lustre/osd-ldiskfs/osd_oi.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) 2007, 2010, Oracle and/or its affiliates. All rights reserved.
  24  * Use is subject to license terms.
  25  *
  26  * Copyright (c) 2012, 2016, Intel Corporation.
  27  */
  28 /*
  29  * This file is part of Lustre, http://www.lustre.org/
  30  * Lustre is a trademark of Sun Microsystems, Inc.
  31  *
  32  * lustre/osd/osd_oi.c
  33  *
  34  * Object Index.
  35  *
  36  * Author: Nikita Danilov <nikita@clusterfs.com>
  37  */
  38
  39 #define DEBUG_SUBSYSTEM S_OSD
  40
  41 #include <linux/module.h>
  42
  43 /*
  44  * struct OBD_{ALLOC,FREE}*()
  45  * OBD_FAIL_CHECK
  46  */
  47 #include <obd.h>
  48 #include <obd_support.h>
  49
  50 /* fid_cpu_to_be() */
  51 #include <lustre_fid.h>
  52 #include <dt_object.h>
  53
  54 #include "osd_oi.h"
  55 /* osd_lookup(), struct osd_thread_info */
  56 #include "osd_internal.h"
  57 #include "osd_scrub.h"
  58
  59 static unsigned int osd_oi_count = OSD_OI_FID_NR;
  60 module_param(osd_oi_count, int, 0444);
  61 MODULE_PARM_DESC(osd_oi_count, "Number of Object Index containers to be created, it's only valid for new filesystem.");
  62
  63 /** to serialize concurrent OI index initialization */
  64 static struct mutex oi_init_lock;
  65
  66 static struct dt_index_features oi_feat = {
  67         .dif_flags       = DT_IND_UPDATE,
  68         .dif_recsize_min = sizeof(struct osd_inode_id),
  69         .dif_recsize_max = sizeof(struct osd_inode_id),
  70         .dif_ptrsize     = 4
  71 };
  72
  73 #define OSD_OI_NAME_BASE        "oi.16"
  74
  75 static void osd_oi_table_put(struct osd_thread_info *info,
  76                              struct osd_oi **oi_table, unsigned oi_count)
  77 {
  78         struct iam_container *bag;
  79         int                   i;
  80
  81         for (i = 0; i < oi_count; i++) {
  82                 if (oi_table[i] == NULL)
  83                         continue;
  84
  85                 LASSERT(oi_table[i]->oi_inode != NULL);
  86
  87                 bag = &(oi_table[i]->oi_dir.od_container);
  88                 if (bag->ic_object == oi_table[i]->oi_inode)
  89                         iam_container_fini(bag);
  90                 iput(oi_table[i]->oi_inode);
  91                 oi_table[i]->oi_inode = NULL;
  92                 OBD_FREE_PTR(oi_table[i]);
  93                 oi_table[i] = NULL;
  94         }
  95 }
  96
  97 static int osd_oi_index_create_one(struct osd_thread_info *info,
  98                                    struct osd_device *osd, const char *name,
  99                                    struct dt_index_features *feat)
 100 {
 101         const struct lu_env             *env = info->oti_env;
 102         struct osd_inode_id             *id  = &info->oti_id;
 103         struct buffer_head              *bh;
 104         struct inode                    *inode;
 105         struct ldiskfs_dir_entry_2      *de;
 106         struct dentry                   *dentry;
 107         struct super_block              *sb  = osd_sb(osd);
 108         struct inode                    *dir = sb->s_root->d_inode;
 109         handle_t                        *jh;
 110         int                              rc;
 111
 112         dentry = osd_child_dentry_by_inode(env, dir, name, strlen(name));
 113         bh = osd_ldiskfs_find_entry(dir, &dentry->d_name, &de, NULL, NULL);
 114         if (!IS_ERR(bh)) {
 115                 osd_id_gen(id, le32_to_cpu(de->inode), OSD_OII_NOGEN);
 116                 brelse(bh);
 117                 inode = osd_iget(info, osd, id);
 118                 if (!IS_ERR(inode)) {
 119                         iput(inode);
 120                         inode = ERR_PTR(-EEXIST);
 121                 }
 122                 return PTR_ERR(inode);
 123         }
 124
 125         jh = osd_journal_start_sb(sb, LDISKFS_HT_MISC, 100);
 126         if (IS_ERR(jh))
 127                 return PTR_ERR(jh);
 128
 129         inode = ldiskfs_create_inode(jh, dir, (S_IFREG | S_IRUGO | S_IWUSR));
 130         if (IS_ERR(inode)) {
 131                 ldiskfs_journal_stop(jh);
 132                 return PTR_ERR(inode);
 133         }
 134
 135         ldiskfs_set_inode_state(inode, LDISKFS_STATE_LUSTRE_NOSCRUB);
 136         unlock_new_inode(inode);
 137
 138         if (feat->dif_flags & DT_IND_VARKEY)
 139                 rc = iam_lvar_create(inode, feat->dif_keysize_max,
 140                                      feat->dif_ptrsize, feat->dif_recsize_max,
 141                                      jh);
 142         else
 143                 rc = iam_lfix_create(inode, feat->dif_keysize_max,
 144                                      feat->dif_ptrsize, feat->dif_recsize_max,
 145                                      jh);
 146         dentry = osd_child_dentry_by_inode(env, dir, name, strlen(name));
 147         rc = osd_ldiskfs_add_entry(info, osd, jh, dentry, inode, NULL);
 148         ldiskfs_journal_stop(jh);
 149         iput(inode);
 150         return rc;
 151 }
 152
 153 static struct inode *osd_oi_index_open(struct osd_thread_info *info,
 154                                        struct osd_device *osd,
 155                                        const char *name,
 156                                        struct dt_index_features *f,
 157                                        bool create)
 158 {
 159         struct dentry *dentry;
 160         struct inode  *inode;
 161         int            rc;
 162
 163         dentry = ll_lookup_one_len(name, osd_sb(osd)->s_root, strlen(name));
 164         if (IS_ERR(dentry))
 165                 return (void *) dentry;
 166
 167         if (dentry->d_inode) {
 168                 LASSERT(!is_bad_inode(dentry->d_inode));
 169                 inode = dentry->d_inode;
 170                 atomic_inc(&inode->i_count);
 171                 dput(dentry);
 172                 return inode;
 173         }
 174
 175         /* create */
 176         dput(dentry);
 177         shrink_dcache_parent(osd_sb(osd)->s_root);
 178         if (!create)
 179                 return ERR_PTR(-ENOENT);
 180
 181         rc = osd_oi_index_create_one(info, osd, name, f);
 182         if (rc)
 183                 return ERR_PTR(rc);
 184
 185         dentry = ll_lookup_one_len(name, osd_sb(osd)->s_root, strlen(name));
 186         if (IS_ERR(dentry))
 187                 return (void *) dentry;
 188
 189         if (dentry->d_inode) {
 190                 LASSERT(!is_bad_inode(dentry->d_inode));
 191                 inode = dentry->d_inode;
 192                 atomic_inc(&inode->i_count);
 193                 dput(dentry);
 194                 return inode;
 195         }
 196
 197         return ERR_PTR(-ENOENT);
 198 }
 199
 200 /**
 201  * Open an OI(Ojbect Index) container.
 202  *
 203  * \param       name    Name of OI container
 204  * \param       objp    Pointer of returned OI
 205  *
 206  * \retval      0       success
 207  * \retval      -ve     failure
 208  */
 209 static int osd_oi_open(struct osd_thread_info *info, struct osd_device *osd,
 210                        char *name, struct osd_oi **oi_slot, bool create)
 211 {
 212         struct osd_directory *dir;
 213         struct iam_container *bag;
 214         struct inode         *inode;
 215         struct osd_oi        *oi;
 216         int                   rc;
 217
 218         ENTRY;
 219
 220         oi_feat.dif_keysize_min = sizeof(struct lu_fid);
 221         oi_feat.dif_keysize_max = sizeof(struct lu_fid);
 222
 223         inode = osd_oi_index_open(info, osd, name, &oi_feat, create);
 224         if (IS_ERR(inode))
 225                 RETURN(PTR_ERR(inode));
 226
 227         /* 'What the @fid is' is not imporatant, because these objects
 228          * have no OI mappings, and only are visible inside the OSD.*/
 229         lu_igif_build(&info->oti_fid, inode->i_ino, inode->i_generation);
 230         rc = osd_ea_fid_set(info, inode, &info->oti_fid, LMAC_NOT_IN_OI, 0);
 231         if (rc != 0)
 232                 GOTO(out_inode, rc);
 233
 234         OBD_ALLOC_PTR(oi);
 235         if (oi == NULL)
 236                 GOTO(out_inode, rc = -ENOMEM);
 237
 238         oi->oi_inode = inode;
 239         dir = &oi->oi_dir;
 240
 241         bag = &dir->od_container;
 242         rc = iam_container_init(bag, &dir->od_descr, inode);
 243         if (rc < 0)
 244                 GOTO(out_free, rc);
 245
 246         rc = iam_container_setup(bag);
 247         if (rc < 0)
 248                 GOTO(out_container, rc);
 249
 250         *oi_slot = oi;
 251         RETURN(0);
 252
 253 out_container:
 254         iam_container_fini(bag);
 255 out_free:
 256         OBD_FREE_PTR(oi);
 257 out_inode:
 258         iput(inode);
 259         return rc;
 260 }
 261
 262 /**
 263  * Open OI(Object Index) table.
 264  * If \a oi_count is zero, which means caller doesn't know how many OIs there
 265  * will be, this function can either return 0 for new filesystem, or number
 266  * of OIs on existed filesystem.
 267  *
 268  * If \a oi_count is non-zero, which means caller does know number of OIs on
 269  * filesystem, this function should return the exactly same number on
 270  * success, or error code in failure.
 271  *
 272  * \param     oi_count  Number of expected OI containers
 273  * \param     create    Create OIs if doesn't exist
 274  *
 275  * \retval    +ve       number of opened OI containers
 276  * \retval      0       no OI containers found
 277  * \retval    -ve       failure
 278  */
 279 static int
 280 osd_oi_table_open(struct osd_thread_info *info, struct osd_device *osd,
 281                   struct osd_oi **oi_table, unsigned oi_count, bool create)
 282 {
 283         struct scrub_file *sf = &osd->od_scrub.os_file;
 284         int                count = 0;
 285         int                rc = 0;
 286         int                i;
 287         ENTRY;
 288
 289         /* NB: oi_count != 0 means that we have already created/known all OIs
 290          * and have known exact number of OIs. */
 291         LASSERT(oi_count <= OSD_OI_FID_NR_MAX);
 292
 293         for (i = 0; i < (oi_count != 0 ? oi_count : OSD_OI_FID_NR_MAX); i++) {
 294                 char name[12];
 295
 296                 if (oi_table[i] != NULL) {
 297                         count++;
 298                         continue;
 299                 }
 300
 301                 sprintf(name, "%s.%d", OSD_OI_NAME_BASE, i);
 302                 rc = osd_oi_open(info, osd, name, &oi_table[i], create);
 303                 if (rc == 0) {
 304                         count++;
 305                         continue;
 306                 }
 307
 308                 if (rc == -ENOENT && create == false) {
 309                         if (oi_count == 0)
 310                                 return count;
 311
 312                         rc = 0;
 313                         ldiskfs_set_bit(i, sf->sf_oi_bitmap);
 314                         continue;
 315                 }
 316
 317                 CERROR("%s: can't open %s: rc = %d\n",
 318                        osd_dev2name(osd), name, rc);
 319                 if (oi_count > 0)
 320                         CERROR("%s: expect to open total %d OI files.\n",
 321                                osd_dev2name(osd), oi_count);
 322                 break;
 323         }
 324
 325         if (rc < 0) {
 326                 osd_oi_table_put(info, oi_table, oi_count > 0 ? oi_count : i);
 327                 count = rc;
 328         }
 329
 330         RETURN(count);
 331 }
 332
 333 static int osd_remove_oi_one(struct dentry *parent, const char *name,
 334                              int namelen)
 335 {
 336         struct dentry *child;
 337         int rc;
 338
 339         child = ll_lookup_one_len(name, parent, namelen);
 340         if (IS_ERR(child)) {
 341                 rc = PTR_ERR(child);
 342         } else {
 343                 rc = ll_vfs_unlink(parent->d_inode, child);
 344                 dput(child);
 345         }
 346
 347         return rc == -ENOENT ? 0 : rc;
 348 }
 349
 350 static int osd_remove_ois(struct osd_thread_info *info, struct osd_device *osd)
 351 {
 352         char name[16];
 353         int namelen;
 354         int rc;
 355         int i;
 356
 357         for (i = 0; i < osd->od_scrub.os_file.sf_oi_count; i++) {
 358                 namelen = snprintf(name, sizeof(name), "%s.%d",
 359                                    OSD_OI_NAME_BASE, i);
 360                 rc = osd_remove_oi_one(osd_sb(osd)->s_root, name, namelen);
 361                 if (rc != 0) {
 362                         CERROR("%s: fail to remove the stale OI file %s: "
 363                                "rc = %d\n", osd_dev2name(osd), name, rc);
 364                         return rc;
 365                 }
 366         }
 367
 368         namelen = snprintf(name, sizeof(name), "%s", OSD_OI_NAME_BASE);
 369         rc = osd_remove_oi_one(osd_sb(osd)->s_root, name, namelen);
 370         if (rc != 0)
 371                 CERROR("%s: fail to remove the stale OI file %s: rc = %d\n",
 372                        osd_dev2name(osd), name, rc);
 373
 374         return rc;
 375 }
 376
 377 int osd_oi_init(struct osd_thread_info *info, struct osd_device *osd,
 378                 bool restored)
 379 {
 380         struct osd_scrub  *scrub = &osd->od_scrub;
 381         struct scrub_file *sf = &scrub->os_file;
 382         struct osd_oi    **oi;
 383         int                rc;
 384         ENTRY;
 385
 386         if (restored) {
 387                 rc = osd_remove_ois(info, osd);
 388                 if (rc != 0)
 389                         return rc;
 390         }
 391
 392         OBD_ALLOC(oi, sizeof(*oi) * OSD_OI_FID_NR_MAX);
 393         if (oi == NULL)
 394                 RETURN(-ENOMEM);
 395
 396         mutex_lock(&oi_init_lock);
 397         /* try to open existing multiple OIs first */
 398         rc = osd_oi_table_open(info, osd, oi, sf->sf_oi_count, false);
 399         if (rc < 0)
 400                 GOTO(out, rc);
 401
 402         if (rc > 0) {
 403                 if (rc == sf->sf_oi_count || sf->sf_oi_count == 0)
 404                         GOTO(out, rc);
 405
 406                 osd_scrub_file_reset(scrub,
 407                                      LDISKFS_SB(osd_sb(osd))->s_es->s_uuid,
 408                                      SF_RECREATED);
 409                 osd_oi_count = sf->sf_oi_count;
 410                 goto create;
 411         }
 412
 413         /* if previous failed then try found single OI from old filesystem */
 414         rc = osd_oi_open(info, osd, OSD_OI_NAME_BASE, &oi[0], false);
 415         if (rc == 0) { /* found single OI from old filesystem */
 416                 ldiskfs_clear_bit(0, sf->sf_oi_bitmap);
 417                 if (sf->sf_success_count == 0)
 418                         /* XXX: There is one corner case that if the OI_scrub
 419                          *      file crashed or lost and we regard it upgrade,
 420                          *      then we allow IGIF lookup to bypass OI files.
 421                          *
 422                          *      The risk is that osd_fid_lookup() may found
 423                          *      a wrong inode with the given IGIF especially
 424                          *      when the MDT has performed file-level backup
 425                          *      and restored after former upgrading from 1.8
 426                          *      to 2.x. Fortunately, the osd_fid_lookup()can
 427                          *      verify the inode to decrease the risk. */
 428                         osd_scrub_file_reset(scrub,
 429                                         LDISKFS_SB(osd_sb(osd))->s_es->s_uuid,
 430                                         SF_UPGRADE);
 431                 GOTO(out, rc = 1);
 432         } else if (rc != -ENOENT) {
 433                 CERROR("%s: can't open %s: rc = %d\n",
 434                        osd_dev2name(osd), OSD_OI_NAME_BASE, rc);
 435                 GOTO(out, rc);
 436         }
 437
 438         if (sf->sf_oi_count > 0) {
 439                 int i;
 440
 441                 memset(sf->sf_oi_bitmap, 0, SCRUB_OI_BITMAP_SIZE);
 442                 for (i = 0; i < osd_oi_count; i++)
 443                         ldiskfs_set_bit(i, sf->sf_oi_bitmap);
 444                 osd_scrub_file_reset(scrub,
 445                                      LDISKFS_SB(osd_sb(osd))->s_es->s_uuid,
 446                                      SF_RECREATED);
 447         }
 448         sf->sf_oi_count = osd_oi_count;
 449
 450 create:
 451         rc = osd_scrub_file_store(scrub);
 452         if (rc < 0) {
 453                 osd_oi_table_put(info, oi, sf->sf_oi_count);
 454                 GOTO(out, rc);
 455         }
 456
 457         /* No OIs exist, new filesystem, create OI objects */
 458         rc = osd_oi_table_open(info, osd, oi, osd_oi_count, true);
 459         LASSERT(ergo(rc >= 0, rc == osd_oi_count));
 460
 461         GOTO(out, rc);
 462
 463 out:
 464         if (rc < 0) {
 465                 OBD_FREE(oi, sizeof(*oi) * OSD_OI_FID_NR_MAX);
 466         } else {
 467                 LASSERT((rc & (rc - 1)) == 0);
 468                 osd->od_oi_table = oi;
 469                 osd->od_oi_count = rc;
 470                 if (sf->sf_oi_count != rc) {
 471                         sf->sf_oi_count = rc;
 472                         rc = osd_scrub_file_store(scrub);
 473                         if (rc < 0) {
 474                                 osd_oi_table_put(info, oi, sf->sf_oi_count);
 475                                 OBD_FREE(oi, sizeof(*oi) * OSD_OI_FID_NR_MAX);
 476                         }
 477                 } else {
 478                         rc = 0;
 479                 }
 480         }
 481
 482         mutex_unlock(&oi_init_lock);
 483         return rc;
 484 }
 485
 486 void osd_oi_fini(struct osd_thread_info *info, struct osd_device *osd)
 487 {
 488         if (unlikely(osd->od_oi_table == NULL))
 489                 return;
 490
 491         osd_oi_table_put(info, osd->od_oi_table, osd->od_oi_count);
 492
 493         OBD_FREE(osd->od_oi_table,
 494                  sizeof(*(osd->od_oi_table)) * OSD_OI_FID_NR_MAX);
 495         osd->od_oi_table = NULL;
 496 }
 497
 498 static inline int fid_is_fs_root(const struct lu_fid *fid)
 499 {
 500         /* Map root inode to special local object FID */
 501         return (unlikely(fid_seq(fid) == FID_SEQ_LOCAL_FILE &&
 502                          fid_oid(fid) == OSD_FS_ROOT_OID));
 503 }
 504
 505 static int osd_oi_iam_lookup(struct osd_thread_info *oti,
 506                              struct osd_oi *oi, struct dt_rec *rec,
 507                              const struct dt_key *key)
 508 {
 509         struct iam_container  *bag;
 510         struct iam_iterator   *it = &oti->oti_idx_it;
 511         struct iam_path_descr *ipd;
 512         int                    rc;
 513         ENTRY;
 514
 515         LASSERT(oi);
 516         LASSERT(oi->oi_inode);
 517
 518         bag = &oi->oi_dir.od_container;
 519         ipd = osd_idx_ipd_get(oti->oti_env, bag);
 520         if (IS_ERR(ipd))
 521                 RETURN(-ENOMEM);
 522
 523         /* got ipd now we can start iterator. */
 524         iam_it_init(it, bag, 0, ipd);
 525
 526         rc = iam_it_get(it, (struct iam_key *)key);
 527         if (rc > 0)
 528                 iam_reccpy(&it->ii_path.ip_leaf, (struct iam_rec *)rec);
 529         iam_it_put(it);
 530         iam_it_fini(it);
 531         osd_ipd_put(oti->oti_env, bag, ipd);
 532
 533         LINVRNT(osd_invariant(obj));
 534
 535         RETURN(rc);
 536 }
 537
 538 int fid_is_on_ost(struct osd_thread_info *info, struct osd_device *osd,
 539                   const struct lu_fid *fid, enum oi_check_flags flags)
 540 {
 541         struct lu_seq_range     *range = &info->oti_seq_range;
 542         int                     rc;
 543         ENTRY;
 544
 545         if (flags & OI_KNOWN_ON_OST)
 546                 RETURN(1);
 547
 548         if (unlikely(fid_is_local_file(fid) || fid_is_igif(fid) ||
 549                      fid_is_llog(fid)) || fid_is_name_llog(fid) ||
 550                      fid_is_quota(fid))
 551                 RETURN(0);
 552
 553         if (fid_is_idif(fid) || fid_is_last_id(fid))
 554                 RETURN(1);
 555
 556         if (!(flags & OI_CHECK_FLD))
 557                 RETURN(0);
 558
 559         if (osd_seq_site(osd)->ss_server_fld == NULL)
 560                 RETURN(0);
 561
 562         rc = osd_fld_lookup(info->oti_env, osd, fid_seq(fid), range);
 563         if (rc != 0) {
 564                 /* During upgrade, OST FLDB might not be loaded because
 565                  * OST FLDB is not created until 2.6, so if some DNE
 566                  * filesystem upgrade from 2.5 to 2.7/2.8, they will
 567                  * not be able to find the sequence from local FLDB
 568                  * cache see fld_index_init(). */
 569                 if (rc == -ENOENT && osd->od_is_ost)
 570                         RETURN(1);
 571
 572                 if (rc != -ENOENT)
 573                         CERROR("%s: lookup FLD "DFID": rc = %d\n",
 574                                osd_name(osd), PFID(fid), rc);
 575                 RETURN(0);
 576         }
 577
 578         if (fld_range_is_ost(range))
 579                 RETURN(1);
 580
 581         RETURN(0);
 582 }
 583
 584 static int __osd_oi_lookup(struct osd_thread_info *info, struct osd_device *osd,
 585                            const struct lu_fid *fid, struct osd_inode_id *id)
 586 {
 587         struct lu_fid *oi_fid = &info->oti_fid2;
 588         int            rc;
 589
 590         fid_cpu_to_be(oi_fid, fid);
 591         rc = osd_oi_iam_lookup(info, osd_fid2oi(osd, fid), (struct dt_rec *)id,
 592                                (const struct dt_key *)oi_fid);
 593         if (rc > 0) {
 594                 osd_id_unpack(id, id);
 595                 rc = 0;
 596         } else if (rc == 0) {
 597                 rc = -ENOENT;
 598         }
 599         return rc;
 600 }
 601
 602 int osd_oi_lookup(struct osd_thread_info *info, struct osd_device *osd,
 603                   const struct lu_fid *fid, struct osd_inode_id *id,
 604                   enum oi_check_flags flags)
 605 {
 606         if (unlikely(fid_is_last_id(fid)))
 607                 return osd_obj_spec_lookup(info, osd, fid, id);
 608
 609         if (fid_is_on_ost(info, osd, fid, flags) || fid_is_llog(fid))
 610                 return osd_obj_map_lookup(info, osd, fid, id);
 611
 612
 613         if (unlikely(fid_seq(fid) == FID_SEQ_LOCAL_FILE)) {
 614                 int rc;
 615                 if (fid_is_fs_root(fid)) {
 616                         osd_id_gen(id, osd_sb(osd)->s_root->d_inode->i_ino,
 617                                    osd_sb(osd)->s_root->d_inode->i_generation);
 618                         return 0;
 619                 }
 620                 if (unlikely(fid_is_acct(fid)))
 621                         return osd_acct_obj_lookup(info, osd, fid, id);
 622
 623                 /* For other special FIDs, try OI first, then do spec lookup */
 624                 rc = __osd_oi_lookup(info, osd, fid, id);
 625                 if (rc == -ENOENT)
 626                         return osd_obj_spec_lookup(info, osd, fid, id);
 627                 return rc;
 628         }
 629
 630         if (!osd->od_igif_inoi && fid_is_igif(fid)) {
 631                 osd_id_gen(id, lu_igif_ino(fid), lu_igif_gen(fid));
 632                 return 0;
 633         }
 634
 635         return __osd_oi_lookup(info, osd, fid, id);
 636 }
 637
 638 static int osd_oi_iam_refresh(struct osd_thread_info *oti, struct osd_oi *oi,
 639                              const struct dt_rec *rec, const struct dt_key *key,
 640                              handle_t *th, bool insert)
 641 {
 642         struct iam_container    *bag;
 643         struct iam_path_descr   *ipd;
 644         int                     rc;
 645         ENTRY;
 646
 647         LASSERT(oi);
 648         LASSERT(oi->oi_inode);
 649         ll_vfs_dq_init(oi->oi_inode);
 650
 651         bag = &oi->oi_dir.od_container;
 652         ipd = osd_idx_ipd_get(oti->oti_env, bag);
 653         if (unlikely(ipd == NULL))
 654                 RETURN(-ENOMEM);
 655
 656         LASSERT(th != NULL);
 657         LASSERT(th->h_transaction != NULL);
 658         if (insert)
 659                 rc = iam_insert(th, bag, (const struct iam_key *)key,
 660                                 (const struct iam_rec *)rec, ipd);
 661         else
 662                 rc = iam_update(th, bag, (const struct iam_key *)key,
 663                                 (const struct iam_rec *)rec, ipd);
 664         osd_ipd_put(oti->oti_env, bag, ipd);
 665         LINVRNT(osd_invariant(obj));
 666         RETURN(rc);
 667 }
 668
 669 int osd_oi_insert(struct osd_thread_info *info, struct osd_device *osd,
 670                   const struct lu_fid *fid, const struct osd_inode_id *id,
 671                   handle_t *th, enum oi_check_flags flags, bool *exist)
 672 {
 673         struct lu_fid       *oi_fid = &info->oti_fid2;
 674         struct osd_inode_id *oi_id  = &info->oti_id2;
 675         int                  rc     = 0;
 676
 677         if (unlikely(fid_is_last_id(fid)))
 678                 return osd_obj_spec_insert(info, osd, fid, id, th);
 679
 680         if (fid_is_on_ost(info, osd, fid, flags) || fid_is_llog(fid))
 681                 return osd_obj_map_insert(info, osd, fid, id, th);
 682
 683         fid_cpu_to_be(oi_fid, fid);
 684         osd_id_pack(oi_id, id);
 685         rc = osd_oi_iam_refresh(info, osd_fid2oi(osd, fid),
 686                                (const struct dt_rec *)oi_id,
 687                                (const struct dt_key *)oi_fid, th, true);
 688         if (rc != 0) {
 689                 struct inode *inode;
 690                 struct lustre_mdt_attrs *lma = &info->oti_ost_attrs.loa_lma;
 691
 692                 if (rc != -EEXIST)
 693                         return rc;
 694
 695                 rc = osd_oi_lookup(info, osd, fid, oi_id, 0);
 696                 if (rc != 0)
 697                         return rc;
 698
 699                 if (unlikely(osd_id_eq(id, oi_id)))
 700                         return 1;
 701
 702                 /* Check whether the mapping for oi_id is valid or not. */
 703                 inode = osd_iget(info, osd, oi_id);
 704                 if (IS_ERR(inode)) {
 705                         rc = PTR_ERR(inode);
 706                         if (rc == -ENOENT || rc == -ESTALE)
 707                                 goto update;
 708                         return rc;
 709                 }
 710
 711                 /* The EA inode should NOT be in OI, old OI scrub may added
 712                  * such OI mapping by wrong, replace it. */
 713                 if (unlikely(osd_is_ea_inode(inode))) {
 714                         iput(inode);
 715                         goto update;
 716                 }
 717
 718                 rc = osd_get_lma(info, inode, &info->oti_obj_dentry,
 719                                  &info->oti_ost_attrs);
 720                 iput(inode);
 721                 if (rc == -ENODATA)
 722                         goto update;
 723
 724                 if (rc != 0)
 725                         return rc;
 726
 727                 if (!(lma->lma_compat & LMAC_NOT_IN_OI) &&
 728                     lu_fid_eq(fid, &lma->lma_self_fid)) {
 729                         CERROR("%s: the FID "DFID" is used by two objects: "
 730                                "%u/%u %u/%u\n", osd_dev2name(osd),
 731                                PFID(fid), oi_id->oii_ino, oi_id->oii_gen,
 732                                id->oii_ino, id->oii_gen);
 733                         return -EEXIST;
 734                 }
 735
 736 update:
 737                 osd_id_pack(oi_id, id);
 738                 rc = osd_oi_iam_refresh(info, osd_fid2oi(osd, fid),
 739                                         (const struct dt_rec *)oi_id,
 740                                         (const struct dt_key *)oi_fid, th, false);
 741                 if (rc != 0)
 742                         return rc;
 743
 744                 if (exist != NULL)
 745                         *exist = true;
 746         }
 747
 748         if (unlikely(fid_seq(fid) == FID_SEQ_LOCAL_FILE))
 749                 rc = osd_obj_spec_insert(info, osd, fid, id, th);
 750         return rc;
 751 }
 752
 753 static int osd_oi_iam_delete(struct osd_thread_info *oti, struct osd_oi *oi,
 754                              const struct dt_key *key, handle_t *th)
 755 {
 756         struct iam_container    *bag;
 757         struct iam_path_descr   *ipd;
 758         int                      rc;
 759         ENTRY;
 760
 761         LASSERT(oi);
 762         LASSERT(oi->oi_inode);
 763         ll_vfs_dq_init(oi->oi_inode);
 764
 765         bag = &oi->oi_dir.od_container;
 766         ipd = osd_idx_ipd_get(oti->oti_env, bag);
 767         if (unlikely(ipd == NULL))
 768                 RETURN(-ENOMEM);
 769
 770         LASSERT(th != NULL);
 771         LASSERT(th->h_transaction != NULL);
 772
 773         rc = iam_delete(th, bag, (const struct iam_key *)key, ipd);
 774         osd_ipd_put(oti->oti_env, bag, ipd);
 775         LINVRNT(osd_invariant(obj));
 776         RETURN(rc);
 777 }
 778
 779 int osd_oi_delete(struct osd_thread_info *info,
 780                   struct osd_device *osd, const struct lu_fid *fid,
 781                   handle_t *th, enum oi_check_flags flags)
 782 {
 783         struct lu_fid *oi_fid = &info->oti_fid2;
 784
 785         /* clear idmap cache */
 786         if (lu_fid_eq(fid, &info->oti_cache.oic_fid))
 787                 fid_zero(&info->oti_cache.oic_fid);
 788
 789         if (fid_is_last_id(fid))
 790                 return 0;
 791
 792         if (fid_is_on_ost(info, osd, fid, flags) || fid_is_llog(fid))
 793                 return osd_obj_map_delete(info, osd, fid, th);
 794
 795         fid_cpu_to_be(oi_fid, fid);
 796         return osd_oi_iam_delete(info, osd_fid2oi(osd, fid),
 797                                  (const struct dt_key *)oi_fid, th);
 798 }
 799
 800 int osd_oi_update(struct osd_thread_info *info, struct osd_device *osd,
 801                   const struct lu_fid *fid, const struct osd_inode_id *id,
 802                   handle_t *th, enum oi_check_flags flags)
 803 {
 804         struct lu_fid       *oi_fid = &info->oti_fid2;
 805         struct osd_inode_id *oi_id  = &info->oti_id2;
 806         int                  rc     = 0;
 807
 808         if (unlikely(fid_is_last_id(fid)))
 809                 return osd_obj_spec_update(info, osd, fid, id, th);
 810
 811         if (fid_is_on_ost(info, osd, fid, flags) || fid_is_llog(fid))
 812                 return osd_obj_map_update(info, osd, fid, id, th);
 813
 814         fid_cpu_to_be(oi_fid, fid);
 815         osd_id_pack(oi_id, id);
 816         rc = osd_oi_iam_refresh(info, osd_fid2oi(osd, fid),
 817                                (const struct dt_rec *)oi_id,
 818                                (const struct dt_key *)oi_fid, th, false);
 819         if (rc != 0)
 820                 return rc;
 821
 822         if (unlikely(fid_seq(fid) == FID_SEQ_LOCAL_FILE))
 823                 rc = osd_obj_spec_update(info, osd, fid, id, th);
 824         return rc;
 825 }
 826
 827 int osd_oi_mod_init(void)
 828 {
 829         if (osd_oi_count == 0 || osd_oi_count > OSD_OI_FID_NR_MAX)
 830                 osd_oi_count = OSD_OI_FID_NR;
 831
 832         if ((osd_oi_count & (osd_oi_count - 1)) != 0) {
 833                 LCONSOLE_WARN("Round up oi_count %d to power2 %d\n",
 834                               osd_oi_count, size_roundup_power2(osd_oi_count));
 835                 osd_oi_count = size_roundup_power2(osd_oi_count);
 836         }
 837
 838         mutex_init(&oi_init_lock);
 839         return 0;
 840 }