[thirdparty/kernel/linux.git] / fs / eventfd.c

// SPDX-License-Identifier: GPL-2.0-only
/*
 *  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/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". If EFD_SEMAPHORE flag was not
	 * specified, 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;
};

__u64 eventfd_signal_mask(struct eventfd_ctx *ctx, __u64 n, __poll_t mask)
{
	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))
		return 0;

	spin_lock_irqsave(&ctx->wqh.lock, flags);
	current->in_eventfd = 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 | mask);
	current->in_eventfd = 0;
	spin_unlock_irqrestore(&ctx->wqh.lock, flags);

	return n;
}

/**
 * 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)
{
	return eventfd_signal_mask(ctx, n, 0);
}
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) && ctx->count) ? 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;

	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;
		}

		if (wait_event_interruptible_locked_irq(ctx->wqh, ctx->count)) {
			spin_unlock_irq(&ctx->wqh.lock);
			return -ERESTARTSYS;
		}
	}
	eventfd_ctx_do_read(ctx, &ucnt);
	current->in_eventfd = 1;
	if (waitqueue_active(&ctx->wqh))
		wake_up_locked_poll(&ctx->wqh, EPOLLOUT);
	current->in_eventfd = 0;
	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;

	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)) {
		res = wait_event_interruptible_locked_irq(ctx->wqh,
				ULLONG_MAX - ctx->count > ucnt);
		if (!res)
			res = sizeof(ucnt);
	}
	if (likely(res > 0)) {
		ctx->count += ucnt;
		current->in_eventfd = 1;
		if (waitqueue_active(&ctx->wqh))
			wake_up_locked_poll(&ctx->wqh, EPOLLIN);
		current->in_eventfd = 0;
	}
	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);
	seq_printf(m, "eventfd-semaphore: %d\n",
		   !!(ctx->flags & EFD_SEMAPHORE));
}
#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 <davidel@xmailserver.org>
	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
33d8b5d7 WY	36	* wakeup is performed on "wqh". If EFD_SEMAPHORE flag was not
	37	* specified, a read(2) will return the "count" value to userspace,
	38	* and will reset "count" to zero. The kernel side eventfd_signal()
	39	* also, adds to the "count" counter and issue a wakeup.
e1ad7468 DL	40	*/
e1ad7468 DL	41	__u64 count;
bcd0b235	42	unsigned int flags;
b556db17	43	int id;
e1ad7468 DL	44	};
e1ad7468 DL	45
38f1755a	46	__u64 eventfd_signal_mask(struct eventfd_ctx *ctx, __u64 n, __poll_t mask)
e1ad7468	47	{
e1ad7468 DL	48	unsigned long flags;
e1ad7468 DL	49
b5e683d5 JA	50	/*
	51	* Deadlock or stack overflow issues can happen if we recurse here
	52	* through waitqueue wakeup handlers. If the caller users potentially
	53	* nested waitqueues with custom wakeup handlers, then it should
b542e383 TG	54	* check eventfd_signal_allowed() before calling this function. If
b542e383 TG	55	* it returns false, the eventfd_signal() call should be deferred to a
b5e683d5 JA	56	* safe context.
b5e683d5 JA	57	*/
9f0deaa1	58	if (WARN_ON_ONCE(current->in_eventfd))
b5e683d5 JA	59	return 0;
b5e683d5 JA	60
d48eb233	61	spin_lock_irqsave(&ctx->wqh.lock, flags);
9f0deaa1	62	current->in_eventfd = 1;
e1ad7468	63	if (ULLONG_MAX - ctx->count < n)
ee62c6b2	64	n = ULLONG_MAX - ctx->count;
e1ad7468 DL	65	ctx->count += n;
e1ad7468 DL	66	if (waitqueue_active(&ctx->wqh))
03e02acd	67	wake_up_locked_poll(&ctx->wqh, EPOLLIN \| mask);
9f0deaa1	68	current->in_eventfd = 0;
d48eb233	69	spin_unlock_irqrestore(&ctx->wqh.lock, flags);
e1ad7468 DL	70
	71	return n;
	72	}
03e02acd JA	73
	74	/**
	75	* eventfd_signal - Adds @n to the eventfd counter.
	76	* @ctx: [in] Pointer to the eventfd context.
	77	* @n: [in] Value of the counter to be added to the eventfd internal counter.
	78	* The value cannot be negative.
	79	*
	80	* This function is supposed to be called by the kernel in paths that do not
	81	* allow sleeping. In this function we allow the counter to reach the ULLONG_MAX
	82	* value, and we signal this as overflow condition by returning a EPOLLERR
	83	* to poll(2).
	84	*
	85	* Returns the amount by which the counter was incremented. This will be less
	86	* than @n if the counter has overflowed.
	87	*/
	88	__u64 eventfd_signal(struct eventfd_ctx *ctx, __u64 n)
	89	{
	90	return eventfd_signal_mask(ctx, n, 0);
	91	}
5718607b	92	EXPORT_SYMBOL_GPL(eventfd_signal);
e1ad7468	93
562787a5 DL	94	static void eventfd_free_ctx(struct eventfd_ctx *ctx)
562787a5 DL	95	{
b556db17 MY	96	if (ctx->id >= 0)
b556db17 MY	97	ida_simple_remove(&eventfd_ida, ctx->id);
562787a5 DL	98	kfree(ctx);
	99	}
	100
13389010 DL	101	static void eventfd_free(struct kref *kref)
	102	{
	103	struct eventfd_ctx *ctx = container_of(kref, struct eventfd_ctx, kref);
	104
562787a5	105	eventfd_free_ctx(ctx);
13389010 DL	106	}
13389010 DL	107
13389010 DL	108	/**
	109	* eventfd_ctx_put - Releases a reference to the internal eventfd context.
	110	* @ctx: [in] Pointer to eventfd context.
	111	*
	112	* The eventfd context reference must have been previously acquired either
105f2b70	113	* with eventfd_ctx_fdget() or eventfd_ctx_fileget().
13389010 DL	114	*/
	115	void eventfd_ctx_put(struct eventfd_ctx *ctx)
	116	{
	117	kref_put(&ctx->kref, eventfd_free);
	118	}
	119	EXPORT_SYMBOL_GPL(eventfd_ctx_put);
	120
e1ad7468 DL	121	static int eventfd_release(struct inode inode, struct file file)
e1ad7468 DL	122	{
13389010 DL	123	struct eventfd_ctx *ctx = file->private_data;
13389010 DL	124
a9a08845	125	wake_up_poll(&ctx->wqh, EPOLLHUP);
13389010	126	eventfd_ctx_put(ctx);
e1ad7468 DL	127	return 0;
	128	}
	129
a11e1d43	130	static __poll_t eventfd_poll(struct file file, poll_table wait)
e1ad7468 DL	131	{
e1ad7468 DL	132	struct eventfd_ctx *ctx = file->private_data;
076ccb76	133	__poll_t events = 0;
e22553e2	134	u64 count;
e1ad7468	135
a11e1d43 LT	136	poll_wait(file, &ctx->wqh, wait);
a11e1d43 LT	137
a484c3dd PB	138	/*
	139	* All writes to ctx->count occur within ctx->wqh.lock. This read
	140	* can be done outside ctx->wqh.lock because we know that poll_wait
	141	* takes that lock (through add_wait_queue) if our caller will sleep.
	142	*
	143	* The read _can_ therefore seep into add_wait_queue's critical
	144	* section, but cannot move above it! add_wait_queue's spin_lock acts
	145	* as an acquire barrier and ensures that the read be ordered properly
	146	* against the writes. The following CAN happen and is safe:
	147	*
	148	* poll write
	149	* ----------------- ------------
	150	* lock ctx->wqh.lock (in poll_wait)
	151	* count = ctx->count
	152	* __add_wait_queue
	153	* unlock ctx->wqh.lock
	154	* lock ctx->qwh.lock
	155	* ctx->count += n
	156	* if (waitqueue_active)
	157	* wake_up_locked_poll
	158	* unlock ctx->qwh.lock
	159	* eventfd_poll returns 0
	160	*
	161	* but the following, which would miss a wakeup, cannot happen:
	162	*
	163	* poll write
	164	* ----------------- ------------
	165	* count = ctx->count (INVALID!)
	166	* lock ctx->qwh.lock
	167	* ctx->count += n
	168	* waitqueue_active is false
	169	* no wake_up_locked_poll!
	170	* unlock ctx->qwh.lock
	171	* lock ctx->wqh.lock (in poll_wait)
	172	* __add_wait_queue
	173	* unlock ctx->wqh.lock
	174	* eventfd_poll returns 0
	175	*/
	176	count = READ_ONCE(ctx->count);
e1ad7468	177
e22553e2	178	if (count > 0)
a11e1d43	179	events \|= EPOLLIN;
e22553e2	180	if (count == ULLONG_MAX)
a9a08845	181	events \|= EPOLLERR;
e22553e2	182	if (ULLONG_MAX - 1 > count)
a11e1d43	183	events \|= EPOLLOUT;
e1ad7468 DL	184
	185	return events;
	186	}
	187
28f13267	188	void eventfd_ctx_do_read(struct eventfd_ctx ctx, __u64 cnt)
cb289d62	189	{
28f13267 DW	190	lockdep_assert_held(&ctx->wqh.lock);
28f13267 DW	191
758b4920	192	*cnt = ((ctx->flags & EFD_SEMAPHORE) && ctx->count) ? 1 : ctx->count;
cb289d62 DL	193	ctx->count -= *cnt;
cb289d62 DL	194	}
28f13267	195	EXPORT_SYMBOL_GPL(eventfd_ctx_do_read);
cb289d62 DL	196
	197	/**
	198	* eventfd_ctx_remove_wait_queue - Read the current counter and removes wait queue.
	199	* @ctx: [in] Pointer to eventfd context.
	200	* @wait: [in] Wait queue to be removed.
36182185	201	* @cnt: [out] Pointer to the 64-bit counter value.
cb289d62	202	*
36182185	203	* Returns %0 if successful, or the following error codes:
cb289d62 DL	204	*
	205	* -EAGAIN : The operation would have blocked.
	206	*
	207	* This is used to atomically remove a wait queue entry from the eventfd wait
	208	* queue head, and read/reset the counter value.
	209	*/
ac6424b9	210	int eventfd_ctx_remove_wait_queue(struct eventfd_ctx ctx, wait_queue_entry_t wait,
cb289d62 DL	211	__u64 *cnt)
	212	{
	213	unsigned long flags;
	214
	215	spin_lock_irqsave(&ctx->wqh.lock, flags);
	216	eventfd_ctx_do_read(ctx, cnt);
	217	__remove_wait_queue(&ctx->wqh, wait);
	218	if (*cnt != 0 && waitqueue_active(&ctx->wqh))
a9a08845	219	wake_up_locked_poll(&ctx->wqh, EPOLLOUT);
cb289d62 DL	220	spin_unlock_irqrestore(&ctx->wqh.lock, flags);
	221
	222	return *cnt != 0 ? 0 : -EAGAIN;
	223	}
	224	EXPORT_SYMBOL_GPL(eventfd_ctx_remove_wait_queue);
	225
12aceb89	226	static ssize_t eventfd_read(struct kiocb iocb, struct iov_iter to)
e1ad7468	227	{
12aceb89	228	struct file *file = iocb->ki_filp;
b6364572	229	struct eventfd_ctx *ctx = file->private_data;
b6364572	230	__u64 ucnt = 0;
e1ad7468	231
12aceb89	232	if (iov_iter_count(to) < sizeof(ucnt))
b6364572	233	return -EINVAL;
d48eb233	234	spin_lock_irq(&ctx->wqh.lock);
12aceb89 JA	235	if (!ctx->count) {
	236	if ((file->f_flags & O_NONBLOCK) \|\|
	237	(iocb->ki_flags & IOCB_NOWAIT)) {
	238	spin_unlock_irq(&ctx->wqh.lock);
	239	return -EAGAIN;
	240	}
113348a4 WY	241
113348a4 WY	242	if (wait_event_interruptible_locked_irq(ctx->wqh, ctx->count)) {
d48eb233	243	spin_unlock_irq(&ctx->wqh.lock);
113348a4	244	return -ERESTARTSYS;
e1ad7468	245	}
e1ad7468	246	}
12aceb89	247	eventfd_ctx_do_read(ctx, &ucnt);
9f0deaa1	248	current->in_eventfd = 1;
12aceb89 JA	249	if (waitqueue_active(&ctx->wqh))
12aceb89 JA	250	wake_up_locked_poll(&ctx->wqh, EPOLLOUT);
9f0deaa1	251	current->in_eventfd = 0;
d48eb233	252	spin_unlock_irq(&ctx->wqh.lock);
12aceb89	253	if (unlikely(copy_to_iter(&ucnt, sizeof(ucnt), to) != sizeof(ucnt)))
b6364572	254	return -EFAULT;
cb289d62	255
12aceb89	256	return sizeof(ucnt);
cb289d62	257	}
e1ad7468 DL	258
	259	static ssize_t eventfd_write(struct file file, const char __user buf, size_t count,
	260	loff_t *ppos)
	261	{
	262	struct eventfd_ctx *ctx = file->private_data;
	263	ssize_t res;
	264	__u64 ucnt;
e1ad7468 DL	265
	266	if (count < sizeof(ucnt))
	267	return -EINVAL;
	268	if (copy_from_user(&ucnt, buf, sizeof(ucnt)))
	269	return -EFAULT;
	270	if (ucnt == ULLONG_MAX)
	271	return -EINVAL;
d48eb233	272	spin_lock_irq(&ctx->wqh.lock);
e1ad7468 DL	273	res = -EAGAIN;
	274	if (ULLONG_MAX - ctx->count > ucnt)
	275	res = sizeof(ucnt);
	276	else if (!(file->f_flags & O_NONBLOCK)) {
113348a4 WY	277	res = wait_event_interruptible_locked_irq(ctx->wqh,
	278	ULLONG_MAX - ctx->count > ucnt);
	279	if (!res)
	280	res = sizeof(ucnt);
e1ad7468	281	}
bcd0b235	282	if (likely(res > 0)) {
e1ad7468	283	ctx->count += ucnt;
9f0deaa1	284	current->in_eventfd = 1;
e1ad7468	285	if (waitqueue_active(&ctx->wqh))
a9a08845	286	wake_up_locked_poll(&ctx->wqh, EPOLLIN);
9f0deaa1	287	current->in_eventfd = 0;
e1ad7468	288	}
d48eb233	289	spin_unlock_irq(&ctx->wqh.lock);
e1ad7468 DL	290
	291	return res;
	292	}
	293
cbac5542	294	#ifdef CONFIG_PROC_FS
a3816ab0	295	static void eventfd_show_fdinfo(struct seq_file m, struct file f)
cbac5542 CG	296	{
cbac5542 CG	297	struct eventfd_ctx *ctx = f->private_data;
cbac5542 CG	298
cbac5542 CG	299	spin_lock_irq(&ctx->wqh.lock);
a3816ab0 JP	300	seq_printf(m, "eventfd-count: %16llx\n",
a3816ab0 JP	301	(unsigned long long)ctx->count);
cbac5542	302	spin_unlock_irq(&ctx->wqh.lock);
b556db17	303	seq_printf(m, "eventfd-id: %d\n", ctx->id);
33d8b5d7 WY	304	seq_printf(m, "eventfd-semaphore: %d\n",
33d8b5d7 WY	305	!!(ctx->flags & EFD_SEMAPHORE));
cbac5542 CG	306	}
	307	#endif
	308
e1ad7468	309	static const struct file_operations eventfd_fops = {
cbac5542 CG	310	#ifdef CONFIG_PROC_FS
	311	.show_fdinfo = eventfd_show_fdinfo,
	312	#endif
e1ad7468	313	.release = eventfd_release,
a11e1d43	314	.poll = eventfd_poll,
12aceb89	315	.read_iter = eventfd_read,
e1ad7468	316	.write = eventfd_write,
6038f373	317	.llseek = noop_llseek,
e1ad7468 DL	318	};
e1ad7468 DL	319
13389010 DL	320	/**
	321	* eventfd_fget - Acquire a reference of an eventfd file descriptor.
	322	* @fd: [in] Eventfd file descriptor.
	323	*
	324	* Returns a pointer to the eventfd file structure in case of success, or the
	325	* following error pointer:
	326	*
	327	* -EBADF : Invalid @fd file descriptor.
	328	* -EINVAL : The @fd file descriptor is not an eventfd file.
	329	*/
e1ad7468 DL	330	struct file *eventfd_fget(int fd)
	331	{
	332	struct file *file;
	333
	334	file = fget(fd);
	335	if (!file)
	336	return ERR_PTR(-EBADF);
	337	if (file->f_op != &eventfd_fops) {
	338	fput(file);
	339	return ERR_PTR(-EINVAL);
	340	}
	341
	342	return file;
	343	}
5718607b	344	EXPORT_SYMBOL_GPL(eventfd_fget);
e1ad7468	345
13389010 DL	346	/**
	347	* eventfd_ctx_fdget - Acquires a reference to the internal eventfd context.
	348	* @fd: [in] Eventfd file descriptor.
	349	*
	350	* Returns a pointer to the internal eventfd context, otherwise the error
	351	* pointers returned by the following functions:
	352	*
	353	* eventfd_fget
	354	*/
	355	struct eventfd_ctx *eventfd_ctx_fdget(int fd)
	356	{
13389010	357	struct eventfd_ctx *ctx;
36a74117 AV	358	struct fd f = fdget(fd);
	359	if (!f.file)
	360	return ERR_PTR(-EBADF);
	361	ctx = eventfd_ctx_fileget(f.file);
	362	fdput(f);
13389010 DL	363	return ctx;
	364	}
	365	EXPORT_SYMBOL_GPL(eventfd_ctx_fdget);
	366
	367	/**
	368	* eventfd_ctx_fileget - Acquires a reference to the internal eventfd context.
	369	* @file: [in] Eventfd file pointer.
	370	*
	371	* Returns a pointer to the internal eventfd context, otherwise the error
	372	* pointer:
	373	*
	374	* -EINVAL : The @fd file descriptor is not an eventfd file.
	375	*/
	376	struct eventfd_ctx eventfd_ctx_fileget(struct file file)
	377	{
105f2b70 EB	378	struct eventfd_ctx *ctx;
105f2b70 EB	379
13389010 DL	380	if (file->f_op != &eventfd_fops)
	381	return ERR_PTR(-EINVAL);
	382
105f2b70 EB	383	ctx = file->private_data;
	384	kref_get(&ctx->kref);
	385	return ctx;
13389010 DL	386	}
	387	EXPORT_SYMBOL_GPL(eventfd_ctx_fileget);
	388
2fc96f83	389	static int do_eventfd(unsigned int count, int flags)
e1ad7468	390	{
e1ad7468	391	struct eventfd_ctx *ctx;
12aceb89	392	struct file *file;
7d815165	393	int fd;
e1ad7468	394
e38b36f3 UD	395	/* Check the EFD_* constants for consistency. */
	396	BUILD_BUG_ON(EFD_CLOEXEC != O_CLOEXEC);
	397	BUILD_BUG_ON(EFD_NONBLOCK != O_NONBLOCK);
	398
bcd0b235	399	if (flags & ~EFD_FLAGS_SET)
7d815165	400	return -EINVAL;
b087498e	401
e1ad7468 DL	402	ctx = kmalloc(sizeof(*ctx), GFP_KERNEL);
e1ad7468 DL	403	if (!ctx)
7d815165	404	return -ENOMEM;
e1ad7468	405
13389010	406	kref_init(&ctx->kref);
e1ad7468	407	init_waitqueue_head(&ctx->wqh);
e1ad7468	408	ctx->count = count;
bcd0b235	409	ctx->flags = flags;
b556db17	410	ctx->id = ida_simple_get(&eventfd_ida, 0, 0, GFP_KERNEL);
e1ad7468	411
12aceb89 JA	412	flags &= EFD_SHARED_FCNTL_FLAGS;
	413	flags \|= O_RDWR;
	414	fd = get_unused_fd_flags(flags);
7d815165	415	if (fd < 0)
12aceb89 JA	416	goto err;
	417
	418	file = anon_inode_getfile("[eventfd]", &eventfd_fops, ctx, flags);
	419	if (IS_ERR(file)) {
	420	put_unused_fd(fd);
	421	fd = PTR_ERR(file);
	422	goto err;
	423	}
562787a5	424
12aceb89 JA	425	file->f_mode \|= FMODE_NOWAIT;
	426	fd_install(fd, file);
	427	return fd;
	428	err:
	429	eventfd_free_ctx(ctx);
2030a42c	430	return fd;
e1ad7468 DL	431	}
e1ad7468 DL	432
2fc96f83 DB	433	SYSCALL_DEFINE2(eventfd2, unsigned int, count, int, flags)
	434	{
	435	return do_eventfd(count, flags);
	436	}
	437
d4e82042	438	SYSCALL_DEFINE1(eventfd, unsigned int, count)
b087498e	439	{
2fc96f83	440	return do_eventfd(count, 0);
b087498e	441	}
bcd0b235	442