?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Patrick Welche <prlw1@cam.ac.uk>
[netbsd-mini2440.git] / external / cddl / osnet / dist / uts / common / fs / zfs / zfs_replay.c
blobdd94a02a1605dee8427802c69c6a48fd01fb090c
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 2008 Sun Microsystems, Inc. All rights reserved.
23 * Use is subject to license terms.
24 */
25
26 #pragma ident "%Z%%M% %I% %E% SMI"
27
28 #include <sys/types.h>
29 #include <sys/param.h>
30 #include <sys/systm.h>
31 #include <sys/sysmacros.h>
32 #include <sys/cmn_err.h>
33 #include <sys/kmem.h>
34 #include <sys/file.h>
35 #include <sys/fcntl.h>
36 #include <sys/vfs.h>
37 #include <sys/fs/zfs.h>
38 #include <sys/zfs_znode.h>
39 #include <sys/zfs_dir.h>
40 #include <sys/zfs_acl.h>
41 #include <sys/zfs_fuid.h>
42 #include <sys/spa.h>
43 #include <sys/zil.h>
44 #include <sys/byteorder.h>
45 #include <sys/stat.h>
46 #include <sys/acl.h>
47 #include <sys/atomic.h>
48 #include <sys/cred.h>
49 #include <sys/namei.h>
50
51 /*
52 * Functions to replay ZFS intent log (ZIL) records
53 * The functions are called through a function vector (zfs_replay_vector)
54 * which is indexed by the transaction type.
55 */
56
57 static void
58 zfs_init_vattr(vattr_t *vap, uint64_t mask, uint64_t mode,
59 uint64_t uid, uint64_t gid, uint64_t rdev, uint64_t nodeid)
60 {
61 VATTR_NULL(vap);
62 vap->va_mask = (uint_t)mask;
63 vap->va_type = IFTOVT(mode);
64 vap->va_mode = mode & MODEMASK;
65 vap->va_uid = (uid_t)(IS_EPHEMERAL(uid)) ? -1 : uid;
66 vap->va_gid = (gid_t)(IS_EPHEMERAL(gid)) ? -1 : gid;
67 vap->va_rdev = zfs_cmpldev(rdev);
68 vap->va_nodeid = nodeid;
69 }
70
71 /* ARGSUSED */
72 static int
73 zfs_replay_error(zfsvfs_t *zfsvfs, lr_t *lr, boolean_t byteswap)
74 {
75 return (ENOTSUP);
76 }
77
78 static void
79 zfs_replay_xvattr(lr_attr_t *lrattr, xvattr_t *xvap)
80 {
81 xoptattr_t *xoap = NULL;
82 uint64_t *attrs;
83 uint64_t *crtime;
84 uint32_t *bitmap;
85 void *scanstamp;
86 int i;
87
88 xvap->xva_vattr.va_mask |= AT_XVATTR;
89 if ((xoap = xva_getxoptattr(xvap)) == NULL) {
90 xvap->xva_vattr.va_mask &= ~AT_XVATTR; /* shouldn't happen */
91 return;
92 }
93
94 ASSERT(lrattr->lr_attr_masksize == xvap->xva_mapsize);
95
96 bitmap = &lrattr->lr_attr_bitmap;
97 for (i = 0; i != lrattr->lr_attr_masksize; i++, bitmap++)
98 xvap->xva_reqattrmap[i] = *bitmap;
99
100 attrs = (uint64_t *)(lrattr + lrattr->lr_attr_masksize - 1);
101 crtime = attrs + 1;
102 scanstamp = (caddr_t)(crtime + 2);
103
104 if (XVA_ISSET_REQ(xvap, XAT_HIDDEN))
105 xoap->xoa_hidden = ((*attrs & XAT0_HIDDEN) != 0);
106 if (XVA_ISSET_REQ(xvap, XAT_SYSTEM))
107 xoap->xoa_system = ((*attrs & XAT0_SYSTEM) != 0);
108 if (XVA_ISSET_REQ(xvap, XAT_ARCHIVE))
109 xoap->xoa_archive = ((*attrs & XAT0_ARCHIVE) != 0);
110 if (XVA_ISSET_REQ(xvap, XAT_READONLY))
111 xoap->xoa_readonly = ((*attrs & XAT0_READONLY) != 0);
112 if (XVA_ISSET_REQ(xvap, XAT_IMMUTABLE))
113 xoap->xoa_immutable = ((*attrs & XAT0_IMMUTABLE) != 0);
114 if (XVA_ISSET_REQ(xvap, XAT_NOUNLINK))
115 xoap->xoa_nounlink = ((*attrs & XAT0_NOUNLINK) != 0);
116 if (XVA_ISSET_REQ(xvap, XAT_APPENDONLY))
117 xoap->xoa_appendonly = ((*attrs & XAT0_APPENDONLY) != 0);
118 if (XVA_ISSET_REQ(xvap, XAT_NODUMP))
119 xoap->xoa_nodump = ((*attrs & XAT0_NODUMP) != 0);
120 if (XVA_ISSET_REQ(xvap, XAT_OPAQUE))
121 xoap->xoa_opaque = ((*attrs & XAT0_OPAQUE) != 0);
122 if (XVA_ISSET_REQ(xvap, XAT_AV_MODIFIED))
123 xoap->xoa_av_modified = ((*attrs & XAT0_AV_MODIFIED) != 0);
124 if (XVA_ISSET_REQ(xvap, XAT_AV_QUARANTINED))
125 xoap->xoa_av_quarantined =
126 ((*attrs & XAT0_AV_QUARANTINED) != 0);
127 if (XVA_ISSET_REQ(xvap, XAT_CREATETIME))
128 ZFS_TIME_DECODE(&xoap->xoa_createtime, crtime);
129 if (XVA_ISSET_REQ(xvap, XAT_AV_SCANSTAMP))
130 bcopy(scanstamp, xoap->xoa_av_scanstamp, AV_SCANSTAMP_SZ);
131 }
132
133 static int
134 zfs_replay_domain_cnt(uint64_t uid, uint64_t gid)
135 {
136 uint64_t uid_idx;
137 uint64_t gid_idx;
138 int domcnt = 0;
139
140 uid_idx = FUID_INDEX(uid);
141 gid_idx = FUID_INDEX(gid);
142 if (uid_idx)
143 domcnt++;
144 if (gid_idx > 0 && gid_idx != uid_idx)
145 domcnt++;
146
147 return (domcnt);
148 }
149
150 static void *
151 zfs_replay_fuid_domain_common(zfs_fuid_info_t *fuid_infop, void *start,
152 int domcnt)
153 {
154 int i;
155
156 for (i = 0; i != domcnt; i++) {
157 fuid_infop->z_domain_table[i] = start;
158 start = (caddr_t)start + strlen(start) + 1;
159 }
160
161 return (start);
162 }
163
164 /*
165 * Set the uid/gid in the fuid_info structure.
166 */
167 static void
168 zfs_replay_fuid_ugid(zfs_fuid_info_t *fuid_infop, uint64_t uid, uint64_t gid)
169 {
170 /*
171 * If owner or group are log specific FUIDs then slurp up
172 * domain information and build zfs_fuid_info_t
173 */
174 if (IS_EPHEMERAL(uid))
175 fuid_infop->z_fuid_owner = uid;
176
177 if (IS_EPHEMERAL(gid))
178 fuid_infop->z_fuid_group = gid;
179 }
180
181 /*
182 * Load fuid domains into fuid_info_t
183 */
184 static zfs_fuid_info_t *
185 zfs_replay_fuid_domain(void *buf, void **end, uint64_t uid, uint64_t gid)
186 {
187 int domcnt;
188
189 zfs_fuid_info_t *fuid_infop;
190
191 fuid_infop = zfs_fuid_info_alloc();
192
193 domcnt = zfs_replay_domain_cnt(uid, gid);
194
195 if (domcnt == 0)
196 return (fuid_infop);
197
198 fuid_infop->z_domain_table =
199 kmem_zalloc(domcnt * sizeof (char **), KM_SLEEP);
200
201 zfs_replay_fuid_ugid(fuid_infop, uid, gid);
202
203 fuid_infop->z_domain_cnt = domcnt;
204 *end = zfs_replay_fuid_domain_common(fuid_infop, buf, domcnt);
205 return (fuid_infop);
206 }
207
208 /*
209 * load zfs_fuid_t's and fuid_domains into fuid_info_t
210 */
211 static zfs_fuid_info_t *
212 zfs_replay_fuids(void *start, void **end, int idcnt, int domcnt, uint64_t uid,
213 uint64_t gid)
214 {
215 uint64_t *log_fuid = (uint64_t *)start;
216 zfs_fuid_info_t *fuid_infop;
217 int i;
218
219 fuid_infop = zfs_fuid_info_alloc();
220 fuid_infop->z_domain_cnt = domcnt;
221
222 fuid_infop->z_domain_table =
223 kmem_zalloc(domcnt * sizeof (char **), KM_SLEEP);
224
225 for (i = 0; i != idcnt; i++) {
226 zfs_fuid_t *zfuid;
227
228 zfuid = kmem_alloc(sizeof (zfs_fuid_t), KM_SLEEP);
229 zfuid->z_logfuid = *log_fuid;
230 zfuid->z_id = -1;
231 zfuid->z_domidx = 0;
232 list_insert_tail(&fuid_infop->z_fuids, zfuid);
233 log_fuid++;
234 }
235
236 zfs_replay_fuid_ugid(fuid_infop, uid, gid);
237
238 *end = zfs_replay_fuid_domain_common(fuid_infop, log_fuid, domcnt);
239 return (fuid_infop);
240 }
241
242 static void
243 zfs_replay_swap_attrs(lr_attr_t *lrattr)
244 {
245 /* swap the lr_attr structure */
246 byteswap_uint32_array(lrattr, sizeof (*lrattr));
247 /* swap the bitmap */
248 byteswap_uint32_array(lrattr + 1, (lrattr->lr_attr_masksize - 1) *
249 sizeof (uint32_t));
250 /* swap the attributes, create time + 64 bit word for attributes */
251 byteswap_uint64_array((caddr_t)(lrattr + 1) + (sizeof (uint32_t) *
252 (lrattr->lr_attr_masksize - 1)), 3 * sizeof (uint64_t));
253 }
254
255 /*
256 * Replay file create with optional ACL, xvattr information as well
257 * as option FUID information.
258 */
259 static int
260 zfs_replay_create_acl(zfsvfs_t *zfsvfs,
261 lr_acl_create_t *lracl, boolean_t byteswap)
262 {
263 char *name = NULL; /* location determined later */
264 lr_create_t *lr = (lr_create_t *)lracl;
265 znode_t *dzp;
266 vnode_t *vp = NULL;
267 xvattr_t xva;
268 int vflg = 0;
269 vsecattr_t vsec = { 0 };
270 lr_attr_t *lrattr;
271 void *aclstart;
272 void *fuidstart;
273 size_t xvatlen = 0;
274 uint64_t txtype;
275 int error;
276
277 if (byteswap) {
278 byteswap_uint64_array(lracl, sizeof (*lracl));
279 txtype = (int)lr->lr_common.lrc_txtype;
280 if (txtype == TX_CREATE_ACL_ATTR ||
281 txtype == TX_MKDIR_ACL_ATTR) {
282 lrattr = (lr_attr_t *)(caddr_t)(lracl + 1);
283 zfs_replay_swap_attrs(lrattr);
284 xvatlen = ZIL_XVAT_SIZE(lrattr->lr_attr_masksize);
285 }
286
287 aclstart = (caddr_t)(lracl + 1) + xvatlen;
288 zfs_ace_byteswap(aclstart, lracl->lr_acl_bytes, B_FALSE);
289 /* swap fuids */
290 if (lracl->lr_fuidcnt) {
291 byteswap_uint64_array((caddr_t)aclstart +
292 ZIL_ACE_LENGTH(lracl->lr_acl_bytes),
293 lracl->lr_fuidcnt * sizeof (uint64_t));
294 }
295 }
296
297 if ((error = zfs_zget(zfsvfs, lr->lr_doid, &dzp)) != 0)
298 return (error);
299
300 xva_init(&xva);
301 zfs_init_vattr(&xva.xva_vattr, AT_TYPE | AT_MODE | AT_UID | AT_GID,
302 lr->lr_mode, lr->lr_uid, lr->lr_gid, lr->lr_rdev, lr->lr_foid);
303
304 /*
305 * All forms of zfs create (create, mkdir, mkxattrdir, symlink)
306 * eventually end up in zfs_mknode(), which assigns the object's
307 * creation time and generation number. The generic VOP_CREATE()
308 * doesn't have either concept, so we smuggle the values inside
309 * the vattr's otherwise unused va_ctime and va_nblocks fields.
310 */
311 ZFS_TIME_DECODE(&xva.xva_vattr.va_ctime, lr->lr_crtime);
312 xva.xva_vattr.va_nblocks = lr->lr_gen;
313
314 error = dmu_object_info(zfsvfs->z_os, lr->lr_foid, NULL);
315 if (error != ENOENT)
316 goto bail;
317
318 if (lr->lr_common.lrc_txtype & TX_CI)
319 vflg |= FIGNORECASE;
320 switch ((int)lr->lr_common.lrc_txtype) {
321 case TX_CREATE_ACL:
322 aclstart = (caddr_t)(lracl + 1);
323 fuidstart = (caddr_t)aclstart +
324 ZIL_ACE_LENGTH(lracl->lr_acl_bytes);
325 zfsvfs->z_fuid_replay = zfs_replay_fuids(fuidstart,
326 (void *)&name, lracl->lr_fuidcnt, lracl->lr_domcnt,
327 lr->lr_uid, lr->lr_gid);
328 /*FALLTHROUGH*/
329 case TX_CREATE_ACL_ATTR:
330 if (name == NULL) {
331 lrattr = (lr_attr_t *)(caddr_t)(lracl + 1);
332 xvatlen = ZIL_XVAT_SIZE(lrattr->lr_attr_masksize);
333 xva.xva_vattr.va_mask |= AT_XVATTR;
334 zfs_replay_xvattr(lrattr, &xva);
335 }
336 vsec.vsa_mask = VSA_ACE | VSA_ACE_ACLFLAGS;
337 vsec.vsa_aclentp = (caddr_t)(lracl + 1) + xvatlen;
338 vsec.vsa_aclcnt = lracl->lr_aclcnt;
339 vsec.vsa_aclentsz = lracl->lr_acl_bytes;
340 vsec.vsa_aclflags = lracl->lr_acl_flags;
341 if (zfsvfs->z_fuid_replay == NULL) {
342 fuidstart = (caddr_t)(lracl + 1) + xvatlen +
343 ZIL_ACE_LENGTH(lracl->lr_acl_bytes);
344 zfsvfs->z_fuid_replay =
345 zfs_replay_fuids(fuidstart,
346 (void *)&name, lracl->lr_fuidcnt, lracl->lr_domcnt,
347 lr->lr_uid, lr->lr_gid);
348 }
349
350 #ifdef TODO
351 error = VOP_CREATE(ZTOV(dzp), name, &xva.xva_vattr,
352 0, 0, &vp, kcred, vflg, NULL, &vsec);
353 #else
354 panic("%s:%u: unsupported condition", __func__, __LINE__);
355 #endif
356 break;
357 case TX_MKDIR_ACL:
358 aclstart = (caddr_t)(lracl + 1);
359 fuidstart = (caddr_t)aclstart +
360 ZIL_ACE_LENGTH(lracl->lr_acl_bytes);
361 zfsvfs->z_fuid_replay = zfs_replay_fuids(fuidstart,
362 (void *)&name, lracl->lr_fuidcnt, lracl->lr_domcnt,
363 lr->lr_uid, lr->lr_gid);
364 /*FALLTHROUGH*/
365 case TX_MKDIR_ACL_ATTR:
366 if (name == NULL) {
367 lrattr = (lr_attr_t *)(caddr_t)(lracl + 1);
368 xvatlen = ZIL_XVAT_SIZE(lrattr->lr_attr_masksize);
369 zfs_replay_xvattr(lrattr, &xva);
370 }
371 vsec.vsa_mask = VSA_ACE | VSA_ACE_ACLFLAGS;
372 vsec.vsa_aclentp = (caddr_t)(lracl + 1) + xvatlen;
373 vsec.vsa_aclcnt = lracl->lr_aclcnt;
374 vsec.vsa_aclentsz = lracl->lr_acl_bytes;
375 vsec.vsa_aclflags = lracl->lr_acl_flags;
376 if (zfsvfs->z_fuid_replay == NULL) {
377 fuidstart = (caddr_t)(lracl + 1) + xvatlen +
378 ZIL_ACE_LENGTH(lracl->lr_acl_bytes);
379 zfsvfs->z_fuid_replay =
380 zfs_replay_fuids(fuidstart,
381 (void *)&name, lracl->lr_fuidcnt, lracl->lr_domcnt,
382 lr->lr_uid, lr->lr_gid);
383 }
384 #ifdef TODO
385 error = VOP_MKDIR(ZTOV(dzp), name, &xva.xva_vattr,
386 &vp, kcred, NULL, vflg, &vsec);
387 #else
388 panic("%s:%u: unsupported condition", __func__, __LINE__);
389 #endif
390 break;
391 default:
392 error = ENOTSUP;
393 }
394
395 bail:
396 if (error == 0 && vp != NULL)
397 VN_RELE(vp);
398
399 VN_RELE(ZTOV(dzp));
400
401 zfs_fuid_info_free(zfsvfs->z_fuid_replay);
402 zfsvfs->z_fuid_replay = NULL;
403
404 return (error);
405 }
406
407 static int
408 zfs_replay_create(zfsvfs_t *zfsvfs, lr_create_t *lr, boolean_t byteswap)
409 {
410 char *name = NULL; /* location determined later */
411 char *link; /* symlink content follows name */
412 znode_t *dzp;
413 vnode_t *vp = NULL;
414 xvattr_t xva;
415 int vflg = 0;
416 size_t lrsize = sizeof (lr_create_t);
417 lr_attr_t *lrattr;
418 void *start;
419 size_t xvatlen;
420 uint64_t txtype;
421 struct componentname cn;
422 int error;
423
424 if (byteswap) {
425 byteswap_uint64_array(lr, sizeof (*lr));
426 txtype = (int)lr->lr_common.lrc_txtype;
427 if (txtype == TX_CREATE_ATTR || txtype == TX_MKDIR_ATTR)
428 zfs_replay_swap_attrs((lr_attr_t *)(lr + 1));
429 }
430
431
432 if ((error = zfs_zget(zfsvfs, lr->lr_doid, &dzp)) != 0)
433 return (error);
434
435 xva_init(&xva);
436 zfs_init_vattr(&xva.xva_vattr, AT_TYPE | AT_MODE | AT_UID | AT_GID,
437 lr->lr_mode, lr->lr_uid, lr->lr_gid, lr->lr_rdev, lr->lr_foid);
438
439 /*
440 * All forms of zfs create (create, mkdir, mkxattrdir, symlink)
441 * eventually end up in zfs_mknode(), which assigns the object's
442 * creation time and generation number. The generic VOP_CREATE()
443 * doesn't have either concept, so we smuggle the values inside
444 * the vattr's otherwise unused va_ctime and va_nblocks fields.
445 */
446 ZFS_TIME_DECODE(&xva.xva_vattr.va_ctime, lr->lr_crtime);
447 xva.xva_vattr.va_nblocks = lr->lr_gen;
448
449 error = dmu_object_info(zfsvfs->z_os, lr->lr_foid, NULL);
450 if (error != ENOENT)
451 goto out;
452
453 if (lr->lr_common.lrc_txtype & TX_CI)
454 vflg |= FIGNORECASE;
455
456 /*
457 * Symlinks don't have fuid info, and CIFS never creates
458 * symlinks.
459 *
460 * The _ATTR versions will grab the fuid info in their subcases.
461 */
462 if ((int)lr->lr_common.lrc_txtype != TX_SYMLINK &&
463 (int)lr->lr_common.lrc_txtype != TX_MKDIR_ATTR &&
464 (int)lr->lr_common.lrc_txtype != TX_CREATE_ATTR) {
465 start = (lr + 1);
466 zfsvfs->z_fuid_replay =
467 zfs_replay_fuid_domain(start, &start,
468 lr->lr_uid, lr->lr_gid);
469 }
470
471 cn.cn_cred = kcred;
472 cn.cn_flags = SAVENAME;
473
474 vn_lock(ZTOV(dzp), LK_EXCLUSIVE | LK_RETRY);
475 switch ((int)lr->lr_common.lrc_txtype) {
476 case TX_CREATE_ATTR:
477 lrattr = (lr_attr_t *)(caddr_t)(lr + 1);
478 xvatlen = ZIL_XVAT_SIZE(lrattr->lr_attr_masksize);
479 zfs_replay_xvattr((lr_attr_t *)((caddr_t)lr + lrsize), &xva);
480 start = (caddr_t)(lr + 1) + xvatlen;
481 zfsvfs->z_fuid_replay =
482 zfs_replay_fuid_domain(start, &start,
483 lr->lr_uid, lr->lr_gid);
484 name = (char *)start;
485
486 /*FALLTHROUGH*/
487 case TX_CREATE:
488 if (name == NULL)
489 name = (char *)start;
490
491 cn.cn_nameptr = name;
492 error = VOP_CREATE(ZTOV(dzp), &vp, &cn, &xva.xva_vattr /*,vflg*/);
493 break;
494 case TX_MKDIR_ATTR:
495 lrattr = (lr_attr_t *)(caddr_t)(lr + 1);
496 xvatlen = ZIL_XVAT_SIZE(lrattr->lr_attr_masksize);
497 zfs_replay_xvattr((lr_attr_t *)((caddr_t)lr + lrsize), &xva);
498 start = (caddr_t)(lr + 1) + xvatlen;
499 zfsvfs->z_fuid_replay =
500 zfs_replay_fuid_domain(start, &start,
501 lr->lr_uid, lr->lr_gid);
502 name = (char *)start;
503
504 /*FALLTHROUGH*/
505 case TX_MKDIR:
506 if (name == NULL)
507 name = (char *)(lr + 1);
508
509 cn.cn_nameptr = name;
510 error = VOP_MKDIR(ZTOV(dzp), &vp, &cn, &xva.xva_vattr /*,vflg*/);
511 break;
512 case TX_MKXATTR:
513 name = (char *)(lr + 1);
514 error = zfs_make_xattrdir(dzp, &xva.xva_vattr, &vp, kcred);
515 break;
516 case TX_SYMLINK:
517 name = (char *)(lr + 1);
518 link = name + strlen(name) + 1;
519 cn.cn_nameptr = name;
520 error = VOP_SYMLINK(ZTOV(dzp), &vp, &cn, &xva.xva_vattr, link /*,vflg*/);
521 break;
522 default:
523 error = ENOTSUP;
524 }
525 VOP_UNLOCK(ZTOV(dzp), 0);
526
527 out:
528 if (error == 0 && vp != NULL) {
529 VOP_UNLOCK(vp, 0);
530 VN_RELE(vp);
531 }
532
533 VN_RELE(ZTOV(dzp));
534
535 if (zfsvfs->z_fuid_replay)
536 zfs_fuid_info_free(zfsvfs->z_fuid_replay);
537 zfsvfs->z_fuid_replay = NULL;
538 return (error);
539 }
540
541 static int
542 zfs_replay_remove(zfsvfs_t *zfsvfs, lr_remove_t *lr, boolean_t byteswap)
543 {
544 char *name = (char *)(lr + 1); /* name follows lr_remove_t */
545 znode_t *dzp;
546 struct componentname cn;
547 vnode_t *vp;
548 int error;
549 int vflg = 0;
550
551 if (byteswap)
552 byteswap_uint64_array(lr, sizeof (*lr));
553
554 if ((error = zfs_zget(zfsvfs, lr->lr_doid, &dzp)) != 0)
555 return (error);
556
557 if (lr->lr_common.lrc_txtype & TX_CI)
558 vflg |= FIGNORECASE;
559 cn.cn_nameptr = name;
560 cn.cn_namelen = strlen(name);
561 cn.cn_nameiop = DELETE;
562 cn.cn_flags = ISLASTCN | SAVENAME;
563 //cn.cn_lkflags = LK_EXCLUSIVE | LK_RETRY;
564 cn.cn_cred = kcred;
565 vn_lock(ZTOV(dzp), LK_EXCLUSIVE | LK_RETRY);
566 error = VOP_LOOKUP(ZTOV(dzp), &vp, &cn);
567 if (error != 0) {
568 VOP_UNLOCK(ZTOV(dzp), 0);
569 goto fail;
570 }
571
572 switch ((int)lr->lr_common.lrc_txtype) {
573 case TX_REMOVE:
574 error = VOP_REMOVE(ZTOV(dzp), vp, &cn /*,vflg*/);
575 break;
576 case TX_RMDIR:
577 error = VOP_RMDIR(ZTOV(dzp), vp, &cn /*,vflg*/);
578 break;
579 default:
580 error = ENOTSUP;
581 }
582 vput(vp);
583 VOP_UNLOCK(ZTOV(dzp), 0);
584 fail:
585 VN_RELE(ZTOV(dzp));
586
587 return (error);
588 }
589
590 static int
591 zfs_replay_link(zfsvfs_t *zfsvfs, lr_link_t *lr, boolean_t byteswap)
592 {
593 char *name = (char *)(lr + 1); /* name follows lr_link_t */
594 znode_t *dzp, *zp;
595 struct componentname cn;
596 int error;
597 int vflg = 0;
598
599 if (byteswap)
600 byteswap_uint64_array(lr, sizeof (*lr));
601
602 if ((error = zfs_zget(zfsvfs, lr->lr_doid, &dzp)) != 0)
603 return (error);
604
605 if ((error = zfs_zget(zfsvfs, lr->lr_link_obj, &zp)) != 0) {
606 VN_RELE(ZTOV(dzp));
607 return (error);
608 }
609
610 if (lr->lr_common.lrc_txtype & TX_CI)
611 vflg |= FIGNORECASE;
612 cn.cn_nameptr = name;
613 cn.cn_cred = kcred;
614 cn.cn_flags = SAVENAME;
615
616 vn_lock(ZTOV(dzp), LK_EXCLUSIVE | LK_RETRY);
617 vn_lock(ZTOV(zp), LK_EXCLUSIVE | LK_RETRY);
618 error = VOP_LINK(ZTOV(dzp), ZTOV(zp), &cn /*,vflg*/);
619 VOP_UNLOCK(ZTOV(zp), 0);
620 VOP_UNLOCK(ZTOV(dzp), 0);
621
622 VN_RELE(ZTOV(zp));
623 VN_RELE(ZTOV(dzp));
624
625 return (error);
626 }
627
628 static int
629 zfs_replay_rename(zfsvfs_t *zfsvfs, lr_rename_t *lr, boolean_t byteswap)
630 {
631 char *sname = (char *)(lr + 1); /* sname and tname follow lr_rename_t */
632 char *tname = sname + strlen(sname) + 1;
633 znode_t *sdzp, *tdzp;
634 struct componentname scn, tcn;
635 vnode_t *svp, *tvp;
636 kthread_t *td = curthread;
637 int error;
638 int vflg = 0;
639
640 if (byteswap)
641 byteswap_uint64_array(lr, sizeof (*lr));
642
643 if ((error = zfs_zget(zfsvfs, lr->lr_sdoid, &sdzp)) != 0)
644 return (error);
645
646 if ((error = zfs_zget(zfsvfs, lr->lr_tdoid, &tdzp)) != 0) {
647 VN_RELE(ZTOV(sdzp));
648 return (error);
649 }
650
651 if (lr->lr_common.lrc_txtype & TX_CI)
652 vflg |= FIGNORECASE;
653 svp = tvp = NULL;
654
655 scn.cn_nameptr = sname;
656 scn.cn_namelen = strlen(sname);
657 scn.cn_nameiop = DELETE;
658 scn.cn_flags = ISLASTCN | SAVENAME;
659 // scn.cn_lkflags = LK_EXCLUSIVE | LK_RETRY;
660 scn.cn_cred = kcred;
661 vn_lock(ZTOV(sdzp), LK_EXCLUSIVE | LK_RETRY);
662 error = VOP_LOOKUP(ZTOV(sdzp), &svp, &scn);
663 VOP_UNLOCK(ZTOV(sdzp), 0);
664 if (error != 0)
665 goto fail;
666 VOP_UNLOCK(svp, 0);
667
668 tcn.cn_nameptr = tname;
669 tcn.cn_namelen = strlen(tname);
670 tcn.cn_nameiop = RENAME;
671 tcn.cn_flags = ISLASTCN | SAVENAME;
672 // tcn.cn_lkflags = LK_EXCLUSIVE | LK_RETRY;
673 tcn.cn_cred = kcred;
674 vn_lock(ZTOV(tdzp), LK_EXCLUSIVE | LK_RETRY);
675 error = VOP_LOOKUP(ZTOV(tdzp), &tvp, &tcn);
676 if (error == EJUSTRETURN)
677 tvp = NULL;
678 else if (error != 0) {
679 VOP_UNLOCK(ZTOV(tdzp), 0);
680 goto fail;
681 }
682
683 error = VOP_RENAME(ZTOV(sdzp), svp, &scn, ZTOV(tdzp), tvp, &tcn /*,vflg*/);
684 return (error);
685 fail:
686 if (svp != NULL)
687 vrele(svp);
688 if (tvp != NULL)
689 vrele(tvp);
690 VN_RELE(ZTOV(tdzp));
691 VN_RELE(ZTOV(sdzp));
692
693 return (error);
694 }
695
696 static int
697 zfs_replay_write(zfsvfs_t *zfsvfs, lr_write_t *lr, boolean_t byteswap)
698 {
699 char *data = (char *)(lr + 1); /* data follows lr_write_t */
700 znode_t *zp;
701 int error;
702 ssize_t resid;
703
704 if (byteswap)
705 byteswap_uint64_array(lr, sizeof (*lr));
706
707 if ((error = zfs_zget(zfsvfs, lr->lr_foid, &zp)) != 0) {
708 /*
709 * As we can log writes out of order, it's possible the
710 * file has been removed. In this case just drop the write
711 * and return success.
712 */
713 if (error == ENOENT)
714 error = 0;
715 return (error);
716 }
717
718 error = vn_rdwr(UIO_WRITE, ZTOV(zp), data, lr->lr_length,
719 lr->lr_offset, UIO_SYSSPACE, 0, RLIM64_INFINITY, kcred, &resid);
720
721 VN_RELE(ZTOV(zp));
722
723 return (error);
724 }
725
726 static int
727 zfs_replay_truncate(zfsvfs_t *zfsvfs, lr_truncate_t *lr, boolean_t byteswap)
728 {
729
730 ZFS_LOG(0, "Unexpected code path, report to pjd@FreeBSD.org");
731 return (EOPNOTSUPP);
732 }
733
734 static int
735 zfs_replay_setattr(zfsvfs_t *zfsvfs, lr_setattr_t *lr, boolean_t byteswap)
736 {
737 znode_t *zp;
738 xvattr_t xva;
739 vattr_t *vap = &xva.xva_vattr;
740 vnode_t *vp;
741 int error;
742 void *start;
743
744 xva_init(&xva);
745 if (byteswap) {
746 byteswap_uint64_array(lr, sizeof (*lr));
747
748 if ((lr->lr_mask & AT_XVATTR) &&
749 zfsvfs->z_version >= ZPL_VERSION_INITIAL)
750 zfs_replay_swap_attrs((lr_attr_t *)(lr + 1));
751 }
752
753 if ((error = zfs_zget(zfsvfs, lr->lr_foid, &zp)) != 0) {
754 /*
755 * As we can log setattrs out of order, it's possible the
756 * file has been removed. In this case just drop the setattr
757 * and return success.
758 */
759 if (error == ENOENT)
760 error = 0;
761 return (error);
762 }
763
764 zfs_init_vattr(vap, lr->lr_mask, lr->lr_mode,
765 lr->lr_uid, lr->lr_gid, 0, lr->lr_foid);
766
767 vap->va_size = lr->lr_size;
768 ZFS_TIME_DECODE(&vap->va_atime, lr->lr_atime);
769 ZFS_TIME_DECODE(&vap->va_mtime, lr->lr_mtime);
770
771 /*
772 * Fill in xvattr_t portions if necessary.
773 */
774
775 start = (lr_setattr_t *)(lr + 1);
776 if (vap->va_mask & AT_XVATTR) {
777 zfs_replay_xvattr((lr_attr_t *)start, &xva);
778 start = (caddr_t)start +
779 ZIL_XVAT_SIZE(((lr_attr_t *)start)->lr_attr_masksize);
780 } else
781 xva.xva_vattr.va_mask &= ~AT_XVATTR;
782
783 zfsvfs->z_fuid_replay = zfs_replay_fuid_domain(start, &start,
784 lr->lr_uid, lr->lr_gid);
785
786 vp = ZTOV(zp);
787 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
788 error = VOP_SETATTR(vp, vap, kcred);
789 VOP_UNLOCK(vp, 0);
790
791 zfs_fuid_info_free(zfsvfs->z_fuid_replay);
792 zfsvfs->z_fuid_replay = NULL;
793 VN_RELE(vp);
794
795 return (error);
796 }
797
798 static int
799 zfs_replay_acl_v0(zfsvfs_t *zfsvfs, lr_acl_v0_t *lr, boolean_t byteswap)
800 {
801 ace_t *ace = (ace_t *)(lr + 1); /* ace array follows lr_acl_t */
802 vsecattr_t vsa;
803 znode_t *zp;
804 int error;
805
806 if (byteswap) {
807 byteswap_uint64_array(lr, sizeof (*lr));
808 zfs_oldace_byteswap(ace, lr->lr_aclcnt);
809 }
810
811 if ((error = zfs_zget(zfsvfs, lr->lr_foid, &zp)) != 0) {
812 /*
813 * As we can log acls out of order, it's possible the
814 * file has been removed. In this case just drop the acl
815 * and return success.
816 */
817 if (error == ENOENT)
818 error = 0;
819 return (error);
820 }
821
822 bzero(&vsa, sizeof (vsa));
823 vsa.vsa_mask = VSA_ACE | VSA_ACECNT;
824 vsa.vsa_aclcnt = lr->lr_aclcnt;
825 vsa.vsa_aclentsz = sizeof (ace_t) * vsa.vsa_aclcnt;
826 vsa.vsa_aclflags = 0;
827 vsa.vsa_aclentp = ace;
828
829 #ifdef TODO
830 error = VOP_SETSECATTR(ZTOV(zp), &vsa, 0, kcred, NULL);
831 #else
832 panic("%s:%u: unsupported condition", __func__, __LINE__);
833 #endif
834
835 VN_RELE(ZTOV(zp));
836
837 return (error);
838 }
839
840 /*
841 * Replaying ACLs is complicated by FUID support.
842 * The log record may contain some optional data
843 * to be used for replaying FUID's. These pieces
844 * are the actual FUIDs that were created initially.
845 * The FUID table index may no longer be valid and
846 * during zfs_create() a new index may be assigned.
847 * Because of this the log will contain the original
848 * doman+rid in order to create a new FUID.
849 *
850 * The individual ACEs may contain an ephemeral uid/gid which is no
851 * longer valid and will need to be replaced with an actual FUID.
852 *
853 */
854 static int
855 zfs_replay_acl(zfsvfs_t *zfsvfs, lr_acl_t *lr, boolean_t byteswap)
856 {
857 ace_t *ace = (ace_t *)(lr + 1);
858 vsecattr_t vsa;
859 znode_t *zp;
860 int error;
861
862 if (byteswap) {
863 byteswap_uint64_array(lr, sizeof (*lr));
864 zfs_ace_byteswap(ace, lr->lr_acl_bytes, B_FALSE);
865 if (lr->lr_fuidcnt) {
866 byteswap_uint64_array((caddr_t)ace +
867 ZIL_ACE_LENGTH(lr->lr_acl_bytes),
868 lr->lr_fuidcnt * sizeof (uint64_t));
869 }
870 }
871
872 if ((error = zfs_zget(zfsvfs, lr->lr_foid, &zp)) != 0) {
873 /*
874 * As we can log acls out of order, it's possible the
875 * file has been removed. In this case just drop the acl
876 * and return success.
877 */
878 if (error == ENOENT)
879 error = 0;
880 return (error);
881 }
882
883 #ifdef TODO
884 bzero(&vsa, sizeof (vsa));
885 vsa.vsa_mask = VSA_ACE | VSA_ACECNT | VSA_ACE_ACLFLAGS;
886 vsa.vsa_aclcnt = lr->lr_aclcnt;
887 vsa.vsa_aclentp = ace;
888 vsa.vsa_aclentsz = lr->lr_acl_bytes;
889 vsa.vsa_aclflags = lr->lr_acl_flags;
890
891 if (lr->lr_fuidcnt) {
892 void *fuidstart = (caddr_t)ace +
893 ZIL_ACE_LENGTH(lr->lr_acl_bytes);
894
895 zfsvfs->z_fuid_replay =
896 zfs_replay_fuids(fuidstart, &fuidstart,
897 lr->lr_fuidcnt, lr->lr_domcnt, 0, 0);
898 }
899
900 error = VOP_SETSECATTR(ZTOV(zp), &vsa, 0, kcred, NULL);
901
902 if (zfsvfs->z_fuid_replay)
903 zfs_fuid_info_free(zfsvfs->z_fuid_replay);
904 #else
905 error = EOPNOTSUPP;
906 #endif
907
908 zfsvfs->z_fuid_replay = NULL;
909 VN_RELE(ZTOV(zp));
910
911 return (error);
912 }
913
914 /*
915 * Callback vectors for replaying records
916 */
917 zil_replay_func_t *zfs_replay_vector[TX_MAX_TYPE] = {
918 zfs_replay_error, /* 0 no such transaction type */
919 zfs_replay_create, /* TX_CREATE */
920 zfs_replay_create, /* TX_MKDIR */
921 zfs_replay_create, /* TX_MKXATTR */
922 zfs_replay_create, /* TX_SYMLINK */
923 zfs_replay_remove, /* TX_REMOVE */
924 zfs_replay_remove, /* TX_RMDIR */
925 zfs_replay_link, /* TX_LINK */
926 zfs_replay_rename, /* TX_RENAME */
927 zfs_replay_write, /* TX_WRITE */
928 zfs_replay_truncate, /* TX_TRUNCATE */
929 zfs_replay_setattr, /* TX_SETATTR */
930 zfs_replay_acl_v0, /* TX_ACL_V0 */
931 zfs_replay_acl, /* TX_ACL */
932 zfs_replay_create_acl, /* TX_CREATE_ACL */
933 zfs_replay_create, /* TX_CREATE_ATTR */
934 zfs_replay_create_acl, /* TX_CREATE_ACL_ATTR */
935 zfs_replay_create_acl, /* TX_MKDIR_ACL */
936 zfs_replay_create, /* TX_MKDIR_ATTR */
937 zfs_replay_create_acl, /* TX_MKDIR_ACL_ATTR */
938 };
NetBSD on the FriendlyARM MINI2440