Line data Source code
1 : // SPDX-License-Identifier: GPL-2.0
2 : /*
3 : * Ldisc rw semaphore
4 : *
5 : * The ldisc semaphore is semantically a rw_semaphore but which enforces
6 : * an alternate policy, namely:
7 : * 1) Supports lock wait timeouts
8 : * 2) Write waiter has priority
9 : * 3) Downgrading is not supported
10 : *
11 : * Implementation notes:
12 : * 1) Upper half of semaphore count is a wait count (differs from rwsem
13 : * in that rwsem normalizes the upper half to the wait bias)
14 : * 2) Lacks overflow checking
15 : *
16 : * The generic counting was copied and modified from include/asm-generic/rwsem.h
17 : * by Paul Mackerras <[email protected]>.
18 : *
19 : * The scheduling policy was copied and modified from lib/rwsem.c
20 : * Written by David Howells ([email protected]).
21 : *
22 : * This implementation incorporates the write lock stealing work of
23 : * Michel Lespinasse <[email protected]>.
24 : *
25 : * Copyright (C) 2013 Peter Hurley <[email protected]>
26 : */
27 :
28 : #include <linux/list.h>
29 : #include <linux/spinlock.h>
30 : #include <linux/atomic.h>
31 : #include <linux/tty.h>
32 : #include <linux/sched.h>
33 : #include <linux/sched/debug.h>
34 : #include <linux/sched/task.h>
35 :
36 :
37 : #if BITS_PER_LONG == 64
38 : # define LDSEM_ACTIVE_MASK 0xffffffffL
39 : #else
40 : # define LDSEM_ACTIVE_MASK 0x0000ffffL
41 : #endif
42 :
43 : #define LDSEM_UNLOCKED 0L
44 : #define LDSEM_ACTIVE_BIAS 1L
45 : #define LDSEM_WAIT_BIAS (-LDSEM_ACTIVE_MASK-1)
46 : #define LDSEM_READ_BIAS LDSEM_ACTIVE_BIAS
47 : #define LDSEM_WRITE_BIAS (LDSEM_WAIT_BIAS + LDSEM_ACTIVE_BIAS)
48 :
49 : struct ldsem_waiter {
50 : struct list_head list;
51 : struct task_struct *task;
52 : };
53 :
54 : /*
55 : * Initialize an ldsem:
56 : */
57 6 : void __init_ldsem(struct ld_semaphore *sem, const char *name,
58 : struct lock_class_key *key)
59 : {
60 : #ifdef CONFIG_DEBUG_LOCK_ALLOC
61 : /*
62 : * Make sure we are not reinitializing a held semaphore:
63 : */
64 6 : debug_check_no_locks_freed((void *)sem, sizeof(*sem));
65 6 : lockdep_init_map(&sem->dep_map, name, key, 0);
66 : #endif
67 6 : atomic_long_set(&sem->count, LDSEM_UNLOCKED);
68 6 : sem->wait_readers = 0;
69 6 : raw_spin_lock_init(&sem->wait_lock);
70 6 : INIT_LIST_HEAD(&sem->read_wait);
71 6 : INIT_LIST_HEAD(&sem->write_wait);
72 6 : }
73 :
74 0 : static void __ldsem_wake_readers(struct ld_semaphore *sem)
75 : {
76 0 : struct ldsem_waiter *waiter, *next;
77 0 : struct task_struct *tsk;
78 0 : long adjust, count;
79 :
80 : /*
81 : * Try to grant read locks to all readers on the read wait list.
82 : * Note the 'active part' of the count is incremented by
83 : * the number of readers before waking any processes up.
84 : */
85 0 : adjust = sem->wait_readers * (LDSEM_ACTIVE_BIAS - LDSEM_WAIT_BIAS);
86 0 : count = atomic_long_add_return(adjust, &sem->count);
87 0 : do {
88 0 : if (count > 0)
89 0 : break;
90 0 : if (atomic_long_try_cmpxchg(&sem->count, &count, count - adjust))
91 0 : return;
92 0 : } while (1);
93 :
94 0 : list_for_each_entry_safe(waiter, next, &sem->read_wait, list) {
95 0 : tsk = waiter->task;
96 0 : smp_store_release(&waiter->task, NULL);
97 0 : wake_up_process(tsk);
98 0 : put_task_struct(tsk);
99 0 : }
100 0 : INIT_LIST_HEAD(&sem->read_wait);
101 0 : sem->wait_readers = 0;
102 0 : }
103 :
104 0 : static inline int writer_trylock(struct ld_semaphore *sem)
105 : {
106 : /*
107 : * Only wake this writer if the active part of the count can be
108 : * transitioned from 0 -> 1
109 : */
110 0 : long count = atomic_long_add_return(LDSEM_ACTIVE_BIAS, &sem->count);
111 0 : do {
112 0 : if ((count & LDSEM_ACTIVE_MASK) == LDSEM_ACTIVE_BIAS)
113 0 : return 1;
114 0 : if (atomic_long_try_cmpxchg(&sem->count, &count, count - LDSEM_ACTIVE_BIAS))
115 0 : return 0;
116 0 : } while (1);
117 0 : }
118 :
119 0 : static void __ldsem_wake_writer(struct ld_semaphore *sem)
120 : {
121 0 : struct ldsem_waiter *waiter;
122 :
123 0 : waiter = list_entry(sem->write_wait.next, struct ldsem_waiter, list);
124 0 : wake_up_process(waiter->task);
125 0 : }
126 :
127 : /*
128 : * handle the lock release when processes blocked on it that can now run
129 : * - if we come here from up_xxxx(), then:
130 : * - the 'active part' of count (&0x0000ffff) reached 0 (but may have changed)
131 : * - the 'waiting part' of count (&0xffff0000) is -ve (and will still be so)
132 : * - the spinlock must be held by the caller
133 : * - woken process blocks are discarded from the list after having task zeroed
134 : */
135 0 : static void __ldsem_wake(struct ld_semaphore *sem)
136 : {
137 0 : if (!list_empty(&sem->write_wait))
138 0 : __ldsem_wake_writer(sem);
139 0 : else if (!list_empty(&sem->read_wait))
140 0 : __ldsem_wake_readers(sem);
141 0 : }
142 :
143 0 : static void ldsem_wake(struct ld_semaphore *sem)
144 : {
145 0 : unsigned long flags;
146 :
147 0 : raw_spin_lock_irqsave(&sem->wait_lock, flags);
148 0 : __ldsem_wake(sem);
149 0 : raw_spin_unlock_irqrestore(&sem->wait_lock, flags);
150 0 : }
151 :
152 : /*
153 : * wait for the read lock to be granted
154 : */
155 : static struct ld_semaphore __sched *
156 0 : down_read_failed(struct ld_semaphore *sem, long count, long timeout)
157 : {
158 0 : struct ldsem_waiter waiter;
159 0 : long adjust = -LDSEM_ACTIVE_BIAS + LDSEM_WAIT_BIAS;
160 :
161 : /* set up my own style of waitqueue */
162 0 : raw_spin_lock_irq(&sem->wait_lock);
163 :
164 : /*
165 : * Try to reverse the lock attempt but if the count has changed
166 : * so that reversing fails, check if there are no waiters,
167 : * and early-out if not
168 : */
169 0 : do {
170 0 : if (atomic_long_try_cmpxchg(&sem->count, &count, count + adjust)) {
171 0 : count += adjust;
172 0 : break;
173 : }
174 0 : if (count > 0) {
175 0 : raw_spin_unlock_irq(&sem->wait_lock);
176 0 : return sem;
177 : }
178 0 : } while (1);
179 :
180 0 : list_add_tail(&waiter.list, &sem->read_wait);
181 0 : sem->wait_readers++;
182 :
183 0 : waiter.task = current;
184 0 : get_task_struct(current);
185 :
186 : /* if there are no active locks, wake the new lock owner(s) */
187 0 : if ((count & LDSEM_ACTIVE_MASK) == 0)
188 0 : __ldsem_wake(sem);
189 :
190 0 : raw_spin_unlock_irq(&sem->wait_lock);
191 :
192 : /* wait to be given the lock */
193 0 : for (;;) {
194 0 : set_current_state(TASK_UNINTERRUPTIBLE);
195 :
196 0 : if (!smp_load_acquire(&waiter.task))
197 0 : break;
198 0 : if (!timeout)
199 0 : break;
200 0 : timeout = schedule_timeout(timeout);
201 : }
202 :
203 0 : __set_current_state(TASK_RUNNING);
204 :
205 0 : if (!timeout) {
206 : /*
207 : * Lock timed out but check if this task was just
208 : * granted lock ownership - if so, pretend there
209 : * was no timeout; otherwise, cleanup lock wait.
210 : */
211 0 : raw_spin_lock_irq(&sem->wait_lock);
212 0 : if (waiter.task) {
213 0 : atomic_long_add_return(-LDSEM_WAIT_BIAS, &sem->count);
214 0 : sem->wait_readers--;
215 0 : list_del(&waiter.list);
216 0 : raw_spin_unlock_irq(&sem->wait_lock);
217 0 : put_task_struct(waiter.task);
218 0 : return NULL;
219 : }
220 0 : raw_spin_unlock_irq(&sem->wait_lock);
221 0 : }
222 :
223 0 : return sem;
224 0 : }
225 :
226 : /*
227 : * wait for the write lock to be granted
228 : */
229 : static struct ld_semaphore __sched *
230 0 : down_write_failed(struct ld_semaphore *sem, long count, long timeout)
231 : {
232 0 : struct ldsem_waiter waiter;
233 0 : long adjust = -LDSEM_ACTIVE_BIAS;
234 0 : int locked = 0;
235 :
236 : /* set up my own style of waitqueue */
237 0 : raw_spin_lock_irq(&sem->wait_lock);
238 :
239 : /*
240 : * Try to reverse the lock attempt but if the count has changed
241 : * so that reversing fails, check if the lock is now owned,
242 : * and early-out if so.
243 : */
244 0 : do {
245 0 : if (atomic_long_try_cmpxchg(&sem->count, &count, count + adjust))
246 0 : break;
247 0 : if ((count & LDSEM_ACTIVE_MASK) == LDSEM_ACTIVE_BIAS) {
248 0 : raw_spin_unlock_irq(&sem->wait_lock);
249 0 : return sem;
250 : }
251 0 : } while (1);
252 :
253 0 : list_add_tail(&waiter.list, &sem->write_wait);
254 :
255 0 : waiter.task = current;
256 :
257 0 : set_current_state(TASK_UNINTERRUPTIBLE);
258 0 : for (;;) {
259 0 : if (!timeout)
260 0 : break;
261 0 : raw_spin_unlock_irq(&sem->wait_lock);
262 0 : timeout = schedule_timeout(timeout);
263 0 : raw_spin_lock_irq(&sem->wait_lock);
264 0 : set_current_state(TASK_UNINTERRUPTIBLE);
265 0 : locked = writer_trylock(sem);
266 0 : if (locked)
267 0 : break;
268 : }
269 :
270 0 : if (!locked)
271 0 : atomic_long_add_return(-LDSEM_WAIT_BIAS, &sem->count);
272 0 : list_del(&waiter.list);
273 :
274 : /*
275 : * In case of timeout, wake up every reader who gave the right of way
276 : * to writer. Prevent separation readers into two groups:
277 : * one that helds semaphore and another that sleeps.
278 : * (in case of no contention with a writer)
279 : */
280 0 : if (!locked && list_empty(&sem->write_wait))
281 0 : __ldsem_wake_readers(sem);
282 :
283 0 : raw_spin_unlock_irq(&sem->wait_lock);
284 :
285 0 : __set_current_state(TASK_RUNNING);
286 :
287 : /* lock wait may have timed out */
288 0 : if (!locked)
289 0 : return NULL;
290 0 : return sem;
291 0 : }
292 :
293 :
294 :
295 159 : static int __ldsem_down_read_nested(struct ld_semaphore *sem,
296 : int subclass, long timeout)
297 : {
298 159 : long count;
299 :
300 159 : rwsem_acquire_read(&sem->dep_map, subclass, 0, _RET_IP_);
301 :
302 159 : count = atomic_long_add_return(LDSEM_READ_BIAS, &sem->count);
303 159 : if (count <= 0) {
304 0 : lock_contended(&sem->dep_map, _RET_IP_);
305 0 : if (!down_read_failed(sem, count, timeout)) {
306 0 : rwsem_release(&sem->dep_map, _RET_IP_);
307 0 : return 0;
308 : }
309 0 : }
310 159 : lock_acquired(&sem->dep_map, _RET_IP_);
311 159 : return 1;
312 159 : }
313 :
314 12 : static int __ldsem_down_write_nested(struct ld_semaphore *sem,
315 : int subclass, long timeout)
316 : {
317 12 : long count;
318 :
319 12 : rwsem_acquire(&sem->dep_map, subclass, 0, _RET_IP_);
320 :
321 12 : count = atomic_long_add_return(LDSEM_WRITE_BIAS, &sem->count);
322 12 : if ((count & LDSEM_ACTIVE_MASK) != LDSEM_ACTIVE_BIAS) {
323 0 : lock_contended(&sem->dep_map, _RET_IP_);
324 0 : if (!down_write_failed(sem, count, timeout)) {
325 0 : rwsem_release(&sem->dep_map, _RET_IP_);
326 0 : return 0;
327 : }
328 0 : }
329 12 : lock_acquired(&sem->dep_map, _RET_IP_);
330 12 : return 1;
331 12 : }
332 :
333 :
334 : /*
335 : * lock for reading -- returns 1 if successful, 0 if timed out
336 : */
337 159 : int __sched ldsem_down_read(struct ld_semaphore *sem, long timeout)
338 : {
339 159 : might_sleep();
340 159 : return __ldsem_down_read_nested(sem, 0, timeout);
341 : }
342 :
343 : /*
344 : * trylock for reading -- returns 1 if successful, 0 if contention
345 : */
346 35 : int ldsem_down_read_trylock(struct ld_semaphore *sem)
347 : {
348 35 : long count = atomic_long_read(&sem->count);
349 :
350 35 : while (count >= 0) {
351 35 : if (atomic_long_try_cmpxchg(&sem->count, &count, count + LDSEM_READ_BIAS)) {
352 35 : rwsem_acquire_read(&sem->dep_map, 0, 1, _RET_IP_);
353 35 : lock_acquired(&sem->dep_map, _RET_IP_);
354 35 : return 1;
355 : }
356 : }
357 0 : return 0;
358 35 : }
359 :
360 : /*
361 : * lock for writing -- returns 1 if successful, 0 if timed out
362 : */
363 12 : int __sched ldsem_down_write(struct ld_semaphore *sem, long timeout)
364 : {
365 12 : might_sleep();
366 12 : return __ldsem_down_write_nested(sem, 0, timeout);
367 : }
368 :
369 : /*
370 : * release a read lock
371 : */
372 195 : void ldsem_up_read(struct ld_semaphore *sem)
373 : {
374 195 : long count;
375 :
376 195 : rwsem_release(&sem->dep_map, _RET_IP_);
377 :
378 195 : count = atomic_long_add_return(-LDSEM_READ_BIAS, &sem->count);
379 195 : if (count < 0 && (count & LDSEM_ACTIVE_MASK) == 0)
380 0 : ldsem_wake(sem);
381 195 : }
382 :
383 : /*
384 : * release a write lock
385 : */
386 12 : void ldsem_up_write(struct ld_semaphore *sem)
387 : {
388 12 : long count;
389 :
390 12 : rwsem_release(&sem->dep_map, _RET_IP_);
391 :
392 12 : count = atomic_long_add_return(-LDSEM_WRITE_BIAS, &sem->count);
393 12 : if (count < 0)
394 0 : ldsem_wake(sem);
395 12 : }
396 :
397 :
398 : #ifdef CONFIG_DEBUG_LOCK_ALLOC
399 :
400 0 : int ldsem_down_read_nested(struct ld_semaphore *sem, int subclass, long timeout)
401 : {
402 0 : might_sleep();
403 0 : return __ldsem_down_read_nested(sem, subclass, timeout);
404 : }
405 :
406 0 : int ldsem_down_write_nested(struct ld_semaphore *sem, int subclass,
407 : long timeout)
408 : {
409 0 : might_sleep();
410 0 : return __ldsem_down_write_nested(sem, subclass, timeout);
411 : }
412 :
413 : #endif
|
