Any modification of buf_pool.page_hash is supposed to be protected
by buf_pool.mutex and page_hash_latch::write_lock(). The buffer pool
watch mechanism of the InnoDB change buffer was violating that
ever since commit b1ab211d (MDEV-15053).

buf_pool_t::watch_set(): Extend the critical section of buf_pool.mutex.

buf_pool_t::watch_unset(): Define non-inline, because
calls are infrequent and this function became larger.
If we have to detach a sentinel from page_hash,
do it while holding both the mutex and the exclusive hash latch.

buf_pool_t::watch_remove(): Assert that the mutex is being held.

buf_page_init_for_read(): Remove some work-arounds for
previously encountered race conditions related to buf_pool.watch.

buf_load(): Wait for the submitted reads to finish before updating
innodb_buffer_pool_load_status.

buf_read_page_background(): Remove the parameter "bool sync"
and always actually initiate a page read in the background.

buf_load(): Always submit asynchronous reads. This allows
page checksums to be verified in concurrent threads as
soon as the reads are completed.

MDEV-25684 Crash in THD::find_temporary_table while calling spider_direct_sql UDF without temporary table created

The server crashed when SPIDER_DIRECT_SQL UDF was called with
non-existing temporary table.

The bug has been introduced by 91ffdc83. The commit removed
the check, from THD::open_temporary_table(), which ensure that
the target temporary tables exist.

We can fix the bug by adding the check before the call of
THD::open_temporary_table().

This is a side-effect of my_large_malloc() introduction,MDEV-18851

It removed a cast to size_t to variable 'blocks' in
multiplication blocks * keycache->key_cache_block_size , creating ulong value
instead of correct size_t.


Replaced a couple of ulongs with appropriate data type, which is size_t.

Also, fixed casts to ulongs in crash handler messages, so that people would
not be confused by that, too.

Interestingly, aria did not expose the same problem even if it contains
copied and pasted code in ma_pagecache, because Aria had some ulongs removed
when fixing a similar problem in MDEV-9256.

Allow the caller to have current_thd. Also do not store
PSI_CALL_get_thread() in the new THD, it is a thread local storage variable
that can become invalid any time, we do not control the lifetime of the
caller's thread.

Other things:
- Don't allocate an IO_CACHE for scanning tables of type BLOCK
  (It was never used in this case)
- Fixed bug in page cache that cased a hang when trying to read a
  not existing S3 block.

Other things
- Cleaned up error messages for CHECK, REPAIR and OPTIMIZE

These options was needed in some cases, like when using minio that require
the port option, to be able to connect to the S3 storage.
The sympthom was that one could get the error
"Table t1.MAI doesn't exist in s3"
even if the table did exits.

Other things:
- Improved error message for non existing S3 files

Apparently, in bootstrap this could crash when creating new THDs

Change logic to only allocate purge thds at startup, or
in pre-shutdown (for slow shutdown).

THD::copy_db_to(): Always return true if the output parameter
was left uninitialized. This fixes a regression that was caused
by commit 7d0d934c (MDEV-16473).

MariaDB Server 10.3 and later were unaffected by this bug
thanks to commit a7e352b5.

Possibly this bug only affects mysql_list_fields()
in the Embedded Server (libmysqld).

This bug was found by GCC 11.2.0 in CMAKE_BUILD_TYPE=RelWithDebInfo.

init_mutex_v1_t: Stop lying that the mutex parameter is const.
GCC 11.2.0 assumes that it is and could complain about any mysql_mutex_t
being uninitialized even after mysql_mutex_init() as long as
PLUGIN_PERFSCHEMA is enabled.

init_rwlock_v1_t, init_cond_v1_t: Remove untruthful const qualifiers.

Note: init_socket_v1_t is expecting that the socket fd has already
been created before PSI_SOCKET_CALL(init_socket), and therefore that
parameter really is being treated as a pointer to const.

if all options from a combination from the combinations file are already
present in the server's list of options, then don't try to run tests
in other combinations from this file.

old behavior was: if at least one option from a combination is
already present in the list...

replaced CPACK_RPM_PACKAGE_VERSION with SERVER_VERSION.
CPACK_* variables are empty and can't be used until INCLUDE(CPack) is
called.
SERVER_VERSION is the safest option because other variables may be
overwritten from submodules

Thanks to Theodore Brockman on Zulip for noticing
on an OSX ARM64 and testing this patch.

Per https://github.com/google/cpu_features/pull/150/files
CMAKE_SYSTEM_PROCESSOR is arm64 on Apple.

Without this, compulation error:

[ 80%] Building CXX object storage/rocksdb/CMakeFiles/rocksdblib.dir/rocksdb/util/crc32c.cc.o
/mariadb/storage/rocksdb/rocksdb/util/crc32c.cc:500:18: error: use of undeclared identifier 'isSSE42'
  has_fast_crc = isSSE42();
                 ^
/mariadb/storage/rocksdb/rocksdb/util/crc32c.cc:1230:7: error: use of undeclared identifier 'isSSE42'
  if (isSSE42()) {
      ^
/mariadb/storage/rocksdb/rocksdb/util/crc32c.cc:1231:9: error: use of undeclared identifier 'isPCLMULQDQ'
    if (isPCLMULQDQ()) {
        ^

This can be reverted when the RocksDB submodule is updated.

https://github.com/facebook/rocksdb/commit/ee4bd4780b321ddb5f92a0f4eb956f2a2ebd60dc

The problem was that a PREARE followed by a non prepared statement
using DEFAULT NEXT_VALUE() could change table->next_local to point to
a not persitent memory aria. The next EXECUTE would then try to use
the wrong pointer, which could cause a crash.
Fixed by reseting the pointer to it's old value when doing EXECUTE.

Backported patch from MariaDB 10.6
The issue was that the using session_mem_used to break a test does not
guarantee where the test breaks, which gives different results
depending on the environment or how MariaDB is compield.

Buggy sepdebugcrcfix in CentOS 7 (rpm-4.11.3) does not restore
SUID bit after editing the binary. This is fixed in rpm-4.12.

Still let's not set SUID bit when installing auth_pam_tool
and use rpm spec %attr directive instead.

In a rebase of the merge, two preceding commits were accidentally reverted:
commit 112b2396 (MDEV-26308)
commit ac2857a5 (MDEV-25717)

Thanks to Daniele Sciascia for noticing this.

Changes:
- Don't include port in connection parameters with 'localhost' hostname

More info:

The hostname, if not specified or specified as '' or 'localhost', will default
to a MySQL server running on the local machine using the default for the UNIX socket.
To connect to a MySQL server on the local machine via TCP, you must specify the
loopback IP address (127.0.0.1) as the host.

To comply with what the perl module expects, we omit the port for a
host=localhost connection.

Reported issue: https://bugzilla.redhat.com/show_bug.cgi?id=1976224

FLUSH PRIVILEGES hasn't been needed for very many years.

Changes:
- Don't include port in connection parameters with 'localhost' hostname

More info:
The hostname, if not specified or specified as '' or 'localhost', will default
to a MySQL server running on the local machine using the default for the UNIX socket.
To connect to a MySQL server on the local machine via TCP, you must specify the
loopback IP address (127.0.0.1) as the host.

Reported issue: https://bugzilla.redhat.com/show_bug.cgi?id=1976224

Currently @@version_compile_machine is used by mtr
to determine if the compiled executable is 32 or 64
bits.

We extend that logic by ensuring that if the DEFAUT_MACHINE
name doesn't have "64" in its string, "-64bits" is appended
to ensure these test pass.

trx_purge_rseg_get_next_history_log(): Fix a race condition that
was introduced in commit e46f76c9
(MDEV-15912). The buffer pool page contents must not be accessed
while not holding a page latch. The page latch was released by
mtr_t::commit().

This race resulted in an ASAN heap-use-after-poison during a stress test.