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, and tn_status under shared 203 * vnode lock. 204 */ 205 struct mtx tn_interlock; 206 207 /* Identify if current node has vnode assiocate with 208 * or allocating vnode. 209 */ 210 int tn_vpstate; 211 212 /* misc data field for different tn_type node */ 213 union { 214 /* Valid when tn_type == VBLK || tn_type == VCHR. */ 215 dev_t tn_rdev; 216 217 /* Valid when tn_type == VDIR. */ 218 struct tn_dir { 219 /* Pointer to the parent directory. The root 220 * directory has a pointer to itself in this field; 221 * this property identifies the root node. */ 222 struct tmpfs_node * tn_parent; 223 224 /* Head of a tree that links the contents of 225 * the directory together. */ 226 struct tmpfs_dir tn_dirhead; 227 228 /* Head of a list the contains fake directory entries 229 * heads, i.e. entries with TMPFS_DIRCOOKIE_DUPHEAD 230 * flag. */ 231 struct tmpfs_dir_duphead tn_dupindex; 232 233 /* Number and pointer of the first directory entry 234 * returned by the readdir operation if it were 235 * called again to continue reading data from the 236 * same directory as before. This is used to speed 237 * up reads of long directories, assuming that no 238 * more than one read is in progress at a given time. 239 * Otherwise, these values are discarded. */ 240 off_t tn_readdir_lastn; 241 struct tmpfs_dirent * tn_readdir_lastp; 242 } tn_dir; 243 244 /* Valid when tn_type == VLNK. */ 245 /* The link's target, allocated from a string pool. */ 246 char * tn_link; 247 248 /* Valid when tn_type == VREG. */ 249 struct tn_reg { 250 /* The contents of regular files stored in a tmpfs 251 * file system are represented by a single anonymous 252 * memory object (aobj, for short). The aobj provides 253 * direct access to any position within the file, 254 * because its contents are always mapped in a 255 * contiguous region of virtual memory. It is a task 256 * of the memory management subsystem (see uvm(9)) to 257 * issue the required page ins or page outs whenever 258 * a position within the file is accessed. */ 259 vm_object_t tn_aobj; 260 261 }tn_reg; 262 263 /* Valid when tn_type = VFIFO */ 264 struct tn_fifo { 265 fo_rdwr_t *tn_fo_read; 266 fo_rdwr_t *tn_fo_write; 267 }tn_fifo; 268 }tn_spec; 269 }; 270 LIST_HEAD(tmpfs_node_list, tmpfs_node); 271 272 #define tn_rdev tn_spec.tn_rdev 273 #define tn_dir tn_spec.tn_dir 274 #define tn_link tn_spec.tn_link 275 #define tn_reg tn_spec.tn_reg 276 #define tn_fifo tn_spec.tn_fifo 277 278 #define TMPFS_NODE_LOCK(node) mtx_lock(&(node)->tn_interlock) 279 #define TMPFS_NODE_UNLOCK(node) mtx_unlock(&(node)->tn_interlock) 280 #define TMPFS_NODE_MTX(node) (&(node)->tn_interlock) 281 #define TMPFS_NODE_ASSERT_LOCKED(node) mtx_assert(TMPFS_NODE_MTX(node), \ 282 MA_OWNED) 283 284 #ifdef INVARIANTS 285 #define TMPFS_ASSERT_LOCKED(node) do { \ 286 MPASS(node != NULL); \ 287 MPASS(node->tn_vnode != NULL); \ 288 if (!VOP_ISLOCKED(node->tn_vnode) && \ 289 !mtx_owned(TMPFS_NODE_MTX(node))) \ 290 panic("tmpfs: node is not locked: %p", node); \ 291 } while (0) 292 #define TMPFS_ASSERT_ELOCKED(node) do { \ 293 MPASS((node) != NULL); \ 294 MPASS((node)->tn_vnode != NULL); \ 295 mtx_assert(TMPFS_NODE_MTX(node), MA_OWNED); \ 296 ASSERT_VOP_LOCKED((node)->tn_vnode, "tmpfs"); \ 297 } while (0) 298 #else 299 #define TMPFS_ASSERT_LOCKED(node) (void)0 300 #define TMPFS_ASSERT_ELOCKED(node) (void)0 301 #endif 302 303 #define TMPFS_VNODE_ALLOCATING 1 304 #define TMPFS_VNODE_WANT 2 305 #define TMPFS_VNODE_DOOMED 4 306 #define TMPFS_VNODE_WRECLAIM 8 307 308 /* 309 * Internal representation of a tmpfs mount point. 310 */ 311 struct tmpfs_mount { 312 /* Maximum number of memory pages available for use by the file 313 * system, set during mount time. This variable must never be 314 * used directly as it may be bigger than the current amount of 315 * free memory; in the extreme case, it will hold the ULONG_MAX 316 * value. */ 317 u_long tm_pages_max; 318 319 /* Number of pages in use by the file system. */ 320 u_long tm_pages_used; 321 322 /* Pointer to the node representing the root directory of this 323 * file system. */ 324 struct tmpfs_node * tm_root; 325 326 /* Maximum number of possible nodes for this file system; set 327 * during mount time. We need a hard limit on the maximum number 328 * of nodes to avoid allocating too much of them; their objects 329 * cannot be released until the file system is unmounted. 330 * Otherwise, we could easily run out of memory by creating lots 331 * of empty files and then simply removing them. */ 332 ino_t tm_nodes_max; 333 334 /* unrhdr used to allocate inode numbers */ 335 struct unrhdr * tm_ino_unr; 336 337 /* Number of nodes currently that are in use. */ 338 ino_t tm_nodes_inuse; 339 340 /* maximum representable file size */ 341 u_int64_t tm_maxfilesize; 342 343 /* Nodes are organized in two different lists. The used list 344 * contains all nodes that are currently used by the file system; 345 * i.e., they refer to existing files. The available list contains 346 * all nodes that are currently available for use by new files. 347 * Nodes must be kept in this list (instead of deleting them) 348 * because we need to keep track of their generation number (tn_gen 349 * field). 350 * 351 * Note that nodes are lazily allocated: if the available list is 352 * empty and we have enough space to create more nodes, they will be 353 * created and inserted in the used list. Once these are released, 354 * they will go into the available list, remaining alive until the 355 * file system is unmounted. */ 356 struct tmpfs_node_list tm_nodes_used; 357 358 /* All node lock to protect the node list and tmp_pages_used */ 359 struct mtx allnode_lock; 360 361 /* Pools used to store file system meta data. These are not shared 362 * across several instances of tmpfs for the reasons described in 363 * tmpfs_pool.c. */ 364 uma_zone_t tm_dirent_pool; 365 uma_zone_t tm_node_pool; 366 367 /* Read-only status. */ 368 int tm_ronly; 369 }; 370 #define TMPFS_LOCK(tm) mtx_lock(&(tm)->allnode_lock) 371 #define TMPFS_UNLOCK(tm) mtx_unlock(&(tm)->allnode_lock) 372 373 /* 374 * This structure maps a file identifier to a tmpfs node. Used by the 375 * NFS code. 376 */ 377 struct tmpfs_fid { 378 uint16_t tf_len; 379 uint16_t tf_pad; 380 ino_t tf_id; 381 unsigned long tf_gen; 382 }; 383 384 #ifdef _KERNEL 385 /* 386 * Prototypes for tmpfs_subr.c. 387 */ 388 389 int tmpfs_alloc_node(struct mount *mp, struct tmpfs_mount *, enum vtype, 390 uid_t uid, gid_t gid, mode_t mode, struct tmpfs_node *, 391 char *, dev_t, struct tmpfs_node **); 392 void tmpfs_free_node(struct tmpfs_mount *, struct tmpfs_node *); 393 int tmpfs_alloc_dirent(struct tmpfs_mount *, struct tmpfs_node *, 394 const char *, u_int, struct tmpfs_dirent **); 395 void tmpfs_free_dirent(struct tmpfs_mount *, struct tmpfs_dirent *); 396 void tmpfs_dirent_init(struct tmpfs_dirent *, const char *, u_int); 397 void tmpfs_destroy_vobject(struct vnode *vp, vm_object_t obj); 398 int tmpfs_alloc_vp(struct mount *, struct tmpfs_node *, int, 399 struct vnode **); 400 void tmpfs_free_vp(struct vnode *); 401 int tmpfs_alloc_file(struct vnode *, struct vnode **, struct vattr *, 402 struct componentname *, char *); 403 void tmpfs_check_mtime(struct vnode *); 404 void tmpfs_dir_attach(struct vnode *, struct tmpfs_dirent *); 405 void tmpfs_dir_detach(struct vnode *, struct tmpfs_dirent *); 406 void tmpfs_dir_destroy(struct tmpfs_mount *, struct tmpfs_node *); 407 struct tmpfs_dirent * tmpfs_dir_lookup(struct tmpfs_node *node, 408 struct tmpfs_node *f, 409 struct componentname *cnp); 410 int tmpfs_dir_getdents(struct tmpfs_node *, struct uio *, int, 411 u_long *, int *); 412 int tmpfs_dir_whiteout_add(struct vnode *, struct componentname *); 413 void tmpfs_dir_whiteout_remove(struct vnode *, struct componentname *); 414 int tmpfs_reg_resize(struct vnode *, off_t, boolean_t); 415 int tmpfs_chflags(struct vnode *, u_long, struct ucred *, struct thread *); 416 int tmpfs_chmod(struct vnode *, mode_t, struct ucred *, struct thread *); 417 int tmpfs_chown(struct vnode *, uid_t, gid_t, struct ucred *, 418 struct thread *); 419 int tmpfs_chsize(struct vnode *, u_quad_t, struct ucred *, struct thread *); 420 int tmpfs_chtimes(struct vnode *, struct vattr *, struct ucred *cred, 421 struct thread *); 422 void tmpfs_itimes(struct vnode *, const struct timespec *, 423 const struct timespec *); 424 425 void tmpfs_set_status(struct tmpfs_node *node, int status); 426 void tmpfs_update(struct vnode *); 427 int tmpfs_truncate(struct vnode *, off_t); 428 429 /* 430 * Convenience macros to simplify some logical expressions. 431 */ 432 #define IMPLIES(a, b) (!(a) || (b)) 433 #define IFF(a, b) (IMPLIES(a, b) && IMPLIES(b, a)) 434 435 /* 436 * Checks that the directory entry pointed by 'de' matches the name 'name' 437 * with a length of 'len'. 438 */ 439 #define TMPFS_DIRENT_MATCHES(de, name, len) \ 440 (de->td_namelen == len && \ 441 bcmp((de)->ud.td_name, (name), (de)->td_namelen) == 0) 442 443 /* 444 * Ensures that the node pointed by 'node' is a directory and that its 445 * contents are consistent with respect to directories. 446 */ 447 #define TMPFS_VALIDATE_DIR(node) do { \ 448 MPASS((node)->tn_type == VDIR); \ 449 MPASS((node)->tn_size % sizeof(struct tmpfs_dirent) == 0); \ 450 } while (0) 451 452 /* 453 * Memory management stuff. 454 */ 455 456 /* 457 * Amount of memory pages to reserve for the system (e.g., to not use by 458 * tmpfs). 459 */ 460 #define TMPFS_PAGES_MINRESERVED (4 * 1024 * 1024 / PAGE_SIZE) 461 462 size_t tmpfs_mem_avail(void); 463 464 size_t tmpfs_pages_used(struct tmpfs_mount *tmp); 465 466 #endif 467 468 /* 469 * Macros/functions to convert from generic data structures to tmpfs 470 * specific ones. 471 */ 472 473 static inline 474 struct tmpfs_mount * 475 VFS_TO_TMPFS(struct mount *mp) 476 { 477 struct tmpfs_mount *tmp; 478 479 MPASS((mp) != NULL && (mp)->mnt_data != NULL); 480 tmp = (struct tmpfs_mount *)(mp)->mnt_data; 481 return tmp; 482 } 483 484 static inline 485 struct tmpfs_node * 486 VP_TO_TMPFS_NODE(struct vnode *vp) 487 { 488 struct tmpfs_node *node; 489 490 MPASS((vp) != NULL && (vp)->v_data != NULL); 491 node = (struct tmpfs_node *)vp->v_data; 492 return node; 493 } 494 495 static inline 496 struct tmpfs_node * 497 VP_TO_TMPFS_DIR(struct vnode *vp) 498 { 499 struct tmpfs_node *node; 500 501 node = VP_TO_TMPFS_NODE(vp); 502 TMPFS_VALIDATE_DIR(node); 503 return node; 504 } 505 506 #endif /* _FS_TMPFS_TMPFS_H_ */ 507