Commit | Line | Data |
---|---|---|
a6537be9 SR |
1 | RT-mutex subsystem with PI support |
2 | ---------------------------------- | |
3 | ||
4 | RT-mutexes with priority inheritance are used to support PI-futexes, | |
5 | which enable pthread_mutex_t priority inheritance attributes | |
6 | (PTHREAD_PRIO_INHERIT). [See Documentation/pi-futex.txt for more details | |
7 | about PI-futexes.] | |
8 | ||
9 | This technology was developed in the -rt tree and streamlined for | |
10 | pthread_mutex support. | |
11 | ||
12 | Basic principles: | |
13 | ----------------- | |
14 | ||
15 | RT-mutexes extend the semantics of simple mutexes by the priority | |
16 | inheritance protocol. | |
17 | ||
18 | A low priority owner of a rt-mutex inherits the priority of a higher | |
19 | priority waiter until the rt-mutex is released. If the temporarily | |
20 | boosted owner blocks on a rt-mutex itself it propagates the priority | |
21 | boosting to the owner of the other rt_mutex it gets blocked on. The | |
22 | priority boosting is immediately removed once the rt_mutex has been | |
23 | unlocked. | |
24 | ||
25 | This approach allows us to shorten the block of high-prio tasks on | |
26 | mutexes which protect shared resources. Priority inheritance is not a | |
27 | magic bullet for poorly designed applications, but it allows | |
28 | well-designed applications to use userspace locks in critical parts of | |
29 | an high priority thread, without losing determinism. | |
30 | ||
68a1e349 | 31 | The enqueueing of the waiters into the rtmutex waiter tree is done in |
a6537be9 SR |
32 | priority order. For same priorities FIFO order is chosen. For each |
33 | rtmutex, only the top priority waiter is enqueued into the owner's | |
68a1e349 | 34 | priority waiters tree. This tree too queues in priority order. Whenever |
a6537be9 | 35 | the top priority waiter of a task changes (for example it timed out or |
68a1e349 AS |
36 | got a signal), the priority of the owner task is readjusted. The |
37 | priority enqueueing is handled by "pi_waiters". | |
a6537be9 SR |
38 | |
39 | RT-mutexes are optimized for fastpath operations and have no internal | |
40 | locking overhead when locking an uncontended mutex or unlocking a mutex | |
41 | without waiters. The optimized fastpath operations require cmpxchg | |
42 | support. [If that is not available then the rt-mutex internal spinlock | |
43 | is used] | |
44 | ||
45 | The state of the rt-mutex is tracked via the owner field of the rt-mutex | |
46 | structure: | |
47 | ||
68a1e349 AS |
48 | lock->owner holds the task_struct pointer of the owner. Bit 0 is used to |
49 | keep track of the "lock has waiters" state. | |
a6537be9 | 50 | |
68a1e349 AS |
51 | owner bit0 |
52 | NULL 0 lock is free (fast acquire possible) | |
53 | NULL 1 lock is free and has waiters and the top waiter | |
54 | is going to take the lock* | |
55 | taskpointer 0 lock is held (fast release possible) | |
56 | taskpointer 1 lock is held and has waiters** | |
a6537be9 | 57 | |
68a1e349 AS |
58 | The fast atomic compare exchange based acquire and release is only |
59 | possible when bit 0 of lock->owner is 0. | |
a6537be9 | 60 | |
68a1e349 AS |
61 | (*) It also can be a transitional state when grabbing the lock |
62 | with ->wait_lock is held. To prevent any fast path cmpxchg to the lock, | |
63 | we need to set the bit0 before looking at the lock, and the owner may be | |
64 | NULL in this small time, hence this can be a transitional state. | |
a6537be9 | 65 | |
68a1e349 AS |
66 | (**) There is a small time when bit 0 is set but there are no |
67 | waiters. This can happen when grabbing the lock in the slow path. | |
68 | To prevent a cmpxchg of the owner releasing the lock, we need to | |
69 | set this bit before looking at the lock. | |
70 | ||
71 | BTW, there is still technically a "Pending Owner", it's just not called | |
72 | that anymore. The pending owner happens to be the top_waiter of a lock | |
73 | that has no owner and has been woken up to grab the lock. |