不同进程的线程之间的通信是否有保护机制？

Question

Given that threads (of the same process) run in a shared memory space and synchronizing the access to the shared data within the process gains unprecedented importance, how does this apply to threads from different processes? What are the communication methods available for them?

Answer 1

Threads from different processes are generally treated just like independent processes themselves, because they share no state with each other and operate in separate memory spaces.

As a consequence, they use the standard interprocess communication mechanisms, such as Unix sockets, pipes, and other file descriptors; SysV or POSIX IPC mechanisms such as shared memory and semaphores; locking; and signals. It may be convenient to have a thread in a process that is dedicated to interacting with other processes, but IPC can be done with single-threaded processes as well or in multithreaded processes without dedicated IPC threads.

Since the risk of accidentally mutating other threads' data is not possible across processes, sometimes synchronization is handled automatically by the kernel, such as with pipes, but it is possible to need to coordinate across processes. File locks, such as with fcntl(2) and flock(2) , can prevent co-operating processes from corrupting files; semaphores can prevent contention or limit concurrency for shared resources; and it's even possible to share a mutex for some shared memory cases. The tools you use depend on the resources you're sharing across processes.