4 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
6 * This program is free software; you can redistribute it and/or modify
7 * it under the terms of the GNU General Public License version 2 only,
8 * as published by the Free Software Foundation.
10 * This program is distributed in the hope that it will be useful, but
11 * WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13 * General Public License version 2 for more details (a copy is included
14 * in the LICENSE file that accompanied this code).
16 * You should have received a copy of the GNU General Public License
17 * version 2 along with this program; If not, see
18 * http://www.gnu.org/licenses/gpl-2.0.html
23 * Copyright (c) 2007, 2010, Oracle and/or its affiliates. All rights reserved.
24 * Use is subject to license terms.
26 * Copyright (c) 2011, 2017, Intel Corporation.
29 * This file is part of Lustre, http://www.lustre.org/
30 * Lustre is a trademark of Sun Microsystems, Inc.
32 * lustre/mdd/mdd_dir.c
34 * Lustre Metadata Server (mdd) routines
36 * Author: Wang Di <wangdi@intel.com>
39 #define DEBUG_SUBSYSTEM S_MDS
41 #include <linux/kthread.h>
43 #include <obd_class.h>
44 #include <obd_support.h>
45 #include <lustre_mds.h>
46 #include <lustre_fid.h>
48 #include "mdd_internal.h"
50 static const char dot[] = ".";
51 static const char dotdot[] = "..";
53 static struct lu_name lname_dotdot = {
54 .ln_name = (char *) dotdot,
55 .ln_namelen = sizeof(dotdot) - 1,
59 mdd_name_check(struct mdd_device *m, const struct lu_name *ln)
61 if (!lu_name_is_valid(ln))
63 else if (ln->ln_namelen > m->mdd_dt_conf.ddp_max_name_len)
69 /* Get FID from name and parent */
71 __mdd_lookup(const struct lu_env *env, struct md_object *pobj,
72 const struct lu_attr *pattr, const struct lu_name *lname,
73 struct lu_fid* fid, int mask)
75 const char *name = lname->ln_name;
76 const struct dt_key *key = (const struct dt_key *)name;
77 struct mdd_object *mdd_obj = md2mdd_obj(pobj);
78 struct mdd_device *m = mdo2mdd(pobj);
79 struct dt_object *dir = mdd_object_child(mdd_obj);
83 if (unlikely(mdd_is_dead_obj(mdd_obj)))
86 if (!mdd_object_exists(mdd_obj))
89 if (mdd_object_remote(mdd_obj)) {
90 CDEBUG(D_INFO, "%s: Object "DFID" locates on remote server\n",
91 mdd2obd_dev(m)->obd_name, PFID(mdo2fid(mdd_obj)));
94 rc = mdd_permission_internal_locked(env, mdd_obj, pattr, mask,
99 if (likely(S_ISDIR(mdd_object_type(mdd_obj)) &&
100 dt_try_as_dir(env, dir)))
101 rc = dt_lookup(env, dir, (struct dt_rec *)fid, key);
108 int mdd_lookup(const struct lu_env *env,
109 struct md_object *pobj, const struct lu_name *lname,
110 struct lu_fid *fid, struct md_op_spec *spec)
112 struct lu_attr *pattr = MDD_ENV_VAR(env, pattr);
116 rc = mdd_la_get(env, md2mdd_obj(pobj), pattr);
120 rc = __mdd_lookup(env, pobj, pattr, lname, fid,
121 (spec != NULL && spec->sp_permitted) ? 0 : MAY_EXEC);
125 /** Read the link EA into a temp buffer.
126 * Uses the mdd_thread_info::mti_big_buf since it is generally large.
127 * A pointer to the buffer is stored in \a ldata::ld_buf.
131 static int __mdd_links_read(const struct lu_env *env,
132 struct mdd_object *mdd_obj,
133 struct linkea_data *ldata)
137 if (!mdd_object_exists(mdd_obj))
140 /* First try a small buf */
141 LASSERT(env != NULL);
142 ldata->ld_buf = lu_buf_check_and_alloc(&mdd_env_info(env)->mti_link_buf,
144 if (ldata->ld_buf->lb_buf == NULL)
147 rc = mdo_xattr_get(env, mdd_obj, ldata->ld_buf, XATTR_NAME_LINK);
149 /* Buf was too small, figure out what we need. */
150 lu_buf_free(ldata->ld_buf);
151 rc = mdo_xattr_get(env, mdd_obj, ldata->ld_buf,
155 ldata->ld_buf = lu_buf_check_and_alloc(ldata->ld_buf, rc);
156 if (ldata->ld_buf->lb_buf == NULL)
158 rc = mdo_xattr_get(env, mdd_obj, ldata->ld_buf,
162 lu_buf_free(ldata->ld_buf);
163 ldata->ld_buf = NULL;
167 return linkea_init(ldata);
170 static int mdd_links_read(const struct lu_env *env,
171 struct mdd_object *mdd_obj,
172 struct linkea_data *ldata)
176 rc = __mdd_links_read(env, mdd_obj, ldata);
178 rc = linkea_init(ldata);
183 static int mdd_links_read_with_rec(const struct lu_env *env,
184 struct mdd_object *mdd_obj,
185 struct linkea_data *ldata)
189 rc = __mdd_links_read(env, mdd_obj, ldata);
191 rc = linkea_init_with_rec(ldata);
197 * Get parent FID of the directory
199 * Read parent FID from linkEA, if that fails, then do lookup
200 * dotdot to get the parent FID.
202 * \param[in] env execution environment
203 * \param[in] obj object from which to find the parent FID
204 * \param[in] attr attribute of the object
205 * \param[out] fid fid to get the parent FID
207 * \retval 0 if getting the parent FID succeeds.
208 * \retval negative errno if getting the parent FID fails.
210 static inline int mdd_parent_fid(const struct lu_env *env,
211 struct mdd_object *obj,
212 const struct lu_attr *attr,
215 struct mdd_thread_info *info = mdd_env_info(env);
216 struct linkea_data ldata = { NULL };
217 struct lu_buf *buf = &info->mti_link_buf;
218 struct lu_name lname;
223 LASSERT(S_ISDIR(mdd_object_type(obj)));
225 buf = lu_buf_check_and_alloc(buf, PATH_MAX);
226 if (buf->lb_buf == NULL)
227 GOTO(lookup, rc = 0);
230 rc = mdd_links_read_with_rec(env, obj, &ldata);
234 LASSERT(ldata.ld_leh != NULL);
235 /* Directory should only have 1 parent */
236 if (ldata.ld_leh->leh_reccount > 1)
239 ldata.ld_lee = (struct link_ea_entry *)(ldata.ld_leh + 1);
241 linkea_entry_unpack(ldata.ld_lee, &ldata.ld_reclen, &lname, fid);
242 if (likely(fid_is_sane(fid)))
245 rc = __mdd_lookup(env, &obj->mod_obj, attr, &lname_dotdot, fid, 0);
250 * For root fid use special function, which does not compare version component
251 * of fid. Version component is different for root fids on all MDTs.
253 int mdd_is_root(struct mdd_device *mdd, const struct lu_fid *fid)
255 return fid_seq(&mdd->mdd_root_fid) == fid_seq(fid) &&
256 fid_oid(&mdd->mdd_root_fid) == fid_oid(fid);
260 * return 1: if lf is the fid of the ancestor of p1;
263 * return -EREMOTE: if remote object is found, in this
264 * case fid of remote object is saved to @pf;
266 * otherwise: values < 0, errors.
268 static int mdd_is_parent(const struct lu_env *env,
269 struct mdd_device *mdd,
270 struct mdd_object *p1,
271 const struct lu_attr *attr,
272 const struct lu_fid *lf,
275 struct mdd_object *parent = NULL;
280 LASSERT(!lu_fid_eq(mdo2fid(p1), lf));
281 pfid = &mdd_env_info(env)->mti_fid;
283 /* Check for root first. */
284 if (mdd_is_root(mdd, mdo2fid(p1)))
288 /* this is done recursively */
289 rc = mdd_parent_fid(env, p1, attr, pfid);
292 if (mdd_is_root(mdd, pfid))
294 if (lu_fid_eq(pfid, &mdd->mdd_local_root_fid))
296 if (lu_fid_eq(pfid, lf))
299 mdd_object_put(env, parent);
301 parent = mdd_object_find(env, mdd, pfid);
303 GOTO(out, rc = PTR_ERR(parent));
305 if (!mdd_object_exists(parent))
306 GOTO(out, rc = -EINVAL);
312 if (parent && !IS_ERR(parent))
313 mdd_object_put(env, parent);
318 * No permission check is needed.
320 * returns 1: if fid is ancestor of @mo;
321 * returns 0: if fid is not an ancestor of @mo;
323 * returns EREMOTE if remote object is found, fid of remote object is saved to
326 * returns < 0: if error
328 int mdd_is_subdir(const struct lu_env *env, struct md_object *mo,
329 const struct lu_fid *fid, struct lu_fid *sfid)
331 struct mdd_device *mdd = mdo2mdd(mo);
332 struct lu_attr *attr = MDD_ENV_VAR(env, cattr);
336 if (!S_ISDIR(mdd_object_type(md2mdd_obj(mo))))
339 rc = mdd_la_get(env, md2mdd_obj(mo), attr);
343 rc = mdd_is_parent(env, mdd, md2mdd_obj(mo), attr, fid, sfid);
347 } else if (rc == 1) {
348 /* found @fid is parent */
356 * Check that @dir contains no entries except (possibly) dot and dotdot.
361 * -ENOTDIR not a directory object
362 * -ENOTEMPTY not empty
366 static int mdd_dir_is_empty(const struct lu_env *env,
367 struct mdd_object *dir)
370 struct dt_object *obj;
371 const struct dt_it_ops *iops;
375 obj = mdd_object_child(dir);
376 if (!dt_try_as_dir(env, obj))
379 iops = &obj->do_index_ops->dio_it;
380 it = iops->init(env, obj, LUDA_64BITHASH);
382 result = iops->get(env, it, (const struct dt_key *)"");
385 for (result = 0, i = 0; result == 0 && i < 3; ++i)
386 result = iops->next(env, it);
389 else if (result == 1)
391 } else if (result == 0)
393 * Huh? Index contains no zero key?
400 result = PTR_ERR(it);
405 * Determine if the target object can be hard linked, and right now it only
406 * checks if the link count reach the maximum limit. Note: for ldiskfs, the
407 * directory nlink count might exceed the maximum link count(see
408 * osd_object_ref_add), so it only check nlink for non-directories.
410 * \param[in] env thread environment
411 * \param[in] obj object being linked to
412 * \param[in] la attributes of \a obj
414 * \retval 0 if \a obj can be hard linked
415 * \retval negative error if \a obj is a directory or has too
418 static int __mdd_may_link(const struct lu_env *env, struct mdd_object *obj,
419 const struct lu_attr *la)
421 struct mdd_device *m = mdd_obj2mdd_dev(obj);
426 /* Subdir count limitation can be broken through
427 * (see osd_object_ref_add), so only check non-directory here. */
428 if (!S_ISDIR(la->la_mode) &&
429 la->la_nlink >= m->mdd_dt_conf.ddp_max_nlink)
436 * Check whether it may create the cobj under the pobj.
438 * \param[in] env execution environment
439 * \param[in] pobj the parent directory
440 * \param[in] pattr the attribute of the parent directory
441 * \param[in] cobj the child to be created
442 * \param[in] check_perm if check WRITE|EXEC permission for parent
444 * \retval = 0 create the child under this dir is allowed
445 * \retval negative errno create the child under this dir is
448 int mdd_may_create(const struct lu_env *env, struct mdd_object *pobj,
449 const struct lu_attr *pattr, struct mdd_object *cobj,
455 if (cobj && mdd_object_exists(cobj))
458 if (mdd_is_dead_obj(pobj))
462 rc = mdd_permission_internal_locked(env, pobj, pattr,
463 MAY_WRITE | MAY_EXEC,
469 * Check whether can unlink from the pobj in the case of "cobj == NULL".
471 int mdd_may_unlink(const struct lu_env *env, struct mdd_object *pobj,
472 const struct lu_attr *pattr, const struct lu_attr *attr)
477 if (mdd_is_dead_obj(pobj))
480 if (attr->la_flags & (LUSTRE_APPEND_FL | LUSTRE_IMMUTABLE_FL))
483 rc = mdd_permission_internal_locked(env, pobj, pattr,
484 MAY_WRITE | MAY_EXEC,
489 if (pattr->la_flags & LUSTRE_APPEND_FL)
496 * pobj == NULL is remote ops case, under such case, pobj's
497 * VTX feature has been checked already, no need check again.
499 static inline int mdd_is_sticky(const struct lu_env *env,
500 struct mdd_object *pobj,
501 const struct lu_attr *pattr,
502 struct mdd_object *cobj,
503 const struct lu_attr *cattr)
505 struct lu_ucred *uc = lu_ucred_assert(env);
508 LASSERT(pattr != NULL);
509 if (!(pattr->la_mode & S_ISVTX) ||
510 (pattr->la_uid == uc->uc_fsuid))
514 LASSERT(cattr != NULL);
515 if (cattr->la_uid == uc->uc_fsuid)
518 return !md_capable(uc, CFS_CAP_FOWNER);
521 static int mdd_may_delete_entry(const struct lu_env *env,
522 struct mdd_object *pobj,
523 const struct lu_attr *pattr,
528 LASSERT(pobj != NULL);
529 if (!mdd_object_exists(pobj))
532 if (mdd_is_dead_obj(pobj))
537 rc = mdd_permission_internal_locked(env, pobj, pattr,
538 MAY_WRITE | MAY_EXEC,
544 if (pattr->la_flags & LUSTRE_APPEND_FL)
551 * Check whether it may delete the cobj from the pobj.
554 int mdd_may_delete(const struct lu_env *env, struct mdd_object *tpobj,
555 const struct lu_attr *tpattr, struct mdd_object *tobj,
556 const struct lu_attr *tattr, const struct lu_attr *cattr,
557 int check_perm, int check_empty)
563 LASSERT(tpattr != NULL);
564 rc = mdd_may_delete_entry(env, tpobj, tpattr, check_perm);
572 if (!mdd_object_exists(tobj))
575 if (mdd_is_dead_obj(tobj))
578 if (mdd_is_sticky(env, tpobj, tpattr, tobj, tattr))
581 if (tattr->la_flags & (LUSTRE_APPEND_FL | LUSTRE_IMMUTABLE_FL))
584 /* additional check the rename case */
586 if (S_ISDIR(cattr->la_mode)) {
587 struct mdd_device *mdd = mdo2mdd(&tobj->mod_obj);
589 if (!S_ISDIR(tattr->la_mode))
592 if (lu_fid_eq(mdo2fid(tobj), &mdd->mdd_root_fid))
594 } else if (S_ISDIR(tattr->la_mode))
598 if (S_ISDIR(tattr->la_mode) && check_empty)
599 rc = mdd_dir_is_empty(env, tobj);
605 * Check whether it can create the link file(linked to @src_obj) under
606 * the target directory(@tgt_obj), and src_obj has been locked by
609 * \param[in] env execution environment
610 * \param[in] tgt_obj the target directory
611 * \param[in] tattr attributes of target directory
612 * \param[in] lname the link name
613 * \param[in] src_obj source object for link
614 * \param[in] cattr attributes for source object
616 * \retval = 0 it is allowed to create the link file under tgt_obj
617 * \retval negative error not allowed to create the link file
619 static int mdd_link_sanity_check(const struct lu_env *env,
620 struct mdd_object *tgt_obj,
621 const struct lu_attr *tattr,
622 const struct lu_name *lname,
623 struct mdd_object *src_obj,
624 const struct lu_attr *cattr)
626 struct mdd_device *m = mdd_obj2mdd_dev(src_obj);
630 if (!mdd_object_exists(src_obj))
633 if (mdd_is_dead_obj(src_obj))
636 /* Local ops, no lookup before link, check filename length here. */
637 rc = mdd_name_check(m, lname);
641 if (cattr->la_flags & (LUSTRE_IMMUTABLE_FL | LUSTRE_APPEND_FL))
644 if (S_ISDIR(mdd_object_type(src_obj)))
647 LASSERT(src_obj != tgt_obj);
648 rc = mdd_may_create(env, tgt_obj, tattr, NULL, true);
652 rc = __mdd_may_link(env, src_obj, cattr);
657 static int __mdd_index_delete_only(const struct lu_env *env, struct mdd_object *pobj,
658 const char *name, struct thandle *handle)
660 struct dt_object *next = mdd_object_child(pobj);
664 if (dt_try_as_dir(env, next))
665 rc = dt_delete(env, next, (struct dt_key *)name, handle);
672 static int __mdd_index_insert_only(const struct lu_env *env,
673 struct mdd_object *pobj,
674 const struct lu_fid *lf, __u32 type,
676 struct thandle *handle)
678 struct dt_object *next = mdd_object_child(pobj);
682 if (dt_try_as_dir(env, next)) {
683 struct dt_insert_rec *rec = &mdd_env_info(env)->mti_dt_rec;
684 struct lu_ucred *uc = lu_ucred_check(env);
688 rec->rec_type = type;
689 ignore_quota = uc ? uc->uc_cap & CFS_CAP_SYS_RESOURCE_MASK : 1;
690 rc = dt_insert(env, next, (const struct dt_rec *)rec,
691 (const struct dt_key *)name, handle,
699 /* insert named index, add reference if isdir */
700 static int __mdd_index_insert(const struct lu_env *env, struct mdd_object *pobj,
701 const struct lu_fid *lf, __u32 type,
702 const char *name, struct thandle *handle)
707 rc = __mdd_index_insert_only(env, pobj, lf, type, name, handle);
708 if (rc == 0 && S_ISDIR(type)) {
709 mdd_write_lock(env, pobj, MOR_TGT_PARENT);
710 mdo_ref_add(env, pobj, handle);
711 mdd_write_unlock(env, pobj);
717 /* delete named index, drop reference if isdir */
718 static int __mdd_index_delete(const struct lu_env *env, struct mdd_object *pobj,
719 const char *name, int is_dir,
720 struct thandle *handle)
725 rc = __mdd_index_delete_only(env, pobj, name, handle);
726 if (rc == 0 && is_dir) {
727 mdd_write_lock(env, pobj, MOR_TGT_PARENT);
728 mdo_ref_del(env, pobj, handle);
729 mdd_write_unlock(env, pobj);
735 static int mdd_llog_record_calc_size(const struct lu_env *env,
736 const struct lu_name *tname,
737 const struct lu_name *sname)
739 const struct lu_ucred *uc = lu_ucred(env);
740 enum changelog_rec_flags crf = CLF_EXTRA_FLAGS;
741 enum changelog_rec_extra_flags crfe = CLFE_UIDGID;
746 if (uc != NULL && uc->uc_jobid[0] != '\0')
749 return llog_data_len(LLOG_CHANGELOG_HDR_SZ +
750 changelog_rec_offset(crf, crfe) +
751 (tname != NULL ? tname->ln_namelen : 0) +
752 (sname != NULL ? 1 + sname->ln_namelen : 0));
755 int mdd_declare_changelog_store(const struct lu_env *env,
756 struct mdd_device *mdd,
757 const struct lu_name *tname,
758 const struct lu_name *sname,
759 struct thandle *handle)
761 struct obd_device *obd = mdd2obd_dev(mdd);
762 struct llog_ctxt *ctxt;
763 struct llog_changelog_rec *rec;
765 struct thandle *llog_th;
770 if (!(mdd->mdd_cl.mc_flags & CLM_ON))
773 reclen = mdd_llog_record_calc_size(env, tname, sname);
774 buf = lu_buf_check_and_alloc(&mdd_env_info(env)->mti_big_buf, reclen);
775 if (buf->lb_buf == NULL)
779 rec->cr_hdr.lrh_len = reclen;
780 rec->cr_hdr.lrh_type = CHANGELOG_REC;
782 ctxt = llog_get_context(obd, LLOG_CHANGELOG_ORIG_CTXT);
786 llog_th = thandle_get_sub(env, handle, ctxt->loc_handle->lgh_obj);
788 GOTO(out_put, rc = PTR_ERR(llog_th));
790 rc = llog_declare_add(env, ctxt->loc_handle, &rec->cr_hdr, llog_th);
798 struct mdd_changelog_gc {
799 struct mdd_device *mcgc_mdd;
802 __u64 mcgc_maxindexes;
806 /* return first registered ChangeLog user idle since too long
807 * use ChangeLog's user plain LLOG mtime for this */
808 static int mdd_changelog_gc_cb(const struct lu_env *env,
809 struct llog_handle *llh,
810 struct llog_rec_hdr *hdr, void *data)
812 struct llog_changelog_user_rec *rec;
813 struct mdd_changelog_gc *mcgc = (struct mdd_changelog_gc *)data;
814 struct mdd_device *mdd = mcgc->mcgc_mdd;
817 if ((llh->lgh_hdr->llh_flags & LLOG_F_IS_PLAIN) == 0)
820 rec = container_of(hdr, struct llog_changelog_user_rec,
823 /* find oldest idle user, based on last record update/cancel time (new
824 * behavior), or for old user records, last record index vs current
825 * ChangeLog index. Late users with old record format will be treated
826 * first as we assume they could be idle since longer
828 if (rec->cur_time != 0) {
829 __u32 time_now = (__u32)get_seconds();
830 __u32 time_out = rec->cur_time +
831 mdd->mdd_changelog_max_idle_time;
832 __u32 idle_time = time_now - rec->cur_time;
834 /* treat oldest idle user first, and if no old format user
835 * has been already selected
837 if (time_after32(time_now, time_out) &&
838 idle_time > mcgc->mcgc_maxtime &&
839 mcgc->mcgc_maxindexes == 0) {
840 mcgc->mcgc_maxtime = idle_time;
841 mcgc->mcgc_id = rec->cur_id;
842 mcgc->mcgc_found = true;
845 /* old user record with no idle time stamp, so use empirical
846 * method based on its current index/position
850 idle_indexes = mdd->mdd_cl.mc_index - rec->cur_endrec;
852 /* treat user with the oldest/smallest current index first */
853 if (idle_indexes >= mdd->mdd_changelog_max_idle_indexes &&
854 idle_indexes > mcgc->mcgc_maxindexes) {
855 mcgc->mcgc_maxindexes = idle_indexes;
856 mcgc->mcgc_id = rec->cur_id;
857 mcgc->mcgc_found = true;
864 /* recover space from long-term inactive ChangeLog users */
865 static int mdd_chlg_garbage_collect(void *data)
867 struct mdd_device *mdd = (struct mdd_device *)data;
868 struct lu_env *env = NULL;
870 struct llog_ctxt *ctxt;
871 struct mdd_changelog_gc mcgc = {
875 .mcgc_maxindexes = 0,
879 CDEBUG(D_HA, "%s: ChangeLog garbage collect thread start\n",
880 mdd2obd_dev(mdd)->obd_name);
884 GOTO(out, rc = -ENOMEM);
886 rc = lu_env_init(env, LCT_MD_THREAD);
891 ctxt = llog_get_context(mdd2obd_dev(mdd),
892 LLOG_CHANGELOG_USER_ORIG_CTXT);
894 (ctxt->loc_handle->lgh_hdr->llh_flags & LLOG_F_IS_CAT) == 0)
895 GOTO(out_env, rc = -ENXIO);
897 rc = llog_cat_process(env, ctxt->loc_handle,
898 mdd_changelog_gc_cb, &mcgc, 0, 0);
899 if (rc != 0 || mcgc.mcgc_found == false)
903 CWARN("%s: Force deregister of ChangeLog user cl%d idle more "
904 "than %us\n", mdd2obd_dev(mdd)->obd_name, mcgc.mcgc_id,
907 mdd_changelog_user_purge(env, mdd, mcgc.mcgc_id);
909 /* try again to search for another candidate */
910 mcgc.mcgc_found = false;
911 mcgc.mcgc_maxtime = 0;
912 mcgc.mcgc_maxindexes = 0;
924 mdd->mdd_cl.mc_gc_task = NULL;
928 /** Add a changelog entry \a rec to the changelog llog
931 * \param handle - currently ignored since llogs start their own transaction;
932 * this will hopefully be fixed in llog rewrite
935 int mdd_changelog_store(const struct lu_env *env, struct mdd_device *mdd,
936 struct llog_changelog_rec *rec, struct thandle *th)
938 struct obd_device *obd = mdd2obd_dev(mdd);
939 struct llog_ctxt *ctxt;
940 struct thandle *llog_th;
944 rec->cr_hdr.lrh_len = llog_data_len(sizeof(*rec) +
945 changelog_rec_varsize(&rec->cr));
947 /* llog_lvfs_write_rec sets the llog tail len */
948 rec->cr_hdr.lrh_type = CHANGELOG_REC;
949 rec->cr.cr_time = cl_time();
951 spin_lock(&mdd->mdd_cl.mc_lock);
952 /* NB: I suppose it's possible llog_add adds out of order wrt cr_index,
953 * but as long as the MDD transactions are ordered correctly for e.g.
954 * rename conflicts, I don't think this should matter. */
955 rec->cr.cr_index = ++mdd->mdd_cl.mc_index;
956 spin_unlock(&mdd->mdd_cl.mc_lock);
958 ctxt = llog_get_context(obd, LLOG_CHANGELOG_ORIG_CTXT);
962 llog_th = thandle_get_sub(env, th, ctxt->loc_handle->lgh_obj);
964 GOTO(out_put, rc = PTR_ERR(llog_th));
966 /* nested journal transaction */
967 rc = llog_add(env, ctxt->loc_handle, &rec->cr_hdr, NULL, llog_th);
969 /* time to recover some space ?? */
970 spin_lock(&mdd->mdd_cl.mc_lock);
971 if (unlikely(mdd->mdd_changelog_gc && (ktime_get_real_seconds() -
972 mdd->mdd_cl.mc_gc_time > mdd->mdd_changelog_min_gc_interval) &&
973 mdd->mdd_cl.mc_gc_task == NULL &&
974 llog_cat_free_space(ctxt->loc_handle) <=
975 mdd->mdd_changelog_min_free_cat_entries)) {
976 CWARN("%s: low on changelog_catalog free entries, starting "
977 "ChangeLog garbage collection thread\n", obd->obd_name);
979 /* indicate further kthread run will occur outside right after
982 mdd->mdd_cl.mc_gc_task = (struct task_struct *)(-1);
985 spin_unlock(&mdd->mdd_cl.mc_lock);
987 struct task_struct *gc_task;
989 gc_task = kthread_run(mdd_chlg_garbage_collect, mdd,
991 if (IS_ERR(gc_task)) {
992 CERROR("%s: cannot start ChangeLog garbage collection "
993 "thread: rc = %ld\n", obd->obd_name,
995 mdd->mdd_cl.mc_gc_task = NULL;
997 CDEBUG(D_HA, "%s: ChangeLog garbage collection thread "
998 "has started with Pid %d\n", obd->obd_name,
1000 mdd->mdd_cl.mc_gc_task = gc_task;
1001 mdd->mdd_cl.mc_gc_time = ktime_get_real_seconds();
1005 llog_ctxt_put(ctxt);
1011 static void mdd_changelog_rec_ext_rename(struct changelog_rec *rec,
1012 const struct lu_fid *sfid,
1013 const struct lu_fid *spfid,
1014 const struct lu_name *sname)
1016 struct changelog_ext_rename *rnm = changelog_rec_rename(rec);
1017 size_t extsize = sname->ln_namelen + 1;
1019 LASSERT(sfid != NULL);
1020 LASSERT(spfid != NULL);
1021 LASSERT(sname != NULL);
1023 rnm->cr_sfid = *sfid;
1024 rnm->cr_spfid = *spfid;
1026 changelog_rec_name(rec)[rec->cr_namelen] = '\0';
1027 strlcpy(changelog_rec_sname(rec), sname->ln_name, extsize);
1028 rec->cr_namelen += extsize;
1031 void mdd_changelog_rec_ext_jobid(struct changelog_rec *rec, const char *jobid)
1033 struct changelog_ext_jobid *jid = changelog_rec_jobid(rec);
1035 if (jobid == NULL || jobid[0] == '\0')
1038 strlcpy(jid->cr_jobid, jobid, sizeof(jid->cr_jobid));
1041 void mdd_changelog_rec_ext_extra_flags(struct changelog_rec *rec, __u64 eflags)
1043 struct changelog_ext_extra_flags *ef = changelog_rec_extra_flags(rec);
1045 ef->cr_extra_flags = eflags;
1048 void mdd_changelog_rec_extra_uidgid(struct changelog_rec *rec,
1049 __u64 uid, __u64 gid)
1051 struct changelog_ext_uidgid *uidgid = changelog_rec_uidgid(rec);
1053 uidgid->cr_uid = uid;
1054 uidgid->cr_gid = gid;
1057 /** Store a namespace change changelog record
1058 * If this fails, we must fail the whole transaction; we don't
1059 * want the change to commit without the log entry.
1060 * \param target - mdd_object of change
1061 * \param tpfid - target parent dir/object fid
1062 * \param sfid - source object fid
1063 * \param spfid - source parent fid
1064 * \param tname - target name string
1065 * \param sname - source name string
1066 * \param handle - transaction handle
1068 int mdd_changelog_ns_store(const struct lu_env *env,
1069 struct mdd_device *mdd,
1070 enum changelog_rec_type type,
1071 enum changelog_rec_flags crf,
1072 struct mdd_object *target,
1073 const struct lu_fid *tpfid,
1074 const struct lu_fid *sfid,
1075 const struct lu_fid *spfid,
1076 const struct lu_name *tname,
1077 const struct lu_name *sname,
1078 struct thandle *handle)
1080 const struct lu_ucred *uc = lu_ucred(env);
1081 struct llog_changelog_rec *rec;
1084 __u64 xflags = CLFE_INVALID;
1089 if (!(mdd->mdd_cl.mc_flags & CLM_ON))
1092 if ((mdd->mdd_cl.mc_mask & (1 << type)) == 0)
1095 LASSERT(tpfid != NULL);
1096 LASSERT(tname != NULL);
1097 LASSERT(handle != NULL);
1099 reclen = mdd_llog_record_calc_size(env, tname, sname);
1100 buf = lu_buf_check_and_alloc(&mdd_env_info(env)->mti_big_buf, reclen);
1101 if (buf->lb_buf == NULL)
1105 crf &= CLF_FLAGMASK;
1106 crf |= CLF_EXTRA_FLAGS;
1109 if (uc->uc_jobid[0] != '\0')
1111 xflags |= CLFE_UIDGID;
1119 rec->cr.cr_flags = crf;
1121 if (crf & CLF_EXTRA_FLAGS) {
1122 mdd_changelog_rec_ext_extra_flags(&rec->cr, xflags);
1123 if (xflags & CLFE_UIDGID)
1124 mdd_changelog_rec_extra_uidgid(&rec->cr,
1125 uc->uc_uid, uc->uc_gid);
1128 rec->cr.cr_type = (__u32)type;
1129 rec->cr.cr_pfid = *tpfid;
1130 rec->cr.cr_namelen = tname->ln_namelen;
1131 memcpy(changelog_rec_name(&rec->cr), tname->ln_name, tname->ln_namelen);
1133 if (crf & CLF_RENAME)
1134 mdd_changelog_rec_ext_rename(&rec->cr, sfid, spfid, sname);
1136 if (crf & CLF_JOBID)
1137 mdd_changelog_rec_ext_jobid(&rec->cr, uc->uc_jobid);
1139 if (likely(target != NULL)) {
1140 rec->cr.cr_tfid = *mdo2fid(target);
1141 target->mod_cltime = ktime_get();
1143 fid_zero(&rec->cr.cr_tfid);
1146 rc = mdd_changelog_store(env, mdd, rec, handle);
1148 CERROR("%s: cannot store changelog record: type = %d, "
1149 "name = '%s', t = "DFID", p = "DFID": rc = %d\n",
1150 mdd2obd_dev(mdd)->obd_name, type, tname->ln_name,
1151 PFID(&rec->cr.cr_tfid), PFID(&rec->cr.cr_pfid), rc);
1158 static int __mdd_links_add(const struct lu_env *env,
1159 struct mdd_object *mdd_obj,
1160 struct linkea_data *ldata,
1161 const struct lu_name *lname,
1162 const struct lu_fid *pfid,
1163 int first, int check)
1167 if (ldata->ld_leh == NULL) {
1168 rc = first ? -ENODATA : mdd_links_read(env, mdd_obj, ldata);
1172 rc = linkea_data_new(ldata,
1173 &mdd_env_info(env)->mti_link_buf);
1180 rc = linkea_links_find(ldata, lname, pfid);
1181 if (rc && rc != -ENOENT)
1187 if (OBD_FAIL_CHECK(OBD_FAIL_LFSCK_LINKEA_MORE)) {
1188 struct lu_fid *tfid = &mdd_env_info(env)->mti_fid2;
1192 linkea_add_buf(ldata, lname, tfid);
1195 if (OBD_FAIL_CHECK(OBD_FAIL_LFSCK_LINKEA_MORE2))
1196 linkea_add_buf(ldata, lname, pfid);
1198 return linkea_add_buf(ldata, lname, pfid);
1201 static int __mdd_links_del(const struct lu_env *env,
1202 struct mdd_object *mdd_obj,
1203 struct linkea_data *ldata,
1204 const struct lu_name *lname,
1205 const struct lu_fid *pfid)
1209 if (ldata->ld_leh == NULL) {
1210 rc = mdd_links_read(env, mdd_obj, ldata);
1215 rc = linkea_links_find(ldata, lname, pfid);
1219 linkea_del_buf(ldata, lname);
1223 static int mdd_linkea_prepare(const struct lu_env *env,
1224 struct mdd_object *mdd_obj,
1225 const struct lu_fid *oldpfid,
1226 const struct lu_name *oldlname,
1227 const struct lu_fid *newpfid,
1228 const struct lu_name *newlname,
1229 int first, int check,
1230 struct linkea_data *ldata)
1235 if (OBD_FAIL_CHECK(OBD_FAIL_FID_IGIF))
1238 LASSERT(oldpfid != NULL || newpfid != NULL);
1240 if (mdd_obj->mod_flags & DEAD_OBJ)
1241 /* Unnecessary to update linkEA for dead object. */
1244 if (oldpfid != NULL) {
1245 rc = __mdd_links_del(env, mdd_obj, ldata, oldlname, oldpfid);
1247 if ((check == 1) || (rc != -ENODATA && rc != -ENOENT))
1250 /* No changes done. */
1255 /* If renaming, add the new record */
1256 if (newpfid != NULL)
1257 rc = __mdd_links_add(env, mdd_obj, ldata, newlname, newpfid,
1263 int mdd_links_rename(const struct lu_env *env,
1264 struct mdd_object *mdd_obj,
1265 const struct lu_fid *oldpfid,
1266 const struct lu_name *oldlname,
1267 const struct lu_fid *newpfid,
1268 const struct lu_name *newlname,
1269 struct thandle *handle,
1270 struct linkea_data *ldata,
1271 int first, int check)
1276 if (ldata == NULL) {
1277 ldata = &mdd_env_info(env)->mti_link_data;
1278 memset(ldata, 0, sizeof(*ldata));
1279 rc = mdd_linkea_prepare(env, mdd_obj, oldpfid, oldlname,
1280 newpfid, newlname, first, check, ldata);
1285 if (!(mdd_obj->mod_flags & DEAD_OBJ))
1286 rc = mdd_links_write(env, mdd_obj, ldata, handle);
1292 if (newlname == NULL)
1293 CERROR("link_ea add failed %d "DFID"\n",
1294 rc, PFID(mdd_object_fid(mdd_obj)));
1295 else if (oldpfid == NULL)
1296 CERROR("link_ea add '%.*s' failed %d "DFID"\n",
1297 newlname->ln_namelen, newlname->ln_name, rc,
1298 PFID(mdd_object_fid(mdd_obj)));
1299 else if (newpfid == NULL)
1300 CERROR("link_ea del '%.*s' failed %d "DFID"\n",
1301 oldlname->ln_namelen, oldlname->ln_name, rc,
1302 PFID(mdd_object_fid(mdd_obj)));
1304 CERROR("link_ea rename '%.*s'->'%.*s' failed %d "DFID
1305 "\n", oldlname->ln_namelen, oldlname->ln_name,
1306 newlname->ln_namelen, newlname->ln_name, rc,
1307 PFID(mdd_object_fid(mdd_obj)));
1310 if (is_vmalloc_addr(ldata->ld_buf))
1311 /* if we vmalloced a large buffer drop it */
1312 lu_buf_free(ldata->ld_buf);
1317 static inline int mdd_links_add(const struct lu_env *env,
1318 struct mdd_object *mdd_obj,
1319 const struct lu_fid *pfid,
1320 const struct lu_name *lname,
1321 struct thandle *handle,
1322 struct linkea_data *ldata, int first)
1324 return mdd_links_rename(env, mdd_obj, NULL, NULL,
1325 pfid, lname, handle, ldata, first, 0);
1328 static inline int mdd_links_del(const struct lu_env *env,
1329 struct mdd_object *mdd_obj,
1330 const struct lu_fid *pfid,
1331 const struct lu_name *lname,
1332 struct thandle *handle)
1334 return mdd_links_rename(env, mdd_obj, pfid, lname,
1335 NULL, NULL, handle, NULL, 0, 0);
1338 /** Read the link EA into a temp buffer.
1339 * Uses the name_buf since it is generally large.
1340 * \retval IS_ERR err
1341 * \retval ptr to \a lu_buf (always \a mti_big_buf)
1343 struct lu_buf *mdd_links_get(const struct lu_env *env,
1344 struct mdd_object *mdd_obj)
1346 struct linkea_data ldata = { NULL };
1349 rc = mdd_links_read(env, mdd_obj, &ldata);
1350 return rc ? ERR_PTR(rc) : ldata.ld_buf;
1353 int mdd_links_write(const struct lu_env *env, struct mdd_object *mdd_obj,
1354 struct linkea_data *ldata, struct thandle *handle)
1356 const struct lu_buf *buf;
1359 if (ldata == NULL || ldata->ld_buf == NULL ||
1360 ldata->ld_leh == NULL)
1363 if (OBD_FAIL_CHECK(OBD_FAIL_LFSCK_NO_LINKEA))
1367 buf = mdd_buf_get_const(env, ldata->ld_buf->lb_buf,
1368 ldata->ld_leh->leh_len);
1369 rc = mdo_xattr_set(env, mdd_obj, buf, XATTR_NAME_LINK, 0, handle);
1370 if (unlikely(rc == -ENOSPC)) {
1371 rc = linkea_overflow_shrink(ldata);
1379 static int mdd_declare_links_add(const struct lu_env *env,
1380 struct mdd_object *mdd_obj,
1381 struct thandle *handle,
1382 struct linkea_data *ldata)
1388 if (ldata != NULL && ldata->ld_leh != NULL) {
1389 ea_len = ldata->ld_leh->leh_len;
1390 linkea = ldata->ld_buf->lb_buf;
1392 ea_len = MAX_LINKEA_SIZE;
1396 rc = mdo_declare_xattr_set(env, mdd_obj,
1397 mdd_buf_get_const(env, linkea, ea_len),
1398 XATTR_NAME_LINK, 0, handle);
1403 static inline int mdd_declare_links_del(const struct lu_env *env,
1404 struct mdd_object *c,
1405 struct thandle *handle)
1409 /* For directory, the linkEA will be removed together
1410 * with the object. */
1411 if (!S_ISDIR(mdd_object_type(c)))
1412 rc = mdd_declare_links_add(env, c, handle, NULL);
1417 static int mdd_declare_link(const struct lu_env *env,
1418 struct mdd_device *mdd,
1419 struct mdd_object *p,
1420 struct mdd_object *c,
1421 const struct lu_name *name,
1422 struct thandle *handle,
1424 struct linkea_data *data)
1426 struct lu_fid tfid = *mdo2fid(c);
1429 if (OBD_FAIL_CHECK(OBD_FAIL_LFSCK_DANGLING3))
1430 tfid.f_oid = cfs_fail_val;
1432 rc = mdo_declare_index_insert(env, p, &tfid, mdd_object_type(c),
1433 name->ln_name, handle);
1437 rc = mdo_declare_ref_add(env, c, handle);
1441 la->la_valid = LA_CTIME | LA_MTIME;
1442 rc = mdo_declare_attr_set(env, p, la, handle);
1446 la->la_valid = LA_CTIME;
1447 rc = mdo_declare_attr_set(env, c, la, handle);
1451 rc = mdd_declare_links_add(env, c, handle, data);
1455 rc = mdd_declare_changelog_store(env, mdd, name, NULL, handle);
1460 static int mdd_link(const struct lu_env *env, struct md_object *tgt_obj,
1461 struct md_object *src_obj, const struct lu_name *lname,
1464 const char *name = lname->ln_name;
1465 struct lu_attr *la = &mdd_env_info(env)->mti_la_for_fix;
1466 struct mdd_object *mdd_tobj = md2mdd_obj(tgt_obj);
1467 struct mdd_object *mdd_sobj = md2mdd_obj(src_obj);
1468 struct lu_attr *cattr = MDD_ENV_VAR(env, cattr);
1469 struct lu_attr *tattr = MDD_ENV_VAR(env, tattr);
1470 struct mdd_device *mdd = mdo2mdd(src_obj);
1471 struct thandle *handle;
1472 struct lu_fid *tfid = &mdd_env_info(env)->mti_fid2;
1473 struct linkea_data *ldata = &mdd_env_info(env)->mti_link_data;
1477 rc = mdd_la_get(env, mdd_sobj, cattr);
1481 rc = mdd_la_get(env, mdd_tobj, tattr);
1486 * If we are using project inheritance, we only allow hard link
1487 * creation in our tree when the project IDs are the same;
1488 * otherwise the tree quota mechanism could be circumvented.
1490 if ((tattr->la_flags & LUSTRE_PROJINHERIT_FL) &&
1491 (tattr->la_projid != cattr->la_projid))
1494 handle = mdd_trans_create(env, mdd);
1496 GOTO(out_pending, rc = PTR_ERR(handle));
1498 memset(ldata, 0, sizeof(*ldata));
1500 LASSERT(ma->ma_attr.la_valid & LA_CTIME);
1501 la->la_ctime = la->la_mtime = ma->ma_attr.la_ctime;
1503 /* Note: even this function will change ldata, but it comes from
1504 * thread_info, which is completely temporary and only seen in
1505 * this function, so we do not need reset ldata once it fails.*/
1506 rc = mdd_linkea_prepare(env, mdd_sobj, NULL, NULL, mdo2fid(mdd_tobj),
1507 lname, 0, 0, ldata);
1511 rc = mdd_declare_link(env, mdd, mdd_tobj, mdd_sobj, lname, handle,
1516 rc = mdd_trans_start(env, mdd, handle);
1520 mdd_write_lock(env, mdd_sobj, MOR_TGT_CHILD);
1521 rc = mdd_link_sanity_check(env, mdd_tobj, tattr, lname, mdd_sobj,
1524 GOTO(out_unlock, rc);
1526 if (!OBD_FAIL_CHECK(OBD_FAIL_LFSCK_LESS_NLINK)) {
1527 rc = mdo_ref_add(env, mdd_sobj, handle);
1529 GOTO(out_unlock, rc);
1532 *tfid = *mdo2fid(mdd_sobj);
1533 if (OBD_FAIL_CHECK(OBD_FAIL_LFSCK_DANGLING3))
1534 tfid->f_oid = cfs_fail_val;
1536 rc = __mdd_index_insert_only(env, mdd_tobj, tfid,
1537 mdd_object_type(mdd_sobj), name, handle);
1539 mdo_ref_del(env, mdd_sobj, handle);
1540 GOTO(out_unlock, rc);
1543 la->la_valid = LA_CTIME | LA_MTIME;
1544 rc = mdd_update_time(env, mdd_tobj, tattr, la, handle);
1546 GOTO(out_unlock, rc);
1548 la->la_valid = LA_CTIME;
1549 rc = mdd_update_time(env, mdd_sobj, cattr, la, handle);
1551 /* Note: The failure of links_add should not cause the
1552 * link failure, so do not check return value. */
1553 mdd_links_add(env, mdd_sobj, mdo2fid(mdd_tobj),
1554 lname, handle, ldata, 0);
1558 mdd_write_unlock(env, mdd_sobj);
1560 rc = mdd_changelog_ns_store(env, mdd, CL_HARDLINK, 0, mdd_sobj,
1561 mdo2fid(mdd_tobj), NULL, NULL,
1562 lname, NULL, handle);
1564 rc = mdd_trans_stop(env, mdd, rc, handle);
1565 if (is_vmalloc_addr(ldata->ld_buf))
1566 /* if we vmalloced a large buffer drop it */
1567 lu_buf_free(ldata->ld_buf);
1572 static int mdd_mark_orphan_object(const struct lu_env *env,
1573 struct mdd_object *obj, struct thandle *handle,
1576 struct lu_attr *attr = MDD_ENV_VAR(env, la_for_start);
1579 if (!S_ISDIR(mdd_object_type(obj)))
1582 attr->la_valid = LA_FLAGS;
1583 attr->la_flags = LUSTRE_ORPHAN_FL;
1586 rc = mdo_declare_attr_set(env, obj, attr, handle);
1588 rc = mdo_attr_set(env, obj, attr, handle);
1593 static int mdd_declare_finish_unlink(const struct lu_env *env,
1594 struct mdd_object *obj,
1595 struct thandle *handle)
1599 /* Sigh, we do not know if the unlink object will become orphan in
1600 * declare phase, but fortunately the flags here does not matter
1601 * in current declare implementation */
1602 rc = mdd_mark_orphan_object(env, obj, handle, true);
1606 rc = mdo_declare_destroy(env, obj, handle);
1610 rc = orph_declare_index_insert(env, obj, mdd_object_type(obj), handle);
1614 return mdd_declare_links_del(env, obj, handle);
1617 /* caller should take a lock before calling */
1618 int mdd_finish_unlink(const struct lu_env *env,
1619 struct mdd_object *obj, struct md_attr *ma,
1620 const struct mdd_object *pobj,
1621 const struct lu_name *lname,
1625 int is_dir = S_ISDIR(ma->ma_attr.la_mode);
1628 LASSERT(mdd_write_locked(env, obj) != 0);
1630 if (ma->ma_attr.la_nlink == 0 || is_dir) {
1631 /* add new orphan and the object
1632 * will be deleted during mdd_close() */
1633 obj->mod_flags |= DEAD_OBJ;
1634 if (obj->mod_count) {
1635 rc = __mdd_orphan_add(env, obj, th);
1637 CDEBUG(D_HA, "Object "DFID" is inserted into "
1638 "orphan list, open count = %d\n",
1639 PFID(mdd_object_fid(obj)),
1642 CERROR("Object "DFID" fail to be an orphan, "
1643 "open count = %d, maybe cause failed "
1645 PFID(mdd_object_fid(obj)),
1648 /* mark object as an orphan here, not
1649 * before __mdd_orphan_add() as racing
1650 * mdd_la_get() may propagate ORPHAN_OBJ
1651 * causing the asserition */
1652 rc = mdd_mark_orphan_object(env, obj, th, false);
1654 rc = mdo_destroy(env, obj, th);
1656 } else if (!is_dir) {
1657 /* old files may not have link ea; ignore errors */
1658 mdd_links_del(env, obj, mdo2fid(pobj), lname, th);
1666 * has mdd_write_lock on cobj already, but not on pobj yet
1668 int mdd_unlink_sanity_check(const struct lu_env *env, struct mdd_object *pobj,
1669 const struct lu_attr *pattr,
1670 struct mdd_object *cobj,
1671 const struct lu_attr *cattr)
1676 rc = mdd_may_delete(env, pobj, pattr, cobj, cattr, NULL, 1, 1);
1681 static int mdd_declare_unlink(const struct lu_env *env, struct mdd_device *mdd,
1682 struct mdd_object *p, struct mdd_object *c,
1683 const struct lu_name *name, struct md_attr *ma,
1684 struct thandle *handle, int no_name, int is_dir)
1686 struct lu_attr *la = &mdd_env_info(env)->mti_la_for_fix;
1689 if (!OBD_FAIL_CHECK(OBD_FAIL_LFSCK_DANGLING2)) {
1690 if (likely(no_name == 0)) {
1691 rc = mdo_declare_index_delete(env, p, name->ln_name,
1698 rc = mdo_declare_ref_del(env, p, handle);
1704 LASSERT(ma->ma_attr.la_valid & LA_CTIME);
1705 la->la_ctime = la->la_mtime = ma->ma_attr.la_ctime;
1706 la->la_valid = LA_CTIME | LA_MTIME;
1707 rc = mdo_declare_attr_set(env, p, la, handle);
1712 rc = mdo_declare_ref_del(env, c, handle);
1716 rc = mdo_declare_ref_del(env, c, handle);
1720 la->la_valid = LA_CTIME;
1721 rc = mdo_declare_attr_set(env, c, la, handle);
1725 rc = mdd_declare_finish_unlink(env, c, handle);
1729 /* FIXME: need changelog for remove entry */
1730 rc = mdd_declare_changelog_store(env, mdd, name, NULL, handle);
1737 * test if a file has an HSM archive
1738 * if HSM attributes are not found in ma update them from
1741 static bool mdd_hsm_archive_exists(const struct lu_env *env,
1742 struct mdd_object *obj,
1747 if (!(ma->ma_valid & MA_HSM)) {
1748 /* no HSM MD provided, read xattr */
1749 struct lu_buf *hsm_buf;
1750 const size_t buflen = sizeof(struct hsm_attrs);
1753 hsm_buf = mdd_buf_get(env, NULL, 0);
1754 lu_buf_alloc(hsm_buf, buflen);
1755 rc = mdo_xattr_get(env, obj, hsm_buf, XATTR_NAME_HSM);
1756 rc = lustre_buf2hsm(hsm_buf->lb_buf, rc, &ma->ma_hsm);
1757 lu_buf_free(hsm_buf);
1761 ma->ma_valid |= MA_HSM;
1763 if (ma->ma_hsm.mh_flags & HS_EXISTS)
1769 * Delete name entry and the object.
1770 * Note: no_name == 1 means it only destory the object, i.e. name_entry
1771 * does not exist for this object, and it could only happen during resending
1772 * of remote unlink. see the comments in mdt_reint_unlink. Unfortunately, lname
1773 * is also needed in this case(needed by changelog), so we have to add another
1774 * parameter(no_name)here. XXX: this is only needed in DNE phase I, on Phase II,
1775 * the ENOENT failure should be able to be fixed by redo mechanism.
1777 static int mdd_unlink(const struct lu_env *env, struct md_object *pobj,
1778 struct md_object *cobj, const struct lu_name *lname,
1779 struct md_attr *ma, int no_name)
1781 const char *name = lname->ln_name;
1782 struct lu_attr *pattr = MDD_ENV_VAR(env, pattr);
1783 struct lu_attr *cattr = MDD_ENV_VAR(env, cattr);
1784 struct lu_attr *la = &mdd_env_info(env)->mti_la_for_fix;
1785 struct mdd_object *mdd_pobj = md2mdd_obj(pobj);
1786 struct mdd_object *mdd_cobj = NULL;
1787 struct mdd_device *mdd = mdo2mdd(pobj);
1788 struct thandle *handle;
1789 int rc, is_dir = 0, cl_flags = 0;
1792 /* cobj == NULL means only delete name entry */
1793 if (likely(cobj != NULL)) {
1794 mdd_cobj = md2mdd_obj(cobj);
1795 if (mdd_object_exists(mdd_cobj) == 0)
1799 rc = mdd_la_get(env, mdd_pobj, pattr);
1803 if (likely(mdd_cobj != NULL)) {
1805 rc = mdd_la_get(env, mdd_cobj, cattr);
1809 is_dir = S_ISDIR(cattr->la_mode);
1810 /* search for an existing archive.
1811 * we should check ahead as the object
1812 * can be destroyed in this transaction */
1813 if (mdd_hsm_archive_exists(env, mdd_cobj, ma))
1814 cl_flags |= CLF_UNLINK_HSM_EXISTS;
1817 rc = mdd_unlink_sanity_check(env, mdd_pobj, pattr, mdd_cobj, cattr);
1821 handle = mdd_trans_create(env, mdd);
1823 RETURN(PTR_ERR(handle));
1825 rc = mdd_declare_unlink(env, mdd, mdd_pobj, mdd_cobj,
1826 lname, ma, handle, no_name, is_dir);
1830 rc = mdd_trans_start(env, mdd, handle);
1834 if (likely(mdd_cobj != NULL))
1835 mdd_write_lock(env, mdd_cobj, MOR_TGT_CHILD);
1837 if (likely(no_name == 0) && !OBD_FAIL_CHECK(OBD_FAIL_LFSCK_DANGLING2)) {
1838 rc = __mdd_index_delete(env, mdd_pobj, name, is_dir, handle);
1843 if (OBD_FAIL_CHECK(OBD_FAIL_LFSCK_MUL_REF) ||
1844 OBD_FAIL_CHECK(OBD_FAIL_LFSCK_NO_NAMEENTRY))
1845 GOTO(cleanup, rc = 0);
1847 if (likely(mdd_cobj != NULL)) {
1848 rc = mdo_ref_del(env, mdd_cobj, handle);
1850 __mdd_index_insert_only(env, mdd_pobj,
1852 mdd_object_type(mdd_cobj),
1859 mdo_ref_del(env, mdd_cobj, handle);
1861 /* fetch updated nlink */
1862 rc = mdd_la_get(env, mdd_cobj, cattr);
1867 LASSERT(ma->ma_attr.la_valid & LA_CTIME);
1868 la->la_ctime = la->la_mtime = ma->ma_attr.la_ctime;
1870 la->la_valid = LA_CTIME | LA_MTIME;
1871 rc = mdd_update_time(env, mdd_pobj, pattr, la, handle);
1875 /* Enough for only unlink the entry */
1876 if (unlikely(mdd_cobj == NULL))
1879 if (cattr->la_nlink > 0 || mdd_cobj->mod_count > 0) {
1880 /* update ctime of an unlinked file only if it is still
1881 * opened or a link still exists */
1882 la->la_valid = LA_CTIME;
1883 rc = mdd_update_time(env, mdd_cobj, cattr, la, handle);
1888 /* XXX: this transfer to ma will be removed with LOD/OSP */
1889 ma->ma_attr = *cattr;
1890 ma->ma_valid |= MA_INODE;
1891 rc = mdd_finish_unlink(env, mdd_cobj, ma, mdd_pobj, lname, handle);
1895 /* fetch updated nlink */
1896 rc = mdd_la_get(env, mdd_cobj, cattr);
1897 /* if object is removed then we can't get its attrs,
1899 if (rc == -ENOENT) {
1900 cattr->la_nlink = 0;
1904 if (cattr->la_nlink == 0) {
1905 ma->ma_attr = *cattr;
1906 ma->ma_valid |= MA_INODE;
1911 if (likely(mdd_cobj != NULL))
1912 mdd_write_unlock(env, mdd_cobj);
1915 if (cattr->la_nlink == 0)
1916 cl_flags |= CLF_UNLINK_LAST;
1918 cl_flags &= ~CLF_UNLINK_HSM_EXISTS;
1920 rc = mdd_changelog_ns_store(env, mdd,
1921 is_dir ? CL_RMDIR : CL_UNLINK, cl_flags,
1922 mdd_cobj, mdo2fid(mdd_pobj), NULL, NULL, lname, NULL,
1927 rc = mdd_trans_stop(env, mdd, rc, handle);
1933 * The permission has been checked when obj created, no need check again.
1935 static int mdd_cd_sanity_check(const struct lu_env *env,
1936 struct mdd_object *obj)
1941 if (!obj || mdd_is_dead_obj(obj))
1947 static int mdd_create_data(const struct lu_env *env,
1948 struct md_object *pobj,
1949 struct md_object *cobj,
1950 const struct md_op_spec *spec,
1953 struct mdd_device *mdd = mdo2mdd(cobj);
1954 struct mdd_object *mdd_pobj = md2mdd_obj(pobj);
1955 struct mdd_object *son = md2mdd_obj(cobj);
1956 struct thandle *handle;
1957 const struct lu_buf *buf;
1958 struct lu_attr *attr = MDD_ENV_VAR(env, cattr);
1959 struct dt_allocation_hint *hint = &mdd_env_info(env)->mti_hint;
1963 rc = mdd_cd_sanity_check(env, son);
1967 if (!md_should_create(spec->sp_cr_flags))
1971 * there are following use cases for this function:
1972 * 1) late striping - file was created with MDS_OPEN_DELAY_CREATE
1973 * striping can be specified or not
1976 rc = mdd_la_get(env, son, attr);
1980 /* calling ->ah_make_hint() is used to transfer information from parent */
1981 mdd_object_make_hint(env, mdd_pobj, son, attr, spec, hint);
1983 handle = mdd_trans_create(env, mdd);
1985 GOTO(out_free, rc = PTR_ERR(handle));
1988 * XXX: Setting the lov ea is not locked but setting the attr is locked?
1989 * Should this be fixed?
1991 CDEBUG(D_OTHER, "ea %p/%u, cr_flags %#llo, no_create %u\n",
1992 spec->u.sp_ea.eadata, spec->u.sp_ea.eadatalen,
1993 spec->sp_cr_flags, spec->no_create);
1995 if (spec->no_create || (spec->sp_cr_flags & MDS_OPEN_HAS_EA)) {
1996 buf = mdd_buf_get_const(env, spec->u.sp_ea.eadata,
1997 spec->u.sp_ea.eadatalen);
2002 rc = dt_declare_xattr_set(env, mdd_object_child(son), buf,
2003 XATTR_NAME_LOV, 0, handle);
2007 rc = mdd_declare_changelog_store(env, mdd, NULL, NULL, handle);
2011 rc = mdd_trans_start(env, mdd, handle);
2015 rc = dt_xattr_set(env, mdd_object_child(son), buf, XATTR_NAME_LOV,
2021 rc = mdd_changelog_data_store(env, mdd, CL_LAYOUT, 0, son, handle);
2024 rc = mdd_trans_stop(env, mdd, rc, handle);
2030 static int mdd_declare_object_initialize(const struct lu_env *env,
2031 struct mdd_object *parent,
2032 struct mdd_object *child,
2033 const struct lu_attr *attr,
2034 struct thandle *handle)
2039 LASSERT(attr->la_valid & (LA_MODE | LA_TYPE));
2040 if (!S_ISDIR(attr->la_mode))
2043 rc = mdo_declare_index_insert(env, child, mdo2fid(child), S_IFDIR,
2048 rc = mdo_declare_ref_add(env, child, handle);
2052 rc = mdo_declare_index_insert(env, child, mdo2fid(parent), S_IFDIR,
2058 static int mdd_object_initialize(const struct lu_env *env,
2059 const struct lu_fid *pfid,
2060 struct mdd_object *child,
2061 struct lu_attr *attr, struct thandle *handle,
2062 const struct md_op_spec *spec)
2067 if (S_ISDIR(attr->la_mode)) {
2068 /* Add "." and ".." for newly created dir */
2069 mdo_ref_add(env, child, handle);
2070 rc = __mdd_index_insert_only(env, child, mdo2fid(child),
2071 S_IFDIR, dot, handle);
2073 rc = __mdd_index_insert_only(env, child, pfid, S_IFDIR,
2076 mdo_ref_del(env, child, handle);
2083 * This function checks whether it can create a file/dir under the
2084 * directory(@pobj). The directory(@pobj) is not being locked by
2087 * \param[in] env execution environment
2088 * \param[in] pobj the directory to create files
2089 * \param[in] pattr the attributes of the directory
2090 * \param[in] lname the name of the created file/dir
2091 * \param[in] cattr the attributes of the file/dir
2092 * \param[in] spec create specification
2094 * \retval = 0 it is allowed to create file/dir under
2096 * \retval negative error not allowed to create file/dir
2097 * under the directory
2099 static int mdd_create_sanity_check(const struct lu_env *env,
2100 struct md_object *pobj,
2101 const struct lu_attr *pattr,
2102 const struct lu_name *lname,
2103 struct lu_attr *cattr,
2104 struct md_op_spec *spec)
2106 struct mdd_thread_info *info = mdd_env_info(env);
2107 struct lu_fid *fid = &info->mti_fid;
2108 struct mdd_object *obj = md2mdd_obj(pobj);
2109 struct mdd_device *m = mdo2mdd(pobj);
2110 bool check_perm = true;
2115 if (mdd_is_dead_obj(obj))
2119 * In some cases this lookup is not needed - we know before if name
2120 * exists or not because MDT performs lookup for it.
2121 * name length check is done in lookup.
2123 if (spec->sp_cr_lookup) {
2125 * Check if the name already exist, though it will be checked in
2126 * _index_insert also, for avoiding rolling back if exists
2129 rc = __mdd_lookup(env, pobj, pattr, lname, fid,
2130 MAY_WRITE | MAY_EXEC);
2132 RETURN(rc ? : -EEXIST);
2134 /* Permission is already being checked in mdd_lookup */
2138 if (S_ISDIR(cattr->la_mode) &&
2139 unlikely(spec != NULL && spec->sp_cr_flags & MDS_OPEN_HAS_EA) &&
2140 spec->u.sp_ea.eadata != NULL && spec->u.sp_ea.eadatalen > 0) {
2141 const struct lmv_user_md *lum = spec->u.sp_ea.eadata;
2143 if (le32_to_cpu(lum->lum_magic) != LMV_USER_MAGIC &&
2144 le32_to_cpu(lum->lum_magic) != LMV_USER_MAGIC_SPECIFIC &&
2145 le32_to_cpu(lum->lum_magic) != LMV_USER_MAGIC_V0) {
2147 CERROR("%s: invalid lmv_user_md: magic = %x, "
2148 "stripe_offset = %d, stripe_count = %u: "
2149 "rc = %d\n", mdd2obd_dev(m)->obd_name,
2150 le32_to_cpu(lum->lum_magic),
2151 (int)le32_to_cpu(lum->lum_stripe_offset),
2152 le32_to_cpu(lum->lum_stripe_count), rc);
2157 rc = mdd_may_create(env, obj, pattr, NULL, check_perm);
2162 if (pattr->la_mode & S_ISGID) {
2163 cattr->la_gid = pattr->la_gid;
2164 if (S_ISDIR(cattr->la_mode)) {
2165 cattr->la_mode |= S_ISGID;
2166 cattr->la_valid |= LA_MODE;
2170 /* Inherit project ID from parent directory */
2171 if (pattr->la_flags & LUSTRE_PROJINHERIT_FL) {
2172 cattr->la_projid = pattr->la_projid;
2173 if (S_ISDIR(cattr->la_mode)) {
2174 cattr->la_flags |= LUSTRE_PROJINHERIT_FL;
2175 cattr->la_valid |= LA_FLAGS;
2177 cattr->la_valid |= LA_PROJID;
2180 rc = mdd_name_check(m, lname);
2184 switch (cattr->la_mode & S_IFMT) {
2186 unsigned int symlen = strlen(spec->u.sp_symname) + 1;
2188 if (symlen > m->mdd_dt_conf.ddp_symlink_max)
2189 RETURN(-ENAMETOOLONG);
2208 static int mdd_declare_create_object(const struct lu_env *env,
2209 struct mdd_device *mdd,
2210 struct mdd_object *p, struct mdd_object *c,
2211 struct lu_attr *attr,
2212 struct thandle *handle,
2213 const struct md_op_spec *spec,
2214 struct lu_buf *def_acl_buf,
2215 struct lu_buf *acl_buf,
2216 struct dt_allocation_hint *hint)
2218 const struct lu_buf *buf;
2221 rc = mdd_declare_create_object_internal(env, p, c, attr, handle, spec,
2226 #ifdef CONFIG_FS_POSIX_ACL
2227 if (def_acl_buf->lb_len > 0 && S_ISDIR(attr->la_mode)) {
2228 /* if dir, then can inherit default ACl */
2229 rc = mdo_declare_xattr_set(env, c, def_acl_buf,
2230 XATTR_NAME_ACL_DEFAULT,
2236 if (acl_buf->lb_len > 0) {
2237 rc = mdo_declare_attr_set(env, c, attr, handle);
2241 rc = mdo_declare_xattr_set(env, c, acl_buf,
2242 XATTR_NAME_ACL_ACCESS, 0, handle);
2247 rc = mdd_declare_object_initialize(env, p, c, attr, handle);
2251 /* replay case, create LOV EA from client data */
2252 if ((!(spec->sp_cr_flags & MDS_OPEN_DELAY_CREATE) && spec->no_create) ||
2253 (spec->sp_cr_flags & MDS_OPEN_HAS_EA && S_ISREG(attr->la_mode))) {
2254 buf = mdd_buf_get_const(env, spec->u.sp_ea.eadata,
2255 spec->u.sp_ea.eadatalen);
2256 rc = mdo_declare_xattr_set(env, c, buf, XATTR_NAME_LOV, 0,
2262 if (S_ISLNK(attr->la_mode)) {
2263 const char *target_name = spec->u.sp_symname;
2264 int sym_len = strlen(target_name);
2265 const struct lu_buf *buf;
2267 buf = mdd_buf_get_const(env, target_name, sym_len);
2268 rc = dt_declare_record_write(env, mdd_object_child(c),
2274 if (spec->sp_cr_file_secctx_name != NULL) {
2275 buf = mdd_buf_get_const(env, spec->sp_cr_file_secctx,
2276 spec->sp_cr_file_secctx_size);
2277 rc = mdo_declare_xattr_set(env, c, buf,
2278 spec->sp_cr_file_secctx_name, 0,
2287 static int mdd_declare_create(const struct lu_env *env, struct mdd_device *mdd,
2288 struct mdd_object *p, struct mdd_object *c,
2289 const struct lu_name *name,
2290 struct lu_attr *attr,
2291 struct thandle *handle,
2292 const struct md_op_spec *spec,
2293 struct linkea_data *ldata,
2294 struct lu_buf *def_acl_buf,
2295 struct lu_buf *acl_buf,
2296 struct dt_allocation_hint *hint)
2300 rc = mdd_declare_create_object(env, mdd, p, c, attr, handle, spec,
2301 def_acl_buf, acl_buf, hint);
2305 if (S_ISDIR(attr->la_mode)) {
2306 rc = mdo_declare_ref_add(env, p, handle);
2311 if (unlikely(spec->sp_cr_flags & MDS_OPEN_VOLATILE)) {
2312 rc = orph_declare_index_insert(env, c, attr->la_mode, handle);
2316 struct lu_attr *la = &mdd_env_info(env)->mti_la_for_fix;
2318 rc = mdo_declare_index_insert(env, p, mdo2fid(c), attr->la_mode,
2319 name->ln_name, handle);
2323 rc = mdd_declare_links_add(env, c, handle, ldata);
2328 la->la_valid = LA_CTIME | LA_MTIME;
2329 rc = mdo_declare_attr_set(env, p, la, handle);
2333 rc = mdd_declare_changelog_store(env, mdd, name, NULL, handle);
2341 static int mdd_acl_init(const struct lu_env *env, struct mdd_object *pobj,
2342 struct lu_attr *la, struct lu_buf *def_acl_buf,
2343 struct lu_buf *acl_buf)
2348 if (S_ISLNK(la->la_mode)) {
2349 acl_buf->lb_len = 0;
2350 def_acl_buf->lb_len = 0;
2354 mdd_read_lock(env, pobj, MOR_TGT_PARENT);
2355 rc = mdo_xattr_get(env, pobj, def_acl_buf,
2356 XATTR_NAME_ACL_DEFAULT);
2357 mdd_read_unlock(env, pobj);
2359 /* If there are default ACL, fix mode/ACL by default ACL */
2360 def_acl_buf->lb_len = rc;
2361 LASSERT(def_acl_buf->lb_len <= acl_buf->lb_len);
2362 memcpy(acl_buf->lb_buf, def_acl_buf->lb_buf, rc);
2363 acl_buf->lb_len = rc;
2364 rc = __mdd_fix_mode_acl(env, acl_buf, &la->la_mode);
2367 } else if (rc == -ENODATA || rc == -EOPNOTSUPP) {
2368 /* If there are no default ACL, fix mode by mask */
2369 struct lu_ucred *uc = lu_ucred(env);
2371 /* The create triggered by MDT internal events, such as
2372 * LFSCK reset, will not contain valid "uc". */
2373 if (unlikely(uc != NULL))
2374 la->la_mode &= ~uc->uc_umask;
2376 acl_buf->lb_len = 0;
2377 def_acl_buf->lb_len = 0;
2384 * Create a metadata object and initialize it, set acl, xattr.
2386 static int mdd_create_object(const struct lu_env *env, struct mdd_object *pobj,
2387 struct mdd_object *son, struct lu_attr *attr,
2388 struct md_op_spec *spec, struct lu_buf *acl_buf,
2389 struct lu_buf *def_acl_buf,
2390 struct dt_allocation_hint *hint,
2391 struct thandle *handle)
2393 const struct lu_buf *buf;
2396 mdd_write_lock(env, son, MOR_TGT_CHILD);
2397 rc = mdd_create_object_internal(env, NULL, son, attr, handle, spec,
2402 /* Note: In DNE phase I, for striped dir, though sub-stripes will be
2403 * created in declare phase, they also needs to be added to master
2404 * object as sub-directory entry. So it has to initialize the master
2405 * object, then set dir striped EA.(in mdo_xattr_set) */
2406 rc = mdd_object_initialize(env, mdo2fid(pobj), son, attr, handle,
2409 GOTO(err_destroy, rc);
2412 * in case of replay we just set LOVEA provided by the client
2413 * XXX: I think it would be interesting to try "old" way where
2414 * MDT calls this xattr_set(LOV) in a different transaction.
2415 * probably this way we code can be made better.
2418 /* During creation, there are only a few cases we need do xattr_set to
2420 * 1. regular file: see comments above.
2421 * 2. dir: inherit default striping or pool settings from parent.
2422 * 3. create striped directory with provided stripeEA.
2423 * 4. create striped directory because inherit default layout from the
2426 if (spec->no_create ||
2427 (S_ISREG(attr->la_mode) && spec->sp_cr_flags & MDS_OPEN_HAS_EA) ||
2428 S_ISDIR(attr->la_mode)) {
2429 buf = mdd_buf_get_const(env, spec->u.sp_ea.eadata,
2430 spec->u.sp_ea.eadatalen);
2431 rc = mdo_xattr_set(env, son, buf,
2432 S_ISDIR(attr->la_mode) ? XATTR_NAME_LMV :
2436 GOTO(err_destroy, rc);
2439 #ifdef CONFIG_FS_POSIX_ACL
2440 if (def_acl_buf != NULL && def_acl_buf->lb_len > 0 &&
2441 S_ISDIR(attr->la_mode)) {
2442 /* set default acl */
2443 rc = mdo_xattr_set(env, son, def_acl_buf,
2444 XATTR_NAME_ACL_DEFAULT, 0,
2447 GOTO(err_destroy, rc);
2449 /* set its own acl */
2450 if (acl_buf != NULL && acl_buf->lb_len > 0) {
2451 rc = mdo_xattr_set(env, son, acl_buf,
2452 XATTR_NAME_ACL_ACCESS,
2455 GOTO(err_destroy, rc);
2459 if (S_ISLNK(attr->la_mode)) {
2460 struct lu_ucred *uc = lu_ucred_assert(env);
2461 struct dt_object *dt = mdd_object_child(son);
2462 const char *target_name = spec->u.sp_symname;
2463 int sym_len = strlen(target_name);
2466 buf = mdd_buf_get_const(env, target_name, sym_len);
2467 rc = dt->do_body_ops->dbo_write(env, dt, buf, &pos, handle,
2469 CFS_CAP_SYS_RESOURCE_MASK);
2474 GOTO(err_initlized, rc = -EFAULT);
2477 if (spec->sp_cr_file_secctx_name != NULL) {
2478 buf = mdd_buf_get_const(env, spec->sp_cr_file_secctx,
2479 spec->sp_cr_file_secctx_size);
2480 rc = mdo_xattr_set(env, son, buf, spec->sp_cr_file_secctx_name,
2483 GOTO(err_initlized, rc);
2487 if (unlikely(rc != 0)) {
2489 if (S_ISDIR(attr->la_mode)) {
2490 /* Drop the reference, no need to delete "."/"..",
2491 * because the object to be destroied directly. */
2492 rc2 = mdo_ref_del(env, son, handle);
2496 rc2 = mdo_ref_del(env, son, handle);
2500 mdo_destroy(env, son, handle);
2503 mdd_write_unlock(env, son);
2507 static int mdd_index_delete(const struct lu_env *env,
2508 struct mdd_object *mdd_pobj,
2509 struct lu_attr *cattr,
2510 const struct lu_name *lname)
2512 struct mdd_device *mdd = mdo2mdd(&mdd_pobj->mod_obj);
2513 struct thandle *handle;
2517 handle = mdd_trans_create(env, mdd);
2519 RETURN(PTR_ERR(handle));
2521 rc = mdo_declare_index_delete(env, mdd_pobj, lname->ln_name,
2526 if (S_ISDIR(cattr->la_mode)) {
2527 rc = mdo_declare_ref_del(env, mdd_pobj, handle);
2532 /* Since this will only be used in the error handler path,
2533 * Let's set the thandle to be local and not mess the transno */
2534 handle->th_local = 1;
2535 rc = mdd_trans_start(env, mdd, handle);
2539 rc = __mdd_index_delete(env, mdd_pobj, lname->ln_name,
2540 S_ISDIR(cattr->la_mode), handle);
2544 rc = mdd_trans_stop(env, mdd, rc, handle);
2550 * Create object and insert it into namespace.
2552 * Two operations have to be performed:
2554 * - an allocation of a new object (->do_create()), and
2555 * - an insertion into a parent index (->dio_insert()).
2557 * Due to locking, operation order is not important, when both are
2558 * successful, *but* error handling cases are quite different:
2560 * - if insertion is done first, and following object creation fails,
2561 * insertion has to be rolled back, but this operation might fail
2562 * also leaving us with dangling index entry.
2564 * - if creation is done first, is has to be undone if insertion fails,
2565 * leaving us with leaked space, which is not good but not fatal.
2567 * It seems that creation-first is simplest solution, but it is sub-optimal
2573 * case, because second mkdir is bound to create object, only to
2574 * destroy it immediately.
2576 * To avoid this follow local file systems that do double lookup:
2578 * 0. lookup -> -EEXIST (mdd_create_sanity_check())
2579 * 1. create (mdd_create_object_internal())
2580 * 2. insert (__mdd_index_insert(), lookup again)
2582 * \param[in] pobj parent object
2583 * \param[in] lname name of child being created
2584 * \param[in,out] child child object being created
2585 * \param[in] spec additional create parameters
2586 * \param[in] ma attributes for new child object
2588 * \retval 0 on success
2589 * \retval negative errno on failure
2591 static int mdd_create(const struct lu_env *env, struct md_object *pobj,
2592 const struct lu_name *lname, struct md_object *child,
2593 struct md_op_spec *spec, struct md_attr *ma)
2595 struct mdd_thread_info *info = mdd_env_info(env);
2596 struct lu_attr *la = &info->mti_la_for_fix;
2597 struct mdd_object *mdd_pobj = md2mdd_obj(pobj);
2598 struct mdd_object *son = md2mdd_obj(child);
2599 struct mdd_device *mdd = mdo2mdd(pobj);
2600 struct lu_attr *attr = &ma->ma_attr;
2601 struct thandle *handle;
2602 struct lu_attr *pattr = &info->mti_pattr;
2603 struct lu_buf acl_buf;
2604 struct lu_buf def_acl_buf;
2605 struct linkea_data *ldata = &info->mti_link_data;
2606 const char *name = lname->ln_name;
2607 struct dt_allocation_hint *hint = &mdd_env_info(env)->mti_hint;
2612 rc = mdd_la_get(env, mdd_pobj, pattr);
2616 /* Sanity checks before big job. */
2617 rc = mdd_create_sanity_check(env, pobj, pattr, lname, attr, spec);
2621 if (OBD_FAIL_CHECK(OBD_FAIL_MDS_DQACQ_NET))
2622 GOTO(out_free, rc = -EINPROGRESS);
2624 handle = mdd_trans_create(env, mdd);
2626 GOTO(out_free, rc = PTR_ERR(handle));
2628 lu_buf_check_and_alloc(&info->mti_xattr_buf,
2629 mdd->mdd_dt_conf.ddp_max_ea_size);
2630 acl_buf = info->mti_xattr_buf;
2631 def_acl_buf.lb_buf = info->mti_key;
2632 def_acl_buf.lb_len = sizeof(info->mti_key);
2633 rc = mdd_acl_init(env, mdd_pobj, attr, &def_acl_buf, &acl_buf);
2637 mdd_object_make_hint(env, mdd_pobj, son, attr, spec, hint);
2639 memset(ldata, 0, sizeof(*ldata));
2640 if (OBD_FAIL_CHECK(OBD_FAIL_LFSCK_BAD_PARENT)) {
2641 struct lu_fid tfid = *mdd_object_fid(mdd_pobj);
2644 rc = mdd_linkea_prepare(env, son, NULL, NULL,
2645 &tfid, lname, 1, 0, ldata);
2647 rc = mdd_linkea_prepare(env, son, NULL, NULL,
2648 mdd_object_fid(mdd_pobj),
2649 lname, 1, 0, ldata);
2652 rc = mdd_declare_create(env, mdd, mdd_pobj, son, lname, attr,
2653 handle, spec, ldata, &def_acl_buf, &acl_buf,
2658 rc = mdd_trans_start(env, mdd, handle);
2662 rc = mdd_create_object(env, mdd_pobj, son, attr, spec, &acl_buf,
2663 &def_acl_buf, hint, handle);
2667 if (unlikely(spec->sp_cr_flags & MDS_OPEN_VOLATILE)) {
2668 mdd_write_lock(env, son, MOR_TGT_CHILD);
2669 son->mod_flags |= VOLATILE_OBJ;
2670 rc = __mdd_orphan_add(env, son, handle);
2671 GOTO(out_volatile, rc);
2673 rc = __mdd_index_insert(env, mdd_pobj, mdo2fid(son),
2674 attr->la_mode, name, handle);
2676 GOTO(err_created, rc);
2678 mdd_links_add(env, son, mdo2fid(mdd_pobj), lname, handle,
2681 /* update parent directory mtime/ctime */
2683 la->la_valid = LA_CTIME | LA_MTIME;
2684 rc = mdd_update_time(env, mdd_pobj, pattr, la, handle);
2686 GOTO(err_insert, rc);
2692 if (spec->sp_cr_flags & MDS_OPEN_VOLATILE)
2693 rc2 = __mdd_orphan_del(env, son, handle);
2695 rc2 = __mdd_index_delete(env, mdd_pobj, name,
2696 S_ISDIR(attr->la_mode),
2702 mdd_write_lock(env, son, MOR_TGT_CHILD);
2703 if (S_ISDIR(attr->la_mode)) {
2704 /* Drop the reference, no need to delete "."/"..",
2705 * because the object is to be destroyed directly. */
2706 rc2 = mdo_ref_del(env, son, handle);
2708 mdd_write_unlock(env, son);
2713 /* For volatile files drop one link immediately, since there is
2714 * no filename in the namespace, and save any error returned. */
2715 rc2 = mdo_ref_del(env, son, handle);
2717 mdd_write_unlock(env, son);
2718 if (unlikely(rc == 0))
2723 /* Don't destroy the volatile object on success */
2724 if (likely(rc != 0))
2725 mdo_destroy(env, son, handle);
2726 mdd_write_unlock(env, son);
2729 if (rc == 0 && fid_is_namespace_visible(mdo2fid(son)) &&
2730 likely((spec->sp_cr_flags & MDS_OPEN_VOLATILE) == 0))
2731 rc = mdd_changelog_ns_store(env, mdd,
2732 S_ISDIR(attr->la_mode) ? CL_MKDIR :
2733 S_ISREG(attr->la_mode) ? CL_CREATE :
2734 S_ISLNK(attr->la_mode) ? CL_SOFTLINK : CL_MKNOD,
2735 0, son, mdo2fid(mdd_pobj), NULL, NULL, lname,
2738 rc2 = mdd_trans_stop(env, mdd, rc, handle);
2740 /* If creation fails, it is most likely due to the remote update
2741 * failure, because local transaction will mostly succeed at
2742 * this stage. There is no easy way to rollback all of previous
2743 * updates, so let's remove the object from namespace, and
2744 * LFSCK should handle the orphan object. */
2745 if (rc2 < 0 && !mdd_object_remote(mdd_pobj))
2746 mdd_index_delete(env, mdd_pobj, attr, lname);
2750 if (is_vmalloc_addr(ldata->ld_buf))
2751 /* if we vmalloced a large buffer drop it */
2752 lu_buf_free(ldata->ld_buf);
2754 /* The child object shouldn't be cached anymore */
2756 set_bit(LU_OBJECT_HEARD_BANSHEE,
2757 &child->mo_lu.lo_header->loh_flags);
2762 * Get locks on parents in proper order
2763 * RETURN: < 0 - error, rename_order if successful
2771 static int mdd_rename_order(const struct lu_env *env,
2772 struct mdd_device *mdd,
2773 struct mdd_object *src_pobj,
2774 const struct lu_attr *pattr,
2775 struct mdd_object *tgt_pobj)
2777 /* order of locking, 1 - tgt-src, 0 - src-tgt*/
2781 if (src_pobj == tgt_pobj)
2782 RETURN(MDD_RN_SAME);
2784 /* compared the parent child relationship of src_p&tgt_p */
2785 if (lu_fid_eq(&mdd->mdd_root_fid, mdo2fid(src_pobj))){
2787 } else if (lu_fid_eq(&mdd->mdd_root_fid, mdo2fid(tgt_pobj))) {
2790 rc = mdd_is_parent(env, mdd, src_pobj, pattr, mdo2fid(tgt_pobj),
2804 /* has not mdd_write{read}_lock on any obj yet. */
2805 static int mdd_rename_sanity_check(const struct lu_env *env,
2806 struct mdd_object *src_pobj,
2807 const struct lu_attr *pattr,
2808 struct mdd_object *tgt_pobj,
2809 const struct lu_attr *tpattr,
2810 struct mdd_object *sobj,
2811 const struct lu_attr *cattr,
2812 struct mdd_object *tobj,
2813 const struct lu_attr *tattr)
2818 /* XXX: when get here, sobj must NOT be NULL,
2819 * the other case has been processed in cld_rename
2820 * before mdd_rename and enable MDS_PERM_BYPASS. */
2824 * If we are using project inheritance, we only allow renames
2825 * into our tree when the project IDs are the same; otherwise
2826 * tree quota mechanism would be circumvented.
2828 if (((tpattr->la_flags & LUSTRE_PROJINHERIT_FL) &&
2829 tpattr->la_projid != cattr->la_projid) ||
2830 ((pattr->la_flags & LUSTRE_PROJINHERIT_FL) &&
2831 (pattr->la_projid != tpattr->la_projid)))
2834 rc = mdd_may_delete(env, src_pobj, pattr, sobj, cattr, NULL, 1, 0);
2838 /* XXX: when get here, "tobj == NULL" means tobj must
2839 * NOT exist (neither on remote MDS, such case has been
2840 * processed in cld_rename before mdd_rename and enable
2842 * So check may_create, but not check may_unlink. */
2844 rc = mdd_may_create(env, tgt_pobj, tpattr, NULL,
2845 (src_pobj != tgt_pobj));
2847 rc = mdd_may_delete(env, tgt_pobj, tpattr, tobj, tattr, cattr,
2848 (src_pobj != tgt_pobj), 1);
2850 if (!rc && !tobj && (src_pobj != tgt_pobj) && S_ISDIR(cattr->la_mode))
2851 rc = __mdd_may_link(env, tgt_pobj, tpattr);
2856 static int mdd_declare_rename(const struct lu_env *env,
2857 struct mdd_device *mdd,
2858 struct mdd_object *mdd_spobj,
2859 struct mdd_object *mdd_tpobj,
2860 struct mdd_object *mdd_sobj,
2861 struct mdd_object *mdd_tobj,
2862 const struct lu_name *sname,
2863 const struct lu_name *tname,
2865 struct linkea_data *ldata,
2866 struct thandle *handle)
2868 struct lu_attr *la = &mdd_env_info(env)->mti_la_for_fix;
2871 LASSERT(ma->ma_attr.la_valid & LA_CTIME);
2872 la->la_ctime = la->la_mtime = ma->ma_attr.la_ctime;
2878 /* name from source dir */
2879 rc = mdo_declare_index_delete(env, mdd_spobj, sname->ln_name, handle);
2883 /* .. from source child */
2884 if (S_ISDIR(mdd_object_type(mdd_sobj))) {
2885 /* source child can be directory,
2886 * counted by source dir's nlink */
2887 rc = mdo_declare_ref_del(env, mdd_spobj, handle);
2890 if (mdd_spobj != mdd_tpobj) {
2891 rc = mdo_declare_index_delete(env, mdd_sobj, dotdot,
2896 rc = mdo_declare_index_insert(env, mdd_sobj,
2898 S_IFDIR, dotdot, handle);
2903 /* new target child can be directory,
2904 * counted by target dir's nlink */
2905 rc = mdo_declare_ref_add(env, mdd_tpobj, handle);
2910 la->la_valid = LA_CTIME | LA_MTIME;
2911 rc = mdo_declare_attr_set(env, mdd_spobj, la, handle);
2915 rc = mdo_declare_attr_set(env, mdd_tpobj, la, handle);
2919 la->la_valid = LA_CTIME;
2920 rc = mdo_declare_attr_set(env, mdd_sobj, la, handle);
2924 rc = mdd_declare_links_add(env, mdd_sobj, handle, ldata);
2929 rc = mdo_declare_index_insert(env, mdd_tpobj, mdo2fid(mdd_sobj),
2930 mdd_object_type(mdd_sobj),
2931 tname->ln_name, handle);
2935 if (mdd_tobj && mdd_object_exists(mdd_tobj)) {
2936 /* delete target child in target parent directory */
2937 rc = mdo_declare_index_delete(env, mdd_tpobj, tname->ln_name,
2942 rc = mdo_declare_ref_del(env, mdd_tobj, handle);
2946 if (S_ISDIR(mdd_object_type(mdd_tobj))) {
2947 /* target child can be directory,
2948 * delete "." reference in target child directory */
2949 rc = mdo_declare_ref_del(env, mdd_tobj, handle);
2953 /* delete ".." reference in target parent directory */
2954 rc = mdo_declare_ref_del(env, mdd_tpobj, handle);
2959 la->la_valid = LA_CTIME;
2960 rc = mdo_declare_attr_set(env, mdd_tobj, la, handle);
2964 rc = mdd_declare_finish_unlink(env, mdd_tobj, handle);
2969 rc = mdd_declare_changelog_store(env, mdd, tname, sname, handle);
2976 /* src object can be remote that is why we use only fid and type of object */
2977 static int mdd_rename(const struct lu_env *env,
2978 struct md_object *src_pobj, struct md_object *tgt_pobj,
2979 const struct lu_fid *lf, const struct lu_name *lsname,
2980 struct md_object *tobj, const struct lu_name *ltname,
2983 const char *sname = lsname->ln_name;
2984 const char *tname = ltname->ln_name;
2985 struct lu_attr *la = &mdd_env_info(env)->mti_la_for_fix;
2986 struct mdd_object *mdd_spobj = md2mdd_obj(src_pobj); /* source parent */
2987 struct mdd_object *mdd_tpobj = md2mdd_obj(tgt_pobj);
2988 struct mdd_device *mdd = mdo2mdd(src_pobj);
2989 struct mdd_object *mdd_sobj = NULL; /* source object */
2990 struct mdd_object *mdd_tobj = NULL;
2991 struct lu_attr *cattr = MDD_ENV_VAR(env, cattr);
2992 struct lu_attr *pattr = MDD_ENV_VAR(env, pattr);
2993 struct lu_attr *tattr = MDD_ENV_VAR(env, tattr);
2994 struct lu_attr *tpattr = MDD_ENV_VAR(env, tpattr);
2995 struct thandle *handle;
2996 struct linkea_data *ldata = &mdd_env_info(env)->mti_link_data;
2997 const struct lu_fid *tpobj_fid = mdo2fid(mdd_tpobj);
2998 const struct lu_fid *spobj_fid = mdo2fid(mdd_spobj);
3001 bool tobj_locked = 0;
3002 unsigned cl_flags = 0;
3007 mdd_tobj = md2mdd_obj(tobj);
3009 mdd_sobj = mdd_object_find(env, mdd, lf);
3010 if (IS_ERR(mdd_sobj))
3011 RETURN(PTR_ERR(mdd_sobj));
3013 rc = mdd_la_get(env, mdd_sobj, cattr);
3015 GOTO(out_pending, rc);
3017 rc = mdd_la_get(env, mdd_spobj, pattr);
3019 GOTO(out_pending, rc);
3022 rc = mdd_la_get(env, mdd_tobj, tattr);
3024 GOTO(out_pending, rc);
3025 /* search for an existing archive.
3026 * we should check ahead as the object
3027 * can be destroyed in this transaction */
3028 if (mdd_hsm_archive_exists(env, mdd_tobj, ma))
3029 cl_flags |= CLF_RENAME_LAST_EXISTS;
3032 rc = mdd_la_get(env, mdd_tpobj, tpattr);
3034 GOTO(out_pending, rc);
3036 rc = mdd_rename_sanity_check(env, mdd_spobj, pattr, mdd_tpobj, tpattr,
3037 mdd_sobj, cattr, mdd_tobj, tattr);
3039 GOTO(out_pending, rc);
3041 rc = mdd_name_check(mdd, ltname);
3043 GOTO(out_pending, rc);
3045 /* FIXME: Should consider tobj and sobj too in rename_lock. */
3046 rc = mdd_rename_order(env, mdd, mdd_spobj, pattr, mdd_tpobj);
3048 GOTO(out_pending, rc);
3050 handle = mdd_trans_create(env, mdd);
3052 GOTO(out_pending, rc = PTR_ERR(handle));
3054 memset(ldata, 0, sizeof(*ldata));
3055 rc = mdd_linkea_prepare(env, mdd_sobj, mdd_object_fid(mdd_spobj),
3056 lsname, mdd_object_fid(mdd_tpobj), ltname,
3061 rc = mdd_declare_rename(env, mdd, mdd_spobj, mdd_tpobj, mdd_sobj,
3062 mdd_tobj, lsname, ltname, ma, ldata, handle);
3066 rc = mdd_trans_start(env, mdd, handle);
3070 is_dir = S_ISDIR(cattr->la_mode);
3072 /* Remove source name from source directory */
3073 rc = __mdd_index_delete(env, mdd_spobj, sname, is_dir, handle);
3077 /* "mv dir1 dir2" needs "dir1/.." link update */
3078 if (is_dir && !lu_fid_eq(spobj_fid, tpobj_fid)) {
3079 rc = __mdd_index_delete_only(env, mdd_sobj, dotdot, handle);
3081 GOTO(fixup_spobj2, rc);
3083 rc = __mdd_index_insert_only(env, mdd_sobj, tpobj_fid, S_IFDIR,
3086 GOTO(fixup_spobj, rc);
3089 if (mdd_tobj != NULL && mdd_object_exists(mdd_tobj)) {
3090 rc = __mdd_index_delete(env, mdd_tpobj, tname, is_dir, handle);
3092 /* tname might been renamed to something else */
3093 GOTO(fixup_spobj, rc);
3096 /* Insert new fid with target name into target dir */
3097 rc = __mdd_index_insert(env, mdd_tpobj, lf, cattr->la_mode,
3100 GOTO(fixup_tpobj, rc);
3102 LASSERT(ma->ma_attr.la_valid & LA_CTIME);
3103 la->la_ctime = la->la_mtime = ma->ma_attr.la_ctime;
3105 /* XXX: mdd_sobj must be local one if it is NOT NULL. */
3106 la->la_valid = LA_CTIME;
3107 rc = mdd_update_time(env, mdd_sobj, cattr, la, handle);
3109 GOTO(fixup_tpobj, rc);
3111 /* Update the linkEA for the source object */
3112 mdd_write_lock(env, mdd_sobj, MOR_SRC_CHILD);
3113 rc = mdd_links_rename(env, mdd_sobj, mdo2fid(mdd_spobj), lsname,
3114 mdo2fid(mdd_tpobj), ltname, handle, ldata,
3117 /* Old files might not have EA entry */
3118 mdd_links_add(env, mdd_sobj, mdo2fid(mdd_spobj),
3119 lsname, handle, NULL, 0);
3120 mdd_write_unlock(env, mdd_sobj);
3121 /* We don't fail the transaction if the link ea can't be
3122 updated -- fid2path will use alternate lookup method. */
3125 /* Remove old target object
3126 * For tobj is remote case cmm layer has processed
3127 * and set tobj to NULL then. So when tobj is NOT NULL,
3128 * it must be local one.
3130 if (tobj && mdd_object_exists(mdd_tobj)) {
3131 mdd_write_lock(env, mdd_tobj, MOR_TGT_CHILD);
3133 if (mdd_is_dead_obj(mdd_tobj)) {
3134 /* shld not be dead, something is wrong */
3135 CERROR("tobj is dead, something is wrong\n");
3139 mdo_ref_del(env, mdd_tobj, handle);
3141 /* Remove dot reference. */
3142 if (S_ISDIR(tattr->la_mode))
3143 mdo_ref_del(env, mdd_tobj, handle);
3146 /* fetch updated nlink */
3147 rc = mdd_la_get(env, mdd_tobj, tattr);
3149 CERROR("%s: Failed to get nlink for tobj "
3151 mdd2obd_dev(mdd)->obd_name,
3152 PFID(tpobj_fid), rc);
3153 GOTO(fixup_tpobj, rc);
3156 la->la_valid = LA_CTIME;
3157 rc = mdd_update_time(env, mdd_tobj, tattr, la, handle);
3159 CERROR("%s: Failed to set ctime for tobj "
3161 mdd2obd_dev(mdd)->obd_name,
3162 PFID(tpobj_fid), rc);
3163 GOTO(fixup_tpobj, rc);
3166 /* XXX: this transfer to ma will be removed with LOD/OSP */
3167 ma->ma_attr = *tattr;
3168 ma->ma_valid |= MA_INODE;
3169 rc = mdd_finish_unlink(env, mdd_tobj, ma, mdd_tpobj, ltname,
3172 CERROR("%s: Failed to unlink tobj "
3174 mdd2obd_dev(mdd)->obd_name,
3175 PFID(tpobj_fid), rc);
3176 GOTO(fixup_tpobj, rc);
3179 /* fetch updated nlink */
3180 rc = mdd_la_get(env, mdd_tobj, tattr);
3181 if (rc == -ENOENT) {
3182 /* the object got removed, let's
3183 * return the latest known attributes */
3184 tattr->la_nlink = 0;
3186 } else if (rc != 0) {
3187 CERROR("%s: Failed to get nlink for tobj "
3189 mdd2obd_dev(mdd)->obd_name,
3190 PFID(tpobj_fid), rc);
3191 GOTO(fixup_tpobj, rc);
3193 /* XXX: this transfer to ma will be removed with LOD/OSP */
3194 ma->ma_attr = *tattr;
3195 ma->ma_valid |= MA_INODE;
3197 if (tattr->la_nlink == 0)
3198 cl_flags |= CLF_RENAME_LAST;
3200 cl_flags &= ~CLF_RENAME_LAST_EXISTS;
3203 la->la_valid = LA_CTIME | LA_MTIME;
3204 rc = mdd_update_time(env, mdd_spobj, pattr, la, handle);
3206 GOTO(fixup_tpobj, rc);
3208 if (mdd_spobj != mdd_tpobj) {
3209 la->la_valid = LA_CTIME | LA_MTIME;
3210 rc = mdd_update_time(env, mdd_tpobj, tpattr, la, handle);
3212 GOTO(fixup_tpobj, rc);
3219 rc2 = __mdd_index_delete(env, mdd_tpobj, tname, is_dir, handle);
3221 CWARN("tp obj fix error %d\n",rc2);
3223 if (mdd_tobj && mdd_object_exists(mdd_tobj) &&
3224 !mdd_is_dead_obj(mdd_tobj)) {
3226 mdo_ref_add(env, mdd_tobj, handle);
3228 mdo_ref_add(env, mdd_tobj, handle);
3231 rc2 = __mdd_index_insert(env, mdd_tpobj,
3233 mdd_object_type(mdd_tobj),
3236 CWARN("tp obj fix error: rc = %d\n", rc2);
3241 if (rc && is_dir && mdd_sobj && mdd_spobj != mdd_tpobj) {
3242 rc2 = __mdd_index_delete_only(env, mdd_sobj, dotdot, handle);
3244 CWARN("%s: sp obj dotdot delete error: rc = %d\n",
3245 mdd2obd_dev(mdd)->obd_name, rc2);
3248 rc2 = __mdd_index_insert_only(env, mdd_sobj, spobj_fid, S_IFDIR,
3251 CWARN("%s: sp obj dotdot insert error: rc = %d\n",
3252 mdd2obd_dev(mdd)->obd_name, rc2);
3257 rc2 = __mdd_index_insert(env, mdd_spobj, lf,
3258 mdd_object_type(mdd_sobj), sname,
3261 CWARN("sp obj fix error: rc = %d\n", rc2);
3266 mdd_write_unlock(env, mdd_tobj);
3269 rc = mdd_changelog_ns_store(env, mdd, CL_RENAME, cl_flags,
3270 mdd_tobj, tpobj_fid, lf, spobj_fid,
3271 ltname, lsname, handle);
3274 rc = mdd_trans_stop(env, mdd, rc, handle);
3277 mdd_object_put(env, mdd_sobj);
3282 * During migration once the parent FID has been changed,
3283 * we need update the parent FID in linkea.
3285 static int mdd_linkea_update_child_internal(const struct lu_env *env,
3286 struct mdd_object *parent,
3287 struct mdd_object *newparent,
3288 struct mdd_object *child,
3289 const char *name, int namelen,
3290 struct thandle *handle,
3293 struct mdd_thread_info *info = mdd_env_info(env);
3294 struct linkea_data ldata = { NULL };
3295 struct lu_buf *buf = &info->mti_link_buf;
3301 buf = lu_buf_check_and_alloc(buf, PATH_MAX);
3302 if (buf->lb_buf == NULL)
3306 rc = mdd_links_read(env, child, &ldata);
3308 if (rc == -ENOENT || rc == -ENODATA)
3313 LASSERT(ldata.ld_leh != NULL);
3314 ldata.ld_lee = (struct link_ea_entry *)(ldata.ld_leh + 1);
3315 for (count = 0; count < ldata.ld_leh->leh_reccount; count++) {
3316 struct mdd_device *mdd = mdo2mdd(&child->mod_obj);
3317 struct lu_name lname;
3320 linkea_entry_unpack(ldata.ld_lee, &ldata.ld_reclen,
3323 if (strncmp(lname.ln_name, name, namelen) != 0 ||
3324 !lu_fid_eq(&fid, mdd_object_fid(parent))) {
3325 ldata.ld_lee = (struct link_ea_entry *)
3326 ((char *)ldata.ld_lee +
3331 CDEBUG(D_INFO, "%s: update "DFID" with %.*s:"DFID"\n",
3332 mdd2obd_dev(mdd)->obd_name, PFID(mdd_object_fid(child)),
3333 lname.ln_namelen, lname.ln_name,
3334 PFID(mdd_object_fid(newparent)));
3335 /* update to the new parent fid */
3336 linkea_entry_pack(ldata.ld_lee, &lname,
3337 mdd_object_fid(newparent));
3339 rc = mdd_declare_links_add(env, child, handle, &ldata);
3341 rc = mdd_links_write(env, child, &ldata, handle);
3347 static int mdd_linkea_declare_update_child(const struct lu_env *env,
3348 struct mdd_object *parent,
3349 struct mdd_object *newparent,
3350 struct mdd_object *child,
3351 const char *name, int namelen,
3352 struct thandle *handle)
3354 return mdd_linkea_update_child_internal(env, parent, newparent,
3356 namelen, handle, true);
3359 static int mdd_linkea_update_child(const struct lu_env *env,
3360 struct mdd_object *parent,
3361 struct mdd_object *newparent,
3362 struct mdd_object *child,
3363 const char *name, int namelen,
3364 struct thandle *handle)
3366 return mdd_linkea_update_child_internal(env, parent, newparent,
3368 namelen, handle, false);
3371 static int mdd_update_linkea_internal(const struct lu_env *env,
3372 struct mdd_object *mdd_pobj,
3373 struct mdd_object *mdd_sobj,
3374 struct mdd_object *mdd_tobj,
3375 const struct lu_name *child_name,
3376 struct linkea_data *ldata,
3377 struct thandle *handle,
3380 struct mdd_thread_info *info = mdd_env_info(env);
3385 LASSERT(ldata->ld_buf != NULL);
3386 LASSERT(ldata->ld_leh != NULL);
3388 /* If it is mulitple links file, we need update the name entry for
3390 ldata->ld_lee = (struct link_ea_entry *)(ldata->ld_leh + 1);
3391 for (count = 0; count < ldata->ld_leh->leh_reccount; count++) {
3392 struct mdd_device *mdd = mdo2mdd(&mdd_sobj->mod_obj);
3393 struct mdd_object *pobj;
3394 struct lu_name lname;
3397 linkea_entry_unpack(ldata->ld_lee, &ldata->ld_reclen,
3399 pobj = mdd_object_find(env, mdd, &fid);
3401 CWARN("%s: cannot find obj "DFID": rc = %ld\n",
3402 mdd2obd_dev(mdd)->obd_name, PFID(&fid),
3407 if (!mdd_object_exists(pobj)) {
3408 CDEBUG(D_INFO, "%s: obj "DFID" does not exist\n",
3409 mdd2obd_dev(mdd)->obd_name, PFID(&fid));
3413 if (pobj == mdd_pobj &&
3414 lname.ln_namelen == child_name->ln_namelen &&
3415 strncmp(lname.ln_name, child_name->ln_name,
3416 lname.ln_namelen) == 0) {
3417 CDEBUG(D_INFO, "%s: skip its own %s: "DFID"\n",
3418 mdd2obd_dev(mdd)->obd_name, child_name->ln_name,
3423 CDEBUG(D_INFO, "%s: update "DFID" with "DNAME":"DFID"\n",
3424 mdd2obd_dev(mdd)->obd_name, PFID(mdd_object_fid(pobj)),
3425 PNAME(&lname), PFID(mdd_object_fid(mdd_tobj)));
3428 /* Remove source name from source directory */
3429 /* Insert new fid with target name into target dir */
3430 rc = mdo_declare_index_delete(env, pobj, lname.ln_name,
3435 rc = mdo_declare_index_insert(env, pobj,
3436 mdd_object_fid(mdd_tobj),
3437 mdd_object_type(mdd_tobj),
3438 lname.ln_name, handle);
3442 rc = mdo_declare_ref_add(env, mdd_tobj, handle);
3446 rc = mdo_declare_ref_del(env, mdd_sobj, handle);
3450 char *tmp_name = info->mti_key;
3452 if (lname.ln_namelen >= sizeof(info->mti_key)) {
3453 /* lnamelen is too big(> NAME_MAX + 16),
3454 * something wrong about this linkea, let's
3456 CWARN("%s: the name %.*s is too long under "
3457 DFID"\n", mdd2obd_dev(mdd)->obd_name,
3458 lname.ln_namelen, lname.ln_name,
3463 /* Note: lname might be without \0 at the end, see
3464 * linkea_entry_unpack(), let's add extra \0 by
3466 snprintf(tmp_name, sizeof(info->mti_key), "%.*s",
3467 lname.ln_namelen, lname.ln_name);
3468 lname.ln_name = tmp_name;
3470 /* Let's check if this linkEA still valid, before
3471 * it might be packed into the RPC buffer. */
3472 rc = mdd_lookup(env, &pobj->mod_obj, &lname,
3473 &info->mti_fid, NULL);
3474 if (rc < 0 || !lu_fid_eq(&info->mti_fid,
3475 mdd_object_fid(mdd_sobj)))
3476 GOTO(next_put, rc == -ENOENT ? 0 : rc);
3478 rc = __mdd_index_delete(env, pobj, tmp_name, 0, handle);
3482 rc = __mdd_index_insert(env, pobj,
3483 mdd_object_fid(mdd_tobj),
3484 mdd_object_type(mdd_tobj),
3489 mdd_write_lock(env, mdd_tobj, MOR_SRC_CHILD);
3490 rc = mdo_ref_add(env, mdd_tobj, handle);
3491 mdd_write_unlock(env, mdd_tobj);
3495 mdd_write_lock(env, mdd_sobj, MOR_TGT_CHILD);
3496 mdo_ref_del(env, mdd_sobj, handle);
3497 mdd_write_unlock(env, mdd_sobj);
3500 mdd_object_put(env, pobj);
3504 ldata->ld_lee = (struct link_ea_entry *)((char *)ldata->ld_lee +
3511 static int mdd_migrate_xattrs(const struct lu_env *env,
3512 struct mdd_object *mdd_sobj,
3513 struct mdd_object *mdd_tobj)
3515 struct mdd_thread_info *info = mdd_env_info(env);
3516 struct mdd_device *mdd = mdo2mdd(&mdd_sobj->mod_obj);
3518 struct thandle *handle;
3524 struct lu_buf list_xbuf;
3527 /* retrieve xattr list from the old object */
3528 list_xsize = mdo_xattr_list(env, mdd_sobj, &LU_BUF_NULL);
3529 if (list_xsize == -ENODATA)
3535 lu_buf_check_and_alloc(&info->mti_big_buf, list_xsize);
3536 if (info->mti_big_buf.lb_buf == NULL)
3539 list_xbuf.lb_buf = info->mti_big_buf.lb_buf;
3540 list_xbuf.lb_len = list_xsize;
3541 rc = mdo_xattr_list(env, mdd_sobj, &list_xbuf);
3546 xname = list_xbuf.lb_buf;
3548 xlen = strnlen(xname, rem - 1) + 1;
3549 if (strcmp(XATTR_NAME_LMA, xname) == 0 ||
3550 strcmp(XATTR_NAME_LMV, xname) == 0)
3553 /* For directory, if there are default layout, migrate here */
3554 if (strcmp(XATTR_NAME_LOV, xname) == 0 &&
3555 !S_ISDIR(lu_object_attr(&mdd_sobj->mod_obj.mo_lu)))
3558 xsize = mdo_xattr_get(env, mdd_sobj, &LU_BUF_NULL, xname);
3559 if (xsize == -ENODATA)
3564 lu_buf_check_and_alloc(&info->mti_link_buf, xsize);
3565 if (info->mti_link_buf.lb_buf == NULL)
3566 GOTO(out, rc = -ENOMEM);
3568 xbuf.lb_len = xsize;
3569 xbuf.lb_buf = info->mti_link_buf.lb_buf;
3570 rc = mdo_xattr_get(env, mdd_sobj, &xbuf, xname);
3576 handle = mdd_trans_create(env, mdd);
3578 GOTO(out, rc = PTR_ERR(handle));
3580 rc = mdo_declare_xattr_set(env, mdd_tobj, &xbuf, xname, 0,
3583 GOTO(stop_trans, rc);
3584 /* Note: this transaction is part of migration, and it is not
3585 * the last step of migration, so we set th_local = 1 to avoid
3586 * update last rcvd for this transaction */
3587 handle->th_local = 1;
3588 rc = mdd_trans_start(env, mdd, handle);
3590 GOTO(stop_trans, rc);
3593 rc = mdo_xattr_set(env, mdd_tobj, &xbuf, xname, 0, handle);
3595 GOTO(stop_trans, rc = 0);
3597 if (unlikely(rc == -ENOSPC &&
3598 strcmp(xname, XATTR_NAME_LINK) == 0)) {
3599 rc = linkea_overflow_shrink(
3600 (struct linkea_data *)(xbuf.lb_buf));
3601 if (likely(rc > 0)) {
3608 GOTO(stop_trans, rc);
3610 rc = mdd_trans_stop(env, mdd, rc, handle);
3615 memmove(xname, xname + xlen, rem);
3621 static int mdd_declare_migrate_create(const struct lu_env *env,
3622 struct mdd_object *mdd_pobj,
3623 struct mdd_object *mdd_sobj,
3624 struct mdd_object *mdd_tobj,
3625 struct md_op_spec *spec,
3627 union lmv_mds_md *mgr_ea,
3628 struct linkea_data *ldata,
3629 struct thandle *handle)
3631 struct lu_attr *la_flag = MDD_ENV_VAR(env, la_for_fix);
3632 const struct lu_buf *buf;
3636 rc = mdd_declare_create_object_internal(env, mdd_pobj, mdd_tobj, la,
3637 handle, spec, NULL);
3641 rc = mdd_declare_object_initialize(env, mdd_pobj, mdd_tobj, la,
3646 if (S_ISLNK(la->la_mode)) {
3647 const char *target_name = spec->u.sp_symname;
3648 int sym_len = strlen(target_name);
3649 const struct lu_buf *buf;
3651 buf = mdd_buf_get_const(env, target_name, sym_len);
3652 rc = dt_declare_record_write(env, mdd_object_child(mdd_tobj),
3656 } else if (S_ISDIR(la->la_mode) && ldata != NULL) {
3657 rc = mdd_declare_links_add(env, mdd_tobj, handle, ldata);
3662 if (spec->u.sp_ea.eadata != NULL && spec->u.sp_ea.eadatalen != 0) {
3663 buf = mdd_buf_get_const(env, spec->u.sp_ea.eadata,
3664 spec->u.sp_ea.eadatalen);
3665 rc = mdo_declare_xattr_set(env, mdd_tobj, buf, XATTR_NAME_LOV,
3671 mgr_easize = lmv_mds_md_size(2, LMV_MAGIC_V1);
3672 buf = mdd_buf_get_const(env, mgr_ea, mgr_easize);
3673 rc = mdo_declare_xattr_set(env, mdd_sobj, buf, XATTR_NAME_LMV,
3678 la_flag->la_valid = LA_FLAGS;
3679 la_flag->la_flags = la->la_flags | LUSTRE_IMMUTABLE_FL;
3680 rc = mdo_declare_attr_set(env, mdd_sobj, la_flag, handle);
3685 static int mdd_migrate_create(const struct lu_env *env,
3686 struct mdd_object *mdd_pobj,
3687 struct mdd_object *mdd_sobj,
3688 struct mdd_object *mdd_tobj,
3689 const struct lu_name *lname,
3692 struct mdd_thread_info *info = mdd_env_info(env);
3693 struct mdd_device *mdd = mdo2mdd(&mdd_sobj->mod_obj);
3694 struct md_op_spec *spec = &info->mti_spec;
3695 struct lu_buf lmm_buf = { NULL };
3696 struct lu_buf link_buf = { NULL };
3697 struct lu_buf mgr_buf;
3698 struct thandle *handle;
3699 struct lmv_mds_md_v1 *mgr_ea;
3700 struct lu_attr *la_flag = MDD_ENV_VAR(env, la_for_fix);
3701 struct dt_allocation_hint *hint = &mdd_env_info(env)->mti_hint;
3703 struct linkea_data *ldata = &mdd_env_info(env)->mti_link_data;
3707 /* prepare spec for create */
3708 memset(spec, 0, sizeof(*spec));
3709 spec->sp_cr_lookup = 0;
3710 spec->sp_feat = &dt_directory_features;
3711 if (S_ISLNK(la->la_mode)) {
3712 const struct lu_buf *buf;
3714 buf = lu_buf_check_and_alloc(
3715 &mdd_env_info(env)->mti_big_buf,
3718 link_buf.lb_len = la->la_size + 1;
3719 memset(link_buf.lb_buf, 0, link_buf.lb_len);
3720 rc = mdd_readlink(env, &mdd_sobj->mod_obj, &link_buf);
3722 rc = rc != 0 ? rc : -EFAULT;
3723 CERROR("%s: "DFID" readlink failed: rc = %d\n",
3724 mdd2obd_dev(mdd)->obd_name,
3725 PFID(mdd_object_fid(mdd_sobj)), rc);
3728 spec->u.sp_symname = link_buf.lb_buf;
3729 } else if (S_ISREG(la->la_mode)) {
3730 /* retrieve lov of the old object */
3731 rc = mdd_get_lov_ea(env, mdd_sobj, &lmm_buf);
3732 if (rc != 0 && rc != -ENODATA)
3734 if (lmm_buf.lb_buf != NULL && lmm_buf.lb_len != 0) {
3735 spec->u.sp_ea.eadata = lmm_buf.lb_buf;
3736 spec->u.sp_ea.eadatalen = lmm_buf.lb_len;
3737 spec->sp_cr_flags |= MDS_OPEN_HAS_EA;
3739 } else if (S_ISDIR(la->la_mode)) {
3740 rc = mdd_links_read_with_rec(env, mdd_sobj, ldata);
3741 if (rc == -ENODATA) {
3742 /* ignore the non-linkEA error */
3750 mgr_easize = lmv_mds_md_size(2, LMV_MAGIC_V1);
3751 lu_buf_check_and_alloc(&info->mti_xattr_buf, mgr_easize);
3752 mgr_buf.lb_buf = info->mti_xattr_buf.lb_buf;
3753 mgr_buf.lb_len = mgr_easize;
3754 mgr_ea = mgr_buf.lb_buf;
3755 memset(mgr_ea, 0, sizeof(*mgr_ea));
3756 mgr_ea->lmv_magic = cpu_to_le32(LMV_MAGIC_V1);
3757 mgr_ea->lmv_stripe_count = cpu_to_le32(2);
3758 mgr_ea->lmv_master_mdt_index = mdd_seq_site(mdd)->ss_node_id;
3759 mgr_ea->lmv_hash_type = cpu_to_le32(LMV_HASH_FLAG_MIGRATION);
3760 fid_cpu_to_le(&mgr_ea->lmv_stripe_fids[0], mdd_object_fid(mdd_sobj));
3761 fid_cpu_to_le(&mgr_ea->lmv_stripe_fids[1], mdd_object_fid(mdd_tobj));
3763 mdd_object_make_hint(env, mdd_pobj, mdd_tobj, la, spec, hint);
3765 handle = mdd_trans_create(env, mdd);
3767 GOTO(out_free, rc = PTR_ERR(handle));
3769 /* Note: this transaction is part of migration, and it is not
3770 * the last step of migration, so we set th_local = 1 to avoid
3771 * update last rcvd for this transaction */
3772 handle->th_local = 1;
3773 rc = mdd_declare_migrate_create(env, mdd_pobj, mdd_sobj, mdd_tobj, spec,
3774 la, mgr_buf.lb_buf, ldata, handle);
3776 GOTO(stop_trans, rc);
3778 rc = mdd_trans_start(env, mdd, handle);
3780 GOTO(stop_trans, rc);
3782 /* don't set nlink from the original object */
3783 la->la_valid &= ~LA_NLINK;
3785 /* create the target object */
3786 rc = mdd_create_object(env, mdd_pobj, mdd_tobj, la, spec, NULL, NULL,
3789 GOTO(stop_trans, rc);
3791 if (S_ISDIR(la->la_mode) && ldata != NULL) {
3792 rc = mdd_links_write(env, mdd_tobj, ldata, handle);
3794 GOTO(stop_trans, rc);
3797 /* Set MIGRATE EA on the source inode, so once the migration needs
3798 * to be re-done during failover, the re-do process can locate the
3799 * target object which is already being created. */
3800 rc = mdo_xattr_set(env, mdd_sobj, &mgr_buf, XATTR_NAME_LMV, 0, handle);
3802 GOTO(stop_trans, rc);
3804 /* Set immutable flag, so any modification is disabled until
3805 * the migration is done. Once the migration is interrupted,
3806 * if the resume process find the migrating object has both
3807 * IMMUTALBE flag and MIGRATE EA, it need to clear IMMUTABLE
3808 * flag and approve the migration */
3809 la_flag->la_valid = LA_FLAGS;
3810 la_flag->la_flags = la->la_flags | LUSTRE_IMMUTABLE_FL;
3811 rc = mdo_attr_set(env, mdd_sobj, la_flag, handle);
3814 rc = mdd_trans_stop(env, mdd, rc, handle);
3816 if (lmm_buf.lb_buf != NULL)
3817 OBD_FREE(lmm_buf.lb_buf, lmm_buf.lb_len);
3821 static int mdd_migrate_entries(const struct lu_env *env,
3822 struct mdd_object *mdd_sobj,
3823 struct mdd_object *mdd_tobj)
3825 struct dt_object *next = mdd_object_child(mdd_sobj);
3826 struct mdd_device *mdd = mdo2mdd(&mdd_sobj->mod_obj);
3827 struct dt_object *dt_tobj = mdd_object_child(mdd_tobj);
3828 struct thandle *handle;
3830 const struct dt_it_ops *iops;
3832 struct lu_dirent *ent;
3836 OBD_ALLOC(ent, NAME_MAX + sizeof(*ent) + 1);
3840 if (!dt_try_as_dir(env, next))
3841 GOTO(out_ent, rc = -ENOTDIR);
3843 * iterate directories
3845 iops = &next->do_index_ops->dio_it;
3846 it = iops->init(env, next, LUDA_FID | LUDA_TYPE);
3848 GOTO(out_ent, rc = PTR_ERR(it));
3850 rc = iops->load(env, it, 0);
3852 rc = iops->next(env, it);
3856 * At this point and across for-loop:
3858 * rc == 0 -> ok, proceed.
3859 * rc > 0 -> end of directory.
3863 struct mdd_object *child;
3864 char *name = mdd_env_info(env)->mti_key;
3867 bool target_exist = false;
3869 len = iops->key_size(env, it);
3873 result = iops->rec(env, it, (struct dt_rec *)ent,
3874 LUDA_FID | LUDA_TYPE);
3875 if (result == -ESTALE)
3882 fid_le_to_cpu(&ent->lde_fid, &ent->lde_fid);
3884 /* Insert new fid with target name into target dir */
3885 if ((ent->lde_namelen == 1 && ent->lde_name[0] == '.') ||
3886 (ent->lde_namelen == 2 && ent->lde_name[0] == '.' &&
3887 ent->lde_name[1] == '.'))
3890 child = mdd_object_find(env, mdd, &ent->lde_fid);
3892 GOTO(out, rc = PTR_ERR(child));
3894 /* child may not exist, but lu_object_attr will assert this,
3895 * get type from loh_attr directly */
3896 is_dir = S_ISDIR(child->mod_obj.mo_lu.lo_header->loh_attr);
3898 mdd_write_lock(env, child, MOR_SRC_CHILD);
3900 snprintf(name, ent->lde_namelen + 1, "%s", ent->lde_name);
3902 /* Check whether the name has been inserted to the target */
3903 if (dt_try_as_dir(env, dt_tobj)) {
3904 struct lu_fid *fid = &mdd_env_info(env)->mti_fid2;
3906 rc = dt_lookup(env, dt_tobj, (struct dt_rec *)fid,
3907 (struct dt_key *)name);
3908 if (unlikely(rc == 0))
3909 target_exist = true;
3912 handle = mdd_trans_create(env, mdd);
3914 GOTO(out_put, rc = PTR_ERR(handle));
3916 /* Note: this transaction is part of migration, and it is not
3917 * the last step of migration, so we set th_local = 1 to avoid
3918 * updating last rcvd for this transaction */
3919 handle->th_local = 1;
3920 if (likely(!target_exist)) {
3921 rc = mdo_declare_index_insert(env, mdd_tobj,
3923 child->mod_obj.mo_lu.lo_header->loh_attr,
3929 rc = mdo_declare_ref_add(env, mdd_tobj, handle);
3935 rc = mdo_declare_index_delete(env, mdd_sobj, name, handle);
3940 rc = mdo_declare_ref_del(env, mdd_sobj, handle);
3944 /* Update .. for child */
3945 rc = mdo_declare_index_delete(env, child, dotdot,
3950 rc = mdo_declare_index_insert(env, child,
3951 mdd_object_fid(mdd_tobj),
3952 S_IFDIR, dotdot, handle);
3957 rc = mdd_linkea_declare_update_child(env, mdd_sobj,mdd_tobj,
3964 rc = mdd_trans_start(env, mdd, handle);
3966 CERROR("%s: transaction start failed: rc = %d\n",
3967 mdd2obd_dev(mdd)->obd_name, rc);
3971 if (likely(!target_exist)) {
3972 rc = __mdd_index_insert(env, mdd_tobj, &ent->lde_fid,
3973 child->mod_obj.mo_lu.lo_header->loh_attr, name,
3979 rc = __mdd_index_delete(env, mdd_sobj, name, is_dir, handle);
3984 rc = __mdd_index_delete_only(env, child, dotdot,
3989 rc = __mdd_index_insert_only(env, child,
3990 mdd_object_fid(mdd_tobj), S_IFDIR,
3996 rc = mdd_linkea_update_child(env, mdd_sobj, mdd_tobj,
3998 strlen(name), handle);
4001 mdd_write_unlock(env, child);
4002 mdd_object_put(env, child);
4003 rc = mdd_trans_stop(env, mdd, rc, handle);
4007 result = iops->next(env, it);
4008 if (OBD_FAIL_CHECK(OBD_FAIL_MIGRATE_ENTRIES))
4009 GOTO(out, rc = -EINTR);
4011 if (result == -ESTALE)
4013 } while (result == 0);
4016 iops->fini(env, it);
4018 OBD_FREE(ent, NAME_MAX + sizeof(*ent) + 1);
4022 static int mdd_declare_update_linkea(const struct lu_env *env,
4023 struct mdd_object *mdd_pobj,
4024 struct mdd_object *mdd_sobj,
4025 struct mdd_object *mdd_tobj,
4026 const struct lu_name *child_name,
4027 struct linkea_data *ldata,
4028 struct thandle *handle)
4030 return mdd_update_linkea_internal(env, mdd_pobj, mdd_sobj, mdd_tobj,
4031 child_name, ldata, handle, 1);
4034 static int mdd_update_linkea(const struct lu_env *env,
4035 struct mdd_object *mdd_pobj,
4036 struct mdd_object *mdd_sobj,
4037 struct mdd_object *mdd_tobj,
4038 const struct lu_name *child_name,
4039 struct linkea_data *ldata,
4040 struct thandle *handle)
4042 return mdd_update_linkea_internal(env, mdd_pobj, mdd_sobj, mdd_tobj,
4043 child_name, ldata, handle, 0);
4046 static int mdd_declare_migrate_update_name(const struct lu_env *env,
4047 struct mdd_object *mdd_pobj,
4048 struct mdd_object *mdd_sobj,
4049 struct mdd_object *mdd_tobj,
4050 const struct lu_name *lname,
4052 struct lu_attr *parent_la,
4053 struct linkea_data *ldata,
4054 struct thandle *handle)
4056 struct mdd_device *mdd = mdo2mdd(&mdd_sobj->mod_obj);
4057 struct lu_attr *la_flag = MDD_ENV_VAR(env, tattr);
4060 /* Revert IMMUTABLE flag */
4061 la_flag->la_valid = LA_FLAGS;
4062 la_flag->la_flags = la->la_flags & ~LUSTRE_IMMUTABLE_FL;
4063 rc = mdo_declare_attr_set(env, mdd_sobj, la_flag, handle);
4067 /* delete entry from source dir */
4068 rc = mdo_declare_index_delete(env, mdd_pobj, lname->ln_name, handle);
4072 if (ldata->ld_buf != NULL) {
4073 rc = mdd_declare_update_linkea(env, mdd_pobj, mdd_sobj,
4074 mdd_tobj, lname, ldata, handle);
4079 if (S_ISREG(mdd_object_type(mdd_sobj))) {
4080 rc = mdo_declare_xattr_del(env, mdd_sobj, XATTR_NAME_LOV,
4085 handle->th_complex = 1;
4086 rc = mdo_declare_xattr_set(env, mdd_tobj, NULL,
4088 LU_XATTR_REPLACE, handle);
4093 if (S_ISDIR(mdd_object_type(mdd_sobj))) {
4094 rc = mdo_declare_ref_del(env, mdd_pobj, handle);
4100 rc = mdo_declare_index_insert(env, mdd_pobj, mdo2fid(mdd_tobj),
4101 mdd_object_type(mdd_tobj),
4102 lname->ln_name, handle);
4106 rc = mdd_declare_links_add(env, mdd_tobj, handle, NULL);
4110 if (S_ISDIR(mdd_object_type(mdd_sobj))) {
4111 rc = mdo_declare_ref_add(env, mdd_pobj, handle);
4116 /* delete old object */
4117 rc = mdo_declare_ref_del(env, mdd_sobj, handle);
4121 if (S_ISDIR(mdd_object_type(mdd_sobj))) {
4122 /* delete old object */
4123 rc = mdo_declare_ref_del(env, mdd_sobj, handle);
4126 /* set nlink to 0 */
4127 rc = mdo_declare_attr_set(env, mdd_sobj, la, handle);
4132 rc = mdd_declare_finish_unlink(env, mdd_sobj, handle);
4136 rc = mdo_declare_attr_set(env, mdd_pobj, parent_la, handle);
4140 rc = mdd_declare_changelog_store(env, mdd, lname, NULL, handle);
4145 static int mdd_migrate_update_name(const struct lu_env *env,
4146 struct mdd_object *mdd_pobj,
4147 struct mdd_object *mdd_sobj,
4148 struct mdd_object *mdd_tobj,
4149 const struct lu_name *lname,
4152 struct lu_attr *p_la = MDD_ENV_VAR(env, la_for_fix);
4153 struct lu_attr *so_attr = MDD_ENV_VAR(env, cattr);
4154 struct lu_attr *la_flag = MDD_ENV_VAR(env, tattr);
4155 struct mdd_device *mdd = mdo2mdd(&mdd_sobj->mod_obj);
4156 struct linkea_data *ldata = &mdd_env_info(env)->mti_link_data;
4157 struct thandle *handle;
4158 int is_dir = S_ISDIR(mdd_object_type(mdd_sobj));
4159 const char *name = lname->ln_name;
4163 /* update time for parent */
4164 LASSERT(ma->ma_attr.la_valid & LA_CTIME);
4165 p_la->la_ctime = p_la->la_mtime = ma->ma_attr.la_ctime;
4166 p_la->la_valid = LA_CTIME;
4168 rc = mdd_la_get(env, mdd_sobj, so_attr);
4172 ldata->ld_buf = NULL;
4173 rc = mdd_links_read(env, mdd_sobj, ldata);
4174 if (rc != 0 && rc != -ENOENT && rc != -ENODATA)
4177 handle = mdd_trans_create(env, mdd);
4179 RETURN(PTR_ERR(handle));
4181 rc = mdd_declare_migrate_update_name(env, mdd_pobj, mdd_sobj, mdd_tobj,
4182 lname, so_attr, p_la, ldata,
4185 /* If the migration can not be fit in one transaction, just
4186 * leave it in the original MDT */
4188 GOTO(stop_trans, rc = 0);
4190 GOTO(stop_trans, rc);
4193 CDEBUG(D_INFO, "%s: update "DFID"/"DFID" with %s:"DFID"\n",
4194 mdd2obd_dev(mdd)->obd_name, PFID(mdd_object_fid(mdd_pobj)),
4195 PFID(mdd_object_fid(mdd_sobj)), lname->ln_name,
4196 PFID(mdd_object_fid(mdd_tobj)));
4198 rc = mdd_trans_start(env, mdd, handle);
4200 GOTO(stop_trans, rc);
4202 /* Revert IMMUTABLE flag */
4203 la_flag->la_valid = LA_FLAGS;
4204 la_flag->la_flags = so_attr->la_flags & ~LUSTRE_IMMUTABLE_FL;
4205 rc = mdo_attr_set(env, mdd_sobj, la_flag, handle);
4207 GOTO(stop_trans, rc);
4209 /* Remove source name from source directory */
4210 rc = __mdd_index_delete(env, mdd_pobj, name, is_dir, handle);
4212 GOTO(stop_trans, rc);
4214 if (ldata->ld_buf != NULL) {
4215 rc = mdd_update_linkea(env, mdd_pobj, mdd_sobj, mdd_tobj,
4216 lname, ldata, handle);
4218 GOTO(stop_trans, rc);
4220 /* linkea update might decrease the source object
4221 * nlink, let's get the attr again after ref_del */
4222 rc = mdd_la_get(env, mdd_sobj, so_attr);
4224 GOTO(stop_trans, rc);
4227 if (S_ISREG(so_attr->la_mode)) {
4228 if (so_attr->la_nlink == 1) {
4229 rc = mdo_xattr_del(env, mdd_sobj, XATTR_NAME_LOV,
4231 if (rc != 0 && rc != -ENODATA)
4232 GOTO(stop_trans, rc);
4234 rc = mdo_xattr_set(env, mdd_tobj, NULL,
4236 LU_XATTR_REPLACE, handle);
4238 GOTO(stop_trans, rc);
4242 /* Insert new fid with target name into target dir */
4243 rc = __mdd_index_insert(env, mdd_pobj, mdd_object_fid(mdd_tobj),
4244 mdd_object_type(mdd_tobj), name, handle);
4246 GOTO(stop_trans, rc);
4248 mdd_write_lock(env, mdd_sobj, MOR_TGT_CHILD);
4250 mdd_sobj->mod_flags |= DEAD_OBJ;
4251 rc = mdd_mark_orphan_object(env, mdd_sobj, handle, false);
4253 GOTO(out_unlock, rc);
4255 rc = __mdd_orphan_add(env, mdd_sobj, handle);
4257 GOTO(out_unlock, rc);
4259 mdo_ref_del(env, mdd_sobj, handle);
4261 mdo_ref_del(env, mdd_sobj, handle);
4263 /* Get the attr again after ref_del */
4264 rc = mdd_la_get(env, mdd_sobj, so_attr);
4266 GOTO(out_unlock, rc);
4268 ma->ma_attr = *so_attr;
4269 ma->ma_valid |= MA_INODE;
4271 rc = mdd_attr_set_internal(env, mdd_pobj, p_la, handle, 0);
4273 GOTO(out_unlock, rc);
4275 rc = mdd_changelog_ns_store(env, mdd, CL_MIGRATE, 0, mdd_tobj,
4276 mdo2fid(mdd_pobj), mdo2fid(mdd_sobj),
4277 mdo2fid(mdd_pobj), lname, lname, handle);
4279 CWARN("%s: changelog for migrate %s "DFID
4280 "under "DFID" failed: rc = %d\n",
4281 mdd2obd_dev(mdd)->obd_name, lname->ln_name,
4282 PFID(mdd_object_fid(mdd_sobj)),
4283 PFID(mdd_object_fid(mdd_pobj)), rc);
4284 /* Sigh, there are no easy way to migrate back the object, so
4285 * let's reset the result to 0 for now XXX */
4289 mdd_write_unlock(env, mdd_sobj);
4292 rc = mdd_trans_stop(env, mdd, rc, handle);
4297 static int mdd_fld_lookup(const struct lu_env *env, struct mdd_device *mdd,
4298 const struct lu_fid *fid, __u32 *mdt_index)
4300 struct lu_seq_range *range = &mdd_env_info(env)->mti_range;
4301 struct seq_server_site *ss;
4304 ss = mdd->mdd_md_dev.md_lu_dev.ld_site->ld_seq_site;
4306 range->lsr_flags = LU_SEQ_RANGE_MDT;
4307 rc = fld_server_lookup(env, ss->ss_server_fld, fid->f_seq, range);
4311 *mdt_index = range->lsr_index;
4316 * Check whether we should migrate the file/dir
4318 * < 0 permission check failed or other error.
4319 * = 0 the file can be migrated.
4320 * > 0 the file does not need to be migrated, mostly
4321 * for multiple link file
4323 static int mdd_migrate_sanity_check(const struct lu_env *env,
4324 struct mdd_object *pobj,
4325 const struct lu_attr *pattr,
4326 struct mdd_object *sobj,
4327 struct lu_attr *sattr)
4329 struct mdd_thread_info *info = mdd_env_info(env);
4330 struct linkea_data *ldata = &info->mti_link_data;
4331 struct mdd_device *mdd = mdo2mdd(&pobj->mod_obj);
4333 struct lu_buf *mgr_buf;
4339 mgr_easize = lmv_mds_md_size(2, LMV_MAGIC_V1);
4340 mgr_buf = lu_buf_check_and_alloc(&info->mti_big_buf, mgr_easize);
4341 if (mgr_buf->lb_buf == NULL)
4344 rc = mdo_xattr_get(env, sobj, mgr_buf, XATTR_NAME_LMV);
4346 union lmv_mds_md *lmm = mgr_buf->lb_buf;
4348 /* If the object has migrateEA, it means IMMUTE flag
4349 * is being set by previous migration process, so it
4350 * needs to override the IMMUTE flag, otherwise the
4351 * following sanity check will fail */
4352 if (le32_to_cpu(lmm->lmv_md_v1.lmv_hash_type) &
4353 LMV_HASH_FLAG_MIGRATION) {
4354 struct mdd_device *mdd = mdo2mdd(&sobj->mod_obj);
4356 sattr->la_flags &= ~LUSTRE_IMMUTABLE_FL;
4357 CDEBUG(D_HA, "%s: "DFID" override IMMUTE FLAG\n",
4358 mdd2obd_dev(mdd)->obd_name,
4359 PFID(mdd_object_fid(sobj)));
4363 rc = mdd_rename_sanity_check(env, pobj, pattr, pobj, pattr,
4364 sobj, sattr, NULL, NULL);
4368 /* Then it will check if the file should be migrated. If the file
4369 * has mulitple links, we only need migrate the file if all of its
4370 * entries has been migrated to the remote MDT */
4371 if (!S_ISREG(sattr->la_mode) || sattr->la_nlink < 2)
4374 rc = mdd_links_read(env, sobj, ldata);
4376 /* For multiple links files, if there are no linkEA data at all,
4377 * means the file might be created before linkEA is enabled, and
4378 * all of its links should not be migrated yet, otherwise it
4379 * should have some linkEA there */
4380 if (rc == -ENOENT || rc == -ENODATA)
4385 mdt_index = mdd->mdd_md_dev.md_lu_dev.ld_site->ld_seq_site->ss_node_id;
4386 /* If there are still links locally, then the file will not be
4388 LASSERT(ldata->ld_leh != NULL);
4390 /* If the linkEA is overflow, then means there are some unknown name
4391 * entries under unknown parents, that will prevent the migration. */
4392 if (unlikely(ldata->ld_leh->leh_overflow_time))
4395 ldata->ld_lee = (struct link_ea_entry *)(ldata->ld_leh + 1);
4396 for (count = 0; count < ldata->ld_leh->leh_reccount; count++) {
4397 struct lu_name lname;
4399 __u32 parent_mdt_index;
4401 linkea_entry_unpack(ldata->ld_lee, &ldata->ld_reclen,
4403 ldata->ld_lee = (struct link_ea_entry *)((char *)ldata->ld_lee +
4406 rc = mdd_fld_lookup(env, mdd, &fid, &parent_mdt_index);
4410 /* Migrate the object only if none of its parents are on the
4412 if (parent_mdt_index != mdt_index)
4415 CDEBUG(D_INFO, DFID"still has local entry %.*s "DFID"\n",
4416 PFID(mdd_object_fid(sobj)), lname.ln_namelen,
4417 lname.ln_name, PFID(&fid));
4425 static int mdd_migrate(const struct lu_env *env, struct md_object *pobj,
4426 struct md_object *sobj, const struct lu_name *lname,
4427 struct md_object *tobj, struct md_attr *ma)
4429 struct mdd_object *mdd_pobj = md2mdd_obj(pobj);
4430 struct mdd_device *mdd = mdo2mdd(pobj);
4431 struct mdd_object *mdd_sobj = md2mdd_obj(sobj);
4432 struct mdd_object *mdd_tobj = md2mdd_obj(tobj);
4433 struct lu_attr *so_attr = MDD_ENV_VAR(env, cattr);
4434 struct lu_attr *pattr = MDD_ENV_VAR(env, pattr);
4435 bool created = false;
4439 /* If the file will being migrated, it will check whether
4440 * the file is being opened by someone else right now */
4441 mdd_read_lock(env, mdd_sobj, MOR_SRC_CHILD);
4442 if (mdd_sobj->mod_count > 0) {
4444 "%s: "DFID"%s is already opened count %d: rc = %d\n",
4445 mdd2obd_dev(mdd)->obd_name,
4446 PFID(mdd_object_fid(mdd_sobj)), lname->ln_name,
4447 mdd_sobj->mod_count, -EBUSY);
4448 mdd_read_unlock(env, mdd_sobj);
4449 GOTO(put, rc = -EBUSY);
4451 mdd_read_unlock(env, mdd_sobj);
4453 rc = mdd_la_get(env, mdd_sobj, so_attr);
4457 rc = mdd_la_get(env, mdd_pobj, pattr);
4461 rc = mdd_migrate_sanity_check(env, mdd_pobj, pattr, mdd_sobj, so_attr);
4468 /* Sigh, it is impossible to finish all of migration in a single
4469 * transaction, for example migrating big directory entries to the
4470 * new MDT, it needs insert all of name entries of children in the
4473 * So migration will be done in multiple steps and transactions.
4475 * 1. create an orphan object on the remote MDT in one transaction.
4476 * 2. migrate extend attributes to the new target file/directory.
4477 * 3. For directory, migrate the entries to the new MDT and update
4478 * linkEA of each children. Because we can not migrate all entries
4479 * in a single transaction, so the migrating directory will become
4480 * a striped directory during migration, so once the process is
4481 * interrupted, the directory is still accessible. (During lookup,
4482 * client will locate the name by searching both original and target
4484 * 4. Finally, update the name/FID to point to the new file/directory
4485 * in a separate transaction.
4488 /* step 1: Check whether the orphan object has been created, and create
4489 * orphan object on the remote MDT if needed */
4490 if (!mdd_object_exists(mdd_tobj)) {
4491 rc = mdd_migrate_create(env, mdd_pobj, mdd_sobj, mdd_tobj,
4498 LASSERT(mdd_object_exists(mdd_tobj));
4499 /* step 2: migrate xattr */
4500 rc = mdd_migrate_xattrs(env, mdd_sobj, mdd_tobj);
4504 /* step 3: migrate name entries to the orphan object */
4505 if (S_ISDIR(lu_object_attr(&mdd_sobj->mod_obj.mo_lu))) {
4506 rc = mdd_migrate_entries(env, mdd_sobj, mdd_tobj);
4509 if (unlikely(OBD_FAIL_CHECK_RESET(OBD_FAIL_MIGRATE_NET_REP,
4510 OBD_FAIL_MDS_REINT_NET_REP)))
4513 OBD_FAIL_TIMEOUT(OBD_FAIL_MIGRATE_DELAY, cfs_fail_val);
4516 LASSERT(mdd_object_exists(mdd_tobj));
4517 /* step 4: update name entry to the new object */
4518 rc = mdd_migrate_update_name(env, mdd_pobj, mdd_sobj, mdd_tobj, lname,
4523 /* newly created target was not locked, don't cache its attributes */
4525 mdd_invalidate(env, tobj);
4530 const struct md_dir_operations mdd_dir_ops = {
4531 .mdo_is_subdir = mdd_is_subdir,
4532 .mdo_lookup = mdd_lookup,
4533 .mdo_create = mdd_create,
4534 .mdo_rename = mdd_rename,
4535 .mdo_link = mdd_link,
4536 .mdo_unlink = mdd_unlink,
4537 .mdo_create_data = mdd_create_data,
4538 .mdo_migrate = mdd_migrate,