module/zfs/zpl_super.c

   1 /*
   2  * CDDL HEADER START
   3  *
   4  * The contents of this file are subject to the terms of the
   5  * Common Development and Distribution License (the "License").
   6  * You may not use this file except in compliance with the License.
   7  *
   8  * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
   9  * or http://www.opensolaris.org/os/licensing.
  10  * See the License for the specific language governing permissions
  11  * and limitations under the License.
  12  *
  13  * When distributing Covered Code, include this CDDL HEADER in each
  14  * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
  15  * If applicable, add the following below this CDDL HEADER, with the
  16  * fields enclosed by brackets "[]" replaced with your own identifying
  17  * information: Portions Copyright [yyyy] [name of copyright owner]
  18  *
  19  * CDDL HEADER END
  20  */
  21 /*
  22  * Copyright (c) 2011, Lawrence Livermore National Security, LLC.
  23  */
  24
  25
  26 #include <sys/zfs_vfsops.h>
  27 #include <sys/zfs_vnops.h>
  28 #include <sys/zfs_znode.h>
  29 #include <sys/zfs_ctldir.h>
  30 #include <sys/zpl.h>
  31
  32
  33 static struct inode *
  34 zpl_inode_alloc(struct super_block *sb)
  35 {
  36         struct inode *ip;
  37
  38         VERIFY3S(zfs_inode_alloc(sb, &ip), ==, 0);
  39         ip->i_version = 1;
  40
  41         return (ip);
  42 }
  43
  44 static void
  45 zpl_inode_destroy(struct inode *ip)
  46 {
  47         ASSERT(atomic_read(&ip->i_count) == 0);
  48         zfs_inode_destroy(ip);
  49 }
  50
  51 /*
  52  * Called from __mark_inode_dirty() to reflect that something in the
  53  * inode has changed.  We use it to ensure the znode system attributes
  54  * are always strictly update to date with respect to the inode.
  55  */
  56 #ifdef HAVE_DIRTY_INODE_WITH_FLAGS
  57 static void
  58 zpl_dirty_inode(struct inode *ip, int flags)
  59 {
  60         fstrans_cookie_t cookie;
  61
  62         cookie = spl_fstrans_mark();
  63         zfs_dirty_inode(ip, flags);
  64         spl_fstrans_unmark(cookie);
  65 }
  66 #else
  67 static void
  68 zpl_dirty_inode(struct inode *ip)
  69 {
  70         fstrans_cookie_t cookie;
  71
  72         cookie = spl_fstrans_mark();
  73         zfs_dirty_inode(ip, 0);
  74         spl_fstrans_unmark(cookie);
  75 }
  76 #endif /* HAVE_DIRTY_INODE_WITH_FLAGS */
  77
  78 /*
  79  * When ->drop_inode() is called its return value indicates if the
  80  * inode should be evicted from the inode cache.  If the inode is
  81  * unhashed and has no links the default policy is to evict it
  82  * immediately.
  83  *
  84  * Prior to 2.6.36 this eviction was accomplished by the vfs calling
  85  * ->delete_inode().  It was ->delete_inode()'s responsibility to
  86  * truncate the inode pages and call clear_inode().  The call to
  87  * clear_inode() synchronously invalidates all the buffers and
  88  * calls ->clear_inode().  It was ->clear_inode()'s responsibility
  89  * to cleanup and filesystem specific data before freeing the inode.
  90  *
  91  * This elaborate mechanism was replaced by ->evict_inode() which
  92  * does the job of both ->delete_inode() and ->clear_inode().  It
  93  * will be called exactly once, and when it returns the inode must
  94  * be in a state where it can simply be freed.i
  95  *
  96  * The ->evict_inode() callback must minimally truncate the inode pages,
  97  * and call clear_inode().  For 2.6.35 and later kernels this will
  98  * simply update the inode state, with the sync occurring before the
  99  * truncate in evict().  For earlier kernels clear_inode() maps to
 100  * end_writeback() which is responsible for completing all outstanding
 101  * write back.  In either case, once this is done it is safe to cleanup
 102  * any remaining inode specific data via zfs_inactive().
 103  * remaining filesystem specific data.
 104  */
 105 #ifdef HAVE_EVICT_INODE
 106 static void
 107 zpl_evict_inode(struct inode *ip)
 108 {
 109         fstrans_cookie_t cookie;
 110
 111         cookie = spl_fstrans_mark();
 112         truncate_setsize(ip, 0);
 113         clear_inode(ip);
 114         zfs_inactive(ip);
 115         spl_fstrans_unmark(cookie);
 116 }
 117
 118 #else
 119
 120 static void
 121 zpl_drop_inode(struct inode *ip)
 122 {
 123         generic_delete_inode(ip);
 124 }
 125
 126 static void
 127 zpl_clear_inode(struct inode *ip)
 128 {
 129         fstrans_cookie_t cookie;
 130
 131         cookie = spl_fstrans_mark();
 132         zfs_inactive(ip);
 133         spl_fstrans_unmark(cookie);
 134 }
 135
 136 static void
 137 zpl_inode_delete(struct inode *ip)
 138 {
 139         truncate_setsize(ip, 0);
 140         clear_inode(ip);
 141 }
 142 #endif /* HAVE_EVICT_INODE */
 143
 144 static void
 145 zpl_put_super(struct super_block *sb)
 146 {
 147         fstrans_cookie_t cookie;
 148         int error;
 149
 150         cookie = spl_fstrans_mark();
 151         error = -zfs_umount(sb);
 152         spl_fstrans_unmark(cookie);
 153         ASSERT3S(error, <=, 0);
 154 }
 155
 156 static int
 157 zpl_sync_fs(struct super_block *sb, int wait)
 158 {
 159         fstrans_cookie_t cookie;
 160         cred_t *cr = CRED();
 161         int error;
 162
 163         crhold(cr);
 164         cookie = spl_fstrans_mark();
 165         error = -zfs_sync(sb, wait, cr);
 166         spl_fstrans_unmark(cookie);
 167         crfree(cr);
 168         ASSERT3S(error, <=, 0);
 169
 170         return (error);
 171 }
 172
 173 static int
 174 zpl_statfs(struct dentry *dentry, struct kstatfs *statp)
 175 {
 176         fstrans_cookie_t cookie;
 177         int error;
 178
 179         cookie = spl_fstrans_mark();
 180         error = -zfs_statvfs(dentry, statp);
 181         spl_fstrans_unmark(cookie);
 182         ASSERT3S(error, <=, 0);
 183
 184         return (error);
 185 }
 186
 187 enum {
 188         TOKEN_RO,
 189         TOKEN_RW,
 190         TOKEN_SETUID,
 191         TOKEN_NOSETUID,
 192         TOKEN_EXEC,
 193         TOKEN_NOEXEC,
 194         TOKEN_DEVICES,
 195         TOKEN_NODEVICES,
 196         TOKEN_DIRXATTR,
 197         TOKEN_SAXATTR,
 198         TOKEN_XATTR,
 199         TOKEN_NOXATTR,
 200         TOKEN_ATIME,
 201         TOKEN_NOATIME,
 202         TOKEN_RELATIME,
 203         TOKEN_NORELATIME,
 204         TOKEN_NBMAND,
 205         TOKEN_NONBMAND,
 206         TOKEN_MNTPOINT,
 207         TOKEN_LAST,
 208 };
 209
 210 static const match_table_t zpl_tokens = {
 211         { TOKEN_RO,             MNTOPT_RO },
 212         { TOKEN_RW,             MNTOPT_RW },
 213         { TOKEN_SETUID,         MNTOPT_SETUID },
 214         { TOKEN_NOSETUID,       MNTOPT_NOSETUID },
 215         { TOKEN_EXEC,           MNTOPT_EXEC },
 216         { TOKEN_NOEXEC,         MNTOPT_NOEXEC },
 217         { TOKEN_DEVICES,        MNTOPT_DEVICES },
 218         { TOKEN_NODEVICES,      MNTOPT_NODEVICES },
 219         { TOKEN_DIRXATTR,       MNTOPT_DIRXATTR },
 220         { TOKEN_SAXATTR,        MNTOPT_SAXATTR },
 221         { TOKEN_XATTR,          MNTOPT_XATTR },
 222         { TOKEN_NOXATTR,        MNTOPT_NOXATTR },
 223         { TOKEN_ATIME,          MNTOPT_ATIME },
 224         { TOKEN_NOATIME,        MNTOPT_NOATIME },
 225         { TOKEN_RELATIME,       MNTOPT_RELATIME },
 226         { TOKEN_NORELATIME,     MNTOPT_NORELATIME },
 227         { TOKEN_NBMAND,         MNTOPT_NBMAND },
 228         { TOKEN_NONBMAND,       MNTOPT_NONBMAND },
 229         { TOKEN_MNTPOINT,       MNTOPT_MNTPOINT "=%s" },
 230         { TOKEN_LAST,           NULL },
 231 };
 232
 233 static int
 234 zpl_parse_option(char *option, int token, substring_t *args, zfs_mntopts_t *zmo)
 235 {
 236         switch (token) {
 237         case TOKEN_RO:
 238                 zmo->z_readonly = B_TRUE;
 239                 zmo->z_do_readonly = B_TRUE;
 240                 break;
 241         case TOKEN_RW:
 242                 zmo->z_readonly = B_FALSE;
 243                 zmo->z_do_readonly = B_TRUE;
 244                 break;
 245         case TOKEN_SETUID:
 246                 zmo->z_setuid = B_TRUE;
 247                 zmo->z_do_setuid = B_TRUE;
 248                 break;
 249         case TOKEN_NOSETUID:
 250                 zmo->z_setuid = B_FALSE;
 251                 zmo->z_do_setuid = B_TRUE;
 252                 break;
 253         case TOKEN_EXEC:
 254                 zmo->z_exec = B_TRUE;
 255                 zmo->z_do_exec = B_TRUE;
 256                 break;
 257         case TOKEN_NOEXEC:
 258                 zmo->z_exec = B_FALSE;
 259                 zmo->z_do_exec = B_TRUE;
 260                 break;
 261         case TOKEN_DEVICES:
 262                 zmo->z_devices = B_TRUE;
 263                 zmo->z_do_devices = B_TRUE;
 264                 break;
 265         case TOKEN_NODEVICES:
 266                 zmo->z_devices = B_FALSE;
 267                 zmo->z_do_devices = B_TRUE;
 268                 break;
 269         case TOKEN_DIRXATTR:
 270                 zmo->z_xattr = ZFS_XATTR_DIR;
 271                 zmo->z_do_xattr = B_TRUE;
 272                 break;
 273         case TOKEN_SAXATTR:
 274                 zmo->z_xattr = ZFS_XATTR_SA;
 275                 zmo->z_do_xattr = B_TRUE;
 276                 break;
 277         case TOKEN_XATTR:
 278                 zmo->z_xattr = ZFS_XATTR_DIR;
 279                 zmo->z_do_xattr = B_TRUE;
 280                 break;
 281         case TOKEN_NOXATTR:
 282                 zmo->z_xattr = ZFS_XATTR_OFF;
 283                 zmo->z_do_xattr = B_TRUE;
 284                 break;
 285         case TOKEN_ATIME:
 286                 zmo->z_atime = B_TRUE;
 287                 zmo->z_do_atime = B_TRUE;
 288                 break;
 289         case TOKEN_NOATIME:
 290                 zmo->z_atime = B_FALSE;
 291                 zmo->z_do_atime = B_TRUE;
 292                 break;
 293         case TOKEN_RELATIME:
 294                 zmo->z_relatime = B_TRUE;
 295                 zmo->z_do_relatime = B_TRUE;
 296                 break;
 297         case TOKEN_NORELATIME:
 298                 zmo->z_relatime = B_FALSE;
 299                 zmo->z_do_relatime = B_TRUE;
 300                 break;
 301         case TOKEN_NBMAND:
 302                 zmo->z_nbmand = B_TRUE;
 303                 zmo->z_do_nbmand = B_TRUE;
 304                 break;
 305         case TOKEN_NONBMAND:
 306                 zmo->z_nbmand = B_FALSE;
 307                 zmo->z_do_nbmand = B_TRUE;
 308                 break;
 309         case TOKEN_MNTPOINT:
 310                 zmo->z_mntpoint = match_strdup(&args[0]);
 311                 if (zmo->z_mntpoint == NULL)
 312                         return (-ENOMEM);
 313
 314                 break;
 315         default:
 316                 break;
 317         }
 318
 319         return (0);
 320 }
 321
 322 /*
 323  * Parse the mntopts string storing the results in provided zmo argument.
 324  * If an error occurs the zmo argument will not be modified.  The caller
 325  * needs to set isremount when recycling an existing zfs_mntopts_t.
 326  */
 327 static int
 328 zpl_parse_options(char *osname, char *mntopts, zfs_mntopts_t *zmo,
 329     boolean_t isremount)
 330 {
 331         zfs_mntopts_t *tmp_zmo;
 332         int error;
 333
 334         tmp_zmo = zfs_mntopts_alloc();
 335         tmp_zmo->z_osname = strdup(osname);
 336
 337         if (mntopts) {
 338                 substring_t args[MAX_OPT_ARGS];
 339                 char *tmp_mntopts, *p, *t;
 340                 int token;
 341
 342                 t = tmp_mntopts = strdup(mntopts);
 343
 344                 while ((p = strsep(&t, ",")) != NULL) {
 345                         if (!*p)
 346                                 continue;
 347
 348                         args[0].to = args[0].from = NULL;
 349                         token = match_token(p, zpl_tokens, args);
 350                         error = zpl_parse_option(p, token, args, tmp_zmo);
 351                         if (error) {
 352                                 zfs_mntopts_free(tmp_zmo);
 353                                 strfree(tmp_mntopts);
 354                                 return (error);
 355                         }
 356                 }
 357
 358                 strfree(tmp_mntopts);
 359         }
 360
 361         if (isremount == B_TRUE) {
 362                 if (zmo->z_osname)
 363                         strfree(zmo->z_osname);
 364
 365                 if (zmo->z_mntpoint)
 366                         strfree(zmo->z_mntpoint);
 367         } else {
 368                 ASSERT3P(zmo->z_osname, ==, NULL);
 369                 ASSERT3P(zmo->z_mntpoint, ==, NULL);
 370         }
 371
 372         memcpy(zmo, tmp_zmo, sizeof (zfs_mntopts_t));
 373         kmem_free(tmp_zmo, sizeof (zfs_mntopts_t));
 374
 375         return (0);
 376 }
 377
 378 static int
 379 zpl_remount_fs(struct super_block *sb, int *flags, char *data)
 380 {
 381         zfs_sb_t *zsb = sb->s_fs_info;
 382         fstrans_cookie_t cookie;
 383         int error;
 384
 385         error = zpl_parse_options(zsb->z_mntopts->z_osname, data,
 386             zsb->z_mntopts, B_TRUE);
 387         if (error)
 388                 return (error);
 389
 390         cookie = spl_fstrans_mark();
 391         error = -zfs_remount(sb, flags, zsb->z_mntopts);
 392         spl_fstrans_unmark(cookie);
 393         ASSERT3S(error, <=, 0);
 394
 395         return (error);
 396 }
 397
 398 static int
 399 __zpl_show_options(struct seq_file *seq, zfs_sb_t *zsb)
 400 {
 401         seq_printf(seq, ",%s", zsb->z_flags & ZSB_XATTR ? "xattr" : "noxattr");
 402
 403 #ifdef CONFIG_FS_POSIX_ACL
 404         switch (zsb->z_acl_type) {
 405         case ZFS_ACLTYPE_POSIXACL:
 406                 seq_puts(seq, ",posixacl");
 407                 break;
 408         default:
 409                 seq_puts(seq, ",noacl");
 410                 break;
 411         }
 412 #endif /* CONFIG_FS_POSIX_ACL */
 413
 414         return (0);
 415 }
 416
 417 #ifdef HAVE_SHOW_OPTIONS_WITH_DENTRY
 418 static int
 419 zpl_show_options(struct seq_file *seq, struct dentry *root)
 420 {
 421         return (__zpl_show_options(seq, root->d_sb->s_fs_info));
 422 }
 423 #else
 424 static int
 425 zpl_show_options(struct seq_file *seq, struct vfsmount *vfsp)
 426 {
 427         return (__zpl_show_options(seq, vfsp->mnt_sb->s_fs_info));
 428 }
 429 #endif /* HAVE_SHOW_OPTIONS_WITH_DENTRY */
 430
 431 static int
 432 zpl_fill_super(struct super_block *sb, void *data, int silent)
 433 {
 434         zfs_mntopts_t *zmo = (zfs_mntopts_t *)data;
 435         fstrans_cookie_t cookie;
 436         int error;
 437
 438         cookie = spl_fstrans_mark();
 439         error = -zfs_domount(sb, zmo, silent);
 440         spl_fstrans_unmark(cookie);
 441         ASSERT3S(error, <=, 0);
 442
 443         return (error);
 444 }
 445
 446 #ifdef HAVE_MOUNT_NODEV
 447 static struct dentry *
 448 zpl_mount(struct file_system_type *fs_type, int flags,
 449     const char *osname, void *data)
 450 {
 451         zfs_mntopts_t *zmo = zfs_mntopts_alloc();
 452         int error;
 453
 454         error = zpl_parse_options((char *)osname, (char *)data, zmo, B_FALSE);
 455         if (error) {
 456                 zfs_mntopts_free(zmo);
 457                 return (ERR_PTR(error));
 458         }
 459
 460         return (mount_nodev(fs_type, flags, zmo, zpl_fill_super));
 461 }
 462 #else
 463 static int
 464 zpl_get_sb(struct file_system_type *fs_type, int flags,
 465     const char *osname, void *data, struct vfsmount *mnt)
 466 {
 467         zfs_mntopts_t *zmo = zfs_mntopts_alloc();
 468         int error;
 469
 470         error = zpl_parse_options((char *)osname, (char *)data, zmo, B_FALSE);
 471         if (error) {
 472                 zfs_mntopts_free(zmo);
 473                 return (error);
 474         }
 475
 476         return (get_sb_nodev(fs_type, flags, zmo, zpl_fill_super, mnt));
 477 }
 478 #endif /* HAVE_MOUNT_NODEV */
 479
 480 static void
 481 zpl_kill_sb(struct super_block *sb)
 482 {
 483         zfs_preumount(sb);
 484         kill_anon_super(sb);
 485
 486 #ifdef HAVE_S_INSTANCES_LIST_HEAD
 487         sb->s_instances.next = &(zpl_fs_type.fs_supers);
 488 #endif /* HAVE_S_INSTANCES_LIST_HEAD */
 489 }
 490
 491 void
 492 zpl_prune_sb(int64_t nr_to_scan, void *arg)
 493 {
 494         struct super_block *sb = (struct super_block *)arg;
 495         int objects = 0;
 496
 497         (void) -zfs_sb_prune(sb, nr_to_scan, &objects);
 498 }
 499
 500 #ifdef HAVE_NR_CACHED_OBJECTS
 501 static int
 502 zpl_nr_cached_objects(struct super_block *sb)
 503 {
 504         return (0);
 505 }
 506 #endif /* HAVE_NR_CACHED_OBJECTS */
 507
 508 #ifdef HAVE_FREE_CACHED_OBJECTS
 509 static void
 510 zpl_free_cached_objects(struct super_block *sb, int nr_to_scan)
 511 {
 512         /* noop */
 513 }
 514 #endif /* HAVE_FREE_CACHED_OBJECTS */
 515
 516 const struct super_operations zpl_super_operations = {
 517         .alloc_inode            = zpl_inode_alloc,
 518         .destroy_inode          = zpl_inode_destroy,
 519         .dirty_inode            = zpl_dirty_inode,
 520         .write_inode            = NULL,
 521 #ifdef HAVE_EVICT_INODE
 522         .evict_inode            = zpl_evict_inode,
 523 #else
 524         .drop_inode             = zpl_drop_inode,
 525         .clear_inode            = zpl_clear_inode,
 526         .delete_inode           = zpl_inode_delete,
 527 #endif /* HAVE_EVICT_INODE */
 528         .put_super              = zpl_put_super,
 529         .sync_fs                = zpl_sync_fs,
 530         .statfs                 = zpl_statfs,
 531         .remount_fs             = zpl_remount_fs,
 532         .show_options           = zpl_show_options,
 533         .show_stats             = NULL,
 534 #ifdef HAVE_NR_CACHED_OBJECTS
 535         .nr_cached_objects      = zpl_nr_cached_objects,
 536 #endif /* HAVE_NR_CACHED_OBJECTS */
 537 #ifdef HAVE_FREE_CACHED_OBJECTS
 538         .free_cached_objects    = zpl_free_cached_objects,
 539 #endif /* HAVE_FREE_CACHED_OBJECTS */
 540 };
 541
 542 struct file_system_type zpl_fs_type = {
 543         .owner                  = THIS_MODULE,
 544         .name                   = ZFS_DRIVER,
 545 #ifdef HAVE_MOUNT_NODEV
 546         .mount                  = zpl_mount,
 547 #else
 548         .get_sb                 = zpl_get_sb,
 549 #endif /* HAVE_MOUNT_NODEV */
 550         .kill_sb                = zpl_kill_sb,
 551 };