[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/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);
		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);
		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,
};

asmlinkage long sys_signalfd4(int ufd, sigset_t __user *user_mask,
			      size_t sizemask, int flags)
{
	sigset_t sigmask;
	struct signalfd_ctx *ctx;

	if (flags & ~SFD_CLOEXEC)
		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,
				       flags & O_CLOEXEC);
		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;
}

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