[linux-2.6-block.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>

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 memebers 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 is just in case for now ... */
		err |= __put_user(kinfo->si_pid, &uinfo->ssi_pid);
		err |= __put_user(kinfo->si_uid, &uinfo->ssi_uid);
		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_signalfd(int ufd, sigset_t __user *user_mask, size_t sizemask)
{
	int error;
	sigset_t sigmask;
	struct signalfd_ctx *ctx;
	struct file *file;
	struct inode *inode;

	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.
		 */
		error = anon_inode_getfd(&ufd, &inode, &file, "[signalfd]",
					 &signalfd_fops, ctx);
		if (error)
			goto err_fdalloc;
	} else {
		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;

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