1 /* SPDX-License-Identifier: BSD-3-Clause
2 * Copyright(c) 2019 Arm Limited
5 #ifndef _RTE_MCSLOCK_H_
6 #define _RTE_MCSLOCK_H_
13 * This file defines the main data structure and APIs for MCS queued lock.
15 * The MCS lock (proposed by John M. Mellor-Crummey and Michael L. Scott)
16 * provides scalability by spinning on a CPU/thread local variable which
17 * avoids expensive cache bouncings. It provides fairness by maintaining
18 * a list of acquirers and passing the lock to each CPU/thread in the order
19 * they acquired the lock.
22 #include <rte_lcore.h>
23 #include <rte_common.h>
24 #include <rte_pause.h>
25 #include <rte_branch_prediction.h>
28 * The rte_mcslock_t type.
30 typedef struct rte_mcslock {
31 struct rte_mcslock *next;
32 int locked; /* 1 if the queue locked, 0 otherwise */
39 * A pointer to the pointer of a MCS lock.
40 * When the lock is initialized or declared, the msl pointer should be
43 * A pointer to a new node of MCS lock. Each CPU/thread acquiring the
44 * lock should use its 'own node'.
47 rte_mcslock_lock(rte_mcslock_t **msl, rte_mcslock_t *me)
52 __atomic_store_n(&me->locked, 1, __ATOMIC_RELAXED);
53 __atomic_store_n(&me->next, NULL, __ATOMIC_RELAXED);
55 /* If the queue is empty, the exchange operation is enough to acquire
56 * the lock. Hence, the exchange operation requires acquire semantics.
57 * The store to me->next above should complete before the node is
58 * visible to other CPUs/threads. Hence, the exchange operation requires
59 * release semantics as well.
61 prev = __atomic_exchange_n(msl, me, __ATOMIC_ACQ_REL);
62 if (likely(prev == NULL)) {
63 /* Queue was empty, no further action required,
64 * proceed with lock taken.
68 /* The store to me->next above should also complete before the node is
69 * visible to predecessor thread releasing the lock. Hence, the store
70 * prev->next also requires release semantics. Note that, for example,
71 * on ARM, the release semantics in the exchange operation is not
72 * strong as a release fence and is not sufficient to enforce the
75 __atomic_store_n(&prev->next, me, __ATOMIC_RELEASE);
77 /* The while-load of me->locked should not move above the previous
78 * store to prev->next. Otherwise it will cause a deadlock. Need a
81 __atomic_thread_fence(__ATOMIC_ACQ_REL);
82 /* If the lock has already been acquired, it first atomically
83 * places the node at the end of the queue and then proceeds
84 * to spin on me->locked until the previous lock holder resets
85 * the me->locked using mcslock_unlock().
87 rte_wait_until_equal_32((uint32_t *)&me->locked, 0, __ATOMIC_ACQUIRE);
91 * Release the MCS lock.
94 * A pointer to the pointer of a MCS lock.
96 * A pointer to the node of MCS lock passed in rte_mcslock_lock.
99 rte_mcslock_unlock(rte_mcslock_t **msl, rte_mcslock_t *me)
101 /* Check if there are more nodes in the queue. */
102 if (likely(__atomic_load_n(&me->next, __ATOMIC_RELAXED) == NULL)) {
103 /* No, last member in the queue. */
104 rte_mcslock_t *save_me = __atomic_load_n(&me, __ATOMIC_RELAXED);
106 /* Release the lock by setting it to NULL */
107 if (likely(__atomic_compare_exchange_n(msl, &save_me, NULL, 0,
108 __ATOMIC_RELEASE, __ATOMIC_RELAXED)))
111 /* Speculative execution would be allowed to read in the
112 * while-loop first. This has the potential to cause a
113 * deadlock. Need a load barrier.
115 __atomic_thread_fence(__ATOMIC_ACQUIRE);
116 /* More nodes added to the queue by other CPUs.
117 * Wait until the next pointer is set.
120 next = (uintptr_t *)&me->next;
121 RTE_WAIT_UNTIL_MASKED(next, UINTPTR_MAX, !=, 0,
125 /* Pass lock to next waiter. */
126 __atomic_store_n(&me->next->locked, 0, __ATOMIC_RELEASE);
130 * Try to take the lock.
133 * A pointer to the pointer of a MCS lock.
135 * A pointer to a new node of MCS lock.
137 * 1 if the lock is successfully taken; 0 otherwise.
140 rte_mcslock_trylock(rte_mcslock_t **msl, rte_mcslock_t *me)
143 __atomic_store_n(&me->next, NULL, __ATOMIC_RELAXED);
146 rte_mcslock_t *expected = NULL;
148 /* The lock can be taken only when the queue is empty. Hence,
149 * the compare-exchange operation requires acquire semantics.
150 * The store to me->next above should complete before the node
151 * is visible to other CPUs/threads. Hence, the compare-exchange
152 * operation requires release semantics as well.
154 return __atomic_compare_exchange_n(msl, &expected, me, 0,
155 __ATOMIC_ACQ_REL, __ATOMIC_RELAXED);
159 * Test if the lock is taken.
162 * A pointer to a MCS lock node.
164 * 1 if the lock is currently taken; 0 otherwise.
167 rte_mcslock_is_locked(rte_mcslock_t *msl)
169 return (__atomic_load_n(&msl, __ATOMIC_RELAXED) != NULL);
172 #endif /* _RTE_MCSLOCK_H_ */