#include <hard_sync_mutex.h>

Inheritance diagram for fpmas::synchro::hard::HardSyncMutex< T >:

Collaboration diagram for fpmas::synchro::hard::HardSyncMutex< T >:

Public Member Functions
	HardSyncMutex (fpmas::api::graph::DistributedNode< T > *node, MutexClient &mutex_client, MutexServer &mutex_server)

void	pushRequest (Request request) override

std::queue< Request >	requestsToProcess () override

T &	data () override

const T &	data () const override

const T &	read () override

void	releaseRead () override

T &	acquire () override

void	releaseAcquire () override

void	lock () override

void	unlock () override

bool	locked () const override

void	lockShared () override

void	unlockShared () override

int	sharedLockCount () const override

void	synchronize () override

virtual void	pushRequest (MutexRequest request)=0

virtual std::queue< MutexRequest >	requestsToProcess ()=0

virtual T &	data ()=0

virtual const T &	data () const =0

virtual const T &	read ()=0

virtual void	releaseRead ()=0

virtual T &	acquire ()=0

virtual void	releaseAcquire ()=0

virtual void	lock ()=0

virtual void	unlock ()=0

virtual bool	locked () const =0

virtual void	lockShared ()=0

virtual void	unlockShared ()=0

virtual int	sharedLockCount () const =0

virtual void	synchronize ()=0

Additional Inherited Members
virtual void	_lock ()=0

virtual void	_lockShared ()=0

virtual void	_unlock ()=0

virtual void	_unlockShared ()=0

Detailed Description

template<typename T>
class fpmas::synchro::hard::HardSyncMutex< T >

HardSyncMode Mutex implementation.

Each operation is added to a queue and potentially transmitted to DISTANT processes to ensure a strict global concurrent access management.

Constructor & Destructor Documentation

◆ HardSyncMutex()

template<typename T >

fpmas::synchro::hard::HardSyncMutex< T >::HardSyncMutex	(	fpmas::api::graph::DistributedNode< T > *	node,
		MutexClient &	mutex_client,
		MutexServer &	mutex_server
	)

inline

HardSyncMutex constructor.

Parameters

node	node associated to this mutex
mutex_client	client used to transmit requests to other processes
mutex_server	server used to handle incoming mutex requests

Member Function Documentation

◆ pushRequest()

template<typename T >

void fpmas::synchro::hard::HardSyncMutex< T >::pushRequest ( Request request )

overridevirtual

Pushes a pending request into the waiting queue.

Parameters

request mutex request to push

Implements fpmas::synchro::hard::api::HardSyncMutex< T >.

◆ requestsToProcess()

template<typename T >

std::queue< typename HardSyncMutex< T >::Request > fpmas::synchro::hard::HardSyncMutex< T >::requestsToProcess ( )

overridevirtual

Returns an ordered list of requests that can be handled, depending on the current mutex status.

For example, if the mutex is locked, an empty queue might be returned. Else, read or lock shared requests can be returned, until the next pending acquire or lock request.

In any case, the exact behavior is implementation defined and might depend about what priority is given to each request. (e.g. : https://en.wikipedia.org/wiki/Readers%E2%80%93writers_problem)

Returns: an ordered queue of request that can safely be handled at the time of the call, depending on the current mutex status

Implements fpmas::synchro::hard::api::HardSyncMutex< T >.

◆ data() [1/2]

template<typename T >

T & fpmas::synchro::hard::HardSyncMutex< T >::data ( )

inlineoverridevirtual

Direct access to internal data representation.

Does not guarantee any concurrent access or data update management.

Returns: direct reference to internal data

Implements fpmas::api::synchro::Mutex< T >.

◆ data() [2/2]

template<typename T >

const T & fpmas::synchro::hard::HardSyncMutex< T >::data ( ) const

inlineoverridevirtual

Direct access to internal data representation.

Does not guarantee any concurrent access or data update management.

Returns: direct reference to internal data

Implements fpmas::api::synchro::Mutex< T >.

◆ read()

template<typename T >

const T & fpmas::synchro::hard::HardSyncMutex< T >::read

overridevirtual

Provides a read access to the internal data.

The functions blocks until data is effectively read.

Read data must be released using releaseRead().

Returns: reference to read data

Implementation details: If the protected node is LOCAL and locked, adds the local request to the queue and waits for the request to complete. Else if its unlocked, directly returns the read data (which is already local).

: If the node is DISTANT, transmits the request using the MutexClient.

See also: MutexServer::wait(); MutexClient::read()

Implements fpmas::api::synchro::Mutex< T >.

◆ releaseRead()

template<typename T >

void fpmas::synchro::hard::HardSyncMutex< T >::releaseRead

overridevirtual

Releases a mutex previously accessed with read().

Implementation details: If the protected node is LOCAL, releases a shared lock from the mutex. If this unlocks the resource, the MutexServer is notified and pending requests are handled.

: If the node is DISTANT, transmits the request using the MutexClient.

See also: MutexServer::notify(); MutexClient::releaseRead()

Implements fpmas::api::synchro::Mutex< T >.

◆ acquire()

template<typename T >

T & fpmas::synchro::hard::HardSyncMutex< T >::acquire

overridevirtual

Releases a mutex previously accessed with acquire().

Semantically, this operation also "commits" eventual write operations that were performed on the acquired data. However, the actual behavior is implementation defined.

Implementation details: If the protected node is LOCAL and locked, adds the local request to the queue and waits for the request to complete. Else if its unlocked, directly returns the acquired data (which is already local).

: If the node is DISTANT, transmits the request using the MutexClient.

See also: MutexServer::wait(); MutexClient::acquire()

Implements fpmas::api::synchro::Mutex< T >.

◆ releaseAcquire()

template<typename T >

void fpmas::synchro::hard::HardSyncMutex< T >::releaseAcquire

overridevirtual

Releases a mutex previously accessed with acquire().

Semantically, this operation also "commits" eventual write operations that were performed on the acquired data. However, the actual behavior is implementation defined.

Implementation details: If the protected node is LOCAL, releases the exclusive lock from the mutex and notifies the MutexServer so that pending requests are handled.

: If the node is DISTANT, transmits the request with local data updates using the MutexClient.

See also: MutexServer::notify(); MutexClient::releaseAcquire()

Implements fpmas::api::synchro::Mutex< T >.

◆ lock()

template<typename T >

void fpmas::synchro::hard::HardSyncMutex< T >::lock

overridevirtual

Locks the mutex.

This function blocks until internal data is effectively locked. Upon return, an exclusive access to the internal data is guaranteed. However, contrary to the acquire() function, this operation does not involve any data update or write operation handling.

A locked mutex must be unlocked using unlock().

Implementation details: If the protected node is LOCAL and locked, adds the local request to the queue and waits for the request to complete. Else if its unlocked, directly locks the mutex.

: If the node is DISTANT, transmits the request using the MutexClient.

See also: MutexServer::wait(); MutexClient::lock()

Implements fpmas::api::synchro::Mutex< T >.

◆ unlock()

template<typename T >

void fpmas::synchro::hard::HardSyncMutex< T >::unlock

overridevirtual

Unlocks a mutex previously locked with lock().

Implementation details: If the protected node is LOCAL, releases the exclusive lock from the mutex and notifies the MutexServer so that pending requests are handled.

: If the node is DISTANT, transmits the request using the MutexClient.

See also: MutexServer::notify(); MutexClient::unlock()

Implements fpmas::api::synchro::Mutex< T >.

◆ locked()

template<typename T >

bool fpmas::synchro::hard::HardSyncMutex< T >::locked ( ) const

inlineoverridevirtual

Returns true if and only if the mutex has been locked and not yet unlocked.

Returns: true iff mutex is locked

Implements fpmas::api::synchro::Mutex< T >.

◆ lockShared()

template<typename T >

void fpmas::synchro::hard::HardSyncMutex< T >::lockShared

overridevirtual

Obtains a shared lock access to the mutex.

This function blocks until a shared lock access to mutex is effectively obtained. Contrary to lock(), multiple threads are allowed to simultaneously obtain a shared lock to the mutex.

The mutex must then be unlocked using unlockShared().

Implementation details: If the protected node is LOCAL and locked, adds the local request to the queue and waits for the request to complete. Else if its unlocked (or shared lock), adds a share lock to the mutex.

: If the node is DISTANT, transmits the request using the MutexClient.

See also: MutexServer::wait(); MutexClient::lockShared()

Implements fpmas::api::synchro::Mutex< T >.

◆ unlockShared()

template<typename T >

void fpmas::synchro::hard::HardSyncMutex< T >::unlockShared

overridevirtual

Unlocks a mutex previously locked with lockShared().

Semantically, all threads that called lockShared() must call unlockShared() to finally unlock the mutex.

Implementation details: If the protected node is LOCAL, releases a shared lock from the mutex. If this unlocks the resource, the MutexServer is notified and pending requests are handled.

: If the node is DISTANT, transmits the request using the MutexClient.

See also: MutexServer::notify(); MutexClient::unlockShared()

Implements fpmas::api::synchro::Mutex< T >.

◆ sharedLockCount()

template<typename T >

int fpmas::synchro::hard::HardSyncMutex< T >::sharedLockCount ( ) const

inlineoverridevirtual

Returns the count of threads that currently own a shared lock access to the mutex.

Returns: shared lock count

Implements fpmas::api::synchro::Mutex< T >.

◆ synchronize()

template<typename T >

void fpmas::synchro::hard::HardSyncMutex< T >::synchronize ( )

inlineoverridevirtual

Method eventually called when data synchronization is performed on this Mutex.

The behavior of this method depends on the currently implemented SynchronizationMode.

Implements fpmas::api::synchro::Mutex< T >.

The documentation for this class was generated from the following file:

src/fpmas/synchro/hard/hard_sync_mutex.h

Public Member Functions

Additional Inherited Members

Detailed Description

Constructor & Destructor Documentation

◆ HardSyncMutex()

Member Function Documentation

◆ pushRequest()

◆ requestsToProcess()

◆ data() [1/2]

◆ data() [2/2]

◆ read()

◆ releaseRead()

◆ acquire()

◆ releaseAcquire()

◆ lock()

◆ unlock()

◆ locked()

◆ lockShared()

◆ unlockShared()

◆ sharedLockCount()

◆ synchronize()