Locking
[Transactions, Locking and Logging]

Collaboration diagram for Locking:

Functions
static concurrency_t	ss_m::xct_lock_level ()
	Get the locking granularity for the attached transaction.
static void	ss_m::set_xct_lock_level (concurrency_t l)
	Set the default locking level for the attached transaction.
static rc_t	ss_m::lock (const lockid_t &n, lock_mode_t m, lock_duration_t d=t_long, timeout_in_ms timeout=WAIT_SPECIFIED_BY_XCT)
	Acquire a lock.
static rc_t	ss_m::unlock (const lockid_t &n)
	Release a lock.
static rc_t	ss_m::dont_escalate (const lockid_t &n, bool passOnToDescendants=true)
	Disable lock escalation on the given entity.
static rc_t	ss_m::get_escalation_thresholds (w_base_t::int4_t &toPage, w_base_t::int4_t &toStore, w_base_t::int4_t &toVolume)
	Find the storage-manager-wide escalation thresholds Default values (used for all transactions until they change their per-transaction thresholds) are determined by the storage-manager-wide options. See List of Run-Time Options.
static rc_t	ss_m::set_escalation_thresholds (w_base_t::int4_t toPage, w_base_t::int4_t toStore, w_base_t::int4_t toVolume)
	Change the storage-manager-wide escalation thresholds Default values (used for all transactions until they change their per-transaction thresholds) are determined by the storage-manager-wide options. See List of Run-Time Options.
static rc_t	ss_m::query_lock (const lockid_t &n, lock_mode_t &m, bool implicit=false)
	Find out if the attached transaction has an entity locked.
static rc_t	ss_m::set_lock_cache_enable (bool enable)
	Control lock caching for attached transaction.
static rc_t	ss_m::lock_cache_enabled (bool &enabled)
	True if lock cache is enabled for the attached transaction.

---

Function Documentation

static void ss_m::set_xct_lock_level

(

concurrency_t

l

)

[static, inherited]

Set the default locking level for the attached transaction.

Parameters:

[in]

l

The level to use for the balance of this transaction. Legitimate values are t_cc_record, t_cc_page, t_cc_file.

Note:

Only one thread may be attached to the transaction when this is called. If more than one thread is attached, a fatal error will ensue.

static rc_t ss_m::lock	(	const lockid_t &	n,
		lock_mode_t	m,
		lock_duration_t	d = t_long,
		timeout_in_ms	timeout = WAIT_SPECIFIED_BY_XCT
	)			[static, inherited]

Acquire a lock.

Parameters:

[in]	n	Lock id of the entity to lock. There are conversions from record ids, volume ids, store ids, and page ids to lockid_t.
[in]	m	Desired lock mode. Values: EX, SH.
[in]	d	Desired duration. Values: t_very_long : Held across transaction boundaries; cannot be released by unlock() t_long : Released at commit; cannot be released by unlock() t_medium : May be released early by explicit unlock() t_short : May be released early by explicit unlock() t_instant : Not held: acquired and released immediately. Useful to see if any other transaction holds an incompatible lock.
[in]	timeout	Milliseconds willing to block. See timeout_in_ms.

The lock manager is written with these durations in mind, but the only durations used by the storage manager are t_instant and t_long. Medium-duration locks are used internally in a one place.

Durations other than long and instant are not well-tested.

static rc_t ss_m::unlock

(

const lockid_t &

n

)

[static, inherited]

Release a lock.

Parameters:

[in]

n

Lock id of the entity to lock. There are conversions from record ids, volume ids, store ids, and page ids to lockid_t.

static rc_t ss_m::dont_escalate	(	const lockid_t &	n,
		bool	passOnToDescendants = true
	)			[static, inherited]

Disable lock escalation on the given entity.

Parameters:

[in]	n	Lock id of the entity to lock. There are conversions from record ids, volume ids, store ids, and page ids to lockid_t.
[in]	passOnToDescendants	If true, apply this to the descendants of n.

static rc_t ss_m::query_lock	(	const lockid_t &	n,
		lock_mode_t &	m,
		bool	implicit = false
	)			[static, inherited]

Find out if the attached transaction has an entity locked.

Parameters:

[in]	n	Lock id of the entity to lock. There are conversions from record ids, volume ids, store ids, and page ids to lockid_t.
[out]	m	Mode of lock held. NL if none.
[in]	implicit	If "true" the query will returns a lock mode if an implicit lock is held, otherwise the lock must be held explicitly.

static rc_t ss_m::set_lock_cache_enable

(

bool

enable

)

[static, inherited]

Control lock caching for attached transaction.

Parameters:

[in]

enable

Set to true if you want to turn on lock caching for the attached transaction. The default is that it is turned on.

Only long-duration locks are cached. Lock caching can be turned off by default using the sm_lock_caching option. Even with it turned off by default, it can be turned on for a given transcation with this method.

static rc_t ss_m::lock_cache_enabled

(

bool &

enabled

)

[static, inherited]

True if lock cache is enabled for the attached transaction.

Parameters:

[out]

enabled

Will be set to true if the attached transaction has lock caching enabled, false otherwise.

---