This should be a non-functional change (apart from removing the
redundant data field).

In the merge of MDEV-11623 from 10.1 to 10.2, we added a call to
fsp_flags_try_adjust() which causes the data file to be opened
via the buffer pool while fil_ibd_open() is holding another open
handle to the file.

Problem was that on import with buffered input/output
trim is not possible as we write several pages in
one write.

fail in buildbot.

Fixed test results.

The fields st_select_lex::cond_pushed_into_where and
st_select_lex::cond_pushed_into_having should be re-initialized
for the unit specifying a derived table at every re-execution
of the query that uses this derived table, because the result
of condition pushdown may be different for different executions.

        more fixes.

        Typenames made into proper character case.

        json_no_table.test ported.

Before "MDEV-10709 Expressions as parameters to Dynamic SQL" only
user variables were syntactically allowed as EXECUTE parameters.
User variables were OK as both IN and OUT parameters.
When Item_param was bound to an actual parameter (a user variable),
it automatically meant that the bound Item was settable.
The DBUG_ASSERT() in Protocol_text::send_out_parameters() guarded that
the actual parameter is really settable.

After MDEV-10709, any kind of expressions are allowed as EXECUTE IN parameters.
But the patch for MDEV-10709 forgot to check that only descendants of
Settable_routine_parameter should be allowed as OUT parameters.
So an attempt to pass a non-settable parameter as an OUT parameter
made server crash on the above mentioned DBUG_ASSERT.

This patch changes Item_param::get_settable_routine_parameter(),
which previously always returned "this". Now, when Item_param is bound
to some Item, it caches if the bound Item is settable.
Item_param::get_settable_routine_parameter() now returns "this" only
if the bound actual parameter is settable, and returns NULL otherwise.

Move --leaf under -e to avoid duplicate option and retain
-l for --log (to maintain compatibility with MySQL 5.7).

Problem was that implementation merged from 10.1 was incompatible
with InnoDB 5.7.

buf0buf.cc: Add functions to return should we punch hole and
how big.

buf0flu.cc: Add written page to IORequest

fil0fil.cc: Remove unneeded status call and add test is
sparse files and punch hole supported by file system when
tablespace is created. Add call to get file system
block size. Used file node is added to IORequest. Added
functions to check is punch hole supported and setting
punch hole.

ha_innodb.cc: Remove unneeded status variables (trim512-32768)
and trim_op_saved. Deprecate innodb_use_trim and
set it ON by default. Add function to set innodb-use-trim
dynamically.

dberr.h: Add error code DB_IO_NO_PUNCH_HOLE
if punch hole operation fails.

fil0fil.h: Add punch_hole variable to fil_space_t and
block size to fil_node_t.

os0api.h: Header to helper functions on buf0buf.cc and
fil0fil.cc for os0file.h

os0file.h: Remove unneeded m_block_size from IORequest
and add bpage to IORequest to know actual size of
the block and m_fil_node to know tablespace file
system block size and does it support punch hole.

os0file.cc: Add function punch_hole() to IORequest
to do punch_hole operation,
get the file system block size and determine
does file system support sparse files (for punch hole).

page0size.h: remove implicit copy disable and
use this implicit copy to implement copy_from()
function.

buf0dblwr.cc, buf0flu.cc, buf0rea.cc, fil0fil.cc, fil0fil.h,
os0file.h, os0file.cc, log0log.cc, log0recv.cc:
Remove unneeded write_size parameter from fil_io
calls.

srv0mon.h, srv0srv.h, srv0mon.cc: Remove unneeded
trim512-trim32678 status variables. Removed
these from monitor tests.

        ST_AsGeoJSON and ST_GeomFromGeoJSON functions implemented.

MDEV-11108: Assertion `uniq_tuple_length_arg <= table->file->max_key_length()' failed in SJ_TMP_TABLE::create_sj_weedout_tmp_table

Removed the assert from the if clause to the else clause.

MDEV-11134 Assertion `fixed' failed in Item::const_charset_converter(THD*, CHARSET_INFO*, bool, const char*)

Problem: Item_param::basic_const_item() returned true when fixed==false.
This unexpected combination made Item::const_charset_converter() crash
on asserts.

Fix:
- Changing all Item_param::set_xxx() to set "fixed" to true.
  This fixes the problem.
- Additionally, changing all Item_param::set_xxx() to set
  Item_param::item_type, to avoid duplicate code, and for consistency,
  to make the code symmetric between different constant types.
  Before this patch only set_null() set item_type.
- Moving Item_param::state and Item_param::item_type from public to private,
  to make sure easier that these members are in sync with "fixed" and to
  each other.
- Adding a new argument "unsigned_arg" to Item::set_decimal(),
  and reusing it in two places instead of duplicate code.
- Adding a new method Item_param::fix_temporal() and reusing it in two places.
- Adding methods has_no_value(), has_long_data_value(), has_int_value(),
  instead of direct access to Item_param::state.

Fixed the typo, updated result file and suppression in the test
file which did not work anyway

==== Description ====

Flashback can rollback the instances/databases/tables to an old snapshot.
It's implement on Server-Level by full image format binary logs (--binlog-row-image=FULL), so it supports all engines.
Currently, it’s a feature inside mysqlbinlog tool (with --flashback arguments).

Because the flashback binlog events will store in the memory, you should check if there is enough memory in your machine.

==== New Arguments to mysqlbinlog ====

--flashback (-B)
It will let mysqlbinlog to work on FLASHBACK mode.

==== New Arguments to mysqld ====

--flashback

Setup the server to use flashback. This enables binary log in row mode
and will enable extra logging for DDL's needed by flashback feature

==== Example ====

I have a table "t" in database "test", we can compare the output with "--flashback" and without.

#client/mysqlbinlog /data/mysqldata_10.0/binlog/mysql-bin.000001 -vv -d test -T t --start-datetime="2013-03-27 14:54:00" > /tmp/1.sql
#client/mysqlbinlog /data/mysqldata_10.0/binlog/mysql-bin.000001 -vv -d test -T t --start-datetime="2013-03-27 14:54:00" -B > /tmp/2.sql

Then, importing the output flashback file (/tmp/2.log), it can flashback your database/table to the special time (--start-datetime).
And if you know the exact postion, "--start-postion" is also works, mysqlbinlog will output the flashback logs that can flashback to "--start-postion" position.

==== Implement ====

1. As we know, if binlog_format is ROW (binlog-row-image=FULL in 10.1 and later), all columns value are store in the row event, so we can get the data before mis-operation.

2. Just do following things:

  2.1 Change Event Type, INSERT->DELETE, DELETE->INSERT.
  For example:
    INSERT INTO t VALUES (...)  ---> DELETE FROM t WHERE ...
    DELETE FROM t ... ---> INSERT INTO t VALUES (...)

  2.2 For Update_Event, swapping the SET part and WHERE part.
  For example:
    UPDATE t SET cols1 = vals1 WHERE cols2 = vals2
    --->
    UPDATE t SET cols2 = vals2 WHERE cols1 = vals1

  2.3 For Multi-Rows Event, reverse the rows sequence, from the last row to the first row.
  For example:
    DELETE FROM t WHERE id=1; DELETE FROM t WHERE id=2; ...; DELETE FROM t WHERE id=n;
    --->
    DELETE FROM t WHERE id=n; ...; DELETE FROM t WHERE id=2; DELETE FROM t WHERE id=1;

  2.4 Output those events from the last one to the first one which mis-operation happened.
  For example:

Fix 32-bit rdiffs to get rid of failed hunks

Most notably, this includes MDEV-11623, which includes a fix and
an upgrade procedure for the InnoDB file format incompatibility
that is present in MariaDB Server 10.1.0 through 10.1.20.

In other words, this merge should address
MDEV-11202 InnoDB 10.1 -> 10.2 migration does not work

When a query containing a WITH clause is printed by EXPLAIN
EXTENDED command there should not be any data expansion in
the query specifications of the WITH elements of this WITH
clause.

This is the final preparation for the merge of MDEV-11623
from MariaDB Server 10.1 (correcting the assignment of
FSP_SPACE_FLAGS in MariaDB Server 10.1).

We must avoid reading FSP_SPACE_FLAGS directly from page 0,
because the flags that affect the uncompressed page size
cannot be trusted if we are upgrading from MariaDB 10.1.

We will pass fil_space_t* instead of a numeric tablespace ID
in many low-level functions, so that fil_space_t::flags
will be available. This will also reduce the amount o
tablespace ID lookups and the pressure on fil_system->mutex.

fil_space_is_being_truncated(), fil_space_is_redo_skipped(),
PageCallback::set_page_size(), fsp_header_get_page_size(): Remove.

fil_node_open_file(), fil_space_get_crypt_data(): Use space->flags.

fsp_free_extent(), fseg_get_first_extent(), fsp_get_space_header(),
xdes_get_descriptor_with_space_hdr(), xdes_get_descriptor(),
xdes_lst_get_descriptor(), fsp_space_modify_check(),
fsp_init_file_page(), fsp_alloc_free_extent(), fsp_page_create(),
fsp_alloc_free_page(), fsp_free_page(), fsp_alloc_seg_inode_page(),
fsp_alloc_seg_inode(), fsp_free_seg_inode(), fseg_fill_free_list(),
fseg_mark_page_used(), fseg_free_page_low(), fseg_free_extent():
Take fil_space_t* as a parameter, instead of taking a numeric ID.

InnoDB keeps track of buffer-fixed buf_block_t or acquired rw_lock_t
within a mini-transaction. There are some memo_contains assertions
in the code that document when certain blocks or rw_locks must be held.
But, these assertions only check the mini-transaction memo, not the fact
whether the rw_lock_t are actually being held by the caller.

btr_pcur_store_position(): Remove #ifdef, and assert that the block
is always buffer-fixed.

rtr_pcur_getnext_from_path(), rtr_pcur_open_low(),
ibuf_rec_get_page_no_func(), ibuf_rec_get_space_func(),
ibuf_rec_get_info_func(), ibuf_rec_get_op_type_func(),
ibuf_build_entry_from_ibuf_rec_func(), ibuf_rec_get_volume_func(),
ibuf_get_merge_page_nos_func(), ibuf_get_volume_buffered_count_func()
ibuf_get_entry_counter_low_func(), page_set_ssn_id(),
row_vers_old_has_index_entry(), row_vers_build_for_consistent_read(),
row_vers_build_for_semi_consistent_read(),
trx_undo_prev_version_build():
Make use of mtr_memo_contains_page_flagged().

mtr_t::memo_contains(): Take a const memo. Assert rw_lock_own().

FindPage, FlaggedCheck: Assert rw_lock_own_flagged().

MySQL Compression code was removed but missed this test case.

Modified test data to be more easily compressed and
fixed mtr_suppression if compression fails.

Remove the testcase. It turns out, the crash still happens, just not
on all platforms.

innodb_file_format=Barracuda is the default in MariaDB 10.2.
Do not set it, because the option will be removed in MariaDB 10.3.

Also, do not set innodb_file_per_table=1 because it is the default.

Note that MDEV-11828 should fix the test innodb.innodb-64k
already in 10.1.

When MySQL 5.7.9 (and MariaDB Server 10.2) introduced
innodb_default_row_format and made ROW_FORMAT=DYNAMIC the default,
it became possible to create any ROW_FORMAT tables in the InnoDB
system tablespace, except ROW_FORMAT=COMPRESSED.

In MySQL 5.7, it is possible to create ROW_FORMAT=DYNAMIC
tables when TABLESPACE=innodb_system is explicitly specified.
Because MariaDB Server 10.2 does not support the MySQL 5.7
TABLESPACE=innodb_system attribute for tables, we should allow
ROW_FORMAT=DYNAMIC when innodb_file_per_table=0.

Also, remove the test innodb_zip.innodb-create-options, which was
an outdated copy of innodb_zip.create_options.

MySQL 5.7 allows temporary tables to be created in ROW_FORMAT=COMPRESSED.
The usefulness of this is questionable. WL#7899 in MySQL 8.0.0
prevents the creation of such compressed tables, so that all InnoDB
temporary tables will be located inside the predefined
InnoDB temporary tablespace.

Pick up and adjust some tests from MySQL 5.7 and 8.0.

dict_tf_to_fsp_flags(): Remove the parameter is_temp.
fsp_flags_init(): Remove the parameter is_temporary.

row_mysql_drop_temp_tables(): Remove. There cannot be any temporary
tables in InnoDB. (This never removed #sql* tables in the datadir
which were created by DDL.)

dict_table_t::dir_path_of_temp_table: Remove.

create_table_info_t::m_temp_path: Remove.

create_table_info_t::create_options_are_invalid(): Do not allow
ROW_FORMAT=COMPRESSED or KEY_BLOCK_SIZE for temporary tables.

create_table_info_t::innobase_table_flags(): Do not unnecessarily
prevent CREATE TEMPORARY TABLE with SPATIAL INDEX.
(MySQL 5.7 does allow this.)

fil_space_belongs_in_lru(): The only FIL_TYPE_TEMPORARY tablespace
is never subjected to closing least-recently-used files.

MySQL 5.7 introduced partial support for user-created shared tablespaces
(for example, import and export are not supported).

MariaDB Server does not support tablespaces at this point of time.
Let us remove most InnoDB code and data structures that is related
to shared tablespaces.

The MariaDB 10.1 page_compression is incompatible with the Oracle
implementation that was introduced in MySQL 5.7 later.

Remove the Oracle implementation. Also remove the remaining traces of
MYSQL_ENCRYPTION.

This will also remove traces of PUNCH_HOLE until it is implemented
better. The only effective call to os_file_punch_hole() was in
fil_node_create_low() to test if the operation is supported for the file.

In other words, it looks like page_compression is not working in
MariaDB 10.2, because no code equivalent to the 10.1 os_file_trim()
is enabled.