[net-next-2.6.git] / fs / timerfd.c

/*
 *  fs/timerfd.c
 *
 *  Copyright (C) 2007  Davide Libenzi <davidel@xmailserver.org>
 *
 *
 *  Thanks to Thomas Gleixner for code reviews and useful comments.
 *
 */

#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/time.h>
#include <linux/hrtimer.h>
#include <linux/anon_inodes.h>
#include <linux/timerfd.h>
#include <linux/syscalls.h>

struct timerfd_ctx {
	struct hrtimer tmr;
	ktime_t tintv;
	wait_queue_head_t wqh;
	u64 ticks;
	int expired;
	int clockid;
};

/*
 * This gets called when the timer event triggers. We set the "expired"
 * flag, but we do not re-arm the timer (in case it's necessary,
 * tintv.tv64 != 0) until the timer is accessed.
 */
static enum hrtimer_restart timerfd_tmrproc(struct hrtimer *htmr)
{
	struct timerfd_ctx *ctx = container_of(htmr, struct timerfd_ctx, tmr);
	unsigned long flags;

	spin_lock_irqsave(&ctx->wqh.lock, flags);
	ctx->expired = 1;
	ctx->ticks++;
	wake_up_locked(&ctx->wqh);
	spin_unlock_irqrestore(&ctx->wqh.lock, flags);

	return HRTIMER_NORESTART;
}

static ktime_t timerfd_get_remaining(struct timerfd_ctx *ctx)
{
	ktime_t now, remaining;

	now = ctx->tmr.base->get_time();
	remaining = ktime_sub(ctx->tmr.expires, now);

	return remaining.tv64 < 0 ? ktime_set(0, 0): remaining;
}

static void timerfd_setup(struct timerfd_ctx *ctx, int flags,
			  const struct itimerspec *ktmr)
{
	enum hrtimer_mode htmode;
	ktime_t texp;

	htmode = (flags & TFD_TIMER_ABSTIME) ?
		HRTIMER_MODE_ABS: HRTIMER_MODE_REL;

	texp = timespec_to_ktime(ktmr->it_value);
	ctx->expired = 0;
	ctx->ticks = 0;
	ctx->tintv = timespec_to_ktime(ktmr->it_interval);
	hrtimer_init(&ctx->tmr, ctx->clockid, htmode);
	ctx->tmr.expires = texp;
	ctx->tmr.function = timerfd_tmrproc;
	if (texp.tv64 != 0)
		hrtimer_start(&ctx->tmr, texp, htmode);
}

static int timerfd_release(struct inode *inode, struct file *file)
{
	struct timerfd_ctx *ctx = file->private_data;

	hrtimer_cancel(&ctx->tmr);
	kfree(ctx);
	return 0;
}

static unsigned int timerfd_poll(struct file *file, poll_table *wait)
{
	struct timerfd_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->ticks)
		events |= POLLIN;
	spin_unlock_irqrestore(&ctx->wqh.lock, flags);

	return events;
}

static ssize_t timerfd_read(struct file *file, char __user *buf, size_t count,
			    loff_t *ppos)
{
	struct timerfd_ctx *ctx = file->private_data;
	ssize_t res;
	u64 ticks = 0;
	DECLARE_WAITQUEUE(wait, current);

	if (count < sizeof(ticks))
		return -EINVAL;
	spin_lock_irq(&ctx->wqh.lock);
	res = -EAGAIN;
	if (!ctx->ticks && !(file->f_flags & O_NONBLOCK)) {
		__add_wait_queue(&ctx->wqh, &wait);
		for (res = 0;;) {
			set_current_state(TASK_INTERRUPTIBLE);
			if (ctx->ticks) {
				res = 0;
				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 (ctx->ticks) {
		ticks = ctx->ticks;
		if (ctx->expired && ctx->tintv.tv64) {
			/*
			 * If tintv.tv64 != 0, this is a periodic timer that
			 * needs to be re-armed. We avoid doing it in the timer
			 * callback to avoid DoS attacks specifying a very
			 * short timer period.
			 */
			ticks += hrtimer_forward_now(&ctx->tmr,
						     ctx->tintv) - 1;
			hrtimer_restart(&ctx->tmr);
		}
		ctx->expired = 0;
		ctx->ticks = 0;
	}
	spin_unlock_irq(&ctx->wqh.lock);
	if (ticks)
		res = put_user(ticks, (u64 __user *) buf) ? -EFAULT: sizeof(ticks);
	return res;
}

static const struct file_operations timerfd_fops = {
	.release	= timerfd_release,
	.poll		= timerfd_poll,
	.read		= timerfd_read,
};

static struct file *timerfd_fget(int fd)
{
	struct file *file;

	file = fget(fd);
	if (!file)
		return ERR_PTR(-EBADF);
	if (file->f_op != &timerfd_fops) {
		fput(file);
		return ERR_PTR(-EINVAL);
	}

	return file;
}

asmlinkage long sys_timerfd_create(int clockid, int flags)
{
	int ufd;
	struct timerfd_ctx *ctx;

	/* Check the TFD_* constants for consistency.  */
	BUILD_BUG_ON(TFD_CLOEXEC != O_CLOEXEC);
	BUILD_BUG_ON(TFD_NONBLOCK != O_NONBLOCK);

	if (flags & ~(TFD_CLOEXEC | TFD_NONBLOCK))
		return -EINVAL;
	if (clockid != CLOCK_MONOTONIC &&
	    clockid != CLOCK_REALTIME)
		return -EINVAL;

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

	init_waitqueue_head(&ctx->wqh);
	ctx->clockid = clockid;
	hrtimer_init(&ctx->tmr, clockid, HRTIMER_MODE_ABS);

	ufd = anon_inode_getfd("[timerfd]", &timerfd_fops, ctx,
			       flags & (O_CLOEXEC | O_NONBLOCK));
	if (ufd < 0)
		kfree(ctx);

	return ufd;
}

asmlinkage long sys_timerfd_settime(int ufd, int flags,
				    const struct itimerspec __user *utmr,
				    struct itimerspec __user *otmr)
{
	struct file *file;
	struct timerfd_ctx *ctx;
	struct itimerspec ktmr, kotmr;

	if (copy_from_user(&ktmr, utmr, sizeof(ktmr)))
		return -EFAULT;

	if (!timespec_valid(&ktmr.it_value) ||
	    !timespec_valid(&ktmr.it_interval))
		return -EINVAL;

	file = timerfd_fget(ufd);
	if (IS_ERR(file))
		return PTR_ERR(file);
	ctx = file->private_data;

	/*
	 * We need to stop the existing timer before reprogramming
	 * it to the new values.
	 */
	for (;;) {
		spin_lock_irq(&ctx->wqh.lock);
		if (hrtimer_try_to_cancel(&ctx->tmr) >= 0)
			break;
		spin_unlock_irq(&ctx->wqh.lock);
		cpu_relax();
	}

	/*
	 * If the timer is expired and it's periodic, we need to advance it
	 * because the caller may want to know the previous expiration time.
	 * We do not update "ticks" and "expired" since the timer will be
	 * re-programmed again in the following timerfd_setup() call.
	 */
	if (ctx->expired && ctx->tintv.tv64)
		hrtimer_forward_now(&ctx->tmr, ctx->tintv);

	kotmr.it_value = ktime_to_timespec(timerfd_get_remaining(ctx));
	kotmr.it_interval = ktime_to_timespec(ctx->tintv);

	/*
	 * Re-program the timer to the new value ...
	 */
	timerfd_setup(ctx, flags, &ktmr);

	spin_unlock_irq(&ctx->wqh.lock);
	fput(file);
	if (otmr && copy_to_user(otmr, &kotmr, sizeof(kotmr)))
		return -EFAULT;

	return 0;
}

asmlinkage long sys_timerfd_gettime(int ufd, struct itimerspec __user *otmr)
{
	struct file *file;
	struct timerfd_ctx *ctx;
	struct itimerspec kotmr;

	file = timerfd_fget(ufd);
	if (IS_ERR(file))
		return PTR_ERR(file);
	ctx = file->private_data;

	spin_lock_irq(&ctx->wqh.lock);
	if (ctx->expired && ctx->tintv.tv64) {
		ctx->expired = 0;
		ctx->ticks +=
			hrtimer_forward_now(&ctx->tmr, ctx->tintv) - 1;
		hrtimer_restart(&ctx->tmr);
	}
	kotmr.it_value = ktime_to_timespec(timerfd_get_remaining(ctx));
	kotmr.it_interval = ktime_to_timespec(ctx->tintv);
	spin_unlock_irq(&ctx->wqh.lock);
	fput(file);

	return copy_to_user(otmr, &kotmr, sizeof(kotmr)) ? -EFAULT: 0;
}
Commit	Line	Data
b215e283 DL	1	/*
	2	* fs/timerfd.c
	3	*
	4	* Copyright (C) 2007 Davide Libenzi <davidel@xmailserver.org>
	5	*
	6	*
	7	* Thanks to Thomas Gleixner for code reviews and useful comments.
	8	*
	9	*/
	10
	11	#include <linux/file.h>
	12	#include <linux/poll.h>
	13	#include <linux/init.h>
	14	#include <linux/fs.h>
	15	#include <linux/sched.h>
	16	#include <linux/kernel.h>
	17	#include <linux/list.h>
	18	#include <linux/spinlock.h>
	19	#include <linux/time.h>
	20	#include <linux/hrtimer.h>
	21	#include <linux/anon_inodes.h>
	22	#include <linux/timerfd.h>
45cc2b96	23	#include <linux/syscalls.h>
b215e283 DL	24
	25	struct timerfd_ctx {
	26	struct hrtimer tmr;
	27	ktime_t tintv;
b215e283	28	wait_queue_head_t wqh;
4d672e7a	29	u64 ticks;
b215e283	30	int expired;
4d672e7a	31	int clockid;
b215e283 DL	32	};
	33
	34	/*
	35	* This gets called when the timer event triggers. We set the "expired"
	36	* flag, but we do not re-arm the timer (in case it's necessary,
4d672e7a	37	* tintv.tv64 != 0) until the timer is accessed.
b215e283 DL	38	*/
	39	static enum hrtimer_restart timerfd_tmrproc(struct hrtimer *htmr)
	40	{
	41	struct timerfd_ctx *ctx = container_of(htmr, struct timerfd_ctx, tmr);
	42	unsigned long flags;
	43
18963c01	44	spin_lock_irqsave(&ctx->wqh.lock, flags);
b215e283	45	ctx->expired = 1;
4d672e7a	46	ctx->ticks++;
b215e283	47	wake_up_locked(&ctx->wqh);
18963c01	48	spin_unlock_irqrestore(&ctx->wqh.lock, flags);
b215e283 DL	49
	50	return HRTIMER_NORESTART;
	51	}
	52
4d672e7a DL	53	static ktime_t timerfd_get_remaining(struct timerfd_ctx *ctx)
	54	{
	55	ktime_t now, remaining;
	56
	57	now = ctx->tmr.base->get_time();
	58	remaining = ktime_sub(ctx->tmr.expires, now);
	59
	60	return remaining.tv64 < 0 ? ktime_set(0, 0): remaining;
	61	}
	62
	63	static void timerfd_setup(struct timerfd_ctx *ctx, int flags,
b215e283 DL	64	const struct itimerspec *ktmr)
	65	{
	66	enum hrtimer_mode htmode;
	67	ktime_t texp;
	68
	69	htmode = (flags & TFD_TIMER_ABSTIME) ?
	70	HRTIMER_MODE_ABS: HRTIMER_MODE_REL;
	71
	72	texp = timespec_to_ktime(ktmr->it_value);
	73	ctx->expired = 0;
4d672e7a	74	ctx->ticks = 0;
b215e283	75	ctx->tintv = timespec_to_ktime(ktmr->it_interval);
4d672e7a	76	hrtimer_init(&ctx->tmr, ctx->clockid, htmode);
b215e283 DL	77	ctx->tmr.expires = texp;
	78	ctx->tmr.function = timerfd_tmrproc;
	79	if (texp.tv64 != 0)
	80	hrtimer_start(&ctx->tmr, texp, htmode);
	81	}
	82
	83	static int timerfd_release(struct inode inode, struct file file)
	84	{
	85	struct timerfd_ctx *ctx = file->private_data;
	86
	87	hrtimer_cancel(&ctx->tmr);
	88	kfree(ctx);
	89	return 0;
	90	}
	91
	92	static unsigned int timerfd_poll(struct file file, poll_table wait)
	93	{
	94	struct timerfd_ctx *ctx = file->private_data;
	95	unsigned int events = 0;
	96	unsigned long flags;
	97
	98	poll_wait(file, &ctx->wqh, wait);
	99
18963c01	100	spin_lock_irqsave(&ctx->wqh.lock, flags);
4d672e7a	101	if (ctx->ticks)
b215e283	102	events \|= POLLIN;
18963c01	103	spin_unlock_irqrestore(&ctx->wqh.lock, flags);
b215e283 DL	104
	105	return events;
	106	}
	107
	108	static ssize_t timerfd_read(struct file file, char __user buf, size_t count,
	109	loff_t *ppos)
	110	{
	111	struct timerfd_ctx *ctx = file->private_data;
	112	ssize_t res;
09828402	113	u64 ticks = 0;
b215e283 DL	114	DECLARE_WAITQUEUE(wait, current);
	115
	116	if (count < sizeof(ticks))
	117	return -EINVAL;
18963c01	118	spin_lock_irq(&ctx->wqh.lock);
b215e283	119	res = -EAGAIN;
4d672e7a	120	if (!ctx->ticks && !(file->f_flags & O_NONBLOCK)) {
b215e283 DL	121	__add_wait_queue(&ctx->wqh, &wait);
	122	for (res = 0;;) {
	123	set_current_state(TASK_INTERRUPTIBLE);
4d672e7a	124	if (ctx->ticks) {
b215e283 DL	125	res = 0;
	126	break;
	127	}
	128	if (signal_pending(current)) {
	129	res = -ERESTARTSYS;
	130	break;
	131	}
18963c01	132	spin_unlock_irq(&ctx->wqh.lock);
b215e283	133	schedule();
18963c01	134	spin_lock_irq(&ctx->wqh.lock);
b215e283 DL	135	}
	136	__remove_wait_queue(&ctx->wqh, &wait);
	137	__set_current_state(TASK_RUNNING);
	138	}
4d672e7a DL	139	if (ctx->ticks) {
	140	ticks = ctx->ticks;
	141	if (ctx->expired && ctx->tintv.tv64) {
b215e283 DL	142	/*
	143	* If tintv.tv64 != 0, this is a periodic timer that
	144	* needs to be re-armed. We avoid doing it in the timer
	145	* callback to avoid DoS attacks specifying a very
	146	* short timer period.
	147	*/
4d672e7a DL	148	ticks += hrtimer_forward_now(&ctx->tmr,
4d672e7a DL	149	ctx->tintv) - 1;
b215e283	150	hrtimer_restart(&ctx->tmr);
4d672e7a DL	151	}
	152	ctx->expired = 0;
	153	ctx->ticks = 0;
b215e283	154	}
18963c01	155	spin_unlock_irq(&ctx->wqh.lock);
b215e283	156	if (ticks)
09828402	157	res = put_user(ticks, (u64 __user *) buf) ? -EFAULT: sizeof(ticks);
b215e283 DL	158	return res;
	159	}
	160
	161	static const struct file_operations timerfd_fops = {
	162	.release = timerfd_release,
	163	.poll = timerfd_poll,
	164	.read = timerfd_read,
	165	};
	166
4d672e7a DL	167	static struct file *timerfd_fget(int fd)
	168	{
	169	struct file *file;
	170
	171	file = fget(fd);
	172	if (!file)
	173	return ERR_PTR(-EBADF);
	174	if (file->f_op != &timerfd_fops) {
	175	fput(file);
	176	return ERR_PTR(-EINVAL);
	177	}
	178
	179	return file;
	180	}
	181
	182	asmlinkage long sys_timerfd_create(int clockid, int flags)
b215e283	183	{
2030a42c	184	int ufd;
b215e283	185	struct timerfd_ctx *ctx;
b215e283	186
e38b36f3 UD	187	/* Check the TFD_* constants for consistency. */
	188	BUILD_BUG_ON(TFD_CLOEXEC != O_CLOEXEC);
	189	BUILD_BUG_ON(TFD_NONBLOCK != O_NONBLOCK);
	190
6b1ef0e6	191	if (flags & ~(TFD_CLOEXEC \| TFD_NONBLOCK))
4d672e7a	192	return -EINVAL;
b215e283 DL	193	if (clockid != CLOCK_MONOTONIC &&
	194	clockid != CLOCK_REALTIME)
	195	return -EINVAL;
4d672e7a DL	196
	197	ctx = kzalloc(sizeof(*ctx), GFP_KERNEL);
	198	if (!ctx)
	199	return -ENOMEM;
	200
	201	init_waitqueue_head(&ctx->wqh);
	202	ctx->clockid = clockid;
	203	hrtimer_init(&ctx->tmr, clockid, HRTIMER_MODE_ABS);
	204
11fcb6c1	205	ufd = anon_inode_getfd("[timerfd]", &timerfd_fops, ctx,
6b1ef0e6	206	flags & (O_CLOEXEC \| O_NONBLOCK));
2030a42c	207	if (ufd < 0)
4d672e7a	208	kfree(ctx);
4d672e7a DL	209
	210	return ufd;
	211	}
	212
	213	asmlinkage long sys_timerfd_settime(int ufd, int flags,
	214	const struct itimerspec __user *utmr,
	215	struct itimerspec __user *otmr)
	216	{
	217	struct file *file;
	218	struct timerfd_ctx *ctx;
	219	struct itimerspec ktmr, kotmr;
	220
	221	if (copy_from_user(&ktmr, utmr, sizeof(ktmr)))
	222	return -EFAULT;
	223
b215e283 DL	224	if (!timespec_valid(&ktmr.it_value) \|\|
	225	!timespec_valid(&ktmr.it_interval))
	226	return -EINVAL;
	227
4d672e7a DL	228	file = timerfd_fget(ufd);
	229	if (IS_ERR(file))
	230	return PTR_ERR(file);
	231	ctx = file->private_data;
b215e283	232
4d672e7a DL	233	/*
	234	* We need to stop the existing timer before reprogramming
	235	* it to the new values.
	236	*/
	237	for (;;) {
	238	spin_lock_irq(&ctx->wqh.lock);
	239	if (hrtimer_try_to_cancel(&ctx->tmr) >= 0)
	240	break;
18963c01	241	spin_unlock_irq(&ctx->wqh.lock);
4d672e7a	242	cpu_relax();
b215e283 DL	243	}
b215e283 DL	244
4d672e7a DL	245	/*
	246	* If the timer is expired and it's periodic, we need to advance it
	247	* because the caller may want to know the previous expiration time.
	248	* We do not update "ticks" and "expired" since the timer will be
	249	* re-programmed again in the following timerfd_setup() call.
	250	*/
	251	if (ctx->expired && ctx->tintv.tv64)
	252	hrtimer_forward_now(&ctx->tmr, ctx->tintv);
b215e283	253
4d672e7a DL	254	kotmr.it_value = ktime_to_timespec(timerfd_get_remaining(ctx));
	255	kotmr.it_interval = ktime_to_timespec(ctx->tintv);
	256
	257	/*
	258	* Re-program the timer to the new value ...
	259	*/
	260	timerfd_setup(ctx, flags, &ktmr);
	261
	262	spin_unlock_irq(&ctx->wqh.lock);
	263	fput(file);
	264	if (otmr && copy_to_user(otmr, &kotmr, sizeof(kotmr)))
	265	return -EFAULT;
	266
	267	return 0;
	268	}
	269
	270	asmlinkage long sys_timerfd_gettime(int ufd, struct itimerspec __user *otmr)
	271	{
	272	struct file *file;
	273	struct timerfd_ctx *ctx;
	274	struct itimerspec kotmr;
	275
	276	file = timerfd_fget(ufd);
	277	if (IS_ERR(file))
	278	return PTR_ERR(file);
	279	ctx = file->private_data;
	280
	281	spin_lock_irq(&ctx->wqh.lock);
	282	if (ctx->expired && ctx->tintv.tv64) {
	283	ctx->expired = 0;
	284	ctx->ticks +=
	285	hrtimer_forward_now(&ctx->tmr, ctx->tintv) - 1;
	286	hrtimer_restart(&ctx->tmr);
	287	}
	288	kotmr.it_value = ktime_to_timespec(timerfd_get_remaining(ctx));
	289	kotmr.it_interval = ktime_to_timespec(ctx->tintv);
	290	spin_unlock_irq(&ctx->wqh.lock);
	291	fput(file);
	292
	293	return copy_to_user(otmr, &kotmr, sizeof(kotmr)) ? -EFAULT: 0;
b215e283 DL	294	}
b215e283 DL	295