1 /* $NetBSD: tmpfs.h,v 1.26 2007/02/22 06:37:00 thorpej Exp $ */ 2 3 /*- 4 * Copyright (c) 2005, 2006 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 * $FreeBSD$ 33 */ 34 35 #ifndef _FS_TMPFS_TMPFS_H_ 36 #define _FS_TMPFS_TMPFS_H_ 37 38 #include <sys/dirent.h> 39 #include <sys/mount.h> 40 #include <sys/queue.h> 41 #include <sys/vnode.h> 42 #include <sys/file.h> 43 #include <sys/lock.h> 44 #include <sys/mutex.h> 45 46 #include <sys/malloc.h> 47 #include <sys/systm.h> 48 #include <sys/tree.h> 49 #include <sys/vmmeter.h> 50 #include <vm/swap_pager.h> 51 52 MALLOC_DECLARE(M_TMPFSMNT); 53 MALLOC_DECLARE(M_TMPFSNAME); 54 55 /* 56 * Internal representation of a tmpfs directory entry. 57 */ 58 59 LIST_HEAD(tmpfs_dir_duphead, tmpfs_dirent); 60 61 struct tmpfs_dirent { 62 /* 63 * Depending on td_cookie flag entry can be of 3 types: 64 * - regular -- no hash collisions, stored in RB-Tree 65 * - duphead -- synthetic linked list head for dup entries 66 * - dup -- stored in linked list instead of RB-Tree 67 */ 68 union { 69 /* regular and duphead entry types */ 70 RB_ENTRY(tmpfs_dirent) td_entries; 71 72 /* dup entry type */ 73 struct { 74 LIST_ENTRY(tmpfs_dirent) entries; 75 LIST_ENTRY(tmpfs_dirent) index_entries; 76 } td_dup; 77 } uh; 78 79 uint32_t td_cookie; 80 uint32_t td_hash; 81 u_int td_namelen; 82 83 /* Pointer to the node this entry refers to. In case this field 84 * is NULL, the node is a whiteout. */ 85 struct tmpfs_node * td_node; 86 87 union { 88 /* 89 * The name of the entry, allocated from a string pool. This 90 * string is not required to be zero-terminated. 91 */ 92 char * td_name; /* regular, dup */ 93 struct tmpfs_dir_duphead td_duphead; /* duphead */ 94 } ud; 95 }; 96 97 /* A directory in tmpfs holds a list of directory entries, which in 98 * turn point to other files (which can be directories themselves). 99 * 100 * In tmpfs, this list is managed by a RB-Tree, whose head is defined by 101 * the struct tmpfs_dir type. 102 * 103 * It is important to notice that directories do not have entries for . and 104 * .. as other file systems do. These can be generated when requested 105 * based on information available by other means, such as the pointer to 106 * the node itself in the former case or the pointer to the parent directory 107 * in the latter case. This is done to simplify tmpfs's code and, more 108 * importantly, to remove redundancy. */ 109 RB_HEAD(tmpfs_dir, tmpfs_dirent); 110 111 /* Each entry in a directory has a cookie that identifies it. Cookies 112 * supersede offsets within directories because, given how tmpfs stores 113 * directories in memory, there is no such thing as an offset. 114 * 115 * The '.', '..' and the end of directory markers have fixed cookies which 116 * cannot collide with the cookies generated by other entries. The cookies 117 * for the other entries are generated based on the file name hash value or 118 * unique number in case of name hash collision. 119 * 120 * To preserve compatibility cookies are limited to 31 bits. 121 */ 122 123 #define TMPFS_DIRCOOKIE_DOT 0 124 #define TMPFS_DIRCOOKIE_DOTDOT 1 125 #define TMPFS_DIRCOOKIE_EOF 2 126 #define TMPFS_DIRCOOKIE_MASK ((off_t)0x3fffffffU) 127 #define TMPFS_DIRCOOKIE_MIN ((off_t)0x00000004U) 128 #define TMPFS_DIRCOOKIE_DUP ((off_t)0x40000000U) 129 #define TMPFS_DIRCOOKIE_DUPHEAD ((off_t)0x80000000U) 130 #define TMPFS_DIRCOOKIE_DUP_MIN TMPFS_DIRCOOKIE_DUP 131 #define TMPFS_DIRCOOKIE_DUP_MAX \ 132 (TMPFS_DIRCOOKIE_DUP | TMPFS_DIRCOOKIE_MASK) 133 134 /* 135 * Internal representation of a tmpfs file system node. 136 * 137 * This structure is splitted in two parts: one holds attributes common 138 * to all file types and the other holds data that is only applicable to 139 * a particular type. The code must be careful to only access those 140 * attributes that are actually allowed by the node's type. 141 * 142 * 143 * Below is the key of locks used to protected the fields in the following 144 * structures. 145 * 146 */ 147 struct tmpfs_node { 148 /* Doubly-linked list entry which links all existing nodes for a 149 * single file system. This is provided to ease the removal of 150 * all nodes during the unmount operation. */ 151 LIST_ENTRY(tmpfs_node) tn_entries; 152 153 /* The node's type. Any of 'VBLK', 'VCHR', 'VDIR', 'VFIFO', 154 * 'VLNK', 'VREG' and 'VSOCK' is allowed. The usage of vnode 155 * types instead of a custom enumeration is to make things simpler 156 * and faster, as we do not need to convert between two types. */ 157 enum vtype tn_type; 158 159 /* Node identifier. */ 160 ino_t tn_id; 161 162 /* Node's internal status. This is used by several file system 163 * operations to do modifications to the node in a delayed 164 * fashion. */ 165 int tn_status; 166 #define TMPFS_NODE_ACCESSED (1 << 1) 167 #define TMPFS_NODE_MODIFIED (1 << 2) 168 #define TMPFS_NODE_CHANGED (1 << 3) 169 170 /* The node size. It does not necessarily match the real amount 171 * of memory consumed by it. */ 172 off_t tn_size; 173 174 /* Generic node attributes. */ 175 uid_t tn_uid; 176 gid_t tn_gid; 177 mode_t tn_mode; 178 u_long tn_flags; 179 nlink_t tn_links; 180 struct timespec tn_atime; 181 struct timespec tn_mtime; 182 struct timespec tn_ctime; 183 struct timespec tn_birthtime; 184 unsigned long tn_gen; 185 186 /* As there is a single vnode for each active file within the 187 * system, care has to be taken to avoid allocating more than one 188 * vnode per file. In order to do this, a bidirectional association 189 * is kept between vnodes and nodes. 190 * 191 * Whenever a vnode is allocated, its v_data field is updated to 192 * point to the node it references. At the same time, the node's 193 * tn_vnode field is modified to point to the new vnode representing 194 * it. Further attempts to allocate a vnode for this same node will 195 * result in returning a new reference to the value stored in 196 * tn_vnode. 197 * 198 * May be NULL when the node is unused (that is, no vnode has been 199 * allocated for it or it has been reclaimed). */ 200 struct vnode * tn_vnode; 201 202 /* interlock to protect tn_vpstate */ 203 struct mtx tn_interlock; 204 205 /* Identify if current node has vnode assiocate with 206 * or allocating vnode. 207 */ 208 int tn_vpstate; 209 210 /* misc data field for different tn_type node */ 211 union { 212 /* Valid when tn_type == VBLK || tn_type == VCHR. */ 213 dev_t tn_rdev; 214 215 /* Valid when tn_type == VDIR. */ 216 struct tn_dir { 217 /* Pointer to the parent directory. The root 218 * directory has a pointer to itself in this field; 219 * this property identifies the root node. */ 220 struct tmpfs_node * tn_parent; 221 222 /* Head of a tree that links the contents of 223 * the directory together. */ 224 struct tmpfs_dir tn_dirhead; 225 226 /* Head of a list the contains fake directory entries 227 * heads, i.e. entries with TMPFS_DIRCOOKIE_DUPHEAD 228 * flag. */ 229 struct tmpfs_dir_duphead tn_dupindex; 230 231 /* Number and pointer of the first directory entry 232 * returned by the readdir operation if it were 233 * called again to continue reading data from the 234 * same directory as before. This is used to speed 235 * up reads of long directories, assuming that no 236 * more than one read is in progress at a given time. 237 * Otherwise, these values are discarded. */ 238 off_t tn_readdir_lastn; 239 struct tmpfs_dirent * tn_readdir_lastp; 240 } tn_dir; 241 242 /* Valid when tn_type == VLNK. */ 243 /* The link's target, allocated from a string pool. */ 244 char * tn_link; 245 246 /* Valid when tn_type == VREG. */ 247 struct tn_reg { 248 /* The contents of regular files stored in a tmpfs 249 * file system are represented by a single anonymous 250 * memory object (aobj, for short). The aobj provides 251 * direct access to any position within the file, 252 * because its contents are always mapped in a 253 * contiguous region of virtual memory. It is a task 254 * of the memory management subsystem (see uvm(9)) to 255 * issue the required page ins or page outs whenever 256 * a position within the file is accessed. */ 257 vm_object_t tn_aobj; 258 259 }tn_reg; 260 261 /* Valid when tn_type = VFIFO */ 262 struct tn_fifo { 263 fo_rdwr_t *tn_fo_read; 264 fo_rdwr_t *tn_fo_write; 265 }tn_fifo; 266 }tn_spec; 267 }; 268 LIST_HEAD(tmpfs_node_list, tmpfs_node); 269 270 #define tn_rdev tn_spec.tn_rdev 271 #define tn_dir tn_spec.tn_dir 272 #define tn_link tn_spec.tn_link 273 #define tn_reg tn_spec.tn_reg 274 #define tn_fifo tn_spec.tn_fifo 275 276 #define TMPFS_NODE_LOCK(node) mtx_lock(&(node)->tn_interlock) 277 #define TMPFS_NODE_UNLOCK(node) mtx_unlock(&(node)->tn_interlock) 278 #define TMPFS_NODE_MTX(node) (&(node)->tn_interlock) 279 #define TMPFS_NODE_ASSERT_LOCKED(node) mtx_assert(TMPFS_NODE_MTX(node), \ 280 MA_OWNED) 281 282 #ifdef INVARIANTS 283 #define TMPFS_ASSERT_LOCKED(node) do { \ 284 MPASS(node != NULL); \ 285 MPASS(node->tn_vnode != NULL); \ 286 if (!VOP_ISLOCKED(node->tn_vnode) && \ 287 !mtx_owned(TMPFS_NODE_MTX(node))) \ 288 panic("tmpfs: node is not locked: %p", node); \ 289 } while (0) 290 #define TMPFS_ASSERT_ELOCKED(node) do { \ 291 MPASS((node) != NULL); \ 292 MPASS((node)->tn_vnode != NULL); \ 293 mtx_assert(TMPFS_NODE_MTX(node), MA_OWNED); \ 294 ASSERT_VOP_LOCKED((node)->tn_vnode, "tmpfs"); \ 295 } while (0) 296 #else 297 #define TMPFS_ASSERT_LOCKED(node) (void)0 298 #define TMPFS_ASSERT_ELOCKED(node) (void)0 299 #endif 300 301 #define TMPFS_VNODE_ALLOCATING 1 302 #define TMPFS_VNODE_WANT 2 303 #define TMPFS_VNODE_DOOMED 4 304 #define TMPFS_VNODE_WRECLAIM 8 305 306 /* 307 * Internal representation of a tmpfs mount point. 308 */ 309 struct tmpfs_mount { 310 /* Maximum number of memory pages available for use by the file 311 * system, set during mount time. This variable must never be 312 * used directly as it may be bigger than the current amount of 313 * free memory; in the extreme case, it will hold the SIZE_MAX 314 * value. */ 315 size_t tm_pages_max; 316 317 /* Number of pages in use by the file system. */ 318 size_t tm_pages_used; 319 320 /* Pointer to the node representing the root directory of this 321 * file system. */ 322 struct tmpfs_node * tm_root; 323 324 /* Maximum number of possible nodes for this file system; set 325 * during mount time. We need a hard limit on the maximum number 326 * of nodes to avoid allocating too much of them; their objects 327 * cannot be released until the file system is unmounted. 328 * Otherwise, we could easily run out of memory by creating lots 329 * of empty files and then simply removing them. */ 330 ino_t tm_nodes_max; 331 332 /* unrhdr used to allocate inode numbers */ 333 struct unrhdr * tm_ino_unr; 334 335 /* Number of nodes currently that are in use. */ 336 ino_t tm_nodes_inuse; 337 338 /* maximum representable file size */ 339 u_int64_t tm_maxfilesize; 340 341 /* Nodes are organized in two different lists. The used list 342 * contains all nodes that are currently used by the file system; 343 * i.e., they refer to existing files. The available list contains 344 * all nodes that are currently available for use by new files. 345 * Nodes must be kept in this list (instead of deleting them) 346 * because we need to keep track of their generation number (tn_gen 347 * field). 348 * 349 * Note that nodes are lazily allocated: if the available list is 350 * empty and we have enough space to create more nodes, they will be 351 * created and inserted in the used list. Once these are released, 352 * they will go into the available list, remaining alive until the 353 * file system is unmounted. */ 354 struct tmpfs_node_list tm_nodes_used; 355 356 /* All node lock to protect the node list and tmp_pages_used */ 357 struct mtx allnode_lock; 358 359 /* Pools used to store file system meta data. These are not shared 360 * across several instances of tmpfs for the reasons described in 361 * tmpfs_pool.c. */ 362 uma_zone_t tm_dirent_pool; 363 uma_zone_t tm_node_pool; 364 365 /* Read-only status. */ 366 int tm_ronly; 367 }; 368 #define TMPFS_LOCK(tm) mtx_lock(&(tm)->allnode_lock) 369 #define TMPFS_UNLOCK(tm) mtx_unlock(&(tm)->allnode_lock) 370 371 /* 372 * This structure maps a file identifier to a tmpfs node. Used by the 373 * NFS code. 374 */ 375 struct tmpfs_fid { 376 uint16_t tf_len; 377 uint16_t tf_pad; 378 ino_t tf_id; 379 unsigned long tf_gen; 380 }; 381 382 #ifdef _KERNEL 383 /* 384 * Prototypes for tmpfs_subr.c. 385 */ 386 387 int tmpfs_alloc_node(struct mount *mp, struct tmpfs_mount *, enum vtype, 388 uid_t uid, gid_t gid, mode_t mode, struct tmpfs_node *, 389 char *, dev_t, struct tmpfs_node **); 390 void tmpfs_free_node(struct tmpfs_mount *, struct tmpfs_node *); 391 int tmpfs_alloc_dirent(struct tmpfs_mount *, struct tmpfs_node *, 392 const char *, u_int, struct tmpfs_dirent **); 393 void tmpfs_free_dirent(struct tmpfs_mount *, struct tmpfs_dirent *); 394 void tmpfs_dirent_init(struct tmpfs_dirent *, const char *, u_int); 395 void tmpfs_destroy_vobject(struct vnode *vp, vm_object_t obj); 396 int tmpfs_alloc_vp(struct mount *, struct tmpfs_node *, int, 397 struct vnode **); 398 void tmpfs_free_vp(struct vnode *); 399 int tmpfs_alloc_file(struct vnode *, struct vnode **, struct vattr *, 400 struct componentname *, char *); 401 void tmpfs_check_mtime(struct vnode *); 402 void tmpfs_dir_attach(struct vnode *, struct tmpfs_dirent *); 403 void tmpfs_dir_detach(struct vnode *, struct tmpfs_dirent *); 404 void tmpfs_dir_destroy(struct tmpfs_mount *, struct tmpfs_node *); 405 struct tmpfs_dirent * tmpfs_dir_lookup(struct tmpfs_node *node, 406 struct tmpfs_node *f, 407 struct componentname *cnp); 408 int tmpfs_dir_getdents(struct tmpfs_node *, struct uio *, int, 409 u_long *, int *); 410 int tmpfs_dir_whiteout_add(struct vnode *, struct componentname *); 411 void tmpfs_dir_whiteout_remove(struct vnode *, struct componentname *); 412 int tmpfs_reg_resize(struct vnode *, off_t, boolean_t); 413 int tmpfs_chflags(struct vnode *, u_long, struct ucred *, struct thread *); 414 int tmpfs_chmod(struct vnode *, mode_t, struct ucred *, struct thread *); 415 int tmpfs_chown(struct vnode *, uid_t, gid_t, struct ucred *, 416 struct thread *); 417 int tmpfs_chsize(struct vnode *, u_quad_t, struct ucred *, struct thread *); 418 int tmpfs_chtimes(struct vnode *, struct vattr *, struct ucred *cred, 419 struct thread *); 420 void tmpfs_itimes(struct vnode *, const struct timespec *, 421 const struct timespec *); 422 423 void tmpfs_update(struct vnode *); 424 int tmpfs_truncate(struct vnode *, off_t); 425 426 /* 427 * Convenience macros to simplify some logical expressions. 428 */ 429 #define IMPLIES(a, b) (!(a) || (b)) 430 #define IFF(a, b) (IMPLIES(a, b) && IMPLIES(b, a)) 431 432 /* 433 * Checks that the directory entry pointed by 'de' matches the name 'name' 434 * with a length of 'len'. 435 */ 436 #define TMPFS_DIRENT_MATCHES(de, name, len) \ 437 (de->td_namelen == len && \ 438 bcmp((de)->ud.td_name, (name), (de)->td_namelen) == 0) 439 440 /* 441 * Ensures that the node pointed by 'node' is a directory and that its 442 * contents are consistent with respect to directories. 443 */ 444 #define TMPFS_VALIDATE_DIR(node) do { \ 445 MPASS((node)->tn_type == VDIR); \ 446 MPASS((node)->tn_size % sizeof(struct tmpfs_dirent) == 0); \ 447 } while (0) 448 449 /* 450 * Memory management stuff. 451 */ 452 453 /* 454 * Amount of memory pages to reserve for the system (e.g., to not use by 455 * tmpfs). 456 */ 457 #define TMPFS_PAGES_MINRESERVED (4 * 1024 * 1024 / PAGE_SIZE) 458 459 size_t tmpfs_mem_avail(void); 460 461 size_t tmpfs_pages_used(struct tmpfs_mount *tmp); 462 463 #endif 464 465 /* 466 * Macros/functions to convert from generic data structures to tmpfs 467 * specific ones. 468 */ 469 470 static inline 471 struct tmpfs_mount * 472 VFS_TO_TMPFS(struct mount *mp) 473 { 474 struct tmpfs_mount *tmp; 475 476 MPASS((mp) != NULL && (mp)->mnt_data != NULL); 477 tmp = (struct tmpfs_mount *)(mp)->mnt_data; 478 return tmp; 479 } 480 481 static inline 482 struct tmpfs_node * 483 VP_TO_TMPFS_NODE(struct vnode *vp) 484 { 485 struct tmpfs_node *node; 486 487 MPASS((vp) != NULL && (vp)->v_data != NULL); 488 node = (struct tmpfs_node *)vp->v_data; 489 return node; 490 } 491 492 static inline 493 struct tmpfs_node * 494 VP_TO_TMPFS_DIR(struct vnode *vp) 495 { 496 struct tmpfs_node *node; 497 498 node = VP_TO_TMPFS_NODE(vp); 499 TMPFS_VALIDATE_DIR(node); 500 return node; 501 } 502 503 #endif /* _FS_TMPFS_TMPFS_H_ */ 504