[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>

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;

	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, EPOLLIN);
	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;
}

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_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 file *file, char __user *buf, size_t count,
			    loff_t *ppos)
{
	struct eventfd_ctx *ctx = file->private_data;
	ssize_t res;
	__u64 ucnt = 0;
	DECLARE_WAITQUEUE(wait, current);

	if (count < sizeof(ucnt))
		return -EINVAL;

	spin_lock_irq(&ctx->wqh.lock);
	res = -EAGAIN;
	if (ctx->count > 0)
		res = sizeof(ucnt);
	else if (!(file->f_flags & O_NONBLOCK)) {
		__add_wait_queue(&ctx->wqh, &wait);
		for (;;) {
			set_current_state(TASK_INTERRUPTIBLE);
			if (ctx->count > 0) {
				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)) {
		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 (res > 0 && put_user(ucnt, (__u64 __user *)buf))
		return -EFAULT;

	return res;
}

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		= 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;
	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);

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