1 /* 2 * Copyright (c) 2003,2004 The DragonFly Project. All rights reserved. 3 * 4 * This code is derived from software contributed to The DragonFly Project 5 * by Matthew Dillon <dillon@backplane.com> 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 11 * 1. Redistributions of source code must retain the above copyright 12 * notice, this list of conditions and the following disclaimer. 13 * 2. Redistributions in binary form must reproduce the above copyright 14 * notice, this list of conditions and the following disclaimer in 15 * the documentation and/or other materials provided with the 16 * distribution. 17 * 3. Neither the name of The DragonFly Project nor the names of its 18 * contributors may be used to endorse or promote products derived 19 * from this software without specific, prior written permission. 20 * 21 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS 22 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT 23 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS 24 * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE 25 * COPYRIGHT HOLDERS OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, 26 * INCIDENTAL, SPECIAL, EXEMPLARY OR CONSEQUENTIAL DAMAGES (INCLUDING, 27 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; 28 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED 29 * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, 30 * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT 31 * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 32 * SUCH DAMAGE. 33 * 34 * 35 * Copyright (c) 1989, 1993 36 * The Regents of the University of California. All rights reserved. 37 * 38 * This code is derived from software contributed 39 * to Berkeley by John Heidemann of the UCLA Ficus project. 40 * 41 * Source: * @(#)i405_init.c 2.10 92/04/27 UCLA Ficus project 42 * 43 * Redistribution and use in source and binary forms, with or without 44 * modification, are permitted provided that the following conditions 45 * are met: 46 * 1. Redistributions of source code must retain the above copyright 47 * notice, this list of conditions and the following disclaimer. 48 * 2. Redistributions in binary form must reproduce the above copyright 49 * notice, this list of conditions and the following disclaimer in the 50 * documentation and/or other materials provided with the distribution. 51 * 3. All advertising materials mentioning features or use of this software 52 * must display the following acknowledgement: 53 * This product includes software developed by the University of 54 * California, Berkeley and its contributors. 55 * 4. Neither the name of the University nor the names of its contributors 56 * may be used to endorse or promote products derived from this software 57 * without specific prior written permission. 58 * 59 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 60 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 61 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 62 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 63 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 64 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 65 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 66 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 67 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 68 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 69 * SUCH DAMAGE. 70 * 71 * @(#)vfs_init.c 8.3 (Berkeley) 1/4/94 72 * $FreeBSD: src/sys/kern/vfs_init.c,v 1.59 2002/04/30 18:44:32 dillon Exp $ 73 * $DragonFly: src/sys/kern/vfs_init.c,v 1.11 2006/03/24 18:35:33 dillon Exp $ 74 */ 75 /* 76 * Manage vnode VOP operations vectors 77 */ 78 #include <sys/param.h> 79 #include <sys/systm.h> 80 #include <sys/kernel.h> 81 #include <sys/mount.h> 82 #include <sys/sysctl.h> 83 #include <sys/vnode.h> 84 #include <sys/malloc.h> 85 #include <vm/vm_zone.h> 86 87 static MALLOC_DEFINE(M_VNODEOP, "vnodeops", "vnode operations vectors"); 88 89 /* 90 * Zone for namei 91 */ 92 struct vm_zone *namei_zone; 93 94 /* 95 * vfs_init() will set maxvfsconf 96 * to the highest defined type number. 97 */ 98 int maxvfsconf; 99 struct vfsconf *vfsconf; 100 101 static TAILQ_HEAD(, vnodeopv_node) vnodeopv_list; 102 static void vfs_recalc_vnodeops(void); 103 104 /* 105 * Add a vnode operations (vnops) vector to the global list. 106 */ 107 void 108 vfs_add_vnodeops_sysinit(const void *data) 109 { 110 const struct vnodeopv_desc *vdesc = data; 111 112 vfs_add_vnodeops(NULL, vdesc->opv_desc_vector, 113 vdesc->opv_desc_ops, vdesc->opv_flags); 114 } 115 116 /* 117 * Unlink previously added vnode operations vector. 118 */ 119 void 120 vfs_rm_vnodeops_sysinit(const void *data) 121 { 122 const struct vnodeopv_desc *vdesc = data; 123 124 vfs_rm_vnodeops(vdesc->opv_desc_vector); 125 } 126 127 void 128 vfs_add_vnodeops(struct mount *mp, struct vop_ops **vops_pp, 129 struct vnodeopv_entry_desc *descs, int flags) 130 { 131 struct vnodeopv_node *node; 132 struct vop_ops *ops; 133 134 node = malloc(sizeof(*node), M_VNODEOP, M_ZERO|M_WAITOK); 135 KKASSERT(*vops_pp == NULL); 136 if ((ops = *vops_pp) == NULL) { 137 ops = malloc(sizeof(struct vop_ops), 138 M_VNODEOP, M_ZERO|M_WAITOK); 139 *vops_pp = ops; 140 } 141 node->ops = ops; 142 node->descs = descs; 143 ops->vv_mount = mp; 144 ops->vv_flags |= flags; 145 146 /* 147 * Journal and coherency ops inherit normal ops flags 148 */ 149 if (vops_pp == &mp->mnt_vn_coherency_ops && mp->mnt_vn_norm_ops) 150 ops->vv_flags |= mp->mnt_vn_norm_ops->vv_flags; 151 if (vops_pp == &mp->mnt_vn_journal_ops && mp->mnt_vn_norm_ops) 152 ops->vv_flags |= mp->mnt_vn_norm_ops->vv_flags; 153 154 ++ops->vv_refs; 155 TAILQ_INSERT_TAIL(&vnodeopv_list, node, entry); 156 157 vfs_recalc_vnodeops(); 158 159 if (mp) { 160 if (mp->mnt_vn_coherency_ops) 161 mp->mnt_vn_use_ops = mp->mnt_vn_coherency_ops; 162 else if (mp->mnt_vn_journal_ops) 163 mp->mnt_vn_use_ops = mp->mnt_vn_journal_ops; 164 else 165 mp->mnt_vn_use_ops = mp->mnt_vn_norm_ops; 166 } 167 } 168 169 void 170 vfs_rm_vnodeops(struct vop_ops **vops_pp) 171 { 172 struct vop_ops *ops = *vops_pp; 173 struct vnodeopv_node *node; 174 struct mount *mp; 175 176 if (ops == NULL) 177 return; 178 179 TAILQ_FOREACH(node, &vnodeopv_list, entry) { 180 if (node->ops == ops) 181 break; 182 } 183 if (node == NULL) { 184 printf("vfs_rm_vnodeops: unable to find ops: %p\n", ops); 185 return; 186 } 187 TAILQ_REMOVE(&vnodeopv_list, node, entry); 188 free(node, M_VNODEOP); 189 KKASSERT(ops != NULL && ops->vv_refs > 0); 190 if (--ops->vv_refs == 0) { 191 *vops_pp = NULL; 192 if ((mp = ops->vv_mount) != NULL) { 193 if (mp->mnt_vn_coherency_ops) 194 mp->mnt_vn_use_ops = mp->mnt_vn_coherency_ops; 195 else if (mp->mnt_vn_journal_ops) 196 mp->mnt_vn_use_ops = mp->mnt_vn_journal_ops; 197 else 198 mp->mnt_vn_use_ops = mp->mnt_vn_norm_ops; 199 } 200 free(ops, M_VNODEOP); 201 } 202 vfs_recalc_vnodeops(); 203 } 204 205 /* 206 * Recalculate VFS operations vectors 207 */ 208 static void 209 vfs_recalc_vnodeops(void) 210 { 211 struct vnodeopv_node *node; 212 struct vnodeopv_entry_desc *desc; 213 struct vop_ops *ops; 214 struct vop_ops *vnew; 215 int off; 216 217 /* 218 * Because vop_ops may be active we can't just blow them away, we 219 * have to generate new vop_ops and then copy them into the running 220 * vop_ops. Any missing entries will be assigned to the default 221 * entry. If the default entry itself is missing it will be assigned 222 * to vop_eopnotsupp. 223 */ 224 TAILQ_FOREACH(node, &vnodeopv_list, entry) { 225 ops = node->ops; 226 if ((vnew = ops->vv_new) == NULL) { 227 vnew = malloc(sizeof(struct vop_ops), 228 M_VNODEOP, M_ZERO|M_WAITOK); 229 ops->vv_new = vnew; 230 vnew->vop_default = vop_eopnotsupp; 231 } 232 for (desc = node->descs; desc->opve_op; ++desc) { 233 off = desc->opve_op->vdesc_offset; 234 *(void **)((char *)vnew + off) = desc->opve_func; 235 } 236 for (off = __offsetof(struct vop_ops, vop_ops_first_field); 237 off <= __offsetof(struct vop_ops, vop_ops_last_field); 238 off += sizeof(void **) 239 ) { 240 if (*(void **)((char *)vnew + off) == NULL) 241 *(void **)((char *)vnew + off) = vnew->vop_default; 242 } 243 } 244 245 /* 246 * Copy the temporary ops into the running configuration and then 247 * delete them. 248 */ 249 TAILQ_FOREACH(node, &vnodeopv_list, entry) { 250 ops = node->ops; 251 if ((vnew = ops->vv_new) == NULL) 252 continue; 253 for (off = __offsetof(struct vop_ops, vop_ops_first_field); 254 off <= __offsetof(struct vop_ops, vop_ops_last_field); 255 off += sizeof(void **) 256 ) { 257 *(void **)((char *)ops + off) = 258 *(void **)((char *)vnew + off); 259 } 260 ops->vv_new = NULL; 261 free(vnew, M_VNODEOP); 262 } 263 } 264 265 /* 266 * Routines having to do with the management of the vnode table. 267 */ 268 struct vattr va_null; 269 270 /* 271 * Initialize the vnode structures and initialize each file system type. 272 */ 273 /* ARGSUSED*/ 274 static void 275 vfsinit(void *dummy) 276 { 277 TAILQ_INIT(&vnodeopv_list); 278 namei_zone = zinit("NAMEI", MAXPATHLEN, 0, 0, 2); 279 280 /* 281 * Initialize the vnode table 282 */ 283 vfs_subr_init(); 284 vfs_mount_init(); 285 vfs_lock_init(); 286 vfs_sync_init(); 287 /* 288 * Initialize the vnode name cache 289 */ 290 nchinit(); 291 /* 292 * Initialize each file system type. 293 * Vfs type numbers must be distinct from VFS_GENERIC (and VFS_VFSCONF). 294 */ 295 vattr_null(&va_null); 296 maxvfsconf = VFS_GENERIC + 1; 297 } 298 SYSINIT(vfs, SI_SUB_VFS, SI_ORDER_FIRST, vfsinit, NULL) 299 300 /* 301 * Register a VFS. 302 * 303 * After doing general initialisation, this function will 304 * call the filesystem specific initialisation vector op, 305 * i.e. vfsops->vfs_init(). 306 */ 307 int 308 vfs_register(struct vfsconf *vfc) 309 { 310 struct sysctl_oid *oidp; 311 struct vfsconf *vfsp; 312 struct vfsops *vfsops = NULL; 313 314 vfsp = NULL; 315 if (vfsconf) 316 for (vfsp = vfsconf; vfsp->vfc_next; vfsp = vfsp->vfc_next) 317 if (strcmp(vfc->vfc_name, vfsp->vfc_name) == 0) 318 return EEXIST; 319 320 vfc->vfc_typenum = maxvfsconf++; 321 if (vfsp) 322 vfsp->vfc_next = vfc; 323 else 324 vfsconf = vfc; 325 vfc->vfc_next = NULL; 326 327 /* 328 * If this filesystem has a sysctl node under vfs 329 * (i.e. vfs.xxfs), then change the oid number of that node to 330 * match the filesystem's type number. This allows user code 331 * which uses the type number to read sysctl variables defined 332 * by the filesystem to continue working. Since the oids are 333 * in a sorted list, we need to make sure the order is 334 * preserved by re-registering the oid after modifying its 335 * number. 336 */ 337 SLIST_FOREACH(oidp, &sysctl__vfs_children, oid_link) 338 if (strcmp(oidp->oid_name, vfc->vfc_name) == 0) { 339 sysctl_unregister_oid(oidp); 340 oidp->oid_number = vfc->vfc_typenum; 341 sysctl_register_oid(oidp); 342 } 343 344 /* 345 * Initialise unused fields in the file system's vfsops vector. 346 * 347 * NOTE the file system should provide the mount and unmount ops 348 * at the least. In order for unmount to succeed, we also need 349 * the file system to provide us with vfsops->vfs_root otherwise 350 * the unmount(2) operation will not succeed. 351 */ 352 vfsops = vfc->vfc_vfsops; 353 KKASSERT(vfc->vfc_vfsops != NULL); 354 KKASSERT(vfsops->vfs_mount != NULL); 355 KKASSERT(vfsops->vfs_root != NULL); 356 KKASSERT(vfsops->vfs_unmount != NULL); 357 358 if (vfsops->vfs_root == NULL) { 359 /* return file system's root vnode */ 360 vfsops->vfs_root = vfs_stdroot; 361 } 362 if (vfsops->vfs_start == NULL) { 363 /* 364 * Make file system operational before first use. This 365 * routine is called at mount-time for initialising MFS, 366 * not used by other file systems. 367 */ 368 vfsops->vfs_start = vfs_stdstart; 369 } 370 if (vfsops->vfs_quotactl == NULL) { 371 /* quota control */ 372 vfsops->vfs_quotactl = vfs_stdquotactl; 373 } 374 if (vfsops->vfs_statfs == NULL) { 375 /* return file system's status */ 376 vfsops->vfs_statfs = vfs_stdstatfs; 377 } 378 if (vfsops->vfs_sync == NULL) { 379 /* 380 * Flush dirty buffers. File systems can use vfs_stdsync() 381 * by explicitly setting it in the vfsops->vfs_sync vector 382 * entry. 383 */ 384 vfsops->vfs_sync = vfs_stdnosync; 385 } 386 if (vfsops->vfs_vget == NULL) { 387 /* convert an inode number to a vnode */ 388 vfsops->vfs_vget = vfs_stdvget; 389 } 390 if (vfsops->vfs_fhtovp == NULL) { 391 /* turn an NFS file handle into a vnode */ 392 vfsops->vfs_fhtovp = vfs_stdfhtovp; 393 } 394 if (vfsops->vfs_checkexp == NULL) { 395 /* check if file system is exported */ 396 vfsops->vfs_checkexp = vfs_stdcheckexp; 397 } 398 if (vfsops->vfs_vptofh == NULL) { 399 /* turn a vnode into an NFS file handle */ 400 vfsops->vfs_vptofh = vfs_stdvptofh; 401 } 402 if (vfsops->vfs_init == NULL) { 403 /* file system specific initialisation */ 404 vfsops->vfs_init = vfs_stdinit; 405 } 406 if (vfsops->vfs_uninit == NULL) { 407 /* file system specific uninitialisation */ 408 vfsops->vfs_uninit = vfs_stduninit; 409 } 410 if (vfsops->vfs_extattrctl == NULL) { 411 /* extended attribute control */ 412 vfsops->vfs_extattrctl = vfs_stdextattrctl; 413 } 414 415 /* 416 * Call init function for this VFS... 417 */ 418 (*(vfc->vfc_vfsops->vfs_init))(vfc); 419 420 return 0; 421 } 422 423 424 /* 425 * Remove previously registered VFS. 426 * 427 * After doing general de-registration like removing sysctl 428 * nodes etc, it will call the filesystem specific vector 429 * op, i.e. vfsops->vfs_uninit(). 430 * 431 */ 432 int 433 vfs_unregister(struct vfsconf *vfc) 434 { 435 struct vfsconf *vfsp, *prev_vfsp; 436 int error, i, maxtypenum; 437 438 i = vfc->vfc_typenum; 439 440 prev_vfsp = NULL; 441 for (vfsp = vfsconf; vfsp; 442 prev_vfsp = vfsp, vfsp = vfsp->vfc_next) { 443 if (!strcmp(vfc->vfc_name, vfsp->vfc_name)) 444 break; 445 } 446 if (vfsp == NULL) 447 return EINVAL; 448 if (vfsp->vfc_refcount) 449 return EBUSY; 450 if (vfc->vfc_vfsops->vfs_uninit != NULL) { 451 error = (*vfc->vfc_vfsops->vfs_uninit)(vfsp); 452 if (error) 453 return (error); 454 } 455 if (prev_vfsp) 456 prev_vfsp->vfc_next = vfsp->vfc_next; 457 else 458 vfsconf = vfsp->vfc_next; 459 maxtypenum = VFS_GENERIC; 460 for (vfsp = vfsconf; vfsp != NULL; vfsp = vfsp->vfc_next) 461 if (maxtypenum < vfsp->vfc_typenum) 462 maxtypenum = vfsp->vfc_typenum; 463 maxvfsconf = maxtypenum + 1; 464 return 0; 465 } 466 467 int 468 vfs_modevent(module_t mod, int type, void *data) 469 { 470 struct vfsconf *vfc; 471 int error = 0; 472 473 vfc = (struct vfsconf *)data; 474 475 switch (type) { 476 case MOD_LOAD: 477 if (vfc) 478 error = vfs_register(vfc); 479 break; 480 481 case MOD_UNLOAD: 482 if (vfc) 483 error = vfs_unregister(vfc); 484 break; 485 default: /* including MOD_SHUTDOWN */ 486 break; 487 } 488 return (error); 489 } 490