[linux.git] / fs / fcntl.c

/*
 *  linux/fs/fcntl.c
 *
 *  Copyright (C) 1991, 1992  Linus Torvalds
 */

#include <linux/syscalls.h>
#include <linux/init.h>
#include <linux/mm.h>
#include <linux/fs.h>
#include <linux/file.h>
#include <linux/fdtable.h>
#include <linux/capability.h>
#include <linux/dnotify.h>
#include <linux/slab.h>
#include <linux/module.h>
#include <linux/security.h>
#include <linux/ptrace.h>
#include <linux/signal.h>
#include <linux/rcupdate.h>
#include <linux/pid_namespace.h>
#include <linux/smp_lock.h>

#include <asm/poll.h>
#include <asm/siginfo.h>
#include <asm/uaccess.h>

void set_close_on_exec(unsigned int fd, int flag)
{
	struct files_struct *files = current->files;
	struct fdtable *fdt;
	spin_lock(&files->file_lock);
	fdt = files_fdtable(files);
	if (flag)
		FD_SET(fd, fdt->close_on_exec);
	else
		FD_CLR(fd, fdt->close_on_exec);
	spin_unlock(&files->file_lock);
}

static int get_close_on_exec(unsigned int fd)
{
	struct files_struct *files = current->files;
	struct fdtable *fdt;
	int res;
	rcu_read_lock();
	fdt = files_fdtable(files);
	res = FD_ISSET(fd, fdt->close_on_exec);
	rcu_read_unlock();
	return res;
}

SYSCALL_DEFINE3(dup3, unsigned int, oldfd, unsigned int, newfd, int, flags)
{
	int err = -EBADF;
	struct file * file, *tofree;
	struct files_struct * files = current->files;
	struct fdtable *fdt;

	if ((flags & ~O_CLOEXEC) != 0)
		return -EINVAL;

	if (unlikely(oldfd == newfd))
		return -EINVAL;

	spin_lock(&files->file_lock);
	err = expand_files(files, newfd);
	file = fcheck(oldfd);
	if (unlikely(!file))
		goto Ebadf;
	if (unlikely(err < 0)) {
		if (err == -EMFILE)
			goto Ebadf;
		goto out_unlock;
	}
	/*
	 * We need to detect attempts to do dup2() over allocated but still
	 * not finished descriptor.  NB: OpenBSD avoids that at the price of
	 * extra work in their equivalent of fget() - they insert struct
	 * file immediately after grabbing descriptor, mark it larval if
	 * more work (e.g. actual opening) is needed and make sure that
	 * fget() treats larval files as absent.  Potentially interesting,
	 * but while extra work in fget() is trivial, locking implications
	 * and amount of surgery on open()-related paths in VFS are not.
	 * FreeBSD fails with -EBADF in the same situation, NetBSD "solution"
	 * deadlocks in rather amusing ways, AFAICS.  All of that is out of
	 * scope of POSIX or SUS, since neither considers shared descriptor
	 * tables and this condition does not arise without those.
	 */
	err = -EBUSY;
	fdt = files_fdtable(files);
	tofree = fdt->fd[newfd];
	if (!tofree && FD_ISSET(newfd, fdt->open_fds))
		goto out_unlock;
	get_file(file);
	rcu_assign_pointer(fdt->fd[newfd], file);
	FD_SET(newfd, fdt->open_fds);
	if (flags & O_CLOEXEC)
		FD_SET(newfd, fdt->close_on_exec);
	else
		FD_CLR(newfd, fdt->close_on_exec);
	spin_unlock(&files->file_lock);

	if (tofree)
		filp_close(tofree, files);

	return newfd;

Ebadf:
	err = -EBADF;
out_unlock:
	spin_unlock(&files->file_lock);
	return err;
}

SYSCALL_DEFINE2(dup2, unsigned int, oldfd, unsigned int, newfd)
{
	if (unlikely(newfd == oldfd)) { /* corner case */
		struct files_struct *files = current->files;
		rcu_read_lock();
		if (!fcheck_files(files, oldfd))
			oldfd = -EBADF;
		rcu_read_unlock();
		return oldfd;
	}
	return sys_dup3(oldfd, newfd, 0);
}

SYSCALL_DEFINE1(dup, unsigned int, fildes)
{
	int ret = -EBADF;
	struct file *file = fget(fildes);

	if (file) {
		ret = get_unused_fd();
		if (ret >= 0)
			fd_install(ret, file);
		else
			fput(file);
	}
	return ret;
}

#define SETFL_MASK (O_APPEND | O_NONBLOCK | O_NDELAY | FASYNC | O_DIRECT | O_NOATIME)

static int setfl(int fd, struct file * filp, unsigned long arg)
{
	struct inode * inode = filp->f_path.dentry->d_inode;
	int error = 0;

	/*
	 * O_APPEND cannot be cleared if the file is marked as append-only
	 * and the file is open for write.
	 */
	if (((arg ^ filp->f_flags) & O_APPEND) && IS_APPEND(inode))
		return -EPERM;

	/* O_NOATIME can only be set by the owner or superuser */
	if ((arg & O_NOATIME) && !(filp->f_flags & O_NOATIME))
		if (!is_owner_or_cap(inode))
			return -EPERM;

	/* required for strict SunOS emulation */
	if (O_NONBLOCK != O_NDELAY)
	       if (arg & O_NDELAY)
		   arg |= O_NONBLOCK;

	if (arg & O_DIRECT) {
		if (!filp->f_mapping || !filp->f_mapping->a_ops ||
			!filp->f_mapping->a_ops->direct_IO)
				return -EINVAL;
	}

	if (filp->f_op && filp->f_op->check_flags)
		error = filp->f_op->check_flags(arg);
	if (error)
		return error;

	/*
	 * We still need a lock here for now to keep multiple FASYNC calls
	 * from racing with each other.
	 */
	lock_kernel();
	if ((arg ^ filp->f_flags) & FASYNC) {
		if (filp->f_op && filp->f_op->fasync) {
			error = filp->f_op->fasync(fd, filp, (arg & FASYNC) != 0);
			if (error < 0)
				goto out;
		}
	}

	filp->f_flags = (arg & SETFL_MASK) | (filp->f_flags & ~SETFL_MASK);
 out:
	unlock_kernel();
	return error;
}

static void f_modown(struct file *filp, struct pid *pid, enum pid_type type,
                     uid_t uid, uid_t euid, int force)
{
	write_lock_irq(&filp->f_owner.lock);
	if (force || !filp->f_owner.pid) {
		put_pid(filp->f_owner.pid);
		filp->f_owner.pid = get_pid(pid);
		filp->f_owner.pid_type = type;
		filp->f_owner.uid = uid;
		filp->f_owner.euid = euid;
	}
	write_unlock_irq(&filp->f_owner.lock);
}

int __f_setown(struct file *filp, struct pid *pid, enum pid_type type,
		int force)
{
	const struct cred *cred = current_cred();
	int err;
	
	err = security_file_set_fowner(filp);
	if (err)
		return err;

	f_modown(filp, pid, type, cred->uid, cred->euid, force);
	return 0;
}
EXPORT_SYMBOL(__f_setown);

int f_setown(struct file *filp, unsigned long arg, int force)
{
	enum pid_type type;
	struct pid *pid;
	int who = arg;
	int result;
	type = PIDTYPE_PID;
	if (who < 0) {
		type = PIDTYPE_PGID;
		who = -who;
	}
	rcu_read_lock();
	pid = find_vpid(who);
	result = __f_setown(filp, pid, type, force);
	rcu_read_unlock();
	return result;
}
EXPORT_SYMBOL(f_setown);

void f_delown(struct file *filp)
{
	f_modown(filp, NULL, PIDTYPE_PID, 0, 0, 1);
}

pid_t f_getown(struct file *filp)
{
	pid_t pid;
	read_lock(&filp->f_owner.lock);
	pid = pid_vnr(filp->f_owner.pid);
	if (filp->f_owner.pid_type == PIDTYPE_PGID)
		pid = -pid;
	read_unlock(&filp->f_owner.lock);
	return pid;
}

static long do_fcntl(int fd, unsigned int cmd, unsigned long arg,
		struct file *filp)
{
	long err = -EINVAL;

	switch (cmd) {
	case F_DUPFD:
	case F_DUPFD_CLOEXEC:
		if (arg >= current->signal->rlim[RLIMIT_NOFILE].rlim_cur)
			break;
		err = alloc_fd(arg, cmd == F_DUPFD_CLOEXEC ? O_CLOEXEC : 0);
		if (err >= 0) {
			get_file(filp);
			fd_install(err, filp);
		}
		break;
	case F_GETFD:
		err = get_close_on_exec(fd) ? FD_CLOEXEC : 0;
		break;
	case F_SETFD:
		err = 0;
		set_close_on_exec(fd, arg & FD_CLOEXEC);
		break;
	case F_GETFL:
		err = filp->f_flags;
		break;
	case F_SETFL:
		err = setfl(fd, filp, arg);
		break;
	case F_GETLK:
		err = fcntl_getlk(filp, (struct flock __user *) arg);
		break;
	case F_SETLK:
	case F_SETLKW:
		err = fcntl_setlk(fd, filp, cmd, (struct flock __user *) arg);
		break;
	case F_GETOWN:
		/*
		 * XXX If f_owner is a process group, the
		 * negative return value will get converted
		 * into an error.  Oops.  If we keep the
		 * current syscall conventions, the only way
		 * to fix this will be in libc.
		 */
		err = f_getown(filp);
		force_successful_syscall_return();
		break;
	case F_SETOWN:
		err = f_setown(filp, arg, 1);
		break;
	case F_GETSIG:
		err = filp->f_owner.signum;
		break;
	case F_SETSIG:
		/* arg == 0 restores default behaviour. */
		if (!valid_signal(arg)) {
			break;
		}
		err = 0;
		filp->f_owner.signum = arg;
		break;
	case F_GETLEASE:
		err = fcntl_getlease(filp);
		break;
	case F_SETLEASE:
		err = fcntl_setlease(fd, filp, arg);
		break;
	case F_NOTIFY:
		err = fcntl_dirnotify(fd, filp, arg);
		break;
	default:
		break;
	}
	return err;
}

SYSCALL_DEFINE3(fcntl, unsigned int, fd, unsigned int, cmd, unsigned long, arg)
{	
	struct file *filp;
	long err = -EBADF;

	filp = fget(fd);
	if (!filp)
		goto out;

	err = security_file_fcntl(filp, cmd, arg);
	if (err) {
		fput(filp);
		return err;
	}

	err = do_fcntl(fd, cmd, arg, filp);

 	fput(filp);
out:
	return err;
}

#if BITS_PER_LONG == 32
SYSCALL_DEFINE3(fcntl64, unsigned int, fd, unsigned int, cmd,
		unsigned long, arg)
{	
	struct file * filp;
	long err;

	err = -EBADF;
	filp = fget(fd);
	if (!filp)
		goto out;

	err = security_file_fcntl(filp, cmd, arg);
	if (err) {
		fput(filp);
		return err;
	}
	err = -EBADF;
	
	switch (cmd) {
		case F_GETLK64:
			err = fcntl_getlk64(filp, (struct flock64 __user *) arg);
			break;
		case F_SETLK64:
		case F_SETLKW64:
			err = fcntl_setlk64(fd, filp, cmd,
					(struct flock64 __user *) arg);
			break;
		default:
			err = do_fcntl(fd, cmd, arg, filp);
			break;
	}
	fput(filp);
out:
	return err;
}
#endif

/* Table to convert sigio signal codes into poll band bitmaps */

static const long band_table[NSIGPOLL] = {
	POLLIN | POLLRDNORM,			/* POLL_IN */
	POLLOUT | POLLWRNORM | POLLWRBAND,	/* POLL_OUT */
	POLLIN | POLLRDNORM | POLLMSG,		/* POLL_MSG */
	POLLERR,				/* POLL_ERR */
	POLLPRI | POLLRDBAND,			/* POLL_PRI */
	POLLHUP | POLLERR			/* POLL_HUP */
};

static inline int sigio_perm(struct task_struct *p,
                             struct fown_struct *fown, int sig)
{
	const struct cred *cred;
	int ret;

	rcu_read_lock();
	cred = __task_cred(p);
	ret = ((fown->euid == 0 ||
		fown->euid == cred->suid || fown->euid == cred->uid ||
		fown->uid  == cred->suid || fown->uid  == cred->uid) &&
	       !security_file_send_sigiotask(p, fown, sig));
	rcu_read_unlock();
	return ret;
}

static void send_sigio_to_task(struct task_struct *p,
			       struct fown_struct *fown, 
			       int fd,
			       int reason)
{
	if (!sigio_perm(p, fown, fown->signum))
		return;

	switch (fown->signum) {
		siginfo_t si;
		default:
			/* Queue a rt signal with the appropriate fd as its
			   value.  We use SI_SIGIO as the source, not 
			   SI_KERNEL, since kernel signals always get 
			   delivered even if we can't queue.  Failure to
			   queue in this case _should_ be reported; we fall
			   back to SIGIO in that case. --sct */
			si.si_signo = fown->signum;
			si.si_errno = 0;
		        si.si_code  = reason;
			/* Make sure we are called with one of the POLL_*
			   reasons, otherwise we could leak kernel stack into
			   userspace.  */
			BUG_ON((reason & __SI_MASK) != __SI_POLL);
			if (reason - POLL_IN >= NSIGPOLL)
				si.si_band  = ~0L;
			else
				si.si_band = band_table[reason - POLL_IN];
			si.si_fd    = fd;
			if (!group_send_sig_info(fown->signum, &si, p))
				break;
		/* fall-through: fall back on the old plain SIGIO signal */
		case 0:
			group_send_sig_info(SIGIO, SEND_SIG_PRIV, p);
	}
}

void send_sigio(struct fown_struct *fown, int fd, int band)
{
	struct task_struct *p;
	enum pid_type type;
	struct pid *pid;
	
	read_lock(&fown->lock);
	type = fown->pid_type;
	pid = fown->pid;
	if (!pid)
		goto out_unlock_fown;
	
	read_lock(&tasklist_lock);
	do_each_pid_task(pid, type, p) {
		send_sigio_to_task(p, fown, fd, band);
	} while_each_pid_task(pid, type, p);
	read_unlock(&tasklist_lock);
 out_unlock_fown:
	read_unlock(&fown->lock);
}

static void send_sigurg_to_task(struct task_struct *p,
                                struct fown_struct *fown)
{
	if (sigio_perm(p, fown, SIGURG))
		group_send_sig_info(SIGURG, SEND_SIG_PRIV, p);
}

int send_sigurg(struct fown_struct *fown)
{
	struct task_struct *p;
	enum pid_type type;
	struct pid *pid;
	int ret = 0;
	
	read_lock(&fown->lock);
	type = fown->pid_type;
	pid = fown->pid;
	if (!pid)
		goto out_unlock_fown;

	ret = 1;
	
	read_lock(&tasklist_lock);
	do_each_pid_task(pid, type, p) {
		send_sigurg_to_task(p, fown);
	} while_each_pid_task(pid, type, p);
	read_unlock(&tasklist_lock);
 out_unlock_fown:
	read_unlock(&fown->lock);
	return ret;
}

static DEFINE_RWLOCK(fasync_lock);
static struct kmem_cache *fasync_cache __read_mostly;

/*
 * fasync_helper() is used by some character device drivers (mainly mice)
 * to set up the fasync queue. It returns negative on error, 0 if it did
 * no changes and positive if it added/deleted the entry.
 */
int fasync_helper(int fd, struct file * filp, int on, struct fasync_struct **fapp)
{
	struct fasync_struct *fa, **fp;
	struct fasync_struct *new = NULL;
	int result = 0;

	if (on) {
		new = kmem_cache_alloc(fasync_cache, GFP_KERNEL);
		if (!new)
			return -ENOMEM;
	}
	write_lock_irq(&fasync_lock);
	for (fp = fapp; (fa = *fp) != NULL; fp = &fa->fa_next) {
		if (fa->fa_file == filp) {
			if(on) {
				fa->fa_fd = fd;
				kmem_cache_free(fasync_cache, new);
			} else {
				*fp = fa->fa_next;
				kmem_cache_free(fasync_cache, fa);
				result = 1;
			}
			goto out;
		}
	}

	if (on) {
		new->magic = FASYNC_MAGIC;
		new->fa_file = filp;
		new->fa_fd = fd;
		new->fa_next = *fapp;
		*fapp = new;
		result = 1;
	}
out:
	write_unlock_irq(&fasync_lock);
	return result;
}

EXPORT_SYMBOL(fasync_helper);

void __kill_fasync(struct fasync_struct *fa, int sig, int band)
{
	while (fa) {
		struct fown_struct * fown;
		if (fa->magic != FASYNC_MAGIC) {
			printk(KERN_ERR "kill_fasync: bad magic number in "
			       "fasync_struct!\n");
			return;
		}
		fown = &fa->fa_file->f_owner;
		/* Don't send SIGURG to processes which have not set a
		   queued signum: SIGURG has its own default signalling
		   mechanism. */
		if (!(sig == SIGURG && fown->signum == 0))
			send_sigio(fown, fa->fa_fd, band);
		fa = fa->fa_next;
	}
}

EXPORT_SYMBOL(__kill_fasync);

void kill_fasync(struct fasync_struct **fp, int sig, int band)
{
	/* First a quick test without locking: usually
	 * the list is empty.
	 */
	if (*fp) {
		read_lock(&fasync_lock);
		/* reread *fp after obtaining the lock */
		__kill_fasync(*fp, sig, band);
		read_unlock(&fasync_lock);
	}
}
EXPORT_SYMBOL(kill_fasync);

static int __init fasync_init(void)
{
	fasync_cache = kmem_cache_create("fasync_cache",
		sizeof(struct fasync_struct), 0, SLAB_PANIC, NULL);
	return 0;
}

module_init(fasync_init)
Commit	Line	Data
1da177e4 LT	1	/*
	2	* linux/fs/fcntl.c
	3	*
	4	* Copyright (C) 1991, 1992 Linus Torvalds
	5	*/
	6
	7	#include <linux/syscalls.h>
	8	#include <linux/init.h>
	9	#include <linux/mm.h>
	10	#include <linux/fs.h>
	11	#include <linux/file.h>
9f3acc31	12	#include <linux/fdtable.h>
16f7e0fe	13	#include <linux/capability.h>
1da177e4	14	#include <linux/dnotify.h>
1da177e4 LT	15	#include <linux/slab.h>
	16	#include <linux/module.h>
	17	#include <linux/security.h>
	18	#include <linux/ptrace.h>
7ed20e1a	19	#include <linux/signal.h>
ab2af1f5	20	#include <linux/rcupdate.h>
b488893a	21	#include <linux/pid_namespace.h>
218d11a8	22	#include <linux/smp_lock.h>
1da177e4 LT	23
	24	#include <asm/poll.h>
	25	#include <asm/siginfo.h>
	26	#include <asm/uaccess.h>
	27
fc9b52cd	28	void set_close_on_exec(unsigned int fd, int flag)
1da177e4 LT	29	{
1da177e4 LT	30	struct files_struct *files = current->files;
badf1662	31	struct fdtable *fdt;
1da177e4	32	spin_lock(&files->file_lock);
badf1662	33	fdt = files_fdtable(files);
1da177e4	34	if (flag)
badf1662	35	FD_SET(fd, fdt->close_on_exec);
1da177e4	36	else
badf1662	37	FD_CLR(fd, fdt->close_on_exec);
1da177e4 LT	38	spin_unlock(&files->file_lock);
	39	}
	40
858119e1	41	static int get_close_on_exec(unsigned int fd)
1da177e4 LT	42	{
1da177e4 LT	43	struct files_struct *files = current->files;
badf1662	44	struct fdtable *fdt;
1da177e4	45	int res;
b835996f	46	rcu_read_lock();
badf1662 DS	47	fdt = files_fdtable(files);
badf1662 DS	48	res = FD_ISSET(fd, fdt->close_on_exec);
b835996f	49	rcu_read_unlock();
1da177e4 LT	50	return res;
	51	}
	52
a26eab24	53	SYSCALL_DEFINE3(dup3, unsigned int, oldfd, unsigned int, newfd, int, flags)
1da177e4 LT	54	{
	55	int err = -EBADF;
	56	struct file * file, *tofree;
	57	struct files_struct * files = current->files;
badf1662	58	struct fdtable *fdt;
1da177e4	59
336dd1f7 UD	60	if ((flags & ~O_CLOEXEC) != 0)
	61	return -EINVAL;
	62
6c5d0512 AV	63	if (unlikely(oldfd == newfd))
	64	return -EINVAL;
	65
1da177e4	66	spin_lock(&files->file_lock);
1da177e4	67	err = expand_files(files, newfd);
1b7e190b AV	68	file = fcheck(oldfd);
	69	if (unlikely(!file))
	70	goto Ebadf;
4e1e018e AV	71	if (unlikely(err < 0)) {
4e1e018e AV	72	if (err == -EMFILE)
1b7e190b AV	73	goto Ebadf;
1b7e190b AV	74	goto out_unlock;
4e1e018e	75	}
1b7e190b AV	76	/*
	77	* We need to detect attempts to do dup2() over allocated but still
	78	* not finished descriptor. NB: OpenBSD avoids that at the price of
	79	* extra work in their equivalent of fget() - they insert struct
	80	* file immediately after grabbing descriptor, mark it larval if
	81	* more work (e.g. actual opening) is needed and make sure that
	82	* fget() treats larval files as absent. Potentially interesting,
	83	* but while extra work in fget() is trivial, locking implications
	84	* and amount of surgery on open()-related paths in VFS are not.
	85	* FreeBSD fails with -EBADF in the same situation, NetBSD "solution"
	86	* deadlocks in rather amusing ways, AFAICS. All of that is out of
	87	* scope of POSIX or SUS, since neither considers shared descriptor
	88	* tables and this condition does not arise without those.
	89	*/
1da177e4	90	err = -EBUSY;
badf1662 DS	91	fdt = files_fdtable(files);
	92	tofree = fdt->fd[newfd];
	93	if (!tofree && FD_ISSET(newfd, fdt->open_fds))
1b7e190b AV	94	goto out_unlock;
1b7e190b AV	95	get_file(file);
ab2af1f5	96	rcu_assign_pointer(fdt->fd[newfd], file);
badf1662	97	FD_SET(newfd, fdt->open_fds);
336dd1f7 UD	98	if (flags & O_CLOEXEC)
	99	FD_SET(newfd, fdt->close_on_exec);
	100	else
	101	FD_CLR(newfd, fdt->close_on_exec);
1da177e4 LT	102	spin_unlock(&files->file_lock);
	103
	104	if (tofree)
	105	filp_close(tofree, files);
1da177e4	106
1b7e190b AV	107	return newfd;
	108
	109	Ebadf:
	110	err = -EBADF;
	111	out_unlock:
1da177e4	112	spin_unlock(&files->file_lock);
1b7e190b	113	return err;
1da177e4	114	}
336dd1f7	115
a26eab24	116	SYSCALL_DEFINE2(dup2, unsigned int, oldfd, unsigned int, newfd)
336dd1f7	117	{
6c5d0512 AV	118	if (unlikely(newfd == oldfd)) { /* corner case */
	119	struct files_struct *files = current->files;
	120	rcu_read_lock();
	121	if (!fcheck_files(files, oldfd))
	122	oldfd = -EBADF;
	123	rcu_read_unlock();
	124	return oldfd;
	125	}
336dd1f7 UD	126	return sys_dup3(oldfd, newfd, 0);
336dd1f7 UD	127	}
1da177e4	128
a26eab24	129	SYSCALL_DEFINE1(dup, unsigned int, fildes)
1da177e4 LT	130	{
1da177e4 LT	131	int ret = -EBADF;
1027abe8 AV	132	struct file *file = fget(fildes);
	133
	134	if (file) {
	135	ret = get_unused_fd();
	136	if (ret >= 0)
	137	fd_install(ret, file);
	138	else
	139	fput(file);
	140	}
1da177e4 LT	141	return ret;
	142	}
	143
	144	#define SETFL_MASK (O_APPEND \| O_NONBLOCK \| O_NDELAY \| FASYNC \| O_DIRECT \| O_NOATIME)
	145
	146	static int setfl(int fd, struct file * filp, unsigned long arg)
	147	{
0f7fc9e4	148	struct inode * inode = filp->f_path.dentry->d_inode;
1da177e4 LT	149	int error = 0;
1da177e4 LT	150
7d95c8f2	151	/*
	152	* O_APPEND cannot be cleared if the file is marked as append-only
	153	* and the file is open for write.
	154	*/
	155	if (((arg ^ filp->f_flags) & O_APPEND) && IS_APPEND(inode))
1da177e4 LT	156	return -EPERM;
	157
	158	/* O_NOATIME can only be set by the owner or superuser */
	159	if ((arg & O_NOATIME) && !(filp->f_flags & O_NOATIME))
3bd858ab	160	if (!is_owner_or_cap(inode))
1da177e4 LT	161	return -EPERM;
	162
	163	/* required for strict SunOS emulation */
	164	if (O_NONBLOCK != O_NDELAY)
	165	if (arg & O_NDELAY)
	166	arg \|= O_NONBLOCK;
	167
	168	if (arg & O_DIRECT) {
	169	if (!filp->f_mapping \|\| !filp->f_mapping->a_ops \|\|
	170	!filp->f_mapping->a_ops->direct_IO)
	171	return -EINVAL;
	172	}
	173
	174	if (filp->f_op && filp->f_op->check_flags)
	175	error = filp->f_op->check_flags(arg);
	176	if (error)
	177	return error;
	178
218d11a8 JC	179	/*
	180	* We still need a lock here for now to keep multiple FASYNC calls
	181	* from racing with each other.
	182	*/
	183	lock_kernel();
1da177e4 LT	184	if ((arg ^ filp->f_flags) & FASYNC) {
	185	if (filp->f_op && filp->f_op->fasync) {
	186	error = filp->f_op->fasync(fd, filp, (arg & FASYNC) != 0);
	187	if (error < 0)
	188	goto out;
	189	}
	190	}
	191
	192	filp->f_flags = (arg & SETFL_MASK) \| (filp->f_flags & ~SETFL_MASK);
	193	out:
218d11a8	194	unlock_kernel();
1da177e4 LT	195	return error;
	196	}
	197
609d7fa9	198	static void f_modown(struct file filp, struct pid pid, enum pid_type type,
1da177e4 LT	199	uid_t uid, uid_t euid, int force)
	200	{
	201	write_lock_irq(&filp->f_owner.lock);
	202	if (force \|\| !filp->f_owner.pid) {
609d7fa9 EB	203	put_pid(filp->f_owner.pid);
	204	filp->f_owner.pid = get_pid(pid);
	205	filp->f_owner.pid_type = type;
1da177e4 LT	206	filp->f_owner.uid = uid;
	207	filp->f_owner.euid = euid;
	208	}
	209	write_unlock_irq(&filp->f_owner.lock);
	210	}
	211
609d7fa9 EB	212	int __f_setown(struct file filp, struct pid pid, enum pid_type type,
609d7fa9 EB	213	int force)
1da177e4	214	{
86a264ab	215	const struct cred *cred = current_cred();
1da177e4 LT	216	int err;
	217
	218	err = security_file_set_fowner(filp);
	219	if (err)
	220	return err;
	221
86a264ab	222	f_modown(filp, pid, type, cred->uid, cred->euid, force);
1da177e4 LT	223	return 0;
1da177e4 LT	224	}
609d7fa9	225	EXPORT_SYMBOL(__f_setown);
1da177e4	226
609d7fa9 EB	227	int f_setown(struct file *filp, unsigned long arg, int force)
	228	{
	229	enum pid_type type;
	230	struct pid *pid;
	231	int who = arg;
	232	int result;
	233	type = PIDTYPE_PID;
	234	if (who < 0) {
	235	type = PIDTYPE_PGID;
	236	who = -who;
	237	}
	238	rcu_read_lock();
b488893a	239	pid = find_vpid(who);
609d7fa9 EB	240	result = __f_setown(filp, pid, type, force);
	241	rcu_read_unlock();
	242	return result;
	243	}
1da177e4 LT	244	EXPORT_SYMBOL(f_setown);
	245
	246	void f_delown(struct file *filp)
	247	{
609d7fa9 EB	248	f_modown(filp, NULL, PIDTYPE_PID, 0, 0, 1);
	249	}
	250
	251	pid_t f_getown(struct file *filp)
	252	{
	253	pid_t pid;
43fa1adb	254	read_lock(&filp->f_owner.lock);
6c5f3e7b	255	pid = pid_vnr(filp->f_owner.pid);
609d7fa9 EB	256	if (filp->f_owner.pid_type == PIDTYPE_PGID)
609d7fa9 EB	257	pid = -pid;
43fa1adb	258	read_unlock(&filp->f_owner.lock);
609d7fa9	259	return pid;
1da177e4 LT	260	}
	261
	262	static long do_fcntl(int fd, unsigned int cmd, unsigned long arg,
	263	struct file *filp)
	264	{
	265	long err = -EINVAL;
	266
	267	switch (cmd) {
	268	case F_DUPFD:
22d2b35b	269	case F_DUPFD_CLOEXEC:
4e1e018e AV	270	if (arg >= current->signal->rlim[RLIMIT_NOFILE].rlim_cur)
4e1e018e AV	271	break;
1027abe8 AV	272	err = alloc_fd(arg, cmd == F_DUPFD_CLOEXEC ? O_CLOEXEC : 0);
	273	if (err >= 0) {
	274	get_file(filp);
	275	fd_install(err, filp);
	276	}
1da177e4 LT	277	break;
	278	case F_GETFD:
	279	err = get_close_on_exec(fd) ? FD_CLOEXEC : 0;
	280	break;
	281	case F_SETFD:
	282	err = 0;
	283	set_close_on_exec(fd, arg & FD_CLOEXEC);
	284	break;
	285	case F_GETFL:
	286	err = filp->f_flags;
	287	break;
	288	case F_SETFL:
	289	err = setfl(fd, filp, arg);
	290	break;
	291	case F_GETLK:
	292	err = fcntl_getlk(filp, (struct flock __user *) arg);
	293	break;
	294	case F_SETLK:
	295	case F_SETLKW:
c293621b	296	err = fcntl_setlk(fd, filp, cmd, (struct flock __user *) arg);
1da177e4 LT	297	break;
	298	case F_GETOWN:
	299	/*
	300	* XXX If f_owner is a process group, the
	301	* negative return value will get converted
	302	* into an error. Oops. If we keep the
	303	* current syscall conventions, the only way
	304	* to fix this will be in libc.
	305	*/
609d7fa9	306	err = f_getown(filp);
1da177e4 LT	307	force_successful_syscall_return();
	308	break;
	309	case F_SETOWN:
	310	err = f_setown(filp, arg, 1);
	311	break;
	312	case F_GETSIG:
	313	err = filp->f_owner.signum;
	314	break;
	315	case F_SETSIG:
	316	/* arg == 0 restores default behaviour. */
7ed20e1a	317	if (!valid_signal(arg)) {
1da177e4 LT	318	break;
	319	}
	320	err = 0;
	321	filp->f_owner.signum = arg;
	322	break;
	323	case F_GETLEASE:
	324	err = fcntl_getlease(filp);
	325	break;
	326	case F_SETLEASE:
	327	err = fcntl_setlease(fd, filp, arg);
	328	break;
	329	case F_NOTIFY:
	330	err = fcntl_dirnotify(fd, filp, arg);
	331	break;
	332	default:
	333	break;
	334	}
	335	return err;
	336	}
	337
a26eab24	338	SYSCALL_DEFINE3(fcntl, unsigned int, fd, unsigned int, cmd, unsigned long, arg)
1da177e4 LT	339	{
	340	struct file *filp;
	341	long err = -EBADF;
	342
	343	filp = fget(fd);
	344	if (!filp)
	345	goto out;
	346
	347	err = security_file_fcntl(filp, cmd, arg);
	348	if (err) {
	349	fput(filp);
	350	return err;
	351	}
	352
	353	err = do_fcntl(fd, cmd, arg, filp);
	354
	355	fput(filp);
	356	out:
	357	return err;
	358	}
	359
	360	#if BITS_PER_LONG == 32
a26eab24 HC	361	SYSCALL_DEFINE3(fcntl64, unsigned int, fd, unsigned int, cmd,
a26eab24 HC	362	unsigned long, arg)
1da177e4 LT	363	{
	364	struct file * filp;
	365	long err;
	366
	367	err = -EBADF;
	368	filp = fget(fd);
	369	if (!filp)
	370	goto out;
	371
	372	err = security_file_fcntl(filp, cmd, arg);
	373	if (err) {
	374	fput(filp);
	375	return err;
	376	}
	377	err = -EBADF;
	378
	379	switch (cmd) {
	380	case F_GETLK64:
	381	err = fcntl_getlk64(filp, (struct flock64 __user *) arg);
	382	break;
	383	case F_SETLK64:
	384	case F_SETLKW64:
c293621b PS	385	err = fcntl_setlk64(fd, filp, cmd,
c293621b PS	386	(struct flock64 __user *) arg);
1da177e4 LT	387	break;
	388	default:
	389	err = do_fcntl(fd, cmd, arg, filp);
	390	break;
	391	}
	392	fput(filp);
	393	out:
	394	return err;
	395	}
	396	#endif
	397
	398	/* Table to convert sigio signal codes into poll band bitmaps */
	399
fa3536cc	400	static const long band_table[NSIGPOLL] = {
1da177e4 LT	401	POLLIN \| POLLRDNORM, /* POLL_IN */
	402	POLLOUT \| POLLWRNORM \| POLLWRBAND, /* POLL_OUT */
	403	POLLIN \| POLLRDNORM \| POLLMSG, /* POLL_MSG */
	404	POLLERR, /* POLL_ERR */
	405	POLLPRI \| POLLRDBAND, /* POLL_PRI */
	406	POLLHUP \| POLLERR /* POLL_HUP */
	407	};
	408
	409	static inline int sigio_perm(struct task_struct *p,
	410	struct fown_struct *fown, int sig)
	411	{
c69e8d9c DH	412	const struct cred *cred;
	413	int ret;
	414
	415	rcu_read_lock();
	416	cred = __task_cred(p);
	417	ret = ((fown->euid == 0 \|\|
	418	fown->euid == cred->suid \|\| fown->euid == cred->uid \|\|
	419	fown->uid == cred->suid \|\| fown->uid == cred->uid) &&
	420	!security_file_send_sigiotask(p, fown, sig));
	421	rcu_read_unlock();
	422	return ret;
1da177e4 LT	423	}
	424
	425	static void send_sigio_to_task(struct task_struct *p,
	426	struct fown_struct *fown,
	427	int fd,
	428	int reason)
	429	{
	430	if (!sigio_perm(p, fown, fown->signum))
	431	return;
	432
	433	switch (fown->signum) {
	434	siginfo_t si;
	435	default:
	436	/* Queue a rt signal with the appropriate fd as its
	437	value. We use SI_SIGIO as the source, not
	438	SI_KERNEL, since kernel signals always get
	439	delivered even if we can't queue. Failure to
	440	queue in this case _should_ be reported; we fall
	441	back to SIGIO in that case. --sct */
	442	si.si_signo = fown->signum;
	443	si.si_errno = 0;
	444	si.si_code = reason;
	445	/* Make sure we are called with one of the POLL_*
	446	reasons, otherwise we could leak kernel stack into
	447	userspace. */
f6298aab	448	BUG_ON((reason & __SI_MASK) != __SI_POLL);
1da177e4 LT	449	if (reason - POLL_IN >= NSIGPOLL)
	450	si.si_band = ~0L;
	451	else
	452	si.si_band = band_table[reason - POLL_IN];
	453	si.si_fd = fd;
850d6fbe	454	if (!group_send_sig_info(fown->signum, &si, p))
1da177e4 LT	455	break;
	456	/* fall-through: fall back on the old plain SIGIO signal */
	457	case 0:
850d6fbe	458	group_send_sig_info(SIGIO, SEND_SIG_PRIV, p);
1da177e4 LT	459	}
	460	}
	461
	462	void send_sigio(struct fown_struct *fown, int fd, int band)
	463	{
	464	struct task_struct *p;
609d7fa9 EB	465	enum pid_type type;
609d7fa9 EB	466	struct pid *pid;
1da177e4 LT	467
1da177e4 LT	468	read_lock(&fown->lock);
609d7fa9	469	type = fown->pid_type;
1da177e4 LT	470	pid = fown->pid;
	471	if (!pid)
	472	goto out_unlock_fown;
	473
	474	read_lock(&tasklist_lock);
609d7fa9 EB	475	do_each_pid_task(pid, type, p) {
	476	send_sigio_to_task(p, fown, fd, band);
	477	} while_each_pid_task(pid, type, p);
1da177e4 LT	478	read_unlock(&tasklist_lock);
	479	out_unlock_fown:
	480	read_unlock(&fown->lock);
	481	}
	482
	483	static void send_sigurg_to_task(struct task_struct *p,
	484	struct fown_struct *fown)
	485	{
	486	if (sigio_perm(p, fown, SIGURG))
850d6fbe	487	group_send_sig_info(SIGURG, SEND_SIG_PRIV, p);
1da177e4 LT	488	}
	489
	490	int send_sigurg(struct fown_struct *fown)
	491	{
	492	struct task_struct *p;
609d7fa9 EB	493	enum pid_type type;
	494	struct pid *pid;
	495	int ret = 0;
1da177e4 LT	496
1da177e4 LT	497	read_lock(&fown->lock);
609d7fa9	498	type = fown->pid_type;
1da177e4 LT	499	pid = fown->pid;
	500	if (!pid)
	501	goto out_unlock_fown;
	502
	503	ret = 1;
	504
	505	read_lock(&tasklist_lock);
609d7fa9 EB	506	do_each_pid_task(pid, type, p) {
	507	send_sigurg_to_task(p, fown);
	508	} while_each_pid_task(pid, type, p);
1da177e4 LT	509	read_unlock(&tasklist_lock);
	510	out_unlock_fown:
	511	read_unlock(&fown->lock);
	512	return ret;
	513	}
	514
	515	static DEFINE_RWLOCK(fasync_lock);
e18b890b	516	static struct kmem_cache *fasync_cache __read_mostly;
1da177e4 LT	517
	518	/*
	519	* fasync_helper() is used by some character device drivers (mainly mice)
	520	* to set up the fasync queue. It returns negative on error, 0 if it did
	521	* no changes and positive if it added/deleted the entry.
	522	*/
	523	int fasync_helper(int fd, struct file * filp, int on, struct fasync_struct **fapp)
	524	{
	525	struct fasync_struct fa, *fp;
	526	struct fasync_struct *new = NULL;
	527	int result = 0;
	528
	529	if (on) {
e94b1766	530	new = kmem_cache_alloc(fasync_cache, GFP_KERNEL);
1da177e4 LT	531	if (!new)
	532	return -ENOMEM;
	533	}
	534	write_lock_irq(&fasync_lock);
	535	for (fp = fapp; (fa = *fp) != NULL; fp = &fa->fa_next) {
	536	if (fa->fa_file == filp) {
	537	if(on) {
	538	fa->fa_fd = fd;
	539	kmem_cache_free(fasync_cache, new);
	540	} else {
	541	*fp = fa->fa_next;
	542	kmem_cache_free(fasync_cache, fa);
	543	result = 1;
	544	}
	545	goto out;
	546	}
	547	}
	548
	549	if (on) {
	550	new->magic = FASYNC_MAGIC;
	551	new->fa_file = filp;
	552	new->fa_fd = fd;
	553	new->fa_next = *fapp;
	554	*fapp = new;
	555	result = 1;
	556	}
	557	out:
	558	write_unlock_irq(&fasync_lock);
	559	return result;
	560	}
	561
	562	EXPORT_SYMBOL(fasync_helper);
	563
	564	void __kill_fasync(struct fasync_struct *fa, int sig, int band)
	565	{
	566	while (fa) {
	567	struct fown_struct * fown;
	568	if (fa->magic != FASYNC_MAGIC) {
	569	printk(KERN_ERR "kill_fasync: bad magic number in "
	570	"fasync_struct!\n");
	571	return;
	572	}
	573	fown = &fa->fa_file->f_owner;
	574	/* Don't send SIGURG to processes which have not set a
	575	queued signum: SIGURG has its own default signalling
	576	mechanism. */
	577	if (!(sig == SIGURG && fown->signum == 0))
	578	send_sigio(fown, fa->fa_fd, band);
	579	fa = fa->fa_next;
	580	}
	581	}
	582
	583	EXPORT_SYMBOL(__kill_fasync);
	584
	585	void kill_fasync(struct fasync_struct **fp, int sig, int band)
	586	{
	587	/* First a quick test without locking: usually
	588	* the list is empty.
	589	*/
	590	if (*fp) {
	591	read_lock(&fasync_lock);
	592	/* reread fp after obtaining the lock /
	593	__kill_fasync(*fp, sig, band);
	594	read_unlock(&fasync_lock);
595	}
596	}
597	EXPORT_SYMBOL(kill_fasync);
598
599	static int __init fasync_init(void)
600	{
601	fasync_cache = kmem_cache_create("fasync_cache",
20c2df83	602	sizeof(struct fasync_struct), 0, SLAB_PANIC, NULL);
1da177e4 LT	603	return 0;
	604	}
	605
	606	module_init(fasync_init)