module/zfs/zpl_inode.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/vfs.h>
  30 #include <sys/zpl.h>
  31
  32
  33 static struct dentry *
  34 #ifdef HAVE_LOOKUP_NAMEIDATA
  35 zpl_lookup(struct inode *dir, struct dentry *dentry, struct nameidata *nd)
  36 #else
  37 zpl_lookup(struct inode *dir, struct dentry *dentry, unsigned int flags)
  38 #endif
  39 {
  40         cred_t *cr = CRED();
  41         struct inode *ip;
  42         int error;
  43
  44         if (dlen(dentry) > ZFS_MAXNAMELEN)
  45                 return ERR_PTR(-ENAMETOOLONG);
  46
  47         crhold(cr);
  48         error = -zfs_lookup(dir, dname(dentry), &ip, 0, cr, NULL, NULL);
  49         ASSERT3S(error, <=, 0);
  50         crfree(cr);
  51
  52         spin_lock(&dentry->d_lock);
  53         dentry->d_time = jiffies;
  54 #ifndef HAVE_S_D_OP
  55         d_set_d_op(dentry, &zpl_dentry_operations);
  56 #endif /* HAVE_S_D_OP */
  57         spin_unlock(&dentry->d_lock);
  58
  59         if (error) {
  60                 if (error == -ENOENT)
  61                         return d_splice_alias(NULL, dentry);
  62                 else
  63                         return ERR_PTR(error);
  64         }
  65
  66         return d_splice_alias(ip, dentry);
  67 }
  68
  69 void
  70 zpl_vap_init(vattr_t *vap, struct inode *dir, zpl_umode_t mode, cred_t *cr)
  71 {
  72         vap->va_mask = ATTR_MODE;
  73         vap->va_mode = mode;
  74         vap->va_uid = crgetfsuid(cr);
  75
  76         if (dir && dir->i_mode & S_ISGID) {
  77                 vap->va_gid = dir->i_gid;
  78                 if (S_ISDIR(mode))
  79                         vap->va_mode |= S_ISGID;
  80         } else {
  81                 vap->va_gid = crgetfsgid(cr);
  82         }
  83 }
  84
  85 static int
  86 #ifdef HAVE_CREATE_NAMEIDATA
  87 zpl_create(struct inode *dir, struct dentry *dentry, zpl_umode_t mode,
  88     struct nameidata *nd)
  89 #else
  90 zpl_create(struct inode *dir, struct dentry *dentry, zpl_umode_t mode,
  91     bool flag)
  92 #endif
  93 {
  94         cred_t *cr = CRED();
  95         struct inode *ip;
  96         vattr_t *vap;
  97         int error;
  98
  99         crhold(cr);
 100         vap = kmem_zalloc(sizeof(vattr_t), KM_SLEEP);
 101         zpl_vap_init(vap, dir, mode, cr);
 102
 103         error = -zfs_create(dir, dname(dentry), vap, 0, mode, &ip, cr, 0, NULL);
 104         if (error == 0) {
 105                 error = zpl_xattr_security_init(ip, dir, &dentry->d_name);
 106                 VERIFY3S(error, ==, 0);
 107                 d_instantiate(dentry, ip);
 108         }
 109
 110         kmem_free(vap, sizeof(vattr_t));
 111         crfree(cr);
 112         ASSERT3S(error, <=, 0);
 113
 114         return (error);
 115 }
 116
 117 static int
 118 zpl_mknod(struct inode *dir, struct dentry *dentry, zpl_umode_t mode,
 119     dev_t rdev)
 120 {
 121         cred_t *cr = CRED();
 122         struct inode *ip;
 123         vattr_t *vap;
 124         int error;
 125
 126         /*
 127          * We currently expect Linux to supply rdev=0 for all sockets
 128          * and fifos, but we want to know if this behavior ever changes.
 129          */
 130         if (S_ISSOCK(mode) || S_ISFIFO(mode))
 131                 ASSERT(rdev == 0);
 132
 133         crhold(cr);
 134         vap = kmem_zalloc(sizeof(vattr_t), KM_SLEEP);
 135         zpl_vap_init(vap, dir, mode, cr);
 136         vap->va_rdev = rdev;
 137
 138         error = -zfs_create(dir, dname(dentry), vap, 0, mode, &ip, cr, 0, NULL);
 139         if (error == 0)
 140                 d_instantiate(dentry, ip);
 141
 142         kmem_free(vap, sizeof(vattr_t));
 143         crfree(cr);
 144         ASSERT3S(error, <=, 0);
 145
 146         return (-error);
 147 }
 148
 149 static int
 150 zpl_unlink(struct inode *dir, struct dentry *dentry)
 151 {
 152         cred_t *cr = CRED();
 153         int error;
 154
 155         crhold(cr);
 156         error = -zfs_remove(dir, dname(dentry), cr);
 157         crfree(cr);
 158         ASSERT3S(error, <=, 0);
 159
 160         return (error);
 161 }
 162
 163 static int
 164 zpl_mkdir(struct inode *dir, struct dentry *dentry, zpl_umode_t mode)
 165 {
 166         cred_t *cr = CRED();
 167         vattr_t *vap;
 168         struct inode *ip;
 169         int error;
 170
 171         crhold(cr);
 172         vap = kmem_zalloc(sizeof(vattr_t), KM_SLEEP);
 173         zpl_vap_init(vap, dir, mode | S_IFDIR, cr);
 174
 175         error = -zfs_mkdir(dir, dname(dentry), vap, &ip, cr, 0, NULL);
 176         if (error == 0)
 177                 d_instantiate(dentry, ip);
 178
 179         kmem_free(vap, sizeof(vattr_t));
 180         crfree(cr);
 181         ASSERT3S(error, <=, 0);
 182
 183         return (error);
 184 }
 185
 186 static int
 187 zpl_rmdir(struct inode * dir, struct dentry *dentry)
 188 {
 189         cred_t *cr = CRED();
 190         int error;
 191
 192         crhold(cr);
 193         error = -zfs_rmdir(dir, dname(dentry), NULL, cr, 0);
 194         crfree(cr);
 195         ASSERT3S(error, <=, 0);
 196
 197         return (error);
 198 }
 199
 200 static int
 201 zpl_getattr(struct vfsmount *mnt, struct dentry *dentry, struct kstat *stat)
 202 {
 203         boolean_t issnap = ITOZSB(dentry->d_inode)->z_issnap;
 204         int error;
 205
 206         /*
 207          * Ensure MNT_SHRINKABLE is set on snapshots to ensure they are
 208          * unmounted automatically with the parent file system.  This
 209          * is done on the first getattr because it's not easy to get the
 210          * vfsmount structure at mount time.  This call path is explicitly
 211          * marked unlikely to avoid any performance impact.  FWIW, ext4
 212          * resorts to a similar trick for sysadmin convenience.
 213          */
 214         if (unlikely(issnap && !(mnt->mnt_flags & MNT_SHRINKABLE)))
 215                 mnt->mnt_flags |= MNT_SHRINKABLE;
 216
 217         error = -zfs_getattr_fast(dentry->d_inode, stat);
 218         ASSERT3S(error, <=, 0);
 219
 220         return (error);
 221 }
 222
 223 static int
 224 zpl_setattr(struct dentry *dentry, struct iattr *ia)
 225 {
 226         cred_t *cr = CRED();
 227         vattr_t *vap;
 228         int error;
 229
 230         error = inode_change_ok(dentry->d_inode, ia);
 231         if (error)
 232                 return (error);
 233
 234         crhold(cr);
 235         vap = kmem_zalloc(sizeof(vattr_t), KM_SLEEP);
 236         vap->va_mask = ia->ia_valid & ATTR_IATTR_MASK;
 237         vap->va_mode = ia->ia_mode;
 238         vap->va_uid = ia->ia_uid;
 239         vap->va_gid = ia->ia_gid;
 240         vap->va_size = ia->ia_size;
 241         vap->va_atime = ia->ia_atime;
 242         vap->va_mtime = ia->ia_mtime;
 243         vap->va_ctime = ia->ia_ctime;
 244
 245         error = -zfs_setattr(dentry->d_inode, vap, 0, cr);
 246
 247         kmem_free(vap, sizeof(vattr_t));
 248         crfree(cr);
 249         ASSERT3S(error, <=, 0);
 250
 251         return (error);
 252 }
 253
 254 static int
 255 zpl_rename(struct inode *sdip, struct dentry *sdentry,
 256     struct inode *tdip, struct dentry *tdentry)
 257 {
 258         cred_t *cr = CRED();
 259         int error;
 260
 261         crhold(cr);
 262         error = -zfs_rename(sdip, dname(sdentry), tdip, dname(tdentry), cr, 0);
 263         crfree(cr);
 264         ASSERT3S(error, <=, 0);
 265
 266         return (error);
 267 }
 268
 269 static int
 270 zpl_symlink(struct inode *dir, struct dentry *dentry, const char *name)
 271 {
 272         cred_t *cr = CRED();
 273         vattr_t *vap;
 274         struct inode *ip;
 275         int error;
 276
 277         crhold(cr);
 278         vap = kmem_zalloc(sizeof(vattr_t), KM_SLEEP);
 279         zpl_vap_init(vap, dir, S_IFLNK | S_IRWXUGO, cr);
 280
 281         error = -zfs_symlink(dir, dname(dentry), vap, (char *)name, &ip, cr, 0);
 282         if (error == 0)
 283                 d_instantiate(dentry, ip);
 284
 285         kmem_free(vap, sizeof(vattr_t));
 286         crfree(cr);
 287         ASSERT3S(error, <=, 0);
 288
 289         return (error);
 290 }
 291
 292 static void *
 293 zpl_follow_link(struct dentry *dentry, struct nameidata *nd)
 294 {
 295         cred_t *cr = CRED();
 296         struct inode *ip = dentry->d_inode;
 297         struct iovec iov;
 298         uio_t uio;
 299         char *link;
 300         int error;
 301
 302         crhold(cr);
 303
 304         iov.iov_len = MAXPATHLEN;
 305         iov.iov_base = link = kmem_zalloc(MAXPATHLEN, KM_SLEEP);
 306
 307         uio.uio_iov = &iov;
 308         uio.uio_iovcnt = 1;
 309         uio.uio_resid = (MAXPATHLEN - 1);
 310         uio.uio_segflg = UIO_SYSSPACE;
 311
 312         error = -zfs_readlink(ip, &uio, cr);
 313         if (error) {
 314                 kmem_free(link, MAXPATHLEN);
 315                 nd_set_link(nd, ERR_PTR(error));
 316         } else {
 317                 nd_set_link(nd, link);
 318         }
 319
 320         crfree(cr);
 321         return (NULL);
 322 }
 323
 324 static void
 325 zpl_put_link(struct dentry *dentry, struct nameidata *nd, void *ptr)
 326 {
 327         const char *link = nd_get_link(nd);
 328
 329         if (!IS_ERR(link))
 330                 kmem_free(link, MAXPATHLEN);
 331 }
 332
 333 static int
 334 zpl_link(struct dentry *old_dentry, struct inode *dir, struct dentry *dentry)
 335 {
 336         cred_t *cr = CRED();
 337         struct inode *ip = old_dentry->d_inode;
 338         int error;
 339
 340         if (ip->i_nlink >= ZFS_LINK_MAX)
 341                 return -EMLINK;
 342
 343         crhold(cr);
 344         ip->i_ctime = CURRENT_TIME_SEC;
 345         igrab(ip); /* Use ihold() if available */
 346
 347         error = -zfs_link(dir, ip, dname(dentry), cr);
 348         if (error) {
 349                 iput(ip);
 350                 goto out;
 351         }
 352
 353         d_instantiate(dentry, ip);
 354 out:
 355         crfree(cr);
 356         ASSERT3S(error, <=, 0);
 357
 358         return (error);
 359 }
 360
 361 #ifdef HAVE_INODE_TRUNCATE_RANGE
 362 static void
 363 zpl_truncate_range(struct inode* ip, loff_t start, loff_t end)
 364 {
 365         cred_t *cr = CRED();
 366         flock64_t bf;
 367
 368         ASSERT3S(start, <=, end);
 369
 370         /*
 371          * zfs_freesp() will interpret (len == 0) as meaning "truncate until
 372          * the end of the file". We don't want that.
 373          */
 374         if (start == end)
 375                 return;
 376
 377         crhold(cr);
 378
 379         bf.l_type = F_WRLCK;
 380         bf.l_whence = 0;
 381         bf.l_start = start;
 382         bf.l_len = end - start;
 383         bf.l_pid = 0;
 384         zfs_space(ip, F_FREESP, &bf, FWRITE, start, cr);
 385
 386         crfree(cr);
 387 }
 388 #endif /* HAVE_INODE_TRUNCATE_RANGE */
 389
 390 #ifdef HAVE_INODE_FALLOCATE
 391 static long
 392 zpl_fallocate(struct inode *ip, int mode, loff_t offset, loff_t len)
 393 {
 394         return zpl_fallocate_common(ip, mode, offset, len);
 395 }
 396 #endif /* HAVE_INODE_FALLOCATE */
 397
 398 static int
 399 #ifdef HAVE_D_REVALIDATE_NAMEIDATA
 400 zpl_revalidate(struct dentry *dentry, struct nameidata *nd)
 401 {
 402         unsigned int flags = (nd ? nd->flags : 0);
 403 #else
 404 zpl_revalidate(struct dentry *dentry, unsigned int flags)
 405 {
 406 #endif /* HAVE_D_REVALIDATE_NAMEIDATA */
 407         zfs_sb_t *zsb = dentry->d_sb->s_fs_info;
 408         int error;
 409
 410         if (flags & LOOKUP_RCU)
 411                 return (-ECHILD);
 412
 413         /*
 414          * After a rollback negative dentries created before the rollback
 415          * time must be invalidated.  Otherwise they can obscure files which
 416          * are only present in the rolled back dataset.
 417          */
 418         if (dentry->d_inode == NULL) {
 419                 spin_lock(&dentry->d_lock);
 420                 error = time_before(dentry->d_time, zsb->z_rollback_time);
 421                 spin_unlock(&dentry->d_lock);
 422
 423                 if (error)
 424                         return (0);
 425         }
 426
 427         /*
 428          * The dentry may reference a stale inode if a mounted file system
 429          * was rolled back to a point in time where the object didn't exist.
 430          */
 431         if (dentry->d_inode && ITOZ(dentry->d_inode)->z_is_stale)
 432                 return (0);
 433
 434         return (1);
 435 }
 436
 437 const struct inode_operations zpl_inode_operations = {
 438         .create         = zpl_create,
 439         .link           = zpl_link,
 440         .unlink         = zpl_unlink,
 441         .symlink        = zpl_symlink,
 442         .mkdir          = zpl_mkdir,
 443         .rmdir          = zpl_rmdir,
 444         .mknod          = zpl_mknod,
 445         .rename         = zpl_rename,
 446         .setattr        = zpl_setattr,
 447         .getattr        = zpl_getattr,
 448         .setxattr       = generic_setxattr,
 449         .getxattr       = generic_getxattr,
 450         .removexattr    = generic_removexattr,
 451         .listxattr      = zpl_xattr_list,
 452 #ifdef HAVE_INODE_TRUNCATE_RANGE
 453         .truncate_range = zpl_truncate_range,
 454 #endif /* HAVE_INODE_TRUNCATE_RANGE */
 455 #ifdef HAVE_INODE_FALLOCATE
 456         .fallocate      = zpl_fallocate,
 457 #endif /* HAVE_INODE_FALLOCATE */
 458 };
 459
 460 const struct inode_operations zpl_dir_inode_operations = {
 461         .create         = zpl_create,
 462         .lookup         = zpl_lookup,
 463         .link           = zpl_link,
 464         .unlink         = zpl_unlink,
 465         .symlink        = zpl_symlink,
 466         .mkdir          = zpl_mkdir,
 467         .rmdir          = zpl_rmdir,
 468         .mknod          = zpl_mknod,
 469         .rename         = zpl_rename,
 470         .setattr        = zpl_setattr,
 471         .getattr        = zpl_getattr,
 472         .setxattr       = generic_setxattr,
 473         .getxattr       = generic_getxattr,
 474         .removexattr    = generic_removexattr,
 475         .listxattr      = zpl_xattr_list,
 476 };
 477
 478 const struct inode_operations zpl_symlink_inode_operations = {
 479         .readlink       = generic_readlink,
 480         .follow_link    = zpl_follow_link,
 481         .put_link       = zpl_put_link,
 482         .setattr        = zpl_setattr,
 483         .getattr        = zpl_getattr,
 484         .setxattr       = generic_setxattr,
 485         .getxattr       = generic_getxattr,
 486         .removexattr    = generic_removexattr,
 487         .listxattr      = zpl_xattr_list,
 488 };
 489
 490 const struct inode_operations zpl_special_inode_operations = {
 491         .setattr        = zpl_setattr,
 492         .getattr        = zpl_getattr,
 493         .setxattr       = generic_setxattr,
 494         .getxattr       = generic_getxattr,
 495         .removexattr    = generic_removexattr,
 496         .listxattr      = zpl_xattr_list,
 497 };
 498
 499 dentry_operations_t zpl_dentry_operations = {
 500         .d_revalidate   = zpl_revalidate,
 501 };