fs/btrfs/inode-map.c

   1 /*
   2  * Copyright (C) 2007 Oracle.  All rights reserved.
   3  *
   4  * This program is free software; you can redistribute it and/or
   5  * modify it under the terms of the GNU General Public
   6  * License v2 as published by the Free Software Foundation.
   7  *
   8  * This program is distributed in the hope that it will be useful,
   9  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  10  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  11  * General Public License for more details.
  12  *
  13  * You should have received a copy of the GNU General Public
  14  * License along with this program; if not, write to the
  15  * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  16  * Boston, MA 021110-1307, USA.
  17  */
  18
  19 #include <linux/delay.h>
  20 #include <linux/kthread.h>
  21 #include <linux/pagemap.h>
  22
  23 #include "ctree.h"
  24 #include "disk-io.h"
  25 #include "free-space-cache.h"
  26 #include "inode-map.h"
  27 #include "transaction.h"
  28
  29 static int caching_kthread(void *data)
  30 {
  31         struct btrfs_root *root = data;
  32         struct btrfs_fs_info *fs_info = root->fs_info;
  33         struct btrfs_free_space_ctl *ctl = root->free_ino_ctl;
  34         struct btrfs_key key;
  35         struct btrfs_path *path;
  36         struct extent_buffer *leaf;
  37         u64 last = (u64)-1;
  38         int slot;
  39         int ret;
  40
  41         if (!btrfs_test_opt(fs_info, INODE_MAP_CACHE))
  42                 return 0;
  43
  44         path = btrfs_alloc_path();
  45         if (!path)
  46                 return -ENOMEM;
  47
  48         /* Since the commit root is read-only, we can safely skip locking. */
  49         path->skip_locking = 1;
  50         path->search_commit_root = 1;
  51         path->reada = READA_FORWARD;
  52
  53         key.objectid = BTRFS_FIRST_FREE_OBJECTID;
  54         key.offset = 0;
  55         key.type = BTRFS_INODE_ITEM_KEY;
  56 again:
  57         /* need to make sure the commit_root doesn't disappear */
  58         down_read(&fs_info->commit_root_sem);
  59
  60         ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
  61         if (ret < 0)
  62                 goto out;
  63
  64         while (1) {
  65                 if (btrfs_fs_closing(fs_info))
  66                         goto out;
  67
  68                 leaf = path->nodes[0];
  69                 slot = path->slots[0];
  70                 if (slot >= btrfs_header_nritems(leaf)) {
  71                         ret = btrfs_next_leaf(root, path);
  72                         if (ret < 0)
  73                                 goto out;
  74                         else if (ret > 0)
  75                                 break;
  76
  77                         if (need_resched() ||
  78                             btrfs_transaction_in_commit(fs_info)) {
  79                                 leaf = path->nodes[0];
  80
  81                                 if (WARN_ON(btrfs_header_nritems(leaf) == 0))
  82                                         break;
  83
  84                                 /*
  85                                  * Save the key so we can advances forward
  86                                  * in the next search.
  87                                  */
  88                                 btrfs_item_key_to_cpu(leaf, &key, 0);
  89                                 btrfs_release_path(path);
  90                                 root->ino_cache_progress = last;
  91                                 up_read(&fs_info->commit_root_sem);
  92                                 schedule_timeout(1);
  93                                 goto again;
  94                         } else
  95                                 continue;
  96                 }
  97
  98                 btrfs_item_key_to_cpu(leaf, &key, slot);
  99
 100                 if (key.type != BTRFS_INODE_ITEM_KEY)
 101                         goto next;
 102
 103                 if (key.objectid >= root->highest_objectid)
 104                         break;
 105
 106                 if (last != (u64)-1 && last + 1 != key.objectid) {
 107                         __btrfs_add_free_space(fs_info, ctl, last + 1,
 108                                                key.objectid - last - 1);
 109                         wake_up(&root->ino_cache_wait);
 110                 }
 111
 112                 last = key.objectid;
 113 next:
 114                 path->slots[0]++;
 115         }
 116
 117         if (last < root->highest_objectid - 1) {
 118                 __btrfs_add_free_space(fs_info, ctl, last + 1,
 119                                        root->highest_objectid - last - 1);
 120         }
 121
 122         spin_lock(&root->ino_cache_lock);
 123         root->ino_cache_state = BTRFS_CACHE_FINISHED;
 124         spin_unlock(&root->ino_cache_lock);
 125
 126         root->ino_cache_progress = (u64)-1;
 127         btrfs_unpin_free_ino(root);
 128 out:
 129         wake_up(&root->ino_cache_wait);
 130         up_read(&fs_info->commit_root_sem);
 131
 132         btrfs_free_path(path);
 133
 134         return ret;
 135 }
 136
 137 static void start_caching(struct btrfs_root *root)
 138 {
 139         struct btrfs_fs_info *fs_info = root->fs_info;
 140         struct btrfs_free_space_ctl *ctl = root->free_ino_ctl;
 141         struct task_struct *tsk;
 142         int ret;
 143         u64 objectid;
 144
 145         if (!btrfs_test_opt(fs_info, INODE_MAP_CACHE))
 146                 return;
 147
 148         spin_lock(&root->ino_cache_lock);
 149         if (root->ino_cache_state != BTRFS_CACHE_NO) {
 150                 spin_unlock(&root->ino_cache_lock);
 151                 return;
 152         }
 153
 154         root->ino_cache_state = BTRFS_CACHE_STARTED;
 155         spin_unlock(&root->ino_cache_lock);
 156
 157         ret = load_free_ino_cache(fs_info, root);
 158         if (ret == 1) {
 159                 spin_lock(&root->ino_cache_lock);
 160                 root->ino_cache_state = BTRFS_CACHE_FINISHED;
 161                 spin_unlock(&root->ino_cache_lock);
 162                 return;
 163         }
 164
 165         /*
 166          * It can be quite time-consuming to fill the cache by searching
 167          * through the extent tree, and this can keep ino allocation path
 168          * waiting. Therefore at start we quickly find out the highest
 169          * inode number and we know we can use inode numbers which fall in
 170          * [highest_ino + 1, BTRFS_LAST_FREE_OBJECTID].
 171          */
 172         ret = btrfs_find_free_objectid(root, &objectid);
 173         if (!ret && objectid <= BTRFS_LAST_FREE_OBJECTID) {
 174                 __btrfs_add_free_space(fs_info, ctl, objectid,
 175                                        BTRFS_LAST_FREE_OBJECTID - objectid + 1);
 176         }
 177
 178         tsk = kthread_run(caching_kthread, root, "btrfs-ino-cache-%llu",
 179                           root->root_key.objectid);
 180         if (IS_ERR(tsk)) {
 181                 btrfs_warn(fs_info, "failed to start inode caching task");
 182                 btrfs_clear_pending_and_info(fs_info, INODE_MAP_CACHE,
 183                                              "disabling inode map caching");
 184         }
 185 }
 186
 187 int btrfs_find_free_ino(struct btrfs_root *root, u64 *objectid)
 188 {
 189         if (!btrfs_test_opt(root->fs_info, INODE_MAP_CACHE))
 190                 return btrfs_find_free_objectid(root, objectid);
 191
 192 again:
 193         *objectid = btrfs_find_ino_for_alloc(root);
 194
 195         if (*objectid != 0)
 196                 return 0;
 197
 198         start_caching(root);
 199
 200         wait_event(root->ino_cache_wait,
 201                    root->ino_cache_state == BTRFS_CACHE_FINISHED ||
 202                    root->free_ino_ctl->free_space > 0);
 203
 204         if (root->ino_cache_state == BTRFS_CACHE_FINISHED &&
 205             root->free_ino_ctl->free_space == 0)
 206                 return -ENOSPC;
 207         else
 208                 goto again;
 209 }
 210
 211 void btrfs_return_ino(struct btrfs_root *root, u64 objectid)
 212 {
 213         struct btrfs_fs_info *fs_info = root->fs_info;
 214         struct btrfs_free_space_ctl *pinned = root->free_ino_pinned;
 215
 216         if (!btrfs_test_opt(fs_info, INODE_MAP_CACHE))
 217                 return;
 218 again:
 219         if (root->ino_cache_state == BTRFS_CACHE_FINISHED) {
 220                 __btrfs_add_free_space(fs_info, pinned, objectid, 1);
 221         } else {
 222                 down_write(&fs_info->commit_root_sem);
 223                 spin_lock(&root->ino_cache_lock);
 224                 if (root->ino_cache_state == BTRFS_CACHE_FINISHED) {
 225                         spin_unlock(&root->ino_cache_lock);
 226                         up_write(&fs_info->commit_root_sem);
 227                         goto again;
 228                 }
 229                 spin_unlock(&root->ino_cache_lock);
 230
 231                 start_caching(root);
 232
 233                 __btrfs_add_free_space(fs_info, pinned, objectid, 1);
 234
 235                 up_write(&fs_info->commit_root_sem);
 236         }
 237 }
 238
 239 /*
 240  * When a transaction is committed, we'll move those inode numbers which are
 241  * smaller than root->ino_cache_progress from pinned tree to free_ino tree, and
 242  * others will just be dropped, because the commit root we were searching has
 243  * changed.
 244  *
 245  * Must be called with root->fs_info->commit_root_sem held
 246  */
 247 void btrfs_unpin_free_ino(struct btrfs_root *root)
 248 {
 249         struct btrfs_free_space_ctl *ctl = root->free_ino_ctl;
 250         struct rb_root *rbroot = &root->free_ino_pinned->free_space_offset;
 251         spinlock_t *rbroot_lock = &root->free_ino_pinned->tree_lock;
 252         struct btrfs_free_space *info;
 253         struct rb_node *n;
 254         u64 count;
 255
 256         if (!btrfs_test_opt(root->fs_info, INODE_MAP_CACHE))
 257                 return;
 258
 259         while (1) {
 260                 bool add_to_ctl = true;
 261
 262                 spin_lock(rbroot_lock);
 263                 n = rb_first(rbroot);
 264                 if (!n) {
 265                         spin_unlock(rbroot_lock);
 266                         break;
 267                 }
 268
 269                 info = rb_entry(n, struct btrfs_free_space, offset_index);
 270                 BUG_ON(info->bitmap); /* Logic error */
 271
 272                 if (info->offset > root->ino_cache_progress)
 273                         add_to_ctl = false;
 274                 else if (info->offset + info->bytes > root->ino_cache_progress)
 275                         count = root->ino_cache_progress - info->offset + 1;
 276                 else
 277                         count = info->bytes;
 278
 279                 rb_erase(&info->offset_index, rbroot);
 280                 spin_unlock(rbroot_lock);
 281                 if (add_to_ctl)
 282                         __btrfs_add_free_space(root->fs_info, ctl,
 283                                                info->offset, count);
 284                 kmem_cache_free(btrfs_free_space_cachep, info);
 285         }
 286 }
 287
 288 #define INIT_THRESHOLD  ((SZ_32K / 2) / sizeof(struct btrfs_free_space))
 289 #define INODES_PER_BITMAP (PAGE_SIZE * 8)
 290
 291 /*
 292  * The goal is to keep the memory used by the free_ino tree won't
 293  * exceed the memory if we use bitmaps only.
 294  */
 295 static void recalculate_thresholds(struct btrfs_free_space_ctl *ctl)
 296 {
 297         struct btrfs_free_space *info;
 298         struct rb_node *n;
 299         int max_ino;
 300         int max_bitmaps;
 301
 302         n = rb_last(&ctl->free_space_offset);
 303         if (!n) {
 304                 ctl->extents_thresh = INIT_THRESHOLD;
 305                 return;
 306         }
 307         info = rb_entry(n, struct btrfs_free_space, offset_index);
 308
 309         /*
 310          * Find the maximum inode number in the filesystem. Note we
 311          * ignore the fact that this can be a bitmap, because we are
 312          * not doing precise calculation.
 313          */
 314         max_ino = info->bytes - 1;
 315
 316         max_bitmaps = ALIGN(max_ino, INODES_PER_BITMAP) / INODES_PER_BITMAP;
 317         if (max_bitmaps <= ctl->total_bitmaps) {
 318                 ctl->extents_thresh = 0;
 319                 return;
 320         }
 321
 322         ctl->extents_thresh = (max_bitmaps - ctl->total_bitmaps) *
 323                                 PAGE_SIZE / sizeof(*info);
 324 }
 325
 326 /*
 327  * We don't fall back to bitmap, if we are below the extents threshold
 328  * or this chunk of inode numbers is a big one.
 329  */
 330 static bool use_bitmap(struct btrfs_free_space_ctl *ctl,
 331                        struct btrfs_free_space *info)
 332 {
 333         if (ctl->free_extents < ctl->extents_thresh ||
 334             info->bytes > INODES_PER_BITMAP / 10)
 335                 return false;
 336
 337         return true;
 338 }
 339
 340 static const struct btrfs_free_space_op free_ino_op = {
 341         .recalc_thresholds      = recalculate_thresholds,
 342         .use_bitmap             = use_bitmap,
 343 };
 344
 345 static void pinned_recalc_thresholds(struct btrfs_free_space_ctl *ctl)
 346 {
 347 }
 348
 349 static bool pinned_use_bitmap(struct btrfs_free_space_ctl *ctl,
 350                               struct btrfs_free_space *info)
 351 {
 352         /*
 353          * We always use extents for two reasons:
 354          *
 355          * - The pinned tree is only used during the process of caching
 356          *   work.
 357          * - Make code simpler. See btrfs_unpin_free_ino().
 358          */
 359         return false;
 360 }
 361
 362 static const struct btrfs_free_space_op pinned_free_ino_op = {
 363         .recalc_thresholds      = pinned_recalc_thresholds,
 364         .use_bitmap             = pinned_use_bitmap,
 365 };
 366
 367 void btrfs_init_free_ino_ctl(struct btrfs_root *root)
 368 {
 369         struct btrfs_free_space_ctl *ctl = root->free_ino_ctl;
 370         struct btrfs_free_space_ctl *pinned = root->free_ino_pinned;
 371
 372         spin_lock_init(&ctl->tree_lock);
 373         ctl->unit = 1;
 374         ctl->start = 0;
 375         ctl->private = NULL;
 376         ctl->op = &free_ino_op;
 377         INIT_LIST_HEAD(&ctl->trimming_ranges);
 378         mutex_init(&ctl->cache_writeout_mutex);
 379
 380         /*
 381          * Initially we allow to use 16K of ram to cache chunks of
 382          * inode numbers before we resort to bitmaps. This is somewhat
 383          * arbitrary, but it will be adjusted in runtime.
 384          */
 385         ctl->extents_thresh = INIT_THRESHOLD;
 386
 387         spin_lock_init(&pinned->tree_lock);
 388         pinned->unit = 1;
 389         pinned->start = 0;
 390         pinned->private = NULL;
 391         pinned->extents_thresh = 0;
 392         pinned->op = &pinned_free_ino_op;
 393 }
 394
 395 int btrfs_save_ino_cache(struct btrfs_root *root,
 396                          struct btrfs_trans_handle *trans)
 397 {
 398         struct btrfs_fs_info *fs_info = root->fs_info;
 399         struct btrfs_free_space_ctl *ctl = root->free_ino_ctl;
 400         struct btrfs_path *path;
 401         struct inode *inode;
 402         struct btrfs_block_rsv *rsv;
 403         struct extent_changeset *data_reserved = NULL;
 404         u64 num_bytes;
 405         u64 alloc_hint = 0;
 406         int ret;
 407         int prealloc;
 408         bool retry = false;
 409
 410         /* only fs tree and subvol/snap needs ino cache */
 411         if (root->root_key.objectid != BTRFS_FS_TREE_OBJECTID &&
 412             (root->root_key.objectid < BTRFS_FIRST_FREE_OBJECTID ||
 413              root->root_key.objectid > BTRFS_LAST_FREE_OBJECTID))
 414                 return 0;
 415
 416         /* Don't save inode cache if we are deleting this root */
 417         if (btrfs_root_refs(&root->root_item) == 0)
 418                 return 0;
 419
 420         if (!btrfs_test_opt(fs_info, INODE_MAP_CACHE))
 421                 return 0;
 422
 423         path = btrfs_alloc_path();
 424         if (!path)
 425                 return -ENOMEM;
 426
 427         rsv = trans->block_rsv;
 428         trans->block_rsv = &fs_info->trans_block_rsv;
 429
 430         num_bytes = trans->bytes_reserved;
 431         /*
 432          * 1 item for inode item insertion if need
 433          * 4 items for inode item update (in the worst case)
 434          * 1 items for slack space if we need do truncation
 435          * 1 item for free space object
 436          * 3 items for pre-allocation
 437          */
 438         trans->bytes_reserved = btrfs_calc_trans_metadata_size(fs_info, 10);
 439         ret = btrfs_block_rsv_add(root, trans->block_rsv,
 440                                   trans->bytes_reserved,
 441                                   BTRFS_RESERVE_NO_FLUSH);
 442         if (ret)
 443                 goto out;
 444         trace_btrfs_space_reservation(fs_info, "ino_cache", trans->transid,
 445                                       trans->bytes_reserved, 1);
 446 again:
 447         inode = lookup_free_ino_inode(root, path);
 448         if (IS_ERR(inode) && (PTR_ERR(inode) != -ENOENT || retry)) {
 449                 ret = PTR_ERR(inode);
 450                 goto out_release;
 451         }
 452
 453         if (IS_ERR(inode)) {
 454                 BUG_ON(retry); /* Logic error */
 455                 retry = true;
 456
 457                 ret = create_free_ino_inode(root, trans, path);
 458                 if (ret)
 459                         goto out_release;
 460                 goto again;
 461         }
 462
 463         BTRFS_I(inode)->generation = 0;
 464         ret = btrfs_update_inode(trans, root, inode);
 465         if (ret) {
 466                 btrfs_abort_transaction(trans, ret);
 467                 goto out_put;
 468         }
 469
 470         if (i_size_read(inode) > 0) {
 471                 ret = btrfs_truncate_free_space_cache(trans, NULL, inode);
 472                 if (ret) {
 473                         if (ret != -ENOSPC)
 474                                 btrfs_abort_transaction(trans, ret);
 475                         goto out_put;
 476                 }
 477         }
 478
 479         spin_lock(&root->ino_cache_lock);
 480         if (root->ino_cache_state != BTRFS_CACHE_FINISHED) {
 481                 ret = -1;
 482                 spin_unlock(&root->ino_cache_lock);
 483                 goto out_put;
 484         }
 485         spin_unlock(&root->ino_cache_lock);
 486
 487         spin_lock(&ctl->tree_lock);
 488         prealloc = sizeof(struct btrfs_free_space) * ctl->free_extents;
 489         prealloc = ALIGN(prealloc, PAGE_SIZE);
 490         prealloc += ctl->total_bitmaps * PAGE_SIZE;
 491         spin_unlock(&ctl->tree_lock);
 492
 493         /* Just to make sure we have enough space */
 494         prealloc += 8 * PAGE_SIZE;
 495
 496         ret = btrfs_delalloc_reserve_space(inode, &data_reserved, 0, prealloc);
 497         if (ret)
 498                 goto out_put;
 499
 500         ret = btrfs_prealloc_file_range_trans(inode, trans, 0, 0, prealloc,
 501                                               prealloc, prealloc, &alloc_hint);
 502         if (ret) {
 503                 btrfs_delalloc_release_extents(BTRFS_I(inode), prealloc);
 504                 goto out_put;
 505         }
 506
 507         ret = btrfs_write_out_ino_cache(root, trans, path, inode);
 508         btrfs_delalloc_release_extents(BTRFS_I(inode), prealloc);
 509 out_put:
 510         iput(inode);
 511 out_release:
 512         trace_btrfs_space_reservation(fs_info, "ino_cache", trans->transid,
 513                                       trans->bytes_reserved, 0);
 514         btrfs_block_rsv_release(fs_info, trans->block_rsv,
 515                                 trans->bytes_reserved);
 516 out:
 517         trans->block_rsv = rsv;
 518         trans->bytes_reserved = num_bytes;
 519
 520         btrfs_free_path(path);
 521         extent_changeset_free(data_reserved);
 522         return ret;
 523 }
 524
 525 int btrfs_find_highest_objectid(struct btrfs_root *root, u64 *objectid)
 526 {
 527         struct btrfs_path *path;
 528         int ret;
 529         struct extent_buffer *l;
 530         struct btrfs_key search_key;
 531         struct btrfs_key found_key;
 532         int slot;
 533
 534         path = btrfs_alloc_path();
 535         if (!path)
 536                 return -ENOMEM;
 537
 538         search_key.objectid = BTRFS_LAST_FREE_OBJECTID;
 539         search_key.type = -1;
 540         search_key.offset = (u64)-1;
 541         ret = btrfs_search_slot(NULL, root, &search_key, path, 0, 0);
 542         if (ret < 0)
 543                 goto error;
 544         BUG_ON(ret == 0); /* Corruption */
 545         if (path->slots[0] > 0) {
 546                 slot = path->slots[0] - 1;
 547                 l = path->nodes[0];
 548                 btrfs_item_key_to_cpu(l, &found_key, slot);
 549                 *objectid = max_t(u64, found_key.objectid,
 550                                   BTRFS_FIRST_FREE_OBJECTID - 1);
 551         } else {
 552                 *objectid = BTRFS_FIRST_FREE_OBJECTID - 1;
 553         }
 554         ret = 0;
 555 error:
 556         btrfs_free_path(path);
 557         return ret;
 558 }
 559
 560 int btrfs_find_free_objectid(struct btrfs_root *root, u64 *objectid)
 561 {
 562         int ret;
 563         mutex_lock(&root->objectid_mutex);
 564
 565         if (unlikely(root->highest_objectid >= BTRFS_LAST_FREE_OBJECTID)) {
 566                 btrfs_warn(root->fs_info,
 567                            "the objectid of root %llu reaches its highest value",
 568                            root->root_key.objectid);
 569                 ret = -ENOSPC;
 570                 goto out;
 571         }
 572
 573         *objectid = ++root->highest_objectid;
 574         ret = 0;
 575 out:
 576         mutex_unlock(&root->objectid_mutex);
 577         return ret;
 578 }