revision-id: 673e253724979fd9fe43a4a22bd7e1b2c3a5269e
Author: Kristian Nielsen

Fix missing memory barrier in wait_for_commit.

The function wait_for_commit::wait_for_prior_commit() has a fast path where it
checks without locks if wakeup_subsequent_commits() has already been called.
This check was missing a memory barrier. The waitee thread does two writes to
variables `waitee' and `wakeup_error', and if the waiting thread sees the
first write it _must_ also see the second or incorrect behavior will occur.
This requires memory barriers between both the writes (release semantics) and
the reads (acquire semantics) of those two variables.

Other accesses to these variables are done under lock or where only one thread
will be accessing them, and can be done without barriers (relaxed semantics).

btr_create(), btr_root_raise_and_insert(): Write a MLOG_MEMSET record
to set FIL_PAGE_PREV,FIL_PAGE_NEXT to FIL_NULL, instead of writing
two MLOG_4BYTES records.

For ROW_FORMAT=COMPRESSED pages, we will not use MLOG_MEMSET
because we want the crash-downgrade to earlier 10.4 releases to succeed.

mlog_parse_nbytes(): Relax the too strict assertion. There is no problem
with MLOG_MEMSET records that affect the uncompressed header of
ROW_FORMAT=COMPRESSED index pages.

Fix incorrect change introduced in the fix for MDEV-20109.

The patch tried to compute a more precise estimate for the record_count
value in SJ-Materialization-Scan strategy (in
Sj_materialization_picker::check_qep).  However the new formula is worse
as it produces extremely optimistic results in common cases where
SJ-Materialization-Scan should be used)

The old formula produces pessimistic results in cases when Sj-Materialization-
Scan is unlikely to be a good choice anyway. So, the old behavior is better.

memo_block_unfix(), memo_latch_release(): Merge to ReleaseLatches.

memo_slot_release(), ReleaseAll: Clean up the formatting.

A search with PAGE_CUR_GE may land on the supremum record on
a leaf page that is not the rightmost leaf page.
This could occur when all keys on the current page are
smaller than the search key, and the smallest key on the
successor page is larger than the search key.

ibuf_delete_recs(): Correct the debug assertion accordingly.

Basically, use more List<T>::iterator. This patch required adding two more
overloads to new iterator for convenience.

It's just added regression tests.

Closes #1407

mtr_t::Impl, mtr_t::Command: Merge to mtr_t.

MTR_MAGIC_N: Remove.

MTR_STATE_COMMITTING: Remove. This state was only being set
internally during mtr_t::commit().

mtr_t::Command::m_locks_released: Remove (set-and-never-read member).

mtr_t::Command::m_start_lsn: Replaced with the return value of
finish_write() and a parameter to release_blocks().

mtr_t::Command::m_end_lsn: Removed as a duplicate of mtr_t::m_commit_lsn.

mtr_t::Command::prepare_write(): Replace a switch () with a
comparison against 0. Only 2 m_log_mode are allowed.

Avoid creating std::vector, and use single instead of double traversal.

Problem:
========
CURRENT_TEST: binlog_encryption.rpl_corruption

mysqltest: In included file "./include/wait_for_slave_io_error.inc":
...
At line 72: Slave stopped with wrong error code
**** Slave stopped with wrong error code: 1743 (expected 1595,1913) ****

Analysis:
========
The test emulates the corruption at the various stages of replication for
example in binlog file, in network and in relay log etc. It verifies that all
corruption cases are handled through appropriate error messages.

The test cases which emulate network failure expect following errors.
--ER_SLAVE_RELAY_LOG_WRITE_FAILURE (1595)
--ER_NETWORK_READ_EVENT_CHECKSUM_FAILURE (1743)

Ideally test should expect error codes as 1595 and 1743.
But the test actually waits on incorrect error code 1595,1913

Fix:
===
Added appropriate error code for 'ER_NETWORK_READ_EVENT_CHECKSUM_FAILURE'.
Replaced 1913 with 1743.

New iterator has the fastest possible implementation: just moves one pointer.
It's faster that List_iterator and List_iterator_fast: both do more on increment.

Overall patch brings:
1) work compile times
2) possibly(!) worse debug build performance
3) definitely better optimized build performance
4) ability to write less code
5) ability to write less bug-prone code

fseg_create(): Initialize FSEG_FRAG_ARRY by a single MLOG_MEMSET record.

flst_zero_addr(), flst_init(): Optimize away redundant writes.

fseg_free_page_low(): Write FIL_NULL by MLOG_MEMSET.

The XDES_CLEAN_BIT is always set for every element of
the page allocation bitmap in the extent descriptor pages.
Do not bother touching it, to avoid redundant writes.

The DICT_HDR_MAX_SPACE_ID was already zero-initialized at page allocation.

page_rec_write_field(): Remove.

dict_create_index_tree_step(): If the SYS_INDEXES.PAGE does not change,
do not update it in the data dictionary. Typically, all index page numbers
would be unchanged before and after IMPORT TABLESPACE, except if some
secondary indexes were created after loading some data.

btr_root_fseg_adjust_on_import(): Remove the redundant mtr_t* parameter.
Redo logging is disabled during the page adjustments that IMPORT TABLESPACE
is performing.

The root page must never have any siblings, so it is unnecessary
to clear those fields.

fsp_alloc_seg_inode_page(): Ever since
commit 3926673c
all newly allocated pages are zero-initialized.
Assert that this is the case for the FSEG_ID fields.

btr_store_big_rec_extern_fields(): Remove the redundant initialization
of the most significant 32 bits of BTR_EXTERN_LEN. InnoDB never supported
BLOBs that are longer than 4GiB. In fact, dtuple_convert_big_rec()
would write emit an error message if a clustered index record tuple would
exceed 1,000,000,000 bytes in length.

The BTR_EXTERN_LEN in the BLOB pointers in clustered index leaf page
records is zero-initialized at least since
commit 41bb3537

Remove the unnecessary retrieval and null-modifications of
the preceding page.

Remove the redundant parameter mtr_t*.

Make use of page_has_prev(), page_has_next() whenever possible.

Remove the unnecessary retrieval and null-modifications of
the preceding page.

MDEV-19376 Repl_semi_sync_master::commit_trx assertion failure: ... || !m_active_tranxs->is_tranx_end_pos(trx_wait_binlog_name, trx_wait_binlog_pos)

The assert indicates that the current transaction got caught uncleaned from
the semisync master's cache when it is signaled to proceed upon its
ack receive.

The reason of missed cleanup turns out to be a flaw in the gtid
connect mode.
A submitted by connecting slave value of its last received event's
binlog file *name* was adopted into
{{Repl_semi_sync_master::m_reply_file_name}} as a part of semisync
initialization.

Notice that the initialization still refines the position part of the
submitted last received event's binlog coordinates.
The master side binlog filename:pos refinement is
specific to the gtid connect mode for purpose of computing the latest
binlog file to resume slave feeding from.
Effectively in the gtid connect mode the computed resumption filename:pos
may appear smaller in which case a new post-connect time committing
transaction may be logged with its filename:pos also less than the
submitted coordinates and that triggers the assert.

Fixed with making the semisync initialization to use the refined filename:pos.
It is guaranteed to be less than any new generated transaction's binlog:pos.

The function mach_read_ulint() is a wrapper for the lower-level
functions mach_read_from_1(), mach_read_from_2(), mach_read_from_8().
Invoke those functions directly, for better readability of the code.

mtr_t::read_ulint(), mtr_read_ulint(): Remove. Yes, we will lose the
ability to assert that the read is covered by the mini-transaction.
We would still check that on writes, and any writes that
wrongly bypass mini-transaction logging would likely be caught by
stress testing with Mariabackup.