We require synchronization because of the fact that, without any intervention, threads can concurrently execute critical sections where shared writeable data (shared state) is being worked upon. To defeat concurrency, we need to get rid of parallelism, and we need to serialize code that's within the critical section – the place where the shared data is being worked upon (for reading and/or writing).
To force a code path to become serialized, a common technique is to use a lock. Essentially, a lock works by guaranteeing that precisely one thread of execution can "take" or own the lock at any given point in time. Thus, using a lock to protect a critical section in your code will give you what we're after – running the critical section's code exclusively (and perhaps atomically; more on this to come):
