[deliverable/linux.git] / fs / notify / fanotify / fanotify_user.c

#include <linux/fanotify.h>
#include <linux/fcntl.h>
#include <linux/file.h>
#include <linux/fs.h>
#include <linux/anon_inodes.h>
#include <linux/fsnotify_backend.h>
#include <linux/init.h>
#include <linux/mount.h>
#include <linux/namei.h>
#include <linux/poll.h>
#include <linux/security.h>
#include <linux/syscalls.h>
#include <linux/types.h>
#include <linux/uaccess.h>

#include <asm/ioctls.h>

extern const struct fsnotify_ops fanotify_fsnotify_ops;

static struct kmem_cache *fanotify_mark_cache __read_mostly;

/*
 * Get an fsnotify notification event if one exists and is small
 * enough to fit in "count". Return an error pointer if the count
 * is not large enough.
 *
 * Called with the group->notification_mutex held.
 */
static struct fsnotify_event *get_one_event(struct fsnotify_group *group,
					    size_t count)
{
	BUG_ON(!mutex_is_locked(&group->notification_mutex));

	pr_debug("%s: group=%p count=%zd\n", __func__, group, count);

	if (fsnotify_notify_queue_is_empty(group))
		return NULL;

	if (FAN_EVENT_METADATA_LEN > count)
		return ERR_PTR(-EINVAL);

	/* held the notification_mutex the whole time, so this is the
	 * same event we peeked above */
	return fsnotify_remove_notify_event(group);
}

static int create_fd(struct fsnotify_group *group, struct fsnotify_event *event)
{
	int client_fd;
	struct dentry *dentry;
	struct vfsmount *mnt;
	struct file *new_file;

	pr_debug("%s: group=%p event=%p\n", __func__, group, event);

	client_fd = get_unused_fd();
	if (client_fd < 0)
		return client_fd;

	if (event->data_type != FSNOTIFY_EVENT_PATH) {
		WARN_ON(1);
		put_unused_fd(client_fd);
		return -EINVAL;
	}

	/*
	 * we need a new file handle for the userspace program so it can read even if it was
	 * originally opened O_WRONLY.
	 */
	dentry = dget(event->path.dentry);
	mnt = mntget(event->path.mnt);
	/* it's possible this event was an overflow event.  in that case dentry and mnt
	 * are NULL;  That's fine, just don't call dentry open */
	if (dentry && mnt)
		new_file = dentry_open(dentry, mnt,
				       O_RDONLY | O_LARGEFILE | FMODE_NONOTIFY,
				       current_cred());
	else
		new_file = ERR_PTR(-EOVERFLOW);
	if (IS_ERR(new_file)) {
		/*
		 * we still send an event even if we can't open the file.  this
		 * can happen when say tasks are gone and we try to open their
		 * /proc files or we try to open a WRONLY file like in sysfs
		 * we just send the errno to userspace since there isn't much
		 * else we can do.
		 */
		put_unused_fd(client_fd);
		client_fd = PTR_ERR(new_file);
	} else {
		fd_install(client_fd, new_file);
	}

	return client_fd;
}

static ssize_t fill_event_metadata(struct fsnotify_group *group,
				   struct fanotify_event_metadata *metadata,
				   struct fsnotify_event *event)
{
	pr_debug("%s: group=%p metadata=%p event=%p\n", __func__,
		 group, metadata, event);

	metadata->event_len = FAN_EVENT_METADATA_LEN;
	metadata->vers = FANOTIFY_METADATA_VERSION;
	metadata->mask = event->mask & FAN_ALL_OUTGOING_EVENTS;
	metadata->pid = pid_vnr(event->tgid);
	metadata->fd = create_fd(group, event);

	return metadata->fd;
}

static ssize_t copy_event_to_user(struct fsnotify_group *group,
				  struct fsnotify_event *event,
				  char __user *buf)
{
	struct fanotify_event_metadata fanotify_event_metadata;
	int ret;

	pr_debug("%s: group=%p event=%p\n", __func__, group, event);

	ret = fill_event_metadata(group, &fanotify_event_metadata, event);
	if (ret < 0)
		return ret;

	if (copy_to_user(buf, &fanotify_event_metadata, FAN_EVENT_METADATA_LEN))
		return -EFAULT;

	return FAN_EVENT_METADATA_LEN;
}

/* intofiy userspace file descriptor functions */
static unsigned int fanotify_poll(struct file *file, poll_table *wait)
{
	struct fsnotify_group *group = file->private_data;
	int ret = 0;

	poll_wait(file, &group->notification_waitq, wait);
	mutex_lock(&group->notification_mutex);
	if (!fsnotify_notify_queue_is_empty(group))
		ret = POLLIN | POLLRDNORM;
	mutex_unlock(&group->notification_mutex);

	return ret;
}

static ssize_t fanotify_read(struct file *file, char __user *buf,
			     size_t count, loff_t *pos)
{
	struct fsnotify_group *group;
	struct fsnotify_event *kevent;
	char __user *start;
	int ret;
	DEFINE_WAIT(wait);

	start = buf;
	group = file->private_data;

	pr_debug("%s: group=%p\n", __func__, group);

	while (1) {
		prepare_to_wait(&group->notification_waitq, &wait, TASK_INTERRUPTIBLE);

		mutex_lock(&group->notification_mutex);
		kevent = get_one_event(group, count);
		mutex_unlock(&group->notification_mutex);

		if (kevent) {
			ret = PTR_ERR(kevent);
			if (IS_ERR(kevent))
				break;
			ret = copy_event_to_user(group, kevent, buf);
			fsnotify_put_event(kevent);
			if (ret < 0)
				break;
			buf += ret;
			count -= ret;
			continue;
		}

		ret = -EAGAIN;
		if (file->f_flags & O_NONBLOCK)
			break;
		ret = -EINTR;
		if (signal_pending(current))
			break;

		if (start != buf)
			break;

		schedule();
	}

	finish_wait(&group->notification_waitq, &wait);
	if (start != buf && ret != -EFAULT)
		ret = buf - start;
	return ret;
}

static int fanotify_release(struct inode *ignored, struct file *file)
{
	struct fsnotify_group *group = file->private_data;

	pr_debug("%s: file=%p group=%p\n", __func__, file, group);

	/* matches the fanotify_init->fsnotify_alloc_group */
	fsnotify_put_group(group);

	return 0;
}

static long fanotify_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
{
	struct fsnotify_group *group;
	struct fsnotify_event_holder *holder;
	void __user *p;
	int ret = -ENOTTY;
	size_t send_len = 0;

	group = file->private_data;

	p = (void __user *) arg;

	switch (cmd) {
	case FIONREAD:
		mutex_lock(&group->notification_mutex);
		list_for_each_entry(holder, &group->notification_list, event_list)
			send_len += FAN_EVENT_METADATA_LEN;
		mutex_unlock(&group->notification_mutex);
		ret = put_user(send_len, (int __user *) p);
		break;
	}

	return ret;
}

static const struct file_operations fanotify_fops = {
	.poll		= fanotify_poll,
	.read		= fanotify_read,
	.fasync		= NULL,
	.release	= fanotify_release,
	.unlocked_ioctl	= fanotify_ioctl,
	.compat_ioctl	= fanotify_ioctl,
};

static void fanotify_free_mark(struct fsnotify_mark *fsn_mark)
{
	kmem_cache_free(fanotify_mark_cache, fsn_mark);
}

static int fanotify_find_path(int dfd, const char __user *filename,
			      struct path *path, unsigned int flags)
{
	int ret;

	pr_debug("%s: dfd=%d filename=%p flags=%x\n", __func__,
		 dfd, filename, flags);

	if (filename == NULL) {
		struct file *file;
		int fput_needed;

		ret = -EBADF;
		file = fget_light(dfd, &fput_needed);
		if (!file)
			goto out;

		ret = -ENOTDIR;
		if ((flags & FAN_MARK_ONLYDIR) &&
		    !(S_ISDIR(file->f_path.dentry->d_inode->i_mode))) {
			fput_light(file, fput_needed);
			goto out;
		}

		*path = file->f_path;
		path_get(path);
		fput_light(file, fput_needed);
	} else {
		unsigned int lookup_flags = 0;

		if (!(flags & FAN_MARK_DONT_FOLLOW))
			lookup_flags |= LOOKUP_FOLLOW;
		if (flags & FAN_MARK_ONLYDIR)
			lookup_flags |= LOOKUP_DIRECTORY;

		ret = user_path_at(dfd, filename, lookup_flags, path);
		if (ret)
			goto out;
	}

	/* you can only watch an inode if you have read permissions on it */
	ret = inode_permission(path->dentry->d_inode, MAY_READ);
	if (ret)
		path_put(path);
out:
	return ret;
}

static __u32 fanotify_mark_remove_from_mask(struct fsnotify_mark *fsn_mark,
					    __u32 mask,
					    unsigned int flags)
{
	__u32 oldmask;

	spin_lock(&fsn_mark->lock);
	if (!(flags & FAN_MARK_IGNORED_MASK)) {
		oldmask = fsn_mark->mask;
		fsnotify_set_mark_mask_locked(fsn_mark, (oldmask & ~mask));
	} else {
		oldmask = fsn_mark->ignored_mask;
		fsnotify_set_mark_ignored_mask_locked(fsn_mark, (oldmask & ~mask));
	}
	spin_unlock(&fsn_mark->lock);

	if (!(oldmask & ~mask))
		fsnotify_destroy_mark(fsn_mark);

	return mask & oldmask;
}

static int fanotify_remove_vfsmount_mark(struct fsnotify_group *group,
					 struct vfsmount *mnt, __u32 mask,
					 unsigned int flags)
{
	struct fsnotify_mark *fsn_mark = NULL;
	__u32 removed;

	fsn_mark = fsnotify_find_vfsmount_mark(group, mnt);
	if (!fsn_mark)
		return -ENOENT;

	removed = fanotify_mark_remove_from_mask(fsn_mark, mask, flags);
	fsnotify_put_mark(fsn_mark);
	if (removed & group->mask)
		fsnotify_recalc_group_mask(group);
	if (removed & mnt->mnt_fsnotify_mask)
		fsnotify_recalc_vfsmount_mask(mnt);

	return 0;
}

static int fanotify_remove_inode_mark(struct fsnotify_group *group,
				      struct inode *inode, __u32 mask,
				      unsigned int flags)
{
	struct fsnotify_mark *fsn_mark = NULL;
	__u32 removed;

	fsn_mark = fsnotify_find_inode_mark(group, inode);
	if (!fsn_mark)
		return -ENOENT;

	removed = fanotify_mark_remove_from_mask(fsn_mark, mask, flags);
	/* matches the fsnotify_find_inode_mark() */
	fsnotify_put_mark(fsn_mark);

	if (removed & group->mask)
		fsnotify_recalc_group_mask(group);
	if (removed & inode->i_fsnotify_mask)
		fsnotify_recalc_inode_mask(inode);

	return 0;
}

static __u32 fanotify_mark_add_to_mask(struct fsnotify_mark *fsn_mark,
				       __u32 mask,
				       unsigned int flags)
{
	__u32 oldmask;

	spin_lock(&fsn_mark->lock);
	if (!(flags & FAN_MARK_IGNORED_MASK)) {
		oldmask = fsn_mark->mask;
		fsnotify_set_mark_mask_locked(fsn_mark, (oldmask | mask));
	} else {
		oldmask = fsn_mark->ignored_mask;
		fsnotify_set_mark_ignored_mask_locked(fsn_mark, (oldmask | mask));
		if (flags & FAN_MARK_IGNORED_SURV_MODIFY)
			fsn_mark->flags |= FSNOTIFY_MARK_FLAG_IGNORED_SURV_MODIFY;
	}
	spin_unlock(&fsn_mark->lock);

	return mask & ~oldmask;
}

static int fanotify_add_vfsmount_mark(struct fsnotify_group *group,
				      struct vfsmount *mnt, __u32 mask,
				      unsigned int flags)
{
	struct fsnotify_mark *fsn_mark;
	__u32 added;

	fsn_mark = fsnotify_find_vfsmount_mark(group, mnt);
	if (!fsn_mark) {
		int ret;

		fsn_mark = kmem_cache_alloc(fanotify_mark_cache, GFP_KERNEL);
		if (!fsn_mark)
			return -ENOMEM;

		fsnotify_init_mark(fsn_mark, fanotify_free_mark);
		ret = fsnotify_add_mark(fsn_mark, group, NULL, mnt, 0);
		if (ret) {
			fanotify_free_mark(fsn_mark);
			return ret;
		}
	}
	added = fanotify_mark_add_to_mask(fsn_mark, mask, flags);
	fsnotify_put_mark(fsn_mark);
	if (added) {
		if (added & ~group->mask)
			fsnotify_recalc_group_mask(group);
		if (added & ~mnt->mnt_fsnotify_mask)
			fsnotify_recalc_vfsmount_mask(mnt);
	}
	return 0;
}

static int fanotify_add_inode_mark(struct fsnotify_group *group,
				   struct inode *inode, __u32 mask,
				   unsigned int flags)
{
	struct fsnotify_mark *fsn_mark;
	__u32 added;

	pr_debug("%s: group=%p inode=%p\n", __func__, group, inode);

	fsn_mark = fsnotify_find_inode_mark(group, inode);
	if (!fsn_mark) {
		int ret;

		fsn_mark = kmem_cache_alloc(fanotify_mark_cache, GFP_KERNEL);
		if (!fsn_mark)
			return -ENOMEM;

		fsnotify_init_mark(fsn_mark, fanotify_free_mark);
		ret = fsnotify_add_mark(fsn_mark, group, inode, NULL, 0);
		if (ret) {
			fanotify_free_mark(fsn_mark);
			return ret;
		}
	}
	added = fanotify_mark_add_to_mask(fsn_mark, mask, flags);
	fsnotify_put_mark(fsn_mark);
	if (added) {
		if (added & ~group->mask)
			fsnotify_recalc_group_mask(group);
		if (added & ~inode->i_fsnotify_mask)
			fsnotify_recalc_inode_mask(inode);
	}
	return 0;
}

/* fanotify syscalls */
SYSCALL_DEFINE3(fanotify_init, unsigned int, flags, unsigned int, event_f_flags,
		unsigned int, priority)
{
	struct fsnotify_group *group;
	int f_flags, fd;

	pr_debug("%s: flags=%d event_f_flags=%d priority=%d\n",
		__func__, flags, event_f_flags, priority);

	if (event_f_flags)
		return -EINVAL;
	if (priority)
		return -EINVAL;

	if (!capable(CAP_SYS_ADMIN))
		return -EACCES;

	if (flags & ~FAN_ALL_INIT_FLAGS)
		return -EINVAL;

	f_flags = (O_RDONLY | FMODE_NONOTIFY);
	if (flags & FAN_CLOEXEC)
		f_flags |= O_CLOEXEC;
	if (flags & FAN_NONBLOCK)
		f_flags |= O_NONBLOCK;

	/* fsnotify_alloc_group takes a ref.  Dropped in fanotify_release */
	group = fsnotify_alloc_group(&fanotify_fsnotify_ops);
	if (IS_ERR(group))
		return PTR_ERR(group);

	fd = anon_inode_getfd("[fanotify]", &fanotify_fops, group, f_flags);
	if (fd < 0)
		goto out_put_group;

	return fd;

out_put_group:
	fsnotify_put_group(group);
	return fd;
}

SYSCALL_DEFINE(fanotify_mark)(int fanotify_fd, unsigned int flags,
			      __u64 mask, int dfd,
			      const char  __user * pathname)
{
	struct inode *inode = NULL;
	struct vfsmount *mnt = NULL;
	struct fsnotify_group *group;
	struct file *filp;
	struct path path;
	int ret, fput_needed;

	pr_debug("%s: fanotify_fd=%d flags=%x dfd=%d pathname=%p mask=%llx\n",
		 __func__, fanotify_fd, flags, dfd, pathname, mask);

	/* we only use the lower 32 bits as of right now. */
	if (mask & ((__u64)0xffffffff << 32))
		return -EINVAL;

	if (flags & ~FAN_ALL_MARK_FLAGS)
		return -EINVAL;
	switch (flags & (FAN_MARK_ADD | FAN_MARK_REMOVE | FAN_MARK_FLUSH)) {
	case FAN_MARK_ADD:
	case FAN_MARK_REMOVE:
	case FAN_MARK_FLUSH:
		break;
	default:
		return -EINVAL;
	}
	if (mask & ~(FAN_ALL_EVENTS | FAN_EVENT_ON_CHILD))
		return -EINVAL;

	filp = fget_light(fanotify_fd, &fput_needed);
	if (unlikely(!filp))
		return -EBADF;

	/* verify that this is indeed an fanotify instance */
	ret = -EINVAL;
	if (unlikely(filp->f_op != &fanotify_fops))
		goto fput_and_out;

	ret = fanotify_find_path(dfd, pathname, &path, flags);
	if (ret)
		goto fput_and_out;

	/* inode held in place by reference to path; group by fget on fd */
	if (!(flags & FAN_MARK_MOUNT))
		inode = path.dentry->d_inode;
	else
		mnt = path.mnt;
	group = filp->private_data;

	/* create/update an inode mark */
	switch (flags & (FAN_MARK_ADD | FAN_MARK_REMOVE | FAN_MARK_FLUSH)) {
	case FAN_MARK_ADD:
		if (flags & FAN_MARK_MOUNT)
			ret = fanotify_add_vfsmount_mark(group, mnt, mask, flags);
		else
			ret = fanotify_add_inode_mark(group, inode, mask, flags);
		break;
	case FAN_MARK_REMOVE:
		if (flags & FAN_MARK_MOUNT)
			ret = fanotify_remove_vfsmount_mark(group, mnt, mask, flags);
		else
			ret = fanotify_remove_inode_mark(group, inode, mask, flags);
		break;
	case FAN_MARK_FLUSH:
		if (flags & FAN_MARK_MOUNT)
			fsnotify_clear_vfsmount_marks_by_group(group);
		else
			fsnotify_clear_inode_marks_by_group(group);
		fsnotify_recalc_group_mask(group);
		break;
	default:
		ret = -EINVAL;
	}

	path_put(&path);
fput_and_out:
	fput_light(filp, fput_needed);
	return ret;
}

#ifdef CONFIG_HAVE_SYSCALL_WRAPPERS
asmlinkage long SyS_fanotify_mark(long fanotify_fd, long flags, __u64 mask,
				  long dfd, long pathname)
{
	return SYSC_fanotify_mark((int) fanotify_fd, (unsigned int) flags,
				  mask, (int) dfd,
				  (const char  __user *) pathname);
}
SYSCALL_ALIAS(sys_fanotify_mark, SyS_fanotify_mark);
#endif

/*
 * fanotify_user_setup - Our initialization function.  Note that we cannnot return
 * error because we have compiled-in VFS hooks.  So an (unlikely) failure here
 * must result in panic().
 */
static int __init fanotify_user_setup(void)
{
	fanotify_mark_cache = KMEM_CACHE(fsnotify_mark, SLAB_PANIC);

	return 0;
}
device_initcall(fanotify_user_setup);
Commit	Line	Data
	1	#include <linux/fanotify.h>
	2	#include <linux/fcntl.h>
	3	#include <linux/file.h>
	4	#include <linux/fs.h>
	5	#include <linux/anon_inodes.h>
	6	#include <linux/fsnotify_backend.h>
	7	#include <linux/init.h>
	8	#include <linux/mount.h>
	9	#include <linux/namei.h>
	10	#include <linux/poll.h>
	11	#include <linux/security.h>
	12	#include <linux/syscalls.h>
	13	#include <linux/types.h>
	14	#include <linux/uaccess.h>
	15
	16	#include <asm/ioctls.h>
	17
	18	extern const struct fsnotify_ops fanotify_fsnotify_ops;
	19
	20	static struct kmem_cache *fanotify_mark_cache __read_mostly;
	21
	22	/*
	23	* Get an fsnotify notification event if one exists and is small
	24	* enough to fit in "count". Return an error pointer if the count
	25	* is not large enough.
	26	*
	27	* Called with the group->notification_mutex held.
	28	*/
	29	static struct fsnotify_event get_one_event(struct fsnotify_group group,
	30	size_t count)
	31	{
	32	BUG_ON(!mutex_is_locked(&group->notification_mutex));
	33
	34	pr_debug("%s: group=%p count=%zd\n", __func__, group, count);
	35
	36	if (fsnotify_notify_queue_is_empty(group))
	37	return NULL;
	38
	39	if (FAN_EVENT_METADATA_LEN > count)
	40	return ERR_PTR(-EINVAL);
	41
	42	/* held the notification_mutex the whole time, so this is the
	43	* same event we peeked above */
	44	return fsnotify_remove_notify_event(group);
	45	}
	46
	47	static int create_fd(struct fsnotify_group group, struct fsnotify_event event)
	48	{
	49	int client_fd;
	50	struct dentry *dentry;
	51	struct vfsmount *mnt;
	52	struct file *new_file;
	53
	54	pr_debug("%s: group=%p event=%p\n", __func__, group, event);
	55
	56	client_fd = get_unused_fd();
	57	if (client_fd < 0)
	58	return client_fd;
	59
	60	if (event->data_type != FSNOTIFY_EVENT_PATH) {
	61	WARN_ON(1);
	62	put_unused_fd(client_fd);
	63	return -EINVAL;
	64	}
	65
	66	/*
	67	* we need a new file handle for the userspace program so it can read even if it was
	68	* originally opened O_WRONLY.
	69	*/
	70	dentry = dget(event->path.dentry);
	71	mnt = mntget(event->path.mnt);
	72	/* it's possible this event was an overflow event. in that case dentry and mnt
	73	* are NULL; That's fine, just don't call dentry open */
	74	if (dentry && mnt)
	75	new_file = dentry_open(dentry, mnt,
	76	O_RDONLY \| O_LARGEFILE \| FMODE_NONOTIFY,
	77	current_cred());
	78	else
	79	new_file = ERR_PTR(-EOVERFLOW);
	80	if (IS_ERR(new_file)) {
	81	/*
	82	* we still send an event even if we can't open the file. this
	83	* can happen when say tasks are gone and we try to open their
	84	* /proc files or we try to open a WRONLY file like in sysfs
	85	* we just send the errno to userspace since there isn't much
	86	* else we can do.
	87	*/
	88	put_unused_fd(client_fd);
	89	client_fd = PTR_ERR(new_file);
	90	} else {
	91	fd_install(client_fd, new_file);
	92	}
	93
	94	return client_fd;
	95	}
	96
	97	static ssize_t fill_event_metadata(struct fsnotify_group *group,
	98	struct fanotify_event_metadata *metadata,
	99	struct fsnotify_event *event)
	100	{
	101	pr_debug("%s: group=%p metadata=%p event=%p\n", __func__,
	102	group, metadata, event);
	103
	104	metadata->event_len = FAN_EVENT_METADATA_LEN;
	105	metadata->vers = FANOTIFY_METADATA_VERSION;
	106	metadata->mask = event->mask & FAN_ALL_OUTGOING_EVENTS;
	107	metadata->pid = pid_vnr(event->tgid);
	108	metadata->fd = create_fd(group, event);
	109
	110	return metadata->fd;
	111	}
	112
	113	static ssize_t copy_event_to_user(struct fsnotify_group *group,
	114	struct fsnotify_event *event,
	115	char __user *buf)
	116	{
	117	struct fanotify_event_metadata fanotify_event_metadata;
	118	int ret;
	119
	120	pr_debug("%s: group=%p event=%p\n", __func__, group, event);
	121
	122	ret = fill_event_metadata(group, &fanotify_event_metadata, event);
	123	if (ret < 0)
	124	return ret;
	125
	126	if (copy_to_user(buf, &fanotify_event_metadata, FAN_EVENT_METADATA_LEN))
	127	return -EFAULT;
	128
	129	return FAN_EVENT_METADATA_LEN;
	130	}
	131
	132	/* intofiy userspace file descriptor functions */
	133	static unsigned int fanotify_poll(struct file file, poll_table wait)
	134	{
	135	struct fsnotify_group *group = file->private_data;
	136	int ret = 0;
	137
	138	poll_wait(file, &group->notification_waitq, wait);
	139	mutex_lock(&group->notification_mutex);
	140	if (!fsnotify_notify_queue_is_empty(group))
	141	ret = POLLIN \| POLLRDNORM;
	142	mutex_unlock(&group->notification_mutex);
	143
	144	return ret;
	145	}
	146
	147	static ssize_t fanotify_read(struct file file, char __user buf,
	148	size_t count, loff_t *pos)
	149	{
	150	struct fsnotify_group *group;
	151	struct fsnotify_event *kevent;
	152	char __user *start;
	153	int ret;
	154	DEFINE_WAIT(wait);
	155
	156	start = buf;
	157	group = file->private_data;
	158
	159	pr_debug("%s: group=%p\n", __func__, group);
	160
	161	while (1) {
	162	prepare_to_wait(&group->notification_waitq, &wait, TASK_INTERRUPTIBLE);
	163
	164	mutex_lock(&group->notification_mutex);
	165	kevent = get_one_event(group, count);
	166	mutex_unlock(&group->notification_mutex);
	167
	168	if (kevent) {
	169	ret = PTR_ERR(kevent);
	170	if (IS_ERR(kevent))
	171	break;
	172	ret = copy_event_to_user(group, kevent, buf);
	173	fsnotify_put_event(kevent);
	174	if (ret < 0)
	175	break;
	176	buf += ret;
	177	count -= ret;
	178	continue;
	179	}
	180
	181	ret = -EAGAIN;
	182	if (file->f_flags & O_NONBLOCK)
	183	break;
	184	ret = -EINTR;
	185	if (signal_pending(current))
	186	break;
	187
	188	if (start != buf)
	189	break;
	190
	191	schedule();
	192	}
	193
	194	finish_wait(&group->notification_waitq, &wait);
	195	if (start != buf && ret != -EFAULT)
	196	ret = buf - start;
	197	return ret;
	198	}
	199
	200	static int fanotify_release(struct inode ignored, struct file file)
	201	{
	202	struct fsnotify_group *group = file->private_data;
	203
	204	pr_debug("%s: file=%p group=%p\n", __func__, file, group);
	205
	206	/* matches the fanotify_init->fsnotify_alloc_group */
	207	fsnotify_put_group(group);
	208
	209	return 0;
	210	}
	211
	212	static long fanotify_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
	213	{
	214	struct fsnotify_group *group;
	215	struct fsnotify_event_holder *holder;
	216	void __user *p;
	217	int ret = -ENOTTY;
	218	size_t send_len = 0;
	219
	220	group = file->private_data;
	221
	222	p = (void __user *) arg;
	223
	224	switch (cmd) {
	225	case FIONREAD:
	226	mutex_lock(&group->notification_mutex);
	227	list_for_each_entry(holder, &group->notification_list, event_list)
	228	send_len += FAN_EVENT_METADATA_LEN;
	229	mutex_unlock(&group->notification_mutex);
	230	ret = put_user(send_len, (int __user *) p);
	231	break;
	232	}
	233
	234	return ret;
	235	}
	236
	237	static const struct file_operations fanotify_fops = {
	238	.poll = fanotify_poll,
	239	.read = fanotify_read,
	240	.fasync = NULL,
	241	.release = fanotify_release,
	242	.unlocked_ioctl = fanotify_ioctl,
	243	.compat_ioctl = fanotify_ioctl,
	244	};
	245
	246	static void fanotify_free_mark(struct fsnotify_mark *fsn_mark)
	247	{
	248	kmem_cache_free(fanotify_mark_cache, fsn_mark);
	249	}
	250
	251	static int fanotify_find_path(int dfd, const char __user *filename,
	252	struct path *path, unsigned int flags)
	253	{
	254	int ret;
	255
	256	pr_debug("%s: dfd=%d filename=%p flags=%x\n", __func__,
	257	dfd, filename, flags);
	258
	259	if (filename == NULL) {
	260	struct file *file;
	261	int fput_needed;
	262
	263	ret = -EBADF;
	264	file = fget_light(dfd, &fput_needed);
	265	if (!file)
	266	goto out;
	267
	268	ret = -ENOTDIR;
	269	if ((flags & FAN_MARK_ONLYDIR) &&
	270	!(S_ISDIR(file->f_path.dentry->d_inode->i_mode))) {
	271	fput_light(file, fput_needed);
	272	goto out;
	273	}
	274
	275	*path = file->f_path;
	276	path_get(path);
	277	fput_light(file, fput_needed);
	278	} else {
	279	unsigned int lookup_flags = 0;
	280
	281	if (!(flags & FAN_MARK_DONT_FOLLOW))
	282	lookup_flags \|= LOOKUP_FOLLOW;
	283	if (flags & FAN_MARK_ONLYDIR)
	284	lookup_flags \|= LOOKUP_DIRECTORY;
	285
	286	ret = user_path_at(dfd, filename, lookup_flags, path);
	287	if (ret)
	288	goto out;
	289	}
	290
	291	/* you can only watch an inode if you have read permissions on it */
	292	ret = inode_permission(path->dentry->d_inode, MAY_READ);
	293	if (ret)
	294	path_put(path);
	295	out:
	296	return ret;
	297	}
	298
	299	static __u32 fanotify_mark_remove_from_mask(struct fsnotify_mark *fsn_mark,
	300	__u32 mask,
	301	unsigned int flags)
	302	{
	303	__u32 oldmask;
	304
	305	spin_lock(&fsn_mark->lock);
	306	if (!(flags & FAN_MARK_IGNORED_MASK)) {
	307	oldmask = fsn_mark->mask;
	308	fsnotify_set_mark_mask_locked(fsn_mark, (oldmask & ~mask));
	309	} else {
	310	oldmask = fsn_mark->ignored_mask;
	311	fsnotify_set_mark_ignored_mask_locked(fsn_mark, (oldmask & ~mask));
	312	}
	313	spin_unlock(&fsn_mark->lock);
	314
	315	if (!(oldmask & ~mask))
	316	fsnotify_destroy_mark(fsn_mark);
	317
	318	return mask & oldmask;
	319	}
	320
	321	static int fanotify_remove_vfsmount_mark(struct fsnotify_group *group,
	322	struct vfsmount *mnt, __u32 mask,
	323	unsigned int flags)
	324	{
	325	struct fsnotify_mark *fsn_mark = NULL;
	326	__u32 removed;
	327
	328	fsn_mark = fsnotify_find_vfsmount_mark(group, mnt);
	329	if (!fsn_mark)
	330	return -ENOENT;
	331
	332	removed = fanotify_mark_remove_from_mask(fsn_mark, mask, flags);
	333	fsnotify_put_mark(fsn_mark);
	334	if (removed & group->mask)
	335	fsnotify_recalc_group_mask(group);
	336	if (removed & mnt->mnt_fsnotify_mask)
	337	fsnotify_recalc_vfsmount_mask(mnt);
	338
	339	return 0;
	340	}
	341
	342	static int fanotify_remove_inode_mark(struct fsnotify_group *group,
	343	struct inode *inode, __u32 mask,
	344	unsigned int flags)
	345	{
	346	struct fsnotify_mark *fsn_mark = NULL;
	347	__u32 removed;
	348
	349	fsn_mark = fsnotify_find_inode_mark(group, inode);
	350	if (!fsn_mark)
	351	return -ENOENT;
	352
	353	removed = fanotify_mark_remove_from_mask(fsn_mark, mask, flags);
	354	/* matches the fsnotify_find_inode_mark() */
	355	fsnotify_put_mark(fsn_mark);
	356
	357	if (removed & group->mask)
	358	fsnotify_recalc_group_mask(group);
	359	if (removed & inode->i_fsnotify_mask)
	360	fsnotify_recalc_inode_mask(inode);
	361
	362	return 0;
	363	}
	364
	365	static __u32 fanotify_mark_add_to_mask(struct fsnotify_mark *fsn_mark,
	366	__u32 mask,
	367	unsigned int flags)
	368	{
	369	__u32 oldmask;
	370
	371	spin_lock(&fsn_mark->lock);
	372	if (!(flags & FAN_MARK_IGNORED_MASK)) {
	373	oldmask = fsn_mark->mask;
	374	fsnotify_set_mark_mask_locked(fsn_mark, (oldmask \| mask));
	375	} else {
	376	oldmask = fsn_mark->ignored_mask;
	377	fsnotify_set_mark_ignored_mask_locked(fsn_mark, (oldmask \| mask));
	378	if (flags & FAN_MARK_IGNORED_SURV_MODIFY)
	379	fsn_mark->flags \|= FSNOTIFY_MARK_FLAG_IGNORED_SURV_MODIFY;
	380	}
	381	spin_unlock(&fsn_mark->lock);
	382
	383	return mask & ~oldmask;
	384	}
	385
	386	static int fanotify_add_vfsmount_mark(struct fsnotify_group *group,
	387	struct vfsmount *mnt, __u32 mask,
	388	unsigned int flags)
	389	{
	390	struct fsnotify_mark *fsn_mark;
	391	__u32 added;
	392
	393	fsn_mark = fsnotify_find_vfsmount_mark(group, mnt);
	394	if (!fsn_mark) {
	395	int ret;
	396
	397	fsn_mark = kmem_cache_alloc(fanotify_mark_cache, GFP_KERNEL);
	398	if (!fsn_mark)
	399	return -ENOMEM;
	400
	401	fsnotify_init_mark(fsn_mark, fanotify_free_mark);
	402	ret = fsnotify_add_mark(fsn_mark, group, NULL, mnt, 0);
	403	if (ret) {
	404	fanotify_free_mark(fsn_mark);
	405	return ret;
	406	}
	407	}
	408	added = fanotify_mark_add_to_mask(fsn_mark, mask, flags);
	409	fsnotify_put_mark(fsn_mark);
	410	if (added) {
	411	if (added & ~group->mask)
	412	fsnotify_recalc_group_mask(group);
	413	if (added & ~mnt->mnt_fsnotify_mask)
	414	fsnotify_recalc_vfsmount_mask(mnt);
	415	}
	416	return 0;
	417	}
	418
	419	static int fanotify_add_inode_mark(struct fsnotify_group *group,
	420	struct inode *inode, __u32 mask,
	421	unsigned int flags)
	422	{
	423	struct fsnotify_mark *fsn_mark;
	424	__u32 added;
	425
	426	pr_debug("%s: group=%p inode=%p\n", __func__, group, inode);
	427
	428	fsn_mark = fsnotify_find_inode_mark(group, inode);
	429	if (!fsn_mark) {
	430	int ret;
	431
	432	fsn_mark = kmem_cache_alloc(fanotify_mark_cache, GFP_KERNEL);
	433	if (!fsn_mark)
	434	return -ENOMEM;
	435
	436	fsnotify_init_mark(fsn_mark, fanotify_free_mark);
	437	ret = fsnotify_add_mark(fsn_mark, group, inode, NULL, 0);
	438	if (ret) {
	439	fanotify_free_mark(fsn_mark);
	440	return ret;
	441	}
	442	}
	443	added = fanotify_mark_add_to_mask(fsn_mark, mask, flags);
	444	fsnotify_put_mark(fsn_mark);
	445	if (added) {
	446	if (added & ~group->mask)
	447	fsnotify_recalc_group_mask(group);
	448	if (added & ~inode->i_fsnotify_mask)
	449	fsnotify_recalc_inode_mask(inode);
	450	}
	451	return 0;
	452	}
	453
	454	/* fanotify syscalls */
	455	SYSCALL_DEFINE3(fanotify_init, unsigned int, flags, unsigned int, event_f_flags,
	456	unsigned int, priority)
	457	{
	458	struct fsnotify_group *group;
	459	int f_flags, fd;
	460
	461	pr_debug("%s: flags=%d event_f_flags=%d priority=%d\n",
	462	__func__, flags, event_f_flags, priority);
	463
	464	if (event_f_flags)
	465	return -EINVAL;
	466	if (priority)
	467	return -EINVAL;
	468
	469	if (!capable(CAP_SYS_ADMIN))
	470	return -EACCES;
	471
	472	if (flags & ~FAN_ALL_INIT_FLAGS)
	473	return -EINVAL;
	474
	475	f_flags = (O_RDONLY \| FMODE_NONOTIFY);
	476	if (flags & FAN_CLOEXEC)
	477	f_flags \|= O_CLOEXEC;
	478	if (flags & FAN_NONBLOCK)
	479	f_flags \|= O_NONBLOCK;
	480
	481	/* fsnotify_alloc_group takes a ref. Dropped in fanotify_release */
	482	group = fsnotify_alloc_group(&fanotify_fsnotify_ops);
	483	if (IS_ERR(group))
	484	return PTR_ERR(group);
	485
	486	fd = anon_inode_getfd("[fanotify]", &fanotify_fops, group, f_flags);
	487	if (fd < 0)
	488	goto out_put_group;
	489
	490	return fd;
	491
	492	out_put_group:
	493	fsnotify_put_group(group);
	494	return fd;
	495	}
	496
	497	SYSCALL_DEFINE(fanotify_mark)(int fanotify_fd, unsigned int flags,
	498	__u64 mask, int dfd,
	499	const char __user * pathname)
	500	{
	501	struct inode *inode = NULL;
	502	struct vfsmount *mnt = NULL;
	503	struct fsnotify_group *group;
	504	struct file *filp;
	505	struct path path;
	506	int ret, fput_needed;
	507
	508	pr_debug("%s: fanotify_fd=%d flags=%x dfd=%d pathname=%p mask=%llx\n",
	509	__func__, fanotify_fd, flags, dfd, pathname, mask);
	510
	511	/* we only use the lower 32 bits as of right now. */
	512	if (mask & ((__u64)0xffffffff << 32))
	513	return -EINVAL;
	514
	515	if (flags & ~FAN_ALL_MARK_FLAGS)
	516	return -EINVAL;
	517	switch (flags & (FAN_MARK_ADD \| FAN_MARK_REMOVE \| FAN_MARK_FLUSH)) {
	518	case FAN_MARK_ADD:
	519	case FAN_MARK_REMOVE:
	520	case FAN_MARK_FLUSH:
	521	break;
	522	default:
	523	return -EINVAL;
	524	}
	525	if (mask & ~(FAN_ALL_EVENTS \| FAN_EVENT_ON_CHILD))
	526	return -EINVAL;
	527
	528	filp = fget_light(fanotify_fd, &fput_needed);
	529	if (unlikely(!filp))
	530	return -EBADF;
	531
	532	/* verify that this is indeed an fanotify instance */
	533	ret = -EINVAL;
	534	if (unlikely(filp->f_op != &fanotify_fops))
	535	goto fput_and_out;
	536
	537	ret = fanotify_find_path(dfd, pathname, &path, flags);
	538	if (ret)
	539	goto fput_and_out;
	540
	541	/* inode held in place by reference to path; group by fget on fd */
	542	if (!(flags & FAN_MARK_MOUNT))
	543	inode = path.dentry->d_inode;
	544	else
	545	mnt = path.mnt;
	546	group = filp->private_data;
	547
	548	/* create/update an inode mark */
	549	switch (flags & (FAN_MARK_ADD \| FAN_MARK_REMOVE \| FAN_MARK_FLUSH)) {
	550	case FAN_MARK_ADD:
	551	if (flags & FAN_MARK_MOUNT)
	552	ret = fanotify_add_vfsmount_mark(group, mnt, mask, flags);
	553	else
	554	ret = fanotify_add_inode_mark(group, inode, mask, flags);
	555	break;
	556	case FAN_MARK_REMOVE:
	557	if (flags & FAN_MARK_MOUNT)
	558	ret = fanotify_remove_vfsmount_mark(group, mnt, mask, flags);
	559	else
	560	ret = fanotify_remove_inode_mark(group, inode, mask, flags);
	561	break;
	562	case FAN_MARK_FLUSH:
	563	if (flags & FAN_MARK_MOUNT)
	564	fsnotify_clear_vfsmount_marks_by_group(group);
	565	else
	566	fsnotify_clear_inode_marks_by_group(group);
	567	fsnotify_recalc_group_mask(group);
	568	break;
	569	default:
	570	ret = -EINVAL;
	571	}
	572
	573	path_put(&path);
	574	fput_and_out:
	575	fput_light(filp, fput_needed);
	576	return ret;
	577	}
	578
	579	#ifdef CONFIG_HAVE_SYSCALL_WRAPPERS
	580	asmlinkage long SyS_fanotify_mark(long fanotify_fd, long flags, __u64 mask,
	581	long dfd, long pathname)
	582	{
	583	return SYSC_fanotify_mark((int) fanotify_fd, (unsigned int) flags,
	584	mask, (int) dfd,
	585	(const char __user *) pathname);
	586	}
	587	SYSCALL_ALIAS(sys_fanotify_mark, SyS_fanotify_mark);
	588	#endif
	589
	590	/*
	591	* fanotify_user_setup - Our initialization function. Note that we cannnot return
	592	* error because we have compiled-in VFS hooks. So an (unlikely) failure here
	593	* must result in panic().
	594	*/
	595	static int __init fanotify_user_setup(void)
	596	{
	597	fanotify_mark_cache = KMEM_CACHE(fsnotify_mark, SLAB_PANIC);
	598
	599	return 0;
	600	}
	601	device_initcall(fanotify_user_setup);