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

/*
 *  fs/signalfd.c
 *
 *  Copyright (C) 2003  Linus Torvalds
 *
 *  Mon Mar 5, 2007: Davide Libenzi <davidel@xmailserver.org>
 *      Changed ->read() to return a siginfo strcture instead of signal number.
 *      Fixed locking in ->poll().
 *      Added sighand-detach notification.
 *      Added fd re-use in sys_signalfd() syscall.
 *      Now using anonymous inode source.
 *      Thanks to Oleg Nesterov for useful code review and suggestions.
 *      More comments and suggestions from Arnd Bergmann.
 *  Sat May 19, 2007: Davi E. M. Arnaut <davi@haxent.com.br>
 *      Retrieve multiple signals with one read() call
 *  Sun Jul 15, 2007: Davide Libenzi <davidel@xmailserver.org>
 *      Attach to the sighand only during read() and poll().
 */

#include <linux/file.h>
#include <linux/poll.h>
#include <linux/init.h>
#include <linux/fs.h>
#include <linux/sched.h>
#include <linux/slab.h>
#include <linux/kernel.h>
#include <linux/signal.h>
#include <linux/list.h>
#include <linux/anon_inodes.h>
#include <linux/signalfd.h>
#include <linux/syscalls.h>

struct signalfd_ctx {
	sigset_t sigmask;
};

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

static unsigned int signalfd_poll(struct file *file, poll_table *wait)
{
	struct signalfd_ctx *ctx = file->private_data;
	unsigned int events = 0;

	poll_wait(file, &current->sighand->signalfd_wqh, wait);

	spin_lock_irq(&current->sighand->siglock);
	if (next_signal(&current->pending, &ctx->sigmask) ||
	    next_signal(&current->signal->shared_pending,
			&ctx->sigmask))
		events |= POLLIN;
	spin_unlock_irq(&current->sighand->siglock);

	return events;
}

/*
 * Copied from copy_siginfo_to_user() in kernel/signal.c
 */
static int signalfd_copyinfo(struct signalfd_siginfo __user *uinfo,
			     siginfo_t const *kinfo)
{
	long err;

	BUILD_BUG_ON(sizeof(struct signalfd_siginfo) != 128);

	/*
	 * Unused members should be zero ...
	 */
	err = __clear_user(uinfo, sizeof(*uinfo));

	/*
	 * If you change siginfo_t structure, please be sure
	 * this code is fixed accordingly.
	 */
	err |= __put_user(kinfo->si_signo, &uinfo->ssi_signo);
	err |= __put_user(kinfo->si_errno, &uinfo->ssi_errno);
	err |= __put_user((short) kinfo->si_code, &uinfo->ssi_code);
	switch (kinfo->si_code & __SI_MASK) {
	case __SI_KILL:
		err |= __put_user(kinfo->si_pid, &uinfo->ssi_pid);
		err |= __put_user(kinfo->si_uid, &uinfo->ssi_uid);
		break;
	case __SI_TIMER:
		 err |= __put_user(kinfo->si_tid, &uinfo->ssi_tid);
		 err |= __put_user(kinfo->si_overrun, &uinfo->ssi_overrun);
		 err |= __put_user((long) kinfo->si_ptr, &uinfo->ssi_ptr);
		 err |= __put_user(kinfo->si_int, &uinfo->ssi_int);
		break;
	case __SI_POLL:
		err |= __put_user(kinfo->si_band, &uinfo->ssi_band);
		err |= __put_user(kinfo->si_fd, &uinfo->ssi_fd);
		break;
	case __SI_FAULT:
		err |= __put_user((long) kinfo->si_addr, &uinfo->ssi_addr);
#ifdef __ARCH_SI_TRAPNO
		err |= __put_user(kinfo->si_trapno, &uinfo->ssi_trapno);
#endif
		break;
	case __SI_CHLD:
		err |= __put_user(kinfo->si_pid, &uinfo->ssi_pid);
		err |= __put_user(kinfo->si_uid, &uinfo->ssi_uid);
		err |= __put_user(kinfo->si_status, &uinfo->ssi_status);
		err |= __put_user(kinfo->si_utime, &uinfo->ssi_utime);
		err |= __put_user(kinfo->si_stime, &uinfo->ssi_stime);
		break;
	case __SI_RT: /* This is not generated by the kernel as of now. */
	case __SI_MESGQ: /* But this is */
		err |= __put_user(kinfo->si_pid, &uinfo->ssi_pid);
		err |= __put_user(kinfo->si_uid, &uinfo->ssi_uid);
		err |= __put_user((long) kinfo->si_ptr, &uinfo->ssi_ptr);
		err |= __put_user(kinfo->si_int, &uinfo->ssi_int);
		break;
	default:
		/*
		 * This case catches also the signals queued by sigqueue().
		 */
		err |= __put_user(kinfo->si_pid, &uinfo->ssi_pid);
		err |= __put_user(kinfo->si_uid, &uinfo->ssi_uid);
		err |= __put_user((long) kinfo->si_ptr, &uinfo->ssi_ptr);
		err |= __put_user(kinfo->si_int, &uinfo->ssi_int);
		break;
	}

	return err ? -EFAULT: sizeof(*uinfo);
}

static ssize_t signalfd_dequeue(struct signalfd_ctx *ctx, siginfo_t *info,
				int nonblock)
{
	ssize_t ret;
	DECLARE_WAITQUEUE(wait, current);

	spin_lock_irq(&current->sighand->siglock);
	ret = dequeue_signal(current, &ctx->sigmask, info);
	switch (ret) {
	case 0:
		if (!nonblock)
			break;
		ret = -EAGAIN;
	default:
		spin_unlock_irq(&current->sighand->siglock);
		return ret;
	}

	add_wait_queue(&current->sighand->signalfd_wqh, &wait);
	for (;;) {
		set_current_state(TASK_INTERRUPTIBLE);
		ret = dequeue_signal(current, &ctx->sigmask, info);
		if (ret != 0)
			break;
		if (signal_pending(current)) {
			ret = -ERESTARTSYS;
			break;
		}
		spin_unlock_irq(&current->sighand->siglock);
		schedule();
		spin_lock_irq(&current->sighand->siglock);
	}
	spin_unlock_irq(&current->sighand->siglock);

	remove_wait_queue(&current->sighand->signalfd_wqh, &wait);
	__set_current_state(TASK_RUNNING);

	return ret;
}

/*
 * Returns a multiple of the size of a "struct signalfd_siginfo", or a negative
 * error code. The "count" parameter must be at least the size of a
 * "struct signalfd_siginfo".
 */
static ssize_t signalfd_read(struct file *file, char __user *buf, size_t count,
			     loff_t *ppos)
{
	struct signalfd_ctx *ctx = file->private_data;
	struct signalfd_siginfo __user *siginfo;
	int nonblock = file->f_flags & O_NONBLOCK;
	ssize_t ret, total = 0;
	siginfo_t info;

	count /= sizeof(struct signalfd_siginfo);
	if (!count)
		return -EINVAL;

	siginfo = (struct signalfd_siginfo __user *) buf;
	do {
		ret = signalfd_dequeue(ctx, &info, nonblock);
		if (unlikely(ret <= 0))
			break;
		ret = signalfd_copyinfo(siginfo, &info);
		if (ret < 0)
			break;
		siginfo++;
		total += ret;
		nonblock = 1;
	} while (--count);

	return total ? total: ret;
}

static const struct file_operations signalfd_fops = {
	.release	= signalfd_release,
	.poll		= signalfd_poll,
	.read		= signalfd_read,
};

SYSCALL_DEFINE4(signalfd4, int, ufd, sigset_t __user *, user_mask,
		size_t, sizemask, int, flags)
{
	sigset_t sigmask;
	struct signalfd_ctx *ctx;

	/* Check the SFD_* constants for consistency.  */
	BUILD_BUG_ON(SFD_CLOEXEC != O_CLOEXEC);
	BUILD_BUG_ON(SFD_NONBLOCK != O_NONBLOCK);

	if (flags & ~(SFD_CLOEXEC | SFD_NONBLOCK))
		return -EINVAL;

	if (sizemask != sizeof(sigset_t) ||
	    copy_from_user(&sigmask, user_mask, sizeof(sigmask)))
		return -EINVAL;
	sigdelsetmask(&sigmask, sigmask(SIGKILL) | sigmask(SIGSTOP));
	signotset(&sigmask);

	if (ufd == -1) {
		ctx = kmalloc(sizeof(*ctx), GFP_KERNEL);
		if (!ctx)
			return -ENOMEM;

		ctx->sigmask = sigmask;

		/*
		 * When we call this, the initialization must be complete, since
		 * anon_inode_getfd() will install the fd.
		 */
		ufd = anon_inode_getfd("[signalfd]", &signalfd_fops, ctx,
				       O_RDWR | (flags & (O_CLOEXEC | O_NONBLOCK)));
		if (ufd < 0)
			kfree(ctx);
	} else {
		struct file *file = fget(ufd);
		if (!file)
			return -EBADF;
		ctx = file->private_data;
		if (file->f_op != &signalfd_fops) {
			fput(file);
			return -EINVAL;
		}
		spin_lock_irq(&current->sighand->siglock);
		ctx->sigmask = sigmask;
		spin_unlock_irq(&current->sighand->siglock);

		wake_up(&current->sighand->signalfd_wqh);
		fput(file);
	}

	return ufd;
}

SYSCALL_DEFINE3(signalfd, int, ufd, sigset_t __user *, user_mask,
		size_t, sizemask)
{
	return sys_signalfd4(ufd, user_mask, sizemask, 0);
}
Commit	Line	Data
fba2afaa DL	1	/*
	2	* fs/signalfd.c
	3	*
	4	* Copyright (C) 2003 Linus Torvalds
	5	*
	6	* Mon Mar 5, 2007: Davide Libenzi <davidel@xmailserver.org>
	7	* Changed ->read() to return a siginfo strcture instead of signal number.
	8	* Fixed locking in ->poll().
	9	* Added sighand-detach notification.
	10	* Added fd re-use in sys_signalfd() syscall.
	11	* Now using anonymous inode source.
	12	* Thanks to Oleg Nesterov for useful code review and suggestions.
	13	* More comments and suggestions from Arnd Bergmann.
b8fceee1	14	* Sat May 19, 2007: Davi E. M. Arnaut <davi@haxent.com.br>
b3762bfc	15	* Retrieve multiple signals with one read() call
b8fceee1 DL	16	* Sun Jul 15, 2007: Davide Libenzi <davidel@xmailserver.org>
b8fceee1 DL	17	* Attach to the sighand only during read() and poll().
fba2afaa DL	18	*/
	19
	20	#include <linux/file.h>
	21	#include <linux/poll.h>
	22	#include <linux/init.h>
	23	#include <linux/fs.h>
	24	#include <linux/sched.h>
5a0e3ad6	25	#include <linux/slab.h>
fba2afaa DL	26	#include <linux/kernel.h>
	27	#include <linux/signal.h>
	28	#include <linux/list.h>
	29	#include <linux/anon_inodes.h>
	30	#include <linux/signalfd.h>
7ec37dfd	31	#include <linux/syscalls.h>
fba2afaa DL	32
fba2afaa DL	33	struct signalfd_ctx {
fba2afaa	34	sigset_t sigmask;
fba2afaa DL	35	};
fba2afaa DL	36
fba2afaa DL	37	static int signalfd_release(struct inode inode, struct file file)
fba2afaa DL	38	{
b8fceee1	39	kfree(file->private_data);
fba2afaa DL	40	return 0;
	41	}
	42
	43	static unsigned int signalfd_poll(struct file file, poll_table wait)
	44	{
	45	struct signalfd_ctx *ctx = file->private_data;
	46	unsigned int events = 0;
fba2afaa	47
b8fceee1	48	poll_wait(file, &current->sighand->signalfd_wqh, wait);
fba2afaa	49
b8fceee1 DL	50	spin_lock_irq(&current->sighand->siglock);
	51	if (next_signal(&current->pending, &ctx->sigmask) \|\|
	52	next_signal(&current->signal->shared_pending,
	53	&ctx->sigmask))
fba2afaa	54	events \|= POLLIN;
b8fceee1	55	spin_unlock_irq(&current->sighand->siglock);
fba2afaa DL	56
	57	return events;
	58	}
	59
	60	/*
	61	* Copied from copy_siginfo_to_user() in kernel/signal.c
	62	*/
	63	static int signalfd_copyinfo(struct signalfd_siginfo __user *uinfo,
	64	siginfo_t const *kinfo)
	65	{
	66	long err;
	67
	68	BUILD_BUG_ON(sizeof(struct signalfd_siginfo) != 128);
	69
	70	/*
14e4a0f2	71	* Unused members should be zero ...
fba2afaa DL	72	*/
	73	err = __clear_user(uinfo, sizeof(*uinfo));
	74
	75	/*
	76	* If you change siginfo_t structure, please be sure
	77	* this code is fixed accordingly.
	78	*/
96358de6 DL	79	err \|= __put_user(kinfo->si_signo, &uinfo->ssi_signo);
	80	err \|= __put_user(kinfo->si_errno, &uinfo->ssi_errno);
	81	err \|= __put_user((short) kinfo->si_code, &uinfo->ssi_code);
fba2afaa DL	82	switch (kinfo->si_code & __SI_MASK) {
fba2afaa DL	83	case __SI_KILL:
96358de6 DL	84	err \|= __put_user(kinfo->si_pid, &uinfo->ssi_pid);
96358de6 DL	85	err \|= __put_user(kinfo->si_uid, &uinfo->ssi_uid);
fba2afaa DL	86	break;
fba2afaa DL	87	case __SI_TIMER:
96358de6 DL	88	err \|= __put_user(kinfo->si_tid, &uinfo->ssi_tid);
	89	err \|= __put_user(kinfo->si_overrun, &uinfo->ssi_overrun);
	90	err \|= __put_user((long) kinfo->si_ptr, &uinfo->ssi_ptr);
a2a20c41	91	err \|= __put_user(kinfo->si_int, &uinfo->ssi_int);
fba2afaa DL	92	break;
fba2afaa DL	93	case __SI_POLL:
96358de6 DL	94	err \|= __put_user(kinfo->si_band, &uinfo->ssi_band);
96358de6 DL	95	err \|= __put_user(kinfo->si_fd, &uinfo->ssi_fd);
fba2afaa DL	96	break;
fba2afaa DL	97	case __SI_FAULT:
96358de6	98	err \|= __put_user((long) kinfo->si_addr, &uinfo->ssi_addr);
fba2afaa	99	#ifdef __ARCH_SI_TRAPNO
96358de6	100	err \|= __put_user(kinfo->si_trapno, &uinfo->ssi_trapno);
fba2afaa DL	101	#endif
	102	break;
	103	case __SI_CHLD:
96358de6 DL	104	err \|= __put_user(kinfo->si_pid, &uinfo->ssi_pid);
	105	err \|= __put_user(kinfo->si_uid, &uinfo->ssi_uid);
	106	err \|= __put_user(kinfo->si_status, &uinfo->ssi_status);
	107	err \|= __put_user(kinfo->si_utime, &uinfo->ssi_utime);
	108	err \|= __put_user(kinfo->si_stime, &uinfo->ssi_stime);
fba2afaa DL	109	break;
	110	case __SI_RT: /* This is not generated by the kernel as of now. */
	111	case __SI_MESGQ: /* But this is */
96358de6 DL	112	err \|= __put_user(kinfo->si_pid, &uinfo->ssi_pid);
	113	err \|= __put_user(kinfo->si_uid, &uinfo->ssi_uid);
	114	err \|= __put_user((long) kinfo->si_ptr, &uinfo->ssi_ptr);
a2a20c41	115	err \|= __put_user(kinfo->si_int, &uinfo->ssi_int);
fba2afaa	116	break;
0859ab59 DL	117	default:
	118	/*
	119	* This case catches also the signals queued by sigqueue().
	120	*/
96358de6 DL	121	err \|= __put_user(kinfo->si_pid, &uinfo->ssi_pid);
96358de6 DL	122	err \|= __put_user(kinfo->si_uid, &uinfo->ssi_uid);
0859ab59 DL	123	err \|= __put_user((long) kinfo->si_ptr, &uinfo->ssi_ptr);
0859ab59 DL	124	err \|= __put_user(kinfo->si_int, &uinfo->ssi_int);
fba2afaa DL	125	break;
	126	}
	127
	128	return err ? -EFAULT: sizeof(*uinfo);
	129	}
	130
b3762bfc DA	131	static ssize_t signalfd_dequeue(struct signalfd_ctx ctx, siginfo_t info,
	132	int nonblock)
	133	{
	134	ssize_t ret;
b3762bfc DA	135	DECLARE_WAITQUEUE(wait, current);
b3762bfc DA	136
b8fceee1 DL	137	spin_lock_irq(&current->sighand->siglock);
b8fceee1 DL	138	ret = dequeue_signal(current, &ctx->sigmask, info);
b3762bfc DA	139	switch (ret) {
	140	case 0:
	141	if (!nonblock)
	142	break;
	143	ret = -EAGAIN;
	144	default:
b8fceee1	145	spin_unlock_irq(&current->sighand->siglock);
b3762bfc DA	146	return ret;
	147	}
	148
b8fceee1	149	add_wait_queue(&current->sighand->signalfd_wqh, &wait);
b3762bfc DA	150	for (;;) {
b3762bfc DA	151	set_current_state(TASK_INTERRUPTIBLE);
b8fceee1	152	ret = dequeue_signal(current, &ctx->sigmask, info);
b3762bfc DA	153	if (ret != 0)
	154	break;
	155	if (signal_pending(current)) {
	156	ret = -ERESTARTSYS;
	157	break;
	158	}
b8fceee1	159	spin_unlock_irq(&current->sighand->siglock);
b3762bfc	160	schedule();
b8fceee1	161	spin_lock_irq(&current->sighand->siglock);
b3762bfc	162	}
b8fceee1	163	spin_unlock_irq(&current->sighand->siglock);
b3762bfc	164
b8fceee1	165	remove_wait_queue(&current->sighand->signalfd_wqh, &wait);
b3762bfc DA	166	__set_current_state(TASK_RUNNING);
	167
	168	return ret;
	169	}
	170
fba2afaa	171	/*
b8fceee1 DL	172	* Returns a multiple of the size of a "struct signalfd_siginfo", or a negative
	173	* error code. The "count" parameter must be at least the size of a
	174	* "struct signalfd_siginfo".
fba2afaa DL	175	*/
	176	static ssize_t signalfd_read(struct file file, char __user buf, size_t count,
	177	loff_t *ppos)
	178	{
	179	struct signalfd_ctx *ctx = file->private_data;
b3762bfc DA	180	struct signalfd_siginfo __user *siginfo;
	181	int nonblock = file->f_flags & O_NONBLOCK;
	182	ssize_t ret, total = 0;
fba2afaa	183	siginfo_t info;
fba2afaa	184
b3762bfc DA	185	count /= sizeof(struct signalfd_siginfo);
b3762bfc DA	186	if (!count)
fba2afaa	187	return -EINVAL;
fba2afaa	188
b3762bfc	189	siginfo = (struct signalfd_siginfo __user *) buf;
b3762bfc DA	190	do {
	191	ret = signalfd_dequeue(ctx, &info, nonblock);
	192	if (unlikely(ret <= 0))
	193	break;
	194	ret = signalfd_copyinfo(siginfo, &info);
	195	if (ret < 0)
	196	break;
	197	siginfo++;
	198	total += ret;
	199	nonblock = 1;
	200	} while (--count);
	201
b8fceee1	202	return total ? total: ret;
fba2afaa DL	203	}
	204
	205	static const struct file_operations signalfd_fops = {
	206	.release = signalfd_release,
	207	.poll = signalfd_poll,
	208	.read = signalfd_read,
	209	};
	210
836f92ad HC	211	SYSCALL_DEFINE4(signalfd4, int, ufd, sigset_t __user *, user_mask,
836f92ad HC	212	size_t, sizemask, int, flags)
fba2afaa	213	{
fba2afaa DL	214	sigset_t sigmask;
fba2afaa DL	215	struct signalfd_ctx *ctx;
fba2afaa	216
e38b36f3 UD	217	/* Check the SFD_* constants for consistency. */
	218	BUILD_BUG_ON(SFD_CLOEXEC != O_CLOEXEC);
	219	BUILD_BUG_ON(SFD_NONBLOCK != O_NONBLOCK);
	220
5fb5e049	221	if (flags & ~(SFD_CLOEXEC \| SFD_NONBLOCK))
9deb27ba UD	222	return -EINVAL;
9deb27ba UD	223
fba2afaa DL	224	if (sizemask != sizeof(sigset_t) \|\|
fba2afaa DL	225	copy_from_user(&sigmask, user_mask, sizeof(sigmask)))
f50cadaa	226	return -EINVAL;
fba2afaa DL	227	sigdelsetmask(&sigmask, sigmask(SIGKILL) \| sigmask(SIGSTOP));
	228	signotset(&sigmask);
	229
	230	if (ufd == -1) {
	231	ctx = kmalloc(sizeof(*ctx), GFP_KERNEL);
	232	if (!ctx)
	233	return -ENOMEM;
	234
fba2afaa	235	ctx->sigmask = sigmask;
fba2afaa DL	236
	237	/*
	238	* When we call this, the initialization must be complete, since
	239	* anon_inode_getfd() will install the fd.
	240	*/
7d9dbca3	241	ufd = anon_inode_getfd("[signalfd]", &signalfd_fops, ctx,
628ff7c1	242	O_RDWR \| (flags & (O_CLOEXEC \| O_NONBLOCK)));
2030a42c AV	243	if (ufd < 0)
2030a42c AV	244	kfree(ctx);
fba2afaa	245	} else {
2030a42c	246	struct file *file = fget(ufd);
fba2afaa DL	247	if (!file)
	248	return -EBADF;
	249	ctx = file->private_data;
	250	if (file->f_op != &signalfd_fops) {
	251	fput(file);
	252	return -EINVAL;
	253	}
b8fceee1 DL	254	spin_lock_irq(&current->sighand->siglock);
	255	ctx->sigmask = sigmask;
	256	spin_unlock_irq(&current->sighand->siglock);
	257
	258	wake_up(&current->sighand->signalfd_wqh);
fba2afaa DL	259	fput(file);
	260	}
	261
	262	return ufd;
fba2afaa	263	}
9deb27ba	264
836f92ad HC	265	SYSCALL_DEFINE3(signalfd, int, ufd, sigset_t __user *, user_mask,
836f92ad HC	266	size_t, sizemask)
9deb27ba UD	267	{
	268	return sys_signalfd4(ufd, user_mask, sizemask, 0);
	269	}