Home | History | Annotate | Line # | Download | only in tmpfs
tmpfs_vnops.c revision 1.66.4.2
      1 /*	$NetBSD: tmpfs_vnops.c,v 1.66.4.2 2010/05/30 05:17:56 rmind Exp $	*/
      2 
      3 /*
      4  * Copyright (c) 2005, 2006, 2007 The NetBSD Foundation, Inc.
      5  * All rights reserved.
      6  *
      7  * This code is derived from software contributed to The NetBSD Foundation
      8  * by Julio M. Merino Vidal, developed as part of Google's Summer of Code
      9  * 2005 program.
     10  *
     11  * Redistribution and use in source and binary forms, with or without
     12  * modification, are permitted provided that the following conditions
     13  * are met:
     14  * 1. Redistributions of source code must retain the above copyright
     15  *    notice, this list of conditions and the following disclaimer.
     16  * 2. Redistributions in binary form must reproduce the above copyright
     17  *    notice, this list of conditions and the following disclaimer in the
     18  *    documentation and/or other materials provided with the distribution.
     19  *
     20  * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
     21  * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
     22  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
     23  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
     24  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
     25  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
     26  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
     27  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
     28  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
     29  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
     30  * POSSIBILITY OF SUCH DAMAGE.
     31  */
     32 
     33 /*
     34  * tmpfs vnode interface.
     35  */
     36 
     37 #include <sys/cdefs.h>
     38 __KERNEL_RCSID(0, "$NetBSD: tmpfs_vnops.c,v 1.66.4.2 2010/05/30 05:17:56 rmind Exp $");
     39 
     40 #include <sys/param.h>
     41 #include <sys/dirent.h>
     42 #include <sys/fcntl.h>
     43 #include <sys/event.h>
     44 #include <sys/malloc.h>
     45 #include <sys/namei.h>
     46 #include <sys/proc.h>
     47 #include <sys/stat.h>
     48 #include <sys/uio.h>
     49 #include <sys/unistd.h>
     50 #include <sys/vnode.h>
     51 #include <sys/lockf.h>
     52 #include <sys/kauth.h>
     53 
     54 #include <uvm/uvm.h>
     55 
     56 #include <miscfs/fifofs/fifo.h>
     57 #include <miscfs/genfs/genfs.h>
     58 #include <fs/tmpfs/tmpfs_vnops.h>
     59 #include <fs/tmpfs/tmpfs.h>
     60 
     61 /* --------------------------------------------------------------------- */
     62 
     63 /*
     64  * vnode operations vector used for files stored in a tmpfs file system.
     65  */
     66 int (**tmpfs_vnodeop_p)(void *);
     67 const struct vnodeopv_entry_desc tmpfs_vnodeop_entries[] = {
     68 	{ &vop_default_desc,		vn_default_error },
     69 	{ &vop_lookup_desc,		tmpfs_lookup },
     70 	{ &vop_create_desc,		tmpfs_create },
     71 	{ &vop_mknod_desc,		tmpfs_mknod },
     72 	{ &vop_open_desc,		tmpfs_open },
     73 	{ &vop_close_desc,		tmpfs_close },
     74 	{ &vop_access_desc,		tmpfs_access },
     75 	{ &vop_getattr_desc,		tmpfs_getattr },
     76 	{ &vop_setattr_desc,		tmpfs_setattr },
     77 	{ &vop_read_desc,		tmpfs_read },
     78 	{ &vop_write_desc,		tmpfs_write },
     79 	{ &vop_ioctl_desc,		tmpfs_ioctl },
     80 	{ &vop_fcntl_desc,		tmpfs_fcntl },
     81 	{ &vop_poll_desc,		tmpfs_poll },
     82 	{ &vop_kqfilter_desc,		tmpfs_kqfilter },
     83 	{ &vop_revoke_desc,		tmpfs_revoke },
     84 	{ &vop_mmap_desc,		tmpfs_mmap },
     85 	{ &vop_fsync_desc,		tmpfs_fsync },
     86 	{ &vop_seek_desc,		tmpfs_seek },
     87 	{ &vop_remove_desc,		tmpfs_remove },
     88 	{ &vop_link_desc,		tmpfs_link },
     89 	{ &vop_rename_desc,		tmpfs_rename },
     90 	{ &vop_mkdir_desc,		tmpfs_mkdir },
     91 	{ &vop_rmdir_desc,		tmpfs_rmdir },
     92 	{ &vop_symlink_desc,		tmpfs_symlink },
     93 	{ &vop_readdir_desc,		tmpfs_readdir },
     94 	{ &vop_readlink_desc,		tmpfs_readlink },
     95 	{ &vop_abortop_desc,		tmpfs_abortop },
     96 	{ &vop_inactive_desc,		tmpfs_inactive },
     97 	{ &vop_reclaim_desc,		tmpfs_reclaim },
     98 	{ &vop_lock_desc,		tmpfs_lock },
     99 	{ &vop_unlock_desc,		tmpfs_unlock },
    100 	{ &vop_bmap_desc,		tmpfs_bmap },
    101 	{ &vop_strategy_desc,		tmpfs_strategy },
    102 	{ &vop_print_desc,		tmpfs_print },
    103 	{ &vop_pathconf_desc,		tmpfs_pathconf },
    104 	{ &vop_islocked_desc,		tmpfs_islocked },
    105 	{ &vop_advlock_desc,		tmpfs_advlock },
    106 	{ &vop_bwrite_desc,		tmpfs_bwrite },
    107 	{ &vop_getpages_desc,		tmpfs_getpages },
    108 	{ &vop_putpages_desc,		tmpfs_putpages },
    109 	{ NULL, NULL }
    110 };
    111 const struct vnodeopv_desc tmpfs_vnodeop_opv_desc =
    112 	{ &tmpfs_vnodeop_p, tmpfs_vnodeop_entries };
    113 
    114 /* --------------------------------------------------------------------- */
    115 
    116 int
    117 tmpfs_lookup(void *v)
    118 {
    119 	struct vnode *dvp = ((struct vop_lookup_args *)v)->a_dvp;
    120 	struct vnode **vpp = ((struct vop_lookup_args *)v)->a_vpp;
    121 	struct componentname *cnp = ((struct vop_lookup_args *)v)->a_cnp;
    122 
    123 	int error;
    124 	struct tmpfs_dirent *de;
    125 	struct tmpfs_node *dnode;
    126 
    127 	KASSERT(VOP_ISLOCKED(dvp));
    128 
    129 	dnode = VP_TO_TMPFS_DIR(dvp);
    130 	*vpp = NULL;
    131 
    132 	/* Check accessibility of requested node as a first step. */
    133 	error = VOP_ACCESS(dvp, VEXEC, cnp->cn_cred);
    134 	if (error != 0)
    135 		goto out;
    136 
    137 	/* If requesting the last path component on a read-only file system
    138 	 * with a write operation, deny it. */
    139 	if ((cnp->cn_flags & ISLASTCN) &&
    140 	    (dvp->v_mount->mnt_flag & MNT_RDONLY) &&
    141 	    (cnp->cn_nameiop == DELETE || cnp->cn_nameiop == RENAME)) {
    142 		error = EROFS;
    143 		goto out;
    144 	}
    145 
    146 	/* Avoid doing a linear scan of the directory if the requested
    147 	 * directory/name couple is already in the cache. */
    148 	error = cache_lookup(dvp, vpp, cnp);
    149 	if (error >= 0)
    150 		goto out;
    151 
    152 	/* We cannot be requesting the parent directory of the root node. */
    153 	KASSERT(IMPLIES(dnode->tn_type == VDIR &&
    154 	    dnode->tn_spec.tn_dir.tn_parent == dnode,
    155 	    !(cnp->cn_flags & ISDOTDOT)));
    156 
    157 	if (cnp->cn_flags & ISDOTDOT) {
    158 		VOP_UNLOCK(dvp, 0);
    159 
    160 		/* Allocate a new vnode on the matching entry. */
    161 		error = tmpfs_alloc_vp(dvp->v_mount,
    162 		    dnode->tn_spec.tn_dir.tn_parent, vpp);
    163 
    164 		vn_lock(dvp, LK_EXCLUSIVE | LK_RETRY);
    165 	} else if (cnp->cn_namelen == 1 && cnp->cn_nameptr[0] == '.') {
    166 		vref(dvp);
    167 		*vpp = dvp;
    168 		error = 0;
    169 	} else {
    170 		de = tmpfs_dir_lookup(dnode, cnp);
    171 		if (de == NULL) {
    172 			/* The entry was not found in the directory.
    173 			 * This is OK iff we are creating or renaming an
    174 			 * entry and are working on the last component of
    175 			 * the path name. */
    176 			if ((cnp->cn_flags & ISLASTCN) &&
    177 			    (cnp->cn_nameiop == CREATE || \
    178 			    cnp->cn_nameiop == RENAME)) {
    179 				error = VOP_ACCESS(dvp, VWRITE, cnp->cn_cred);
    180 				if (error != 0)
    181 					goto out;
    182 
    183 				/* Keep the component name in the buffer for
    184 				 * future uses. */
    185 				cnp->cn_flags |= SAVENAME;
    186 
    187 				error = EJUSTRETURN;
    188 			} else
    189 				error = ENOENT;
    190 		} else {
    191 			struct tmpfs_node *tnode;
    192 
    193 			/* The entry was found, so get its associated
    194 			 * tmpfs_node. */
    195 			tnode = de->td_node;
    196 
    197 			/* If we are not at the last path component and
    198 			 * found a non-directory or non-link entry (which
    199 			 * may itself be pointing to a directory), raise
    200 			 * an error. */
    201 			if ((tnode->tn_type != VDIR &&
    202 			    tnode->tn_type != VLNK) &&
    203 			    !(cnp->cn_flags & ISLASTCN)) {
    204 				error = ENOTDIR;
    205 				goto out;
    206 			}
    207 
    208 			/* Check permissions */
    209 			if ((cnp->cn_flags & ISLASTCN) &&
    210 			    (cnp->cn_nameiop == DELETE ||
    211 			    cnp->cn_nameiop == RENAME)) {
    212 				kauth_action_t action = 0;
    213 
    214 				/* This is the file-system's decision. */
    215 				if ((dnode->tn_mode & S_ISTXT) != 0 &&
    216 				    kauth_cred_geteuid(cnp->cn_cred) != dnode->tn_uid &&
    217 				    kauth_cred_geteuid(cnp->cn_cred) != tnode->tn_uid)
    218 					error = EPERM;
    219 				else
    220 					error = 0;
    221 
    222 				/* Only bother if we're not already failing it. */
    223 				if (!error) {
    224 					error = VOP_ACCESS(dvp, VWRITE, cnp->cn_cred);
    225 				}
    226 
    227 				if (cnp->cn_nameiop == DELETE)
    228 					action |= KAUTH_VNODE_DELETE;
    229 				else /* if (cnp->cn_nameiop == RENAME) */
    230 					action |= KAUTH_VNODE_RENAME;
    231 
    232 				error = kauth_authorize_vnode(cnp->cn_cred,
    233 				    action, *vpp, dvp, error);
    234 				if (error != 0)
    235 					goto out;
    236 
    237 				cnp->cn_flags |= SAVENAME;
    238 			} else
    239 				de = NULL;
    240 
    241 			/* Allocate a new vnode on the matching entry. */
    242 			error = tmpfs_alloc_vp(dvp->v_mount, tnode, vpp);
    243 		}
    244 	}
    245 
    246 	/* Store the result of this lookup in the cache.  Avoid this if the
    247 	 * request was for creation, as it does not improve timings on
    248 	 * emprical tests. */
    249 	if ((cnp->cn_flags & MAKEENTRY) && cnp->cn_nameiop != CREATE &&
    250 	    (cnp->cn_flags & ISDOTDOT) == 0)
    251 		cache_enter(dvp, *vpp, cnp);
    252 
    253 out:
    254 	/* If there were no errors, *vpp cannot be null and it must be
    255 	 * locked. */
    256 	KASSERT(IFF(error == 0, *vpp != NULL && VOP_ISLOCKED(*vpp)));
    257 
    258 	/* dvp must always be locked. */
    259 	KASSERT(VOP_ISLOCKED(dvp));
    260 
    261 	return error;
    262 }
    263 
    264 /* --------------------------------------------------------------------- */
    265 
    266 int
    267 tmpfs_create(void *v)
    268 {
    269 	struct vnode *dvp = ((struct vop_create_args *)v)->a_dvp;
    270 	struct vnode **vpp = ((struct vop_create_args *)v)->a_vpp;
    271 	struct componentname *cnp = ((struct vop_create_args *)v)->a_cnp;
    272 	struct vattr *vap = ((struct vop_create_args *)v)->a_vap;
    273 
    274 	KASSERT(vap->va_type == VREG || vap->va_type == VSOCK);
    275 
    276 	return tmpfs_alloc_file(dvp, vpp, vap, cnp, NULL);
    277 }
    278 /* --------------------------------------------------------------------- */
    279 
    280 int
    281 tmpfs_mknod(void *v)
    282 {
    283 	struct vnode *dvp = ((struct vop_mknod_args *)v)->a_dvp;
    284 	struct vnode **vpp = ((struct vop_mknod_args *)v)->a_vpp;
    285 	struct componentname *cnp = ((struct vop_mknod_args *)v)->a_cnp;
    286 	struct vattr *vap = ((struct vop_mknod_args *)v)->a_vap;
    287 
    288 	if (vap->va_type != VBLK && vap->va_type != VCHR &&
    289 	    vap->va_type != VFIFO) {
    290 		vput(dvp);
    291 		return EINVAL;
    292 	}
    293 
    294 	return tmpfs_alloc_file(dvp, vpp, vap, cnp, NULL);
    295 }
    296 
    297 /* --------------------------------------------------------------------- */
    298 
    299 int
    300 tmpfs_open(void *v)
    301 {
    302 	struct vnode *vp = ((struct vop_open_args *)v)->a_vp;
    303 	int mode = ((struct vop_open_args *)v)->a_mode;
    304 
    305 	int error;
    306 	struct tmpfs_node *node;
    307 
    308 	KASSERT(VOP_ISLOCKED(vp));
    309 
    310 	node = VP_TO_TMPFS_NODE(vp);
    311 
    312 	/* The file is still active but all its names have been removed
    313 	 * (e.g. by a "rmdir $(pwd)").  It cannot be opened any more as
    314 	 * it is about to die. */
    315 	if (node->tn_links < 1) {
    316 		error = ENOENT;
    317 		goto out;
    318 	}
    319 
    320 	/* If the file is marked append-only, deny write requests. */
    321 	if (node->tn_flags & APPEND && (mode & (FWRITE | O_APPEND)) == FWRITE)
    322 		error = EPERM;
    323 	else
    324 		error = 0;
    325 
    326 out:
    327 	KASSERT(VOP_ISLOCKED(vp));
    328 
    329 	return error;
    330 }
    331 
    332 /* --------------------------------------------------------------------- */
    333 
    334 int
    335 tmpfs_close(void *v)
    336 {
    337 	struct vnode *vp = ((struct vop_close_args *)v)->a_vp;
    338 
    339 	struct tmpfs_node *node;
    340 
    341 	KASSERT(VOP_ISLOCKED(vp));
    342 
    343 	node = VP_TO_TMPFS_NODE(vp);
    344 
    345 	if (node->tn_links > 0) {
    346 		/* Update node times.  No need to do it if the node has
    347 		 * been deleted, because it will vanish after we return. */
    348 		tmpfs_update(vp, NULL, NULL, NULL, UPDATE_CLOSE);
    349 	}
    350 
    351 	return 0;
    352 }
    353 
    354 /* --------------------------------------------------------------------- */
    355 
    356 static int
    357 tmpfs_check_possible(struct vnode *vp, struct tmpfs_node *node, mode_t mode)
    358 {
    359 	int error = 0;
    360 
    361 	switch (vp->v_type) {
    362 	case VDIR:
    363 		/* FALLTHROUGH */
    364 	case VLNK:
    365 		/* FALLTHROUGH */
    366 	case VREG:
    367 		if (mode & VWRITE && vp->v_mount->mnt_flag & MNT_RDONLY) {
    368 			error = EROFS;
    369 			goto out;
    370 		}
    371 		break;
    372 
    373 	case VBLK:
    374 		/* FALLTHROUGH */
    375 	case VCHR:
    376 		/* FALLTHROUGH */
    377 	case VSOCK:
    378 		/* FALLTHROUGH */
    379 	case VFIFO:
    380 		break;
    381 
    382 	default:
    383 		error = EINVAL;
    384 		goto out;
    385 	}
    386 
    387 	if (mode & VWRITE && node->tn_flags & IMMUTABLE) {
    388 		error = EPERM;
    389 		goto out;
    390 	}
    391 
    392  out:
    393 	return error;
    394 }
    395 
    396 static int
    397 tmpfs_check_permitted(struct vnode *vp, struct tmpfs_node *node, mode_t mode,
    398     kauth_cred_t cred)
    399 {
    400 
    401 	return genfs_can_access(vp->v_type, node->tn_mode, node->tn_uid,
    402 	    node->tn_gid, mode, cred);
    403 }
    404 
    405 int
    406 tmpfs_access(void *v)
    407 {
    408 	struct vnode *vp = ((struct vop_access_args *)v)->a_vp;
    409 	int mode = ((struct vop_access_args *)v)->a_mode;
    410 	kauth_cred_t cred = ((struct vop_access_args *)v)->a_cred;
    411 
    412 	int error;
    413 	struct tmpfs_node *node;
    414 
    415 	KASSERT(VOP_ISLOCKED(vp));
    416 
    417 	node = VP_TO_TMPFS_NODE(vp);
    418 
    419 	error = tmpfs_check_possible(vp, node, mode);
    420 	if (error)
    421 		goto out;
    422 
    423 	error = tmpfs_check_permitted(vp, node, mode, cred);
    424 
    425 	error = kauth_authorize_vnode(cred, kauth_mode_to_action(mode), vp,
    426 	    NULL, error);
    427 
    428 out:
    429 	KASSERT(VOP_ISLOCKED(vp));
    430 
    431 	return error;
    432 }
    433 
    434 /* --------------------------------------------------------------------- */
    435 
    436 int
    437 tmpfs_getattr(void *v)
    438 {
    439 	struct vnode *vp = ((struct vop_getattr_args *)v)->a_vp;
    440 	struct vattr *vap = ((struct vop_getattr_args *)v)->a_vap;
    441 
    442 	struct tmpfs_node *node;
    443 
    444 	node = VP_TO_TMPFS_NODE(vp);
    445 
    446 	vattr_null(vap);
    447 
    448 	tmpfs_itimes(vp, NULL, NULL, NULL);
    449 
    450 	vap->va_type = vp->v_type;
    451 	vap->va_mode = node->tn_mode;
    452 	vap->va_nlink = node->tn_links;
    453 	vap->va_uid = node->tn_uid;
    454 	vap->va_gid = node->tn_gid;
    455 	vap->va_fsid = vp->v_mount->mnt_stat.f_fsidx.__fsid_val[0];
    456 	vap->va_fileid = node->tn_id;
    457 	vap->va_size = node->tn_size;
    458 	vap->va_blocksize = PAGE_SIZE;
    459 	vap->va_atime = node->tn_atime;
    460 	vap->va_mtime = node->tn_mtime;
    461 	vap->va_ctime = node->tn_ctime;
    462 	vap->va_birthtime = node->tn_birthtime;
    463 	vap->va_gen = node->tn_gen;
    464 	vap->va_flags = node->tn_flags;
    465 	vap->va_rdev = (vp->v_type == VBLK || vp->v_type == VCHR) ?
    466 		node->tn_spec.tn_dev.tn_rdev : VNOVAL;
    467 	vap->va_bytes = round_page(node->tn_size);
    468 	vap->va_filerev = VNOVAL;
    469 	vap->va_vaflags = 0;
    470 	vap->va_spare = VNOVAL; /* XXX */
    471 
    472 	return 0;
    473 }
    474 
    475 /* --------------------------------------------------------------------- */
    476 
    477 #define GOODTIME(tv)	((tv)->tv_sec != VNOVAL || (tv)->tv_nsec != VNOVAL)
    478 /* XXX Should this operation be atomic?  I think it should, but code in
    479  * XXX other places (e.g., ufs) doesn't seem to be... */
    480 int
    481 tmpfs_setattr(void *v)
    482 {
    483 	struct vnode *vp = ((struct vop_setattr_args *)v)->a_vp;
    484 	struct vattr *vap = ((struct vop_setattr_args *)v)->a_vap;
    485 	kauth_cred_t cred = ((struct vop_setattr_args *)v)->a_cred;
    486 	struct lwp *l = curlwp;
    487 
    488 	int error;
    489 
    490 	KASSERT(VOP_ISLOCKED(vp));
    491 
    492 	error = 0;
    493 
    494 	/* Abort if any unsettable attribute is given. */
    495 	if (vap->va_type != VNON ||
    496 	    vap->va_nlink != VNOVAL ||
    497 	    vap->va_fsid != VNOVAL ||
    498 	    vap->va_fileid != VNOVAL ||
    499 	    vap->va_blocksize != VNOVAL ||
    500 	    GOODTIME(&vap->va_ctime) ||
    501 	    vap->va_gen != VNOVAL ||
    502 	    vap->va_rdev != VNOVAL ||
    503 	    vap->va_bytes != VNOVAL)
    504 		error = EINVAL;
    505 
    506 	if (error == 0 && (vap->va_flags != VNOVAL))
    507 		error = tmpfs_chflags(vp, vap->va_flags, cred, l);
    508 
    509 	if (error == 0 && (vap->va_size != VNOVAL))
    510 		error = tmpfs_chsize(vp, vap->va_size, cred, l);
    511 
    512 	if (error == 0 && (vap->va_uid != VNOVAL || vap->va_gid != VNOVAL))
    513 		error = tmpfs_chown(vp, vap->va_uid, vap->va_gid, cred, l);
    514 
    515 	if (error == 0 && (vap->va_mode != VNOVAL))
    516 		error = tmpfs_chmod(vp, vap->va_mode, cred, l);
    517 
    518 	if (error == 0 && (GOODTIME(&vap->va_atime) || GOODTIME(&vap->va_mtime)
    519 	    || GOODTIME(&vap->va_birthtime)))
    520 		if ((error = tmpfs_chtimes(vp, &vap->va_atime, &vap->va_mtime,
    521 		    &vap->va_birthtime, vap->va_vaflags, cred, l)) == 0)
    522 			return 0;
    523 
    524 	/* Update the node times.  We give preference to the error codes
    525 	 * generated by this function rather than the ones that may arise
    526 	 * from tmpfs_update. */
    527 	tmpfs_update(vp, NULL, NULL, NULL, 0);
    528 
    529 	KASSERT(VOP_ISLOCKED(vp));
    530 
    531 	return error;
    532 }
    533 
    534 /* --------------------------------------------------------------------- */
    535 
    536 int
    537 tmpfs_read(void *v)
    538 {
    539 	struct vnode *vp = ((struct vop_read_args *)v)->a_vp;
    540 	struct uio *uio = ((struct vop_read_args *)v)->a_uio;
    541 	int ioflag = ((struct vop_read_args *)v)->a_ioflag;
    542 
    543 	int error;
    544 	struct tmpfs_node *node;
    545 	struct uvm_object *uobj;
    546 
    547 	KASSERT(VOP_ISLOCKED(vp));
    548 
    549 	node = VP_TO_TMPFS_NODE(vp);
    550 
    551 	if (vp->v_type != VREG) {
    552 		error = EISDIR;
    553 		goto out;
    554 	}
    555 
    556 	if (uio->uio_offset < 0) {
    557 		error = EINVAL;
    558 		goto out;
    559 	}
    560 
    561 	node->tn_status |= TMPFS_NODE_ACCESSED;
    562 
    563 	uobj = node->tn_spec.tn_reg.tn_aobj;
    564 	error = 0;
    565 	while (error == 0 && uio->uio_resid > 0) {
    566 		vsize_t len;
    567 
    568 		if (node->tn_size <= uio->uio_offset)
    569 			break;
    570 
    571 		len = MIN(node->tn_size - uio->uio_offset, uio->uio_resid);
    572 		if (len == 0)
    573 			break;
    574 
    575 		error = ubc_uiomove(uobj, uio, len, IO_ADV_DECODE(ioflag),
    576 		    UBC_READ | UBC_PARTIALOK | UBC_UNMAP_FLAG(vp));
    577 	}
    578 
    579 out:
    580 	KASSERT(VOP_ISLOCKED(vp));
    581 
    582 	return error;
    583 }
    584 
    585 /* --------------------------------------------------------------------- */
    586 
    587 int
    588 tmpfs_write(void *v)
    589 {
    590 	struct vnode *vp = ((struct vop_write_args *)v)->a_vp;
    591 	struct uio *uio = ((struct vop_write_args *)v)->a_uio;
    592 	int ioflag = ((struct vop_write_args *)v)->a_ioflag;
    593 
    594 	bool extended;
    595 	int error;
    596 	off_t oldsize;
    597 	struct tmpfs_node *node;
    598 	struct uvm_object *uobj;
    599 
    600 	KASSERT(VOP_ISLOCKED(vp));
    601 
    602 	node = VP_TO_TMPFS_NODE(vp);
    603 	oldsize = node->tn_size;
    604 
    605 	if (uio->uio_offset < 0 || vp->v_type != VREG) {
    606 		error = EINVAL;
    607 		goto out;
    608 	}
    609 
    610 	if (uio->uio_resid == 0) {
    611 		error = 0;
    612 		goto out;
    613 	}
    614 
    615 	if (ioflag & IO_APPEND)
    616 		uio->uio_offset = node->tn_size;
    617 
    618 	extended = uio->uio_offset + uio->uio_resid > node->tn_size;
    619 	if (extended) {
    620 		error = tmpfs_reg_resize(vp, uio->uio_offset + uio->uio_resid);
    621 		if (error != 0)
    622 			goto out;
    623 	}
    624 
    625 	uobj = node->tn_spec.tn_reg.tn_aobj;
    626 	error = 0;
    627 	while (error == 0 && uio->uio_resid > 0) {
    628 		vsize_t len;
    629 
    630 		len = MIN(node->tn_size - uio->uio_offset, uio->uio_resid);
    631 		if (len == 0)
    632 			break;
    633 
    634 		error = ubc_uiomove(uobj, uio, len, IO_ADV_DECODE(ioflag),
    635 		    UBC_WRITE | UBC_UNMAP_FLAG(vp));
    636 	}
    637 
    638 	node->tn_status |= TMPFS_NODE_ACCESSED | TMPFS_NODE_MODIFIED |
    639 	    (extended ? TMPFS_NODE_CHANGED : 0);
    640 
    641 	if (error != 0)
    642 		(void)tmpfs_reg_resize(vp, oldsize);
    643 
    644 	VN_KNOTE(vp, NOTE_WRITE);
    645 
    646 out:
    647 	KASSERT(VOP_ISLOCKED(vp));
    648 	KASSERT(IMPLIES(error == 0, uio->uio_resid == 0));
    649 	KASSERT(IMPLIES(error != 0, oldsize == node->tn_size));
    650 
    651 	return error;
    652 }
    653 
    654 /* --------------------------------------------------------------------- */
    655 
    656 int
    657 tmpfs_fsync(void *v)
    658 {
    659 	struct vnode *vp = ((struct vop_fsync_args *)v)->a_vp;
    660 
    661 	KASSERT(VOP_ISLOCKED(vp));
    662 
    663 	tmpfs_update(vp, NULL, NULL, NULL, 0);
    664 
    665 	return 0;
    666 }
    667 
    668 /* --------------------------------------------------------------------- */
    669 
    670 int
    671 tmpfs_remove(void *v)
    672 {
    673 	struct vnode *dvp = ((struct vop_remove_args *)v)->a_dvp;
    674 	struct vnode *vp = ((struct vop_remove_args *)v)->a_vp;
    675 	struct componentname *cnp = (((struct vop_remove_args *)v)->a_cnp);
    676 
    677 	int error;
    678 	struct tmpfs_dirent *de;
    679 	struct tmpfs_mount *tmp;
    680 	struct tmpfs_node *dnode;
    681 	struct tmpfs_node *node;
    682 
    683 	KASSERT(VOP_ISLOCKED(dvp));
    684 	KASSERT(VOP_ISLOCKED(vp));
    685 
    686 	if (vp->v_type == VDIR) {
    687 		error = EPERM;
    688 		goto out;
    689 	}
    690 
    691 	dnode = VP_TO_TMPFS_DIR(dvp);
    692 	node = VP_TO_TMPFS_NODE(vp);
    693 	tmp = VFS_TO_TMPFS(vp->v_mount);
    694 	de = tmpfs_dir_lookup(dnode, cnp);
    695 	KASSERT(de);
    696 	KASSERT(de->td_node == node);
    697 
    698 	/* Files marked as immutable or append-only cannot be deleted. */
    699 	if (node->tn_flags & (IMMUTABLE | APPEND)) {
    700 		error = EPERM;
    701 		goto out;
    702 	}
    703 
    704 	/* Remove the entry from the directory; as it is a file, we do not
    705 	 * have to change the number of hard links of the directory. */
    706 	tmpfs_dir_detach(dvp, de);
    707 
    708 	/* Free the directory entry we just deleted.  Note that the node
    709 	 * referred by it will not be removed until the vnode is really
    710 	 * reclaimed. */
    711 	tmpfs_free_dirent(tmp, de, true);
    712 
    713 	error = 0;
    714 
    715 out:
    716 	vput(vp);
    717 	if (dvp == vp)
    718 		vrele(dvp);
    719 	else
    720 		vput(dvp);
    721 	if (cnp->cn_flags & HASBUF) {
    722 		PNBUF_PUT(cnp->cn_pnbuf);
    723 		cnp->cn_flags &= ~HASBUF;
    724 	}
    725 
    726 	return error;
    727 }
    728 
    729 /* --------------------------------------------------------------------- */
    730 
    731 int
    732 tmpfs_link(void *v)
    733 {
    734 	struct vnode *dvp = ((struct vop_link_args *)v)->a_dvp;
    735 	struct vnode *vp = ((struct vop_link_args *)v)->a_vp;
    736 	struct componentname *cnp = ((struct vop_link_args *)v)->a_cnp;
    737 
    738 	int error;
    739 	struct tmpfs_dirent *de;
    740 	struct tmpfs_node *dnode;
    741 	struct tmpfs_node *node;
    742 
    743 	KASSERT(VOP_ISLOCKED(dvp));
    744 	KASSERT(cnp->cn_flags & HASBUF);
    745 	KASSERT(dvp != vp); /* XXX When can this be false? */
    746 
    747 	dnode = VP_TO_TMPFS_DIR(dvp);
    748 	node = VP_TO_TMPFS_NODE(vp);
    749 
    750 	/* Lock vp because we will need to run tmpfs_update over it, which
    751 	 * needs the vnode to be locked. */
    752 	vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
    753 
    754 	/* XXX: Why aren't the following two tests done by the caller? */
    755 
    756 	/* Hard links of directories are forbidden. */
    757 	if (vp->v_type == VDIR) {
    758 		error = EPERM;
    759 		goto out;
    760 	}
    761 
    762 	/* Cannot create cross-device links. */
    763 	if (dvp->v_mount != vp->v_mount) {
    764 		error = EXDEV;
    765 		goto out;
    766 	}
    767 
    768 	/* Ensure that we do not overflow the maximum number of links imposed
    769 	 * by the system. */
    770 	KASSERT(node->tn_links <= LINK_MAX);
    771 	if (node->tn_links == LINK_MAX) {
    772 		error = EMLINK;
    773 		goto out;
    774 	}
    775 
    776 	/* We cannot create links of files marked immutable or append-only. */
    777 	if (node->tn_flags & (IMMUTABLE | APPEND)) {
    778 		error = EPERM;
    779 		goto out;
    780 	}
    781 
    782 	/* Allocate a new directory entry to represent the node. */
    783 	error = tmpfs_alloc_dirent(VFS_TO_TMPFS(vp->v_mount), node,
    784 	    cnp->cn_nameptr, cnp->cn_namelen, &de);
    785 	if (error != 0)
    786 		goto out;
    787 
    788 	/* Insert the new directory entry into the appropriate directory. */
    789 	tmpfs_dir_attach(dvp, de);
    790 
    791 	/* vp link count has changed, so update node times. */
    792 	node->tn_status |= TMPFS_NODE_CHANGED;
    793 	tmpfs_update(vp, NULL, NULL, NULL, 0);
    794 
    795 	error = 0;
    796 
    797 out:
    798 	VOP_UNLOCK(vp, 0);
    799 	PNBUF_PUT(cnp->cn_pnbuf);
    800 	vput(dvp);
    801 
    802 	return error;
    803 }
    804 
    805 /*
    806  * tmpfs_rename: rename routine.
    807  *
    808  * Arguments: fdvp (from-parent vnode), fvp (from-leaf), tdvp (to-parent)
    809  * and tvp (to-leaf), if exists (NULL if not).
    810  *
    811  * => Caller holds a reference on fdvp and fvp, they are unlocked.
    812  *    Note: fdvp and fvp can refer to the same object (i.e. when it is root).
    813  *
    814  * => Both tdvp and tvp are referenced and locked.  It is our responsibility
    815  *    to release the references and unlock them (or destroy).
    816  */
    817 int
    818 tmpfs_rename(void *v)
    819 {
    820 	struct vnode *fdvp = ((struct vop_rename_args *)v)->a_fdvp;
    821 	struct vnode *fvp = ((struct vop_rename_args *)v)->a_fvp;
    822 	struct componentname *fcnp = ((struct vop_rename_args *)v)->a_fcnp;
    823 	struct vnode *tdvp = ((struct vop_rename_args *)v)->a_tdvp;
    824 	struct vnode *tvp = ((struct vop_rename_args *)v)->a_tvp;
    825 	struct componentname *tcnp = ((struct vop_rename_args *)v)->a_tcnp;
    826 
    827 	char *newname;
    828 	int error;
    829 	struct tmpfs_dirent *de, *de2;
    830 	struct tmpfs_mount *tmp;
    831 	struct tmpfs_node *fdnode;
    832 	struct tmpfs_node *fnode;
    833 	struct tmpfs_node *tnode;
    834 	struct tmpfs_node *tdnode;
    835 	size_t namelen;
    836 
    837 	KASSERT(VOP_ISLOCKED(tdvp));
    838 	KASSERT(IMPLIES(tvp != NULL, VOP_ISLOCKED(tvp) == LK_EXCLUSIVE));
    839 	KASSERT(fcnp->cn_flags & HASBUF);
    840 	KASSERT(tcnp->cn_flags & HASBUF);
    841 
    842 	newname = NULL;
    843 	namelen = 0;
    844 	tmp = NULL;
    845 
    846 	/* Disallow cross-device renames. */
    847 	if (fvp->v_mount != tdvp->v_mount ||
    848 	    (tvp != NULL && fvp->v_mount != tvp->v_mount)) {
    849 		error = EXDEV;
    850 		goto out_unlocked;
    851 	}
    852 
    853 	fnode = VP_TO_TMPFS_NODE(fvp);
    854 	fdnode = VP_TO_TMPFS_DIR(fdvp);
    855 	tnode = (tvp == NULL) ? NULL : VP_TO_TMPFS_NODE(tvp);
    856 	tdnode = VP_TO_TMPFS_DIR(tdvp);
    857 	tmp = VFS_TO_TMPFS(tdvp->v_mount);
    858 
    859 	if (fdvp == tvp) {
    860 		error = 0;
    861 		goto out_unlocked;
    862 	}
    863 
    864 	/* If we need to move the directory between entries, lock the
    865 	 * source so that we can safely operate on it. */
    866 
    867 	/* XXX: this is a potential locking order violation! */
    868 	if (fdnode != tdnode) {
    869 		vn_lock(fdvp, LK_EXCLUSIVE | LK_RETRY);
    870 	}
    871 
    872 	/*
    873 	 * If the node we were renaming has scarpered, just give up.
    874 	 */
    875 	de = tmpfs_dir_lookup(fdnode, fcnp);
    876 	if (de == NULL || de->td_node != fnode) {
    877 		error = ENOENT;
    878 		goto out;
    879 	}
    880 
    881 	/* If source and target is the same vnode, remove the source link. */
    882 	if (fvp == tvp) {
    883 		/*
    884 		 * Detach and free the directory entry.  Drops the link
    885 		 * count on the node.
    886 		 */
    887 		tmpfs_dir_detach(fdvp, de);
    888 		tmpfs_free_dirent(VFS_TO_TMPFS(fvp->v_mount), de, true);
    889 		VN_KNOTE(fdvp, NOTE_WRITE);
    890 		goto out_ok;
    891 	}
    892 
    893 	/* If replacing an existing entry, ensure we can do the operation. */
    894 	if (tvp != NULL) {
    895 		KASSERT(tnode != NULL);
    896 		if (fnode->tn_type == VDIR && tnode->tn_type == VDIR) {
    897 			if (tnode->tn_size > 0) {
    898 				error = ENOTEMPTY;
    899 				goto out;
    900 			}
    901 		} else if (fnode->tn_type == VDIR && tnode->tn_type != VDIR) {
    902 			error = ENOTDIR;
    903 			goto out;
    904 		} else if (fnode->tn_type != VDIR && tnode->tn_type == VDIR) {
    905 			error = EISDIR;
    906 			goto out;
    907 		} else {
    908 			KASSERT(fnode->tn_type != VDIR &&
    909 			        tnode->tn_type != VDIR);
    910 		}
    911 	}
    912 
    913 	/* Ensure that we have enough memory to hold the new name, if it
    914 	 * has to be changed. */
    915 	namelen = tcnp->cn_namelen;
    916 	if (fcnp->cn_namelen != tcnp->cn_namelen ||
    917 	    memcmp(fcnp->cn_nameptr, tcnp->cn_nameptr, fcnp->cn_namelen) != 0) {
    918 		newname = tmpfs_str_pool_get(&tmp->tm_str_pool, namelen, 0);
    919 		if (newname == NULL) {
    920 			error = ENOSPC;
    921 			goto out;
    922 		}
    923 	}
    924 
    925 	/* If the node is being moved to another directory, we have to do
    926 	 * the move. */
    927 	if (fdnode != tdnode) {
    928 		/* In case we are moving a directory, we have to adjust its
    929 		 * parent to point to the new parent. */
    930 		if (de->td_node->tn_type == VDIR) {
    931 			struct tmpfs_node *n;
    932 
    933 			/* Ensure the target directory is not a child of the
    934 			 * directory being moved.  Otherwise, we'd end up
    935 			 * with stale nodes. */
    936 			n = tdnode;
    937 			while (n != n->tn_spec.tn_dir.tn_parent) {
    938 				if (n == fnode) {
    939 					error = EINVAL;
    940 					goto out;
    941 				}
    942 				n = n->tn_spec.tn_dir.tn_parent;
    943 			}
    944 
    945 			/* Adjust the parent pointer. */
    946 			TMPFS_VALIDATE_DIR(fnode);
    947 			de->td_node->tn_spec.tn_dir.tn_parent = tdnode;
    948 
    949 			/* As a result of changing the target of the '..'
    950 			 * entry, the link count of the source and target
    951 			 * directories has to be adjusted. */
    952 			fdnode->tn_links--;
    953 			tdnode->tn_links++;
    954 		}
    955 
    956 		/* Do the move: just remove the entry from the source directory
    957 		 * and insert it into the target one. */
    958 		tmpfs_dir_detach(fdvp, de);
    959 		tmpfs_dir_attach(tdvp, de);
    960 
    961 		/* Notify listeners of fdvp about the change in the directory.
    962 		 * We can do it at this point because we aren't touching fdvp
    963 		 * any more below. */
    964 		VN_KNOTE(fdvp, NOTE_WRITE);
    965 	}
    966 
    967 	/* If we are overwriting an entry, we have to remove the old one
    968 	 * from the target directory. */
    969 	if (tvp != NULL) {
    970 		KASSERT(tnode != NULL);
    971 
    972 		/* Remove the old entry from the target directory.
    973 		 * Note! This relies on tmpfs_dir_attach() putting the new
    974 		 * node on the end of the target's node list. */
    975 		de2 = tmpfs_dir_lookup(tdnode, tcnp);
    976 		KASSERT(de2 != NULL);
    977 		KASSERT(de2->td_node == tnode);
    978 		tmpfs_dir_detach(tdvp, de2);
    979 
    980 		/* Free the directory entry we just deleted.  Note that the
    981 		 * node referred by it will not be removed until the vnode is
    982 		 * really reclaimed. */
    983 		tmpfs_free_dirent(VFS_TO_TMPFS(tvp->v_mount), de2, true);
    984 	}
    985 
    986 	/* If the name has changed, we need to make it effective by changing
    987 	 * it in the directory entry. */
    988 	if (newname != NULL) {
    989 		KASSERT(tcnp->cn_namelen < MAXNAMLEN);
    990 		KASSERT(tcnp->cn_namelen < 0xffff);
    991 
    992 		tmpfs_str_pool_put(&tmp->tm_str_pool, de->td_name,
    993 		    de->td_namelen);
    994 		de->td_namelen = (uint16_t)namelen;
    995 		memcpy(newname, tcnp->cn_nameptr, namelen);
    996 		de->td_name = newname;
    997 		newname = NULL;
    998 
    999 		fnode->tn_status |= TMPFS_NODE_CHANGED;
   1000 		tdnode->tn_status |= TMPFS_NODE_MODIFIED;
   1001 	}
   1002  out_ok:
   1003 	/* Notify listeners of tdvp about the change in the directory (either
   1004 	 * because a new entry was added or because one was removed) and
   1005 	 * listeners of fvp about the rename. */
   1006 	VN_KNOTE(tdvp, NOTE_WRITE);
   1007 	VN_KNOTE(fvp, NOTE_RENAME);
   1008 
   1009 	error = 0;
   1010 
   1011  out:
   1012 	if (fdnode != tdnode)
   1013 		VOP_UNLOCK(fdvp, 0);
   1014 
   1015  out_unlocked:
   1016 	/* Release target nodes. */
   1017 	if (tdvp == tvp)
   1018 		vrele(tdvp);
   1019 	else
   1020 		vput(tdvp);
   1021 	if (tvp != NULL)
   1022 		vput(tvp);
   1023 
   1024 	/* Release source nodes. */
   1025 	vrele(fdvp);
   1026 	vrele(fvp);
   1027 
   1028 	if (newname != NULL)
   1029 		tmpfs_str_pool_put(&tmp->tm_str_pool, newname, namelen);
   1030 
   1031 	return error;
   1032 }
   1033 
   1034 /* --------------------------------------------------------------------- */
   1035 
   1036 int
   1037 tmpfs_mkdir(void *v)
   1038 {
   1039 	struct vnode *dvp = ((struct vop_mkdir_args *)v)->a_dvp;
   1040 	struct vnode **vpp = ((struct vop_mkdir_args *)v)->a_vpp;
   1041 	struct componentname *cnp = ((struct vop_mkdir_args *)v)->a_cnp;
   1042 	struct vattr *vap = ((struct vop_mkdir_args *)v)->a_vap;
   1043 
   1044 	KASSERT(vap->va_type == VDIR);
   1045 
   1046 	return tmpfs_alloc_file(dvp, vpp, vap, cnp, NULL);
   1047 }
   1048 
   1049 /* --------------------------------------------------------------------- */
   1050 
   1051 int
   1052 tmpfs_rmdir(void *v)
   1053 {
   1054 	struct vnode *dvp = ((struct vop_rmdir_args *)v)->a_dvp;
   1055 	struct vnode *vp = ((struct vop_rmdir_args *)v)->a_vp;
   1056 	struct componentname *cnp = ((struct vop_rmdir_args *)v)->a_cnp;
   1057 
   1058 	int error;
   1059 	struct tmpfs_dirent *de;
   1060 	struct tmpfs_mount *tmp;
   1061 	struct tmpfs_node *dnode;
   1062 	struct tmpfs_node *node;
   1063 
   1064 	KASSERT(VOP_ISLOCKED(dvp));
   1065 	KASSERT(VOP_ISLOCKED(vp));
   1066 
   1067 	tmp = VFS_TO_TMPFS(dvp->v_mount);
   1068 	dnode = VP_TO_TMPFS_DIR(dvp);
   1069 	node = VP_TO_TMPFS_DIR(vp);
   1070 	error = 0;
   1071 
   1072 	/* Directories with more than two entries ('.' and '..') cannot be
   1073 	 * removed. */
   1074 	if (node->tn_size > 0) {
   1075 		error = ENOTEMPTY;
   1076 		goto out;
   1077 	}
   1078 
   1079 	/* This invariant holds only if we are not trying to remove "..".
   1080 	 * We checked for that above so this is safe now. */
   1081 	KASSERT(node->tn_spec.tn_dir.tn_parent == dnode);
   1082 
   1083 	/* Get the directory entry associated with node (vp). */
   1084 	de = tmpfs_dir_lookup(dnode, cnp);
   1085 	KASSERT(de);
   1086 	KASSERT(de->td_node == node);
   1087 
   1088 	/* Check flags to see if we are allowed to remove the directory. */
   1089 	if (dnode->tn_flags & APPEND || node->tn_flags & (IMMUTABLE | APPEND)) {
   1090 		error = EPERM;
   1091 		goto out;
   1092 	}
   1093 
   1094 	/* Detach the directory entry from the directory (dnode). */
   1095 	tmpfs_dir_detach(dvp, de);
   1096 
   1097 	node->tn_links--;
   1098 	node->tn_status |= TMPFS_NODE_ACCESSED | TMPFS_NODE_CHANGED | \
   1099 	    TMPFS_NODE_MODIFIED;
   1100 	node->tn_spec.tn_dir.tn_parent->tn_links--;
   1101 	node->tn_spec.tn_dir.tn_parent->tn_status |= TMPFS_NODE_ACCESSED | \
   1102 	    TMPFS_NODE_CHANGED | TMPFS_NODE_MODIFIED;
   1103 
   1104 	/* Release the parent. */
   1105 	cache_purge(dvp); /* XXX Is this needed? */
   1106 
   1107 	/* Free the directory entry we just deleted.  Note that the node
   1108 	 * referred by it will not be removed until the vnode is really
   1109 	 * reclaimed. */
   1110 	tmpfs_free_dirent(tmp, de, true);
   1111 
   1112 	KASSERT(node->tn_links == 0);
   1113  out:
   1114 	/* Release the nodes. */
   1115 	vput(dvp);
   1116 	vput(vp);
   1117 	PNBUF_PUT(cnp->cn_pnbuf);
   1118 
   1119 	return error;
   1120 }
   1121 
   1122 /* --------------------------------------------------------------------- */
   1123 
   1124 int
   1125 tmpfs_symlink(void *v)
   1126 {
   1127 	struct vnode *dvp = ((struct vop_symlink_args *)v)->a_dvp;
   1128 	struct vnode **vpp = ((struct vop_symlink_args *)v)->a_vpp;
   1129 	struct componentname *cnp = ((struct vop_symlink_args *)v)->a_cnp;
   1130 	struct vattr *vap = ((struct vop_symlink_args *)v)->a_vap;
   1131 	char *target = ((struct vop_symlink_args *)v)->a_target;
   1132 
   1133 	KASSERT(vap->va_type == VLNK);
   1134 
   1135 	return tmpfs_alloc_file(dvp, vpp, vap, cnp, target);
   1136 }
   1137 
   1138 /* --------------------------------------------------------------------- */
   1139 
   1140 int
   1141 tmpfs_readdir(void *v)
   1142 {
   1143 	struct vnode *vp = ((struct vop_readdir_args *)v)->a_vp;
   1144 	struct uio *uio = ((struct vop_readdir_args *)v)->a_uio;
   1145 	int *eofflag = ((struct vop_readdir_args *)v)->a_eofflag;
   1146 	off_t **cookies = ((struct vop_readdir_args *)v)->a_cookies;
   1147 	int *ncookies = ((struct vop_readdir_args *)v)->a_ncookies;
   1148 
   1149 	int error;
   1150 	off_t startoff;
   1151 	off_t cnt;
   1152 	struct tmpfs_node *node;
   1153 
   1154 	KASSERT(VOP_ISLOCKED(vp));
   1155 
   1156 	/* This operation only makes sense on directory nodes. */
   1157 	if (vp->v_type != VDIR) {
   1158 		error = ENOTDIR;
   1159 		goto out;
   1160 	}
   1161 
   1162 	node = VP_TO_TMPFS_DIR(vp);
   1163 
   1164 	startoff = uio->uio_offset;
   1165 
   1166 	cnt = 0;
   1167 	if (uio->uio_offset == TMPFS_DIRCOOKIE_DOT) {
   1168 		error = tmpfs_dir_getdotdent(node, uio);
   1169 		if (error == -1) {
   1170 			error = 0;
   1171 			goto outok;
   1172 		} else if (error != 0)
   1173 			goto outok;
   1174 		cnt++;
   1175 	}
   1176 
   1177 	if (uio->uio_offset == TMPFS_DIRCOOKIE_DOTDOT) {
   1178 		error = tmpfs_dir_getdotdotdent(node, uio);
   1179 		if (error == -1) {
   1180 			error = 0;
   1181 			goto outok;
   1182 		} else if (error != 0)
   1183 			goto outok;
   1184 		cnt++;
   1185 	}
   1186 
   1187 	error = tmpfs_dir_getdents(node, uio, &cnt);
   1188 	if (error == -1)
   1189 		error = 0;
   1190 	KASSERT(error >= 0);
   1191 
   1192 outok:
   1193 	/* This label assumes that startoff has been
   1194 	 * initialized.  If the compiler didn't spit out warnings, we'd
   1195 	 * simply make this one be 'out' and drop 'outok'. */
   1196 
   1197 	if (eofflag != NULL)
   1198 		*eofflag =
   1199 		    (error == 0 && uio->uio_offset == TMPFS_DIRCOOKIE_EOF);
   1200 
   1201 	/* Update NFS-related variables. */
   1202 	if (error == 0 && cookies != NULL && ncookies != NULL) {
   1203 		off_t i;
   1204 		off_t off = startoff;
   1205 		struct tmpfs_dirent *de = NULL;
   1206 
   1207 		*ncookies = cnt;
   1208 		*cookies = malloc(cnt * sizeof(off_t), M_TEMP, M_WAITOK);
   1209 
   1210 		for (i = 0; i < cnt; i++) {
   1211 			KASSERT(off != TMPFS_DIRCOOKIE_EOF);
   1212 			if (off == TMPFS_DIRCOOKIE_DOT) {
   1213 				off = TMPFS_DIRCOOKIE_DOTDOT;
   1214 			} else {
   1215 				if (off == TMPFS_DIRCOOKIE_DOTDOT) {
   1216 					de = TAILQ_FIRST(&node->tn_spec.
   1217 					    tn_dir.tn_dir);
   1218 				} else if (de != NULL) {
   1219 					de = TAILQ_NEXT(de, td_entries);
   1220 				} else {
   1221 					de = tmpfs_dir_lookupbycookie(node,
   1222 					    off);
   1223 					KASSERT(de != NULL);
   1224 					de = TAILQ_NEXT(de, td_entries);
   1225 				}
   1226 				if (de == NULL) {
   1227 					off = TMPFS_DIRCOOKIE_EOF;
   1228 				} else {
   1229 					off = tmpfs_dircookie(de);
   1230 				}
   1231 			}
   1232 
   1233 			(*cookies)[i] = off;
   1234 		}
   1235 		KASSERT(uio->uio_offset == off);
   1236 	}
   1237 
   1238 out:
   1239 	KASSERT(VOP_ISLOCKED(vp));
   1240 
   1241 	return error;
   1242 }
   1243 
   1244 /* --------------------------------------------------------------------- */
   1245 
   1246 int
   1247 tmpfs_readlink(void *v)
   1248 {
   1249 	struct vnode *vp = ((struct vop_readlink_args *)v)->a_vp;
   1250 	struct uio *uio = ((struct vop_readlink_args *)v)->a_uio;
   1251 
   1252 	int error;
   1253 	struct tmpfs_node *node;
   1254 
   1255 	KASSERT(VOP_ISLOCKED(vp));
   1256 	KASSERT(uio->uio_offset == 0);
   1257 	KASSERT(vp->v_type == VLNK);
   1258 
   1259 	node = VP_TO_TMPFS_NODE(vp);
   1260 
   1261 	error = uiomove(node->tn_spec.tn_lnk.tn_link,
   1262 	    MIN(node->tn_size, uio->uio_resid), uio);
   1263 	node->tn_status |= TMPFS_NODE_ACCESSED;
   1264 
   1265 	KASSERT(VOP_ISLOCKED(vp));
   1266 
   1267 	return error;
   1268 }
   1269 
   1270 /* --------------------------------------------------------------------- */
   1271 
   1272 int
   1273 tmpfs_inactive(void *v)
   1274 {
   1275 	struct vnode *vp = ((struct vop_inactive_args *)v)->a_vp;
   1276 
   1277 	struct tmpfs_node *node;
   1278 
   1279 	KASSERT(VOP_ISLOCKED(vp));
   1280 
   1281 	node = VP_TO_TMPFS_NODE(vp);
   1282 	*((struct vop_inactive_args *)v)->a_recycle = (node->tn_links == 0);
   1283 	VOP_UNLOCK(vp, 0);
   1284 
   1285 	return 0;
   1286 }
   1287 
   1288 /* --------------------------------------------------------------------- */
   1289 
   1290 int
   1291 tmpfs_reclaim(void *v)
   1292 {
   1293 	struct vnode *vp = ((struct vop_reclaim_args *)v)->a_vp;
   1294 
   1295 	struct tmpfs_mount *tmp;
   1296 	struct tmpfs_node *node;
   1297 
   1298 	node = VP_TO_TMPFS_NODE(vp);
   1299 	tmp = VFS_TO_TMPFS(vp->v_mount);
   1300 
   1301 	cache_purge(vp);
   1302 	tmpfs_free_vp(vp);
   1303 
   1304 	/* If the node referenced by this vnode was deleted by the user,
   1305 	 * we must free its associated data structures (now that the vnode
   1306 	 * is being reclaimed). */
   1307 	if (node->tn_links == 0)
   1308 		tmpfs_free_node(tmp, node);
   1309 
   1310 	KASSERT(vp->v_data == NULL);
   1311 
   1312 	return 0;
   1313 }
   1314 
   1315 /* --------------------------------------------------------------------- */
   1316 
   1317 int
   1318 tmpfs_print(void *v)
   1319 {
   1320 	struct vnode *vp = ((struct vop_print_args *)v)->a_vp;
   1321 
   1322 	struct tmpfs_node *node;
   1323 
   1324 	node = VP_TO_TMPFS_NODE(vp);
   1325 
   1326 	printf("tag VT_TMPFS, tmpfs_node %p, flags 0x%x, links %d\n",
   1327 	    node, node->tn_flags, node->tn_links);
   1328 	printf("\tmode 0%o, owner %d, group %d, size %" PRIdMAX
   1329 	    ", status 0x%x",
   1330 	    node->tn_mode, node->tn_uid, node->tn_gid,
   1331 	    (uintmax_t)node->tn_size, node->tn_status);
   1332 	if (vp->v_type == VFIFO)
   1333 		VOCALL(fifo_vnodeop_p, VOFFSET(vop_print), v);
   1334 	printf("\n");
   1335 
   1336 	return 0;
   1337 }
   1338 
   1339 /* --------------------------------------------------------------------- */
   1340 
   1341 int
   1342 tmpfs_pathconf(void *v)
   1343 {
   1344 	int name = ((struct vop_pathconf_args *)v)->a_name;
   1345 	register_t *retval = ((struct vop_pathconf_args *)v)->a_retval;
   1346 
   1347 	int error;
   1348 
   1349 	error = 0;
   1350 
   1351 	switch (name) {
   1352 	case _PC_LINK_MAX:
   1353 		*retval = LINK_MAX;
   1354 		break;
   1355 
   1356 	case _PC_NAME_MAX:
   1357 		*retval = NAME_MAX;
   1358 		break;
   1359 
   1360 	case _PC_PATH_MAX:
   1361 		*retval = PATH_MAX;
   1362 		break;
   1363 
   1364 	case _PC_PIPE_BUF:
   1365 		*retval = PIPE_BUF;
   1366 		break;
   1367 
   1368 	case _PC_CHOWN_RESTRICTED:
   1369 		*retval = 1;
   1370 		break;
   1371 
   1372 	case _PC_NO_TRUNC:
   1373 		*retval = 1;
   1374 		break;
   1375 
   1376 	case _PC_SYNC_IO:
   1377 		*retval = 1;
   1378 		break;
   1379 
   1380 	case _PC_FILESIZEBITS:
   1381 		*retval = 0; /* XXX Don't know which value should I return. */
   1382 		break;
   1383 
   1384 	default:
   1385 		error = EINVAL;
   1386 	}
   1387 
   1388 	return error;
   1389 }
   1390 
   1391 /* --------------------------------------------------------------------- */
   1392 
   1393 int
   1394 tmpfs_advlock(void *v)
   1395 {
   1396 	struct vnode *vp = ((struct vop_advlock_args *)v)->a_vp;
   1397 
   1398 	struct tmpfs_node *node;
   1399 
   1400 	node = VP_TO_TMPFS_NODE(vp);
   1401 
   1402 	return lf_advlock(v, &node->tn_lockf, node->tn_size);
   1403 }
   1404 
   1405 /* --------------------------------------------------------------------- */
   1406 
   1407 int
   1408 tmpfs_getpages(void *v)
   1409 {
   1410 	struct vnode *vp = ((struct vop_getpages_args *)v)->a_vp;
   1411 	voff_t offset = ((struct vop_getpages_args *)v)->a_offset;
   1412 	struct vm_page **m = ((struct vop_getpages_args *)v)->a_m;
   1413 	int *count = ((struct vop_getpages_args *)v)->a_count;
   1414 	int centeridx = ((struct vop_getpages_args *)v)->a_centeridx;
   1415 	vm_prot_t access_type = ((struct vop_getpages_args *)v)->a_access_type;
   1416 	int advice = ((struct vop_getpages_args *)v)->a_advice;
   1417 	int flags = ((struct vop_getpages_args *)v)->a_flags;
   1418 
   1419 	int error;
   1420 	int i;
   1421 	struct tmpfs_node *node;
   1422 	struct uvm_object *uobj;
   1423 	int npages = *count;
   1424 
   1425 	KASSERT(vp->v_type == VREG);
   1426 	KASSERT(mutex_owned(vp->v_interlock));
   1427 
   1428 	node = VP_TO_TMPFS_NODE(vp);
   1429 	uobj = node->tn_spec.tn_reg.tn_aobj;
   1430 
   1431 	/* We currently don't rely on PGO_PASTEOF. */
   1432 
   1433 	if (vp->v_size <= offset + (centeridx << PAGE_SHIFT)) {
   1434 		if ((flags & PGO_LOCKED) == 0)
   1435 			mutex_exit(vp->v_interlock);
   1436 		return EINVAL;
   1437 	}
   1438 
   1439 	if (vp->v_size < offset + (npages << PAGE_SHIFT)) {
   1440 		npages = (round_page(vp->v_size) - offset) >> PAGE_SHIFT;
   1441 	}
   1442 
   1443 	if ((flags & PGO_LOCKED) != 0)
   1444 		return EBUSY;
   1445 
   1446 	if ((flags & PGO_NOTIMESTAMP) == 0) {
   1447 		if ((vp->v_mount->mnt_flag & MNT_NOATIME) == 0)
   1448 			node->tn_status |= TMPFS_NODE_ACCESSED;
   1449 
   1450 		if ((access_type & VM_PROT_WRITE) != 0)
   1451 			node->tn_status |= TMPFS_NODE_MODIFIED;
   1452 	}
   1453 
   1454 	mutex_exit(vp->v_interlock);
   1455 
   1456 	/*
   1457 	 * Make sure that the array on which we will store the
   1458 	 * gotten pages is clean.  Otherwise uao_get (pointed to by
   1459 	 * the pgo_get below) gets confused and does not return the
   1460 	 * appropriate pages.
   1461 	 *
   1462 	 * XXX This shall be revisited when kern/32166 is addressed
   1463 	 * because the loop to clean m[i] will most likely be redundant
   1464 	 * as well as the PGO_ALLPAGES flag.
   1465 	 */
   1466 	if (m != NULL)
   1467 		for (i = 0; i < npages; i++)
   1468 			m[i] = NULL;
   1469 	mutex_enter(uobj->vmobjlock);
   1470 	error = (*uobj->pgops->pgo_get)(uobj, offset, m, &npages, centeridx,
   1471 	    access_type, advice, flags | PGO_ALLPAGES);
   1472 #if defined(DEBUG)
   1473 	{
   1474 		/* Make sure that all the pages we return are valid. */
   1475 		int dbgi;
   1476 		if (error == 0 && m != NULL)
   1477 			for (dbgi = 0; dbgi < npages; dbgi++)
   1478 				KASSERT(m[dbgi] != NULL);
   1479 	}
   1480 #endif
   1481 
   1482 	return error;
   1483 }
   1484 
   1485 /* --------------------------------------------------------------------- */
   1486 
   1487 int
   1488 tmpfs_putpages(void *v)
   1489 {
   1490 	struct vnode *vp = ((struct vop_putpages_args *)v)->a_vp;
   1491 	voff_t offlo = ((struct vop_putpages_args *)v)->a_offlo;
   1492 	voff_t offhi = ((struct vop_putpages_args *)v)->a_offhi;
   1493 	int flags = ((struct vop_putpages_args *)v)->a_flags;
   1494 
   1495 	int error;
   1496 	struct tmpfs_node *node;
   1497 	struct uvm_object *uobj;
   1498 
   1499 	KASSERT(mutex_owned(vp->v_interlock));
   1500 
   1501 	node = VP_TO_TMPFS_NODE(vp);
   1502 
   1503 	if (vp->v_type != VREG) {
   1504 		mutex_exit(vp->v_interlock);
   1505 		return 0;
   1506 	}
   1507 
   1508 	uobj = node->tn_spec.tn_reg.tn_aobj;
   1509 	mutex_exit(vp->v_interlock);
   1510 
   1511 	mutex_enter(uobj->vmobjlock);
   1512 	error = (*uobj->pgops->pgo_put)(uobj, offlo, offhi, flags);
   1513 
   1514 	/* XXX mtime */
   1515 
   1516 	return error;
   1517 }
   1518