[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>

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;
};

/**
 * 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)
{
	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);
}
#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;

	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>
e1ad7468 DL	24
e1ad7468 DL	25	struct eventfd_ctx {
13389010	26	struct kref kref;
e1ad7468 DL	27	wait_queue_head_t wqh;
	28	/*
	29	* Every time that a write(2) is performed on an eventfd, the
	30	* value of the __u64 being written is added to "count" and a
	31	* wakeup is performed on "wqh". A read(2) will return the "count"
	32	* value to userspace, and will reset "count" to zero. The kernel
13389010	33	* side eventfd_signal() also, adds to the "count" counter and
e1ad7468 DL	34	* issue a wakeup.
	35	*/
	36	__u64 count;
bcd0b235	37	unsigned int flags;
e1ad7468 DL	38	};
e1ad7468 DL	39
13389010 DL	40	/**
	41	* eventfd_signal - Adds @n to the eventfd counter.
	42	* @ctx: [in] Pointer to the eventfd context.
	43	* @n: [in] Value of the counter to be added to the eventfd internal counter.
	44	* The value cannot be negative.
	45	*
	46	* This function is supposed to be called by the kernel in paths that do not
	47	* allow sleeping. In this function we allow the counter to reach the ULLONG_MAX
a9a08845	48	* value, and we signal this as overflow condition by returning a EPOLLERR
13389010 DL	49	* to poll(2).
13389010 DL	50	*
20d5a865	51	* Returns the amount by which the counter was incremented. This will be less
ee62c6b2	52	* than @n if the counter has overflowed.
e1ad7468	53	*/
ee62c6b2	54	__u64 eventfd_signal(struct eventfd_ctx *ctx, __u64 n)
e1ad7468	55	{
e1ad7468 DL	56	unsigned long flags;
e1ad7468 DL	57
d48eb233	58	spin_lock_irqsave(&ctx->wqh.lock, flags);
e1ad7468	59	if (ULLONG_MAX - ctx->count < n)
ee62c6b2	60	n = ULLONG_MAX - ctx->count;
e1ad7468 DL	61	ctx->count += n;
e1ad7468 DL	62	if (waitqueue_active(&ctx->wqh))
a9a08845	63	wake_up_locked_poll(&ctx->wqh, EPOLLIN);
d48eb233	64	spin_unlock_irqrestore(&ctx->wqh.lock, flags);
e1ad7468 DL	65
	66	return n;
	67	}
5718607b	68	EXPORT_SYMBOL_GPL(eventfd_signal);
e1ad7468	69
562787a5 DL	70	static void eventfd_free_ctx(struct eventfd_ctx *ctx)
	71	{
	72	kfree(ctx);
	73	}
	74
13389010 DL	75	static void eventfd_free(struct kref *kref)
	76	{
	77	struct eventfd_ctx *ctx = container_of(kref, struct eventfd_ctx, kref);
	78
562787a5	79	eventfd_free_ctx(ctx);
13389010 DL	80	}
13389010 DL	81
13389010 DL	82	/**
	83	* eventfd_ctx_put - Releases a reference to the internal eventfd context.
	84	* @ctx: [in] Pointer to eventfd context.
	85	*
	86	* The eventfd context reference must have been previously acquired either
105f2b70	87	* with eventfd_ctx_fdget() or eventfd_ctx_fileget().
13389010 DL	88	*/
	89	void eventfd_ctx_put(struct eventfd_ctx *ctx)
	90	{
	91	kref_put(&ctx->kref, eventfd_free);
	92	}
	93	EXPORT_SYMBOL_GPL(eventfd_ctx_put);
	94
e1ad7468 DL	95	static int eventfd_release(struct inode inode, struct file file)
e1ad7468 DL	96	{
13389010 DL	97	struct eventfd_ctx *ctx = file->private_data;
13389010 DL	98
a9a08845	99	wake_up_poll(&ctx->wqh, EPOLLHUP);
13389010	100	eventfd_ctx_put(ctx);
e1ad7468 DL	101	return 0;
	102	}
	103
a11e1d43	104	static __poll_t eventfd_poll(struct file file, poll_table wait)
e1ad7468 DL	105	{
e1ad7468 DL	106	struct eventfd_ctx *ctx = file->private_data;
076ccb76	107	__poll_t events = 0;
e22553e2	108	u64 count;
e1ad7468	109
a11e1d43 LT	110	poll_wait(file, &ctx->wqh, wait);
a11e1d43 LT	111
a484c3dd PB	112	/*
	113	* All writes to ctx->count occur within ctx->wqh.lock. This read
	114	* can be done outside ctx->wqh.lock because we know that poll_wait
	115	* takes that lock (through add_wait_queue) if our caller will sleep.
	116	*
	117	* The read _can_ therefore seep into add_wait_queue's critical
	118	* section, but cannot move above it! add_wait_queue's spin_lock acts
	119	* as an acquire barrier and ensures that the read be ordered properly
	120	* against the writes. The following CAN happen and is safe:
	121	*
	122	* poll write
	123	* ----------------- ------------
	124	* lock ctx->wqh.lock (in poll_wait)
	125	* count = ctx->count
	126	* __add_wait_queue
	127	* unlock ctx->wqh.lock
	128	* lock ctx->qwh.lock
	129	* ctx->count += n
	130	* if (waitqueue_active)
	131	* wake_up_locked_poll
	132	* unlock ctx->qwh.lock
	133	* eventfd_poll returns 0
	134	*
	135	* but the following, which would miss a wakeup, cannot happen:
	136	*
	137	* poll write
	138	* ----------------- ------------
	139	* count = ctx->count (INVALID!)
	140	* lock ctx->qwh.lock
	141	* ctx->count += n
	142	* waitqueue_active is false
	143	* no wake_up_locked_poll!
	144	* unlock ctx->qwh.lock
	145	* lock ctx->wqh.lock (in poll_wait)
	146	* __add_wait_queue
	147	* unlock ctx->wqh.lock
	148	* eventfd_poll returns 0
	149	*/
	150	count = READ_ONCE(ctx->count);
e1ad7468	151
e22553e2	152	if (count > 0)
a11e1d43	153	events \|= EPOLLIN;
e22553e2	154	if (count == ULLONG_MAX)
a9a08845	155	events \|= EPOLLERR;
e22553e2	156	if (ULLONG_MAX - 1 > count)
a11e1d43	157	events \|= EPOLLOUT;
e1ad7468 DL	158
	159	return events;
	160	}
	161
cb289d62 DL	162	static void eventfd_ctx_do_read(struct eventfd_ctx ctx, __u64 cnt)
	163	{
	164	*cnt = (ctx->flags & EFD_SEMAPHORE) ? 1 : ctx->count;
	165	ctx->count -= *cnt;
	166	}
	167
	168	/**
	169	* eventfd_ctx_remove_wait_queue - Read the current counter and removes wait queue.
	170	* @ctx: [in] Pointer to eventfd context.
	171	* @wait: [in] Wait queue to be removed.
36182185	172	* @cnt: [out] Pointer to the 64-bit counter value.
cb289d62	173	*
36182185	174	* Returns %0 if successful, or the following error codes:
cb289d62 DL	175	*
	176	* -EAGAIN : The operation would have blocked.
	177	*
	178	* This is used to atomically remove a wait queue entry from the eventfd wait
	179	* queue head, and read/reset the counter value.
	180	*/
ac6424b9	181	int eventfd_ctx_remove_wait_queue(struct eventfd_ctx ctx, wait_queue_entry_t wait,
cb289d62 DL	182	__u64 *cnt)
	183	{
	184	unsigned long flags;
	185
	186	spin_lock_irqsave(&ctx->wqh.lock, flags);
	187	eventfd_ctx_do_read(ctx, cnt);
	188	__remove_wait_queue(&ctx->wqh, wait);
	189	if (*cnt != 0 && waitqueue_active(&ctx->wqh))
a9a08845	190	wake_up_locked_poll(&ctx->wqh, EPOLLOUT);
cb289d62 DL	191	spin_unlock_irqrestore(&ctx->wqh.lock, flags);
	192
	193	return *cnt != 0 ? 0 : -EAGAIN;
	194	}
	195	EXPORT_SYMBOL_GPL(eventfd_ctx_remove_wait_queue);
	196
b6364572 EB	197	static ssize_t eventfd_read(struct file file, char __user buf, size_t count,
b6364572 EB	198	loff_t *ppos)
e1ad7468	199	{
b6364572	200	struct eventfd_ctx *ctx = file->private_data;
e1ad7468	201	ssize_t res;
b6364572	202	__u64 ucnt = 0;
e1ad7468 DL	203	DECLARE_WAITQUEUE(wait, current);
e1ad7468 DL	204
b6364572 EB	205	if (count < sizeof(ucnt))
	206	return -EINVAL;
	207
d48eb233	208	spin_lock_irq(&ctx->wqh.lock);
e1ad7468	209	res = -EAGAIN;
bcd0b235	210	if (ctx->count > 0)
b6364572 EB	211	res = sizeof(ucnt);
b6364572 EB	212	else if (!(file->f_flags & O_NONBLOCK)) {
e1ad7468	213	__add_wait_queue(&ctx->wqh, &wait);
cb289d62	214	for (;;) {
e1ad7468 DL	215	set_current_state(TASK_INTERRUPTIBLE);
e1ad7468 DL	216	if (ctx->count > 0) {
b6364572	217	res = sizeof(ucnt);
e1ad7468 DL	218	break;
	219	}
	220	if (signal_pending(current)) {
	221	res = -ERESTARTSYS;
	222	break;
	223	}
d48eb233	224	spin_unlock_irq(&ctx->wqh.lock);
e1ad7468	225	schedule();
d48eb233	226	spin_lock_irq(&ctx->wqh.lock);
e1ad7468 DL	227	}
	228	__remove_wait_queue(&ctx->wqh, &wait);
	229	__set_current_state(TASK_RUNNING);
	230	}
b6364572 EB	231	if (likely(res > 0)) {
b6364572 EB	232	eventfd_ctx_do_read(ctx, &ucnt);
e1ad7468	233	if (waitqueue_active(&ctx->wqh))
a9a08845	234	wake_up_locked_poll(&ctx->wqh, EPOLLOUT);
e1ad7468	235	}
d48eb233	236	spin_unlock_irq(&ctx->wqh.lock);
e1ad7468	237
b6364572 EB	238	if (res > 0 && put_user(ucnt, (__u64 __user *)buf))
b6364572 EB	239	return -EFAULT;
cb289d62	240
b6364572	241	return res;
cb289d62	242	}
e1ad7468 DL	243
	244	static ssize_t eventfd_write(struct file file, const char __user buf, size_t count,
	245	loff_t *ppos)
	246	{
	247	struct eventfd_ctx *ctx = file->private_data;
	248	ssize_t res;
	249	__u64 ucnt;
	250	DECLARE_WAITQUEUE(wait, current);
	251
	252	if (count < sizeof(ucnt))
	253	return -EINVAL;
	254	if (copy_from_user(&ucnt, buf, sizeof(ucnt)))
	255	return -EFAULT;
	256	if (ucnt == ULLONG_MAX)
	257	return -EINVAL;
d48eb233	258	spin_lock_irq(&ctx->wqh.lock);
e1ad7468 DL	259	res = -EAGAIN;
	260	if (ULLONG_MAX - ctx->count > ucnt)
	261	res = sizeof(ucnt);
	262	else if (!(file->f_flags & O_NONBLOCK)) {
	263	__add_wait_queue(&ctx->wqh, &wait);
	264	for (res = 0;;) {
	265	set_current_state(TASK_INTERRUPTIBLE);
	266	if (ULLONG_MAX - ctx->count > ucnt) {
	267	res = sizeof(ucnt);
	268	break;
	269	}
	270	if (signal_pending(current)) {
	271	res = -ERESTARTSYS;
	272	break;
	273	}
d48eb233	274	spin_unlock_irq(&ctx->wqh.lock);
e1ad7468	275	schedule();
d48eb233	276	spin_lock_irq(&ctx->wqh.lock);
e1ad7468 DL	277	}
	278	__remove_wait_queue(&ctx->wqh, &wait);
	279	__set_current_state(TASK_RUNNING);
	280	}
bcd0b235	281	if (likely(res > 0)) {
e1ad7468 DL	282	ctx->count += ucnt;
e1ad7468 DL	283	if (waitqueue_active(&ctx->wqh))
a9a08845	284	wake_up_locked_poll(&ctx->wqh, EPOLLIN);
e1ad7468	285	}
d48eb233	286	spin_unlock_irq(&ctx->wqh.lock);
e1ad7468 DL	287
	288	return res;
	289	}
	290
cbac5542	291	#ifdef CONFIG_PROC_FS
a3816ab0	292	static void eventfd_show_fdinfo(struct seq_file m, struct file f)
cbac5542 CG	293	{
cbac5542 CG	294	struct eventfd_ctx *ctx = f->private_data;
cbac5542 CG	295
cbac5542 CG	296	spin_lock_irq(&ctx->wqh.lock);
a3816ab0 JP	297	seq_printf(m, "eventfd-count: %16llx\n",
a3816ab0 JP	298	(unsigned long long)ctx->count);
cbac5542	299	spin_unlock_irq(&ctx->wqh.lock);
cbac5542 CG	300	}
	301	#endif
	302
e1ad7468	303	static const struct file_operations eventfd_fops = {
cbac5542 CG	304	#ifdef CONFIG_PROC_FS
	305	.show_fdinfo = eventfd_show_fdinfo,
	306	#endif
e1ad7468	307	.release = eventfd_release,
a11e1d43	308	.poll = eventfd_poll,
e1ad7468 DL	309	.read = eventfd_read,
e1ad7468 DL	310	.write = eventfd_write,
6038f373	311	.llseek = noop_llseek,
e1ad7468 DL	312	};
e1ad7468 DL	313
13389010 DL	314	/**
	315	* eventfd_fget - Acquire a reference of an eventfd file descriptor.
	316	* @fd: [in] Eventfd file descriptor.
	317	*
	318	* Returns a pointer to the eventfd file structure in case of success, or the
	319	* following error pointer:
	320	*
	321	* -EBADF : Invalid @fd file descriptor.
	322	* -EINVAL : The @fd file descriptor is not an eventfd file.
	323	*/
e1ad7468 DL	324	struct file *eventfd_fget(int fd)
	325	{
	326	struct file *file;
	327
	328	file = fget(fd);
	329	if (!file)
	330	return ERR_PTR(-EBADF);
	331	if (file->f_op != &eventfd_fops) {
	332	fput(file);
	333	return ERR_PTR(-EINVAL);
	334	}
	335
	336	return file;
	337	}
5718607b	338	EXPORT_SYMBOL_GPL(eventfd_fget);
e1ad7468	339
13389010 DL	340	/**
	341	* eventfd_ctx_fdget - Acquires a reference to the internal eventfd context.
	342	* @fd: [in] Eventfd file descriptor.
	343	*
	344	* Returns a pointer to the internal eventfd context, otherwise the error
	345	* pointers returned by the following functions:
	346	*
	347	* eventfd_fget
	348	*/
	349	struct eventfd_ctx *eventfd_ctx_fdget(int fd)
	350	{
13389010	351	struct eventfd_ctx *ctx;
36a74117 AV	352	struct fd f = fdget(fd);
	353	if (!f.file)
	354	return ERR_PTR(-EBADF);
	355	ctx = eventfd_ctx_fileget(f.file);
	356	fdput(f);
13389010 DL	357	return ctx;
	358	}
	359	EXPORT_SYMBOL_GPL(eventfd_ctx_fdget);
	360
	361	/**
	362	* eventfd_ctx_fileget - Acquires a reference to the internal eventfd context.
	363	* @file: [in] Eventfd file pointer.
	364	*
	365	* Returns a pointer to the internal eventfd context, otherwise the error
	366	* pointer:
	367	*
	368	* -EINVAL : The @fd file descriptor is not an eventfd file.
	369	*/
	370	struct eventfd_ctx eventfd_ctx_fileget(struct file file)
	371	{
105f2b70 EB	372	struct eventfd_ctx *ctx;
105f2b70 EB	373
13389010 DL	374	if (file->f_op != &eventfd_fops)
	375	return ERR_PTR(-EINVAL);
	376
105f2b70 EB	377	ctx = file->private_data;
	378	kref_get(&ctx->kref);
	379	return ctx;
13389010 DL	380	}
	381	EXPORT_SYMBOL_GPL(eventfd_ctx_fileget);
	382
2fc96f83	383	static int do_eventfd(unsigned int count, int flags)
e1ad7468	384	{
e1ad7468	385	struct eventfd_ctx *ctx;
7d815165	386	int fd;
e1ad7468	387
e38b36f3 UD	388	/* Check the EFD_* constants for consistency. */
	389	BUILD_BUG_ON(EFD_CLOEXEC != O_CLOEXEC);
	390	BUILD_BUG_ON(EFD_NONBLOCK != O_NONBLOCK);
	391
bcd0b235	392	if (flags & ~EFD_FLAGS_SET)
7d815165	393	return -EINVAL;
b087498e	394
e1ad7468 DL	395	ctx = kmalloc(sizeof(*ctx), GFP_KERNEL);
e1ad7468 DL	396	if (!ctx)
7d815165	397	return -ENOMEM;
e1ad7468	398
13389010	399	kref_init(&ctx->kref);
e1ad7468	400	init_waitqueue_head(&ctx->wqh);
e1ad7468	401	ctx->count = count;
bcd0b235	402	ctx->flags = flags;
e1ad7468	403
7d815165 EB	404	fd = anon_inode_getfd("[eventfd]", &eventfd_fops, ctx,
	405	O_RDWR \| (flags & EFD_SHARED_FCNTL_FLAGS));
	406	if (fd < 0)
562787a5 DL	407	eventfd_free_ctx(ctx);
562787a5 DL	408
2030a42c	409	return fd;
e1ad7468 DL	410	}
e1ad7468 DL	411
2fc96f83 DB	412	SYSCALL_DEFINE2(eventfd2, unsigned int, count, int, flags)
	413	{
	414	return do_eventfd(count, flags);
	415	}
	416
d4e82042	417	SYSCALL_DEFINE1(eventfd, unsigned int, count)
b087498e	418	{
2fc96f83	419	return do_eventfd(count, 0);
b087498e	420	}
bcd0b235	421