CODEWITHSUNDEEP
cpthreads
user422005
user422005

Reputation: 2031

Can I let pthread_t instance go out of scope?

I am programming with C and pthreads. I have a long running function which I want to run in a seperate thread: 

void long_running_function(void * arg) {  
  ...
}


void start_long_running_function(void * arg) {  
  pthread_t thread;  
  pthread_create( &thread , NULL , long_running_function , arg);
  /* What about the thread variable? */  
}

When leaving the start_long_running_function() function the local variable 'thread' will go out of scope. Is this OK - or can I risk problems e.g. when the long_running_function() is complete?

I have tried the approach illustrated in my code, and it seems to work - but maybe that is only luck?

Regards Joakim

Upvotes: 4

Views: 965

Answers (2)

spraff
spraff

Reputation: 33385

It's a C structure, Plain Old Data, so there's no destructor to introduce side-effects when it goes out of scope. The only implication of losing scope is that you can't see it any more.

I know yours is a C question, but a lot of thread implementations solve the problem with something like this:

class Thread {
    pthread_t handle;

    static void * start (void * self) {
        static_cast <Thread *> (self) -> run ();
    }

    protected: void run () = 0;

    public: void start () {
        pthread_create (&handle, NULL, start, this);
    }

    ~ Thread () {
        pthread_join (&handle, NULL);
    }
};

You can do something analogous with C, the arg is a pointer to a malloced struct which contains the thread handle; the thread function frees it on termination.

Upvotes: 0

Owen
Owen

Reputation: 39356

Yes -- it's safe to let the variable go out of scope. But remember you have to at some point do one of two things:

1) pthread_detach() it so the kernel will free some kind of stuff that's associated with it.

2) pthread_join() it which has as a side-effect detaching it.

If you don't, I think this will be a resource leak.

Upvotes: 4

Related Questions