Change runner label 'macos-11' to 'macos-11-self-hosted' and 'macos-12'
to 'macos-12-self-hosted' to distinguish between self-hosted and
GitHub-hosted runners.
We want to use only self-hosted macOS runners because of test-run problems
with python3.11 on GitHub-hosted macOS runners.

NO_DOC=ci
NO_TEST=ci
NO_CHANGELOG=ci

(cherry picked from commit f991f7c0)

Closes #8043

NO_DOC=bug fix

(cherry picked from commit 19dcdd34)

The coverage workflow is a part of the default testing, so there is no
sense to run this workflow when the 'full-ci' label is set. Moreover, it
cancelled the run of the default testing and started the run of the full
testing due to workflow `concurrency`.

By default, a workflow only runs when a `pull_request` event's activity
type is `opened`, `synchronize`, or `reopened`. That's why there is no
sense in the following construction:

  pull_request:
    types: [ opened, reopened, synchronize ]

So just removed the line related to event's activity type.

NO_DOC=ci
NO_TEST=ci
NO_CHANGELOG=ci

(cherry picked from commit 182034dc)

Add workflow for integration testing of ddl
and crud modules.

Resolves tarantool/tarantool#6619
Resolves tarantool/tarantool#6620

NO_DOC=ci
NO_TEST=ci
NO_CHANGELOG=ci

(cherry picked from commit 53f41130)

A jiggle in tarantool's initialization code can lead to trace numbers
above 9. We can either:

* accept trace numbers above 9 in the test
* or drop traces from the initialization code in the test and assume
  that the new trace will have number 1.

Igor Munkin suggested to stick with the second approach to avoid
dependency on the initialization code.

NO_DOC=test fixup, no user visible changes
NO_CHANGELOG=see NO_DOC

Co-authored-by: Igor Munkin <imun@tarantool.org>
(cherry picked from commit 92794ddc)

Currently all non-main threads have all the signals blocked, however
according to `man pthread_sigmask':

> If SIGBUS, SIGFPE, SIGILL, or SIGSEGV are generated while they
> are blocked, the result is undefined, unless the signal was
> generated by kill(2), sigqueue(3), or raise(3).

On macOS they are actually blocked, causing the faulting instruction
to loop indefinitely. While on Linux they are not blocked, however the
signal handler registered by sigaction is not executed. Don't block them.

Closes #8023
Closes #8083

NO_DOC=bugfix

(cherry picked from commit 706bfea4)

There are 4 fatal signals that cannot be blocked by sigmask if they
are caused by the CPU exception (rather than kill, sigqueue or raise):
SIGILL, SIGBUS, SIGFPE, SIGSEGV. Currently the crash module handles
only SIGSEGV and SIGFPE. This patch adds handlers for SIGBUS and SIGILL.
SIGBUS is usually raised by macOS on access to the unmapped memory, and
SIGILL is possible, for example, while running AVX version of memcpy on
a CPU without AVX support.

Faulting address siginfo->si_addr is valid for all these signals, so
print it unconditionally.

Part of #8023
Part of #8083

NO_DOC=See next commit
NO_CHANGELOG=See next commit

(cherry picked from commit 6c3ce7a6)

Currently, core constraints are dropped on commit. That is why
it is impossible to drop constraint and drop objects it references to
at the same transaction. Let's drop constraints in two steps - detach
them when DDL occurs, then reattach on rollback or delete on commit.

Closes #7339

NO_DOC=bugfix

(cherry picked from commit f72efbc2)

Use O_CLOEXEC flag instead.

If Tarantool is forked before executing box.cfg{}, e.g. using io.popen(),
the child process could start with stdin linked to /dev/urandom.
This happens because wal_writer_singleton and vy_log_writer are not yet
initialized, i.e. `fd' fields are 0, then atfork child handler wal_atfork()
is called. It checks that xlog is opened (the check succeeded as 0 != -1)
and closes its fd 0, in fact closing stdin (twice). Next, Tarantool opens
the file /dev/urandom during initialization, and it receives the lowest
unused file descriptor, which is 0. Then luaL_loadfile() loads stdin as a
Lua chunk, in effect reading random numbers.

This happens on glibc 2.28 and older, as newer versions do not invoke
atfork handlers during io.popen():
https://sourceware.org/bugzilla/show_bug.cgi?id=17490

Closes #7886

NO_DOC=bugfix
NO_TEST=Tested by test/app-luatest/gh_5747_crash_multiple_args_test.lua

(cherry picked from commit ec1af129)

This patch fixes an issue where collation could change the computed
type of a built-in function argument when choosing a function
implementation.

Closes #7992

NO_DOC=bugfix

(cherry picked from commit 1c80ce2d)

Workflow submodule_update now brings latest commits in stable branches
further to tarantool/sdk, where integration tests can run with other
enterprise products and modules.

In tarantool/tarantool-ee a pull request is still opened, so the
maintainers can merge it at any time.

In tarantool/sdk the test branches are named in the following pattern:
`TarantoolBot/test-ce-master`
`TarantoolBot/test-ce-2.10`
and so on. This pattern can be easily distinguished by workflows,
which run integration tests and alert developers on failures.
It is also consistent with branches with the same purpose of integration
testing, that are created by the sdk_test workflow:
`TarantoolBot/test-ce-{feature_branch_name}`

NO_DOC=CI
NO_TEST=CI
NO_CHANGELOG=CI

Resolves tarantool/sdk#355

(cherry picked from commit 60a2ee11)

JIT has been disabled for these 3 tests on arm64 to avoid weird fails, but
they are still flaky in CI. Perhaps the JIT traces from other tests affect
them. Let's try to do jit.flush() together with jit.off().

Also disable JIT for pagination_netbox_test.lua, which suffers from the
same problem as pagination_test.lua.

Part of #6599
Part of #7739
Part of #8011

NO_DOC=testing stuff
NO_CHANGELOG=testing stuff

(cherry picked from commit 2695f66a)

Some internal modules have been recently copied to luatest repo [1] and
now they can be safely removed, and the corresponding functionality from
luatest can be used instead.

Affected modules:

- test/luatest_helpers/cluster.lua

[1] tarantool/luatest#271

Closes tarantool/luatest#237
Closes tarantool/luatest#269

NO_DOC=testing stuff
NO_TEST=testing stuff
NO_CHANGELOG=testing stuff

(cherry picked from commit 62ffc72c)

Bump test-run to new version with the following improvements:

- Bump luatest to 0.5.7-20-g48c406a [1]

[1] tarantool/test-run#366

Part of tarantool/luatest#237

NO_DOC=testing stuff
NO_TEST=testing stuff
NO_CHANGELOG=testing stuff

(cherry picked from commit 903211df)

Before the patch, fiber->storage.lua.stack is used for
`panic` calls. However, some fibers don't have any Lua
state saved in their storage (for example, space
triggers).

After the patch, the Lua state pointed by `cur_L` is
used to make those calls, as it is always present.

Closes #6647
NO_DOC=bugfix

(cherry picked from commit dfe79b9a)

The 'copy' ptr is checked for NULL after dereferencing. Found by PVS.

https://jira.vk.team/browse/TNT-698

It can hardly result in a null-ptr dereference, because vy_stmt_dup()
uses malloc(), which doesn't normally fail (instead, the OOM killer just
kills the process if there's a severe shortage of memory, see #3534).
Still, PVS rightfully complains about it so let's fix it.

Fixes commit 902d212b ("vinyl: zap vy_write_iterator->format")

NO_DOC=bug fix
NO_TEST=unreproducible
NO_CHANGELOG=unreproducible

(cherry picked from commit de175b97)

ChristopherHX/github-act-runner@0.4.1, which we use on FreeBSD,
is incompatible with newest actions/checkout@v3.2.0, aliased to @v3.
Until this is resolved, FreeBSD workflows will use fixed version
actions/checkout@v3.1.0.

actions/checkout@v3.2.0 fails with an unclear error:

>   Run actions/checkout@v3
>     git clone 'https://github.com/actions/checkout' # ref=v3
> Error: Unable to resolve v3: unsupported object type
> Error: Unable to resolve v3: unsupported object type
> Error:     Failure - actions/checkout@v3
> Error: unsupported object type

NO_DOC=CI
NO_TEST=CI
NO_CHANGELOG=CI

(cherry picked from commit a79d32b1)

Support for OpenSUSE 15.1/15.2 has been recently dropped.
There is no sense to keep the dead code. So removing it.

Follows up #8021

NO_DOC=cd
NO_TEST=cd
NO_CHANGELOG=cd

(cherry picked from commit e5704327)

The `recursive_rmdir` can start execution while tarantool instance
started by the tarantoolctl is in the process of shutdown. As a result
the instance pid file can be available at the name collection, but by
the time of its removal the instance deletes it by itself.

NO_DOC=test
NO_CHANGELOG=test

(cherry picked from commit 8f3fc2ef)

Two more places where replica didn't wait until data is replicated.

Follow-up tarantool/tarantool-qa#274

NO_CHANGELOG=test fix
NO_DOC=test fix

(cherry picked from commit e27ddf70)

* Prepare workspace before actions/checkout to prevent
  git-related workflow failures.
* Replace tarantool/actions/cleanup, because it solves
  just the same problem.

Resolves tarantool/tarantool-qa#285

NO_DOC=ci
NO_TEST=ci
NO_CHANGELOG=ci

(cherry picked from commit f6ee11e9)

It was decided not to proceed with support for these distros.

Closes tarantool/infra#162

NO_DOC=ci/cd
NO_TEST=ci/cd
NO_CHANGELOG=ci/cd

(cherry picked from commit 739e6eb5)

Currently, we stop without checking the authentication request body if
the specified user doesn't exist. As a result, one can check if a user
exists by sending a malformed authentication request: for an existing
user, the server will reply with ER_INVALID_MSGPACK error while if the
user doesn't exist, the error will be set to ER_CREDS_MISMATCH. To fix
this, let's parse the authentication request body even if the user
doesn't exist.

Closes https://github.com/tarantool/security/issues/21

NO_DOC=bug fix

(cherry picked from commit 42badb87)

The triggers are supposed to run on any authentication attempt,
successful or not. Without it, the admin may not notice a malefactor
enumerating user names.

Closes #8017

NO_DOC=bug fix

(cherry picked from commit 8a2f1653)

The guest can't have a non-empty password - it's explicitly prevented by
the _user trigger callback:

https://github.com/tarantool/tarantool/blob/2ed06651c317f2d46631b0d55e4b88f2c8a6c0a0/src/box/alter.cc#L2930-L2936

So there's no need to check the guest user password in authenticate.

NO_DOC=code cleanup
NO_TEST=code cleanup
NO_CHANGELOG=code cleanup

(cherry picked from commit 26f69561)

1. Fixed modification of a variable passed to the logging function
   (both log formats)

   tarantool> a = {foo = 'bar', file = 'c://autorun.bat'}
   tarantool> log.info(a)
   [...]
   tarantool> a
   (before) - foo: bar
    (after) - foo: bar
              file: c://autorun.bat

2. Fixed the drop of fields with reserved internal names (plain log format)

   tarantool> log.info({foo = 'bar', pid = 666})
   (before) {"foo":"bar"}
    (after) {"foo":"bar","pid":666}

3. Now if 'message' field is absent, it is set to the first field of the
   table (JSON log format)

   tarantool> log.info({'Hello, world', key = 'value'})
   (before) {[...], "1":"Hello, world", "key":"value", [...]}
    (after) {[...], "message":"Hello, world", "key":"value", [...]}

4. Fixed assertion on a table without string keys (JSON log format)

   If a Lua table, passed to say(), contains only integer indexes,
   json.encode() will encode it as an array rather than a map, thereafter
   say_format_json() will produce improperly formatted JSON string, or get
   assertion failure. Fix it by encoding all kinds of tables as maps.

Closes #3853
Closes #7955

NO_DOC=Mostly bug fixes, the format of the JSON log is not documented.

(cherry picked from commit 85ef1118)

Turn LuaJIT off for this test on AArch64 in the scope of the issue [1].

Also increase grep_log_bytes. Currently the message is 235 byte long,
which is pretty close to 256. If one more item is added to the traceback,
the test will fail.

```
C> Potentially long select from space 'test_memtx' (512)
 stack traceback:
        builtin/box/schema.lua:2528: in function 'check_select_safety'
        builtin/box/schema.lua:2545: in function 'select'
        eval:1: in main chunk
        [C]: at 0x556e8ffd5c
```

[1] Related to #8011
Closes tarantool/tarantool-qa#264

NO_DOC=test fix
NO_CHANGELOG=test fix

(cherry picked from commit 84306671)

* Fix narrowing of unary minus.
* Don't compile math.modf() anymore.
* OSX/ARM64: Fix external unwinding.
* Fix build with busybox grep.
* BSD: Fix build with BSD grep.
* OSX/ARM64: Disable unwind info.
* ARM64: Reorder interpreter stack frame and fix unwinding.
* OSX/ARM64: Disable external unwinding for now.
* OSX: Fix build by hardcoding external frame unwinding.
* Cleanup and enable external unwinding for more platforms.
* test: remove TAP side effects in getmetrics tests
* test: relax JIT setup in misc.getmetrics test
* test: relax JIT setup in lj-430-maxirconst test
* GC64: enable sysprof support

Closes #6096
Closes #6976
Closes #7919
Relates to #781
Relates to #7762
Part of #7230

NO_DOC=LuaJIT submodule bump
NO_TEST=LuaJIT submodule bump

Update luacheckrc to suppress all warnings from lua loop vars with `_`
prefix. This allows to make loops more verbose.

Closes #8006

NO_CHANGELOG=luacheckrc fix
NO_DOC=luacheckrc fix
NO_TEST=luacheckrc fix

(cherry picked from commit 2ed06651)

Bump test-run to new version with the following improvements:

- Bump luatest to 0.5.7-18-gcb28d93 [1]

[1] tarantool/test-run#364

NO_DOC=testing stuff
NO_TEST=testing stuff
NO_CHANGELOG=testing stuff

(cherry picked from commit be73ba47)

Some internal modules have been recently copied to luatest repo [1,2]
and now they can be safely removed, and the corresponding functionality
from luatest can be used instead.

Affected modules:

- test/luatest_helpers/server.lua

[1] tarantool/luatest#258
[2] tarantool/luatest#266

Closes tarantool/luatest#239

NO_DOC=testing stuff
NO_TEST=testing stuff
NO_CHANGELOG=testing stuff

(cherry picked from commit 21fc0770)

Bump test-run to new version with the following improvements:

- Bump luatest to 0.5.7-17-g1387aa8 [1]

[1] tarantool/test-run#360

Part of tarantool/luatest#239

NO_DOC=testing stuff
NO_TEST=testing stuff
NO_CHANGELOG=testing stuff

(cherry picked from commit 9687656f)

The function play_wal_until_synchro_queue_is_busy() was used in a
few tests copy-pasted since it was considered to be too specific
for a few rare tests. But apparently it is going to be used again
in a new test in a future commit.

The patch makes this function a method of server object to reuse
it properly.

Needed for #7253

NO_DOC=refactoring
NO_CHANGELOG=refactoring

(cherry picked from commit 135fd0ff)

This patch removes an old CMake kludge, which sets CMAKE_SYSTEM_PROCESSOR
to "x86_64" on all 64-bit architectures, even on ARM (on Darwin).

Closes #7495

NO_DOC=bugfix

(cherry picked from commit d76be7ef)

backtrace_collect() is suffering from the same issue as coro_transfer(),
see commit 215630e6 ("coro: update sp before saving registers to a
stack frame").

It stores x19-x30 and d8-d15 registers to the stack, but only after that
it updates the stack pointer. If a SIGALRM signal is delivered during
the execution of backtrace_collect, the signal handler will use the
stack starting from current sp, thus corrupting the saved registers.

Fix this by updating the stack pointer at the beginning of
backtrace_collect. The constraint of the input operand `bt` is changed
from "m" to "r" to force `bt` always be passed to the inline assembly
via the register. Passing on stack will not work after this fix.

Closes #7985

NO_DOC=bugfix
NO_TEST=Hard to create a stable reproducer

(cherry picked from commit 9adc4aa7)

During rollback read trackers are rebound to newer or older story, if
present. The problem with the current logic is that read trackers are
basically rebound in the first available index: all information about key
parts read from other indexes gets lost — rebind read trackers in all
indexes.

Closes #7828

NO_DOC=bugfix

(cherry picked from commit 559b27db)

We assumed that the successor tuple's story could not get garbage collected
on clarify of result tuple in `tree_iterator_start`, since they coincide in
case of regular iterators. But this is not the case for reverse iterators:
the result tuple is of-by-one from the successor, which means the
successor's story can get garbage collected along with the tuple itself
getting deleted, leading to use-after-free of successor: remove garbage
collection from `memtx_tx_tuple_clarify` and call it manually.

The crash in #7756 revealed that the `put` in transaction manager's story
hash table was performed incorrectly: fix it and add an assertion that
nothing was replaced.

Closes #7755
Closes #7756

NO_DOC=bugfix

(cherry picked from commit 651535bc)

Firstly, there is an assertion that checks if replica_id haven't been
changed after it's registration. Let's replace this assertion with warning
message about id being altered - new id will be used for affected replica.

Secondly, if before_replace trigger, set on space _cluster, returns nil,
tarantool will crash with segmentation fault. This patch handles the case.

Closes #7846

NO_DOC=bugfix

(cherry picked from commit a470bb63)

The test fails with:

  not ok 2 ...ty_or_nil_select.test_log_entry_presence_for_memtx_user_space
  gh_6539_log_user_space_empty_or_nil_select_test.lua:122: log must contain
  a critical entry about `box.space.test_memtx:select({0}, {limit = 1001,
  iterator = "ALL"})` call on a memtx user space
  expected: a value evaluating to true, actual: nil

Looks like it happens, because g.server:grep_log() is executed before
g.server:eval(call) completes writing to the log.
In other tests grep_log() is wrapped into t.helpers.retrying() in order
to avoid such issues, so do the same here.

May close tarantool/tarantool-qa#264

NO_DOC=test fix
NO_CHANGELOG=test fix

(cherry picked from commit e4f1c891)

The read iterator skips DELETE statements. If a skipped DELETE statement
was read from the transaction write set, the read iterator breaks the
current cache chain by clearing vy_read_iterator::last_cached, because
the deleted tuple may actually be present at a lower level so adding
a gap there until the transaction is committed would be incorrect, see
commit dd926790 ("vinyl: fix a bug in adding of a cache chain").

The problem is vy_cache_add() assumes that if last_cached is NULL
the tuple added to the cache must be the first tuple that meets the
iteration criteria and sets the cache node boundary level accordingly.
This may lead to the previous tuple being erroneously skipped later
in case the transaction that attempted to delete it is rolled back.

Fix this issue by passing the flag indicating if the current tuple is
the first result of the iteration explicitly to vy_cache_add().

Closes #7947

NO_DOC=bug fix

(cherry picked from commit e00f16e4)