Fiber attributes are a way to specify parameters that is different
from the default. When a fiber is created using fiber_new_ex(),
an attribute object can be specified to configure custom values for
some options. Attributes are specified only at fiber creation time;
they cannot be altered while the fiber is being used.

Currently only stack_size attribute is supported. Fibers with
non-default stack size won't be recycled via fiber pool.

API overview:

* fiber_new_ex() creates a fiber with custom attributes.
* fiber_attr_new()/fiber_attr_delete() creates/destroys attributes.
* fiber_attr_setstacksize()/fiber_attr_getstacksize() sets/gets
  the fiber stack size.
* fiber_self() returns current running fiber.

All new functions are available from public C API for modules.

See #2438

Generate random names for temporary FFI structures.

Follow up #1701

During converting upsert into replace the
vy_stmt_new_replace_from_upsert doesn't copy field_map of the
source statement. It led to segfault in comparators during access
to the broken field_map.

Fixes #2461

Now Tarantool doesn't support NaN's on indexed positions. Lets
allow their indexing.

According to the IEEE 754 the double has the following format:

+------+----------+----------+
| sign | exponent | fraction |
+------+----------+----------+
 1 bit   11 bit     52 bit

If the exponent is  0x7FF, the value is special one whether of sign bit.
Special value can be NaN, +inf or -inf.

If the fraction == 0, the value is inf, which sign depends on the first
bit.

If the first bit of the fraction is 1, the value is quiet NaN (qNaN),
and else the signaling NaN (sNaN).

Lets compare their in the order:
sNaN < qNaN < -inf < normal number < +inf;
sNaN == sNaN;
qNaN == qNaN;
aNaN != qNaN;
-inf == -inf;
+inf == +inf;

Closes #1701

The quota may be exceeded even if all in-memory trees are empty, because
quota is used by pending transactions. If this happens, the scheduler
will keep dumping empty trees in a tight loop w/o yielding, thus not
even letting pending transactions to fail on timeout. Fix this by adding
a sanity check to vy_scheduler_peek_dump() assuring there's something we
can dump before proceeding to scheduling a task.

vy_index->space is zeroed on index drop, so we have to be extremely
careful about how to use it, e.g. we can't dereference it in a dump
task, while currently we do, which can result in a crash. Actually,
vy_index->space is only used for retrieving the primary key, so having
a pointer to the whole space in a vy_index is kinda superfluous - a
primary key would be enough. That being said, this patch replaces the
space field with the pk field in the vy_index struct. vy_index->pk is
referenced by each secondary index so it's safe to use it wherever we
want. Also, add a test dropping a space in the middle of dump - w/o
this patch it leads to a crash.

There's no point in preallocating memory on creation of a bitset_index
object, it only encumbers usage of it in functions that cannot fail,
e.g. on memtx index creation.

In some cases a xlog file can have some data after EOF marker.
Set XLOG_CURSOR_EOF flag for xlog cursor only when there are no
more data after EOF marker.

Closes #2460

Check properly the return value of a xlog_cursor_find_tx_magic().

See #2460

Negating unsigned size_t value and then casting it to signed off_t type
is very bad idea. On 32-bit platforms size_t is unsigned 32-bit type,
whereas off_t is signed 64-bit type.

See #2460

Slices make range size estimation inaccurate. This may lead to the
coalesce test hang waiting for all ranges to be coalesced in case the
estimated sum range size turns out to be greater than the threshold
(range_size / 2). To make sure it never happens, let's make the test
more aggressive: let it not only delete 90% keys, but also strip the
rest of padding.

We don't have any intermediate log level between INFO and DEBUG.
INFO level should be used for rare events, such as becoming a
master or accepting a replica join. INFO is the default level.

We lack a log level for "hard to debug" situations, usable
by system administrators.

Introduce a new new log level for this purpose, VERBOSE,
for verbose output.

This change should not affect any existing installations.

Update module-api.c test.

Fixes gh-2467.

Follow up #2254

Since we've switched to Type=notify, forking isn't more needed.
Disable `background = true` in tarantoolctl to avoid potential
problems with fork detection by systemd.

Closes #2437

Fixes #2423

vy_mem_iterator_restore() is incorrect when the following
two conditions are both met:

a) Some key was prepared, but hasn't been committed yet
b) vy_mem_iterator is started and hasn't finished yet

Try to fix it by adding a new condition to mem_restore invariant.

See #2423

* Fix compilation warnings on abs()
* Fix problems on OS X and CentOS 6
* Disable this test with -DENABLE_GCOV=ON

Closes #2429

Remove fiber from cord->dead and cord->alive lists on fiber_destroy().

Refactoring for #2437

* Merge `struct tarantool_coro` into `struct fiber`
* Delete coro.c and coro.h files.

Refactoring for #2438

Original behavior can be enable with `./test-run -j -1`

See https://github.com/tarantool/test-run/issues/56

Follow up previous commits. No semantic changes.

Disable tuple_hash() optimization for non-sequential keys.

See #2084

* Refactor templates in tuple_compare.cc to generate optimized
  versions for various field types
* Add a function pointer in key_def to select optimized
  version at runtime

Fixes #2084

* Implement tuple_extract_key_sequential() for sequential keys
* Add a function pointer to in key_def to select optimized
  version at runtime

Closes #2048

It uses error injection which is compiled out for release builds.
Also, improve the test output by adding s:count() after each space
modification for checking if the statement was actually inserted.

* update comments, messages
* add a test for changing index id

Allow to alter any index opts which don't change key_def of
the index.

Closes #1931
Closes #2109
Closes #2149

Needed for extractiong write iterator to a separate file.
Also move stat account from the function to caller code.

Change-Id: I73af73e6a34f9d7431a3c8056030c9858226b0ad

time_t is long on all major platforms, not int.

Fixes #2443

Change-Id: I7ca029ccfe87124bdc7cdd2ee1db06404f38fe39

Simplify check_param_table() arguments in index.create() and index.alter().

Closes #2148

Change-Id: I10d8d93a12e8bbd91c36f6eefc0f0e8e07636de5

Currently, only keys from vy_log_key_mask[type] may be present in a
record of the given type. Extra keys will result in an error. In
order to rework space truncate, we will have to add a new key for
VY_LOG_CREATE_INDEX record (space_version). To be able to start from
vylog generated by an older version, we need to allow extra keys. So
this patch deletes vy_log_key_mask[] and makes vy_log_record_encode()
encode only those keys whose value is different from the default one
(similarly to request_encode()). Checking for mandatory keys is now up
to vy_recovery_process_record() (currently there's the only that needs
to be checked - VY_LOG_KEY_DEF)

vy_recovery_iterate() doesn't clean vy_log_record before proceeding to
the next log entry. As a result, a run record passed to the callback by
vy_recovery_iterate() contains extra info left from the index this run
is for: index_lsn, index_id and space_id. We use this in gc and backup
callbacks to format file name. The problem is vy_recovery_iterate() is
also used internally for log rotation. Including extra keys in records
doesn't result in writing them to the log file on rotation, because per
each record type we have a mask of keys corresponding to the record
(vy_log_key_mask). In order to allow optional keys in vylog, the
following patch will change the meaning of the mask so that it only
contains mandatory keys, while a key will be written to the log only if
its value differs from the default (similarly to request_encode). Thus,
to avoid writing keys not relevant to a record type we need to clean
vy_log_record within vy_recovery_iterate() before jumping to the next
record. So this patch weans gc and backup from exploiting this feature -
let them save index_id and space_id in the context, as we do on
replication and recovery.

Info_append_double is needed to print vinyl index.info.bloom_fpr later.

It was helpful when a vinyl index could have a custom path. Currently,
it's forbidden, so we can format index path in place.

We can format it in place when needed.