lustre/obdclass/scrub.c

   1 /*
   2  * GPL HEADER START
   3  *
   4  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
   5  *
   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.
   9  *
  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).
  15  *
  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
  19  *
  20  * GPL HEADER END
  21  */
  22 /*
  23  * Copyright (c) 2017, Intel Corporation.
  24  */
  25 /*
  26  * lustre/obdclass/scrub.c
  27  *
  28  * The OI scrub is used for checking and (re)building Object Index files
  29  * that are usually backend special. Here are some general scrub related
  30  * functions that can be shared by different backends for OI scrub.
  31  *
  32  * Author: Fan Yong <fan.yong@intel.com>
  33  */
  34
  35 #define DEBUG_SUBSYSTEM S_LFSCK
  36
  37 #include <linux/kthread.h>
  38 #include <lustre_scrub.h>
  39 #include <lustre_lib.h>
  40 #include <lustre_fid.h>
  41
  42 static inline struct dt_device *scrub_obj2dev(struct dt_object *obj)
  43 {
  44         return container_of_safe(obj->do_lu.lo_dev, struct dt_device,
  45                                  dd_lu_dev);
  46 }
  47
  48 static void scrub_file_to_cpu(struct scrub_file *des, struct scrub_file *src)
  49 {
  50         uuid_copy(&des->sf_uuid, &src->sf_uuid);
  51         des->sf_flags   = le64_to_cpu(src->sf_flags);
  52         des->sf_magic   = le32_to_cpu(src->sf_magic);
  53         des->sf_status  = le16_to_cpu(src->sf_status);
  54         des->sf_param   = le16_to_cpu(src->sf_param);
  55         des->sf_time_last_complete      =
  56                                 le64_to_cpu(src->sf_time_last_complete);
  57         des->sf_time_latest_start       =
  58                                 le64_to_cpu(src->sf_time_latest_start);
  59         des->sf_time_last_checkpoint    =
  60                                 le64_to_cpu(src->sf_time_last_checkpoint);
  61         des->sf_pos_latest_start        =
  62                                 le64_to_cpu(src->sf_pos_latest_start);
  63         des->sf_pos_last_checkpoint     =
  64                                 le64_to_cpu(src->sf_pos_last_checkpoint);
  65         des->sf_pos_first_inconsistent  =
  66                                 le64_to_cpu(src->sf_pos_first_inconsistent);
  67         des->sf_items_checked           =
  68                                 le64_to_cpu(src->sf_items_checked);
  69         des->sf_items_updated           =
  70                                 le64_to_cpu(src->sf_items_updated);
  71         des->sf_items_failed            =
  72                                 le64_to_cpu(src->sf_items_failed);
  73         des->sf_items_updated_prior     =
  74                                 le64_to_cpu(src->sf_items_updated_prior);
  75         des->sf_run_time        = le32_to_cpu(src->sf_run_time);
  76         des->sf_success_count   = le32_to_cpu(src->sf_success_count);
  77         des->sf_oi_count        = le16_to_cpu(src->sf_oi_count);
  78         des->sf_internal_flags  = le16_to_cpu(src->sf_internal_flags);
  79         memcpy(des->sf_oi_bitmap, src->sf_oi_bitmap, SCRUB_OI_BITMAP_SIZE);
  80 }
  81
  82 static void scrub_file_to_le(struct scrub_file *des, struct scrub_file *src)
  83 {
  84         uuid_copy(&des->sf_uuid, &src->sf_uuid);
  85         des->sf_flags   = cpu_to_le64(src->sf_flags);
  86         des->sf_magic   = cpu_to_le32(src->sf_magic);
  87         des->sf_status  = cpu_to_le16(src->sf_status);
  88         des->sf_param   = cpu_to_le16(src->sf_param);
  89         des->sf_time_last_complete      =
  90                                 cpu_to_le64(src->sf_time_last_complete);
  91         des->sf_time_latest_start       =
  92                                 cpu_to_le64(src->sf_time_latest_start);
  93         des->sf_time_last_checkpoint    =
  94                                 cpu_to_le64(src->sf_time_last_checkpoint);
  95         des->sf_pos_latest_start        =
  96                                 cpu_to_le64(src->sf_pos_latest_start);
  97         des->sf_pos_last_checkpoint     =
  98                                 cpu_to_le64(src->sf_pos_last_checkpoint);
  99         des->sf_pos_first_inconsistent  =
 100                                 cpu_to_le64(src->sf_pos_first_inconsistent);
 101         des->sf_items_checked           =
 102                                 cpu_to_le64(src->sf_items_checked);
 103         des->sf_items_updated           =
 104                                 cpu_to_le64(src->sf_items_updated);
 105         des->sf_items_failed            =
 106                                 cpu_to_le64(src->sf_items_failed);
 107         des->sf_items_updated_prior     =
 108                                 cpu_to_le64(src->sf_items_updated_prior);
 109         des->sf_run_time        = cpu_to_le32(src->sf_run_time);
 110         des->sf_success_count   = cpu_to_le32(src->sf_success_count);
 111         des->sf_oi_count        = cpu_to_le16(src->sf_oi_count);
 112         des->sf_internal_flags  = cpu_to_le16(src->sf_internal_flags);
 113         memcpy(des->sf_oi_bitmap, src->sf_oi_bitmap, SCRUB_OI_BITMAP_SIZE);
 114 }
 115
 116 void scrub_file_init(struct lustre_scrub *scrub, uuid_t uuid)
 117 {
 118         struct scrub_file *sf = &scrub->os_file;
 119
 120         memset(sf, 0, sizeof(*sf));
 121         uuid_copy(&sf->sf_uuid, &uuid);
 122         sf->sf_magic = SCRUB_MAGIC_V1;
 123         sf->sf_status = SS_INIT;
 124 }
 125 EXPORT_SYMBOL(scrub_file_init);
 126
 127 void scrub_file_reset(struct lustre_scrub *scrub, uuid_t uuid, u64 flags)
 128 {
 129         struct scrub_file *sf = &scrub->os_file;
 130
 131         CDEBUG(D_LFSCK, "%s: reset OI scrub file, old flags = "
 132                "%#llx, add flags = %#llx\n",
 133                scrub->os_name, sf->sf_flags, flags);
 134
 135         uuid_copy(&sf->sf_uuid, &uuid);
 136         sf->sf_status = SS_INIT;
 137         sf->sf_flags |= flags;
 138         sf->sf_flags &= ~SF_AUTO;
 139         sf->sf_run_time = 0;
 140         sf->sf_time_latest_start = 0;
 141         sf->sf_time_last_checkpoint = 0;
 142         sf->sf_pos_latest_start = 0;
 143         sf->sf_pos_last_checkpoint = 0;
 144         sf->sf_pos_first_inconsistent = 0;
 145         sf->sf_items_checked = 0;
 146         sf->sf_items_updated = 0;
 147         sf->sf_items_failed = 0;
 148         sf->sf_items_noscrub = 0;
 149         sf->sf_items_igif = 0;
 150         if (!scrub->os_in_join)
 151                 sf->sf_items_updated_prior = 0;
 152 }
 153 EXPORT_SYMBOL(scrub_file_reset);
 154
 155 int scrub_file_load(const struct lu_env *env, struct lustre_scrub *scrub)
 156 {
 157         struct scrub_file *sf = &scrub->os_file;
 158         struct lu_buf buf = {
 159                 .lb_buf = &scrub->os_file_disk,
 160                 .lb_len = sizeof(scrub->os_file_disk)
 161         };
 162         loff_t pos = 0;
 163         int rc;
 164
 165         rc = dt_read(env, scrub->os_obj, &buf, &pos);
 166         /* failure */
 167         if (rc < 0) {
 168                 CERROR("%s: fail to load scrub file: rc = %d\n",
 169                        scrub->os_name, rc);
 170                 return rc;
 171         }
 172
 173         /* empty */
 174         if (!rc)
 175                 return -ENOENT;
 176
 177         /* corrupted */
 178         if (rc < buf.lb_len) {
 179                 CDEBUG(D_LFSCK, "%s: fail to load scrub file, "
 180                        "expected = %d: rc = %d\n",
 181                        scrub->os_name, (int)buf.lb_len, rc);
 182                 return -EFAULT;
 183         }
 184
 185         scrub_file_to_cpu(sf, &scrub->os_file_disk);
 186         if (sf->sf_magic != SCRUB_MAGIC_V1) {
 187                 CDEBUG(D_LFSCK, "%s: invalid scrub magic 0x%x != 0x%x\n",
 188                        scrub->os_name, sf->sf_magic, SCRUB_MAGIC_V1);
 189                 return -EFAULT;
 190         }
 191
 192         return 0;
 193 }
 194 EXPORT_SYMBOL(scrub_file_load);
 195
 196 int scrub_file_store(const struct lu_env *env, struct lustre_scrub *scrub)
 197 {
 198         struct scrub_file *sf = &scrub->os_file_disk;
 199         struct dt_object *obj = scrub->os_obj;
 200         struct dt_device *dev = scrub_obj2dev(obj);
 201         struct lu_buf buf = {
 202                 .lb_buf = sf,
 203                 .lb_len = sizeof(*sf)
 204         };
 205         struct thandle *th;
 206         loff_t pos = 0;
 207         int rc;
 208         ENTRY;
 209
 210         /* Skip store under rdonly mode. */
 211         if (dev->dd_rdonly)
 212                 RETURN(0);
 213
 214         scrub_file_to_le(sf, &scrub->os_file);
 215         th = dt_trans_create(env, dev);
 216         if (IS_ERR(th))
 217                 GOTO(log, rc = PTR_ERR(th));
 218
 219         rc = dt_declare_record_write(env, obj, &buf, pos, th);
 220         if (rc)
 221                 GOTO(stop, rc);
 222
 223         rc = dt_trans_start_local(env, dev, th);
 224         if (rc)
 225                 GOTO(stop, rc);
 226
 227         rc = dt_record_write(env, obj, &buf, &pos, th);
 228
 229         GOTO(stop, rc);
 230
 231 stop:
 232         dt_trans_stop(env, dev, th);
 233
 234 log:
 235         if (rc)
 236                 CERROR("%s: store scrub file: rc = %d\n",
 237                        scrub->os_name, rc);
 238         else
 239                 CDEBUG(D_LFSCK, "%s: store scrub file: rc = %d\n",
 240                        scrub->os_name, rc);
 241
 242         scrub->os_time_last_checkpoint = ktime_get_seconds();
 243         scrub->os_time_next_checkpoint = scrub->os_time_last_checkpoint +
 244                                          SCRUB_CHECKPOINT_INTERVAL;
 245         return rc;
 246 }
 247 EXPORT_SYMBOL(scrub_file_store);
 248
 249 int scrub_checkpoint(const struct lu_env *env, struct lustre_scrub *scrub)
 250 {
 251         struct scrub_file *sf = &scrub->os_file;
 252         time64_t now = ktime_get_seconds();
 253         int rc;
 254
 255         if (likely(now < scrub->os_time_next_checkpoint ||
 256                    scrub->os_new_checked == 0))
 257                 return 0;
 258
 259         CDEBUG(D_LFSCK, "%s: OI scrub checkpoint at pos %llu\n",
 260                scrub->os_name, scrub->os_pos_current);
 261
 262         down_write(&scrub->os_rwsem);
 263         sf->sf_items_checked += scrub->os_new_checked;
 264         scrub->os_new_checked = 0;
 265         sf->sf_pos_last_checkpoint = scrub->os_pos_current;
 266         sf->sf_time_last_checkpoint = ktime_get_real_seconds();
 267         sf->sf_run_time += now - scrub->os_time_last_checkpoint;
 268         rc = scrub_file_store(env, scrub);
 269         up_write(&scrub->os_rwsem);
 270
 271         return rc;
 272 }
 273 EXPORT_SYMBOL(scrub_checkpoint);
 274
 275 int scrub_thread_prep(const struct lu_env *env, struct lustre_scrub *scrub,
 276                       uuid_t uuid, u64 start)
 277 {
 278         struct scrub_file *sf = &scrub->os_file;
 279         u32 flags = scrub->os_start_flags;
 280         bool drop_dryrun = false;
 281         int rc;
 282
 283         ENTRY;
 284         CDEBUG(D_LFSCK, "%s: OI scrub prep, flags = 0x%x\n",
 285                scrub->os_name, flags);
 286
 287         down_write(&scrub->os_rwsem);
 288         if (flags & SS_SET_FAILOUT)
 289                 sf->sf_param |= SP_FAILOUT;
 290         else if (flags & SS_CLEAR_FAILOUT)
 291                 sf->sf_param &= ~SP_FAILOUT;
 292
 293         if (flags & SS_SET_DRYRUN) {
 294                 sf->sf_param |= SP_DRYRUN;
 295         } else if (flags & SS_CLEAR_DRYRUN && sf->sf_param & SP_DRYRUN) {
 296                 sf->sf_param &= ~SP_DRYRUN;
 297                 drop_dryrun = true;
 298         }
 299
 300         if (flags & SS_RESET)
 301                 scrub_file_reset(scrub, uuid, 0);
 302
 303         spin_lock(&scrub->os_lock);
 304         scrub->os_partial_scan = 0;
 305         if (flags & SS_AUTO_FULL) {
 306                 scrub->os_full_speed = 1;
 307                 sf->sf_flags |= SF_AUTO;
 308         } else if (flags & SS_AUTO_PARTIAL) {
 309                 scrub->os_full_speed = 0;
 310                 scrub->os_partial_scan = 1;
 311                 sf->sf_flags |= SF_AUTO;
 312         } else if (sf->sf_flags & (SF_RECREATED | SF_INCONSISTENT |
 313                                    SF_UPGRADE)) {
 314                 scrub->os_full_speed = 1;
 315         } else {
 316                 scrub->os_full_speed = 0;
 317         }
 318
 319         scrub->os_in_prior = 0;
 320         scrub->os_waiting = 0;
 321         scrub->os_paused = 0;
 322         scrub->os_in_join = 0;
 323         scrub->os_full_scrub = 0;
 324         spin_unlock(&scrub->os_lock);
 325         scrub->os_new_checked = 0;
 326         if (drop_dryrun && sf->sf_pos_first_inconsistent != 0)
 327                 sf->sf_pos_latest_start = sf->sf_pos_first_inconsistent;
 328         else if (sf->sf_pos_last_checkpoint != 0)
 329                 sf->sf_pos_latest_start = sf->sf_pos_last_checkpoint + 1;
 330         else
 331                 sf->sf_pos_latest_start = start;
 332
 333         scrub->os_pos_current = sf->sf_pos_latest_start;
 334         sf->sf_status = SS_SCANNING;
 335         sf->sf_time_latest_start = ktime_get_real_seconds();
 336         sf->sf_time_last_checkpoint = sf->sf_time_latest_start;
 337         sf->sf_pos_last_checkpoint = sf->sf_pos_latest_start - 1;
 338         rc = scrub_file_store(env, scrub);
 339         if (rc == 0) {
 340                 spin_lock(&scrub->os_lock);
 341                 scrub->os_running = 1;
 342                 spin_unlock(&scrub->os_lock);
 343                 wake_up_var(scrub);
 344         }
 345         up_write(&scrub->os_rwsem);
 346
 347         RETURN(rc);
 348 }
 349 EXPORT_SYMBOL(scrub_thread_prep);
 350
 351 int scrub_thread_post(const struct lu_env *env, struct lustre_scrub *scrub,
 352                       int result)
 353 {
 354         struct scrub_file *sf = &scrub->os_file;
 355         int rc;
 356         ENTRY;
 357
 358         CDEBUG(D_LFSCK, "%s: OI scrub post with result = %d\n",
 359                scrub->os_name, result);
 360
 361         down_write(&scrub->os_rwsem);
 362         spin_lock(&scrub->os_lock);
 363         scrub->os_running = 0;
 364         spin_unlock(&scrub->os_lock);
 365         if (scrub->os_new_checked > 0) {
 366                 sf->sf_items_checked += scrub->os_new_checked;
 367                 scrub->os_new_checked = 0;
 368                 sf->sf_pos_last_checkpoint = scrub->os_pos_current;
 369         }
 370         sf->sf_time_last_checkpoint = ktime_get_real_seconds();
 371         if (result > 0) {
 372                 sf->sf_status = SS_COMPLETED;
 373                 if (!(sf->sf_param & SP_DRYRUN)) {
 374                         memset(sf->sf_oi_bitmap, 0, SCRUB_OI_BITMAP_SIZE);
 375                         sf->sf_flags &= ~(SF_RECREATED | SF_INCONSISTENT |
 376                                           SF_UPGRADE | SF_AUTO);
 377                 }
 378                 sf->sf_time_last_complete = sf->sf_time_last_checkpoint;
 379                 sf->sf_success_count++;
 380         } else if (result == 0) {
 381                 if (scrub->os_paused)
 382                         sf->sf_status = SS_PAUSED;
 383                 else
 384                         sf->sf_status = SS_STOPPED;
 385         } else {
 386                 sf->sf_status = SS_FAILED;
 387         }
 388         sf->sf_run_time += ktime_get_seconds() -
 389                            scrub->os_time_last_checkpoint;
 390
 391         rc = scrub_file_store(env, scrub);
 392         up_write(&scrub->os_rwsem);
 393
 394         RETURN(rc < 0 ? rc : result);
 395 }
 396 EXPORT_SYMBOL(scrub_thread_post);
 397
 398 int scrub_start(int (*threadfn)(void *data), struct lustre_scrub *scrub,
 399                 void *data, __u32 flags)
 400 {
 401         struct task_struct *task;
 402         int rc;
 403         ENTRY;
 404
 405         if (scrub->os_task)
 406                 RETURN(-EALREADY);
 407
 408         if (scrub->os_file.sf_status == SS_COMPLETED) {
 409                 if (!(flags & SS_SET_FAILOUT))
 410                         flags |= SS_CLEAR_FAILOUT;
 411
 412                 if (!(flags & SS_SET_DRYRUN))
 413                         flags |= SS_CLEAR_DRYRUN;
 414
 415                 flags |= SS_RESET;
 416         }
 417
 418         task = kthread_create(threadfn, data, "OI_scrub");
 419         if (IS_ERR(task)) {
 420                 rc = PTR_ERR(task);
 421                 CERROR("%s: cannot start iteration thread: rc = %d\n",
 422                        scrub->os_name, rc);
 423                 RETURN(rc);
 424         }
 425         spin_lock(&scrub->os_lock);
 426         if (scrub->os_task) {
 427                 /* Lost a race */
 428                 spin_unlock(&scrub->os_lock);
 429                 kthread_stop(task);
 430                 RETURN(-EALREADY);
 431         }
 432         scrub->os_start_flags = flags;
 433         scrub->os_task = task;
 434         wake_up_process(task);
 435         spin_unlock(&scrub->os_lock);
 436         wait_var_event(scrub, scrub->os_running || !scrub->os_task);
 437
 438         RETURN(0);
 439 }
 440 EXPORT_SYMBOL(scrub_start);
 441
 442 void scrub_stop(struct lustre_scrub *scrub)
 443 {
 444         struct task_struct *task;
 445
 446         spin_lock(&scrub->os_lock);
 447         scrub->os_running = 0;
 448         spin_unlock(&scrub->os_lock);
 449         task = xchg(&scrub->os_task, NULL);
 450         if (task)
 451                 kthread_stop(task);
 452 }
 453 EXPORT_SYMBOL(scrub_stop);
 454
 455 const char *const scrub_status_names[] = {
 456         "init",
 457         "scanning",
 458         "completed",
 459         "failed",
 460         "stopped",
 461         "paused",
 462         "crashed",
 463         NULL
 464 };
 465
 466 const char *const scrub_flags_names[] = {
 467         "recreated",
 468         "inconsistent",
 469         "auto",
 470         "upgrade",
 471         NULL
 472 };
 473
 474 const char *const scrub_param_names[] = {
 475         "failout",
 476         "dryrun",
 477         NULL
 478 };
 479
 480 static void scrub_bits_dump(struct seq_file *m, int bits,
 481                             const char *const names[],
 482                             const char *prefix)
 483 {
 484         int flag;
 485         int i;
 486
 487         seq_printf(m, "%s:%c", prefix, bits != 0 ? ' ' : '\n');
 488
 489         for (i = 0, flag = 1; bits != 0; i++, flag = BIT(i)) {
 490                 if (flag & bits) {
 491                         bits &= ~flag;
 492                         seq_printf(m, "%s%c", names[i],
 493                                    bits != 0 ? ',' : '\n');
 494                 }
 495         }
 496 }
 497
 498 static void scrub_time_dump(struct seq_file *m, time64_t time,
 499                             const char *prefix)
 500 {
 501         if (time != 0)
 502                 seq_printf(m, "%s: %llu seconds\n", prefix,
 503                            ktime_get_real_seconds() - time);
 504         else
 505                 seq_printf(m, "%s: N/A\n", prefix);
 506 }
 507
 508 static void scrub_pos_dump(struct seq_file *m, __u64 pos, const char *prefix)
 509 {
 510         if (pos != 0)
 511                 seq_printf(m, "%s: %llu\n", prefix, pos);
 512         else
 513                 seq_printf(m, "%s: N/A\n", prefix);
 514 }
 515
 516 void scrub_dump(struct seq_file *m, struct lustre_scrub *scrub)
 517 {
 518         struct scrub_file *sf = &scrub->os_file;
 519         u64 checked;
 520         s64 speed;
 521
 522         down_read(&scrub->os_rwsem);
 523         seq_printf(m, "name: OI_scrub\n"
 524                    "magic: 0x%x\n"
 525                    "oi_files: %d\n"
 526                    "status: %s\n",
 527                    sf->sf_magic, (int)sf->sf_oi_count,
 528                    scrub_status_names[sf->sf_status]);
 529
 530         scrub_bits_dump(m, sf->sf_flags, scrub_flags_names, "flags");
 531
 532         scrub_bits_dump(m, sf->sf_param, scrub_param_names, "param");
 533
 534         scrub_time_dump(m, sf->sf_time_last_complete,
 535                         "time_since_last_completed");
 536
 537         scrub_time_dump(m, sf->sf_time_latest_start,
 538                         "time_since_latest_start");
 539
 540         scrub_time_dump(m, sf->sf_time_last_checkpoint,
 541                         "time_since_last_checkpoint");
 542
 543         scrub_pos_dump(m, sf->sf_pos_latest_start,
 544                         "latest_start_position");
 545
 546         scrub_pos_dump(m, sf->sf_pos_last_checkpoint,
 547                         "last_checkpoint_position");
 548
 549         scrub_pos_dump(m, sf->sf_pos_first_inconsistent,
 550                         "first_failure_position");
 551
 552         checked = sf->sf_items_checked + scrub->os_new_checked;
 553         seq_printf(m, "checked: %llu\n"
 554                    "%s: %llu\n"
 555                    "failed: %llu\n"
 556                    "prior_%s: %llu\n"
 557                    "noscrub: %llu\n"
 558                    "igif: %llu\n"
 559                    "success_count: %u\n",
 560                    checked,
 561                    sf->sf_param & SP_DRYRUN ? "inconsistent" : "updated",
 562                    sf->sf_items_updated, sf->sf_items_failed,
 563                    sf->sf_param & SP_DRYRUN ? "inconsistent" : "updated",
 564                    sf->sf_items_updated_prior, sf->sf_items_noscrub,
 565                    sf->sf_items_igif, sf->sf_success_count);
 566
 567         speed = checked;
 568         if (scrub->os_running) {
 569                 s64 new_checked = scrub->os_new_checked;
 570                 time64_t duration;
 571                 time64_t rtime;
 572
 573                 /* Since the time resolution is in seconds for new system
 574                  * or small devices it ismore likely that duration will be
 575                  * zero which will lead to inaccurate results.
 576                  */
 577                 duration = ktime_get_seconds() -
 578                            scrub->os_time_last_checkpoint;
 579                 if (duration != 0)
 580                         new_checked = div_s64(new_checked, duration);
 581
 582                 rtime = sf->sf_run_time + duration;
 583                 if (rtime != 0)
 584                         speed = div_s64(speed, rtime);
 585
 586                 seq_printf(m, "run_time: %lld seconds\n"
 587                            "average_speed: %lld objects/sec\n"
 588                            "real_time_speed: %lld objects/sec\n"
 589                            "current_position: %llu\n"
 590                            "scrub_in_prior: %s\n"
 591                            "scrub_full_speed: %s\n"
 592                            "partial_scan: %s\n",
 593                            rtime, speed, new_checked,
 594                            scrub->os_pos_current,
 595                            scrub->os_in_prior ? "yes" : "no",
 596                            scrub->os_full_speed ? "yes" : "no",
 597                            scrub->os_partial_scan ? "yes" : "no");
 598         } else {
 599                 if (sf->sf_run_time != 0)
 600                         speed = div_s64(speed, sf->sf_run_time);
 601                 seq_printf(m, "run_time: %d seconds\n"
 602                            "average_speed: %lld objects/sec\n"
 603                            "real_time_speed: N/A\n"
 604                            "current_position: N/A\n",
 605                            sf->sf_run_time, speed);
 606         }
 607
 608         up_read(&scrub->os_rwsem);
 609 }
 610 EXPORT_SYMBOL(scrub_dump);
 611
 612 int lustre_liru_new(struct list_head *head, const struct lu_fid *pfid,
 613                     const struct lu_fid *cfid, __u64 child,
 614                     const char *name, int namelen)
 615 {
 616         struct lustre_index_restore_unit *liru;
 617         int len = sizeof(*liru) + namelen + 1;
 618
 619         OBD_ALLOC(liru, len);
 620         if (!liru)
 621                 return -ENOMEM;
 622
 623         INIT_LIST_HEAD(&liru->liru_link);
 624         liru->liru_pfid = *pfid;
 625         liru->liru_cfid = *cfid;
 626         liru->liru_clid = child;
 627         liru->liru_len = len;
 628         memcpy(liru->liru_name, name, namelen);
 629         liru->liru_name[namelen] = 0;
 630         list_add_tail(&liru->liru_link, head);
 631
 632         return 0;
 633 }
 634 EXPORT_SYMBOL(lustre_liru_new);
 635
 636 int lustre_index_register(struct dt_device *dev, const char *devname,
 637                           struct list_head *head, spinlock_t *lock, int *guard,
 638                           const struct lu_fid *fid,
 639                           __u32 keysize, __u32 recsize)
 640 {
 641         struct lustre_index_backup_unit *libu, *pos;
 642         int rc = 0;
 643         ENTRY;
 644
 645         if (dev->dd_rdonly || *guard)
 646                 RETURN(1);
 647
 648         OBD_ALLOC_PTR(libu);
 649         if (!libu)
 650                 RETURN(-ENOMEM);
 651
 652         INIT_LIST_HEAD(&libu->libu_link);
 653         libu->libu_keysize = keysize;
 654         libu->libu_recsize = recsize;
 655         libu->libu_fid = *fid;
 656
 657         spin_lock(lock);
 658         if (unlikely(*guard)) {
 659                 spin_unlock(lock);
 660                 OBD_FREE_PTR(libu);
 661
 662                 RETURN(1);
 663         }
 664
 665         list_for_each_entry_reverse(pos, head, libu_link) {
 666                 rc = lu_fid_cmp(&pos->libu_fid, fid);
 667                 if (rc < 0) {
 668                         list_add(&libu->libu_link, &pos->libu_link);
 669                         spin_unlock(lock);
 670
 671                         RETURN(0);
 672                 }
 673
 674                 if (!rc) {
 675                         /* Registered already. But the former registered one
 676                          * has different keysize/recsize. It may because that
 677                          * the former values are from disk and corrupted, then
 678                          * replace it with new values. */
 679                         if (unlikely(keysize != pos->libu_keysize ||
 680                                      recsize != pos->libu_recsize)) {
 681                                 CWARN("%s: the index "DFID" has registered "
 682                                       "with %u/%u, may be invalid, replace "
 683                                       "with %u/%u\n",
 684                                       devname, PFID(fid), pos->libu_keysize,
 685                                       pos->libu_recsize, keysize, recsize);
 686
 687                                 pos->libu_keysize = keysize;
 688                                 pos->libu_recsize = recsize;
 689                         } else {
 690                                 rc = 1;
 691                         }
 692
 693                         spin_unlock(lock);
 694                         OBD_FREE_PTR(libu);
 695
 696                         RETURN(rc);
 697                 }
 698         }
 699
 700         list_add(&libu->libu_link, head);
 701         spin_unlock(lock);
 702
 703         RETURN(0);
 704 }
 705 EXPORT_SYMBOL(lustre_index_register);
 706
 707 static void lustre_index_degister(struct list_head *head, spinlock_t *lock,
 708                                   const struct lu_fid *fid)
 709 {
 710         struct lustre_index_backup_unit *libu;
 711         int rc = -ENOENT;
 712
 713         spin_lock(lock);
 714         list_for_each_entry_reverse(libu, head, libu_link) {
 715                 rc = lu_fid_cmp(&libu->libu_fid, fid);
 716                 /* NOT registered. */
 717                 if (rc < 0)
 718                         break;
 719
 720                 if (!rc) {
 721                         list_del(&libu->libu_link);
 722                         break;
 723                 }
 724         }
 725         spin_unlock(lock);
 726
 727         if (!rc)
 728                 OBD_FREE_PTR(libu);
 729 }
 730
 731 static void
 732 lustre_index_backup_make_header(struct lustre_index_backup_header *header,
 733                                 __u32 keysize, __u32 recsize,
 734                                 const struct lu_fid *fid, __u32 count)
 735 {
 736         memset(header, 0, sizeof(*header));
 737         header->libh_magic = cpu_to_le32(INDEX_BACKUP_MAGIC_V1);
 738         header->libh_count = cpu_to_le32(count);
 739         header->libh_keysize = cpu_to_le32(keysize);
 740         header->libh_recsize = cpu_to_le32(recsize);
 741         fid_cpu_to_le(&header->libh_owner, fid);
 742 }
 743
 744 static int lustre_index_backup_body(const struct lu_env *env,
 745                                     struct dt_object *obj, loff_t *pos,
 746                                     void *buf, int bufsize)
 747 {
 748         struct dt_device *dev = lu2dt_dev(obj->do_lu.lo_dev);
 749         struct thandle *th;
 750         struct lu_buf lbuf = {
 751                 .lb_buf = buf,
 752                 .lb_len = bufsize
 753         };
 754         int rc;
 755         ENTRY;
 756
 757         th = dt_trans_create(env, dev);
 758         if (IS_ERR(th))
 759                 RETURN(PTR_ERR(th));
 760
 761         rc = dt_declare_record_write(env, obj, &lbuf, *pos, th);
 762         if (rc)
 763                 GOTO(stop, rc);
 764
 765         rc = dt_trans_start_local(env, dev, th);
 766         if (rc)
 767                 GOTO(stop, rc);
 768
 769         rc = dt_record_write(env, obj, &lbuf, pos, th);
 770
 771         GOTO(stop, rc);
 772
 773 stop:
 774         dt_trans_stop(env, dev, th);
 775         return rc;
 776 }
 777
 778 static int lustre_index_backup_header(const struct lu_env *env,
 779                                       struct dt_object *obj,
 780                                       const struct lu_fid *tgt_fid,
 781                                       __u32 keysize, __u32 recsize,
 782                                       void *buf, int bufsize, int count)
 783 {
 784         struct dt_device *dev = lu2dt_dev(obj->do_lu.lo_dev);
 785         struct lustre_index_backup_header *header = buf;
 786         struct lu_attr *la = buf;
 787         struct thandle *th;
 788         struct lu_buf lbuf = {
 789                 .lb_buf = header,
 790                 .lb_len = sizeof(*header)
 791         };
 792         loff_t size = sizeof(*header) + (keysize + recsize) * count;
 793         loff_t pos = 0;
 794         int rc;
 795         bool punch = false;
 796         ENTRY;
 797
 798         LASSERT(sizeof(*la) <= bufsize);
 799         LASSERT(sizeof(*header) <= bufsize);
 800
 801         rc = dt_attr_get(env, obj, la);
 802         if (rc)
 803                 RETURN(rc);
 804
 805         if (la->la_size > size)
 806                 punch = true;
 807
 808         lustre_index_backup_make_header(header, keysize, recsize,
 809                                         tgt_fid, count);
 810         th = dt_trans_create(env, dev);
 811         if (IS_ERR(th))
 812                 RETURN(PTR_ERR(th));
 813
 814         rc = dt_declare_record_write(env, obj, &lbuf, pos, th);
 815         if (rc)
 816                 GOTO(stop, rc);
 817
 818         if (punch) {
 819                 rc = dt_declare_punch(env, obj, size, OBD_OBJECT_EOF, th);
 820                 if (rc)
 821                         GOTO(stop, rc);
 822         }
 823
 824         rc = dt_trans_start_local(env, dev, th);
 825         if (rc)
 826                 GOTO(stop, rc);
 827
 828         rc = dt_record_write(env, obj, &lbuf, &pos, th);
 829         if (!rc && punch)
 830                 rc = dt_punch(env, obj, size, OBD_OBJECT_EOF, th);
 831
 832         GOTO(stop, rc);
 833
 834 stop:
 835         dt_trans_stop(env, dev, th);
 836         return rc;
 837 }
 838
 839 static int lustre_index_update_lma(const struct lu_env *env,
 840                                    struct dt_object *obj,
 841                                    void *buf, int bufsize)
 842 {
 843         struct dt_device *dev = lu2dt_dev(obj->do_lu.lo_dev);
 844         struct lustre_mdt_attrs *lma = buf;
 845         struct lu_buf lbuf = {
 846                 .lb_buf = lma,
 847                 .lb_len = sizeof(struct lustre_ost_attrs)
 848         };
 849         struct thandle *th;
 850         int fl = LU_XATTR_REPLACE;
 851         int rc;
 852         ENTRY;
 853
 854         LASSERT(bufsize >= lbuf.lb_len);
 855
 856         rc = dt_xattr_get(env, obj, &lbuf, XATTR_NAME_LMA);
 857         if (unlikely(rc == -ENODATA)) {
 858                 fl = LU_XATTR_CREATE;
 859                 lustre_lma_init(lma, lu_object_fid(&obj->do_lu),
 860                                 LMAC_IDX_BACKUP, 0);
 861                 rc = sizeof(*lma);
 862         } else if (rc < sizeof(*lma)) {
 863                 RETURN(rc < 0 ? rc : -EFAULT);
 864         } else {
 865                 lustre_lma_swab(lma);
 866                 if (lma->lma_compat & LMAC_IDX_BACKUP)
 867                         RETURN(0);
 868
 869                 lma->lma_compat |= LMAC_IDX_BACKUP;
 870         }
 871
 872         lustre_lma_swab(lma);
 873         lbuf.lb_len = rc;
 874         th = dt_trans_create(env, dev);
 875         if (IS_ERR(th))
 876                 RETURN(rc);
 877
 878         rc = dt_declare_xattr_set(env, obj, &lbuf, XATTR_NAME_LMA, fl, th);
 879         if (rc)
 880                 GOTO(stop, rc);
 881
 882         rc = dt_trans_start_local(env, dev, th);
 883         if (rc)
 884                 GOTO(stop, rc);
 885
 886         rc = dt_xattr_set(env, obj, &lbuf, XATTR_NAME_LMA, fl, th);
 887
 888         GOTO(stop, rc);
 889
 890 stop:
 891         dt_trans_stop(env, dev, th);
 892         return rc;
 893 }
 894
 895 static int lustre_index_backup_one(const struct lu_env *env,
 896                                    struct local_oid_storage *los,
 897                                    struct dt_object *parent,
 898                                    struct lustre_index_backup_unit *libu,
 899                                    char *buf, int bufsize)
 900 {
 901         struct dt_device *dev = scrub_obj2dev(parent);
 902         struct dt_object *tgt_obj = NULL;
 903         struct dt_object *bak_obj = NULL;
 904         const struct dt_it_ops *iops;
 905         struct dt_it *di;
 906         loff_t pos = sizeof(struct lustre_index_backup_header);
 907         int count = 0;
 908         int size = 0;
 909         int rc;
 910         ENTRY;
 911
 912         tgt_obj = lu2dt(lu_object_find_slice(env, &dev->dd_lu_dev,
 913                                              &libu->libu_fid, NULL));
 914         if (IS_ERR_OR_NULL(tgt_obj))
 915                 GOTO(out, rc = tgt_obj ? PTR_ERR(tgt_obj) : -ENOENT);
 916
 917         if (!dt_object_exists(tgt_obj))
 918                 GOTO(out, rc = 0);
 919
 920         if (!tgt_obj->do_index_ops) {
 921                 struct dt_index_features feat;
 922
 923                 feat.dif_flags = DT_IND_UPDATE;
 924                 feat.dif_keysize_min = libu->libu_keysize;
 925                 feat.dif_keysize_max = libu->libu_keysize;
 926                 feat.dif_recsize_min = libu->libu_recsize;
 927                 feat.dif_recsize_max = libu->libu_recsize;
 928                 feat.dif_ptrsize = 4;
 929                 rc = tgt_obj->do_ops->do_index_try(env, tgt_obj, &feat);
 930                 if (rc)
 931                         GOTO(out, rc);
 932         }
 933
 934         lustre_fid2lbx(buf, &libu->libu_fid, bufsize);
 935         bak_obj = local_file_find_or_create(env, los, parent, buf,
 936                                             S_IFREG | S_IRUGO | S_IWUSR);
 937         if (IS_ERR_OR_NULL(bak_obj))
 938                 GOTO(out, rc = bak_obj ? PTR_ERR(bak_obj) : -ENOENT);
 939
 940         iops = &tgt_obj->do_index_ops->dio_it;
 941         di = iops->init(env, tgt_obj, 0);
 942         if (IS_ERR(di))
 943                 GOTO(out, rc = PTR_ERR(di));
 944
 945         rc = iops->load(env, di, 0);
 946         if (!rc)
 947                 rc = iops->next(env, di);
 948         else if (rc > 0)
 949                 rc = 0;
 950
 951         while (!rc) {
 952                 void *key;
 953                 void *rec;
 954
 955                 key = iops->key(env, di);
 956                 memcpy(&buf[size], key, libu->libu_keysize);
 957                 size += libu->libu_keysize;
 958                 rec = &buf[size];
 959                 rc = iops->rec(env, di, rec, 0);
 960                 if (rc)
 961                         GOTO(fini, rc);
 962
 963                 size += libu->libu_recsize;
 964                 count++;
 965                 if (size + libu->libu_keysize + libu->libu_recsize > bufsize) {
 966                         rc = lustre_index_backup_body(env, bak_obj, &pos,
 967                                                       buf, size);
 968                         if (rc)
 969                                 GOTO(fini, rc);
 970
 971                         size = 0;
 972                 }
 973
 974                 rc = iops->next(env, di);
 975         }
 976
 977         if (rc >= 0 && size > 0)
 978                 rc = lustre_index_backup_body(env, bak_obj, &pos, buf, size);
 979
 980         if (rc < 0)
 981                 GOTO(fini, rc);
 982
 983         rc = lustre_index_backup_header(env, bak_obj, &libu->libu_fid,
 984                                         libu->libu_keysize, libu->libu_recsize,
 985                                         buf, bufsize, count);
 986         if (!rc)
 987                 rc = lustre_index_update_lma(env, tgt_obj, buf, bufsize);
 988
 989         if (!rc && OBD_FAIL_CHECK(OBD_FAIL_OSD_INDEX_CRASH)) {
 990                 LASSERT(bufsize >= 512);
 991
 992                 pos = 0;
 993                 memset(buf, 0, 512);
 994                 lustre_index_backup_body(env, tgt_obj, &pos, buf, 512);
 995         }
 996
 997         GOTO(fini, rc);
 998
 999 fini:
1000         iops->fini(env, di);
1001 out:
1002         if (!IS_ERR_OR_NULL(tgt_obj))
1003                 dt_object_put_nocache(env, tgt_obj);
1004         if (!IS_ERR_OR_NULL(bak_obj))
1005                 dt_object_put_nocache(env, bak_obj);
1006         return rc;
1007 }
1008
1009 void lustre_index_backup(const struct lu_env *env, struct dt_device *dev,
1010                          const char *devname, struct list_head *head,
1011                          spinlock_t *lock, int *guard, bool backup)
1012 {
1013         struct lustre_index_backup_unit *libu;
1014         struct local_oid_storage *los = NULL;
1015         struct dt_object *parent = NULL;
1016         char *buf = NULL;
1017         struct lu_fid fid;
1018         int rc;
1019         ENTRY;
1020
1021         if (dev->dd_rdonly || *guard)
1022                 RETURN_EXIT;
1023
1024         spin_lock(lock);
1025         *guard = 1;
1026         spin_unlock(lock);
1027
1028         if (list_empty(head))
1029                 RETURN_EXIT;
1030
1031         /* Handle kinds of failures during mount process. */
1032         if (!dev->dd_lu_dev.ld_site || !dev->dd_lu_dev.ld_site->ls_top_dev)
1033                 backup = false;
1034
1035         if (backup) {
1036                 OBD_ALLOC_LARGE(buf, INDEX_BACKUP_BUFSIZE);
1037                 if (!buf) {
1038                         backup = false;
1039                         goto scan;
1040                 }
1041
1042                 lu_local_obj_fid(&fid, INDEX_BACKUP_OID);
1043                 parent = lu2dt(lu_object_find_slice(env, &dev->dd_lu_dev,
1044                                                     &fid, NULL));
1045                 if (IS_ERR_OR_NULL(parent)) {
1046                         CERROR("%s: failed to locate backup dir: rc = %ld\n",
1047                                devname, parent ? PTR_ERR(parent) : -ENOENT);
1048                         backup = false;
1049                         goto scan;
1050                 }
1051
1052                 lu_local_name_obj_fid(&fid, 1);
1053                 rc = local_oid_storage_init(env, dev, &fid, &los);
1054                 if (rc) {
1055                         CERROR("%s: failed to init local storage: rc = %d\n",
1056                                devname, rc);
1057                         backup = false;
1058                 }
1059         }
1060
1061 scan:
1062         spin_lock(lock);
1063         while (!list_empty(head)) {
1064                 libu = list_entry(head->next,
1065                                   struct lustre_index_backup_unit, libu_link);
1066                 list_del_init(&libu->libu_link);
1067                 spin_unlock(lock);
1068
1069                 if (backup) {
1070                         rc = lustre_index_backup_one(env, los, parent, libu,
1071                                                      buf, INDEX_BACKUP_BUFSIZE);
1072                         CDEBUG(D_WARNING, "%s: backup index "DFID": rc = %d\n",
1073                                devname, PFID(&libu->libu_fid), rc);
1074                 }
1075
1076                 OBD_FREE_PTR(libu);
1077                 spin_lock(lock);
1078         }
1079         spin_unlock(lock);
1080
1081         if (los)
1082                 local_oid_storage_fini(env, los);
1083         if (parent)
1084                 dt_object_put_nocache(env, parent);
1085         if (buf)
1086                 OBD_FREE_LARGE(buf, INDEX_BACKUP_BUFSIZE);
1087
1088         EXIT;
1089 }
1090 EXPORT_SYMBOL(lustre_index_backup);
1091
1092 int lustre_index_restore(const struct lu_env *env, struct dt_device *dev,
1093                          const struct lu_fid *parent_fid,
1094                          const struct lu_fid *tgt_fid,
1095                          const struct lu_fid *bak_fid, const char *name,
1096                          struct list_head *head, spinlock_t *lock,
1097                          char *buf, int bufsize)
1098 {
1099         struct dt_object *parent_obj = NULL;
1100         struct dt_object *tgt_obj = NULL;
1101         struct dt_object *bak_obj = NULL;
1102         struct lustre_index_backup_header *header;
1103         struct dt_index_features *feat;
1104         struct dt_object_format *dof;
1105         struct lu_attr *la;
1106         struct thandle *th;
1107         struct lu_object_conf conf;
1108         struct dt_insert_rec ent;
1109         struct lu_buf lbuf;
1110         struct lu_fid tfid;
1111         loff_t pos = 0;
1112         __u32 keysize;
1113         __u32 recsize;
1114         __u32 pairsize;
1115         int count;
1116         int rc;
1117         bool registered = false;
1118         ENTRY;
1119
1120         LASSERT(bufsize >= sizeof(*la) + sizeof(*dof) +
1121                 sizeof(*feat) + sizeof(*header));
1122
1123         memset(buf, 0, bufsize);
1124         la = (struct lu_attr *)buf;
1125         dof = (void *)la + sizeof(*la);
1126         feat = (void *)dof + sizeof(*dof);
1127         header = (void *)feat + sizeof(*feat);
1128         lbuf.lb_buf = header;
1129         lbuf.lb_len = sizeof(*header);
1130
1131         tgt_obj = lu2dt(lu_object_find_slice(env, &dev->dd_lu_dev,
1132                                              tgt_fid, NULL));
1133         if (IS_ERR_OR_NULL(tgt_obj))
1134                 GOTO(out, rc = tgt_obj ? PTR_ERR(tgt_obj) : -ENOENT);
1135
1136         bak_obj = lu2dt(lu_object_find_slice(env, &dev->dd_lu_dev,
1137                                              bak_fid, NULL));
1138         if (IS_ERR_OR_NULL(bak_obj))
1139                 GOTO(out, rc = bak_obj ? PTR_ERR(bak_obj) : -ENOENT);
1140
1141         if (!dt_object_exists(bak_obj))
1142                 GOTO(out, rc = -ENOENT);
1143
1144         parent_obj = lu2dt(lu_object_find_slice(env, &dev->dd_lu_dev,
1145                                                 parent_fid, NULL));
1146         if (IS_ERR_OR_NULL(parent_obj))
1147                 GOTO(out, rc = parent_obj ? PTR_ERR(parent_obj) : -ENOENT);
1148
1149         LASSERT(dt_object_exists(parent_obj));
1150
1151         if (unlikely(!dt_try_as_dir(env, parent_obj)))
1152                 GOTO(out, rc = -ENOTDIR);
1153
1154         rc = dt_attr_get(env, tgt_obj, la);
1155         if (rc)
1156                 GOTO(out, rc);
1157
1158         rc = dt_record_read(env, bak_obj, &lbuf, &pos);
1159         if (rc)
1160                 GOTO(out, rc);
1161
1162         if (le32_to_cpu(header->libh_magic) != INDEX_BACKUP_MAGIC_V1)
1163                 GOTO(out, rc = -EINVAL);
1164
1165         fid_le_to_cpu(&tfid, &header->libh_owner);
1166         if (unlikely(!lu_fid_eq(tgt_fid, &tfid)))
1167                 GOTO(out, rc = -EINVAL);
1168
1169         keysize = le32_to_cpu(header->libh_keysize);
1170         recsize = le32_to_cpu(header->libh_recsize);
1171         pairsize = keysize + recsize;
1172
1173         memset(feat, 0, sizeof(*feat));
1174         feat->dif_flags = DT_IND_UPDATE;
1175         feat->dif_keysize_min = feat->dif_keysize_max = keysize;
1176         feat->dif_recsize_min = feat->dif_recsize_max = recsize;
1177         feat->dif_ptrsize = 4;
1178
1179         /* T1: remove old name entry and destroy old index. */
1180         th = dt_trans_create(env, dev);
1181         if (IS_ERR(th))
1182                 GOTO(out, rc = PTR_ERR(th));
1183
1184         rc = dt_declare_delete(env, parent_obj,
1185                                (const struct dt_key *)name, th);
1186         if (rc)
1187                 GOTO(stop, rc);
1188
1189         rc = dt_declare_ref_del(env, tgt_obj, th);
1190         if (rc)
1191                 GOTO(stop, rc);
1192
1193         rc = dt_declare_destroy(env, tgt_obj, th);
1194         if (rc)
1195                 GOTO(stop, rc);
1196
1197         rc = dt_trans_start_local(env, dev, th);
1198         if (rc)
1199                 GOTO(stop, rc);
1200
1201         rc = dt_delete(env, parent_obj, (const struct dt_key *)name, th);
1202         if (rc)
1203                 GOTO(stop, rc);
1204
1205         dt_write_lock(env, tgt_obj, 0);
1206         rc = dt_ref_del(env, tgt_obj, th);
1207         if (rc == 0) {
1208                 if (S_ISDIR(tgt_obj->do_lu.lo_header->loh_attr))
1209                         dt_ref_del(env, tgt_obj, th);
1210                 rc = dt_destroy(env, tgt_obj, th);
1211         }
1212         dt_write_unlock(env, tgt_obj);
1213         dt_trans_stop(env, dev, th);
1214         if (rc)
1215                 GOTO(out, rc);
1216
1217         la->la_valid = LA_MODE | LA_UID | LA_GID;
1218         conf.loc_flags = LOC_F_NEW;
1219         dof->u.dof_idx.di_feat = feat;
1220         dof->dof_type = DFT_INDEX;
1221         ent.rec_type = S_IFREG;
1222         ent.rec_fid = tgt_fid;
1223
1224         /* Drop cache before re-create it. */
1225         dt_object_put_nocache(env, tgt_obj);
1226         tgt_obj = lu2dt(lu_object_find_slice(env, &dev->dd_lu_dev,
1227                                              tgt_fid, &conf));
1228         if (IS_ERR_OR_NULL(tgt_obj))
1229                 GOTO(out, rc = tgt_obj ? PTR_ERR(tgt_obj) : -ENOENT);
1230
1231         LASSERT(!dt_object_exists(tgt_obj));
1232
1233         /* T2: create new index and insert new name entry. */
1234         th = dt_trans_create(env, dev);
1235         if (IS_ERR(th))
1236                 GOTO(out, rc = PTR_ERR(th));
1237
1238         rc = dt_declare_create(env, tgt_obj, la, NULL, dof, th);
1239         if (rc)
1240                 GOTO(stop, rc);
1241
1242         rc = dt_declare_insert(env, parent_obj, (const struct dt_rec *)&ent,
1243                                (const struct dt_key *)name, th);
1244         if (rc)
1245                 GOTO(stop, rc);
1246
1247         rc = dt_trans_start_local(env, dev, th);
1248         if (rc)
1249                 GOTO(stop, rc);
1250
1251         dt_write_lock(env, tgt_obj, 0);
1252         rc = dt_create(env, tgt_obj, la, NULL, dof, th);
1253         dt_write_unlock(env, tgt_obj);
1254         if (rc)
1255                 GOTO(stop, rc);
1256
1257         rc = dt_insert(env, parent_obj, (const struct dt_rec *)&ent,
1258                        (const struct dt_key *)name, th);
1259         dt_trans_stop(env, dev, th);
1260         /* Some index name may has been inserted by OSD
1261          * automatically when create the index object. */
1262         if (unlikely(rc == -EEXIST))
1263                 rc = 0;
1264         if (rc)
1265                 GOTO(out, rc);
1266
1267         /* The new index will register via index_try. */
1268         rc = tgt_obj->do_ops->do_index_try(env, tgt_obj, feat);
1269         if (rc)
1270                 GOTO(out, rc);
1271
1272         registered = true;
1273         count = le32_to_cpu(header->libh_count);
1274         while (!rc && count > 0) {
1275                 int size = pairsize * count;
1276                 int items = count;
1277                 int i;
1278
1279                 if (size > bufsize) {
1280                         items = bufsize / pairsize;
1281                         size = pairsize * items;
1282                 }
1283
1284                 lbuf.lb_buf = buf;
1285                 lbuf.lb_len = size;
1286                 rc = dt_record_read(env, bak_obj, &lbuf, &pos);
1287                 for (i = 0; i < items && !rc; i++) {
1288                         void *key = &buf[i * pairsize];
1289                         void *rec = &buf[i * pairsize + keysize];
1290
1291                         /* Tn: restore the records. */
1292                         th = dt_trans_create(env, dev);
1293                         if (!th)
1294                                 GOTO(out, rc = -ENOMEM);
1295
1296                         rc = dt_declare_insert(env, tgt_obj, rec, key, th);
1297                         if (rc)
1298                                 GOTO(stop, rc);
1299
1300                         rc = dt_trans_start_local(env, dev, th);
1301                         if (rc)
1302                                 GOTO(stop, rc);
1303
1304                         rc = dt_insert(env, tgt_obj, rec, key, th);
1305                         if (unlikely(rc == -EEXIST))
1306                                 rc = 0;
1307
1308                         dt_trans_stop(env, dev, th);
1309                 }
1310
1311                 count -= items;
1312         }
1313
1314         GOTO(out, rc);
1315
1316 stop:
1317         dt_trans_stop(env, dev, th);
1318         if (rc && registered)
1319                 /* Degister the index to avoid overwriting the backup. */
1320                 lustre_index_degister(head, lock, tgt_fid);
1321
1322 out:
1323         if (!IS_ERR_OR_NULL(tgt_obj))
1324                 dt_object_put_nocache(env, tgt_obj);
1325         if (!IS_ERR_OR_NULL(bak_obj))
1326                 dt_object_put_nocache(env, bak_obj);
1327         if (!IS_ERR_OR_NULL(parent_obj))
1328                 dt_object_put_nocache(env, parent_obj);
1329         return rc;
1330 }
1331 EXPORT_SYMBOL(lustre_index_restore);