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