buf_LRU_get_free_block(): Increment the counter after acquiring
buf_pool.mutex.

buf_pool.stat.LRU_waits: Replaces export_vars.innodb_buffer_pool_wait_free
and srv_stats.buf_pool_wait_free.

Reads of the counter will remain dirty, not acquiring the mutex.

In commit 3a9a3be1 (MDEV-23855)
some previous logic was replaced with the condition
dirty_pct < srv_max_dirty_pages_pct_lwm, which caused
the default value of the parameter innodb_max_dirty_pages_pct_lwm=0
to lose its special meaning: 'refer to innodb_max_dirty_pages_pct instead'.

This implicit special meaning was visible in the function
af_get_pct_for_dirty(), which was removed in
commit f0c295e2 (MDEV-24369).

page_cleaner_flush_pages_recommendation(): Restore the special
meaning that was removed in MDEV-24369.

buf_flush_page_cleaner(): If srv_max_dirty_pages_pct_lwm==0.0,
refer to srv_max_buf_pool_modified_pct. This fixes the observed
performance regression due to excessive page flushing.

buf_pool_t::page_cleaner_wakeup(): Revise the wakeup condition.

innodb_init(): Do initialize srv_max_io_capacity in Mariabackup.
It was previously constantly 0, which caused mariadb-backup --prepare
to hang in buf_flush_sync(), making no progress.

Related: MDEV-24124

Partial upstreaming of Debian packaging commit
https://salsa.debian.org/mariadb-team/mariadb-10.5/-/commit/93a02bbe848ef9810d44635a3b4d409e71bd0628

Adopt the same install-build-deps.sh the upstream Salsa-CI has started
using. Since we have our own custom build step (to support autobake-deb.sh)
we need to maintain in like this and adopt to upstream changes.

This change will make the stretch-backports build pass and use the
backported libzstd-dev 1.3.8 library (to satisfy the > 1.3.3 requirement).

Also clean away excess autopkgtest stanza and allow missing-breaks to
fail so that in total Salsa-CI would now pass.

- Add libarchive-dev and libkrb5-dev as they are defined as dependencies
  in debian/control
- Update curl dependency to libcurl4-openssl-dev, available in Bionic and Focal
- Remove libstemmer-dev that was added without documented motivations in
  https://github.com/ottok/mariadb/commit/e9ada862651fdb98f84871e4ec59bef1bb36646d
- Remove libdistro-info-perl that was added due to need by autobake-deb.sh
  in https://github.com/ottok/mariadb/commit/eb04ee5c9d2416aa45acd1353dff19f569e96317
  but which is not needed anymore

Keep the user and password definitions as e.g. dbconfig-common expects to
find them there. Extend the file to document (in context) why it is about
to be obsoleted to facilitate users migrating away from it.

Upstreamed from https://salsa.debian.org/mariadb-team/mariadb-10.5/-/commit/a6583c15226ad9a0ca37861e1bd01057bcff32a5

- Add cracklib-runtime and libarchive-dev as build dependencies
- Update Debian policy standards version to 4.5.0
- Add libssl-dev to libmariadb-dev run-time dependency
- Add "Multi-Arch: same" to packages that have it in Debian
- Sync README.Debian
- Sync debian/rules formatting
- Sync autopkgtests

Upstreamed from https://salsa.debian.org/mariadb-team/mariadb-10.5/-/commit/13e3a7903e5c72e70c02b98a3e51a6de70b9b6ef

Also clean away impossible/unnecessary "Multi-Arch: same" stanzas.
Mostly upstreamed from https://salsa.debian.org/mariadb-team/mariadb-10.5/-/commit/f0ba31e1565feaca9c2a6f2534e55bfc29af7021

Omitted 'libzstd-dev (>= 1.3.3)' as the version requirement would need
stretch-backports to be available on buildbot.askmonty.org builders and
they are not yet.

The mariadb.service systemd file indeed does contain:
  ExecStartPost=/etc/mysql/debian-start

So don't claim it does not use debian-start.

Upstreamed from https://salsa.debian.org/mariadb-team/mariadb-10.5/-/commit/af869e703014b25795b0891e3450520c599d3650

- no space after comment
- align equal signs in groups

Upstreamed from: https://salsa.debian.org/mariadb-team/mariadb-10.5/-/commit/a8c67e15c9a58199f085ae84cdb97c542667767d

Let us use Intel intrinsic functions in WolfSSL whenever possible.
This allows such code to be compiled WITH_MSAN.

Move 'encryption' tests to another job, since the 'binlog' and 'rpl'
tests are so slow and often make the job timeout (after 50 minutes).

Allow failure in ppc64el as it frequently fails due to out-of-memory.
A simple restart often fixes it, but we can't depend on restarts.

Also re-enable arm64 as MDEV-23955 is now fixed.

MERGING: This commit is OK to merge 10.6 and upwards.

dbug_print_rec() functions used to print data inside GDB.

If PLUGIN_COLUMNSTORE is not defined, ColumnStore will build automatically
by default on supported architectures as defined in its CMakeFile.txt.

Thus there should not be any need to inject this build flag at any point
and it can be removed to keep thing lean and clean.

ColumnStore seems to build by default, so it must be explicitly disabled
with a build flag, so that it does not build at all and thus build machine
disk space and CPU will be spared.

This reverts commit 113f1868.

Remove the unused dbug_print_rec() functions because they break
the clang build due to -Wreturn-stack-address

Fix a trivial error in the fix for MDEV-21958: check the key in the right
table.

Add missed PSI_CALL_drop_table_share().

Galera replication does not support XA transactions yet. Reject any
attempt to `XA START` a transaction, if Galera is enabled.
Reviewed-by: Jan Lindström <jan.lindstrom@mariadb.com>

Add history row outside of compare_record() check. For TRX_ID
versioning we have to fail can_compare_record to force InnoDB update
which adds history row; and there in ha_innobase::update_row() is
additional "row changed" check where we force history row anyway.

MDEV-23644 Assertion on evaluating foreign referential action for self-reference in system versioned table

First part of the fix (row0mysql.cc) addresses external columns when adding history
row on referential action. The full data must be retrieved before the
row is inserted.

Second part of the fix (the rest) avoids duplicate primary key error between
the history row generated on referential action and the history row
generated by SQL command. Both command and referential action can
happen on same table since foreign key can be self-reference (parent
and child tables are same). Moreover, the self-reference can refer
multiple rows when the key is non-unique. In such case history is
generated by referential action occured on first row but processed all
rows by a matched key. The second round is when the next row is
processed by a command but history already exists. In such case we
check TRX_ID of existing history row and if it is the same we assume
the above situation and skip adding one more history row or failing
the command.

First part (row0mysql.cc) fixes ins_node_set_new_row() usage workflow
as it is designed to operate on empty row (see row_get_prebuilt_insert_row()
for example).

Second part (row0ins.cc) fixes duplicate key error in FTS_DOC_ID_INDEX
since history rows must not generate entries in that index. We detect
FTS_DOC_ID_INDEX by a number of attributes and skip it if the row is
historical.

Misc fixes:

row_build_index_entry_low() does not accept non-NULL tuple
for FTS index (subject assertion fails), assertion (index->type !=
DICT_FTS) adds code understanding.

Now as historical_row is copied in row_update_vers_insert() there is
no need to copy the row twice: ROW_COPY_POINTERS is used to build
historical_row initially.

dbug_print_rec() debug functions.

MDEV-22178 Assertion `info->alias.str' failed in partition_info::check_partition_info instead of ER_VERS_WRONG_PARTS

Assign create_info->alias for ALTER TABLE since it is NULL and later
accessed for printing error message.

Refactor previous commit to fix mistake revealed by Buildbot. We can't
have a structure where PLUGIN_COLUMNSTORE would ever be 'YES' on an arch
that does not support it, as the flag overrides any potential platform
detection code and builds on non-amd64 would all fail.

- Use new Salsa-CI templates just like downstream does
- Also apply adaptations needed for native MariaDB to
  build correctly on Salsa-CI
- Adapt for mysql-8.0 now in Debian Sid, keep also mysql-5.7 test
- Test TLSv1.3 to ensure it does not regress
- Remove Snappy compression from autopkgtest as it not longer works

ColumnStore files and debian/control stanza was removed in 1edd2243, and
thus will not be included in a native build. Also adapt the debian/rules
to follow this same policy and only build ColumnStore in builds triggered
from autobake-deb.sh. Avoiding building ColumnStore in vain saves a lot of
build time and disk space.

Maintaining the same deterministic file order in both upstream (this) and
downstream (Debian) repositories makes it easier to compare the delta.