Line data Source code
1 : // SPDX-License-Identifier: GPL-2.0
2 : #include <linux/percpu.h>
3 : #include <linux/sched.h>
4 : #include <linux/osq_lock.h>
5 :
6 : /*
7 : * An MCS like lock especially tailored for optimistic spinning for sleeping
8 : * lock implementations (mutex, rwsem, etc).
9 : *
10 : * Using a single mcs node per CPU is safe because sleeping locks should not be
11 : * called from interrupt context and we have preemption disabled while
12 : * spinning.
13 : */
14 : static DEFINE_PER_CPU_SHARED_ALIGNED(struct optimistic_spin_node, osq_node);
15 :
16 : /*
17 : * We use the value 0 to represent "no CPU", thus the encoded value
18 : * will be the CPU number incremented by 1.
19 : */
20 2024 : static inline int encode_cpu(int cpu_nr)
21 : {
22 2024 : return cpu_nr + 1;
23 : }
24 :
25 2608 : static inline int node_cpu(struct optimistic_spin_node *node)
26 : {
27 2608 : return node->cpu - 1;
28 : }
29 :
30 6 : static inline struct optimistic_spin_node *decode_cpu(int encoded_cpu_val)
31 : {
32 6 : int cpu_nr = encoded_cpu_val - 1;
33 :
34 6 : return per_cpu_ptr(&osq_node, cpu_nr);
35 : }
36 :
37 : /*
38 : * Get a stable @node->next pointer, either for unlock() or unqueue() purposes.
39 : * Can return NULL in case we were the last queued and we updated @lock instead.
40 : */
41 : static inline struct optimistic_spin_node *
42 0 : osq_wait_next(struct optimistic_spin_queue *lock,
43 : struct optimistic_spin_node *node,
44 : struct optimistic_spin_node *prev)
45 : {
46 0 : struct optimistic_spin_node *next = NULL;
47 0 : int curr = encode_cpu(smp_processor_id());
48 0 : int old;
49 :
50 : /*
51 : * If there is a prev node in queue, then the 'old' value will be
52 : * the prev node's CPU #, else it's set to OSQ_UNLOCKED_VAL since if
53 : * we're currently last in queue, then the queue will then become empty.
54 : */
55 0 : old = prev ? prev->cpu : OSQ_UNLOCKED_VAL;
56 :
57 0 : for (;;) {
58 0 : if (atomic_read(&lock->tail) == curr &&
59 0 : atomic_cmpxchg_acquire(&lock->tail, curr, old) == curr) {
60 : /*
61 : * We were the last queued, we moved @lock back. @prev
62 : * will now observe @lock and will complete its
63 : * unlock()/unqueue().
64 : */
65 : break;
66 : }
67 :
68 : /*
69 : * We must xchg() the @node->next value, because if we were to
70 : * leave it in, a concurrent unlock()/unqueue() from
71 : * @node->next might complete Step-A and think its @prev is
72 : * still valid.
73 : *
74 : * If the concurrent unlock()/unqueue() wins the race, we'll
75 : * wait for either @lock to point to us, through its Step-B, or
76 : * wait for a new @node->next from its Step-C.
77 : */
78 0 : if (node->next) {
79 0 : next = xchg(&node->next, NULL);
80 0 : if (next)
81 : break;
82 : }
83 :
84 0 : cpu_relax();
85 : }
86 :
87 0 : return next;
88 : }
89 :
90 1012 : bool osq_lock(struct optimistic_spin_queue *lock)
91 : {
92 1012 : struct optimistic_spin_node *node = this_cpu_ptr(&osq_node);
93 1012 : struct optimistic_spin_node *prev, *next;
94 1012 : int curr = encode_cpu(smp_processor_id());
95 1012 : int old;
96 :
97 1012 : node->locked = 0;
98 1012 : node->next = NULL;
99 1012 : node->cpu = curr;
100 :
101 : /*
102 : * We need both ACQUIRE (pairs with corresponding RELEASE in
103 : * unlock() uncontended, or fastpath) and RELEASE (to publish
104 : * the node fields we just initialised) semantics when updating
105 : * the lock tail.
106 : */
107 1012 : old = atomic_xchg(&lock->tail, curr);
108 1012 : if (old == OSQ_UNLOCKED_VAL)
109 : return true;
110 :
111 6 : prev = decode_cpu(old);
112 6 : node->prev = prev;
113 :
114 : /*
115 : * osq_lock() unqueue
116 : *
117 : * node->prev = prev osq_wait_next()
118 : * WMB MB
119 : * prev->next = node next->prev = prev // unqueue-C
120 : *
121 : * Here 'node->prev' and 'next->prev' are the same variable and we need
122 : * to ensure these stores happen in-order to avoid corrupting the list.
123 : */
124 6 : smp_wmb();
125 :
126 6 : WRITE_ONCE(prev->next, node);
127 :
128 : /*
129 : * Normally @prev is untouchable after the above store; because at that
130 : * moment unlock can proceed and wipe the node element from stack.
131 : *
132 : * However, since our nodes are static per-cpu storage, we're
133 : * guaranteed their existence -- this allows us to apply
134 : * cmpxchg in an attempt to undo our queueing.
135 : */
136 :
137 : /*
138 : * Wait to acquire the lock or cancelation. Note that need_resched()
139 : * will come with an IPI, which will wake smp_cond_load_relaxed() if it
140 : * is implemented with a monitor-wait. vcpu_is_preempted() relies on
141 : * polling, be careful.
142 : */
143 5222 : if (smp_cond_load_relaxed(&node->locked, VAL || need_resched() ||
144 : vcpu_is_preempted(node_cpu(node->prev))))
145 : return true;
146 :
147 : /* unqueue */
148 : /*
149 : * Step - A -- stabilize @prev
150 : *
151 : * Undo our @prev->next assignment; this will make @prev's
152 : * unlock()/unqueue() wait for a next pointer since @lock points to us
153 : * (or later).
154 : */
155 :
156 0 : for (;;) {
157 : /*
158 : * cpu_relax() below implies a compiler barrier which would
159 : * prevent this comparison being optimized away.
160 : */
161 0 : if (data_race(prev->next) == node &&
162 0 : cmpxchg(&prev->next, node, NULL) == node)
163 : break;
164 :
165 : /*
166 : * We can only fail the cmpxchg() racing against an unlock(),
167 : * in which case we should observe @node->locked becomming
168 : * true.
169 : */
170 0 : if (smp_load_acquire(&node->locked))
171 : return true;
172 :
173 0 : cpu_relax();
174 :
175 : /*
176 : * Or we race against a concurrent unqueue()'s step-B, in which
177 : * case its step-C will write us a new @node->prev pointer.
178 : */
179 0 : prev = READ_ONCE(node->prev);
180 : }
181 :
182 : /*
183 : * Step - B -- stabilize @next
184 : *
185 : * Similar to unlock(), wait for @node->next or move @lock from @node
186 : * back to @prev.
187 : */
188 :
189 0 : next = osq_wait_next(lock, node, prev);
190 0 : if (!next)
191 : return false;
192 :
193 : /*
194 : * Step - C -- unlink
195 : *
196 : * @prev is stable because its still waiting for a new @prev->next
197 : * pointer, @next is stable because our @node->next pointer is NULL and
198 : * it will wait in Step-A.
199 : */
200 :
201 0 : WRITE_ONCE(next->prev, prev);
202 0 : WRITE_ONCE(prev->next, next);
203 :
204 0 : return false;
205 : }
206 :
207 1012 : void osq_unlock(struct optimistic_spin_queue *lock)
208 : {
209 1012 : struct optimistic_spin_node *node, *next;
210 1012 : int curr = encode_cpu(smp_processor_id());
211 :
212 : /*
213 : * Fast path for the uncontended case.
214 : */
215 2024 : if (likely(atomic_cmpxchg_release(&lock->tail, curr,
216 : OSQ_UNLOCKED_VAL) == curr))
217 : return;
218 :
219 : /*
220 : * Second most likely case.
221 : */
222 6 : node = this_cpu_ptr(&osq_node);
223 6 : next = xchg(&node->next, NULL);
224 6 : if (next) {
225 6 : WRITE_ONCE(next->locked, 1);
226 6 : return;
227 : }
228 :
229 0 : next = osq_wait_next(lock, node, NULL);
230 0 : if (next)
231 0 : WRITE_ONCE(next->locked, 1);
232 : }
|
