File:  [DragonFly] / src / sys / vfs / gnu / ext2fs / ext2_vnops.c
Revision 1.10: download - view: text, annotated - select for diffs
Mon Mar 1 06:33:20 2004 UTC (10 years, 1 month ago) by dillon
Branches: MAIN
CVS tags: HEAD
Newtoken commit.  Change the token implementation as follows:  (1) Obtaining
a token no longer enters a critical section.  (2) tokens can be held through
schedular switches and blocking conditions and are effectively released and
reacquired on resume.  Thus tokens serialize access only while the thread
is actually running.  Serialization is not broken by preemptive interrupts.
That is, interrupt threads which preempt do no release the preempted thread's
tokens.  (3) Unlike spl's, tokens will interlock w/ interrupt threads on
the same or on a different cpu.

The vnode interlock code has been rewritten and the API has changed.  The
mountlist vnode scanning code has been consolidated and all known races have
been fixed.  The vnode interlock is now a pool token.

The code that frees unreferenced vnodes whos last VM page has been freed has
been moved out of the low level vm_page_free() code and moved to the
periodic filesystem sycer code in vfs_msycn().

The SMP startup code and the IPI code has been cleaned up considerably.
Certain early token interactions on AP cpus have been moved to the BSP.

The LWKT rwlock API has been cleaned up and turned on.

Major testing by: David Rhodus

/*
 *  modified for EXT2FS support in Lites 1.1
 *
 *  Aug 1995, Godmar Back (gback@cs.utah.edu)
 *  University of Utah, Department of Computer Science
 */
/*
 * Copyright (c) 1982, 1986, 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. All advertising materials mentioning features or use of this software
 *    must display the following acknowledgement:
 *	This product includes software developed by the University of
 *	California, Berkeley and its contributors.
 * 4. 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.
 *
 *	@(#)ufs_vnops.c 8.27 (Berkeley) 5/27/95
 *	@(#)ext2_vnops.c	8.7 (Berkeley) 2/3/94
 * $FreeBSD: src/sys/gnu/ext2fs/ext2_vnops.c,v 1.51.2.2 2003/01/02 17:26:18 bde Exp $
 * $DragonFly: src/sys/vfs/gnu/ext2fs/ext2_vnops.c,v 1.10 2004/03/01 06:33:20 dillon Exp $
 */

#include "opt_quota.h"
#include "opt_suiddir.h"

#include <sys/param.h>
#include <sys/systm.h>
#include <sys/resourcevar.h>
#include <sys/kernel.h>
#include <sys/stat.h>
#include <sys/buf.h>
#include <sys/proc.h>
#include <sys/mount.h>
#include <sys/time.h>
#include <sys/vnode.h>
#include <sys/namei.h>

#include <vm/vm.h>
#include <vm/vm_extern.h>
#include <vm/vm_zone.h>
#include <vm/vnode_pager.h>
#include <sys/buf2.h>

#include <sys/signalvar.h>
#include <vfs/ufs/dir.h>
#include <vfs/ufs/quota.h>
#include <vfs/ufs/inode.h>
#include <vfs/ufs/ufsmount.h>
#include <vfs/ufs/ufs_extern.h>

#include "ext2_fs_sb.h"
#include "fs.h"
#include "ext2_extern.h"
#include "ext2_fs.h"

static int ext2_makeinode (int mode, struct vnode *, struct vnode **, struct componentname *);

static int ext2_fsync (struct vop_fsync_args *);
static int ext2_read (struct vop_read_args *);
static int ext2_write (struct vop_write_args *);
static int ext2_remove (struct vop_remove_args *);
static int ext2_link (struct vop_link_args *);
static int ext2_rename (struct vop_rename_args *);
static int ext2_mkdir (struct vop_mkdir_args *);
static int ext2_rmdir (struct vop_rmdir_args *);
static int ext2_create (struct vop_create_args *);
static int ext2_mknod (struct vop_mknod_args *);
static int ext2_symlink (struct vop_symlink_args *);
static int ext2_getpages (struct vop_getpages_args *);
static int ext2_putpages (struct vop_putpages_args *);

/* Global vfs data structures for ufs. */
vop_t **ext2_vnodeop_p;
static struct vnodeopv_entry_desc ext2_vnodeop_entries[] = {
	{ &vop_default_desc,		(vop_t *) ufs_vnoperate },
	{ &vop_cachedlookup_desc,	(vop_t *) ext2_lookup },
	{ &vop_fsync_desc,		(vop_t *) ext2_fsync },
	{ &vop_inactive_desc,		(vop_t *) ext2_inactive },
	{ &vop_lookup_desc,		(vop_t *) vfs_cache_lookup },
	{ &vop_read_desc,		(vop_t *) ext2_read },
	{ &vop_readdir_desc,		(vop_t *) ext2_readdir },
	{ &vop_reallocblks_desc,	(vop_t *) ext2_reallocblks },
	{ &vop_write_desc,		(vop_t *) ext2_write },
	{ &vop_remove_desc,		(vop_t *) ext2_remove },
	{ &vop_link_desc,		(vop_t *) ext2_link },
	{ &vop_rename_desc,		(vop_t *) ext2_rename },
	{ &vop_mkdir_desc,		(vop_t *) ext2_mkdir },
	{ &vop_rmdir_desc,		(vop_t *) ext2_rmdir },
	{ &vop_create_desc,		(vop_t *) ext2_create },
	{ &vop_mknod_desc,		(vop_t *) ext2_mknod },
	{ &vop_symlink_desc,		(vop_t *) ext2_symlink },
	{ &vop_getpages_desc,		(vop_t *) ext2_getpages },
	{ &vop_putpages_desc,		(vop_t *) ext2_putpages },
	{ NULL, NULL }
};
static struct vnodeopv_desc ext2fs_vnodeop_opv_desc =
	{ &ext2_vnodeop_p, ext2_vnodeop_entries };

vop_t **ext2_specop_p;
static struct vnodeopv_entry_desc ext2_specop_entries[] = {
	{ &vop_default_desc,		(vop_t *) ufs_vnoperatespec },
	{ &vop_fsync_desc,		(vop_t *) ext2_fsync },
	{ &vop_inactive_desc,		(vop_t *) ext2_inactive },
	{ NULL, NULL }
};
static struct vnodeopv_desc ext2fs_specop_opv_desc =
	{ &ext2_specop_p, ext2_specop_entries };

vop_t **ext2_fifoop_p;
static struct vnodeopv_entry_desc ext2_fifoop_entries[] = {
	{ &vop_default_desc,		(vop_t *) ufs_vnoperatefifo },
	{ &vop_fsync_desc,		(vop_t *) ext2_fsync },
	{ &vop_inactive_desc,		(vop_t *) ext2_inactive },
	{ NULL, NULL }
};
static struct vnodeopv_desc ext2fs_fifoop_opv_desc =
	{ &ext2_fifoop_p, ext2_fifoop_entries };

	VNODEOP_SET(ext2fs_vnodeop_opv_desc);
	VNODEOP_SET(ext2fs_specop_opv_desc);
	VNODEOP_SET(ext2fs_fifoop_opv_desc);

#include "ext2_readwrite.c"

/*
 * A virgin directory (no blushing please).
 * Note that the type and namlen fields are reversed relative to ufs.
 * Also, we don't use `struct odirtemplate', since it would just cause
 * endianness problems.
 */
static struct dirtemplate mastertemplate = {
	0, 12, 1, EXT2_FT_DIR, ".",
	0, DIRBLKSIZ - 12, 2, EXT2_FT_DIR, ".."
};
static struct dirtemplate omastertemplate = {
	0, 12, 1, EXT2_FT_UNKNOWN, ".",
	0, DIRBLKSIZ - 12, 2, EXT2_FT_UNKNOWN, ".."
};

/*
 * Create a regular file
 */
static int
ext2_create(ap)
	struct vop_create_args /* {
		struct vnode *a_dvp;
		struct vnode **a_vpp;
		struct componentname *a_cnp;
		struct vattr *a_vap;
	} */ *ap;
{
	int error;

	error =
	    ext2_makeinode(MAKEIMODE(ap->a_vap->va_type, ap->a_vap->va_mode),
	    ap->a_dvp, ap->a_vpp, ap->a_cnp);
	if (error)
		return (error);
	return (0);
}

/*
 * Synch an open file.
 */
/* ARGSUSED */
static int
ext2_fsync(ap)
	struct vop_fsync_args /* {
		struct vnode *a_vp;
		struct ucred *a_cred;
		int a_waitfor;
		struct proc *a_p;
	} */ *ap;
{
	struct vnode *vp = ap->a_vp;
	struct buf *bp;
	struct buf *nbp;
	int s;

	/* 
	 * XXX why is all this fs specific?
	 */

	/*
	 * Flush all dirty buffers associated with a vnode.
	 */
	ext2_discard_prealloc(VTOI(vp));

loop:
	s = splbio();
	for (bp = TAILQ_FIRST(&vp->v_dirtyblkhd); bp; bp = nbp) {
		nbp = TAILQ_NEXT(bp, b_vnbufs);
		if (BUF_LOCK(bp, LK_EXCLUSIVE | LK_NOWAIT))
			continue;
		if ((bp->b_flags & B_DELWRI) == 0)
			panic("ext2_fsync: not dirty");
		bremfree(bp);
		splx(s);
		/*
		 * Wait for I/O associated with indirect blocks to complete,
		 * since there is no way to quickly wait for them below.
		 */
		if (bp->b_vp == vp || ap->a_waitfor == MNT_NOWAIT)
			(void) bawrite(bp);
		else
			(void) bwrite(bp);
		goto loop;
	}
	if (ap->a_waitfor == MNT_WAIT) {
		while (vp->v_numoutput) {
			vp->v_flag |= VBWAIT;
			tsleep(&vp->v_numoutput, 0, "e2fsyn", 0);
		}
#if DIAGNOSTIC
		if (!TAILQ_EMPTY(&vp->v_dirtyblkhd)) {
			vprint("ext2_fsync: dirty", vp);
			goto loop;
		}
#endif
	}
	splx(s);
	return (UFS_UPDATE(ap->a_vp, ap->a_waitfor == MNT_WAIT));
}

/*
 * Mknod vnode call
 */
/* ARGSUSED */
static int
ext2_mknod(ap)
	struct vop_mknod_args /* {
		struct vnode *a_dvp;
		struct vnode **a_vpp;
		struct componentname *a_cnp;
		struct vattr *a_vap;
	} */ *ap;
{
	struct vattr *vap = ap->a_vap;
	struct vnode **vpp = ap->a_vpp;
	struct inode *ip;
	ino_t ino;
	int error;

	error = ext2_makeinode(MAKEIMODE(vap->va_type, vap->va_mode),
	    ap->a_dvp, vpp, ap->a_cnp);
	if (error)
		return (error);
	ip = VTOI(*vpp);
	ip->i_flag |= IN_ACCESS | IN_CHANGE | IN_UPDATE;
	if (vap->va_rdev != VNOVAL) {
		/*
		 * Want to be able to use this to make badblock
		 * inodes, so don't truncate the dev number.
		 */
		ip->i_rdev = vap->va_rdev;
	}
	/*
	 * Remove inode, then reload it through VFS_VGET so it is
	 * checked to see if it is an alias of an existing entry in
	 * the inode cache.
	 */
	vput(*vpp);
	(*vpp)->v_type = VNON;
	ino = ip->i_number;	/* Save this before vgone() invalidates ip. */
	vgone(*vpp);
	error = VFS_VGET(ap->a_dvp->v_mount, ino, vpp);
	if (error) {
		*vpp = NULL;
		return (error);
	}
	return (0);
}

static int
ext2_remove(ap)
	struct vop_remove_args /* {
		struct vnode *a_dvp;
		struct vnode *a_vp;
		struct componentname *a_cnp;
	} */ *ap;
{
	struct inode *ip;
	struct vnode *vp = ap->a_vp;
	struct vnode *dvp = ap->a_dvp;
	int error;

	ip = VTOI(vp);
	if ((ip->i_flags & (NOUNLINK | IMMUTABLE | APPEND)) ||
	    (VTOI(dvp)->i_flags & APPEND)) {
		error = EPERM;
		goto out;
	}
	error = ext2_dirremove(dvp, ap->a_cnp);
	if (error == 0) {
		ip->i_nlink--;
		ip->i_flag |= IN_CHANGE;
	}
out:
	return (error);
}

/*
 * link vnode call
 */
static int
ext2_link(ap)
	struct vop_link_args /* {
		struct vnode *a_tdvp;
		struct vnode *a_vp;
		struct componentname *a_cnp;
	} */ *ap;
{
	struct vnode *vp = ap->a_vp;
	struct vnode *tdvp = ap->a_tdvp;
	struct componentname *cnp = ap->a_cnp;
	struct thread *td = cnp->cn_td;
	struct inode *ip;
	int error;

#ifdef DIAGNOSTIC
	if ((cnp->cn_flags & CNP_HASBUF) == 0)
		panic("ufs_link: no name");
#endif
	if (tdvp->v_mount != vp->v_mount) {
		error = EXDEV;
		goto out2;
	}
	if (tdvp != vp && (error = vn_lock(vp, NULL, LK_EXCLUSIVE, td))) {
		goto out2;
	}
	ip = VTOI(vp);
	if ((nlink_t)ip->i_nlink >= LINK_MAX) {
		error = EMLINK;
		goto out1;
	}
	if (ip->i_flags & (IMMUTABLE | APPEND)) {
		error = EPERM;
		goto out1;
	}
	ip->i_nlink++;
	ip->i_flag |= IN_CHANGE;
	error = UFS_UPDATE(vp, 1);
	if (!error)
		error = ext2_direnter(ip, tdvp, cnp);
	if (error) {
		ip->i_nlink--;
		ip->i_flag |= IN_CHANGE;
	}
out1:
	if (tdvp != vp)
		VOP_UNLOCK(vp, NULL, 0, td);
out2:
	return (error);
}

/*
 * Rename system call.
 *   See comments in sys/ufs/ufs/ufs_vnops.c
 */
static int
ext2_rename(ap)
	struct vop_rename_args  /* {
		struct vnode *a_fdvp;
		struct vnode *a_fvp;
		struct componentname *a_fcnp;
		struct vnode *a_tdvp;
		struct vnode *a_tvp;
		struct componentname *a_tcnp;
	} */ *ap;
{
	struct vnode *tvp = ap->a_tvp;
	struct vnode *tdvp = ap->a_tdvp;
	struct vnode *fvp = ap->a_fvp;
	struct vnode *fdvp = ap->a_fdvp;
	struct componentname *tcnp = ap->a_tcnp;
	struct componentname *fcnp = ap->a_fcnp;
	struct thread *td = fcnp->cn_td;
	struct inode *ip, *xp, *dp;
	struct dirtemplate dirbuf;
	int doingdirectory = 0, oldparent = 0, newparent = 0;
	int error = 0;
	u_char namlen;

#ifdef DIAGNOSTIC
	if ((tcnp->cn_flags & CNP_HASBUF) == 0 ||
	    (fcnp->cn_flags & CNP_HASBUF) == 0)
		panic("ufs_rename: no name");
#endif
	/*
	 * Check for cross-device rename.
	 */
	if ((fvp->v_mount != tdvp->v_mount) ||
	    (tvp && (fvp->v_mount != tvp->v_mount))) {
		error = EXDEV;
abortit:
		if (tdvp == tvp)
			vrele(tdvp);
		else
			vput(tdvp);
		if (tvp)
			vput(tvp);
		vrele(fdvp);
		vrele(fvp);
		return (error);
	}

	if (tvp && ((VTOI(tvp)->i_flags & (NOUNLINK | IMMUTABLE | APPEND)) ||
	    (VTOI(tdvp)->i_flags & APPEND))) {
		error = EPERM;
		goto abortit;
	}

	/*
	 * Renaming a file to itself has no effect.  The upper layers should
	 * not call us in that case.  Temporarily just warn if they do.
	 */
	if (fvp == tvp) {
		printf("ext2_rename: fvp == tvp (can't happen)\n");
		error = 0;
		goto abortit;
	}

	if ((error = vn_lock(fvp, NULL, LK_EXCLUSIVE, td)) != 0)
		goto abortit;
	dp = VTOI(fdvp);
	ip = VTOI(fvp);
 	if (ip->i_nlink >= LINK_MAX) {
 		VOP_UNLOCK(fvp, NULL, 0, td);
 		error = EMLINK;
 		goto abortit;
 	}
	if ((ip->i_flags & (NOUNLINK | IMMUTABLE | APPEND))
	    || (dp->i_flags & APPEND)) {
		VOP_UNLOCK(fvp, NULL, 0, td);
		error = EPERM;
		goto abortit;
	}
	if ((ip->i_mode & IFMT) == IFDIR) {
		/*
		 * Avoid ".", "..", and aliases of "." for obvious reasons.
		 */
		if ((fcnp->cn_namelen == 1 && fcnp->cn_nameptr[0] == '.') ||
		    dp == ip || (fcnp->cn_flags | tcnp->cn_flags) & CNP_ISDOTDOT ||
		    (ip->i_flag & IN_RENAME)) {
			VOP_UNLOCK(fvp, NULL, 0, td);
			error = EINVAL;
			goto abortit;
		}
		ip->i_flag |= IN_RENAME;
		oldparent = dp->i_number;
		doingdirectory++;
	}
	vrele(fdvp);

	/*
	 * When the target exists, both the directory
	 * and target vnodes are returned locked.
	 */
	dp = VTOI(tdvp);
	xp = NULL;
	if (tvp)
		xp = VTOI(tvp);

	/*
	 * 1) Bump link count while we're moving stuff
	 *    around.  If we crash somewhere before
	 *    completing our work, the link count
	 *    may be wrong, but correctable.
	 */
	ip->i_nlink++;
	ip->i_flag |= IN_CHANGE;
	if ((error = UFS_UPDATE(fvp, 1)) != 0) {
		VOP_UNLOCK(fvp, NULL, 0, td);
		goto bad;
	}

	/*
	 * If ".." must be changed (ie the directory gets a new
	 * parent) then the source directory must not be in the
	 * directory heirarchy above the target, as this would
	 * orphan everything below the source directory. Also
	 * the user must have write permission in the source so
	 * as to be able to change "..". We must repeat the call
	 * to namei, as the parent directory is unlocked by the
	 * call to checkpath().
	 */
	error = VOP_ACCESS(fvp, VWRITE, tcnp->cn_cred, tcnp->cn_td);
	VOP_UNLOCK(fvp, NULL, 0, td);
	if (oldparent != dp->i_number)
		newparent = dp->i_number;
	if (doingdirectory && newparent) {
		if (error)	/* write access check above */
			goto bad;
		if (xp != NULL)
			vput(tvp);
		error = ext2_checkpath(ip, dp, tcnp->cn_cred);
		if (error)
			goto out;
		VREF(tdvp);
		error = relookup(tdvp, &tvp, tcnp);
		if (error)
			goto out;
		vrele(tdvp);
		dp = VTOI(tdvp);
		xp = NULL;
		if (tvp)
			xp = VTOI(tvp);
	}
	/*
	 * 2) If target doesn't exist, link the target
	 *    to the source and unlink the source.
	 *    Otherwise, rewrite the target directory
	 *    entry to reference the source inode and
	 *    expunge the original entry's existence.
	 */
	if (xp == NULL) {
		if (dp->i_dev != ip->i_dev)
			panic("ufs_rename: EXDEV");
		/*
		 * Account for ".." in new directory.
		 * When source and destination have the same
		 * parent we don't fool with the link count.
		 */
		if (doingdirectory && newparent) {
			if ((nlink_t)dp->i_nlink >= LINK_MAX) {
				error = EMLINK;
				goto bad;
			}
			dp->i_nlink++;
			dp->i_flag |= IN_CHANGE;
			error = UFS_UPDATE(tdvp, 1);
			if (error)
				goto bad;
		}
		error = ext2_direnter(ip, tdvp, tcnp);
		if (error) {
			if (doingdirectory && newparent) {
				dp->i_nlink--;
				dp->i_flag |= IN_CHANGE;
				(void)UFS_UPDATE(tdvp, 1);
			}
			goto bad;
		}
		vput(tdvp);
	} else {
		if (xp->i_dev != dp->i_dev || xp->i_dev != ip->i_dev)
			panic("ufs_rename: EXDEV");
		/*
		 * Short circuit rename(foo, foo).
		 */
		if (xp->i_number == ip->i_number)
			panic("ufs_rename: same file");
		/*
		 * If the parent directory is "sticky", then the user must
		 * own the parent directory, or the destination of the rename,
		 * otherwise the destination may not be changed (except by
		 * root). This implements append-only directories.
		 */
		if ((dp->i_mode & S_ISTXT) && tcnp->cn_cred->cr_uid != 0 &&
		    tcnp->cn_cred->cr_uid != dp->i_uid &&
		    xp->i_uid != tcnp->cn_cred->cr_uid) {
			error = EPERM;
			goto bad;
		}
		/*
		 * Target must be empty if a directory and have no links
		 * to it. Also, ensure source and target are compatible
		 * (both directories, or both not directories).
		 */
		if ((xp->i_mode&IFMT) == IFDIR) {
			if (! ext2_dirempty(xp, dp->i_number, tcnp->cn_cred) || 
			    xp->i_nlink > 2) {
				error = ENOTEMPTY;
				goto bad;
			}
			if (!doingdirectory) {
				error = ENOTDIR;
				goto bad;
			}
			cache_purge(tdvp);
		} else if (doingdirectory) {
			error = EISDIR;
			goto bad;
		}
		error = ext2_dirrewrite(dp, ip, tcnp);
		if (error)
			goto bad;
		/*
		 * If the target directory is in the same
		 * directory as the source directory,
		 * decrement the link count on the parent
		 * of the target directory.
		 */
		 if (doingdirectory && !newparent) {
			dp->i_nlink--;
			dp->i_flag |= IN_CHANGE;
		}
		vput(tdvp);
		/*
		 * Adjust the link count of the target to
		 * reflect the dirrewrite above.  If this is
		 * a directory it is empty and there are
		 * no links to it, so we can squash the inode and
		 * any space associated with it.  We disallowed
		 * renaming over top of a directory with links to
		 * it above, as the remaining link would point to
		 * a directory without "." or ".." entries.
		 */
		xp->i_nlink--;
		if (doingdirectory) {
			if (--xp->i_nlink != 0)
				panic("ufs_rename: linked directory");
			error = UFS_TRUNCATE(tvp, (off_t)0, IO_SYNC,
			    tcnp->cn_cred, tcnp->cn_td);
		}
		xp->i_flag |= IN_CHANGE;
		vput(tvp);
		xp = NULL;
	}

	/*
	 * 3) Unlink the source.
	 */
	fcnp->cn_flags &= ~CNP_MODMASK;
	fcnp->cn_flags |= CNP_LOCKPARENT | CNP_LOCKLEAF;
	VREF(fdvp);
	error = relookup(fdvp, &fvp, fcnp);
	if (error == 0)
		vrele(fdvp);
	if (fvp != NULL) {
		xp = VTOI(fvp);
		dp = VTOI(fdvp);
	} else {
		/*
		 * From name has disappeared.
		 */
		if (doingdirectory)
			panic("ufs_rename: lost dir entry");
		vrele(ap->a_fvp);
		return (0);
	}
	/*
	 * Ensure that the directory entry still exists and has not
	 * changed while the new name has been entered. If the source is
	 * a file then the entry may have been unlinked or renamed. In
	 * either case there is no further work to be done. If the source
	 * is a directory then it cannot have been rmdir'ed; its link
	 * count of three would cause a rmdir to fail with ENOTEMPTY.
	 * The IN_RENAME flag ensures that it cannot be moved by another
	 * rename.
	 */
	if (xp != ip) {
		if (doingdirectory)
			panic("ufs_rename: lost dir entry");
	} else {
		/*
		 * If the source is a directory with a
		 * new parent, the link count of the old
		 * parent directory must be decremented
		 * and ".." set to point to the new parent.
		 */
		if (doingdirectory && newparent) {
			dp->i_nlink--;
			dp->i_flag |= IN_CHANGE;
			error = vn_rdwr(UIO_READ, fvp, (caddr_t)&dirbuf,
				sizeof (struct dirtemplate), (off_t)0,
				UIO_SYSSPACE, IO_NODELOCKED,
				tcnp->cn_cred, (int *)0, NULL);
			if (error == 0) {
				/* Like ufs little-endian: */
				namlen = dirbuf.dotdot_type;
				if (namlen != 2 ||
				    dirbuf.dotdot_name[0] != '.' ||
				    dirbuf.dotdot_name[1] != '.') {
					ufs_dirbad(xp, (doff_t)12,
					    "rename: mangled dir");
				} else {
					dirbuf.dotdot_ino = newparent;
					(void) vn_rdwr(UIO_WRITE, fvp,
					    (caddr_t)&dirbuf,
					    sizeof (struct dirtemplate),
					    (off_t)0, UIO_SYSSPACE,
					    IO_NODELOCKED|IO_SYNC,
					    tcnp->cn_cred, (int *)0,
					    NULL);
					cache_purge(fdvp);
				}
			}
		}
		error = ext2_dirremove(fdvp, fcnp);
		if (!error) {
			xp->i_nlink--;
			xp->i_flag |= IN_CHANGE;
		}
		xp->i_flag &= ~IN_RENAME;
	}
	if (dp)
		vput(fdvp);
	if (xp)
		vput(fvp);
	vrele(ap->a_fvp);
	return (error);

bad:
	if (xp)
		vput(ITOV(xp));
	vput(ITOV(dp));
out:
	if (doingdirectory)
		ip->i_flag &= ~IN_RENAME;
	if (vn_lock(fvp, NULL, LK_EXCLUSIVE, td) == 0) {
		ip->i_nlink--;
		ip->i_flag |= IN_CHANGE;
		ip->i_flag &= ~IN_RENAME;
		vput(fvp);
	} else
		vrele(fvp);
	return (error);
}

/*
 * Mkdir system call
 */
static int
ext2_mkdir(ap)
	struct vop_mkdir_args /* {
		struct vnode *a_dvp;
		struct vnode **a_vpp;
		struct componentname *a_cnp;
		struct vattr *a_vap;
	} */ *ap;
{
	struct vnode *dvp = ap->a_dvp;
	struct vattr *vap = ap->a_vap;
	struct componentname *cnp = ap->a_cnp;
	struct inode *ip, *dp;
	struct vnode *tvp;
	struct dirtemplate dirtemplate, *dtp;
	int error, dmode;

#ifdef DIAGNOSTIC
	if ((cnp->cn_flags & CNP_HASBUF) == 0)
		panic("ufs_mkdir: no name");
#endif
	dp = VTOI(dvp);
	if ((nlink_t)dp->i_nlink >= LINK_MAX) {
		error = EMLINK;
		goto out;
	}
	dmode = vap->va_mode & 0777;
	dmode |= IFDIR;
	/*
	 * Must simulate part of ext2_makeinode here to acquire the inode,
	 * but not have it entered in the parent directory. The entry is
	 * made later after writing "." and ".." entries.
	 */
	error = UFS_VALLOC(dvp, dmode, cnp->cn_cred, &tvp);
	if (error)
		goto out;
	ip = VTOI(tvp);
	ip->i_gid = dp->i_gid;
#ifdef SUIDDIR
	{
#ifdef QUOTA
		struct ucred ucred, *ucp;
		ucp = cnp->cn_cred;
#endif			I
		/*
		 * if we are hacking owners here, (only do this where told to)
		 * and we are not giving it TOO root, (would subvert quotas)
		 * then go ahead and give it to the other user.
		 * The new directory also inherits the SUID bit. 
		 * If user's UID and dir UID are the same,
		 * 'give it away' so that the SUID is still forced on.
		 */
		if ( (dvp->v_mount->mnt_flag & MNT_SUIDDIR) &&
		   (dp->i_mode & ISUID) && dp->i_uid) {
			dmode |= ISUID;
			ip->i_uid = dp->i_uid;
#ifdef QUOTA
			if (dp->i_uid != cnp->cn_cred->cr_uid) {
				/*
				 * make sure the correct user gets charged
				 * for the space.
				 * Make a dummy credential for the victim.
				 * XXX This seems to never be accessed out of
				 * our context so a stack variable is ok.
				 */
				ucred.cr_ref = 1;
				ucred.cr_uid = ip->i_uid;
				ucred.cr_ngroups = 1;
				ucred.cr_groups[0] = dp->i_gid;
				ucp = &ucred;
			}
#endif			I
		} else {
			ip->i_uid = cnp->cn_cred->cr_uid;
		}
#ifdef QUOTA
		if ((error = getinoquota(ip)) ||
	    	(error = chkiq(ip, 1, ucp, 0))) {
			UFS_VFREE(tvp, ip->i_number, dmode);
			vput(tvp);
			return (error);
		}
#endif
	}
#else
	ip->i_uid = cnp->cn_cred->cr_uid;
#ifdef QUOTA
	if ((error = getinoquota(ip)) ||
	    (error = chkiq(ip, 1, cnp->cn_cred, 0))) {
		UFS_VFREE(tvp, ip->i_number, dmode);
		vput(tvp);
		return (error);
	}
#endif
#endif
	ip->i_flag |= IN_ACCESS | IN_CHANGE | IN_UPDATE;
	ip->i_mode = dmode;
	tvp->v_type = VDIR;	/* Rest init'd in getnewvnode(). */
	ip->i_nlink = 2;
	if (cnp->cn_flags & CNP_ISWHITEOUT)
		ip->i_flags |= UF_OPAQUE;
	error = UFS_UPDATE(tvp, 1);

	/*
	 * Bump link count in parent directory
	 * to reflect work done below.  Should
	 * be done before reference is created
	 * so reparation is possible if we crash.
	 */
	dp->i_nlink++;
	dp->i_flag |= IN_CHANGE;
	error = UFS_UPDATE(dvp, 1);
	if (error)
		goto bad;

	/* Initialize directory with "." and ".." from static template. */
	if (EXT2_HAS_INCOMPAT_FEATURE(ip->i_e2fs->s_es,
	    EXT2_FEATURE_INCOMPAT_FILETYPE))
		dtp = &mastertemplate;
	else
		dtp = &omastertemplate;
	dirtemplate = *dtp;
	dirtemplate.dot_ino = ip->i_number;
	dirtemplate.dotdot_ino = dp->i_number;
	/* note that in ext2 DIRBLKSIZ == blocksize, not DEV_BSIZE 
	 * so let's just redefine it - for this function only
	 */
#undef  DIRBLKSIZ 
#define DIRBLKSIZ  VTOI(dvp)->i_e2fs->s_blocksize
	dirtemplate.dotdot_reclen = DIRBLKSIZ - 12;
	error = vn_rdwr(UIO_WRITE, tvp, (caddr_t)&dirtemplate,
	    sizeof (dirtemplate), (off_t)0, UIO_SYSSPACE,
	    IO_NODELOCKED|IO_SYNC, cnp->cn_cred, (int *)0, NULL);
	if (error) {
		dp->i_nlink--;
		dp->i_flag |= IN_CHANGE;
		goto bad;
	}
	if (DIRBLKSIZ > VFSTOUFS(dvp->v_mount)->um_mountp->mnt_stat.f_bsize)
		panic("ufs_mkdir: blksize"); /* XXX should grow with balloc() */
	else {
		ip->i_size = DIRBLKSIZ;
		ip->i_flag |= IN_CHANGE;
	}

	/* Directory set up, now install its entry in the parent directory. */
	error = ext2_direnter(ip, dvp, cnp);
	if (error) {
		dp->i_nlink--;
		dp->i_flag |= IN_CHANGE;
	}
bad:
	/*
	 * No need to do an explicit VOP_TRUNCATE here, vrele will do this
	 * for us because we set the link count to 0.
	 */
	if (error) {
		ip->i_nlink = 0;
		ip->i_flag |= IN_CHANGE;
		vput(tvp);
	} else
		*ap->a_vpp = tvp;
out:
	return (error);
#undef  DIRBLKSIZ
#define DIRBLKSIZ  DEV_BSIZE
}

/*
 * Rmdir system call.
 */
static int
ext2_rmdir(ap)
	struct vop_rmdir_args /* {
		struct vnode *a_dvp;
		struct vnode *a_vp;
		struct componentname *a_cnp;
	} */ *ap;
{
	struct vnode *vp = ap->a_vp;
	struct vnode *dvp = ap->a_dvp;
	struct componentname *cnp = ap->a_cnp;
	struct thread *td = cnp->cn_td;
	struct inode *ip, *dp;
	int error;

	ip = VTOI(vp);
	dp = VTOI(dvp);

	/*
	 * Verify the directory is empty (and valid).
	 * (Rmdir ".." won't be valid since
	 *  ".." will contain a reference to
	 *  the current directory and thus be
	 *  non-empty.)
	 */
	error = 0;
	if (ip->i_nlink != 2 || !ext2_dirempty(ip, dp->i_number, cnp->cn_cred)) {
		error = ENOTEMPTY;
		goto out;
	}
	if ((dp->i_flags & APPEND)
	    || (ip->i_flags & (NOUNLINK | IMMUTABLE | APPEND))) {
		error = EPERM;
		goto out;
	}
	/*
	 * Delete reference to directory before purging
	 * inode.  If we crash in between, the directory
	 * will be reattached to lost+found,
	 */
	error = ext2_dirremove(dvp, cnp);
	if (error)
		goto out;
	dp->i_nlink--;
	dp->i_flag |= IN_CHANGE;
	cache_purge(dvp);
	VOP_UNLOCK(dvp, NULL, 0, td);
	/*
	 * Truncate inode.  The only stuff left
	 * in the directory is "." and "..".  The
	 * "." reference is inconsequential since
	 * we're quashing it.  The ".." reference
	 * has already been adjusted above.  We've
	 * removed the "." reference and the reference
	 * in the parent directory, but there may be
	 * other hard links so decrement by 2 and
	 * worry about them later.
	 */
	ip->i_nlink -= 2;
	error = UFS_TRUNCATE(vp, (off_t)0, IO_SYNC, cnp->cn_cred, td);
	cache_purge(ITOV(ip));
	vn_lock(dvp, NULL, LK_EXCLUSIVE | LK_RETRY, td);
out:
	return (error);
}

/*
 * symlink -- make a symbolic link
 */
static int
ext2_symlink(ap)
	struct vop_symlink_args /* {
		struct vnode *a_dvp;
		struct vnode **a_vpp;
		struct componentname *a_cnp;
		struct vattr *a_vap;
		char *a_target;
	} */ *ap;
{
	struct vnode *vp, **vpp = ap->a_vpp;
	struct inode *ip;
	int len, error;

	error = ext2_makeinode(IFLNK | ap->a_vap->va_mode, ap->a_dvp,
	    vpp, ap->a_cnp);
	if (error)
		return (error);
	vp = *vpp;
	len = strlen(ap->a_target);
	if (len < vp->v_mount->mnt_maxsymlinklen) {
		ip = VTOI(vp);
		bcopy(ap->a_target, (char *)ip->i_shortlink, len);
		ip->i_size = len;
		ip->i_flag |= IN_CHANGE | IN_UPDATE;
	} else
		error = vn_rdwr(UIO_WRITE, vp, ap->a_target, len, (off_t)0,
		    UIO_SYSSPACE, IO_NODELOCKED, ap->a_cnp->cn_cred, (int *)0,
		    NULL);
	if (error)
		vput(vp);
	return (error);
}

/*
 * Allocate a new inode.
 */
static int
ext2_makeinode(mode, dvp, vpp, cnp)
	int mode;
	struct vnode *dvp;
	struct vnode **vpp;
	struct componentname *cnp;
{
	struct inode *ip, *pdir;
	struct vnode *tvp;
	int error;

	pdir = VTOI(dvp);
#ifdef DIAGNOSTIC
	if ((cnp->cn_flags & CNP_HASBUF) == 0)
		panic("ext2_makeinode: no name");
#endif
	*vpp = NULL;
	if ((mode & IFMT) == 0)
		mode |= IFREG;

	error = UFS_VALLOC(dvp, mode, cnp->cn_cred, &tvp);
	if (error) {
		return (error);
	}
	ip = VTOI(tvp);
	ip->i_gid = pdir->i_gid;
#ifdef SUIDDIR
	{
#ifdef QUOTA
		struct ucred ucred, *ucp;
		ucp = cnp->cn_cred;
#endif			I
		/*
		 * if we are
		 * not the owner of the directory,
		 * and we are hacking owners here, (only do this where told to)
		 * and we are not giving it TOO root, (would subvert quotas)
		 * then go ahead and give it to the other user.
		 * Note that this drops off the execute bits for security.
		 */
		if ( (dvp->v_mount->mnt_flag & MNT_SUIDDIR) &&
		     (pdir->i_mode & ISUID) &&
		     (pdir->i_uid != cnp->cn_cred->cr_uid) && pdir->i_uid) {
			ip->i_uid = pdir->i_uid;
			mode &= ~07111;
#ifdef QUOTA
			/*
			 * make sure the correct user gets charged
			 * for the space.
			 * Quickly knock up a dummy credential for the victim.
			 * XXX This seems to never be accessed out of our
			 * context so a stack variable is ok.
			 */
			ucred.cr_ref = 1;
			ucred.cr_uid = ip->i_uid;
			ucred.cr_ngroups = 1;
			ucred.cr_groups[0] = pdir->i_gid;
			ucp = &ucred;
#endif			I
		} else {
			ip->i_uid = cnp->cn_cred->cr_uid;
		}
	
#ifdef QUOTA
		if ((error = getinoquota(ip)) ||
	    	(error = chkiq(ip, 1, ucp, 0))) {
			UFS_VFREE(tvp, ip->i_number, mode);
			vput(tvp);
			return (error);
		}
#endif
	}
#else
	ip->i_uid = cnp->cn_cred->cr_uid;
#ifdef QUOTA
	if ((error = getinoquota(ip)) ||
	    (error = chkiq(ip, 1, cnp->cn_cred, 0))) {
		UFS_VFREE(tvp, ip->i_number, mode);
		vput(tvp);
		return (error);
	}
#endif
#endif
	ip->i_flag |= IN_ACCESS | IN_CHANGE | IN_UPDATE;
	ip->i_mode = mode;
	tvp->v_type = IFTOVT(mode);	/* Rest init'd in getnewvnode(). */
	ip->i_nlink = 1;
	if ((ip->i_mode & ISGID) && !groupmember(ip->i_gid, cnp->cn_cred) &&
	    suser_cred(cnp->cn_cred, PRISON_ROOT))
		ip->i_mode &= ~ISGID;

	if (cnp->cn_flags & CNP_ISWHITEOUT)
		ip->i_flags |= UF_OPAQUE;

	/*
	 * Make sure inode goes to disk before directory entry.
	 */
	error = UFS_UPDATE(tvp, 1);
	if (error)
		goto bad;
	error = ext2_direnter(ip, dvp, cnp);
	if (error)
		goto bad;

	*vpp = tvp;
	return (0);

bad:
	/*
	 * Write error occurred trying to update the inode
	 * or the directory so must deallocate the inode.
	 */
	ip->i_nlink = 0;
	ip->i_flag |= IN_CHANGE;
	vput(tvp);
	return (error);
}

/*
 * get page routine
 *
 * XXX By default, wimp out... note that a_offset is ignored (and always
 * XXX has been).
 */
static int
ext2_getpages(ap)
	struct vop_getpages_args *ap;
{
	return (vnode_pager_generic_getpages(ap->a_vp, ap->a_m, ap->a_count,
		ap->a_reqpage));
}

/*
 * put page routine
 *
 * XXX By default, wimp out... note that a_offset is ignored (and always
 * XXX has been).
 */
static int
ext2_putpages(ap)
	struct vop_putpages_args *ap;
{
	return (vnode_pager_generic_putpages(ap->a_vp, ap->a_m, ap->a_count,
		ap->a_sync, ap->a_rtvals));
}