Hi, For a given shared memory object, is construction and use of a mapped_region threadsafe? For example, is the following going to be threadsafe: namespace ipc = boost::interprocess; ipc::file_mapping fmap("file.bin", ipc::read_write); void threadfunc(unsigned idx) { ipc::mapped_region r(fmap, ipc::read_write, idx*1024, 1024); ... } for (unsigned i=0; i<10; ++i) start_thread(boost::bind(threadfunc,i)); --- or, in OpenMP-speak: --- ipc::file_mapping fmap("file.bin", ipc::read_write); int idx; #pragma omp parallel for private(idx) for (idx=0; idx<10; ++idx) { ipc::mapped_region r(fmap, ipc::read_write, idx*1024, 1024); ... } Thanks, -bryan