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 static int
 188 zpl_remount_fs(struct super_block *sb, int *flags, char *data)
 189 {
 190         fstrans_cookie_t cookie;
 191         int error;
 192
 193         cookie = spl_fstrans_mark();
 194         error = -zfs_remount(sb, flags, data);
 195         spl_fstrans_unmark(cookie);
 196         ASSERT3S(error, <=, 0);
 197
 198         return (error);
 199 }
 200
 201 static void
 202 zpl_umount_begin(struct super_block *sb)
 203 {
 204         zfs_sb_t *zsb = sb->s_fs_info;
 205         int count;
 206
 207         /*
 208          * Best effort to unmount snapshots in .zfs/snapshot/.  Normally this
 209          * isn't required because snapshots have the MNT_SHRINKABLE flag set.
 210          */
 211         if (zsb->z_ctldir)
 212                 (void) zfsctl_unmount_snapshots(zsb, MNT_FORCE, &count);
 213 }
 214
 215 /*
 216  * ZFS specific features must be explicitly handled here, the VFS will
 217  * automatically handled the following generic functionality.
 218  *
 219  *   MNT_NOSUID,
 220  *   MNT_NODEV,
 221  *   MNT_NOEXEC,
 222  *   MNT_NOATIME,
 223  *   MNT_NODIRATIME,
 224  *   MNT_READONLY,
 225  *   MNT_STRICTATIME,
 226  *   MS_SYNCHRONOUS,
 227  *   MS_DIRSYNC,
 228  *   MS_MANDLOCK.
 229  */
 230 static int
 231 __zpl_show_options(struct seq_file *seq, zfs_sb_t *zsb)
 232 {
 233         seq_printf(seq, ",%s", zsb->z_flags & ZSB_XATTR ? "xattr" : "noxattr");
 234
 235 #ifdef CONFIG_FS_POSIX_ACL
 236         switch (zsb->z_acl_type) {
 237         case ZFS_ACLTYPE_POSIXACL:
 238                 seq_puts(seq, ",posixacl");
 239                 break;
 240         default:
 241                 seq_puts(seq, ",noacl");
 242                 break;
 243         }
 244 #endif /* CONFIG_FS_POSIX_ACL */
 245
 246         return (0);
 247 }
 248
 249 #ifdef HAVE_SHOW_OPTIONS_WITH_DENTRY
 250 static int
 251 zpl_show_options(struct seq_file *seq, struct dentry *root)
 252 {
 253         return (__zpl_show_options(seq, root->d_sb->s_fs_info));
 254 }
 255 #else
 256 static int
 257 zpl_show_options(struct seq_file *seq, struct vfsmount *vfsp)
 258 {
 259         return (__zpl_show_options(seq, vfsp->mnt_sb->s_fs_info));
 260 }
 261 #endif /* HAVE_SHOW_OPTIONS_WITH_DENTRY */
 262
 263 static int
 264 zpl_fill_super(struct super_block *sb, void *data, int silent)
 265 {
 266         fstrans_cookie_t cookie;
 267         int error;
 268
 269         cookie = spl_fstrans_mark();
 270         error = -zfs_domount(sb, data, silent);
 271         spl_fstrans_unmark(cookie);
 272         ASSERT3S(error, <=, 0);
 273
 274         return (error);
 275 }
 276
 277 #ifdef HAVE_MOUNT_NODEV
 278 static struct dentry *
 279 zpl_mount(struct file_system_type *fs_type, int flags,
 280     const char *osname, void *data)
 281 {
 282         zpl_mount_data_t zmd = { osname, data };
 283
 284         return (mount_nodev(fs_type, flags, &zmd, zpl_fill_super));
 285 }
 286 #else
 287 static int
 288 zpl_get_sb(struct file_system_type *fs_type, int flags,
 289     const char *osname, void *data, struct vfsmount *mnt)
 290 {
 291         zpl_mount_data_t zmd = { osname, data };
 292
 293         return (get_sb_nodev(fs_type, flags, &zmd, zpl_fill_super, mnt));
 294 }
 295 #endif /* HAVE_MOUNT_NODEV */
 296
 297 static void
 298 zpl_kill_sb(struct super_block *sb)
 299 {
 300         zfs_preumount(sb);
 301         kill_anon_super(sb);
 302
 303 #ifdef HAVE_S_INSTANCES_LIST_HEAD
 304         sb->s_instances.next = &(zpl_fs_type.fs_supers);
 305 #endif /* HAVE_S_INSTANCES_LIST_HEAD */
 306 }
 307
 308 void
 309 zpl_prune_sb(int64_t nr_to_scan, void *arg)
 310 {
 311         struct super_block *sb = (struct super_block *)arg;
 312         int objects = 0;
 313
 314         (void) -zfs_sb_prune(sb, nr_to_scan, &objects);
 315 }
 316
 317 #ifdef HAVE_NR_CACHED_OBJECTS
 318 static int
 319 zpl_nr_cached_objects(struct super_block *sb)
 320 {
 321         zfs_sb_t *zsb = sb->s_fs_info;
 322         int nr;
 323
 324         mutex_enter(&zsb->z_znodes_lock);
 325         nr = zsb->z_nr_znodes;
 326         mutex_exit(&zsb->z_znodes_lock);
 327
 328         return (nr);
 329 }
 330 #endif /* HAVE_NR_CACHED_OBJECTS */
 331
 332 #ifdef HAVE_FREE_CACHED_OBJECTS
 333 /*
 334  * Attempt to evict some meta data from the cache.  The ARC operates in
 335  * terms of bytes while the Linux VFS uses objects.  Now because this is
 336  * just a best effort eviction and the exact values aren't critical so we
 337  * extrapolate from an object count to a byte size using the znode_t size.
 338  */
 339 static void
 340 zpl_free_cached_objects(struct super_block *sb, int nr_to_scan)
 341 {
 342         /* noop */
 343 }
 344 #endif /* HAVE_FREE_CACHED_OBJECTS */
 345
 346 const struct super_operations zpl_super_operations = {
 347         .alloc_inode            = zpl_inode_alloc,
 348         .destroy_inode          = zpl_inode_destroy,
 349         .dirty_inode            = zpl_dirty_inode,
 350         .write_inode            = NULL,
 351 #ifdef HAVE_EVICT_INODE
 352         .evict_inode            = zpl_evict_inode,
 353 #else
 354         .drop_inode             = zpl_drop_inode,
 355         .clear_inode            = zpl_clear_inode,
 356         .delete_inode           = zpl_inode_delete,
 357 #endif /* HAVE_EVICT_INODE */
 358         .put_super              = zpl_put_super,
 359         .sync_fs                = zpl_sync_fs,
 360         .statfs                 = zpl_statfs,
 361         .remount_fs             = zpl_remount_fs,
 362         .umount_begin           = zpl_umount_begin,
 363         .show_options           = zpl_show_options,
 364         .show_stats             = NULL,
 365 #ifdef HAVE_NR_CACHED_OBJECTS
 366         .nr_cached_objects      = zpl_nr_cached_objects,
 367 #endif /* HAVE_NR_CACHED_OBJECTS */
 368 #ifdef HAVE_FREE_CACHED_OBJECTS
 369         .free_cached_objects    = zpl_free_cached_objects,
 370 #endif /* HAVE_FREE_CACHED_OBJECTS */
 371 };
 372
 373 struct file_system_type zpl_fs_type = {
 374         .owner                  = THIS_MODULE,
 375         .name                   = ZFS_DRIVER,
 376 #ifdef HAVE_MOUNT_NODEV
 377         .mount                  = zpl_mount,
 378 #else
 379         .get_sb                 = zpl_get_sb,
 380 #endif /* HAVE_MOUNT_NODEV */
 381         .kill_sb                = zpl_kill_sb,
 382 };