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) 2014, 2015, Intel Corporation.
27 * lustre/target/update_recovery.c
29 * This file implement the methods to handle the update recovery.
31 * During DNE recovery, the recovery thread will redo the operation according
32 * to the transaction no, and these replay are either from client replay req
33 * or update replay records(for distribute transaction) in the update log.
34 * For distribute transaction replay, the replay thread will call
35 * distribute_txn_replay_handle() to handle the updates.
37 * After the Master MDT restarts, it will retrieve the update records from all
38 * of MDTs, for each distributed operation, it will check updates on all MDTs,
39 * if some updates records are missing on some MDTs, the replay thread will redo
40 * updates on these MDTs.
42 * Author: Di Wang <di.wang@intel.com>
44 #define DEBUG_SUBSYSTEM S_CLASS
46 #include <lu_target.h>
47 #include <lustre_update.h>
48 #include <lustre_swab.h>
49 #include <md_object.h>
51 #include <obd_class.h>
52 #include "tgt_internal.h"
55 * Lookup distribute_txn_replay req
57 * Lookup distribute_txn_replay in the replay list by batchid.
58 * It is assumed the list has been locked before calling this function.
60 * \param[in] tdtd distribute_txn_data, which holds the replay
62 * \param[in] batchid batchid used by lookup.
64 * \retval pointer of the replay if succeeds.
65 * \retval NULL if can not find it.
67 static struct distribute_txn_replay_req *
68 dtrq_lookup(struct target_distribute_txn_data *tdtd, __u64 batchid)
70 struct distribute_txn_replay_req *tmp;
71 struct distribute_txn_replay_req *dtrq = NULL;
73 list_for_each_entry(tmp, &tdtd->tdtd_replay_list, dtrq_list) {
74 if (tmp->dtrq_batchid == batchid) {
83 * insert distribute txn replay req
85 * Insert distribute txn replay to the replay list, and it assumes the
86 * list has been looked. Note: the replay list is a sorted list, which
87 * is sorted by master transno. It is assumed the replay list has been
88 * locked before calling this function.
90 * \param[in] tdtd target distribute txn data where replay list is
91 * \param[in] new distribute txn replay to be inserted
93 * \retval 0 if insertion succeeds
94 * \retval EEXIST if the dtrq already exists
96 static int dtrq_insert(struct target_distribute_txn_data *tdtd,
97 struct distribute_txn_replay_req *new)
99 struct distribute_txn_replay_req *iter;
101 /* Check if the dtrq has been added to the list */
102 iter = dtrq_lookup(tdtd, new->dtrq_batchid);
106 list_for_each_entry_reverse(iter, &tdtd->tdtd_replay_list, dtrq_list) {
107 if (iter->dtrq_master_transno > new->dtrq_master_transno)
110 /* If there are mulitple replay req with same transno, then
111 * sort them with batchid */
112 if (iter->dtrq_master_transno == new->dtrq_master_transno &&
113 iter->dtrq_batchid > new->dtrq_batchid)
116 list_add(&new->dtrq_list, &iter->dtrq_list);
120 if (list_empty(&new->dtrq_list))
121 list_add(&new->dtrq_list, &tdtd->tdtd_replay_list);
127 * create distribute txn replay req
129 * Allocate distribute txn replay req according to the update records.
131 * \param[in] tdtd target distribute txn data where replay list is.
132 * \param[in] record update records from the update log.
134 * \retval the pointer of distribute txn replay req if
135 * the creation succeeds.
136 * \retval NULL if the creation fails.
138 static struct distribute_txn_replay_req *
139 dtrq_create(struct target_distribute_txn_data *tdtd,
140 struct llog_update_record *lur)
142 struct distribute_txn_replay_req *new;
146 RETURN(ERR_PTR(-ENOMEM));
148 new->dtrq_lur_size = llog_update_record_size(lur);
149 OBD_ALLOC_LARGE(new->dtrq_lur, new->dtrq_lur_size);
150 if (new->dtrq_lur == NULL) {
152 RETURN(ERR_PTR(-ENOMEM));
155 memcpy(new->dtrq_lur, lur, new->dtrq_lur_size);
157 /* If the transno in the update record is 0, it means the
158 * update are from master MDT, and it will use the master
159 * last committed transno as its master transno. Later, if
160 * the update records are gotten from slave MDTs, then these
161 * transno will be replaced.
162 * See insert_update_records_to_replay_list(). */
163 if (lur->lur_update_rec.ur_master_transno == 0) {
164 new->dtrq_lur->lur_update_rec.ur_master_transno =
165 tdtd->tdtd_lut->lut_obd->obd_last_committed;
166 new->dtrq_master_transno =
167 tdtd->tdtd_lut->lut_obd->obd_last_committed;
169 new->dtrq_master_transno =
170 lur->lur_update_rec.ur_master_transno;
173 new->dtrq_batchid = lur->lur_update_rec.ur_batchid;
175 spin_lock_init(&new->dtrq_sub_list_lock);
176 INIT_LIST_HEAD(&new->dtrq_sub_list);
177 INIT_LIST_HEAD(&new->dtrq_list);
183 * Lookup distribute sub replay
185 * Lookup distribute sub replay in the sub list of distribute_txn_replay by
188 * \param[in] distribute_txn_replay_req the distribute txn replay req to lookup
189 * \param[in] mdt_index the mdt_index as the key of lookup
191 * \retval the pointer of sub replay if it can be found.
192 * \retval NULL if it can not find.
194 struct distribute_txn_replay_req_sub *
195 dtrq_sub_lookup(struct distribute_txn_replay_req *dtrq, __u32 mdt_index)
197 struct distribute_txn_replay_req_sub *dtrqs = NULL;
198 struct distribute_txn_replay_req_sub *tmp;
200 list_for_each_entry(tmp, &dtrq->dtrq_sub_list, dtrqs_list) {
201 if (tmp->dtrqs_mdt_index == mdt_index) {
210 * Try to add cookie to sub distribute txn request
212 * Check if the update log cookie has been added to the request, if not,
213 * add it to the dtrqs_cookie_list.
215 * \param[in] dtrqs sub replay req where cookies to be added.
216 * \param[in] cookie cookie to be added.
218 * \retval 0 if the cookie is adding succeeds.
219 * \retval negative errno if adding fails.
221 static int dtrq_sub_add_cookie(struct distribute_txn_replay_req_sub *dtrqs,
222 struct llog_cookie *cookie)
224 struct sub_thandle_cookie *new;
230 INIT_LIST_HEAD(&new->stc_list);
231 new->stc_cookie = *cookie;
232 /* Note: only single thread will access one sub_request each time,
233 * so no need lock here */
234 list_add(&new->stc_list, &dtrqs->dtrqs_cookie_list);
240 * Insert distribute txn sub req replay
242 * Allocate sub replay req and insert distribute txn replay list.
244 * \param[in] dtrq d to be added
245 * \param[in] cookie the cookie of the update record
246 * \param[in] mdt_index the mdt_index of the update record
248 * \retval 0 if the adding succeeds.
249 * \retval negative errno if the adding fails.
252 dtrq_sub_create_and_insert(struct distribute_txn_replay_req *dtrq,
253 struct llog_cookie *cookie,
256 struct distribute_txn_replay_req_sub *dtrqs = NULL;
257 struct distribute_txn_replay_req_sub *new;
261 spin_lock(&dtrq->dtrq_sub_list_lock);
262 dtrqs = dtrq_sub_lookup(dtrq, mdt_index);
263 spin_unlock(&dtrq->dtrq_sub_list_lock);
265 rc = dtrq_sub_add_cookie(dtrqs, cookie);
273 INIT_LIST_HEAD(&new->dtrqs_list);
274 INIT_LIST_HEAD(&new->dtrqs_cookie_list);
275 new->dtrqs_mdt_index = mdt_index;
276 spin_lock(&dtrq->dtrq_sub_list_lock);
277 dtrqs = dtrq_sub_lookup(dtrq, mdt_index);
279 list_add(&new->dtrqs_list, &dtrq->dtrq_sub_list);
284 spin_unlock(&dtrq->dtrq_sub_list_lock);
286 rc = dtrq_sub_add_cookie(dtrqs, cookie);
292 * append updates to the current replay updates
294 * Append more updates to the existent replay update. And this is only
295 * used when combining mulitple updates into one large updates during
298 * \param[in] dtrq the update replay request where the new update
299 * records will be added.
300 * \param[in] lur the new update record.
302 * \retval 0 if appending succeeds.
303 * \retval negative errno if appending fails.
305 static int dtrq_append_updates(struct distribute_txn_replay_req *dtrq,
306 struct update_records *record)
308 struct llog_update_record *new_lur;
309 size_t lur_size = dtrq->dtrq_lur_size;
313 /* Because several threads might retrieve the same records from
314 * different targets, and we only need one copy of records. So
315 * we will check if the records is in the next one, if not, just
317 spin_lock(&dtrq->dtrq_sub_list_lock);
318 if (dtrq->dtrq_lur->lur_update_rec.ur_index + 1 != record->ur_index) {
319 spin_unlock(&dtrq->dtrq_sub_list_lock);
322 dtrq->dtrq_lur->lur_update_rec.ur_index++;
323 spin_unlock(&dtrq->dtrq_sub_list_lock);
325 lur_size += update_records_size(record);
326 OBD_ALLOC_LARGE(new_lur, lur_size);
327 if (new_lur == NULL) {
328 spin_lock(&dtrq->dtrq_sub_list_lock);
329 dtrq->dtrq_lur->lur_update_rec.ur_index--;
330 spin_unlock(&dtrq->dtrq_sub_list_lock);
334 /* Copy the old and new records to the new allocated buffer */
335 memcpy(new_lur, dtrq->dtrq_lur, dtrq->dtrq_lur_size);
336 ptr = (char *)&new_lur->lur_update_rec +
337 update_records_size(&new_lur->lur_update_rec);
338 memcpy(ptr, &record->ur_ops,
339 update_records_size(record) -
340 offsetof(struct update_records, ur_ops));
342 new_lur->lur_update_rec.ur_update_count += record->ur_update_count;
343 new_lur->lur_update_rec.ur_param_count += record->ur_param_count;
344 new_lur->lur_hdr.lrh_len = llog_update_record_size(new_lur);
346 /* Replace the records */
347 OBD_FREE_LARGE(dtrq->dtrq_lur, dtrq->dtrq_lur_size);
348 dtrq->dtrq_lur = new_lur;
349 dtrq->dtrq_lur_size = lur_size;
350 dtrq->dtrq_lur->lur_update_rec.ur_flags = record->ur_flags;
351 update_records_dump(&new_lur->lur_update_rec, D_INFO, true);
356 * Insert update records to the replay list.
358 * Allocate distribute txn replay req and insert it into the replay
359 * list, then insert the update records into the replay req.
361 * \param[in] tdtd distribute txn replay data where the replay list
363 * \param[in] record the update record
364 * \param[in] cookie cookie of the record
365 * \param[in] index mdt index of the record
367 * \retval 0 if the adding succeeds.
368 * \retval negative errno if the adding fails.
371 insert_update_records_to_replay_list(struct target_distribute_txn_data *tdtd,
372 struct llog_update_record *lur,
373 struct llog_cookie *cookie,
376 struct distribute_txn_replay_req *dtrq;
377 struct update_records *record = &lur->lur_update_rec;
378 bool replace_record = false;
382 CDEBUG(D_HA, "%s: insert record batchid = "LPU64" transno = "LPU64
383 " mdt_index %u\n", tdtd->tdtd_lut->lut_obd->obd_name,
384 record->ur_batchid, record->ur_master_transno, mdt_index);
386 /* Update batchid if necessary */
387 spin_lock(&tdtd->tdtd_batchid_lock);
388 if (record->ur_batchid >= tdtd->tdtd_batchid) {
389 CDEBUG(D_HA, "%s update batchid from "LPU64 " to "LPU64"\n",
390 tdtd->tdtd_lut->lut_obd->obd_name,
391 tdtd->tdtd_batchid, record->ur_batchid);
392 tdtd->tdtd_batchid = record->ur_batchid + 1;
394 spin_unlock(&tdtd->tdtd_batchid_lock);
397 spin_lock(&tdtd->tdtd_replay_list_lock);
398 /* First try to build the replay update request with the records */
399 dtrq = dtrq_lookup(tdtd, record->ur_batchid);
401 spin_unlock(&tdtd->tdtd_replay_list_lock);
402 dtrq = dtrq_create(tdtd, lur);
404 RETURN(PTR_ERR(dtrq));
406 spin_lock(&tdtd->tdtd_replay_list_lock);
407 rc = dtrq_insert(tdtd, dtrq);
409 spin_unlock(&tdtd->tdtd_replay_list_lock);
416 /* If the master transno in update header is not
417 * matched with the one in the record, then it means
418 * the dtrq is originally created by master record,
419 * so we need update master transno and reposition
420 * the dtrq(by master transno) in the list and also
421 * replace update record */
422 if (record->ur_master_transno != 0 &&
423 dtrq->dtrq_master_transno != record->ur_master_transno &&
424 dtrq->dtrq_lur != NULL) {
425 list_del_init(&dtrq->dtrq_list);
426 dtrq->dtrq_lur->lur_update_rec.ur_master_transno =
427 record->ur_master_transno;
429 dtrq->dtrq_master_transno = record->ur_master_transno;
430 replace_record = true;
431 /* try to insert again */
432 rc = dtrq_insert(tdtd, dtrq);
434 spin_unlock(&tdtd->tdtd_replay_list_lock);
440 spin_unlock(&tdtd->tdtd_replay_list_lock);
442 /* Because there should be only thread access the update record, so
443 * we do not need lock here */
444 if (replace_record) {
445 /* Replace the update record and master transno */
446 OBD_FREE_LARGE(dtrq->dtrq_lur, dtrq->dtrq_lur_size);
447 dtrq->dtrq_lur = NULL;
448 dtrq->dtrq_lur_size = llog_update_record_size(lur);
449 OBD_ALLOC_LARGE(dtrq->dtrq_lur, dtrq->dtrq_lur_size);
450 if (dtrq->dtrq_lur == NULL)
453 memcpy(dtrq->dtrq_lur, lur, dtrq->dtrq_lur_size);
456 /* This is a partial update records, let's try to append
457 * the record to the current replay request */
458 if (record->ur_flags & UPDATE_RECORD_CONTINUE)
459 rc = dtrq_append_updates(dtrq, record);
461 /* Then create and add sub update request */
462 rc = dtrq_sub_create_and_insert(dtrq, cookie, mdt_index);
466 EXPORT_SYMBOL(insert_update_records_to_replay_list);
469 * Dump updates of distribute txns.
471 * Output all of recovery updates in the distribute txn list to the
474 * \param[in] tdtd distribute txn data where all of distribute txn
476 * \param[in] mask debug mask
478 void dtrq_list_dump(struct target_distribute_txn_data *tdtd, unsigned int mask)
480 struct distribute_txn_replay_req *dtrq;
482 spin_lock(&tdtd->tdtd_replay_list_lock);
483 list_for_each_entry(dtrq, &tdtd->tdtd_replay_list, dtrq_list)
484 update_records_dump(&dtrq->dtrq_lur->lur_update_rec, mask,
486 spin_unlock(&tdtd->tdtd_replay_list_lock);
488 EXPORT_SYMBOL(dtrq_list_dump);
491 * Destroy distribute txn replay req
493 * Destroy distribute txn replay req and all of subs.
495 * \param[in] dtrq distribute txn replqy req to be destroyed.
497 void dtrq_destroy(struct distribute_txn_replay_req *dtrq)
499 struct distribute_txn_replay_req_sub *dtrqs;
500 struct distribute_txn_replay_req_sub *tmp;
502 LASSERT(list_empty(&dtrq->dtrq_list));
503 spin_lock(&dtrq->dtrq_sub_list_lock);
504 list_for_each_entry_safe(dtrqs, tmp, &dtrq->dtrq_sub_list, dtrqs_list) {
505 struct sub_thandle_cookie *stc;
506 struct sub_thandle_cookie *tmp;
508 list_del(&dtrqs->dtrqs_list);
509 list_for_each_entry_safe(stc, tmp, &dtrqs->dtrqs_cookie_list,
511 list_del(&stc->stc_list);
516 spin_unlock(&dtrq->dtrq_sub_list_lock);
518 if (dtrq->dtrq_lur != NULL)
519 OBD_FREE_LARGE(dtrq->dtrq_lur, dtrq->dtrq_lur_size);
523 EXPORT_SYMBOL(dtrq_destroy);
526 * Destroy all of replay req.
528 * Destroy all of replay req in the replay list.
530 * \param[in] tdtd target distribute txn data where the replay list is.
532 void dtrq_list_destroy(struct target_distribute_txn_data *tdtd)
534 struct distribute_txn_replay_req *dtrq;
535 struct distribute_txn_replay_req *tmp;
537 spin_lock(&tdtd->tdtd_replay_list_lock);
538 list_for_each_entry_safe(dtrq, tmp, &tdtd->tdtd_replay_list,
540 list_del_init(&dtrq->dtrq_list);
543 list_for_each_entry_safe(dtrq, tmp, &tdtd->tdtd_replay_finish_list,
545 list_del_init(&dtrq->dtrq_list);
548 spin_unlock(&tdtd->tdtd_replay_list_lock);
550 EXPORT_SYMBOL(dtrq_list_destroy);
553 * Get next req in the replay list
555 * Get next req needs to be replayed, since it is a sorted list
556 * (by master MDT transno)
558 * \param[in] tdtd distribute txn data where the replay list is
560 * \retval the pointer of update recovery header
562 struct distribute_txn_replay_req *
563 distribute_txn_get_next_req(struct target_distribute_txn_data *tdtd)
565 struct distribute_txn_replay_req *dtrq = NULL;
567 spin_lock(&tdtd->tdtd_replay_list_lock);
568 if (!list_empty(&tdtd->tdtd_replay_list)) {
569 dtrq = list_entry(tdtd->tdtd_replay_list.next,
570 struct distribute_txn_replay_req, dtrq_list);
571 list_del_init(&dtrq->dtrq_list);
573 spin_unlock(&tdtd->tdtd_replay_list_lock);
577 EXPORT_SYMBOL(distribute_txn_get_next_req);
580 * Get next transno in the replay list, because this is the sorted
581 * list, so it will return the transno of next req in the list.
583 * \param[in] tdtd distribute txn data where the replay list is
585 * \retval the transno of next update in the list
587 __u64 distribute_txn_get_next_transno(struct target_distribute_txn_data *tdtd)
589 struct distribute_txn_replay_req *dtrq = NULL;
592 spin_lock(&tdtd->tdtd_replay_list_lock);
593 if (!list_empty(&tdtd->tdtd_replay_list)) {
594 dtrq = list_entry(tdtd->tdtd_replay_list.next,
595 struct distribute_txn_replay_req, dtrq_list);
596 transno = dtrq->dtrq_master_transno;
598 spin_unlock(&tdtd->tdtd_replay_list_lock);
600 CDEBUG(D_HA, "%s: Next update transno "LPU64"\n",
601 tdtd->tdtd_lut->lut_obd->obd_name, transno);
604 EXPORT_SYMBOL(distribute_txn_get_next_transno);
606 struct distribute_txn_replay_req *
607 distribute_txn_lookup_finish_list(struct target_distribute_txn_data *tdtd,
610 struct distribute_txn_replay_req *dtrq = NULL;
611 struct distribute_txn_replay_req *iter;
613 spin_lock(&tdtd->tdtd_replay_list_lock);
614 list_for_each_entry(iter, &tdtd->tdtd_replay_finish_list, dtrq_list) {
615 if (iter->dtrq_xid == xid) {
620 spin_unlock(&tdtd->tdtd_replay_list_lock);
624 bool is_req_replayed_by_update(struct ptlrpc_request *req)
626 struct lu_target *tgt = class_exp2tgt(req->rq_export);
627 struct distribute_txn_replay_req *dtrq;
629 if (tgt->lut_tdtd == NULL)
632 dtrq = distribute_txn_lookup_finish_list(tgt->lut_tdtd, req->rq_xid);
638 EXPORT_SYMBOL(is_req_replayed_by_update);
641 * Check if the update of one object is committed
643 * Check whether the update for the object is committed by checking whether
644 * the correspondent sub exists in the replay req. If it is committed, mark
645 * the committed flag in correspondent the sub thandle.
647 * \param[in] env execution environment
648 * \param[in] dtrq replay request
649 * \param[in] dt_obj object for the update
650 * \param[in] top_th top thandle
651 * \param[in] sub_th sub thandle which the update belongs to
653 * \retval 1 if the update is not committed.
654 * \retval 0 if the update is committed.
655 * \retval negative errno if some other failures happen.
657 static int update_is_committed(const struct lu_env *env,
658 struct distribute_txn_replay_req *dtrq,
659 struct dt_object *dt_obj,
660 struct top_thandle *top_th,
661 struct sub_thandle *st)
663 struct seq_server_site *seq_site;
664 const struct lu_fid *fid = lu_object_fid(&dt_obj->do_lu);
665 struct distribute_txn_replay_req_sub *dtrqs;
669 if (st->st_sub_th != NULL)
672 if (st->st_committed)
675 seq_site = lu_site2seq(dt_obj->do_lu.lo_dev->ld_site);
676 if (fid_is_update_log(fid) || fid_is_update_log_dir(fid)) {
677 mdt_index = fid_oid(fid);
678 } else if (!fid_seq_in_fldb(fid_seq(fid))) {
679 mdt_index = seq_site->ss_node_id;
681 struct lu_server_fld *fld;
682 struct lu_seq_range range = {0};
685 fld = seq_site->ss_server_fld;
686 fld_range_set_type(&range, LU_SEQ_RANGE_MDT);
687 LASSERT(fld->lsf_seq_lookup != NULL);
688 rc = fld->lsf_seq_lookup(env, fld, fid_seq(fid),
692 mdt_index = range.lsr_index;
695 dtrqs = dtrq_sub_lookup(dtrq, mdt_index);
696 if (dtrqs != NULL || top_th->tt_multiple_thandle->tmt_committed) {
697 st->st_committed = 1;
699 struct sub_thandle_cookie *stc;
700 struct sub_thandle_cookie *tmp;
702 list_for_each_entry_safe(stc, tmp,
703 &dtrqs->dtrqs_cookie_list,
705 list_move(&stc->stc_list, &st->st_cookie_list);
710 CDEBUG(D_HA, "Update of "DFID "on MDT%u is not committed\n", PFID(fid),
717 * Implementation of different update methods for update recovery.
719 * These following functions update_recovery_$(update_name) implement
720 * different updates recovery methods. They will extract the parameters
721 * from the common parameters area and call correspondent dt API to redo
724 * \param[in] env execution environment
725 * \param[in] op update operation to be replayed
726 * \param[in] params common update parameters which holds all parameters
728 * \param[in] th transaction handle
729 * \param[in] declare indicate it will do declare or real execution, true
730 * means declare, false means real execution
732 * \retval 0 if it succeeds.
733 * \retval negative errno if it fails.
735 static int update_recovery_create(const struct lu_env *env,
736 struct dt_object *dt_obj,
737 const struct update_op *op,
738 const struct update_params *params,
739 struct thandle_exec_args *ta,
742 struct update_thread_info *uti = update_env_info(env);
743 struct llog_update_record *lur = uti->uti_dtrq->dtrq_lur;
744 struct lu_attr *attr = &uti->uti_attr;
746 struct obdo *lobdo = &uti->uti_obdo;
747 struct dt_object_format dof;
749 unsigned int param_count;
753 if (dt_object_exists(dt_obj))
756 param_count = lur->lur_update_rec.ur_param_count;
757 wobdo = update_params_get_param_buf(params, op->uop_params_off[0],
761 if (size != sizeof(*wobdo))
764 if (LLOG_REC_HDR_NEEDS_SWABBING(&lur->lur_hdr))
765 lustre_swab_obdo(wobdo);
767 lustre_get_wire_obdo(NULL, lobdo, wobdo);
768 la_from_obdo(attr, lobdo, lobdo->o_valid);
770 dof.dof_type = dt_mode_to_dft(attr->la_mode);
772 rc = out_tx_create(env, dt_obj, attr, NULL, &dof,
778 static int update_recovery_destroy(const struct lu_env *env,
779 struct dt_object *dt_obj,
780 const struct update_op *op,
781 const struct update_params *params,
782 struct thandle_exec_args *ta,
788 rc = out_tx_destroy(env, dt_obj, ta, th, NULL, 0);
793 static int update_recovery_ref_add(const struct lu_env *env,
794 struct dt_object *dt_obj,
795 const struct update_op *op,
796 const struct update_params *params,
797 struct thandle_exec_args *ta,
803 rc = out_tx_ref_add(env, dt_obj, ta, th, NULL, 0);
808 static int update_recovery_ref_del(const struct lu_env *env,
809 struct dt_object *dt_obj,
810 const struct update_op *op,
811 const struct update_params *params,
812 struct thandle_exec_args *ta,
818 rc = out_tx_ref_del(env, dt_obj, ta, th, NULL, 0);
823 static int update_recovery_attr_set(const struct lu_env *env,
824 struct dt_object *dt_obj,
825 const struct update_op *op,
826 const struct update_params *params,
827 struct thandle_exec_args *ta,
830 struct update_thread_info *uti = update_env_info(env);
831 struct llog_update_record *lur = uti->uti_dtrq->dtrq_lur;
833 struct obdo *lobdo = &uti->uti_obdo;
834 struct lu_attr *attr = &uti->uti_attr;
836 unsigned int param_count;
840 param_count = lur->lur_update_rec.ur_param_count;
841 wobdo = update_params_get_param_buf(params, op->uop_params_off[0],
845 if (size != sizeof(*wobdo))
848 if (LLOG_REC_HDR_NEEDS_SWABBING(&lur->lur_hdr))
849 lustre_swab_obdo(wobdo);
851 lustre_get_wire_obdo(NULL, lobdo, wobdo);
852 la_from_obdo(attr, lobdo, lobdo->o_valid);
854 rc = out_tx_attr_set(env, dt_obj, attr, ta, th, NULL, 0);
859 static int update_recovery_xattr_set(const struct lu_env *env,
860 struct dt_object *dt_obj,
861 const struct update_op *op,
862 const struct update_params *params,
863 struct thandle_exec_args *ta,
866 struct update_thread_info *uti = update_env_info(env);
875 param_count = uti->uti_dtrq->dtrq_lur->lur_update_rec.ur_param_count;
876 name = update_params_get_param_buf(params,
877 op->uop_params_off[0],
882 buf = update_params_get_param_buf(params,
883 op->uop_params_off[1],
888 uti->uti_buf.lb_buf = buf;
889 uti->uti_buf.lb_len = (size_t)size;
891 buf = update_params_get_param_buf(params, op->uop_params_off[2],
895 if (size != sizeof(fl))
898 fl = le32_to_cpu(*(int *)buf);
900 rc = out_tx_xattr_set(env, dt_obj, &uti->uti_buf, name, fl, ta, th,
906 static int update_recovery_index_insert(const struct lu_env *env,
907 struct dt_object *dt_obj,
908 const struct update_op *op,
909 const struct update_params *params,
910 struct thandle_exec_args *ta,
913 struct update_thread_info *uti = update_env_info(env);
923 param_count = uti->uti_dtrq->dtrq_lur->lur_update_rec.ur_param_count;
924 name = update_params_get_param_buf(params, op->uop_params_off[0],
929 fid = update_params_get_param_buf(params, op->uop_params_off[1],
933 if (size != sizeof(*fid))
936 fid_le_to_cpu(fid, fid);
938 ptype = update_params_get_param_buf(params, op->uop_params_off[2],
942 if (size != sizeof(*ptype))
944 type = le32_to_cpu(*ptype);
946 if (dt_try_as_dir(env, dt_obj) == 0)
949 uti->uti_rec.rec_fid = fid;
950 uti->uti_rec.rec_type = type;
952 rc = out_tx_index_insert(env, dt_obj,
953 (const struct dt_rec *)&uti->uti_rec,
954 (const struct dt_key *)name, ta, th,
960 static int update_recovery_index_delete(const struct lu_env *env,
961 struct dt_object *dt_obj,
962 const struct update_op *op,
963 const struct update_params *params,
964 struct thandle_exec_args *ta,
967 struct update_thread_info *uti = update_env_info(env);
974 param_count = uti->uti_dtrq->dtrq_lur->lur_update_rec.ur_param_count;
975 name = update_params_get_param_buf(params, op->uop_params_off[0],
980 if (dt_try_as_dir(env, dt_obj) == 0)
983 rc = out_tx_index_delete(env, dt_obj,
984 (const struct dt_key *)name, ta, th, NULL, 0);
989 static int update_recovery_write(const struct lu_env *env,
990 struct dt_object *dt_obj,
991 const struct update_op *op,
992 const struct update_params *params,
993 struct thandle_exec_args *ta,
996 struct update_thread_info *uti = update_env_info(env);
1004 param_count = uti->uti_dtrq->dtrq_lur->lur_update_rec.ur_param_count;
1005 buf = update_params_get_param_buf(params, op->uop_params_off[0],
1006 param_count, &size);
1010 uti->uti_buf.lb_buf = buf;
1011 uti->uti_buf.lb_len = size;
1013 buf = update_params_get_param_buf(params, op->uop_params_off[1],
1014 param_count, &size);
1018 pos = le64_to_cpu(*(__u64 *)buf);
1020 rc = out_tx_write(env, dt_obj, &uti->uti_buf, pos,
1026 static int update_recovery_xattr_del(const struct lu_env *env,
1027 struct dt_object *dt_obj,
1028 const struct update_op *op,
1029 const struct update_params *params,
1030 struct thandle_exec_args *ta,
1033 struct update_thread_info *uti = update_env_info(env);
1040 param_count = uti->uti_dtrq->dtrq_lur->lur_update_rec.ur_param_count;
1041 name = update_params_get_param_buf(params, op->uop_params_off[0],
1042 param_count, &size);
1046 rc = out_tx_xattr_del(env, dt_obj, name, ta, th, NULL, 0);
1052 * Update session information
1054 * Update session information so tgt_txn_stop_cb()->tgt_last_rcvd_update()
1055 * can be called correctly during update replay.
1057 * \param[in] env execution environment.
1058 * \param[in] tdtd distribute data structure of the recovering tgt.
1059 * \param[in] th thandle of this update replay.
1060 * \param[in] master_th master sub thandle.
1061 * \param[in] ta_arg the tx arg structure to hold the update for updating
1064 static void update_recovery_update_ses(struct lu_env *env,
1065 struct target_distribute_txn_data *tdtd,
1067 struct thandle *master_th,
1068 struct distribute_txn_replay_req *dtrq,
1069 struct tx_arg *ta_arg)
1071 struct tgt_session_info *tsi;
1072 struct lu_target *lut = tdtd->tdtd_lut;
1073 struct obd_export *export;
1074 struct cfs_hash *hash;
1075 struct top_thandle *top_th;
1076 struct lsd_reply_data *lrd;
1079 tsi = tgt_ses_info(env);
1080 if (tsi->tsi_exp != NULL)
1083 size = ta_arg->u.write.buf.lb_len;
1084 lrd = ta_arg->u.write.buf.lb_buf;
1085 if (size != sizeof(*lrd) || lrd == NULL)
1088 lrd->lrd_transno = le64_to_cpu(lrd->lrd_transno);
1089 lrd->lrd_xid = le64_to_cpu(lrd->lrd_xid);
1090 lrd->lrd_data = le64_to_cpu(lrd->lrd_data);
1091 lrd->lrd_result = le32_to_cpu(lrd->lrd_result);
1092 lrd->lrd_client_gen = le32_to_cpu(lrd->lrd_client_gen);
1094 if (lrd->lrd_transno != tgt_th_info(env)->tti_transno)
1097 hash = cfs_hash_getref(lut->lut_obd->obd_gen_hash);
1101 export = cfs_hash_lookup(hash, &lrd->lrd_client_gen);
1102 if (export == NULL) {
1103 cfs_hash_putref(hash);
1107 tsi->tsi_exp = export;
1108 tsi->tsi_xid = lrd->lrd_xid;
1109 tsi->tsi_opdata = lrd->lrd_data;
1110 tsi->tsi_result = lrd->lrd_result;
1111 tsi->tsi_client_gen = lrd->lrd_client_gen;
1112 dtrq->dtrq_xid = lrd->lrd_xid;
1113 top_th = container_of(th, struct top_thandle, tt_super);
1114 top_th->tt_master_sub_thandle = master_th;
1115 cfs_hash_putref(hash);
1119 * Execute updates in the update replay records
1121 * Declare distribute txn replay by update records and add the updates
1122 * to the execution list. Note: it will check if the update has been
1123 * committed, and only execute the updates if it is not committed to
1126 * \param[in] env execution environment
1127 * \param[in] tdtd distribute txn replay data which hold all of replay
1128 * reqs and all replay parameters.
1129 * \param[in] dtrq distribute transaction replay req.
1130 * \param[in] ta thandle execute args.
1132 * \retval 0 if declare succeeds.
1133 * \retval negative errno if declare fails.
1135 static int update_recovery_exec(const struct lu_env *env,
1136 struct target_distribute_txn_data *tdtd,
1137 struct distribute_txn_replay_req *dtrq,
1138 struct thandle_exec_args *ta)
1140 struct llog_update_record *lur = dtrq->dtrq_lur;
1141 struct update_records *records = &lur->lur_update_rec;
1142 struct update_ops *ops = &records->ur_ops;
1143 struct update_params *params = update_records_get_params(records);
1144 struct top_thandle *top_th = container_of(ta->ta_handle,
1147 struct top_multiple_thandle *tmt = top_th->tt_multiple_thandle;
1148 struct update_op *op;
1153 /* These records have been swabbed in llog_cat_process() */
1154 for (i = 0, op = &ops->uops_op[0]; i < records->ur_update_count;
1155 i++, op = update_op_next_op(op)) {
1156 struct lu_fid *fid = &op->uop_fid;
1157 struct dt_object *dt_obj;
1158 struct dt_object *sub_dt_obj;
1159 struct dt_device *sub_dt;
1160 struct sub_thandle *st;
1162 if (op->uop_type == OUT_NOOP)
1165 dt_obj = dt_locate(env, tdtd->tdtd_dt, fid);
1166 if (IS_ERR(dt_obj)) {
1167 rc = PTR_ERR(dt_obj);
1170 sub_dt_obj = dt_object_child(dt_obj);
1172 /* Create sub thandle if not */
1173 sub_dt = lu2dt_dev(sub_dt_obj->do_lu.lo_dev);
1174 st = lookup_sub_thandle(tmt, sub_dt);
1176 st = create_sub_thandle(tmt, sub_dt);
1178 GOTO(next, rc = PTR_ERR(st));
1181 /* check if updates on the OSD/OSP are committed */
1182 rc = update_is_committed(env, dtrq, dt_obj, top_th, st);
1184 /* If this is committed, goto next */
1190 /* Create thandle for sub thandle if needed */
1191 if (st->st_sub_th == NULL) {
1192 rc = sub_thandle_trans_create(env, top_th, st);
1197 CDEBUG(D_HA, "replay %uth update\n", i);
1198 switch (op->uop_type) {
1200 rc = update_recovery_create(env, sub_dt_obj,
1205 rc = update_recovery_destroy(env, sub_dt_obj,
1210 rc = update_recovery_ref_add(env, sub_dt_obj,
1215 rc = update_recovery_ref_del(env, sub_dt_obj,
1220 rc = update_recovery_attr_set(env, sub_dt_obj,
1225 rc = update_recovery_xattr_set(env, sub_dt_obj,
1229 case OUT_INDEX_INSERT:
1230 rc = update_recovery_index_insert(env, sub_dt_obj,
1234 case OUT_INDEX_DELETE:
1235 rc = update_recovery_index_delete(env, sub_dt_obj,
1240 rc = update_recovery_write(env, sub_dt_obj,
1245 rc = update_recovery_xattr_del(env, sub_dt_obj,
1250 CERROR("Unknown update type %u\n", (__u32)op->uop_type);
1255 lu_object_put(env, &dt_obj->do_lu);
1260 ta->ta_handle->th_result = rc;
1265 * redo updates on MDT if needed.
1267 * During DNE recovery, the recovery thread (target_recovery_thread) will call
1268 * this function to replay distribute txn updates on all MDTs. It only replay
1269 * updates on the MDT where the update record is missing.
1271 * If the update already exists on the MDT, then it does not need replay the
1272 * updates on that MDT, and only mark the sub transaction has been committed
1275 * \param[in] env execution environment
1276 * \param[in] tdtd target distribute txn data, which holds the replay list
1277 * and all parameters needed by replay process.
1278 * \param[in] dtrq distribute txn replay req.
1280 * \retval 0 if replay succeeds.
1281 * \retval negative errno if replay failes.
1283 int distribute_txn_replay_handle(struct lu_env *env,
1284 struct target_distribute_txn_data *tdtd,
1285 struct distribute_txn_replay_req *dtrq)
1287 struct update_records *records = &dtrq->dtrq_lur->lur_update_rec;
1288 struct thandle_exec_args *ta;
1289 struct lu_context session_env;
1290 struct thandle *th = NULL;
1291 struct top_thandle *top_th;
1292 struct top_multiple_thandle *tmt;
1293 struct thandle_update_records *tur = NULL;
1298 /* initialize session, it is needed for the handler of target */
1299 rc = lu_context_init(&session_env, LCT_SERVER_SESSION | LCT_NOREF);
1301 CERROR("%s: failure to initialize session: rc = %d\n",
1302 tdtd->tdtd_lut->lut_obd->obd_name, rc);
1305 lu_context_enter(&session_env);
1306 env->le_ses = &session_env;
1308 update_records_dump(records, D_HA, true);
1309 th = top_trans_create(env, NULL);
1311 GOTO(exit_session, rc = PTR_ERR(th));
1313 ta = &update_env_info(env)->uti_tea;
1316 update_env_info(env)->uti_dtrq = dtrq;
1317 /* Create distribute transaction structure for this top thandle */
1318 top_th = container_of(th, struct top_thandle, tt_super);
1319 rc = top_trans_create_tmt(env, top_th);
1321 GOTO(stop_trans, rc);
1323 th->th_dev = tdtd->tdtd_dt;
1326 /* check if the distribute transaction has been committed */
1327 tmt = top_th->tt_multiple_thandle;
1328 tmt->tmt_master_sub_dt = tdtd->tdtd_lut->lut_bottom;
1329 tmt->tmt_batchid = dtrq->dtrq_batchid;
1330 tgt_th_info(env)->tti_transno = dtrq->dtrq_master_transno;
1332 if (tmt->tmt_batchid <= tdtd->tdtd_committed_batchid)
1333 tmt->tmt_committed = 1;
1335 rc = update_recovery_exec(env, tdtd, dtrq, ta);
1337 GOTO(stop_trans, rc);
1339 /* If no updates are needed to be replayed, then mark this records as
1340 * committed, so commit thread distribute_txn_commit_thread() will
1341 * delete the record */
1342 if (ta->ta_argno == 0)
1343 tmt->tmt_committed = 1;
1345 tur = &update_env_info(env)->uti_tur;
1346 tur->tur_update_records = dtrq->dtrq_lur;
1347 tur->tur_update_records_buf_size = dtrq->dtrq_lur_size;
1348 tur->tur_update_params = NULL;
1349 tur->tur_update_param_count = 0;
1350 tmt->tmt_update_records = tur;
1352 distribute_txn_insert_by_batchid(tmt);
1353 rc = top_trans_start(env, NULL, th);
1355 GOTO(stop_trans, rc);
1357 for (i = 0; i < ta->ta_argno; i++) {
1358 struct tx_arg *ta_arg;
1359 struct dt_object *dt_obj;
1360 struct dt_device *sub_dt;
1361 struct sub_thandle *st;
1363 ta_arg = ta->ta_args[i];
1364 dt_obj = ta_arg->object;
1366 LASSERT(tmt->tmt_committed == 0);
1367 sub_dt = lu2dt_dev(dt_obj->do_lu.lo_dev);
1368 st = lookup_sub_thandle(tmt, sub_dt);
1370 LASSERT(st != NULL);
1371 LASSERT(st->st_sub_th != NULL);
1372 rc = ta->ta_args[i]->exec_fn(env, st->st_sub_th,
1375 /* If the update is to update the reply data, then
1376 * we need set the session information, so
1377 * tgt_last_rcvd_update() can be called correctly */
1378 if (rc == 0 && dt_obj == tdtd->tdtd_lut->lut_reply_data)
1379 update_recovery_update_ses(env, tdtd, th,
1380 st->st_sub_th, dtrq, ta_arg);
1382 if (unlikely(rc < 0)) {
1383 CDEBUG(D_HA, "error during execution of #%u from"
1384 " %s:%d: rc = %d\n", i, ta->ta_args[i]->file,
1385 ta->ta_args[i]->line, rc);
1387 if (ta->ta_args[i]->undo_fn != NULL) {
1388 dt_obj = ta->ta_args[i]->object;
1390 lu2dt_dev(dt_obj->do_lu.lo_dev);
1391 st = lookup_sub_thandle(tmt, sub_dt);
1392 LASSERT(st != NULL);
1393 LASSERT(st->st_sub_th != NULL);
1395 ta->ta_args[i]->undo_fn(env,
1399 CERROR("%s: undo for %s:%d: rc = %d\n",
1400 dt_obd_name(ta->ta_handle->th_dev),
1401 ta->ta_args[i]->file,
1402 ta->ta_args[i]->line, -ENOTSUPP);
1407 CDEBUG(D_HA, "%s: executed %u/%u: rc = %d\n",
1408 dt_obd_name(sub_dt), i, ta->ta_argno, rc);
1414 rc = top_trans_stop(env, tdtd->tdtd_dt, th);
1415 for (i = 0; i < ta->ta_argno; i++) {
1416 if (ta->ta_args[i]->object != NULL) {
1417 lu_object_put(env, &ta->ta_args[i]->object->do_lu);
1418 ta->ta_args[i]->object = NULL;
1423 tur->tur_update_records = NULL;
1425 if (tgt_ses_info(env)->tsi_exp != NULL) {
1426 class_export_put(tgt_ses_info(env)->tsi_exp);
1427 tgt_ses_info(env)->tsi_exp = NULL;
1430 lu_context_exit(&session_env);
1431 lu_context_fini(&session_env);
1434 EXPORT_SYMBOL(distribute_txn_replay_handle);