[mirror_ubuntu-jammy-kernel.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>

DEFINE_PER_CPU(int, eventfd_wake_count);

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_count() before calling this function. If
	 * it returns true, the eventfd_signal() call should be deferred to a
	 * safe context.
	 */
	if (WARN_ON_ONCE(this_cpu_read(eventfd_wake_count)))
		return 0;

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