[mirror_ubuntu-focal-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>

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 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 <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 MY	25	#include <linux/idr.h>
b556db17 MY	26
91858f98 JA	27	DEFINE_PER_CPU(int, eventfd_wake_count);
91858f98 JA	28
ce528c4c	29	static DEFINE_IDA(eventfd_ida);
e1ad7468 DL	30
e1ad7468 DL	31	struct eventfd_ctx {
13389010	32	struct kref kref;
e1ad7468 DL	33	wait_queue_head_t wqh;
	34	/*
	35	* Every time that a write(2) is performed on an eventfd, the
	36	* value of the __u64 being written is added to "count" and a
	37	* wakeup is performed on "wqh". A read(2) will return the "count"
	38	* value to userspace, and will reset "count" to zero. The kernel
13389010	39	* side eventfd_signal() also, adds to the "count" counter and
e1ad7468 DL	40	* issue a wakeup.
	41	*/
	42	__u64 count;
bcd0b235	43	unsigned int flags;
b556db17	44	int id;
e1ad7468 DL	45	};
e1ad7468 DL	46
13389010 DL	47	/**
	48	* eventfd_signal - Adds @n to the eventfd counter.
	49	* @ctx: [in] Pointer to the eventfd context.
	50	* @n: [in] Value of the counter to be added to the eventfd internal counter.
	51	* The value cannot be negative.
	52	*
	53	* This function is supposed to be called by the kernel in paths that do not
	54	* allow sleeping. In this function we allow the counter to reach the ULLONG_MAX
a9a08845	55	* value, and we signal this as overflow condition by returning a EPOLLERR
13389010 DL	56	* to poll(2).
13389010 DL	57	*
20d5a865	58	* Returns the amount by which the counter was incremented. This will be less
ee62c6b2	59	* than @n if the counter has overflowed.
e1ad7468	60	*/
ee62c6b2	61	__u64 eventfd_signal(struct eventfd_ctx *ctx, __u64 n)
e1ad7468	62	{
e1ad7468 DL	63	unsigned long flags;
e1ad7468 DL	64
91858f98 JA	65	/*
	66	* Deadlock or stack overflow issues can happen if we recurse here
	67	* through waitqueue wakeup handlers. If the caller users potentially
	68	* nested waitqueues with custom wakeup handlers, then it should
	69	* check eventfd_signal_count() before calling this function. If
	70	* it returns true, the eventfd_signal() call should be deferred to a
	71	* safe context.
	72	*/
	73	if (WARN_ON_ONCE(this_cpu_read(eventfd_wake_count)))
	74	return 0;
	75
d48eb233	76	spin_lock_irqsave(&ctx->wqh.lock, flags);
91858f98	77	this_cpu_inc(eventfd_wake_count);
e1ad7468	78	if (ULLONG_MAX - ctx->count < n)
ee62c6b2	79	n = ULLONG_MAX - ctx->count;
e1ad7468 DL	80	ctx->count += n;
e1ad7468 DL	81	if (waitqueue_active(&ctx->wqh))
a9a08845	82	wake_up_locked_poll(&ctx->wqh, EPOLLIN);
91858f98	83	this_cpu_dec(eventfd_wake_count);
d48eb233	84	spin_unlock_irqrestore(&ctx->wqh.lock, flags);
e1ad7468 DL	85
	86	return n;
	87	}
5718607b	88	EXPORT_SYMBOL_GPL(eventfd_signal);
e1ad7468	89
562787a5 DL	90	static void eventfd_free_ctx(struct eventfd_ctx *ctx)
562787a5 DL	91	{
b556db17 MY	92	if (ctx->id >= 0)
b556db17 MY	93	ida_simple_remove(&eventfd_ida, ctx->id);
562787a5 DL	94	kfree(ctx);
	95	}
	96
13389010 DL	97	static void eventfd_free(struct kref *kref)
	98	{
	99	struct eventfd_ctx *ctx = container_of(kref, struct eventfd_ctx, kref);
	100
562787a5	101	eventfd_free_ctx(ctx);
13389010 DL	102	}
13389010 DL	103
13389010 DL	104	/**
	105	* eventfd_ctx_put - Releases a reference to the internal eventfd context.
	106	* @ctx: [in] Pointer to eventfd context.
	107	*
	108	* The eventfd context reference must have been previously acquired either
105f2b70	109	* with eventfd_ctx_fdget() or eventfd_ctx_fileget().
13389010 DL	110	*/
	111	void eventfd_ctx_put(struct eventfd_ctx *ctx)
	112	{
	113	kref_put(&ctx->kref, eventfd_free);
	114	}
	115	EXPORT_SYMBOL_GPL(eventfd_ctx_put);
	116
e1ad7468 DL	117	static int eventfd_release(struct inode inode, struct file file)
e1ad7468 DL	118	{
13389010 DL	119	struct eventfd_ctx *ctx = file->private_data;
13389010 DL	120
a9a08845	121	wake_up_poll(&ctx->wqh, EPOLLHUP);
13389010	122	eventfd_ctx_put(ctx);
e1ad7468 DL	123	return 0;
	124	}
	125
a11e1d43	126	static __poll_t eventfd_poll(struct file file, poll_table wait)
e1ad7468 DL	127	{
e1ad7468 DL	128	struct eventfd_ctx *ctx = file->private_data;
076ccb76	129	__poll_t events = 0;
e22553e2	130	u64 count;
e1ad7468	131
a11e1d43 LT	132	poll_wait(file, &ctx->wqh, wait);
a11e1d43 LT	133
a484c3dd PB	134	/*
	135	* All writes to ctx->count occur within ctx->wqh.lock. This read
	136	* can be done outside ctx->wqh.lock because we know that poll_wait
	137	* takes that lock (through add_wait_queue) if our caller will sleep.
	138	*
	139	* The read _can_ therefore seep into add_wait_queue's critical
	140	* section, but cannot move above it! add_wait_queue's spin_lock acts
	141	* as an acquire barrier and ensures that the read be ordered properly
	142	* against the writes. The following CAN happen and is safe:
	143	*
	144	* poll write
	145	* ----------------- ------------
	146	* lock ctx->wqh.lock (in poll_wait)
	147	* count = ctx->count
	148	* __add_wait_queue
	149	* unlock ctx->wqh.lock
	150	* lock ctx->qwh.lock
	151	* ctx->count += n
	152	* if (waitqueue_active)
	153	* wake_up_locked_poll
	154	* unlock ctx->qwh.lock
	155	* eventfd_poll returns 0
	156	*
	157	* but the following, which would miss a wakeup, cannot happen:
	158	*
	159	* poll write
	160	* ----------------- ------------
	161	* count = ctx->count (INVALID!)
	162	* lock ctx->qwh.lock
	163	* ctx->count += n
	164	* waitqueue_active is false
	165	* no wake_up_locked_poll!
	166	* unlock ctx->qwh.lock
	167	* lock ctx->wqh.lock (in poll_wait)
	168	* __add_wait_queue
	169	* unlock ctx->wqh.lock
	170	* eventfd_poll returns 0
	171	*/
	172	count = READ_ONCE(ctx->count);
e1ad7468	173
e22553e2	174	if (count > 0)
a11e1d43	175	events \|= EPOLLIN;
e22553e2	176	if (count == ULLONG_MAX)
a9a08845	177	events \|= EPOLLERR;
e22553e2	178	if (ULLONG_MAX - 1 > count)
a11e1d43	179	events \|= EPOLLOUT;
e1ad7468 DL	180
	181	return events;
	182	}
	183
cb289d62 DL	184	static void eventfd_ctx_do_read(struct eventfd_ctx ctx, __u64 cnt)
	185	{
	186	*cnt = (ctx->flags & EFD_SEMAPHORE) ? 1 : ctx->count;
	187	ctx->count -= *cnt;
	188	}
	189
	190	/**
	191	* eventfd_ctx_remove_wait_queue - Read the current counter and removes wait queue.
	192	* @ctx: [in] Pointer to eventfd context.
	193	* @wait: [in] Wait queue to be removed.
36182185	194	* @cnt: [out] Pointer to the 64-bit counter value.
cb289d62	195	*
36182185	196	* Returns %0 if successful, or the following error codes:
cb289d62 DL	197	*
	198	* -EAGAIN : The operation would have blocked.
	199	*
	200	* This is used to atomically remove a wait queue entry from the eventfd wait
	201	* queue head, and read/reset the counter value.
	202	*/
ac6424b9	203	int eventfd_ctx_remove_wait_queue(struct eventfd_ctx ctx, wait_queue_entry_t wait,
cb289d62 DL	204	__u64 *cnt)
	205	{
	206	unsigned long flags;
	207
	208	spin_lock_irqsave(&ctx->wqh.lock, flags);
	209	eventfd_ctx_do_read(ctx, cnt);
	210	__remove_wait_queue(&ctx->wqh, wait);
	211	if (*cnt != 0 && waitqueue_active(&ctx->wqh))
a9a08845	212	wake_up_locked_poll(&ctx->wqh, EPOLLOUT);
cb289d62 DL	213	spin_unlock_irqrestore(&ctx->wqh.lock, flags);
	214
	215	return *cnt != 0 ? 0 : -EAGAIN;
	216	}
	217	EXPORT_SYMBOL_GPL(eventfd_ctx_remove_wait_queue);
	218
b6364572 EB	219	static ssize_t eventfd_read(struct file file, char __user buf, size_t count,
b6364572 EB	220	loff_t *ppos)
e1ad7468	221	{
b6364572	222	struct eventfd_ctx *ctx = file->private_data;
e1ad7468	223	ssize_t res;
b6364572	224	__u64 ucnt = 0;
e1ad7468 DL	225	DECLARE_WAITQUEUE(wait, current);
e1ad7468 DL	226
b6364572 EB	227	if (count < sizeof(ucnt))
	228	return -EINVAL;
	229
d48eb233	230	spin_lock_irq(&ctx->wqh.lock);
e1ad7468	231	res = -EAGAIN;
bcd0b235	232	if (ctx->count > 0)
b6364572 EB	233	res = sizeof(ucnt);
b6364572 EB	234	else if (!(file->f_flags & O_NONBLOCK)) {
e1ad7468	235	__add_wait_queue(&ctx->wqh, &wait);
cb289d62	236	for (;;) {
e1ad7468 DL	237	set_current_state(TASK_INTERRUPTIBLE);
e1ad7468 DL	238	if (ctx->count > 0) {
b6364572	239	res = sizeof(ucnt);
e1ad7468 DL	240	break;
	241	}
	242	if (signal_pending(current)) {
	243	res = -ERESTARTSYS;
	244	break;
	245	}
d48eb233	246	spin_unlock_irq(&ctx->wqh.lock);
e1ad7468	247	schedule();
d48eb233	248	spin_lock_irq(&ctx->wqh.lock);
e1ad7468 DL	249	}
	250	__remove_wait_queue(&ctx->wqh, &wait);
	251	__set_current_state(TASK_RUNNING);
	252	}
b6364572 EB	253	if (likely(res > 0)) {
b6364572 EB	254	eventfd_ctx_do_read(ctx, &ucnt);
e1ad7468	255	if (waitqueue_active(&ctx->wqh))
a9a08845	256	wake_up_locked_poll(&ctx->wqh, EPOLLOUT);
e1ad7468	257	}
d48eb233	258	spin_unlock_irq(&ctx->wqh.lock);
e1ad7468	259
b6364572 EB	260	if (res > 0 && put_user(ucnt, (__u64 __user *)buf))
b6364572 EB	261	return -EFAULT;
cb289d62	262
b6364572	263	return res;
cb289d62	264	}
e1ad7468 DL	265
	266	static ssize_t eventfd_write(struct file file, const char __user buf, size_t count,
	267	loff_t *ppos)
	268	{
	269	struct eventfd_ctx *ctx = file->private_data;
	270	ssize_t res;
	271	__u64 ucnt;
	272	DECLARE_WAITQUEUE(wait, current);
	273
	274	if (count < sizeof(ucnt))
	275	return -EINVAL;
	276	if (copy_from_user(&ucnt, buf, sizeof(ucnt)))
	277	return -EFAULT;
	278	if (ucnt == ULLONG_MAX)
	279	return -EINVAL;
d48eb233	280	spin_lock_irq(&ctx->wqh.lock);
e1ad7468 DL	281	res = -EAGAIN;
	282	if (ULLONG_MAX - ctx->count > ucnt)
	283	res = sizeof(ucnt);
	284	else if (!(file->f_flags & O_NONBLOCK)) {
	285	__add_wait_queue(&ctx->wqh, &wait);
	286	for (res = 0;;) {
	287	set_current_state(TASK_INTERRUPTIBLE);
	288	if (ULLONG_MAX - ctx->count > ucnt) {
	289	res = sizeof(ucnt);
	290	break;
	291	}
	292	if (signal_pending(current)) {
	293	res = -ERESTARTSYS;
	294	break;
	295	}
d48eb233	296	spin_unlock_irq(&ctx->wqh.lock);
e1ad7468	297	schedule();
d48eb233	298	spin_lock_irq(&ctx->wqh.lock);
e1ad7468 DL	299	}
	300	__remove_wait_queue(&ctx->wqh, &wait);
	301	__set_current_state(TASK_RUNNING);
	302	}
bcd0b235	303	if (likely(res > 0)) {
e1ad7468 DL	304	ctx->count += ucnt;
e1ad7468 DL	305	if (waitqueue_active(&ctx->wqh))
a9a08845	306	wake_up_locked_poll(&ctx->wqh, EPOLLIN);
e1ad7468	307	}
d48eb233	308	spin_unlock_irq(&ctx->wqh.lock);
e1ad7468 DL	309
	310	return res;
	311	}
	312
cbac5542	313	#ifdef CONFIG_PROC_FS
a3816ab0	314	static void eventfd_show_fdinfo(struct seq_file m, struct file f)
cbac5542 CG	315	{
cbac5542 CG	316	struct eventfd_ctx *ctx = f->private_data;
cbac5542 CG	317
cbac5542 CG	318	spin_lock_irq(&ctx->wqh.lock);
a3816ab0 JP	319	seq_printf(m, "eventfd-count: %16llx\n",
a3816ab0 JP	320	(unsigned long long)ctx->count);
cbac5542	321	spin_unlock_irq(&ctx->wqh.lock);
b556db17	322	seq_printf(m, "eventfd-id: %d\n", ctx->id);
cbac5542 CG	323	}
	324	#endif
	325
e1ad7468	326	static const struct file_operations eventfd_fops = {
cbac5542 CG	327	#ifdef CONFIG_PROC_FS
	328	.show_fdinfo = eventfd_show_fdinfo,
	329	#endif
e1ad7468	330	.release = eventfd_release,
a11e1d43	331	.poll = eventfd_poll,
e1ad7468 DL	332	.read = eventfd_read,
e1ad7468 DL	333	.write = eventfd_write,
6038f373	334	.llseek = noop_llseek,
e1ad7468 DL	335	};
e1ad7468 DL	336
13389010 DL	337	/**
	338	* eventfd_fget - Acquire a reference of an eventfd file descriptor.
	339	* @fd: [in] Eventfd file descriptor.
	340	*
	341	* Returns a pointer to the eventfd file structure in case of success, or the
	342	* following error pointer:
	343	*
	344	* -EBADF : Invalid @fd file descriptor.
	345	* -EINVAL : The @fd file descriptor is not an eventfd file.
	346	*/
e1ad7468 DL	347	struct file *eventfd_fget(int fd)
	348	{
	349	struct file *file;
	350
	351	file = fget(fd);
	352	if (!file)
	353	return ERR_PTR(-EBADF);
	354	if (file->f_op != &eventfd_fops) {
	355	fput(file);
	356	return ERR_PTR(-EINVAL);
	357	}
	358
	359	return file;
	360	}
5718607b	361	EXPORT_SYMBOL_GPL(eventfd_fget);
e1ad7468	362
13389010 DL	363	/**
	364	* eventfd_ctx_fdget - Acquires a reference to the internal eventfd context.
	365	* @fd: [in] Eventfd file descriptor.
	366	*
	367	* Returns a pointer to the internal eventfd context, otherwise the error
	368	* pointers returned by the following functions:
	369	*
	370	* eventfd_fget
	371	*/
	372	struct eventfd_ctx *eventfd_ctx_fdget(int fd)
	373	{
13389010	374	struct eventfd_ctx *ctx;
36a74117 AV	375	struct fd f = fdget(fd);
	376	if (!f.file)
	377	return ERR_PTR(-EBADF);
	378	ctx = eventfd_ctx_fileget(f.file);
	379	fdput(f);
13389010 DL	380	return ctx;
	381	}
	382	EXPORT_SYMBOL_GPL(eventfd_ctx_fdget);
	383
	384	/**
	385	* eventfd_ctx_fileget - Acquires a reference to the internal eventfd context.
	386	* @file: [in] Eventfd file pointer.
	387	*
	388	* Returns a pointer to the internal eventfd context, otherwise the error
	389	* pointer:
	390	*
	391	* -EINVAL : The @fd file descriptor is not an eventfd file.
	392	*/
	393	struct eventfd_ctx eventfd_ctx_fileget(struct file file)
	394	{
105f2b70 EB	395	struct eventfd_ctx *ctx;
105f2b70 EB	396
13389010 DL	397	if (file->f_op != &eventfd_fops)
	398	return ERR_PTR(-EINVAL);
	399
105f2b70 EB	400	ctx = file->private_data;
	401	kref_get(&ctx->kref);
	402	return ctx;
13389010 DL	403	}
	404	EXPORT_SYMBOL_GPL(eventfd_ctx_fileget);
	405
2fc96f83	406	static int do_eventfd(unsigned int count, int flags)
e1ad7468	407	{
e1ad7468	408	struct eventfd_ctx *ctx;
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
7d815165 EB	428	fd = anon_inode_getfd("[eventfd]", &eventfd_fops, ctx,
	429	O_RDWR \| (flags & EFD_SHARED_FCNTL_FLAGS));
	430	if (fd < 0)
562787a5 DL	431	eventfd_free_ctx(ctx);
562787a5 DL	432
2030a42c	433	return fd;
e1ad7468 DL	434	}
e1ad7468 DL	435
2fc96f83 DB	436	SYSCALL_DEFINE2(eventfd2, unsigned int, count, int, flags)
	437	{
	438	return do_eventfd(count, flags);
	439	}
	440
d4e82042	441	SYSCALL_DEFINE1(eventfd, unsigned int, count)
b087498e	442	{
2fc96f83	443	return do_eventfd(count, 0);
b087498e	444	}
bcd0b235	445