1 /* $NetBSD: tmpfs_vnops.c,v 1.39 2007/07/23 15:41:01 jmmv Exp $ */ 2 3 /*- 4 * SPDX-License-Identifier: BSD-2-Clause-NetBSD 5 * 6 * Copyright (c) 2005, 2006 The NetBSD Foundation, Inc. 7 * All rights reserved. 8 * 9 * This code is derived from software contributed to The NetBSD Foundation 10 * by Julio M. Merino Vidal, developed as part of Google's Summer of Code 11 * 2005 program. 12 * 13 * Redistribution and use in source and binary forms, with or without 14 * modification, are permitted provided that the following conditions 15 * are met: 16 * 1. Redistributions of source code must retain the above copyright 17 * notice, this list of conditions and the following disclaimer. 18 * 2. Redistributions in binary form must reproduce the above copyright 19 * notice, this list of conditions and the following disclaimer in the 20 * documentation and/or other materials provided with the distribution. 21 * 22 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 23 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 24 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 25 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 26 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 27 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 28 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 29 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 30 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 31 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 32 * POSSIBILITY OF SUCH DAMAGE. 33 */ 34 35 /* 36 * tmpfs vnode interface. 37 */ 38 #include <sys/cdefs.h> 39 __FBSDID("$FreeBSD$"); 40 41 #include <sys/param.h> 42 #include <sys/systm.h> 43 #include <sys/dirent.h> 44 #include <sys/fcntl.h> 45 #include <sys/file.h> 46 #include <sys/limits.h> 47 #include <sys/lockf.h> 48 #include <sys/lock.h> 49 #include <sys/mount.h> 50 #include <sys/namei.h> 51 #include <sys/priv.h> 52 #include <sys/proc.h> 53 #include <sys/rwlock.h> 54 #include <sys/sched.h> 55 #include <sys/stat.h> 56 #include <sys/sysctl.h> 57 #include <sys/unistd.h> 58 #include <sys/vnode.h> 59 #include <sys/smr.h> 60 #include <security/audit/audit.h> 61 #include <security/mac/mac_framework.h> 62 63 #include <vm/vm.h> 64 #include <vm/vm_param.h> 65 #include <vm/vm_object.h> 66 67 #include <fs/tmpfs/tmpfs_vnops.h> 68 #include <fs/tmpfs/tmpfs.h> 69 70 SYSCTL_DECL(_vfs_tmpfs); 71 VFS_SMR_DECLARE; 72 73 static volatile int tmpfs_rename_restarts; 74 SYSCTL_INT(_vfs_tmpfs, OID_AUTO, rename_restarts, CTLFLAG_RD, 75 __DEVOLATILE(int *, &tmpfs_rename_restarts), 0, 76 "Times rename had to restart due to lock contention"); 77 78 static int 79 tmpfs_vn_get_ino_alloc(struct mount *mp, void *arg, int lkflags, 80 struct vnode **rvp) 81 { 82 83 return (tmpfs_alloc_vp(mp, arg, lkflags, rvp)); 84 } 85 86 static int 87 tmpfs_lookup1(struct vnode *dvp, struct vnode **vpp, struct componentname *cnp) 88 { 89 struct tmpfs_dirent *de; 90 struct tmpfs_node *dnode, *pnode; 91 struct tmpfs_mount *tm; 92 int error; 93 94 /* Caller assumes responsibility for ensuring access (VEXEC). */ 95 dnode = VP_TO_TMPFS_DIR(dvp); 96 *vpp = NULLVP; 97 98 /* We cannot be requesting the parent directory of the root node. */ 99 MPASS(IMPLIES(dnode->tn_type == VDIR && 100 dnode->tn_dir.tn_parent == dnode, 101 !(cnp->cn_flags & ISDOTDOT))); 102 103 TMPFS_ASSERT_LOCKED(dnode); 104 if (dnode->tn_dir.tn_parent == NULL) { 105 error = ENOENT; 106 goto out; 107 } 108 if (cnp->cn_flags & ISDOTDOT) { 109 tm = VFS_TO_TMPFS(dvp->v_mount); 110 pnode = dnode->tn_dir.tn_parent; 111 tmpfs_ref_node(pnode); 112 error = vn_vget_ino_gen(dvp, tmpfs_vn_get_ino_alloc, 113 pnode, cnp->cn_lkflags, vpp); 114 tmpfs_free_node(tm, pnode); 115 if (error != 0) 116 goto out; 117 } else if (cnp->cn_namelen == 1 && cnp->cn_nameptr[0] == '.') { 118 VREF(dvp); 119 *vpp = dvp; 120 error = 0; 121 } else { 122 de = tmpfs_dir_lookup(dnode, NULL, cnp); 123 if (de != NULL && de->td_node == NULL) 124 cnp->cn_flags |= ISWHITEOUT; 125 if (de == NULL || de->td_node == NULL) { 126 /* 127 * The entry was not found in the directory. 128 * This is OK if we are creating or renaming an 129 * entry and are working on the last component of 130 * the path name. 131 */ 132 if ((cnp->cn_flags & ISLASTCN) && 133 (cnp->cn_nameiop == CREATE || \ 134 cnp->cn_nameiop == RENAME || 135 (cnp->cn_nameiop == DELETE && 136 cnp->cn_flags & DOWHITEOUT && 137 cnp->cn_flags & ISWHITEOUT))) { 138 error = VOP_ACCESS(dvp, VWRITE, cnp->cn_cred, 139 curthread); 140 if (error != 0) 141 goto out; 142 143 /* 144 * Keep the component name in the buffer for 145 * future uses. 146 */ 147 cnp->cn_flags |= SAVENAME; 148 149 error = EJUSTRETURN; 150 } else 151 error = ENOENT; 152 } else { 153 struct tmpfs_node *tnode; 154 155 /* 156 * The entry was found, so get its associated 157 * tmpfs_node. 158 */ 159 tnode = de->td_node; 160 161 /* 162 * If we are not at the last path component and 163 * found a non-directory or non-link entry (which 164 * may itself be pointing to a directory), raise 165 * an error. 166 */ 167 if ((tnode->tn_type != VDIR && 168 tnode->tn_type != VLNK) && 169 !(cnp->cn_flags & ISLASTCN)) { 170 error = ENOTDIR; 171 goto out; 172 } 173 174 /* 175 * If we are deleting or renaming the entry, keep 176 * track of its tmpfs_dirent so that it can be 177 * easily deleted later. 178 */ 179 if ((cnp->cn_flags & ISLASTCN) && 180 (cnp->cn_nameiop == DELETE || 181 cnp->cn_nameiop == RENAME)) { 182 error = VOP_ACCESS(dvp, VWRITE, cnp->cn_cred, 183 curthread); 184 if (error != 0) 185 goto out; 186 187 /* Allocate a new vnode on the matching entry. */ 188 error = tmpfs_alloc_vp(dvp->v_mount, tnode, 189 cnp->cn_lkflags, vpp); 190 if (error != 0) 191 goto out; 192 193 if ((dnode->tn_mode & S_ISTXT) && 194 VOP_ACCESS(dvp, VADMIN, cnp->cn_cred, 195 curthread) && VOP_ACCESS(*vpp, VADMIN, 196 cnp->cn_cred, curthread)) { 197 error = EPERM; 198 vput(*vpp); 199 *vpp = NULL; 200 goto out; 201 } 202 cnp->cn_flags |= SAVENAME; 203 } else { 204 error = tmpfs_alloc_vp(dvp->v_mount, tnode, 205 cnp->cn_lkflags, vpp); 206 if (error != 0) 207 goto out; 208 } 209 } 210 } 211 212 /* 213 * Store the result of this lookup in the cache. Avoid this if the 214 * request was for creation, as it does not improve timings on 215 * emprical tests. 216 */ 217 if ((cnp->cn_flags & MAKEENTRY) != 0 && tmpfs_use_nc(dvp)) 218 cache_enter(dvp, *vpp, cnp); 219 220 out: 221 /* 222 * If there were no errors, *vpp cannot be null and it must be 223 * locked. 224 */ 225 MPASS(IFF(error == 0, *vpp != NULLVP && VOP_ISLOCKED(*vpp))); 226 227 return (error); 228 } 229 230 static int 231 tmpfs_cached_lookup(struct vop_cachedlookup_args *v) 232 { 233 234 return (tmpfs_lookup1(v->a_dvp, v->a_vpp, v->a_cnp)); 235 } 236 237 static int 238 tmpfs_lookup(struct vop_lookup_args *v) 239 { 240 struct vnode *dvp = v->a_dvp; 241 struct vnode **vpp = v->a_vpp; 242 struct componentname *cnp = v->a_cnp; 243 int error; 244 245 /* Check accessibility of requested node as a first step. */ 246 error = vn_dir_check_exec(dvp, cnp); 247 if (error != 0) 248 return (error); 249 250 return (tmpfs_lookup1(dvp, vpp, cnp)); 251 } 252 253 static int 254 tmpfs_create(struct vop_create_args *v) 255 { 256 struct vnode *dvp = v->a_dvp; 257 struct vnode **vpp = v->a_vpp; 258 struct componentname *cnp = v->a_cnp; 259 struct vattr *vap = v->a_vap; 260 int error; 261 262 MPASS(vap->va_type == VREG || vap->va_type == VSOCK); 263 264 error = tmpfs_alloc_file(dvp, vpp, vap, cnp, NULL); 265 if (error == 0 && (cnp->cn_flags & MAKEENTRY) != 0 && tmpfs_use_nc(dvp)) 266 cache_enter(dvp, *vpp, cnp); 267 return (error); 268 } 269 270 static int 271 tmpfs_mknod(struct vop_mknod_args *v) 272 { 273 struct vnode *dvp = v->a_dvp; 274 struct vnode **vpp = v->a_vpp; 275 struct componentname *cnp = v->a_cnp; 276 struct vattr *vap = v->a_vap; 277 278 if (vap->va_type != VBLK && vap->va_type != VCHR && 279 vap->va_type != VFIFO) 280 return (EINVAL); 281 282 return (tmpfs_alloc_file(dvp, vpp, vap, cnp, NULL)); 283 } 284 285 struct fileops tmpfs_fnops; 286 287 static int 288 tmpfs_open(struct vop_open_args *v) 289 { 290 struct vnode *vp; 291 struct tmpfs_node *node; 292 struct file *fp; 293 int error, mode; 294 295 vp = v->a_vp; 296 mode = v->a_mode; 297 node = VP_TO_TMPFS_NODE(vp); 298 299 /* 300 * The file is still active but all its names have been removed 301 * (e.g. by a "rmdir $(pwd)"). It cannot be opened any more as 302 * it is about to die. 303 */ 304 if (node->tn_links < 1) 305 return (ENOENT); 306 307 /* If the file is marked append-only, deny write requests. */ 308 if (node->tn_flags & APPEND && (mode & (FWRITE | O_APPEND)) == FWRITE) 309 error = EPERM; 310 else { 311 error = 0; 312 /* For regular files, the call below is nop. */ 313 KASSERT(vp->v_type != VREG || (node->tn_reg.tn_aobj->flags & 314 OBJ_DEAD) == 0, ("dead object")); 315 vnode_create_vobject(vp, node->tn_size, v->a_td); 316 } 317 318 fp = v->a_fp; 319 MPASS(fp == NULL || fp->f_data == NULL); 320 if (error == 0 && fp != NULL && vp->v_type == VREG) { 321 tmpfs_ref_node(node); 322 finit_vnode(fp, mode, node, &tmpfs_fnops); 323 } 324 325 return (error); 326 } 327 328 static int 329 tmpfs_close(struct vop_close_args *v) 330 { 331 struct vnode *vp = v->a_vp; 332 333 /* Update node times. */ 334 tmpfs_update(vp); 335 336 return (0); 337 } 338 339 int 340 tmpfs_fo_close(struct file *fp, struct thread *td) 341 { 342 struct tmpfs_node *node; 343 344 node = fp->f_data; 345 if (node != NULL) { 346 MPASS(node->tn_type == VREG); 347 tmpfs_free_node(node->tn_reg.tn_tmp, node); 348 } 349 return (vnops.fo_close(fp, td)); 350 } 351 352 /* 353 * VOP_FPLOOKUP_VEXEC routines are subject to special circumstances, see 354 * the comment above cache_fplookup for details. 355 */ 356 int 357 tmpfs_fplookup_vexec(struct vop_fplookup_vexec_args *v) 358 { 359 struct vnode *vp; 360 struct tmpfs_node *node; 361 struct ucred *cred; 362 mode_t all_x, mode; 363 364 vp = v->a_vp; 365 node = VP_TO_TMPFS_NODE_SMR(vp); 366 if (__predict_false(node == NULL)) 367 return (EAGAIN); 368 369 all_x = S_IXUSR | S_IXGRP | S_IXOTH; 370 mode = atomic_load_short(&node->tn_mode); 371 if (__predict_true((mode & all_x) == all_x)) 372 return (0); 373 374 cred = v->a_cred; 375 return (vaccess_vexec_smr(mode, node->tn_uid, node->tn_gid, cred)); 376 } 377 378 int 379 tmpfs_access(struct vop_access_args *v) 380 { 381 struct vnode *vp = v->a_vp; 382 accmode_t accmode = v->a_accmode; 383 struct ucred *cred = v->a_cred; 384 mode_t all_x = S_IXUSR | S_IXGRP | S_IXOTH; 385 int error; 386 struct tmpfs_node *node; 387 388 MPASS(VOP_ISLOCKED(vp)); 389 390 node = VP_TO_TMPFS_NODE(vp); 391 392 /* 393 * Common case path lookup. 394 */ 395 if (__predict_true(accmode == VEXEC && (node->tn_mode & all_x) == all_x)) 396 return (0); 397 398 switch (vp->v_type) { 399 case VDIR: 400 /* FALLTHROUGH */ 401 case VLNK: 402 /* FALLTHROUGH */ 403 case VREG: 404 if (accmode & VWRITE && vp->v_mount->mnt_flag & MNT_RDONLY) { 405 error = EROFS; 406 goto out; 407 } 408 break; 409 410 case VBLK: 411 /* FALLTHROUGH */ 412 case VCHR: 413 /* FALLTHROUGH */ 414 case VSOCK: 415 /* FALLTHROUGH */ 416 case VFIFO: 417 break; 418 419 default: 420 error = EINVAL; 421 goto out; 422 } 423 424 if (accmode & VWRITE && node->tn_flags & IMMUTABLE) { 425 error = EPERM; 426 goto out; 427 } 428 429 error = vaccess(vp->v_type, node->tn_mode, node->tn_uid, node->tn_gid, 430 accmode, cred); 431 432 out: 433 MPASS(VOP_ISLOCKED(vp)); 434 435 return (error); 436 } 437 438 int 439 tmpfs_stat(struct vop_stat_args *v) 440 { 441 struct vnode *vp = v->a_vp; 442 struct stat *sb = v->a_sb; 443 vm_object_t obj; 444 struct tmpfs_node *node; 445 int error; 446 447 node = VP_TO_TMPFS_NODE(vp); 448 449 tmpfs_update_getattr(vp); 450 451 error = vop_stat_helper_pre(v); 452 if (__predict_false(error)) 453 return (error); 454 455 sb->st_dev = vp->v_mount->mnt_stat.f_fsid.val[0]; 456 sb->st_ino = node->tn_id; 457 sb->st_mode = node->tn_mode | VTTOIF(vp->v_type); 458 sb->st_nlink = node->tn_links; 459 sb->st_uid = node->tn_uid; 460 sb->st_gid = node->tn_gid; 461 sb->st_rdev = (vp->v_type == VBLK || vp->v_type == VCHR) ? 462 node->tn_rdev : NODEV; 463 sb->st_size = node->tn_size; 464 sb->st_atim.tv_sec = node->tn_atime.tv_sec; 465 sb->st_atim.tv_nsec = node->tn_atime.tv_nsec; 466 sb->st_mtim.tv_sec = node->tn_mtime.tv_sec; 467 sb->st_mtim.tv_nsec = node->tn_mtime.tv_nsec; 468 sb->st_ctim.tv_sec = node->tn_ctime.tv_sec; 469 sb->st_ctim.tv_nsec = node->tn_ctime.tv_nsec; 470 sb->st_birthtim.tv_sec = node->tn_birthtime.tv_sec; 471 sb->st_birthtim.tv_nsec = node->tn_birthtime.tv_nsec; 472 sb->st_blksize = PAGE_SIZE; 473 sb->st_flags = node->tn_flags; 474 sb->st_gen = node->tn_gen; 475 if (vp->v_type == VREG) { 476 obj = node->tn_reg.tn_aobj; 477 sb->st_blocks = (u_quad_t)obj->resident_page_count * PAGE_SIZE; 478 } else 479 sb->st_blocks = node->tn_size; 480 sb->st_blocks /= S_BLKSIZE; 481 return (vop_stat_helper_post(v, error)); 482 } 483 484 int 485 tmpfs_getattr(struct vop_getattr_args *v) 486 { 487 struct vnode *vp = v->a_vp; 488 struct vattr *vap = v->a_vap; 489 vm_object_t obj; 490 struct tmpfs_node *node; 491 492 node = VP_TO_TMPFS_NODE(vp); 493 494 tmpfs_update_getattr(vp); 495 496 vap->va_type = vp->v_type; 497 vap->va_mode = node->tn_mode; 498 vap->va_nlink = node->tn_links; 499 vap->va_uid = node->tn_uid; 500 vap->va_gid = node->tn_gid; 501 vap->va_fsid = vp->v_mount->mnt_stat.f_fsid.val[0]; 502 vap->va_fileid = node->tn_id; 503 vap->va_size = node->tn_size; 504 vap->va_blocksize = PAGE_SIZE; 505 vap->va_atime = node->tn_atime; 506 vap->va_mtime = node->tn_mtime; 507 vap->va_ctime = node->tn_ctime; 508 vap->va_birthtime = node->tn_birthtime; 509 vap->va_gen = node->tn_gen; 510 vap->va_flags = node->tn_flags; 511 vap->va_rdev = (vp->v_type == VBLK || vp->v_type == VCHR) ? 512 node->tn_rdev : NODEV; 513 if (vp->v_type == VREG) { 514 obj = node->tn_reg.tn_aobj; 515 vap->va_bytes = (u_quad_t)obj->resident_page_count * PAGE_SIZE; 516 } else 517 vap->va_bytes = node->tn_size; 518 vap->va_filerev = 0; 519 520 return (0); 521 } 522 523 int 524 tmpfs_setattr(struct vop_setattr_args *v) 525 { 526 struct vnode *vp = v->a_vp; 527 struct vattr *vap = v->a_vap; 528 struct ucred *cred = v->a_cred; 529 struct thread *td = curthread; 530 531 int error; 532 533 MPASS(VOP_ISLOCKED(vp)); 534 ASSERT_VOP_IN_SEQC(vp); 535 536 error = 0; 537 538 /* Abort if any unsettable attribute is given. */ 539 if (vap->va_type != VNON || 540 vap->va_nlink != VNOVAL || 541 vap->va_fsid != VNOVAL || 542 vap->va_fileid != VNOVAL || 543 vap->va_blocksize != VNOVAL || 544 vap->va_gen != VNOVAL || 545 vap->va_rdev != VNOVAL || 546 vap->va_bytes != VNOVAL) 547 error = EINVAL; 548 549 if (error == 0 && (vap->va_flags != VNOVAL)) 550 error = tmpfs_chflags(vp, vap->va_flags, cred, td); 551 552 if (error == 0 && (vap->va_size != VNOVAL)) 553 error = tmpfs_chsize(vp, vap->va_size, cred, td); 554 555 if (error == 0 && (vap->va_uid != VNOVAL || vap->va_gid != VNOVAL)) 556 error = tmpfs_chown(vp, vap->va_uid, vap->va_gid, cred, td); 557 558 if (error == 0 && (vap->va_mode != (mode_t)VNOVAL)) 559 error = tmpfs_chmod(vp, vap->va_mode, cred, td); 560 561 if (error == 0 && ((vap->va_atime.tv_sec != VNOVAL && 562 vap->va_atime.tv_nsec != VNOVAL) || 563 (vap->va_mtime.tv_sec != VNOVAL && 564 vap->va_mtime.tv_nsec != VNOVAL) || 565 (vap->va_birthtime.tv_sec != VNOVAL && 566 vap->va_birthtime.tv_nsec != VNOVAL))) 567 error = tmpfs_chtimes(vp, vap, cred, td); 568 569 /* 570 * Update the node times. We give preference to the error codes 571 * generated by this function rather than the ones that may arise 572 * from tmpfs_update. 573 */ 574 tmpfs_update(vp); 575 576 MPASS(VOP_ISLOCKED(vp)); 577 578 return (error); 579 } 580 581 static int 582 tmpfs_read(struct vop_read_args *v) 583 { 584 struct vnode *vp; 585 struct uio *uio; 586 struct tmpfs_node *node; 587 588 vp = v->a_vp; 589 if (vp->v_type != VREG) 590 return (EISDIR); 591 uio = v->a_uio; 592 if (uio->uio_offset < 0) 593 return (EINVAL); 594 node = VP_TO_TMPFS_NODE(vp); 595 tmpfs_set_accessed(VFS_TO_TMPFS(vp->v_mount), node); 596 return (uiomove_object(node->tn_reg.tn_aobj, node->tn_size, uio)); 597 } 598 599 static int 600 tmpfs_read_pgcache(struct vop_read_pgcache_args *v) 601 { 602 struct vnode *vp; 603 struct tmpfs_node *node; 604 vm_object_t object; 605 off_t size; 606 int error; 607 608 vp = v->a_vp; 609 VNPASS((vn_irflag_read(vp) & VIRF_PGREAD) != 0, vp); 610 611 if (v->a_uio->uio_offset < 0) 612 return (EINVAL); 613 614 error = EJUSTRETURN; 615 vfs_smr_enter(); 616 617 node = VP_TO_TMPFS_NODE_SMR(vp); 618 if (node == NULL) 619 goto out_smr; 620 MPASS(node->tn_type == VREG); 621 MPASS(node->tn_refcount >= 1); 622 object = node->tn_reg.tn_aobj; 623 if (object == NULL) 624 goto out_smr; 625 626 MPASS(object->type == tmpfs_pager_type); 627 MPASS((object->flags & (OBJ_ANON | OBJ_DEAD | OBJ_SWAP)) == 628 OBJ_SWAP); 629 if (!VN_IS_DOOMED(vp)) { 630 /* size cannot become shorter due to rangelock. */ 631 size = node->tn_size; 632 tmpfs_set_accessed(node->tn_reg.tn_tmp, node); 633 vfs_smr_exit(); 634 error = uiomove_object(object, size, v->a_uio); 635 return (error); 636 } 637 out_smr: 638 vfs_smr_exit(); 639 return (error); 640 } 641 642 static int 643 tmpfs_write(struct vop_write_args *v) 644 { 645 struct vnode *vp; 646 struct uio *uio; 647 struct tmpfs_node *node; 648 off_t oldsize; 649 int error, ioflag; 650 mode_t newmode; 651 652 vp = v->a_vp; 653 uio = v->a_uio; 654 ioflag = v->a_ioflag; 655 error = 0; 656 node = VP_TO_TMPFS_NODE(vp); 657 oldsize = node->tn_size; 658 659 if (uio->uio_offset < 0 || vp->v_type != VREG) 660 return (EINVAL); 661 if (uio->uio_resid == 0) 662 return (0); 663 if (ioflag & IO_APPEND) 664 uio->uio_offset = node->tn_size; 665 if (uio->uio_offset + uio->uio_resid > 666 VFS_TO_TMPFS(vp->v_mount)->tm_maxfilesize) 667 return (EFBIG); 668 if (vn_rlimit_fsize(vp, uio, uio->uio_td)) 669 return (EFBIG); 670 if (uio->uio_offset + uio->uio_resid > node->tn_size) { 671 error = tmpfs_reg_resize(vp, uio->uio_offset + uio->uio_resid, 672 FALSE); 673 if (error != 0) 674 goto out; 675 } 676 677 error = uiomove_object(node->tn_reg.tn_aobj, node->tn_size, uio); 678 node->tn_status |= TMPFS_NODE_MODIFIED | TMPFS_NODE_CHANGED; 679 node->tn_accessed = true; 680 if (node->tn_mode & (S_ISUID | S_ISGID)) { 681 if (priv_check_cred(v->a_cred, PRIV_VFS_RETAINSUGID)) { 682 newmode = node->tn_mode & ~(S_ISUID | S_ISGID); 683 vn_seqc_write_begin(vp); 684 atomic_store_short(&node->tn_mode, newmode); 685 vn_seqc_write_end(vp); 686 } 687 } 688 if (error != 0) 689 (void)tmpfs_reg_resize(vp, oldsize, TRUE); 690 691 out: 692 MPASS(IMPLIES(error == 0, uio->uio_resid == 0)); 693 MPASS(IMPLIES(error != 0, oldsize == node->tn_size)); 694 695 return (error); 696 } 697 698 static int 699 tmpfs_deallocate(struct vop_deallocate_args *v) 700 { 701 return (tmpfs_reg_punch_hole(v->a_vp, v->a_offset, v->a_len)); 702 } 703 704 static int 705 tmpfs_fsync(struct vop_fsync_args *v) 706 { 707 struct vnode *vp = v->a_vp; 708 709 MPASS(VOP_ISLOCKED(vp)); 710 711 tmpfs_check_mtime(vp); 712 tmpfs_update(vp); 713 714 return (0); 715 } 716 717 static int 718 tmpfs_remove(struct vop_remove_args *v) 719 { 720 struct vnode *dvp = v->a_dvp; 721 struct vnode *vp = v->a_vp; 722 723 int error; 724 struct tmpfs_dirent *de; 725 struct tmpfs_mount *tmp; 726 struct tmpfs_node *dnode; 727 struct tmpfs_node *node; 728 729 MPASS(VOP_ISLOCKED(dvp)); 730 MPASS(VOP_ISLOCKED(vp)); 731 732 if (vp->v_type == VDIR) { 733 error = EISDIR; 734 goto out; 735 } 736 737 dnode = VP_TO_TMPFS_DIR(dvp); 738 node = VP_TO_TMPFS_NODE(vp); 739 tmp = VFS_TO_TMPFS(vp->v_mount); 740 de = tmpfs_dir_lookup(dnode, node, v->a_cnp); 741 MPASS(de != NULL); 742 743 /* Files marked as immutable or append-only cannot be deleted. */ 744 if ((node->tn_flags & (IMMUTABLE | APPEND | NOUNLINK)) || 745 (dnode->tn_flags & APPEND)) { 746 error = EPERM; 747 goto out; 748 } 749 750 /* Remove the entry from the directory; as it is a file, we do not 751 * have to change the number of hard links of the directory. */ 752 tmpfs_dir_detach(dvp, de); 753 if (v->a_cnp->cn_flags & DOWHITEOUT) 754 tmpfs_dir_whiteout_add(dvp, v->a_cnp); 755 756 /* Free the directory entry we just deleted. Note that the node 757 * referred by it will not be removed until the vnode is really 758 * reclaimed. */ 759 tmpfs_free_dirent(tmp, de); 760 761 node->tn_status |= TMPFS_NODE_CHANGED; 762 node->tn_accessed = true; 763 error = 0; 764 765 out: 766 return (error); 767 } 768 769 static int 770 tmpfs_link(struct vop_link_args *v) 771 { 772 struct vnode *dvp = v->a_tdvp; 773 struct vnode *vp = v->a_vp; 774 struct componentname *cnp = v->a_cnp; 775 776 int error; 777 struct tmpfs_dirent *de; 778 struct tmpfs_node *node; 779 780 MPASS(VOP_ISLOCKED(dvp)); 781 MPASS(cnp->cn_flags & HASBUF); 782 MPASS(dvp != vp); /* XXX When can this be false? */ 783 node = VP_TO_TMPFS_NODE(vp); 784 785 /* Ensure that we do not overflow the maximum number of links imposed 786 * by the system. */ 787 MPASS(node->tn_links <= TMPFS_LINK_MAX); 788 if (node->tn_links == TMPFS_LINK_MAX) { 789 error = EMLINK; 790 goto out; 791 } 792 793 /* We cannot create links of files marked immutable or append-only. */ 794 if (node->tn_flags & (IMMUTABLE | APPEND)) { 795 error = EPERM; 796 goto out; 797 } 798 799 /* Allocate a new directory entry to represent the node. */ 800 error = tmpfs_alloc_dirent(VFS_TO_TMPFS(vp->v_mount), node, 801 cnp->cn_nameptr, cnp->cn_namelen, &de); 802 if (error != 0) 803 goto out; 804 805 /* Insert the new directory entry into the appropriate directory. */ 806 if (cnp->cn_flags & ISWHITEOUT) 807 tmpfs_dir_whiteout_remove(dvp, cnp); 808 tmpfs_dir_attach(dvp, de); 809 810 /* vp link count has changed, so update node times. */ 811 node->tn_status |= TMPFS_NODE_CHANGED; 812 tmpfs_update(vp); 813 814 error = 0; 815 816 out: 817 return (error); 818 } 819 820 /* 821 * We acquire all but fdvp locks using non-blocking acquisitions. If we 822 * fail to acquire any lock in the path we will drop all held locks, 823 * acquire the new lock in a blocking fashion, and then release it and 824 * restart the rename. This acquire/release step ensures that we do not 825 * spin on a lock waiting for release. On error release all vnode locks 826 * and decrement references the way tmpfs_rename() would do. 827 */ 828 static int 829 tmpfs_rename_relock(struct vnode *fdvp, struct vnode **fvpp, 830 struct vnode *tdvp, struct vnode **tvpp, 831 struct componentname *fcnp, struct componentname *tcnp) 832 { 833 struct vnode *nvp; 834 struct mount *mp; 835 struct tmpfs_dirent *de; 836 int error, restarts = 0; 837 838 VOP_UNLOCK(tdvp); 839 if (*tvpp != NULL && *tvpp != tdvp) 840 VOP_UNLOCK(*tvpp); 841 mp = fdvp->v_mount; 842 843 relock: 844 restarts += 1; 845 error = vn_lock(fdvp, LK_EXCLUSIVE); 846 if (error) 847 goto releout; 848 if (vn_lock(tdvp, LK_EXCLUSIVE | LK_NOWAIT) != 0) { 849 VOP_UNLOCK(fdvp); 850 error = vn_lock(tdvp, LK_EXCLUSIVE); 851 if (error) 852 goto releout; 853 VOP_UNLOCK(tdvp); 854 goto relock; 855 } 856 /* 857 * Re-resolve fvp to be certain it still exists and fetch the 858 * correct vnode. 859 */ 860 de = tmpfs_dir_lookup(VP_TO_TMPFS_DIR(fdvp), NULL, fcnp); 861 if (de == NULL) { 862 VOP_UNLOCK(fdvp); 863 VOP_UNLOCK(tdvp); 864 if ((fcnp->cn_flags & ISDOTDOT) != 0 || 865 (fcnp->cn_namelen == 1 && fcnp->cn_nameptr[0] == '.')) 866 error = EINVAL; 867 else 868 error = ENOENT; 869 goto releout; 870 } 871 error = tmpfs_alloc_vp(mp, de->td_node, LK_EXCLUSIVE | LK_NOWAIT, &nvp); 872 if (error != 0) { 873 VOP_UNLOCK(fdvp); 874 VOP_UNLOCK(tdvp); 875 if (error != EBUSY) 876 goto releout; 877 error = tmpfs_alloc_vp(mp, de->td_node, LK_EXCLUSIVE, &nvp); 878 if (error != 0) 879 goto releout; 880 VOP_UNLOCK(nvp); 881 /* 882 * Concurrent rename race. 883 */ 884 if (nvp == tdvp) { 885 vrele(nvp); 886 error = EINVAL; 887 goto releout; 888 } 889 vrele(*fvpp); 890 *fvpp = nvp; 891 goto relock; 892 } 893 vrele(*fvpp); 894 *fvpp = nvp; 895 VOP_UNLOCK(*fvpp); 896 /* 897 * Re-resolve tvp and acquire the vnode lock if present. 898 */ 899 de = tmpfs_dir_lookup(VP_TO_TMPFS_DIR(tdvp), NULL, tcnp); 900 /* 901 * If tvp disappeared we just carry on. 902 */ 903 if (de == NULL && *tvpp != NULL) { 904 vrele(*tvpp); 905 *tvpp = NULL; 906 } 907 /* 908 * Get the tvp ino if the lookup succeeded. We may have to restart 909 * if the non-blocking acquire fails. 910 */ 911 if (de != NULL) { 912 nvp = NULL; 913 error = tmpfs_alloc_vp(mp, de->td_node, 914 LK_EXCLUSIVE | LK_NOWAIT, &nvp); 915 if (*tvpp != NULL) 916 vrele(*tvpp); 917 *tvpp = nvp; 918 if (error != 0) { 919 VOP_UNLOCK(fdvp); 920 VOP_UNLOCK(tdvp); 921 if (error != EBUSY) 922 goto releout; 923 error = tmpfs_alloc_vp(mp, de->td_node, LK_EXCLUSIVE, 924 &nvp); 925 if (error != 0) 926 goto releout; 927 VOP_UNLOCK(nvp); 928 /* 929 * fdvp contains fvp, thus tvp (=fdvp) is not empty. 930 */ 931 if (nvp == fdvp) { 932 error = ENOTEMPTY; 933 goto releout; 934 } 935 goto relock; 936 } 937 } 938 tmpfs_rename_restarts += restarts; 939 940 return (0); 941 942 releout: 943 vrele(fdvp); 944 vrele(*fvpp); 945 vrele(tdvp); 946 if (*tvpp != NULL) 947 vrele(*tvpp); 948 tmpfs_rename_restarts += restarts; 949 950 return (error); 951 } 952 953 static int 954 tmpfs_rename(struct vop_rename_args *v) 955 { 956 struct vnode *fdvp = v->a_fdvp; 957 struct vnode *fvp = v->a_fvp; 958 struct componentname *fcnp = v->a_fcnp; 959 struct vnode *tdvp = v->a_tdvp; 960 struct vnode *tvp = v->a_tvp; 961 struct componentname *tcnp = v->a_tcnp; 962 char *newname; 963 struct tmpfs_dirent *de; 964 struct tmpfs_mount *tmp; 965 struct tmpfs_node *fdnode; 966 struct tmpfs_node *fnode; 967 struct tmpfs_node *tnode; 968 struct tmpfs_node *tdnode; 969 int error; 970 bool want_seqc_end; 971 972 MPASS(VOP_ISLOCKED(tdvp)); 973 MPASS(IMPLIES(tvp != NULL, VOP_ISLOCKED(tvp))); 974 MPASS(fcnp->cn_flags & HASBUF); 975 MPASS(tcnp->cn_flags & HASBUF); 976 977 want_seqc_end = false; 978 979 /* 980 * Disallow cross-device renames. 981 * XXX Why isn't this done by the caller? 982 */ 983 if (fvp->v_mount != tdvp->v_mount || 984 (tvp != NULL && fvp->v_mount != tvp->v_mount)) { 985 error = EXDEV; 986 goto out; 987 } 988 989 /* If source and target are the same file, there is nothing to do. */ 990 if (fvp == tvp) { 991 error = 0; 992 goto out; 993 } 994 995 /* 996 * If we need to move the directory between entries, lock the 997 * source so that we can safely operate on it. 998 */ 999 if (fdvp != tdvp && fdvp != tvp) { 1000 if (vn_lock(fdvp, LK_EXCLUSIVE | LK_NOWAIT) != 0) { 1001 error = tmpfs_rename_relock(fdvp, &fvp, tdvp, &tvp, 1002 fcnp, tcnp); 1003 if (error != 0) 1004 return (error); 1005 ASSERT_VOP_ELOCKED(fdvp, 1006 "tmpfs_rename: fdvp not locked"); 1007 ASSERT_VOP_ELOCKED(tdvp, 1008 "tmpfs_rename: tdvp not locked"); 1009 if (tvp != NULL) 1010 ASSERT_VOP_ELOCKED(tvp, 1011 "tmpfs_rename: tvp not locked"); 1012 if (fvp == tvp) { 1013 error = 0; 1014 goto out_locked; 1015 } 1016 } 1017 } 1018 1019 if (tvp != NULL) 1020 vn_seqc_write_begin(tvp); 1021 vn_seqc_write_begin(tdvp); 1022 vn_seqc_write_begin(fvp); 1023 vn_seqc_write_begin(fdvp); 1024 want_seqc_end = true; 1025 1026 tmp = VFS_TO_TMPFS(tdvp->v_mount); 1027 tdnode = VP_TO_TMPFS_DIR(tdvp); 1028 tnode = (tvp == NULL) ? NULL : VP_TO_TMPFS_NODE(tvp); 1029 fdnode = VP_TO_TMPFS_DIR(fdvp); 1030 fnode = VP_TO_TMPFS_NODE(fvp); 1031 de = tmpfs_dir_lookup(fdnode, fnode, fcnp); 1032 1033 /* 1034 * Entry can disappear before we lock fdvp, 1035 * also avoid manipulating '.' and '..' entries. 1036 */ 1037 if (de == NULL) { 1038 if ((fcnp->cn_flags & ISDOTDOT) != 0 || 1039 (fcnp->cn_namelen == 1 && fcnp->cn_nameptr[0] == '.')) 1040 error = EINVAL; 1041 else 1042 error = ENOENT; 1043 goto out_locked; 1044 } 1045 MPASS(de->td_node == fnode); 1046 1047 /* 1048 * If re-naming a directory to another preexisting directory 1049 * ensure that the target directory is empty so that its 1050 * removal causes no side effects. 1051 * Kern_rename guarantees the destination to be a directory 1052 * if the source is one. 1053 */ 1054 if (tvp != NULL) { 1055 MPASS(tnode != NULL); 1056 1057 if ((tnode->tn_flags & (NOUNLINK | IMMUTABLE | APPEND)) || 1058 (tdnode->tn_flags & (APPEND | IMMUTABLE))) { 1059 error = EPERM; 1060 goto out_locked; 1061 } 1062 1063 if (fnode->tn_type == VDIR && tnode->tn_type == VDIR) { 1064 if (tnode->tn_size > 0) { 1065 error = ENOTEMPTY; 1066 goto out_locked; 1067 } 1068 } else if (fnode->tn_type == VDIR && tnode->tn_type != VDIR) { 1069 error = ENOTDIR; 1070 goto out_locked; 1071 } else if (fnode->tn_type != VDIR && tnode->tn_type == VDIR) { 1072 error = EISDIR; 1073 goto out_locked; 1074 } else { 1075 MPASS(fnode->tn_type != VDIR && 1076 tnode->tn_type != VDIR); 1077 } 1078 } 1079 1080 if ((fnode->tn_flags & (NOUNLINK | IMMUTABLE | APPEND)) 1081 || (fdnode->tn_flags & (APPEND | IMMUTABLE))) { 1082 error = EPERM; 1083 goto out_locked; 1084 } 1085 1086 /* 1087 * Ensure that we have enough memory to hold the new name, if it 1088 * has to be changed. 1089 */ 1090 if (fcnp->cn_namelen != tcnp->cn_namelen || 1091 bcmp(fcnp->cn_nameptr, tcnp->cn_nameptr, fcnp->cn_namelen) != 0) { 1092 newname = malloc(tcnp->cn_namelen, M_TMPFSNAME, M_WAITOK); 1093 } else 1094 newname = NULL; 1095 1096 /* 1097 * If the node is being moved to another directory, we have to do 1098 * the move. 1099 */ 1100 if (fdnode != tdnode) { 1101 /* 1102 * In case we are moving a directory, we have to adjust its 1103 * parent to point to the new parent. 1104 */ 1105 if (de->td_node->tn_type == VDIR) { 1106 struct tmpfs_node *n; 1107 1108 /* 1109 * Ensure the target directory is not a child of the 1110 * directory being moved. Otherwise, we'd end up 1111 * with stale nodes. 1112 */ 1113 n = tdnode; 1114 /* 1115 * TMPFS_LOCK guaranties that no nodes are freed while 1116 * traversing the list. Nodes can only be marked as 1117 * removed: tn_parent == NULL. 1118 */ 1119 TMPFS_LOCK(tmp); 1120 TMPFS_NODE_LOCK(n); 1121 while (n != n->tn_dir.tn_parent) { 1122 struct tmpfs_node *parent; 1123 1124 if (n == fnode) { 1125 TMPFS_NODE_UNLOCK(n); 1126 TMPFS_UNLOCK(tmp); 1127 error = EINVAL; 1128 if (newname != NULL) 1129 free(newname, M_TMPFSNAME); 1130 goto out_locked; 1131 } 1132 parent = n->tn_dir.tn_parent; 1133 TMPFS_NODE_UNLOCK(n); 1134 if (parent == NULL) { 1135 n = NULL; 1136 break; 1137 } 1138 TMPFS_NODE_LOCK(parent); 1139 if (parent->tn_dir.tn_parent == NULL) { 1140 TMPFS_NODE_UNLOCK(parent); 1141 n = NULL; 1142 break; 1143 } 1144 n = parent; 1145 } 1146 TMPFS_UNLOCK(tmp); 1147 if (n == NULL) { 1148 error = EINVAL; 1149 if (newname != NULL) 1150 free(newname, M_TMPFSNAME); 1151 goto out_locked; 1152 } 1153 TMPFS_NODE_UNLOCK(n); 1154 1155 /* Adjust the parent pointer. */ 1156 TMPFS_VALIDATE_DIR(fnode); 1157 TMPFS_NODE_LOCK(de->td_node); 1158 de->td_node->tn_dir.tn_parent = tdnode; 1159 TMPFS_NODE_UNLOCK(de->td_node); 1160 1161 /* 1162 * As a result of changing the target of the '..' 1163 * entry, the link count of the source and target 1164 * directories has to be adjusted. 1165 */ 1166 TMPFS_NODE_LOCK(tdnode); 1167 TMPFS_ASSERT_LOCKED(tdnode); 1168 tdnode->tn_links++; 1169 TMPFS_NODE_UNLOCK(tdnode); 1170 1171 TMPFS_NODE_LOCK(fdnode); 1172 TMPFS_ASSERT_LOCKED(fdnode); 1173 fdnode->tn_links--; 1174 TMPFS_NODE_UNLOCK(fdnode); 1175 } 1176 } 1177 1178 /* 1179 * Do the move: just remove the entry from the source directory 1180 * and insert it into the target one. 1181 */ 1182 tmpfs_dir_detach(fdvp, de); 1183 1184 if (fcnp->cn_flags & DOWHITEOUT) 1185 tmpfs_dir_whiteout_add(fdvp, fcnp); 1186 if (tcnp->cn_flags & ISWHITEOUT) 1187 tmpfs_dir_whiteout_remove(tdvp, tcnp); 1188 1189 /* 1190 * If the name has changed, we need to make it effective by changing 1191 * it in the directory entry. 1192 */ 1193 if (newname != NULL) { 1194 MPASS(tcnp->cn_namelen <= MAXNAMLEN); 1195 1196 free(de->ud.td_name, M_TMPFSNAME); 1197 de->ud.td_name = newname; 1198 tmpfs_dirent_init(de, tcnp->cn_nameptr, tcnp->cn_namelen); 1199 1200 fnode->tn_status |= TMPFS_NODE_CHANGED; 1201 tdnode->tn_status |= TMPFS_NODE_MODIFIED; 1202 } 1203 1204 /* 1205 * If we are overwriting an entry, we have to remove the old one 1206 * from the target directory. 1207 */ 1208 if (tvp != NULL) { 1209 struct tmpfs_dirent *tde; 1210 1211 /* Remove the old entry from the target directory. */ 1212 tde = tmpfs_dir_lookup(tdnode, tnode, tcnp); 1213 tmpfs_dir_detach(tdvp, tde); 1214 1215 /* 1216 * Free the directory entry we just deleted. Note that the 1217 * node referred by it will not be removed until the vnode is 1218 * really reclaimed. 1219 */ 1220 tmpfs_free_dirent(VFS_TO_TMPFS(tvp->v_mount), tde); 1221 } 1222 1223 tmpfs_dir_attach(tdvp, de); 1224 1225 if (tmpfs_use_nc(fvp)) { 1226 cache_vop_rename(fdvp, fvp, tdvp, tvp, fcnp, tcnp); 1227 } 1228 1229 error = 0; 1230 1231 out_locked: 1232 if (fdvp != tdvp && fdvp != tvp) 1233 VOP_UNLOCK(fdvp); 1234 1235 out: 1236 if (want_seqc_end) { 1237 if (tvp != NULL) 1238 vn_seqc_write_end(tvp); 1239 vn_seqc_write_end(tdvp); 1240 vn_seqc_write_end(fvp); 1241 vn_seqc_write_end(fdvp); 1242 } 1243 1244 /* 1245 * Release target nodes. 1246 * XXX: I don't understand when tdvp can be the same as tvp, but 1247 * other code takes care of this... 1248 */ 1249 if (tdvp == tvp) 1250 vrele(tdvp); 1251 else 1252 vput(tdvp); 1253 if (tvp != NULL) 1254 vput(tvp); 1255 1256 /* Release source nodes. */ 1257 vrele(fdvp); 1258 vrele(fvp); 1259 1260 return (error); 1261 } 1262 1263 static int 1264 tmpfs_mkdir(struct vop_mkdir_args *v) 1265 { 1266 struct vnode *dvp = v->a_dvp; 1267 struct vnode **vpp = v->a_vpp; 1268 struct componentname *cnp = v->a_cnp; 1269 struct vattr *vap = v->a_vap; 1270 1271 MPASS(vap->va_type == VDIR); 1272 1273 return (tmpfs_alloc_file(dvp, vpp, vap, cnp, NULL)); 1274 } 1275 1276 static int 1277 tmpfs_rmdir(struct vop_rmdir_args *v) 1278 { 1279 struct vnode *dvp = v->a_dvp; 1280 struct vnode *vp = v->a_vp; 1281 1282 int error; 1283 struct tmpfs_dirent *de; 1284 struct tmpfs_mount *tmp; 1285 struct tmpfs_node *dnode; 1286 struct tmpfs_node *node; 1287 1288 MPASS(VOP_ISLOCKED(dvp)); 1289 MPASS(VOP_ISLOCKED(vp)); 1290 1291 tmp = VFS_TO_TMPFS(dvp->v_mount); 1292 dnode = VP_TO_TMPFS_DIR(dvp); 1293 node = VP_TO_TMPFS_DIR(vp); 1294 1295 /* Directories with more than two entries ('.' and '..') cannot be 1296 * removed. */ 1297 if (node->tn_size > 0) { 1298 error = ENOTEMPTY; 1299 goto out; 1300 } 1301 1302 if ((dnode->tn_flags & APPEND) 1303 || (node->tn_flags & (NOUNLINK | IMMUTABLE | APPEND))) { 1304 error = EPERM; 1305 goto out; 1306 } 1307 1308 /* This invariant holds only if we are not trying to remove "..". 1309 * We checked for that above so this is safe now. */ 1310 MPASS(node->tn_dir.tn_parent == dnode); 1311 1312 /* Get the directory entry associated with node (vp). This was 1313 * filled by tmpfs_lookup while looking up the entry. */ 1314 de = tmpfs_dir_lookup(dnode, node, v->a_cnp); 1315 MPASS(TMPFS_DIRENT_MATCHES(de, 1316 v->a_cnp->cn_nameptr, 1317 v->a_cnp->cn_namelen)); 1318 1319 /* Check flags to see if we are allowed to remove the directory. */ 1320 if ((dnode->tn_flags & APPEND) != 0 || 1321 (node->tn_flags & (NOUNLINK | IMMUTABLE | APPEND)) != 0) { 1322 error = EPERM; 1323 goto out; 1324 } 1325 1326 /* Detach the directory entry from the directory (dnode). */ 1327 tmpfs_dir_detach(dvp, de); 1328 if (v->a_cnp->cn_flags & DOWHITEOUT) 1329 tmpfs_dir_whiteout_add(dvp, v->a_cnp); 1330 1331 /* No vnode should be allocated for this entry from this point */ 1332 TMPFS_NODE_LOCK(node); 1333 node->tn_links--; 1334 node->tn_dir.tn_parent = NULL; 1335 node->tn_status |= TMPFS_NODE_CHANGED | TMPFS_NODE_MODIFIED; 1336 node->tn_accessed = true; 1337 1338 TMPFS_NODE_UNLOCK(node); 1339 1340 TMPFS_NODE_LOCK(dnode); 1341 dnode->tn_links--; 1342 dnode->tn_status |= TMPFS_NODE_CHANGED | TMPFS_NODE_MODIFIED; 1343 dnode->tn_accessed = true; 1344 TMPFS_NODE_UNLOCK(dnode); 1345 1346 if (tmpfs_use_nc(dvp)) { 1347 cache_vop_rmdir(dvp, vp); 1348 } 1349 1350 /* Free the directory entry we just deleted. Note that the node 1351 * referred by it will not be removed until the vnode is really 1352 * reclaimed. */ 1353 tmpfs_free_dirent(tmp, de); 1354 1355 /* Release the deleted vnode (will destroy the node, notify 1356 * interested parties and clean it from the cache). */ 1357 1358 dnode->tn_status |= TMPFS_NODE_CHANGED; 1359 tmpfs_update(dvp); 1360 1361 error = 0; 1362 1363 out: 1364 return (error); 1365 } 1366 1367 static int 1368 tmpfs_symlink(struct vop_symlink_args *v) 1369 { 1370 struct vnode *dvp = v->a_dvp; 1371 struct vnode **vpp = v->a_vpp; 1372 struct componentname *cnp = v->a_cnp; 1373 struct vattr *vap = v->a_vap; 1374 const char *target = v->a_target; 1375 1376 #ifdef notyet /* XXX FreeBSD BUG: kern_symlink is not setting VLNK */ 1377 MPASS(vap->va_type == VLNK); 1378 #else 1379 vap->va_type = VLNK; 1380 #endif 1381 1382 return (tmpfs_alloc_file(dvp, vpp, vap, cnp, target)); 1383 } 1384 1385 static int 1386 tmpfs_readdir(struct vop_readdir_args *va) 1387 { 1388 struct vnode *vp; 1389 struct uio *uio; 1390 struct tmpfs_mount *tm; 1391 struct tmpfs_node *node; 1392 uint64_t **cookies; 1393 int *eofflag, *ncookies; 1394 ssize_t startresid; 1395 int error, maxcookies; 1396 1397 vp = va->a_vp; 1398 uio = va->a_uio; 1399 eofflag = va->a_eofflag; 1400 cookies = va->a_cookies; 1401 ncookies = va->a_ncookies; 1402 1403 /* This operation only makes sense on directory nodes. */ 1404 if (vp->v_type != VDIR) 1405 return (ENOTDIR); 1406 1407 maxcookies = 0; 1408 node = VP_TO_TMPFS_DIR(vp); 1409 tm = VFS_TO_TMPFS(vp->v_mount); 1410 1411 startresid = uio->uio_resid; 1412 1413 /* Allocate cookies for NFS and compat modules. */ 1414 if (cookies != NULL && ncookies != NULL) { 1415 maxcookies = howmany(node->tn_size, 1416 sizeof(struct tmpfs_dirent)) + 2; 1417 *cookies = malloc(maxcookies * sizeof(**cookies), M_TEMP, 1418 M_WAITOK); 1419 *ncookies = 0; 1420 } 1421 1422 if (cookies == NULL) 1423 error = tmpfs_dir_getdents(tm, node, uio, 0, NULL, NULL); 1424 else 1425 error = tmpfs_dir_getdents(tm, node, uio, maxcookies, *cookies, 1426 ncookies); 1427 1428 /* Buffer was filled without hitting EOF. */ 1429 if (error == EJUSTRETURN) 1430 error = (uio->uio_resid != startresid) ? 0 : EINVAL; 1431 1432 if (error != 0 && cookies != NULL && ncookies != NULL) { 1433 free(*cookies, M_TEMP); 1434 *cookies = NULL; 1435 *ncookies = 0; 1436 } 1437 1438 if (eofflag != NULL) 1439 *eofflag = 1440 (error == 0 && uio->uio_offset == TMPFS_DIRCOOKIE_EOF); 1441 1442 return (error); 1443 } 1444 1445 static int 1446 tmpfs_readlink(struct vop_readlink_args *v) 1447 { 1448 struct vnode *vp = v->a_vp; 1449 struct uio *uio = v->a_uio; 1450 1451 int error; 1452 struct tmpfs_node *node; 1453 1454 MPASS(uio->uio_offset == 0); 1455 MPASS(vp->v_type == VLNK); 1456 1457 node = VP_TO_TMPFS_NODE(vp); 1458 1459 error = uiomove(node->tn_link_target, MIN(node->tn_size, uio->uio_resid), 1460 uio); 1461 tmpfs_set_accessed(VFS_TO_TMPFS(vp->v_mount), node); 1462 1463 return (error); 1464 } 1465 1466 /* 1467 * VOP_FPLOOKUP_SYMLINK routines are subject to special circumstances, see 1468 * the comment above cache_fplookup for details. 1469 * 1470 * Check tmpfs_alloc_node for tmpfs-specific synchronisation notes. 1471 */ 1472 static int 1473 tmpfs_fplookup_symlink(struct vop_fplookup_symlink_args *v) 1474 { 1475 struct vnode *vp; 1476 struct tmpfs_node *node; 1477 char *symlink; 1478 1479 vp = v->a_vp; 1480 node = VP_TO_TMPFS_NODE_SMR(vp); 1481 if (__predict_false(node == NULL)) 1482 return (EAGAIN); 1483 if (!atomic_load_char(&node->tn_link_smr)) 1484 return (EAGAIN); 1485 symlink = atomic_load_ptr(&node->tn_link_target); 1486 if (symlink == NULL) 1487 return (EAGAIN); 1488 1489 return (cache_symlink_resolve(v->a_fpl, symlink, node->tn_size)); 1490 } 1491 1492 static int 1493 tmpfs_inactive(struct vop_inactive_args *v) 1494 { 1495 struct vnode *vp; 1496 struct tmpfs_node *node; 1497 1498 vp = v->a_vp; 1499 node = VP_TO_TMPFS_NODE(vp); 1500 if (node->tn_links == 0) 1501 vrecycle(vp); 1502 else 1503 tmpfs_check_mtime(vp); 1504 return (0); 1505 } 1506 1507 static int 1508 tmpfs_need_inactive(struct vop_need_inactive_args *ap) 1509 { 1510 struct vnode *vp; 1511 struct tmpfs_node *node; 1512 struct vm_object *obj; 1513 1514 vp = ap->a_vp; 1515 node = VP_TO_TMPFS_NODE(vp); 1516 if (node->tn_links == 0) 1517 goto need; 1518 if (vp->v_type == VREG) { 1519 obj = vp->v_object; 1520 if (obj->generation != obj->cleangeneration) 1521 goto need; 1522 } 1523 return (0); 1524 need: 1525 return (1); 1526 } 1527 1528 int 1529 tmpfs_reclaim(struct vop_reclaim_args *v) 1530 { 1531 struct vnode *vp; 1532 struct tmpfs_mount *tmp; 1533 struct tmpfs_node *node; 1534 bool unlock; 1535 1536 vp = v->a_vp; 1537 node = VP_TO_TMPFS_NODE(vp); 1538 tmp = VFS_TO_TMPFS(vp->v_mount); 1539 1540 if (vp->v_type == VREG) 1541 tmpfs_destroy_vobject(vp, node->tn_reg.tn_aobj); 1542 vp->v_object = NULL; 1543 1544 TMPFS_LOCK(tmp); 1545 TMPFS_NODE_LOCK(node); 1546 tmpfs_free_vp(vp); 1547 1548 /* 1549 * If the node referenced by this vnode was deleted by the user, 1550 * we must free its associated data structures (now that the vnode 1551 * is being reclaimed). 1552 */ 1553 unlock = true; 1554 if (node->tn_links == 0 && 1555 (node->tn_vpstate & TMPFS_VNODE_ALLOCATING) == 0) { 1556 node->tn_vpstate = TMPFS_VNODE_DOOMED; 1557 unlock = !tmpfs_free_node_locked(tmp, node, true); 1558 } 1559 1560 if (unlock) { 1561 TMPFS_NODE_UNLOCK(node); 1562 TMPFS_UNLOCK(tmp); 1563 } 1564 1565 MPASS(vp->v_data == NULL); 1566 return (0); 1567 } 1568 1569 int 1570 tmpfs_print(struct vop_print_args *v) 1571 { 1572 struct vnode *vp = v->a_vp; 1573 1574 struct tmpfs_node *node; 1575 1576 node = VP_TO_TMPFS_NODE(vp); 1577 1578 printf("tag VT_TMPFS, tmpfs_node %p, flags 0x%lx, links %jd\n", 1579 node, node->tn_flags, (uintmax_t)node->tn_links); 1580 printf("\tmode 0%o, owner %d, group %d, size %jd, status 0x%x\n", 1581 node->tn_mode, node->tn_uid, node->tn_gid, 1582 (intmax_t)node->tn_size, node->tn_status); 1583 1584 if (vp->v_type == VFIFO) 1585 fifo_printinfo(vp); 1586 1587 printf("\n"); 1588 1589 return (0); 1590 } 1591 1592 int 1593 tmpfs_pathconf(struct vop_pathconf_args *v) 1594 { 1595 struct vnode *vp = v->a_vp; 1596 int name = v->a_name; 1597 long *retval = v->a_retval; 1598 1599 int error; 1600 1601 error = 0; 1602 1603 switch (name) { 1604 case _PC_LINK_MAX: 1605 *retval = TMPFS_LINK_MAX; 1606 break; 1607 1608 case _PC_SYMLINK_MAX: 1609 *retval = MAXPATHLEN; 1610 break; 1611 1612 case _PC_NAME_MAX: 1613 *retval = NAME_MAX; 1614 break; 1615 1616 case _PC_PIPE_BUF: 1617 if (vp->v_type == VDIR || vp->v_type == VFIFO) 1618 *retval = PIPE_BUF; 1619 else 1620 error = EINVAL; 1621 break; 1622 1623 case _PC_CHOWN_RESTRICTED: 1624 *retval = 1; 1625 break; 1626 1627 case _PC_NO_TRUNC: 1628 *retval = 1; 1629 break; 1630 1631 case _PC_SYNC_IO: 1632 *retval = 1; 1633 break; 1634 1635 case _PC_FILESIZEBITS: 1636 *retval = 64; 1637 break; 1638 1639 default: 1640 error = vop_stdpathconf(v); 1641 } 1642 1643 return (error); 1644 } 1645 1646 static int 1647 tmpfs_vptofh(struct vop_vptofh_args *ap) 1648 /* 1649 vop_vptofh { 1650 IN struct vnode *a_vp; 1651 IN struct fid *a_fhp; 1652 }; 1653 */ 1654 { 1655 struct tmpfs_fid_data tfd; 1656 struct tmpfs_node *node; 1657 struct fid *fhp; 1658 1659 node = VP_TO_TMPFS_NODE(ap->a_vp); 1660 fhp = ap->a_fhp; 1661 fhp->fid_len = sizeof(tfd); 1662 1663 /* 1664 * Copy into fid_data from the stack to avoid unaligned pointer use. 1665 * See the comment in sys/mount.h on struct fid for details. 1666 */ 1667 tfd.tfd_id = node->tn_id; 1668 tfd.tfd_gen = node->tn_gen; 1669 memcpy(fhp->fid_data, &tfd, fhp->fid_len); 1670 1671 return (0); 1672 } 1673 1674 static int 1675 tmpfs_whiteout(struct vop_whiteout_args *ap) 1676 { 1677 struct vnode *dvp = ap->a_dvp; 1678 struct componentname *cnp = ap->a_cnp; 1679 struct tmpfs_dirent *de; 1680 1681 switch (ap->a_flags) { 1682 case LOOKUP: 1683 return (0); 1684 case CREATE: 1685 de = tmpfs_dir_lookup(VP_TO_TMPFS_DIR(dvp), NULL, cnp); 1686 if (de != NULL) 1687 return (de->td_node == NULL ? 0 : EEXIST); 1688 return (tmpfs_dir_whiteout_add(dvp, cnp)); 1689 case DELETE: 1690 tmpfs_dir_whiteout_remove(dvp, cnp); 1691 return (0); 1692 default: 1693 panic("tmpfs_whiteout: unknown op"); 1694 } 1695 } 1696 1697 static int 1698 tmpfs_vptocnp_dir(struct tmpfs_node *tn, struct tmpfs_node *tnp, 1699 struct tmpfs_dirent **pde) 1700 { 1701 struct tmpfs_dir_cursor dc; 1702 struct tmpfs_dirent *de; 1703 1704 for (de = tmpfs_dir_first(tnp, &dc); de != NULL; 1705 de = tmpfs_dir_next(tnp, &dc)) { 1706 if (de->td_node == tn) { 1707 *pde = de; 1708 return (0); 1709 } 1710 } 1711 return (ENOENT); 1712 } 1713 1714 static int 1715 tmpfs_vptocnp_fill(struct vnode *vp, struct tmpfs_node *tn, 1716 struct tmpfs_node *tnp, char *buf, size_t *buflen, struct vnode **dvp) 1717 { 1718 struct tmpfs_dirent *de; 1719 int error, i; 1720 1721 error = vn_vget_ino_gen(vp, tmpfs_vn_get_ino_alloc, tnp, LK_SHARED, 1722 dvp); 1723 if (error != 0) 1724 return (error); 1725 error = tmpfs_vptocnp_dir(tn, tnp, &de); 1726 if (error == 0) { 1727 i = *buflen; 1728 i -= de->td_namelen; 1729 if (i < 0) { 1730 error = ENOMEM; 1731 } else { 1732 bcopy(de->ud.td_name, buf + i, de->td_namelen); 1733 *buflen = i; 1734 } 1735 } 1736 if (error == 0) { 1737 if (vp != *dvp) 1738 VOP_UNLOCK(*dvp); 1739 } else { 1740 if (vp != *dvp) 1741 vput(*dvp); 1742 else 1743 vrele(vp); 1744 } 1745 return (error); 1746 } 1747 1748 static int 1749 tmpfs_vptocnp(struct vop_vptocnp_args *ap) 1750 { 1751 struct vnode *vp, **dvp; 1752 struct tmpfs_node *tn, *tnp, *tnp1; 1753 struct tmpfs_dirent *de; 1754 struct tmpfs_mount *tm; 1755 char *buf; 1756 size_t *buflen; 1757 int error; 1758 1759 vp = ap->a_vp; 1760 dvp = ap->a_vpp; 1761 buf = ap->a_buf; 1762 buflen = ap->a_buflen; 1763 1764 tm = VFS_TO_TMPFS(vp->v_mount); 1765 tn = VP_TO_TMPFS_NODE(vp); 1766 if (tn->tn_type == VDIR) { 1767 tnp = tn->tn_dir.tn_parent; 1768 if (tnp == NULL) 1769 return (ENOENT); 1770 tmpfs_ref_node(tnp); 1771 error = tmpfs_vptocnp_fill(vp, tn, tn->tn_dir.tn_parent, buf, 1772 buflen, dvp); 1773 tmpfs_free_node(tm, tnp); 1774 return (error); 1775 } 1776 restart: 1777 TMPFS_LOCK(tm); 1778 restart_locked: 1779 LIST_FOREACH_SAFE(tnp, &tm->tm_nodes_used, tn_entries, tnp1) { 1780 if (tnp->tn_type != VDIR) 1781 continue; 1782 TMPFS_NODE_LOCK(tnp); 1783 tmpfs_ref_node(tnp); 1784 1785 /* 1786 * tn_vnode cannot be instantiated while we hold the 1787 * node lock, so the directory cannot be changed while 1788 * we iterate over it. Do this to avoid instantiating 1789 * vnode for directories which cannot point to our 1790 * node. 1791 */ 1792 error = tnp->tn_vnode == NULL ? tmpfs_vptocnp_dir(tn, tnp, 1793 &de) : 0; 1794 1795 if (error == 0) { 1796 TMPFS_NODE_UNLOCK(tnp); 1797 TMPFS_UNLOCK(tm); 1798 error = tmpfs_vptocnp_fill(vp, tn, tnp, buf, buflen, 1799 dvp); 1800 if (error == 0) { 1801 tmpfs_free_node(tm, tnp); 1802 return (0); 1803 } 1804 if (VN_IS_DOOMED(vp)) { 1805 tmpfs_free_node(tm, tnp); 1806 return (ENOENT); 1807 } 1808 TMPFS_LOCK(tm); 1809 TMPFS_NODE_LOCK(tnp); 1810 } 1811 if (tmpfs_free_node_locked(tm, tnp, false)) { 1812 goto restart; 1813 } else { 1814 KASSERT(tnp->tn_refcount > 0, 1815 ("node %p refcount zero", tnp)); 1816 if (tnp->tn_attached) { 1817 tnp1 = LIST_NEXT(tnp, tn_entries); 1818 TMPFS_NODE_UNLOCK(tnp); 1819 } else { 1820 TMPFS_NODE_UNLOCK(tnp); 1821 goto restart_locked; 1822 } 1823 } 1824 } 1825 TMPFS_UNLOCK(tm); 1826 return (ENOENT); 1827 } 1828 1829 /* 1830 * Vnode operations vector used for files stored in a tmpfs file system. 1831 */ 1832 struct vop_vector tmpfs_vnodeop_entries = { 1833 .vop_default = &default_vnodeops, 1834 .vop_lookup = vfs_cache_lookup, 1835 .vop_cachedlookup = tmpfs_cached_lookup, 1836 .vop_create = tmpfs_create, 1837 .vop_mknod = tmpfs_mknod, 1838 .vop_open = tmpfs_open, 1839 .vop_close = tmpfs_close, 1840 .vop_fplookup_vexec = tmpfs_fplookup_vexec, 1841 .vop_fplookup_symlink = tmpfs_fplookup_symlink, 1842 .vop_access = tmpfs_access, 1843 .vop_stat = tmpfs_stat, 1844 .vop_getattr = tmpfs_getattr, 1845 .vop_setattr = tmpfs_setattr, 1846 .vop_read = tmpfs_read, 1847 .vop_read_pgcache = tmpfs_read_pgcache, 1848 .vop_write = tmpfs_write, 1849 .vop_deallocate = tmpfs_deallocate, 1850 .vop_fsync = tmpfs_fsync, 1851 .vop_remove = tmpfs_remove, 1852 .vop_link = tmpfs_link, 1853 .vop_rename = tmpfs_rename, 1854 .vop_mkdir = tmpfs_mkdir, 1855 .vop_rmdir = tmpfs_rmdir, 1856 .vop_symlink = tmpfs_symlink, 1857 .vop_readdir = tmpfs_readdir, 1858 .vop_readlink = tmpfs_readlink, 1859 .vop_inactive = tmpfs_inactive, 1860 .vop_need_inactive = tmpfs_need_inactive, 1861 .vop_reclaim = tmpfs_reclaim, 1862 .vop_print = tmpfs_print, 1863 .vop_pathconf = tmpfs_pathconf, 1864 .vop_vptofh = tmpfs_vptofh, 1865 .vop_whiteout = tmpfs_whiteout, 1866 .vop_bmap = VOP_EOPNOTSUPP, 1867 .vop_vptocnp = tmpfs_vptocnp, 1868 .vop_lock1 = vop_lock, 1869 .vop_unlock = vop_unlock, 1870 .vop_islocked = vop_islocked, 1871 .vop_add_writecount = vop_stdadd_writecount_nomsync, 1872 }; 1873 VFS_VOP_VECTOR_REGISTER(tmpfs_vnodeop_entries); 1874 1875 /* 1876 * Same vector for mounts which do not use namecache. 1877 */ 1878 struct vop_vector tmpfs_vnodeop_nonc_entries = { 1879 .vop_default = &tmpfs_vnodeop_entries, 1880 .vop_lookup = tmpfs_lookup, 1881 }; 1882 VFS_VOP_VECTOR_REGISTER(tmpfs_vnodeop_nonc_entries); 1883