[linux.git] / fs / eventfd.c

/*
 *  fs/eventfd.c
 *
 *  Copyright (C) 2007  Davide Libenzi <[email protected]>
 *
 */

#include <linux/file.h>
#include <linux/poll.h>
#include <linux/init.h>
#include <linux/fs.h>
#include <linux/sched.h>
#include <linux/kernel.h>
#include <linux/list.h>
#include <linux/spinlock.h>
#include <linux/anon_inodes.h>
#include <linux/eventfd.h>
#include <linux/syscalls.h>

struct eventfd_ctx {
	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
	 * size eventfd_signal() also, adds to the "count" counter and
	 * issue a wakeup.
	 */
	__u64 count;
};

/*
 * Adds "n" to the eventfd counter "count". Returns "n" in case of
 * success, or a value lower then "n" in case of coutner overflow.
 * 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 returining a POLLERR to poll(2).
 */
int eventfd_signal(struct file *file, int n)
{
	struct eventfd_ctx *ctx = file->private_data;
	unsigned long flags;

	if (n < 0)
		return -EINVAL;
	spin_lock_irqsave(&ctx->wqh.lock, flags);
	if (ULLONG_MAX - ctx->count < n)
		n = (int) (ULLONG_MAX - ctx->count);
	ctx->count += n;
	if (waitqueue_active(&ctx->wqh))
		wake_up_locked(&ctx->wqh);
	spin_unlock_irqrestore(&ctx->wqh.lock, flags);

	return n;
}

static int eventfd_release(struct inode *inode, struct file *file)
{
	kfree(file->private_data);
	return 0;
}

static unsigned int eventfd_poll(struct file *file, poll_table *wait)
{
	struct eventfd_ctx *ctx = file->private_data;
	unsigned int events = 0;
	unsigned long flags;

	poll_wait(file, &ctx->wqh, wait);

	spin_lock_irqsave(&ctx->wqh.lock, flags);
	if (ctx->count > 0)
		events |= POLLIN;
	if (ctx->count == ULLONG_MAX)
		events |= POLLERR;
	if (ULLONG_MAX - 1 > ctx->count)
		events |= POLLOUT;
	spin_unlock_irqrestore(&ctx->wqh.lock, flags);

	return events;
}

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;
	DECLARE_WAITQUEUE(wait, current);

	if (count < sizeof(ucnt))
		return -EINVAL;
	spin_lock_irq(&ctx->wqh.lock);
	res = -EAGAIN;
	ucnt = ctx->count;
	if (ucnt > 0)
		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 (ctx->count > 0) {
				ucnt = ctx->count;
				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 (res > 0) {
		ctx->count = 0;
		if (waitqueue_active(&ctx->wqh))
			wake_up_locked(&ctx->wqh);
	}
	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 (res > 0) {
		ctx->count += ucnt;
		if (waitqueue_active(&ctx->wqh))
			wake_up_locked(&ctx->wqh);
	}
	spin_unlock_irq(&ctx->wqh.lock);

	return res;
}

static const struct file_operations eventfd_fops = {
	.release	= eventfd_release,
	.poll		= eventfd_poll,
	.read		= eventfd_read,
	.write		= eventfd_write,
};

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;
}

asmlinkage long sys_eventfd2(unsigned int count, int flags)
{
	int fd;
	struct eventfd_ctx *ctx;

	/* Check the EFD_* constants for consistency.  */
	BUILD_BUG_ON(EFD_CLOEXEC != O_CLOEXEC);
	BUILD_BUG_ON(EFD_NONBLOCK != O_NONBLOCK);

	if (flags & ~(EFD_CLOEXEC | EFD_NONBLOCK))
		return -EINVAL;

	ctx = kmalloc(sizeof(*ctx), GFP_KERNEL);
	if (!ctx)
		return -ENOMEM;

	init_waitqueue_head(&ctx->wqh);
	ctx->count = count;

	/*
	 * When we call this, the initialization must be complete, since
	 * anon_inode_getfd() will install the fd.
	 */
	fd = anon_inode_getfd("[eventfd]", &eventfd_fops, ctx,
			      flags & (O_CLOEXEC | O_NONBLOCK));
	if (fd < 0)
		kfree(ctx);
	return fd;
}

asmlinkage long sys_eventfd(unsigned int count)
{
	return sys_eventfd2(count, 0);
}
Commit	Line	Data
e1ad7468 DL	1	/*
	2	* fs/eventfd.c
	3	*
	4	* Copyright (C) 2007 Davide Libenzi <[email protected]>
	5	*
	6	*/
	7
	8	#include <linux/file.h>
	9	#include <linux/poll.h>
	10	#include <linux/init.h>
	11	#include <linux/fs.h>
	12	#include <linux/sched.h>
	13	#include <linux/kernel.h>
	14	#include <linux/list.h>
	15	#include <linux/spinlock.h>
	16	#include <linux/anon_inodes.h>
	17	#include <linux/eventfd.h>
7747cdb2	18	#include <linux/syscalls.h>
e1ad7468 DL	19
e1ad7468 DL	20	struct eventfd_ctx {
e1ad7468 DL	21	wait_queue_head_t wqh;
	22	/*
	23	* Every time that a write(2) is performed on an eventfd, the
	24	* value of the __u64 being written is added to "count" and a
	25	* wakeup is performed on "wqh". A read(2) will return the "count"
	26	* value to userspace, and will reset "count" to zero. The kernel
	27	* size eventfd_signal() also, adds to the "count" counter and
	28	* issue a wakeup.
	29	*/
	30	__u64 count;
	31	};
	32
	33	/*
	34	* Adds "n" to the eventfd counter "count". Returns "n" in case of
	35	* success, or a value lower then "n" in case of coutner overflow.
	36	* This function is supposed to be called by the kernel in paths
	37	* that do not allow sleeping. In this function we allow the counter
	38	* to reach the ULLONG_MAX value, and we signal this as overflow
	39	* condition by returining a POLLERR to poll(2).
	40	*/
	41	int eventfd_signal(struct file *file, int n)
	42	{
	43	struct eventfd_ctx *ctx = file->private_data;
	44	unsigned long flags;
	45
	46	if (n < 0)
	47	return -EINVAL;
d48eb233	48	spin_lock_irqsave(&ctx->wqh.lock, flags);
e1ad7468 DL	49	if (ULLONG_MAX - ctx->count < n)
	50	n = (int) (ULLONG_MAX - ctx->count);
	51	ctx->count += n;
	52	if (waitqueue_active(&ctx->wqh))
	53	wake_up_locked(&ctx->wqh);
d48eb233	54	spin_unlock_irqrestore(&ctx->wqh.lock, flags);
e1ad7468 DL	55
	56	return n;
	57	}
	58
	59	static int eventfd_release(struct inode inode, struct file file)
	60	{
	61	kfree(file->private_data);
	62	return 0;
	63	}
	64
	65	static unsigned int eventfd_poll(struct file file, poll_table wait)
	66	{
	67	struct eventfd_ctx *ctx = file->private_data;
	68	unsigned int events = 0;
	69	unsigned long flags;
	70
	71	poll_wait(file, &ctx->wqh, wait);
	72
d48eb233	73	spin_lock_irqsave(&ctx->wqh.lock, flags);
e1ad7468 DL	74	if (ctx->count > 0)
	75	events \|= POLLIN;
	76	if (ctx->count == ULLONG_MAX)
	77	events \|= POLLERR;
	78	if (ULLONG_MAX - 1 > ctx->count)
	79	events \|= POLLOUT;
d48eb233	80	spin_unlock_irqrestore(&ctx->wqh.lock, flags);
e1ad7468 DL	81
	82	return events;
	83	}
	84
	85	static ssize_t eventfd_read(struct file file, char __user buf, size_t count,
	86	loff_t *ppos)
	87	{
	88	struct eventfd_ctx *ctx = file->private_data;
	89	ssize_t res;
	90	__u64 ucnt;
	91	DECLARE_WAITQUEUE(wait, current);
	92
	93	if (count < sizeof(ucnt))
	94	return -EINVAL;
d48eb233	95	spin_lock_irq(&ctx->wqh.lock);
e1ad7468 DL	96	res = -EAGAIN;
	97	ucnt = ctx->count;
	98	if (ucnt > 0)
	99	res = sizeof(ucnt);
	100	else if (!(file->f_flags & O_NONBLOCK)) {
	101	__add_wait_queue(&ctx->wqh, &wait);
	102	for (res = 0;;) {
	103	set_current_state(TASK_INTERRUPTIBLE);
	104	if (ctx->count > 0) {
	105	ucnt = ctx->count;
	106	res = sizeof(ucnt);
	107	break;
	108	}
	109	if (signal_pending(current)) {
	110	res = -ERESTARTSYS;
	111	break;
	112	}
d48eb233	113	spin_unlock_irq(&ctx->wqh.lock);
e1ad7468	114	schedule();
d48eb233	115	spin_lock_irq(&ctx->wqh.lock);
e1ad7468 DL	116	}
	117	__remove_wait_queue(&ctx->wqh, &wait);
	118	__set_current_state(TASK_RUNNING);
	119	}
	120	if (res > 0) {
	121	ctx->count = 0;
	122	if (waitqueue_active(&ctx->wqh))
	123	wake_up_locked(&ctx->wqh);
	124	}
d48eb233	125	spin_unlock_irq(&ctx->wqh.lock);
e1ad7468 DL	126	if (res > 0 && put_user(ucnt, (__u64 __user *) buf))
	127	return -EFAULT;
	128
	129	return res;
	130	}
	131
	132	static ssize_t eventfd_write(struct file file, const char __user buf, size_t count,
	133	loff_t *ppos)
	134	{
	135	struct eventfd_ctx *ctx = file->private_data;
	136	ssize_t res;
	137	__u64 ucnt;
	138	DECLARE_WAITQUEUE(wait, current);
	139
	140	if (count < sizeof(ucnt))
	141	return -EINVAL;
	142	if (copy_from_user(&ucnt, buf, sizeof(ucnt)))
	143	return -EFAULT;
	144	if (ucnt == ULLONG_MAX)
	145	return -EINVAL;
d48eb233	146	spin_lock_irq(&ctx->wqh.lock);
e1ad7468 DL	147	res = -EAGAIN;
	148	if (ULLONG_MAX - ctx->count > ucnt)
	149	res = sizeof(ucnt);
	150	else if (!(file->f_flags & O_NONBLOCK)) {
	151	__add_wait_queue(&ctx->wqh, &wait);
	152	for (res = 0;;) {
	153	set_current_state(TASK_INTERRUPTIBLE);
	154	if (ULLONG_MAX - ctx->count > ucnt) {
	155	res = sizeof(ucnt);
	156	break;
	157	}
	158	if (signal_pending(current)) {
	159	res = -ERESTARTSYS;
	160	break;
	161	}
d48eb233	162	spin_unlock_irq(&ctx->wqh.lock);
e1ad7468	163	schedule();
d48eb233	164	spin_lock_irq(&ctx->wqh.lock);
e1ad7468 DL	165	}
	166	__remove_wait_queue(&ctx->wqh, &wait);
	167	__set_current_state(TASK_RUNNING);
	168	}
	169	if (res > 0) {
	170	ctx->count += ucnt;
	171	if (waitqueue_active(&ctx->wqh))
	172	wake_up_locked(&ctx->wqh);
	173	}
d48eb233	174	spin_unlock_irq(&ctx->wqh.lock);
e1ad7468 DL	175
	176	return res;
	177	}
	178
	179	static const struct file_operations eventfd_fops = {
	180	.release = eventfd_release,
	181	.poll = eventfd_poll,
	182	.read = eventfd_read,
	183	.write = eventfd_write,
	184	};
	185
	186	struct file *eventfd_fget(int fd)
	187	{
	188	struct file *file;
	189
	190	file = fget(fd);
	191	if (!file)
	192	return ERR_PTR(-EBADF);
	193	if (file->f_op != &eventfd_fops) {
	194	fput(file);
	195	return ERR_PTR(-EINVAL);
	196	}
	197
	198	return file;
	199	}
	200
b087498e	201	asmlinkage long sys_eventfd2(unsigned int count, int flags)
e1ad7468	202	{
2030a42c	203	int fd;
e1ad7468	204	struct eventfd_ctx *ctx;
e1ad7468	205
e38b36f3 UD	206	/* Check the EFD_* constants for consistency. */
	207	BUILD_BUG_ON(EFD_CLOEXEC != O_CLOEXEC);
	208	BUILD_BUG_ON(EFD_NONBLOCK != O_NONBLOCK);
	209
e7d476df	210	if (flags & ~(EFD_CLOEXEC \| EFD_NONBLOCK))
b087498e UD	211	return -EINVAL;
b087498e UD	212
e1ad7468 DL	213	ctx = kmalloc(sizeof(*ctx), GFP_KERNEL);
	214	if (!ctx)
	215	return -ENOMEM;
	216
	217	init_waitqueue_head(&ctx->wqh);
e1ad7468 DL	218	ctx->count = count;
	219
	220	/*
	221	* When we call this, the initialization must be complete, since
	222	* anon_inode_getfd() will install the fd.
	223	*/
b087498e	224	fd = anon_inode_getfd("[eventfd]", &eventfd_fops, ctx,
e7d476df	225	flags & (O_CLOEXEC \| O_NONBLOCK));
2030a42c AV	226	if (fd < 0)
	227	kfree(ctx);
	228	return fd;
e1ad7468 DL	229	}
e1ad7468 DL	230
b087498e UD	231	asmlinkage long sys_eventfd(unsigned int count)
	232	{
	233	return sys_eventfd2(count, 0);
	234	}
	235