fs/btrfs/free-space-cache.c

   1 /*
   2  * Copyright (C) 2008 Red Hat.  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/pagemap.h>
  20 #include <linux/sched.h>
  21 #include <linux/sched/signal.h>
  22 #include <linux/slab.h>
  23 #include <linux/math64.h>
  24 #include <linux/ratelimit.h>
  25 #include "ctree.h"
  26 #include "free-space-cache.h"
  27 #include "transaction.h"
  28 #include "disk-io.h"
  29 #include "extent_io.h"
  30 #include "inode-map.h"
  31 #include "volumes.h"
  32
  33 #define BITS_PER_BITMAP         (PAGE_SIZE * 8UL)
  34 #define MAX_CACHE_BYTES_PER_GIG SZ_32K
  35
  36 struct btrfs_trim_range {
  37         u64 start;
  38         u64 bytes;
  39         struct list_head list;
  40 };
  41
  42 static int link_free_space(struct btrfs_free_space_ctl *ctl,
  43                            struct btrfs_free_space *info);
  44 static void unlink_free_space(struct btrfs_free_space_ctl *ctl,
  45                               struct btrfs_free_space *info);
  46 static int btrfs_wait_cache_io_root(struct btrfs_root *root,
  47                              struct btrfs_trans_handle *trans,
  48                              struct btrfs_io_ctl *io_ctl,
  49                              struct btrfs_path *path);
  50
  51 static struct inode *__lookup_free_space_inode(struct btrfs_root *root,
  52                                                struct btrfs_path *path,
  53                                                u64 offset)
  54 {
  55         struct btrfs_fs_info *fs_info = root->fs_info;
  56         struct btrfs_key key;
  57         struct btrfs_key location;
  58         struct btrfs_disk_key disk_key;
  59         struct btrfs_free_space_header *header;
  60         struct extent_buffer *leaf;
  61         struct inode *inode = NULL;
  62         int ret;
  63
  64         key.objectid = BTRFS_FREE_SPACE_OBJECTID;
  65         key.offset = offset;
  66         key.type = 0;
  67
  68         ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
  69         if (ret < 0)
  70                 return ERR_PTR(ret);
  71         if (ret > 0) {
  72                 btrfs_release_path(path);
  73                 return ERR_PTR(-ENOENT);
  74         }
  75
  76         leaf = path->nodes[0];
  77         header = btrfs_item_ptr(leaf, path->slots[0],
  78                                 struct btrfs_free_space_header);
  79         btrfs_free_space_key(leaf, header, &disk_key);
  80         btrfs_disk_key_to_cpu(&location, &disk_key);
  81         btrfs_release_path(path);
  82
  83         inode = btrfs_iget(fs_info->sb, &location, root, NULL);
  84         if (IS_ERR(inode))
  85                 return inode;
  86         if (is_bad_inode(inode)) {
  87                 iput(inode);
  88                 return ERR_PTR(-ENOENT);
  89         }
  90
  91         mapping_set_gfp_mask(inode->i_mapping,
  92                         mapping_gfp_constraint(inode->i_mapping,
  93                         ~(__GFP_FS | __GFP_HIGHMEM)));
  94
  95         return inode;
  96 }
  97
  98 struct inode *lookup_free_space_inode(struct btrfs_fs_info *fs_info,
  99                                       struct btrfs_block_group_cache
 100                                       *block_group, struct btrfs_path *path)
 101 {
 102         struct inode *inode = NULL;
 103         u32 flags = BTRFS_INODE_NODATASUM | BTRFS_INODE_NODATACOW;
 104
 105         spin_lock(&block_group->lock);
 106         if (block_group->inode)
 107                 inode = igrab(block_group->inode);
 108         spin_unlock(&block_group->lock);
 109         if (inode)
 110                 return inode;
 111
 112         inode = __lookup_free_space_inode(fs_info->tree_root, path,
 113                                           block_group->key.objectid);
 114         if (IS_ERR(inode))
 115                 return inode;
 116
 117         spin_lock(&block_group->lock);
 118         if (!((BTRFS_I(inode)->flags & flags) == flags)) {
 119                 btrfs_info(fs_info, "Old style space inode found, converting.");
 120                 BTRFS_I(inode)->flags |= BTRFS_INODE_NODATASUM |
 121                         BTRFS_INODE_NODATACOW;
 122                 block_group->disk_cache_state = BTRFS_DC_CLEAR;
 123         }
 124
 125         if (!block_group->iref) {
 126                 block_group->inode = igrab(inode);
 127                 block_group->iref = 1;
 128         }
 129         spin_unlock(&block_group->lock);
 130
 131         return inode;
 132 }
 133
 134 static int __create_free_space_inode(struct btrfs_root *root,
 135                                      struct btrfs_trans_handle *trans,
 136                                      struct btrfs_path *path,
 137                                      u64 ino, u64 offset)
 138 {
 139         struct btrfs_key key;
 140         struct btrfs_disk_key disk_key;
 141         struct btrfs_free_space_header *header;
 142         struct btrfs_inode_item *inode_item;
 143         struct extent_buffer *leaf;
 144         u64 flags = BTRFS_INODE_NOCOMPRESS | BTRFS_INODE_PREALLOC;
 145         int ret;
 146
 147         ret = btrfs_insert_empty_inode(trans, root, path, ino);
 148         if (ret)
 149                 return ret;
 150
 151         /* We inline crc's for the free disk space cache */
 152         if (ino != BTRFS_FREE_INO_OBJECTID)
 153                 flags |= BTRFS_INODE_NODATASUM | BTRFS_INODE_NODATACOW;
 154
 155         leaf = path->nodes[0];
 156         inode_item = btrfs_item_ptr(leaf, path->slots[0],
 157                                     struct btrfs_inode_item);
 158         btrfs_item_key(leaf, &disk_key, path->slots[0]);
 159         memzero_extent_buffer(leaf, (unsigned long)inode_item,
 160                              sizeof(*inode_item));
 161         btrfs_set_inode_generation(leaf, inode_item, trans->transid);
 162         btrfs_set_inode_size(leaf, inode_item, 0);
 163         btrfs_set_inode_nbytes(leaf, inode_item, 0);
 164         btrfs_set_inode_uid(leaf, inode_item, 0);
 165         btrfs_set_inode_gid(leaf, inode_item, 0);
 166         btrfs_set_inode_mode(leaf, inode_item, S_IFREG | 0600);
 167         btrfs_set_inode_flags(leaf, inode_item, flags);
 168         btrfs_set_inode_nlink(leaf, inode_item, 1);
 169         btrfs_set_inode_transid(leaf, inode_item, trans->transid);
 170         btrfs_set_inode_block_group(leaf, inode_item, offset);
 171         btrfs_mark_buffer_dirty(leaf);
 172         btrfs_release_path(path);
 173
 174         key.objectid = BTRFS_FREE_SPACE_OBJECTID;
 175         key.offset = offset;
 176         key.type = 0;
 177         ret = btrfs_insert_empty_item(trans, root, path, &key,
 178                                       sizeof(struct btrfs_free_space_header));
 179         if (ret < 0) {
 180                 btrfs_release_path(path);
 181                 return ret;
 182         }
 183
 184         leaf = path->nodes[0];
 185         header = btrfs_item_ptr(leaf, path->slots[0],
 186                                 struct btrfs_free_space_header);
 187         memzero_extent_buffer(leaf, (unsigned long)header, sizeof(*header));
 188         btrfs_set_free_space_key(leaf, header, &disk_key);
 189         btrfs_mark_buffer_dirty(leaf);
 190         btrfs_release_path(path);
 191
 192         return 0;
 193 }
 194
 195 int create_free_space_inode(struct btrfs_fs_info *fs_info,
 196                             struct btrfs_trans_handle *trans,
 197                             struct btrfs_block_group_cache *block_group,
 198                             struct btrfs_path *path)
 199 {
 200         int ret;
 201         u64 ino;
 202
 203         ret = btrfs_find_free_objectid(fs_info->tree_root, &ino);
 204         if (ret < 0)
 205                 return ret;
 206
 207         return __create_free_space_inode(fs_info->tree_root, trans, path, ino,
 208                                          block_group->key.objectid);
 209 }
 210
 211 int btrfs_check_trunc_cache_free_space(struct btrfs_fs_info *fs_info,
 212                                        struct btrfs_block_rsv *rsv)
 213 {
 214         u64 needed_bytes;
 215         int ret;
 216
 217         /* 1 for slack space, 1 for updating the inode */
 218         needed_bytes = btrfs_calc_trunc_metadata_size(fs_info, 1) +
 219                 btrfs_calc_trans_metadata_size(fs_info, 1);
 220
 221         spin_lock(&rsv->lock);
 222         if (rsv->reserved < needed_bytes)
 223                 ret = -ENOSPC;
 224         else
 225                 ret = 0;
 226         spin_unlock(&rsv->lock);
 227         return ret;
 228 }
 229
 230 int btrfs_truncate_free_space_cache(struct btrfs_trans_handle *trans,
 231                                     struct btrfs_block_group_cache *block_group,
 232                                     struct inode *inode)
 233 {
 234         struct btrfs_root *root = BTRFS_I(inode)->root;
 235         int ret = 0;
 236         bool locked = false;
 237
 238         if (block_group) {
 239                 struct btrfs_path *path = btrfs_alloc_path();
 240
 241                 if (!path) {
 242                         ret = -ENOMEM;
 243                         goto fail;
 244                 }
 245                 locked = true;
 246                 mutex_lock(&trans->transaction->cache_write_mutex);
 247                 if (!list_empty(&block_group->io_list)) {
 248                         list_del_init(&block_group->io_list);
 249
 250                         btrfs_wait_cache_io(trans, block_group, path);
 251                         btrfs_put_block_group(block_group);
 252                 }
 253
 254                 /*
 255                  * now that we've truncated the cache away, its no longer
 256                  * setup or written
 257                  */
 258                 spin_lock(&block_group->lock);
 259                 block_group->disk_cache_state = BTRFS_DC_CLEAR;
 260                 spin_unlock(&block_group->lock);
 261                 btrfs_free_path(path);
 262         }
 263
 264         btrfs_i_size_write(BTRFS_I(inode), 0);
 265         truncate_pagecache(inode, 0);
 266
 267         /*
 268          * We don't need an orphan item because truncating the free space cache
 269          * will never be split across transactions.
 270          * We don't need to check for -EAGAIN because we're a free space
 271          * cache inode
 272          */
 273         ret = btrfs_truncate_inode_items(trans, root, inode,
 274                                          0, BTRFS_EXTENT_DATA_KEY);
 275         if (ret)
 276                 goto fail;
 277
 278         ret = btrfs_update_inode(trans, root, inode);
 279
 280 fail:
 281         if (locked)
 282                 mutex_unlock(&trans->transaction->cache_write_mutex);
 283         if (ret)
 284                 btrfs_abort_transaction(trans, ret);
 285
 286         return ret;
 287 }
 288
 289 static void readahead_cache(struct inode *inode)
 290 {
 291         struct file_ra_state *ra;
 292         unsigned long last_index;
 293
 294         ra = kzalloc(sizeof(*ra), GFP_NOFS);
 295         if (!ra)
 296                 return;
 297
 298         file_ra_state_init(ra, inode->i_mapping);
 299         last_index = (i_size_read(inode) - 1) >> PAGE_SHIFT;
 300
 301         page_cache_sync_readahead(inode->i_mapping, ra, NULL, 0, last_index);
 302
 303         kfree(ra);
 304 }
 305
 306 static int io_ctl_init(struct btrfs_io_ctl *io_ctl, struct inode *inode,
 307                        int write)
 308 {
 309         int num_pages;
 310         int check_crcs = 0;
 311
 312         num_pages = DIV_ROUND_UP(i_size_read(inode), PAGE_SIZE);
 313
 314         if (btrfs_ino(BTRFS_I(inode)) != BTRFS_FREE_INO_OBJECTID)
 315                 check_crcs = 1;
 316
 317         /* Make sure we can fit our crcs into the first page */
 318         if (write && check_crcs &&
 319             (num_pages * sizeof(u32)) >= PAGE_SIZE)
 320                 return -ENOSPC;
 321
 322         memset(io_ctl, 0, sizeof(struct btrfs_io_ctl));
 323
 324         io_ctl->pages = kcalloc(num_pages, sizeof(struct page *), GFP_NOFS);
 325         if (!io_ctl->pages)
 326                 return -ENOMEM;
 327
 328         io_ctl->num_pages = num_pages;
 329         io_ctl->fs_info = btrfs_sb(inode->i_sb);
 330         io_ctl->check_crcs = check_crcs;
 331         io_ctl->inode = inode;
 332
 333         return 0;
 334 }
 335
 336 static void io_ctl_free(struct btrfs_io_ctl *io_ctl)
 337 {
 338         kfree(io_ctl->pages);
 339         io_ctl->pages = NULL;
 340 }
 341
 342 static void io_ctl_unmap_page(struct btrfs_io_ctl *io_ctl)
 343 {
 344         if (io_ctl->cur) {
 345                 io_ctl->cur = NULL;
 346                 io_ctl->orig = NULL;
 347         }
 348 }
 349
 350 static void io_ctl_map_page(struct btrfs_io_ctl *io_ctl, int clear)
 351 {
 352         ASSERT(io_ctl->index < io_ctl->num_pages);
 353         io_ctl->page = io_ctl->pages[io_ctl->index++];
 354         io_ctl->cur = page_address(io_ctl->page);
 355         io_ctl->orig = io_ctl->cur;
 356         io_ctl->size = PAGE_SIZE;
 357         if (clear)
 358                 clear_page(io_ctl->cur);
 359 }
 360
 361 static void io_ctl_drop_pages(struct btrfs_io_ctl *io_ctl)
 362 {
 363         int i;
 364
 365         io_ctl_unmap_page(io_ctl);
 366
 367         for (i = 0; i < io_ctl->num_pages; i++) {
 368                 if (io_ctl->pages[i]) {
 369                         ClearPageChecked(io_ctl->pages[i]);
 370                         unlock_page(io_ctl->pages[i]);
 371                         put_page(io_ctl->pages[i]);
 372                 }
 373         }
 374 }
 375
 376 static int io_ctl_prepare_pages(struct btrfs_io_ctl *io_ctl, struct inode *inode,
 377                                 int uptodate)
 378 {
 379         struct page *page;
 380         gfp_t mask = btrfs_alloc_write_mask(inode->i_mapping);
 381         int i;
 382
 383         for (i = 0; i < io_ctl->num_pages; i++) {
 384                 page = find_or_create_page(inode->i_mapping, i, mask);
 385                 if (!page) {
 386                         io_ctl_drop_pages(io_ctl);
 387                         return -ENOMEM;
 388                 }
 389                 io_ctl->pages[i] = page;
 390                 if (uptodate && !PageUptodate(page)) {
 391                         btrfs_readpage(NULL, page);
 392                         lock_page(page);
 393                         if (!PageUptodate(page)) {
 394                                 btrfs_err(BTRFS_I(inode)->root->fs_info,
 395                                            "error reading free space cache");
 396                                 io_ctl_drop_pages(io_ctl);
 397                                 return -EIO;
 398                         }
 399                 }
 400         }
 401
 402         for (i = 0; i < io_ctl->num_pages; i++) {
 403                 clear_page_dirty_for_io(io_ctl->pages[i]);
 404                 set_page_extent_mapped(io_ctl->pages[i]);
 405         }
 406
 407         return 0;
 408 }
 409
 410 static void io_ctl_set_generation(struct btrfs_io_ctl *io_ctl, u64 generation)
 411 {
 412         __le64 *val;
 413
 414         io_ctl_map_page(io_ctl, 1);
 415
 416         /*
 417          * Skip the csum areas.  If we don't check crcs then we just have a
 418          * 64bit chunk at the front of the first page.
 419          */
 420         if (io_ctl->check_crcs) {
 421                 io_ctl->cur += (sizeof(u32) * io_ctl->num_pages);
 422                 io_ctl->size -= sizeof(u64) + (sizeof(u32) * io_ctl->num_pages);
 423         } else {
 424                 io_ctl->cur += sizeof(u64);
 425                 io_ctl->size -= sizeof(u64) * 2;
 426         }
 427
 428         val = io_ctl->cur;
 429         *val = cpu_to_le64(generation);
 430         io_ctl->cur += sizeof(u64);
 431 }
 432
 433 static int io_ctl_check_generation(struct btrfs_io_ctl *io_ctl, u64 generation)
 434 {
 435         __le64 *gen;
 436
 437         /*
 438          * Skip the crc area.  If we don't check crcs then we just have a 64bit
 439          * chunk at the front of the first page.
 440          */
 441         if (io_ctl->check_crcs) {
 442                 io_ctl->cur += sizeof(u32) * io_ctl->num_pages;
 443                 io_ctl->size -= sizeof(u64) +
 444                         (sizeof(u32) * io_ctl->num_pages);
 445         } else {
 446                 io_ctl->cur += sizeof(u64);
 447                 io_ctl->size -= sizeof(u64) * 2;
 448         }
 449
 450         gen = io_ctl->cur;
 451         if (le64_to_cpu(*gen) != generation) {
 452                 btrfs_err_rl(io_ctl->fs_info,
 453                         "space cache generation (%llu) does not match inode (%llu)",
 454                                 *gen, generation);
 455                 io_ctl_unmap_page(io_ctl);
 456                 return -EIO;
 457         }
 458         io_ctl->cur += sizeof(u64);
 459         return 0;
 460 }
 461
 462 static void io_ctl_set_crc(struct btrfs_io_ctl *io_ctl, int index)
 463 {
 464         u32 *tmp;
 465         u32 crc = ~(u32)0;
 466         unsigned offset = 0;
 467
 468         if (!io_ctl->check_crcs) {
 469                 io_ctl_unmap_page(io_ctl);
 470                 return;
 471         }
 472
 473         if (index == 0)
 474                 offset = sizeof(u32) * io_ctl->num_pages;
 475
 476         crc = btrfs_csum_data(io_ctl->orig + offset, crc,
 477                               PAGE_SIZE - offset);
 478         btrfs_csum_final(crc, (u8 *)&crc);
 479         io_ctl_unmap_page(io_ctl);
 480         tmp = page_address(io_ctl->pages[0]);
 481         tmp += index;
 482         *tmp = crc;
 483 }
 484
 485 static int io_ctl_check_crc(struct btrfs_io_ctl *io_ctl, int index)
 486 {
 487         u32 *tmp, val;
 488         u32 crc = ~(u32)0;
 489         unsigned offset = 0;
 490
 491         if (!io_ctl->check_crcs) {
 492                 io_ctl_map_page(io_ctl, 0);
 493                 return 0;
 494         }
 495
 496         if (index == 0)
 497                 offset = sizeof(u32) * io_ctl->num_pages;
 498
 499         tmp = page_address(io_ctl->pages[0]);
 500         tmp += index;
 501         val = *tmp;
 502
 503         io_ctl_map_page(io_ctl, 0);
 504         crc = btrfs_csum_data(io_ctl->orig + offset, crc,
 505                               PAGE_SIZE - offset);
 506         btrfs_csum_final(crc, (u8 *)&crc);
 507         if (val != crc) {
 508                 btrfs_err_rl(io_ctl->fs_info,
 509                         "csum mismatch on free space cache");
 510                 io_ctl_unmap_page(io_ctl);
 511                 return -EIO;
 512         }
 513
 514         return 0;
 515 }
 516
 517 static int io_ctl_add_entry(struct btrfs_io_ctl *io_ctl, u64 offset, u64 bytes,
 518                             void *bitmap)
 519 {
 520         struct btrfs_free_space_entry *entry;
 521
 522         if (!io_ctl->cur)
 523                 return -ENOSPC;
 524
 525         entry = io_ctl->cur;
 526         entry->offset = cpu_to_le64(offset);
 527         entry->bytes = cpu_to_le64(bytes);
 528         entry->type = (bitmap) ? BTRFS_FREE_SPACE_BITMAP :
 529                 BTRFS_FREE_SPACE_EXTENT;
 530         io_ctl->cur += sizeof(struct btrfs_free_space_entry);
 531         io_ctl->size -= sizeof(struct btrfs_free_space_entry);
 532
 533         if (io_ctl->size >= sizeof(struct btrfs_free_space_entry))
 534                 return 0;
 535
 536         io_ctl_set_crc(io_ctl, io_ctl->index - 1);
 537
 538         /* No more pages to map */
 539         if (io_ctl->index >= io_ctl->num_pages)
 540                 return 0;
 541
 542         /* map the next page */
 543         io_ctl_map_page(io_ctl, 1);
 544         return 0;
 545 }
 546
 547 static int io_ctl_add_bitmap(struct btrfs_io_ctl *io_ctl, void *bitmap)
 548 {
 549         if (!io_ctl->cur)
 550                 return -ENOSPC;
 551
 552         /*
 553          * If we aren't at the start of the current page, unmap this one and
 554          * map the next one if there is any left.
 555          */
 556         if (io_ctl->cur != io_ctl->orig) {
 557                 io_ctl_set_crc(io_ctl, io_ctl->index - 1);
 558                 if (io_ctl->index >= io_ctl->num_pages)
 559                         return -ENOSPC;
 560                 io_ctl_map_page(io_ctl, 0);
 561         }
 562
 563         memcpy(io_ctl->cur, bitmap, PAGE_SIZE);
 564         io_ctl_set_crc(io_ctl, io_ctl->index - 1);
 565         if (io_ctl->index < io_ctl->num_pages)
 566                 io_ctl_map_page(io_ctl, 0);
 567         return 0;
 568 }
 569
 570 static void io_ctl_zero_remaining_pages(struct btrfs_io_ctl *io_ctl)
 571 {
 572         /*
 573          * If we're not on the boundary we know we've modified the page and we
 574          * need to crc the page.
 575          */
 576         if (io_ctl->cur != io_ctl->orig)
 577                 io_ctl_set_crc(io_ctl, io_ctl->index - 1);
 578         else
 579                 io_ctl_unmap_page(io_ctl);
 580
 581         while (io_ctl->index < io_ctl->num_pages) {
 582                 io_ctl_map_page(io_ctl, 1);
 583                 io_ctl_set_crc(io_ctl, io_ctl->index - 1);
 584         }
 585 }
 586
 587 static int io_ctl_read_entry(struct btrfs_io_ctl *io_ctl,
 588                             struct btrfs_free_space *entry, u8 *type)
 589 {
 590         struct btrfs_free_space_entry *e;
 591         int ret;
 592
 593         if (!io_ctl->cur) {
 594                 ret = io_ctl_check_crc(io_ctl, io_ctl->index);
 595                 if (ret)
 596                         return ret;
 597         }
 598
 599         e = io_ctl->cur;
 600         entry->offset = le64_to_cpu(e->offset);
 601         entry->bytes = le64_to_cpu(e->bytes);
 602         *type = e->type;
 603         io_ctl->cur += sizeof(struct btrfs_free_space_entry);
 604         io_ctl->size -= sizeof(struct btrfs_free_space_entry);
 605
 606         if (io_ctl->size >= sizeof(struct btrfs_free_space_entry))
 607                 return 0;
 608
 609         io_ctl_unmap_page(io_ctl);
 610
 611         return 0;
 612 }
 613
 614 static int io_ctl_read_bitmap(struct btrfs_io_ctl *io_ctl,
 615                               struct btrfs_free_space *entry)
 616 {
 617         int ret;
 618
 619         ret = io_ctl_check_crc(io_ctl, io_ctl->index);
 620         if (ret)
 621                 return ret;
 622
 623         memcpy(entry->bitmap, io_ctl->cur, PAGE_SIZE);
 624         io_ctl_unmap_page(io_ctl);
 625
 626         return 0;
 627 }
 628
 629 /*
 630  * Since we attach pinned extents after the fact we can have contiguous sections
 631  * of free space that are split up in entries.  This poses a problem with the
 632  * tree logging stuff since it could have allocated across what appears to be 2
 633  * entries since we would have merged the entries when adding the pinned extents
 634  * back to the free space cache.  So run through the space cache that we just
 635  * loaded and merge contiguous entries.  This will make the log replay stuff not
 636  * blow up and it will make for nicer allocator behavior.
 637  */
 638 static void merge_space_tree(struct btrfs_free_space_ctl *ctl)
 639 {
 640         struct btrfs_free_space *e, *prev = NULL;
 641         struct rb_node *n;
 642
 643 again:
 644         spin_lock(&ctl->tree_lock);
 645         for (n = rb_first(&ctl->free_space_offset); n; n = rb_next(n)) {
 646                 e = rb_entry(n, struct btrfs_free_space, offset_index);
 647                 if (!prev)
 648                         goto next;
 649                 if (e->bitmap || prev->bitmap)
 650                         goto next;
 651                 if (prev->offset + prev->bytes == e->offset) {
 652                         unlink_free_space(ctl, prev);
 653                         unlink_free_space(ctl, e);
 654                         prev->bytes += e->bytes;
 655                         kmem_cache_free(btrfs_free_space_cachep, e);
 656                         link_free_space(ctl, prev);
 657                         prev = NULL;
 658                         spin_unlock(&ctl->tree_lock);
 659                         goto again;
 660                 }
 661 next:
 662                 prev = e;
 663         }
 664         spin_unlock(&ctl->tree_lock);
 665 }
 666
 667 static int __load_free_space_cache(struct btrfs_root *root, struct inode *inode,
 668                                    struct btrfs_free_space_ctl *ctl,
 669                                    struct btrfs_path *path, u64 offset)
 670 {
 671         struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
 672         struct btrfs_free_space_header *header;
 673         struct extent_buffer *leaf;
 674         struct btrfs_io_ctl io_ctl;
 675         struct btrfs_key key;
 676         struct btrfs_free_space *e, *n;
 677         LIST_HEAD(bitmaps);
 678         u64 num_entries;
 679         u64 num_bitmaps;
 680         u64 generation;
 681         u8 type;
 682         int ret = 0;
 683
 684         /* Nothing in the space cache, goodbye */
 685         if (!i_size_read(inode))
 686                 return 0;
 687
 688         key.objectid = BTRFS_FREE_SPACE_OBJECTID;
 689         key.offset = offset;
 690         key.type = 0;
 691
 692         ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
 693         if (ret < 0)
 694                 return 0;
 695         else if (ret > 0) {
 696                 btrfs_release_path(path);
 697                 return 0;
 698         }
 699
 700         ret = -1;
 701
 702         leaf = path->nodes[0];
 703         header = btrfs_item_ptr(leaf, path->slots[0],
 704                                 struct btrfs_free_space_header);
 705         num_entries = btrfs_free_space_entries(leaf, header);
 706         num_bitmaps = btrfs_free_space_bitmaps(leaf, header);
 707         generation = btrfs_free_space_generation(leaf, header);
 708         btrfs_release_path(path);
 709
 710         if (!BTRFS_I(inode)->generation) {
 711                 btrfs_info(fs_info,
 712                            "The free space cache file (%llu) is invalid. skip it\n",
 713                            offset);
 714                 return 0;
 715         }
 716
 717         if (BTRFS_I(inode)->generation != generation) {
 718                 btrfs_err(fs_info,
 719                           "free space inode generation (%llu) did not match free space cache generation (%llu)",
 720                           BTRFS_I(inode)->generation, generation);
 721                 return 0;
 722         }
 723
 724         if (!num_entries)
 725                 return 0;
 726
 727         ret = io_ctl_init(&io_ctl, inode, 0);
 728         if (ret)
 729                 return ret;
 730
 731         readahead_cache(inode);
 732
 733         ret = io_ctl_prepare_pages(&io_ctl, inode, 1);
 734         if (ret)
 735                 goto out;
 736
 737         ret = io_ctl_check_crc(&io_ctl, 0);
 738         if (ret)
 739                 goto free_cache;
 740
 741         ret = io_ctl_check_generation(&io_ctl, generation);
 742         if (ret)
 743                 goto free_cache;
 744
 745         while (num_entries) {
 746                 e = kmem_cache_zalloc(btrfs_free_space_cachep,
 747                                       GFP_NOFS);
 748                 if (!e)
 749                         goto free_cache;
 750
 751                 ret = io_ctl_read_entry(&io_ctl, e, &type);
 752                 if (ret) {
 753                         kmem_cache_free(btrfs_free_space_cachep, e);
 754                         goto free_cache;
 755                 }
 756
 757                 if (!e->bytes) {
 758                         kmem_cache_free(btrfs_free_space_cachep, e);
 759                         goto free_cache;
 760                 }
 761
 762                 if (type == BTRFS_FREE_SPACE_EXTENT) {
 763                         spin_lock(&ctl->tree_lock);
 764                         ret = link_free_space(ctl, e);
 765                         spin_unlock(&ctl->tree_lock);
 766                         if (ret) {
 767                                 btrfs_err(fs_info,
 768                                         "Duplicate entries in free space cache, dumping");
 769                                 kmem_cache_free(btrfs_free_space_cachep, e);
 770                                 goto free_cache;
 771                         }
 772                 } else {
 773                         ASSERT(num_bitmaps);
 774                         num_bitmaps--;
 775                         e->bitmap = kzalloc(PAGE_SIZE, GFP_NOFS);
 776                         if (!e->bitmap) {
 777                                 kmem_cache_free(
 778                                         btrfs_free_space_cachep, e);
 779                                 goto free_cache;
 780                         }
 781                         spin_lock(&ctl->tree_lock);
 782                         ret = link_free_space(ctl, e);
 783                         ctl->total_bitmaps++;
 784                         ctl->op->recalc_thresholds(ctl);
 785                         spin_unlock(&ctl->tree_lock);
 786                         if (ret) {
 787                                 btrfs_err(fs_info,
 788                                         "Duplicate entries in free space cache, dumping");
 789                                 kmem_cache_free(btrfs_free_space_cachep, e);
 790                                 goto free_cache;
 791                         }
 792                         list_add_tail(&e->list, &bitmaps);
 793                 }
 794
 795                 num_entries--;
 796         }
 797
 798         io_ctl_unmap_page(&io_ctl);
 799
 800         /*
 801          * We add the bitmaps at the end of the entries in order that
 802          * the bitmap entries are added to the cache.
 803          */
 804         list_for_each_entry_safe(e, n, &bitmaps, list) {
 805                 list_del_init(&e->list);
 806                 ret = io_ctl_read_bitmap(&io_ctl, e);
 807                 if (ret)
 808                         goto free_cache;
 809         }
 810
 811         io_ctl_drop_pages(&io_ctl);
 812         merge_space_tree(ctl);
 813         ret = 1;
 814 out:
 815         io_ctl_free(&io_ctl);
 816         return ret;
 817 free_cache:
 818         io_ctl_drop_pages(&io_ctl);
 819         __btrfs_remove_free_space_cache(ctl);
 820         goto out;
 821 }
 822
 823 int load_free_space_cache(struct btrfs_fs_info *fs_info,
 824                           struct btrfs_block_group_cache *block_group)
 825 {
 826         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
 827         struct inode *inode;
 828         struct btrfs_path *path;
 829         int ret = 0;
 830         bool matched;
 831         u64 used = btrfs_block_group_used(&block_group->item);
 832
 833         /*
 834          * If this block group has been marked to be cleared for one reason or
 835          * another then we can't trust the on disk cache, so just return.
 836          */
 837         spin_lock(&block_group->lock);
 838         if (block_group->disk_cache_state != BTRFS_DC_WRITTEN) {
 839                 spin_unlock(&block_group->lock);
 840                 return 0;
 841         }
 842         spin_unlock(&block_group->lock);
 843
 844         path = btrfs_alloc_path();
 845         if (!path)
 846                 return 0;
 847         path->search_commit_root = 1;
 848         path->skip_locking = 1;
 849
 850         inode = lookup_free_space_inode(fs_info, block_group, path);
 851         if (IS_ERR(inode)) {
 852                 btrfs_free_path(path);
 853                 return 0;
 854         }
 855
 856         /* We may have converted the inode and made the cache invalid. */
 857         spin_lock(&block_group->lock);
 858         if (block_group->disk_cache_state != BTRFS_DC_WRITTEN) {
 859                 spin_unlock(&block_group->lock);
 860                 btrfs_free_path(path);
 861                 goto out;
 862         }
 863         spin_unlock(&block_group->lock);
 864
 865         ret = __load_free_space_cache(fs_info->tree_root, inode, ctl,
 866                                       path, block_group->key.objectid);
 867         btrfs_free_path(path);
 868         if (ret <= 0)
 869                 goto out;
 870
 871         spin_lock(&ctl->tree_lock);
 872         matched = (ctl->free_space == (block_group->key.offset - used -
 873                                        block_group->bytes_super));
 874         spin_unlock(&ctl->tree_lock);
 875
 876         if (!matched) {
 877                 __btrfs_remove_free_space_cache(ctl);
 878                 btrfs_warn(fs_info,
 879                            "block group %llu has wrong amount of free space",
 880                            block_group->key.objectid);
 881                 ret = -1;
 882         }
 883 out:
 884         if (ret < 0) {
 885                 /* This cache is bogus, make sure it gets cleared */
 886                 spin_lock(&block_group->lock);
 887                 block_group->disk_cache_state = BTRFS_DC_CLEAR;
 888                 spin_unlock(&block_group->lock);
 889                 ret = 0;
 890
 891                 btrfs_warn(fs_info,
 892                            "failed to load free space cache for block group %llu, rebuilding it now",
 893                            block_group->key.objectid);
 894         }
 895
 896         iput(inode);
 897         return ret;
 898 }
 899
 900 static noinline_for_stack
 901 int write_cache_extent_entries(struct btrfs_io_ctl *io_ctl,
 902                               struct btrfs_free_space_ctl *ctl,
 903                               struct btrfs_block_group_cache *block_group,
 904                               int *entries, int *bitmaps,
 905                               struct list_head *bitmap_list)
 906 {
 907         int ret;
 908         struct btrfs_free_cluster *cluster = NULL;
 909         struct btrfs_free_cluster *cluster_locked = NULL;
 910         struct rb_node *node = rb_first(&ctl->free_space_offset);
 911         struct btrfs_trim_range *trim_entry;
 912
 913         /* Get the cluster for this block_group if it exists */
 914         if (block_group && !list_empty(&block_group->cluster_list)) {
 915                 cluster = list_entry(block_group->cluster_list.next,
 916                                      struct btrfs_free_cluster,
 917                                      block_group_list);
 918         }
 919
 920         if (!node && cluster) {
 921                 cluster_locked = cluster;
 922                 spin_lock(&cluster_locked->lock);
 923                 node = rb_first(&cluster->root);
 924                 cluster = NULL;
 925         }
 926
 927         /* Write out the extent entries */
 928         while (node) {
 929                 struct btrfs_free_space *e;
 930
 931                 e = rb_entry(node, struct btrfs_free_space, offset_index);
 932                 *entries += 1;
 933
 934                 ret = io_ctl_add_entry(io_ctl, e->offset, e->bytes,
 935                                        e->bitmap);
 936                 if (ret)
 937                         goto fail;
 938
 939                 if (e->bitmap) {
 940                         list_add_tail(&e->list, bitmap_list);
 941                         *bitmaps += 1;
 942                 }
 943                 node = rb_next(node);
 944                 if (!node && cluster) {
 945                         node = rb_first(&cluster->root);
 946                         cluster_locked = cluster;
 947                         spin_lock(&cluster_locked->lock);
 948                         cluster = NULL;
 949                 }
 950         }
 951         if (cluster_locked) {
 952                 spin_unlock(&cluster_locked->lock);
 953                 cluster_locked = NULL;
 954         }
 955
 956         /*
 957          * Make sure we don't miss any range that was removed from our rbtree
 958          * because trimming is running. Otherwise after a umount+mount (or crash
 959          * after committing the transaction) we would leak free space and get
 960          * an inconsistent free space cache report from fsck.
 961          */
 962         list_for_each_entry(trim_entry, &ctl->trimming_ranges, list) {
 963                 ret = io_ctl_add_entry(io_ctl, trim_entry->start,
 964                                        trim_entry->bytes, NULL);
 965                 if (ret)
 966                         goto fail;
 967                 *entries += 1;
 968         }
 969
 970         return 0;
 971 fail:
 972         if (cluster_locked)
 973                 spin_unlock(&cluster_locked->lock);
 974         return -ENOSPC;
 975 }
 976
 977 static noinline_for_stack int
 978 update_cache_item(struct btrfs_trans_handle *trans,
 979                   struct btrfs_root *root,
 980                   struct inode *inode,
 981                   struct btrfs_path *path, u64 offset,
 982                   int entries, int bitmaps)
 983 {
 984         struct btrfs_key key;
 985         struct btrfs_free_space_header *header;
 986         struct extent_buffer *leaf;
 987         int ret;
 988
 989         key.objectid = BTRFS_FREE_SPACE_OBJECTID;
 990         key.offset = offset;
 991         key.type = 0;
 992
 993         ret = btrfs_search_slot(trans, root, &key, path, 0, 1);
 994         if (ret < 0) {
 995                 clear_extent_bit(&BTRFS_I(inode)->io_tree, 0, inode->i_size - 1,
 996                                  EXTENT_DIRTY | EXTENT_DELALLOC, 0, 0, NULL,
 997                                  GFP_NOFS);
 998                 goto fail;
 999         }
1000         leaf = path->nodes[0];
1001         if (ret > 0) {
1002                 struct btrfs_key found_key;
1003                 ASSERT(path->slots[0]);
1004                 path->slots[0]--;
1005                 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
1006                 if (found_key.objectid != BTRFS_FREE_SPACE_OBJECTID ||
1007                     found_key.offset != offset) {
1008                         clear_extent_bit(&BTRFS_I(inode)->io_tree, 0,
1009                                          inode->i_size - 1,
1010                                          EXTENT_DIRTY | EXTENT_DELALLOC, 0, 0,
1011                                          NULL, GFP_NOFS);
1012                         btrfs_release_path(path);
1013                         goto fail;
1014                 }
1015         }
1016
1017         BTRFS_I(inode)->generation = trans->transid;
1018         header = btrfs_item_ptr(leaf, path->slots[0],
1019                                 struct btrfs_free_space_header);
1020         btrfs_set_free_space_entries(leaf, header, entries);
1021         btrfs_set_free_space_bitmaps(leaf, header, bitmaps);
1022         btrfs_set_free_space_generation(leaf, header, trans->transid);
1023         btrfs_mark_buffer_dirty(leaf);
1024         btrfs_release_path(path);
1025
1026         return 0;
1027
1028 fail:
1029         return -1;
1030 }
1031
1032 static noinline_for_stack int
1033 write_pinned_extent_entries(struct btrfs_fs_info *fs_info,
1034                             struct btrfs_block_group_cache *block_group,
1035                             struct btrfs_io_ctl *io_ctl,
1036                             int *entries)
1037 {
1038         u64 start, extent_start, extent_end, len;
1039         struct extent_io_tree *unpin = NULL;
1040         int ret;
1041
1042         if (!block_group)
1043                 return 0;
1044
1045         /*
1046          * We want to add any pinned extents to our free space cache
1047          * so we don't leak the space
1048          *
1049          * We shouldn't have switched the pinned extents yet so this is the
1050          * right one
1051          */
1052         unpin = fs_info->pinned_extents;
1053
1054         start = block_group->key.objectid;
1055
1056         while (start < block_group->key.objectid + block_group->key.offset) {
1057                 ret = find_first_extent_bit(unpin, start,
1058                                             &extent_start, &extent_end,
1059                                             EXTENT_DIRTY, NULL);
1060                 if (ret)
1061                         return 0;
1062
1063                 /* This pinned extent is out of our range */
1064                 if (extent_start >= block_group->key.objectid +
1065                     block_group->key.offset)
1066                         return 0;
1067
1068                 extent_start = max(extent_start, start);
1069                 extent_end = min(block_group->key.objectid +
1070                                  block_group->key.offset, extent_end + 1);
1071                 len = extent_end - extent_start;
1072
1073                 *entries += 1;
1074                 ret = io_ctl_add_entry(io_ctl, extent_start, len, NULL);
1075                 if (ret)
1076                         return -ENOSPC;
1077
1078                 start = extent_end;
1079         }
1080
1081         return 0;
1082 }
1083
1084 static noinline_for_stack int
1085 write_bitmap_entries(struct btrfs_io_ctl *io_ctl, struct list_head *bitmap_list)
1086 {
1087         struct btrfs_free_space *entry, *next;
1088         int ret;
1089
1090         /* Write out the bitmaps */
1091         list_for_each_entry_safe(entry, next, bitmap_list, list) {
1092                 ret = io_ctl_add_bitmap(io_ctl, entry->bitmap);
1093                 if (ret)
1094                         return -ENOSPC;
1095                 list_del_init(&entry->list);
1096         }
1097
1098         return 0;
1099 }
1100
1101 static int flush_dirty_cache(struct inode *inode)
1102 {
1103         int ret;
1104
1105         ret = btrfs_wait_ordered_range(inode, 0, (u64)-1);
1106         if (ret)
1107                 clear_extent_bit(&BTRFS_I(inode)->io_tree, 0, inode->i_size - 1,
1108                                  EXTENT_DIRTY | EXTENT_DELALLOC, 0, 0, NULL,
1109                                  GFP_NOFS);
1110
1111         return ret;
1112 }
1113
1114 static void noinline_for_stack
1115 cleanup_bitmap_list(struct list_head *bitmap_list)
1116 {
1117         struct btrfs_free_space *entry, *next;
1118
1119         list_for_each_entry_safe(entry, next, bitmap_list, list)
1120                 list_del_init(&entry->list);
1121 }
1122
1123 static void noinline_for_stack
1124 cleanup_write_cache_enospc(struct inode *inode,
1125                            struct btrfs_io_ctl *io_ctl,
1126                            struct extent_state **cached_state)
1127 {
1128         io_ctl_drop_pages(io_ctl);
1129         unlock_extent_cached(&BTRFS_I(inode)->io_tree, 0,
1130                              i_size_read(inode) - 1, cached_state,
1131                              GFP_NOFS);
1132 }
1133
1134 static int __btrfs_wait_cache_io(struct btrfs_root *root,
1135                                  struct btrfs_trans_handle *trans,
1136                                  struct btrfs_block_group_cache *block_group,
1137                                  struct btrfs_io_ctl *io_ctl,
1138                                  struct btrfs_path *path, u64 offset)
1139 {
1140         int ret;
1141         struct inode *inode = io_ctl->inode;
1142         struct btrfs_fs_info *fs_info;
1143
1144         if (!inode)
1145                 return 0;
1146
1147         fs_info = btrfs_sb(inode->i_sb);
1148
1149         /* Flush the dirty pages in the cache file. */
1150         ret = flush_dirty_cache(inode);
1151         if (ret)
1152                 goto out;
1153
1154         /* Update the cache item to tell everyone this cache file is valid. */
1155         ret = update_cache_item(trans, root, inode, path, offset,
1156                                 io_ctl->entries, io_ctl->bitmaps);
1157 out:
1158         io_ctl_free(io_ctl);
1159         if (ret) {
1160                 invalidate_inode_pages2(inode->i_mapping);
1161                 BTRFS_I(inode)->generation = 0;
1162                 if (block_group) {
1163 #ifdef DEBUG
1164                         btrfs_err(fs_info,
1165                                   "failed to write free space cache for block group %llu",
1166                                   block_group->key.objectid);
1167 #endif
1168                 }
1169         }
1170         btrfs_update_inode(trans, root, inode);
1171
1172         if (block_group) {
1173                 /* the dirty list is protected by the dirty_bgs_lock */
1174                 spin_lock(&trans->transaction->dirty_bgs_lock);
1175
1176                 /* the disk_cache_state is protected by the block group lock */
1177                 spin_lock(&block_group->lock);
1178
1179                 /*
1180                  * only mark this as written if we didn't get put back on
1181                  * the dirty list while waiting for IO.   Otherwise our
1182                  * cache state won't be right, and we won't get written again
1183                  */
1184                 if (!ret && list_empty(&block_group->dirty_list))
1185                         block_group->disk_cache_state = BTRFS_DC_WRITTEN;
1186                 else if (ret)
1187                         block_group->disk_cache_state = BTRFS_DC_ERROR;
1188
1189                 spin_unlock(&block_group->lock);
1190                 spin_unlock(&trans->transaction->dirty_bgs_lock);
1191                 io_ctl->inode = NULL;
1192                 iput(inode);
1193         }
1194
1195         return ret;
1196
1197 }
1198
1199 static int btrfs_wait_cache_io_root(struct btrfs_root *root,
1200                                     struct btrfs_trans_handle *trans,
1201                                     struct btrfs_io_ctl *io_ctl,
1202                                     struct btrfs_path *path)
1203 {
1204         return __btrfs_wait_cache_io(root, trans, NULL, io_ctl, path, 0);
1205 }
1206
1207 int btrfs_wait_cache_io(struct btrfs_trans_handle *trans,
1208                         struct btrfs_block_group_cache *block_group,
1209                         struct btrfs_path *path)
1210 {
1211         return __btrfs_wait_cache_io(block_group->fs_info->tree_root, trans,
1212                                      block_group, &block_group->io_ctl,
1213                                      path, block_group->key.objectid);
1214 }
1215
1216 /**
1217  * __btrfs_write_out_cache - write out cached info to an inode
1218  * @root - the root the inode belongs to
1219  * @ctl - the free space cache we are going to write out
1220  * @block_group - the block_group for this cache if it belongs to a block_group
1221  * @trans - the trans handle
1222  *
1223  * This function writes out a free space cache struct to disk for quick recovery
1224  * on mount.  This will return 0 if it was successful in writing the cache out,
1225  * or an errno if it was not.
1226  */
1227 static int __btrfs_write_out_cache(struct btrfs_root *root, struct inode *inode,
1228                                    struct btrfs_free_space_ctl *ctl,
1229                                    struct btrfs_block_group_cache *block_group,
1230                                    struct btrfs_io_ctl *io_ctl,
1231                                    struct btrfs_trans_handle *trans)
1232 {
1233         struct btrfs_fs_info *fs_info = root->fs_info;
1234         struct extent_state *cached_state = NULL;
1235         LIST_HEAD(bitmap_list);
1236         int entries = 0;
1237         int bitmaps = 0;
1238         int ret;
1239         int must_iput = 0;
1240
1241         if (!i_size_read(inode))
1242                 return -EIO;
1243
1244         WARN_ON(io_ctl->pages);
1245         ret = io_ctl_init(io_ctl, inode, 1);
1246         if (ret)
1247                 return ret;
1248
1249         if (block_group && (block_group->flags & BTRFS_BLOCK_GROUP_DATA)) {
1250                 down_write(&block_group->data_rwsem);
1251                 spin_lock(&block_group->lock);
1252                 if (block_group->delalloc_bytes) {
1253                         block_group->disk_cache_state = BTRFS_DC_WRITTEN;
1254                         spin_unlock(&block_group->lock);
1255                         up_write(&block_group->data_rwsem);
1256                         BTRFS_I(inode)->generation = 0;
1257                         ret = 0;
1258                         must_iput = 1;
1259                         goto out;
1260                 }
1261                 spin_unlock(&block_group->lock);
1262         }
1263
1264         /* Lock all pages first so we can lock the extent safely. */
1265         ret = io_ctl_prepare_pages(io_ctl, inode, 0);
1266         if (ret)
1267                 goto out;
1268
1269         lock_extent_bits(&BTRFS_I(inode)->io_tree, 0, i_size_read(inode) - 1,
1270                          &cached_state);
1271
1272         io_ctl_set_generation(io_ctl, trans->transid);
1273
1274         mutex_lock(&ctl->cache_writeout_mutex);
1275         /* Write out the extent entries in the free space cache */
1276         spin_lock(&ctl->tree_lock);
1277         ret = write_cache_extent_entries(io_ctl, ctl,
1278                                          block_group, &entries, &bitmaps,
1279                                          &bitmap_list);
1280         if (ret)
1281                 goto out_nospc_locked;
1282
1283         /*
1284          * Some spaces that are freed in the current transaction are pinned,
1285          * they will be added into free space cache after the transaction is
1286          * committed, we shouldn't lose them.
1287          *
1288          * If this changes while we are working we'll get added back to
1289          * the dirty list and redo it.  No locking needed
1290          */
1291         ret = write_pinned_extent_entries(fs_info, block_group,
1292                                           io_ctl, &entries);
1293         if (ret)
1294                 goto out_nospc_locked;
1295
1296         /*
1297          * At last, we write out all the bitmaps and keep cache_writeout_mutex
1298          * locked while doing it because a concurrent trim can be manipulating
1299          * or freeing the bitmap.
1300          */
1301         ret = write_bitmap_entries(io_ctl, &bitmap_list);
1302         spin_unlock(&ctl->tree_lock);
1303         mutex_unlock(&ctl->cache_writeout_mutex);
1304         if (ret)
1305                 goto out_nospc;
1306
1307         /* Zero out the rest of the pages just to make sure */
1308         io_ctl_zero_remaining_pages(io_ctl);
1309
1310         /* Everything is written out, now we dirty the pages in the file. */
1311         ret = btrfs_dirty_pages(inode, io_ctl->pages, io_ctl->num_pages, 0,
1312                                 i_size_read(inode), &cached_state);
1313         if (ret)
1314                 goto out_nospc;
1315
1316         if (block_group && (block_group->flags & BTRFS_BLOCK_GROUP_DATA))
1317                 up_write(&block_group->data_rwsem);
1318         /*
1319          * Release the pages and unlock the extent, we will flush
1320          * them out later
1321          */
1322         io_ctl_drop_pages(io_ctl);
1323
1324         unlock_extent_cached(&BTRFS_I(inode)->io_tree, 0,
1325                              i_size_read(inode) - 1, &cached_state, GFP_NOFS);
1326
1327         /*
1328          * at this point the pages are under IO and we're happy,
1329          * The caller is responsible for waiting on them and updating the
1330          * the cache and the inode
1331          */
1332         io_ctl->entries = entries;
1333         io_ctl->bitmaps = bitmaps;
1334
1335         ret = btrfs_fdatawrite_range(inode, 0, (u64)-1);
1336         if (ret)
1337                 goto out;
1338
1339         return 0;
1340
1341 out:
1342         io_ctl->inode = NULL;
1343         io_ctl_free(io_ctl);
1344         if (ret) {
1345                 invalidate_inode_pages2(inode->i_mapping);
1346                 BTRFS_I(inode)->generation = 0;
1347         }
1348         btrfs_update_inode(trans, root, inode);
1349         if (must_iput)
1350                 iput(inode);
1351         return ret;
1352
1353 out_nospc_locked:
1354         cleanup_bitmap_list(&bitmap_list);
1355         spin_unlock(&ctl->tree_lock);
1356         mutex_unlock(&ctl->cache_writeout_mutex);
1357
1358 out_nospc:
1359         cleanup_write_cache_enospc(inode, io_ctl, &cached_state);
1360
1361         if (block_group && (block_group->flags & BTRFS_BLOCK_GROUP_DATA))
1362                 up_write(&block_group->data_rwsem);
1363
1364         goto out;
1365 }
1366
1367 int btrfs_write_out_cache(struct btrfs_fs_info *fs_info,
1368                           struct btrfs_trans_handle *trans,
1369                           struct btrfs_block_group_cache *block_group,
1370                           struct btrfs_path *path)
1371 {
1372         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
1373         struct inode *inode;
1374         int ret = 0;
1375
1376         spin_lock(&block_group->lock);
1377         if (block_group->disk_cache_state < BTRFS_DC_SETUP) {
1378                 spin_unlock(&block_group->lock);
1379                 return 0;
1380         }
1381         spin_unlock(&block_group->lock);
1382
1383         inode = lookup_free_space_inode(fs_info, block_group, path);
1384         if (IS_ERR(inode))
1385                 return 0;
1386
1387         ret = __btrfs_write_out_cache(fs_info->tree_root, inode, ctl,
1388                                 block_group, &block_group->io_ctl, trans);
1389         if (ret) {
1390 #ifdef DEBUG
1391                 btrfs_err(fs_info,
1392                           "failed to write free space cache for block group %llu",
1393                           block_group->key.objectid);
1394 #endif
1395                 spin_lock(&block_group->lock);
1396                 block_group->disk_cache_state = BTRFS_DC_ERROR;
1397                 spin_unlock(&block_group->lock);
1398
1399                 block_group->io_ctl.inode = NULL;
1400                 iput(inode);
1401         }
1402
1403         /*
1404          * if ret == 0 the caller is expected to call btrfs_wait_cache_io
1405          * to wait for IO and put the inode
1406          */
1407
1408         return ret;
1409 }
1410
1411 static inline unsigned long offset_to_bit(u64 bitmap_start, u32 unit,
1412                                           u64 offset)
1413 {
1414         ASSERT(offset >= bitmap_start);
1415         offset -= bitmap_start;
1416         return (unsigned long)(div_u64(offset, unit));
1417 }
1418
1419 static inline unsigned long bytes_to_bits(u64 bytes, u32 unit)
1420 {
1421         return (unsigned long)(div_u64(bytes, unit));
1422 }
1423
1424 static inline u64 offset_to_bitmap(struct btrfs_free_space_ctl *ctl,
1425                                    u64 offset)
1426 {
1427         u64 bitmap_start;
1428         u64 bytes_per_bitmap;
1429
1430         bytes_per_bitmap = BITS_PER_BITMAP * ctl->unit;
1431         bitmap_start = offset - ctl->start;
1432         bitmap_start = div64_u64(bitmap_start, bytes_per_bitmap);
1433         bitmap_start *= bytes_per_bitmap;
1434         bitmap_start += ctl->start;
1435
1436         return bitmap_start;
1437 }
1438
1439 static int tree_insert_offset(struct rb_root *root, u64 offset,
1440                               struct rb_node *node, int bitmap)
1441 {
1442         struct rb_node **p = &root->rb_node;
1443         struct rb_node *parent = NULL;
1444         struct btrfs_free_space *info;
1445
1446         while (*p) {
1447                 parent = *p;
1448                 info = rb_entry(parent, struct btrfs_free_space, offset_index);
1449
1450                 if (offset < info->offset) {
1451                         p = &(*p)->rb_left;
1452                 } else if (offset > info->offset) {
1453                         p = &(*p)->rb_right;
1454                 } else {
1455                         /*
1456                          * we could have a bitmap entry and an extent entry
1457                          * share the same offset.  If this is the case, we want
1458                          * the extent entry to always be found first if we do a
1459                          * linear search through the tree, since we want to have
1460                          * the quickest allocation time, and allocating from an
1461                          * extent is faster than allocating from a bitmap.  So
1462                          * if we're inserting a bitmap and we find an entry at
1463                          * this offset, we want to go right, or after this entry
1464                          * logically.  If we are inserting an extent and we've
1465                          * found a bitmap, we want to go left, or before
1466                          * logically.
1467                          */
1468                         if (bitmap) {
1469                                 if (info->bitmap) {
1470                                         WARN_ON_ONCE(1);
1471                                         return -EEXIST;
1472                                 }
1473                                 p = &(*p)->rb_right;
1474                         } else {
1475                                 if (!info->bitmap) {
1476                                         WARN_ON_ONCE(1);
1477                                         return -EEXIST;
1478                                 }
1479                                 p = &(*p)->rb_left;
1480                         }
1481                 }
1482         }
1483
1484         rb_link_node(node, parent, p);
1485         rb_insert_color(node, root);
1486
1487         return 0;
1488 }
1489
1490 /*
1491  * searches the tree for the given offset.
1492  *
1493  * fuzzy - If this is set, then we are trying to make an allocation, and we just
1494  * want a section that has at least bytes size and comes at or after the given
1495  * offset.
1496  */
1497 static struct btrfs_free_space *
1498 tree_search_offset(struct btrfs_free_space_ctl *ctl,
1499                    u64 offset, int bitmap_only, int fuzzy)
1500 {
1501         struct rb_node *n = ctl->free_space_offset.rb_node;
1502         struct btrfs_free_space *entry, *prev = NULL;
1503
1504         /* find entry that is closest to the 'offset' */
1505         while (1) {
1506                 if (!n) {
1507                         entry = NULL;
1508                         break;
1509                 }
1510
1511                 entry = rb_entry(n, struct btrfs_free_space, offset_index);
1512                 prev = entry;
1513
1514                 if (offset < entry->offset)
1515                         n = n->rb_left;
1516                 else if (offset > entry->offset)
1517                         n = n->rb_right;
1518                 else
1519                         break;
1520         }
1521
1522         if (bitmap_only) {
1523                 if (!entry)
1524                         return NULL;
1525                 if (entry->bitmap)
1526                         return entry;
1527
1528                 /*
1529                  * bitmap entry and extent entry may share same offset,
1530                  * in that case, bitmap entry comes after extent entry.
1531                  */
1532                 n = rb_next(n);
1533                 if (!n)
1534                         return NULL;
1535                 entry = rb_entry(n, struct btrfs_free_space, offset_index);
1536                 if (entry->offset != offset)
1537                         return NULL;
1538
1539                 WARN_ON(!entry->bitmap);
1540                 return entry;
1541         } else if (entry) {
1542                 if (entry->bitmap) {
1543                         /*
1544                          * if previous extent entry covers the offset,
1545                          * we should return it instead of the bitmap entry
1546                          */
1547                         n = rb_prev(&entry->offset_index);
1548                         if (n) {
1549                                 prev = rb_entry(n, struct btrfs_free_space,
1550                                                 offset_index);
1551                                 if (!prev->bitmap &&
1552                                     prev->offset + prev->bytes > offset)
1553                                         entry = prev;
1554                         }
1555                 }
1556                 return entry;
1557         }
1558
1559         if (!prev)
1560                 return NULL;
1561
1562         /* find last entry before the 'offset' */
1563         entry = prev;
1564         if (entry->offset > offset) {
1565                 n = rb_prev(&entry->offset_index);
1566                 if (n) {
1567                         entry = rb_entry(n, struct btrfs_free_space,
1568                                         offset_index);
1569                         ASSERT(entry->offset <= offset);
1570                 } else {
1571                         if (fuzzy)
1572                                 return entry;
1573                         else
1574                                 return NULL;
1575                 }
1576         }
1577
1578         if (entry->bitmap) {
1579                 n = rb_prev(&entry->offset_index);
1580                 if (n) {
1581                         prev = rb_entry(n, struct btrfs_free_space,
1582                                         offset_index);
1583                         if (!prev->bitmap &&
1584                             prev->offset + prev->bytes > offset)
1585                                 return prev;
1586                 }
1587                 if (entry->offset + BITS_PER_BITMAP * ctl->unit > offset)
1588                         return entry;
1589         } else if (entry->offset + entry->bytes > offset)
1590                 return entry;
1591
1592         if (!fuzzy)
1593                 return NULL;
1594
1595         while (1) {
1596                 if (entry->bitmap) {
1597                         if (entry->offset + BITS_PER_BITMAP *
1598                             ctl->unit > offset)
1599                                 break;
1600                 } else {
1601                         if (entry->offset + entry->bytes > offset)
1602                                 break;
1603                 }
1604
1605                 n = rb_next(&entry->offset_index);
1606                 if (!n)
1607                         return NULL;
1608                 entry = rb_entry(n, struct btrfs_free_space, offset_index);
1609         }
1610         return entry;
1611 }
1612
1613 static inline void
1614 __unlink_free_space(struct btrfs_free_space_ctl *ctl,
1615                     struct btrfs_free_space *info)
1616 {
1617         rb_erase(&info->offset_index, &ctl->free_space_offset);
1618         ctl->free_extents--;
1619 }
1620
1621 static void unlink_free_space(struct btrfs_free_space_ctl *ctl,
1622                               struct btrfs_free_space *info)
1623 {
1624         __unlink_free_space(ctl, info);
1625         ctl->free_space -= info->bytes;
1626 }
1627
1628 static int link_free_space(struct btrfs_free_space_ctl *ctl,
1629                            struct btrfs_free_space *info)
1630 {
1631         int ret = 0;
1632
1633         ASSERT(info->bytes || info->bitmap);
1634         ret = tree_insert_offset(&ctl->free_space_offset, info->offset,
1635                                  &info->offset_index, (info->bitmap != NULL));
1636         if (ret)
1637                 return ret;
1638
1639         ctl->free_space += info->bytes;
1640         ctl->free_extents++;
1641         return ret;
1642 }
1643
1644 static void recalculate_thresholds(struct btrfs_free_space_ctl *ctl)
1645 {
1646         struct btrfs_block_group_cache *block_group = ctl->private;
1647         u64 max_bytes;
1648         u64 bitmap_bytes;
1649         u64 extent_bytes;
1650         u64 size = block_group->key.offset;
1651         u64 bytes_per_bg = BITS_PER_BITMAP * ctl->unit;
1652         u64 max_bitmaps = div64_u64(size + bytes_per_bg - 1, bytes_per_bg);
1653
1654         max_bitmaps = max_t(u64, max_bitmaps, 1);
1655
1656         ASSERT(ctl->total_bitmaps <= max_bitmaps);
1657
1658         /*
1659          * The goal is to keep the total amount of memory used per 1gb of space
1660          * at or below 32k, so we need to adjust how much memory we allow to be
1661          * used by extent based free space tracking
1662          */
1663         if (size < SZ_1G)
1664                 max_bytes = MAX_CACHE_BYTES_PER_GIG;
1665         else
1666                 max_bytes = MAX_CACHE_BYTES_PER_GIG * div_u64(size, SZ_1G);
1667
1668         /*
1669          * we want to account for 1 more bitmap than what we have so we can make
1670          * sure we don't go over our overall goal of MAX_CACHE_BYTES_PER_GIG as
1671          * we add more bitmaps.
1672          */
1673         bitmap_bytes = (ctl->total_bitmaps + 1) * ctl->unit;
1674
1675         if (bitmap_bytes >= max_bytes) {
1676                 ctl->extents_thresh = 0;
1677                 return;
1678         }
1679
1680         /*
1681          * we want the extent entry threshold to always be at most 1/2 the max
1682          * bytes we can have, or whatever is less than that.
1683          */
1684         extent_bytes = max_bytes - bitmap_bytes;
1685         extent_bytes = min_t(u64, extent_bytes, max_bytes >> 1);
1686
1687         ctl->extents_thresh =
1688                 div_u64(extent_bytes, sizeof(struct btrfs_free_space));
1689 }
1690
1691 static inline void __bitmap_clear_bits(struct btrfs_free_space_ctl *ctl,
1692                                        struct btrfs_free_space *info,
1693                                        u64 offset, u64 bytes)
1694 {
1695         unsigned long start, count;
1696
1697         start = offset_to_bit(info->offset, ctl->unit, offset);
1698         count = bytes_to_bits(bytes, ctl->unit);
1699         ASSERT(start + count <= BITS_PER_BITMAP);
1700
1701         bitmap_clear(info->bitmap, start, count);
1702
1703         info->bytes -= bytes;
1704 }
1705
1706 static void bitmap_clear_bits(struct btrfs_free_space_ctl *ctl,
1707                               struct btrfs_free_space *info, u64 offset,
1708                               u64 bytes)
1709 {
1710         __bitmap_clear_bits(ctl, info, offset, bytes);
1711         ctl->free_space -= bytes;
1712 }
1713
1714 static void bitmap_set_bits(struct btrfs_free_space_ctl *ctl,
1715                             struct btrfs_free_space *info, u64 offset,
1716                             u64 bytes)
1717 {
1718         unsigned long start, count;
1719
1720         start = offset_to_bit(info->offset, ctl->unit, offset);
1721         count = bytes_to_bits(bytes, ctl->unit);
1722         ASSERT(start + count <= BITS_PER_BITMAP);
1723
1724         bitmap_set(info->bitmap, start, count);
1725
1726         info->bytes += bytes;
1727         ctl->free_space += bytes;
1728 }
1729
1730 /*
1731  * If we can not find suitable extent, we will use bytes to record
1732  * the size of the max extent.
1733  */
1734 static int search_bitmap(struct btrfs_free_space_ctl *ctl,
1735                          struct btrfs_free_space *bitmap_info, u64 *offset,
1736                          u64 *bytes, bool for_alloc)
1737 {
1738         unsigned long found_bits = 0;
1739         unsigned long max_bits = 0;
1740         unsigned long bits, i;
1741         unsigned long next_zero;
1742         unsigned long extent_bits;
1743
1744         /*
1745          * Skip searching the bitmap if we don't have a contiguous section that
1746          * is large enough for this allocation.
1747          */
1748         if (for_alloc &&
1749             bitmap_info->max_extent_size &&
1750             bitmap_info->max_extent_size < *bytes) {
1751                 *bytes = bitmap_info->max_extent_size;
1752                 return -1;
1753         }
1754
1755         i = offset_to_bit(bitmap_info->offset, ctl->unit,
1756                           max_t(u64, *offset, bitmap_info->offset));
1757         bits = bytes_to_bits(*bytes, ctl->unit);
1758
1759         for_each_set_bit_from(i, bitmap_info->bitmap, BITS_PER_BITMAP) {
1760                 if (for_alloc && bits == 1) {
1761                         found_bits = 1;
1762                         break;
1763                 }
1764                 next_zero = find_next_zero_bit(bitmap_info->bitmap,
1765                                                BITS_PER_BITMAP, i);
1766                 extent_bits = next_zero - i;
1767                 if (extent_bits >= bits) {
1768                         found_bits = extent_bits;
1769                         break;
1770                 } else if (extent_bits > max_bits) {
1771                         max_bits = extent_bits;
1772                 }
1773                 i = next_zero;
1774         }
1775
1776         if (found_bits) {
1777                 *offset = (u64)(i * ctl->unit) + bitmap_info->offset;
1778                 *bytes = (u64)(found_bits) * ctl->unit;
1779                 return 0;
1780         }
1781
1782         *bytes = (u64)(max_bits) * ctl->unit;
1783         bitmap_info->max_extent_size = *bytes;
1784         return -1;
1785 }
1786
1787 /* Cache the size of the max extent in bytes */
1788 static struct btrfs_free_space *
1789 find_free_space(struct btrfs_free_space_ctl *ctl, u64 *offset, u64 *bytes,
1790                 unsigned long align, u64 *max_extent_size)
1791 {
1792         struct btrfs_free_space *entry;
1793         struct rb_node *node;
1794         u64 tmp;
1795         u64 align_off;
1796         int ret;
1797
1798         if (!ctl->free_space_offset.rb_node)
1799                 goto out;
1800
1801         entry = tree_search_offset(ctl, offset_to_bitmap(ctl, *offset), 0, 1);
1802         if (!entry)
1803                 goto out;
1804
1805         for (node = &entry->offset_index; node; node = rb_next(node)) {
1806                 entry = rb_entry(node, struct btrfs_free_space, offset_index);
1807                 if (entry->bytes < *bytes) {
1808                         if (entry->bytes > *max_extent_size)
1809                                 *max_extent_size = entry->bytes;
1810                         continue;
1811                 }
1812
1813                 /* make sure the space returned is big enough
1814                  * to match our requested alignment
1815                  */
1816                 if (*bytes >= align) {
1817                         tmp = entry->offset - ctl->start + align - 1;
1818                         tmp = div64_u64(tmp, align);
1819                         tmp = tmp * align + ctl->start;
1820                         align_off = tmp - entry->offset;
1821                 } else {
1822                         align_off = 0;
1823                         tmp = entry->offset;
1824                 }
1825
1826                 if (entry->bytes < *bytes + align_off) {
1827                         if (entry->bytes > *max_extent_size)
1828                                 *max_extent_size = entry->bytes;
1829                         continue;
1830                 }
1831
1832                 if (entry->bitmap) {
1833                         u64 size = *bytes;
1834
1835                         ret = search_bitmap(ctl, entry, &tmp, &size, true);
1836                         if (!ret) {
1837                                 *offset = tmp;
1838                                 *bytes = size;
1839                                 return entry;
1840                         } else if (size > *max_extent_size) {
1841                                 *max_extent_size = size;
1842                         }
1843                         continue;
1844                 }
1845
1846                 *offset = tmp;
1847                 *bytes = entry->bytes - align_off;
1848                 return entry;
1849         }
1850 out:
1851         return NULL;
1852 }
1853
1854 static void add_new_bitmap(struct btrfs_free_space_ctl *ctl,
1855                            struct btrfs_free_space *info, u64 offset)
1856 {
1857         info->offset = offset_to_bitmap(ctl, offset);
1858         info->bytes = 0;
1859         INIT_LIST_HEAD(&info->list);
1860         link_free_space(ctl, info);
1861         ctl->total_bitmaps++;
1862
1863         ctl->op->recalc_thresholds(ctl);
1864 }
1865
1866 static void free_bitmap(struct btrfs_free_space_ctl *ctl,
1867                         struct btrfs_free_space *bitmap_info)
1868 {
1869         unlink_free_space(ctl, bitmap_info);
1870         kfree(bitmap_info->bitmap);
1871         kmem_cache_free(btrfs_free_space_cachep, bitmap_info);
1872         ctl->total_bitmaps--;
1873         ctl->op->recalc_thresholds(ctl);
1874 }
1875
1876 static noinline int remove_from_bitmap(struct btrfs_free_space_ctl *ctl,
1877                               struct btrfs_free_space *bitmap_info,
1878                               u64 *offset, u64 *bytes)
1879 {
1880         u64 end;
1881         u64 search_start, search_bytes;
1882         int ret;
1883
1884 again:
1885         end = bitmap_info->offset + (u64)(BITS_PER_BITMAP * ctl->unit) - 1;
1886
1887         /*
1888          * We need to search for bits in this bitmap.  We could only cover some
1889          * of the extent in this bitmap thanks to how we add space, so we need
1890          * to search for as much as it as we can and clear that amount, and then
1891          * go searching for the next bit.
1892          */
1893         search_start = *offset;
1894         search_bytes = ctl->unit;
1895         search_bytes = min(search_bytes, end - search_start + 1);
1896         ret = search_bitmap(ctl, bitmap_info, &search_start, &search_bytes,
1897                             false);
1898         if (ret < 0 || search_start != *offset)
1899                 return -EINVAL;
1900
1901         /* We may have found more bits than what we need */
1902         search_bytes = min(search_bytes, *bytes);
1903
1904         /* Cannot clear past the end of the bitmap */
1905         search_bytes = min(search_bytes, end - search_start + 1);
1906
1907         bitmap_clear_bits(ctl, bitmap_info, search_start, search_bytes);
1908         *offset += search_bytes;
1909         *bytes -= search_bytes;
1910
1911         if (*bytes) {
1912                 struct rb_node *next = rb_next(&bitmap_info->offset_index);
1913                 if (!bitmap_info->bytes)
1914                         free_bitmap(ctl, bitmap_info);
1915
1916                 /*
1917                  * no entry after this bitmap, but we still have bytes to
1918                  * remove, so something has gone wrong.
1919                  */
1920                 if (!next)
1921                         return -EINVAL;
1922
1923                 bitmap_info = rb_entry(next, struct btrfs_free_space,
1924                                        offset_index);
1925
1926                 /*
1927                  * if the next entry isn't a bitmap we need to return to let the
1928                  * extent stuff do its work.
1929                  */
1930                 if (!bitmap_info->bitmap)
1931                         return -EAGAIN;
1932
1933                 /*
1934                  * Ok the next item is a bitmap, but it may not actually hold
1935                  * the information for the rest of this free space stuff, so
1936                  * look for it, and if we don't find it return so we can try
1937                  * everything over again.
1938                  */
1939                 search_start = *offset;
1940                 search_bytes = ctl->unit;
1941                 ret = search_bitmap(ctl, bitmap_info, &search_start,
1942                                     &search_bytes, false);
1943                 if (ret < 0 || search_start != *offset)
1944                         return -EAGAIN;
1945
1946                 goto again;
1947         } else if (!bitmap_info->bytes)
1948                 free_bitmap(ctl, bitmap_info);
1949
1950         return 0;
1951 }
1952
1953 static u64 add_bytes_to_bitmap(struct btrfs_free_space_ctl *ctl,
1954                                struct btrfs_free_space *info, u64 offset,
1955                                u64 bytes)
1956 {
1957         u64 bytes_to_set = 0;
1958         u64 end;
1959
1960         end = info->offset + (u64)(BITS_PER_BITMAP * ctl->unit);
1961
1962         bytes_to_set = min(end - offset, bytes);
1963
1964         bitmap_set_bits(ctl, info, offset, bytes_to_set);
1965
1966         /*
1967          * We set some bytes, we have no idea what the max extent size is
1968          * anymore.
1969          */
1970         info->max_extent_size = 0;
1971
1972         return bytes_to_set;
1973
1974 }
1975
1976 static bool use_bitmap(struct btrfs_free_space_ctl *ctl,
1977                       struct btrfs_free_space *info)
1978 {
1979         struct btrfs_block_group_cache *block_group = ctl->private;
1980         struct btrfs_fs_info *fs_info = block_group->fs_info;
1981         bool forced = false;
1982
1983 #ifdef CONFIG_BTRFS_DEBUG
1984         if (btrfs_should_fragment_free_space(block_group))
1985                 forced = true;
1986 #endif
1987
1988         /*
1989          * If we are below the extents threshold then we can add this as an
1990          * extent, and don't have to deal with the bitmap
1991          */
1992         if (!forced && ctl->free_extents < ctl->extents_thresh) {
1993                 /*
1994                  * If this block group has some small extents we don't want to
1995                  * use up all of our free slots in the cache with them, we want
1996                  * to reserve them to larger extents, however if we have plenty
1997                  * of cache left then go ahead an dadd them, no sense in adding
1998                  * the overhead of a bitmap if we don't have to.
1999                  */
2000                 if (info->bytes <= fs_info->sectorsize * 4) {
2001                         if (ctl->free_extents * 2 <= ctl->extents_thresh)
2002                                 return false;
2003                 } else {
2004                         return false;
2005                 }
2006         }
2007
2008         /*
2009          * The original block groups from mkfs can be really small, like 8
2010          * megabytes, so don't bother with a bitmap for those entries.  However
2011          * some block groups can be smaller than what a bitmap would cover but
2012          * are still large enough that they could overflow the 32k memory limit,
2013          * so allow those block groups to still be allowed to have a bitmap
2014          * entry.
2015          */
2016         if (((BITS_PER_BITMAP * ctl->unit) >> 1) > block_group->key.offset)
2017                 return false;
2018
2019         return true;
2020 }
2021
2022 static const struct btrfs_free_space_op free_space_op = {
2023         .recalc_thresholds      = recalculate_thresholds,
2024         .use_bitmap             = use_bitmap,
2025 };
2026
2027 static int insert_into_bitmap(struct btrfs_free_space_ctl *ctl,
2028                               struct btrfs_free_space *info)
2029 {
2030         struct btrfs_free_space *bitmap_info;
2031         struct btrfs_block_group_cache *block_group = NULL;
2032         int added = 0;
2033         u64 bytes, offset, bytes_added;
2034         int ret;
2035
2036         bytes = info->bytes;
2037         offset = info->offset;
2038
2039         if (!ctl->op->use_bitmap(ctl, info))
2040                 return 0;
2041
2042         if (ctl->op == &free_space_op)
2043                 block_group = ctl->private;
2044 again:
2045         /*
2046          * Since we link bitmaps right into the cluster we need to see if we
2047          * have a cluster here, and if so and it has our bitmap we need to add
2048          * the free space to that bitmap.
2049          */
2050         if (block_group && !list_empty(&block_group->cluster_list)) {
2051                 struct btrfs_free_cluster *cluster;
2052                 struct rb_node *node;
2053                 struct btrfs_free_space *entry;
2054
2055                 cluster = list_entry(block_group->cluster_list.next,
2056                                      struct btrfs_free_cluster,
2057                                      block_group_list);
2058                 spin_lock(&cluster->lock);
2059                 node = rb_first(&cluster->root);
2060                 if (!node) {
2061                         spin_unlock(&cluster->lock);
2062                         goto no_cluster_bitmap;
2063                 }
2064
2065                 entry = rb_entry(node, struct btrfs_free_space, offset_index);
2066                 if (!entry->bitmap) {
2067                         spin_unlock(&cluster->lock);
2068                         goto no_cluster_bitmap;
2069                 }
2070
2071                 if (entry->offset == offset_to_bitmap(ctl, offset)) {
2072                         bytes_added = add_bytes_to_bitmap(ctl, entry,
2073                                                           offset, bytes);
2074                         bytes -= bytes_added;
2075                         offset += bytes_added;
2076                 }
2077                 spin_unlock(&cluster->lock);
2078                 if (!bytes) {
2079                         ret = 1;
2080                         goto out;
2081                 }
2082         }
2083
2084 no_cluster_bitmap:
2085         bitmap_info = tree_search_offset(ctl, offset_to_bitmap(ctl, offset),
2086                                          1, 0);
2087         if (!bitmap_info) {
2088                 ASSERT(added == 0);
2089                 goto new_bitmap;
2090         }
2091
2092         bytes_added = add_bytes_to_bitmap(ctl, bitmap_info, offset, bytes);
2093         bytes -= bytes_added;
2094         offset += bytes_added;
2095         added = 0;
2096
2097         if (!bytes) {
2098                 ret = 1;
2099                 goto out;
2100         } else
2101                 goto again;
2102
2103 new_bitmap:
2104         if (info && info->bitmap) {
2105                 add_new_bitmap(ctl, info, offset);
2106                 added = 1;
2107                 info = NULL;
2108                 goto again;
2109         } else {
2110                 spin_unlock(&ctl->tree_lock);
2111
2112                 /* no pre-allocated info, allocate a new one */
2113                 if (!info) {
2114                         info = kmem_cache_zalloc(btrfs_free_space_cachep,
2115                                                  GFP_NOFS);
2116                         if (!info) {
2117                                 spin_lock(&ctl->tree_lock);
2118                                 ret = -ENOMEM;
2119                                 goto out;
2120                         }
2121                 }
2122
2123                 /* allocate the bitmap */
2124                 info->bitmap = kzalloc(PAGE_SIZE, GFP_NOFS);
2125                 spin_lock(&ctl->tree_lock);
2126                 if (!info->bitmap) {
2127                         ret = -ENOMEM;
2128                         goto out;
2129                 }
2130                 goto again;
2131         }
2132
2133 out:
2134         if (info) {
2135                 if (info->bitmap)
2136                         kfree(info->bitmap);
2137                 kmem_cache_free(btrfs_free_space_cachep, info);
2138         }
2139
2140         return ret;
2141 }
2142
2143 static bool try_merge_free_space(struct btrfs_free_space_ctl *ctl,
2144                           struct btrfs_free_space *info, bool update_stat)
2145 {
2146         struct btrfs_free_space *left_info;
2147         struct btrfs_free_space *right_info;
2148         bool merged = false;
2149         u64 offset = info->offset;
2150         u64 bytes = info->bytes;
2151
2152         /*
2153          * first we want to see if there is free space adjacent to the range we
2154          * are adding, if there is remove that struct and add a new one to
2155          * cover the entire range
2156          */
2157         right_info = tree_search_offset(ctl, offset + bytes, 0, 0);
2158         if (right_info && rb_prev(&right_info->offset_index))
2159                 left_info = rb_entry(rb_prev(&right_info->offset_index),
2160                                      struct btrfs_free_space, offset_index);
2161         else
2162                 left_info = tree_search_offset(ctl, offset - 1, 0, 0);
2163
2164         if (right_info && !right_info->bitmap) {
2165                 if (update_stat)
2166                         unlink_free_space(ctl, right_info);
2167                 else
2168                         __unlink_free_space(ctl, right_info);
2169                 info->bytes += right_info->bytes;
2170                 kmem_cache_free(btrfs_free_space_cachep, right_info);
2171                 merged = true;
2172         }
2173
2174         if (left_info && !left_info->bitmap &&
2175             left_info->offset + left_info->bytes == offset) {
2176                 if (update_stat)
2177                         unlink_free_space(ctl, left_info);
2178                 else
2179                         __unlink_free_space(ctl, left_info);
2180                 info->offset = left_info->offset;
2181                 info->bytes += left_info->bytes;
2182                 kmem_cache_free(btrfs_free_space_cachep, left_info);
2183                 merged = true;
2184         }
2185
2186         return merged;
2187 }
2188
2189 static bool steal_from_bitmap_to_end(struct btrfs_free_space_ctl *ctl,
2190                                      struct btrfs_free_space *info,
2191                                      bool update_stat)
2192 {
2193         struct btrfs_free_space *bitmap;
2194         unsigned long i;
2195         unsigned long j;
2196         const u64 end = info->offset + info->bytes;
2197         const u64 bitmap_offset = offset_to_bitmap(ctl, end);
2198         u64 bytes;
2199
2200         bitmap = tree_search_offset(ctl, bitmap_offset, 1, 0);
2201         if (!bitmap)
2202                 return false;
2203
2204         i = offset_to_bit(bitmap->offset, ctl->unit, end);
2205         j = find_next_zero_bit(bitmap->bitmap, BITS_PER_BITMAP, i);
2206         if (j == i)
2207                 return false;
2208         bytes = (j - i) * ctl->unit;
2209         info->bytes += bytes;
2210
2211         if (update_stat)
2212                 bitmap_clear_bits(ctl, bitmap, end, bytes);
2213         else
2214                 __bitmap_clear_bits(ctl, bitmap, end, bytes);
2215
2216         if (!bitmap->bytes)
2217                 free_bitmap(ctl, bitmap);
2218
2219         return true;
2220 }
2221
2222 static bool steal_from_bitmap_to_front(struct btrfs_free_space_ctl *ctl,
2223                                        struct btrfs_free_space *info,
2224                                        bool update_stat)
2225 {
2226         struct btrfs_free_space *bitmap;
2227         u64 bitmap_offset;
2228         unsigned long i;
2229         unsigned long j;
2230         unsigned long prev_j;
2231         u64 bytes;
2232
2233         bitmap_offset = offset_to_bitmap(ctl, info->offset);
2234         /* If we're on a boundary, try the previous logical bitmap. */
2235         if (bitmap_offset == info->offset) {
2236                 if (info->offset == 0)
2237                         return false;
2238                 bitmap_offset = offset_to_bitmap(ctl, info->offset - 1);
2239         }
2240
2241         bitmap = tree_search_offset(ctl, bitmap_offset, 1, 0);
2242         if (!bitmap)
2243                 return false;
2244
2245         i = offset_to_bit(bitmap->offset, ctl->unit, info->offset) - 1;
2246         j = 0;
2247         prev_j = (unsigned long)-1;
2248         for_each_clear_bit_from(j, bitmap->bitmap, BITS_PER_BITMAP) {
2249                 if (j > i)
2250                         break;
2251                 prev_j = j;
2252         }
2253         if (prev_j == i)
2254                 return false;
2255
2256         if (prev_j == (unsigned long)-1)
2257                 bytes = (i + 1) * ctl->unit;
2258         else
2259                 bytes = (i - prev_j) * ctl->unit;
2260
2261         info->offset -= bytes;
2262         info->bytes += bytes;
2263
2264         if (update_stat)
2265                 bitmap_clear_bits(ctl, bitmap, info->offset, bytes);
2266         else
2267                 __bitmap_clear_bits(ctl, bitmap, info->offset, bytes);
2268
2269         if (!bitmap->bytes)
2270                 free_bitmap(ctl, bitmap);
2271
2272         return true;
2273 }
2274
2275 /*
2276  * We prefer always to allocate from extent entries, both for clustered and
2277  * non-clustered allocation requests. So when attempting to add a new extent
2278  * entry, try to see if there's adjacent free space in bitmap entries, and if
2279  * there is, migrate that space from the bitmaps to the extent.
2280  * Like this we get better chances of satisfying space allocation requests
2281  * because we attempt to satisfy them based on a single cache entry, and never
2282  * on 2 or more entries - even if the entries represent a contiguous free space
2283  * region (e.g. 1 extent entry + 1 bitmap entry starting where the extent entry
2284  * ends).
2285  */
2286 static void steal_from_bitmap(struct btrfs_free_space_ctl *ctl,
2287                               struct btrfs_free_space *info,
2288                               bool update_stat)
2289 {
2290         /*
2291          * Only work with disconnected entries, as we can change their offset,
2292          * and must be extent entries.
2293          */
2294         ASSERT(!info->bitmap);
2295         ASSERT(RB_EMPTY_NODE(&info->offset_index));
2296
2297         if (ctl->total_bitmaps > 0) {
2298                 bool stole_end;
2299                 bool stole_front = false;
2300
2301                 stole_end = steal_from_bitmap_to_end(ctl, info, update_stat);
2302                 if (ctl->total_bitmaps > 0)
2303                         stole_front = steal_from_bitmap_to_front(ctl, info,
2304                                                                  update_stat);
2305
2306                 if (stole_end || stole_front)
2307                         try_merge_free_space(ctl, info, update_stat);
2308         }
2309 }
2310
2311 int __btrfs_add_free_space(struct btrfs_fs_info *fs_info,
2312                            struct btrfs_free_space_ctl *ctl,
2313                            u64 offset, u64 bytes)
2314 {
2315         struct btrfs_free_space *info;
2316         int ret = 0;
2317
2318         info = kmem_cache_zalloc(btrfs_free_space_cachep, GFP_NOFS);
2319         if (!info)
2320                 return -ENOMEM;
2321
2322         info->offset = offset;
2323         info->bytes = bytes;
2324         RB_CLEAR_NODE(&info->offset_index);
2325
2326         spin_lock(&ctl->tree_lock);
2327
2328         if (try_merge_free_space(ctl, info, true))
2329                 goto link;
2330
2331         /*
2332          * There was no extent directly to the left or right of this new
2333          * extent then we know we're going to have to allocate a new extent, so
2334          * before we do that see if we need to drop this into a bitmap
2335          */
2336         ret = insert_into_bitmap(ctl, info);
2337         if (ret < 0) {
2338                 goto out;
2339         } else if (ret) {
2340                 ret = 0;
2341                 goto out;
2342         }
2343 link:
2344         /*
2345          * Only steal free space from adjacent bitmaps if we're sure we're not
2346          * going to add the new free space to existing bitmap entries - because
2347          * that would mean unnecessary work that would be reverted. Therefore
2348          * attempt to steal space from bitmaps if we're adding an extent entry.
2349          */
2350         steal_from_bitmap(ctl, info, true);
2351
2352         ret = link_free_space(ctl, info);
2353         if (ret)
2354                 kmem_cache_free(btrfs_free_space_cachep, info);
2355 out:
2356         spin_unlock(&ctl->tree_lock);
2357
2358         if (ret) {
2359                 btrfs_crit(fs_info, "unable to add free space :%d", ret);
2360                 ASSERT(ret != -EEXIST);
2361         }
2362
2363         return ret;
2364 }
2365
2366 int btrfs_remove_free_space(struct btrfs_block_group_cache *block_group,
2367                             u64 offset, u64 bytes)
2368 {
2369         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2370         struct btrfs_free_space *info;
2371         int ret;
2372         bool re_search = false;
2373
2374         spin_lock(&ctl->tree_lock);
2375
2376 again:
2377         ret = 0;
2378         if (!bytes)
2379                 goto out_lock;
2380
2381         info = tree_search_offset(ctl, offset, 0, 0);
2382         if (!info) {
2383                 /*
2384                  * oops didn't find an extent that matched the space we wanted
2385                  * to remove, look for a bitmap instead
2386                  */
2387                 info = tree_search_offset(ctl, offset_to_bitmap(ctl, offset),
2388                                           1, 0);
2389                 if (!info) {
2390                         /*
2391                          * If we found a partial bit of our free space in a
2392                          * bitmap but then couldn't find the other part this may
2393                          * be a problem, so WARN about it.
2394                          */
2395                         WARN_ON(re_search);
2396                         goto out_lock;
2397                 }
2398         }
2399
2400         re_search = false;
2401         if (!info->bitmap) {
2402                 unlink_free_space(ctl, info);
2403                 if (offset == info->offset) {
2404                         u64 to_free = min(bytes, info->bytes);
2405
2406                         info->bytes -= to_free;
2407                         info->offset += to_free;
2408                         if (info->bytes) {
2409                                 ret = link_free_space(ctl, info);
2410                                 WARN_ON(ret);
2411                         } else {
2412                                 kmem_cache_free(btrfs_free_space_cachep, info);
2413                         }
2414
2415                         offset += to_free;
2416                         bytes -= to_free;
2417                         goto again;
2418                 } else {
2419                         u64 old_end = info->bytes + info->offset;
2420
2421                         info->bytes = offset - info->offset;
2422                         ret = link_free_space(ctl, info);
2423                         WARN_ON(ret);
2424                         if (ret)
2425                                 goto out_lock;
2426
2427                         /* Not enough bytes in this entry to satisfy us */
2428                         if (old_end < offset + bytes) {
2429                                 bytes -= old_end - offset;
2430                                 offset = old_end;
2431                                 goto again;
2432                         } else if (old_end == offset + bytes) {
2433                                 /* all done */
2434                                 goto out_lock;
2435                         }
2436                         spin_unlock(&ctl->tree_lock);
2437
2438                         ret = btrfs_add_free_space(block_group, offset + bytes,
2439                                                    old_end - (offset + bytes));
2440                         WARN_ON(ret);
2441                         goto out;
2442                 }
2443         }
2444
2445         ret = remove_from_bitmap(ctl, info, &offset, &bytes);
2446         if (ret == -EAGAIN) {
2447                 re_search = true;
2448                 goto again;
2449         }
2450 out_lock:
2451         spin_unlock(&ctl->tree_lock);
2452 out:
2453         return ret;
2454 }
2455
2456 void btrfs_dump_free_space(struct btrfs_block_group_cache *block_group,
2457                            u64 bytes)
2458 {
2459         struct btrfs_fs_info *fs_info = block_group->fs_info;
2460         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2461         struct btrfs_free_space *info;
2462         struct rb_node *n;
2463         int count = 0;
2464
2465         for (n = rb_first(&ctl->free_space_offset); n; n = rb_next(n)) {
2466                 info = rb_entry(n, struct btrfs_free_space, offset_index);
2467                 if (info->bytes >= bytes && !block_group->ro)
2468                         count++;
2469                 btrfs_crit(fs_info, "entry offset %llu, bytes %llu, bitmap %s",
2470                            info->offset, info->bytes,
2471                        (info->bitmap) ? "yes" : "no");
2472         }
2473         btrfs_info(fs_info, "block group has cluster?: %s",
2474                list_empty(&block_group->cluster_list) ? "no" : "yes");
2475         btrfs_info(fs_info,
2476                    "%d blocks of free space at or bigger than bytes is", count);
2477 }
2478
2479 void btrfs_init_free_space_ctl(struct btrfs_block_group_cache *block_group)
2480 {
2481         struct btrfs_fs_info *fs_info = block_group->fs_info;
2482         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2483
2484         spin_lock_init(&ctl->tree_lock);
2485         ctl->unit = fs_info->sectorsize;
2486         ctl->start = block_group->key.objectid;
2487         ctl->private = block_group;
2488         ctl->op = &free_space_op;
2489         INIT_LIST_HEAD(&ctl->trimming_ranges);
2490         mutex_init(&ctl->cache_writeout_mutex);
2491
2492         /*
2493          * we only want to have 32k of ram per block group for keeping
2494          * track of free space, and if we pass 1/2 of that we want to
2495          * start converting things over to using bitmaps
2496          */
2497         ctl->extents_thresh = (SZ_32K / 2) / sizeof(struct btrfs_free_space);
2498 }
2499
2500 /*
2501  * for a given cluster, put all of its extents back into the free
2502  * space cache.  If the block group passed doesn't match the block group
2503  * pointed to by the cluster, someone else raced in and freed the
2504  * cluster already.  In that case, we just return without changing anything
2505  */
2506 static int
2507 __btrfs_return_cluster_to_free_space(
2508                              struct btrfs_block_group_cache *block_group,
2509                              struct btrfs_free_cluster *cluster)
2510 {
2511         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2512         struct btrfs_free_space *entry;
2513         struct rb_node *node;
2514
2515         spin_lock(&cluster->lock);
2516         if (cluster->block_group != block_group)
2517                 goto out;
2518
2519         cluster->block_group = NULL;
2520         cluster->window_start = 0;
2521         list_del_init(&cluster->block_group_list);
2522
2523         node = rb_first(&cluster->root);
2524         while (node) {
2525                 bool bitmap;
2526
2527                 entry = rb_entry(node, struct btrfs_free_space, offset_index);
2528                 node = rb_next(&entry->offset_index);
2529                 rb_erase(&entry->offset_index, &cluster->root);
2530                 RB_CLEAR_NODE(&entry->offset_index);
2531
2532                 bitmap = (entry->bitmap != NULL);
2533                 if (!bitmap) {
2534                         try_merge_free_space(ctl, entry, false);
2535                         steal_from_bitmap(ctl, entry, false);
2536                 }
2537                 tree_insert_offset(&ctl->free_space_offset,
2538                                    entry->offset, &entry->offset_index, bitmap);
2539         }
2540         cluster->root = RB_ROOT;
2541
2542 out:
2543         spin_unlock(&cluster->lock);
2544         btrfs_put_block_group(block_group);
2545         return 0;
2546 }
2547
2548 static void __btrfs_remove_free_space_cache_locked(
2549                                 struct btrfs_free_space_ctl *ctl)
2550 {
2551         struct btrfs_free_space *info;
2552         struct rb_node *node;
2553
2554         while ((node = rb_last(&ctl->free_space_offset)) != NULL) {
2555                 info = rb_entry(node, struct btrfs_free_space, offset_index);
2556                 if (!info->bitmap) {
2557                         unlink_free_space(ctl, info);
2558                         kmem_cache_free(btrfs_free_space_cachep, info);
2559                 } else {
2560                         free_bitmap(ctl, info);
2561                 }
2562
2563                 cond_resched_lock(&ctl->tree_lock);
2564         }
2565 }
2566
2567 void __btrfs_remove_free_space_cache(struct btrfs_free_space_ctl *ctl)
2568 {
2569         spin_lock(&ctl->tree_lock);
2570         __btrfs_remove_free_space_cache_locked(ctl);
2571         spin_unlock(&ctl->tree_lock);
2572 }
2573
2574 void btrfs_remove_free_space_cache(struct btrfs_block_group_cache *block_group)
2575 {
2576         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2577         struct btrfs_free_cluster *cluster;
2578         struct list_head *head;
2579
2580         spin_lock(&ctl->tree_lock);
2581         while ((head = block_group->cluster_list.next) !=
2582                &block_group->cluster_list) {
2583                 cluster = list_entry(head, struct btrfs_free_cluster,
2584                                      block_group_list);
2585
2586                 WARN_ON(cluster->block_group != block_group);
2587                 __btrfs_return_cluster_to_free_space(block_group, cluster);
2588
2589                 cond_resched_lock(&ctl->tree_lock);
2590         }
2591         __btrfs_remove_free_space_cache_locked(ctl);
2592         spin_unlock(&ctl->tree_lock);
2593
2594 }
2595
2596 u64 btrfs_find_space_for_alloc(struct btrfs_block_group_cache *block_group,
2597                                u64 offset, u64 bytes, u64 empty_size,
2598                                u64 *max_extent_size)
2599 {
2600         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2601         struct btrfs_free_space *entry = NULL;
2602         u64 bytes_search = bytes + empty_size;
2603         u64 ret = 0;
2604         u64 align_gap = 0;
2605         u64 align_gap_len = 0;
2606
2607         spin_lock(&ctl->tree_lock);
2608         entry = find_free_space(ctl, &offset, &bytes_search,
2609                                 block_group->full_stripe_len, max_extent_size);
2610         if (!entry)
2611                 goto out;
2612
2613         ret = offset;
2614         if (entry->bitmap) {
2615                 bitmap_clear_bits(ctl, entry, offset, bytes);
2616                 if (!entry->bytes)
2617                         free_bitmap(ctl, entry);
2618         } else {
2619                 unlink_free_space(ctl, entry);
2620                 align_gap_len = offset - entry->offset;
2621                 align_gap = entry->offset;
2622
2623                 entry->offset = offset + bytes;
2624                 WARN_ON(entry->bytes < bytes + align_gap_len);
2625
2626                 entry->bytes -= bytes + align_gap_len;
2627                 if (!entry->bytes)
2628                         kmem_cache_free(btrfs_free_space_cachep, entry);
2629                 else
2630                         link_free_space(ctl, entry);
2631         }
2632 out:
2633         spin_unlock(&ctl->tree_lock);
2634
2635         if (align_gap_len)
2636                 __btrfs_add_free_space(block_group->fs_info, ctl,
2637                                        align_gap, align_gap_len);
2638         return ret;
2639 }
2640
2641 /*
2642  * given a cluster, put all of its extents back into the free space
2643  * cache.  If a block group is passed, this function will only free
2644  * a cluster that belongs to the passed block group.
2645  *
2646  * Otherwise, it'll get a reference on the block group pointed to by the
2647  * cluster and remove the cluster from it.
2648  */
2649 int btrfs_return_cluster_to_free_space(
2650                                struct btrfs_block_group_cache *block_group,
2651                                struct btrfs_free_cluster *cluster)
2652 {
2653         struct btrfs_free_space_ctl *ctl;
2654         int ret;
2655
2656         /* first, get a safe pointer to the block group */
2657         spin_lock(&cluster->lock);
2658         if (!block_group) {
2659                 block_group = cluster->block_group;
2660                 if (!block_group) {
2661                         spin_unlock(&cluster->lock);
2662                         return 0;
2663                 }
2664         } else if (cluster->block_group != block_group) {
2665                 /* someone else has already freed it don't redo their work */
2666                 spin_unlock(&cluster->lock);
2667                 return 0;
2668         }
2669         atomic_inc(&block_group->count);
2670         spin_unlock(&cluster->lock);
2671
2672         ctl = block_group->free_space_ctl;
2673
2674         /* now return any extents the cluster had on it */
2675         spin_lock(&ctl->tree_lock);
2676         ret = __btrfs_return_cluster_to_free_space(block_group, cluster);
2677         spin_unlock(&ctl->tree_lock);
2678
2679         /* finally drop our ref */
2680         btrfs_put_block_group(block_group);
2681         return ret;
2682 }
2683
2684 static u64 btrfs_alloc_from_bitmap(struct btrfs_block_group_cache *block_group,
2685                                    struct btrfs_free_cluster *cluster,
2686                                    struct btrfs_free_space *entry,
2687                                    u64 bytes, u64 min_start,
2688                                    u64 *max_extent_size)
2689 {
2690         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2691         int err;
2692         u64 search_start = cluster->window_start;
2693         u64 search_bytes = bytes;
2694         u64 ret = 0;
2695
2696         search_start = min_start;
2697         search_bytes = bytes;
2698
2699         err = search_bitmap(ctl, entry, &search_start, &search_bytes, true);
2700         if (err) {
2701                 if (search_bytes > *max_extent_size)
2702                         *max_extent_size = search_bytes;
2703                 return 0;
2704         }
2705
2706         ret = search_start;
2707         __bitmap_clear_bits(ctl, entry, ret, bytes);
2708
2709         return ret;
2710 }
2711
2712 /*
2713  * given a cluster, try to allocate 'bytes' from it, returns 0
2714  * if it couldn't find anything suitably large, or a logical disk offset
2715  * if things worked out
2716  */
2717 u64 btrfs_alloc_from_cluster(struct btrfs_block_group_cache *block_group,
2718                              struct btrfs_free_cluster *cluster, u64 bytes,
2719                              u64 min_start, u64 *max_extent_size)
2720 {
2721         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2722         struct btrfs_free_space *entry = NULL;
2723         struct rb_node *node;
2724         u64 ret = 0;
2725
2726         spin_lock(&cluster->lock);
2727         if (bytes > cluster->max_size)
2728                 goto out;
2729
2730         if (cluster->block_group != block_group)
2731                 goto out;
2732
2733         node = rb_first(&cluster->root);
2734         if (!node)
2735                 goto out;
2736
2737         entry = rb_entry(node, struct btrfs_free_space, offset_index);
2738         while (1) {
2739                 if (entry->bytes < bytes && entry->bytes > *max_extent_size)
2740                         *max_extent_size = entry->bytes;
2741
2742                 if (entry->bytes < bytes ||
2743                     (!entry->bitmap && entry->offset < min_start)) {
2744                         node = rb_next(&entry->offset_index);
2745                         if (!node)
2746                                 break;
2747                         entry = rb_entry(node, struct btrfs_free_space,
2748                                          offset_index);
2749                         continue;
2750                 }
2751
2752                 if (entry->bitmap) {
2753                         ret = btrfs_alloc_from_bitmap(block_group,
2754                                                       cluster, entry, bytes,
2755                                                       cluster->window_start,
2756                                                       max_extent_size);
2757                         if (ret == 0) {
2758                                 node = rb_next(&entry->offset_index);
2759                                 if (!node)
2760                                         break;
2761                                 entry = rb_entry(node, struct btrfs_free_space,
2762                                                  offset_index);
2763                                 continue;
2764                         }
2765                         cluster->window_start += bytes;
2766                 } else {
2767                         ret = entry->offset;
2768
2769                         entry->offset += bytes;
2770                         entry->bytes -= bytes;
2771                 }
2772
2773                 if (entry->bytes == 0)
2774                         rb_erase(&entry->offset_index, &cluster->root);
2775                 break;
2776         }
2777 out:
2778         spin_unlock(&cluster->lock);
2779
2780         if (!ret)
2781                 return 0;
2782
2783         spin_lock(&ctl->tree_lock);
2784
2785         ctl->free_space -= bytes;
2786         if (entry->bytes == 0) {
2787                 ctl->free_extents--;
2788                 if (entry->bitmap) {
2789                         kfree(entry->bitmap);
2790                         ctl->total_bitmaps--;
2791                         ctl->op->recalc_thresholds(ctl);
2792                 }
2793                 kmem_cache_free(btrfs_free_space_cachep, entry);
2794         }
2795
2796         spin_unlock(&ctl->tree_lock);
2797
2798         return ret;
2799 }
2800
2801 static int btrfs_bitmap_cluster(struct btrfs_block_group_cache *block_group,
2802                                 struct btrfs_free_space *entry,
2803                                 struct btrfs_free_cluster *cluster,
2804                                 u64 offset, u64 bytes,
2805                                 u64 cont1_bytes, u64 min_bytes)
2806 {
2807         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2808         unsigned long next_zero;
2809         unsigned long i;
2810         unsigned long want_bits;
2811         unsigned long min_bits;
2812         unsigned long found_bits;
2813         unsigned long max_bits = 0;
2814         unsigned long start = 0;
2815         unsigned long total_found = 0;
2816         int ret;
2817
2818         i = offset_to_bit(entry->offset, ctl->unit,
2819                           max_t(u64, offset, entry->offset));
2820         want_bits = bytes_to_bits(bytes, ctl->unit);
2821         min_bits = bytes_to_bits(min_bytes, ctl->unit);
2822
2823         /*
2824          * Don't bother looking for a cluster in this bitmap if it's heavily
2825          * fragmented.
2826          */
2827         if (entry->max_extent_size &&
2828             entry->max_extent_size < cont1_bytes)
2829                 return -ENOSPC;
2830 again:
2831         found_bits = 0;
2832         for_each_set_bit_from(i, entry->bitmap, BITS_PER_BITMAP) {
2833                 next_zero = find_next_zero_bit(entry->bitmap,
2834                                                BITS_PER_BITMAP, i);
2835                 if (next_zero - i >= min_bits) {
2836                         found_bits = next_zero - i;
2837                         if (found_bits > max_bits)
2838                                 max_bits = found_bits;
2839                         break;
2840                 }
2841                 if (next_zero - i > max_bits)
2842                         max_bits = next_zero - i;
2843                 i = next_zero;
2844         }
2845
2846         if (!found_bits) {
2847                 entry->max_extent_size = (u64)max_bits * ctl->unit;
2848                 return -ENOSPC;
2849         }
2850
2851         if (!total_found) {
2852                 start = i;
2853                 cluster->max_size = 0;
2854         }
2855
2856         total_found += found_bits;
2857
2858         if (cluster->max_size < found_bits * ctl->unit)
2859                 cluster->max_size = found_bits * ctl->unit;
2860
2861         if (total_found < want_bits || cluster->max_size < cont1_bytes) {
2862                 i = next_zero + 1;
2863                 goto again;
2864         }
2865
2866         cluster->window_start = start * ctl->unit + entry->offset;
2867         rb_erase(&entry->offset_index, &ctl->free_space_offset);
2868         ret = tree_insert_offset(&cluster->root, entry->offset,
2869                                  &entry->offset_index, 1);
2870         ASSERT(!ret); /* -EEXIST; Logic error */
2871
2872         trace_btrfs_setup_cluster(block_group, cluster,
2873                                   total_found * ctl->unit, 1);
2874         return 0;
2875 }
2876
2877 /*
2878  * This searches the block group for just extents to fill the cluster with.
2879  * Try to find a cluster with at least bytes total bytes, at least one
2880  * extent of cont1_bytes, and other clusters of at least min_bytes.
2881  */
2882 static noinline int
2883 setup_cluster_no_bitmap(struct btrfs_block_group_cache *block_group,
2884                         struct btrfs_free_cluster *cluster,
2885                         struct list_head *bitmaps, u64 offset, u64 bytes,
2886                         u64 cont1_bytes, u64 min_bytes)
2887 {
2888         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2889         struct btrfs_free_space *first = NULL;
2890         struct btrfs_free_space *entry = NULL;
2891         struct btrfs_free_space *last;
2892         struct rb_node *node;
2893         u64 window_free;
2894         u64 max_extent;
2895         u64 total_size = 0;
2896
2897         entry = tree_search_offset(ctl, offset, 0, 1);
2898         if (!entry)
2899                 return -ENOSPC;
2900
2901         /*
2902          * We don't want bitmaps, so just move along until we find a normal
2903          * extent entry.
2904          */
2905         while (entry->bitmap || entry->bytes < min_bytes) {
2906                 if (entry->bitmap && list_empty(&entry->list))
2907                         list_add_tail(&entry->list, bitmaps);
2908                 node = rb_next(&entry->offset_index);
2909                 if (!node)
2910                         return -ENOSPC;
2911                 entry = rb_entry(node, struct btrfs_free_space, offset_index);
2912         }
2913
2914         window_free = entry->bytes;
2915         max_extent = entry->bytes;
2916         first = entry;
2917         last = entry;
2918
2919         for (node = rb_next(&entry->offset_index); node;
2920              node = rb_next(&entry->offset_index)) {
2921                 entry = rb_entry(node, struct btrfs_free_space, offset_index);
2922
2923                 if (entry->bitmap) {
2924                         if (list_empty(&entry->list))
2925                                 list_add_tail(&entry->list, bitmaps);
2926                         continue;
2927                 }
2928
2929                 if (entry->bytes < min_bytes)
2930                         continue;
2931
2932                 last = entry;
2933                 window_free += entry->bytes;
2934                 if (entry->bytes > max_extent)
2935                         max_extent = entry->bytes;
2936         }
2937
2938         if (window_free < bytes || max_extent < cont1_bytes)
2939                 return -ENOSPC;
2940
2941         cluster->window_start = first->offset;
2942
2943         node = &first->offset_index;
2944
2945         /*
2946          * now we've found our entries, pull them out of the free space
2947          * cache and put them into the cluster rbtree
2948          */
2949         do {
2950                 int ret;
2951
2952                 entry = rb_entry(node, struct btrfs_free_space, offset_index);
2953                 node = rb_next(&entry->offset_index);
2954                 if (entry->bitmap || entry->bytes < min_bytes)
2955                         continue;
2956
2957                 rb_erase(&entry->offset_index, &ctl->free_space_offset);
2958                 ret = tree_insert_offset(&cluster->root, entry->offset,
2959                                          &entry->offset_index, 0);
2960                 total_size += entry->bytes;
2961                 ASSERT(!ret); /* -EEXIST; Logic error */
2962         } while (node && entry != last);
2963
2964         cluster->max_size = max_extent;
2965         trace_btrfs_setup_cluster(block_group, cluster, total_size, 0);
2966         return 0;
2967 }
2968
2969 /*
2970  * This specifically looks for bitmaps that may work in the cluster, we assume
2971  * that we have already failed to find extents that will work.
2972  */
2973 static noinline int
2974 setup_cluster_bitmap(struct btrfs_block_group_cache *block_group,
2975                      struct btrfs_free_cluster *cluster,
2976                      struct list_head *bitmaps, u64 offset, u64 bytes,
2977                      u64 cont1_bytes, u64 min_bytes)
2978 {
2979         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2980         struct btrfs_free_space *entry = NULL;
2981         int ret = -ENOSPC;
2982         u64 bitmap_offset = offset_to_bitmap(ctl, offset);
2983
2984         if (ctl->total_bitmaps == 0)
2985                 return -ENOSPC;
2986
2987         /*
2988          * The bitmap that covers offset won't be in the list unless offset
2989          * is just its start offset.
2990          */
2991         if (!list_empty(bitmaps))
2992                 entry = list_first_entry(bitmaps, struct btrfs_free_space, list);
2993
2994         if (!entry || entry->offset != bitmap_offset) {
2995                 entry = tree_search_offset(ctl, bitmap_offset, 1, 0);
2996                 if (entry && list_empty(&entry->list))
2997                         list_add(&entry->list, bitmaps);
2998         }
2999
3000         list_for_each_entry(entry, bitmaps, list) {
3001                 if (entry->bytes < bytes)
3002                         continue;
3003                 ret = btrfs_bitmap_cluster(block_group, entry, cluster, offset,
3004                                            bytes, cont1_bytes, min_bytes);
3005                 if (!ret)
3006                         return 0;
3007         }
3008
3009         /*
3010          * The bitmaps list has all the bitmaps that record free space
3011          * starting after offset, so no more search is required.
3012          */
3013         return -ENOSPC;
3014 }
3015
3016 /*
3017  * here we try to find a cluster of blocks in a block group.  The goal
3018  * is to find at least bytes+empty_size.
3019  * We might not find them all in one contiguous area.
3020  *
3021  * returns zero and sets up cluster if things worked out, otherwise
3022  * it returns -enospc
3023  */
3024 int btrfs_find_space_cluster(struct btrfs_fs_info *fs_info,
3025                              struct btrfs_block_group_cache *block_group,
3026                              struct btrfs_free_cluster *cluster,
3027                              u64 offset, u64 bytes, u64 empty_size)
3028 {
3029         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3030         struct btrfs_free_space *entry, *tmp;
3031         LIST_HEAD(bitmaps);
3032         u64 min_bytes;
3033         u64 cont1_bytes;
3034         int ret;
3035
3036         /*
3037          * Choose the minimum extent size we'll require for this
3038          * cluster.  For SSD_SPREAD, don't allow any fragmentation.
3039          * For metadata, allow allocates with smaller extents.  For
3040          * data, keep it dense.
3041          */
3042         if (btrfs_test_opt(fs_info, SSD_SPREAD)) {
3043                 cont1_bytes = min_bytes = bytes + empty_size;
3044         } else if (block_group->flags & BTRFS_BLOCK_GROUP_METADATA) {
3045                 cont1_bytes = bytes;
3046                 min_bytes = fs_info->sectorsize;
3047         } else {
3048                 cont1_bytes = max(bytes, (bytes + empty_size) >> 2);
3049                 min_bytes = fs_info->sectorsize;
3050         }
3051
3052         spin_lock(&ctl->tree_lock);
3053
3054         /*
3055          * If we know we don't have enough space to make a cluster don't even
3056          * bother doing all the work to try and find one.
3057          */
3058         if (ctl->free_space < bytes) {
3059                 spin_unlock(&ctl->tree_lock);
3060                 return -ENOSPC;
3061         }
3062
3063         spin_lock(&cluster->lock);
3064
3065         /* someone already found a cluster, hooray */
3066         if (cluster->block_group) {
3067                 ret = 0;
3068                 goto out;
3069         }
3070
3071         trace_btrfs_find_cluster(block_group, offset, bytes, empty_size,
3072                                  min_bytes);
3073
3074         ret = setup_cluster_no_bitmap(block_group, cluster, &bitmaps, offset,
3075                                       bytes + empty_size,
3076                                       cont1_bytes, min_bytes);
3077         if (ret)
3078                 ret = setup_cluster_bitmap(block_group, cluster, &bitmaps,
3079                                            offset, bytes + empty_size,
3080                                            cont1_bytes, min_bytes);
3081
3082         /* Clear our temporary list */
3083         list_for_each_entry_safe(entry, tmp, &bitmaps, list)
3084                 list_del_init(&entry->list);
3085
3086         if (!ret) {
3087                 atomic_inc(&block_group->count);
3088                 list_add_tail(&cluster->block_group_list,
3089                               &block_group->cluster_list);
3090                 cluster->block_group = block_group;
3091         } else {
3092                 trace_btrfs_failed_cluster_setup(block_group);
3093         }
3094 out:
3095         spin_unlock(&cluster->lock);
3096         spin_unlock(&ctl->tree_lock);
3097
3098         return ret;
3099 }
3100
3101 /*
3102  * simple code to zero out a cluster
3103  */
3104 void btrfs_init_free_cluster(struct btrfs_free_cluster *cluster)
3105 {
3106         spin_lock_init(&cluster->lock);
3107         spin_lock_init(&cluster->refill_lock);
3108         cluster->root = RB_ROOT;
3109         cluster->max_size = 0;
3110         cluster->fragmented = false;
3111         INIT_LIST_HEAD(&cluster->block_group_list);
3112         cluster->block_group = NULL;
3113 }
3114
3115 static int do_trimming(struct btrfs_block_group_cache *block_group,
3116                        u64 *total_trimmed, u64 start, u64 bytes,
3117                        u64 reserved_start, u64 reserved_bytes,
3118                        struct btrfs_trim_range *trim_entry)
3119 {
3120         struct btrfs_space_info *space_info = block_group->space_info;
3121         struct btrfs_fs_info *fs_info = block_group->fs_info;
3122         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3123         int ret;
3124         int update = 0;
3125         u64 trimmed = 0;
3126
3127         spin_lock(&space_info->lock);
3128         spin_lock(&block_group->lock);
3129         if (!block_group->ro) {
3130                 block_group->reserved += reserved_bytes;
3131                 space_info->bytes_reserved += reserved_bytes;
3132                 update = 1;
3133         }
3134         spin_unlock(&block_group->lock);
3135         spin_unlock(&space_info->lock);
3136
3137         ret = btrfs_discard_extent(fs_info, start, bytes, &trimmed);
3138         if (!ret)
3139                 *total_trimmed += trimmed;
3140
3141         mutex_lock(&ctl->cache_writeout_mutex);
3142         btrfs_add_free_space(block_group, reserved_start, reserved_bytes);
3143         list_del(&trim_entry->list);
3144         mutex_unlock(&ctl->cache_writeout_mutex);
3145
3146         if (update) {
3147                 spin_lock(&space_info->lock);
3148                 spin_lock(&block_group->lock);
3149                 if (block_group->ro)
3150                         space_info->bytes_readonly += reserved_bytes;
3151                 block_group->reserved -= reserved_bytes;
3152                 space_info->bytes_reserved -= reserved_bytes;
3153                 spin_unlock(&space_info->lock);
3154                 spin_unlock(&block_group->lock);
3155         }
3156
3157         return ret;
3158 }
3159
3160 static int trim_no_bitmap(struct btrfs_block_group_cache *block_group,
3161                           u64 *total_trimmed, u64 start, u64 end, u64 minlen)
3162 {
3163         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3164         struct btrfs_free_space *entry;
3165         struct rb_node *node;
3166         int ret = 0;
3167         u64 extent_start;
3168         u64 extent_bytes;
3169         u64 bytes;
3170
3171         while (start < end) {
3172                 struct btrfs_trim_range trim_entry;
3173
3174                 mutex_lock(&ctl->cache_writeout_mutex);
3175                 spin_lock(&ctl->tree_lock);
3176
3177                 if (ctl->free_space < minlen) {
3178                         spin_unlock(&ctl->tree_lock);
3179                         mutex_unlock(&ctl->cache_writeout_mutex);
3180                         break;
3181                 }
3182
3183                 entry = tree_search_offset(ctl, start, 0, 1);
3184                 if (!entry) {
3185                         spin_unlock(&ctl->tree_lock);
3186                         mutex_unlock(&ctl->cache_writeout_mutex);
3187                         break;
3188                 }
3189
3190                 /* skip bitmaps */
3191                 while (entry->bitmap) {
3192                         node = rb_next(&entry->offset_index);
3193                         if (!node) {
3194                                 spin_unlock(&ctl->tree_lock);
3195                                 mutex_unlock(&ctl->cache_writeout_mutex);
3196                                 goto out;
3197                         }
3198                         entry = rb_entry(node, struct btrfs_free_space,
3199                                          offset_index);
3200                 }
3201
3202                 if (entry->offset >= end) {
3203                         spin_unlock(&ctl->tree_lock);
3204                         mutex_unlock(&ctl->cache_writeout_mutex);
3205                         break;
3206                 }
3207
3208                 extent_start = entry->offset;
3209                 extent_bytes = entry->bytes;
3210                 start = max(start, extent_start);
3211                 bytes = min(extent_start + extent_bytes, end) - start;
3212                 if (bytes < minlen) {
3213                         spin_unlock(&ctl->tree_lock);
3214                         mutex_unlock(&ctl->cache_writeout_mutex);
3215                         goto next;
3216                 }
3217
3218                 unlink_free_space(ctl, entry);
3219                 kmem_cache_free(btrfs_free_space_cachep, entry);
3220
3221                 spin_unlock(&ctl->tree_lock);
3222                 trim_entry.start = extent_start;
3223                 trim_entry.bytes = extent_bytes;
3224                 list_add_tail(&trim_entry.list, &ctl->trimming_ranges);
3225                 mutex_unlock(&ctl->cache_writeout_mutex);
3226
3227                 ret = do_trimming(block_group, total_trimmed, start, bytes,
3228                                   extent_start, extent_bytes, &trim_entry);
3229                 if (ret)
3230                         break;
3231 next:
3232                 start += bytes;
3233
3234                 if (fatal_signal_pending(current)) {
3235                         ret = -ERESTARTSYS;
3236                         break;
3237                 }
3238
3239                 cond_resched();
3240         }
3241 out:
3242         return ret;
3243 }
3244
3245 static int trim_bitmaps(struct btrfs_block_group_cache *block_group,
3246                         u64 *total_trimmed, u64 start, u64 end, u64 minlen)
3247 {
3248         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3249         struct btrfs_free_space *entry;
3250         int ret = 0;
3251         int ret2;
3252         u64 bytes;
3253         u64 offset = offset_to_bitmap(ctl, start);
3254
3255         while (offset < end) {
3256                 bool next_bitmap = false;
3257                 struct btrfs_trim_range trim_entry;
3258
3259                 mutex_lock(&ctl->cache_writeout_mutex);
3260                 spin_lock(&ctl->tree_lock);
3261
3262                 if (ctl->free_space < minlen) {
3263                         spin_unlock(&ctl->tree_lock);
3264                         mutex_unlock(&ctl->cache_writeout_mutex);
3265                         break;
3266                 }
3267
3268                 entry = tree_search_offset(ctl, offset, 1, 0);
3269                 if (!entry) {
3270                         spin_unlock(&ctl->tree_lock);
3271                         mutex_unlock(&ctl->cache_writeout_mutex);
3272                         next_bitmap = true;
3273                         goto next;
3274                 }
3275
3276                 bytes = minlen;
3277                 ret2 = search_bitmap(ctl, entry, &start, &bytes, false);
3278                 if (ret2 || start >= end) {
3279                         spin_unlock(&ctl->tree_lock);
3280                         mutex_unlock(&ctl->cache_writeout_mutex);
3281                         next_bitmap = true;
3282                         goto next;
3283                 }
3284
3285                 bytes = min(bytes, end - start);
3286                 if (bytes < minlen) {
3287                         spin_unlock(&ctl->tree_lock);
3288                         mutex_unlock(&ctl->cache_writeout_mutex);
3289                         goto next;
3290                 }
3291
3292                 bitmap_clear_bits(ctl, entry, start, bytes);
3293                 if (entry->bytes == 0)
3294                         free_bitmap(ctl, entry);
3295
3296                 spin_unlock(&ctl->tree_lock);
3297                 trim_entry.start = start;
3298                 trim_entry.bytes = bytes;
3299                 list_add_tail(&trim_entry.list, &ctl->trimming_ranges);
3300                 mutex_unlock(&ctl->cache_writeout_mutex);
3301
3302                 ret = do_trimming(block_group, total_trimmed, start, bytes,
3303                                   start, bytes, &trim_entry);
3304                 if (ret)
3305                         break;
3306 next:
3307                 if (next_bitmap) {
3308                         offset += BITS_PER_BITMAP * ctl->unit;
3309                 } else {
3310                         start += bytes;
3311                         if (start >= offset + BITS_PER_BITMAP * ctl->unit)
3312                                 offset += BITS_PER_BITMAP * ctl->unit;
3313                 }
3314
3315                 if (fatal_signal_pending(current)) {
3316                         ret = -ERESTARTSYS;
3317                         break;
3318                 }
3319
3320                 cond_resched();
3321         }
3322
3323         return ret;
3324 }
3325
3326 void btrfs_get_block_group_trimming(struct btrfs_block_group_cache *cache)
3327 {
3328         atomic_inc(&cache->trimming);
3329 }
3330
3331 void btrfs_put_block_group_trimming(struct btrfs_block_group_cache *block_group)
3332 {
3333         struct btrfs_fs_info *fs_info = block_group->fs_info;
3334         struct extent_map_tree *em_tree;
3335         struct extent_map *em;
3336         bool cleanup;
3337
3338         spin_lock(&block_group->lock);
3339         cleanup = (atomic_dec_and_test(&block_group->trimming) &&
3340                    block_group->removed);
3341         spin_unlock(&block_group->lock);
3342
3343         if (cleanup) {
3344                 mutex_lock(&fs_info->chunk_mutex);
3345                 em_tree = &fs_info->mapping_tree.map_tree;
3346                 write_lock(&em_tree->lock);
3347                 em = lookup_extent_mapping(em_tree, block_group->key.objectid,
3348                                            1);
3349                 BUG_ON(!em); /* logic error, can't happen */
3350                 /*
3351                  * remove_extent_mapping() will delete us from the pinned_chunks
3352                  * list, which is protected by the chunk mutex.
3353                  */
3354                 remove_extent_mapping(em_tree, em);
3355                 write_unlock(&em_tree->lock);
3356                 mutex_unlock(&fs_info->chunk_mutex);
3357
3358                 /* once for us and once for the tree */
3359                 free_extent_map(em);
3360                 free_extent_map(em);
3361
3362                 /*
3363                  * We've left one free space entry and other tasks trimming
3364                  * this block group have left 1 entry each one. Free them.
3365                  */
3366                 __btrfs_remove_free_space_cache(block_group->free_space_ctl);
3367         }
3368 }
3369
3370 int btrfs_trim_block_group(struct btrfs_block_group_cache *block_group,
3371                            u64 *trimmed, u64 start, u64 end, u64 minlen)
3372 {
3373         int ret;
3374
3375         *trimmed = 0;
3376
3377         spin_lock(&block_group->lock);
3378         if (block_group->removed) {
3379                 spin_unlock(&block_group->lock);
3380                 return 0;
3381         }
3382         btrfs_get_block_group_trimming(block_group);
3383         spin_unlock(&block_group->lock);
3384
3385         ret = trim_no_bitmap(block_group, trimmed, start, end, minlen);
3386         if (ret)
3387                 goto out;
3388
3389         ret = trim_bitmaps(block_group, trimmed, start, end, minlen);
3390 out:
3391         btrfs_put_block_group_trimming(block_group);
3392         return ret;
3393 }
3394
3395 /*
3396  * Find the left-most item in the cache tree, and then return the
3397  * smallest inode number in the item.
3398  *
3399  * Note: the returned inode number may not be the smallest one in
3400  * the tree, if the left-most item is a bitmap.
3401  */
3402 u64 btrfs_find_ino_for_alloc(struct btrfs_root *fs_root)
3403 {
3404         struct btrfs_free_space_ctl *ctl = fs_root->free_ino_ctl;
3405         struct btrfs_free_space *entry = NULL;
3406         u64 ino = 0;
3407
3408         spin_lock(&ctl->tree_lock);
3409
3410         if (RB_EMPTY_ROOT(&ctl->free_space_offset))
3411                 goto out;
3412
3413         entry = rb_entry(rb_first(&ctl->free_space_offset),
3414                          struct btrfs_free_space, offset_index);
3415
3416         if (!entry->bitmap) {
3417                 ino = entry->offset;
3418
3419                 unlink_free_space(ctl, entry);
3420                 entry->offset++;
3421                 entry->bytes--;
3422                 if (!entry->bytes)
3423                         kmem_cache_free(btrfs_free_space_cachep, entry);
3424                 else
3425                         link_free_space(ctl, entry);
3426         } else {
3427                 u64 offset = 0;
3428                 u64 count = 1;
3429                 int ret;
3430
3431                 ret = search_bitmap(ctl, entry, &offset, &count, true);
3432                 /* Logic error; Should be empty if it can't find anything */
3433                 ASSERT(!ret);
3434
3435                 ino = offset;
3436                 bitmap_clear_bits(ctl, entry, offset, 1);
3437                 if (entry->bytes == 0)
3438                         free_bitmap(ctl, entry);
3439         }
3440 out:
3441         spin_unlock(&ctl->tree_lock);
3442
3443         return ino;
3444 }
3445
3446 struct inode *lookup_free_ino_inode(struct btrfs_root *root,
3447                                     struct btrfs_path *path)
3448 {
3449         struct inode *inode = NULL;
3450
3451         spin_lock(&root->ino_cache_lock);
3452         if (root->ino_cache_inode)
3453                 inode = igrab(root->ino_cache_inode);
3454         spin_unlock(&root->ino_cache_lock);
3455         if (inode)
3456                 return inode;
3457
3458         inode = __lookup_free_space_inode(root, path, 0);
3459         if (IS_ERR(inode))
3460                 return inode;
3461
3462         spin_lock(&root->ino_cache_lock);
3463         if (!btrfs_fs_closing(root->fs_info))
3464                 root->ino_cache_inode = igrab(inode);
3465         spin_unlock(&root->ino_cache_lock);
3466
3467         return inode;
3468 }
3469
3470 int create_free_ino_inode(struct btrfs_root *root,
3471                           struct btrfs_trans_handle *trans,
3472                           struct btrfs_path *path)
3473 {
3474         return __create_free_space_inode(root, trans, path,
3475                                          BTRFS_FREE_INO_OBJECTID, 0);
3476 }
3477
3478 int load_free_ino_cache(struct btrfs_fs_info *fs_info, struct btrfs_root *root)
3479 {
3480         struct btrfs_free_space_ctl *ctl = root->free_ino_ctl;
3481         struct btrfs_path *path;
3482         struct inode *inode;
3483         int ret = 0;
3484         u64 root_gen = btrfs_root_generation(&root->root_item);
3485
3486         if (!btrfs_test_opt(fs_info, INODE_MAP_CACHE))
3487                 return 0;
3488
3489         /*
3490          * If we're unmounting then just return, since this does a search on the
3491          * normal root and not the commit root and we could deadlock.
3492          */
3493         if (btrfs_fs_closing(fs_info))
3494                 return 0;
3495
3496         path = btrfs_alloc_path();
3497         if (!path)
3498                 return 0;
3499
3500         inode = lookup_free_ino_inode(root, path);
3501         if (IS_ERR(inode))
3502                 goto out;
3503
3504         if (root_gen != BTRFS_I(inode)->generation)
3505                 goto out_put;
3506
3507         ret = __load_free_space_cache(root, inode, ctl, path, 0);
3508
3509         if (ret < 0)
3510                 btrfs_err(fs_info,
3511                         "failed to load free ino cache for root %llu",
3512                         root->root_key.objectid);
3513 out_put:
3514         iput(inode);
3515 out:
3516         btrfs_free_path(path);
3517         return ret;
3518 }
3519
3520 int btrfs_write_out_ino_cache(struct btrfs_root *root,
3521                               struct btrfs_trans_handle *trans,
3522                               struct btrfs_path *path,
3523                               struct inode *inode)
3524 {
3525         struct btrfs_fs_info *fs_info = root->fs_info;
3526         struct btrfs_free_space_ctl *ctl = root->free_ino_ctl;
3527         int ret;
3528         struct btrfs_io_ctl io_ctl;
3529         bool release_metadata = true;
3530
3531         if (!btrfs_test_opt(fs_info, INODE_MAP_CACHE))
3532                 return 0;
3533
3534         memset(&io_ctl, 0, sizeof(io_ctl));
3535         ret = __btrfs_write_out_cache(root, inode, ctl, NULL, &io_ctl, trans);
3536         if (!ret) {
3537                 /*
3538                  * At this point writepages() didn't error out, so our metadata
3539                  * reservation is released when the writeback finishes, at
3540                  * inode.c:btrfs_finish_ordered_io(), regardless of it finishing
3541                  * with or without an error.
3542                  */
3543                 release_metadata = false;
3544                 ret = btrfs_wait_cache_io_root(root, trans, &io_ctl, path);
3545         }
3546
3547         if (ret) {
3548                 if (release_metadata)
3549                         btrfs_delalloc_release_metadata(BTRFS_I(inode),
3550                                         inode->i_size);
3551 #ifdef DEBUG
3552                 btrfs_err(fs_info,
3553                           "failed to write free ino cache for root %llu",
3554                           root->root_key.objectid);
3555 #endif
3556         }
3557
3558         return ret;
3559 }
3560
3561 #ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
3562 /*
3563  * Use this if you need to make a bitmap or extent entry specifically, it
3564  * doesn't do any of the merging that add_free_space does, this acts a lot like
3565  * how the free space cache loading stuff works, so you can get really weird
3566  * configurations.
3567  */
3568 int test_add_free_space_entry(struct btrfs_block_group_cache *cache,
3569                               u64 offset, u64 bytes, bool bitmap)
3570 {
3571         struct btrfs_free_space_ctl *ctl = cache->free_space_ctl;
3572         struct btrfs_free_space *info = NULL, *bitmap_info;
3573         void *map = NULL;
3574         u64 bytes_added;
3575         int ret;
3576
3577 again:
3578         if (!info) {
3579                 info = kmem_cache_zalloc(btrfs_free_space_cachep, GFP_NOFS);
3580                 if (!info)
3581                         return -ENOMEM;
3582         }
3583
3584         if (!bitmap) {
3585                 spin_lock(&ctl->tree_lock);
3586                 info->offset = offset;
3587                 info->bytes = bytes;
3588                 info->max_extent_size = 0;
3589                 ret = link_free_space(ctl, info);
3590                 spin_unlock(&ctl->tree_lock);
3591                 if (ret)
3592                         kmem_cache_free(btrfs_free_space_cachep, info);
3593                 return ret;
3594         }
3595
3596         if (!map) {
3597                 map = kzalloc(PAGE_SIZE, GFP_NOFS);
3598                 if (!map) {
3599                         kmem_cache_free(btrfs_free_space_cachep, info);
3600                         return -ENOMEM;
3601                 }
3602         }
3603
3604         spin_lock(&ctl->tree_lock);
3605         bitmap_info = tree_search_offset(ctl, offset_to_bitmap(ctl, offset),
3606                                          1, 0);
3607         if (!bitmap_info) {
3608                 info->bitmap = map;
3609                 map = NULL;
3610                 add_new_bitmap(ctl, info, offset);
3611                 bitmap_info = info;
3612                 info = NULL;
3613         }
3614
3615         bytes_added = add_bytes_to_bitmap(ctl, bitmap_info, offset, bytes);
3616
3617         bytes -= bytes_added;
3618         offset += bytes_added;
3619         spin_unlock(&ctl->tree_lock);
3620
3621         if (bytes)
3622                 goto again;
3623
3624         if (info)
3625                 kmem_cache_free(btrfs_free_space_cachep, info);
3626         if (map)
3627                 kfree(map);
3628         return 0;
3629 }
3630
3631 /*
3632  * Checks to see if the given range is in the free space cache.  This is really
3633  * just used to check the absence of space, so if there is free space in the
3634  * range at all we will return 1.
3635  */
3636 int test_check_exists(struct btrfs_block_group_cache *cache,
3637                       u64 offset, u64 bytes)
3638 {
3639         struct btrfs_free_space_ctl *ctl = cache->free_space_ctl;
3640         struct btrfs_free_space *info;
3641         int ret = 0;
3642
3643         spin_lock(&ctl->tree_lock);
3644         info = tree_search_offset(ctl, offset, 0, 0);
3645         if (!info) {
3646                 info = tree_search_offset(ctl, offset_to_bitmap(ctl, offset),
3647                                           1, 0);
3648                 if (!info)
3649                         goto out;
3650         }
3651
3652 have_info:
3653         if (info->bitmap) {
3654                 u64 bit_off, bit_bytes;
3655                 struct rb_node *n;
3656                 struct btrfs_free_space *tmp;
3657
3658                 bit_off = offset;
3659                 bit_bytes = ctl->unit;
3660                 ret = search_bitmap(ctl, info, &bit_off, &bit_bytes, false);
3661                 if (!ret) {
3662                         if (bit_off == offset) {
3663                                 ret = 1;
3664                                 goto out;
3665                         } else if (bit_off > offset &&
3666                                    offset + bytes > bit_off) {
3667                                 ret = 1;
3668                                 goto out;
3669                         }
3670                 }
3671
3672                 n = rb_prev(&info->offset_index);
3673                 while (n) {
3674                         tmp = rb_entry(n, struct btrfs_free_space,
3675                                        offset_index);
3676                         if (tmp->offset + tmp->bytes < offset)
3677                                 break;
3678                         if (offset + bytes < tmp->offset) {
3679                                 n = rb_prev(&tmp->offset_index);
3680                                 continue;
3681                         }
3682                         info = tmp;
3683                         goto have_info;
3684                 }
3685
3686                 n = rb_next(&info->offset_index);
3687                 while (n) {
3688                         tmp = rb_entry(n, struct btrfs_free_space,
3689                                        offset_index);
3690                         if (offset + bytes < tmp->offset)
3691                                 break;
3692                         if (tmp->offset + tmp->bytes < offset) {
3693                                 n = rb_next(&tmp->offset_index);
3694                                 continue;
3695                         }
3696                         info = tmp;
3697                         goto have_info;
3698                 }
3699
3700                 ret = 0;
3701                 goto out;
3702         }
3703
3704         if (info->offset == offset) {
3705                 ret = 1;
3706                 goto out;
3707         }
3708
3709         if (offset > info->offset && offset < info->offset + info->bytes)
3710                 ret = 1;
3711 out:
3712         spin_unlock(&ctl->tree_lock);
3713         return ret;
3714 }
3715 #endif /* CONFIG_BTRFS_FS_RUN_SANITY_TESTS */