Is atomicity guaranteed on data residing in shared memory?

 

E.g. Processes allocate shared memory as follows using either create_only and truncate or open_only:

 

#define SHARED_MEM_SIZE sizeof(boost::uint32_t)

#define SHARED_MEM_NAME "SHMTest"

 

boost::shared_ptr<shared_memory_object> shm_obj;

boost::shared_ptr<mapped_region> mapping;

std::string sharedMemoryName(SHARED_MEM_NAME);

 

shm_obj.reset(new shared_memory_object(create_only, sharedMemoryName.c_str(), read_write));

 

shm_obj->truncate(SHARED_MEM_SIZE);

 

mapping.reset(new mapped_region(*shm_obj, read_write, 0, SHARED_MEM_SIZE));

 

Then, they use the boost atomic functions to read and write to the memory:

 

volatile boost::uint32_t* temp = static_cast<volatile boost::uint32_t*>(mapping->get_address());

 

unsigned int value = boost::interprocess::detail::atomic_read32(temp);

 

boost::interprocess::detail::atomic_write32(temp, value);

 

Are these memory accesses guaranteed to be atomic? Platform is Linux if that makes a difference.

 

Thanks