[deliverable/linux.git] / fs / eventfd.c

/*
 *  fs/eventfd.c
 *
 *  Copyright (C) 2007  Davide Libenzi <davidel@xmailserver.org>
 *
 */

#include <linux/file.h>
#include <linux/poll.h>
#include <linux/init.h>
#include <linux/fs.h>
#include <linux/sched.h>
#include <linux/kernel.h>
#include <linux/slab.h>
#include <linux/list.h>
#include <linux/spinlock.h>
#include <linux/anon_inodes.h>
#include <linux/syscalls.h>
#include <linux/export.h>
#include <linux/kref.h>
#include <linux/eventfd.h>
#include <linux/proc_fs.h>
#include <linux/seq_file.h>

struct eventfd_ctx {
	struct kref kref;
	wait_queue_head_t wqh;
	/*
	 * Every time that a write(2) is performed on an eventfd, the
	 * value of the __u64 being written is added to "count" and a
	 * wakeup is performed on "wqh". A read(2) will return the "count"
	 * value to userspace, and will reset "count" to zero. The kernel
	 * side eventfd_signal() also, adds to the "count" counter and
	 * issue a wakeup.
	 */
	__u64 count;
	unsigned int flags;
};

/**
 * eventfd_signal - Adds @n to the eventfd counter.
 * @ctx: [in] Pointer to the eventfd context.
 * @n: [in] Value of the counter to be added to the eventfd internal counter.
 *          The value cannot be negative.
 *
 * This function is supposed to be called by the kernel in paths that do not
 * allow sleeping. In this function we allow the counter to reach the ULLONG_MAX
 * value, and we signal this as overflow condition by returning a POLLERR
 * to poll(2).
 *
 * Returns the amount by which the counter was incremented.  This will be less
 * than @n if the counter has overflowed.
 */
__u64 eventfd_signal(struct eventfd_ctx *ctx, __u64 n)
{
	unsigned long flags;

	spin_lock_irqsave(&ctx->wqh.lock, flags);
	if (ULLONG_MAX - ctx->count < n)
		n = ULLONG_MAX - ctx->count;
	ctx->count += n;
	if (waitqueue_active(&ctx->wqh))
		wake_up_locked_poll(&ctx->wqh, POLLIN);
	spin_unlock_irqrestore(&ctx->wqh.lock, flags);

	return n;
}
EXPORT_SYMBOL_GPL(eventfd_signal);

static void eventfd_free_ctx(struct eventfd_ctx *ctx)
{
	kfree(ctx);
}

static void eventfd_free(struct kref *kref)
{
	struct eventfd_ctx *ctx = container_of(kref, struct eventfd_ctx, kref);

	eventfd_free_ctx(ctx);
}

/**
 * eventfd_ctx_get - Acquires a reference to the internal eventfd context.
 * @ctx: [in] Pointer to the eventfd context.
 *
 * Returns: In case of success, returns a pointer to the eventfd context.
 */
struct eventfd_ctx *eventfd_ctx_get(struct eventfd_ctx *ctx)
{
	kref_get(&ctx->kref);
	return ctx;
}
EXPORT_SYMBOL_GPL(eventfd_ctx_get);

/**
 * eventfd_ctx_put - Releases a reference to the internal eventfd context.
 * @ctx: [in] Pointer to eventfd context.
 *
 * The eventfd context reference must have been previously acquired either
 * with eventfd_ctx_get() or eventfd_ctx_fdget().
 */
void eventfd_ctx_put(struct eventfd_ctx *ctx)
{
	kref_put(&ctx->kref, eventfd_free);
}
EXPORT_SYMBOL_GPL(eventfd_ctx_put);

static int eventfd_release(struct inode *inode, struct file *file)
{
	struct eventfd_ctx *ctx = file->private_data;

	wake_up_poll(&ctx->wqh, POLLHUP);
	eventfd_ctx_put(ctx);
	return 0;
}

static unsigned int eventfd_poll(struct file *file, poll_table *wait)
{
	struct eventfd_ctx *ctx = file->private_data;
	unsigned int events = 0;
	u64 count;

	poll_wait(file, &ctx->wqh, wait);

	/*
	 * All writes to ctx->count occur within ctx->wqh.lock.  This read
	 * can be done outside ctx->wqh.lock because we know that poll_wait
	 * takes that lock (through add_wait_queue) if our caller will sleep.
	 *
	 * The read _can_ therefore seep into add_wait_queue's critical
	 * section, but cannot move above it!  add_wait_queue's spin_lock acts
	 * as an acquire barrier and ensures that the read be ordered properly
	 * against the writes.  The following CAN happen and is safe:
	 *
	 *     poll                               write
	 *     -----------------                  ------------
	 *     lock ctx->wqh.lock (in poll_wait)
	 *     count = ctx->count
	 *     __add_wait_queue
	 *     unlock ctx->wqh.lock
	 *                                        lock ctx->qwh.lock
	 *                                        ctx->count += n
	 *                                        if (waitqueue_active)
	 *                                          wake_up_locked_poll
	 *                                        unlock ctx->qwh.lock
	 *     eventfd_poll returns 0
	 *
	 * but the following, which would miss a wakeup, cannot happen:
	 *
	 *     poll                               write
	 *     -----------------                  ------------
	 *     count = ctx->count (INVALID!)
	 *                                        lock ctx->qwh.lock
	 *                                        ctx->count += n
	 *                                        **waitqueue_active is false**
	 *                                        **no wake_up_locked_poll!**
	 *                                        unlock ctx->qwh.lock
	 *     lock ctx->wqh.lock (in poll_wait)
	 *     __add_wait_queue
	 *     unlock ctx->wqh.lock
	 *     eventfd_poll returns 0
	 */
	count = READ_ONCE(ctx->count);

	if (count > 0)
		events |= POLLIN;
	if (count == ULLONG_MAX)
		events |= POLLERR;
	if (ULLONG_MAX - 1 > count)
		events |= POLLOUT;

	return events;
}

static void eventfd_ctx_do_read(struct eventfd_ctx *ctx, __u64 *cnt)
{
	*cnt = (ctx->flags & EFD_SEMAPHORE) ? 1 : ctx->count;
	ctx->count -= *cnt;
}

/**
 * eventfd_ctx_remove_wait_queue - Read the current counter and removes wait queue.
 * @ctx: [in] Pointer to eventfd context.
 * @wait: [in] Wait queue to be removed.
 * @cnt: [out] Pointer to the 64-bit counter value.
 *
 * Returns %0 if successful, or the following error codes:
 *
 * -EAGAIN      : The operation would have blocked.
 *
 * This is used to atomically remove a wait queue entry from the eventfd wait
 * queue head, and read/reset the counter value.
 */
int eventfd_ctx_remove_wait_queue(struct eventfd_ctx *ctx, wait_queue_t *wait,
				  __u64 *cnt)
{
	unsigned long flags;

	spin_lock_irqsave(&ctx->wqh.lock, flags);
	eventfd_ctx_do_read(ctx, cnt);
	__remove_wait_queue(&ctx->wqh, wait);
	if (*cnt != 0 && waitqueue_active(&ctx->wqh))
		wake_up_locked_poll(&ctx->wqh, POLLOUT);
	spin_unlock_irqrestore(&ctx->wqh.lock, flags);

	return *cnt != 0 ? 0 : -EAGAIN;
}
EXPORT_SYMBOL_GPL(eventfd_ctx_remove_wait_queue);

/**
 * eventfd_ctx_read - Reads the eventfd counter or wait if it is zero.
 * @ctx: [in] Pointer to eventfd context.
 * @no_wait: [in] Different from zero if the operation should not block.
 * @cnt: [out] Pointer to the 64-bit counter value.
 *
 * Returns %0 if successful, or the following error codes:
 *
 * -EAGAIN      : The operation would have blocked but @no_wait was non-zero.
 * -ERESTARTSYS : A signal interrupted the wait operation.
 *
 * If @no_wait is zero, the function might sleep until the eventfd internal
 * counter becomes greater than zero.
 */
ssize_t eventfd_ctx_read(struct eventfd_ctx *ctx, int no_wait, __u64 *cnt)
{
	ssize_t res;
	DECLARE_WAITQUEUE(wait, current);

	spin_lock_irq(&ctx->wqh.lock);
	*cnt = 0;
	res = -EAGAIN;
	if (ctx->count > 0)
		res = 0;
	else if (!no_wait) {
		__add_wait_queue(&ctx->wqh, &wait);
		for (;;) {
			set_current_state(TASK_INTERRUPTIBLE);
			if (ctx->count > 0) {
				res = 0;
				break;
			}
			if (signal_pending(current)) {
				res = -ERESTARTSYS;
				break;
			}
			spin_unlock_irq(&ctx->wqh.lock);
			schedule();
			spin_lock_irq(&ctx->wqh.lock);
		}
		__remove_wait_queue(&ctx->wqh, &wait);
		__set_current_state(TASK_RUNNING);
	}
	if (likely(res == 0)) {
		eventfd_ctx_do_read(ctx, cnt);
		if (waitqueue_active(&ctx->wqh))
			wake_up_locked_poll(&ctx->wqh, POLLOUT);
	}
	spin_unlock_irq(&ctx->wqh.lock);

	return res;
}
EXPORT_SYMBOL_GPL(eventfd_ctx_read);

static ssize_t eventfd_read(struct file *file, char __user *buf, size_t count,
			    loff_t *ppos)
{
	struct eventfd_ctx *ctx = file->private_data;
	ssize_t res;
	__u64 cnt;

	if (count < sizeof(cnt))
		return -EINVAL;
	res = eventfd_ctx_read(ctx, file->f_flags & O_NONBLOCK, &cnt);
	if (res < 0)
		return res;

	return put_user(cnt, (__u64 __user *) buf) ? -EFAULT : sizeof(cnt);
}

static ssize_t eventfd_write(struct file *file, const char __user *buf, size_t count,
			     loff_t *ppos)
{
	struct eventfd_ctx *ctx = file->private_data;
	ssize_t res;
	__u64 ucnt;
	DECLARE_WAITQUEUE(wait, current);

	if (count < sizeof(ucnt))
		return -EINVAL;
	if (copy_from_user(&ucnt, buf, sizeof(ucnt)))
		return -EFAULT;
	if (ucnt == ULLONG_MAX)
		return -EINVAL;
	spin_lock_irq(&ctx->wqh.lock);
	res = -EAGAIN;
	if (ULLONG_MAX - ctx->count > ucnt)
		res = sizeof(ucnt);
	else if (!(file->f_flags & O_NONBLOCK)) {
		__add_wait_queue(&ctx->wqh, &wait);
		for (res = 0;;) {
			set_current_state(TASK_INTERRUPTIBLE);
			if (ULLONG_MAX - ctx->count > ucnt) {
				res = sizeof(ucnt);
				break;
			}
			if (signal_pending(current)) {
				res = -ERESTARTSYS;
				break;
			}
			spin_unlock_irq(&ctx->wqh.lock);
			schedule();
			spin_lock_irq(&ctx->wqh.lock);
		}
		__remove_wait_queue(&ctx->wqh, &wait);
		__set_current_state(TASK_RUNNING);
	}
	if (likely(res > 0)) {
		ctx->count += ucnt;
		if (waitqueue_active(&ctx->wqh))
			wake_up_locked_poll(&ctx->wqh, POLLIN);
	}
	spin_unlock_irq(&ctx->wqh.lock);

	return res;
}

#ifdef CONFIG_PROC_FS
static void eventfd_show_fdinfo(struct seq_file *m, struct file *f)
{
	struct eventfd_ctx *ctx = f->private_data;

	spin_lock_irq(&ctx->wqh.lock);
	seq_printf(m, "eventfd-count: %16llx\n",
		   (unsigned long long)ctx->count);
	spin_unlock_irq(&ctx->wqh.lock);
}
#endif

static const struct file_operations eventfd_fops = {
#ifdef CONFIG_PROC_FS
	.show_fdinfo	= eventfd_show_fdinfo,
#endif
	.release	= eventfd_release,
	.poll		= eventfd_poll,
	.read		= eventfd_read,
	.write		= eventfd_write,
	.llseek		= noop_llseek,
};

/**
 * eventfd_fget - Acquire a reference of an eventfd file descriptor.
 * @fd: [in] Eventfd file descriptor.
 *
 * Returns a pointer to the eventfd file structure in case of success, or the
 * following error pointer:
 *
 * -EBADF    : Invalid @fd file descriptor.
 * -EINVAL   : The @fd file descriptor is not an eventfd file.
 */
struct file *eventfd_fget(int fd)
{
	struct file *file;

	file = fget(fd);
	if (!file)
		return ERR_PTR(-EBADF);
	if (file->f_op != &eventfd_fops) {
		fput(file);
		return ERR_PTR(-EINVAL);
	}

	return file;
}
EXPORT_SYMBOL_GPL(eventfd_fget);

/**
 * eventfd_ctx_fdget - Acquires a reference to the internal eventfd context.
 * @fd: [in] Eventfd file descriptor.
 *
 * Returns a pointer to the internal eventfd context, otherwise the error
 * pointers returned by the following functions:
 *
 * eventfd_fget
 */
struct eventfd_ctx *eventfd_ctx_fdget(int fd)
{
	struct eventfd_ctx *ctx;
	struct fd f = fdget(fd);
	if (!f.file)
		return ERR_PTR(-EBADF);
	ctx = eventfd_ctx_fileget(f.file);
	fdput(f);
	return ctx;
}
EXPORT_SYMBOL_GPL(eventfd_ctx_fdget);

/**
 * eventfd_ctx_fileget - Acquires a reference to the internal eventfd context.
 * @file: [in] Eventfd file pointer.
 *
 * Returns a pointer to the internal eventfd context, otherwise the error
 * pointer:
 *
 * -EINVAL   : The @fd file descriptor is not an eventfd file.
 */
struct eventfd_ctx *eventfd_ctx_fileget(struct file *file)
{
	if (file->f_op != &eventfd_fops)
		return ERR_PTR(-EINVAL);

	return eventfd_ctx_get(file->private_data);
}
EXPORT_SYMBOL_GPL(eventfd_ctx_fileget);

/**
 * eventfd_file_create - Creates an eventfd file pointer.
 * @count: Initial eventfd counter value.
 * @flags: Flags for the eventfd file.
 *
 * This function creates an eventfd file pointer, w/out installing it into
 * the fd table. This is useful when the eventfd file is used during the
 * initialization of data structures that require extra setup after the eventfd
 * creation. So the eventfd creation is split into the file pointer creation
 * phase, and the file descriptor installation phase.
 * In this way races with userspace closing the newly installed file descriptor
 * can be avoided.
 * Returns an eventfd file pointer, or a proper error pointer.
 */
struct file *eventfd_file_create(unsigned int count, int flags)
{
	struct file *file;
	struct eventfd_ctx *ctx;

	/* Check the EFD_* constants for consistency.  */
	BUILD_BUG_ON(EFD_CLOEXEC != O_CLOEXEC);
	BUILD_BUG_ON(EFD_NONBLOCK != O_NONBLOCK);

	if (flags & ~EFD_FLAGS_SET)
		return ERR_PTR(-EINVAL);

	ctx = kmalloc(sizeof(*ctx), GFP_KERNEL);
	if (!ctx)
		return ERR_PTR(-ENOMEM);

	kref_init(&ctx->kref);
	init_waitqueue_head(&ctx->wqh);
	ctx->count = count;
	ctx->flags = flags;

	file = anon_inode_getfile("[eventfd]", &eventfd_fops, ctx,
				  O_RDWR | (flags & EFD_SHARED_FCNTL_FLAGS));
	if (IS_ERR(file))
		eventfd_free_ctx(ctx);

	return file;
}

SYSCALL_DEFINE2(eventfd2, unsigned int, count, int, flags)
{
	int fd, error;
	struct file *file;

	error = get_unused_fd_flags(flags & EFD_SHARED_FCNTL_FLAGS);
	if (error < 0)
		return error;
	fd = error;

	file = eventfd_file_create(count, flags);
	if (IS_ERR(file)) {
		error = PTR_ERR(file);
		goto err_put_unused_fd;
	}
	fd_install(fd, file);

	return fd;

err_put_unused_fd:
	put_unused_fd(fd);

	return error;
}

SYSCALL_DEFINE1(eventfd, unsigned int, count)
{
	return sys_eventfd2(count, 0);
}
Commit	Line	Data
e1ad7468 DL	1	/*
	2	* fs/eventfd.c
	3	*
	4	* Copyright (C) 2007 Davide Libenzi <davidel@xmailserver.org>
	5	*
	6	*/
	7
	8	#include <linux/file.h>
	9	#include <linux/poll.h>
	10	#include <linux/init.h>
	11	#include <linux/fs.h>
	12	#include <linux/sched.h>
	13	#include <linux/kernel.h>
5a0e3ad6	14	#include <linux/slab.h>
e1ad7468 DL	15	#include <linux/list.h>
	16	#include <linux/spinlock.h>
	17	#include <linux/anon_inodes.h>
7747cdb2	18	#include <linux/syscalls.h>
630d9c47	19	#include <linux/export.h>
13389010 DL	20	#include <linux/kref.h>
13389010 DL	21	#include <linux/eventfd.h>
cbac5542 CG	22	#include <linux/proc_fs.h>
cbac5542 CG	23	#include <linux/seq_file.h>
e1ad7468 DL	24
e1ad7468 DL	25	struct eventfd_ctx {
13389010	26	struct kref kref;
e1ad7468 DL	27	wait_queue_head_t wqh;
	28	/*
	29	* Every time that a write(2) is performed on an eventfd, the
	30	* value of the __u64 being written is added to "count" and a
	31	* wakeup is performed on "wqh". A read(2) will return the "count"
	32	* value to userspace, and will reset "count" to zero. The kernel
13389010	33	* side eventfd_signal() also, adds to the "count" counter and
e1ad7468 DL	34	* issue a wakeup.
	35	*/
	36	__u64 count;
bcd0b235	37	unsigned int flags;
e1ad7468 DL	38	};
e1ad7468 DL	39
13389010 DL	40	/**
	41	* eventfd_signal - Adds @n to the eventfd counter.
	42	* @ctx: [in] Pointer to the eventfd context.
	43	* @n: [in] Value of the counter to be added to the eventfd internal counter.
	44	* The value cannot be negative.
	45	*
	46	* This function is supposed to be called by the kernel in paths that do not
	47	* allow sleeping. In this function we allow the counter to reach the ULLONG_MAX
20d5a865	48	* value, and we signal this as overflow condition by returning a POLLERR
13389010 DL	49	* to poll(2).
13389010 DL	50	*
20d5a865	51	* Returns the amount by which the counter was incremented. This will be less
ee62c6b2	52	* than @n if the counter has overflowed.
e1ad7468	53	*/
ee62c6b2	54	__u64 eventfd_signal(struct eventfd_ctx *ctx, __u64 n)
e1ad7468	55	{
e1ad7468 DL	56	unsigned long flags;
e1ad7468 DL	57
d48eb233	58	spin_lock_irqsave(&ctx->wqh.lock, flags);
e1ad7468	59	if (ULLONG_MAX - ctx->count < n)
ee62c6b2	60	n = ULLONG_MAX - ctx->count;
e1ad7468 DL	61	ctx->count += n;
e1ad7468 DL	62	if (waitqueue_active(&ctx->wqh))
39510888	63	wake_up_locked_poll(&ctx->wqh, POLLIN);
d48eb233	64	spin_unlock_irqrestore(&ctx->wqh.lock, flags);
e1ad7468 DL	65
	66	return n;
	67	}
5718607b	68	EXPORT_SYMBOL_GPL(eventfd_signal);
e1ad7468	69
562787a5 DL	70	static void eventfd_free_ctx(struct eventfd_ctx *ctx)
	71	{
	72	kfree(ctx);
	73	}
	74
13389010 DL	75	static void eventfd_free(struct kref *kref)
	76	{
	77	struct eventfd_ctx *ctx = container_of(kref, struct eventfd_ctx, kref);
	78
562787a5	79	eventfd_free_ctx(ctx);
13389010 DL	80	}
	81
	82	/**
	83	* eventfd_ctx_get - Acquires a reference to the internal eventfd context.
	84	* @ctx: [in] Pointer to the eventfd context.
	85	*
	86	* Returns: In case of success, returns a pointer to the eventfd context.
	87	*/
	88	struct eventfd_ctx eventfd_ctx_get(struct eventfd_ctx ctx)
	89	{
	90	kref_get(&ctx->kref);
	91	return ctx;
	92	}
	93	EXPORT_SYMBOL_GPL(eventfd_ctx_get);
	94
	95	/**
	96	* eventfd_ctx_put - Releases a reference to the internal eventfd context.
	97	* @ctx: [in] Pointer to eventfd context.
	98	*
	99	* The eventfd context reference must have been previously acquired either
36182185	100	* with eventfd_ctx_get() or eventfd_ctx_fdget().
13389010 DL	101	*/
	102	void eventfd_ctx_put(struct eventfd_ctx *ctx)
	103	{
	104	kref_put(&ctx->kref, eventfd_free);
	105	}
	106	EXPORT_SYMBOL_GPL(eventfd_ctx_put);
	107
e1ad7468 DL	108	static int eventfd_release(struct inode inode, struct file file)
e1ad7468 DL	109	{
13389010 DL	110	struct eventfd_ctx *ctx = file->private_data;
	111
	112	wake_up_poll(&ctx->wqh, POLLHUP);
	113	eventfd_ctx_put(ctx);
e1ad7468 DL	114	return 0;
	115	}
	116
	117	static unsigned int eventfd_poll(struct file file, poll_table wait)
	118	{
	119	struct eventfd_ctx *ctx = file->private_data;
	120	unsigned int events = 0;
e22553e2	121	u64 count;
e1ad7468 DL	122
e1ad7468 DL	123	poll_wait(file, &ctx->wqh, wait);
a484c3dd PB	124
	125	/*
	126	* All writes to ctx->count occur within ctx->wqh.lock. This read
	127	* can be done outside ctx->wqh.lock because we know that poll_wait
	128	* takes that lock (through add_wait_queue) if our caller will sleep.
	129	*
	130	* The read _can_ therefore seep into add_wait_queue's critical
	131	* section, but cannot move above it! add_wait_queue's spin_lock acts
	132	* as an acquire barrier and ensures that the read be ordered properly
	133	* against the writes. The following CAN happen and is safe:
	134	*
	135	* poll write
	136	* ----------------- ------------
	137	* lock ctx->wqh.lock (in poll_wait)
	138	* count = ctx->count
	139	* __add_wait_queue
	140	* unlock ctx->wqh.lock
	141	* lock ctx->qwh.lock
	142	* ctx->count += n
	143	* if (waitqueue_active)
	144	* wake_up_locked_poll
	145	* unlock ctx->qwh.lock
	146	* eventfd_poll returns 0
	147	*
	148	* but the following, which would miss a wakeup, cannot happen:
	149	*
	150	* poll write
	151	* ----------------- ------------
	152	* count = ctx->count (INVALID!)
	153	* lock ctx->qwh.lock
	154	* ctx->count += n
	155	* waitqueue_active is false
	156	* no wake_up_locked_poll!
	157	* unlock ctx->qwh.lock
	158	* lock ctx->wqh.lock (in poll_wait)
	159	* __add_wait_queue
	160	* unlock ctx->wqh.lock
	161	* eventfd_poll returns 0
	162	*/
	163	count = READ_ONCE(ctx->count);
e1ad7468	164
e22553e2	165	if (count > 0)
e1ad7468	166	events \|= POLLIN;
e22553e2	167	if (count == ULLONG_MAX)
e1ad7468	168	events \|= POLLERR;
e22553e2	169	if (ULLONG_MAX - 1 > count)
e1ad7468	170	events \|= POLLOUT;
e1ad7468 DL	171
	172	return events;
	173	}
	174
cb289d62 DL	175	static void eventfd_ctx_do_read(struct eventfd_ctx ctx, __u64 cnt)
	176	{
	177	*cnt = (ctx->flags & EFD_SEMAPHORE) ? 1 : ctx->count;
	178	ctx->count -= *cnt;
	179	}
	180
	181	/**
	182	* eventfd_ctx_remove_wait_queue - Read the current counter and removes wait queue.
	183	* @ctx: [in] Pointer to eventfd context.
	184	* @wait: [in] Wait queue to be removed.
36182185	185	* @cnt: [out] Pointer to the 64-bit counter value.
cb289d62	186	*
36182185	187	* Returns %0 if successful, or the following error codes:
cb289d62 DL	188	*
	189	* -EAGAIN : The operation would have blocked.
	190	*
	191	* This is used to atomically remove a wait queue entry from the eventfd wait
	192	* queue head, and read/reset the counter value.
	193	*/
	194	int eventfd_ctx_remove_wait_queue(struct eventfd_ctx ctx, wait_queue_t wait,
	195	__u64 *cnt)
	196	{
	197	unsigned long flags;
	198
	199	spin_lock_irqsave(&ctx->wqh.lock, flags);
	200	eventfd_ctx_do_read(ctx, cnt);
	201	__remove_wait_queue(&ctx->wqh, wait);
	202	if (*cnt != 0 && waitqueue_active(&ctx->wqh))
	203	wake_up_locked_poll(&ctx->wqh, POLLOUT);
	204	spin_unlock_irqrestore(&ctx->wqh.lock, flags);
	205
	206	return *cnt != 0 ? 0 : -EAGAIN;
	207	}
	208	EXPORT_SYMBOL_GPL(eventfd_ctx_remove_wait_queue);
	209
	210	/**
	211	* eventfd_ctx_read - Reads the eventfd counter or wait if it is zero.
	212	* @ctx: [in] Pointer to eventfd context.
	213	* @no_wait: [in] Different from zero if the operation should not block.
36182185	214	* @cnt: [out] Pointer to the 64-bit counter value.
cb289d62	215	*
36182185	216	* Returns %0 if successful, or the following error codes:
cb289d62	217	*
36182185	218	* -EAGAIN : The operation would have blocked but @no_wait was non-zero.
cb289d62 DL	219	* -ERESTARTSYS : A signal interrupted the wait operation.
	220	*
	221	* If @no_wait is zero, the function might sleep until the eventfd internal
	222	* counter becomes greater than zero.
	223	*/
	224	ssize_t eventfd_ctx_read(struct eventfd_ctx ctx, int no_wait, __u64 cnt)
e1ad7468	225	{
e1ad7468	226	ssize_t res;
e1ad7468 DL	227	DECLARE_WAITQUEUE(wait, current);
e1ad7468 DL	228
d48eb233	229	spin_lock_irq(&ctx->wqh.lock);
cb289d62	230	*cnt = 0;
e1ad7468	231	res = -EAGAIN;
bcd0b235	232	if (ctx->count > 0)
cb289d62 DL	233	res = 0;
cb289d62 DL	234	else if (!no_wait) {
e1ad7468	235	__add_wait_queue(&ctx->wqh, &wait);
cb289d62	236	for (;;) {
e1ad7468 DL	237	set_current_state(TASK_INTERRUPTIBLE);
e1ad7468 DL	238	if (ctx->count > 0) {
cb289d62	239	res = 0;
e1ad7468 DL	240	break;
	241	}
	242	if (signal_pending(current)) {
	243	res = -ERESTARTSYS;
	244	break;
	245	}
d48eb233	246	spin_unlock_irq(&ctx->wqh.lock);
e1ad7468	247	schedule();
d48eb233	248	spin_lock_irq(&ctx->wqh.lock);
e1ad7468 DL	249	}
	250	__remove_wait_queue(&ctx->wqh, &wait);
	251	__set_current_state(TASK_RUNNING);
	252	}
cb289d62 DL	253	if (likely(res == 0)) {
cb289d62 DL	254	eventfd_ctx_do_read(ctx, cnt);
e1ad7468	255	if (waitqueue_active(&ctx->wqh))
39510888	256	wake_up_locked_poll(&ctx->wqh, POLLOUT);
e1ad7468	257	}
d48eb233	258	spin_unlock_irq(&ctx->wqh.lock);
e1ad7468 DL	259
	260	return res;
	261	}
cb289d62 DL	262	EXPORT_SYMBOL_GPL(eventfd_ctx_read);
	263
	264	static ssize_t eventfd_read(struct file file, char __user buf, size_t count,
	265	loff_t *ppos)
	266	{
	267	struct eventfd_ctx *ctx = file->private_data;
	268	ssize_t res;
	269	__u64 cnt;
	270
	271	if (count < sizeof(cnt))
	272	return -EINVAL;
	273	res = eventfd_ctx_read(ctx, file->f_flags & O_NONBLOCK, &cnt);
	274	if (res < 0)
	275	return res;
	276
	277	return put_user(cnt, (__u64 __user *) buf) ? -EFAULT : sizeof(cnt);
	278	}
e1ad7468 DL	279
	280	static ssize_t eventfd_write(struct file file, const char __user buf, size_t count,
	281	loff_t *ppos)
	282	{
	283	struct eventfd_ctx *ctx = file->private_data;
	284	ssize_t res;
	285	__u64 ucnt;
	286	DECLARE_WAITQUEUE(wait, current);
	287
	288	if (count < sizeof(ucnt))
	289	return -EINVAL;
	290	if (copy_from_user(&ucnt, buf, sizeof(ucnt)))
	291	return -EFAULT;
	292	if (ucnt == ULLONG_MAX)
	293	return -EINVAL;
d48eb233	294	spin_lock_irq(&ctx->wqh.lock);
e1ad7468 DL	295	res = -EAGAIN;
	296	if (ULLONG_MAX - ctx->count > ucnt)
	297	res = sizeof(ucnt);
	298	else if (!(file->f_flags & O_NONBLOCK)) {
	299	__add_wait_queue(&ctx->wqh, &wait);
	300	for (res = 0;;) {
	301	set_current_state(TASK_INTERRUPTIBLE);
	302	if (ULLONG_MAX - ctx->count > ucnt) {
	303	res = sizeof(ucnt);
	304	break;
	305	}
	306	if (signal_pending(current)) {
	307	res = -ERESTARTSYS;
	308	break;
	309	}
d48eb233	310	spin_unlock_irq(&ctx->wqh.lock);
e1ad7468	311	schedule();
d48eb233	312	spin_lock_irq(&ctx->wqh.lock);
e1ad7468 DL	313	}
	314	__remove_wait_queue(&ctx->wqh, &wait);
	315	__set_current_state(TASK_RUNNING);
	316	}
bcd0b235	317	if (likely(res > 0)) {
e1ad7468 DL	318	ctx->count += ucnt;
e1ad7468 DL	319	if (waitqueue_active(&ctx->wqh))
39510888	320	wake_up_locked_poll(&ctx->wqh, POLLIN);
e1ad7468	321	}
d48eb233	322	spin_unlock_irq(&ctx->wqh.lock);
e1ad7468 DL	323
	324	return res;
	325	}
	326
cbac5542	327	#ifdef CONFIG_PROC_FS
a3816ab0	328	static void eventfd_show_fdinfo(struct seq_file m, struct file f)
cbac5542 CG	329	{
cbac5542 CG	330	struct eventfd_ctx *ctx = f->private_data;
cbac5542 CG	331
cbac5542 CG	332	spin_lock_irq(&ctx->wqh.lock);
a3816ab0 JP	333	seq_printf(m, "eventfd-count: %16llx\n",
a3816ab0 JP	334	(unsigned long long)ctx->count);
cbac5542	335	spin_unlock_irq(&ctx->wqh.lock);
cbac5542 CG	336	}
	337	#endif
	338
e1ad7468	339	static const struct file_operations eventfd_fops = {
cbac5542 CG	340	#ifdef CONFIG_PROC_FS
	341	.show_fdinfo = eventfd_show_fdinfo,
	342	#endif
e1ad7468 DL	343	.release = eventfd_release,
	344	.poll = eventfd_poll,
	345	.read = eventfd_read,
	346	.write = eventfd_write,
6038f373	347	.llseek = noop_llseek,
e1ad7468 DL	348	};
e1ad7468 DL	349
13389010 DL	350	/**
	351	* eventfd_fget - Acquire a reference of an eventfd file descriptor.
	352	* @fd: [in] Eventfd file descriptor.
	353	*
	354	* Returns a pointer to the eventfd file structure in case of success, or the
	355	* following error pointer:
	356	*
	357	* -EBADF : Invalid @fd file descriptor.
	358	* -EINVAL : The @fd file descriptor is not an eventfd file.
	359	*/
e1ad7468 DL	360	struct file *eventfd_fget(int fd)
	361	{
	362	struct file *file;
	363
	364	file = fget(fd);
	365	if (!file)
	366	return ERR_PTR(-EBADF);
	367	if (file->f_op != &eventfd_fops) {
	368	fput(file);
	369	return ERR_PTR(-EINVAL);
	370	}
	371
	372	return file;
	373	}
5718607b	374	EXPORT_SYMBOL_GPL(eventfd_fget);
e1ad7468	375
13389010 DL	376	/**
	377	* eventfd_ctx_fdget - Acquires a reference to the internal eventfd context.
	378	* @fd: [in] Eventfd file descriptor.
	379	*
	380	* Returns a pointer to the internal eventfd context, otherwise the error
	381	* pointers returned by the following functions:
	382	*
	383	* eventfd_fget
	384	*/
	385	struct eventfd_ctx *eventfd_ctx_fdget(int fd)
	386	{
13389010	387	struct eventfd_ctx *ctx;
36a74117 AV	388	struct fd f = fdget(fd);
	389	if (!f.file)
	390	return ERR_PTR(-EBADF);
	391	ctx = eventfd_ctx_fileget(f.file);
	392	fdput(f);
13389010 DL	393	return ctx;
	394	}
	395	EXPORT_SYMBOL_GPL(eventfd_ctx_fdget);
	396
	397	/**
	398	* eventfd_ctx_fileget - Acquires a reference to the internal eventfd context.
	399	* @file: [in] Eventfd file pointer.
	400	*
	401	* Returns a pointer to the internal eventfd context, otherwise the error
	402	* pointer:
	403	*
	404	* -EINVAL : The @fd file descriptor is not an eventfd file.
	405	*/
	406	struct eventfd_ctx eventfd_ctx_fileget(struct file file)
	407	{
	408	if (file->f_op != &eventfd_fops)
	409	return ERR_PTR(-EINVAL);
	410
	411	return eventfd_ctx_get(file->private_data);
	412	}
	413	EXPORT_SYMBOL_GPL(eventfd_ctx_fileget);
	414
562787a5 DL	415	/**
	416	* eventfd_file_create - Creates an eventfd file pointer.
	417	* @count: Initial eventfd counter value.
	418	* @flags: Flags for the eventfd file.
	419	*
	420	* This function creates an eventfd file pointer, w/out installing it into
	421	* the fd table. This is useful when the eventfd file is used during the
	422	* initialization of data structures that require extra setup after the eventfd
	423	* creation. So the eventfd creation is split into the file pointer creation
	424	* phase, and the file descriptor installation phase.
	425	* In this way races with userspace closing the newly installed file descriptor
	426	* can be avoided.
	427	* Returns an eventfd file pointer, or a proper error pointer.
	428	*/
	429	struct file *eventfd_file_create(unsigned int count, int flags)
e1ad7468	430	{
562787a5	431	struct file *file;
e1ad7468	432	struct eventfd_ctx *ctx;
e1ad7468	433
e38b36f3 UD	434	/* Check the EFD_* constants for consistency. */
	435	BUILD_BUG_ON(EFD_CLOEXEC != O_CLOEXEC);
	436	BUILD_BUG_ON(EFD_NONBLOCK != O_NONBLOCK);
	437
bcd0b235	438	if (flags & ~EFD_FLAGS_SET)
562787a5	439	return ERR_PTR(-EINVAL);
b087498e	440
e1ad7468 DL	441	ctx = kmalloc(sizeof(*ctx), GFP_KERNEL);
e1ad7468 DL	442	if (!ctx)
562787a5	443	return ERR_PTR(-ENOMEM);
e1ad7468	444
13389010	445	kref_init(&ctx->kref);
e1ad7468	446	init_waitqueue_head(&ctx->wqh);
e1ad7468	447	ctx->count = count;
bcd0b235	448	ctx->flags = flags;
e1ad7468	449
562787a5	450	file = anon_inode_getfile("[eventfd]", &eventfd_fops, ctx,
628ff7c1	451	O_RDWR \| (flags & EFD_SHARED_FCNTL_FLAGS));
562787a5 DL	452	if (IS_ERR(file))
	453	eventfd_free_ctx(ctx);
	454
	455	return file;
	456	}
	457
	458	SYSCALL_DEFINE2(eventfd2, unsigned int, count, int, flags)
	459	{
	460	int fd, error;
	461	struct file *file;
	462
	463	error = get_unused_fd_flags(flags & EFD_SHARED_FCNTL_FLAGS);
	464	if (error < 0)
	465	return error;
	466	fd = error;
	467
	468	file = eventfd_file_create(count, flags);
	469	if (IS_ERR(file)) {
	470	error = PTR_ERR(file);
	471	goto err_put_unused_fd;
	472	}
	473	fd_install(fd, file);
	474
2030a42c	475	return fd;
562787a5 DL	476
	477	err_put_unused_fd:
	478	put_unused_fd(fd);
	479
	480	return error;
e1ad7468 DL	481	}
e1ad7468 DL	482
d4e82042	483	SYSCALL_DEFINE1(eventfd, unsigned int, count)
b087498e UD	484	{
	485	return sys_eventfd2(count, 0);
	486	}
bcd0b235	487