Simple Read-write Lock uses a counter and a boolean flag to keep track of multiple readers and a writer, but does not prioritize writers. A common lock is used to ensure internal updates happen atomically and a common condition is used for indicating either "no reader" or "no writer".
Acquiring the read lock involves holding the common lock, waiting until there is no writer, and finally incrementing the readers count. Releasing the read lock involves holding the common lock, decrementing the reader count, and signalling any writer/readers.
Acquiring the write lock involves holding the common lock, waiting until there are no writers and readers, and finally indicating presence of a writer. Releasing the write lock involves involves holding the common lock, indicating absence of writer, and signalling any writer/readers.
Even though the algorithm is correct, it is not quite satisfactory. If readers are much more frequent than writers, as is usually the case, the writers could be locked out for a long period of time by a continual stream of readers. Due to this lack of writer prioritization, this type of lock is generally only suitable for educational purposes.
Course: Concurrent Data Structures, Monsoon 2020
Taught by: Prof. Govindarajulu Regeti
readLock().lock():
1. Acquire common lock.
2. Wait until there is no writer.
3. Increment readers count.
4. Release common lock.
readLock().unlock():
1. Acquire common lock.
2. Decrement readers count.
3. If no readers, signal any writer/readers.
4. Release common lock.
writeLock().lock():
1. Acquire common lock.
2. Wait until there is no writer, reader.
3. Indicate presence of writer.
4. Release common lock.
writeLock().unlock():
1. Acquire common lock.
2. Indicate absence of writer.
3. Signal any writer/readers.
4. Release common lock.
See SimpleReadWriteLock.java for code, Main.java for test, and repl.it for output.