]> git.proxmox.com Git - mirror_zfs-debian.git/blame - module/zfs/zpl_inode.c
Fix 'zfs rollback' on mounted file systems
[mirror_zfs-debian.git] / module / zfs / zpl_inode.c
CommitLineData
ee154f01
BB
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>
ebe7e575 28#include <sys/zfs_znode.h>
ee154f01
BB
29#include <sys/vfs.h>
30#include <sys/zpl.h>
31
32
33static struct dentry *
8f195a90 34#ifdef HAVE_LOOKUP_NAMEIDATA
ee154f01 35zpl_lookup(struct inode *dir, struct dentry *dentry, struct nameidata *nd)
8f195a90
YS
36#else
37zpl_lookup(struct inode *dir, struct dentry *dentry, unsigned int flags)
38#endif
ee154f01 39{
81e97e21 40 cred_t *cr = CRED();
ee154f01 41 struct inode *ip;
ee154f01
BB
42 int error;
43
81e97e21 44 crhold(cr);
ee154f01
BB
45 error = -zfs_lookup(dir, dname(dentry), &ip, 0, cr, NULL, NULL);
46 ASSERT3S(error, <=, 0);
81e97e21 47 crfree(cr);
ee154f01 48
7b3e34ba
BB
49 spin_lock(&dentry->d_lock);
50 dentry->d_time = jiffies;
51 spin_unlock(&dentry->d_lock);
52
ee154f01
BB
53 if (error) {
54 if (error == -ENOENT)
55 return d_splice_alias(NULL, dentry);
56 else
57 return ERR_PTR(error);
58 }
59
60 return d_splice_alias(ip, dentry);
61}
62
ebe7e575 63void
7b3e34ba 64zpl_vap_init(vattr_t *vap, struct inode *dir, zpl_umode_t mode, cred_t *cr)
9fd91dae
BB
65{
66 vap->va_mask = ATTR_MODE;
67 vap->va_mode = mode;
9fd91dae
BB
68 vap->va_uid = crgetfsuid(cr);
69
70 if (dir && dir->i_mode & S_ISGID) {
71 vap->va_gid = dir->i_gid;
72 if (S_ISDIR(mode))
73 vap->va_mode |= S_ISGID;
74 } else {
75 vap->va_gid = crgetfsgid(cr);
76 }
77}
78
ee154f01 79static int
558ef6d0 80#ifdef HAVE_CREATE_NAMEIDATA
b39d3b9f 81zpl_create(struct inode *dir, struct dentry *dentry, zpl_umode_t mode,
ee154f01 82 struct nameidata *nd)
558ef6d0
YS
83#else
84zpl_create(struct inode *dir, struct dentry *dentry, zpl_umode_t mode,
85 bool flag)
86#endif
ee154f01 87{
81e97e21 88 cred_t *cr = CRED();
ee154f01
BB
89 struct inode *ip;
90 vattr_t *vap;
91 int error;
92
81e97e21 93 crhold(cr);
ee154f01 94 vap = kmem_zalloc(sizeof(vattr_t), KM_SLEEP);
7b3e34ba 95 zpl_vap_init(vap, dir, mode, cr);
ee154f01 96
e89260a1
BB
97 error = -zfs_create(dir, dname(dentry), vap, 0, mode, &ip, cr, 0, NULL);
98 if (error == 0) {
99 error = zpl_xattr_security_init(ip, dir, &dentry->d_name);
100 VERIFY3S(error, ==, 0);
7b3e34ba
BB
101 d_instantiate(dentry, ip);
102 d_set_d_op(dentry, &zpl_dentry_operations);
e89260a1
BB
103 }
104
ee154f01 105 kmem_free(vap, sizeof(vattr_t));
81e97e21 106 crfree(cr);
ee154f01
BB
107 ASSERT3S(error, <=, 0);
108
109 return (error);
110}
111
112static int
b39d3b9f
BB
113zpl_mknod(struct inode *dir, struct dentry *dentry, zpl_umode_t mode,
114 dev_t rdev)
ee154f01 115{
81e97e21 116 cred_t *cr = CRED();
ee154f01
BB
117 struct inode *ip;
118 vattr_t *vap;
119 int error;
120
aa6d8c10
NB
121 /*
122 * We currently expect Linux to supply rdev=0 for all sockets
123 * and fifos, but we want to know if this behavior ever changes.
124 */
125 if (S_ISSOCK(mode) || S_ISFIFO(mode))
126 ASSERT(rdev == 0);
127
81e97e21 128 crhold(cr);
ee154f01 129 vap = kmem_zalloc(sizeof(vattr_t), KM_SLEEP);
7b3e34ba 130 zpl_vap_init(vap, dir, mode, cr);
ee154f01 131 vap->va_rdev = rdev;
ee154f01 132
7b3e34ba
BB
133 error = -zfs_create(dir, dname(dentry), vap, 0, mode, &ip, cr, 0, NULL);
134 if (error == 0) {
135 d_instantiate(dentry, ip);
136 d_set_d_op(dentry, &zpl_dentry_operations);
137 }
138
ee154f01 139 kmem_free(vap, sizeof(vattr_t));
81e97e21 140 crfree(cr);
ee154f01
BB
141 ASSERT3S(error, <=, 0);
142
143 return (-error);
144}
145
146static int
147zpl_unlink(struct inode *dir, struct dentry *dentry)
148{
81e97e21 149 cred_t *cr = CRED();
ee154f01
BB
150 int error;
151
81e97e21 152 crhold(cr);
ee154f01 153 error = -zfs_remove(dir, dname(dentry), cr);
81e97e21 154 crfree(cr);
ee154f01
BB
155 ASSERT3S(error, <=, 0);
156
157 return (error);
158}
159
160static int
b39d3b9f 161zpl_mkdir(struct inode *dir, struct dentry *dentry, zpl_umode_t mode)
ee154f01 162{
81e97e21 163 cred_t *cr = CRED();
ee154f01
BB
164 vattr_t *vap;
165 struct inode *ip;
166 int error;
167
81e97e21 168 crhold(cr);
ee154f01 169 vap = kmem_zalloc(sizeof(vattr_t), KM_SLEEP);
7b3e34ba 170 zpl_vap_init(vap, dir, mode | S_IFDIR, cr);
ee154f01
BB
171
172 error = -zfs_mkdir(dir, dname(dentry), vap, &ip, cr, 0, NULL);
7b3e34ba
BB
173 if (error == 0) {
174 d_instantiate(dentry, ip);
175 d_set_d_op(dentry, &zpl_dentry_operations);
176 }
177
ee154f01 178 kmem_free(vap, sizeof(vattr_t));
81e97e21 179 crfree(cr);
ee154f01
BB
180 ASSERT3S(error, <=, 0);
181
182 return (error);
183}
184
185static int
186zpl_rmdir(struct inode * dir, struct dentry *dentry)
187{
81e97e21 188 cred_t *cr = CRED();
ee154f01
BB
189 int error;
190
81e97e21 191 crhold(cr);
ee154f01 192 error = -zfs_rmdir(dir, dname(dentry), NULL, cr, 0);
81e97e21 193 crfree(cr);
ee154f01
BB
194 ASSERT3S(error, <=, 0);
195
196 return (error);
197}
198
199static int
200zpl_getattr(struct vfsmount *mnt, struct dentry *dentry, struct kstat *stat)
201{
ebe7e575 202 boolean_t issnap = ITOZSB(dentry->d_inode)->z_issnap;
ee154f01
BB
203 int error;
204
ebe7e575
BB
205 /*
206 * Ensure MNT_SHRINKABLE is set on snapshots to ensure they are
207 * unmounted automatically with the parent file system. This
208 * is done on the first getattr because it's not easy to get the
209 * vfsmount structure at mount time. This call path is explicitly
210 * marked unlikely to avoid any performance impact. FWIW, ext4
211 * resorts to a similar trick for sysadmin convenience.
212 */
213 if (unlikely(issnap && !(mnt->mnt_flags & MNT_SHRINKABLE)))
214 mnt->mnt_flags |= MNT_SHRINKABLE;
215
057e8eee 216 error = -zfs_getattr_fast(dentry->d_inode, stat);
ee154f01
BB
217 ASSERT3S(error, <=, 0);
218
219 return (error);
220}
221
222static int
5484965a 223zpl_setattr(struct dentry *dentry, struct iattr *ia)
ee154f01 224{
81e97e21 225 cred_t *cr = CRED();
5484965a 226 vattr_t *vap;
ee154f01
BB
227 int error;
228
5484965a 229 error = inode_change_ok(dentry->d_inode, ia);
ee154f01
BB
230 if (error)
231 return (error);
232
81e97e21 233 crhold(cr);
5484965a
BB
234 vap = kmem_zalloc(sizeof(vattr_t), KM_SLEEP);
235 vap->va_mask = ia->ia_valid & ATTR_IATTR_MASK;
236 vap->va_mode = ia->ia_mode;
237 vap->va_uid = ia->ia_uid;
238 vap->va_gid = ia->ia_gid;
239 vap->va_size = ia->ia_size;
240 vap->va_atime = ia->ia_atime;
241 vap->va_mtime = ia->ia_mtime;
242 vap->va_ctime = ia->ia_ctime;
243
244 error = -zfs_setattr(dentry->d_inode, vap, 0, cr);
245
246 kmem_free(vap, sizeof(vattr_t));
81e97e21 247 crfree(cr);
ee154f01
BB
248 ASSERT3S(error, <=, 0);
249
5484965a 250 return (error);
ee154f01
BB
251}
252
253static int
254zpl_rename(struct inode *sdip, struct dentry *sdentry,
255 struct inode *tdip, struct dentry *tdentry)
256{
81e97e21 257 cred_t *cr = CRED();
ee154f01
BB
258 int error;
259
81e97e21 260 crhold(cr);
ee154f01 261 error = -zfs_rename(sdip, dname(sdentry), tdip, dname(tdentry), cr, 0);
81e97e21 262 crfree(cr);
ee154f01
BB
263 ASSERT3S(error, <=, 0);
264
265 return (error);
266}
267
268static int
269zpl_symlink(struct inode *dir, struct dentry *dentry, const char *name)
270{
81e97e21 271 cred_t *cr = CRED();
ee154f01
BB
272 vattr_t *vap;
273 struct inode *ip;
274 int error;
275
81e97e21 276 crhold(cr);
ee154f01 277 vap = kmem_zalloc(sizeof(vattr_t), KM_SLEEP);
7b3e34ba 278 zpl_vap_init(vap, dir, S_IFLNK | S_IRWXUGO, cr);
ee154f01
BB
279
280 error = -zfs_symlink(dir, dname(dentry), vap, (char *)name, &ip, cr, 0);
7b3e34ba
BB
281 if (error == 0) {
282 d_instantiate(dentry, ip);
283 d_set_d_op(dentry, &zpl_dentry_operations);
284 }
285
ee154f01 286 kmem_free(vap, sizeof(vattr_t));
81e97e21 287 crfree(cr);
ee154f01
BB
288 ASSERT3S(error, <=, 0);
289
290 return (error);
291}
292
293static void *
294zpl_follow_link(struct dentry *dentry, struct nameidata *nd)
295{
81e97e21 296 cred_t *cr = CRED();
8b4f9a2d
BB
297 struct inode *ip = dentry->d_inode;
298 struct iovec iov;
299 uio_t uio;
300 char *link;
8b4f9a2d
BB
301 int error;
302
81e97e21 303 crhold(cr);
8b4f9a2d
BB
304
305 iov.iov_len = MAXPATHLEN;
306 iov.iov_base = link = kmem_zalloc(MAXPATHLEN, KM_SLEEP);
307
308 uio.uio_iov = &iov;
309 uio.uio_iovcnt = 1;
310 uio.uio_resid = (MAXPATHLEN - 1);
311 uio.uio_segflg = UIO_SYSSPACE;
312
50950001 313 error = -zfs_readlink(ip, &uio, cr);
8b4f9a2d
BB
314 if (error) {
315 kmem_free(link, MAXPATHLEN);
316 nd_set_link(nd, ERR_PTR(error));
317 } else {
318 nd_set_link(nd, link);
319 }
320
81e97e21 321 crfree(cr);
8b4f9a2d 322 return (NULL);
ee154f01
BB
323}
324
325static void
326zpl_put_link(struct dentry *dentry, struct nameidata *nd, void *ptr)
327{
0a6b03d3 328 const char *link = nd_get_link(nd);
ee154f01 329
ee154f01 330 if (!IS_ERR(link))
8b4f9a2d 331 kmem_free(link, MAXPATHLEN);
ee154f01
BB
332}
333
334static int
335zpl_link(struct dentry *old_dentry, struct inode *dir, struct dentry *dentry)
336{
81e97e21 337 cred_t *cr = CRED();
ee154f01 338 struct inode *ip = old_dentry->d_inode;
ee154f01
BB
339 int error;
340
341 if (ip->i_nlink >= ZFS_LINK_MAX)
342 return -EMLINK;
343
81e97e21 344 crhold(cr);
ee154f01
BB
345 ip->i_ctime = CURRENT_TIME_SEC;
346 igrab(ip); /* Use ihold() if available */
347
348 error = -zfs_link(dir, ip, dname(dentry), cr);
349 if (error) {
350 iput(ip);
351 goto out;
352 }
353
354 d_instantiate(dentry, ip);
7b3e34ba 355 d_set_d_op(dentry, &zpl_dentry_operations);
ee154f01 356out:
81e97e21 357 crfree(cr);
ee154f01
BB
358 ASSERT3S(error, <=, 0);
359
360 return (error);
361}
362
ea1fdf46 363#ifdef HAVE_INODE_TRUNCATE_RANGE
5cb63a57
ED
364static void
365zpl_truncate_range(struct inode* ip, loff_t start, loff_t end)
366{
367 cred_t *cr = CRED();
368 flock64_t bf;
369
370 ASSERT3S(start, <=, end);
371
372 /*
373 * zfs_freesp() will interpret (len == 0) as meaning "truncate until
374 * the end of the file". We don't want that.
375 */
376 if (start == end)
377 return;
378
379 crhold(cr);
380
381 bf.l_type = F_WRLCK;
382 bf.l_whence = 0;
383 bf.l_start = start;
384 bf.l_len = end - start;
385 bf.l_pid = 0;
386 zfs_space(ip, F_FREESP, &bf, FWRITE, start, cr);
387
388 crfree(cr);
389}
ea1fdf46 390#endif /* HAVE_INODE_TRUNCATE_RANGE */
5cb63a57 391
cb2d1901
ED
392#ifdef HAVE_INODE_FALLOCATE
393static long
394zpl_fallocate(struct inode *ip, int mode, loff_t offset, loff_t len)
395{
396 return zpl_fallocate_common(ip, mode, offset, len);
397}
398#endif /* HAVE_INODE_FALLOCATE */
399
7b3e34ba
BB
400static int
401#ifdef HAVE_D_REVALIDATE_NAMEIDATA
402zpl_revalidate(struct dentry *dentry, struct nameidata *nd)
403{
404 unsigned int flags = nd->flags;
405#else
406zpl_revalidate(struct dentry *dentry, unsigned int flags)
407{
408#endif /* HAVE_D_REVALIDATE_NAMEIDATA */
409 zfs_sb_t *zsb = dentry->d_sb->s_fs_info;
410 int error;
411
412 if (flags & LOOKUP_RCU)
413 return (-ECHILD);
414
415 /*
416 * After a rollback negative dentries created before the rollback
417 * time must be invalidated. Otherwise they can obscure files which
418 * are only present in the rolled back dataset.
419 */
420 if (dentry->d_inode == NULL) {
421 spin_lock(&dentry->d_lock);
422 error = time_before(dentry->d_time, zsb->z_rollback_time);
423 spin_unlock(&dentry->d_lock);
424
425 if (error)
426 return (0);
427 }
428
429 /*
430 * The dentry may reference a stale inode if a mounted file system
431 * was rolled back to a point in time where the object didn't exist.
432 */
433 if (dentry->d_inode && ITOZ(dentry->d_inode)->z_is_stale)
434 return (0);
435
436 return (1);
437}
cb2d1901 438
ee154f01 439const struct inode_operations zpl_inode_operations = {
ee154f01
BB
440 .create = zpl_create,
441 .link = zpl_link,
442 .unlink = zpl_unlink,
443 .symlink = zpl_symlink,
444 .mkdir = zpl_mkdir,
445 .rmdir = zpl_rmdir,
446 .mknod = zpl_mknod,
447 .rename = zpl_rename,
448 .setattr = zpl_setattr,
449 .getattr = zpl_getattr,
450 .setxattr = generic_setxattr,
451 .getxattr = generic_getxattr,
452 .removexattr = generic_removexattr,
453 .listxattr = zpl_xattr_list,
ea1fdf46 454#ifdef HAVE_INODE_TRUNCATE_RANGE
5cb63a57 455 .truncate_range = zpl_truncate_range,
ea1fdf46 456#endif /* HAVE_INODE_TRUNCATE_RANGE */
cb2d1901
ED
457#ifdef HAVE_INODE_FALLOCATE
458 .fallocate = zpl_fallocate,
459#endif /* HAVE_INODE_FALLOCATE */
ee154f01
BB
460};
461
462const struct inode_operations zpl_dir_inode_operations = {
ee154f01
BB
463 .create = zpl_create,
464 .lookup = zpl_lookup,
465 .link = zpl_link,
466 .unlink = zpl_unlink,
467 .symlink = zpl_symlink,
468 .mkdir = zpl_mkdir,
469 .rmdir = zpl_rmdir,
470 .mknod = zpl_mknod,
471 .rename = zpl_rename,
472 .setattr = zpl_setattr,
a6695d83
BB
473 .getattr = zpl_getattr,
474 .setxattr = generic_setxattr,
475 .getxattr = generic_getxattr,
476 .removexattr = generic_removexattr,
477 .listxattr = zpl_xattr_list,
ee154f01
BB
478};
479
480const struct inode_operations zpl_symlink_inode_operations = {
ee154f01
BB
481 .readlink = generic_readlink,
482 .follow_link = zpl_follow_link,
483 .put_link = zpl_put_link,
6f2255ba
BB
484 .setattr = zpl_setattr,
485 .getattr = zpl_getattr,
f31b3ebe
BB
486 .setxattr = generic_setxattr,
487 .getxattr = generic_getxattr,
488 .removexattr = generic_removexattr,
489 .listxattr = zpl_xattr_list,
ee154f01
BB
490};
491
492const struct inode_operations zpl_special_inode_operations = {
a6695d83
BB
493 .setattr = zpl_setattr,
494 .getattr = zpl_getattr,
495 .setxattr = generic_setxattr,
496 .getxattr = generic_getxattr,
497 .removexattr = generic_removexattr,
498 .listxattr = zpl_xattr_list,
ee154f01 499};
7b3e34ba
BB
500
501dentry_operations_t zpl_dentry_operations = {
502 .d_revalidate = zpl_revalidate,
503};