[linux-2.6-block.git] / fs / eventfd.c

/*
 *  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>

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