Intel(R) Threading Building Blocks Doxygen Documentation  version 4.2.3
Synchronization

Classes

class  tbb::null_mutex
 A mutex which does nothing. More...
 
class  tbb::null_rw_mutex
 A rw mutex which does nothing. More...
 
class  tbb::queuing_mutex
 Queuing mutex with local-only spinning. More...
 
class  tbb::queuing_rw_mutex
 Queuing reader-writer mutex with local-only spinning. More...
 
class  tbb::spin_mutex
 A lock that occupies a single byte. More...
 
class  tbb::spin_rw_mutex_v3
 Fast, unfair, spinning reader-writer lock with backoff and writer-preference. More...
 

Typedefs

typedef interface7::internal::padded_mutex< spin_mutex, false > tbb::speculative_spin_mutex
 A cross-platform spin mutex with speculative lock acquisition. More...
 
typedef interface7::internal::padded_mutex< tbb::spin_rw_mutex, true > tbb::interface8::speculative_spin_rw_mutex
 A cross-platform spin reader/writer mutex with speculative lock acquisition. More...
 

Functions

template<typename T >
struct tbb::__TBB_DEPRECATED_IN_VERBOSE_MODE_MSG ("tbb::atomic is deprecated, use std::atomic") atomic
 Primary template for atomic. More...
 

Detailed Description

Typedef Documentation

◆ speculative_spin_mutex

A cross-platform spin mutex with speculative lock acquisition.

On platforms with proper HW support, this lock may speculatively execute its critical sections, using HW mechanisms to detect real data races and ensure atomicity of the critical sections. In particular, it uses Intel(R) Transactional Synchronization Extensions (Intel(R) TSX). Without such HW support, it behaves like a spin_mutex. It should be used for locking short critical sections where the lock is contended but the data it protects are not. If zero-initialized, the mutex is considered unheld.

Definition at line 205 of file spin_mutex.h.

◆ speculative_spin_rw_mutex

A cross-platform spin reader/writer mutex with speculative lock acquisition.

On platforms with proper HW support, this lock may speculatively execute its critical sections, using HW mechanisms to detect real data races and ensure atomicity of the critical sections. In particular, it uses Intel(R) Transactional Synchronization Extensions (Intel(R) TSX). Without such HW support, it behaves like a spin_rw_mutex. It should be used for locking short critical sections where the lock is contended but the data it protects are not.

Definition at line 245 of file spin_rw_mutex.h.

Function Documentation

◆ __TBB_DEPRECATED_IN_VERBOSE_MODE_MSG()

template<typename T >
struct tbb::__TBB_DEPRECATED_IN_VERBOSE_MODE_MSG ( "tbb::atomic is  deprecated,
use std::atomic"   
)

Primary template for atomic.

Specialization for atomic<T*> with arithmetic and operator->.

See the Reference for details.

Definition at line 417 of file atomic.h.

418  : internal::atomic_impl<T> {
419 #if __TBB_ATOMIC_CTORS
420  atomic() = default;
421  constexpr atomic(T arg): internal::atomic_impl<T>(arg) {}
422  constexpr atomic<T>(const atomic<T>& rhs): internal::atomic_impl<T>(rhs) {}
423 #endif
424  T operator=( T rhs ) {
425  // "this" required here in strict ISO C++ because store_with_release is a dependent name
426  return this->store_with_release(rhs);
427  }
428  atomic<T>& operator=( const atomic<T>& rhs ) {this->store_with_release(rhs); return *this;}
429 };

Copyright © 2005-2020 Intel Corporation. All Rights Reserved.

Intel, Pentium, Intel Xeon, Itanium, Intel XScale and VTune are registered trademarks or trademarks of Intel Corporation or its subsidiaries in the United States and other countries.

* Other names and brands may be claimed as the property of others.