join_cache_level=6+

The patch fixes two similar bugs in the commit 8eeb689e
that added multi_range_read support to partitions. The commit opened
a possibility to join a partition table using BKA+MRR. However in some
cases it could lead to wrong results or even crashes.

This could happened when
- index condition pushdown was used to join the table or
- the joined table was an inner table of an outer join and 'not exist'
  optimization was applied or
- the join table was the inner table of a semi-join and the first match
  optimization was applied

The bugs were in the code of the call-back functions
- partition_multi_range_key_skip_record() and
- partition_multi_range_key_skip_index_tuple().
Each of this function consist only of an invocation of another function.
Yet a wrong parameter was passed at this invocation.

The fix was suggested by Sergey Petrunia and it is apparently in line
with original design.
The corresponding comprehensive test cases demonstrating the problems
caused by the bugs were constructed by me.

No adverse effects since this was made a null function in
6b53f9d7.

This function had the last remaining cmake CMP0026 violation.

cmake enabling -DENABLE_DTRACE=ON is particularlly noisy with CMP0026
errors.

Fixed in the same way as 6b53f9d7

(Backport to 10.3)
Partitioning storage now supports MRR but doesn't support Index Condition
Pushdown (aka ICP). This causes counter-intuitive query plans for queries
that use BKA and conditions that depend on index fields:
- If the condition refers to other tables, BKA's variant of ICP is used
   to handle it.
- If the condition depends on this table only, the optimizer will try to
  use regular ICP for it, which will fail because the storage engine
  doesn't support ICP.

Make the optimizer be smarter in the second case: if we were not able to
use regular ICP, use BKA's variant of ICP..

No functional changes.

Apply this in 10.3 so eventually it is merged to 10.4 and 10.5, which
in turn will have separate wrap-and-sort runs but only for new lines,
and not affecting badly next merges from 10.3 to 10.4 to 10.5.

Also remove a duplicate line that was found thanks to wrap-and-sort.

The Debian build tools add all run-time dependencies automatically. For the
mariadb-rocksdb-plugin the "${misc:Depends}, ${shlibs:Depends}" will become
(based on the build dependencies at build time):

  Depends: mariadb-server-10.4 (= 1:10.4.12-1), rocksdb-tools,
  libc6 (>= 2.29), liblz4-1 (>= 0.0~r130), libsnappy1v5 (>= 1.1.8),
  libstdc++6 (>= 7), libzstd1 (>= 1.3.3), zlib1g (>= 1:1.1.4)

There is no need to manually add run-time build dependencies. Only external
dependencies need to be defined, for example rocksdb-tools could come in
question for mariadb-plugin-rocksdb, and it will be eventually added when
downstream official Debian packaging is merged upstream in a later commit.

Context involves semicolon batching, and the error starts 10.2
No reproducible examples were made yet, but TCP trace suggests
multiple packets that are "squeezed" together (e.g overlong OK packet
that has a trailer which is belongs to another packet)

Remove thd->get_stmt_da()->set_skip_flush() when processing a batch.
skip_flush stems from the COM_MULTI code, which was checked in during
10.2 (and is never used)

The fix is confirmed to work, when evaluated by bug reporter (one of them)

We never reproduced it locally, with multiple tries
thus the root cause analysis is still missing.

Problem seems to be the fact that we did not enforce correct
applier thread numbers after every command that effects them.
Test changes only.

* Remove those tests that will not be supported on that release.
* Make sure that correct tests are disabled and have MDEVs
* Sort test names

This should not be merged upwards.

Problem seems to be the fact that we did not enforce correct
applier thread numbers after every command that effects them.
Test changes only.

* Remove those tests that will not be supported on that release.
* Make sure that correct tests are disabled and have MDEVs
* Sort test names

This should not be merged upwards.

The symbol SRV_MASTER_PURGE_INTERVAL became unused in
mysql/mysql-server@42f36919584e82c621dbec1e69fd05ab023c54c6
when separate purge threads were introduced in MySQL 5.6.5.

Do not materialize a semi-join nest if it contains a materialized derived
table /view that potentially can be subject to the split optimization.
Splitting of materialization of such nest would help, but currently there
is no code to support this technique.

- n_ext value may be less than dtuple_get_n_ext(dtuple) when PK is being
updated and new record inherits the externally stored fields from
delete mark old record.

- Added the test case in innodb_fts suite
- Updated copyright year in row0mysql.cc

Problem:
=======
After discarding the table, fts_optimize_thread aborts during shutdown.
InnoDB fails to remove the table from fts_optimize_wq and it leads to
the fts_optimize_thread to lookup for the auxiliary table and fails.

Fix:
====
While discarding the fts table, remove the table from fts_optimize_wq.

buf_flush_or_remove_pages(): Only define BUF_LRU_DROP_SEARCH_SIZE
and dependent code when the adaptive hash index has been enabled.

JIRA:https://jira.mariadb.org/browse/MDEV-17292

Closes #1443

Quotes added to output.

For automatic number of opened files limit take into account number of table instances for table cache

Use of freed memory is better caught by AddressSanitizer,
especially with ASAN_POISON_MEMORY_REGION that is aliased
by MEM_NOACCESS and UNIV_MEM_FREE.

The column INFORMATION_SCHEMA.INNODB_MUTEXES.NAME is not populated ever since
commit 2e814d47 applied the InnoDB changes from
MySQL 5.7.9 to MariaDB Server 10.2.2.

Since the same commit, the view is only providing information about
rw_lock_t, not any mutexes.

For now, let us convert the source code file name and line number of
the rw_lock_t creation into a name. A better option in the future might
be to store the information somewhere where it can be looked up by
mysql_pfs_key_t, and possibly to remove the CREATE_FILE and CREATE_LINE
columns.

Use FLOOR rounding for DECIMAL_RESULT item_expr in partition function.

See also original report:
http://bugs.debian.org/946671

Using mysqlhotcopy, the following error occurs:

DBD::mysql::db do failed: You can't use locks with log tables at
/usr/bin/mysqlhotcopy line 545.

Author:

Paul Szabo psz@maths.usyd.edu.au http://www.maths.usyd.edu.au/u/psz/
School of Mathematics and Statistics University of Sydney Australia

Race condition when innodb_lock_wait_timeout (default 50 seconds)
exceeds for 'send update', but information_schema.innodb_lock_waits
still sees this wait or it my exit by timeout. My occur on overloaded
host.

HASH_REPLACE(): allows to not travel through linked list twice
when HASH_INSERT() happens right after HASH_DELETE()

The merge commit 5ff66fb0
accidentally omitted part of
commit 07e34cdd.