trx-commit=2

The real problem is that when innodb.xa_recovery test intentionally
crashes the server, system tables can be opened and marked as crashed,
and the next test in line gets blamed for the error which appears
in the error log.
Fixed by flushing the tables before crashing the server

Backport of upstream patch. revno: 5696

MDEV-8663: IF Statement returns multiple values erroneously (or Assertion `!null_value' failed in Item::send(Protocol*, String*))

Postreview addons by Bar

Fix: keeping contract: NULL value mean NULL pointer in val_str and val_deciman.

THD::>save_prep_leaf_list was set to true by multi-table update
statements with mergeable selects and never reset.

Make every statement reset it at start.

5.5.44-37.3

Correct fix for this bug.

The problem was that Item_func_group_concat() was calling
setup_order(), passing args as the second argument,
ref_pointer_array. While ref_pointer_array should have free
space at the end, as setup_order() can append elements to it.

In this particular case args[] elements were overwritten when
setup_order() was pushing new elements into ref_pointer_array.

Cherry-pick from 10.0:

  commit 126523d1
  Author: Sergei Golubchik <serg@mariadb.org>
  Date:   Mon Feb 23 20:53:41 2015 +0100

  MDEV-6703 Add "mysqlbinlog --binlog-row-event-max-size" support

Alternative fix that doesn't cause view.test crash in --ps:
Remember when Item_ref was fixed right in the constructor
and did not have a full Item_ref::fix_fields() call. Later
in PS/SP, after Item_ref::cleanup, we use this knowledge
to avoid doing full fix_fields() for items that were never
supposed to be fix_field'ed.

Simplify the test case.

Let mysql_upgrade return zero exit status when installation is up to date.

            execution of PS

GROUP_CONCAT() with ORDER BY column position may crash server on PS reexecution.

The problem was that arguments array of GROUP_CONCAT() was adjusted to point to
temporary elements (resolved ORDER BY fields) during first execution.

This patch expands rev. 08763096 to restore original arguments array as well.

There is several different ways to incorrectly define
foreign key constraint. In many cases earlier MariaDB
versions the error messages produced by these cases
are not very clear and helpful. This patch improves
the warning messages produced by foreign key parsing.

Added better error message that will be printed when foreign key
constraint name in create table is not unique in database.

GET_LOCK() silently accepted negative values and NULL for timeout.
Fixed GET_LOCK() to issue a warning and return NULL in such cases.

Fixed memory loss detected on P8. This can happen when we call after_flush but never call after_rollback() or after_commit().

The old code used pthread_setspecific() to store temporary data used by the thread.
This is not safe when used with thread pool, as the thread may change for the transaction.

The fix is to save the data in THD, which is guaranteed to be properly freed.
I also fixed the code so that we don't do a malloc() for every transaction.

field.cc
- Fixed warning about overlapping memory copy (backport from 10.0)

Item_subselect.cc
- Fixed core dump in main.view
- Problem was that thd->lex->current_select->master_unit()->item was not set, which caused crash in maxr_as_dependent

sql/mysqld.cc
- Got error on shutdown as we where freeing mutex before all THD objects was freed
  (~THD uses some mutex). Fixed by during shutdown freeing THD inside mutex.

sql/log.cc
- log_space_lock and LOCK_log where locked in inconsistenly. Fixed by not having a log_space_lock around purge_logs.

sql/slave.cc
- Remove unnecessary log_space_lock
- Move cond_broadcast inside lock to ensure we don't miss the signal

Analysis: At check_trx_exists function InnoDB allocates
a new trx if no trx is found from thd but this newly
allocated trx is not registered to thd. This is unsafe,
because nothing prevents InnoDB plugin from being uninstalled
while there's active transaction. This can cause crashes, hang
and any other odd behavior. It may also corrupt stack, as
functions pointers are not available after dlclose.

Fix: The fix is to use thd_set_ha_data() when
manipulating per-connection handler data. It does appropriate
plugin locking.

The fix is that if the slave has a different integer size than
the master, then they will assume the master has the same signed/unsigned modifier
as the slave.

This means that one can safely change a coon the slave an int to a bigint
or an unsigned int to an unsigned int.  Changing an unsigned int to an
signed bigint will cause replication failures when the high bit of the
unsigned int is set.

We can't give an error if the signess is different on the master and slave
as the binary log doesn't contain the signess of the column on the master.

The fix is that if the slave has a different integer size than
the master, then they will assume the master has the same signed/unsigned modifier
as the slave.

This means that one can safely change a coon the slave an int to a bigint
or an unsigned int to an unsigned int.  Changing an unsigned int to an
signed bigint will cause replication failures when the high bit of the
unsigned int is set.

We can't give an error if the signess is different on the master and slave
as the binary log doesn't contain the signess of the column on the master.

Fix for MDEV-8301;  Statistics for a thread could be counted twice in SHOW STATUS while thread was ending

Fixed by adding a marker if we have added the thread statistics to the global counters.

- Removing use of calls to current_thd
- More DBUG_PRINT
- Code style changes
- Made some local functions static
Ensure that calls to print_keyuse are locked with mutex to get all lines in same debug packet

SELECT ... WHERE XX IN (SELECT YY)
this was transformed to something like:
SELECT ... WHERE IF_EXISTS(SELECT ... HAVING XX=YY)

The bug was that for normal execution XX was fixed in the original outer SELECT context while in PS it was fixed in the sub query context and this confused the optimizer.

Fixed by ensuring that XX is always fixed in the outer context.