[linux.git] / fs / eventfd.c

// SPDX-License-Identifier: GPL-2.0-only
/*
 *  fs/eventfd.c
 *
 *  Copyright (C) 2007  Davide Libenzi <[email protected]>
 *
 */

#include <linux/file.h>
#include <linux/poll.h>
#include <linux/init.h>
#include <linux/fs.h>
#include <linux/sched/signal.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>
#include <linux/idr.h>
#include <linux/uio.h>

static DEFINE_IDA(eventfd_ida);

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;
	int id;
};

/**
 * 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 EPOLLERR
 * 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;

	/*
	 * Deadlock or stack overflow issues can happen if we recurse here
	 * through waitqueue wakeup handlers. If the caller users potentially
	 * nested waitqueues with custom wakeup handlers, then it should
	 * check eventfd_signal_allowed() before calling this function. If
	 * it returns false, the eventfd_signal() call should be deferred to a
	 * safe context.
	 */
	if (WARN_ON_ONCE(current->in_eventfd_signal))
		return 0;

	spin_lock_irqsave(&ctx->wqh.lock, flags);
	current->in_eventfd_signal = 1;
	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, EPOLLIN);
	current->in_eventfd_signal = 0;
	spin_unlock_irqrestore(&ctx->wqh.lock, flags);

	return n;
}
EXPORT_SYMBOL_GPL(eventfd_signal);

static void eventfd_free_ctx(struct eventfd_ctx *ctx)
{
	if (ctx->id >= 0)
		ida_simple_remove(&eventfd_ida, ctx->id);
	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_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_fdget() or eventfd_ctx_fileget().
 */
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, EPOLLHUP);
	eventfd_ctx_put(ctx);
	return 0;
}

static __poll_t eventfd_poll(struct file *file, poll_table *wait)
{
	struct eventfd_ctx *ctx = file->private_data;
	__poll_t 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 |= EPOLLIN;
	if (count == ULLONG_MAX)
		events |= EPOLLERR;
	if (ULLONG_MAX - 1 > count)
		events |= EPOLLOUT;

	return events;
}

void eventfd_ctx_do_read(struct eventfd_ctx *ctx, __u64 *cnt)
{
	lockdep_assert_held(&ctx->wqh.lock);

	*cnt = (ctx->flags & EFD_SEMAPHORE) ? 1 : ctx->count;
	ctx->count -= *cnt;
}
EXPORT_SYMBOL_GPL(eventfd_ctx_do_read);

/**
 * 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_entry_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, EPOLLOUT);
	spin_unlock_irqrestore(&ctx->wqh.lock, flags);

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

static ssize_t eventfd_read(struct kiocb *iocb, struct iov_iter *to)
{
	struct file *file = iocb->ki_filp;
	struct eventfd_ctx *ctx = file->private_data;
	__u64 ucnt = 0;
	DECLARE_WAITQUEUE(wait, current);

	if (iov_iter_count(to) < sizeof(ucnt))
		return -EINVAL;
	spin_lock_irq(&ctx->wqh.lock);
	if (!ctx->count) {
		if ((file->f_flags & O_NONBLOCK) ||
		    (iocb->ki_flags & IOCB_NOWAIT)) {
			spin_unlock_irq(&ctx->wqh.lock);
			return -EAGAIN;
		}
		__add_wait_queue(&ctx->wqh, &wait);
		for (;;) {
			set_current_state(TASK_INTERRUPTIBLE);
			if (ctx->count)
				break;
			if (signal_pending(current)) {
				__remove_wait_queue(&ctx->wqh, &wait);
				__set_current_state(TASK_RUNNING);
				spin_unlock_irq(&ctx->wqh.lock);
				return -ERESTARTSYS;
			}
			spin_unlock_irq(&ctx->wqh.lock);
			schedule();
			spin_lock_irq(&ctx->wqh.lock);
		}
		__remove_wait_queue(&ctx->wqh, &wait);
		__set_current_state(TASK_RUNNING);
	}
	eventfd_ctx_do_read(ctx, &ucnt);
	if (waitqueue_active(&ctx->wqh))
		wake_up_locked_poll(&ctx->wqh, EPOLLOUT);
	spin_unlock_irq(&ctx->wqh.lock);
	if (unlikely(copy_to_iter(&ucnt, sizeof(ucnt), to) != sizeof(ucnt)))
		return -EFAULT;

	return sizeof(ucnt);
}

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, EPOLLIN);
	}
	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);
	seq_printf(m, "eventfd-id: %d\n", ctx->id);
}
#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_iter	= 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)
{
	struct eventfd_ctx *ctx;

	if (file->f_op != &eventfd_fops)
		return ERR_PTR(-EINVAL);

	ctx = file->private_data;
	kref_get(&ctx->kref);
	return ctx;
}
EXPORT_SYMBOL_GPL(eventfd_ctx_fileget);

static int do_eventfd(unsigned int count, int flags)
{
	struct eventfd_ctx *ctx;
	struct file *file;
	int fd;

	/* 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 -EINVAL;

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

	kref_init(&ctx->kref);
	init_waitqueue_head(&ctx->wqh);
	ctx->count = count;
	ctx->flags = flags;
	ctx->id = ida_simple_get(&eventfd_ida, 0, 0, GFP_KERNEL);

	flags &= EFD_SHARED_FCNTL_FLAGS;
	flags |= O_RDWR;
	fd = get_unused_fd_flags(flags);
	if (fd < 0)
		goto err;

	file = anon_inode_getfile("[eventfd]", &eventfd_fops, ctx, flags);
	if (IS_ERR(file)) {
		put_unused_fd(fd);
		fd = PTR_ERR(file);
		goto err;
	}

	file->f_mode |= FMODE_NOWAIT;
	fd_install(fd, file);
	return fd;
err:
	eventfd_free_ctx(ctx);
	return fd;
}

SYSCALL_DEFINE2(eventfd2, unsigned int, count, int, flags)
{
	return do_eventfd(count, flags);
}

SYSCALL_DEFINE1(eventfd, unsigned int, count)
{
	return do_eventfd(count, 0);
}
Commit	Line	Data
457c8996	1	// SPDX-License-Identifier: GPL-2.0-only
e1ad7468 DL	2	/*
	3	* fs/eventfd.c
	4	*
	5	* Copyright (C) 2007 Davide Libenzi <[email protected]>
	6	*
	7	*/
	8
	9	#include <linux/file.h>
	10	#include <linux/poll.h>
	11	#include <linux/init.h>
	12	#include <linux/fs.h>
174cd4b1	13	#include <linux/sched/signal.h>
e1ad7468	14	#include <linux/kernel.h>
5a0e3ad6	15	#include <linux/slab.h>
e1ad7468 DL	16	#include <linux/list.h>
	17	#include <linux/spinlock.h>
	18	#include <linux/anon_inodes.h>
7747cdb2	19	#include <linux/syscalls.h>
630d9c47	20	#include <linux/export.h>
13389010 DL	21	#include <linux/kref.h>
13389010 DL	22	#include <linux/eventfd.h>
cbac5542 CG	23	#include <linux/proc_fs.h>
cbac5542 CG	24	#include <linux/seq_file.h>
b556db17	25	#include <linux/idr.h>
12aceb89	26	#include <linux/uio.h>
b556db17	27
ce528c4c	28	static DEFINE_IDA(eventfd_ida);
e1ad7468 DL	29
e1ad7468 DL	30	struct eventfd_ctx {
13389010	31	struct kref kref;
e1ad7468 DL	32	wait_queue_head_t wqh;
	33	/*
	34	* Every time that a write(2) is performed on an eventfd, the
	35	* value of the __u64 being written is added to "count" and a
	36	* wakeup is performed on "wqh". A read(2) will return the "count"
	37	* value to userspace, and will reset "count" to zero. The kernel
13389010	38	* side eventfd_signal() also, adds to the "count" counter and
e1ad7468 DL	39	* issue a wakeup.
	40	*/
	41	__u64 count;
bcd0b235	42	unsigned int flags;
b556db17	43	int id;
e1ad7468 DL	44	};
e1ad7468 DL	45
13389010 DL	46	/**
	47	* eventfd_signal - Adds @n to the eventfd counter.
	48	* @ctx: [in] Pointer to the eventfd context.
	49	* @n: [in] Value of the counter to be added to the eventfd internal counter.
	50	* The value cannot be negative.
	51	*
	52	* This function is supposed to be called by the kernel in paths that do not
	53	* allow sleeping. In this function we allow the counter to reach the ULLONG_MAX
a9a08845	54	* value, and we signal this as overflow condition by returning a EPOLLERR
13389010 DL	55	* to poll(2).
13389010 DL	56	*
20d5a865	57	* Returns the amount by which the counter was incremented. This will be less
ee62c6b2	58	* than @n if the counter has overflowed.
e1ad7468	59	*/
ee62c6b2	60	__u64 eventfd_signal(struct eventfd_ctx *ctx, __u64 n)
e1ad7468	61	{
e1ad7468 DL	62	unsigned long flags;
e1ad7468 DL	63
b5e683d5 JA	64	/*
	65	* Deadlock or stack overflow issues can happen if we recurse here
	66	* through waitqueue wakeup handlers. If the caller users potentially
	67	* nested waitqueues with custom wakeup handlers, then it should
b542e383 TG	68	* check eventfd_signal_allowed() before calling this function. If
b542e383 TG	69	* it returns false, the eventfd_signal() call should be deferred to a
b5e683d5 JA	70	* safe context.
b5e683d5 JA	71	*/
b542e383	72	if (WARN_ON_ONCE(current->in_eventfd_signal))
b5e683d5 JA	73	return 0;
b5e683d5 JA	74
d48eb233	75	spin_lock_irqsave(&ctx->wqh.lock, flags);
b542e383	76	current->in_eventfd_signal = 1;
e1ad7468	77	if (ULLONG_MAX - ctx->count < n)
ee62c6b2	78	n = ULLONG_MAX - ctx->count;
e1ad7468 DL	79	ctx->count += n;
e1ad7468 DL	80	if (waitqueue_active(&ctx->wqh))
a9a08845	81	wake_up_locked_poll(&ctx->wqh, EPOLLIN);
b542e383	82	current->in_eventfd_signal = 0;
d48eb233	83	spin_unlock_irqrestore(&ctx->wqh.lock, flags);
e1ad7468 DL	84
	85	return n;
	86	}
5718607b	87	EXPORT_SYMBOL_GPL(eventfd_signal);
e1ad7468	88
562787a5 DL	89	static void eventfd_free_ctx(struct eventfd_ctx *ctx)
562787a5 DL	90	{
b556db17 MY	91	if (ctx->id >= 0)
b556db17 MY	92	ida_simple_remove(&eventfd_ida, ctx->id);
562787a5 DL	93	kfree(ctx);
	94	}
	95
13389010 DL	96	static void eventfd_free(struct kref *kref)
	97	{
	98	struct eventfd_ctx *ctx = container_of(kref, struct eventfd_ctx, kref);
	99
562787a5	100	eventfd_free_ctx(ctx);
13389010 DL	101	}
13389010 DL	102
13389010 DL	103	/**
	104	* eventfd_ctx_put - Releases a reference to the internal eventfd context.
	105	* @ctx: [in] Pointer to eventfd context.
	106	*
	107	* The eventfd context reference must have been previously acquired either
105f2b70	108	* with eventfd_ctx_fdget() or eventfd_ctx_fileget().
13389010 DL	109	*/
	110	void eventfd_ctx_put(struct eventfd_ctx *ctx)
	111	{
	112	kref_put(&ctx->kref, eventfd_free);
	113	}
	114	EXPORT_SYMBOL_GPL(eventfd_ctx_put);
	115
e1ad7468 DL	116	static int eventfd_release(struct inode inode, struct file file)
e1ad7468 DL	117	{
13389010 DL	118	struct eventfd_ctx *ctx = file->private_data;
13389010 DL	119
a9a08845	120	wake_up_poll(&ctx->wqh, EPOLLHUP);
13389010	121	eventfd_ctx_put(ctx);
e1ad7468 DL	122	return 0;
	123	}
	124
a11e1d43	125	static __poll_t eventfd_poll(struct file file, poll_table wait)
e1ad7468 DL	126	{
e1ad7468 DL	127	struct eventfd_ctx *ctx = file->private_data;
076ccb76	128	__poll_t events = 0;
e22553e2	129	u64 count;
e1ad7468	130
a11e1d43 LT	131	poll_wait(file, &ctx->wqh, wait);
a11e1d43 LT	132
a484c3dd PB	133	/*
	134	* All writes to ctx->count occur within ctx->wqh.lock. This read
	135	* can be done outside ctx->wqh.lock because we know that poll_wait
	136	* takes that lock (through add_wait_queue) if our caller will sleep.
	137	*
	138	* The read _can_ therefore seep into add_wait_queue's critical
	139	* section, but cannot move above it! add_wait_queue's spin_lock acts
	140	* as an acquire barrier and ensures that the read be ordered properly
	141	* against the writes. The following CAN happen and is safe:
	142	*
	143	* poll write
	144	* ----------------- ------------
	145	* lock ctx->wqh.lock (in poll_wait)
	146	* count = ctx->count
	147	* __add_wait_queue
	148	* unlock ctx->wqh.lock
	149	* lock ctx->qwh.lock
	150	* ctx->count += n
	151	* if (waitqueue_active)
	152	* wake_up_locked_poll
	153	* unlock ctx->qwh.lock
	154	* eventfd_poll returns 0
	155	*
	156	* but the following, which would miss a wakeup, cannot happen:
	157	*
	158	* poll write
	159	* ----------------- ------------
	160	* count = ctx->count (INVALID!)
	161	* lock ctx->qwh.lock
	162	* ctx->count += n
	163	* waitqueue_active is false
	164	* no wake_up_locked_poll!
	165	* unlock ctx->qwh.lock
	166	* lock ctx->wqh.lock (in poll_wait)
	167	* __add_wait_queue
	168	* unlock ctx->wqh.lock
	169	* eventfd_poll returns 0
	170	*/
	171	count = READ_ONCE(ctx->count);
e1ad7468	172
e22553e2	173	if (count > 0)
a11e1d43	174	events \|= EPOLLIN;
e22553e2	175	if (count == ULLONG_MAX)
a9a08845	176	events \|= EPOLLERR;
e22553e2	177	if (ULLONG_MAX - 1 > count)
a11e1d43	178	events \|= EPOLLOUT;
e1ad7468 DL	179
	180	return events;
	181	}
	182
28f13267	183	void eventfd_ctx_do_read(struct eventfd_ctx ctx, __u64 cnt)
cb289d62	184	{
28f13267 DW	185	lockdep_assert_held(&ctx->wqh.lock);
28f13267 DW	186
cb289d62 DL	187	*cnt = (ctx->flags & EFD_SEMAPHORE) ? 1 : ctx->count;
	188	ctx->count -= *cnt;
	189	}
28f13267	190	EXPORT_SYMBOL_GPL(eventfd_ctx_do_read);
cb289d62 DL	191
	192	/**
	193	* eventfd_ctx_remove_wait_queue - Read the current counter and removes wait queue.
	194	* @ctx: [in] Pointer to eventfd context.
	195	* @wait: [in] Wait queue to be removed.
36182185	196	* @cnt: [out] Pointer to the 64-bit counter value.
cb289d62	197	*
36182185	198	* Returns %0 if successful, or the following error codes:
cb289d62 DL	199	*
	200	* -EAGAIN : The operation would have blocked.
	201	*
	202	* This is used to atomically remove a wait queue entry from the eventfd wait
	203	* queue head, and read/reset the counter value.
	204	*/
ac6424b9	205	int eventfd_ctx_remove_wait_queue(struct eventfd_ctx ctx, wait_queue_entry_t wait,
cb289d62 DL	206	__u64 *cnt)
	207	{
	208	unsigned long flags;
	209
	210	spin_lock_irqsave(&ctx->wqh.lock, flags);
	211	eventfd_ctx_do_read(ctx, cnt);
	212	__remove_wait_queue(&ctx->wqh, wait);
	213	if (*cnt != 0 && waitqueue_active(&ctx->wqh))
a9a08845	214	wake_up_locked_poll(&ctx->wqh, EPOLLOUT);
cb289d62 DL	215	spin_unlock_irqrestore(&ctx->wqh.lock, flags);
	216
	217	return *cnt != 0 ? 0 : -EAGAIN;
	218	}
	219	EXPORT_SYMBOL_GPL(eventfd_ctx_remove_wait_queue);
	220
12aceb89	221	static ssize_t eventfd_read(struct kiocb iocb, struct iov_iter to)
e1ad7468	222	{
12aceb89	223	struct file *file = iocb->ki_filp;
b6364572	224	struct eventfd_ctx *ctx = file->private_data;
b6364572	225	__u64 ucnt = 0;
e1ad7468 DL	226	DECLARE_WAITQUEUE(wait, current);
e1ad7468 DL	227
12aceb89	228	if (iov_iter_count(to) < sizeof(ucnt))
b6364572	229	return -EINVAL;
d48eb233	230	spin_lock_irq(&ctx->wqh.lock);
12aceb89 JA	231	if (!ctx->count) {
	232	if ((file->f_flags & O_NONBLOCK) \|\|
	233	(iocb->ki_flags & IOCB_NOWAIT)) {
	234	spin_unlock_irq(&ctx->wqh.lock);
	235	return -EAGAIN;
	236	}
e1ad7468	237	__add_wait_queue(&ctx->wqh, &wait);
cb289d62	238	for (;;) {
e1ad7468	239	set_current_state(TASK_INTERRUPTIBLE);
12aceb89	240	if (ctx->count)
e1ad7468	241	break;
e1ad7468	242	if (signal_pending(current)) {
12aceb89 JA	243	__remove_wait_queue(&ctx->wqh, &wait);
	244	__set_current_state(TASK_RUNNING);
	245	spin_unlock_irq(&ctx->wqh.lock);
	246	return -ERESTARTSYS;
e1ad7468	247	}
d48eb233	248	spin_unlock_irq(&ctx->wqh.lock);
e1ad7468	249	schedule();
d48eb233	250	spin_lock_irq(&ctx->wqh.lock);
e1ad7468 DL	251	}
	252	__remove_wait_queue(&ctx->wqh, &wait);
	253	__set_current_state(TASK_RUNNING);
	254	}
12aceb89 JA	255	eventfd_ctx_do_read(ctx, &ucnt);
	256	if (waitqueue_active(&ctx->wqh))
	257	wake_up_locked_poll(&ctx->wqh, EPOLLOUT);
d48eb233	258	spin_unlock_irq(&ctx->wqh.lock);
12aceb89	259	if (unlikely(copy_to_iter(&ucnt, sizeof(ucnt), to) != sizeof(ucnt)))
b6364572	260	return -EFAULT;
cb289d62	261
12aceb89	262	return sizeof(ucnt);
cb289d62	263	}
e1ad7468 DL	264
	265	static ssize_t eventfd_write(struct file file, const char __user buf, size_t count,
	266	loff_t *ppos)
	267	{
	268	struct eventfd_ctx *ctx = file->private_data;
	269	ssize_t res;
	270	__u64 ucnt;
	271	DECLARE_WAITQUEUE(wait, current);
	272
	273	if (count < sizeof(ucnt))
	274	return -EINVAL;
	275	if (copy_from_user(&ucnt, buf, sizeof(ucnt)))
	276	return -EFAULT;
	277	if (ucnt == ULLONG_MAX)
	278	return -EINVAL;
d48eb233	279	spin_lock_irq(&ctx->wqh.lock);
e1ad7468 DL	280	res = -EAGAIN;
	281	if (ULLONG_MAX - ctx->count > ucnt)
	282	res = sizeof(ucnt);
	283	else if (!(file->f_flags & O_NONBLOCK)) {
	284	__add_wait_queue(&ctx->wqh, &wait);
	285	for (res = 0;;) {
	286	set_current_state(TASK_INTERRUPTIBLE);
	287	if (ULLONG_MAX - ctx->count > ucnt) {
	288	res = sizeof(ucnt);
	289	break;
	290	}
	291	if (signal_pending(current)) {
	292	res = -ERESTARTSYS;
	293	break;
	294	}
d48eb233	295	spin_unlock_irq(&ctx->wqh.lock);
e1ad7468	296	schedule();
d48eb233	297	spin_lock_irq(&ctx->wqh.lock);
e1ad7468 DL	298	}
	299	__remove_wait_queue(&ctx->wqh, &wait);
	300	__set_current_state(TASK_RUNNING);
	301	}
bcd0b235	302	if (likely(res > 0)) {
e1ad7468 DL	303	ctx->count += ucnt;
e1ad7468 DL	304	if (waitqueue_active(&ctx->wqh))
a9a08845	305	wake_up_locked_poll(&ctx->wqh, EPOLLIN);
e1ad7468	306	}
d48eb233	307	spin_unlock_irq(&ctx->wqh.lock);
e1ad7468 DL	308
	309	return res;
	310	}
	311
cbac5542	312	#ifdef CONFIG_PROC_FS
a3816ab0	313	static void eventfd_show_fdinfo(struct seq_file m, struct file f)
cbac5542 CG	314	{
cbac5542 CG	315	struct eventfd_ctx *ctx = f->private_data;
cbac5542 CG	316
cbac5542 CG	317	spin_lock_irq(&ctx->wqh.lock);
a3816ab0 JP	318	seq_printf(m, "eventfd-count: %16llx\n",
a3816ab0 JP	319	(unsigned long long)ctx->count);
cbac5542	320	spin_unlock_irq(&ctx->wqh.lock);
b556db17	321	seq_printf(m, "eventfd-id: %d\n", ctx->id);
cbac5542 CG	322	}
	323	#endif
	324
e1ad7468	325	static const struct file_operations eventfd_fops = {
cbac5542 CG	326	#ifdef CONFIG_PROC_FS
	327	.show_fdinfo = eventfd_show_fdinfo,
	328	#endif
e1ad7468	329	.release = eventfd_release,
a11e1d43	330	.poll = eventfd_poll,
12aceb89	331	.read_iter = eventfd_read,
e1ad7468	332	.write = eventfd_write,
6038f373	333	.llseek = noop_llseek,
e1ad7468 DL	334	};
e1ad7468 DL	335
13389010 DL	336	/**
	337	* eventfd_fget - Acquire a reference of an eventfd file descriptor.
	338	* @fd: [in] Eventfd file descriptor.
	339	*
	340	* Returns a pointer to the eventfd file structure in case of success, or the
	341	* following error pointer:
	342	*
	343	* -EBADF : Invalid @fd file descriptor.
	344	* -EINVAL : The @fd file descriptor is not an eventfd file.
	345	*/
e1ad7468 DL	346	struct file *eventfd_fget(int fd)
	347	{
	348	struct file *file;
	349
	350	file = fget(fd);
	351	if (!file)
	352	return ERR_PTR(-EBADF);
	353	if (file->f_op != &eventfd_fops) {
	354	fput(file);
	355	return ERR_PTR(-EINVAL);
	356	}
	357
	358	return file;
	359	}
5718607b	360	EXPORT_SYMBOL_GPL(eventfd_fget);
e1ad7468	361
13389010 DL	362	/**
	363	* eventfd_ctx_fdget - Acquires a reference to the internal eventfd context.
	364	* @fd: [in] Eventfd file descriptor.
	365	*
	366	* Returns a pointer to the internal eventfd context, otherwise the error
	367	* pointers returned by the following functions:
	368	*
	369	* eventfd_fget
	370	*/
	371	struct eventfd_ctx *eventfd_ctx_fdget(int fd)
	372	{
13389010	373	struct eventfd_ctx *ctx;
36a74117 AV	374	struct fd f = fdget(fd);
	375	if (!f.file)
	376	return ERR_PTR(-EBADF);
	377	ctx = eventfd_ctx_fileget(f.file);
	378	fdput(f);
13389010 DL	379	return ctx;
	380	}
	381	EXPORT_SYMBOL_GPL(eventfd_ctx_fdget);
	382
	383	/**
	384	* eventfd_ctx_fileget - Acquires a reference to the internal eventfd context.
	385	* @file: [in] Eventfd file pointer.
	386	*
	387	* Returns a pointer to the internal eventfd context, otherwise the error
	388	* pointer:
	389	*
	390	* -EINVAL : The @fd file descriptor is not an eventfd file.
	391	*/
	392	struct eventfd_ctx eventfd_ctx_fileget(struct file file)
	393	{
105f2b70 EB	394	struct eventfd_ctx *ctx;
105f2b70 EB	395
13389010 DL	396	if (file->f_op != &eventfd_fops)
	397	return ERR_PTR(-EINVAL);
	398
105f2b70 EB	399	ctx = file->private_data;
	400	kref_get(&ctx->kref);
	401	return ctx;
13389010 DL	402	}
	403	EXPORT_SYMBOL_GPL(eventfd_ctx_fileget);
	404
2fc96f83	405	static int do_eventfd(unsigned int count, int flags)
e1ad7468	406	{
e1ad7468	407	struct eventfd_ctx *ctx;
12aceb89	408	struct file *file;
7d815165	409	int fd;
e1ad7468	410
e38b36f3 UD	411	/* Check the EFD_* constants for consistency. */
	412	BUILD_BUG_ON(EFD_CLOEXEC != O_CLOEXEC);
	413	BUILD_BUG_ON(EFD_NONBLOCK != O_NONBLOCK);
	414
bcd0b235	415	if (flags & ~EFD_FLAGS_SET)
7d815165	416	return -EINVAL;
b087498e	417
e1ad7468 DL	418	ctx = kmalloc(sizeof(*ctx), GFP_KERNEL);
e1ad7468 DL	419	if (!ctx)
7d815165	420	return -ENOMEM;
e1ad7468	421
13389010	422	kref_init(&ctx->kref);
e1ad7468	423	init_waitqueue_head(&ctx->wqh);
e1ad7468	424	ctx->count = count;
bcd0b235	425	ctx->flags = flags;
b556db17	426	ctx->id = ida_simple_get(&eventfd_ida, 0, 0, GFP_KERNEL);
e1ad7468	427
12aceb89 JA	428	flags &= EFD_SHARED_FCNTL_FLAGS;
	429	flags \|= O_RDWR;
	430	fd = get_unused_fd_flags(flags);
7d815165	431	if (fd < 0)
12aceb89 JA	432	goto err;
	433
	434	file = anon_inode_getfile("[eventfd]", &eventfd_fops, ctx, flags);
	435	if (IS_ERR(file)) {
	436	put_unused_fd(fd);
	437	fd = PTR_ERR(file);
	438	goto err;
	439	}
562787a5	440
12aceb89 JA	441	file->f_mode \|= FMODE_NOWAIT;
	442	fd_install(fd, file);
	443	return fd;
	444	err:
	445	eventfd_free_ctx(ctx);
2030a42c	446	return fd;
e1ad7468 DL	447	}
e1ad7468 DL	448
2fc96f83 DB	449	SYSCALL_DEFINE2(eventfd2, unsigned int, count, int, flags)
	450	{
	451	return do_eventfd(count, flags);
	452	}
	453
d4e82042	454	SYSCALL_DEFINE1(eventfd, unsigned int, count)
b087498e	455	{
2fc96f83	456	return do_eventfd(count, 0);
b087498e	457	}
bcd0b235	458