sys/kern/vfs_init.c

   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 1.15 2008/06/01 19:27:35 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 <sys/objcache.h>
  86
  87 static MALLOC_DEFINE(M_VNODEOP, "vnodeops", "vnode operations vectors");
  88 static MALLOC_DEFINE(M_NAMEI, "nameibufs", "namei path buffers");
  89
  90 /*
  91  * Zone for namei
  92  */
  93 struct objcache *namei_oc;
  94
  95 static TAILQ_HEAD(, vnodeopv_node) vnodeopv_list;
  96 static void vfs_calc_vnodeops(struct vop_ops *ops);
  97
  98
  99 /*
 100  * Add a vnode operations (vnops) vector to the global list.
 101  */
 102 void
 103 vfs_nadd_vnodeops_sysinit(void *data)
 104 {
 105         struct vop_ops *ops = data;
 106
 107         vfs_add_vnodeops(NULL, ops, NULL);      /* mount, template, newcopy */
 108 }
 109
 110 /*
 111  * Unlink previously added vnode operations vector.
 112  */
 113 void
 114 vfs_nrm_vnodeops_sysinit(void *data)
 115 {
 116         struct vop_ops *ops = data;
 117
 118         vfs_rm_vnodeops(NULL, ops, NULL);
 119 }
 120
 121 void
 122 vfs_add_vnodeops(struct mount *mp, struct vop_ops *template,
 123                  struct vop_ops **ops_pp)
 124 {
 125         struct vop_ops *ops;
 126
 127         if (ops_pp) {
 128                 KKASSERT(*ops_pp == NULL);
 129                 *ops_pp = kmalloc(sizeof(*ops), M_VNODEOP, M_WAITOK);
 130                 ops = *ops_pp;
 131                 bcopy(template, ops, sizeof(*ops));
 132         } else {
 133                 ops = template;
 134         }
 135
 136         vfs_calc_vnodeops(ops);
 137         ops->head.vv_mount = mp;
 138
 139         if (mp) {
 140                 if (mp->mnt_vn_coherency_ops)
 141                         mp->mnt_vn_use_ops = mp->mnt_vn_coherency_ops;
 142                 else if (mp->mnt_vn_journal_ops)
 143                         mp->mnt_vn_use_ops = mp->mnt_vn_journal_ops;
 144                 else
 145                         mp->mnt_vn_use_ops = mp->mnt_vn_norm_ops;
 146         }
 147 }
 148
 149 /*
 150  * Remove a previously installed operations vector.
 151  *
 152  * NOTE: Either template or ops_pp may be NULL, but not both.
 153  */
 154 void
 155 vfs_rm_vnodeops(struct mount *mp, struct vop_ops *template,
 156                 struct vop_ops **ops_pp)
 157 {
 158         struct vop_ops *ops;
 159
 160         if (ops_pp) {
 161                 ops = *ops_pp;
 162                 *ops_pp = NULL;
 163         } else {
 164                 ops = template;
 165         }
 166         if (ops == NULL)
 167                 return;
 168         KKASSERT(mp == ops->head.vv_mount);
 169         if (mp) {
 170                 if (mp->mnt_vn_coherency_ops)
 171                         mp->mnt_vn_use_ops = mp->mnt_vn_coherency_ops;
 172                 else if (mp->mnt_vn_journal_ops)
 173                         mp->mnt_vn_use_ops = mp->mnt_vn_journal_ops;
 174                 else
 175                         mp->mnt_vn_use_ops = mp->mnt_vn_norm_ops;
 176         }
 177         if (ops_pp)
 178                 kfree(ops, M_VNODEOP);
 179 }
 180
 181 /*
 182  * Calculate the VFS operations vector array.  This function basically
 183  * replaces any NULL entry with the default entry.
 184  */
 185 static void
 186 vfs_calc_vnodeops(struct vop_ops *ops)
 187 {
 188         int off;
 189
 190         for (off = __offsetof(struct vop_ops, vop_ops_first_field);
 191              off <= __offsetof(struct vop_ops, vop_ops_last_field);
 192              off += sizeof(void *)
 193         ) {
 194                 if (*(void **)((char *)ops + off) == NULL)
 195                     *(void **)((char *)ops + off) = ops->vop_default;
 196         }
 197 }
 198
 199 /*
 200  * Routines having to do with the management of the vnode table.
 201  */
 202 struct vattr va_null;
 203
 204 /*
 205  * Initialize the vnode structures and initialize each file system type.
 206  */
 207 /* ARGSUSED*/
 208 static void
 209 vfsinit(void *dummy)
 210 {
 211         TAILQ_INIT(&vnodeopv_list);
 212         namei_oc = objcache_create_simple(M_NAMEI, MAXPATHLEN);
 213
 214         /*
 215          * Initialize the vnode table
 216          */
 217         vfs_subr_init();
 218         vfs_mount_init();
 219         vfs_lock_init();
 220         vfs_sync_init();
 221         /*
 222          * Initialize the vnode name cache
 223          */
 224         nchinit();
 225         /*
 226          * Initialize each file system type.
 227          * Vfs type numbers must be distinct from VFS_GENERIC (and VFS_VFSCONF).
 228          */
 229         vattr_null(&va_null);
 230 }
 231 SYSINIT(vfs, SI_SUB_VFS, SI_ORDER_FIRST, vfsinit, NULL)
 232
 233 /*
 234  * vfsconf related functions/data.
 235  */
 236
 237 /* highest defined filesystem type */
 238 static int vfsconf_maxtypenum = VFS_GENERIC + 1;
 239
 240 /* head of list of filesystem types */
 241 static STAILQ_HEAD(, vfsconf) vfsconf_list =
 242         STAILQ_HEAD_INITIALIZER(vfsconf_list);
 243
 244 struct vfsconf *
 245 vfsconf_find_by_name(const char *name)
 246 {
 247         struct vfsconf *vfsp;
 248
 249         STAILQ_FOREACH(vfsp, &vfsconf_list, vfc_next) {
 250                 if (strcmp(name, vfsp->vfc_name) == 0)
 251                         break;
 252         }
 253         return vfsp;
 254 }
 255
 256 struct vfsconf *
 257 vfsconf_find_by_typenum(int typenum)
 258 {
 259         struct vfsconf *vfsp;
 260
 261         STAILQ_FOREACH(vfsp, &vfsconf_list, vfc_next) {
 262                 if (typenum == vfsp->vfc_typenum)
 263                         break;
 264         }
 265         return vfsp;
 266 }
 267
 268 static void
 269 vfsconf_add(struct vfsconf *vfc)
 270 {
 271         vfc->vfc_typenum = vfsconf_maxtypenum++;
 272         STAILQ_INSERT_TAIL(&vfsconf_list, vfc, vfc_next);
 273 }
 274
 275 static void
 276 vfsconf_remove(struct vfsconf *vfc)
 277 {
 278         int maxtypenum;
 279
 280         STAILQ_REMOVE(&vfsconf_list, vfc, vfsconf, vfc_next);
 281
 282         maxtypenum = VFS_GENERIC;
 283         STAILQ_FOREACH(vfc, &vfsconf_list, vfc_next) {
 284                 if (maxtypenum < vfc->vfc_typenum)
 285                         maxtypenum = vfc->vfc_typenum;
 286         }
 287         vfsconf_maxtypenum = maxtypenum + 1;
 288 }
 289
 290 int
 291 vfsconf_get_maxtypenum(void)
 292 {
 293         return vfsconf_maxtypenum;
 294 }
 295
 296 /*
 297  * Iterate over all vfsconf entries. Break out of the iterator
 298  * by returning != 0.
 299  */
 300 int
 301 vfsconf_each(int (*iter)(struct vfsconf *element, void *data), void *data)
 302 {
 303         int error;
 304         struct vfsconf *vfsp;
 305
 306         STAILQ_FOREACH(vfsp, &vfsconf_list, vfc_next) {
 307                 error = iter(vfsp, data);
 308                 if (error)
 309                         return (error);
 310         }
 311         return (0);
 312 }
 313
 314 /*
 315  * Register a VFS.
 316  *
 317  * After doing general initialisation, this function will
 318  * call the filesystem specific initialisation vector op,
 319  * i.e. vfsops->vfs_init().
 320  */
 321 int
 322 vfs_register(struct vfsconf *vfc)
 323 {
 324         struct sysctl_oid *oidp;
 325         struct vfsops *vfsops = NULL;
 326
 327         if (vfsconf_find_by_name(vfc->vfc_name) != NULL)
 328                 return EEXIST;
 329
 330         vfsconf_add(vfc);
 331
 332         /*
 333          * If this filesystem has a sysctl node under vfs
 334          * (i.e. vfs.xxfs), then change the oid number of that node to
 335          * match the filesystem's type number.  This allows user code
 336          * which uses the type number to read sysctl variables defined
 337          * by the filesystem to continue working. Since the oids are
 338          * in a sorted list, we need to make sure the order is
 339          * preserved by re-registering the oid after modifying its
 340          * number.
 341          */
 342         SLIST_FOREACH(oidp, &sysctl__vfs_children, oid_link)
 343                 if (strcmp(oidp->oid_name, vfc->vfc_name) == 0) {
 344                         sysctl_unregister_oid(oidp);
 345                         oidp->oid_number = vfc->vfc_typenum;
 346                         sysctl_register_oid(oidp);
 347                 }
 348
 349         /*
 350          * Initialise unused fields in the file system's vfsops vector.
 351          *
 352          * NOTE the file system should provide the mount and unmount ops
 353          * at the least.  In order for unmount to succeed, we also need
 354          * the file system to provide us with vfsops->vfs_root otherwise
 355          * the unmount(2) operation will not succeed.
 356          */
 357         vfsops = vfc->vfc_vfsops;
 358         KKASSERT(vfc->vfc_vfsops != NULL);
 359         KKASSERT(vfsops->vfs_mount != NULL);
 360         KKASSERT(vfsops->vfs_root != NULL);
 361         KKASSERT(vfsops->vfs_unmount != NULL);
 362
 363         if (vfsops->vfs_root == NULL) {
 364                 /* return file system's root vnode */
 365                 vfsops->vfs_root = vfs_stdroot;
 366         }
 367         if (vfsops->vfs_start == NULL) {
 368                 /*
 369                  * Make file system operational before first use.  This
 370                  * routine is called at mount-time for initialising MFS,
 371                  * not used by other file systems.
 372                  */
 373                 vfsops->vfs_start = vfs_stdstart;
 374         }
 375         if (vfsops->vfs_quotactl == NULL) {
 376                 /* quota control */
 377                 vfsops->vfs_quotactl = vfs_stdquotactl;
 378         }
 379         if (vfsops->vfs_statfs == NULL) {
 380                 /* return file system's status */
 381                 vfsops->vfs_statfs = vfs_stdstatfs;
 382         }
 383         if (vfsops->vfs_statvfs == NULL) {
 384                 /* return file system's status */
 385                 vfsops->vfs_statvfs = vfs_stdstatvfs;
 386         }
 387         if (vfsops->vfs_sync == NULL) {
 388                 /*
 389                  * Flush dirty buffers.  File systems can use vfs_stdsync()
 390                  * by explicitly setting it in the vfsops->vfs_sync vector
 391                  * entry.
 392                  */
 393                 vfsops->vfs_sync = vfs_stdnosync;
 394         }
 395         if (vfsops->vfs_vget == NULL) {
 396                 /* convert an inode number to a vnode */
 397                 vfsops->vfs_vget = vfs_stdvget;
 398         }
 399         if (vfsops->vfs_fhtovp == NULL) {
 400                 /* turn an NFS file handle into a vnode */
 401                 vfsops->vfs_fhtovp = vfs_stdfhtovp;
 402         }
 403         if (vfsops->vfs_checkexp == NULL) {
 404                 /* check if file system is exported */
 405                 vfsops->vfs_checkexp = vfs_stdcheckexp;
 406         }
 407         if (vfsops->vfs_vptofh == NULL) {
 408                 /* turn a vnode into an NFS file handle */
 409                 vfsops->vfs_vptofh = vfs_stdvptofh;
 410         }
 411         if (vfsops->vfs_init == NULL) {
 412                 /* file system specific initialisation */
 413                 vfsops->vfs_init = vfs_stdinit;
 414         }
 415         if (vfsops->vfs_uninit == NULL) {
 416                 /* file system specific uninitialisation */
 417                 vfsops->vfs_uninit = vfs_stduninit;
 418         }
 419         if (vfsops->vfs_extattrctl == NULL) {
 420                 /* extended attribute control */
 421                 vfsops->vfs_extattrctl = vfs_stdextattrctl;
 422         }
 423
 424         /*
 425          * Call init function for this VFS...
 426          */
 427         (*(vfc->vfc_vfsops->vfs_init))(vfc);
 428
 429         return 0;
 430 }
 431
 432
 433 /*
 434  * Remove previously registered VFS.
 435  *
 436  * After doing general de-registration like removing sysctl
 437  * nodes etc, it will call the filesystem specific vector
 438  * op, i.e. vfsops->vfs_uninit().
 439  *
 440  */
 441 int
 442 vfs_unregister(struct vfsconf *vfc)
 443 {
 444         struct vfsconf *vfsp;
 445         int error;
 446
 447         vfsp = vfsconf_find_by_name(vfc->vfc_name);
 448
 449         if (vfsp == NULL)
 450                 return EINVAL;
 451
 452         if (vfsp->vfc_refcount != 0)
 453                 return EBUSY;
 454
 455         if (vfc->vfc_vfsops->vfs_uninit != NULL) {
 456                 error = (*vfc->vfc_vfsops->vfs_uninit)(vfsp);
 457                 if (error)
 458                         return (error);
 459         }
 460
 461         vfsconf_remove(vfsp);
 462         return 0;
 463 }
 464
 465 int
 466 vfs_modevent(module_t mod, int type, void *data)
 467 {
 468         struct vfsconf *vfc;
 469         int error = 0;
 470
 471         vfc = (struct vfsconf *)data;
 472
 473         switch (type) {
 474         case MOD_LOAD:
 475                 if (vfc)
 476                         error = vfs_register(vfc);
 477                 break;
 478
 479         case MOD_UNLOAD:
 480                 if (vfc)
 481                         error = vfs_unregister(vfc);
 482                 break;
 483         default:        /* including MOD_SHUTDOWN */
 484                 break;
 485         }
 486         return (error);
 487 }