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.
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.
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]
22 * Copyright 2008 Sun Microsystems, Inc. All rights reserved.
23 * Use is subject to license terms.
26 #pragma ident "%Z%%M% %I% %E% SMI"
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>
35 #include <sys/fcntl.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>
44 #include <sys/byteorder.h>
47 #include <sys/atomic.h>
49 #include <sys/namei.h>
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.
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)
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;
73 zfs_replay_error(zfsvfs_t *zfsvfs, lr_t *lr, boolean_t byteswap)
79 zfs_replay_xvattr(lr_attr_t *lrattr, xvattr_t *xvap)
81 xoptattr_t *xoap = NULL;
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 */
94 ASSERT(lrattr->lr_attr_masksize == xvap->xva_mapsize);
96 bitmap = &lrattr->lr_attr_bitmap;
97 for (i = 0; i != lrattr->lr_attr_masksize; i++, bitmap++)
98 xvap->xva_reqattrmap[i] = *bitmap;
100 attrs = (uint64_t *)(lrattr + lrattr->lr_attr_masksize - 1);
102 scanstamp = (caddr_t)(crtime + 2);
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);
134 zfs_replay_domain_cnt(uint64_t uid, uint64_t gid)
140 uid_idx = FUID_INDEX(uid);
141 gid_idx = FUID_INDEX(gid);
144 if (gid_idx > 0 && gid_idx != uid_idx)
151 zfs_replay_fuid_domain_common(zfs_fuid_info_t *fuid_infop, void *start,
156 for (i = 0; i != domcnt; i++) {
157 fuid_infop->z_domain_table[i] = start;
158 start = (caddr_t)start + strlen(start) + 1;
165 * Set the uid/gid in the fuid_info structure.
168 zfs_replay_fuid_ugid(zfs_fuid_info_t *fuid_infop, uint64_t uid, uint64_t gid)
171 * If owner or group are log specific FUIDs then slurp up
172 * domain information and build zfs_fuid_info_t
174 if (IS_EPHEMERAL(uid))
175 fuid_infop->z_fuid_owner = uid;
177 if (IS_EPHEMERAL(gid))
178 fuid_infop->z_fuid_group = gid;
182 * Load fuid domains into fuid_info_t
184 static zfs_fuid_info_t *
185 zfs_replay_fuid_domain(void *buf, void **end, uint64_t uid, uint64_t gid)
189 zfs_fuid_info_t *fuid_infop;
191 fuid_infop = zfs_fuid_info_alloc();
193 domcnt = zfs_replay_domain_cnt(uid, gid);
198 fuid_infop->z_domain_table =
199 kmem_zalloc(domcnt * sizeof (char **), KM_SLEEP);
201 zfs_replay_fuid_ugid(fuid_infop, uid, gid);
203 fuid_infop->z_domain_cnt = domcnt;
204 *end = zfs_replay_fuid_domain_common(fuid_infop, buf, domcnt);
209 * load zfs_fuid_t's and fuid_domains into fuid_info_t
211 static zfs_fuid_info_t *
212 zfs_replay_fuids(void *start, void **end, int idcnt, int domcnt, uint64_t uid,
215 uint64_t *log_fuid = (uint64_t *)start;
216 zfs_fuid_info_t *fuid_infop;
219 fuid_infop = zfs_fuid_info_alloc();
220 fuid_infop->z_domain_cnt = domcnt;
222 fuid_infop->z_domain_table =
223 kmem_zalloc(domcnt * sizeof (char **), KM_SLEEP);
225 for (i = 0; i != idcnt; i++) {
228 zfuid = kmem_alloc(sizeof (zfs_fuid_t), KM_SLEEP);
229 zfuid->z_logfuid = *log_fuid;
232 list_insert_tail(&fuid_infop->z_fuids, zfuid);
236 zfs_replay_fuid_ugid(fuid_infop, uid, gid);
238 *end = zfs_replay_fuid_domain_common(fuid_infop, log_fuid, domcnt);
243 zfs_replay_swap_attrs(lr_attr_t *lrattr)
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) *
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));
256 * Replay file create with optional ACL, xvattr information as well
257 * as option FUID information.
260 zfs_replay_create_acl(zfsvfs_t *zfsvfs,
261 lr_acl_create_t *lracl, boolean_t byteswap)
263 char *name = NULL; /* location determined later */
264 lr_create_t *lr = (lr_create_t *)lracl;
269 vsecattr_t vsec = { 0 };
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);
287 aclstart = (caddr_t)(lracl + 1) + xvatlen;
288 zfs_ace_byteswap(aclstart, lracl->lr_acl_bytes, B_FALSE);
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));
297 if ((error = zfs_zget(zfsvfs, lr->lr_doid, &dzp)) != 0)
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);
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.
311 ZFS_TIME_DECODE(&xva.xva_vattr.va_ctime, lr->lr_crtime);
312 xva.xva_vattr.va_nblocks = lr->lr_gen;
314 error = dmu_object_info(zfsvfs->z_os, lr->lr_foid, NULL);
318 if (lr->lr_common.lrc_txtype & TX_CI)
320 switch ((int)lr->lr_common.lrc_txtype) {
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);
329 case TX_CREATE_ACL_ATTR:
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);
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);
351 error = VOP_CREATE(ZTOV(dzp), name, &xva.xva_vattr,
352 0, 0, &vp, kcred, vflg, NULL, &vsec);
354 panic("%s:%u: unsupported condition", __func__, __LINE__);
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);
365 case TX_MKDIR_ACL_ATTR:
367 lrattr = (lr_attr_t *)(caddr_t)(lracl + 1);
368 xvatlen = ZIL_XVAT_SIZE(lrattr->lr_attr_masksize);
369 zfs_replay_xvattr(lrattr, &xva);
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);
385 error = VOP_MKDIR(ZTOV(dzp), name, &xva.xva_vattr,
386 &vp, kcred, NULL, vflg, &vsec);
388 panic("%s:%u: unsupported condition", __func__, __LINE__);
396 if (error == 0 && vp != NULL)
401 zfs_fuid_info_free(zfsvfs->z_fuid_replay);
402 zfsvfs->z_fuid_replay = NULL;
408 zfs_replay_create(zfsvfs_t *zfsvfs, lr_create_t *lr, boolean_t byteswap)
410 char *name = NULL; /* location determined later */
411 char *link; /* symlink content follows name */
416 size_t lrsize = sizeof (lr_create_t);
421 struct componentname cn;
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));
432 if ((error = zfs_zget(zfsvfs, lr->lr_doid, &dzp)) != 0)
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);
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.
446 ZFS_TIME_DECODE(&xva.xva_vattr.va_ctime, lr->lr_crtime);
447 xva.xva_vattr.va_nblocks = lr->lr_gen;
449 error = dmu_object_info(zfsvfs->z_os, lr->lr_foid, NULL);
453 if (lr->lr_common.lrc_txtype & TX_CI)
457 * Symlinks don't have fuid info, and CIFS never creates
460 * The _ATTR versions will grab the fuid info in their subcases.
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) {
466 zfsvfs->z_fuid_replay =
467 zfs_replay_fuid_domain(start, &start,
468 lr->lr_uid, lr->lr_gid);
472 cn.cn_thread = curthread;
473 cn.cn_flags = SAVENAME;
475 vn_lock(ZTOV(dzp), LK_EXCLUSIVE | LK_RETRY);
476 switch ((int)lr->lr_common.lrc_txtype) {
478 lrattr = (lr_attr_t *)(caddr_t)(lr + 1);
479 xvatlen = ZIL_XVAT_SIZE(lrattr->lr_attr_masksize);
480 zfs_replay_xvattr((lr_attr_t *)((caddr_t)lr + lrsize), &xva);
481 start = (caddr_t)(lr + 1) + xvatlen;
482 zfsvfs->z_fuid_replay =
483 zfs_replay_fuid_domain(start, &start,
484 lr->lr_uid, lr->lr_gid);
485 name = (char *)start;
490 name = (char *)start;
492 cn.cn_nameptr = name;
493 error = VOP_CREATE(ZTOV(dzp), &vp, &cn, &xva.xva_vattr /*,vflg*/);
496 lrattr = (lr_attr_t *)(caddr_t)(lr + 1);
497 xvatlen = ZIL_XVAT_SIZE(lrattr->lr_attr_masksize);
498 zfs_replay_xvattr((lr_attr_t *)((caddr_t)lr + lrsize), &xva);
499 start = (caddr_t)(lr + 1) + xvatlen;
500 zfsvfs->z_fuid_replay =
501 zfs_replay_fuid_domain(start, &start,
502 lr->lr_uid, lr->lr_gid);
503 name = (char *)start;
508 name = (char *)(lr + 1);
510 cn.cn_nameptr = name;
511 error = VOP_MKDIR(ZTOV(dzp), &vp, &cn, &xva.xva_vattr /*,vflg*/);
514 name = (char *)(lr + 1);
515 error = zfs_make_xattrdir(dzp, &xva.xva_vattr, &vp, kcred);
518 name = (char *)(lr + 1);
519 link = name + strlen(name) + 1;
520 cn.cn_nameptr = name;
521 error = VOP_SYMLINK(ZTOV(dzp), &vp, &cn, &xva.xva_vattr, link /*,vflg*/);
526 VOP_UNLOCK(ZTOV(dzp), 0);
529 if (error == 0 && vp != NULL) {
536 if (zfsvfs->z_fuid_replay)
537 zfs_fuid_info_free(zfsvfs->z_fuid_replay);
538 zfsvfs->z_fuid_replay = NULL;
543 zfs_replay_remove(zfsvfs_t *zfsvfs, lr_remove_t *lr, boolean_t byteswap)
545 char *name = (char *)(lr + 1); /* name follows lr_remove_t */
547 struct componentname cn;
553 byteswap_uint64_array(lr, sizeof (*lr));
555 if ((error = zfs_zget(zfsvfs, lr->lr_doid, &dzp)) != 0)
558 if (lr->lr_common.lrc_txtype & TX_CI)
560 cn.cn_nameptr = name;
561 cn.cn_namelen = strlen(name);
562 cn.cn_nameiop = DELETE;
563 cn.cn_flags = ISLASTCN | SAVENAME;
564 cn.cn_lkflags = LK_EXCLUSIVE | LK_RETRY;
566 cn.cn_thread = curthread;
567 vn_lock(ZTOV(dzp), LK_EXCLUSIVE | LK_RETRY);
568 error = VOP_LOOKUP(ZTOV(dzp), &vp, &cn);
570 VOP_UNLOCK(ZTOV(dzp), 0);
574 switch ((int)lr->lr_common.lrc_txtype) {
576 error = VOP_REMOVE(ZTOV(dzp), vp, &cn /*,vflg*/);
579 error = VOP_RMDIR(ZTOV(dzp), vp, &cn /*,vflg*/);
585 VOP_UNLOCK(ZTOV(dzp), 0);
593 zfs_replay_link(zfsvfs_t *zfsvfs, lr_link_t *lr, boolean_t byteswap)
595 char *name = (char *)(lr + 1); /* name follows lr_link_t */
597 struct componentname cn;
602 byteswap_uint64_array(lr, sizeof (*lr));
604 if ((error = zfs_zget(zfsvfs, lr->lr_doid, &dzp)) != 0)
607 if ((error = zfs_zget(zfsvfs, lr->lr_link_obj, &zp)) != 0) {
612 if (lr->lr_common.lrc_txtype & TX_CI)
614 cn.cn_nameptr = name;
616 cn.cn_thread = curthread;
617 cn.cn_flags = SAVENAME;
619 vn_lock(ZTOV(dzp), LK_EXCLUSIVE | LK_RETRY);
620 vn_lock(ZTOV(zp), LK_EXCLUSIVE | LK_RETRY);
621 error = VOP_LINK(ZTOV(dzp), ZTOV(zp), &cn /*,vflg*/);
622 VOP_UNLOCK(ZTOV(zp), 0);
623 VOP_UNLOCK(ZTOV(dzp), 0);
632 zfs_replay_rename(zfsvfs_t *zfsvfs, lr_rename_t *lr, boolean_t byteswap)
634 char *sname = (char *)(lr + 1); /* sname and tname follow lr_rename_t */
635 char *tname = sname + strlen(sname) + 1;
636 znode_t *sdzp, *tdzp;
637 struct componentname scn, tcn;
639 kthread_t *td = curthread;
644 byteswap_uint64_array(lr, sizeof (*lr));
646 if ((error = zfs_zget(zfsvfs, lr->lr_sdoid, &sdzp)) != 0)
649 if ((error = zfs_zget(zfsvfs, lr->lr_tdoid, &tdzp)) != 0) {
654 if (lr->lr_common.lrc_txtype & TX_CI)
658 scn.cn_nameptr = sname;
659 scn.cn_namelen = strlen(sname);
660 scn.cn_nameiop = DELETE;
661 scn.cn_flags = ISLASTCN | SAVENAME;
662 scn.cn_lkflags = LK_EXCLUSIVE | LK_RETRY;
665 vn_lock(ZTOV(sdzp), LK_EXCLUSIVE | LK_RETRY);
666 error = VOP_LOOKUP(ZTOV(sdzp), &svp, &scn);
667 VOP_UNLOCK(ZTOV(sdzp), 0);
672 tcn.cn_nameptr = tname;
673 tcn.cn_namelen = strlen(tname);
674 tcn.cn_nameiop = RENAME;
675 tcn.cn_flags = ISLASTCN | SAVENAME;
676 tcn.cn_lkflags = LK_EXCLUSIVE | LK_RETRY;
679 vn_lock(ZTOV(tdzp), LK_EXCLUSIVE | LK_RETRY);
680 error = VOP_LOOKUP(ZTOV(tdzp), &tvp, &tcn);
681 if (error == EJUSTRETURN)
683 else if (error != 0) {
684 VOP_UNLOCK(ZTOV(tdzp), 0);
688 error = VOP_RENAME(ZTOV(sdzp), svp, &scn, ZTOV(tdzp), tvp, &tcn /*,vflg*/);
702 zfs_replay_write(zfsvfs_t *zfsvfs, lr_write_t *lr, boolean_t byteswap)
704 char *data = (char *)(lr + 1); /* data follows lr_write_t */
710 byteswap_uint64_array(lr, sizeof (*lr));
712 if ((error = zfs_zget(zfsvfs, lr->lr_foid, &zp)) != 0) {
714 * As we can log writes out of order, it's possible the
715 * file has been removed. In this case just drop the write
716 * and return success.
723 error = vn_rdwr(UIO_WRITE, ZTOV(zp), data, lr->lr_length,
724 lr->lr_offset, UIO_SYSSPACE, 0, RLIM64_INFINITY, kcred, &resid);
732 zfs_replay_truncate(zfsvfs_t *zfsvfs, lr_truncate_t *lr, boolean_t byteswap)
735 ZFS_LOG(0, "Unexpected code path, report to pjd@FreeBSD.org");
740 zfs_replay_setattr(zfsvfs_t *zfsvfs, lr_setattr_t *lr, boolean_t byteswap)
744 vattr_t *vap = &xva.xva_vattr;
751 byteswap_uint64_array(lr, sizeof (*lr));
753 if ((lr->lr_mask & AT_XVATTR) &&
754 zfsvfs->z_version >= ZPL_VERSION_INITIAL)
755 zfs_replay_swap_attrs((lr_attr_t *)(lr + 1));
758 if ((error = zfs_zget(zfsvfs, lr->lr_foid, &zp)) != 0) {
760 * As we can log setattrs out of order, it's possible the
761 * file has been removed. In this case just drop the setattr
762 * and return success.
769 zfs_init_vattr(vap, lr->lr_mask, lr->lr_mode,
770 lr->lr_uid, lr->lr_gid, 0, lr->lr_foid);
772 vap->va_size = lr->lr_size;
773 ZFS_TIME_DECODE(&vap->va_atime, lr->lr_atime);
774 ZFS_TIME_DECODE(&vap->va_mtime, lr->lr_mtime);
777 * Fill in xvattr_t portions if necessary.
780 start = (lr_setattr_t *)(lr + 1);
781 if (vap->va_mask & AT_XVATTR) {
782 zfs_replay_xvattr((lr_attr_t *)start, &xva);
783 start = (caddr_t)start +
784 ZIL_XVAT_SIZE(((lr_attr_t *)start)->lr_attr_masksize);
786 xva.xva_vattr.va_mask &= ~AT_XVATTR;
788 zfsvfs->z_fuid_replay = zfs_replay_fuid_domain(start, &start,
789 lr->lr_uid, lr->lr_gid);
792 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
793 error = VOP_SETATTR(vp, vap, kcred);
796 zfs_fuid_info_free(zfsvfs->z_fuid_replay);
797 zfsvfs->z_fuid_replay = NULL;
804 zfs_replay_acl_v0(zfsvfs_t *zfsvfs, lr_acl_v0_t *lr, boolean_t byteswap)
806 ace_t *ace = (ace_t *)(lr + 1); /* ace array follows lr_acl_t */
812 byteswap_uint64_array(lr, sizeof (*lr));
813 zfs_oldace_byteswap(ace, lr->lr_aclcnt);
816 if ((error = zfs_zget(zfsvfs, lr->lr_foid, &zp)) != 0) {
818 * As we can log acls out of order, it's possible the
819 * file has been removed. In this case just drop the acl
820 * and return success.
827 bzero(&vsa, sizeof (vsa));
828 vsa.vsa_mask = VSA_ACE | VSA_ACECNT;
829 vsa.vsa_aclcnt = lr->lr_aclcnt;
830 vsa.vsa_aclentsz = sizeof (ace_t) * vsa.vsa_aclcnt;
831 vsa.vsa_aclflags = 0;
832 vsa.vsa_aclentp = ace;
835 error = VOP_SETSECATTR(ZTOV(zp), &vsa, 0, kcred, NULL);
837 panic("%s:%u: unsupported condition", __func__, __LINE__);
846 * Replaying ACLs is complicated by FUID support.
847 * The log record may contain some optional data
848 * to be used for replaying FUID's. These pieces
849 * are the actual FUIDs that were created initially.
850 * The FUID table index may no longer be valid and
851 * during zfs_create() a new index may be assigned.
852 * Because of this the log will contain the original
853 * doman+rid in order to create a new FUID.
855 * The individual ACEs may contain an ephemeral uid/gid which is no
856 * longer valid and will need to be replaced with an actual FUID.
860 zfs_replay_acl(zfsvfs_t *zfsvfs, lr_acl_t *lr, boolean_t byteswap)
862 ace_t *ace = (ace_t *)(lr + 1);
868 byteswap_uint64_array(lr, sizeof (*lr));
869 zfs_ace_byteswap(ace, lr->lr_acl_bytes, B_FALSE);
870 if (lr->lr_fuidcnt) {
871 byteswap_uint64_array((caddr_t)ace +
872 ZIL_ACE_LENGTH(lr->lr_acl_bytes),
873 lr->lr_fuidcnt * sizeof (uint64_t));
877 if ((error = zfs_zget(zfsvfs, lr->lr_foid, &zp)) != 0) {
879 * As we can log acls out of order, it's possible the
880 * file has been removed. In this case just drop the acl
881 * and return success.
889 bzero(&vsa, sizeof (vsa));
890 vsa.vsa_mask = VSA_ACE | VSA_ACECNT | VSA_ACE_ACLFLAGS;
891 vsa.vsa_aclcnt = lr->lr_aclcnt;
892 vsa.vsa_aclentp = ace;
893 vsa.vsa_aclentsz = lr->lr_acl_bytes;
894 vsa.vsa_aclflags = lr->lr_acl_flags;
896 if (lr->lr_fuidcnt) {
897 void *fuidstart = (caddr_t)ace +
898 ZIL_ACE_LENGTH(lr->lr_acl_bytes);
900 zfsvfs->z_fuid_replay =
901 zfs_replay_fuids(fuidstart, &fuidstart,
902 lr->lr_fuidcnt, lr->lr_domcnt, 0, 0);
905 error = VOP_SETSECATTR(ZTOV(zp), &vsa, 0, kcred, NULL);
907 if (zfsvfs->z_fuid_replay)
908 zfs_fuid_info_free(zfsvfs->z_fuid_replay);
913 zfsvfs->z_fuid_replay = NULL;
920 * Callback vectors for replaying records
922 zil_replay_func_t *zfs_replay_vector[TX_MAX_TYPE] = {
923 zfs_replay_error, /* 0 no such transaction type */
924 zfs_replay_create, /* TX_CREATE */
925 zfs_replay_create, /* TX_MKDIR */
926 zfs_replay_create, /* TX_MKXATTR */
927 zfs_replay_create, /* TX_SYMLINK */
928 zfs_replay_remove, /* TX_REMOVE */
929 zfs_replay_remove, /* TX_RMDIR */
930 zfs_replay_link, /* TX_LINK */
931 zfs_replay_rename, /* TX_RENAME */
932 zfs_replay_write, /* TX_WRITE */
933 zfs_replay_truncate, /* TX_TRUNCATE */
934 zfs_replay_setattr, /* TX_SETATTR */
935 zfs_replay_acl_v0, /* TX_ACL_V0 */
936 zfs_replay_acl, /* TX_ACL */
937 zfs_replay_create_acl, /* TX_CREATE_ACL */
938 zfs_replay_create, /* TX_CREATE_ATTR */
939 zfs_replay_create_acl, /* TX_CREATE_ACL_ATTR */
940 zfs_replay_create_acl, /* TX_MKDIR_ACL */
941 zfs_replay_create, /* TX_MKDIR_ATTR */
942 zfs_replay_create_acl, /* TX_MKDIR_ACL_ATTR */