What is the priority ceiling protocol and how does it differ from inheritance?

In the priority ceiling protocol, every mutex is assigned a ceiling equal to the highest priority of any task that could ever use it, and when a task acquires that mutex it immediately runs at the ceiling priority, not only when a higher-priority task later blocks on it. The difference from priority inheritance is timing and scope: inheritance is reactive, boosting the holder only when a higher-priority task actually blocks on the mutex, whereas the (immediate) ceiling protocol is proactive, elevating the holder as soon as it takes the lock regardless of whether anyone blocks. The proactive approach has two advantages: it prevents priority inversion (no medium task can preempt a lock holder, since the holder already runs at the ceiling) and it also prevents a class of deadlocks and limits the number of times a task can be blocked, which makes worst-case blocking easier to bound for hard-real-time analysis. The cost is that you must know each mutex's ceiling in advance (which tasks use it), and a task may run at an elevated priority even when no contention exists, slightly reducing concurrency. Inheritance needs no advance ceiling declaration and only elevates under actual contention, which is simpler and is what FreeRTOS uses by default; ceiling protocols appear in RTOSes targeting rigorous real-time guarantees. So both prevent unbounded inversion; ceiling does it proactively with stronger deadlock/analysis properties at the cost of needing ceilings declared.