[net-next-2.6.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.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>
#include <linux/module.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;
	unsigned int flags;
};

/*
 * 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_poll(&ctx->wqh, POLLIN);
	spin_unlock_irqrestore(&ctx->wqh.lock, flags);

	return n;
}
EXPORT_SYMBOL_GPL(eventfd_signal);

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 = 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 (res = 0;;) {
			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)) {
		ucnt = (ctx->flags & EFD_SEMAPHORE) ? 1 : ctx->count;
		ctx->count -= ucnt;
		if (waitqueue_active(&ctx->wqh))
			wake_up_locked_poll(&ctx->wqh, POLLOUT);
	}
	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, POLLIN);
	}
	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;
}
EXPORT_SYMBOL_GPL(eventfd_fget);

SYSCALL_DEFINE2(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_FLAGS_SET)
		return -EINVAL;

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

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

	/*
	 * 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 & EFD_SHARED_FCNTL_FLAGS);
	if (fd < 0)
		kfree(ctx);
	return fd;
}

SYSCALL_DEFINE1(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 <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>
	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>
5718607b	19	#include <linux/module.h>
e1ad7468 DL	20
e1ad7468 DL	21	struct eventfd_ctx {
e1ad7468 DL	22	wait_queue_head_t wqh;
	23	/*
	24	* Every time that a write(2) is performed on an eventfd, the
	25	* value of the __u64 being written is added to "count" and a
	26	* wakeup is performed on "wqh". A read(2) will return the "count"
	27	* value to userspace, and will reset "count" to zero. The kernel
	28	* size eventfd_signal() also, adds to the "count" counter and
	29	* issue a wakeup.
	30	*/
	31	__u64 count;
bcd0b235	32	unsigned int flags;
e1ad7468 DL	33	};
	34
	35	/*
	36	* Adds "n" to the eventfd counter "count". Returns "n" in case of
	37	* success, or a value lower then "n" in case of coutner overflow.
	38	* This function is supposed to be called by the kernel in paths
	39	* that do not allow sleeping. In this function we allow the counter
	40	* to reach the ULLONG_MAX value, and we signal this as overflow
	41	* condition by returining a POLLERR to poll(2).
	42	*/
	43	int eventfd_signal(struct file *file, int n)
	44	{
	45	struct eventfd_ctx *ctx = file->private_data;
	46	unsigned long flags;
	47
	48	if (n < 0)
	49	return -EINVAL;
d48eb233	50	spin_lock_irqsave(&ctx->wqh.lock, flags);
e1ad7468 DL	51	if (ULLONG_MAX - ctx->count < n)
	52	n = (int) (ULLONG_MAX - ctx->count);
	53	ctx->count += n;
	54	if (waitqueue_active(&ctx->wqh))
39510888	55	wake_up_locked_poll(&ctx->wqh, POLLIN);
d48eb233	56	spin_unlock_irqrestore(&ctx->wqh.lock, flags);
e1ad7468 DL	57
	58	return n;
	59	}
5718607b	60	EXPORT_SYMBOL_GPL(eventfd_signal);
e1ad7468 DL	61
	62	static int eventfd_release(struct inode inode, struct file file)
	63	{
	64	kfree(file->private_data);
	65	return 0;
	66	}
	67
	68	static unsigned int eventfd_poll(struct file file, poll_table wait)
	69	{
	70	struct eventfd_ctx *ctx = file->private_data;
	71	unsigned int events = 0;
	72	unsigned long flags;
	73
	74	poll_wait(file, &ctx->wqh, wait);
	75
d48eb233	76	spin_lock_irqsave(&ctx->wqh.lock, flags);
e1ad7468 DL	77	if (ctx->count > 0)
	78	events \|= POLLIN;
	79	if (ctx->count == ULLONG_MAX)
	80	events \|= POLLERR;
	81	if (ULLONG_MAX - 1 > ctx->count)
	82	events \|= POLLOUT;
d48eb233	83	spin_unlock_irqrestore(&ctx->wqh.lock, flags);
e1ad7468 DL	84
	85	return events;
	86	}
	87
	88	static ssize_t eventfd_read(struct file file, char __user buf, size_t count,
	89	loff_t *ppos)
	90	{
	91	struct eventfd_ctx *ctx = file->private_data;
	92	ssize_t res;
bcd0b235	93	__u64 ucnt = 0;
e1ad7468 DL	94	DECLARE_WAITQUEUE(wait, current);
	95
	96	if (count < sizeof(ucnt))
	97	return -EINVAL;
d48eb233	98	spin_lock_irq(&ctx->wqh.lock);
e1ad7468	99	res = -EAGAIN;
bcd0b235	100	if (ctx->count > 0)
e1ad7468 DL	101	res = sizeof(ucnt);
	102	else if (!(file->f_flags & O_NONBLOCK)) {
	103	__add_wait_queue(&ctx->wqh, &wait);
	104	for (res = 0;;) {
	105	set_current_state(TASK_INTERRUPTIBLE);
	106	if (ctx->count > 0) {
e1ad7468 DL	107	res = sizeof(ucnt);
	108	break;
	109	}
	110	if (signal_pending(current)) {
	111	res = -ERESTARTSYS;
	112	break;
	113	}
d48eb233	114	spin_unlock_irq(&ctx->wqh.lock);
e1ad7468	115	schedule();
d48eb233	116	spin_lock_irq(&ctx->wqh.lock);
e1ad7468 DL	117	}
	118	__remove_wait_queue(&ctx->wqh, &wait);
	119	__set_current_state(TASK_RUNNING);
	120	}
bcd0b235 DL	121	if (likely(res > 0)) {
	122	ucnt = (ctx->flags & EFD_SEMAPHORE) ? 1 : ctx->count;
	123	ctx->count -= ucnt;
e1ad7468	124	if (waitqueue_active(&ctx->wqh))
39510888	125	wake_up_locked_poll(&ctx->wqh, POLLOUT);
e1ad7468	126	}
d48eb233	127	spin_unlock_irq(&ctx->wqh.lock);
e1ad7468 DL	128	if (res > 0 && put_user(ucnt, (__u64 __user *) buf))
	129	return -EFAULT;
	130
	131	return res;
	132	}
	133
	134	static ssize_t eventfd_write(struct file file, const char __user buf, size_t count,
	135	loff_t *ppos)
	136	{
	137	struct eventfd_ctx *ctx = file->private_data;
	138	ssize_t res;
	139	__u64 ucnt;
	140	DECLARE_WAITQUEUE(wait, current);
	141
	142	if (count < sizeof(ucnt))
	143	return -EINVAL;
	144	if (copy_from_user(&ucnt, buf, sizeof(ucnt)))
	145	return -EFAULT;
	146	if (ucnt == ULLONG_MAX)
	147	return -EINVAL;
d48eb233	148	spin_lock_irq(&ctx->wqh.lock);
e1ad7468 DL	149	res = -EAGAIN;
	150	if (ULLONG_MAX - ctx->count > ucnt)
	151	res = sizeof(ucnt);
	152	else if (!(file->f_flags & O_NONBLOCK)) {
	153	__add_wait_queue(&ctx->wqh, &wait);
	154	for (res = 0;;) {
	155	set_current_state(TASK_INTERRUPTIBLE);
	156	if (ULLONG_MAX - ctx->count > ucnt) {
	157	res = sizeof(ucnt);
	158	break;
	159	}
	160	if (signal_pending(current)) {
	161	res = -ERESTARTSYS;
	162	break;
	163	}
d48eb233	164	spin_unlock_irq(&ctx->wqh.lock);
e1ad7468	165	schedule();
d48eb233	166	spin_lock_irq(&ctx->wqh.lock);
e1ad7468 DL	167	}
	168	__remove_wait_queue(&ctx->wqh, &wait);
	169	__set_current_state(TASK_RUNNING);
	170	}
bcd0b235	171	if (likely(res > 0)) {
e1ad7468 DL	172	ctx->count += ucnt;
e1ad7468 DL	173	if (waitqueue_active(&ctx->wqh))
39510888	174	wake_up_locked_poll(&ctx->wqh, POLLIN);
e1ad7468	175	}
d48eb233	176	spin_unlock_irq(&ctx->wqh.lock);
e1ad7468 DL	177
	178	return res;
	179	}
	180
	181	static const struct file_operations eventfd_fops = {
	182	.release = eventfd_release,
	183	.poll = eventfd_poll,
	184	.read = eventfd_read,
	185	.write = eventfd_write,
	186	};
	187
	188	struct file *eventfd_fget(int fd)
	189	{
	190	struct file *file;
	191
	192	file = fget(fd);
	193	if (!file)
	194	return ERR_PTR(-EBADF);
	195	if (file->f_op != &eventfd_fops) {
	196	fput(file);
	197	return ERR_PTR(-EINVAL);
	198	}
	199
	200	return file;
	201	}
5718607b	202	EXPORT_SYMBOL_GPL(eventfd_fget);
e1ad7468	203
d4e82042	204	SYSCALL_DEFINE2(eventfd2, unsigned int, count, int, flags)
e1ad7468	205	{
2030a42c	206	int fd;
e1ad7468	207	struct eventfd_ctx *ctx;
e1ad7468	208
e38b36f3 UD	209	/* Check the EFD_* constants for consistency. */
	210	BUILD_BUG_ON(EFD_CLOEXEC != O_CLOEXEC);
	211	BUILD_BUG_ON(EFD_NONBLOCK != O_NONBLOCK);
	212
bcd0b235	213	if (flags & ~EFD_FLAGS_SET)
b087498e UD	214	return -EINVAL;
b087498e UD	215
e1ad7468 DL	216	ctx = kmalloc(sizeof(*ctx), GFP_KERNEL);
	217	if (!ctx)
	218	return -ENOMEM;
	219
	220	init_waitqueue_head(&ctx->wqh);
e1ad7468	221	ctx->count = count;
bcd0b235	222	ctx->flags = flags;
e1ad7468 DL	223
	224	/*
	225	* When we call this, the initialization must be complete, since
	226	* anon_inode_getfd() will install the fd.
	227	*/
b087498e	228	fd = anon_inode_getfd("[eventfd]", &eventfd_fops, ctx,
bcd0b235	229	flags & EFD_SHARED_FCNTL_FLAGS);
2030a42c AV	230	if (fd < 0)
	231	kfree(ctx);
	232	return fd;
e1ad7468 DL	233	}
e1ad7468 DL	234
d4e82042	235	SYSCALL_DEFINE1(eventfd, unsigned int, count)
b087498e UD	236	{
	237	return sys_eventfd2(count, 0);
	238	}
bcd0b235	239