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, 2016, 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 <obd_class.h>
42 #include <obd_support.h>
43 #include <lustre_mds.h>
44 #include <lustre_fid.h>
46 #include "mdd_internal.h"
48 static const char dot[] = ".";
49 static const char dotdot[] = "..";
51 static struct lu_name lname_dotdot = {
52 .ln_name = (char *) dotdot,
53 .ln_namelen = sizeof(dotdot) - 1,
57 mdd_name_check(struct mdd_device *m, const struct lu_name *ln)
59 if (!lu_name_is_valid(ln))
61 else if (ln->ln_namelen > m->mdd_dt_conf.ddp_max_name_len)
67 /* Get FID from name and parent */
69 __mdd_lookup(const struct lu_env *env, struct md_object *pobj,
70 const struct lu_attr *pattr, const struct lu_name *lname,
71 struct lu_fid* fid, int mask)
73 const char *name = lname->ln_name;
74 const struct dt_key *key = (const struct dt_key *)name;
75 struct mdd_object *mdd_obj = md2mdd_obj(pobj);
76 struct mdd_device *m = mdo2mdd(pobj);
77 struct dt_object *dir = mdd_object_child(mdd_obj);
81 if (unlikely(mdd_is_dead_obj(mdd_obj)))
84 if (!mdd_object_exists(mdd_obj))
87 if (mdd_object_remote(mdd_obj)) {
88 CDEBUG(D_INFO, "%s: Object "DFID" locates on remote server\n",
89 mdd2obd_dev(m)->obd_name, PFID(mdo2fid(mdd_obj)));
92 rc = mdd_permission_internal_locked(env, mdd_obj, pattr, mask,
97 if (likely(S_ISDIR(mdd_object_type(mdd_obj)) &&
98 dt_try_as_dir(env, dir)))
99 rc = dt_lookup(env, dir, (struct dt_rec *)fid, key);
106 int mdd_lookup(const struct lu_env *env,
107 struct md_object *pobj, const struct lu_name *lname,
108 struct lu_fid *fid, struct md_op_spec *spec)
110 struct lu_attr *pattr = MDD_ENV_VAR(env, pattr);
114 rc = mdd_la_get(env, md2mdd_obj(pobj), pattr);
118 rc = __mdd_lookup(env, pobj, pattr, lname, fid,
119 (spec != NULL && spec->sp_permitted) ? 0 : MAY_EXEC);
123 /** Read the link EA into a temp buffer.
124 * Uses the mdd_thread_info::mti_big_buf since it is generally large.
125 * A pointer to the buffer is stored in \a ldata::ld_buf.
129 static int __mdd_links_read(const struct lu_env *env,
130 struct mdd_object *mdd_obj,
131 struct linkea_data *ldata)
135 if (!mdd_object_exists(mdd_obj))
138 /* First try a small buf */
139 LASSERT(env != NULL);
140 ldata->ld_buf = lu_buf_check_and_alloc(&mdd_env_info(env)->mti_link_buf,
142 if (ldata->ld_buf->lb_buf == NULL)
145 rc = mdo_xattr_get(env, mdd_obj, ldata->ld_buf, XATTR_NAME_LINK);
147 /* Buf was too small, figure out what we need. */
148 lu_buf_free(ldata->ld_buf);
149 rc = mdo_xattr_get(env, mdd_obj, ldata->ld_buf,
153 ldata->ld_buf = lu_buf_check_and_alloc(ldata->ld_buf, rc);
154 if (ldata->ld_buf->lb_buf == NULL)
156 rc = mdo_xattr_get(env, mdd_obj, ldata->ld_buf,
160 lu_buf_free(ldata->ld_buf);
161 ldata->ld_buf = NULL;
165 return linkea_init(ldata);
168 static int mdd_links_read(const struct lu_env *env,
169 struct mdd_object *mdd_obj,
170 struct linkea_data *ldata)
174 rc = __mdd_links_read(env, mdd_obj, ldata);
176 rc = linkea_init(ldata);
181 static int mdd_links_read_with_rec(const struct lu_env *env,
182 struct mdd_object *mdd_obj,
183 struct linkea_data *ldata)
187 rc = __mdd_links_read(env, mdd_obj, ldata);
189 rc = linkea_init_with_rec(ldata);
195 * Get parent FID of the directory
197 * Read parent FID from linkEA, if that fails, then do lookup
198 * dotdot to get the parent FID.
200 * \param[in] env execution environment
201 * \param[in] obj object from which to find the parent FID
202 * \param[in] attr attribute of the object
203 * \param[out] fid fid to get the parent FID
205 * \retval 0 if getting the parent FID succeeds.
206 * \retval negative errno if getting the parent FID fails.
208 static inline int mdd_parent_fid(const struct lu_env *env,
209 struct mdd_object *obj,
210 const struct lu_attr *attr,
213 struct mdd_thread_info *info = mdd_env_info(env);
214 struct linkea_data ldata = { NULL };
215 struct lu_buf *buf = &info->mti_link_buf;
216 struct lu_name lname;
221 LASSERT(S_ISDIR(mdd_object_type(obj)));
223 buf = lu_buf_check_and_alloc(buf, PATH_MAX);
224 if (buf->lb_buf == NULL)
225 GOTO(lookup, rc = 0);
228 rc = mdd_links_read_with_rec(env, obj, &ldata);
232 LASSERT(ldata.ld_leh != NULL);
233 /* Directory should only have 1 parent */
234 if (ldata.ld_leh->leh_reccount > 1)
237 ldata.ld_lee = (struct link_ea_entry *)(ldata.ld_leh + 1);
239 linkea_entry_unpack(ldata.ld_lee, &ldata.ld_reclen, &lname, fid);
240 if (likely(fid_is_sane(fid)))
243 rc = __mdd_lookup(env, &obj->mod_obj, attr, &lname_dotdot, fid, 0);
248 * For root fid use special function, which does not compare version component
249 * of fid. Version component is different for root fids on all MDTs.
251 int mdd_is_root(struct mdd_device *mdd, const struct lu_fid *fid)
253 return fid_seq(&mdd->mdd_root_fid) == fid_seq(fid) &&
254 fid_oid(&mdd->mdd_root_fid) == fid_oid(fid);
258 * return 1: if lf is the fid of the ancestor of p1;
261 * return -EREMOTE: if remote object is found, in this
262 * case fid of remote object is saved to @pf;
264 * otherwise: values < 0, errors.
266 static int mdd_is_parent(const struct lu_env *env,
267 struct mdd_device *mdd,
268 struct mdd_object *p1,
269 const struct lu_attr *attr,
270 const struct lu_fid *lf,
273 struct mdd_object *parent = NULL;
278 LASSERT(!lu_fid_eq(mdo2fid(p1), lf));
279 pfid = &mdd_env_info(env)->mti_fid;
281 /* Check for root first. */
282 if (mdd_is_root(mdd, mdo2fid(p1)))
286 /* this is done recursively */
287 rc = mdd_parent_fid(env, p1, attr, pfid);
290 if (mdd_is_root(mdd, pfid))
292 if (lu_fid_eq(pfid, &mdd->mdd_local_root_fid))
294 if (lu_fid_eq(pfid, lf))
297 mdd_object_put(env, parent);
299 parent = mdd_object_find(env, mdd, pfid);
301 GOTO(out, rc = PTR_ERR(parent));
303 if (!mdd_object_exists(parent))
304 GOTO(out, rc = -EINVAL);
310 if (parent && !IS_ERR(parent))
311 mdd_object_put(env, parent);
316 * No permission check is needed.
318 * returns 1: if fid is ancestor of @mo;
319 * returns 0: if fid is not an ancestor of @mo;
321 * returns EREMOTE if remote object is found, fid of remote object is saved to
324 * returns < 0: if error
326 int mdd_is_subdir(const struct lu_env *env, struct md_object *mo,
327 const struct lu_fid *fid, struct lu_fid *sfid)
329 struct mdd_device *mdd = mdo2mdd(mo);
330 struct lu_attr *attr = MDD_ENV_VAR(env, cattr);
334 if (!S_ISDIR(mdd_object_type(md2mdd_obj(mo))))
337 rc = mdd_la_get(env, md2mdd_obj(mo), attr);
341 rc = mdd_is_parent(env, mdd, md2mdd_obj(mo), attr, fid, sfid);
345 } else if (rc == 1) {
346 /* found @fid is parent */
354 * Check that @dir contains no entries except (possibly) dot and dotdot.
359 * -ENOTDIR not a directory object
360 * -ENOTEMPTY not empty
364 static int mdd_dir_is_empty(const struct lu_env *env,
365 struct mdd_object *dir)
368 struct dt_object *obj;
369 const struct dt_it_ops *iops;
373 obj = mdd_object_child(dir);
374 if (!dt_try_as_dir(env, obj))
377 iops = &obj->do_index_ops->dio_it;
378 it = iops->init(env, obj, LUDA_64BITHASH);
380 result = iops->get(env, it, (const struct dt_key *)"");
383 for (result = 0, i = 0; result == 0 && i < 3; ++i)
384 result = iops->next(env, it);
387 else if (result == 1)
389 } else if (result == 0)
391 * Huh? Index contains no zero key?
398 result = PTR_ERR(it);
403 * Determine if the target object can be hard linked, and right now it only
404 * checks if the link count reach the maximum limit. Note: for ldiskfs, the
405 * directory nlink count might exceed the maximum link count(see
406 * osd_object_ref_add), so it only check nlink for non-directories.
408 * \param[in] env thread environment
409 * \param[in] obj object being linked to
410 * \param[in] la attributes of \a obj
412 * \retval 0 if \a obj can be hard linked
413 * \retval negative error if \a obj is a directory or has too
416 static int __mdd_may_link(const struct lu_env *env, struct mdd_object *obj,
417 const struct lu_attr *la)
419 struct mdd_device *m = mdd_obj2mdd_dev(obj);
424 /* Subdir count limitation can be broken through
425 * (see osd_object_ref_add), so only check non-directory here. */
426 if (!S_ISDIR(la->la_mode) &&
427 la->la_nlink >= m->mdd_dt_conf.ddp_max_nlink)
434 * Check whether it may create the cobj under the pobj.
436 * \param[in] env execution environment
437 * \param[in] pobj the parent directory
438 * \param[in] pattr the attribute of the parent directory
439 * \param[in] cobj the child to be created
440 * \param[in] check_perm if check WRITE|EXEC permission for parent
442 * \retval = 0 create the child under this dir is allowed
443 * \retval negative errno create the child under this dir is
446 int mdd_may_create(const struct lu_env *env, struct mdd_object *pobj,
447 const struct lu_attr *pattr, struct mdd_object *cobj,
453 if (cobj && mdd_object_exists(cobj))
456 if (mdd_is_dead_obj(pobj))
460 rc = mdd_permission_internal_locked(env, pobj, pattr,
461 MAY_WRITE | MAY_EXEC,
467 * Check whether can unlink from the pobj in the case of "cobj == NULL".
469 int mdd_may_unlink(const struct lu_env *env, struct mdd_object *pobj,
470 const struct lu_attr *pattr, const struct lu_attr *attr)
475 if (mdd_is_dead_obj(pobj))
478 if (attr->la_flags & (LUSTRE_APPEND_FL | LUSTRE_IMMUTABLE_FL))
481 rc = mdd_permission_internal_locked(env, pobj, pattr,
482 MAY_WRITE | MAY_EXEC,
487 if (pattr->la_flags & LUSTRE_APPEND_FL)
494 * pobj == NULL is remote ops case, under such case, pobj's
495 * VTX feature has been checked already, no need check again.
497 static inline int mdd_is_sticky(const struct lu_env *env,
498 struct mdd_object *pobj,
499 const struct lu_attr *pattr,
500 struct mdd_object *cobj,
501 const struct lu_attr *cattr)
503 struct lu_ucred *uc = lu_ucred_assert(env);
506 LASSERT(pattr != NULL);
507 if (!(pattr->la_mode & S_ISVTX) ||
508 (pattr->la_uid == uc->uc_fsuid))
512 LASSERT(cattr != NULL);
513 if (cattr->la_uid == uc->uc_fsuid)
516 return !md_capable(uc, CFS_CAP_FOWNER);
519 static int mdd_may_delete_entry(const struct lu_env *env,
520 struct mdd_object *pobj,
521 const struct lu_attr *pattr,
526 LASSERT(pobj != NULL);
527 if (!mdd_object_exists(pobj))
530 if (mdd_is_dead_obj(pobj))
535 rc = mdd_permission_internal_locked(env, pobj, pattr,
536 MAY_WRITE | MAY_EXEC,
542 if (pattr->la_flags & LUSTRE_APPEND_FL)
549 * Check whether it may delete the cobj from the pobj.
552 int mdd_may_delete(const struct lu_env *env, struct mdd_object *tpobj,
553 const struct lu_attr *tpattr, struct mdd_object *tobj,
554 const struct lu_attr *tattr, const struct lu_attr *cattr,
555 int check_perm, int check_empty)
561 LASSERT(tpattr != NULL);
562 rc = mdd_may_delete_entry(env, tpobj, tpattr, check_perm);
570 if (!mdd_object_exists(tobj))
573 if (mdd_is_dead_obj(tobj))
576 if (mdd_is_sticky(env, tpobj, tpattr, tobj, tattr))
579 if (tattr->la_flags & (LUSTRE_APPEND_FL | LUSTRE_IMMUTABLE_FL))
582 /* additional check the rename case */
584 if (S_ISDIR(cattr->la_mode)) {
585 struct mdd_device *mdd = mdo2mdd(&tobj->mod_obj);
587 if (!S_ISDIR(tattr->la_mode))
590 if (lu_fid_eq(mdo2fid(tobj), &mdd->mdd_root_fid))
592 } else if (S_ISDIR(tattr->la_mode))
596 if (S_ISDIR(tattr->la_mode) && check_empty)
597 rc = mdd_dir_is_empty(env, tobj);
603 * Check whether it can create the link file(linked to @src_obj) under
604 * the target directory(@tgt_obj), and src_obj has been locked by
607 * \param[in] env execution environment
608 * \param[in] tgt_obj the target directory
609 * \param[in] tattr attributes of target directory
610 * \param[in] lname the link name
611 * \param[in] src_obj source object for link
612 * \param[in] cattr attributes for source object
614 * \retval = 0 it is allowed to create the link file under tgt_obj
615 * \retval negative error not allowed to create the link file
617 static int mdd_link_sanity_check(const struct lu_env *env,
618 struct mdd_object *tgt_obj,
619 const struct lu_attr *tattr,
620 const struct lu_name *lname,
621 struct mdd_object *src_obj,
622 const struct lu_attr *cattr)
624 struct mdd_device *m = mdd_obj2mdd_dev(src_obj);
628 if (!mdd_object_exists(src_obj))
631 if (mdd_is_dead_obj(src_obj))
634 /* Local ops, no lookup before link, check filename length here. */
635 rc = mdd_name_check(m, lname);
639 if (cattr->la_flags & (LUSTRE_IMMUTABLE_FL | LUSTRE_APPEND_FL))
642 if (S_ISDIR(mdd_object_type(src_obj)))
645 LASSERT(src_obj != tgt_obj);
646 rc = mdd_may_create(env, tgt_obj, tattr, NULL, true);
650 rc = __mdd_may_link(env, src_obj, cattr);
655 static int __mdd_index_delete_only(const struct lu_env *env, struct mdd_object *pobj,
656 const char *name, struct thandle *handle)
658 struct dt_object *next = mdd_object_child(pobj);
662 if (dt_try_as_dir(env, next))
663 rc = dt_delete(env, next, (struct dt_key *)name, handle);
670 static int __mdd_index_insert_only(const struct lu_env *env,
671 struct mdd_object *pobj,
672 const struct lu_fid *lf, __u32 type,
674 struct thandle *handle)
676 struct dt_object *next = mdd_object_child(pobj);
680 if (dt_try_as_dir(env, next)) {
681 struct dt_insert_rec *rec = &mdd_env_info(env)->mti_dt_rec;
682 struct lu_ucred *uc = lu_ucred_check(env);
686 rec->rec_type = type;
687 ignore_quota = uc ? uc->uc_cap & CFS_CAP_SYS_RESOURCE_MASK : 1;
688 rc = dt_insert(env, next, (const struct dt_rec *)rec,
689 (const struct dt_key *)name, handle,
697 /* insert named index, add reference if isdir */
698 static int __mdd_index_insert(const struct lu_env *env, struct mdd_object *pobj,
699 const struct lu_fid *lf, __u32 type,
700 const char *name, struct thandle *handle)
705 rc = __mdd_index_insert_only(env, pobj, lf, type, name, handle);
706 if (rc == 0 && S_ISDIR(type)) {
707 mdd_write_lock(env, pobj, MOR_TGT_PARENT);
708 mdo_ref_add(env, pobj, handle);
709 mdd_write_unlock(env, pobj);
715 /* delete named index, drop reference if isdir */
716 static int __mdd_index_delete(const struct lu_env *env, struct mdd_object *pobj,
717 const char *name, int is_dir,
718 struct thandle *handle)
723 rc = __mdd_index_delete_only(env, pobj, name, handle);
724 if (rc == 0 && is_dir) {
725 mdd_write_lock(env, pobj, MOR_TGT_PARENT);
726 mdo_ref_del(env, pobj, handle);
727 mdd_write_unlock(env, pobj);
733 static int mdd_llog_record_calc_size(const struct lu_env *env,
734 const struct lu_name *tname,
735 const struct lu_name *sname)
737 const struct lu_ucred *uc = lu_ucred(env);
738 enum changelog_rec_flags crf = 0;
743 if (uc != NULL && uc->uc_jobid[0] != '\0')
746 return llog_data_len(LLOG_CHANGELOG_HDR_SZ + changelog_rec_offset(crf) +
747 (tname != NULL ? tname->ln_namelen : 0) +
748 (sname != NULL ? 1 + sname->ln_namelen : 0));
751 int mdd_declare_changelog_store(const struct lu_env *env,
752 struct mdd_device *mdd,
753 const struct lu_name *tname,
754 const struct lu_name *sname,
755 struct thandle *handle)
757 struct obd_device *obd = mdd2obd_dev(mdd);
758 struct llog_ctxt *ctxt;
759 struct llog_changelog_rec *rec;
761 struct thandle *llog_th;
766 if (!(mdd->mdd_cl.mc_flags & CLM_ON))
769 reclen = mdd_llog_record_calc_size(env, tname, sname);
770 buf = lu_buf_check_and_alloc(&mdd_env_info(env)->mti_big_buf, reclen);
771 if (buf->lb_buf == NULL)
775 rec->cr_hdr.lrh_len = reclen;
776 rec->cr_hdr.lrh_type = CHANGELOG_REC;
778 ctxt = llog_get_context(obd, LLOG_CHANGELOG_ORIG_CTXT);
782 llog_th = thandle_get_sub(env, handle, ctxt->loc_handle->lgh_obj);
784 GOTO(out_put, rc = PTR_ERR(llog_th));
786 rc = llog_declare_add(env, ctxt->loc_handle, &rec->cr_hdr, llog_th);
794 /** Add a changelog entry \a rec to the changelog llog
797 * \param handle - currently ignored since llogs start their own transaction;
798 * this will hopefully be fixed in llog rewrite
801 int mdd_changelog_store(const struct lu_env *env, struct mdd_device *mdd,
802 struct llog_changelog_rec *rec, struct thandle *th)
804 struct obd_device *obd = mdd2obd_dev(mdd);
805 struct llog_ctxt *ctxt;
806 struct thandle *llog_th;
809 rec->cr_hdr.lrh_len = llog_data_len(sizeof(*rec) +
810 changelog_rec_varsize(&rec->cr));
812 /* llog_lvfs_write_rec sets the llog tail len */
813 rec->cr_hdr.lrh_type = CHANGELOG_REC;
814 rec->cr.cr_time = cl_time();
816 spin_lock(&mdd->mdd_cl.mc_lock);
817 /* NB: I suppose it's possible llog_add adds out of order wrt cr_index,
818 * but as long as the MDD transactions are ordered correctly for e.g.
819 * rename conflicts, I don't think this should matter. */
820 rec->cr.cr_index = ++mdd->mdd_cl.mc_index;
821 spin_unlock(&mdd->mdd_cl.mc_lock);
823 ctxt = llog_get_context(obd, LLOG_CHANGELOG_ORIG_CTXT);
827 llog_th = thandle_get_sub(env, th, ctxt->loc_handle->lgh_obj);
829 GOTO(out_put, rc = PTR_ERR(llog_th));
831 /* nested journal transaction */
832 rc = llog_add(env, ctxt->loc_handle, &rec->cr_hdr, NULL, llog_th);
841 static void mdd_changelog_rec_ext_rename(struct changelog_rec *rec,
842 const struct lu_fid *sfid,
843 const struct lu_fid *spfid,
844 const struct lu_name *sname)
846 struct changelog_ext_rename *rnm = changelog_rec_rename(rec);
847 size_t extsize = sname->ln_namelen + 1;
849 LASSERT(sfid != NULL);
850 LASSERT(spfid != NULL);
851 LASSERT(sname != NULL);
853 rnm->cr_sfid = *sfid;
854 rnm->cr_spfid = *spfid;
856 changelog_rec_name(rec)[rec->cr_namelen] = '\0';
857 strlcpy(changelog_rec_sname(rec), sname->ln_name, extsize);
858 rec->cr_namelen += extsize;
861 void mdd_changelog_rec_ext_jobid(struct changelog_rec *rec, const char *jobid)
863 struct changelog_ext_jobid *jid = changelog_rec_jobid(rec);
865 if (jobid == NULL || jobid[0] == '\0')
868 strlcpy(jid->cr_jobid, jobid, sizeof(jid->cr_jobid));
871 /** Store a namespace change changelog record
872 * If this fails, we must fail the whole transaction; we don't
873 * want the change to commit without the log entry.
874 * \param target - mdd_object of change
875 * \param tpfid - target parent dir/object fid
876 * \param sfid - source object fid
877 * \param spfid - source parent fid
878 * \param tname - target name string
879 * \param sname - source name string
880 * \param handle - transaction handle
882 int mdd_changelog_ns_store(const struct lu_env *env,
883 struct mdd_device *mdd,
884 enum changelog_rec_type type,
885 enum changelog_rec_flags crf,
886 struct mdd_object *target,
887 const struct lu_fid *tpfid,
888 const struct lu_fid *sfid,
889 const struct lu_fid *spfid,
890 const struct lu_name *tname,
891 const struct lu_name *sname,
892 struct thandle *handle)
894 const struct lu_ucred *uc = lu_ucred(env);
895 struct llog_changelog_rec *rec;
902 if (!(mdd->mdd_cl.mc_flags & CLM_ON))
905 if ((mdd->mdd_cl.mc_mask & (1 << type)) == 0)
908 LASSERT(tpfid != NULL);
909 LASSERT(tname != NULL);
910 LASSERT(handle != NULL);
912 reclen = mdd_llog_record_calc_size(env, tname, sname);
913 buf = lu_buf_check_and_alloc(&mdd_env_info(env)->mti_big_buf, reclen);
914 if (buf->lb_buf == NULL)
920 if (uc != NULL && uc->uc_jobid[0] != '\0')
928 rec->cr.cr_flags = crf;
929 rec->cr.cr_type = (__u32)type;
930 rec->cr.cr_pfid = *tpfid;
931 rec->cr.cr_namelen = tname->ln_namelen;
932 memcpy(changelog_rec_name(&rec->cr), tname->ln_name, tname->ln_namelen);
934 if (crf & CLF_RENAME)
935 mdd_changelog_rec_ext_rename(&rec->cr, sfid, spfid, sname);
938 mdd_changelog_rec_ext_jobid(&rec->cr, uc->uc_jobid);
940 if (likely(target != NULL)) {
941 rec->cr.cr_tfid = *mdo2fid(target);
942 target->mod_cltime = ktime_get();
944 fid_zero(&rec->cr.cr_tfid);
947 rc = mdd_changelog_store(env, mdd, rec, handle);
949 CERROR("%s: cannot store changelog record: type = %d, "
950 "name = '%s', t = "DFID", p = "DFID": rc = %d\n",
951 mdd2obd_dev(mdd)->obd_name, type, tname->ln_name,
952 PFID(&rec->cr.cr_tfid), PFID(&rec->cr.cr_pfid), rc);
959 static int __mdd_links_add(const struct lu_env *env,
960 struct mdd_object *mdd_obj,
961 struct linkea_data *ldata,
962 const struct lu_name *lname,
963 const struct lu_fid *pfid,
964 int first, int check)
968 if (ldata->ld_leh == NULL) {
969 rc = first ? -ENODATA : mdd_links_read(env, mdd_obj, ldata);
973 rc = linkea_data_new(ldata,
974 &mdd_env_info(env)->mti_link_buf);
981 rc = linkea_links_find(ldata, lname, pfid);
982 if (rc && rc != -ENOENT)
988 if (OBD_FAIL_CHECK(OBD_FAIL_LFSCK_LINKEA_MORE)) {
989 struct lu_fid *tfid = &mdd_env_info(env)->mti_fid2;
993 linkea_add_buf(ldata, lname, tfid);
996 if (OBD_FAIL_CHECK(OBD_FAIL_LFSCK_LINKEA_MORE2))
997 linkea_add_buf(ldata, lname, pfid);
999 return linkea_add_buf(ldata, lname, pfid);
1002 static int __mdd_links_del(const struct lu_env *env,
1003 struct mdd_object *mdd_obj,
1004 struct linkea_data *ldata,
1005 const struct lu_name *lname,
1006 const struct lu_fid *pfid)
1010 if (ldata->ld_leh == NULL) {
1011 rc = mdd_links_read(env, mdd_obj, ldata);
1016 rc = linkea_links_find(ldata, lname, pfid);
1020 linkea_del_buf(ldata, lname);
1024 static int mdd_linkea_prepare(const struct lu_env *env,
1025 struct mdd_object *mdd_obj,
1026 const struct lu_fid *oldpfid,
1027 const struct lu_name *oldlname,
1028 const struct lu_fid *newpfid,
1029 const struct lu_name *newlname,
1030 int first, int check,
1031 struct linkea_data *ldata)
1036 if (OBD_FAIL_CHECK(OBD_FAIL_FID_IGIF))
1039 LASSERT(oldpfid != NULL || newpfid != NULL);
1041 if (mdd_obj->mod_flags & DEAD_OBJ)
1042 /* Unnecessary to update linkEA for dead object. */
1045 if (oldpfid != NULL) {
1046 rc = __mdd_links_del(env, mdd_obj, ldata, oldlname, oldpfid);
1048 if ((check == 1) || (rc != -ENODATA && rc != -ENOENT))
1051 /* No changes done. */
1056 /* If renaming, add the new record */
1057 if (newpfid != NULL)
1058 rc = __mdd_links_add(env, mdd_obj, ldata, newlname, newpfid,
1064 int mdd_links_rename(const struct lu_env *env,
1065 struct mdd_object *mdd_obj,
1066 const struct lu_fid *oldpfid,
1067 const struct lu_name *oldlname,
1068 const struct lu_fid *newpfid,
1069 const struct lu_name *newlname,
1070 struct thandle *handle,
1071 struct linkea_data *ldata,
1072 int first, int check)
1077 if (ldata == NULL) {
1078 ldata = &mdd_env_info(env)->mti_link_data;
1079 memset(ldata, 0, sizeof(*ldata));
1080 rc = mdd_linkea_prepare(env, mdd_obj, oldpfid, oldlname,
1081 newpfid, newlname, first, check, ldata);
1086 if (!(mdd_obj->mod_flags & DEAD_OBJ))
1087 rc = mdd_links_write(env, mdd_obj, ldata, handle);
1093 if (newlname == NULL)
1094 CERROR("link_ea add failed %d "DFID"\n",
1095 rc, PFID(mdd_object_fid(mdd_obj)));
1096 else if (oldpfid == NULL)
1097 CERROR("link_ea add '%.*s' failed %d "DFID"\n",
1098 newlname->ln_namelen, newlname->ln_name, rc,
1099 PFID(mdd_object_fid(mdd_obj)));
1100 else if (newpfid == NULL)
1101 CERROR("link_ea del '%.*s' failed %d "DFID"\n",
1102 oldlname->ln_namelen, oldlname->ln_name, rc,
1103 PFID(mdd_object_fid(mdd_obj)));
1105 CERROR("link_ea rename '%.*s'->'%.*s' failed %d "DFID
1106 "\n", oldlname->ln_namelen, oldlname->ln_name,
1107 newlname->ln_namelen, newlname->ln_name, rc,
1108 PFID(mdd_object_fid(mdd_obj)));
1111 if (is_vmalloc_addr(ldata->ld_buf))
1112 /* if we vmalloced a large buffer drop it */
1113 lu_buf_free(ldata->ld_buf);
1118 static inline int mdd_links_add(const struct lu_env *env,
1119 struct mdd_object *mdd_obj,
1120 const struct lu_fid *pfid,
1121 const struct lu_name *lname,
1122 struct thandle *handle,
1123 struct linkea_data *ldata, int first)
1125 return mdd_links_rename(env, mdd_obj, NULL, NULL,
1126 pfid, lname, handle, ldata, first, 0);
1129 static inline int mdd_links_del(const struct lu_env *env,
1130 struct mdd_object *mdd_obj,
1131 const struct lu_fid *pfid,
1132 const struct lu_name *lname,
1133 struct thandle *handle)
1135 return mdd_links_rename(env, mdd_obj, pfid, lname,
1136 NULL, NULL, handle, NULL, 0, 0);
1139 /** Read the link EA into a temp buffer.
1140 * Uses the name_buf since it is generally large.
1141 * \retval IS_ERR err
1142 * \retval ptr to \a lu_buf (always \a mti_big_buf)
1144 struct lu_buf *mdd_links_get(const struct lu_env *env,
1145 struct mdd_object *mdd_obj)
1147 struct linkea_data ldata = { NULL };
1150 rc = mdd_links_read(env, mdd_obj, &ldata);
1151 return rc ? ERR_PTR(rc) : ldata.ld_buf;
1154 int mdd_links_write(const struct lu_env *env, struct mdd_object *mdd_obj,
1155 struct linkea_data *ldata, struct thandle *handle)
1157 const struct lu_buf *buf;
1160 if (ldata == NULL || ldata->ld_buf == NULL ||
1161 ldata->ld_leh == NULL)
1164 if (OBD_FAIL_CHECK(OBD_FAIL_LFSCK_NO_LINKEA))
1168 buf = mdd_buf_get_const(env, ldata->ld_buf->lb_buf,
1169 ldata->ld_leh->leh_len);
1170 rc = mdo_xattr_set(env, mdd_obj, buf, XATTR_NAME_LINK, 0, handle);
1171 if (unlikely(rc == -ENOSPC)) {
1172 rc = linkea_overflow_shrink(ldata);
1180 static int mdd_declare_links_add(const struct lu_env *env,
1181 struct mdd_object *mdd_obj,
1182 struct thandle *handle,
1183 struct linkea_data *ldata)
1189 if (ldata != NULL && ldata->ld_leh != NULL) {
1190 ea_len = ldata->ld_leh->leh_len;
1191 linkea = ldata->ld_buf->lb_buf;
1193 ea_len = MAX_LINKEA_SIZE;
1197 rc = mdo_declare_xattr_set(env, mdd_obj,
1198 mdd_buf_get_const(env, linkea, ea_len),
1199 XATTR_NAME_LINK, 0, handle);
1204 static inline int mdd_declare_links_del(const struct lu_env *env,
1205 struct mdd_object *c,
1206 struct thandle *handle)
1210 /* For directory, the linkEA will be removed together
1211 * with the object. */
1212 if (!S_ISDIR(mdd_object_type(c)))
1213 rc = mdd_declare_links_add(env, c, handle, NULL);
1218 static int mdd_declare_link(const struct lu_env *env,
1219 struct mdd_device *mdd,
1220 struct mdd_object *p,
1221 struct mdd_object *c,
1222 const struct lu_name *name,
1223 struct thandle *handle,
1225 struct linkea_data *data)
1227 struct lu_fid tfid = *mdo2fid(c);
1230 if (OBD_FAIL_CHECK(OBD_FAIL_LFSCK_DANGLING3))
1231 tfid.f_oid = cfs_fail_val;
1233 rc = mdo_declare_index_insert(env, p, &tfid, mdd_object_type(c),
1234 name->ln_name, handle);
1238 rc = mdo_declare_ref_add(env, c, handle);
1242 la->la_valid = LA_CTIME | LA_MTIME;
1243 rc = mdo_declare_attr_set(env, p, la, handle);
1247 la->la_valid = LA_CTIME;
1248 rc = mdo_declare_attr_set(env, c, la, handle);
1252 rc = mdd_declare_links_add(env, c, handle, data);
1256 rc = mdd_declare_changelog_store(env, mdd, name, NULL, handle);
1261 static int mdd_link(const struct lu_env *env, struct md_object *tgt_obj,
1262 struct md_object *src_obj, const struct lu_name *lname,
1265 const char *name = lname->ln_name;
1266 struct lu_attr *la = &mdd_env_info(env)->mti_la_for_fix;
1267 struct mdd_object *mdd_tobj = md2mdd_obj(tgt_obj);
1268 struct mdd_object *mdd_sobj = md2mdd_obj(src_obj);
1269 struct lu_attr *cattr = MDD_ENV_VAR(env, cattr);
1270 struct lu_attr *tattr = MDD_ENV_VAR(env, tattr);
1271 struct mdd_device *mdd = mdo2mdd(src_obj);
1272 struct thandle *handle;
1273 struct lu_fid *tfid = &mdd_env_info(env)->mti_fid2;
1274 struct linkea_data *ldata = &mdd_env_info(env)->mti_link_data;
1278 rc = mdd_la_get(env, mdd_sobj, cattr);
1282 rc = mdd_la_get(env, mdd_tobj, tattr);
1287 * If we are using project inheritance, we only allow hard link
1288 * creation in our tree when the project IDs are the same;
1289 * otherwise the tree quota mechanism could be circumvented.
1291 if ((tattr->la_flags & LUSTRE_PROJINHERIT_FL) &&
1292 (tattr->la_projid != cattr->la_projid))
1295 handle = mdd_trans_create(env, mdd);
1297 GOTO(out_pending, rc = PTR_ERR(handle));
1299 memset(ldata, 0, sizeof(*ldata));
1301 LASSERT(ma->ma_attr.la_valid & LA_CTIME);
1302 la->la_ctime = la->la_mtime = ma->ma_attr.la_ctime;
1304 /* Note: even this function will change ldata, but it comes from
1305 * thread_info, which is completely temporary and only seen in
1306 * this function, so we do not need reset ldata once it fails.*/
1307 rc = mdd_linkea_prepare(env, mdd_sobj, NULL, NULL, mdo2fid(mdd_tobj),
1308 lname, 0, 0, ldata);
1312 rc = mdd_declare_link(env, mdd, mdd_tobj, mdd_sobj, lname, handle,
1317 rc = mdd_trans_start(env, mdd, handle);
1321 mdd_write_lock(env, mdd_sobj, MOR_TGT_CHILD);
1322 rc = mdd_link_sanity_check(env, mdd_tobj, tattr, lname, mdd_sobj,
1325 GOTO(out_unlock, rc);
1327 if (!OBD_FAIL_CHECK(OBD_FAIL_LFSCK_LESS_NLINK)) {
1328 rc = mdo_ref_add(env, mdd_sobj, handle);
1330 GOTO(out_unlock, rc);
1333 *tfid = *mdo2fid(mdd_sobj);
1334 if (OBD_FAIL_CHECK(OBD_FAIL_LFSCK_DANGLING3))
1335 tfid->f_oid = cfs_fail_val;
1337 rc = __mdd_index_insert_only(env, mdd_tobj, tfid,
1338 mdd_object_type(mdd_sobj), name, handle);
1340 mdo_ref_del(env, mdd_sobj, handle);
1341 GOTO(out_unlock, rc);
1344 la->la_valid = LA_CTIME | LA_MTIME;
1345 rc = mdd_update_time(env, mdd_tobj, tattr, la, handle);
1347 GOTO(out_unlock, rc);
1349 la->la_valid = LA_CTIME;
1350 rc = mdd_update_time(env, mdd_sobj, cattr, la, handle);
1352 /* Note: The failure of links_add should not cause the
1353 * link failure, so do not check return value. */
1354 mdd_links_add(env, mdd_sobj, mdo2fid(mdd_tobj),
1355 lname, handle, ldata, 0);
1359 mdd_write_unlock(env, mdd_sobj);
1361 rc = mdd_changelog_ns_store(env, mdd, CL_HARDLINK, 0, mdd_sobj,
1362 mdo2fid(mdd_tobj), NULL, NULL,
1363 lname, NULL, handle);
1365 rc = mdd_trans_stop(env, mdd, rc, handle);
1366 if (is_vmalloc_addr(ldata->ld_buf))
1367 /* if we vmalloced a large buffer drop it */
1368 lu_buf_free(ldata->ld_buf);
1373 static int mdd_mark_orphan_object(const struct lu_env *env,
1374 struct mdd_object *obj, struct thandle *handle,
1377 struct lu_attr *attr = MDD_ENV_VAR(env, la_for_start);
1380 if (!S_ISDIR(mdd_object_type(obj)))
1383 attr->la_valid = LA_FLAGS;
1384 attr->la_flags = LUSTRE_ORPHAN_FL;
1387 rc = mdo_declare_attr_set(env, obj, attr, handle);
1389 rc = mdo_attr_set(env, obj, attr, handle);
1394 static int mdd_declare_finish_unlink(const struct lu_env *env,
1395 struct mdd_object *obj,
1396 struct thandle *handle)
1400 /* Sigh, we do not know if the unlink object will become orphan in
1401 * declare phase, but fortunately the flags here does not matter
1402 * in current declare implementation */
1403 rc = mdd_mark_orphan_object(env, obj, handle, true);
1407 rc = mdo_declare_destroy(env, obj, handle);
1411 rc = orph_declare_index_insert(env, obj, mdd_object_type(obj), handle);
1415 return mdd_declare_links_del(env, obj, handle);
1418 /* caller should take a lock before calling */
1419 int mdd_finish_unlink(const struct lu_env *env,
1420 struct mdd_object *obj, struct md_attr *ma,
1421 const struct mdd_object *pobj,
1422 const struct lu_name *lname,
1426 int is_dir = S_ISDIR(ma->ma_attr.la_mode);
1429 LASSERT(mdd_write_locked(env, obj) != 0);
1431 if (ma->ma_attr.la_nlink == 0 || is_dir) {
1432 /* add new orphan and the object
1433 * will be deleted during mdd_close() */
1434 obj->mod_flags |= DEAD_OBJ;
1435 if (obj->mod_count) {
1436 rc = __mdd_orphan_add(env, obj, th);
1438 CDEBUG(D_HA, "Object "DFID" is inserted into "
1439 "orphan list, open count = %d\n",
1440 PFID(mdd_object_fid(obj)),
1443 CERROR("Object "DFID" fail to be an orphan, "
1444 "open count = %d, maybe cause failed "
1446 PFID(mdd_object_fid(obj)),
1449 /* mark object as an orphan here, not
1450 * before __mdd_orphan_add() as racing
1451 * mdd_la_get() may propagate ORPHAN_OBJ
1452 * causing the asserition */
1453 rc = mdd_mark_orphan_object(env, obj, th, false);
1455 rc = mdo_destroy(env, obj, th);
1457 } else if (!is_dir) {
1458 /* old files may not have link ea; ignore errors */
1459 mdd_links_del(env, obj, mdo2fid(pobj), lname, th);
1467 * has mdd_write_lock on cobj already, but not on pobj yet
1469 int mdd_unlink_sanity_check(const struct lu_env *env, struct mdd_object *pobj,
1470 const struct lu_attr *pattr,
1471 struct mdd_object *cobj,
1472 const struct lu_attr *cattr)
1477 rc = mdd_may_delete(env, pobj, pattr, cobj, cattr, NULL, 1, 1);
1482 static int mdd_declare_unlink(const struct lu_env *env, struct mdd_device *mdd,
1483 struct mdd_object *p, struct mdd_object *c,
1484 const struct lu_name *name, struct md_attr *ma,
1485 struct thandle *handle, int no_name, int is_dir)
1487 struct lu_attr *la = &mdd_env_info(env)->mti_la_for_fix;
1490 if (!OBD_FAIL_CHECK(OBD_FAIL_LFSCK_DANGLING2)) {
1491 if (likely(no_name == 0)) {
1492 rc = mdo_declare_index_delete(env, p, name->ln_name,
1499 rc = mdo_declare_ref_del(env, p, handle);
1505 LASSERT(ma->ma_attr.la_valid & LA_CTIME);
1506 la->la_ctime = la->la_mtime = ma->ma_attr.la_ctime;
1507 la->la_valid = LA_CTIME | LA_MTIME;
1508 rc = mdo_declare_attr_set(env, p, la, handle);
1513 rc = mdo_declare_ref_del(env, c, handle);
1517 rc = mdo_declare_ref_del(env, c, handle);
1521 la->la_valid = LA_CTIME;
1522 rc = mdo_declare_attr_set(env, c, la, handle);
1526 rc = mdd_declare_finish_unlink(env, c, handle);
1530 /* FIXME: need changelog for remove entry */
1531 rc = mdd_declare_changelog_store(env, mdd, name, NULL, handle);
1538 * test if a file has an HSM archive
1539 * if HSM attributes are not found in ma update them from
1542 static bool mdd_hsm_archive_exists(const struct lu_env *env,
1543 struct mdd_object *obj,
1548 if (!(ma->ma_valid & MA_HSM)) {
1549 /* no HSM MD provided, read xattr */
1550 struct lu_buf *hsm_buf;
1551 const size_t buflen = sizeof(struct hsm_attrs);
1554 hsm_buf = mdd_buf_get(env, NULL, 0);
1555 lu_buf_alloc(hsm_buf, buflen);
1556 rc = mdo_xattr_get(env, obj, hsm_buf, XATTR_NAME_HSM);
1557 rc = lustre_buf2hsm(hsm_buf->lb_buf, rc, &ma->ma_hsm);
1558 lu_buf_free(hsm_buf);
1562 ma->ma_valid |= MA_HSM;
1564 if (ma->ma_hsm.mh_flags & HS_EXISTS)
1570 * Delete name entry and the object.
1571 * Note: no_name == 1 means it only destory the object, i.e. name_entry
1572 * does not exist for this object, and it could only happen during resending
1573 * of remote unlink. see the comments in mdt_reint_unlink. Unfortunately, lname
1574 * is also needed in this case(needed by changelog), so we have to add another
1575 * parameter(no_name)here. XXX: this is only needed in DNE phase I, on Phase II,
1576 * the ENOENT failure should be able to be fixed by redo mechanism.
1578 static int mdd_unlink(const struct lu_env *env, struct md_object *pobj,
1579 struct md_object *cobj, const struct lu_name *lname,
1580 struct md_attr *ma, int no_name)
1582 const char *name = lname->ln_name;
1583 struct lu_attr *pattr = MDD_ENV_VAR(env, pattr);
1584 struct lu_attr *cattr = MDD_ENV_VAR(env, cattr);
1585 struct lu_attr *la = &mdd_env_info(env)->mti_la_for_fix;
1586 struct mdd_object *mdd_pobj = md2mdd_obj(pobj);
1587 struct mdd_object *mdd_cobj = NULL;
1588 struct mdd_device *mdd = mdo2mdd(pobj);
1589 struct thandle *handle;
1590 int rc, is_dir = 0, cl_flags = 0;
1593 /* cobj == NULL means only delete name entry */
1594 if (likely(cobj != NULL)) {
1595 mdd_cobj = md2mdd_obj(cobj);
1596 if (mdd_object_exists(mdd_cobj) == 0)
1600 rc = mdd_la_get(env, mdd_pobj, pattr);
1604 if (likely(mdd_cobj != NULL)) {
1606 rc = mdd_la_get(env, mdd_cobj, cattr);
1610 is_dir = S_ISDIR(cattr->la_mode);
1611 /* search for an existing archive.
1612 * we should check ahead as the object
1613 * can be destroyed in this transaction */
1614 if (mdd_hsm_archive_exists(env, mdd_cobj, ma))
1615 cl_flags |= CLF_UNLINK_HSM_EXISTS;
1618 rc = mdd_unlink_sanity_check(env, mdd_pobj, pattr, mdd_cobj, cattr);
1622 handle = mdd_trans_create(env, mdd);
1624 RETURN(PTR_ERR(handle));
1626 rc = mdd_declare_unlink(env, mdd, mdd_pobj, mdd_cobj,
1627 lname, ma, handle, no_name, is_dir);
1631 rc = mdd_trans_start(env, mdd, handle);
1635 if (likely(mdd_cobj != NULL))
1636 mdd_write_lock(env, mdd_cobj, MOR_TGT_CHILD);
1638 if (likely(no_name == 0) && !OBD_FAIL_CHECK(OBD_FAIL_LFSCK_DANGLING2)) {
1639 rc = __mdd_index_delete(env, mdd_pobj, name, is_dir, handle);
1644 if (OBD_FAIL_CHECK(OBD_FAIL_LFSCK_MUL_REF) ||
1645 OBD_FAIL_CHECK(OBD_FAIL_LFSCK_NO_NAMEENTRY))
1646 GOTO(cleanup, rc = 0);
1648 if (likely(mdd_cobj != NULL)) {
1649 rc = mdo_ref_del(env, mdd_cobj, handle);
1651 __mdd_index_insert_only(env, mdd_pobj,
1653 mdd_object_type(mdd_cobj),
1660 mdo_ref_del(env, mdd_cobj, handle);
1662 /* fetch updated nlink */
1663 rc = mdd_la_get(env, mdd_cobj, cattr);
1668 LASSERT(ma->ma_attr.la_valid & LA_CTIME);
1669 la->la_ctime = la->la_mtime = ma->ma_attr.la_ctime;
1671 la->la_valid = LA_CTIME | LA_MTIME;
1672 rc = mdd_update_time(env, mdd_pobj, pattr, la, handle);
1676 /* Enough for only unlink the entry */
1677 if (unlikely(mdd_cobj == NULL))
1680 if (cattr->la_nlink > 0 || mdd_cobj->mod_count > 0) {
1681 /* update ctime of an unlinked file only if it is still
1682 * opened or a link still exists */
1683 la->la_valid = LA_CTIME;
1684 rc = mdd_update_time(env, mdd_cobj, cattr, la, handle);
1689 /* XXX: this transfer to ma will be removed with LOD/OSP */
1690 ma->ma_attr = *cattr;
1691 ma->ma_valid |= MA_INODE;
1692 rc = mdd_finish_unlink(env, mdd_cobj, ma, mdd_pobj, lname, handle);
1696 /* fetch updated nlink */
1697 rc = mdd_la_get(env, mdd_cobj, cattr);
1698 /* if object is removed then we can't get its attrs,
1700 if (rc == -ENOENT) {
1701 cattr->la_nlink = 0;
1705 if (cattr->la_nlink == 0) {
1706 ma->ma_attr = *cattr;
1707 ma->ma_valid |= MA_INODE;
1712 if (likely(mdd_cobj != NULL))
1713 mdd_write_unlock(env, mdd_cobj);
1716 if (cattr->la_nlink == 0)
1717 cl_flags |= CLF_UNLINK_LAST;
1719 cl_flags &= ~CLF_UNLINK_HSM_EXISTS;
1721 rc = mdd_changelog_ns_store(env, mdd,
1722 is_dir ? CL_RMDIR : CL_UNLINK, cl_flags,
1723 mdd_cobj, mdo2fid(mdd_pobj), NULL, NULL, lname, NULL,
1728 rc = mdd_trans_stop(env, mdd, rc, handle);
1734 * The permission has been checked when obj created, no need check again.
1736 static int mdd_cd_sanity_check(const struct lu_env *env,
1737 struct mdd_object *obj)
1742 if (!obj || mdd_is_dead_obj(obj))
1748 static int mdd_create_data(const struct lu_env *env,
1749 struct md_object *pobj,
1750 struct md_object *cobj,
1751 const struct md_op_spec *spec,
1754 struct mdd_device *mdd = mdo2mdd(cobj);
1755 struct mdd_object *mdd_pobj = md2mdd_obj(pobj);
1756 struct mdd_object *son = md2mdd_obj(cobj);
1757 struct thandle *handle;
1758 const struct lu_buf *buf;
1759 struct lu_attr *attr = MDD_ENV_VAR(env, cattr);
1760 struct dt_allocation_hint *hint = &mdd_env_info(env)->mti_hint;
1764 rc = mdd_cd_sanity_check(env, son);
1768 if (!md_should_create(spec->sp_cr_flags))
1772 * there are following use cases for this function:
1773 * 1) late striping - file was created with MDS_OPEN_DELAY_CREATE
1774 * striping can be specified or not
1777 rc = mdd_la_get(env, son, attr);
1781 /* calling ->ah_make_hint() is used to transfer information from parent */
1782 mdd_object_make_hint(env, mdd_pobj, son, attr, spec, hint);
1784 handle = mdd_trans_create(env, mdd);
1786 GOTO(out_free, rc = PTR_ERR(handle));
1789 * XXX: Setting the lov ea is not locked but setting the attr is locked?
1790 * Should this be fixed?
1792 CDEBUG(D_OTHER, "ea %p/%u, cr_flags %#llo, no_create %u\n",
1793 spec->u.sp_ea.eadata, spec->u.sp_ea.eadatalen,
1794 spec->sp_cr_flags, spec->no_create);
1796 if (spec->no_create || (spec->sp_cr_flags & MDS_OPEN_HAS_EA)) {
1797 buf = mdd_buf_get_const(env, spec->u.sp_ea.eadata,
1798 spec->u.sp_ea.eadatalen);
1803 rc = dt_declare_xattr_set(env, mdd_object_child(son), buf,
1804 XATTR_NAME_LOV, 0, handle);
1808 rc = mdd_declare_changelog_store(env, mdd, NULL, NULL, handle);
1812 rc = mdd_trans_start(env, mdd, handle);
1816 rc = dt_xattr_set(env, mdd_object_child(son), buf, XATTR_NAME_LOV,
1822 rc = mdd_changelog_data_store(env, mdd, CL_LAYOUT, 0, son, handle);
1825 rc = mdd_trans_stop(env, mdd, rc, handle);
1831 static int mdd_declare_object_initialize(const struct lu_env *env,
1832 struct mdd_object *parent,
1833 struct mdd_object *child,
1834 const struct lu_attr *attr,
1835 struct thandle *handle)
1840 LASSERT(attr->la_valid & (LA_MODE | LA_TYPE));
1841 if (!S_ISDIR(attr->la_mode))
1844 rc = mdo_declare_index_insert(env, child, mdo2fid(child), S_IFDIR,
1849 rc = mdo_declare_ref_add(env, child, handle);
1853 rc = mdo_declare_index_insert(env, child, mdo2fid(parent), S_IFDIR,
1859 static int mdd_object_initialize(const struct lu_env *env,
1860 const struct lu_fid *pfid,
1861 struct mdd_object *child,
1862 struct lu_attr *attr, struct thandle *handle,
1863 const struct md_op_spec *spec)
1868 if (S_ISDIR(attr->la_mode)) {
1869 /* Add "." and ".." for newly created dir */
1870 mdo_ref_add(env, child, handle);
1871 rc = __mdd_index_insert_only(env, child, mdo2fid(child),
1872 S_IFDIR, dot, handle);
1874 rc = __mdd_index_insert_only(env, child, pfid, S_IFDIR,
1877 mdo_ref_del(env, child, handle);
1884 * This function checks whether it can create a file/dir under the
1885 * directory(@pobj). The directory(@pobj) is not being locked by
1888 * \param[in] env execution environment
1889 * \param[in] pobj the directory to create files
1890 * \param[in] pattr the attributes of the directory
1891 * \param[in] lname the name of the created file/dir
1892 * \param[in] cattr the attributes of the file/dir
1893 * \param[in] spec create specification
1895 * \retval = 0 it is allowed to create file/dir under
1897 * \retval negative error not allowed to create file/dir
1898 * under the directory
1900 static int mdd_create_sanity_check(const struct lu_env *env,
1901 struct md_object *pobj,
1902 const struct lu_attr *pattr,
1903 const struct lu_name *lname,
1904 struct lu_attr *cattr,
1905 struct md_op_spec *spec)
1907 struct mdd_thread_info *info = mdd_env_info(env);
1908 struct lu_fid *fid = &info->mti_fid;
1909 struct mdd_object *obj = md2mdd_obj(pobj);
1910 struct mdd_device *m = mdo2mdd(pobj);
1911 bool check_perm = true;
1916 if (mdd_is_dead_obj(obj))
1920 * In some cases this lookup is not needed - we know before if name
1921 * exists or not because MDT performs lookup for it.
1922 * name length check is done in lookup.
1924 if (spec->sp_cr_lookup) {
1926 * Check if the name already exist, though it will be checked in
1927 * _index_insert also, for avoiding rolling back if exists
1930 rc = __mdd_lookup(env, pobj, pattr, lname, fid,
1931 MAY_WRITE | MAY_EXEC);
1933 RETURN(rc ? : -EEXIST);
1935 /* Permission is already being checked in mdd_lookup */
1939 if (S_ISDIR(cattr->la_mode) &&
1940 unlikely(spec != NULL && spec->sp_cr_flags & MDS_OPEN_HAS_EA) &&
1941 spec->u.sp_ea.eadata != NULL && spec->u.sp_ea.eadatalen > 0) {
1942 const struct lmv_user_md *lum = spec->u.sp_ea.eadata;
1944 if (unlikely(le32_to_cpu(lum->lum_magic) != LMV_USER_MAGIC) &&
1945 le32_to_cpu(lum->lum_magic) != LMV_USER_MAGIC_V0) {
1947 CERROR("%s: invalid lmv_user_md: magic = %x, "
1948 "stripe_offset = %d, stripe_count = %u: "
1949 "rc = %d\n", mdd2obd_dev(m)->obd_name,
1950 le32_to_cpu(lum->lum_magic),
1951 (int)le32_to_cpu(lum->lum_stripe_offset),
1952 le32_to_cpu(lum->lum_stripe_count), rc);
1957 rc = mdd_may_create(env, obj, pattr, NULL, check_perm);
1962 if (pattr->la_mode & S_ISGID) {
1963 cattr->la_gid = pattr->la_gid;
1964 if (S_ISDIR(cattr->la_mode)) {
1965 cattr->la_mode |= S_ISGID;
1966 cattr->la_valid |= LA_MODE;
1970 /* Inherit project ID from parent directory */
1971 if (pattr->la_flags & LUSTRE_PROJINHERIT_FL) {
1972 cattr->la_projid = pattr->la_projid;
1973 if (S_ISDIR(cattr->la_mode)) {
1974 cattr->la_flags |= LUSTRE_PROJINHERIT_FL;
1975 cattr->la_valid |= LA_FLAGS;
1977 cattr->la_valid |= LA_PROJID;
1980 rc = mdd_name_check(m, lname);
1984 switch (cattr->la_mode & S_IFMT) {
1986 unsigned int symlen = strlen(spec->u.sp_symname) + 1;
1988 if (symlen > m->mdd_dt_conf.ddp_symlink_max)
1989 RETURN(-ENAMETOOLONG);
2008 static int mdd_declare_create_object(const struct lu_env *env,
2009 struct mdd_device *mdd,
2010 struct mdd_object *p, struct mdd_object *c,
2011 struct lu_attr *attr,
2012 struct thandle *handle,
2013 const struct md_op_spec *spec,
2014 struct lu_buf *def_acl_buf,
2015 struct lu_buf *acl_buf,
2016 struct dt_allocation_hint *hint)
2018 const struct lu_buf *buf;
2021 rc = mdd_declare_create_object_internal(env, p, c, attr, handle, spec,
2026 #ifdef CONFIG_FS_POSIX_ACL
2027 if (def_acl_buf->lb_len > 0 && S_ISDIR(attr->la_mode)) {
2028 /* if dir, then can inherit default ACl */
2029 rc = mdo_declare_xattr_set(env, c, def_acl_buf,
2030 XATTR_NAME_ACL_DEFAULT,
2036 if (acl_buf->lb_len > 0) {
2037 rc = mdo_declare_attr_set(env, c, attr, handle);
2041 rc = mdo_declare_xattr_set(env, c, acl_buf,
2042 XATTR_NAME_ACL_ACCESS, 0, handle);
2047 rc = mdd_declare_object_initialize(env, p, c, attr, handle);
2051 /* replay case, create LOV EA from client data */
2052 if (spec->no_create ||
2053 (spec->sp_cr_flags & MDS_OPEN_HAS_EA && S_ISREG(attr->la_mode))) {
2054 buf = mdd_buf_get_const(env, spec->u.sp_ea.eadata,
2055 spec->u.sp_ea.eadatalen);
2056 rc = mdo_declare_xattr_set(env, c, buf, XATTR_NAME_LOV, 0,
2062 if (S_ISLNK(attr->la_mode)) {
2063 const char *target_name = spec->u.sp_symname;
2064 int sym_len = strlen(target_name);
2065 const struct lu_buf *buf;
2067 buf = mdd_buf_get_const(env, target_name, sym_len);
2068 rc = dt_declare_record_write(env, mdd_object_child(c),
2074 if (spec->sp_cr_file_secctx_name != NULL) {
2075 buf = mdd_buf_get_const(env, spec->sp_cr_file_secctx,
2076 spec->sp_cr_file_secctx_size);
2077 rc = mdo_declare_xattr_set(env, c, buf,
2078 spec->sp_cr_file_secctx_name, 0,
2087 static int mdd_declare_create(const struct lu_env *env, struct mdd_device *mdd,
2088 struct mdd_object *p, struct mdd_object *c,
2089 const struct lu_name *name,
2090 struct lu_attr *attr,
2091 struct thandle *handle,
2092 const struct md_op_spec *spec,
2093 struct linkea_data *ldata,
2094 struct lu_buf *def_acl_buf,
2095 struct lu_buf *acl_buf,
2096 struct dt_allocation_hint *hint)
2100 rc = mdd_declare_create_object(env, mdd, p, c, attr, handle, spec,
2101 def_acl_buf, acl_buf, hint);
2105 if (S_ISDIR(attr->la_mode)) {
2106 rc = mdo_declare_ref_add(env, p, handle);
2111 if (unlikely(spec->sp_cr_flags & MDS_OPEN_VOLATILE)) {
2112 rc = orph_declare_index_insert(env, c, attr->la_mode, handle);
2116 struct lu_attr *la = &mdd_env_info(env)->mti_la_for_fix;
2118 rc = mdo_declare_index_insert(env, p, mdo2fid(c), attr->la_mode,
2119 name->ln_name, handle);
2123 rc = mdd_declare_links_add(env, c, handle, ldata);
2128 la->la_valid = LA_CTIME | LA_MTIME;
2129 rc = mdo_declare_attr_set(env, p, la, handle);
2133 rc = mdd_declare_changelog_store(env, mdd, name, NULL, handle);
2141 static int mdd_acl_init(const struct lu_env *env, struct mdd_object *pobj,
2142 struct lu_attr *la, struct lu_buf *def_acl_buf,
2143 struct lu_buf *acl_buf)
2148 if (S_ISLNK(la->la_mode)) {
2149 acl_buf->lb_len = 0;
2150 def_acl_buf->lb_len = 0;
2154 mdd_read_lock(env, pobj, MOR_TGT_PARENT);
2155 rc = mdo_xattr_get(env, pobj, def_acl_buf,
2156 XATTR_NAME_ACL_DEFAULT);
2157 mdd_read_unlock(env, pobj);
2159 /* If there are default ACL, fix mode/ACL by default ACL */
2160 def_acl_buf->lb_len = rc;
2161 LASSERT(def_acl_buf->lb_len <= acl_buf->lb_len);
2162 memcpy(acl_buf->lb_buf, def_acl_buf->lb_buf, rc);
2163 acl_buf->lb_len = rc;
2164 rc = __mdd_fix_mode_acl(env, acl_buf, &la->la_mode);
2167 } else if (rc == -ENODATA || rc == -EOPNOTSUPP) {
2168 /* If there are no default ACL, fix mode by mask */
2169 struct lu_ucred *uc = lu_ucred(env);
2171 /* The create triggered by MDT internal events, such as
2172 * LFSCK reset, will not contain valid "uc". */
2173 if (unlikely(uc != NULL))
2174 la->la_mode &= ~uc->uc_umask;
2176 acl_buf->lb_len = 0;
2177 def_acl_buf->lb_len = 0;
2184 * Create a metadata object and initialize it, set acl, xattr.
2186 static int mdd_create_object(const struct lu_env *env, struct mdd_object *pobj,
2187 struct mdd_object *son, struct lu_attr *attr,
2188 struct md_op_spec *spec, struct lu_buf *acl_buf,
2189 struct lu_buf *def_acl_buf,
2190 struct dt_allocation_hint *hint,
2191 struct thandle *handle)
2193 const struct lu_buf *buf;
2196 mdd_write_lock(env, son, MOR_TGT_CHILD);
2197 rc = mdd_create_object_internal(env, NULL, son, attr, handle, spec,
2202 /* Note: In DNE phase I, for striped dir, though sub-stripes will be
2203 * created in declare phase, they also needs to be added to master
2204 * object as sub-directory entry. So it has to initialize the master
2205 * object, then set dir striped EA.(in mdo_xattr_set) */
2206 rc = mdd_object_initialize(env, mdo2fid(pobj), son, attr, handle,
2209 GOTO(err_destroy, rc);
2212 * in case of replay we just set LOVEA provided by the client
2213 * XXX: I think it would be interesting to try "old" way where
2214 * MDT calls this xattr_set(LOV) in a different transaction.
2215 * probably this way we code can be made better.
2218 /* During creation, there are only a few cases we need do xattr_set to
2220 * 1. regular file: see comments above.
2221 * 2. dir: inherit default striping or pool settings from parent.
2222 * 3. create striped directory with provided stripeEA.
2223 * 4. create striped directory because inherit default layout from the
2226 if (spec->no_create ||
2227 (S_ISREG(attr->la_mode) && spec->sp_cr_flags & MDS_OPEN_HAS_EA) ||
2228 S_ISDIR(attr->la_mode)) {
2229 buf = mdd_buf_get_const(env, spec->u.sp_ea.eadata,
2230 spec->u.sp_ea.eadatalen);
2231 rc = mdo_xattr_set(env, son, buf,
2232 S_ISDIR(attr->la_mode) ? XATTR_NAME_LMV :
2236 GOTO(err_destroy, rc);
2239 #ifdef CONFIG_FS_POSIX_ACL
2240 if (def_acl_buf != NULL && def_acl_buf->lb_len > 0 &&
2241 S_ISDIR(attr->la_mode)) {
2242 /* set default acl */
2243 rc = mdo_xattr_set(env, son, def_acl_buf,
2244 XATTR_NAME_ACL_DEFAULT, 0,
2247 GOTO(err_destroy, rc);
2249 /* set its own acl */
2250 if (acl_buf != NULL && acl_buf->lb_len > 0) {
2251 rc = mdo_xattr_set(env, son, acl_buf,
2252 XATTR_NAME_ACL_ACCESS,
2255 GOTO(err_destroy, rc);
2259 if (S_ISLNK(attr->la_mode)) {
2260 struct lu_ucred *uc = lu_ucred_assert(env);
2261 struct dt_object *dt = mdd_object_child(son);
2262 const char *target_name = spec->u.sp_symname;
2263 int sym_len = strlen(target_name);
2266 buf = mdd_buf_get_const(env, target_name, sym_len);
2267 rc = dt->do_body_ops->dbo_write(env, dt, buf, &pos, handle,
2269 CFS_CAP_SYS_RESOURCE_MASK);
2274 GOTO(err_initlized, rc = -EFAULT);
2277 if (spec->sp_cr_file_secctx_name != NULL) {
2278 buf = mdd_buf_get_const(env, spec->sp_cr_file_secctx,
2279 spec->sp_cr_file_secctx_size);
2280 rc = mdo_xattr_set(env, son, buf, spec->sp_cr_file_secctx_name,
2283 GOTO(err_initlized, rc);
2287 if (unlikely(rc != 0)) {
2289 if (S_ISDIR(attr->la_mode)) {
2290 /* Drop the reference, no need to delete "."/"..",
2291 * because the object to be destroied directly. */
2292 rc2 = mdo_ref_del(env, son, handle);
2296 rc2 = mdo_ref_del(env, son, handle);
2300 mdo_destroy(env, son, handle);
2303 mdd_write_unlock(env, son);
2307 static int mdd_index_delete(const struct lu_env *env,
2308 struct mdd_object *mdd_pobj,
2309 struct lu_attr *cattr,
2310 const struct lu_name *lname)
2312 struct mdd_device *mdd = mdo2mdd(&mdd_pobj->mod_obj);
2313 struct thandle *handle;
2317 handle = mdd_trans_create(env, mdd);
2319 RETURN(PTR_ERR(handle));
2321 rc = mdo_declare_index_delete(env, mdd_pobj, lname->ln_name,
2326 if (S_ISDIR(cattr->la_mode)) {
2327 rc = mdo_declare_ref_del(env, mdd_pobj, handle);
2332 /* Since this will only be used in the error handler path,
2333 * Let's set the thandle to be local and not mess the transno */
2334 handle->th_local = 1;
2335 rc = mdd_trans_start(env, mdd, handle);
2339 rc = __mdd_index_delete(env, mdd_pobj, lname->ln_name,
2340 S_ISDIR(cattr->la_mode), handle);
2344 rc = mdd_trans_stop(env, mdd, rc, handle);
2350 * Create object and insert it into namespace.
2352 * Two operations have to be performed:
2354 * - an allocation of a new object (->do_create()), and
2355 * - an insertion into a parent index (->dio_insert()).
2357 * Due to locking, operation order is not important, when both are
2358 * successful, *but* error handling cases are quite different:
2360 * - if insertion is done first, and following object creation fails,
2361 * insertion has to be rolled back, but this operation might fail
2362 * also leaving us with dangling index entry.
2364 * - if creation is done first, is has to be undone if insertion fails,
2365 * leaving us with leaked space, which is not good but not fatal.
2367 * It seems that creation-first is simplest solution, but it is sub-optimal
2373 * case, because second mkdir is bound to create object, only to
2374 * destroy it immediately.
2376 * To avoid this follow local file systems that do double lookup:
2378 * 0. lookup -> -EEXIST (mdd_create_sanity_check())
2379 * 1. create (mdd_create_object_internal())
2380 * 2. insert (__mdd_index_insert(), lookup again)
2382 * \param[in] pobj parent object
2383 * \param[in] lname name of child being created
2384 * \param[in,out] child child object being created
2385 * \param[in] spec additional create parameters
2386 * \param[in] ma attributes for new child object
2388 * \retval 0 on success
2389 * \retval negative errno on failure
2391 static int mdd_create(const struct lu_env *env, struct md_object *pobj,
2392 const struct lu_name *lname, struct md_object *child,
2393 struct md_op_spec *spec, struct md_attr *ma)
2395 struct mdd_thread_info *info = mdd_env_info(env);
2396 struct lu_attr *la = &info->mti_la_for_fix;
2397 struct mdd_object *mdd_pobj = md2mdd_obj(pobj);
2398 struct mdd_object *son = md2mdd_obj(child);
2399 struct mdd_device *mdd = mdo2mdd(pobj);
2400 struct lu_attr *attr = &ma->ma_attr;
2401 struct thandle *handle;
2402 struct lu_attr *pattr = &info->mti_pattr;
2403 struct lu_buf acl_buf;
2404 struct lu_buf def_acl_buf;
2405 struct linkea_data *ldata = &info->mti_link_data;
2406 const char *name = lname->ln_name;
2407 struct dt_allocation_hint *hint = &mdd_env_info(env)->mti_hint;
2412 rc = mdd_la_get(env, mdd_pobj, pattr);
2416 /* Sanity checks before big job. */
2417 rc = mdd_create_sanity_check(env, pobj, pattr, lname, attr, spec);
2421 if (OBD_FAIL_CHECK(OBD_FAIL_MDS_DQACQ_NET))
2422 GOTO(out_free, rc = -EINPROGRESS);
2424 handle = mdd_trans_create(env, mdd);
2426 GOTO(out_free, rc = PTR_ERR(handle));
2428 lu_buf_check_and_alloc(&info->mti_xattr_buf,
2429 mdd->mdd_dt_conf.ddp_max_ea_size);
2430 acl_buf = info->mti_xattr_buf;
2431 def_acl_buf.lb_buf = info->mti_key;
2432 def_acl_buf.lb_len = sizeof(info->mti_key);
2433 rc = mdd_acl_init(env, mdd_pobj, attr, &def_acl_buf, &acl_buf);
2437 mdd_object_make_hint(env, mdd_pobj, son, attr, spec, hint);
2439 memset(ldata, 0, sizeof(*ldata));
2440 if (OBD_FAIL_CHECK(OBD_FAIL_LFSCK_BAD_PARENT)) {
2441 struct lu_fid tfid = *mdd_object_fid(mdd_pobj);
2444 rc = mdd_linkea_prepare(env, son, NULL, NULL,
2445 &tfid, lname, 1, 0, ldata);
2447 rc = mdd_linkea_prepare(env, son, NULL, NULL,
2448 mdd_object_fid(mdd_pobj),
2449 lname, 1, 0, ldata);
2452 rc = mdd_declare_create(env, mdd, mdd_pobj, son, lname, attr,
2453 handle, spec, ldata, &def_acl_buf, &acl_buf,
2458 rc = mdd_trans_start(env, mdd, handle);
2462 rc = mdd_create_object(env, mdd_pobj, son, attr, spec, &acl_buf,
2463 &def_acl_buf, hint, handle);
2467 if (unlikely(spec->sp_cr_flags & MDS_OPEN_VOLATILE)) {
2468 mdd_write_lock(env, son, MOR_TGT_CHILD);
2469 son->mod_flags |= VOLATILE_OBJ;
2470 rc = __mdd_orphan_add(env, son, handle);
2471 GOTO(out_volatile, rc);
2473 rc = __mdd_index_insert(env, mdd_pobj, mdo2fid(son),
2474 attr->la_mode, name, handle);
2476 GOTO(err_created, rc);
2478 mdd_links_add(env, son, mdo2fid(mdd_pobj), lname, handle,
2481 /* update parent directory mtime/ctime */
2483 la->la_valid = LA_CTIME | LA_MTIME;
2484 rc = mdd_update_time(env, mdd_pobj, pattr, la, handle);
2486 GOTO(err_insert, rc);
2492 if (spec->sp_cr_flags & MDS_OPEN_VOLATILE)
2493 rc2 = __mdd_orphan_del(env, son, handle);
2495 rc2 = __mdd_index_delete(env, mdd_pobj, name,
2496 S_ISDIR(attr->la_mode),
2502 mdd_write_lock(env, son, MOR_TGT_CHILD);
2503 if (S_ISDIR(attr->la_mode)) {
2504 /* Drop the reference, no need to delete "."/"..",
2505 * because the object is to be destroyed directly. */
2506 rc2 = mdo_ref_del(env, son, handle);
2508 mdd_write_unlock(env, son);
2513 /* For volatile files drop one link immediately, since there is
2514 * no filename in the namespace, and save any error returned. */
2515 rc2 = mdo_ref_del(env, son, handle);
2517 mdd_write_unlock(env, son);
2518 if (unlikely(rc == 0))
2523 /* Don't destroy the volatile object on success */
2524 if (likely(rc != 0))
2525 mdo_destroy(env, son, handle);
2526 mdd_write_unlock(env, son);
2529 if (rc == 0 && fid_is_namespace_visible(mdo2fid(son)) &&
2530 likely((spec->sp_cr_flags & MDS_OPEN_VOLATILE) == 0))
2531 rc = mdd_changelog_ns_store(env, mdd,
2532 S_ISDIR(attr->la_mode) ? CL_MKDIR :
2533 S_ISREG(attr->la_mode) ? CL_CREATE :
2534 S_ISLNK(attr->la_mode) ? CL_SOFTLINK : CL_MKNOD,
2535 0, son, mdo2fid(mdd_pobj), NULL, NULL, lname,
2538 rc2 = mdd_trans_stop(env, mdd, rc, handle);
2540 /* If creation fails, it is most likely due to the remote update
2541 * failure, because local transaction will mostly succeed at
2542 * this stage. There is no easy way to rollback all of previous
2543 * updates, so let's remove the object from namespace, and
2544 * LFSCK should handle the orphan object. */
2545 if (rc2 < 0 && !mdd_object_remote(mdd_pobj))
2546 mdd_index_delete(env, mdd_pobj, attr, lname);
2550 if (is_vmalloc_addr(ldata->ld_buf))
2551 /* if we vmalloced a large buffer drop it */
2552 lu_buf_free(ldata->ld_buf);
2554 /* The child object shouldn't be cached anymore */
2556 set_bit(LU_OBJECT_HEARD_BANSHEE,
2557 &child->mo_lu.lo_header->loh_flags);
2562 * Get locks on parents in proper order
2563 * RETURN: < 0 - error, rename_order if successful
2571 static int mdd_rename_order(const struct lu_env *env,
2572 struct mdd_device *mdd,
2573 struct mdd_object *src_pobj,
2574 const struct lu_attr *pattr,
2575 struct mdd_object *tgt_pobj)
2577 /* order of locking, 1 - tgt-src, 0 - src-tgt*/
2581 if (src_pobj == tgt_pobj)
2582 RETURN(MDD_RN_SAME);
2584 /* compared the parent child relationship of src_p&tgt_p */
2585 if (lu_fid_eq(&mdd->mdd_root_fid, mdo2fid(src_pobj))){
2587 } else if (lu_fid_eq(&mdd->mdd_root_fid, mdo2fid(tgt_pobj))) {
2590 rc = mdd_is_parent(env, mdd, src_pobj, pattr, mdo2fid(tgt_pobj),
2604 /* has not mdd_write{read}_lock on any obj yet. */
2605 static int mdd_rename_sanity_check(const struct lu_env *env,
2606 struct mdd_object *src_pobj,
2607 const struct lu_attr *pattr,
2608 struct mdd_object *tgt_pobj,
2609 const struct lu_attr *tpattr,
2610 struct mdd_object *sobj,
2611 const struct lu_attr *cattr,
2612 struct mdd_object *tobj,
2613 const struct lu_attr *tattr)
2618 /* XXX: when get here, sobj must NOT be NULL,
2619 * the other case has been processed in cld_rename
2620 * before mdd_rename and enable MDS_PERM_BYPASS. */
2624 * If we are using project inheritance, we only allow renames
2625 * into our tree when the project IDs are the same; otherwise
2626 * tree quota mechanism would be circumvented.
2628 if (((tpattr->la_flags & LUSTRE_PROJINHERIT_FL) &&
2629 tpattr->la_projid != cattr->la_projid) ||
2630 ((pattr->la_flags & LUSTRE_PROJINHERIT_FL) &&
2631 (pattr->la_projid != tpattr->la_projid)))
2634 rc = mdd_may_delete(env, src_pobj, pattr, sobj, cattr, NULL, 1, 0);
2638 /* XXX: when get here, "tobj == NULL" means tobj must
2639 * NOT exist (neither on remote MDS, such case has been
2640 * processed in cld_rename before mdd_rename and enable
2642 * So check may_create, but not check may_unlink. */
2644 rc = mdd_may_create(env, tgt_pobj, tpattr, NULL,
2645 (src_pobj != tgt_pobj));
2647 rc = mdd_may_delete(env, tgt_pobj, tpattr, tobj, tattr, cattr,
2648 (src_pobj != tgt_pobj), 1);
2650 if (!rc && !tobj && (src_pobj != tgt_pobj) && S_ISDIR(cattr->la_mode))
2651 rc = __mdd_may_link(env, tgt_pobj, tpattr);
2656 static int mdd_declare_rename(const struct lu_env *env,
2657 struct mdd_device *mdd,
2658 struct mdd_object *mdd_spobj,
2659 struct mdd_object *mdd_tpobj,
2660 struct mdd_object *mdd_sobj,
2661 struct mdd_object *mdd_tobj,
2662 const struct lu_name *sname,
2663 const struct lu_name *tname,
2665 struct linkea_data *ldata,
2666 struct thandle *handle)
2668 struct lu_attr *la = &mdd_env_info(env)->mti_la_for_fix;
2671 LASSERT(ma->ma_attr.la_valid & LA_CTIME);
2672 la->la_ctime = la->la_mtime = ma->ma_attr.la_ctime;
2678 /* name from source dir */
2679 rc = mdo_declare_index_delete(env, mdd_spobj, sname->ln_name, handle);
2683 /* .. from source child */
2684 if (S_ISDIR(mdd_object_type(mdd_sobj))) {
2685 /* source child can be directory,
2686 * counted by source dir's nlink */
2687 rc = mdo_declare_ref_del(env, mdd_spobj, handle);
2690 if (mdd_spobj != mdd_tpobj) {
2691 rc = mdo_declare_index_delete(env, mdd_sobj, dotdot,
2696 rc = mdo_declare_index_insert(env, mdd_sobj,
2698 S_IFDIR, dotdot, handle);
2703 /* new target child can be directory,
2704 * counted by target dir's nlink */
2705 rc = mdo_declare_ref_add(env, mdd_tpobj, handle);
2710 la->la_valid = LA_CTIME | LA_MTIME;
2711 rc = mdo_declare_attr_set(env, mdd_spobj, la, handle);
2715 rc = mdo_declare_attr_set(env, mdd_tpobj, la, handle);
2719 la->la_valid = LA_CTIME;
2720 rc = mdo_declare_attr_set(env, mdd_sobj, la, handle);
2724 rc = mdd_declare_links_add(env, mdd_sobj, handle, ldata);
2729 rc = mdo_declare_index_insert(env, mdd_tpobj, mdo2fid(mdd_sobj),
2730 mdd_object_type(mdd_sobj),
2731 tname->ln_name, handle);
2735 if (mdd_tobj && mdd_object_exists(mdd_tobj)) {
2736 /* delete target child in target parent directory */
2737 rc = mdo_declare_index_delete(env, mdd_tpobj, tname->ln_name,
2742 rc = mdo_declare_ref_del(env, mdd_tobj, handle);
2746 if (S_ISDIR(mdd_object_type(mdd_tobj))) {
2747 /* target child can be directory,
2748 * delete "." reference in target child directory */
2749 rc = mdo_declare_ref_del(env, mdd_tobj, handle);
2753 /* delete ".." reference in target parent directory */
2754 rc = mdo_declare_ref_del(env, mdd_tpobj, handle);
2759 la->la_valid = LA_CTIME;
2760 rc = mdo_declare_attr_set(env, mdd_tobj, la, handle);
2764 rc = mdd_declare_finish_unlink(env, mdd_tobj, handle);
2769 rc = mdd_declare_changelog_store(env, mdd, tname, sname, handle);
2776 /* src object can be remote that is why we use only fid and type of object */
2777 static int mdd_rename(const struct lu_env *env,
2778 struct md_object *src_pobj, struct md_object *tgt_pobj,
2779 const struct lu_fid *lf, const struct lu_name *lsname,
2780 struct md_object *tobj, const struct lu_name *ltname,
2783 const char *sname = lsname->ln_name;
2784 const char *tname = ltname->ln_name;
2785 struct lu_attr *la = &mdd_env_info(env)->mti_la_for_fix;
2786 struct mdd_object *mdd_spobj = md2mdd_obj(src_pobj); /* source parent */
2787 struct mdd_object *mdd_tpobj = md2mdd_obj(tgt_pobj);
2788 struct mdd_device *mdd = mdo2mdd(src_pobj);
2789 struct mdd_object *mdd_sobj = NULL; /* source object */
2790 struct mdd_object *mdd_tobj = NULL;
2791 struct lu_attr *cattr = MDD_ENV_VAR(env, cattr);
2792 struct lu_attr *pattr = MDD_ENV_VAR(env, pattr);
2793 struct lu_attr *tattr = MDD_ENV_VAR(env, tattr);
2794 struct lu_attr *tpattr = MDD_ENV_VAR(env, tpattr);
2795 struct thandle *handle;
2796 struct linkea_data *ldata = &mdd_env_info(env)->mti_link_data;
2797 const struct lu_fid *tpobj_fid = mdo2fid(mdd_tpobj);
2798 const struct lu_fid *spobj_fid = mdo2fid(mdd_spobj);
2801 bool tobj_locked = 0;
2802 unsigned cl_flags = 0;
2807 mdd_tobj = md2mdd_obj(tobj);
2809 mdd_sobj = mdd_object_find(env, mdd, lf);
2810 if (IS_ERR(mdd_sobj))
2811 RETURN(PTR_ERR(mdd_sobj));
2813 rc = mdd_la_get(env, mdd_sobj, cattr);
2815 GOTO(out_pending, rc);
2817 rc = mdd_la_get(env, mdd_spobj, pattr);
2819 GOTO(out_pending, rc);
2822 rc = mdd_la_get(env, mdd_tobj, tattr);
2824 GOTO(out_pending, rc);
2825 /* search for an existing archive.
2826 * we should check ahead as the object
2827 * can be destroyed in this transaction */
2828 if (mdd_hsm_archive_exists(env, mdd_tobj, ma))
2829 cl_flags |= CLF_RENAME_LAST_EXISTS;
2832 rc = mdd_la_get(env, mdd_tpobj, tpattr);
2834 GOTO(out_pending, rc);
2836 rc = mdd_rename_sanity_check(env, mdd_spobj, pattr, mdd_tpobj, tpattr,
2837 mdd_sobj, cattr, mdd_tobj, tattr);
2839 GOTO(out_pending, rc);
2841 rc = mdd_name_check(mdd, ltname);
2843 GOTO(out_pending, rc);
2845 /* FIXME: Should consider tobj and sobj too in rename_lock. */
2846 rc = mdd_rename_order(env, mdd, mdd_spobj, pattr, mdd_tpobj);
2848 GOTO(out_pending, rc);
2850 handle = mdd_trans_create(env, mdd);
2852 GOTO(out_pending, rc = PTR_ERR(handle));
2854 memset(ldata, 0, sizeof(*ldata));
2855 rc = mdd_linkea_prepare(env, mdd_sobj, mdd_object_fid(mdd_spobj),
2856 lsname, mdd_object_fid(mdd_tpobj), ltname,
2861 rc = mdd_declare_rename(env, mdd, mdd_spobj, mdd_tpobj, mdd_sobj,
2862 mdd_tobj, lsname, ltname, ma, ldata, handle);
2866 rc = mdd_trans_start(env, mdd, handle);
2870 is_dir = S_ISDIR(cattr->la_mode);
2872 /* Remove source name from source directory */
2873 rc = __mdd_index_delete(env, mdd_spobj, sname, is_dir, handle);
2877 /* "mv dir1 dir2" needs "dir1/.." link update */
2878 if (is_dir && !lu_fid_eq(spobj_fid, tpobj_fid)) {
2879 rc = __mdd_index_delete_only(env, mdd_sobj, dotdot, handle);
2881 GOTO(fixup_spobj2, rc);
2883 rc = __mdd_index_insert_only(env, mdd_sobj, tpobj_fid, S_IFDIR,
2886 GOTO(fixup_spobj, rc);
2889 if (mdd_tobj != NULL && mdd_object_exists(mdd_tobj)) {
2890 rc = __mdd_index_delete(env, mdd_tpobj, tname, is_dir, handle);
2892 /* tname might been renamed to something else */
2893 GOTO(fixup_spobj, rc);
2896 /* Insert new fid with target name into target dir */
2897 rc = __mdd_index_insert(env, mdd_tpobj, lf, cattr->la_mode,
2900 GOTO(fixup_tpobj, rc);
2902 LASSERT(ma->ma_attr.la_valid & LA_CTIME);
2903 la->la_ctime = la->la_mtime = ma->ma_attr.la_ctime;
2905 /* XXX: mdd_sobj must be local one if it is NOT NULL. */
2906 la->la_valid = LA_CTIME;
2907 rc = mdd_update_time(env, mdd_sobj, cattr, la, handle);
2909 GOTO(fixup_tpobj, rc);
2911 /* Update the linkEA for the source object */
2912 mdd_write_lock(env, mdd_sobj, MOR_SRC_CHILD);
2913 rc = mdd_links_rename(env, mdd_sobj, mdo2fid(mdd_spobj), lsname,
2914 mdo2fid(mdd_tpobj), ltname, handle, ldata,
2917 /* Old files might not have EA entry */
2918 mdd_links_add(env, mdd_sobj, mdo2fid(mdd_spobj),
2919 lsname, handle, NULL, 0);
2920 mdd_write_unlock(env, mdd_sobj);
2921 /* We don't fail the transaction if the link ea can't be
2922 updated -- fid2path will use alternate lookup method. */
2925 /* Remove old target object
2926 * For tobj is remote case cmm layer has processed
2927 * and set tobj to NULL then. So when tobj is NOT NULL,
2928 * it must be local one.
2930 if (tobj && mdd_object_exists(mdd_tobj)) {
2931 mdd_write_lock(env, mdd_tobj, MOR_TGT_CHILD);
2933 if (mdd_is_dead_obj(mdd_tobj)) {
2934 /* shld not be dead, something is wrong */
2935 CERROR("tobj is dead, something is wrong\n");
2939 mdo_ref_del(env, mdd_tobj, handle);
2941 /* Remove dot reference. */
2942 if (S_ISDIR(tattr->la_mode))
2943 mdo_ref_del(env, mdd_tobj, handle);
2946 /* fetch updated nlink */
2947 rc = mdd_la_get(env, mdd_tobj, tattr);
2949 CERROR("%s: Failed to get nlink for tobj "
2951 mdd2obd_dev(mdd)->obd_name,
2952 PFID(tpobj_fid), rc);
2953 GOTO(fixup_tpobj, rc);
2956 la->la_valid = LA_CTIME;
2957 rc = mdd_update_time(env, mdd_tobj, tattr, la, handle);
2959 CERROR("%s: Failed to set ctime for tobj "
2961 mdd2obd_dev(mdd)->obd_name,
2962 PFID(tpobj_fid), rc);
2963 GOTO(fixup_tpobj, rc);
2966 /* XXX: this transfer to ma will be removed with LOD/OSP */
2967 ma->ma_attr = *tattr;
2968 ma->ma_valid |= MA_INODE;
2969 rc = mdd_finish_unlink(env, mdd_tobj, ma, mdd_tpobj, ltname,
2972 CERROR("%s: Failed to unlink tobj "
2974 mdd2obd_dev(mdd)->obd_name,
2975 PFID(tpobj_fid), rc);
2976 GOTO(fixup_tpobj, rc);
2979 /* fetch updated nlink */
2980 rc = mdd_la_get(env, mdd_tobj, tattr);
2981 if (rc == -ENOENT) {
2982 /* the object got removed, let's
2983 * return the latest known attributes */
2984 tattr->la_nlink = 0;
2986 } else if (rc != 0) {
2987 CERROR("%s: Failed to get nlink for tobj "
2989 mdd2obd_dev(mdd)->obd_name,
2990 PFID(tpobj_fid), rc);
2991 GOTO(fixup_tpobj, rc);
2993 /* XXX: this transfer to ma will be removed with LOD/OSP */
2994 ma->ma_attr = *tattr;
2995 ma->ma_valid |= MA_INODE;
2997 if (tattr->la_nlink == 0)
2998 cl_flags |= CLF_RENAME_LAST;
3000 cl_flags &= ~CLF_RENAME_LAST_EXISTS;
3003 la->la_valid = LA_CTIME | LA_MTIME;
3004 rc = mdd_update_time(env, mdd_spobj, pattr, la, handle);
3006 GOTO(fixup_tpobj, rc);
3008 if (mdd_spobj != mdd_tpobj) {
3009 la->la_valid = LA_CTIME | LA_MTIME;
3010 rc = mdd_update_time(env, mdd_tpobj, tpattr, la, handle);
3012 GOTO(fixup_tpobj, rc);
3019 rc2 = __mdd_index_delete(env, mdd_tpobj, tname, is_dir, handle);
3021 CWARN("tp obj fix error %d\n",rc2);
3023 if (mdd_tobj && mdd_object_exists(mdd_tobj) &&
3024 !mdd_is_dead_obj(mdd_tobj)) {
3026 mdo_ref_add(env, mdd_tobj, handle);
3028 mdo_ref_add(env, mdd_tobj, handle);
3031 rc2 = __mdd_index_insert(env, mdd_tpobj,
3033 mdd_object_type(mdd_tobj),
3036 CWARN("tp obj fix error: rc = %d\n", rc2);
3041 if (rc && is_dir && mdd_sobj && mdd_spobj != mdd_tpobj) {
3042 rc2 = __mdd_index_delete_only(env, mdd_sobj, dotdot, handle);
3044 CWARN("%s: sp obj dotdot delete error: rc = %d\n",
3045 mdd2obd_dev(mdd)->obd_name, rc2);
3048 rc2 = __mdd_index_insert_only(env, mdd_sobj, spobj_fid, S_IFDIR,
3051 CWARN("%s: sp obj dotdot insert error: rc = %d\n",
3052 mdd2obd_dev(mdd)->obd_name, rc2);
3057 rc2 = __mdd_index_insert(env, mdd_spobj, lf,
3058 mdd_object_type(mdd_sobj), sname,
3061 CWARN("sp obj fix error: rc = %d\n", rc2);
3066 mdd_write_unlock(env, mdd_tobj);
3069 rc = mdd_changelog_ns_store(env, mdd, CL_RENAME, cl_flags,
3070 mdd_tobj, tpobj_fid, lf, spobj_fid,
3071 ltname, lsname, handle);
3074 rc = mdd_trans_stop(env, mdd, rc, handle);
3077 mdd_object_put(env, mdd_sobj);
3082 * During migration once the parent FID has been changed,
3083 * we need update the parent FID in linkea.
3085 static int mdd_linkea_update_child_internal(const struct lu_env *env,
3086 struct mdd_object *parent,
3087 struct mdd_object *newparent,
3088 struct mdd_object *child,
3089 const char *name, int namelen,
3090 struct thandle *handle,
3093 struct mdd_thread_info *info = mdd_env_info(env);
3094 struct linkea_data ldata = { NULL };
3095 struct lu_buf *buf = &info->mti_link_buf;
3101 buf = lu_buf_check_and_alloc(buf, PATH_MAX);
3102 if (buf->lb_buf == NULL)
3106 rc = mdd_links_read(env, child, &ldata);
3108 if (rc == -ENOENT || rc == -ENODATA)
3113 LASSERT(ldata.ld_leh != NULL);
3114 ldata.ld_lee = (struct link_ea_entry *)(ldata.ld_leh + 1);
3115 for (count = 0; count < ldata.ld_leh->leh_reccount; count++) {
3116 struct mdd_device *mdd = mdo2mdd(&child->mod_obj);
3117 struct lu_name lname;
3120 linkea_entry_unpack(ldata.ld_lee, &ldata.ld_reclen,
3123 if (strncmp(lname.ln_name, name, namelen) != 0 ||
3124 !lu_fid_eq(&fid, mdd_object_fid(parent))) {
3125 ldata.ld_lee = (struct link_ea_entry *)
3126 ((char *)ldata.ld_lee +
3131 CDEBUG(D_INFO, "%s: update "DFID" with %.*s:"DFID"\n",
3132 mdd2obd_dev(mdd)->obd_name, PFID(mdd_object_fid(child)),
3133 lname.ln_namelen, lname.ln_name,
3134 PFID(mdd_object_fid(newparent)));
3135 /* update to the new parent fid */
3136 linkea_entry_pack(ldata.ld_lee, &lname,
3137 mdd_object_fid(newparent));
3139 rc = mdd_declare_links_add(env, child, handle, &ldata);
3141 rc = mdd_links_write(env, child, &ldata, handle);
3147 static int mdd_linkea_declare_update_child(const struct lu_env *env,
3148 struct mdd_object *parent,
3149 struct mdd_object *newparent,
3150 struct mdd_object *child,
3151 const char *name, int namelen,
3152 struct thandle *handle)
3154 return mdd_linkea_update_child_internal(env, parent, newparent,
3156 namelen, handle, true);
3159 static int mdd_linkea_update_child(const struct lu_env *env,
3160 struct mdd_object *parent,
3161 struct mdd_object *newparent,
3162 struct mdd_object *child,
3163 const char *name, int namelen,
3164 struct thandle *handle)
3166 return mdd_linkea_update_child_internal(env, parent, newparent,
3168 namelen, handle, false);
3171 static int mdd_update_linkea_internal(const struct lu_env *env,
3172 struct mdd_object *mdd_pobj,
3173 struct mdd_object *mdd_sobj,
3174 struct mdd_object *mdd_tobj,
3175 const struct lu_name *child_name,
3176 struct linkea_data *ldata,
3177 struct thandle *handle,
3180 struct mdd_thread_info *info = mdd_env_info(env);
3185 LASSERT(ldata->ld_buf != NULL);
3186 LASSERT(ldata->ld_leh != NULL);
3188 /* If it is mulitple links file, we need update the name entry for
3190 ldata->ld_lee = (struct link_ea_entry *)(ldata->ld_leh + 1);
3191 for (count = 0; count < ldata->ld_leh->leh_reccount; count++) {
3192 struct mdd_device *mdd = mdo2mdd(&mdd_sobj->mod_obj);
3193 struct mdd_object *pobj;
3194 struct lu_name lname;
3197 linkea_entry_unpack(ldata->ld_lee, &ldata->ld_reclen,
3199 pobj = mdd_object_find(env, mdd, &fid);
3201 CWARN("%s: cannot find obj "DFID": rc = %ld\n",
3202 mdd2obd_dev(mdd)->obd_name, PFID(&fid),
3207 if (!mdd_object_exists(pobj)) {
3208 CDEBUG(D_INFO, "%s: obj "DFID" does not exist\n",
3209 mdd2obd_dev(mdd)->obd_name, PFID(&fid));
3213 if (pobj == mdd_pobj &&
3214 lname.ln_namelen == child_name->ln_namelen &&
3215 strncmp(lname.ln_name, child_name->ln_name,
3216 lname.ln_namelen) == 0) {
3217 CDEBUG(D_INFO, "%s: skip its own %s: "DFID"\n",
3218 mdd2obd_dev(mdd)->obd_name, child_name->ln_name,
3223 CDEBUG(D_INFO, "%s: update "DFID" with "DNAME":"DFID"\n",
3224 mdd2obd_dev(mdd)->obd_name, PFID(mdd_object_fid(pobj)),
3225 PNAME(&lname), PFID(mdd_object_fid(mdd_tobj)));
3228 /* Remove source name from source directory */
3229 /* Insert new fid with target name into target dir */
3230 rc = mdo_declare_index_delete(env, pobj, lname.ln_name,
3235 rc = mdo_declare_index_insert(env, pobj,
3236 mdd_object_fid(mdd_tobj),
3237 mdd_object_type(mdd_tobj),
3238 lname.ln_name, handle);
3242 rc = mdo_declare_ref_add(env, mdd_tobj, handle);
3246 rc = mdo_declare_ref_del(env, mdd_sobj, handle);
3250 char *tmp_name = info->mti_key;
3252 if (lname.ln_namelen >= sizeof(info->mti_key)) {
3253 /* lnamelen is too big(> NAME_MAX + 16),
3254 * something wrong about this linkea, let's
3256 CWARN("%s: the name %.*s is too long under "
3257 DFID"\n", mdd2obd_dev(mdd)->obd_name,
3258 lname.ln_namelen, lname.ln_name,
3263 /* Note: lname might be without \0 at the end, see
3264 * linkea_entry_unpack(), let's add extra \0 by
3266 snprintf(tmp_name, sizeof(info->mti_key), "%.*s",
3267 lname.ln_namelen, lname.ln_name);
3268 lname.ln_name = tmp_name;
3270 /* Let's check if this linkEA still valid, before
3271 * it might be packed into the RPC buffer. */
3272 rc = mdd_lookup(env, &pobj->mod_obj, &lname,
3273 &info->mti_fid, NULL);
3274 if (rc < 0 || !lu_fid_eq(&info->mti_fid,
3275 mdd_object_fid(mdd_sobj)))
3276 GOTO(next_put, rc == -ENOENT ? 0 : rc);
3278 rc = __mdd_index_delete(env, pobj, tmp_name, 0, handle);
3282 rc = __mdd_index_insert(env, pobj,
3283 mdd_object_fid(mdd_tobj),
3284 mdd_object_type(mdd_tobj),
3289 mdd_write_lock(env, mdd_tobj, MOR_SRC_CHILD);
3290 rc = mdo_ref_add(env, mdd_tobj, handle);
3291 mdd_write_unlock(env, mdd_tobj);
3295 mdd_write_lock(env, mdd_sobj, MOR_TGT_CHILD);
3296 mdo_ref_del(env, mdd_sobj, handle);
3297 mdd_write_unlock(env, mdd_sobj);
3300 mdd_object_put(env, pobj);
3304 ldata->ld_lee = (struct link_ea_entry *)((char *)ldata->ld_lee +
3311 static int mdd_migrate_xattrs(const struct lu_env *env,
3312 struct mdd_object *mdd_sobj,
3313 struct mdd_object *mdd_tobj)
3315 struct mdd_thread_info *info = mdd_env_info(env);
3316 struct mdd_device *mdd = mdo2mdd(&mdd_sobj->mod_obj);
3318 struct thandle *handle;
3324 struct lu_buf list_xbuf;
3327 /* retrieve xattr list from the old object */
3328 list_xsize = mdo_xattr_list(env, mdd_sobj, &LU_BUF_NULL);
3329 if (list_xsize == -ENODATA)
3335 lu_buf_check_and_alloc(&info->mti_big_buf, list_xsize);
3336 if (info->mti_big_buf.lb_buf == NULL)
3339 list_xbuf.lb_buf = info->mti_big_buf.lb_buf;
3340 list_xbuf.lb_len = list_xsize;
3341 rc = mdo_xattr_list(env, mdd_sobj, &list_xbuf);
3346 xname = list_xbuf.lb_buf;
3348 xlen = strnlen(xname, rem - 1) + 1;
3349 if (strcmp(XATTR_NAME_LMA, xname) == 0 ||
3350 strcmp(XATTR_NAME_LMV, xname) == 0)
3353 /* For directory, if there are default layout, migrate here */
3354 if (strcmp(XATTR_NAME_LOV, xname) == 0 &&
3355 !S_ISDIR(lu_object_attr(&mdd_sobj->mod_obj.mo_lu)))
3358 xsize = mdo_xattr_get(env, mdd_sobj, &LU_BUF_NULL, xname);
3359 if (xsize == -ENODATA)
3364 lu_buf_check_and_alloc(&info->mti_link_buf, xsize);
3365 if (info->mti_link_buf.lb_buf == NULL)
3366 GOTO(out, rc = -ENOMEM);
3368 xbuf.lb_len = xsize;
3369 xbuf.lb_buf = info->mti_link_buf.lb_buf;
3370 rc = mdo_xattr_get(env, mdd_sobj, &xbuf, xname);
3376 handle = mdd_trans_create(env, mdd);
3378 GOTO(out, rc = PTR_ERR(handle));
3380 rc = mdo_declare_xattr_set(env, mdd_tobj, &xbuf, xname, 0,
3383 GOTO(stop_trans, rc);
3384 /* Note: this transaction is part of migration, and it is not
3385 * the last step of migration, so we set th_local = 1 to avoid
3386 * update last rcvd for this transaction */
3387 handle->th_local = 1;
3388 rc = mdd_trans_start(env, mdd, handle);
3390 GOTO(stop_trans, rc);
3393 rc = mdo_xattr_set(env, mdd_tobj, &xbuf, xname, 0, handle);
3395 GOTO(stop_trans, rc = 0);
3397 if (unlikely(rc == -ENOSPC &&
3398 strcmp(xname, XATTR_NAME_LINK) == 0)) {
3399 rc = linkea_overflow_shrink(
3400 (struct linkea_data *)(xbuf.lb_buf));
3401 if (likely(rc > 0)) {
3408 GOTO(stop_trans, rc);
3410 rc = mdd_trans_stop(env, mdd, rc, handle);
3415 memmove(xname, xname + xlen, rem);
3421 static int mdd_declare_migrate_create(const struct lu_env *env,
3422 struct mdd_object *mdd_pobj,
3423 struct mdd_object *mdd_sobj,
3424 struct mdd_object *mdd_tobj,
3425 struct md_op_spec *spec,
3427 union lmv_mds_md *mgr_ea,
3428 struct linkea_data *ldata,
3429 struct thandle *handle)
3431 struct lu_attr *la_flag = MDD_ENV_VAR(env, la_for_fix);
3432 const struct lu_buf *buf;
3436 rc = mdd_declare_create_object_internal(env, mdd_pobj, mdd_tobj, la,
3437 handle, spec, NULL);
3441 rc = mdd_declare_object_initialize(env, mdd_pobj, mdd_tobj, la,
3446 if (S_ISLNK(la->la_mode)) {
3447 const char *target_name = spec->u.sp_symname;
3448 int sym_len = strlen(target_name);
3449 const struct lu_buf *buf;
3451 buf = mdd_buf_get_const(env, target_name, sym_len);
3452 rc = dt_declare_record_write(env, mdd_object_child(mdd_tobj),
3456 } else if (S_ISDIR(la->la_mode) && ldata != NULL) {
3457 rc = mdd_declare_links_add(env, mdd_tobj, handle, ldata);
3462 if (spec->u.sp_ea.eadata != NULL && spec->u.sp_ea.eadatalen != 0) {
3463 buf = mdd_buf_get_const(env, spec->u.sp_ea.eadata,
3464 spec->u.sp_ea.eadatalen);
3465 rc = mdo_declare_xattr_set(env, mdd_tobj, buf, XATTR_NAME_LOV,
3471 mgr_easize = lmv_mds_md_size(2, LMV_MAGIC_V1);
3472 buf = mdd_buf_get_const(env, mgr_ea, mgr_easize);
3473 rc = mdo_declare_xattr_set(env, mdd_sobj, buf, XATTR_NAME_LMV,
3478 la_flag->la_valid = LA_FLAGS;
3479 la_flag->la_flags = la->la_flags | LUSTRE_IMMUTABLE_FL;
3480 rc = mdo_declare_attr_set(env, mdd_sobj, la_flag, handle);
3485 static int mdd_migrate_create(const struct lu_env *env,
3486 struct mdd_object *mdd_pobj,
3487 struct mdd_object *mdd_sobj,
3488 struct mdd_object *mdd_tobj,
3489 const struct lu_name *lname,
3492 struct mdd_thread_info *info = mdd_env_info(env);
3493 struct mdd_device *mdd = mdo2mdd(&mdd_sobj->mod_obj);
3494 struct md_op_spec *spec = &info->mti_spec;
3495 struct lu_buf lmm_buf = { NULL };
3496 struct lu_buf link_buf = { NULL };
3497 struct lu_buf mgr_buf;
3498 struct thandle *handle;
3499 struct lmv_mds_md_v1 *mgr_ea;
3500 struct lu_attr *la_flag = MDD_ENV_VAR(env, la_for_fix);
3501 struct dt_allocation_hint *hint = &mdd_env_info(env)->mti_hint;
3503 struct linkea_data *ldata = &mdd_env_info(env)->mti_link_data;
3507 /* prepare spec for create */
3508 memset(spec, 0, sizeof(*spec));
3509 spec->sp_cr_lookup = 0;
3510 spec->sp_feat = &dt_directory_features;
3511 if (S_ISLNK(la->la_mode)) {
3512 const struct lu_buf *buf;
3514 buf = lu_buf_check_and_alloc(
3515 &mdd_env_info(env)->mti_big_buf,
3518 link_buf.lb_len = la->la_size + 1;
3519 memset(link_buf.lb_buf, 0, link_buf.lb_len);
3520 rc = mdd_readlink(env, &mdd_sobj->mod_obj, &link_buf);
3522 rc = rc != 0 ? rc : -EFAULT;
3523 CERROR("%s: "DFID" readlink failed: rc = %d\n",
3524 mdd2obd_dev(mdd)->obd_name,
3525 PFID(mdd_object_fid(mdd_sobj)), rc);
3528 spec->u.sp_symname = link_buf.lb_buf;
3529 } else if (S_ISREG(la->la_mode)) {
3530 /* retrieve lov of the old object */
3531 rc = mdd_get_lov_ea(env, mdd_sobj, &lmm_buf);
3532 if (rc != 0 && rc != -ENODATA)
3534 if (lmm_buf.lb_buf != NULL && lmm_buf.lb_len != 0) {
3535 spec->u.sp_ea.eadata = lmm_buf.lb_buf;
3536 spec->u.sp_ea.eadatalen = lmm_buf.lb_len;
3537 spec->sp_cr_flags |= MDS_OPEN_HAS_EA;
3539 } else if (S_ISDIR(la->la_mode)) {
3540 rc = mdd_links_read_with_rec(env, mdd_sobj, ldata);
3541 if (rc == -ENODATA) {
3542 /* ignore the non-linkEA error */
3550 mgr_easize = lmv_mds_md_size(2, LMV_MAGIC_V1);
3551 lu_buf_check_and_alloc(&info->mti_xattr_buf, mgr_easize);
3552 mgr_buf.lb_buf = info->mti_xattr_buf.lb_buf;
3553 mgr_buf.lb_len = mgr_easize;
3554 mgr_ea = mgr_buf.lb_buf;
3555 memset(mgr_ea, 0, sizeof(*mgr_ea));
3556 mgr_ea->lmv_magic = cpu_to_le32(LMV_MAGIC_V1);
3557 mgr_ea->lmv_stripe_count = cpu_to_le32(2);
3558 mgr_ea->lmv_master_mdt_index = mdd_seq_site(mdd)->ss_node_id;
3559 mgr_ea->lmv_hash_type = cpu_to_le32(LMV_HASH_FLAG_MIGRATION);
3560 fid_cpu_to_le(&mgr_ea->lmv_stripe_fids[0], mdd_object_fid(mdd_sobj));
3561 fid_cpu_to_le(&mgr_ea->lmv_stripe_fids[1], mdd_object_fid(mdd_tobj));
3563 mdd_object_make_hint(env, mdd_pobj, mdd_tobj, la, spec, hint);
3565 handle = mdd_trans_create(env, mdd);
3567 GOTO(out_free, rc = PTR_ERR(handle));
3569 /* Note: this transaction is part of migration, and it is not
3570 * the last step of migration, so we set th_local = 1 to avoid
3571 * update last rcvd for this transaction */
3572 handle->th_local = 1;
3573 rc = mdd_declare_migrate_create(env, mdd_pobj, mdd_sobj, mdd_tobj, spec,
3574 la, mgr_buf.lb_buf, ldata, handle);
3576 GOTO(stop_trans, rc);
3578 rc = mdd_trans_start(env, mdd, handle);
3580 GOTO(stop_trans, rc);
3582 /* don't set nlink from the original object */
3583 la->la_valid &= ~LA_NLINK;
3585 /* create the target object */
3586 rc = mdd_create_object(env, mdd_pobj, mdd_tobj, la, spec, NULL, NULL,
3589 GOTO(stop_trans, rc);
3591 if (S_ISDIR(la->la_mode) && ldata != NULL) {
3592 rc = mdd_links_write(env, mdd_tobj, ldata, handle);
3594 GOTO(stop_trans, rc);
3597 /* Set MIGRATE EA on the source inode, so once the migration needs
3598 * to be re-done during failover, the re-do process can locate the
3599 * target object which is already being created. */
3600 rc = mdo_xattr_set(env, mdd_sobj, &mgr_buf, XATTR_NAME_LMV, 0, handle);
3602 GOTO(stop_trans, rc);
3604 /* Set immutable flag, so any modification is disabled until
3605 * the migration is done. Once the migration is interrupted,
3606 * if the resume process find the migrating object has both
3607 * IMMUTALBE flag and MIGRATE EA, it need to clear IMMUTABLE
3608 * flag and approve the migration */
3609 la_flag->la_valid = LA_FLAGS;
3610 la_flag->la_flags = la->la_flags | LUSTRE_IMMUTABLE_FL;
3611 rc = mdo_attr_set(env, mdd_sobj, la_flag, handle);
3614 rc = mdd_trans_stop(env, mdd, rc, handle);
3616 if (lmm_buf.lb_buf != NULL)
3617 OBD_FREE(lmm_buf.lb_buf, lmm_buf.lb_len);
3621 static int mdd_migrate_entries(const struct lu_env *env,
3622 struct mdd_object *mdd_sobj,
3623 struct mdd_object *mdd_tobj)
3625 struct dt_object *next = mdd_object_child(mdd_sobj);
3626 struct mdd_device *mdd = mdo2mdd(&mdd_sobj->mod_obj);
3627 struct dt_object *dt_tobj = mdd_object_child(mdd_tobj);
3628 struct thandle *handle;
3630 const struct dt_it_ops *iops;
3632 struct lu_dirent *ent;
3636 OBD_ALLOC(ent, NAME_MAX + sizeof(*ent) + 1);
3640 if (!dt_try_as_dir(env, next))
3641 GOTO(out_ent, rc = -ENOTDIR);
3643 * iterate directories
3645 iops = &next->do_index_ops->dio_it;
3646 it = iops->init(env, next, LUDA_FID | LUDA_TYPE);
3648 GOTO(out_ent, rc = PTR_ERR(it));
3650 rc = iops->load(env, it, 0);
3652 rc = iops->next(env, it);
3656 * At this point and across for-loop:
3658 * rc == 0 -> ok, proceed.
3659 * rc > 0 -> end of directory.
3663 struct mdd_object *child;
3664 char *name = mdd_env_info(env)->mti_key;
3667 bool target_exist = false;
3669 len = iops->key_size(env, it);
3673 result = iops->rec(env, it, (struct dt_rec *)ent,
3674 LUDA_FID | LUDA_TYPE);
3675 if (result == -ESTALE)
3682 fid_le_to_cpu(&ent->lde_fid, &ent->lde_fid);
3684 /* Insert new fid with target name into target dir */
3685 if ((ent->lde_namelen == 1 && ent->lde_name[0] == '.') ||
3686 (ent->lde_namelen == 2 && ent->lde_name[0] == '.' &&
3687 ent->lde_name[1] == '.'))
3690 child = mdd_object_find(env, mdd, &ent->lde_fid);
3692 GOTO(out, rc = PTR_ERR(child));
3694 mdd_write_lock(env, child, MOR_SRC_CHILD);
3695 is_dir = S_ISDIR(mdd_object_type(child));
3697 snprintf(name, ent->lde_namelen + 1, "%s", ent->lde_name);
3699 /* Check whether the name has been inserted to the target */
3700 if (dt_try_as_dir(env, dt_tobj)) {
3701 struct lu_fid *fid = &mdd_env_info(env)->mti_fid2;
3703 rc = dt_lookup(env, dt_tobj, (struct dt_rec *)fid,
3704 (struct dt_key *)name);
3705 if (unlikely(rc == 0))
3706 target_exist = true;
3709 handle = mdd_trans_create(env, mdd);
3711 GOTO(out_put, rc = PTR_ERR(handle));
3713 /* Note: this transaction is part of migration, and it is not
3714 * the last step of migration, so we set th_local = 1 to avoid
3715 * updating last rcvd for this transaction */
3716 handle->th_local = 1;
3717 if (likely(!target_exist)) {
3718 rc = mdo_declare_index_insert(env, mdd_tobj,
3720 mdd_object_type(child),
3726 rc = mdo_declare_ref_add(env, mdd_tobj, handle);
3732 rc = mdo_declare_index_delete(env, mdd_sobj, name, handle);
3737 rc = mdo_declare_ref_del(env, mdd_sobj, handle);
3741 /* Update .. for child */
3742 rc = mdo_declare_index_delete(env, child, dotdot,
3747 rc = mdo_declare_index_insert(env, child,
3748 mdd_object_fid(mdd_tobj),
3749 S_IFDIR, dotdot, handle);
3754 rc = mdd_linkea_declare_update_child(env, mdd_sobj,mdd_tobj,
3761 rc = mdd_trans_start(env, mdd, handle);
3763 CERROR("%s: transaction start failed: rc = %d\n",
3764 mdd2obd_dev(mdd)->obd_name, rc);
3768 if (likely(!target_exist)) {
3769 rc = __mdd_index_insert(env, mdd_tobj, &ent->lde_fid,
3770 mdd_object_type(child),
3776 rc = __mdd_index_delete(env, mdd_sobj, name, is_dir, handle);
3781 rc = __mdd_index_delete_only(env, child, dotdot,
3786 rc = __mdd_index_insert_only(env, child,
3787 mdd_object_fid(mdd_tobj), S_IFDIR,
3793 rc = mdd_linkea_update_child(env, mdd_sobj, mdd_tobj,
3795 strlen(name), handle);
3798 mdd_write_unlock(env, child);
3799 mdd_object_put(env, child);
3800 rc = mdd_trans_stop(env, mdd, rc, handle);
3804 result = iops->next(env, it);
3805 if (OBD_FAIL_CHECK(OBD_FAIL_MIGRATE_ENTRIES))
3806 GOTO(out, rc = -EINTR);
3808 if (result == -ESTALE)
3810 } while (result == 0);
3813 iops->fini(env, it);
3815 OBD_FREE(ent, NAME_MAX + sizeof(*ent) + 1);
3819 static int mdd_declare_update_linkea(const struct lu_env *env,
3820 struct mdd_object *mdd_pobj,
3821 struct mdd_object *mdd_sobj,
3822 struct mdd_object *mdd_tobj,
3823 const struct lu_name *child_name,
3824 struct linkea_data *ldata,
3825 struct thandle *handle)
3827 return mdd_update_linkea_internal(env, mdd_pobj, mdd_sobj, mdd_tobj,
3828 child_name, ldata, handle, 1);
3831 static int mdd_update_linkea(const struct lu_env *env,
3832 struct mdd_object *mdd_pobj,
3833 struct mdd_object *mdd_sobj,
3834 struct mdd_object *mdd_tobj,
3835 const struct lu_name *child_name,
3836 struct linkea_data *ldata,
3837 struct thandle *handle)
3839 return mdd_update_linkea_internal(env, mdd_pobj, mdd_sobj, mdd_tobj,
3840 child_name, ldata, handle, 0);
3843 static int mdd_declare_migrate_update_name(const struct lu_env *env,
3844 struct mdd_object *mdd_pobj,
3845 struct mdd_object *mdd_sobj,
3846 struct mdd_object *mdd_tobj,
3847 const struct lu_name *lname,
3849 struct lu_attr *parent_la,
3850 struct linkea_data *ldata,
3851 struct thandle *handle)
3853 struct mdd_device *mdd = mdo2mdd(&mdd_sobj->mod_obj);
3854 struct lu_attr *la_flag = MDD_ENV_VAR(env, tattr);
3857 /* Revert IMMUTABLE flag */
3858 la_flag->la_valid = LA_FLAGS;
3859 la_flag->la_flags = la->la_flags & ~LUSTRE_IMMUTABLE_FL;
3860 rc = mdo_declare_attr_set(env, mdd_sobj, la_flag, handle);
3864 /* delete entry from source dir */
3865 rc = mdo_declare_index_delete(env, mdd_pobj, lname->ln_name, handle);
3869 if (ldata->ld_buf != NULL) {
3870 rc = mdd_declare_update_linkea(env, mdd_pobj, mdd_sobj,
3871 mdd_tobj, lname, ldata, handle);
3876 if (S_ISREG(mdd_object_type(mdd_sobj))) {
3877 rc = mdo_declare_xattr_del(env, mdd_sobj, XATTR_NAME_LOV,
3882 handle->th_complex = 1;
3883 rc = mdo_declare_xattr_set(env, mdd_tobj, NULL,
3885 LU_XATTR_REPLACE, handle);
3890 if (S_ISDIR(mdd_object_type(mdd_sobj))) {
3891 rc = mdo_declare_ref_del(env, mdd_pobj, handle);
3897 rc = mdo_declare_index_insert(env, mdd_pobj, mdo2fid(mdd_tobj),
3898 mdd_object_type(mdd_tobj),
3899 lname->ln_name, handle);
3903 rc = mdd_declare_links_add(env, mdd_tobj, handle, NULL);
3907 if (S_ISDIR(mdd_object_type(mdd_sobj))) {
3908 rc = mdo_declare_ref_add(env, mdd_pobj, handle);
3913 /* delete old object */
3914 rc = mdo_declare_ref_del(env, mdd_sobj, handle);
3918 if (S_ISDIR(mdd_object_type(mdd_sobj))) {
3919 /* delete old object */
3920 rc = mdo_declare_ref_del(env, mdd_sobj, handle);
3923 /* set nlink to 0 */
3924 rc = mdo_declare_attr_set(env, mdd_sobj, la, handle);
3929 rc = mdd_declare_finish_unlink(env, mdd_sobj, handle);
3933 rc = mdo_declare_attr_set(env, mdd_pobj, parent_la, handle);
3937 rc = mdd_declare_changelog_store(env, mdd, lname, NULL, handle);
3942 static int mdd_migrate_update_name(const struct lu_env *env,
3943 struct mdd_object *mdd_pobj,
3944 struct mdd_object *mdd_sobj,
3945 struct mdd_object *mdd_tobj,
3946 const struct lu_name *lname,
3949 struct lu_attr *p_la = MDD_ENV_VAR(env, la_for_fix);
3950 struct lu_attr *so_attr = MDD_ENV_VAR(env, cattr);
3951 struct lu_attr *la_flag = MDD_ENV_VAR(env, tattr);
3952 struct mdd_device *mdd = mdo2mdd(&mdd_sobj->mod_obj);
3953 struct linkea_data *ldata = &mdd_env_info(env)->mti_link_data;
3954 struct thandle *handle;
3955 int is_dir = S_ISDIR(mdd_object_type(mdd_sobj));
3956 const char *name = lname->ln_name;
3960 /* update time for parent */
3961 LASSERT(ma->ma_attr.la_valid & LA_CTIME);
3962 p_la->la_ctime = p_la->la_mtime = ma->ma_attr.la_ctime;
3963 p_la->la_valid = LA_CTIME;
3965 rc = mdd_la_get(env, mdd_sobj, so_attr);
3969 ldata->ld_buf = NULL;
3970 rc = mdd_links_read(env, mdd_sobj, ldata);
3971 if (rc != 0 && rc != -ENOENT && rc != -ENODATA)
3974 handle = mdd_trans_create(env, mdd);
3976 RETURN(PTR_ERR(handle));
3978 rc = mdd_declare_migrate_update_name(env, mdd_pobj, mdd_sobj, mdd_tobj,
3979 lname, so_attr, p_la, ldata,
3982 /* If the migration can not be fit in one transaction, just
3983 * leave it in the original MDT */
3985 GOTO(stop_trans, rc = 0);
3987 GOTO(stop_trans, rc);
3990 CDEBUG(D_INFO, "%s: update "DFID"/"DFID" with %s:"DFID"\n",
3991 mdd2obd_dev(mdd)->obd_name, PFID(mdd_object_fid(mdd_pobj)),
3992 PFID(mdd_object_fid(mdd_sobj)), lname->ln_name,
3993 PFID(mdd_object_fid(mdd_tobj)));
3995 rc = mdd_trans_start(env, mdd, handle);
3997 GOTO(stop_trans, rc);
3999 /* Revert IMMUTABLE flag */
4000 la_flag->la_valid = LA_FLAGS;
4001 la_flag->la_flags = so_attr->la_flags & ~LUSTRE_IMMUTABLE_FL;
4002 rc = mdo_attr_set(env, mdd_sobj, la_flag, handle);
4004 GOTO(stop_trans, rc);
4006 /* Remove source name from source directory */
4007 rc = __mdd_index_delete(env, mdd_pobj, name, is_dir, handle);
4009 GOTO(stop_trans, rc);
4011 if (ldata->ld_buf != NULL) {
4012 rc = mdd_update_linkea(env, mdd_pobj, mdd_sobj, mdd_tobj,
4013 lname, ldata, handle);
4015 GOTO(stop_trans, rc);
4017 /* linkea update might decrease the source object
4018 * nlink, let's get the attr again after ref_del */
4019 rc = mdd_la_get(env, mdd_sobj, so_attr);
4021 GOTO(stop_trans, rc);
4024 if (S_ISREG(so_attr->la_mode)) {
4025 if (so_attr->la_nlink == 1) {
4026 rc = mdo_xattr_del(env, mdd_sobj, XATTR_NAME_LOV,
4028 if (rc != 0 && rc != -ENODATA)
4029 GOTO(stop_trans, rc);
4031 rc = mdo_xattr_set(env, mdd_tobj, NULL,
4033 LU_XATTR_REPLACE, handle);
4035 GOTO(stop_trans, rc);
4039 /* Insert new fid with target name into target dir */
4040 rc = __mdd_index_insert(env, mdd_pobj, mdd_object_fid(mdd_tobj),
4041 mdd_object_type(mdd_tobj), name, handle);
4043 GOTO(stop_trans, rc);
4045 mdd_write_lock(env, mdd_sobj, MOR_TGT_CHILD);
4047 mdd_sobj->mod_flags |= DEAD_OBJ;
4048 rc = mdd_mark_orphan_object(env, mdd_sobj, handle, false);
4050 GOTO(out_unlock, rc);
4052 rc = __mdd_orphan_add(env, mdd_sobj, handle);
4054 GOTO(out_unlock, rc);
4056 mdo_ref_del(env, mdd_sobj, handle);
4058 mdo_ref_del(env, mdd_sobj, handle);
4060 /* Get the attr again after ref_del */
4061 rc = mdd_la_get(env, mdd_sobj, so_attr);
4063 GOTO(out_unlock, rc);
4065 ma->ma_attr = *so_attr;
4066 ma->ma_valid |= MA_INODE;
4068 rc = mdd_attr_set_internal(env, mdd_pobj, p_la, handle, 0);
4070 GOTO(out_unlock, rc);
4072 rc = mdd_changelog_ns_store(env, mdd, CL_MIGRATE, 0, mdd_tobj,
4073 mdo2fid(mdd_pobj), mdo2fid(mdd_sobj),
4074 mdo2fid(mdd_pobj), lname, lname, handle);
4076 CWARN("%s: changelog for migrate %s "DFID
4077 "under "DFID" failed: rc = %d\n",
4078 mdd2obd_dev(mdd)->obd_name, lname->ln_name,
4079 PFID(mdd_object_fid(mdd_sobj)),
4080 PFID(mdd_object_fid(mdd_pobj)), rc);
4081 /* Sigh, there are no easy way to migrate back the object, so
4082 * let's reset the result to 0 for now XXX */
4086 mdd_write_unlock(env, mdd_sobj);
4089 rc = mdd_trans_stop(env, mdd, rc, handle);
4094 static int mdd_fld_lookup(const struct lu_env *env, struct mdd_device *mdd,
4095 const struct lu_fid *fid, __u32 *mdt_index)
4097 struct lu_seq_range *range = &mdd_env_info(env)->mti_range;
4098 struct seq_server_site *ss;
4101 ss = mdd->mdd_md_dev.md_lu_dev.ld_site->ld_seq_site;
4103 range->lsr_flags = LU_SEQ_RANGE_MDT;
4104 rc = fld_server_lookup(env, ss->ss_server_fld, fid->f_seq, range);
4108 *mdt_index = range->lsr_index;
4113 * Check whether we should migrate the file/dir
4115 * < 0 permission check failed or other error.
4116 * = 0 the file can be migrated.
4117 * > 0 the file does not need to be migrated, mostly
4118 * for multiple link file
4120 static int mdd_migrate_sanity_check(const struct lu_env *env,
4121 struct mdd_object *pobj,
4122 const struct lu_attr *pattr,
4123 struct mdd_object *sobj,
4124 struct lu_attr *sattr)
4126 struct mdd_thread_info *info = mdd_env_info(env);
4127 struct linkea_data *ldata = &info->mti_link_data;
4128 struct mdd_device *mdd = mdo2mdd(&pobj->mod_obj);
4130 struct lu_buf *mgr_buf;
4136 mgr_easize = lmv_mds_md_size(2, LMV_MAGIC_V1);
4137 mgr_buf = lu_buf_check_and_alloc(&info->mti_big_buf, mgr_easize);
4138 if (mgr_buf->lb_buf == NULL)
4141 rc = mdo_xattr_get(env, sobj, mgr_buf, XATTR_NAME_LMV);
4143 union lmv_mds_md *lmm = mgr_buf->lb_buf;
4145 /* If the object has migrateEA, it means IMMUTE flag
4146 * is being set by previous migration process, so it
4147 * needs to override the IMMUTE flag, otherwise the
4148 * following sanity check will fail */
4149 if (le32_to_cpu(lmm->lmv_md_v1.lmv_hash_type) &
4150 LMV_HASH_FLAG_MIGRATION) {
4151 struct mdd_device *mdd = mdo2mdd(&sobj->mod_obj);
4153 sattr->la_flags &= ~LUSTRE_IMMUTABLE_FL;
4154 CDEBUG(D_HA, "%s: "DFID" override IMMUTE FLAG\n",
4155 mdd2obd_dev(mdd)->obd_name,
4156 PFID(mdd_object_fid(sobj)));
4160 rc = mdd_rename_sanity_check(env, pobj, pattr, pobj, pattr,
4161 sobj, sattr, NULL, NULL);
4165 /* Then it will check if the file should be migrated. If the file
4166 * has mulitple links, we only need migrate the file if all of its
4167 * entries has been migrated to the remote MDT */
4168 if (!S_ISREG(sattr->la_mode) || sattr->la_nlink < 2)
4171 rc = mdd_links_read(env, sobj, ldata);
4173 /* For multiple links files, if there are no linkEA data at all,
4174 * means the file might be created before linkEA is enabled, and
4175 * all of its links should not be migrated yet, otherwise it
4176 * should have some linkEA there */
4177 if (rc == -ENOENT || rc == -ENODATA)
4182 mdt_index = mdd->mdd_md_dev.md_lu_dev.ld_site->ld_seq_site->ss_node_id;
4183 /* If there are still links locally, then the file will not be
4185 LASSERT(ldata->ld_leh != NULL);
4187 /* If the linkEA is overflow, then means there are some unknown name
4188 * entries under unknown parents, that will prevent the migration. */
4189 if (unlikely(ldata->ld_leh->leh_overflow_time))
4192 ldata->ld_lee = (struct link_ea_entry *)(ldata->ld_leh + 1);
4193 for (count = 0; count < ldata->ld_leh->leh_reccount; count++) {
4194 struct lu_name lname;
4196 __u32 parent_mdt_index;
4198 linkea_entry_unpack(ldata->ld_lee, &ldata->ld_reclen,
4200 ldata->ld_lee = (struct link_ea_entry *)((char *)ldata->ld_lee +
4203 rc = mdd_fld_lookup(env, mdd, &fid, &parent_mdt_index);
4207 /* Migrate the object only if none of its parents are on the
4209 if (parent_mdt_index != mdt_index)
4212 CDEBUG(D_INFO, DFID"still has local entry %.*s "DFID"\n",
4213 PFID(mdd_object_fid(sobj)), lname.ln_namelen,
4214 lname.ln_name, PFID(&fid));
4222 static int mdd_migrate(const struct lu_env *env, struct md_object *pobj,
4223 struct md_object *sobj, const struct lu_name *lname,
4224 struct md_object *tobj, struct md_attr *ma)
4226 struct mdd_object *mdd_pobj = md2mdd_obj(pobj);
4227 struct mdd_device *mdd = mdo2mdd(pobj);
4228 struct mdd_object *mdd_sobj = md2mdd_obj(sobj);
4229 struct mdd_object *mdd_tobj = md2mdd_obj(tobj);
4230 struct lu_attr *so_attr = MDD_ENV_VAR(env, cattr);
4231 struct lu_attr *pattr = MDD_ENV_VAR(env, pattr);
4232 bool created = false;
4236 /* If the file will being migrated, it will check whether
4237 * the file is being opened by someone else right now */
4238 mdd_read_lock(env, mdd_sobj, MOR_SRC_CHILD);
4239 if (mdd_sobj->mod_count > 0) {
4241 "%s: "DFID"%s is already opened count %d: rc = %d\n",
4242 mdd2obd_dev(mdd)->obd_name,
4243 PFID(mdd_object_fid(mdd_sobj)), lname->ln_name,
4244 mdd_sobj->mod_count, -EBUSY);
4245 mdd_read_unlock(env, mdd_sobj);
4246 GOTO(put, rc = -EBUSY);
4248 mdd_read_unlock(env, mdd_sobj);
4250 rc = mdd_la_get(env, mdd_sobj, so_attr);
4254 rc = mdd_la_get(env, mdd_pobj, pattr);
4258 rc = mdd_migrate_sanity_check(env, mdd_pobj, pattr, mdd_sobj, so_attr);
4265 /* Sigh, it is impossible to finish all of migration in a single
4266 * transaction, for example migrating big directory entries to the
4267 * new MDT, it needs insert all of name entries of children in the
4270 * So migration will be done in multiple steps and transactions.
4272 * 1. create an orphan object on the remote MDT in one transaction.
4273 * 2. migrate extend attributes to the new target file/directory.
4274 * 3. For directory, migrate the entries to the new MDT and update
4275 * linkEA of each children. Because we can not migrate all entries
4276 * in a single transaction, so the migrating directory will become
4277 * a striped directory during migration, so once the process is
4278 * interrupted, the directory is still accessible. (During lookup,
4279 * client will locate the name by searching both original and target
4281 * 4. Finally, update the name/FID to point to the new file/directory
4282 * in a separate transaction.
4285 /* step 1: Check whether the orphan object has been created, and create
4286 * orphan object on the remote MDT if needed */
4287 if (!mdd_object_exists(mdd_tobj)) {
4288 rc = mdd_migrate_create(env, mdd_pobj, mdd_sobj, mdd_tobj,
4295 LASSERT(mdd_object_exists(mdd_tobj));
4296 /* step 2: migrate xattr */
4297 rc = mdd_migrate_xattrs(env, mdd_sobj, mdd_tobj);
4301 /* step 3: migrate name entries to the orphan object */
4302 if (S_ISDIR(lu_object_attr(&mdd_sobj->mod_obj.mo_lu))) {
4303 rc = mdd_migrate_entries(env, mdd_sobj, mdd_tobj);
4306 if (unlikely(OBD_FAIL_CHECK_RESET(OBD_FAIL_MIGRATE_NET_REP,
4307 OBD_FAIL_MDS_REINT_NET_REP)))
4310 OBD_FAIL_TIMEOUT(OBD_FAIL_MIGRATE_DELAY, cfs_fail_val);
4313 LASSERT(mdd_object_exists(mdd_tobj));
4314 /* step 4: update name entry to the new object */
4315 rc = mdd_migrate_update_name(env, mdd_pobj, mdd_sobj, mdd_tobj, lname,
4320 /* newly created target was not locked, don't cache its attributes */
4322 mdd_invalidate(env, tobj);
4327 const struct md_dir_operations mdd_dir_ops = {
4328 .mdo_is_subdir = mdd_is_subdir,
4329 .mdo_lookup = mdd_lookup,
4330 .mdo_create = mdd_create,
4331 .mdo_rename = mdd_rename,
4332 .mdo_link = mdd_link,
4333 .mdo_unlink = mdd_unlink,
4334 .mdo_create_data = mdd_create_data,
4335 .mdo_migrate = mdd_migrate,