lustre/obdclass/lu_object.c

   1 /* -*- mode: c; c-basic-offset: 8; indent-tabs-mode: nil; -*-
   2  * vim:expandtab:shiftwidth=8:tabstop=8:
   3  *
   4  * Lustre Object.
   5  *
   6  *  Copyright (C) 2006 Cluster File Systems, Inc.
   7  *   Author: Nikita Danilov <nikita@clusterfs.com>
   8  *
   9  *   This file is part of the Lustre file system, http://www.lustre.org
  10  *   Lustre is a trademark of Cluster File Systems, Inc.
  11  *
  12  *   You may have signed or agreed to another license before downloading
  13  *   this software.  If so, you are bound by the terms and conditions
  14  *   of that agreement, and the following does not apply to you.  See the
  15  *   LICENSE file included with this distribution for more information.
  16  *
  17  *   If you did not agree to a different license, then this copy of Lustre
  18  *   is open source software; you can redistribute it and/or modify it
  19  *   under the terms of version 2 of the GNU General Public License as
  20  *   published by the Free Software Foundation.
  21  *
  22  *   In either case, Lustre is distributed in the hope that it will be
  23  *   useful, but WITHOUT ANY WARRANTY; without even the implied warranty
  24  *   of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  25  *   license text for more details.
  26  *
  27  * These are the only exported functions, they provide some generic
  28  * infrastructure for managing object devices
  29  */
  30
  31 #define DEBUG_SUBSYSTEM S_CLASS
  32 #ifndef EXPORT_SYMTAB
  33 # define EXPORT_SYMTAB
  34 #endif
  35
  36 #include <linux/seq_file.h>
  37 #include <linux/module.h>
  38 #include <obd_support.h>
  39 #include <lustre_disk.h>
  40 #include <lustre_fid.h>
  41 #include <lu_object.h>
  42 #include <libcfs/list.h>
  43
  44 static void lu_object_free(const struct lu_env *env, struct lu_object *o);
  45
  46 /*
  47  * Decrease reference counter on object. If last reference is freed, return
  48  * object to the cache, unless lu_object_is_dying(o) holds. In the latter
  49  * case, free object immediately.
  50  */
  51 void lu_object_put(const struct lu_env *env, struct lu_object *o)
  52 {
  53         struct lu_object_header *top;
  54         struct lu_site          *site;
  55         struct lu_object        *orig;
  56         int                      kill_it;
  57
  58         top = o->lo_header;
  59         site = o->lo_dev->ld_site;
  60         orig = o;
  61         kill_it = 0;
  62         spin_lock(&site->ls_guard);
  63         if (-- top->loh_ref == 0) {
  64                 /*
  65                  * When last reference is released, iterate over object
  66                  * layers, and notify them that object is no longer busy.
  67                  */
  68                 list_for_each_entry_reverse(o, &top->loh_layers, lo_linkage) {
  69                         if (o->lo_ops->loo_object_release != NULL)
  70                                 o->lo_ops->loo_object_release(env, o);
  71                 }
  72                 -- site->ls_busy;
  73                 if (lu_object_is_dying(top)) {
  74                         /*
  75                          * If object is dying (will not be cached), removed it
  76                          * from hash table and LRU.
  77                          *
  78                          * This is done with hash table and LRU lists
  79                          * locked. As the only way to acquire first reference
  80                          * to previously unreferenced object is through
  81                          * hash-table lookup (lu_object_find()), or LRU
  82                          * scanning (lu_site_purge()), that are done under
  83                          * hash-table and LRU lock, no race with concurrent
  84                          * object lookup is possible and we can safely destroy
  85                          * object below.
  86                          */
  87                         hlist_del_init(&top->loh_hash);
  88                         list_del_init(&top->loh_lru);
  89                         kill_it = 1;
  90                 }
  91         }
  92         spin_unlock(&site->ls_guard);
  93         if (kill_it)
  94                 /*
  95                  * Object was already removed from hash and lru above, can
  96                  * kill it.
  97                  */
  98                 lu_object_free(env, orig);
  99 }
 100 EXPORT_SYMBOL(lu_object_put);
 101
 102 /*
 103  * Allocate new object.
 104  *
 105  * This follows object creation protocol, described in the comment within
 106  * struct lu_device_operations definition.
 107  */
 108 static struct lu_object *lu_object_alloc(const struct lu_env *env,
 109                                          struct lu_site *s,
 110                                          const struct lu_fid *f,
 111                                          const struct lustre_capa *capa)
 112 {
 113         struct lu_object *scan;
 114         struct lu_object *top;
 115         struct list_head *layers;
 116         int clean;
 117         int result;
 118
 119         /*
 120          * Create top-level object slice. This will also create
 121          * lu_object_header.
 122          */
 123         top = s->ls_top_dev->ld_ops->ldo_object_alloc(env,
 124                                                       NULL, s->ls_top_dev);
 125         if (IS_ERR(top))
 126                 RETURN(top);
 127         s->ls_total ++;
 128         /*
 129          * This is the only place where object fid is assigned. It's constant
 130          * after this point.
 131          */
 132         top->lo_header->loh_fid  = *f;
 133         if (capa == BYPASS_CAPA)
 134                 lu_object_bypass_capa(top);
 135         else
 136                 top->lo_header->loh_capa = *capa;
 137         layers = &top->lo_header->loh_layers;
 138         do {
 139                 /*
 140                  * Call ->loo_object_init() repeatedly, until no more new
 141                  * object slices are created.
 142                  */
 143                 clean = 1;
 144                 list_for_each_entry(scan, layers, lo_linkage) {
 145                         if (scan->lo_flags & LU_OBJECT_ALLOCATED)
 146                                 continue;
 147                         clean = 0;
 148                         scan->lo_header = top->lo_header;
 149                         result = scan->lo_ops->loo_object_init(env, scan);
 150                         if (result != 0) {
 151                                 lu_object_free(env, top);
 152                                 RETURN(ERR_PTR(result));
 153                         }
 154                         scan->lo_flags |= LU_OBJECT_ALLOCATED;
 155                 }
 156         } while (!clean);
 157
 158         list_for_each_entry_reverse(scan, layers, lo_linkage) {
 159                 if (scan->lo_ops->loo_object_start != NULL) {
 160                         result = scan->lo_ops->loo_object_start(env, scan);
 161                         if (result != 0) {
 162                                 lu_object_free(env, top);
 163                                 RETURN(ERR_PTR(result));
 164                         }
 165                 }
 166         }
 167
 168         s->ls_stats.s_created ++;
 169         RETURN(top);
 170 }
 171
 172 /*
 173  * Free object.
 174  */
 175 static void lu_object_free(const struct lu_env *env, struct lu_object *o)
 176 {
 177         struct list_head splice;
 178         struct lu_object *scan;
 179
 180         /*
 181          * First call ->loo_object_delete() method to release all resources.
 182          */
 183         list_for_each_entry_reverse(scan,
 184                                     &o->lo_header->loh_layers, lo_linkage) {
 185                 if (scan->lo_ops->loo_object_delete != NULL)
 186                         scan->lo_ops->loo_object_delete(env, scan);
 187         }
 188         -- o->lo_dev->ld_site->ls_total;
 189         /*
 190          * Then, splice object layers into stand-alone list, and call
 191          * ->loo_object_free() on all layers to free memory. Splice is
 192          * necessary, because lu_object_header is freed together with the
 193          * top-level slice.
 194          */
 195         INIT_LIST_HEAD(&splice);
 196         list_splice_init(&o->lo_header->loh_layers, &splice);
 197         while (!list_empty(&splice)) {
 198                 o = container_of0(splice.next, struct lu_object, lo_linkage);
 199                 list_del_init(&o->lo_linkage);
 200                 LASSERT(o->lo_ops->loo_object_free != NULL);
 201                 o->lo_ops->loo_object_free(env, o);
 202         }
 203 }
 204
 205 /*
 206  * Free @nr objects from the cold end of the site LRU list.
 207  */
 208 void lu_site_purge(const struct lu_env *env, struct lu_site *s, int nr)
 209 {
 210         struct list_head         dispose;
 211         struct lu_object_header *h;
 212         struct lu_object_header *temp;
 213
 214         INIT_LIST_HEAD(&dispose);
 215         /*
 216          * Under LRU list lock, scan LRU list and move unreferenced objects to
 217          * the dispose list, removing them from LRU and hash table.
 218          */
 219         spin_lock(&s->ls_guard);
 220         list_for_each_entry_safe(h, temp, &s->ls_lru, loh_lru) {
 221                 if (nr-- == 0)
 222                         break;
 223                 if (h->loh_ref > 0)
 224                         continue;
 225                 hlist_del_init(&h->loh_hash);
 226                 list_move(&h->loh_lru, &dispose);
 227         }
 228         spin_unlock(&s->ls_guard);
 229         /*
 230          * Free everything on the dispose list. This is safe against races due
 231          * to the reasons described in lu_object_put().
 232          */
 233         while (!list_empty(&dispose)) {
 234                 h = container_of0(dispose.next,
 235                                  struct lu_object_header, loh_lru);
 236                 list_del_init(&h->loh_lru);
 237                 lu_object_free(env, lu_object_top(h));
 238                 s->ls_stats.s_lru_purged ++;
 239         }
 240 }
 241 EXPORT_SYMBOL(lu_site_purge);
 242
 243 /*
 244  * Object printing.
 245  *
 246  * Code below has to jump through certain loops to output object description
 247  * into libcfs_debug_msg-based log. The problem is that lu_object_print()
 248  * composes object description from strings that are parts of _lines_ of
 249  * output (i.e., strings that are not terminated by newline). This doesn't fit
 250  * very well into libcfs_debug_msg() interface that assumes that each message
 251  * supplied to it is a self-contained output line.
 252  *
 253  * To work around this, strings are collected in a temporary buffer
 254  * (implemented as a value of lu_cdebug_key key), until terminating newline
 255  * character is detected.
 256  *
 257  */
 258
 259 enum {
 260         /*
 261          * Maximal line size.
 262          *
 263          * XXX overflow is not handled correctly.
 264          */
 265         LU_CDEBUG_LINE = 256
 266 };
 267
 268 struct lu_cdebug_data {
 269         /*
 270          * Temporary buffer.
 271          */
 272         char lck_area[LU_CDEBUG_LINE];
 273 };
 274
 275 static void *lu_cdebug_key_init(const struct lu_context *ctx,
 276                                 struct lu_context_key *key)
 277 {
 278         struct lu_cdebug_data *value;
 279
 280         OBD_ALLOC_PTR(value);
 281         if (value == NULL)
 282                 value = ERR_PTR(-ENOMEM);
 283         return value;
 284 }
 285
 286 static void lu_cdebug_key_fini(const struct lu_context *ctx,
 287                                struct lu_context_key *key, void *data)
 288 {
 289         struct lu_cdebug_data *value = data;
 290         OBD_FREE_PTR(value);
 291 }
 292
 293 /*
 294  * Key, holding temporary buffer. This key is registered very early by
 295  * lu_global_init().
 296  */
 297 static struct lu_context_key lu_cdebug_key = {
 298         .lct_tags = LCT_MD_THREAD|LCT_DT_THREAD|LCT_CL_THREAD,
 299         .lct_init = lu_cdebug_key_init,
 300         .lct_fini = lu_cdebug_key_fini
 301 };
 302
 303 /*
 304  * Printer function emitting messages through libcfs_debug_msg().
 305  */
 306 int lu_cdebug_printer(const struct lu_env *env,
 307                       void *cookie, const char *format, ...)
 308 {
 309         struct lu_cdebug_print_info *info = cookie;
 310         struct lu_cdebug_data       *key;
 311         int used;
 312         int complete;
 313         va_list args;
 314
 315         va_start(args, format);
 316
 317         key = lu_context_key_get(&env->le_ctx, &lu_cdebug_key);
 318         LASSERT(key != NULL);
 319
 320         used = strlen(key->lck_area);
 321         complete = format[strlen(format) - 1] == '\n';
 322         /*
 323          * Append new chunk to the buffer.
 324          */
 325         vsnprintf(key->lck_area + used,
 326                   ARRAY_SIZE(key->lck_area) - used, format, args);
 327         if (complete) {
 328                 libcfs_debug_msg(NULL, info->lpi_subsys, info->lpi_mask,
 329                                  (char *)info->lpi_file, info->lpi_fn,
 330                                  info->lpi_line, "%s", key->lck_area);
 331                 key->lck_area[0] = 0;
 332         }
 333         va_end(args);
 334         return 0;
 335 }
 336 EXPORT_SYMBOL(lu_cdebug_printer);
 337
 338 /*
 339  * Print object header.
 340  */
 341 static void lu_object_header_print(const struct lu_env *env,
 342                                    void *cookie, lu_printer_t printer,
 343                                    const struct lu_object_header *hdr)
 344 {
 345         (*printer)(env, cookie, "header@%p[%#lx, %d, "DFID"%s%s]",
 346                    hdr, hdr->loh_flags, hdr->loh_ref, PFID(&hdr->loh_fid),
 347                    hlist_unhashed(&hdr->loh_hash) ? "" : " hash",
 348                    list_empty(&hdr->loh_lru) ? "" : " lru");
 349 }
 350
 351 /*
 352  * Print human readable representation of the @o to the @printer.
 353  */
 354 void lu_object_print(const struct lu_env *env, void *cookie,
 355                      lu_printer_t printer, const struct lu_object *o)
 356 {
 357         static const char ruler[] = "........................................";
 358         struct lu_object_header *top;
 359         int depth;
 360
 361         top = o->lo_header;
 362         lu_object_header_print(env, cookie, printer, top);
 363         (*printer)(env, cookie, "\n");
 364         list_for_each_entry(o, &top->loh_layers, lo_linkage) {
 365                 depth = o->lo_depth + 4;
 366                 LASSERT(o->lo_ops->loo_object_print != NULL);
 367                 /*
 368                  * print `.' @depth times.
 369                  */
 370                 (*printer)(env, cookie, "%*.*s", depth, depth, ruler);
 371                 o->lo_ops->loo_object_print(env, cookie, printer, o);
 372                 (*printer)(env, cookie, "\n");
 373         }
 374 }
 375 EXPORT_SYMBOL(lu_object_print);
 376
 377 /*
 378  * Check object consistency.
 379  */
 380 int lu_object_invariant(const struct lu_object *o)
 381 {
 382         struct lu_object_header *top;
 383
 384         top = o->lo_header;
 385         list_for_each_entry(o, &top->loh_layers, lo_linkage) {
 386                 if (o->lo_ops->loo_object_invariant != NULL &&
 387                     !o->lo_ops->loo_object_invariant(o))
 388                         return 0;
 389         }
 390         return 1;
 391 }
 392 EXPORT_SYMBOL(lu_object_invariant);
 393
 394 static struct lu_object *htable_lookup(struct lu_site *s,
 395                                        const struct hlist_head *bucket,
 396                                        const struct lu_fid *f)
 397 {
 398         struct lu_object_header *h;
 399         struct hlist_node *scan;
 400
 401         hlist_for_each_entry(h, scan, bucket, loh_hash) {
 402                 s->ls_stats.s_cache_check ++;
 403                 if (lu_fid_eq(&h->loh_fid, f) && !lu_object_is_dying(h)) {
 404                         /* bump reference count... */
 405                         if (h->loh_ref ++ == 0)
 406                                 ++ s->ls_busy;
 407                         /* and move to the head of the LRU */
 408                         list_move_tail(&h->loh_lru, &s->ls_lru);
 409                         s->ls_stats.s_cache_hit ++;
 410                         return lu_object_top(h);
 411                 }
 412         }
 413         s->ls_stats.s_cache_miss ++;
 414         return NULL;
 415 }
 416
 417 static __u32 fid_hash(const struct lu_fid *f)
 418 {
 419         /* all objects with same id and different versions will belong to same
 420          * collisions list. */
 421         return (fid_seq(f) - 1) * LUSTRE_SEQ_MAX_WIDTH + fid_oid(f);
 422 }
 423
 424 /*
 425  * Search cache for an object with the fid @f. If such object is found, return
 426  * it. Otherwise, create new object, insert it into cache and return it. In
 427  * any case, additional reference is acquired on the returned object.
 428  */
 429 struct lu_object *lu_object_find(const struct lu_env *env,
 430                                  struct lu_site *s, const struct lu_fid *f,
 431                                  struct lustre_capa *capa)
 432 {
 433         struct lu_object  *o;
 434         struct lu_object  *shadow;
 435         struct hlist_head *bucket;
 436         int                rc;
 437
 438         /*
 439          * This uses standard index maintenance protocol:
 440          *
 441          *     - search index under lock, and return object if found;
 442          *     - otherwise, unlock index, allocate new object;
 443          *     - lock index and search again;
 444          *     - if nothing is found (usual case), insert newly created
 445          *       object into index;
 446          *     - otherwise (race: other thread inserted object), free
 447          *       object just allocated.
 448          *     - unlock index;
 449          *     - return object.
 450          */
 451
 452         bucket = s->ls_hash + (fid_hash(f) & s->ls_hash_mask);
 453         spin_lock(&s->ls_guard);
 454         o = htable_lookup(s, bucket, f);
 455
 456         spin_unlock(&s->ls_guard);
 457         if (o != NULL) {
 458                 if (capa == BYPASS_CAPA) {
 459                         o->lo_header->loh_capa_bypass = 1;
 460                 } else {
 461                         rc = lu_object_auth(env, o, capa,
 462                                             CAPA_OPC_INDEX_LOOKUP);
 463                         if (rc)
 464                                 return ERR_PTR(rc);
 465                         if (capa)
 466                                 o->lo_header->loh_capa = *capa;
 467                 }
 468                 return o;
 469         }
 470
 471         /*
 472          * Allocate new object. This may result in rather complicated
 473          * operations, including fld queries, inode loading, etc.
 474          */
 475         o = lu_object_alloc(env, s, f, capa);
 476         if (IS_ERR(o))
 477                 return o;
 478
 479         LASSERT(lu_fid_eq(lu_object_fid(o), f));
 480
 481         spin_lock(&s->ls_guard);
 482         shadow = htable_lookup(s, bucket, f);
 483         if (shadow == NULL) {
 484                 hlist_add_head(&o->lo_header->loh_hash, bucket);
 485                 list_add_tail(&o->lo_header->loh_lru, &s->ls_lru);
 486                 ++ s->ls_busy;
 487                 shadow = o;
 488                 o = NULL;
 489         } else
 490                 s->ls_stats.s_cache_race ++;
 491         spin_unlock(&s->ls_guard);
 492         if (o != NULL)
 493                 lu_object_free(env, o);
 494         return shadow;
 495 }
 496 EXPORT_SYMBOL(lu_object_find);
 497
 498 int lu_object_auth(const struct lu_env *env, const struct lu_object *o,
 499                    struct lustre_capa *capa, __u64 opc)
 500 {
 501         struct lu_object_header *top = o->lo_header;
 502         int rc;
 503
 504         list_for_each_entry(o, &top->loh_layers, lo_linkage) {
 505                 if (o->lo_ops->loo_object_auth) {
 506                         rc = o->lo_ops->loo_object_auth(env, o, capa, opc);
 507                         if (rc)
 508                                 return rc;
 509                 }
 510         }
 511
 512         return 0;
 513 }
 514 EXPORT_SYMBOL(lu_object_auth);
 515
 516 enum {
 517         LU_SITE_HTABLE_BITS = 8,
 518         LU_SITE_HTABLE_SIZE = (1 << LU_SITE_HTABLE_BITS),
 519         LU_SITE_HTABLE_MASK = LU_SITE_HTABLE_SIZE - 1
 520 };
 521
 522 /*
 523  * Initialize site @s, with @d as the top level device.
 524  */
 525 int lu_site_init(struct lu_site *s, struct lu_device *top)
 526 {
 527         int result;
 528         ENTRY;
 529
 530         memset(s, 0, sizeof *s);
 531         spin_lock_init(&s->ls_guard);
 532         CFS_INIT_LIST_HEAD(&s->ls_lru);
 533         s->ls_top_dev = top;
 534         top->ld_site = s;
 535         lu_device_get(top);
 536         /*
 537          * XXX nikita: fixed size hash-table.
 538          */
 539         s->ls_hash_mask = LU_SITE_HTABLE_MASK;
 540         OBD_ALLOC(s->ls_hash, LU_SITE_HTABLE_SIZE * sizeof s->ls_hash[0]);
 541         if (s->ls_hash != NULL) {
 542                 int i;
 543                 for (i = 0; i < LU_SITE_HTABLE_SIZE; i++)
 544                         INIT_HLIST_HEAD(&s->ls_hash[i]);
 545                 result = 0;
 546         } else {
 547                 result = -ENOMEM;
 548         }
 549
 550         RETURN(result);
 551 }
 552 EXPORT_SYMBOL(lu_site_init);
 553
 554 /*
 555  * Finalize @s and release its resources.
 556  */
 557 void lu_site_fini(struct lu_site *s)
 558 {
 559         LASSERT(list_empty(&s->ls_lru));
 560         LASSERT(s->ls_total == 0);
 561         LASSERT(s->ls_busy == 0);
 562
 563         if (s->ls_hash != NULL) {
 564                 int i;
 565                 for (i = 0; i < LU_SITE_HTABLE_SIZE; i++)
 566                         LASSERT(hlist_empty(&s->ls_hash[i]));
 567                 OBD_FREE(s->ls_hash,
 568                          LU_SITE_HTABLE_SIZE * sizeof s->ls_hash[0]);
 569                 s->ls_hash = NULL;
 570        }
 571        if (s->ls_top_dev != NULL) {
 572                s->ls_top_dev->ld_site = NULL;
 573                lu_device_put(s->ls_top_dev);
 574                s->ls_top_dev = NULL;
 575        }
 576  }
 577 EXPORT_SYMBOL(lu_site_fini);
 578
 579 /*
 580  * Acquire additional reference on device @d
 581  */
 582 void lu_device_get(struct lu_device *d)
 583 {
 584         atomic_inc(&d->ld_ref);
 585 }
 586 EXPORT_SYMBOL(lu_device_get);
 587
 588 /*
 589  * Release reference on device @d.
 590  */
 591 void lu_device_put(struct lu_device *d)
 592 {
 593         atomic_dec(&d->ld_ref);
 594 }
 595 EXPORT_SYMBOL(lu_device_put);
 596
 597 /*
 598  * Initialize device @d of type @t.
 599  */
 600 int lu_device_init(struct lu_device *d, struct lu_device_type *t)
 601 {
 602         memset(d, 0, sizeof *d);
 603         atomic_set(&d->ld_ref, 0);
 604         d->ld_type = t;
 605         return 0;
 606 }
 607 EXPORT_SYMBOL(lu_device_init);
 608
 609 /*
 610  * Finalize device @d.
 611  */
 612 void lu_device_fini(struct lu_device *d)
 613 {
 614         if (d->ld_obd != NULL)
 615                 /* finish lprocfs */
 616                 lprocfs_obd_cleanup(d->ld_obd);
 617
 618         LASSERTF(atomic_read(&d->ld_ref) == 0,
 619                  "Refcount is %u\n", atomic_read(&d->ld_ref));
 620 }
 621 EXPORT_SYMBOL(lu_device_fini);
 622
 623 /*
 624  * Initialize object @o that is part of compound object @h and was created by
 625  * device @d.
 626  */
 627 int lu_object_init(struct lu_object *o,
 628                    struct lu_object_header *h, struct lu_device *d)
 629 {
 630         memset(o, 0, sizeof *o);
 631         o->lo_header = h;
 632         o->lo_dev    = d;
 633         lu_device_get(d);
 634         CFS_INIT_LIST_HEAD(&o->lo_linkage);
 635         return 0;
 636 }
 637 EXPORT_SYMBOL(lu_object_init);
 638
 639 /*
 640  * Finalize object and release its resources.
 641  */
 642 void lu_object_fini(struct lu_object *o)
 643 {
 644         LASSERT(list_empty(&o->lo_linkage));
 645
 646         if (o->lo_dev != NULL) {
 647                 lu_device_put(o->lo_dev);
 648                 o->lo_dev = NULL;
 649         }
 650 }
 651 EXPORT_SYMBOL(lu_object_fini);
 652
 653 /*
 654  * Add object @o as first layer of compound object @h
 655  *
 656  * This is typically called by the ->ldo_object_alloc() method of top-level
 657  * device.
 658  */
 659 void lu_object_add_top(struct lu_object_header *h, struct lu_object *o)
 660 {
 661         list_move(&o->lo_linkage, &h->loh_layers);
 662 }
 663 EXPORT_SYMBOL(lu_object_add_top);
 664
 665 /*
 666  * Add object @o as a layer of compound object, going after @before.1
 667  *
 668  * This is typically called by the ->ldo_object_alloc() method of
 669  * @before->lo_dev.
 670  */
 671 void lu_object_add(struct lu_object *before, struct lu_object *o)
 672 {
 673         list_move(&o->lo_linkage, &before->lo_linkage);
 674 }
 675 EXPORT_SYMBOL(lu_object_add);
 676
 677 /*
 678  * Initialize compound object.
 679  */
 680 int lu_object_header_init(struct lu_object_header *h)
 681 {
 682         memset(h, 0, sizeof *h);
 683         h->loh_ref = 1;
 684         INIT_HLIST_NODE(&h->loh_hash);
 685         CFS_INIT_LIST_HEAD(&h->loh_lru);
 686         CFS_INIT_LIST_HEAD(&h->loh_layers);
 687         return 0;
 688 }
 689 EXPORT_SYMBOL(lu_object_header_init);
 690
 691 /*
 692  * Finalize compound object.
 693  */
 694 void lu_object_header_fini(struct lu_object_header *h)
 695 {
 696         LASSERT(list_empty(&h->loh_layers));
 697         LASSERT(list_empty(&h->loh_lru));
 698         LASSERT(hlist_unhashed(&h->loh_hash));
 699 }
 700 EXPORT_SYMBOL(lu_object_header_fini);
 701
 702 /*
 703  * Given a compound object, find its slice, corresponding to the device type
 704  * @dtype.
 705  */
 706 struct lu_object *lu_object_locate(struct lu_object_header *h,
 707                                    struct lu_device_type *dtype)
 708 {
 709         struct lu_object *o;
 710
 711         list_for_each_entry(o, &h->loh_layers, lo_linkage) {
 712                 if (o->lo_dev->ld_type == dtype)
 713                         return o;
 714         }
 715         return NULL;
 716 }
 717 EXPORT_SYMBOL(lu_object_locate);
 718
 719 enum {
 720         /*
 721          * Maximal number of tld slots.
 722          */
 723         LU_CONTEXT_KEY_NR = 16
 724 };
 725
 726 static struct lu_context_key *lu_keys[LU_CONTEXT_KEY_NR] = { NULL, };
 727
 728 static spinlock_t lu_keys_guard = SPIN_LOCK_UNLOCKED;
 729
 730 /*
 731  * Register new key.
 732  */
 733 int lu_context_key_register(struct lu_context_key *key)
 734 {
 735         int result;
 736         int i;
 737
 738         LASSERT(key->lct_init != NULL);
 739         LASSERT(key->lct_fini != NULL);
 740         LASSERT(key->lct_tags != 0);
 741
 742         result = -ENFILE;
 743         spin_lock(&lu_keys_guard);
 744         for (i = 0; i < ARRAY_SIZE(lu_keys); ++i) {
 745                 if (lu_keys[i] == NULL) {
 746                         key->lct_index = i;
 747                         atomic_set(&key->lct_used, 1);
 748                         lu_keys[i] = key;
 749                         result = 0;
 750                         break;
 751                 }
 752         }
 753         spin_unlock(&lu_keys_guard);
 754         return result;
 755 }
 756 EXPORT_SYMBOL(lu_context_key_register);
 757
 758 /*
 759  * Deregister key.
 760  */
 761 void lu_context_key_degister(struct lu_context_key *key)
 762 {
 763         LASSERT(atomic_read(&key->lct_used) >= 1);
 764         LASSERT(0 <= key->lct_index && key->lct_index < ARRAY_SIZE(lu_keys));
 765
 766         if (atomic_read(&key->lct_used) > 1)
 767                 CERROR("key has instances.\n");
 768         spin_lock(&lu_keys_guard);
 769         lu_keys[key->lct_index] = NULL;
 770         spin_unlock(&lu_keys_guard);
 771 }
 772 EXPORT_SYMBOL(lu_context_key_degister);
 773
 774 /*
 775  * Return value associated with key @key in context @ctx.
 776  */
 777 void *lu_context_key_get(const struct lu_context *ctx,
 778                          struct lu_context_key *key)
 779 {
 780         LASSERT(0 <= key->lct_index && key->lct_index < ARRAY_SIZE(lu_keys));
 781         return ctx->lc_value[key->lct_index];
 782 }
 783 EXPORT_SYMBOL(lu_context_key_get);
 784
 785 static void keys_fini(struct lu_context *ctx)
 786 {
 787         int i;
 788
 789         if (ctx->lc_value != NULL) {
 790                 for (i = 0; i < ARRAY_SIZE(lu_keys); ++i) {
 791                         if (ctx->lc_value[i] != NULL) {
 792                                 struct lu_context_key *key;
 793
 794                                 key = lu_keys[i];
 795                                 LASSERT(key != NULL);
 796                                 LASSERT(key->lct_fini != NULL);
 797                                 LASSERT(atomic_read(&key->lct_used) > 1);
 798
 799                                 key->lct_fini(ctx, key, ctx->lc_value[i]);
 800                                 atomic_dec(&key->lct_used);
 801                                 ctx->lc_value[i] = NULL;
 802                         }
 803                 }
 804                 OBD_FREE(ctx->lc_value,
 805                          ARRAY_SIZE(lu_keys) * sizeof ctx->lc_value[0]);
 806                 ctx->lc_value = NULL;
 807         }
 808 }
 809
 810 static int keys_fill(const struct lu_context *ctx)
 811 {
 812         int i;
 813
 814         for (i = 0; i < ARRAY_SIZE(lu_keys); ++i) {
 815                 struct lu_context_key *key;
 816
 817                 key = lu_keys[i];
 818                 if (ctx->lc_value[i] == NULL &&
 819                     key != NULL && key->lct_tags & ctx->lc_tags) {
 820                         void *value;
 821
 822                         LASSERT(key->lct_init != NULL);
 823                         LASSERT(key->lct_index == i);
 824
 825                         value = key->lct_init(ctx, key);
 826                         if (IS_ERR(value))
 827                                 return PTR_ERR(value);
 828                         atomic_inc(&key->lct_used);
 829                         ctx->lc_value[i] = value;
 830                 }
 831         }
 832         return 0;
 833 }
 834
 835 static int keys_init(struct lu_context *ctx)
 836 {
 837         int result;
 838
 839         OBD_ALLOC(ctx->lc_value, ARRAY_SIZE(lu_keys) * sizeof ctx->lc_value[0]);
 840         if (ctx->lc_value != NULL)
 841                 result = keys_fill(ctx);
 842         else
 843                 result = -ENOMEM;
 844
 845         if (result != 0)
 846                 keys_fini(ctx);
 847         return result;
 848 }
 849
 850 /*
 851  * Initialize context data-structure. Create values for all keys.
 852  */
 853 int lu_context_init(struct lu_context *ctx, __u32 tags)
 854 {
 855         memset(ctx, 0, sizeof *ctx);
 856         ctx->lc_tags = tags;
 857         return keys_init(ctx);
 858 }
 859 EXPORT_SYMBOL(lu_context_init);
 860
 861 /*
 862  * Finalize context data-structure. Destroy key values.
 863  */
 864 void lu_context_fini(struct lu_context *ctx)
 865 {
 866         keys_fini(ctx);
 867 }
 868 EXPORT_SYMBOL(lu_context_fini);
 869
 870 /*
 871  * Called before entering context.
 872  */
 873 void lu_context_enter(struct lu_context *ctx)
 874 {
 875 }
 876 EXPORT_SYMBOL(lu_context_enter);
 877
 878 /*
 879  * Called after exiting from @ctx
 880  */
 881 void lu_context_exit(struct lu_context *ctx)
 882 {
 883         int i;
 884
 885         if (ctx->lc_value != NULL) {
 886                 for (i = 0; i < ARRAY_SIZE(lu_keys); ++i) {
 887                         if (ctx->lc_value[i] != NULL) {
 888                                 struct lu_context_key *key;
 889
 890                                 key = lu_keys[i];
 891                                 LASSERT(key != NULL);
 892                                 if (key->lct_exit != NULL)
 893                                         key->lct_exit(ctx,
 894                                                       key, ctx->lc_value[i]);
 895                         }
 896                 }
 897         }
 898 }
 899 EXPORT_SYMBOL(lu_context_exit);
 900
 901 /*
 902  * Allocate for context all missing keys that were registered after context
 903  * creation.
 904  */
 905 int lu_context_refill(const struct lu_context *ctx)
 906 {
 907         LASSERT(ctx->lc_value != NULL);
 908         return keys_fill(ctx);
 909 }
 910 EXPORT_SYMBOL(lu_context_refill);
 911
 912 int lu_env_init(struct lu_env *env, struct lu_context *ses, __u32 tags)
 913 {
 914         int result;
 915
 916         env->le_ses = ses;
 917         result = lu_context_init(&env->le_ctx, tags);
 918         if (result == 0)
 919                 lu_context_enter(&env->le_ctx);
 920         return result;
 921 }
 922 EXPORT_SYMBOL(lu_env_init);
 923
 924 void lu_env_fini(struct lu_env *env)
 925 {
 926         lu_context_exit(&env->le_ctx);
 927         lu_context_fini(&env->le_ctx);
 928         env->le_ses = NULL;
 929 }
 930 EXPORT_SYMBOL(lu_env_fini);
 931
 932 /*
 933  * Initialization of global lu_* data.
 934  */
 935 int lu_global_init(void)
 936 {
 937         int result;
 938
 939         result = lu_context_key_register(&lu_cdebug_key);
 940         return result;
 941 }
 942
 943 /*
 944  * Dual to lu_global_init().
 945  */
 946 void lu_global_fini(void)
 947 {
 948         lu_context_key_degister(&lu_cdebug_key);
 949 }
 950
 951 struct lu_buf LU_BUF_NULL = {
 952         .lb_buf = NULL,
 953         .lb_len = 0
 954 };
 955 EXPORT_SYMBOL(LU_BUF_NULL);