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