percpu-rwsem.c 5.5 KB
Newer Older
1
#include <linux/atomic.h>
2 3 4
#include <linux/rwsem.h>
#include <linux/percpu.h>
#include <linux/wait.h>
5
#include <linux/lockdep.h>
6 7 8 9 10
#include <linux/percpu-rwsem.h>
#include <linux/rcupdate.h>
#include <linux/sched.h>
#include <linux/errno.h>

11 12
int __percpu_init_rwsem(struct percpu_rw_semaphore *brw,
			const char *name, struct lock_class_key *rwsem_key)
13 14 15 16 17
{
	brw->fast_read_ctr = alloc_percpu(int);
	if (unlikely(!brw->fast_read_ctr))
		return -ENOMEM;

18 19
	/* ->rw_sem represents the whole percpu_rw_semaphore for lockdep */
	__init_rwsem(&brw->rw_sem, name, rwsem_key);
20
	rcu_sync_init(&brw->rss, RCU_SCHED_SYNC);
21 22 23 24
	atomic_set(&brw->slow_read_ctr, 0);
	init_waitqueue_head(&brw->write_waitq);
	return 0;
}
25
EXPORT_SYMBOL_GPL(__percpu_init_rwsem);
26 27 28

void percpu_free_rwsem(struct percpu_rw_semaphore *brw)
{
29 30 31 32 33 34 35
	/*
	 * XXX: temporary kludge. The error path in alloc_super()
	 * assumes that percpu_free_rwsem() is safe after kzalloc().
	 */
	if (!brw->fast_read_ctr)
		return;

36
	rcu_sync_dtor(&brw->rss);
37 38 39 40 41 42
	free_percpu(brw->fast_read_ctr);
	brw->fast_read_ctr = NULL; /* catch use after free bugs */
}

/*
 * This is the fast-path for down_read/up_read, it only needs to ensure
43
 * there is no pending writer (atomic_read(write_ctr) == 0) and inc/dec the
44 45 46 47 48 49 50 51 52 53 54 55 56 57 58
 * fast per-cpu counter. The writer uses synchronize_sched_expedited() to
 * serialize with the preempt-disabled section below.
 *
 * The nontrivial part is that we should guarantee acquire/release semantics
 * in case when
 *
 *	R_W: down_write() comes after up_read(), the writer should see all
 *	     changes done by the reader
 * or
 *	W_R: down_read() comes after up_write(), the reader should see all
 *	     changes done by the writer
 *
 * If this helper fails the callers rely on the normal rw_semaphore and
 * atomic_dec_and_test(), so in this case we have the necessary barriers.
 *
59
 * But if it succeeds we do not have any barriers, atomic_read(write_ctr) or
60 61 62 63 64 65
 * __this_cpu_add() below can be reordered with any LOAD/STORE done by the
 * reader inside the critical section. See the comments in down_write and
 * up_write below.
 */
static bool update_fast_ctr(struct percpu_rw_semaphore *brw, unsigned int val)
{
66
	bool success;
67 68

	preempt_disable();
69 70
	success = rcu_sync_is_idle(&brw->rss);
	if (likely(success))
71 72 73 74 75 76 77 78 79
		__this_cpu_add(*brw->fast_read_ctr, val);
	preempt_enable();

	return success;
}

/*
 * Like the normal down_read() this is not recursive, the writer can
 * come after the first percpu_down_read() and create the deadlock.
80 81 82 83
 *
 * Note: returns with lock_is_held(brw->rw_sem) == T for lockdep,
 * percpu_up_read() does rwsem_release(). This pairs with the usage
 * of ->rw_sem in percpu_down/up_write().
84 85 86
 */
void percpu_down_read(struct percpu_rw_semaphore *brw)
{
87 88 89
	might_sleep();
	if (likely(update_fast_ctr(brw, +1))) {
		rwsem_acquire_read(&brw->rw_sem.dep_map, 0, 0, _RET_IP_);
90
		return;
91
	}
92 93 94

	down_read(&brw->rw_sem);
	atomic_inc(&brw->slow_read_ctr);
95 96
	/* avoid up_read()->rwsem_release() */
	__up_read(&brw->rw_sem);
97
}
98
EXPORT_SYMBOL_GPL(percpu_down_read);
99

100 101 102 103 104 105 106 107 108 109 110 111 112
int percpu_down_read_trylock(struct percpu_rw_semaphore *brw)
{
	if (unlikely(!update_fast_ctr(brw, +1))) {
		if (!__down_read_trylock(&brw->rw_sem))
			return 0;
		atomic_inc(&brw->slow_read_ctr);
		__up_read(&brw->rw_sem);
	}

	rwsem_acquire_read(&brw->rw_sem.dep_map, 0, 1, _RET_IP_);
	return 1;
}

113 114
void percpu_up_read(struct percpu_rw_semaphore *brw)
{
115 116
	rwsem_release(&brw->rw_sem.dep_map, 1, _RET_IP_);

117 118 119 120 121 122 123
	if (likely(update_fast_ctr(brw, -1)))
		return;

	/* false-positive is possible but harmless */
	if (atomic_dec_and_test(&brw->slow_read_ctr))
		wake_up_all(&brw->write_waitq);
}
124
EXPORT_SYMBOL_GPL(percpu_up_read);
125 126 127 128 129 130 131 132 133 134 135 136 137 138 139

static int clear_fast_ctr(struct percpu_rw_semaphore *brw)
{
	unsigned int sum = 0;
	int cpu;

	for_each_possible_cpu(cpu) {
		sum += per_cpu(*brw->fast_read_ctr, cpu);
		per_cpu(*brw->fast_read_ctr, cpu) = 0;
	}

	return sum;
}

/*
140 141
 * A writer increments ->write_ctr to force the readers to switch to the
 * slow mode, note the atomic_read() check in update_fast_ctr().
142 143 144 145 146 147 148 149 150 151 152
 *
 * After that the readers can only inc/dec the slow ->slow_read_ctr counter,
 * ->fast_read_ctr is stable. Once the writer moves its sum into the slow
 * counter it represents the number of active readers.
 *
 * Finally the writer takes ->rw_sem for writing and blocks the new readers,
 * then waits until the slow counter becomes zero.
 */
void percpu_down_write(struct percpu_rw_semaphore *brw)
{
	/*
153
	 * 1. Ensures that write_ctr != 0 is visible to any down_read/up_read
154 155 156 157 158 159 160 161 162
	 *    so that update_fast_ctr() can't succeed.
	 *
	 * 2. Ensures we see the result of every previous this_cpu_add() in
	 *    update_fast_ctr().
	 *
	 * 3. Ensures that if any reader has exited its critical section via
	 *    fast-path, it executes a full memory barrier before we return.
	 *    See R_W case in the comment above update_fast_ctr().
	 */
163
	rcu_sync_enter(&brw->rss);
164

165 166 167
	/* exclude other writers, and block the new readers completely */
	down_write(&brw->rw_sem);

168 169 170 171 172 173
	/* nobody can use fast_read_ctr, move its sum into slow_read_ctr */
	atomic_add(clear_fast_ctr(brw), &brw->slow_read_ctr);

	/* wait for all readers to complete their percpu_up_read() */
	wait_event(brw->write_waitq, !atomic_read(&brw->slow_read_ctr));
}
174
EXPORT_SYMBOL_GPL(percpu_down_write);
175 176 177

void percpu_up_write(struct percpu_rw_semaphore *brw)
{
178
	/* release the lock, but the readers can't use the fast-path */
179 180 181 182 183
	up_write(&brw->rw_sem);
	/*
	 * Insert the barrier before the next fast-path in down_read,
	 * see W_R case in the comment above update_fast_ctr().
	 */
184
	rcu_sync_exit(&brw->rss);
185
}
186
EXPORT_SYMBOL_GPL(percpu_up_write);