purge_sys_t::n_submitted: Document that it is only accessed by
srv_purge_coordinator_thread.

purge_sys_t::n_completed: Exclusively use my_atomic access.

srv_task_execute(): Simplify the code.

srv_purge_coordinator_thread(): Test the cheaper condition first.

trx_purge(): Atomically access purge_sys.n_completed.
Remove some code duplication.

trx_purge_wait_for_workers_to_complete(): Atomically access
purge_sys.n_completed. Remove an unnecessary local variable.

trx_purge_stop(): Remove a redundant assignment.

row_ins_sec_index_entry(): Compare a pointer to fil_system.sys_space,
not to a numeric constant. This code was recently changed in MDEV-13637,
and the condition was essentially disabled, potentially causing the
change buffer to grow uncontrollably when something is inserted into
a table that has secondary indexes and resides in the system tablespace.

Thanks to Daniel Black for pointing out that clang 7 flagged a warning
for the comparison of a pointer to an integer.

row_import_for_mysql(): Fix a possible compiler warning.

There is no need for MariaDB Backup to be versioned. Other packages are
very unlikely to need to depend on a certain generation of this package.

Also for MariaDB Backup to run, it does not need the server, but client,
and from client only core parts.

dict_sys_tables_type_valid(): Do not reject NO_ROLLBACK (sequence)
in ROW_FORMAT=REDUNDANT.

dict_load_table_low(): When flagging an error, assign *table = NULL.
Failure to do so could cause a crash if an error was flagged when
accessing INFORMATION_SCHEMA.INNODB_SYS_TABLES.

MDEV-15742 Assertion `table_share->tmp_table != NO_TMP_TABLE ||
m_lock_type == 1' failed in handler::ha_write_row

Renaming methods:
- Field::make_field(Send_field*) to make_send_field(..)
- Item::make_field(THD *,Send_field *) to make_send_field(..)
- Item::init_make_field(Send_field *, enum_field_type) to init_make_send_field(..)

These names looked similar to other functions that are used
for a very different purpose (creating Field instances):
- Public function "Field * make_field(..)"
- Method "Field *Column_defitinion::make_field(..)"

The rename makes it's easier to search the code using "grep".

MDEV-14762 Server crashes in MDL_ticket::has_stronger_or_equal_type upon inserting into temporary sequence

Fix is to not upgrade MDL locks for temporary tables

Found via `codespell -i 3 -w --skip="./debian/po" -I ../mariadb-server-word-whitelist.txt  ./cmake/ ./debian/ ./Docs/ ./include/ ./man/ ./plugin/ ./strings/`

Problem was the Item_field::Item_field(THD*, Field*) had old code
that put a null pointer in orig_field_names. Now, when we have
proper re-prepare if table definition changes, this is not needed
anymore.

It will only build if there is network access, which is a strict no-go
for Linux distributions. Thus the AWS plugin needs to be a custom built
only special plugin that is not built as part of the normal sources.

This fixes the build error in e.g. Launchpad:

  Performing download step (git clone) for 'aws_sdk_cpp'
  cd "/<<PKGBUILDDIR>>/builddir/plugin/aws_key_management" && /usr/bin/cmake ..
  Cloning into 'aws-sdk-cpp'...
  fatal: unable to access 'https://github.com/awslabs/aws-sdk-cpp.git/':
  Could not resolve host: github.com

RocksDB configure step checks that platform is 64-bit and little endian,
so limit the .deb packages to be built only on those platforms.

List of official Debian platforms at
https://www.debian.org/ports/index.en.html

A non-MariaDB.org contributor introduced the epoch 1: in Debian.
There is no way to revert it, so we just need to adapt to it to
ensure no 10.3 installation would downgrade to that version in
Debian/Ubuntu, as 1: is considered higher than any version before.

The crash occurs when a thread that is closing its connection attempts to
access Spider transaction information when another thread has freed that memory
while processing Spider plugin deinit.  This occurs because Spider does not
adjust the plugin's reference count when it sets a transaction information
pointer for the plugin.

The fix I implemented changes the way Spider sets the transaction information
pointer to use thd_set_ha_data() so that Spider's plugin reference counter is
adjusted as well.

Author:
  Jacob Mathew.

Reviewer:
  Kentoku Shiba.

Merged From:
  bb-10.3-MDEV-7914

Increasing the length of MESSAGE_TEXT from 128 to MYSQL_ERRMSG_SIZE which is the max length of the
error message

Use trx_sys_t::trx_list instead.

Replaced "list of transactions created for MySQL" with "list of all
transactions". This simplifies code and allows further removal of
trx_sys.m_views.

trx_free_resurrected(): Remove, unused function

The test innodb.101_compatibility occasionally fails, because the flags
for the tables tdd and tp are not always converted back. Thus, the
second attempt of corrupting the flags will update garbage to garbage,
and cause test failure.

This started failing related to one or two MDEV-12266 commits affecting
the function fsp_flags_try_adjust().

fsp_flags_try_adjust(): If the file has not been opened (space->size==0),
try determining its size by reading the file. Only if the file is not
readable, give up.

Also, avoid dummy writes in the Perl script, and add some instrumentation.

The crash occurs when a thread that is closing its connection attempts to
access Spider transaction information when another thread has freed that memory
while processing Spider plugin deinit.  This occurs because Spider does not
adjust the plugin's reference count when it sets a transaction information
pointer for the plugin.

The fix I implemented changes the way Spider sets the transaction information
pointer to use thd_set_ha_data() so that Spider's plugin reference counter is
adjusted as well.

Author:
  Jacob Mathew.

Reviewer:
  Kentoku Shiba.

- Adding "return true" into LEX::set_system_variable()
  and LEX::set_default_system_variable() after my_error().
  This makes the parser exit on error immediately.
  Previously, the error was caught only in mysql_parser(),
  a few lines after the parse_sql() call.
- Fixing "--error 1272" to "--error ER_VARIABLE_IS_NOT_STRUCT" in tests

log_write_up_to(): Erase the end of the current log block.
Simplify the computation of pad_size.

log_buffer_switch(): Evaluate a condition only once.

- recovered_lsn shouldn't be initialized during xtrabackup_copy_logfile().
If partial redo log read during the end of xtrabackup_copy_logfile() then
recovered_lsn will be different from scanned_lsn. Re-initialization of
recovered_lsn could lead to partial read again. It is a regression of
MDEV-14545

MDEV-10991: Server crashes in spider_udf_direct_sql_create_conn - tests in spider/oracle* suites crash the server

The fix for this bug was automatically merged from 10.1.  However, part of that
fix is unnecessary in 10.3.  This commit rolls back the part of the fix that is
unnecessary in 10.3.

Author:
  Jacob Mathew.

Reviewer:
  Kentoku Shiba.

           Item::derived_field_transformer_for_having

The crash occurred due to an inappropriate handling of multiple equalities
when pushing conditions into materialized views/derived tables. If equalities
extracted from a multiple equality can be pushed into a materialized
view/derived table they should be plainly conjuncted with other pushed
predicates rather than form a separate AND sub-formula.

- Adding class Field_int as a common
  parent for Field_{longlong|long|short|medium|tiny}
- Moving store_decimal(), val_decimal(), get_date(), store_time_dec(),
  get_date(), val_bool() from Field_num to Field_int
- Adding Field_int::val_str_from_long() and reusing it in
  Field_tiny::val_str(), Field_short::val_str(), Field_medium::val_str()
  and Field_long::val_str(). This removes a good amount of duplicate code
- Adding "const" qualifier to "virtual bool Field::optimize_range()".

Handle overlaping "from" and Field_blob_compressed::value for compressed
blobs similarily to regular blobs.

MDEV-14831 CREATE OR REPLACE SEQUENCE under LOCK TABLE corrupts the
sequence, causes ER_KEY_NOT_FOUND

The problem was that sequence_insert didn't properly handle the case
where there where there was a LOCK TABLE while creating the sequence.

Fixed by opening the sequence table, for inserting the first record, in
a new environment without any other open tables.

Found also a bug in Locked_tables_list::reopen_tables() where the lock
structure for the new tables was allocated in THD::mem_root, which causes
crashes. This could cause problems with other create tables done under
LOCK TABLES.

Use high accuracy timer on Windows 8.1+ for system versioning,it needs
accurate high resoution start query time.

Continue to use the inaccurate (but much faster timer function)
GetSystemTimeAsFileTime() where accuracy does not matter, e.g in
set_timespec_time_nsec(),or my_time()

concurrently.

There is a deadlock between

C1 mariabackup's connection that holds MDL locks
C2 Online ALTER TABLE that wants to have MDL exclusively
   and tries to upgrade its mdl lock.
C3 another mariabackup's connection that does FLUSH TABLES (or FTWRL)

C3 waits waits for C2,  which waits for C1, which waits for C3,
thus the deadlock.


MDL locks cannot be released until FLUSH  succeeds, because
otherwise it would allow ALTER to sneak in, causing backup to abort and
breaking lock-ddl-per-table's promise.

The fix here workarounds the deadlock, by killing connections in
"Waiting for metadata lock" status (i.e ALTER). This killing continues
until FTWRL succeeds.

Killing connections is skipped in case --no-locks parameter
was  passed to backup, because there won't be a FLUSH.

For the reference,in Percona's xtrabackup --lock-ddl-per-connection
silently implies --no-lock ie FLUSH is always skipped there.

A rather large part of fix is introducing DBUG capability to start
a query  the new connection at the right moment of backup
compensating somewhat for mariabackup' lack of send_query or DBUG_SYNC.