What is the difference between Lock and RLock

From the docs:

threading.RLock() -- A factory function that returns a new reentrant lock object. A reentrant lock must be released by the thread that acquired it. Once a thread has acquired a reentrant lock, the same thread may acquire it again without blocking; the thread must release it once for each time it has acquired it.

I am not sure why do we need this? what's the difference between Rlock and Lock?

Solution 1:

The main difference is that a Lock can only be acquired once. It cannot be acquired again, until it is released. (After it's been released, it can be re-acaquired by any thread).

An RLock on the other hand, can be acquired multiple times, by the same thread. It needs to be released the same number of times in order to be "unlocked".

Another difference is that an acquired Lock can be released by any thread, while an acquired RLock can only be released by the thread which acquired it.

Here's an example demostrating why RLock is useful at times. Suppose you have:

def f():
  g()
  h()

def g():
  h()
  do_something1()

def h():
  do_something2()

Let's say all of f, g, and h are public (i.e. can be called directly by an external caller), and all of them require syncronization.

Using a Lock, you can do something like:

lock = Lock()

def f():
  with lock:
    _g()
    _h()

def g():
  with lock:
    _g()

def _g():
  _h()
  do_something1()

def h():
  with lock:
    _h()

def _h():
  do_something2()

Basically, since f cannot call g after acquiring the lock, it needs to call a "raw" version of g (i.e. _g). So you end up with a "synced" version and a "raw" version of each function.

Using an RLock elegantly solves the problem:

lock = RLock()

def f():
  with lock:
    g()
    h()

def g():
  with lock:
    h()
    do_something1()

def h():
  with lock:
    do_something2()

Solution 2:

To expand on shx2's answer, the reason why you want to use one vs the other might be the following:

A regular Lock (mutex) is typically faster and safer.

The reason for using RLock is to avoid a dead lock due to e.g. recursion. For instance, let's put a lock in the recursive factorial function. (admittedly somewhat contrived)

from threading import Lock

lock = Lock()

def factorial(n):
    assert n > 0
    if n == 1:
        return 1
    
    with lock:       
        out = n * factorial(n - 1)

    return out

This function will cause a dead lock due to the recursive call. If we use RLock instead, however, the recursive calls can reenter the same lock as many times as needed. Hence the name reentrant (or recursive) lock.