- Do not scan registry to check if TCPIP is supported.
- Do not read registry under HKEY_LOCAL_MACHINE\SOFTWARE\MySQL anymore.
- Do not load threadpool function dynamically, it is available
since Win7.
- simplify win32_init_tcp_ip(), and return error of WSAStartup() fails.
- Correct comment in my_parameter_handler()

No need to do dynamic loading and fallbacks anymore.
We can safely assume Windows 7, and availability of all SRWLock functions.

page_zip_compress(), page_zip_compress_write_log(),
page_zip_copy_recs(): Replace the parameters page,page_zip with block,
and set buf_page_t::init_on_flush on success
if innodb_log_optimize_ddl=OFF.

page_zip_parse_compress_no_data(): Merge with the only caller
recv_parse_or_apply_log_rec_body().

Thanks to MDEV-12699, the doublewrite buffer will only be needed in
those cases when a page is being updated in the data file. If the page
had never been written to the data file since it was initialized,
then recovery will be able to reconstruct the page based solely on
the contents of the redo log files.

The doublewrite buffer is only really needed when recovery needs to read
the page in order to apply redo log.

Note: As noted in MDEV-19739, we cannot safely disable the doublewrite
buffer if any MLOG_INDEX_LOAD records were written in the past or will
be written in the future. These records denote that redo logging was
disabled for some pages in a tablespace. Ideally, we would have
the setting innodb_log_optimize_ddl=OFF by default, and would not allow
it to be set while the server is running. If we wanted to make this
safe, assignments with SET GLOBAL innodb_log_optimize_ddl=...
should not only issue a redo log checkpoint (including a write of all
dirty pages from the entire buffer pool), but it should also wait for
all pending ALTER TABLE activity to complete. We elect not to do this.

Avoiding unnecessary use of the doublewrite buffer should improve the
write performance of InnoDB.

buf_page_t::init_on_flush: A new flag to indicate whether it is safe to
skip doublewrite buffering when writing the page.

fsp_init_file_page(): When writing a MLOG_INIT_FILE_PAGE2 record,
set the init_on_flush flag if innodb_log_optimize_ddl=OFF.
This is the only function that writes that log record.

buf_flush_write_block_low(): Skip doublewrite if init_on_flush is set.

fil_aio_wait(): Clear init_on_flush.

recv_apply_hashed_log_recs(): Refer directly to recs.last->end_lsn
instead of iterating the entire list starting from recs.log.

when plugins_are_initialized we already called unlink_thd()
and that calls server_threads.erase()

GCC crashes during compilation, on a code fragment that was added  to
workaround code generation bug on GCC.

To fix, I randomly moved pragmas around, which seems to make GCC happy.

MDEV-19706 RPM no longer installs init script on systemd systems, but preun script still tries to erase it

MDEV-16249 CHECKSUM TABLE for a spider table is not parallel and saves all data in memory in the spider head by default (#1328)

followup for be5c432a

ha_partition::calculate_checksum() has to invoke calculate_checksum()
for partitions unconditionally, not under (HA_HAS_OLD_CHECKSUM | HA_HAS_NEW_CHECKSUM).
Because the server uses ::info() to ask for a live checksum, while
calculate_checksum() must, precisely, calculate it the slow way,
also for tables that don't have the live checksum at all.

Also, fix the compilation on Windows (ha_checksum/ulonglong type mix).

Introduced a print_key_value function to makes sure that the trace prints data in readable format
for readable characters and the rest of the characters are printed as hexadecimal.

InnoDB crash recovery buffers redo log records in a hash table.
The function recv_read_in_area() would pick a random hash bucket
and then try to submit read requests for a few nearby pages.
Let us replace the recv_sys.addr_hash with a std::map, which will
automatically be iterated in sorted order.

recv_sys_t::pages: Replaces recv_sys_t::addr_hash, recv_sys_t::n_addrs.

recv_sys_t::recs: Replaces most of recv_addr_t.

recv_t: Encapsulate a raw singly-linked list of records. This reduces
overhead compared to std::forward_list. Storage and cache overhead,
because the next-element pointer also points to the data payload.
Processing overhead, because recv_sys_t::recs_t::last will point to
the last record, so that recv_sys_t::add() can append straight to the
end of the list.

RECV_PROCESSED, RECV_DISCARDED: Remove. When a page is fully processed,
it will be deleted from recv_sys.pages.

recv_sys_t::trim(): Replaces recv_addr_trim().

recv_sys_t::add(): Use page_id_t for identifying pages.

recv_fold(), recv_hash(), recv_get_fil_addr_struct(): Remove.

recv_read_in_area(): Simplify the iteration.

Older GCC generates SSE instruction on not-128-bit-aligned data in
Bitmap<128>::buffer

Workaround by forcing GCC not to use SSE on Bitmap<N> template.

MDEV-16249 CHECKSUM TABLE for a spider table is not parallel and saves all data in memory in the spider head by default (#1328)

add checksum_null for setting null value of checksum

Some I/O functions and macros that are declared in os0file.h used to
return a Boolean status code (nonzero on success). In MySQL 5.7, they
were changed to return dberr_t instead. Alas, in MariaDB Server 10.2,
some uses of functions were not adjusted to the changed return value.

Until MDEV-19231, the valid values of dberr_t were always nonzero.
This means that some code that was incorrectly checking for a zero
return value from the functions would never detect a failure.

After MDEV-19231, some tests for ALTER ONLINE TABLE would fail with
cmake -DPLUGIN_PERFSCHEMA=NO. It turned out that the wrappers
pfs_os_file_read_no_error_handling_int_fd_func() and
pfs_os_file_write_int_fd_func() were wrongly returning
bool instead of dberr_t. Also the callers of these functions were
wrongly expecting bool (nonzero on success) instead of dberr_t.

This mistake had been made when the addition of these functions was
merged from MySQL 5.6.36 and 5.7.18 into MariaDB Server 10.2.7.

This fix also reverts commit 40becbc3
which attempted to work around the problem.

Make it visible

MDEV-18788 Live upgrade from MySQL 5.6/5.7 to MariaDB 10.4 fails with "Event Scheduler: An error occurred when initializing system tables"

if columns or indexes are modified/renamed/dropped in an ALTER TABLE,
stat tables must be updated accordingly (e.g. all statistics for a column
should be dropped). But if a stat table doesn't exist, it's not a reason
to fail the whole ALTER TABLE operation - such an error should be ignored.

comments, rename a function to reflect its usage better

https://github.com/MariaDB/server/pull/777#issuecomment-496469366

- Field::compatible_field_size(), and its "Relay_log_info *rli" argument
- Field::pack_length_from_metadata

option values now have to be edited in-place,
instead of replacing an option with another one
that has the same name but a new value.

- --disable-plugin-load or --skip-plugin-load will now reset plugin load
  list. Needed for future changes to mtr tests

- multiple usage of plugin-load-add and optimizer-switch are now included
  in generated config files