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