[net-next-2.6.git] / lib / rwsem.c

/* rwsem.c: R/W semaphores: contention handling functions
 *
 * Written by David Howells (dhowells@redhat.com).
 * Derived from arch/i386/kernel/semaphore.c
 */
#include <linux/rwsem.h>
#include <linux/sched.h>
#include <linux/init.h>
#include <linux/module.h>

/*
 * Initialize an rwsem:
 */
void __init_rwsem(struct rw_semaphore *sem, const char *name,
		  struct lock_class_key *key)
{
#ifdef CONFIG_DEBUG_LOCK_ALLOC
	/*
	 * Make sure we are not reinitializing a held semaphore:
	 */
	debug_check_no_locks_freed((void *)sem, sizeof(*sem));
	lockdep_init_map(&sem->dep_map, name, key, 0);
#endif
	sem->count = RWSEM_UNLOCKED_VALUE;
	spin_lock_init(&sem->wait_lock);
	INIT_LIST_HEAD(&sem->wait_list);
}

EXPORT_SYMBOL(__init_rwsem);

struct rwsem_waiter {
	struct list_head list;
	struct task_struct *task;
	unsigned int flags;
#define RWSEM_WAITING_FOR_READ	0x00000001
#define RWSEM_WAITING_FOR_WRITE	0x00000002
};

/* Wake types for __rwsem_do_wake().  Note that RWSEM_WAKE_NO_ACTIVE and
 * RWSEM_WAKE_READ_OWNED imply that the spinlock must have been kept held
 * since the rwsem value was observed.
 */
#define RWSEM_WAKE_ANY        0 /* Wake whatever's at head of wait list */
#define RWSEM_WAKE_NO_ACTIVE  1 /* rwsem was observed with no active thread */
#define RWSEM_WAKE_READ_OWNED 2 /* rwsem was observed to be read owned */

/*
 * handle the lock release when processes blocked on it that can now run
 * - if we come here from up_xxxx(), then:
 *   - the 'active part' of count (&0x0000ffff) reached 0 (but may have changed)
 *   - the 'waiting part' of count (&0xffff0000) is -ve (and will still be so)
 * - there must be someone on the queue
 * - the spinlock must be held by the caller
 * - woken process blocks are discarded from the list after having task zeroed
 * - writers are only woken if downgrading is false
 */
static struct rw_semaphore *
__rwsem_do_wake(struct rw_semaphore *sem, int wake_type)
{
	struct rwsem_waiter *waiter;
	struct task_struct *tsk;
	struct list_head *next;
	signed long oldcount, woken, loop;

	waiter = list_entry(sem->wait_list.next, struct rwsem_waiter, list);
	if (!(waiter->flags & RWSEM_WAITING_FOR_WRITE))
		goto readers_only;

	if (wake_type == RWSEM_WAKE_READ_OWNED)
		goto out;

	/* There's a writer at the front of the queue - try to grant it the
	 * write lock.  However, we only wake this writer if we can transition
	 * the active part of the count from 0 -> 1
	 */
 try_again_write:
	oldcount = rwsem_atomic_update(RWSEM_ACTIVE_BIAS, sem)
						- RWSEM_ACTIVE_BIAS;
	if (oldcount & RWSEM_ACTIVE_MASK)
		/* Someone grabbed the sem already */
		goto undo_write;

	/* We must be careful not to touch 'waiter' after we set ->task = NULL.
	 * It is an allocated on the waiter's stack and may become invalid at
	 * any time after that point (due to a wakeup from another source).
	 */
	list_del(&waiter->list);
	tsk = waiter->task;
	smp_mb();
	waiter->task = NULL;
	wake_up_process(tsk);
	put_task_struct(tsk);
	goto out;

 readers_only:
	/* If we come here from up_xxxx(), another thread might have reached
	 * rwsem_down_failed_common() before we acquired the spinlock and
	 * woken up a waiter, making it now active.  We prefer to check for
	 * this first in order to not spend too much time with the spinlock
	 * held if we're not going to be able to wake up readers in the end.
	 *
	 * Note that we do not need to update the rwsem count: any writer
	 * trying to acquire rwsem will run rwsem_down_write_failed() due
	 * to the waiting threads and block trying to acquire the spinlock.
	 *
	 * We use a dummy atomic update in order to acquire the cache line
	 * exclusively since we expect to succeed and run the final rwsem
	 * count adjustment pretty soon.
	 */
	if (wake_type == RWSEM_WAKE_ANY &&
	    (rwsem_atomic_update(0, sem) & RWSEM_ACTIVE_MASK))
		/* Someone grabbed the sem already */
		goto out;

	/* Grant an infinite number of read locks to the readers at the front
	 * of the queue.  Note we increment the 'active part' of the count by
	 * the number of readers before waking any processes up.
	 */
	woken = 0;
	do {
		woken++;

		if (waiter->list.next == &sem->wait_list)
			break;

		waiter = list_entry(waiter->list.next,
					struct rwsem_waiter, list);

	} while (waiter->flags & RWSEM_WAITING_FOR_READ);

	loop = woken;
	woken *= RWSEM_ACTIVE_BIAS - RWSEM_WAITING_BIAS;

	rwsem_atomic_add(woken, sem);

	next = sem->wait_list.next;
	for (; loop > 0; loop--) {
		waiter = list_entry(next, struct rwsem_waiter, list);
		next = waiter->list.next;
		tsk = waiter->task;
		smp_mb();
		waiter->task = NULL;
		wake_up_process(tsk);
		put_task_struct(tsk);
	}

	sem->wait_list.next = next;
	next->prev = &sem->wait_list;

 out:
	return sem;

	/* undo the change to the active count, but check for a transition
	 * 1->0 */
 undo_write:
	if (rwsem_atomic_update(-RWSEM_ACTIVE_BIAS, sem) & RWSEM_ACTIVE_MASK)
		goto out;
	goto try_again_write;
}

/*
 * wait for a lock to be granted
 */
static struct rw_semaphore __sched *
rwsem_down_failed_common(struct rw_semaphore *sem,
			struct rwsem_waiter *waiter, signed long adjustment)
{
	struct task_struct *tsk = current;
	signed long count;

	set_task_state(tsk, TASK_UNINTERRUPTIBLE);

	/* set up my own style of waitqueue */
	spin_lock_irq(&sem->wait_lock);
	waiter->task = tsk;
	get_task_struct(tsk);

	list_add_tail(&waiter->list, &sem->wait_list);

	/* we're now waiting on the lock, but no longer actively locking */
	count = rwsem_atomic_update(adjustment, sem);

	/* if there are no active locks, wake the front queued process(es) up */
	if (!(count & RWSEM_ACTIVE_MASK))
		sem = __rwsem_do_wake(sem, RWSEM_WAKE_NO_ACTIVE);

	spin_unlock_irq(&sem->wait_lock);

	/* wait to be given the lock */
	for (;;) {
		if (!waiter->task)
			break;
		schedule();
		set_task_state(tsk, TASK_UNINTERRUPTIBLE);
	}

	tsk->state = TASK_RUNNING;

	return sem;
}

/*
 * wait for the read lock to be granted
 */
asmregparm struct rw_semaphore __sched *
rwsem_down_read_failed(struct rw_semaphore *sem)
{
	struct rwsem_waiter waiter;

	waiter.flags = RWSEM_WAITING_FOR_READ;
	rwsem_down_failed_common(sem, &waiter,
				RWSEM_WAITING_BIAS - RWSEM_ACTIVE_BIAS);
	return sem;
}

/*
 * wait for the write lock to be granted
 */
asmregparm struct rw_semaphore __sched *
rwsem_down_write_failed(struct rw_semaphore *sem)
{
	struct rwsem_waiter waiter;

	waiter.flags = RWSEM_WAITING_FOR_WRITE;
	rwsem_down_failed_common(sem, &waiter, -RWSEM_ACTIVE_BIAS);

	return sem;
}

/*
 * handle waking up a waiter on the semaphore
 * - up_read/up_write has decremented the active part of count if we come here
 */
asmregparm struct rw_semaphore *rwsem_wake(struct rw_semaphore *sem)
{
	unsigned long flags;

	spin_lock_irqsave(&sem->wait_lock, flags);

	/* do nothing if list empty */
	if (!list_empty(&sem->wait_list))
		sem = __rwsem_do_wake(sem, RWSEM_WAKE_ANY);

	spin_unlock_irqrestore(&sem->wait_lock, flags);

	return sem;
}

/*
 * downgrade a write lock into a read lock
 * - caller incremented waiting part of count and discovered it still negative
 * - just wake up any readers at the front of the queue
 */
asmregparm struct rw_semaphore *rwsem_downgrade_wake(struct rw_semaphore *sem)
{
	unsigned long flags;

	spin_lock_irqsave(&sem->wait_lock, flags);

	/* do nothing if list empty */
	if (!list_empty(&sem->wait_list))
		sem = __rwsem_do_wake(sem, RWSEM_WAKE_READ_OWNED);

	spin_unlock_irqrestore(&sem->wait_lock, flags);

	return sem;
}

EXPORT_SYMBOL(rwsem_down_read_failed);
EXPORT_SYMBOL(rwsem_down_write_failed);
EXPORT_SYMBOL(rwsem_wake);
EXPORT_SYMBOL(rwsem_downgrade_wake);
Commit	Line	Data
1da177e4 LT	1	/* rwsem.c: R/W semaphores: contention handling functions
	2	*
	3	* Written by David Howells (dhowells@redhat.com).
	4	* Derived from arch/i386/kernel/semaphore.c
	5	*/
	6	#include <linux/rwsem.h>
	7	#include <linux/sched.h>
	8	#include <linux/init.h>
	9	#include <linux/module.h>
	10
4ea2176d IM	11	/*
	12	* Initialize an rwsem:
	13	*/
	14	void __init_rwsem(struct rw_semaphore sem, const char name,
	15	struct lock_class_key *key)
	16	{
	17	#ifdef CONFIG_DEBUG_LOCK_ALLOC
	18	/*
	19	* Make sure we are not reinitializing a held semaphore:
	20	*/
	21	debug_check_no_locks_freed((void )sem, sizeof(sem));
4dfbb9d8	22	lockdep_init_map(&sem->dep_map, name, key, 0);
4ea2176d IM	23	#endif
	24	sem->count = RWSEM_UNLOCKED_VALUE;
	25	spin_lock_init(&sem->wait_lock);
	26	INIT_LIST_HEAD(&sem->wait_list);
	27	}
	28
	29	EXPORT_SYMBOL(__init_rwsem);
	30
1da177e4 LT	31	struct rwsem_waiter {
	32	struct list_head list;
	33	struct task_struct *task;
	34	unsigned int flags;
	35	#define RWSEM_WAITING_FOR_READ 0x00000001
	36	#define RWSEM_WAITING_FOR_WRITE 0x00000002
	37	};
	38
70bdc6e0 ML	39	/* Wake types for __rwsem_do_wake(). Note that RWSEM_WAKE_NO_ACTIVE and
	40	* RWSEM_WAKE_READ_OWNED imply that the spinlock must have been kept held
	41	* since the rwsem value was observed.
	42	*/
	43	#define RWSEM_WAKE_ANY 0 /* Wake whatever's at head of wait list */
	44	#define RWSEM_WAKE_NO_ACTIVE 1 /* rwsem was observed with no active thread */
	45	#define RWSEM_WAKE_READ_OWNED 2 /* rwsem was observed to be read owned */
	46
1da177e4 LT	47	/*
	48	* handle the lock release when processes blocked on it that can now run
	49	* - if we come here from up_xxxx(), then:
	50	* - the 'active part' of count (&0x0000ffff) reached 0 (but may have changed)
	51	* - the 'waiting part' of count (&0xffff0000) is -ve (and will still be so)
345af7bf	52	* - there must be someone on the queue
1da177e4 LT	53	* - the spinlock must be held by the caller
	54	* - woken process blocks are discarded from the list after having task zeroed
	55	* - writers are only woken if downgrading is false
	56	*/
70bdc6e0 ML	57	static struct rw_semaphore *
70bdc6e0 ML	58	__rwsem_do_wake(struct rw_semaphore *sem, int wake_type)
1da177e4 LT	59	{
	60	struct rwsem_waiter *waiter;
	61	struct task_struct *tsk;
	62	struct list_head *next;
	63	signed long oldcount, woken, loop;
	64
345af7bf ML	65	waiter = list_entry(sem->wait_list.next, struct rwsem_waiter, list);
	66	if (!(waiter->flags & RWSEM_WAITING_FOR_WRITE))
	67	goto readers_only;
	68
70bdc6e0	69	if (wake_type == RWSEM_WAKE_READ_OWNED)
345af7bf	70	goto out;
1da177e4	71
345af7bf ML	72	/* There's a writer at the front of the queue - try to grant it the
	73	* write lock. However, we only wake this writer if we can transition
	74	* the active part of the count from 0 -> 1
1da177e4	75	*/
345af7bf	76	try_again_write:
1da177e4 LT	77	oldcount = rwsem_atomic_update(RWSEM_ACTIVE_BIAS, sem)
	78	- RWSEM_ACTIVE_BIAS;
	79	if (oldcount & RWSEM_ACTIVE_MASK)
345af7bf ML	80	/* Someone grabbed the sem already */
345af7bf ML	81	goto undo_write;
1da177e4 LT	82
	83	/* We must be careful not to touch 'waiter' after we set ->task = NULL.
	84	* It is an allocated on the waiter's stack and may become invalid at
	85	* any time after that point (due to a wakeup from another source).
	86	*/
	87	list_del(&waiter->list);
	88	tsk = waiter->task;
d59dd462	89	smp_mb();
1da177e4 LT	90	waiter->task = NULL;
	91	wake_up_process(tsk);
	92	put_task_struct(tsk);
	93	goto out;
	94
345af7bf	95	readers_only:
70bdc6e0 ML	96	/* If we come here from up_xxxx(), another thread might have reached
	97	* rwsem_down_failed_common() before we acquired the spinlock and
	98	* woken up a waiter, making it now active. We prefer to check for
	99	* this first in order to not spend too much time with the spinlock
	100	* held if we're not going to be able to wake up readers in the end.
	101	*
	102	* Note that we do not need to update the rwsem count: any writer
	103	* trying to acquire rwsem will run rwsem_down_write_failed() due
	104	* to the waiting threads and block trying to acquire the spinlock.
	105	*
	106	* We use a dummy atomic update in order to acquire the cache line
	107	* exclusively since we expect to succeed and run the final rwsem
	108	* count adjustment pretty soon.
	109	*/
	110	if (wake_type == RWSEM_WAKE_ANY &&
	111	(rwsem_atomic_update(0, sem) & RWSEM_ACTIVE_MASK))
345af7bf	112	/* Someone grabbed the sem already */
70bdc6e0	113	goto out;
1da177e4	114
345af7bf ML	115	/* Grant an infinite number of read locks to the readers at the front
	116	* of the queue. Note we increment the 'active part' of the count by
	117	* the number of readers before waking any processes up.
1da177e4	118	*/
1da177e4 LT	119	woken = 0;
	120	do {
	121	woken++;
	122
	123	if (waiter->list.next == &sem->wait_list)
	124	break;
	125
	126	waiter = list_entry(waiter->list.next,
	127	struct rwsem_waiter, list);
	128
	129	} while (waiter->flags & RWSEM_WAITING_FOR_READ);
	130
	131	loop = woken;
	132	woken *= RWSEM_ACTIVE_BIAS - RWSEM_WAITING_BIAS;
1da177e4 LT	133
	134	rwsem_atomic_add(woken, sem);
	135
	136	next = sem->wait_list.next;
	137	for (; loop > 0; loop--) {
	138	waiter = list_entry(next, struct rwsem_waiter, list);
	139	next = waiter->list.next;
	140	tsk = waiter->task;
d59dd462	141	smp_mb();
1da177e4 LT	142	waiter->task = NULL;
	143	wake_up_process(tsk);
	144	put_task_struct(tsk);
	145	}
	146
	147	sem->wait_list.next = next;
	148	next->prev = &sem->wait_list;
	149
	150	out:
1da177e4 LT	151	return sem;
1da177e4 LT	152
91af7081 ML	153	/* undo the change to the active count, but check for a transition
91af7081 ML	154	* 1->0 */
345af7bf ML	155	undo_write:
	156	if (rwsem_atomic_update(-RWSEM_ACTIVE_BIAS, sem) & RWSEM_ACTIVE_MASK)
	157	goto out;
	158	goto try_again_write;
1da177e4 LT	159	}
	160
	161	/*
	162	* wait for a lock to be granted
	163	*/
c7af77b5	164	static struct rw_semaphore __sched *
1da177e4 LT	165	rwsem_down_failed_common(struct rw_semaphore *sem,
	166	struct rwsem_waiter *waiter, signed long adjustment)
	167	{
	168	struct task_struct *tsk = current;
	169	signed long count;
	170
	171	set_task_state(tsk, TASK_UNINTERRUPTIBLE);
	172
	173	/* set up my own style of waitqueue */
	174	spin_lock_irq(&sem->wait_lock);
	175	waiter->task = tsk;
	176	get_task_struct(tsk);
	177
	178	list_add_tail(&waiter->list, &sem->wait_list);
	179
70bdc6e0	180	/* we're now waiting on the lock, but no longer actively locking */
1da177e4 LT	181	count = rwsem_atomic_update(adjustment, sem);
	182
	183	/* if there are no active locks, wake the front queued process(es) up */
	184	if (!(count & RWSEM_ACTIVE_MASK))
70bdc6e0	185	sem = __rwsem_do_wake(sem, RWSEM_WAKE_NO_ACTIVE);
1da177e4 LT	186
	187	spin_unlock_irq(&sem->wait_lock);
	188
	189	/* wait to be given the lock */
	190	for (;;) {
	191	if (!waiter->task)
	192	break;
	193	schedule();
	194	set_task_state(tsk, TASK_UNINTERRUPTIBLE);
	195	}
	196
	197	tsk->state = TASK_RUNNING;
	198
	199	return sem;
	200	}
	201
	202	/*
	203	* wait for the read lock to be granted
	204	*/
d50efc6c	205	asmregparm struct rw_semaphore __sched *
1da177e4 LT	206	rwsem_down_read_failed(struct rw_semaphore *sem)
	207	{
	208	struct rwsem_waiter waiter;
	209
1da177e4 LT	210	waiter.flags = RWSEM_WAITING_FOR_READ;
	211	rwsem_down_failed_common(sem, &waiter,
	212	RWSEM_WAITING_BIAS - RWSEM_ACTIVE_BIAS);
1da177e4 LT	213	return sem;
	214	}
	215
	216	/*
	217	* wait for the write lock to be granted
	218	*/
d50efc6c	219	asmregparm struct rw_semaphore __sched *
1da177e4 LT	220	rwsem_down_write_failed(struct rw_semaphore *sem)
	221	{
	222	struct rwsem_waiter waiter;
	223
1da177e4 LT	224	waiter.flags = RWSEM_WAITING_FOR_WRITE;
	225	rwsem_down_failed_common(sem, &waiter, -RWSEM_ACTIVE_BIAS);
	226
1da177e4 LT	227	return sem;
	228	}
	229
	230	/*
	231	* handle waking up a waiter on the semaphore
	232	* - up_read/up_write has decremented the active part of count if we come here
	233	*/
d50efc6c	234	asmregparm struct rw_semaphore rwsem_wake(struct rw_semaphore sem)
1da177e4 LT	235	{
	236	unsigned long flags;
	237
1da177e4 LT	238	spin_lock_irqsave(&sem->wait_lock, flags);
	239
	240	/* do nothing if list empty */
	241	if (!list_empty(&sem->wait_list))
70bdc6e0	242	sem = __rwsem_do_wake(sem, RWSEM_WAKE_ANY);
1da177e4 LT	243
	244	spin_unlock_irqrestore(&sem->wait_lock, flags);
	245
1da177e4 LT	246	return sem;
	247	}
	248
	249	/*
	250	* downgrade a write lock into a read lock
	251	* - caller incremented waiting part of count and discovered it still negative
	252	* - just wake up any readers at the front of the queue
	253	*/
d50efc6c	254	asmregparm struct rw_semaphore rwsem_downgrade_wake(struct rw_semaphore sem)
1da177e4 LT	255	{
	256	unsigned long flags;
	257
1da177e4 LT	258	spin_lock_irqsave(&sem->wait_lock, flags);
	259
	260	/* do nothing if list empty */
	261	if (!list_empty(&sem->wait_list))
70bdc6e0	262	sem = __rwsem_do_wake(sem, RWSEM_WAKE_READ_OWNED);
1da177e4 LT	263
	264	spin_unlock_irqrestore(&sem->wait_lock, flags);
	265
1da177e4 LT	266	return sem;
	267	}
	268
	269	EXPORT_SYMBOL(rwsem_down_read_failed);
	270	EXPORT_SYMBOL(rwsem_down_write_failed);
	271	EXPORT_SYMBOL(rwsem_wake);
	272	EXPORT_SYMBOL(rwsem_downgrade_wake);