Operating systems · interactive

Producer–consumer

A producer thread makes items and a consumer thread uses them, through a shared buffer with 3 slots. The producer is faster. Without coordination they overwrite and misread; with two semaphores and a lock, each waits its turn.

Producer

···

Consumer

0 corrupted operations so far

Empty buffer with 3 slots. Semaphores: empty = 3, full = 0.

Coordination

tick 0 of 12

Step 2 of 4 · read left to right

1
A shared, bounded buffer
It decouples the two speeds, but both threads touch the same slots and counters.
Logging, job queues, pipelines.
2
Overwrites and empty reads
A full buffer gets overwritten; an empty one gets read. Unsynchronised index updates can also race.
Busy-waiting instead burns CPU.
3
Two semaphores and a lock
empty counts free slots, full counts items. Producers wait(empty) then signal(full); consumers the reverse. A mutex protects the indexes.
wait(empty); lock; put; unlock; signal(full)
4
Use a blocking queue
Languages ship this pattern ready-made: Java's BlockingQueue, Go's buffered channels, Python's queue.Queue.
A full queue is backpressure.

In production

Tools and technologies

Blocking queues implement this pattern for you.

Not tied to one language — use these names when you search docs or design a system.