Remove too strict debug assertion.

Signed-off-by: Jan Lindström <jan.lindstrom@mariadb.com>

Sync waiting before processing SQLCOM_REPLACE was not necessary given that
this case falls through to processing of SQLCOM_INSERT. In case of
SQLCOM_REPLACE, wsrep_sync_wait would be called twice.
Signed-off-by: Jan Lindström <jan.lindstrom@mariadb.com>

The following commands are now subject to wsrep_sync_wait
with bitmask value 8:

SHOW BINARY LOGS
SHOW BINLOG EVENTS
SHOW GRANTS
Signed-off-by: Jan Lindström <jan.lindstrom@mariadb.com>

Previously, setting `wsrep_sync_wait = 1` would have an effect on
both SELECT and SHOW statements.
This patch changes wsrep_sync_wait so that bitmask value 1 is used
for SELECT statements, while bitmask value 8 is reserved for SHOW
statements.
It is still possible to achieve sync wait on both SELECT and SHOW
statements by setting `wsrep_sync_wait = 9`.
Signed-off-by: Jan Lindström <jan.lindstrom@mariadb.com>

Skipping wsrep extra FK check for applier and replayer
Signed-off-by: Jan Lindström <jan.lindstrom@mariadb.com>

Pushed fix for a typo
Signed-off-by: Jan Lindström <jan.lindstrom@mariadb.com>

Signed-off-by: Jan Lindström <jan.lindstrom@mariadb.com>

Signed-off-by: Jan Lindström <jan.lindstrom@mariadb.com>

Signed-off-by: Jan Lindström <jan.lindstrom@mariadb.com>

Inspired by: https://github.com/devexp-db/mariadb/blob/f27/mariadb-covscan-stroverflow.patch

Signed-off-by: Daniel Black <daniel.black@au.ibm.com>

test case

simplify.
add a test case.

MDEV-11240: Server crashes in check_view_single_update or Assertion `derived->table' failed in mysql_derived_merge_for_insert

Move table pointer for single table view (even if it is view over other view) to make the access universal.

        Set name_length properly in Item_idnet constructors.

end_io_call uses uninitialized values from the new_data_cache

As such we the buffer 0 and check this before calling end_io_cache on it.

Thanks Sergey Vojtovich for the review and for this solution.

Found by Coverity (ref 972481).

A few tests assumes that the CYCLE timer is always available,
which is not true on some platforms (e.g. ARM).
Fixing the tests not to reply on the CYCLE availability.

in ab785bfe

Coverity report this as:
CID 971840 (#1 of 1): Operands don't affect result (CONSTANT_EXPRESSION_RESULT)
result_independent_of_operands: 4 | (flags & 1) is always true regardless of the values of its operands. This occurs as the logical first operand of "?:".

The C order of precidence has | of higher precidence than ?:. The
intenting implies an | of the 3 terms.

Adjust to intented meaning.

Avoid using STDERR_FILENO. The server uses freopen(stderr),
so stderr can be on any file descriptor.

field_names[x][y] is a pointer

client/mysql.cc: In function 'void build_completion_hash(bool, bool)':
client/mysql.cc:2855:37: error: invalid conversion from 'char' to 'char*' [-fpermissive]
       field_names[i][num_fields*2]= '\0';
Signed-off-by: Daniel Black <daniel.black@au.ibm.com>

Same MDEV, second bug.
Merge buffer must fit at least MERGEBUFF2 (that is, 15) key values.
Because merge_index() can merge that many buffers, and
merge_many_buff() leaves that many buffers unmerged.

use the correct value for the merge_buffer size, max_in_memory_size
is too small and merge_walk() fails.

also: remove a cast.

already fixed by be55bbc2
just add a test case

only compare field references with an alias (from the SELECT clause)
when this reference doesn't specify an explicit table name part.

remove unused variable

already fixed by a7ed4644.
just add a test case,

upper->item can be NULL if we're referring to an aggregate function

Item_in_subselect::pushed_cond_guards[] array is allocated only when
left_expr->maybe_null. And it is used (for row expressions) when
left_expr->element_index(i)->maybe_null.

For left_expr being a multi-column subquery, its maybe_null is
always false when the subquery doesn't use tables (see
Item_singlerow_subselect::fix_length_and_dec()
and subselect_single_select_engine::fix_length_and_dec()),
otherwise it's always true.

But row elements can be NULL regardless, so let's always allocate
pushed_cond_guards for multi-column subqueries, no matter whether
its maybe_null was forced to true or false.