Upgrade make(1). 1/2

[dragonfly.git] / sys / kern / vfs_syscalls.c

/*
 * Copyright (c) 1989, 1993
 *      The Regents of the University of California.  All rights reserved.
 * (c) UNIX System Laboratories, Inc.
 * All or some portions of this file are derived from material licensed
 * to the University of California by American Telephone and Telegraph
 * Co. or Unix System Laboratories, Inc. and are reproduced herein with
 * the permission of UNIX System Laboratories, Inc.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice, this list of conditions and the following disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 * 3. Neither the name of the University nor the names of its contributors
 *    may be used to endorse or promote products derived from this software
 *    without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
 * SUCH DAMAGE.
 *
 *      @(#)vfs_syscalls.c      8.13 (Berkeley) 4/15/94
 * $FreeBSD: src/sys/kern/vfs_syscalls.c,v 1.151.2.18 2003/04/04 20:35:58 tegge Exp $
 */

#include <sys/param.h>
#include <sys/systm.h>
#include <sys/buf.h>
#include <sys/conf.h>
#include <sys/sysent.h>
#include <sys/malloc.h>
#include <sys/mount.h>
#include <sys/mountctl.h>
#include <sys/sysmsg.h>
#include <sys/filedesc.h>
#include <sys/kernel.h>
#include <sys/fcntl.h>
#include <sys/file.h>
#include <sys/linker.h>
#include <sys/stat.h>
#include <sys/unistd.h>
#include <sys/vnode.h>
#include <sys/proc.h>
#include <sys/priv.h>
#include <sys/jail.h>
#include <sys/namei.h>
#include <sys/nlookup.h>
#include <sys/dirent.h>
#include <sys/extattr.h>
#include <sys/spinlock.h>
#include <sys/kern_syscall.h>
#include <sys/objcache.h>
#include <sys/sysctl.h>

#include <sys/buf2.h>
#include <sys/file2.h>
#include <sys/spinlock2.h>

#include <vm/vm.h>
#include <vm/vm_object.h>
#include <vm/vm_page.h>

#include <machine/limits.h>
#include <machine/stdarg.h>

static void mount_warning(struct mount *mp, const char *ctl, ...)
                __printflike(2, 3);
static int mount_path(struct proc *p, struct mount *mp, char **rb, char **fb);
static int checkvp_chdir (struct vnode *vn, struct thread *td);
static void checkdirs (struct nchandle *old_nch, struct nchandle *new_nch);
static int get_fspriv(const char *);
static int chroot_refuse_vdir_fds (thread_t td, struct filedesc *fdp);
static int chroot_visible_mnt(struct mount *mp, struct proc *p);
static int getutimes (struct timeval *, struct timespec *);
static int getutimens (const struct timespec *, struct timespec *, int *);
static int setfown (struct mount *, struct vnode *, uid_t, gid_t);
static int setfmode (struct vnode *, int);
static int setfflags (struct vnode *, u_long);
static int setutimes (struct vnode *, struct vattr *,
                        const struct timespec *, int);

static int      usermount = 0;  /* if 1, non-root can mount fs. */
SYSCTL_INT(_vfs, OID_AUTO, usermount, CTLFLAG_RW, &usermount, 0,
    "Allow non-root users to mount filesystems");

static int      debug_unmount = 0; /* if 1 loop until unmount success */
SYSCTL_INT(_vfs, OID_AUTO, debug_unmount, CTLFLAG_RW, &debug_unmount, 0,
    "Stall failed unmounts in loop");

static struct krate krate_rename = { 1 };

/*
 * Virtual File System System Calls
 */

/*
 * Mount a file system.
 *
 * mount_args(char *type, char *path, int flags, caddr_t data)
 *
 * MPALMOSTSAFE
 */
int
sys_mount(struct sysmsg *sysmsg, const struct mount_args *uap)
{
        struct thread *td = curthread;
        struct vnode *vp;
        struct nchandle nch;
        struct mount *mp, *nullmp;
        struct vfsconf *vfsp;
        int error, flag = 0, flag2 = 0;
        int hasmount;
        int priv = 0;
        int flags = uap->flags;
        struct vattr va;
        struct nlookupdata nd;
        char fstypename[MFSNAMELEN];
        struct ucred *cred;

        cred = td->td_ucred;

        /* We do not allow user mounts inside a jail for now */
        if (usermount && jailed(cred)) {
                error = EPERM;
                goto done;
        }

        /*
         * Extract the file system type. We need to know this early, to take
         * appropriate actions for jails and nullfs mounts.
         */
        if ((error = copyinstr(uap->type, fstypename, MFSNAMELEN, NULL)) != 0)
                goto done;

        /*
         * Select the correct priv according to the file system type.
         */
        priv = get_fspriv(fstypename);

        if (usermount == 0 && (error = priv_check(td, priv)))
                goto done;

        /*
         * Do not allow NFS export by non-root users.
         */
        if (flags & MNT_EXPORTED) {
                error = priv_check(td, priv);
                if (error)
                        goto done;
        }
        /*
         * Silently enforce MNT_NOSUID and MNT_NODEV for non-root users
         */
        if (priv_check(td, priv))
                flags |= MNT_NOSUID | MNT_NODEV;

        /*
         * Lookup the requested path and extract the nch and vnode.
         */
        error = nlookup_init(&nd, uap->path, UIO_USERSPACE, NLC_FOLLOW);
        if (error == 0) {
                if ((error = nlookup(&nd)) == 0) {
                        if (nd.nl_nch.ncp->nc_vp == NULL)
                                error = ENOENT;
                }
        }
        if (error) {
                nlookup_done(&nd);
                goto done;
        }

        /*
         * If the target filesystem is resolved via a nullfs mount, then
         * nd.nl_nch.mount will be pointing to the nullfs mount structure
         * instead of the target file system. We need it in case we are
         * doing an update.
         */
        nullmp = nd.nl_nch.mount;

        /*
         * Extract the locked+refd ncp and cleanup the nd structure
         */
        nch = nd.nl_nch;
        cache_zero(&nd.nl_nch);
        nlookup_done(&nd);

        if ((nch.ncp->nc_flag & NCF_ISMOUNTPT) &&
            (mp = cache_findmount(&nch)) != NULL) {
                cache_dropmount(mp);
                hasmount = 1;
        } else {
                hasmount = 0;
        }


        /*
         * now we have the locked ref'd nch and unreferenced vnode.
         */
        vp = nch.ncp->nc_vp;
        if ((error = vget(vp, LK_EXCLUSIVE)) != 0) {
                cache_put(&nch);
                goto done;
        }
        cache_unlock(&nch);

        /*
         * Now we have an unlocked ref'd nch and a locked ref'd vp
         */
        if (flags & MNT_UPDATE) {
                if ((vp->v_flag & (VROOT|VPFSROOT)) == 0) {
                        cache_drop(&nch);
                        vput(vp);
                        error = EINVAL;
                        goto done;
                }

                if (strncmp(fstypename, "null", 5) == 0) {
                        KKASSERT(nullmp);
                        mp = nullmp;
                } else {
                        mp = vp->v_mount;
                }

                flag = mp->mnt_flag;
                flag2 = mp->mnt_kern_flag;
                /*
                 * We only allow the filesystem to be reloaded if it
                 * is currently mounted read-only.
                 */
                if ((flags & MNT_RELOAD) &&
                    ((mp->mnt_flag & MNT_RDONLY) == 0)) {
                        cache_drop(&nch);
                        vput(vp);
                        error = EOPNOTSUPP;     /* Needs translation */
                        goto done;
                }
                /*
                 * Only root, or the user that did the original mount is
                 * permitted to update it.
                 */
                if (mp->mnt_stat.f_owner != cred->cr_uid &&
                    (error = priv_check(td, priv))) {
                        cache_drop(&nch);
                        vput(vp);
                        goto done;
                }
                if (vfs_busy(mp, LK_NOWAIT)) {
                        cache_drop(&nch);
                        vput(vp);
                        error = EBUSY;
                        goto done;
                }
                if (hasmount) {
                        cache_drop(&nch);
                        vfs_unbusy(mp);
                        vput(vp);
                        error = EBUSY;
                        goto done;
                }
                mp->mnt_flag |= flags & (MNT_RELOAD | MNT_FORCE | MNT_UPDATE);
                lwkt_gettoken(&mp->mnt_token);
                vn_unlock(vp);
                vfsp = mp->mnt_vfc;
                goto update;
        }

        /*
         * If the user is not root, ensure that they own the directory
         * onto which we are attempting to mount.
         */
        if ((error = VOP_GETATTR(vp, &va)) ||
            (va.va_uid != cred->cr_uid &&
             (error = priv_check(td, priv)))) {
                cache_drop(&nch);
                vput(vp);
                goto done;
        }
        if ((error = vinvalbuf(vp, V_SAVE, 0, 0)) != 0) {
                cache_drop(&nch);
                vput(vp);
                goto done;
        }
        if (vp->v_type != VDIR) {
                cache_drop(&nch);
                vput(vp);
                error = ENOTDIR;
                goto done;
        }
        if (vp->v_mount->mnt_kern_flag & MNTK_NOSTKMNT) {
                cache_drop(&nch);
                vput(vp);
                error = EPERM;
                goto done;
        }
        vfsp = vfsconf_find_by_name(fstypename);
        if (vfsp == NULL) {
                linker_file_t lf;

                /* Only load modules for root (very important!) */
                if ((error = priv_check(td, PRIV_ROOT)) != 0) {
                        cache_drop(&nch);
                        vput(vp);
                        goto done;
                }
                error = linker_load_file(fstypename, &lf);
                if (error || lf == NULL) {
                        cache_drop(&nch);
                        vput(vp);
                        if (lf == NULL)
                                error = ENODEV;
                        goto done;
                }
                lf->userrefs++;
                /* lookup again, see if the VFS was loaded */
                vfsp = vfsconf_find_by_name(fstypename);
                if (vfsp == NULL) {
                        lf->userrefs--;
                        linker_file_unload(lf);
                        cache_drop(&nch);
                        vput(vp);
                        error = ENODEV;
                        goto done;
                }
        }
        if (hasmount) {
                cache_drop(&nch);
                vput(vp);
                error = EBUSY;
                goto done;
        }

        /*
         * Allocate and initialize the filesystem.
         */
        mp = kmalloc(sizeof(struct mount), M_MOUNT, M_ZERO|M_WAITOK);
        mount_init(mp, vfsp->vfc_vfsops);
        vfs_busy(mp, LK_NOWAIT);
        mp->mnt_vfc = vfsp;
        mp->mnt_pbuf_count = nswbuf_kva / NSWBUF_SPLIT;
        vfsp->vfc_refcount++;
        mp->mnt_stat.f_type = vfsp->vfc_typenum;
        mp->mnt_flag |= vfsp->vfc_flags & MNT_VISFLAGMASK;
        strncpy(mp->mnt_stat.f_fstypename, vfsp->vfc_name, MFSNAMELEN);
        mp->mnt_stat.f_owner = cred->cr_uid;
        lwkt_gettoken(&mp->mnt_token);
        vn_unlock(vp);
update:
        /*
         * (per-mount token acquired at this point)
         *
         * Set the mount level flags.
         */
        if (flags & MNT_RDONLY)
                mp->mnt_flag |= MNT_RDONLY;
        else if (mp->mnt_flag & MNT_RDONLY)
                mp->mnt_kern_flag |= MNTK_WANTRDWR;
        mp->mnt_flag &=~ (MNT_NOSUID | MNT_NOEXEC | MNT_NODEV |
            MNT_SYNCHRONOUS | MNT_ASYNC | MNT_NOATIME |
            MNT_NOSYMFOLLOW | MNT_IGNORE | MNT_TRIM |
            MNT_NOCLUSTERR | MNT_NOCLUSTERW | MNT_SUIDDIR |
            MNT_AUTOMOUNTED);
        mp->mnt_flag |= flags & (MNT_NOSUID | MNT_NOEXEC |
            MNT_NODEV | MNT_SYNCHRONOUS | MNT_ASYNC | MNT_FORCE |
            MNT_NOSYMFOLLOW | MNT_IGNORE | MNT_TRIM |
            MNT_NOATIME | MNT_NOCLUSTERR | MNT_NOCLUSTERW | MNT_SUIDDIR |
            MNT_AUTOMOUNTED);

        /*
         * Pre-set the mount's ALL_MPSAFE flags if specified in the vfsconf.
         * This way the initial VFS_MOUNT() call will also be MPSAFE.
         */
        if (vfsp->vfc_flags & VFCF_MPSAFE)
                mp->mnt_kern_flag |= MNTK_ALL_MPSAFE;

        /*
         * Mount the filesystem.
         * XXX The final recipients of VFS_MOUNT just overwrite the ndp they
         * get.
         */
        if (mp->mnt_flag & MNT_UPDATE) {
                error = VFS_MOUNT(mp, uap->path, uap->data, cred);
                if (mp->mnt_kern_flag & MNTK_WANTRDWR)
                        mp->mnt_flag &= ~MNT_RDONLY;
                mp->mnt_flag &=~ (MNT_UPDATE | MNT_RELOAD | MNT_FORCE);
                mp->mnt_kern_flag &=~ MNTK_WANTRDWR;
                if (error) {
                        mp->mnt_flag = flag;
                        mp->mnt_kern_flag = flag2;
                }
                lwkt_reltoken(&mp->mnt_token);
                vfs_unbusy(mp);
                vrele(vp);
                cache_drop(&nch);
                goto done;
        }
        mp->mnt_ncmounton = nch;
        error = VFS_MOUNT(mp, uap->path, uap->data, cred);
        vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);

        /*
         * Put the new filesystem on the mount list after root.  The mount
         * point gets its own mnt_ncmountpt (unless the VFS already set one
         * up) which represents the root of the mount.  The lookup code
         * detects the mount point going forward and checks the root of
         * the mount going backwards.
         *
         * It is not necessary to invalidate or purge the vnode underneath
         * because elements under the mount will be given their own glue
         * namecache record.
         */
        if (!error) {
                if (mp->mnt_ncmountpt.ncp == NULL) {
                        /* 
                         * Allocate, then unlock, but leave the ref intact.
                         * This is the mnt_refs (1) that we will retain
                         * through to the unmount.
                         */
                        cache_allocroot(&mp->mnt_ncmountpt, mp, NULL);
                        cache_unlock(&mp->mnt_ncmountpt);
                }
                vn_unlock(vp);
                cache_lock(&nch);
                nch.ncp->nc_flag |= NCF_ISMOUNTPT;
                cache_unlock(&nch);
                cache_ismounting(mp);
                vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);

                mountlist_insert(mp, MNTINS_LAST);
                vn_unlock(vp);
                checkdirs(&mp->mnt_ncmounton, &mp->mnt_ncmountpt);
                error = vfs_allocate_syncvnode(mp);
                lwkt_reltoken(&mp->mnt_token);
                vfs_unbusy(mp);
                error = VFS_START(mp, 0);
                vrele(vp);
                KNOTE(&fs_klist, VQ_MOUNT);
        } else {
                bzero(&mp->mnt_ncmounton, sizeof(mp->mnt_ncmounton));
                vn_syncer_thr_stop(mp);
                vfs_rm_vnodeops(mp, NULL, &mp->mnt_vn_coherency_ops);
                vfs_rm_vnodeops(mp, NULL, &mp->mnt_vn_journal_ops);
                vfs_rm_vnodeops(mp, NULL, &mp->mnt_vn_norm_ops);
                vfs_rm_vnodeops(mp, NULL, &mp->mnt_vn_spec_ops);
                vfs_rm_vnodeops(mp, NULL, &mp->mnt_vn_fifo_ops);
                if (mp->mnt_cred) {
                        crfree(mp->mnt_cred);
                        mp->mnt_cred = NULL;
                }
                mp->mnt_vfc->vfc_refcount--;
                lwkt_reltoken(&mp->mnt_token);
                vfs_unbusy(mp);
                kfree(mp, M_MOUNT);
                cache_drop(&nch);
                vput(vp);
        }
done:
        return (error);
}

/*
 * Scan all active processes to see if any of them have a current
 * or root directory onto which the new filesystem has just been
 * mounted. If so, replace them with the new mount point.
 *
 * Both old_nch and new_nch are ref'd on call but not locked.
 * new_nch must be temporarily locked so it can be associated with the
 * vnode representing the root of the mount point.
 */
struct checkdirs_info {
        struct nchandle old_nch;
        struct nchandle new_nch;
        struct vnode *old_vp;
        struct vnode *new_vp;
};

static int checkdirs_callback(struct proc *p, void *data);

static void
checkdirs(struct nchandle *old_nch, struct nchandle *new_nch)
{
        struct checkdirs_info info;
        struct vnode *olddp;
        struct vnode *newdp;
        struct mount *mp;

        /*
         * If the old mount point's vnode has a usecount of 1, it is not
         * being held as a descriptor anywhere.
         */
        olddp = old_nch->ncp->nc_vp;
        if (olddp == NULL || VREFCNT(olddp) == 1)
                return;

        /*
         * Force the root vnode of the new mount point to be resolved
         * so we can update any matching processes.
         */
        mp = new_nch->mount;
        if (VFS_ROOT(mp, &newdp))
                panic("mount: lost mount");
        vn_unlock(newdp);
        cache_lock(new_nch);
        vn_lock(newdp, LK_EXCLUSIVE | LK_RETRY);
        cache_setunresolved(new_nch);
        cache_setvp(new_nch, newdp);
        cache_unlock(new_nch);

        /*
         * Special handling of the root node
         */
        if (rootvnode == olddp) {
                vref(newdp);
                vfs_cache_setroot(newdp, cache_hold(new_nch));
        }

        /*
         * Pass newdp separately so the callback does not have to access
         * it via new_nch->ncp->nc_vp.
         */
        info.old_nch = *old_nch;
        info.new_nch = *new_nch;
        info.new_vp = newdp;
        allproc_scan(checkdirs_callback, &info, 0);
        vput(newdp);
}

/*
 * NOTE: callback is not MP safe because the scanned process's filedesc
 * structure can be ripped out from under us, amoung other things.
 */
static int
checkdirs_callback(struct proc *p, void *data)
{
        struct checkdirs_info *info = data;
        struct filedesc *fdp;
        struct nchandle ncdrop1;
        struct nchandle ncdrop2;
        struct vnode *vprele1;
        struct vnode *vprele2;

        if ((fdp = p->p_fd) != NULL) {
                cache_zero(&ncdrop1);
                cache_zero(&ncdrop2);
                vprele1 = NULL;
                vprele2 = NULL;

                /*
                 * MPUNSAFE - XXX fdp can be pulled out from under a
                 * foreign process.
                 *
                 * A shared filedesc is ok, we don't have to copy it
                 * because we are making this change globally.
                 */
                spin_lock(&fdp->fd_spin);
                if (fdp->fd_ncdir.mount == info->old_nch.mount &&
                    fdp->fd_ncdir.ncp == info->old_nch.ncp) {
                        vprele1 = fdp->fd_cdir;
                        vref(info->new_vp);
                        fdp->fd_cdir = info->new_vp;
                        ncdrop1 = fdp->fd_ncdir;
                        cache_copy(&info->new_nch, &fdp->fd_ncdir);
                }
                if (fdp->fd_nrdir.mount == info->old_nch.mount &&
                    fdp->fd_nrdir.ncp == info->old_nch.ncp) {
                        vprele2 = fdp->fd_rdir;
                        vref(info->new_vp);
                        fdp->fd_rdir = info->new_vp;
                        ncdrop2 = fdp->fd_nrdir;
                        cache_copy(&info->new_nch, &fdp->fd_nrdir);
                }
                spin_unlock(&fdp->fd_spin);
                if (ncdrop1.ncp)
                        cache_drop(&ncdrop1);
                if (ncdrop2.ncp)
                        cache_drop(&ncdrop2);
                if (vprele1)
                        vrele(vprele1);
                if (vprele2)
                        vrele(vprele2);
        }
        return(0);
}

/*
 * Unmount a file system.
 *
 * Note: unmount takes a path to the vnode mounted on as argument,
 * not special file (as before).
 *
 * umount_args(char *path, int flags)
 *
 * MPALMOSTSAFE
 */
int
sys_unmount(struct sysmsg *sysmsg, const struct unmount_args *uap)
{
        struct thread *td = curthread;
        struct proc *p __debugvar = td->td_proc;
        struct mount *mp = NULL;
        struct nlookupdata nd;
        char fstypename[MFSNAMELEN];
        int priv = 0;
        int error;
        struct ucred *cred;

        cred = td->td_ucred;

        KKASSERT(p);

        /* We do not allow user umounts inside a jail for now */
        if (usermount && jailed(cred)) {
                error = EPERM;
                goto done;
        }

        error = nlookup_init(&nd, uap->path, UIO_USERSPACE,
                             NLC_FOLLOW | NLC_IGNBADDIR);
        if (error == 0)
                error = nlookup(&nd);
        if (error)
                goto out;

        mp = nd.nl_nch.mount;

        /* Figure out the fsname in order to select proper privs */
        ksnprintf(fstypename, MFSNAMELEN, "%s", mp->mnt_vfc->vfc_name);
        priv = get_fspriv(fstypename);

        if (usermount == 0 && (error = priv_check(td, priv))) {
                nlookup_done(&nd);
                goto done;
        }

        /*
         * Only root, or the user that did the original mount is
         * permitted to unmount this filesystem.
         */
        if ((mp->mnt_stat.f_owner != td->td_ucred->cr_uid) &&
            (error = priv_check(td, priv)))
                goto out;

        /*
         * Don't allow unmounting the root file system.
         */
        if (mp->mnt_flag & MNT_ROOTFS) {
                error = EINVAL;
                goto out;
        }

        /*
         * Must be the root of the filesystem
         */
        if (nd.nl_nch.ncp != mp->mnt_ncmountpt.ncp) {
                error = EINVAL;
                goto out;
        }

        /* Check if this mount belongs to this prison */
        if (jailed(cred) && mp->mnt_cred && (!mp->mnt_cred->cr_prison ||
                mp->mnt_cred->cr_prison != cred->cr_prison)) {
                kprintf("mountpoint %s does not belong to this jail\n",
                    uap->path);
                error = EPERM;
                goto out;
        }

        /*
         * If no error try to issue the unmount.  We lose our cache
         * ref when we call nlookup_done so we must hold the mount point
         * to prevent use-after-free races.
         */
out:
        if (error == 0) {
                mount_hold(mp);
                nlookup_done(&nd);
                error = dounmount(mp, uap->flags, 0);
                mount_drop(mp);
        } else {
                nlookup_done(&nd);
        }
done:
        return (error);
}

/*
 * Do the actual file system unmount (interlocked against the mountlist
 * token and mp->mnt_token).
 */
static int
dounmount_interlock(struct mount *mp)
{
        if (mp->mnt_kern_flag & MNTK_UNMOUNT)
                return (EBUSY);
        mp->mnt_kern_flag |= MNTK_UNMOUNT;
        return(0);
}

static int
unmount_allproc_cb(struct proc *p, void *arg)
{
        struct mount *mp;

        if (p->p_textnch.ncp == NULL)
                return 0;

        mp = (struct mount *)arg;
        if (p->p_textnch.mount == mp)
                cache_drop(&p->p_textnch);

        return 0;
}

/*
 * The guts of the unmount code.  The mount owns one ref and one hold
 * count.  If we successfully interlock the unmount, those refs are ours.
 * (The ref is from mnt_ncmountpt).
 *
 * When halting we shortcut certain mount types such as devfs by not actually
 * issuing the VFS_SYNC() or VFS_UNMOUNT().  They are still disconnected
 * from the mountlist so higher-level filesytems can unmount cleanly.
 *
 * The mount types that allow QUICKHALT are: devfs, tmpfs, procfs.
 */
int
dounmount(struct mount *mp, int flags, int halting)
{
        struct namecache *ncp;
        struct nchandle nch;
        struct vnode *vp;
        int error;
        int async_flag;
        int lflags;
        int freeok = 1;
        int hadsyncer = 0;
        int retry;
        int quickhalt;

        lwkt_gettoken(&mp->mnt_token);

        /*
         * When halting, certain mount points can essentially just
         * be unhooked and otherwise ignored.
         */
        if (halting && (mp->mnt_kern_flag & MNTK_QUICKHALT)) {
                quickhalt = 1;
                freeok = 0;
        } else {
                quickhalt = 0;
        }


        /*
         * Exclusive access for unmounting purposes.
         */
        if ((error = mountlist_interlock(dounmount_interlock, mp)) != 0)
                goto out;

        /*
         * We now 'own' the last mp->mnt_refs
         *
         * Allow filesystems to detect that a forced unmount is in progress.
         */
        if (flags & MNT_FORCE)
                mp->mnt_kern_flag |= MNTK_UNMOUNTF;
        lflags = LK_EXCLUSIVE | ((flags & MNT_FORCE) ? 0 : LK_TIMELOCK);
        error = lockmgr(&mp->mnt_lock, lflags);
        if (error) {
                mp->mnt_kern_flag &= ~(MNTK_UNMOUNT | MNTK_UNMOUNTF);
                if (mp->mnt_kern_flag & MNTK_MWAIT) {
                        mp->mnt_kern_flag &= ~MNTK_MWAIT;
                        wakeup(mp);
                }
                goto out;
        }

        if (mp->mnt_flag & MNT_EXPUBLIC)
                vfs_setpublicfs(NULL, NULL, NULL);

        vfs_msync(mp, MNT_WAIT);
        async_flag = mp->mnt_flag & MNT_ASYNC;
        mp->mnt_flag &=~ MNT_ASYNC;

        /*
         * Decomission our special mnt_syncer vnode.  This also stops
         * the vnlru code.  If we are unable to unmount we recommission
         * the vnode.
         *
         * Then sync the filesystem.
         */
        if ((vp = mp->mnt_syncer) != NULL) {
                mp->mnt_syncer = NULL;
                atomic_set_int(&vp->v_refcnt, VREF_FINALIZE);
                vrele(vp);
                hadsyncer = 1;
        }

        /*
         * Sync normally-mounted filesystem.
         */
        if (quickhalt == 0) {
                if ((mp->mnt_flag & MNT_RDONLY) == 0)
                        VFS_SYNC(mp, MNT_WAIT);
        }

        /*
         * nchandle records ref the mount structure.  Expect a count of 1
         * (our mount->mnt_ncmountpt).
         *
         * Scans can get temporary refs on a mountpoint (thought really
         * heavy duty stuff like cache_findmount() do not).
         */
        for (retry = 0; (retry < 10 || debug_unmount); ++retry) {
                /*
                 * Invalidate the namecache topology under the mount.
                 * nullfs mounts alias a real mount's namecache topology
                 * and it should not be invalidated in that case.
                 */
                if ((mp->mnt_kern_flag & MNTK_NCALIASED) == 0) {
                        cache_lock(&mp->mnt_ncmountpt);
                        cache_inval(&mp->mnt_ncmountpt,
                                    CINV_DESTROY|CINV_CHILDREN);
                        cache_unlock(&mp->mnt_ncmountpt);
                }

                /*
                 * Clear pcpu caches
                 */
                cache_unmounting(mp);
                if (mp->mnt_refs != 1)
                        cache_clearmntcache(mp);

                /*
                 * Break out if we are good.  Don't count ncp refs if the
                 * mount is aliased.
                 */
                ncp = (mp->mnt_kern_flag & MNTK_NCALIASED) ?
                      NULL : mp->mnt_ncmountpt.ncp;
                if (mp->mnt_refs == 1 &&
                    (ncp == NULL || (ncp->nc_refs == 1 &&
                                     TAILQ_FIRST(&ncp->nc_list) == NULL))) {
                        break;
                }

                /*
                 * If forcing the unmount, clean out any p->p_textnch
                 * nchandles that match this mount.
                 */
                if (flags & MNT_FORCE)
                        allproc_scan(&unmount_allproc_cb, mp, 0);

                /*
                 * Sleep and retry.
                 */
                tsleep(&mp->mnt_refs, 0, "mntbsy", hz / 10 + 1);
                if ((retry & 15) == 15) {
                        mount_warning(mp,
                                      "(%p) debug - retry %d, "
                                      "%d namecache refs, %d mount refs",
                                      mp, retry,
                                      (ncp ? ncp->nc_refs - 1 : 0),
                                      mp->mnt_refs - 1);
                }
        }

        error = 0;
        ncp = (mp->mnt_kern_flag & MNTK_NCALIASED) ?
              NULL : mp->mnt_ncmountpt.ncp;
        if (mp->mnt_refs != 1 ||
            (ncp != NULL && (ncp->nc_refs != 1 ||
                             TAILQ_FIRST(&ncp->nc_list)))) {
                mount_warning(mp,
                              "(%p): %d namecache refs, %d mount refs "
                              "still present",
                              mp,
                              (ncp ? ncp->nc_refs - 1 : 0),
                              mp->mnt_refs - 1);
                if (flags & MNT_FORCE) {
                        freeok = 0;
                        mount_warning(mp, "forcing unmount\n");
                } else {
                        error = EBUSY;
                }
        }

        /*
         * So far so good, sync the filesystem once more and
         * call the VFS unmount code if the sync succeeds.
         */
        if (error == 0 && quickhalt == 0) {
                if (mp->mnt_flag & MNT_RDONLY) {
                        error = VFS_UNMOUNT(mp, flags);
                } else {
                        error = VFS_SYNC(mp, MNT_WAIT);
                        if (error == 0 ||               /* no error */
                            error == EOPNOTSUPP ||      /* no sync avail */
                            (flags & MNT_FORCE)) {      /* force anyway */
                                error = VFS_UNMOUNT(mp, flags);
                        }
                }
                if (error) {
                        mount_warning(mp,
                                      "(%p) unmount: vfs refused to unmount, "
                                      "error %d",
                                      mp, error);
                }
        }

        /*
         * If an error occurred we can still recover, restoring the
         * syncer vnode and misc flags.
         */
        if (error) {
                if (mp->mnt_syncer == NULL && hadsyncer)
                        vfs_allocate_syncvnode(mp);
                mp->mnt_kern_flag &= ~(MNTK_UNMOUNT | MNTK_UNMOUNTF);
                mp->mnt_flag |= async_flag;
                lockmgr(&mp->mnt_lock, LK_RELEASE);
                if (mp->mnt_kern_flag & MNTK_MWAIT) {
                        mp->mnt_kern_flag &= ~MNTK_MWAIT;
                        wakeup(mp);
                }
                goto out;
        }
        /*
         * Clean up any journals still associated with the mount after
         * filesystem activity has ceased.
         */
        journal_remove_all_journals(mp, 
            ((flags & MNT_FORCE) ? MC_JOURNAL_STOP_IMM : 0));

        mountlist_remove(mp);

        /*
         * Remove any installed vnode ops here so the individual VFSs don't
         * have to.
         *
         * mnt_refs should go to zero when we scrap mnt_ncmountpt.
         *
         * When quickhalting we have to keep these intact because the
         * underlying vnodes have not been destroyed, and some might be
         * dirty.
         */
        if (quickhalt == 0) {
                vfs_rm_vnodeops(mp, NULL, &mp->mnt_vn_coherency_ops);
                vfs_rm_vnodeops(mp, NULL, &mp->mnt_vn_journal_ops);
                vfs_rm_vnodeops(mp, NULL, &mp->mnt_vn_norm_ops);
                vfs_rm_vnodeops(mp, NULL, &mp->mnt_vn_spec_ops);
                vfs_rm_vnodeops(mp, NULL, &mp->mnt_vn_fifo_ops);
        }

        if (mp->mnt_ncmountpt.ncp != NULL) {
                nch = mp->mnt_ncmountpt;
                cache_zero(&mp->mnt_ncmountpt);
                cache_clrmountpt(&nch);
                cache_drop(&nch);
        }
        if (mp->mnt_ncmounton.ncp != NULL) {
                cache_unmounting(mp);
                nch = mp->mnt_ncmounton;
                cache_zero(&mp->mnt_ncmounton);
                cache_clrmountpt(&nch);
                cache_drop(&nch);
        }

        if (mp->mnt_cred) {
                crfree(mp->mnt_cred);
                mp->mnt_cred = NULL;
        }

        mp->mnt_vfc->vfc_refcount--;

        /*
         * If not quickhalting the mount, we expect there to be no
         * vnodes left.
         */
        if (quickhalt == 0 && !TAILQ_EMPTY(&mp->mnt_nvnodelist))
                panic("unmount: dangling vnode");

        /*
         * Release the lock
         */
        lockmgr(&mp->mnt_lock, LK_RELEASE);
        if (mp->mnt_kern_flag & MNTK_MWAIT) {
                mp->mnt_kern_flag &= ~MNTK_MWAIT;
                wakeup(mp);
        }

        /*
         * If we reach here and freeok != 0 we must free the mount.
         * mnt_refs should already have dropped to 0, so if it is not
         * zero we must cycle the caches and wait.
         *
         * When we are satisfied that the mount has disconnected we can
         * drop the hold on the mp that represented the mount (though the
         * caller might actually have another, so the caller's drop may
         * do the actual free).
         */
        if (freeok) {
                if (mp->mnt_refs > 0)
                        cache_clearmntcache(mp);
                while (mp->mnt_refs > 0) {
                        cache_unmounting(mp);
                        wakeup(mp);
                        tsleep(&mp->mnt_refs, 0, "umntrwait", hz / 10 + 1);
                        cache_clearmntcache(mp);
                }
                lwkt_reltoken(&mp->mnt_token);
                mount_drop(mp);
                mp = NULL;
        } else {
                cache_clearmntcache(mp);
        }
        error = 0;
        KNOTE(&fs_klist, VQ_UNMOUNT);
out:
        if (mp)
                lwkt_reltoken(&mp->mnt_token);
        return (error);
}

static
void
mount_warning(struct mount *mp, const char *ctl, ...)
{
        char *ptr;
        char *buf;
        __va_list va;

        __va_start(va, ctl);
        if (cache_fullpath(NULL, &mp->mnt_ncmounton, NULL,
                           &ptr, &buf, 0) == 0) {
                kprintf("unmount(%s): ", ptr);
                kvprintf(ctl, va);
                kprintf("\n");
                kfree(buf, M_TEMP);
        } else {
                kprintf("unmount(%p", mp);
                if (mp->mnt_ncmounton.ncp && mp->mnt_ncmounton.ncp->nc_name)
                        kprintf(",%s", mp->mnt_ncmounton.ncp->nc_name);
                kprintf("): ");
                kvprintf(ctl, va);
                kprintf("\n");
        }
        __va_end(va);
}

/*
 * Shim cache_fullpath() to handle the case where a process is chrooted into
 * a subdirectory of a mount.  In this case if the root mount matches the
 * process root directory's mount we have to specify the process's root
 * directory instead of the mount point, because the mount point might
 * be above the root directory.
 */
static
int
mount_path(struct proc *p, struct mount *mp, char **rb, char **fb)
{
        struct nchandle *nch;

        if (p && p->p_fd->fd_nrdir.mount == mp)
                nch = &p->p_fd->fd_nrdir;
        else
                nch = &mp->mnt_ncmountpt;
        return(cache_fullpath(p, nch, NULL, rb, fb, 0));
}

/*
 * Sync each mounted filesystem.
 */

#ifdef DEBUG
static int syncprt = 0;
SYSCTL_INT(_debug, OID_AUTO, syncprt, CTLFLAG_RW, &syncprt, 0, "");
#endif /* DEBUG */

static int sync_callback(struct mount *mp, void *data);

int
sys_sync(struct sysmsg *sysmsg, const struct sync_args *uap)
{
        mountlist_scan(sync_callback, NULL, MNTSCAN_FORWARD);
        return (0);
}

static
int
sync_callback(struct mount *mp, void *data __unused)
{
        int asyncflag;

        if ((mp->mnt_flag & MNT_RDONLY) == 0) {
                lwkt_gettoken(&mp->mnt_token);
                asyncflag = mp->mnt_flag & MNT_ASYNC;
                mp->mnt_flag &= ~MNT_ASYNC;
                lwkt_reltoken(&mp->mnt_token);
                vfs_msync(mp, MNT_NOWAIT);
                VFS_SYNC(mp, MNT_NOWAIT);
                lwkt_gettoken(&mp->mnt_token);
                mp->mnt_flag |= asyncflag;
                lwkt_reltoken(&mp->mnt_token);
        }
        return(0);
}

/* XXX PRISON: could be per prison flag */
static int prison_quotas;
#if 0
SYSCTL_INT(_kern_prison, OID_AUTO, quotas, CTLFLAG_RW, &prison_quotas, 0, "");
#endif

/*
 *  quotactl_args(char *path, int fcmd, int uid, caddr_t arg)
 *
 * Change filesystem quotas.
 *
 * MPALMOSTSAFE
 */
int
sys_quotactl(struct sysmsg *sysmsg, const struct quotactl_args *uap)
{
        struct nlookupdata nd;
        struct thread *td;
        struct mount *mp;
        int error;

        td = curthread;
        if (td->td_ucred->cr_prison && !prison_quotas) {
                error = EPERM;
                goto done;
        }

        error = nlookup_init(&nd, uap->path, UIO_USERSPACE, NLC_FOLLOW);
        if (error == 0)
                error = nlookup(&nd);
        if (error == 0) {
                mp = nd.nl_nch.mount;
                error = VFS_QUOTACTL(mp, uap->cmd, uap->uid,
                                    uap->arg, nd.nl_cred);
        }
        nlookup_done(&nd);
done:
        return (error);
}

/*
 * mountctl(char *path, int op, int fd, const void *ctl, int ctllen,
 *              void *buf, int buflen)
 *
 * This function operates on a mount point and executes the specified
 * operation using the specified control data, and possibly returns data.
 *
 * The actual number of bytes stored in the result buffer is returned, 0
 * if none, otherwise an error is returned.
 *
 * MPALMOSTSAFE
 */
int
sys_mountctl(struct sysmsg *sysmsg, const struct mountctl_args *uap)
{
        struct thread *td = curthread;
        struct file *fp;
        void *ctl = NULL;
        void *buf = NULL;
        char *path = NULL;
        int error;

        /*
         * Sanity and permissions checks.  We must be root.
         */
        if (td->td_ucred->cr_prison != NULL)
                return (EPERM);
        if ((uap->op != MOUNTCTL_MOUNTFLAGS) &&
            (error = priv_check(td, PRIV_ROOT)) != 0)
                return (error);

        /*
         * Argument length checks
         */
        if (uap->ctllen < 0 || uap->ctllen > 1024)
                return (EINVAL);
        if (uap->buflen < 0 || uap->buflen > 16 * 1024)
                return (EINVAL);
        if (uap->path == NULL)
                return (EINVAL);

        /*
         * Allocate the necessary buffers and copyin data
         */
        path = objcache_get(namei_oc, M_WAITOK);
        error = copyinstr(uap->path, path, MAXPATHLEN, NULL);
        if (error)
                goto done;

        if (uap->ctllen) {
                ctl = kmalloc(uap->ctllen + 1, M_TEMP, M_WAITOK|M_ZERO);
                error = copyin(uap->ctl, ctl, uap->ctllen);
                if (error)
                        goto done;
        }
        if (uap->buflen)
                buf = kmalloc(uap->buflen + 1, M_TEMP, M_WAITOK|M_ZERO);

        /*
         * Validate the descriptor
         */
        if (uap->fd >= 0) {
                fp = holdfp(td, uap->fd, -1);
                if (fp == NULL) {
                        error = EBADF;
                        goto done;
                }
        } else {
                fp = NULL;
        }

        /*
         * Execute the internal kernel function and clean up.
         */
        error = kern_mountctl(path, uap->op, fp, ctl, uap->ctllen,
                              buf, uap->buflen, &sysmsg->sysmsg_result);
        if (fp)
                dropfp(td, uap->fd, fp);
        if (error == 0 && sysmsg->sysmsg_result > 0)
                error = copyout(buf, uap->buf, sysmsg->sysmsg_result);
done:
        if (path)
                objcache_put(namei_oc, path);
        if (ctl)
                kfree(ctl, M_TEMP);
        if (buf)
                kfree(buf, M_TEMP);
        return (error);
}

/*
 * Execute a mount control operation by resolving the path to a mount point
 * and calling vop_mountctl().  
 *
 * Use the mount point from the nch instead of the vnode so nullfs mounts
 * can properly spike the VOP.
 */
int
kern_mountctl(const char *path, int op, struct file *fp, 
                const void *ctl, int ctllen, 
                void *buf, int buflen, int *res)
{
        struct vnode *vp;
        struct nlookupdata nd;
        struct nchandle nch;
        struct mount *mp;
        int error;

        *res = 0;
        vp = NULL;
        error = nlookup_init(&nd, path, UIO_SYSSPACE, NLC_FOLLOW);
        if (error)
                return (error);
        error = nlookup(&nd);
        if (error) {
                nlookup_done(&nd);
                return (error);
        }
        error = cache_vget(&nd.nl_nch, nd.nl_cred, LK_EXCLUSIVE, &vp);
        if (error) {
                nlookup_done(&nd);
                return (error);
        }

        /*
         * Yes, all this is needed to use the nch.mount below, because
         * we must maintain a ref on the mount to avoid ripouts (e.g.
         * due to heavy mount/unmount use by synth or poudriere).
         */
        nch = nd.nl_nch;
        cache_zero(&nd.nl_nch);
        cache_unlock(&nch);
        nlookup_done(&nd);
        vn_unlock(vp);

        mp = nch.mount;

        /*
         * Must be the root of the filesystem
         */
        if ((vp->v_flag & (VROOT|VPFSROOT)) == 0) {
                cache_drop(&nch);
                vrele(vp);
                return (EINVAL);
        }
        if (mp == NULL || mp->mnt_kern_flag & MNTK_UNMOUNT) {
                kprintf("kern_mountctl: Warning, \"%s\" racing unmount\n",
                        path);
                cache_drop(&nch);
                vrele(vp);
                return (EINVAL);
        }
        error = vop_mountctl(mp->mnt_vn_use_ops, vp, op, fp, ctl, ctllen,
                             buf, buflen, res);
        vrele(vp);
        cache_drop(&nch);

        return (error);
}

int
kern_statfs(struct nlookupdata *nd, struct statfs *buf)
{
        struct thread *td = curthread;
        struct proc *p = td->td_proc;
        struct mount *mp;
        struct statfs *sp;
        char *fullpath, *freepath;
        int error;

        if ((error = nlookup(nd)) != 0)
                return (error);
        mp = nd->nl_nch.mount;
        sp = &mp->mnt_stat;

        /*
         * Ignore refresh error, user should have visibility.
         * This can happen if a NFS mount goes bad (e.g. server
         * revokes perms or goes down).
         */
        error = VFS_STATFS(mp, sp, nd->nl_cred);
        /* ignore error */

        error = mount_path(p, mp, &fullpath, &freepath);
        if (error)
                return(error);
        bzero(sp->f_mntonname, sizeof(sp->f_mntonname));
        strlcpy(sp->f_mntonname, fullpath, sizeof(sp->f_mntonname));
        kfree(freepath, M_TEMP);

        sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
        bcopy(sp, buf, sizeof(*buf));
        /* Only root should have access to the fsid's. */
        if (priv_check(td, PRIV_ROOT))
                buf->f_fsid.val[0] = buf->f_fsid.val[1] = 0;
        return (0);
}

/*
 * statfs_args(char *path, struct statfs *buf)
 *
 * Get filesystem statistics.
 */
int
sys_statfs(struct sysmsg *sysmsg, const struct statfs_args *uap)
{
        struct nlookupdata nd;
        struct statfs buf;
        int error;

        error = nlookup_init(&nd, uap->path, UIO_USERSPACE, NLC_FOLLOW);
        if (error == 0)
                error = kern_statfs(&nd, &buf);
        nlookup_done(&nd);
        if (error == 0)
                error = copyout(&buf, uap->buf, sizeof(*uap->buf));
        return (error);
}

int
kern_fstatfs(int fd, struct statfs *buf)
{
        struct thread *td = curthread;
        struct proc *p = td->td_proc;
        struct file *fp;
        struct mount *mp;
        struct statfs *sp;
        char *fullpath, *freepath;
        int error;

        KKASSERT(p);
        if ((error = holdvnode(td, fd, &fp)) != 0)
                return (error);

        /*
         * Try to use mount info from any overlays rather than the
         * mount info for the underlying vnode, otherwise we will
         * fail when operating on null-mounted paths inside a chroot.
         */
        if ((mp = fp->f_nchandle.mount) == NULL)
                mp = ((struct vnode *)fp->f_data)->v_mount;
        if (mp == NULL) {
                error = EBADF;
                goto done;
        }
        if (fp->f_cred == NULL) {
                error = EINVAL;
                goto done;
        }

        /*
         * Ignore refresh error, user should have visibility.
         * This can happen if a NFS mount goes bad (e.g. server
         * revokes perms or goes down).
         */
        sp = &mp->mnt_stat;
        error = VFS_STATFS(mp, sp, fp->f_cred);

        if ((error = mount_path(p, mp, &fullpath, &freepath)) != 0)
                goto done;
        bzero(sp->f_mntonname, sizeof(sp->f_mntonname));
        strlcpy(sp->f_mntonname, fullpath, sizeof(sp->f_mntonname));
        kfree(freepath, M_TEMP);

        sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
        bcopy(sp, buf, sizeof(*buf));

        /* Only root should have access to the fsid's. */
        if (priv_check(td, PRIV_ROOT))
                buf->f_fsid.val[0] = buf->f_fsid.val[1] = 0;
        error = 0;
done:
        fdrop(fp);
        return (error);
}

/*
 * fstatfs_args(int fd, struct statfs *buf)
 *
 * Get filesystem statistics.
 */
int
sys_fstatfs(struct sysmsg *sysmsg, const struct fstatfs_args *uap)
{
        struct statfs buf;
        int error;

        error = kern_fstatfs(uap->fd, &buf);

        if (error == 0)
                error = copyout(&buf, uap->buf, sizeof(*uap->buf));
        return (error);
}

int
kern_statvfs(struct nlookupdata *nd, struct statvfs *buf)
{
        struct mount *mp;
        struct statvfs *sp;
        int error;

        if ((error = nlookup(nd)) != 0)
                return (error);
        mp = nd->nl_nch.mount;
        sp = &mp->mnt_vstat;
        if ((error = VFS_STATVFS(mp, sp, nd->nl_cred)) != 0)
                return (error);

        sp->f_flag = 0;
        if (mp->mnt_flag & MNT_RDONLY)
                sp->f_flag |= ST_RDONLY;
        if (mp->mnt_flag & MNT_NOSUID)
                sp->f_flag |= ST_NOSUID;
        bcopy(sp, buf, sizeof(*buf));
        return (0);
}

/*
 * statfs_args(char *path, struct statfs *buf)
 *
 * Get filesystem statistics.
 */
int
sys_statvfs(struct sysmsg *sysmsg, const struct statvfs_args *uap)
{
        struct nlookupdata nd;
        struct statvfs buf;
        int error;

        error = nlookup_init(&nd, uap->path, UIO_USERSPACE, NLC_FOLLOW);
        if (error == 0)
                error = kern_statvfs(&nd, &buf);
        nlookup_done(&nd);
        if (error == 0)
                error = copyout(&buf, uap->buf, sizeof(*uap->buf));
        return (error);
}

int
kern_fstatvfs(int fd, struct statvfs *buf)
{
        struct thread *td = curthread;
        struct file *fp;
        struct mount *mp;
        struct statvfs *sp;
        int error;

        if ((error = holdvnode(td, fd, &fp)) != 0)
                return (error);
        if ((mp = fp->f_nchandle.mount) == NULL)
                mp = ((struct vnode *)fp->f_data)->v_mount;
        if (mp == NULL) {
                error = EBADF;
                goto done;
        }
        if (fp->f_cred == NULL) {
                error = EINVAL;
                goto done;
        }
        sp = &mp->mnt_vstat;
        if ((error = VFS_STATVFS(mp, sp, fp->f_cred)) != 0)
                goto done;

        sp->f_flag = 0;
        if (mp->mnt_flag & MNT_RDONLY)
                sp->f_flag |= ST_RDONLY;
        if (mp->mnt_flag & MNT_NOSUID)
                sp->f_flag |= ST_NOSUID;

        bcopy(sp, buf, sizeof(*buf));
        error = 0;
done:
        fdrop(fp);
        return (error);
}

/*
 * fstatfs_args(int fd, struct statfs *buf)
 *
 * Get filesystem statistics.
 */
int
sys_fstatvfs(struct sysmsg *sysmsg, const struct fstatvfs_args *uap)
{
        struct statvfs buf;
        int error;

        error = kern_fstatvfs(uap->fd, &buf);

        if (error == 0)
                error = copyout(&buf, uap->buf, sizeof(*uap->buf));
        return (error);
}

/*
 * getfsstat_args(struct statfs *buf, long bufsize, int flags)
 *
 * Get statistics on all filesystems.
 */

struct getfsstat_info {
        struct statfs *sfsp;
        long count;
        long maxcount;
        int error;
        int flags;
        struct thread *td;
};

static int getfsstat_callback(struct mount *, void *);

int
sys_getfsstat(struct sysmsg *sysmsg, const struct getfsstat_args *uap)
{
        struct thread *td = curthread;
        struct getfsstat_info info;

        bzero(&info, sizeof(info));

        info.maxcount = uap->bufsize / sizeof(struct statfs);
        info.sfsp = uap->buf;
        info.count = 0;
        info.flags = uap->flags;
        info.td = td;

        mountlist_scan(getfsstat_callback, &info, MNTSCAN_FORWARD);
        if (info.sfsp && info.count > info.maxcount)
                sysmsg->sysmsg_result = info.maxcount;
        else
                sysmsg->sysmsg_result = info.count;
        return (info.error);
}

static int
getfsstat_callback(struct mount *mp, void *data)
{
        struct getfsstat_info *info = data;
        struct statfs *sp;
        char *freepath;
        char *fullpath;
        int error;

        if (info->td->td_proc && !chroot_visible_mnt(mp, info->td->td_proc))
                return(0);

        if (info->sfsp && info->count < info->maxcount) {
                sp = &mp->mnt_stat;

                /*
                 * If MNT_NOWAIT or MNT_LAZY is specified, do not
                 * refresh the fsstat cache. MNT_NOWAIT or MNT_LAZY
                 * overrides MNT_WAIT.
                 *
                 * Ignore refresh error, user should have visibility.
                 * This can happen if a NFS mount goes bad (e.g. server
                 * revokes perms or goes down).
                 */
                if (((info->flags & (MNT_LAZY|MNT_NOWAIT)) == 0 ||
                    (info->flags & MNT_WAIT)) &&
                    (error = VFS_STATFS(mp, sp, info->td->td_ucred))) {
                        /* ignore error */
                }
                sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;

                error = mount_path(info->td->td_proc, mp, &fullpath, &freepath);
                if (error) {
                        info->error = error;
                        return(-1);
                }
                bzero(sp->f_mntonname, sizeof(sp->f_mntonname));
                strlcpy(sp->f_mntonname, fullpath, sizeof(sp->f_mntonname));
                kfree(freepath, M_TEMP);

                error = copyout(sp, info->sfsp, sizeof(*sp));
                if (error) {
                        info->error = error;
                        return (-1);
                }
                ++info->sfsp;
        }
        info->count++;
        return(0);
}

/*
 * getvfsstat_args(struct statfs *buf, struct statvfs *vbuf,
                   long bufsize, int flags)
 *
 * Get statistics on all filesystems.
 */

struct getvfsstat_info {
        struct statfs *sfsp;
        struct statvfs *vsfsp;
        long count;
        long maxcount;
        int error;
        int flags;
        struct thread *td;
};

static int getvfsstat_callback(struct mount *, void *);

int
sys_getvfsstat(struct sysmsg *sysmsg, const struct getvfsstat_args *uap)
{
        struct thread *td = curthread;
        struct getvfsstat_info info;

        bzero(&info, sizeof(info));

        info.maxcount = uap->vbufsize / sizeof(struct statvfs);
        info.sfsp = uap->buf;
        info.vsfsp = uap->vbuf;
        info.count = 0;
        info.flags = uap->flags;
        info.td = td;

        mountlist_scan(getvfsstat_callback, &info, MNTSCAN_FORWARD);
        if (info.vsfsp && info.count > info.maxcount)
                sysmsg->sysmsg_result = info.maxcount;
        else
                sysmsg->sysmsg_result = info.count;
        return (info.error);
}

static int
getvfsstat_callback(struct mount *mp, void *data)
{
        struct getvfsstat_info *info = data;
        struct statfs *sp;
        struct statvfs *vsp;
        char *freepath;
        char *fullpath;
        int error;

        if (info->td->td_proc && !chroot_visible_mnt(mp, info->td->td_proc))
                return(0);

        if (info->vsfsp && info->count < info->maxcount) {
                sp = &mp->mnt_stat;
                vsp = &mp->mnt_vstat;

                /*
                 * If MNT_NOWAIT or MNT_LAZY is specified, do not
                 * refresh the fsstat cache. MNT_NOWAIT or MNT_LAZY
                 * overrides MNT_WAIT.
                 *
                 * Ignore refresh error, user should have visibility.
                 * This can happen if a NFS mount goes bad (e.g. server
                 * revokes perms or goes down).
                 */
                if (((info->flags & (MNT_LAZY|MNT_NOWAIT)) == 0 ||
                    (info->flags & MNT_WAIT)) &&
                    (error = VFS_STATFS(mp, sp, info->td->td_ucred))) {
                        /* ignore error */
                }
                sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;

                if (((info->flags & (MNT_LAZY|MNT_NOWAIT)) == 0 ||
                    (info->flags & MNT_WAIT)) &&
                    (error = VFS_STATVFS(mp, vsp, info->td->td_ucred))) {
                        /* ignore error */
                }
                vsp->f_flag = 0;
                if (mp->mnt_flag & MNT_RDONLY)
                        vsp->f_flag |= ST_RDONLY;
                if (mp->mnt_flag & MNT_NOSUID)
                        vsp->f_flag |= ST_NOSUID;

                error = mount_path(info->td->td_proc, mp, &fullpath, &freepath);
                if (error) {
                        info->error = error;
                        return(-1);
                }
                bzero(sp->f_mntonname, sizeof(sp->f_mntonname));
                strlcpy(sp->f_mntonname, fullpath, sizeof(sp->f_mntonname));
                kfree(freepath, M_TEMP);

                error = copyout(sp, info->sfsp, sizeof(*sp));
                if (error == 0)
                        error = copyout(vsp, info->vsfsp, sizeof(*vsp));
                if (error) {
                        info->error = error;
                        return (-1);
                }
                ++info->sfsp;
                ++info->vsfsp;
        }
        info->count++;
        return(0);
}


/*
 * fchdir_args(int fd)
 *
 * Change current working directory to a given file descriptor.
 */
int
sys_fchdir(struct sysmsg *sysmsg, const struct fchdir_args *uap)
{
        struct thread *td = curthread;
        struct proc *p = td->td_proc;
        struct filedesc *fdp = p->p_fd;
        struct vnode *vp, *ovp;
        struct mount *mp;
        struct file *fp;
        struct nchandle nch, onch, tnch;
        int error;

        if ((error = holdvnode(td, uap->fd, &fp)) != 0)
                return (error);
        lwkt_gettoken(&p->p_token);
        vp = (struct vnode *)fp->f_data;
        vref(vp);
        vn_lock(vp, LK_SHARED | LK_RETRY);
        if (fp->f_nchandle.ncp == NULL)
                error = ENOTDIR;
        else
                error = checkvp_chdir(vp, td);
        if (error) {
                vput(vp);
                goto done;
        }
        cache_copy(&fp->f_nchandle, &nch);

        /*
         * If the ncp has become a mount point, traverse through
         * the mount point.
         */

        while (!error && (nch.ncp->nc_flag & NCF_ISMOUNTPT) &&
               (mp = cache_findmount(&nch)) != NULL
        ) {
                error = nlookup_mp(mp, &tnch);
                if (error == 0) {
                        cache_unlock(&tnch);    /* leave ref intact */
                        vput(vp);
                        vp = tnch.ncp->nc_vp;
                        error = vget(vp, LK_SHARED);
                        KKASSERT(error == 0);
                        cache_drop(&nch);
                        nch = tnch;
                }
                cache_dropmount(mp);
        }
        if (error == 0) {
                spin_lock(&fdp->fd_spin);
                ovp = fdp->fd_cdir;
                onch = fdp->fd_ncdir;
                fdp->fd_cdir = vp;
                fdp->fd_ncdir = nch;
                spin_unlock(&fdp->fd_spin);
                vn_unlock(vp);          /* leave ref intact */
                cache_drop(&onch);
                vrele(ovp);
        } else {
                cache_drop(&nch);
                vput(vp);
        }
        fdrop(fp);
done:
        lwkt_reltoken(&p->p_token);
        return (error);
}

int
kern_chdir(struct nlookupdata *nd)
{
        struct thread *td = curthread;
        struct proc *p = td->td_proc;
        struct filedesc *fdp = p->p_fd;
        struct vnode *vp, *ovp;
        struct nchandle onch;
        int error;

        nd->nl_flags |= NLC_SHAREDLOCK;
        if ((error = nlookup(nd)) != 0)
                return (error);
        if ((vp = nd->nl_nch.ncp->nc_vp) == NULL)
                return (ENOENT);
        if ((error = vget(vp, LK_SHARED)) != 0)
                return (error);

        lwkt_gettoken(&p->p_token);
        error = checkvp_chdir(vp, td);
        vn_unlock(vp);
        if (error == 0) {
                spin_lock(&fdp->fd_spin);
                ovp = fdp->fd_cdir;
                onch = fdp->fd_ncdir;
                fdp->fd_ncdir = nd->nl_nch;
                fdp->fd_cdir = vp;
                spin_unlock(&fdp->fd_spin);
                cache_unlock(&nd->nl_nch);      /* leave reference intact */
                cache_drop(&onch);
                vrele(ovp);
                cache_zero(&nd->nl_nch);
        } else {
                vrele(vp);
        }
        lwkt_reltoken(&p->p_token);
        return (error);
}

/*
 * chdir_args(char *path)
 *
 * Change current working directory (``.'').
 */
int
sys_chdir(struct sysmsg *sysmsg, const struct chdir_args *uap)
{
        struct nlookupdata nd;
        int error;

        error = nlookup_init(&nd, uap->path, UIO_USERSPACE, NLC_FOLLOW);
        if (error == 0)
                error = kern_chdir(&nd);
        nlookup_done(&nd);
        return (error);
}

/*
 * Helper function for raised chroot(2) security function:  Refuse if
 * any filedescriptors are open directories.
 */
static int
chroot_refuse_vdir_fds(thread_t td, struct filedesc *fdp)
{
        struct vnode *vp;
        struct file *fp;
        int error;
        int fd;

        for (fd = 0; fd < fdp->fd_nfiles ; fd++) {
                if ((error = holdvnode(td, fd, &fp)) != 0)
                        continue;
                vp = (struct vnode *)fp->f_data;
                if (vp->v_type != VDIR) {
                        fdrop(fp);
                        continue;
                }
                fdrop(fp);
                return(EPERM);
        }
        return (0);
}

/*
 * This sysctl determines if we will allow a process to chroot(2) if it
 * has a directory open:
 *      0: disallowed for all processes.
 *      1: allowed for processes that were not already chroot(2)'ed.
 *      2: allowed for all processes.
 */

static int chroot_allow_open_directories = 1;

SYSCTL_INT(_kern, OID_AUTO, chroot_allow_open_directories, CTLFLAG_RW,
     &chroot_allow_open_directories, 0, "");

/*
 * chroot to the specified namecache entry.  We obtain the vp from the
 * namecache data.  The passed ncp must be locked and referenced and will
 * remain locked and referenced on return.
 */
int
kern_chroot(struct nchandle *nch)
{
        struct thread *td = curthread;
        struct proc *p = td->td_proc;
        struct filedesc *fdp = p->p_fd;
        struct vnode *vp;
        int error;

        /*
         * Only privileged user can chroot
         */
        error = priv_check_cred(td->td_ucred, PRIV_VFS_CHROOT, 0);
        if (error)
                return (error);

        /*
         * Disallow open directory descriptors (fchdir() breakouts).
         */
        if (chroot_allow_open_directories == 0 ||
           (chroot_allow_open_directories == 1 && fdp->fd_rdir != rootvnode)) {
                if ((error = chroot_refuse_vdir_fds(td, fdp)) != 0)
                        return (error);
        }
        if ((vp = nch->ncp->nc_vp) == NULL)
                return (ENOENT);

        if ((error = vget(vp, LK_SHARED)) != 0)
                return (error);

        /*
         * Check the validity of vp as a directory to change to and 
         * associate it with rdir/jdir.
         */
        error = checkvp_chdir(vp, td);
        vn_unlock(vp);                  /* leave reference intact */
        if (error == 0) {
                lwkt_gettoken(&p->p_token);
                vrele(fdp->fd_rdir);
                fdp->fd_rdir = vp;      /* reference inherited by fd_rdir */
                cache_drop(&fdp->fd_nrdir);
                cache_copy(nch, &fdp->fd_nrdir);
                if (fdp->fd_jdir == NULL) {
                        fdp->fd_jdir = vp;
                        vref(fdp->fd_jdir);
                        cache_copy(nch, &fdp->fd_njdir);
                }
                if ((p->p_flags & P_DIDCHROOT) == 0) {
                        p->p_flags |= P_DIDCHROOT;
                        if (p->p_depth <= 65535 - 32)
                                p->p_depth += 32;
                }
                lwkt_reltoken(&p->p_token);
        } else {
                vrele(vp);
        }
        return (error);
}

/*
 * chroot_args(char *path)
 *
 * Change notion of root (``/'') directory.
 */
int
sys_chroot(struct sysmsg *sysmsg, const struct chroot_args *uap)
{
        struct thread *td __debugvar = curthread;
        struct nlookupdata nd;
        int error;

        KKASSERT(td->td_proc);
        error = nlookup_init(&nd, uap->path, UIO_USERSPACE, NLC_FOLLOW);
        if (error == 0) {
                nd.nl_flags |= NLC_EXEC;
                error = nlookup(&nd);
                if (error == 0)
                        error = kern_chroot(&nd.nl_nch);
        }
        nlookup_done(&nd);
        return(error);
}

int
sys_chroot_kernel(struct sysmsg *sysmsg, const struct chroot_kernel_args *uap)
{
        struct thread *td = curthread;
        struct nlookupdata nd;
        struct nchandle *nch;
        struct vnode *vp;
        int error;

        error = nlookup_init(&nd, uap->path, UIO_USERSPACE, NLC_FOLLOW);
        if (error)
                goto error_nond;

        error = nlookup(&nd);
        if (error)
                goto error_out;

        nch = &nd.nl_nch;

        error = priv_check_cred(td->td_ucred, PRIV_VFS_CHROOT, 0);
        if (error)
                goto error_out;

        if ((vp = nch->ncp->nc_vp) == NULL) {
                error = ENOENT;
                goto error_out;
        }

        if ((error = cache_vref(nch, nd.nl_cred, &vp)) != 0)
                goto error_out;

        vfs_cache_setroot(vp, cache_hold(nch));

error_out:
        nlookup_done(&nd);
error_nond:
        return(error);
}

/*
 * Common routine for chroot and chdir.  Given a locked, referenced vnode,
 * determine whether it is legal to chdir to the vnode.  The vnode's state
 * is not changed by this call.
 */
static int
checkvp_chdir(struct vnode *vp, struct thread *td)
{
        int error;

        if (vp->v_type != VDIR)
                error = ENOTDIR;
        else
                error = VOP_EACCESS(vp, VEXEC, td->td_ucred);
        return (error);
}

int
kern_open(struct nlookupdata *nd, int oflags, int mode, int *res)
{
        struct thread *td = curthread;
        struct proc *p = td->td_proc;
        struct lwp *lp = td->td_lwp;
        struct filedesc *fdp = p->p_fd;
        int cmode, flags;
        struct file *nfp;
        struct file *fp;
        struct vnode *vp;
        int type, indx, error = 0;
        struct flock lf;

        if ((oflags & O_ACCMODE) == O_ACCMODE)
                return (EINVAL);
        flags = FFLAGS(oflags);
        error = falloc(lp, &nfp, NULL);
        if (error)
                return (error);
        fp = nfp;
        cmode = ((mode &~ fdp->fd_cmask) & ALLPERMS) & ~S_ISTXT;

        /*
         * XXX p_dupfd is a real mess.  It allows a device to return a
         * file descriptor to be duplicated rather then doing the open
         * itself.
         */
        lp->lwp_dupfd = -1;

        /*
         * Call vn_open() to do the lookup and assign the vnode to the 
         * file pointer.  vn_open() does not change the ref count on fp
         * and the vnode, on success, will be inherited by the file pointer
         * and unlocked.
         *
         * Request a shared lock on the vnode if possible.
         *
         * When NLC_SHAREDLOCK is set we may still need an exclusive vnode
         * lock for O_RDWR opens on executables in order to avoid a VTEXT
         * detection race.  The NLC_EXCLLOCK_IFEXEC handles this case.
         *
         * NOTE: We need a flag to separate terminal vnode locking from
         *       parent locking.  O_CREAT needs parent locking, but O_TRUNC
         *       and O_RDWR only need to lock the terminal vnode exclusively.
         */
        nd->nl_flags |= NLC_LOCKVP;
        if ((flags & (O_CREAT|O_TRUNC)) == 0) {
                nd->nl_flags |= NLC_SHAREDLOCK;
                if (flags & O_RDWR)
                        nd->nl_flags |= NLC_EXCLLOCK_IFEXEC;
        }

        error = vn_open(nd, fp, flags, cmode);
        nlookup_done(nd);

        if (error) {
                /*
                 * handle special fdopen() case.  bleh.  dupfdopen() is
                 * responsible for dropping the old contents of ofiles[indx]
                 * if it succeeds.
                 *
                 * Note that fsetfd() will add a ref to fp which represents
                 * the fd_files[] assignment.  We must still drop our
                 * reference.
                 */
                if ((error == ENODEV || error == ENXIO) && lp->lwp_dupfd >= 0) {
                        if (fdalloc(p, 0, &indx) == 0) {
                                error = dupfdopen(td, indx, lp->lwp_dupfd, flags, error);
                                if (error == 0) {
                                        *res = indx;
                                        fdrop(fp);      /* our ref */
                                        return (0);
                                }
                                fsetfd(fdp, NULL, indx);
                        }
                }
                fdrop(fp);      /* our ref */
                if (error == ERESTART)
                        error = EINTR;
                return (error);
        }

        /*
         * ref the vnode for ourselves so it can't be ripped out from under
         * is.  XXX need an ND flag to request that the vnode be returned
         * anyway.
         *
         * Reserve a file descriptor but do not assign it until the open
         * succeeds.
         */
        vp = (struct vnode *)fp->f_data;
        vref(vp);
        if ((error = fdalloc(p, 0, &indx)) != 0) {
                fdrop(fp);
                vrele(vp);
                return (error);
        }

        /*
         * If no error occurs the vp will have been assigned to the file
         * pointer.
         */
        lp->lwp_dupfd = 0;

        if (flags & (O_EXLOCK | O_SHLOCK)) {
                lf.l_whence = SEEK_SET;
                lf.l_start = 0;
                lf.l_len = 0;
                if (flags & O_EXLOCK)
                        lf.l_type = F_WRLCK;
                else
                        lf.l_type = F_RDLCK;
                if (flags & FNONBLOCK)
                        type = 0;
                else
                        type = F_WAIT;

                if ((error = VOP_ADVLOCK(vp, (caddr_t)fp, F_SETLK, &lf, type)) != 0) {
                        /*
                         * lock request failed.  Clean up the reserved
                         * descriptor.
                         */
                        vrele(vp);
                        fsetfd(fdp, NULL, indx);
                        fdrop(fp);
                        return (error);
                }
                atomic_set_int(&fp->f_flag, FHASLOCK); /* race ok */
        }
#if 0
        /*
         * Assert that all regular file vnodes were created with a object.
         */
        KASSERT(vp->v_type != VREG || vp->v_object != NULL,
                ("open: regular file has no backing object after vn_open"));
#endif

        vrele(vp);

        /*
         * release our private reference, leaving the one associated with the
         * descriptor table intact.
         */
        if (oflags & O_CLOEXEC)
                fdp->fd_files[indx].fileflags |= UF_EXCLOSE;
        fsetfd(fdp, fp, indx);
        fdrop(fp);
        *res = indx;

        return (error);
}

/*
 * open_args(char *path, int flags, int mode)
 *
 * Check permissions, allocate an open file structure,
 * and call the device open routine if any.
 */
int
sys_open(struct sysmsg *sysmsg, const struct open_args *uap)
{
        struct nlookupdata nd;
        int error;

        error = nlookup_init(&nd, uap->path, UIO_USERSPACE, 0);
        if (error == 0) {
                error = kern_open(&nd, uap->flags,
                                    uap->mode, &sysmsg->sysmsg_result);
        }
        nlookup_done(&nd);
        return (error);
}

/*
 * openat_args(int fd, char *path, int flags, int mode)
 */
int
sys_openat(struct sysmsg *sysmsg, const struct openat_args *uap)
{
        struct nlookupdata nd;
        int error;
        struct file *fp;

        error = nlookup_init_at(&nd, &fp, uap->fd, uap->path, UIO_USERSPACE, 0);
        if (error == 0) {
                error = kern_open(&nd, uap->flags, uap->mode, 
                                        &sysmsg->sysmsg_result);
        }
        nlookup_done_at(&nd, fp);
        return (error);
}

int
kern_mknod(struct nlookupdata *nd, int mode, int rmajor, int rminor)
{
        struct thread *td = curthread;
        struct proc *p = td->td_proc;
        struct vnode *vp;
        struct vattr vattr;
        int error;
        int whiteout = 0;

        KKASSERT(p);

        VATTR_NULL(&vattr);
        vattr.va_mode = (mode & ALLPERMS) &~ p->p_fd->fd_cmask;
        vattr.va_rmajor = rmajor;
        vattr.va_rminor = rminor;

        switch (mode & S_IFMT) {
        case S_IFMT:    /* used by badsect to flag bad sectors */
                error = priv_check_cred(td->td_ucred, PRIV_VFS_MKNOD_BAD, 0);
                vattr.va_type = VBAD;
                break;
        case S_IFCHR:
                error = priv_check(td, PRIV_VFS_MKNOD_DEV);
                vattr.va_type = VCHR;
                break;
        case S_IFBLK:
                error = priv_check(td, PRIV_VFS_MKNOD_DEV);
                vattr.va_type = VBLK;
                break;
        case S_IFWHT:
                error = priv_check_cred(td->td_ucred, PRIV_VFS_MKNOD_WHT, 0);
                whiteout = 1;
                break;
        case S_IFDIR:   /* special directories support for HAMMER */
                error = priv_check_cred(td->td_ucred, PRIV_VFS_MKNOD_DIR, 0);
                vattr.va_type = VDIR;
                break;
        default:
                error = EINVAL;
                break;
        }

        if (error)
                return (error);

        bwillinode(1);
        nd->nl_flags |= NLC_CREATE | NLC_REFDVP;
        if ((error = nlookup(nd)) != 0)
                return (error);
        if (nd->nl_nch.ncp->nc_vp)
                return (EEXIST);
        if (nd->nl_dvp == NULL)
                return (EINVAL);
        if ((error = ncp_writechk(&nd->nl_nch)) != 0)
                return (error);

        if (whiteout) {
                error = VOP_NWHITEOUT(&nd->nl_nch, nd->nl_dvp,
                                      nd->nl_cred, NAMEI_CREATE);
        } else {
                vp = NULL;
                error = VOP_NMKNOD(&nd->nl_nch, nd->nl_dvp,
                                   &vp, nd->nl_cred, &vattr);
                if (error == 0)
                        vput(vp);
        }
        return (error);
}

/*
 * mknod_args(char *path, int mode, int dev)
 *
 * Create a special file.
 */
int
sys_mknod(struct sysmsg *sysmsg, const struct mknod_args *uap)
{
        struct nlookupdata nd;
        int error;

        error = nlookup_init(&nd, uap->path, UIO_USERSPACE, 0);
        if (error == 0) {
                error = kern_mknod(&nd, uap->mode,
                                   umajor(uap->dev), uminor(uap->dev));
        }
        nlookup_done(&nd);
        return (error);
}

/*
 * mknodat_args(int fd, char *path, mode_t mode, dev_t dev)
 *
 * Create a special file.  The path is relative to the directory associated
 * with fd.
 */
int
sys_mknodat(struct sysmsg *sysmsg, const struct mknodat_args *uap)
{
        struct nlookupdata nd;
        struct file *fp;
        int error;

        error = nlookup_init_at(&nd, &fp, uap->fd, uap->path, UIO_USERSPACE, 0);
        if (error == 0) {
                error = kern_mknod(&nd, uap->mode,
                                   umajor(uap->dev), uminor(uap->dev));
        }
        nlookup_done_at(&nd, fp);
        return (error);
}

int
kern_mkfifo(struct nlookupdata *nd, int mode)
{
        struct thread *td = curthread;
        struct proc *p = td->td_proc;
        struct vattr vattr;
        struct vnode *vp;
        int error;

        bwillinode(1);

        nd->nl_flags |= NLC_CREATE | NLC_REFDVP;
        if ((error = nlookup(nd)) != 0)
                return (error);
        if (nd->nl_nch.ncp->nc_vp)
                return (EEXIST);
        if (nd->nl_dvp == NULL)
                return (EINVAL);
        if ((error = ncp_writechk(&nd->nl_nch)) != 0)
                return (error);

        VATTR_NULL(&vattr);
        vattr.va_type = VFIFO;
        vattr.va_mode = (mode & ALLPERMS) &~ p->p_fd->fd_cmask;
        vp = NULL;
        error = VOP_NMKNOD(&nd->nl_nch, nd->nl_dvp, &vp, nd->nl_cred, &vattr);
        if (error == 0)
                vput(vp);
        return (error);
}

/*
 * mkfifo_args(char *path, int mode)
 *
 * Create a named pipe.
 */
int
sys_mkfifo(struct sysmsg *sysmsg, const struct mkfifo_args *uap)
{
        struct nlookupdata nd;
        int error;

        error = nlookup_init(&nd, uap->path, UIO_USERSPACE, 0);
        if (error == 0)
                error = kern_mkfifo(&nd, uap->mode);
        nlookup_done(&nd);
        return (error);
}

/*
 * mkfifoat_args(int fd, char *path, mode_t mode)
 *
 * Create a named pipe.  The path is relative to the directory associated
 * with fd.
 */
int
sys_mkfifoat(struct sysmsg *sysmsg, const struct mkfifoat_args *uap)
{
        struct nlookupdata nd;
        struct file *fp;
        int error;

        error = nlookup_init_at(&nd, &fp, uap->fd, uap->path, UIO_USERSPACE, 0);
        if (error == 0)
                error = kern_mkfifo(&nd, uap->mode);
        nlookup_done_at(&nd, fp);
        return (error);
}

static int hardlink_check_uid = 0;
SYSCTL_INT(_security, OID_AUTO, hardlink_check_uid, CTLFLAG_RW,
    &hardlink_check_uid, 0, 
    "Unprivileged processes cannot create hard links to files owned by other "
    "users");
static int hardlink_check_gid = 0;
SYSCTL_INT(_security, OID_AUTO, hardlink_check_gid, CTLFLAG_RW,
    &hardlink_check_gid, 0,
    "Unprivileged processes cannot create hard links to files owned by other "
    "groups");

static int
can_hardlink(struct vnode *vp, struct thread *td, struct ucred *cred)
{
        struct vattr va;
        int error;

        /*
         * Shortcut if disabled
         */
        if (hardlink_check_uid == 0 && hardlink_check_gid == 0)
                return (0);

        /*
         * Privileged user can always hardlink
         */
        if (priv_check_cred(cred, PRIV_VFS_LINK, 0) == 0)
                return (0);

        /*
         * Otherwise only if the originating file is owned by the
         * same user or group.  Note that any group is allowed if
         * the file is owned by the caller.
         */
        error = VOP_GETATTR(vp, &va);
        if (error != 0)
                return (error);
        
        if (hardlink_check_uid) {
                if (cred->cr_uid != va.va_uid)
                        return (EPERM);
        }
        
        if (hardlink_check_gid) {
                if (cred->cr_uid != va.va_uid && !groupmember(va.va_gid, cred))
                        return (EPERM);
        }

        return (0);
}

int
kern_link(struct nlookupdata *nd, struct nlookupdata *linknd)
{
        struct thread *td = curthread;
        struct vnode *vp;
        int error;

        /*
         * Lookup the source and obtained a locked vnode.
         *
         * You may only hardlink a file which you have write permission
         * on or which you own.
         *
         * XXX relookup on vget failure / race ?
         */
        bwillinode(1);
        nd->nl_flags |= NLC_WRITE | NLC_OWN | NLC_HLINK;
        if ((error = nlookup(nd)) != 0)
                return (error);
        vp = nd->nl_nch.ncp->nc_vp;
        KKASSERT(vp != NULL);
        if (vp->v_type == VDIR)
                return (EPERM);         /* POSIX */
        if ((error = ncp_writechk(&nd->nl_nch)) != 0)
                return (error);
        if ((error = vget(vp, LK_EXCLUSIVE)) != 0)
                return (error);

        /*
         * Unlock the source so we can lookup the target without deadlocking
         * (XXX vp is locked already, possible other deadlock?).  The target
         * must not exist.
         */
        KKASSERT(nd->nl_flags & NLC_NCPISLOCKED);
        nd->nl_flags &= ~NLC_NCPISLOCKED;
        cache_unlock(&nd->nl_nch);
        vn_unlock(vp);

        linknd->nl_flags |= NLC_CREATE | NLC_REFDVP;
        if ((error = nlookup(linknd)) != 0) {
                vrele(vp);
                return (error);
        }
        if (linknd->nl_nch.ncp->nc_vp) {
                vrele(vp);
                return (EEXIST);
        }
        if (linknd->nl_dvp == NULL) {
                vrele(vp);
                return (EINVAL);
        }
        VFS_MODIFYING(vp->v_mount);
        error = vn_lock(vp, LK_EXCLUSIVE | LK_RETRY | LK_FAILRECLAIM);
        if (error) {
                vrele(vp);
                return (error);
        }

        /*
         * Finally run the new API VOP.
         */
        error = can_hardlink(vp, td, td->td_ucred);
        if (error == 0) {
                error = VOP_NLINK(&linknd->nl_nch, linknd->nl_dvp,
                                  vp, linknd->nl_cred);
        }
        vput(vp);
        return (error);
}

/*
 * link_args(char *path, char *link)
 *
 * Make a hard file link.
 */
int
sys_link(struct sysmsg *sysmsg, const struct link_args *uap)
{
        struct nlookupdata nd, linknd;
        int error;

        error = nlookup_init(&nd, uap->path, UIO_USERSPACE, NLC_FOLLOW);
        if (error == 0) {
                error = nlookup_init(&linknd, uap->link, UIO_USERSPACE, 0);
                if (error == 0)
                        error = kern_link(&nd, &linknd);
                nlookup_done(&linknd);
        }
        nlookup_done(&nd);
        return (error);
}

/*
 * linkat_args(int fd1, char *path1, int fd2, char *path2, int flags)
 *
 * Make a hard file link. The path1 argument is relative to the directory
 * associated with fd1, and similarly the path2 argument is relative to
 * the directory associated with fd2.
 */
int
sys_linkat(struct sysmsg *sysmsg, const struct linkat_args *uap)
{
        struct nlookupdata nd, linknd;
        struct file *fp1, *fp2;
        int error;

        error = nlookup_init_at(&nd, &fp1, uap->fd1, uap->path1, UIO_USERSPACE,
            (uap->flags & AT_SYMLINK_FOLLOW) ? NLC_FOLLOW : 0);
        if (error == 0) {
                error = nlookup_init_at(&linknd, &fp2, uap->fd2,
                    uap->path2, UIO_USERSPACE, 0);
                if (error == 0)
                        error = kern_link(&nd, &linknd);
                nlookup_done_at(&linknd, fp2);
        }
        nlookup_done_at(&nd, fp1);
        return (error);
}

int
kern_symlink(struct nlookupdata *nd, char *path, int mode)
{
        struct vattr vattr;
        struct vnode *vp;
        struct vnode *dvp;
        int error;

        bwillinode(1);
        nd->nl_flags |= NLC_CREATE | NLC_REFDVP;
        if ((error = nlookup(nd)) != 0)
                return (error);
        if (nd->nl_nch.ncp->nc_vp)
                return (EEXIST);
        if (nd->nl_dvp == NULL)
                return (EINVAL);
        if ((error = ncp_writechk(&nd->nl_nch)) != 0)
                return (error);
        dvp = nd->nl_dvp;
        VATTR_NULL(&vattr);
        vattr.va_mode = mode;
        error = VOP_NSYMLINK(&nd->nl_nch, dvp, &vp, nd->nl_cred, &vattr, path);
        if (error == 0)
                vput(vp);
        return (error);
}

/*
 * symlink(char *path, char *link)
 *
 * Make a symbolic link.
 */
int
sys_symlink(struct sysmsg *sysmsg, const struct symlink_args *uap)
{
        struct thread *td = curthread;
        struct nlookupdata nd;
        char *path;
        int error;
        int mode;

        path = objcache_get(namei_oc, M_WAITOK);
        error = copyinstr(uap->path, path, MAXPATHLEN, NULL);
        if (error == 0) {
                error = nlookup_init(&nd, uap->link, UIO_USERSPACE, 0);
                if (error == 0) {
                        mode = ACCESSPERMS & ~td->td_proc->p_fd->fd_cmask;
                        error = kern_symlink(&nd, path, mode);
                }
                nlookup_done(&nd);
        }
        objcache_put(namei_oc, path);
        return (error);
}

/*
 * symlinkat_args(char *path1, int fd, char *path2)
 *
 * Make a symbolic link.  The path2 argument is relative to the directory
 * associated with fd.
 */
int
sys_symlinkat(struct sysmsg *sysmsg, const struct symlinkat_args *uap)
{
        struct thread *td = curthread;
        struct nlookupdata nd;
        struct file *fp;
        char *path1;
        int error;
        int mode;

        path1 = objcache_get(namei_oc, M_WAITOK);
        error = copyinstr(uap->path1, path1, MAXPATHLEN, NULL);
        if (error == 0) {
                error = nlookup_init_at(&nd, &fp, uap->fd, uap->path2,
                    UIO_USERSPACE, 0);
                if (error == 0) {
                        mode = ACCESSPERMS & ~td->td_proc->p_fd->fd_cmask;
                        error = kern_symlink(&nd, path1, mode);
                }
                nlookup_done_at(&nd, fp);
        }
        objcache_put(namei_oc, path1);
        return (error);
}

/*
 * undelete_args(char *path)
 *
 * Delete a whiteout from the filesystem.
 */
int
sys_undelete(struct sysmsg *sysmsg, const struct undelete_args *uap)
{
        struct nlookupdata nd;
        int error;

        error = nlookup_init(&nd, uap->path, UIO_USERSPACE, 0);
        bwillinode(1);
        nd.nl_flags |= NLC_DELETE | NLC_REFDVP;
        if (error == 0)
                error = nlookup(&nd);
        if (error == 0 && nd.nl_dvp == NULL)
                error = EINVAL;
        if (error == 0)
                error = ncp_writechk(&nd.nl_nch);
        if (error == 0) {
                error = VOP_NWHITEOUT(&nd.nl_nch, nd.nl_dvp, nd.nl_cred,
                                      NAMEI_DELETE);
        }
        nlookup_done(&nd);
        return (error);
}

int
kern_unlink(struct nlookupdata *nd)
{
        int error;

        bwillinode(1);
        nd->nl_flags |= NLC_DELETE | NLC_REFDVP;
        if ((error = nlookup(nd)) != 0)
                return (error);
        if (nd->nl_dvp == NULL)
                return EINVAL;
        if ((error = ncp_writechk(&nd->nl_nch)) != 0)
                return (error);
        error = VOP_NREMOVE(&nd->nl_nch, nd->nl_dvp, nd->nl_cred);
        return (error);
}

/*
 * unlink_args(char *path)
 *
 * Delete a name from the filesystem.
 */
int
sys_unlink(struct sysmsg *sysmsg, const struct unlink_args *uap)
{
        struct nlookupdata nd;
        int error;

        error = nlookup_init(&nd, uap->path, UIO_USERSPACE, 0);
        if (error == 0)
                error = kern_unlink(&nd);
        nlookup_done(&nd);
        return (error);
}


/*
 * unlinkat_args(int fd, char *path, int flags)
 *
 * Delete the file or directory entry pointed to by fd/path.
 */
int
sys_unlinkat(struct sysmsg *sysmsg, const struct unlinkat_args *uap)
{
        struct nlookupdata nd;
        struct file *fp;
        int error;

        if (uap->flags & ~AT_REMOVEDIR)
                return (EINVAL);

        error = nlookup_init_at(&nd, &fp, uap->fd, uap->path, UIO_USERSPACE, 0);
        if (error == 0) {
                if (uap->flags & AT_REMOVEDIR)
                        error = kern_rmdir(&nd);
                else
                        error = kern_unlink(&nd);
        }
        nlookup_done_at(&nd, fp);
        return (error);
}

int
kern_lseek(int fd, off_t offset, int whence, off_t *res)
{
        struct thread *td = curthread;
        struct file *fp;
        struct vnode *vp;
        struct vattr_lite lva;
        off_t new_offset;
        int error;

        fp = holdfp(td, fd, -1);
        if (fp == NULL)
                return (EBADF);
        if (fp->f_type != DTYPE_VNODE) {
                error = ESPIPE;
                goto done;
        }
        vp = (struct vnode *)fp->f_data;

        switch (whence) {
        case L_INCR:
                spin_lock(&fp->f_spin);
                new_offset = fp->f_offset + offset;
                error = 0;
                break;
        case L_XTND:
                error = VOP_GETATTR_LITE(vp, &lva);
                spin_lock(&fp->f_spin);
                new_offset = offset + lva.va_size;
                break;
        case L_SET:
                new_offset = offset;
                error = 0;
                spin_lock(&fp->f_spin);
                break;
        default:
                new_offset = 0;
                error = EINVAL;
                spin_lock(&fp->f_spin);
                break;
        }

        /*
         * Validate the seek position.  Negative offsets are not allowed
         * for regular files or directories.
         *
         * Normally we would also not want to allow negative offsets for
         * character and block-special devices.  However kvm addresses
         * on 64 bit architectures might appear to be negative and must
         * be allowed.
         */
        if (error == 0) {
                if (new_offset < 0 &&
                    (vp->v_type == VREG || vp->v_type == VDIR)) {
                        error = EINVAL;
                } else {
                        fp->f_offset = new_offset;
                }
        }
        *res = fp->f_offset;
        spin_unlock(&fp->f_spin);
done:
        dropfp(td, fd, fp);

        return (error);
}

/*
 * lseek_args(int fd, int pad, off_t offset, int whence)
 *
 * Reposition read/write file offset.
 */
int
sys_lseek(struct sysmsg *sysmsg, const struct lseek_args *uap)
{
        int error;

        error = kern_lseek(uap->fd, uap->offset, uap->whence,
                           &sysmsg->sysmsg_offset);

        return (error);
}

/*
 * Check if current process can access given file.  amode is a bitmask of *_OK
 * access bits.  flags is a bitmask of AT_* flags.
 */
int
kern_access(struct nlookupdata *nd, int amode, int flags)
{
        struct vnode *vp;
        int error, mode;

        if (flags & ~AT_EACCESS)
                return (EINVAL);
        nd->nl_flags |= NLC_SHAREDLOCK;
        if ((error = nlookup(nd)) != 0)
                return (error);
        if ((amode & W_OK) && (error = ncp_writechk(&nd->nl_nch)) != 0)
                return (error);
retry:
        error = cache_vget(&nd->nl_nch, nd->nl_cred, LK_SHARED, &vp);
        if (error)
                return (error);

        /* Flags == 0 means only check for existence. */
        if (amode) {
                mode = 0;
                if (amode & R_OK)
                        mode |= VREAD;
                if (amode & W_OK)
                        mode |= VWRITE;
                if (amode & X_OK)
                        mode |= VEXEC;
                if ((mode & VWRITE) == 0 || 
                    (error = vn_writechk(vp)) == 0) {
                        error = VOP_ACCESS_FLAGS(vp, mode, flags, nd->nl_cred);
                }

                /*
                 * If the file handle is stale we have to re-resolve the
                 * entry with the ncp held exclusively.  This is a hack
                 * at the moment.
                 */
                if (error == ESTALE) {
                        vput(vp);
                        cache_unlock(&nd->nl_nch);
                        cache_lock(&nd->nl_nch);
                        cache_setunresolved(&nd->nl_nch);
                        error = cache_resolve(&nd->nl_nch, nd->nl_cred);
                        if (error == 0) {
                                vp = NULL;
                                goto retry;
                        }
                        return(error);
                }
        }
        vput(vp);
        return (error);
}

/*
 * access_args(char *path, int flags)
 *
 * Check access permissions.
 */
int
sys_access(struct sysmsg *sysmsg, const struct access_args *uap)
{
        struct nlookupdata nd;
        int error;

        error = nlookup_init(&nd, uap->path, UIO_USERSPACE, NLC_FOLLOW);
        if (error == 0)
                error = kern_access(&nd, uap->flags, 0);
        nlookup_done(&nd);
        return (error);
}


/*
 * eaccess_args(char *path, int flags)
 *
 * Check access permissions.
 */
int
sys_eaccess(struct sysmsg *sysmsg, const struct eaccess_args *uap)
{
        struct nlookupdata nd;
        int error;

        error = nlookup_init(&nd, uap->path, UIO_USERSPACE, NLC_FOLLOW);
        if (error == 0)
                error = kern_access(&nd, uap->flags, AT_EACCESS);
        nlookup_done(&nd);
        return (error);
}


/*
 * faccessat_args(int fd, char *path, int amode, int flags)
 *
 * Check access permissions.
 */
int
sys_faccessat(struct sysmsg *sysmsg, const struct faccessat_args *uap)
{
        struct nlookupdata nd;
        struct file *fp;
        int error;

        error = nlookup_init_at(&nd, &fp, uap->fd, uap->path, UIO_USERSPACE, 
                                NLC_FOLLOW);
        if (error == 0)
                error = kern_access(&nd, uap->amode, uap->flags);
        nlookup_done_at(&nd, fp);
        return (error);
}

int
kern_stat(struct nlookupdata *nd, struct stat *st)
{
        int error;
        struct vnode *vp;

        nd->nl_flags |= NLC_SHAREDLOCK;
        if ((error = nlookup(nd)) != 0)
                return (error);
again:
        if ((vp = nd->nl_nch.ncp->nc_vp) == NULL)
                return (ENOENT);

#if 1
        error = cache_vref(&nd->nl_nch, NULL, &vp);
#else
        error = vget(vp, LK_SHARED);
#endif
        if (error)
                return (error);
        error = vn_stat(vp, st, nd->nl_cred);

        /*
         * If the file handle is stale we have to re-resolve the
         * entry with the ncp held exclusively.  This is a hack
         * at the moment.
         */
        if (error == ESTALE) {
#if 1
                vrele(vp);
#else
                vput(vp);
#endif
                cache_unlock(&nd->nl_nch);
                cache_lock(&nd->nl_nch);
                cache_setunresolved(&nd->nl_nch);
                error = cache_resolve(&nd->nl_nch, nd->nl_cred);
                if (error == 0)
                        goto again;
        } else {
#if 1
                vrele(vp);
#else
                vput(vp);
#endif
        }
        return (error);
}

/*
 * stat_args(char *path, struct stat *ub)
 *
 * Get file status; this version follows links.
 */
int
sys_stat(struct sysmsg *sysmsg, const struct stat_args *uap)
{
        struct nlookupdata nd;
        struct stat st;
        int error;

        error = nlookup_init(&nd, uap->path, UIO_USERSPACE, NLC_FOLLOW);
        if (error == 0) {
                error = kern_stat(&nd, &st);
                if (error == 0)
                        error = copyout(&st, uap->ub, sizeof(*uap->ub));
        }
        nlookup_done(&nd);
        return (error);
}

/*
 * lstat_args(char *path, struct stat *ub)
 *
 * Get file status; this version does not follow links.
 */
int
sys_lstat(struct sysmsg *sysmsg, const struct lstat_args *uap)
{
        struct nlookupdata nd;
        struct stat st;
        int error;

        error = nlookup_init(&nd, uap->path, UIO_USERSPACE, 0);
        if (error == 0) {
                error = kern_stat(&nd, &st);
                if (error == 0)
                        error = copyout(&st, uap->ub, sizeof(*uap->ub));
        }
        nlookup_done(&nd);
        return (error);
}

/*
 * fstatat_args(int fd, char *path, struct stat *sb, int flags)
 *
 * Get status of file pointed to by fd/path.
 */
int
sys_fstatat(struct sysmsg *sysmsg, const struct fstatat_args *uap)
{
        struct nlookupdata nd;
        struct stat st;
        int error;
        int flags;
        struct file *fp;

        if (uap->flags & ~AT_SYMLINK_NOFOLLOW)
                return (EINVAL);

        flags = (uap->flags & AT_SYMLINK_NOFOLLOW) ? 0 : NLC_FOLLOW;

        error = nlookup_init_at(&nd, &fp, uap->fd, uap->path, 
                                UIO_USERSPACE, flags);
        if (error == 0) {
                error = kern_stat(&nd, &st);
                if (error == 0)
                        error = copyout(&st, uap->sb, sizeof(*uap->sb));
        }
        nlookup_done_at(&nd, fp);
        return (error);
}

static int
kern_pathconf(char *path, int name, int flags, register_t *sysmsg_regp)
{
        struct nlookupdata nd;
        struct vnode *vp;
        int error;

        vp = NULL;
        error = nlookup_init(&nd, path, UIO_USERSPACE, flags);
        if (error == 0)
                error = nlookup(&nd);
        if (error == 0)
                error = cache_vget(&nd.nl_nch, nd.nl_cred, LK_EXCLUSIVE, &vp);
        nlookup_done(&nd);
        if (error == 0) {
                error = VOP_PATHCONF(vp, name, sysmsg_regp);
                vput(vp);
        }
        return (error);
}

/*
 * pathconf_Args(char *path, int name)
 *
 * Get configurable pathname variables.
 */
int
sys_pathconf(struct sysmsg *sysmsg, const struct pathconf_args *uap)
{
        return (kern_pathconf(uap->path, uap->name, NLC_FOLLOW,
                &sysmsg->sysmsg_reg));
}

/*
 * lpathconf_Args(char *path, int name)
 *
 * Get configurable pathname variables, but don't follow symlinks.
 */
int
sys_lpathconf(struct sysmsg *sysmsg, const struct lpathconf_args *uap)
{
        return (kern_pathconf(uap->path, uap->name, 0, &sysmsg->sysmsg_reg));
}

/*
 * XXX: daver
 * kern_readlink isn't properly split yet.  There is a copyin burried
 * in VOP_READLINK().
 */
int
kern_readlink(struct nlookupdata *nd, char *buf, int count, int *res)
{
        struct thread *td = curthread;
        struct vnode *vp;
        struct iovec aiov;
        struct uio auio;
        int error;

        nd->nl_flags |= NLC_SHAREDLOCK;
        if ((error = nlookup(nd)) != 0)
                return (error);
        error = cache_vget(&nd->nl_nch, nd->nl_cred, LK_SHARED, &vp);
        if (error)
                return (error);
        if (vp->v_type != VLNK) {
                error = EINVAL;
        } else {
                aiov.iov_base = buf;
                aiov.iov_len = count;
                auio.uio_iov = &aiov;
                auio.uio_iovcnt = 1;
                auio.uio_offset = 0;
                auio.uio_rw = UIO_READ;
                auio.uio_segflg = UIO_USERSPACE;
                auio.uio_td = td;
                auio.uio_resid = count;
                error = VOP_READLINK(vp, &auio, td->td_ucred);
        }
        vput(vp);
        *res = count - auio.uio_resid;
        return (error);
}

/*
 * readlink_args(char *path, char *buf, int count)
 *
 * Return target name of a symbolic link.
 */
int
sys_readlink(struct sysmsg *sysmsg, const struct readlink_args *uap)
{
        struct nlookupdata nd;
        int error;

        error = nlookup_init(&nd, uap->path, UIO_USERSPACE, 0);
        if (error == 0) {
                error = kern_readlink(&nd, uap->buf, uap->count,
                                        &sysmsg->sysmsg_result);
        }
        nlookup_done(&nd);
        return (error);
}

/*
 * readlinkat_args(int fd, char *path, char *buf, size_t bufsize)
 *
 * Return target name of a symbolic link.  The path is relative to the
 * directory associated with fd.
 */
int
sys_readlinkat(struct sysmsg *sysmsg, const struct readlinkat_args *uap)
{
        struct nlookupdata nd;
        struct file *fp;
        int error;

        error = nlookup_init_at(&nd, &fp, uap->fd, uap->path, UIO_USERSPACE, 0);
        if (error == 0) {
                error = kern_readlink(&nd, uap->buf, uap->bufsize,
                                        &sysmsg->sysmsg_result);
        }
        nlookup_done_at(&nd, fp);
        return (error);
}

static int
setfflags(struct vnode *vp, u_long flags)
{
        struct thread *td = curthread;
        int error;
        struct vattr vattr;

        /*
         * Prevent non-root users from setting flags on devices.  When
         * a device is reused, users can retain ownership of the device
         * if they are allowed to set flags and programs assume that
         * chown can't fail when done as root.
         */
        if ((vp->v_type == VCHR || vp->v_type == VBLK) && 
            ((error = priv_check_cred(td->td_ucred, PRIV_VFS_CHFLAGS_DEV, 0)) != 0))
                return (error);

        /*
         * note: vget is required for any operation that might mod the vnode
         * so VINACTIVE is properly cleared.
         */
        if ((error = vget(vp, LK_EXCLUSIVE)) == 0) {
                VATTR_NULL(&vattr);
                vattr.va_flags = flags;
                error = VOP_SETATTR(vp, &vattr, td->td_ucred);
                vput(vp);
        }
        return (error);
}

/*
 * chflags(const char *path, u_long flags)
 *
 * Change flags of a file given a path name.
 */
int
sys_chflags(struct sysmsg *sysmsg, const struct chflags_args *uap)
{
        struct nlookupdata nd;
        struct vnode *vp;
        int error;

        vp = NULL;
        error = nlookup_init(&nd, uap->path, UIO_USERSPACE, NLC_FOLLOW);
        if (error == 0)
                error = nlookup(&nd);
        if (error == 0)
                error = ncp_writechk(&nd.nl_nch);
        if (error == 0)
                error = cache_vref(&nd.nl_nch, nd.nl_cred, &vp);
        nlookup_done(&nd);
        if (error == 0) {
                error = setfflags(vp, uap->flags);
                vrele(vp);
        }
        return (error);
}

/*
 * lchflags(const char *path, u_long flags)
 *
 * Change flags of a file given a path name, but don't follow symlinks.
 */
int
sys_lchflags(struct sysmsg *sysmsg, const struct lchflags_args *uap)
{
        struct nlookupdata nd;
        struct vnode *vp;
        int error;

        vp = NULL;
        error = nlookup_init(&nd, uap->path, UIO_USERSPACE, 0);
        if (error == 0)
                error = nlookup(&nd);
        if (error == 0)
                error = ncp_writechk(&nd.nl_nch);
        if (error == 0)
                error = cache_vref(&nd.nl_nch, nd.nl_cred, &vp);
        nlookup_done(&nd);
        if (error == 0) {
                error = setfflags(vp, uap->flags);
                vrele(vp);
        }
        return (error);
}

/*
 * fchflags_args(int fd, u_flags flags)
 *
 * Change flags of a file given a file descriptor.
 */
int
sys_fchflags(struct sysmsg *sysmsg, const struct fchflags_args *uap)
{
        struct thread *td = curthread;
        struct file *fp;
        int error;

        if ((error = holdvnode(td, uap->fd, &fp)) != 0)
                return (error);
        if (fp->f_nchandle.ncp)
                error = ncp_writechk(&fp->f_nchandle);
        if (error == 0)
                error = setfflags((struct vnode *) fp->f_data, uap->flags);
        fdrop(fp);
        return (error);
}

/*
 * chflagsat_args(int fd, const char *path, u_long flags, int atflags)
 * change flags given a pathname relative to a filedescriptor
 */
int
sys_chflagsat(struct sysmsg *sysmsg, const struct chflagsat_args *uap)
{
        struct nlookupdata nd;
        struct vnode *vp;
        struct file *fp;
        int error;
        int lookupflags;

        if (uap->atflags & ~AT_SYMLINK_NOFOLLOW)
                return (EINVAL);

        lookupflags = (uap->atflags & AT_SYMLINK_NOFOLLOW) ? 0 : NLC_FOLLOW;

        vp = NULL;
        error = nlookup_init_at(&nd, &fp, uap->fd,  uap->path, UIO_USERSPACE, lookupflags);
        if (error == 0)
                error = nlookup(&nd);
        if (error == 0)
                error = ncp_writechk(&nd.nl_nch);
        if (error == 0)
                error = cache_vref(&nd.nl_nch, nd.nl_cred, &vp);
        nlookup_done_at(&nd, fp);
        if (error == 0) {
                error = setfflags(vp, uap->flags);
                vrele(vp);
        }
        return (error);
}


static int
setfmode(struct vnode *vp, int mode)
{
        struct thread *td = curthread;
        int error;
        struct vattr vattr;

        /*
         * note: vget is required for any operation that might mod the vnode
         * so VINACTIVE is properly cleared.
         */
        if ((error = vget(vp, LK_EXCLUSIVE)) == 0) {
                VATTR_NULL(&vattr);
                vattr.va_mode = mode & ALLPERMS;
                error = VOP_SETATTR(vp, &vattr, td->td_ucred);
                cache_inval_wxok(vp);
                vput(vp);
        }
        return error;
}

int
kern_chmod(struct nlookupdata *nd, int mode)
{
        struct vnode *vp;
        int error;

        if ((error = nlookup(nd)) != 0)
                return (error);
        if ((error = cache_vref(&nd->nl_nch, nd->nl_cred, &vp)) != 0)
                return (error);
        if ((error = ncp_writechk(&nd->nl_nch)) == 0)
                error = setfmode(vp, mode);
        vrele(vp);
        return (error);
}

/*
 * chmod_args(char *path, int mode)
 *
 * Change mode of a file given path name.
 */
int
sys_chmod(struct sysmsg *sysmsg, const struct chmod_args *uap)
{
        struct nlookupdata nd;
        int error;

        error = nlookup_init(&nd, uap->path, UIO_USERSPACE, NLC_FOLLOW);
        if (error == 0)
                error = kern_chmod(&nd, uap->mode);
        nlookup_done(&nd);
        return (error);
}

/*
 * lchmod_args(char *path, int mode)
 *
 * Change mode of a file given path name (don't follow links.)
 */
int
sys_lchmod(struct sysmsg *sysmsg, const struct lchmod_args *uap)
{
        struct nlookupdata nd;
        int error;

        error = nlookup_init(&nd, uap->path, UIO_USERSPACE, 0);
        if (error == 0)
                error = kern_chmod(&nd, uap->mode);
        nlookup_done(&nd);
        return (error);
}

/*
 * fchmod_args(int fd, int mode)
 *
 * Change mode of a file given a file descriptor.
 */
int
sys_fchmod(struct sysmsg *sysmsg, const struct fchmod_args *uap)
{
        struct thread *td = curthread;
        struct file *fp;
        int error;

        if ((error = holdvnode(td, uap->fd, &fp)) != 0)
                return (error);
        if (fp->f_nchandle.ncp)
                error = ncp_writechk(&fp->f_nchandle);
        if (error == 0)
                error = setfmode((struct vnode *)fp->f_data, uap->mode);
        fdrop(fp);
        return (error);
}

/*
 * fchmodat_args(char *path, int mode)
 *
 * Change mode of a file pointed to by fd/path.
 */
int
sys_fchmodat(struct sysmsg *sysmsg, const struct fchmodat_args *uap)
{
        struct nlookupdata nd;
        struct file *fp;
        int error;
        int flags;

        if (uap->flags & ~AT_SYMLINK_NOFOLLOW)
                return (EINVAL);
        flags = (uap->flags & AT_SYMLINK_NOFOLLOW) ? 0 : NLC_FOLLOW;

        error = nlookup_init_at(&nd, &fp, uap->fd, uap->path, 
                                UIO_USERSPACE, flags);
        if (error == 0)
                error = kern_chmod(&nd, uap->mode);
        nlookup_done_at(&nd, fp);
        return (error);
}

static int
setfown(struct mount *mp, struct vnode *vp, uid_t uid, gid_t gid)
{
        struct thread *td = curthread;
        int error;
        struct vattr vattr;
        uid_t o_uid;
        gid_t o_gid;
        uint64_t size;

        /*
         * note: vget is required for any operation that might mod the vnode
         * so VINACTIVE is properly cleared.
         */
        if ((error = vget(vp, LK_EXCLUSIVE)) == 0) {
                if ((error = VOP_GETATTR(vp, &vattr)) != 0)
                        return error;
                o_uid = vattr.va_uid;
                o_gid = vattr.va_gid;
                size = vattr.va_size;

                VATTR_NULL(&vattr);
                vattr.va_uid = uid;
                vattr.va_gid = gid;
                error = VOP_SETATTR(vp, &vattr, td->td_ucred);
                vput(vp);
        }

        if (error == 0) {
                if (uid == -1)
                        uid = o_uid;
                if (gid == -1)
                        gid = o_gid;
                VFS_ACCOUNT(mp, o_uid, o_gid, -size);
                VFS_ACCOUNT(mp,   uid,   gid,  size);
        }

        return error;
}

int
kern_chown(struct nlookupdata *nd, int uid, int gid)
{
        struct vnode *vp;
        int error;

        if ((error = nlookup(nd)) != 0)
                return (error);
        if ((error = cache_vref(&nd->nl_nch, nd->nl_cred, &vp)) != 0)
                return (error);
        if ((error = ncp_writechk(&nd->nl_nch)) == 0)
                error = setfown(nd->nl_nch.mount, vp, uid, gid);
        vrele(vp);
        return (error);
}

/*
 * chown(char *path, int uid, int gid)
 *
 * Set ownership given a path name.
 */
int
sys_chown(struct sysmsg *sysmsg, const struct chown_args *uap)
{
        struct nlookupdata nd;
        int error;

        error = nlookup_init(&nd, uap->path, UIO_USERSPACE, NLC_FOLLOW);
        if (error == 0)
                error = kern_chown(&nd, uap->uid, uap->gid);
        nlookup_done(&nd);
        return (error);
}

/*
 * lchown_args(char *path, int uid, int gid)
 *
 * Set ownership given a path name, do not cross symlinks.
 */
int
sys_lchown(struct sysmsg *sysmsg, const struct lchown_args *uap)
{
        struct nlookupdata nd;
        int error;

        error = nlookup_init(&nd, uap->path, UIO_USERSPACE, 0);
        if (error == 0)
                error = kern_chown(&nd, uap->uid, uap->gid);
        nlookup_done(&nd);
        return (error);
}

/*
 * fchown_args(int fd, int uid, int gid)
 *
 * Set ownership given a file descriptor.
 */
int
sys_fchown(struct sysmsg *sysmsg, const struct fchown_args *uap)
{
        struct thread *td = curthread;
        struct proc *p = td->td_proc;
        struct file *fp;
        int error;

        if ((error = holdvnode(td, uap->fd, &fp)) != 0)
                return (error);
        if (fp->f_nchandle.ncp)
                error = ncp_writechk(&fp->f_nchandle);
        if (error == 0)
                error = setfown(p->p_fd->fd_ncdir.mount,
                        (struct vnode *)fp->f_data, uap->uid, uap->gid);
        fdrop(fp);
        return (error);
}

/*
 * fchownat(int fd, char *path, int uid, int gid, int flags)
 *
 * Set ownership of file pointed to by fd/path.
 */
int
sys_fchownat(struct sysmsg *sysmsg, const struct fchownat_args *uap)
{
        struct nlookupdata nd;
        struct file *fp;
        int error;
        int flags;

        if (uap->flags & ~AT_SYMLINK_NOFOLLOW)
                return (EINVAL);
        flags = (uap->flags & AT_SYMLINK_NOFOLLOW) ? 0 : NLC_FOLLOW;

        error = nlookup_init_at(&nd, &fp, uap->fd, uap->path, 
                                UIO_USERSPACE, flags);
        if (error == 0)
                error = kern_chown(&nd, uap->uid, uap->gid);
        nlookup_done_at(&nd, fp);
        return (error);
}


static int
getutimes(struct timeval *tvp, struct timespec *tsp)
{
        struct timeval tv[2];
        int error;

        if (tvp == NULL) {
                microtime(&tv[0]);
                TIMEVAL_TO_TIMESPEC(&tv[0], &tsp[0]);
                tsp[1] = tsp[0];
        } else {
                if ((error = itimerfix(tvp)) != 0)
                        return (error);
                TIMEVAL_TO_TIMESPEC(&tvp[0], &tsp[0]);
                TIMEVAL_TO_TIMESPEC(&tvp[1], &tsp[1]);
        }
        return 0;
}

static int
getutimens(const struct timespec *ts, struct timespec *newts, int *nullflag)
{
        struct timespec tsnow;
        int error;

        *nullflag = 0;
        nanotime(&tsnow);
        if (ts == NULL) {
                newts[0] = tsnow;
                newts[1] = tsnow;
                *nullflag = 1;
                return (0);
        }

        newts[0] = ts[0];
        newts[1] = ts[1];
        if (newts[0].tv_nsec == UTIME_OMIT && newts[1].tv_nsec == UTIME_OMIT)
                return (0);
        if (newts[0].tv_nsec == UTIME_NOW && newts[1].tv_nsec == UTIME_NOW)
                *nullflag = 1;

        if (newts[0].tv_nsec == UTIME_OMIT)
                newts[0].tv_sec = VNOVAL;
        else if (newts[0].tv_nsec == UTIME_NOW)
                newts[0] = tsnow;
        else if ((error = itimespecfix(&newts[0])) != 0)
                return (error);

        if (newts[1].tv_nsec == UTIME_OMIT)
                newts[1].tv_sec = VNOVAL;
        else if (newts[1].tv_nsec == UTIME_NOW)
                newts[1] = tsnow;
        else if ((error = itimespecfix(&newts[1])) != 0)
                return (error);

        return (0);
}

static int
setutimes(struct vnode *vp, struct vattr *vattr,
          const struct timespec *ts, int nullflag)
{
        struct thread *td = curthread;
        int error;

        VATTR_NULL(vattr);
        vattr->va_atime = ts[0];
        vattr->va_mtime = ts[1];
        if (nullflag)
                vattr->va_vaflags |= VA_UTIMES_NULL;
        error = VOP_SETATTR(vp, vattr, td->td_ucred);

        return error;
}

int
kern_utimes(struct nlookupdata *nd, struct timeval *tptr)
{
        struct timespec ts[2];
        int error;

        if (tptr) {
                if ((error = getutimes(tptr, ts)) != 0)
                        return (error);
        }
        error = kern_utimensat(nd, tptr ? ts : NULL, 0);
        return (error);
}

/*
 * utimes_args(char *path, struct timeval *tptr)
 *
 * Set the access and modification times of a file.
 */
int
sys_utimes(struct sysmsg *sysmsg, const struct utimes_args *uap)
{
        struct timeval tv[2];
        struct nlookupdata nd;
        int error;

        if (uap->tptr) {
                error = copyin(uap->tptr, tv, sizeof(tv));
                if (error)
                        return (error);
        }
        error = nlookup_init(&nd, uap->path, UIO_USERSPACE, NLC_FOLLOW);
        if (error == 0)
                error = kern_utimes(&nd, uap->tptr ? tv : NULL);
        nlookup_done(&nd);
        return (error);
}

/*
 * lutimes_args(char *path, struct timeval *tptr)
 *
 * Set the access and modification times of a file.
 */
int
sys_lutimes(struct sysmsg *sysmsg, const struct lutimes_args *uap)
{
        struct timeval tv[2];
        struct nlookupdata nd;
        int error;

        if (uap->tptr) {
                error = copyin(uap->tptr, tv, sizeof(tv));
                if (error)
                        return (error);
        }
        error = nlookup_init(&nd, uap->path, UIO_USERSPACE, 0);
        if (error == 0)
                error = kern_utimes(&nd, uap->tptr ? tv : NULL);
        nlookup_done(&nd);
        return (error);
}

/*
 * Set utimes on a file descriptor.  The creds used to open the
 * file are used to determine whether the operation is allowed
 * or not.
 */
int
kern_futimens(int fd, struct timespec *ts)
{
        struct thread *td = curthread;
        struct timespec newts[2];
        struct file *fp;
        struct vnode *vp;
        struct vattr vattr;
        struct vattr_lite lva;
        int nullflag;
        int error;

        error = getutimens(ts, newts, &nullflag);
        if (error)
                return (error);
        if ((error = holdvnode(td, fd, &fp)) != 0)
                return (error);
        if (fp->f_nchandle.ncp)
                error = ncp_writechk(&fp->f_nchandle);
        if (error == 0) {
                vp = fp->f_data;
                error = vget(vp, LK_EXCLUSIVE);
                if (error == 0) {
                        error = VOP_GETATTR_FP(vp, &vattr, fp);
                        if (error == 0) {
                                lva.va_type = vattr.va_type;
                                lva.va_nlink = vattr.va_nlink;
                                lva.va_mode = vattr.va_mode;
                                lva.va_uid = vattr.va_uid;
                                lva.va_gid = vattr.va_gid;
                                lva.va_size = vattr.va_size;
                                lva.va_flags = vattr.va_flags;

                                error = naccess_lva(&lva, NLC_OWN | NLC_WRITE,
                                                   fp->f_cred);
                        }
                        if (error == 0) {
                                error = setutimes(vp, &vattr, newts, nullflag);
                        }
                        vput(vp);
                }
        }
        fdrop(fp);
        return (error);
}

/*
 * futimens_args(int fd, struct timespec *ts)
 *
 * Set the access and modification times of a file.
 */
int
sys_futimens(struct sysmsg *sysmsg, const struct futimens_args *uap)
{
        struct timespec ts[2];
        int error;

        if (uap->ts) {
                error = copyin(uap->ts, ts, sizeof(ts));
                if (error)
                        return (error);
        }
        error = kern_futimens(uap->fd, uap->ts ? ts : NULL);
        return (error);
}

int
kern_futimes(int fd, struct timeval *tptr)
{
        struct timespec ts[2];
        int error;

        if (tptr) {
                if ((error = getutimes(tptr, ts)) != 0)
                        return (error);
        }
        error = kern_futimens(fd, tptr ? ts : NULL);
        return (error);
}

/*
 * futimes_args(int fd, struct timeval *tptr)
 *
 * Set the access and modification times of a file.
 */
int
sys_futimes(struct sysmsg *sysmsg, const struct futimes_args *uap)
{
        struct timeval tv[2];
        int error;

        if (uap->tptr) {
                error = copyin(uap->tptr, tv, sizeof(tv));
                if (error)
                        return (error);
        }
        error = kern_futimes(uap->fd, uap->tptr ? tv : NULL);
        return (error);
}

int
kern_utimensat(struct nlookupdata *nd, const struct timespec *ts, int flags)
{
        struct timespec newts[2];
        struct vnode *vp;
        struct vattr vattr;
        int nullflag;
        int error;

        if (flags & ~AT_SYMLINK_NOFOLLOW)
                return (EINVAL);

        error = getutimens(ts, newts, &nullflag);
        if (error)
                return (error);

        nd->nl_flags |= NLC_OWN | NLC_WRITE;
        if ((error = nlookup(nd)) != 0)
                return (error);
        if ((error = ncp_writechk(&nd->nl_nch)) != 0)
                return (error);
        if ((error = cache_vref(&nd->nl_nch, nd->nl_cred, &vp)) != 0)
                return (error);
        if ((error = vn_writechk(vp)) == 0) {
                error = vget(vp, LK_EXCLUSIVE);
                if (error == 0) {
                        error = setutimes(vp, &vattr, newts, nullflag);
                        vput(vp);
                }
        }
        vrele(vp);
        return (error);
}

/*
 * utimensat_args(int fd, const char *path, const struct timespec *ts, int flags);
 *
 * Set file access and modification times of a file.
 */
int
sys_utimensat(struct sysmsg *sysmsg, const struct utimensat_args *uap)
{
        struct timespec ts[2];
        struct nlookupdata nd;
        struct file *fp;
        int error;
        int flags;

        if (uap->ts) {
                error = copyin(uap->ts, ts, sizeof(ts));
                if (error)
                        return (error);
        }

        flags = (uap->flags & AT_SYMLINK_NOFOLLOW) ? 0 : NLC_FOLLOW;
        error = nlookup_init_at(&nd, &fp, uap->fd, uap->path,
                                UIO_USERSPACE, flags);
        if (error == 0)
                error = kern_utimensat(&nd, uap->ts ? ts : NULL, uap->flags);
        nlookup_done_at(&nd, fp);
        return (error);
}

int
kern_truncate(struct nlookupdata *nd, off_t length)
{
        struct vnode *vp;
        struct vattr vattr;
        int error;
        uid_t uid = 0;
        gid_t gid = 0;
        uint64_t old_size = 0;

        if (length < 0)
                return(EINVAL);
        nd->nl_flags |= NLC_WRITE | NLC_TRUNCATE;
        if ((error = nlookup(nd)) != 0)
                return (error);
        if ((error = ncp_writechk(&nd->nl_nch)) != 0)
                return (error);
        if ((error = cache_vref(&nd->nl_nch, nd->nl_cred, &vp)) != 0)
                return (error);
        error = vn_lock(vp, LK_EXCLUSIVE | LK_RETRY | LK_FAILRECLAIM);
        if (error) {
                vrele(vp);
                return (error);
        }
        if (vp->v_type == VDIR) {
                error = EISDIR;
                goto done;
        }
        if (vfs_quota_enabled) {
                error = VOP_GETATTR(vp, &vattr);
                KASSERT(error == 0, ("kern_truncate(): VOP_GETATTR didn't return 0"));
                uid = vattr.va_uid;
                gid = vattr.va_gid;
                old_size = vattr.va_size;
        }

        if ((error = vn_writechk(vp)) == 0) {
                VATTR_NULL(&vattr);
                vattr.va_size = length;
                error = VOP_SETATTR(vp, &vattr, nd->nl_cred);
                VFS_ACCOUNT(nd->nl_nch.mount, uid, gid, length - old_size);
        }
done:
        vput(vp);
        return (error);
}

/*
 * truncate(char *path, int pad, off_t length)
 *
 * Truncate a file given its path name.
 */
int
sys_truncate(struct sysmsg *sysmsg, const struct truncate_args *uap)
{
        struct nlookupdata nd;
        int error;

        error = nlookup_init(&nd, uap->path, UIO_USERSPACE, NLC_FOLLOW);
        if (error == 0)
                error = kern_truncate(&nd, uap->length);
        nlookup_done(&nd);
        return error;
}

int
kern_ftruncate(int fd, off_t length)
{
        struct thread *td = curthread;
        struct vattr vattr;
        struct vnode *vp;
        struct file *fp;
        int error;
        uid_t uid = 0;
        gid_t gid = 0;
        uint64_t old_size = 0;
        struct mount *mp;

        if (length < 0)
                return(EINVAL);
        if ((error = holdvnode(td, fd, &fp)) != 0)
                return (error);
        if (fp->f_nchandle.ncp) {
                error = ncp_writechk(&fp->f_nchandle);
                if (error)
                        goto done;
        }
        if ((fp->f_flag & FWRITE) == 0) {
                error = EINVAL;
                goto done;
        }
        if (fp->f_flag & FAPPENDONLY) { /* inode was set s/uapnd */
                error = EINVAL;
                goto done;
        }
        vp = (struct vnode *)fp->f_data;
        vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
        if (vp->v_type == VDIR) {
                error = EISDIR;
                vn_unlock(vp);
                goto done;
        }

        if (vfs_quota_enabled) {
                error = VOP_GETATTR_FP(vp, &vattr, fp);
                KASSERT(error == 0, ("kern_ftruncate(): VOP_GETATTR didn't return 0"));
                uid = vattr.va_uid;
                gid = vattr.va_gid;
                old_size = vattr.va_size;
        }

        if ((error = vn_writechk(vp)) == 0) {
                VATTR_NULL(&vattr);
                vattr.va_size = length;
                error = VOP_SETATTR_FP(vp, &vattr, fp->f_cred, fp);
                mp = vq_vptomp(vp);
                VFS_ACCOUNT(mp, uid, gid, length - old_size);
        }
        vn_unlock(vp);
done:
        fdrop(fp);
        return (error);
}

/*
 * ftruncate_args(int fd, int pad, off_t length)
 *
 * Truncate a file given a file descriptor.
 */
int
sys_ftruncate(struct sysmsg *sysmsg, const struct ftruncate_args *uap)
{
        int error;

        error = kern_ftruncate(uap->fd, uap->length);

        return (error);
}

/*
 * fsync(int fd)
 *
 * Sync an open file.
 */
int
sys_fsync(struct sysmsg *sysmsg, const struct fsync_args *uap)
{
        struct thread *td = curthread;
        struct vnode *vp;
        struct file *fp;
        vm_object_t obj;
        int error;

        if ((error = holdvnode(td, uap->fd, &fp)) != 0)
                return (error);
        vp = (struct vnode *)fp->f_data;
        vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
        if ((obj = vp->v_object) != NULL) {
                if (vp->v_mount == NULL ||
                    (vp->v_mount->mnt_kern_flag & MNTK_NOMSYNC) == 0) {
                        vm_object_page_clean(obj, 0, 0, 0);
                }
        }
        error = VOP_FSYNC_FP(vp, MNT_WAIT, VOP_FSYNC_SYSCALL, fp);
        if (error == 0 && vp->v_mount)
                error = buf_fsync(vp);
        vn_unlock(vp);
        fdrop(fp);

        return (error);
}

int
kern_rename(struct nlookupdata *fromnd, struct nlookupdata *tond)
{
        struct nchandle fnchd;
        struct nchandle tnchd;
        struct namecache *ncp;
        struct vnode *fdvp;
        struct vnode *tdvp;
        struct mount *mp;
        struct mount *userenlk;
        int error;
        u_int fncp_gen;
        u_int tncp_gen;

        bwillinode(1);
        fromnd->nl_flags |= NLC_REFDVP | NLC_RENAME_SRC;
        if ((error = nlookup(fromnd)) != 0)
                return (error);
        if ((fnchd.ncp = fromnd->nl_nch.ncp->nc_parent) == NULL)
                return (ENOENT);
        fnchd.mount = fromnd->nl_nch.mount;
        cache_hold(&fnchd);

        /*
         * unlock the source nch so we can lookup the target nch without
         * deadlocking.  The target may or may not exist so we do not check
         * for a target vp like kern_mkdir() and other creation functions do.
         *
         * The source and target directories are ref'd and rechecked after
         * everything is relocked to determine if the source or target file
         * has been renamed.
         */
        KKASSERT(fromnd->nl_flags & NLC_NCPISLOCKED);
        fromnd->nl_flags &= ~NLC_NCPISLOCKED;
        fncp_gen = fromnd->nl_nch.ncp->nc_generation;

        if (fromnd->nl_nch.ncp->nc_vp &&
            fromnd->nl_nch.ncp->nc_vp->v_type == VDIR) {
                userenlk = fnchd.mount;
                cache_unlock(&fromnd->nl_nch);
                lockmgr(&userenlk->mnt_renlock, LK_EXCLUSIVE);
        } else {
                userenlk = NULL;
                cache_unlock(&fromnd->nl_nch);
        }

        /*
         * Lookup target
         */
        tond->nl_flags |= NLC_RENAME_DST | NLC_REFDVP;
        if ((error = nlookup(tond)) != 0) {
                cache_drop(&fnchd);
                goto done;
        }
        tncp_gen = tond->nl_nch.ncp->nc_generation;

        if ((tnchd.ncp = tond->nl_nch.ncp->nc_parent) == NULL) {
                cache_drop(&fnchd);
                error = ENOENT;
                goto done;
        }
        tnchd.mount = tond->nl_nch.mount;
        cache_hold(&tnchd);

        /*
         * If the source and target are the same there is nothing to do
         */
        if (fromnd->nl_nch.ncp == tond->nl_nch.ncp) {
                cache_drop(&fnchd);
                cache_drop(&tnchd);
                error = 0;
                goto done;
        }

        /*
         * Mount points cannot be renamed or overwritten
         */
        if ((fromnd->nl_nch.ncp->nc_flag | tond->nl_nch.ncp->nc_flag) &
            NCF_ISMOUNTPT
        ) {
                cache_drop(&fnchd);
                cache_drop(&tnchd);
                error = EINVAL;
                goto done;
        }

        /*
         * Lock all four namecache entries.  tond is already locked.
         */
        cache_lock4_tondlocked(&fnchd, &fromnd->nl_nch,
                               &tnchd, &tond->nl_nch,
                               fromnd->nl_cred, tond->nl_cred);
        fromnd->nl_flags |= NLC_NCPISLOCKED;

        /*
         * If the namecache generation changed for either fromnd or tond,
         * we must retry.
         */
        if (fromnd->nl_nch.ncp->nc_generation != fncp_gen ||
            tond->nl_nch.ncp->nc_generation != tncp_gen) {
                krateprintf(&krate_rename,
                        "kern_rename: retry due to race on: "
                        "\"%s\" -> \"%s\"\n",
                        fromnd->nl_nch.ncp->nc_name,
                        tond->nl_nch.ncp->nc_name);
                error = EAGAIN;
                goto finish;
        }

        /*
         * If either fromnd or tond are marked destroyed a ripout occured
         * out from under us and we must retry.
         */
        if ((fromnd->nl_nch.ncp->nc_flag & (NCF_DESTROYED | NCF_UNRESOLVED)) ||
            fromnd->nl_nch.ncp->nc_vp == NULL ||
            (tond->nl_nch.ncp->nc_flag & (NCF_DESTROYED | NCF_UNRESOLVED))) {
                krateprintf(&krate_rename,
                        "kern_rename: retry due to ripout on: "
                        "\"%s\" -> \"%s\"\n",
                        fromnd->nl_nch.ncp->nc_name,
                        tond->nl_nch.ncp->nc_name);
                error = EAGAIN;
                goto finish;
        }

        /*
         * Make sure the parent directories linkages are the same.  We have
         * already checked that fromnd and tond are not mount points so this
         * should not loop forever on a cross-mount.
         */
        if (fnchd.ncp != fromnd->nl_nch.ncp->nc_parent ||
            tnchd.ncp != tond->nl_nch.ncp->nc_parent) {
                error = EAGAIN;
                goto finish;
        }

        /*
         * Both the source and target must be within the same filesystem and
         * in the same filesystem as their parent directories within the
         * namecache topology.
         *
         * NOTE: fromnd's nc_mount or nc_vp could be NULL.
         */
        mp = fnchd.mount;
        if (mp != tnchd.mount || mp != fromnd->nl_nch.mount ||
            mp != tond->nl_nch.mount) {
                error = EXDEV;
                goto finish;
        }

        /*
         * Make sure the mount point is writable
         */
        if ((error = ncp_writechk(&tond->nl_nch)) != 0) {
                goto finish;
        }

        /*
         * If the target exists and either the source or target is a directory,
         * then both must be directories.
         *
         * Due to relocking of the source, fromnd->nl_nch.ncp->nc_vp might h
         * have become NULL.
         */
        if (tond->nl_nch.ncp->nc_vp) {
                if (fromnd->nl_nch.ncp->nc_vp == NULL) {
                        error = ENOENT;
                } else if (fromnd->nl_nch.ncp->nc_vp->v_type == VDIR) {
                        if (tond->nl_nch.ncp->nc_vp->v_type != VDIR)
                                error = ENOTDIR;
                } else if (tond->nl_nch.ncp->nc_vp->v_type == VDIR) {
                        error = EISDIR;
                }
        }

        /*
         * You cannot rename a source into itself or a subdirectory of itself.
         * We check this by travsersing the target directory upwards looking
         * for a match against the source.
         *
         * Only required when renaming a directory, in which case userenlk is
         * non-NULL.
         */
        if (__predict_false(userenlk && error == 0)) {
                for (ncp = tnchd.ncp; ncp; ncp = ncp->nc_parent) {
                        if (fromnd->nl_nch.ncp == ncp) {
                                error = EINVAL;
                                break;
                        }
                }
        }

        /*
         * Even though the namespaces are different, they may still represent
         * hardlinks to the same file.  The filesystem might have a hard time
         * with this so we issue a NREMOVE of the source instead of a NRENAME
         * when we detect the situation.
         */
        if (error == 0) {
                fdvp = fromnd->nl_dvp;
                tdvp = tond->nl_dvp;
                if (fdvp == NULL || tdvp == NULL) {
                        error = EPERM;
                } else if (fromnd->nl_nch.ncp->nc_vp == tond->nl_nch.ncp->nc_vp) {
                        error = VOP_NREMOVE(&fromnd->nl_nch, fdvp,
                                            fromnd->nl_cred);
                } else {
                        error = VOP_NRENAME(&fromnd->nl_nch, &tond->nl_nch, 
                                            fdvp, tdvp, tond->nl_cred);
                }
        }
finish:
        cache_put(&tnchd);
        cache_put(&fnchd);
done:
        if (userenlk)
                lockmgr(&userenlk->mnt_renlock, LK_RELEASE);
        return (error);
}

/*
 * rename_args(char *from, char *to)
 *
 * Rename files.  Source and destination must either both be directories,
 * or both not be directories.  If target is a directory, it must be empty.
 */
int
sys_rename(struct sysmsg *sysmsg, const struct rename_args *uap)
{
        struct nlookupdata fromnd, tond;
        int error;

        do {
                error = nlookup_init(&fromnd, uap->from, UIO_USERSPACE, 0);
                if (error == 0) {
                        error = nlookup_init(&tond, uap->to, UIO_USERSPACE, 0);
                        if (error == 0)
                                error = kern_rename(&fromnd, &tond);
                        nlookup_done(&tond);
                }
                nlookup_done(&fromnd);
        } while (error == EAGAIN);
        return (error);
}

/*
 * renameat_args(int oldfd, char *old, int newfd, char *new)
 *
 * Rename files using paths relative to the directories associated with
 * oldfd and newfd.  Source and destination must either both be directories,
 * or both not be directories.  If target is a directory, it must be empty.
 */
int
sys_renameat(struct sysmsg *sysmsg, const struct renameat_args *uap)
{
        struct nlookupdata oldnd, newnd;
        struct file *oldfp, *newfp;
        int error;

        do {
                error = nlookup_init_at(&oldnd, &oldfp,
                                        uap->oldfd, uap->old,
                                        UIO_USERSPACE, 0);
                if (error == 0) {
                        error = nlookup_init_at(&newnd, &newfp,
                                                uap->newfd, uap->new,
                                                UIO_USERSPACE, 0);
                        if (error == 0)
                                error = kern_rename(&oldnd, &newnd);
                        nlookup_done_at(&newnd, newfp);
                }
                nlookup_done_at(&oldnd, oldfp);
        } while (error == EAGAIN);
        return (error);
}

int
kern_mkdir(struct nlookupdata *nd, int mode)
{
        struct thread *td = curthread;
        struct proc *p = td->td_proc;
        struct vnode *vp;
        struct vattr vattr;
        int error;

        bwillinode(1);
        nd->nl_flags |= NLC_WILLBEDIR | NLC_CREATE | NLC_REFDVP;
        if ((error = nlookup(nd)) != 0)
                return (error);

        if (nd->nl_nch.ncp->nc_vp)
                return (EEXIST);
        if (nd->nl_dvp == NULL)
                return (EINVAL);
        if ((error = ncp_writechk(&nd->nl_nch)) != 0)
                return (error);
        VATTR_NULL(&vattr);
        vattr.va_type = VDIR;
        vattr.va_mode = (mode & ACCESSPERMS) &~ p->p_fd->fd_cmask;

        vp = NULL;
        error = VOP_NMKDIR(&nd->nl_nch, nd->nl_dvp, &vp, td->td_ucred, &vattr);
        if (error == 0)
                vput(vp);
        return (error);
}

/*
 * mkdir_args(char *path, int mode)
 *
 * Make a directory file.
 */
int
sys_mkdir(struct sysmsg *sysmsg, const struct mkdir_args *uap)
{
        struct nlookupdata nd;
        int error;

        error = nlookup_init(&nd, uap->path, UIO_USERSPACE, 0);
        if (error == 0)
                error = kern_mkdir(&nd, uap->mode);
        nlookup_done(&nd);
        return (error);
}

/*
 * mkdirat_args(int fd, char *path, mode_t mode)
 *
 * Make a directory file.  The path is relative to the directory associated
 * with fd.
 */
int
sys_mkdirat(struct sysmsg *sysmsg, const struct mkdirat_args *uap)
{
        struct nlookupdata nd;
        struct file *fp;
        int error;

        error = nlookup_init_at(&nd, &fp, uap->fd, uap->path, UIO_USERSPACE, 0);
        if (error == 0)
                error = kern_mkdir(&nd, uap->mode);
        nlookup_done_at(&nd, fp);
        return (error);
}

int
kern_rmdir(struct nlookupdata *nd)
{
        int error;

        bwillinode(1);
        nd->nl_flags |= NLC_DELETE | NLC_REFDVP;
        if ((error = nlookup(nd)) != 0)
                return (error);

        /*
         * Do not allow directories representing mount points to be
         * deleted, even if empty.  Check write perms on mount point
         * in case the vnode is aliased (aka nullfs).
         */
        if (nd->nl_nch.ncp->nc_flag & (NCF_ISMOUNTPT))
                return (EBUSY);
        if (nd->nl_dvp == NULL)
                return (EINVAL);
        if ((error = ncp_writechk(&nd->nl_nch)) != 0)
                return (error);
        error = VOP_NRMDIR(&nd->nl_nch, nd->nl_dvp, nd->nl_cred);
        return (error);
}

/*
 * rmdir_args(char *path)
 *
 * Remove a directory file.
 */
int
sys_rmdir(struct sysmsg *sysmsg, const struct rmdir_args *uap)
{
        struct nlookupdata nd;
        int error;

        error = nlookup_init(&nd, uap->path, UIO_USERSPACE, 0);
        if (error == 0)
                error = kern_rmdir(&nd);
        nlookup_done(&nd);
        return (error);
}

int
kern_getdirentries(int fd, char *buf, u_int count, long *basep, int *res,
                   enum uio_seg direction)
{
        struct thread *td = curthread;
        struct vnode *vp;
        struct file *fp;
        struct uio auio;
        struct iovec aiov;
        off_t loff;
        int error, eofflag;

        if ((error = holdvnode(td, fd, &fp)) != 0)
                return (error);
        if ((fp->f_flag & FREAD) == 0) {
                error = EBADF;
                goto done;
        }
        vp = (struct vnode *)fp->f_data;
        if (vp->v_type != VDIR) {
                error = EINVAL;
                goto done;
        }
        aiov.iov_base = buf;
        aiov.iov_len = count;
        auio.uio_iov = &aiov;
        auio.uio_iovcnt = 1;
        auio.uio_rw = UIO_READ;
        auio.uio_segflg = direction;
        auio.uio_td = td;
        auio.uio_resid = count;
        loff = auio.uio_offset = fp->f_offset;
        error = VOP_READDIR_FP(vp, &auio, fp->f_cred, &eofflag, NULL, NULL, fp);
        fp->f_offset = auio.uio_offset;
        if (error)
                goto done;

        /*
         * WARNING!  *basep may not be wide enough to accomodate the
         * seek offset.   XXX should we hack this to return the upper 32 bits
         * for offsets greater then 4G?
         */
        if (basep) {
                *basep = (long)loff;
        }
        *res = count - auio.uio_resid;
done:
        fdrop(fp);
        return (error);
}

/*
 * getdirentries_args(int fd, char *buf, u_int conut, long *basep)
 *
 * Read a block of directory entries in a file system independent format.
 */
int
sys_getdirentries(struct sysmsg *sysmsg, const struct getdirentries_args *uap)
{
        long base;
        int error;

        error = kern_getdirentries(uap->fd, uap->buf, uap->count, &base,
                                   &sysmsg->sysmsg_result, UIO_USERSPACE);

        if (error == 0 && uap->basep)
                error = copyout(&base, uap->basep, sizeof(*uap->basep));
        return (error);
}

/*
 * getdents_args(int fd, char *buf, size_t count)
 */
int
sys_getdents(struct sysmsg *sysmsg, const struct getdents_args *uap)
{
        int error;

        error = kern_getdirentries(uap->fd, uap->buf, uap->count, NULL,
                                   &sysmsg->sysmsg_result, UIO_USERSPACE);

        return (error);
}

/*
 * Set the mode mask for creation of filesystem nodes.
 *
 * umask(int newmask)
 */
int
sys_umask(struct sysmsg *sysmsg, const struct umask_args *uap)
{
        struct thread *td = curthread;
        struct proc *p = td->td_proc;
        struct filedesc *fdp;

        fdp = p->p_fd;
        sysmsg->sysmsg_result = fdp->fd_cmask;
        fdp->fd_cmask = uap->newmask & ALLPERMS;
        return (0);
}

/*
 * revoke(char *path)
 *
 * Void all references to file by ripping underlying filesystem
 * away from vnode.
 */
int
sys_revoke(struct sysmsg *sysmsg, const struct revoke_args *uap)
{
        struct nlookupdata nd;
        struct vattr vattr;
        struct vnode *vp;
        struct ucred *cred;
        int error;

        vp = NULL;
        error = nlookup_init(&nd, uap->path, UIO_USERSPACE, NLC_FOLLOW);
        if (error == 0)
                error = nlookup(&nd);
        if (error == 0)
                error = cache_vref(&nd.nl_nch, nd.nl_cred, &vp);
        cred = crhold(nd.nl_cred);
        nlookup_done(&nd);
        if (error == 0) {
                if (error == 0)
                        error = VOP_GETATTR(vp, &vattr);
                if (error == 0 && cred->cr_uid != vattr.va_uid)
                        error = priv_check_cred(cred, PRIV_VFS_REVOKE, 0);
                if (error == 0 && (vp->v_type == VCHR || vp->v_type == VBLK)) {
                        if (vcount(vp) > 0)
                                error = vrevoke(vp, cred);
                } else if (error == 0) {
                        error = vrevoke(vp, cred);
                }
                vrele(vp);
        }
        if (cred)
                crfree(cred);
        return (error);
}

/*
 * getfh_args(char *fname, fhandle_t *fhp)
 *
 * Get (NFS) file handle
 *
 * NOTE: We use the fsid of the covering mount, even if it is a nullfs
 * mount.  This allows nullfs mounts to be explicitly exported. 
 *
 * WARNING: nullfs mounts of HAMMER PFS ROOTs are safe.
 *
 *          nullfs mounts of subdirectories are not safe.  That is, it will
 *          work, but you do not really have protection against access to
 *          the related parent directories.
 */
int
sys_getfh(struct sysmsg *sysmsg, const struct getfh_args *uap)
{
        struct thread *td = curthread;
        struct nlookupdata nd;
        fhandle_t fh;
        struct vnode *vp;
        struct mount *mp;
        int error;

        /*
         * Must be super user
         */
        if ((error = priv_check(td, PRIV_ROOT)) != 0)
                return (error);

        vp = NULL;
        error = nlookup_init(&nd, uap->fname, UIO_USERSPACE, NLC_FOLLOW);
        if (error == 0)
                error = nlookup(&nd);
        if (error == 0)
                error = cache_vget(&nd.nl_nch, nd.nl_cred, LK_EXCLUSIVE, &vp);
        mp = nd.nl_nch.mount;
        nlookup_done(&nd);
        if (error == 0) {
                bzero(&fh, sizeof(fh));
                fh.fh_fsid = mp->mnt_stat.f_fsid;
                error = VFS_VPTOFH(vp, &fh.fh_fid);
                vput(vp);
                if (error == 0)
                        error = copyout(&fh, uap->fhp, sizeof(fh));
        }
        return (error);
}

/*
 * fhopen_args(const struct fhandle *u_fhp, int flags)
 *
 * syscall for the rpc.lockd to use to translate a NFS file handle into
 * an open descriptor.
 *
 * warning: do not remove the priv_check() call or this becomes one giant
 * security hole.
 */
int
sys_fhopen(struct sysmsg *sysmsg, const struct fhopen_args *uap)
{
        struct thread *td = curthread;
        struct filedesc *fdp = td->td_proc->p_fd;
        struct mount *mp;
        struct vnode *vp;
        struct fhandle fhp;
        struct vattr vat;
        struct vattr *vap = &vat;
        struct flock lf;
        int fmode, mode, error = 0, type;
        struct file *nfp; 
        struct file *fp;
        int indx;

        /*
         * Must be super user
         */
        error = priv_check(td, PRIV_ROOT);
        if (error)
                return (error);

        fmode = FFLAGS(uap->flags);

        /*
         * Why not allow a non-read/write open for our lockd?
         */
        if (((fmode & (FREAD | FWRITE)) == 0) || (fmode & O_CREAT))
                return (EINVAL);
        error = copyin(uap->u_fhp, &fhp, sizeof(fhp));
        if (error)
                return(error);

        /*
         * Find the mount point
         */
        mp = vfs_getvfs(&fhp.fh_fsid);
        if (mp == NULL) {
                error = ESTALE;
                goto done2;
        }
        /* now give me my vnode, it gets returned to me locked */
        error = VFS_FHTOVP(mp, NULL, &fhp.fh_fid, &vp);
        if (error)
                goto done;
        /*
         * from now on we have to make sure not
         * to forget about the vnode
         * any error that causes an abort must vput(vp) 
         * just set error = err and 'goto bad;'.
         */

        /* 
         * from vn_open 
         */
        if (vp->v_type == VLNK) {
                error = EMLINK;
                goto bad;
        }
        if (vp->v_type == VSOCK) {
                error = EOPNOTSUPP;
                goto bad;
        }
        mode = 0;
        if (fmode & (FWRITE | O_TRUNC)) {
                if (vp->v_type == VDIR) {
                        error = EISDIR;
                        goto bad;
                }
                error = vn_writechk(vp);
                if (error)
                        goto bad;
                mode |= VWRITE;
        }
        if (fmode & FREAD)
                mode |= VREAD;
        if (mode) {
                error = VOP_ACCESS(vp, mode, td->td_ucred);
                if (error)
                        goto bad;
        }
        if (fmode & O_TRUNC) {
                vn_unlock(vp);                          /* XXX */
                vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);   /* XXX */
                VATTR_NULL(vap);
                vap->va_size = 0;
                error = VOP_SETATTR(vp, vap, td->td_ucred);
                if (error)
                        goto bad;
        }

        /*
         * VOP_OPEN needs the file pointer so it can potentially override
         * it.
         *
         * WARNING! no f_nchandle will be associated when fhopen()ing a
         * directory.  XXX
         */
        if ((error = falloc(td->td_lwp, &nfp, &indx)) != 0)
                goto bad;
        fp = nfp;

        error = VOP_OPEN(vp, fmode, td->td_ucred, fp);
        if (error) {
                /*
                 * setting f_ops this way prevents VOP_CLOSE from being
                 * called or fdrop() releasing the vp from v_data.   Since
                 * the VOP_OPEN failed we don't want to VOP_CLOSE.
                 */
                fp->f_ops = &badfileops;
                fp->f_data = NULL;
                goto bad_drop;
        }

        /*
         * The fp is given its own reference, we still have our ref and lock.
         *
         * Assert that all regular files must be created with a VM object.
         */
        if (vp->v_type == VREG && vp->v_object == NULL) {
                kprintf("fhopen: regular file did not "
                        "have VM object: %p\n",
                        vp);
                goto bad_drop;
        }

        /*
         * The open was successful.  Handle any locking requirements.
         */
        if (fmode & (O_EXLOCK | O_SHLOCK)) {
                lf.l_whence = SEEK_SET;
                lf.l_start = 0;
                lf.l_len = 0;
                if (fmode & O_EXLOCK)
                        lf.l_type = F_WRLCK;
                else
                        lf.l_type = F_RDLCK;
                if (fmode & FNONBLOCK)
                        type = 0;
                else
                        type = F_WAIT;
                vn_unlock(vp);
                if ((error = VOP_ADVLOCK(vp, (caddr_t)fp, F_SETLK,
                                         &lf, type)) != 0) {
                        /*
                         * release our private reference.
                         */
                        fsetfd(fdp, NULL, indx);
                        fdrop(fp);
                        vrele(vp);
                        goto done;
                }
                vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
                atomic_set_int(&fp->f_flag, FHASLOCK);  /* race ok */
        }

        /*
         * Clean up.  Associate the file pointer with the previously
         * reserved descriptor and return it.
         */
        vput(vp);
        if (uap->flags & O_CLOEXEC)
                fdp->fd_files[indx].fileflags |= UF_EXCLOSE;
        fsetfd(fdp, fp, indx);
        fdrop(fp);
        sysmsg->sysmsg_result = indx;
        mount_drop(mp);

        return (error);

bad_drop:
        fsetfd(fdp, NULL, indx);
        fdrop(fp);
bad:
        vput(vp);
done:
        mount_drop(mp);
done2:
        return (error);
}

/*
 * fhstat_args(struct fhandle *u_fhp, struct stat *sb)
 */
int
sys_fhstat(struct sysmsg *sysmsg, const struct fhstat_args *uap)
{
        struct thread *td = curthread;
        struct stat sb;
        fhandle_t fh;
        struct mount *mp;
        struct vnode *vp;
        int error;

        /*
         * Must be super user
         */
        error = priv_check(td, PRIV_ROOT);
        if (error)
                return (error);
        
        error = copyin(uap->u_fhp, &fh, sizeof(fhandle_t));
        if (error)
                return (error);

        if ((mp = vfs_getvfs(&fh.fh_fsid)) == NULL)
                error = ESTALE;
        if (error == 0) {
                if ((error = VFS_FHTOVP(mp, NULL, &fh.fh_fid, &vp)) == 0) {
                        error = vn_stat(vp, &sb, td->td_ucred);
                        vput(vp);
                }
        }
        if (error == 0)
                error = copyout(&sb, uap->sb, sizeof(sb));
        if (mp)
                mount_drop(mp);

        return (error);
}

/*
 * fhstatfs_args(struct fhandle *u_fhp, struct statfs *buf)
 */
int
sys_fhstatfs(struct sysmsg *sysmsg, const struct fhstatfs_args *uap)
{
        struct thread *td = curthread;
        struct proc *p = td->td_proc;
        struct statfs *sp;
        struct mount *mp;
        struct vnode *vp;
        struct statfs sb;
        char *fullpath, *freepath;
        fhandle_t fh;
        int error;

        /*
         * Must be super user
         */
        if ((error = priv_check(td, PRIV_ROOT)))
                return (error);

        if ((error = copyin(uap->u_fhp, &fh, sizeof(fhandle_t))) != 0)
                return (error);

        if ((mp = vfs_getvfs(&fh.fh_fsid)) == NULL) {
                error = ESTALE;
                goto done;
        }
        if (p != NULL && !chroot_visible_mnt(mp, p)) {
                error = ESTALE;
                goto done;
        }

        if ((error = VFS_FHTOVP(mp, NULL, &fh.fh_fid, &vp)) != 0)
                goto done;
        mp = vp->v_mount;
        sp = &mp->mnt_stat;
        vput(vp);
        if ((error = VFS_STATFS(mp, sp, td->td_ucred)) != 0)
                goto done;

        error = mount_path(p, mp, &fullpath, &freepath);
        if (error)
                goto done;
        bzero(sp->f_mntonname, sizeof(sp->f_mntonname));
        strlcpy(sp->f_mntonname, fullpath, sizeof(sp->f_mntonname));
        kfree(freepath, M_TEMP);

        sp->f_flags = mp->mnt_flag & MNT_VISFLAGMASK;
        if (priv_check(td, PRIV_ROOT)) {
                bcopy(sp, &sb, sizeof(sb));
                sb.f_fsid.val[0] = sb.f_fsid.val[1] = 0;
                sp = &sb;
        }
        error = copyout(sp, uap->buf, sizeof(*sp));
done:
        if (mp)
                mount_drop(mp);

        return (error);
}

/*
 * fhstatvfs_args(struct fhandle *u_fhp, struct statvfs *buf)
 */
int
sys_fhstatvfs(struct sysmsg *sysmsg, const struct fhstatvfs_args *uap)
{
        struct thread *td = curthread;
        struct proc *p = td->td_proc;
        struct statvfs *sp;
        struct mount *mp;
        struct vnode *vp;
        fhandle_t fh;
        int error;

        /*
         * Must be super user
         */
        if ((error = priv_check(td, PRIV_ROOT)))
                return (error);

        if ((error = copyin(uap->u_fhp, &fh, sizeof(fhandle_t))) != 0)
                return (error);

        if ((mp = vfs_getvfs(&fh.fh_fsid)) == NULL) {
                error = ESTALE;
                goto done;
        }
        if (p != NULL && !chroot_visible_mnt(mp, p)) {
                error = ESTALE;
                goto done;
        }

        if ((error = VFS_FHTOVP(mp, NULL, &fh.fh_fid, &vp)))
                goto done;
        mp = vp->v_mount;
        sp = &mp->mnt_vstat;
        vput(vp);
        if ((error = VFS_STATVFS(mp, sp, td->td_ucred)) != 0)
                goto done;

        sp->f_flag = 0;
        if (mp->mnt_flag & MNT_RDONLY)
                sp->f_flag |= ST_RDONLY;
        if (mp->mnt_flag & MNT_NOSUID)
                sp->f_flag |= ST_NOSUID;
        error = copyout(sp, uap->buf, sizeof(*sp));
done:
        if (mp)
                mount_drop(mp);
        return (error);
}


/*
 * Syscall to push extended attribute configuration information into the
 * VFS.  Accepts a path, which it converts to a mountpoint, as well as
 * a command (int cmd), and attribute name and misc data.  For now, the
 * attribute name is left in userspace for consumption by the VFS_op.
 * It will probably be changed to be copied into sysspace by the
 * syscall in the future, once issues with various consumers of the
 * attribute code have raised their hands.
 *
 * Currently this is used only by UFS Extended Attributes.
 */
int
sys_extattrctl(struct sysmsg *sysmsg, const struct extattrctl_args *uap)
{
        struct nlookupdata nd;
        struct vnode *vp;
        char attrname[EXTATTR_MAXNAMELEN];
        int error;
        size_t size;

        attrname[0] = 0;
        vp = NULL;
        error = 0;

        if (error == 0 && uap->filename) {
                error = nlookup_init(&nd, uap->filename, UIO_USERSPACE,
                                     NLC_FOLLOW);
                if (error == 0)
                        error = nlookup(&nd);
                if (error == 0)
                        error = cache_vref(&nd.nl_nch, nd.nl_cred, &vp);
                nlookup_done(&nd);
        }

        if (error == 0 && uap->attrname) {
                error = copyinstr(uap->attrname, attrname, EXTATTR_MAXNAMELEN,
                                  &size);
        }

        if (error == 0) {
                error = nlookup_init(&nd, uap->path, UIO_USERSPACE, NLC_FOLLOW);
                if (error == 0)
                        error = nlookup(&nd);
                if (error == 0)
                        error = ncp_writechk(&nd.nl_nch);
                if (error == 0) {
                        error = VFS_EXTATTRCTL(nd.nl_nch.mount, uap->cmd, vp,
                                               uap->attrnamespace,
                                               uap->attrname, nd.nl_cred);
                }
                nlookup_done(&nd);
        }

        return (error);
}

/*
 * Syscall to get a named extended attribute on a file or directory.
 */
int
sys_extattr_set_file(struct sysmsg *sysmsg,
                     const struct extattr_set_file_args *uap)
{
        char attrname[EXTATTR_MAXNAMELEN];
        struct nlookupdata nd;
        struct vnode *vp;
        struct uio auio;
        struct iovec aiov;
        int error;

        error = copyin(uap->attrname, attrname, EXTATTR_MAXNAMELEN);
        if (error)
                return (error);

        vp = NULL;

        error = nlookup_init(&nd, uap->path, UIO_USERSPACE, NLC_FOLLOW);
        if (error == 0)
                error = nlookup(&nd);
        if (error == 0)
                error = ncp_writechk(&nd.nl_nch);
        if (error == 0)
                error = cache_vget(&nd.nl_nch, nd.nl_cred, LK_EXCLUSIVE, &vp);
        if (error) {
                nlookup_done(&nd);
                return (error);
        }

        bzero(&auio, sizeof(auio));
        aiov.iov_base = uap->data;
        aiov.iov_len = uap->nbytes;
        auio.uio_iov = &aiov;
        auio.uio_iovcnt = 1;
        auio.uio_offset = 0;
        auio.uio_resid = uap->nbytes;
        auio.uio_rw = UIO_WRITE;
        auio.uio_td = curthread;

        error = VOP_SETEXTATTR(vp, uap->attrnamespace, attrname,
                               &auio, nd.nl_cred);

        vput(vp);
        nlookup_done(&nd);
        return (error);
}

/*
 * Syscall to get a named extended attribute on a file or directory.
 */
int
sys_extattr_get_file(struct sysmsg *sysmsg,
                     const struct extattr_get_file_args *uap)
{
        char attrname[EXTATTR_MAXNAMELEN];
        struct nlookupdata nd;
        struct uio auio;
        struct iovec aiov;
        struct vnode *vp;
        int error;

        error = copyin(uap->attrname, attrname, EXTATTR_MAXNAMELEN);
        if (error)
                return (error);

        vp = NULL;

        error = nlookup_init(&nd, uap->path, UIO_USERSPACE, NLC_FOLLOW);
        if (error == 0)
                error = nlookup(&nd);
        if (error == 0)
                error = cache_vget(&nd.nl_nch, nd.nl_cred, LK_SHARED, &vp);
        if (error) {
                nlookup_done(&nd);
                return (error);
        }

        bzero(&auio, sizeof(auio));
        aiov.iov_base = uap->data;
        aiov.iov_len = uap->nbytes;
        auio.uio_iov = &aiov;
        auio.uio_iovcnt = 1;
        auio.uio_offset = 0;
        auio.uio_resid = uap->nbytes;
        auio.uio_rw = UIO_READ;
        auio.uio_td = curthread;

        error = VOP_GETEXTATTR(vp, uap->attrnamespace, attrname,
                                &auio, nd.nl_cred);
        sysmsg->sysmsg_result = uap->nbytes - auio.uio_resid;

        vput(vp);
        nlookup_done(&nd);
        return(error);
}

/*
 * Syscall to delete a named extended attribute from a file or directory.
 * Accepts attribute name.  The real work happens in VOP_SETEXTATTR().
 */
int
sys_extattr_delete_file(struct sysmsg *sysmsg,
                        const struct extattr_delete_file_args *uap)
{
        char attrname[EXTATTR_MAXNAMELEN];
        struct nlookupdata nd;
        struct vnode *vp;
        int error;

        error = copyin(uap->attrname, attrname, EXTATTR_MAXNAMELEN);
        if (error)
                return(error);

        error = nlookup_init(&nd, uap->path, UIO_USERSPACE, NLC_FOLLOW);
        if (error == 0)
                error = nlookup(&nd);
        if (error == 0)
                error = ncp_writechk(&nd.nl_nch);
        if (error == 0) {
                error = cache_vget(&nd.nl_nch, nd.nl_cred, LK_EXCLUSIVE, &vp);
                if (error == 0) {
                        error = VOP_SETEXTATTR(vp, uap->attrnamespace,
                                               attrname, NULL, nd.nl_cred);
                        vput(vp);
                }
        }
        nlookup_done(&nd);
        return(error);
}

/*
 * Determine if the mount is visible to the process.
 */
static int
chroot_visible_mnt(struct mount *mp, struct proc *p)
{
        struct nchandle nch;

        /*
         * Traverse from the mount point upwards.  If we hit the process
         * root then the mount point is visible to the process.
         */
        nch = mp->mnt_ncmountpt;
        while (nch.ncp) {
                if (nch.mount == p->p_fd->fd_nrdir.mount &&
                    nch.ncp == p->p_fd->fd_nrdir.ncp) {
                        return(1);
                }
                if (nch.ncp == nch.mount->mnt_ncmountpt.ncp) {
                        nch = nch.mount->mnt_ncmounton;
                } else {
                        nch.ncp = nch.ncp->nc_parent;
                }
        }

        /*
         * If the mount point is not visible to the process, but the
         * process root is in a subdirectory of the mount, return
         * TRUE anyway.
         */
        if (p->p_fd->fd_nrdir.mount == mp)
                return(1);

        return(0);
}

/* Sets priv to PRIV_ROOT in case no matching fs */
static int
get_fspriv(const char *fsname)
{

        if (strncmp("null", fsname, 5) == 0) {
                return PRIV_VFS_MOUNT_NULLFS;
        } else if (strncmp(fsname, "tmpfs", 6) == 0) {
                return PRIV_VFS_MOUNT_TMPFS;
        }

        return PRIV_ROOT;
}

int
sys___realpath(struct sysmsg *sysmsg, const struct __realpath_args *uap)
{
        struct nlookupdata nd;
        char *rbuf;
        char *fbuf;
        ssize_t rlen;
        int error;

        /*
         * Invalid length if less than 0.  0 is allowed
         */
        if ((ssize_t)uap->len < 0)
                return EINVAL;

        rbuf = NULL;
        fbuf = NULL;
        error = nlookup_init(&nd, uap->path, UIO_USERSPACE, NLC_FOLLOW);
        if (error)
                goto done;

        nd.nl_flags |= NLC_SHAREDLOCK;
        error = nlookup(&nd);
        if (error)
                goto done;

        if (nd.nl_nch.ncp->nc_vp == NULL) {
                error = ENOENT;
                goto done;
        }

        /*
         * Shortcut test for existence.
         */
        if (uap->len == 0) {
                error = ENAMETOOLONG;
                goto done;
        }

        /*
         * Obtain the path relative to the process root.  The nch must not
         * be locked for the cache_fullpath() call.
         */
        if (nd.nl_flags & NLC_NCPISLOCKED) {
                nd.nl_flags &= ~NLC_NCPISLOCKED;
                cache_unlock(&nd.nl_nch);
        }
        error = cache_fullpath(curproc, &nd.nl_nch, NULL, &rbuf, &fbuf, 0);
        if (error)
                goto done;

        rlen = (ssize_t)strlen(rbuf);
        if (rlen >= uap->len) {
                error = ENAMETOOLONG;
                goto done;
        }
        error = copyout(rbuf, uap->buf, rlen + 1);
        if (error == 0)
                sysmsg->sysmsg_szresult = rlen;
done:
        nlookup_done(&nd);
        if (fbuf)
                kfree(fbuf, M_TEMP);

        return error;
}