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_root *root,
  98                                       struct btrfs_block_group_cache
  99                                       *block_group, struct btrfs_path *path)
 100 {
 101         struct inode *inode = NULL;
 102         struct btrfs_fs_info *fs_info = root->fs_info;
 103         u32 flags = BTRFS_INODE_NODATASUM | BTRFS_INODE_NODATACOW;
 104
 105         spin_lock(&block_group->lock);
 106         if (block_group->inode)
 107                 inode = igrab(block_group->inode);
 108         spin_unlock(&block_group->lock);
 109         if (inode)
 110                 return inode;
 111
 112         inode = __lookup_free_space_inode(root, path,
 113                                           block_group->key.objectid);
 114         if (IS_ERR(inode))
 115                 return inode;
 116
 117         spin_lock(&block_group->lock);
 118         if (!((BTRFS_I(inode)->flags & flags) == flags)) {
 119                 btrfs_info(fs_info, "Old style space inode found, converting.");
 120                 BTRFS_I(inode)->flags |= BTRFS_INODE_NODATASUM |
 121                         BTRFS_INODE_NODATACOW;
 122                 block_group->disk_cache_state = BTRFS_DC_CLEAR;
 123         }
 124
 125         if (!block_group->iref) {
 126                 block_group->inode = igrab(inode);
 127                 block_group->iref = 1;
 128         }
 129         spin_unlock(&block_group->lock);
 130
 131         return inode;
 132 }
 133
 134 static int __create_free_space_inode(struct btrfs_root *root,
 135                                      struct btrfs_trans_handle *trans,
 136                                      struct btrfs_path *path,
 137                                      u64 ino, u64 offset)
 138 {
 139         struct btrfs_key key;
 140         struct btrfs_disk_key disk_key;
 141         struct btrfs_free_space_header *header;
 142         struct btrfs_inode_item *inode_item;
 143         struct extent_buffer *leaf;
 144         u64 flags = BTRFS_INODE_NOCOMPRESS | BTRFS_INODE_PREALLOC;
 145         int ret;
 146
 147         ret = btrfs_insert_empty_inode(trans, root, path, ino);
 148         if (ret)
 149                 return ret;
 150
 151         /* We inline crc's for the free disk space cache */
 152         if (ino != BTRFS_FREE_INO_OBJECTID)
 153                 flags |= BTRFS_INODE_NODATASUM | BTRFS_INODE_NODATACOW;
 154
 155         leaf = path->nodes[0];
 156         inode_item = btrfs_item_ptr(leaf, path->slots[0],
 157                                     struct btrfs_inode_item);
 158         btrfs_item_key(leaf, &disk_key, path->slots[0]);
 159         memzero_extent_buffer(leaf, (unsigned long)inode_item,
 160                              sizeof(*inode_item));
 161         btrfs_set_inode_generation(leaf, inode_item, trans->transid);
 162         btrfs_set_inode_size(leaf, inode_item, 0);
 163         btrfs_set_inode_nbytes(leaf, inode_item, 0);
 164         btrfs_set_inode_uid(leaf, inode_item, 0);
 165         btrfs_set_inode_gid(leaf, inode_item, 0);
 166         btrfs_set_inode_mode(leaf, inode_item, S_IFREG | 0600);
 167         btrfs_set_inode_flags(leaf, inode_item, flags);
 168         btrfs_set_inode_nlink(leaf, inode_item, 1);
 169         btrfs_set_inode_transid(leaf, inode_item, trans->transid);
 170         btrfs_set_inode_block_group(leaf, inode_item, offset);
 171         btrfs_mark_buffer_dirty(leaf);
 172         btrfs_release_path(path);
 173
 174         key.objectid = BTRFS_FREE_SPACE_OBJECTID;
 175         key.offset = offset;
 176         key.type = 0;
 177         ret = btrfs_insert_empty_item(trans, root, path, &key,
 178                                       sizeof(struct btrfs_free_space_header));
 179         if (ret < 0) {
 180                 btrfs_release_path(path);
 181                 return ret;
 182         }
 183
 184         leaf = path->nodes[0];
 185         header = btrfs_item_ptr(leaf, path->slots[0],
 186                                 struct btrfs_free_space_header);
 187         memzero_extent_buffer(leaf, (unsigned long)header, sizeof(*header));
 188         btrfs_set_free_space_key(leaf, header, &disk_key);
 189         btrfs_mark_buffer_dirty(leaf);
 190         btrfs_release_path(path);
 191
 192         return 0;
 193 }
 194
 195 int create_free_space_inode(struct btrfs_root *root,
 196                             struct btrfs_trans_handle *trans,
 197                             struct btrfs_block_group_cache *block_group,
 198                             struct btrfs_path *path)
 199 {
 200         int ret;
 201         u64 ino;
 202
 203         ret = btrfs_find_free_objectid(root, &ino);
 204         if (ret < 0)
 205                 return ret;
 206
 207         return __create_free_space_inode(root, trans, path, ino,
 208                                          block_group->key.objectid);
 209 }
 210
 211 int btrfs_check_trunc_cache_free_space(struct btrfs_fs_info *fs_info,
 212                                        struct btrfs_block_rsv *rsv)
 213 {
 214         u64 needed_bytes;
 215         int ret;
 216
 217         /* 1 for slack space, 1 for updating the inode */
 218         needed_bytes = btrfs_calc_trunc_metadata_size(fs_info, 1) +
 219                 btrfs_calc_trans_metadata_size(fs_info, 1);
 220
 221         spin_lock(&rsv->lock);
 222         if (rsv->reserved < needed_bytes)
 223                 ret = -ENOSPC;
 224         else
 225                 ret = 0;
 226         spin_unlock(&rsv->lock);
 227         return ret;
 228 }
 229
 230 int btrfs_truncate_free_space_cache(struct btrfs_root *root,
 231                                     struct btrfs_trans_handle *trans,
 232                                     struct btrfs_block_group_cache *block_group,
 233                                     struct inode *inode)
 234 {
 235         int ret = 0;
 236         struct btrfs_path *path = btrfs_alloc_path();
 237         bool locked = false;
 238
 239         if (!path) {
 240                 ret = -ENOMEM;
 241                 goto fail;
 242         }
 243
 244         if (block_group) {
 245                 locked = true;
 246                 mutex_lock(&trans->transaction->cache_write_mutex);
 247                 if (!list_empty(&block_group->io_list)) {
 248                         list_del_init(&block_group->io_list);
 249
 250                         btrfs_wait_cache_io(trans, block_group, path);
 251                         btrfs_put_block_group(block_group);
 252                 }
 253
 254                 /*
 255                  * now that we've truncated the cache away, its no longer
 256                  * setup or written
 257                  */
 258                 spin_lock(&block_group->lock);
 259                 block_group->disk_cache_state = BTRFS_DC_CLEAR;
 260                 spin_unlock(&block_group->lock);
 261         }
 262         btrfs_free_path(path);
 263
 264         btrfs_i_size_write(inode, 0);
 265         truncate_pagecache(inode, 0);
 266
 267         /*
 268          * We don't need an orphan item because truncating the free space cache
 269          * will never be split across transactions.
 270          * We don't need to check for -EAGAIN because we're a free space
 271          * cache inode
 272          */
 273         ret = btrfs_truncate_inode_items(trans, root, inode,
 274                                          0, BTRFS_EXTENT_DATA_KEY);
 275         if (ret)
 276                 goto fail;
 277
 278         ret = btrfs_update_inode(trans, root, inode);
 279
 280 fail:
 281         if (locked)
 282                 mutex_unlock(&trans->transaction->cache_write_mutex);
 283         if (ret)
 284                 btrfs_abort_transaction(trans, ret);
 285
 286         return ret;
 287 }
 288
 289 static int readahead_cache(struct inode *inode)
 290 {
 291         struct file_ra_state *ra;
 292         unsigned long last_index;
 293
 294         ra = kzalloc(sizeof(*ra), GFP_NOFS);
 295         if (!ra)
 296                 return -ENOMEM;
 297
 298         file_ra_state_init(ra, inode->i_mapping);
 299         last_index = (i_size_read(inode) - 1) >> PAGE_SHIFT;
 300
 301         page_cache_sync_readahead(inode->i_mapping, ra, NULL, 0, last_index);
 302
 303         kfree(ra);
 304
 305         return 0;
 306 }
 307
 308 static int io_ctl_init(struct btrfs_io_ctl *io_ctl, struct inode *inode,
 309                        int write)
 310 {
 311         int num_pages;
 312         int check_crcs = 0;
 313
 314         num_pages = DIV_ROUND_UP(i_size_read(inode), PAGE_SIZE);
 315
 316         if (btrfs_ino(BTRFS_I(inode)) != BTRFS_FREE_INO_OBJECTID)
 317                 check_crcs = 1;
 318
 319         /* Make sure we can fit our crcs into the first page */
 320         if (write && check_crcs &&
 321             (num_pages * sizeof(u32)) >= PAGE_SIZE)
 322                 return -ENOSPC;
 323
 324         memset(io_ctl, 0, sizeof(struct btrfs_io_ctl));
 325
 326         io_ctl->pages = kcalloc(num_pages, sizeof(struct page *), GFP_NOFS);
 327         if (!io_ctl->pages)
 328                 return -ENOMEM;
 329
 330         io_ctl->num_pages = num_pages;
 331         io_ctl->fs_info = btrfs_sb(inode->i_sb);
 332         io_ctl->check_crcs = check_crcs;
 333         io_ctl->inode = inode;
 334
 335         return 0;
 336 }
 337
 338 static void io_ctl_free(struct btrfs_io_ctl *io_ctl)
 339 {
 340         kfree(io_ctl->pages);
 341         io_ctl->pages = NULL;
 342 }
 343
 344 static void io_ctl_unmap_page(struct btrfs_io_ctl *io_ctl)
 345 {
 346         if (io_ctl->cur) {
 347                 io_ctl->cur = NULL;
 348                 io_ctl->orig = NULL;
 349         }
 350 }
 351
 352 static void io_ctl_map_page(struct btrfs_io_ctl *io_ctl, int clear)
 353 {
 354         ASSERT(io_ctl->index < io_ctl->num_pages);
 355         io_ctl->page = io_ctl->pages[io_ctl->index++];
 356         io_ctl->cur = page_address(io_ctl->page);
 357         io_ctl->orig = io_ctl->cur;
 358         io_ctl->size = PAGE_SIZE;
 359         if (clear)
 360                 memset(io_ctl->cur, 0, PAGE_SIZE);
 361 }
 362
 363 static void io_ctl_drop_pages(struct btrfs_io_ctl *io_ctl)
 364 {
 365         int i;
 366
 367         io_ctl_unmap_page(io_ctl);
 368
 369         for (i = 0; i < io_ctl->num_pages; i++) {
 370                 if (io_ctl->pages[i]) {
 371                         ClearPageChecked(io_ctl->pages[i]);
 372                         unlock_page(io_ctl->pages[i]);
 373                         put_page(io_ctl->pages[i]);
 374                 }
 375         }
 376 }
 377
 378 static int io_ctl_prepare_pages(struct btrfs_io_ctl *io_ctl, struct inode *inode,
 379                                 int uptodate)
 380 {
 381         struct page *page;
 382         gfp_t mask = btrfs_alloc_write_mask(inode->i_mapping);
 383         int i;
 384
 385         for (i = 0; i < io_ctl->num_pages; i++) {
 386                 page = find_or_create_page(inode->i_mapping, i, mask);
 387                 if (!page) {
 388                         io_ctl_drop_pages(io_ctl);
 389                         return -ENOMEM;
 390                 }
 391                 io_ctl->pages[i] = page;
 392                 if (uptodate && !PageUptodate(page)) {
 393                         btrfs_readpage(NULL, page);
 394                         lock_page(page);
 395                         if (!PageUptodate(page)) {
 396                                 btrfs_err(BTRFS_I(inode)->root->fs_info,
 397                                            "error reading free space cache");
 398                                 io_ctl_drop_pages(io_ctl);
 399                                 return -EIO;
 400                         }
 401                 }
 402         }
 403
 404         for (i = 0; i < io_ctl->num_pages; i++) {
 405                 clear_page_dirty_for_io(io_ctl->pages[i]);
 406                 set_page_extent_mapped(io_ctl->pages[i]);
 407         }
 408
 409         return 0;
 410 }
 411
 412 static void io_ctl_set_generation(struct btrfs_io_ctl *io_ctl, u64 generation)
 413 {
 414         __le64 *val;
 415
 416         io_ctl_map_page(io_ctl, 1);
 417
 418         /*
 419          * Skip the csum areas.  If we don't check crcs then we just have a
 420          * 64bit chunk at the front of the first page.
 421          */
 422         if (io_ctl->check_crcs) {
 423                 io_ctl->cur += (sizeof(u32) * io_ctl->num_pages);
 424                 io_ctl->size -= sizeof(u64) + (sizeof(u32) * io_ctl->num_pages);
 425         } else {
 426                 io_ctl->cur += sizeof(u64);
 427                 io_ctl->size -= sizeof(u64) * 2;
 428         }
 429
 430         val = io_ctl->cur;
 431         *val = cpu_to_le64(generation);
 432         io_ctl->cur += sizeof(u64);
 433 }
 434
 435 static int io_ctl_check_generation(struct btrfs_io_ctl *io_ctl, u64 generation)
 436 {
 437         __le64 *gen;
 438
 439         /*
 440          * Skip the crc area.  If we don't check crcs then we just have a 64bit
 441          * chunk at the front of the first page.
 442          */
 443         if (io_ctl->check_crcs) {
 444                 io_ctl->cur += sizeof(u32) * io_ctl->num_pages;
 445                 io_ctl->size -= sizeof(u64) +
 446                         (sizeof(u32) * io_ctl->num_pages);
 447         } else {
 448                 io_ctl->cur += sizeof(u64);
 449                 io_ctl->size -= sizeof(u64) * 2;
 450         }
 451
 452         gen = io_ctl->cur;
 453         if (le64_to_cpu(*gen) != generation) {
 454                 btrfs_err_rl(io_ctl->fs_info,
 455                         "space cache generation (%llu) does not match inode (%llu)",
 456                                 *gen, generation);
 457                 io_ctl_unmap_page(io_ctl);
 458                 return -EIO;
 459         }
 460         io_ctl->cur += sizeof(u64);
 461         return 0;
 462 }
 463
 464 static void io_ctl_set_crc(struct btrfs_io_ctl *io_ctl, int index)
 465 {
 466         u32 *tmp;
 467         u32 crc = ~(u32)0;
 468         unsigned offset = 0;
 469
 470         if (!io_ctl->check_crcs) {
 471                 io_ctl_unmap_page(io_ctl);
 472                 return;
 473         }
 474
 475         if (index == 0)
 476                 offset = sizeof(u32) * io_ctl->num_pages;
 477
 478         crc = btrfs_csum_data(io_ctl->orig + offset, crc,
 479                               PAGE_SIZE - offset);
 480         btrfs_csum_final(crc, (u8 *)&crc);
 481         io_ctl_unmap_page(io_ctl);
 482         tmp = page_address(io_ctl->pages[0]);
 483         tmp += index;
 484         *tmp = crc;
 485 }
 486
 487 static int io_ctl_check_crc(struct btrfs_io_ctl *io_ctl, int index)
 488 {
 489         u32 *tmp, val;
 490         u32 crc = ~(u32)0;
 491         unsigned offset = 0;
 492
 493         if (!io_ctl->check_crcs) {
 494                 io_ctl_map_page(io_ctl, 0);
 495                 return 0;
 496         }
 497
 498         if (index == 0)
 499                 offset = sizeof(u32) * io_ctl->num_pages;
 500
 501         tmp = page_address(io_ctl->pages[0]);
 502         tmp += index;
 503         val = *tmp;
 504
 505         io_ctl_map_page(io_ctl, 0);
 506         crc = btrfs_csum_data(io_ctl->orig + offset, crc,
 507                               PAGE_SIZE - offset);
 508         btrfs_csum_final(crc, (u8 *)&crc);
 509         if (val != crc) {
 510                 btrfs_err_rl(io_ctl->fs_info,
 511                         "csum mismatch on free space cache");
 512                 io_ctl_unmap_page(io_ctl);
 513                 return -EIO;
 514         }
 515
 516         return 0;
 517 }
 518
 519 static int io_ctl_add_entry(struct btrfs_io_ctl *io_ctl, u64 offset, u64 bytes,
 520                             void *bitmap)
 521 {
 522         struct btrfs_free_space_entry *entry;
 523
 524         if (!io_ctl->cur)
 525                 return -ENOSPC;
 526
 527         entry = io_ctl->cur;
 528         entry->offset = cpu_to_le64(offset);
 529         entry->bytes = cpu_to_le64(bytes);
 530         entry->type = (bitmap) ? BTRFS_FREE_SPACE_BITMAP :
 531                 BTRFS_FREE_SPACE_EXTENT;
 532         io_ctl->cur += sizeof(struct btrfs_free_space_entry);
 533         io_ctl->size -= sizeof(struct btrfs_free_space_entry);
 534
 535         if (io_ctl->size >= sizeof(struct btrfs_free_space_entry))
 536                 return 0;
 537
 538         io_ctl_set_crc(io_ctl, io_ctl->index - 1);
 539
 540         /* No more pages to map */
 541         if (io_ctl->index >= io_ctl->num_pages)
 542                 return 0;
 543
 544         /* map the next page */
 545         io_ctl_map_page(io_ctl, 1);
 546         return 0;
 547 }
 548
 549 static int io_ctl_add_bitmap(struct btrfs_io_ctl *io_ctl, void *bitmap)
 550 {
 551         if (!io_ctl->cur)
 552                 return -ENOSPC;
 553
 554         /*
 555          * If we aren't at the start of the current page, unmap this one and
 556          * map the next one if there is any left.
 557          */
 558         if (io_ctl->cur != io_ctl->orig) {
 559                 io_ctl_set_crc(io_ctl, io_ctl->index - 1);
 560                 if (io_ctl->index >= io_ctl->num_pages)
 561                         return -ENOSPC;
 562                 io_ctl_map_page(io_ctl, 0);
 563         }
 564
 565         memcpy(io_ctl->cur, bitmap, PAGE_SIZE);
 566         io_ctl_set_crc(io_ctl, io_ctl->index - 1);
 567         if (io_ctl->index < io_ctl->num_pages)
 568                 io_ctl_map_page(io_ctl, 0);
 569         return 0;
 570 }
 571
 572 static void io_ctl_zero_remaining_pages(struct btrfs_io_ctl *io_ctl)
 573 {
 574         /*
 575          * If we're not on the boundary we know we've modified the page and we
 576          * need to crc the page.
 577          */
 578         if (io_ctl->cur != io_ctl->orig)
 579                 io_ctl_set_crc(io_ctl, io_ctl->index - 1);
 580         else
 581                 io_ctl_unmap_page(io_ctl);
 582
 583         while (io_ctl->index < io_ctl->num_pages) {
 584                 io_ctl_map_page(io_ctl, 1);
 585                 io_ctl_set_crc(io_ctl, io_ctl->index - 1);
 586         }
 587 }
 588
 589 static int io_ctl_read_entry(struct btrfs_io_ctl *io_ctl,
 590                             struct btrfs_free_space *entry, u8 *type)
 591 {
 592         struct btrfs_free_space_entry *e;
 593         int ret;
 594
 595         if (!io_ctl->cur) {
 596                 ret = io_ctl_check_crc(io_ctl, io_ctl->index);
 597                 if (ret)
 598                         return ret;
 599         }
 600
 601         e = io_ctl->cur;
 602         entry->offset = le64_to_cpu(e->offset);
 603         entry->bytes = le64_to_cpu(e->bytes);
 604         *type = e->type;
 605         io_ctl->cur += sizeof(struct btrfs_free_space_entry);
 606         io_ctl->size -= sizeof(struct btrfs_free_space_entry);
 607
 608         if (io_ctl->size >= sizeof(struct btrfs_free_space_entry))
 609                 return 0;
 610
 611         io_ctl_unmap_page(io_ctl);
 612
 613         return 0;
 614 }
 615
 616 static int io_ctl_read_bitmap(struct btrfs_io_ctl *io_ctl,
 617                               struct btrfs_free_space *entry)
 618 {
 619         int ret;
 620
 621         ret = io_ctl_check_crc(io_ctl, io_ctl->index);
 622         if (ret)
 623                 return ret;
 624
 625         memcpy(entry->bitmap, io_ctl->cur, PAGE_SIZE);
 626         io_ctl_unmap_page(io_ctl);
 627
 628         return 0;
 629 }
 630
 631 /*
 632  * Since we attach pinned extents after the fact we can have contiguous sections
 633  * of free space that are split up in entries.  This poses a problem with the
 634  * tree logging stuff since it could have allocated across what appears to be 2
 635  * entries since we would have merged the entries when adding the pinned extents
 636  * back to the free space cache.  So run through the space cache that we just
 637  * loaded and merge contiguous entries.  This will make the log replay stuff not
 638  * blow up and it will make for nicer allocator behavior.
 639  */
 640 static void merge_space_tree(struct btrfs_free_space_ctl *ctl)
 641 {
 642         struct btrfs_free_space *e, *prev = NULL;
 643         struct rb_node *n;
 644
 645 again:
 646         spin_lock(&ctl->tree_lock);
 647         for (n = rb_first(&ctl->free_space_offset); n; n = rb_next(n)) {
 648                 e = rb_entry(n, struct btrfs_free_space, offset_index);
 649                 if (!prev)
 650                         goto next;
 651                 if (e->bitmap || prev->bitmap)
 652                         goto next;
 653                 if (prev->offset + prev->bytes == e->offset) {
 654                         unlink_free_space(ctl, prev);
 655                         unlink_free_space(ctl, e);
 656                         prev->bytes += e->bytes;
 657                         kmem_cache_free(btrfs_free_space_cachep, e);
 658                         link_free_space(ctl, prev);
 659                         prev = NULL;
 660                         spin_unlock(&ctl->tree_lock);
 661                         goto again;
 662                 }
 663 next:
 664                 prev = e;
 665         }
 666         spin_unlock(&ctl->tree_lock);
 667 }
 668
 669 static int __load_free_space_cache(struct btrfs_root *root, struct inode *inode,
 670                                    struct btrfs_free_space_ctl *ctl,
 671                                    struct btrfs_path *path, u64 offset)
 672 {
 673         struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
 674         struct btrfs_free_space_header *header;
 675         struct extent_buffer *leaf;
 676         struct btrfs_io_ctl io_ctl;
 677         struct btrfs_key key;
 678         struct btrfs_free_space *e, *n;
 679         LIST_HEAD(bitmaps);
 680         u64 num_entries;
 681         u64 num_bitmaps;
 682         u64 generation;
 683         u8 type;
 684         int ret = 0;
 685
 686         /* Nothing in the space cache, goodbye */
 687         if (!i_size_read(inode))
 688                 return 0;
 689
 690         key.objectid = BTRFS_FREE_SPACE_OBJECTID;
 691         key.offset = offset;
 692         key.type = 0;
 693
 694         ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
 695         if (ret < 0)
 696                 return 0;
 697         else if (ret > 0) {
 698                 btrfs_release_path(path);
 699                 return 0;
 700         }
 701
 702         ret = -1;
 703
 704         leaf = path->nodes[0];
 705         header = btrfs_item_ptr(leaf, path->slots[0],
 706                                 struct btrfs_free_space_header);
 707         num_entries = btrfs_free_space_entries(leaf, header);
 708         num_bitmaps = btrfs_free_space_bitmaps(leaf, header);
 709         generation = btrfs_free_space_generation(leaf, header);
 710         btrfs_release_path(path);
 711
 712         if (!BTRFS_I(inode)->generation) {
 713                 btrfs_info(fs_info,
 714                            "The free space cache file (%llu) is invalid. skip it\n",
 715                            offset);
 716                 return 0;
 717         }
 718
 719         if (BTRFS_I(inode)->generation != generation) {
 720                 btrfs_err(fs_info,
 721                           "free space inode generation (%llu) did not match free space cache generation (%llu)",
 722                           BTRFS_I(inode)->generation, generation);
 723                 return 0;
 724         }
 725
 726         if (!num_entries)
 727                 return 0;
 728
 729         ret = io_ctl_init(&io_ctl, inode, 0);
 730         if (ret)
 731                 return ret;
 732
 733         ret = readahead_cache(inode);
 734         if (ret)
 735                 goto out;
 736
 737         ret = io_ctl_prepare_pages(&io_ctl, inode, 1);
 738         if (ret)
 739                 goto out;
 740
 741         ret = io_ctl_check_crc(&io_ctl, 0);
 742         if (ret)
 743                 goto free_cache;
 744
 745         ret = io_ctl_check_generation(&io_ctl, generation);
 746         if (ret)
 747                 goto free_cache;
 748
 749         while (num_entries) {
 750                 e = kmem_cache_zalloc(btrfs_free_space_cachep,
 751                                       GFP_NOFS);
 752                 if (!e)
 753                         goto free_cache;
 754
 755                 ret = io_ctl_read_entry(&io_ctl, e, &type);
 756                 if (ret) {
 757                         kmem_cache_free(btrfs_free_space_cachep, e);
 758                         goto free_cache;
 759                 }
 760
 761                 if (!e->bytes) {
 762                         kmem_cache_free(btrfs_free_space_cachep, e);
 763                         goto free_cache;
 764                 }
 765
 766                 if (type == BTRFS_FREE_SPACE_EXTENT) {
 767                         spin_lock(&ctl->tree_lock);
 768                         ret = link_free_space(ctl, e);
 769                         spin_unlock(&ctl->tree_lock);
 770                         if (ret) {
 771                                 btrfs_err(fs_info,
 772                                         "Duplicate entries in free space cache, dumping");
 773                                 kmem_cache_free(btrfs_free_space_cachep, e);
 774                                 goto free_cache;
 775                         }
 776                 } else {
 777                         ASSERT(num_bitmaps);
 778                         num_bitmaps--;
 779                         e->bitmap = kzalloc(PAGE_SIZE, GFP_NOFS);
 780                         if (!e->bitmap) {
 781                                 kmem_cache_free(
 782                                         btrfs_free_space_cachep, e);
 783                                 goto free_cache;
 784                         }
 785                         spin_lock(&ctl->tree_lock);
 786                         ret = link_free_space(ctl, e);
 787                         ctl->total_bitmaps++;
 788                         ctl->op->recalc_thresholds(ctl);
 789                         spin_unlock(&ctl->tree_lock);
 790                         if (ret) {
 791                                 btrfs_err(fs_info,
 792                                         "Duplicate entries in free space cache, dumping");
 793                                 kmem_cache_free(btrfs_free_space_cachep, e);
 794                                 goto free_cache;
 795                         }
 796                         list_add_tail(&e->list, &bitmaps);
 797                 }
 798
 799                 num_entries--;
 800         }
 801
 802         io_ctl_unmap_page(&io_ctl);
 803
 804         /*
 805          * We add the bitmaps at the end of the entries in order that
 806          * the bitmap entries are added to the cache.
 807          */
 808         list_for_each_entry_safe(e, n, &bitmaps, list) {
 809                 list_del_init(&e->list);
 810                 ret = io_ctl_read_bitmap(&io_ctl, e);
 811                 if (ret)
 812                         goto free_cache;
 813         }
 814
 815         io_ctl_drop_pages(&io_ctl);
 816         merge_space_tree(ctl);
 817         ret = 1;
 818 out:
 819         io_ctl_free(&io_ctl);
 820         return ret;
 821 free_cache:
 822         io_ctl_drop_pages(&io_ctl);
 823         __btrfs_remove_free_space_cache(ctl);
 824         goto out;
 825 }
 826
 827 int load_free_space_cache(struct btrfs_fs_info *fs_info,
 828                           struct btrfs_block_group_cache *block_group)
 829 {
 830         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
 831         struct btrfs_root *root = fs_info->tree_root;
 832         struct inode *inode;
 833         struct btrfs_path *path;
 834         int ret = 0;
 835         bool matched;
 836         u64 used = btrfs_block_group_used(&block_group->item);
 837
 838         /*
 839          * If this block group has been marked to be cleared for one reason or
 840          * another then we can't trust the on disk cache, so just return.
 841          */
 842         spin_lock(&block_group->lock);
 843         if (block_group->disk_cache_state != BTRFS_DC_WRITTEN) {
 844                 spin_unlock(&block_group->lock);
 845                 return 0;
 846         }
 847         spin_unlock(&block_group->lock);
 848
 849         path = btrfs_alloc_path();
 850         if (!path)
 851                 return 0;
 852         path->search_commit_root = 1;
 853         path->skip_locking = 1;
 854
 855         inode = lookup_free_space_inode(root, block_group, path);
 856         if (IS_ERR(inode)) {
 857                 btrfs_free_path(path);
 858                 return 0;
 859         }
 860
 861         /* We may have converted the inode and made the cache invalid. */
 862         spin_lock(&block_group->lock);
 863         if (block_group->disk_cache_state != BTRFS_DC_WRITTEN) {
 864                 spin_unlock(&block_group->lock);
 865                 btrfs_free_path(path);
 866                 goto out;
 867         }
 868         spin_unlock(&block_group->lock);
 869
 870         ret = __load_free_space_cache(fs_info->tree_root, inode, ctl,
 871                                       path, block_group->key.objectid);
 872         btrfs_free_path(path);
 873         if (ret <= 0)
 874                 goto out;
 875
 876         spin_lock(&ctl->tree_lock);
 877         matched = (ctl->free_space == (block_group->key.offset - used -
 878                                        block_group->bytes_super));
 879         spin_unlock(&ctl->tree_lock);
 880
 881         if (!matched) {
 882                 __btrfs_remove_free_space_cache(ctl);
 883                 btrfs_warn(fs_info,
 884                            "block group %llu has wrong amount of free space",
 885                            block_group->key.objectid);
 886                 ret = -1;
 887         }
 888 out:
 889         if (ret < 0) {
 890                 /* This cache is bogus, make sure it gets cleared */
 891                 spin_lock(&block_group->lock);
 892                 block_group->disk_cache_state = BTRFS_DC_CLEAR;
 893                 spin_unlock(&block_group->lock);
 894                 ret = 0;
 895
 896                 btrfs_warn(fs_info,
 897                            "failed to load free space cache for block group %llu, rebuilding it now",
 898                            block_group->key.objectid);
 899         }
 900
 901         iput(inode);
 902         return ret;
 903 }
 904
 905 static noinline_for_stack
 906 int write_cache_extent_entries(struct btrfs_io_ctl *io_ctl,
 907                               struct btrfs_free_space_ctl *ctl,
 908                               struct btrfs_block_group_cache *block_group,
 909                               int *entries, int *bitmaps,
 910                               struct list_head *bitmap_list)
 911 {
 912         int ret;
 913         struct btrfs_free_cluster *cluster = NULL;
 914         struct btrfs_free_cluster *cluster_locked = NULL;
 915         struct rb_node *node = rb_first(&ctl->free_space_offset);
 916         struct btrfs_trim_range *trim_entry;
 917
 918         /* Get the cluster for this block_group if it exists */
 919         if (block_group && !list_empty(&block_group->cluster_list)) {
 920                 cluster = list_entry(block_group->cluster_list.next,
 921                                      struct btrfs_free_cluster,
 922                                      block_group_list);
 923         }
 924
 925         if (!node && cluster) {
 926                 cluster_locked = cluster;
 927                 spin_lock(&cluster_locked->lock);
 928                 node = rb_first(&cluster->root);
 929                 cluster = NULL;
 930         }
 931
 932         /* Write out the extent entries */
 933         while (node) {
 934                 struct btrfs_free_space *e;
 935
 936                 e = rb_entry(node, struct btrfs_free_space, offset_index);
 937                 *entries += 1;
 938
 939                 ret = io_ctl_add_entry(io_ctl, e->offset, e->bytes,
 940                                        e->bitmap);
 941                 if (ret)
 942                         goto fail;
 943
 944                 if (e->bitmap) {
 945                         list_add_tail(&e->list, bitmap_list);
 946                         *bitmaps += 1;
 947                 }
 948                 node = rb_next(node);
 949                 if (!node && cluster) {
 950                         node = rb_first(&cluster->root);
 951                         cluster_locked = cluster;
 952                         spin_lock(&cluster_locked->lock);
 953                         cluster = NULL;
 954                 }
 955         }
 956         if (cluster_locked) {
 957                 spin_unlock(&cluster_locked->lock);
 958                 cluster_locked = NULL;
 959         }
 960
 961         /*
 962          * Make sure we don't miss any range that was removed from our rbtree
 963          * because trimming is running. Otherwise after a umount+mount (or crash
 964          * after committing the transaction) we would leak free space and get
 965          * an inconsistent free space cache report from fsck.
 966          */
 967         list_for_each_entry(trim_entry, &ctl->trimming_ranges, list) {
 968                 ret = io_ctl_add_entry(io_ctl, trim_entry->start,
 969                                        trim_entry->bytes, NULL);
 970                 if (ret)
 971                         goto fail;
 972                 *entries += 1;
 973         }
 974
 975         return 0;
 976 fail:
 977         if (cluster_locked)
 978                 spin_unlock(&cluster_locked->lock);
 979         return -ENOSPC;
 980 }
 981
 982 static noinline_for_stack int
 983 update_cache_item(struct btrfs_trans_handle *trans,
 984                   struct btrfs_root *root,
 985                   struct inode *inode,
 986                   struct btrfs_path *path, u64 offset,
 987                   int entries, int bitmaps)
 988 {
 989         struct btrfs_key key;
 990         struct btrfs_free_space_header *header;
 991         struct extent_buffer *leaf;
 992         int ret;
 993
 994         key.objectid = BTRFS_FREE_SPACE_OBJECTID;
 995         key.offset = offset;
 996         key.type = 0;
 997
 998         ret = btrfs_search_slot(trans, root, &key, path, 0, 1);
 999         if (ret < 0) {
1000                 clear_extent_bit(&BTRFS_I(inode)->io_tree, 0, inode->i_size - 1,
1001                                  EXTENT_DIRTY | EXTENT_DELALLOC, 0, 0, NULL,
1002                                  GFP_NOFS);
1003                 goto fail;
1004         }
1005         leaf = path->nodes[0];
1006         if (ret > 0) {
1007                 struct btrfs_key found_key;
1008                 ASSERT(path->slots[0]);
1009                 path->slots[0]--;
1010                 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
1011                 if (found_key.objectid != BTRFS_FREE_SPACE_OBJECTID ||
1012                     found_key.offset != offset) {
1013                         clear_extent_bit(&BTRFS_I(inode)->io_tree, 0,
1014                                          inode->i_size - 1,
1015                                          EXTENT_DIRTY | EXTENT_DELALLOC, 0, 0,
1016                                          NULL, GFP_NOFS);
1017                         btrfs_release_path(path);
1018                         goto fail;
1019                 }
1020         }
1021
1022         BTRFS_I(inode)->generation = trans->transid;
1023         header = btrfs_item_ptr(leaf, path->slots[0],
1024                                 struct btrfs_free_space_header);
1025         btrfs_set_free_space_entries(leaf, header, entries);
1026         btrfs_set_free_space_bitmaps(leaf, header, bitmaps);
1027         btrfs_set_free_space_generation(leaf, header, trans->transid);
1028         btrfs_mark_buffer_dirty(leaf);
1029         btrfs_release_path(path);
1030
1031         return 0;
1032
1033 fail:
1034         return -1;
1035 }
1036
1037 static noinline_for_stack int
1038 write_pinned_extent_entries(struct btrfs_fs_info *fs_info,
1039                             struct btrfs_block_group_cache *block_group,
1040                             struct btrfs_io_ctl *io_ctl,
1041                             int *entries)
1042 {
1043         u64 start, extent_start, extent_end, len;
1044         struct extent_io_tree *unpin = NULL;
1045         int ret;
1046
1047         if (!block_group)
1048                 return 0;
1049
1050         /*
1051          * We want to add any pinned extents to our free space cache
1052          * so we don't leak the space
1053          *
1054          * We shouldn't have switched the pinned extents yet so this is the
1055          * right one
1056          */
1057         unpin = fs_info->pinned_extents;
1058
1059         start = block_group->key.objectid;
1060
1061         while (start < block_group->key.objectid + block_group->key.offset) {
1062                 ret = find_first_extent_bit(unpin, start,
1063                                             &extent_start, &extent_end,
1064                                             EXTENT_DIRTY, NULL);
1065                 if (ret)
1066                         return 0;
1067
1068                 /* This pinned extent is out of our range */
1069                 if (extent_start >= block_group->key.objectid +
1070                     block_group->key.offset)
1071                         return 0;
1072
1073                 extent_start = max(extent_start, start);
1074                 extent_end = min(block_group->key.objectid +
1075                                  block_group->key.offset, extent_end + 1);
1076                 len = extent_end - extent_start;
1077
1078                 *entries += 1;
1079                 ret = io_ctl_add_entry(io_ctl, extent_start, len, NULL);
1080                 if (ret)
1081                         return -ENOSPC;
1082
1083                 start = extent_end;
1084         }
1085
1086         return 0;
1087 }
1088
1089 static noinline_for_stack int
1090 write_bitmap_entries(struct btrfs_io_ctl *io_ctl, struct list_head *bitmap_list)
1091 {
1092         struct btrfs_free_space *entry, *next;
1093         int ret;
1094
1095         /* Write out the bitmaps */
1096         list_for_each_entry_safe(entry, next, bitmap_list, list) {
1097                 ret = io_ctl_add_bitmap(io_ctl, entry->bitmap);
1098                 if (ret)
1099                         return -ENOSPC;
1100                 list_del_init(&entry->list);
1101         }
1102
1103         return 0;
1104 }
1105
1106 static int flush_dirty_cache(struct inode *inode)
1107 {
1108         int ret;
1109
1110         ret = btrfs_wait_ordered_range(inode, 0, (u64)-1);
1111         if (ret)
1112                 clear_extent_bit(&BTRFS_I(inode)->io_tree, 0, inode->i_size - 1,
1113                                  EXTENT_DIRTY | EXTENT_DELALLOC, 0, 0, NULL,
1114                                  GFP_NOFS);
1115
1116         return ret;
1117 }
1118
1119 static void noinline_for_stack
1120 cleanup_bitmap_list(struct list_head *bitmap_list)
1121 {
1122         struct btrfs_free_space *entry, *next;
1123
1124         list_for_each_entry_safe(entry, next, bitmap_list, list)
1125                 list_del_init(&entry->list);
1126 }
1127
1128 static void noinline_for_stack
1129 cleanup_write_cache_enospc(struct inode *inode,
1130                            struct btrfs_io_ctl *io_ctl,
1131                            struct extent_state **cached_state,
1132                            struct list_head *bitmap_list)
1133 {
1134         io_ctl_drop_pages(io_ctl);
1135         unlock_extent_cached(&BTRFS_I(inode)->io_tree, 0,
1136                              i_size_read(inode) - 1, cached_state,
1137                              GFP_NOFS);
1138 }
1139
1140 static int __btrfs_wait_cache_io(struct btrfs_root *root,
1141                                  struct btrfs_trans_handle *trans,
1142                                  struct btrfs_block_group_cache *block_group,
1143                                  struct btrfs_io_ctl *io_ctl,
1144                                  struct btrfs_path *path, u64 offset)
1145 {
1146         int ret;
1147         struct inode *inode = io_ctl->inode;
1148         struct btrfs_fs_info *fs_info;
1149
1150         if (!inode)
1151                 return 0;
1152
1153         fs_info = btrfs_sb(inode->i_sb);
1154
1155         /* Flush the dirty pages in the cache file. */
1156         ret = flush_dirty_cache(inode);
1157         if (ret)
1158                 goto out;
1159
1160         /* Update the cache item to tell everyone this cache file is valid. */
1161         ret = update_cache_item(trans, root, inode, path, offset,
1162                                 io_ctl->entries, io_ctl->bitmaps);
1163 out:
1164         io_ctl_free(io_ctl);
1165         if (ret) {
1166                 invalidate_inode_pages2(inode->i_mapping);
1167                 BTRFS_I(inode)->generation = 0;
1168                 if (block_group) {
1169 #ifdef DEBUG
1170                         btrfs_err(fs_info,
1171                                   "failed to write free space cache for block group %llu",
1172                                   block_group->key.objectid);
1173 #endif
1174                 }
1175         }
1176         btrfs_update_inode(trans, root, inode);
1177
1178         if (block_group) {
1179                 /* the dirty list is protected by the dirty_bgs_lock */
1180                 spin_lock(&trans->transaction->dirty_bgs_lock);
1181
1182                 /* the disk_cache_state is protected by the block group lock */
1183                 spin_lock(&block_group->lock);
1184
1185                 /*
1186                  * only mark this as written if we didn't get put back on
1187                  * the dirty list while waiting for IO.   Otherwise our
1188                  * cache state won't be right, and we won't get written again
1189                  */
1190                 if (!ret && list_empty(&block_group->dirty_list))
1191                         block_group->disk_cache_state = BTRFS_DC_WRITTEN;
1192                 else if (ret)
1193                         block_group->disk_cache_state = BTRFS_DC_ERROR;
1194
1195                 spin_unlock(&block_group->lock);
1196                 spin_unlock(&trans->transaction->dirty_bgs_lock);
1197                 io_ctl->inode = NULL;
1198                 iput(inode);
1199         }
1200
1201         return ret;
1202
1203 }
1204
1205 static int btrfs_wait_cache_io_root(struct btrfs_root *root,
1206                                     struct btrfs_trans_handle *trans,
1207                                     struct btrfs_io_ctl *io_ctl,
1208                                     struct btrfs_path *path)
1209 {
1210         return __btrfs_wait_cache_io(root, trans, NULL, io_ctl, path, 0);
1211 }
1212
1213 int btrfs_wait_cache_io(struct btrfs_trans_handle *trans,
1214                         struct btrfs_block_group_cache *block_group,
1215                         struct btrfs_path *path)
1216 {
1217         return __btrfs_wait_cache_io(block_group->fs_info->tree_root, trans,
1218                                      block_group, &block_group->io_ctl,
1219                                      path, block_group->key.objectid);
1220 }
1221
1222 /**
1223  * __btrfs_write_out_cache - write out cached info to an inode
1224  * @root - the root the inode belongs to
1225  * @ctl - the free space cache we are going to write out
1226  * @block_group - the block_group for this cache if it belongs to a block_group
1227  * @trans - the trans handle
1228  * @path - the path to use
1229  * @offset - the offset for the key we'll insert
1230  *
1231  * This function writes out a free space cache struct to disk for quick recovery
1232  * on mount.  This will return 0 if it was successful in writing the cache out,
1233  * or an errno if it was not.
1234  */
1235 static int __btrfs_write_out_cache(struct btrfs_root *root, struct inode *inode,
1236                                    struct btrfs_free_space_ctl *ctl,
1237                                    struct btrfs_block_group_cache *block_group,
1238                                    struct btrfs_io_ctl *io_ctl,
1239                                    struct btrfs_trans_handle *trans,
1240                                    struct btrfs_path *path, u64 offset)
1241 {
1242         struct btrfs_fs_info *fs_info = root->fs_info;
1243         struct extent_state *cached_state = NULL;
1244         LIST_HEAD(bitmap_list);
1245         int entries = 0;
1246         int bitmaps = 0;
1247         int ret;
1248         int must_iput = 0;
1249
1250         if (!i_size_read(inode))
1251                 return -EIO;
1252
1253         WARN_ON(io_ctl->pages);
1254         ret = io_ctl_init(io_ctl, inode, 1);
1255         if (ret)
1256                 return ret;
1257
1258         if (block_group && (block_group->flags & BTRFS_BLOCK_GROUP_DATA)) {
1259                 down_write(&block_group->data_rwsem);
1260                 spin_lock(&block_group->lock);
1261                 if (block_group->delalloc_bytes) {
1262                         block_group->disk_cache_state = BTRFS_DC_WRITTEN;
1263                         spin_unlock(&block_group->lock);
1264                         up_write(&block_group->data_rwsem);
1265                         BTRFS_I(inode)->generation = 0;
1266                         ret = 0;
1267                         must_iput = 1;
1268                         goto out;
1269                 }
1270                 spin_unlock(&block_group->lock);
1271         }
1272
1273         /* Lock all pages first so we can lock the extent safely. */
1274         ret = io_ctl_prepare_pages(io_ctl, inode, 0);
1275         if (ret)
1276                 goto out;
1277
1278         lock_extent_bits(&BTRFS_I(inode)->io_tree, 0, i_size_read(inode) - 1,
1279                          &cached_state);
1280
1281         io_ctl_set_generation(io_ctl, trans->transid);
1282
1283         mutex_lock(&ctl->cache_writeout_mutex);
1284         /* Write out the extent entries in the free space cache */
1285         spin_lock(&ctl->tree_lock);
1286         ret = write_cache_extent_entries(io_ctl, ctl,
1287                                          block_group, &entries, &bitmaps,
1288                                          &bitmap_list);
1289         if (ret)
1290                 goto out_nospc_locked;
1291
1292         /*
1293          * Some spaces that are freed in the current transaction are pinned,
1294          * they will be added into free space cache after the transaction is
1295          * committed, we shouldn't lose them.
1296          *
1297          * If this changes while we are working we'll get added back to
1298          * the dirty list and redo it.  No locking needed
1299          */
1300         ret = write_pinned_extent_entries(fs_info, block_group,
1301                                           io_ctl, &entries);
1302         if (ret)
1303                 goto out_nospc_locked;
1304
1305         /*
1306          * At last, we write out all the bitmaps and keep cache_writeout_mutex
1307          * locked while doing it because a concurrent trim can be manipulating
1308          * or freeing the bitmap.
1309          */
1310         ret = write_bitmap_entries(io_ctl, &bitmap_list);
1311         spin_unlock(&ctl->tree_lock);
1312         mutex_unlock(&ctl->cache_writeout_mutex);
1313         if (ret)
1314                 goto out_nospc;
1315
1316         /* Zero out the rest of the pages just to make sure */
1317         io_ctl_zero_remaining_pages(io_ctl);
1318
1319         /* Everything is written out, now we dirty the pages in the file. */
1320         ret = btrfs_dirty_pages(inode, io_ctl->pages, io_ctl->num_pages, 0,
1321                                 i_size_read(inode), &cached_state);
1322         if (ret)
1323                 goto out_nospc;
1324
1325         if (block_group && (block_group->flags & BTRFS_BLOCK_GROUP_DATA))
1326                 up_write(&block_group->data_rwsem);
1327         /*
1328          * Release the pages and unlock the extent, we will flush
1329          * them out later
1330          */
1331         io_ctl_drop_pages(io_ctl);
1332
1333         unlock_extent_cached(&BTRFS_I(inode)->io_tree, 0,
1334                              i_size_read(inode) - 1, &cached_state, GFP_NOFS);
1335
1336         /*
1337          * at this point the pages are under IO and we're happy,
1338          * The caller is responsible for waiting on them and updating the
1339          * the cache and the inode
1340          */
1341         io_ctl->entries = entries;
1342         io_ctl->bitmaps = bitmaps;
1343
1344         ret = btrfs_fdatawrite_range(inode, 0, (u64)-1);
1345         if (ret)
1346                 goto out;
1347
1348         return 0;
1349
1350 out:
1351         io_ctl->inode = NULL;
1352         io_ctl_free(io_ctl);
1353         if (ret) {
1354                 invalidate_inode_pages2(inode->i_mapping);
1355                 BTRFS_I(inode)->generation = 0;
1356         }
1357         btrfs_update_inode(trans, root, inode);
1358         if (must_iput)
1359                 iput(inode);
1360         return ret;
1361
1362 out_nospc_locked:
1363         cleanup_bitmap_list(&bitmap_list);
1364         spin_unlock(&ctl->tree_lock);
1365         mutex_unlock(&ctl->cache_writeout_mutex);
1366
1367 out_nospc:
1368         cleanup_write_cache_enospc(inode, io_ctl, &cached_state, &bitmap_list);
1369
1370         if (block_group && (block_group->flags & BTRFS_BLOCK_GROUP_DATA))
1371                 up_write(&block_group->data_rwsem);
1372
1373         goto out;
1374 }
1375
1376 int btrfs_write_out_cache(struct btrfs_fs_info *fs_info,
1377                           struct btrfs_trans_handle *trans,
1378                           struct btrfs_block_group_cache *block_group,
1379                           struct btrfs_path *path)
1380 {
1381         struct btrfs_root *root = fs_info->tree_root;
1382         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
1383         struct inode *inode;
1384         int ret = 0;
1385
1386         spin_lock(&block_group->lock);
1387         if (block_group->disk_cache_state < BTRFS_DC_SETUP) {
1388                 spin_unlock(&block_group->lock);
1389                 return 0;
1390         }
1391         spin_unlock(&block_group->lock);
1392
1393         inode = lookup_free_space_inode(root, block_group, path);
1394         if (IS_ERR(inode))
1395                 return 0;
1396
1397         ret = __btrfs_write_out_cache(root, inode, ctl, block_group,
1398                                       &block_group->io_ctl, trans,
1399                                       path, block_group->key.objectid);
1400         if (ret) {
1401 #ifdef DEBUG
1402                 btrfs_err(fs_info,
1403                           "failed to write free space cache for block group %llu",
1404                           block_group->key.objectid);
1405 #endif
1406                 spin_lock(&block_group->lock);
1407                 block_group->disk_cache_state = BTRFS_DC_ERROR;
1408                 spin_unlock(&block_group->lock);
1409
1410                 block_group->io_ctl.inode = NULL;
1411                 iput(inode);
1412         }
1413
1414         /*
1415          * if ret == 0 the caller is expected to call btrfs_wait_cache_io
1416          * to wait for IO and put the inode
1417          */
1418
1419         return ret;
1420 }
1421
1422 static inline unsigned long offset_to_bit(u64 bitmap_start, u32 unit,
1423                                           u64 offset)
1424 {
1425         ASSERT(offset >= bitmap_start);
1426         offset -= bitmap_start;
1427         return (unsigned long)(div_u64(offset, unit));
1428 }
1429
1430 static inline unsigned long bytes_to_bits(u64 bytes, u32 unit)
1431 {
1432         return (unsigned long)(div_u64(bytes, unit));
1433 }
1434
1435 static inline u64 offset_to_bitmap(struct btrfs_free_space_ctl *ctl,
1436                                    u64 offset)
1437 {
1438         u64 bitmap_start;
1439         u64 bytes_per_bitmap;
1440
1441         bytes_per_bitmap = BITS_PER_BITMAP * ctl->unit;
1442         bitmap_start = offset - ctl->start;
1443         bitmap_start = div64_u64(bitmap_start, bytes_per_bitmap);
1444         bitmap_start *= bytes_per_bitmap;
1445         bitmap_start += ctl->start;
1446
1447         return bitmap_start;
1448 }
1449
1450 static int tree_insert_offset(struct rb_root *root, u64 offset,
1451                               struct rb_node *node, int bitmap)
1452 {
1453         struct rb_node **p = &root->rb_node;
1454         struct rb_node *parent = NULL;
1455         struct btrfs_free_space *info;
1456
1457         while (*p) {
1458                 parent = *p;
1459                 info = rb_entry(parent, struct btrfs_free_space, offset_index);
1460
1461                 if (offset < info->offset) {
1462                         p = &(*p)->rb_left;
1463                 } else if (offset > info->offset) {
1464                         p = &(*p)->rb_right;
1465                 } else {
1466                         /*
1467                          * we could have a bitmap entry and an extent entry
1468                          * share the same offset.  If this is the case, we want
1469                          * the extent entry to always be found first if we do a
1470                          * linear search through the tree, since we want to have
1471                          * the quickest allocation time, and allocating from an
1472                          * extent is faster than allocating from a bitmap.  So
1473                          * if we're inserting a bitmap and we find an entry at
1474                          * this offset, we want to go right, or after this entry
1475                          * logically.  If we are inserting an extent and we've
1476                          * found a bitmap, we want to go left, or before
1477                          * logically.
1478                          */
1479                         if (bitmap) {
1480                                 if (info->bitmap) {
1481                                         WARN_ON_ONCE(1);
1482                                         return -EEXIST;
1483                                 }
1484                                 p = &(*p)->rb_right;
1485                         } else {
1486                                 if (!info->bitmap) {
1487                                         WARN_ON_ONCE(1);
1488                                         return -EEXIST;
1489                                 }
1490                                 p = &(*p)->rb_left;
1491                         }
1492                 }
1493         }
1494
1495         rb_link_node(node, parent, p);
1496         rb_insert_color(node, root);
1497
1498         return 0;
1499 }
1500
1501 /*
1502  * searches the tree for the given offset.
1503  *
1504  * fuzzy - If this is set, then we are trying to make an allocation, and we just
1505  * want a section that has at least bytes size and comes at or after the given
1506  * offset.
1507  */
1508 static struct btrfs_free_space *
1509 tree_search_offset(struct btrfs_free_space_ctl *ctl,
1510                    u64 offset, int bitmap_only, int fuzzy)
1511 {
1512         struct rb_node *n = ctl->free_space_offset.rb_node;
1513         struct btrfs_free_space *entry, *prev = NULL;
1514
1515         /* find entry that is closest to the 'offset' */
1516         while (1) {
1517                 if (!n) {
1518                         entry = NULL;
1519                         break;
1520                 }
1521
1522                 entry = rb_entry(n, struct btrfs_free_space, offset_index);
1523                 prev = entry;
1524
1525                 if (offset < entry->offset)
1526                         n = n->rb_left;
1527                 else if (offset > entry->offset)
1528                         n = n->rb_right;
1529                 else
1530                         break;
1531         }
1532
1533         if (bitmap_only) {
1534                 if (!entry)
1535                         return NULL;
1536                 if (entry->bitmap)
1537                         return entry;
1538
1539                 /*
1540                  * bitmap entry and extent entry may share same offset,
1541                  * in that case, bitmap entry comes after extent entry.
1542                  */
1543                 n = rb_next(n);
1544                 if (!n)
1545                         return NULL;
1546                 entry = rb_entry(n, struct btrfs_free_space, offset_index);
1547                 if (entry->offset != offset)
1548                         return NULL;
1549
1550                 WARN_ON(!entry->bitmap);
1551                 return entry;
1552         } else if (entry) {
1553                 if (entry->bitmap) {
1554                         /*
1555                          * if previous extent entry covers the offset,
1556                          * we should return it instead of the bitmap entry
1557                          */
1558                         n = rb_prev(&entry->offset_index);
1559                         if (n) {
1560                                 prev = rb_entry(n, struct btrfs_free_space,
1561                                                 offset_index);
1562                                 if (!prev->bitmap &&
1563                                     prev->offset + prev->bytes > offset)
1564                                         entry = prev;
1565                         }
1566                 }
1567                 return entry;
1568         }
1569
1570         if (!prev)
1571                 return NULL;
1572
1573         /* find last entry before the 'offset' */
1574         entry = prev;
1575         if (entry->offset > offset) {
1576                 n = rb_prev(&entry->offset_index);
1577                 if (n) {
1578                         entry = rb_entry(n, struct btrfs_free_space,
1579                                         offset_index);
1580                         ASSERT(entry->offset <= offset);
1581                 } else {
1582                         if (fuzzy)
1583                                 return entry;
1584                         else
1585                                 return NULL;
1586                 }
1587         }
1588
1589         if (entry->bitmap) {
1590                 n = rb_prev(&entry->offset_index);
1591                 if (n) {
1592                         prev = rb_entry(n, struct btrfs_free_space,
1593                                         offset_index);
1594                         if (!prev->bitmap &&
1595                             prev->offset + prev->bytes > offset)
1596                                 return prev;
1597                 }
1598                 if (entry->offset + BITS_PER_BITMAP * ctl->unit > offset)
1599                         return entry;
1600         } else if (entry->offset + entry->bytes > offset)
1601                 return entry;
1602
1603         if (!fuzzy)
1604                 return NULL;
1605
1606         while (1) {
1607                 if (entry->bitmap) {
1608                         if (entry->offset + BITS_PER_BITMAP *
1609                             ctl->unit > offset)
1610                                 break;
1611                 } else {
1612                         if (entry->offset + entry->bytes > offset)
1613                                 break;
1614                 }
1615
1616                 n = rb_next(&entry->offset_index);
1617                 if (!n)
1618                         return NULL;
1619                 entry = rb_entry(n, struct btrfs_free_space, offset_index);
1620         }
1621         return entry;
1622 }
1623
1624 static inline void
1625 __unlink_free_space(struct btrfs_free_space_ctl *ctl,
1626                     struct btrfs_free_space *info)
1627 {
1628         rb_erase(&info->offset_index, &ctl->free_space_offset);
1629         ctl->free_extents--;
1630 }
1631
1632 static void unlink_free_space(struct btrfs_free_space_ctl *ctl,
1633                               struct btrfs_free_space *info)
1634 {
1635         __unlink_free_space(ctl, info);
1636         ctl->free_space -= info->bytes;
1637 }
1638
1639 static int link_free_space(struct btrfs_free_space_ctl *ctl,
1640                            struct btrfs_free_space *info)
1641 {
1642         int ret = 0;
1643
1644         ASSERT(info->bytes || info->bitmap);
1645         ret = tree_insert_offset(&ctl->free_space_offset, info->offset,
1646                                  &info->offset_index, (info->bitmap != NULL));
1647         if (ret)
1648                 return ret;
1649
1650         ctl->free_space += info->bytes;
1651         ctl->free_extents++;
1652         return ret;
1653 }
1654
1655 static void recalculate_thresholds(struct btrfs_free_space_ctl *ctl)
1656 {
1657         struct btrfs_block_group_cache *block_group = ctl->private;
1658         u64 max_bytes;
1659         u64 bitmap_bytes;
1660         u64 extent_bytes;
1661         u64 size = block_group->key.offset;
1662         u64 bytes_per_bg = BITS_PER_BITMAP * ctl->unit;
1663         u64 max_bitmaps = div64_u64(size + bytes_per_bg - 1, bytes_per_bg);
1664
1665         max_bitmaps = max_t(u64, max_bitmaps, 1);
1666
1667         ASSERT(ctl->total_bitmaps <= max_bitmaps);
1668
1669         /*
1670          * The goal is to keep the total amount of memory used per 1gb of space
1671          * at or below 32k, so we need to adjust how much memory we allow to be
1672          * used by extent based free space tracking
1673          */
1674         if (size < SZ_1G)
1675                 max_bytes = MAX_CACHE_BYTES_PER_GIG;
1676         else
1677                 max_bytes = MAX_CACHE_BYTES_PER_GIG * div_u64(size, SZ_1G);
1678
1679         /*
1680          * we want to account for 1 more bitmap than what we have so we can make
1681          * sure we don't go over our overall goal of MAX_CACHE_BYTES_PER_GIG as
1682          * we add more bitmaps.
1683          */
1684         bitmap_bytes = (ctl->total_bitmaps + 1) * ctl->unit;
1685
1686         if (bitmap_bytes >= max_bytes) {
1687                 ctl->extents_thresh = 0;
1688                 return;
1689         }
1690
1691         /*
1692          * we want the extent entry threshold to always be at most 1/2 the max
1693          * bytes we can have, or whatever is less than that.
1694          */
1695         extent_bytes = max_bytes - bitmap_bytes;
1696         extent_bytes = min_t(u64, extent_bytes, max_bytes >> 1);
1697
1698         ctl->extents_thresh =
1699                 div_u64(extent_bytes, sizeof(struct btrfs_free_space));
1700 }
1701
1702 static inline void __bitmap_clear_bits(struct btrfs_free_space_ctl *ctl,
1703                                        struct btrfs_free_space *info,
1704                                        u64 offset, u64 bytes)
1705 {
1706         unsigned long start, count;
1707
1708         start = offset_to_bit(info->offset, ctl->unit, offset);
1709         count = bytes_to_bits(bytes, ctl->unit);
1710         ASSERT(start + count <= BITS_PER_BITMAP);
1711
1712         bitmap_clear(info->bitmap, start, count);
1713
1714         info->bytes -= bytes;
1715 }
1716
1717 static void bitmap_clear_bits(struct btrfs_free_space_ctl *ctl,
1718                               struct btrfs_free_space *info, u64 offset,
1719                               u64 bytes)
1720 {
1721         __bitmap_clear_bits(ctl, info, offset, bytes);
1722         ctl->free_space -= bytes;
1723 }
1724
1725 static void bitmap_set_bits(struct btrfs_free_space_ctl *ctl,
1726                             struct btrfs_free_space *info, u64 offset,
1727                             u64 bytes)
1728 {
1729         unsigned long start, count;
1730
1731         start = offset_to_bit(info->offset, ctl->unit, offset);
1732         count = bytes_to_bits(bytes, ctl->unit);
1733         ASSERT(start + count <= BITS_PER_BITMAP);
1734
1735         bitmap_set(info->bitmap, start, count);
1736
1737         info->bytes += bytes;
1738         ctl->free_space += bytes;
1739 }
1740
1741 /*
1742  * If we can not find suitable extent, we will use bytes to record
1743  * the size of the max extent.
1744  */
1745 static int search_bitmap(struct btrfs_free_space_ctl *ctl,
1746                          struct btrfs_free_space *bitmap_info, u64 *offset,
1747                          u64 *bytes, bool for_alloc)
1748 {
1749         unsigned long found_bits = 0;
1750         unsigned long max_bits = 0;
1751         unsigned long bits, i;
1752         unsigned long next_zero;
1753         unsigned long extent_bits;
1754
1755         /*
1756          * Skip searching the bitmap if we don't have a contiguous section that
1757          * is large enough for this allocation.
1758          */
1759         if (for_alloc &&
1760             bitmap_info->max_extent_size &&
1761             bitmap_info->max_extent_size < *bytes) {
1762                 *bytes = bitmap_info->max_extent_size;
1763                 return -1;
1764         }
1765
1766         i = offset_to_bit(bitmap_info->offset, ctl->unit,
1767                           max_t(u64, *offset, bitmap_info->offset));
1768         bits = bytes_to_bits(*bytes, ctl->unit);
1769
1770         for_each_set_bit_from(i, bitmap_info->bitmap, BITS_PER_BITMAP) {
1771                 if (for_alloc && bits == 1) {
1772                         found_bits = 1;
1773                         break;
1774                 }
1775                 next_zero = find_next_zero_bit(bitmap_info->bitmap,
1776                                                BITS_PER_BITMAP, i);
1777                 extent_bits = next_zero - i;
1778                 if (extent_bits >= bits) {
1779                         found_bits = extent_bits;
1780                         break;
1781                 } else if (extent_bits > max_bits) {
1782                         max_bits = extent_bits;
1783                 }
1784                 i = next_zero;
1785         }
1786
1787         if (found_bits) {
1788                 *offset = (u64)(i * ctl->unit) + bitmap_info->offset;
1789                 *bytes = (u64)(found_bits) * ctl->unit;
1790                 return 0;
1791         }
1792
1793         *bytes = (u64)(max_bits) * ctl->unit;
1794         bitmap_info->max_extent_size = *bytes;
1795         return -1;
1796 }
1797
1798 /* Cache the size of the max extent in bytes */
1799 static struct btrfs_free_space *
1800 find_free_space(struct btrfs_free_space_ctl *ctl, u64 *offset, u64 *bytes,
1801                 unsigned long align, u64 *max_extent_size)
1802 {
1803         struct btrfs_free_space *entry;
1804         struct rb_node *node;
1805         u64 tmp;
1806         u64 align_off;
1807         int ret;
1808
1809         if (!ctl->free_space_offset.rb_node)
1810                 goto out;
1811
1812         entry = tree_search_offset(ctl, offset_to_bitmap(ctl, *offset), 0, 1);
1813         if (!entry)
1814                 goto out;
1815
1816         for (node = &entry->offset_index; node; node = rb_next(node)) {
1817                 entry = rb_entry(node, struct btrfs_free_space, offset_index);
1818                 if (entry->bytes < *bytes) {
1819                         if (entry->bytes > *max_extent_size)
1820                                 *max_extent_size = entry->bytes;
1821                         continue;
1822                 }
1823
1824                 /* make sure the space returned is big enough
1825                  * to match our requested alignment
1826                  */
1827                 if (*bytes >= align) {
1828                         tmp = entry->offset - ctl->start + align - 1;
1829                         tmp = div64_u64(tmp, align);
1830                         tmp = tmp * align + ctl->start;
1831                         align_off = tmp - entry->offset;
1832                 } else {
1833                         align_off = 0;
1834                         tmp = entry->offset;
1835                 }
1836
1837                 if (entry->bytes < *bytes + align_off) {
1838                         if (entry->bytes > *max_extent_size)
1839                                 *max_extent_size = entry->bytes;
1840                         continue;
1841                 }
1842
1843                 if (entry->bitmap) {
1844                         u64 size = *bytes;
1845
1846                         ret = search_bitmap(ctl, entry, &tmp, &size, true);
1847                         if (!ret) {
1848                                 *offset = tmp;
1849                                 *bytes = size;
1850                                 return entry;
1851                         } else if (size > *max_extent_size) {
1852                                 *max_extent_size = size;
1853                         }
1854                         continue;
1855                 }
1856
1857                 *offset = tmp;
1858                 *bytes = entry->bytes - align_off;
1859                 return entry;
1860         }
1861 out:
1862         return NULL;
1863 }
1864
1865 static void add_new_bitmap(struct btrfs_free_space_ctl *ctl,
1866                            struct btrfs_free_space *info, u64 offset)
1867 {
1868         info->offset = offset_to_bitmap(ctl, offset);
1869         info->bytes = 0;
1870         INIT_LIST_HEAD(&info->list);
1871         link_free_space(ctl, info);
1872         ctl->total_bitmaps++;
1873
1874         ctl->op->recalc_thresholds(ctl);
1875 }
1876
1877 static void free_bitmap(struct btrfs_free_space_ctl *ctl,
1878                         struct btrfs_free_space *bitmap_info)
1879 {
1880         unlink_free_space(ctl, bitmap_info);
1881         kfree(bitmap_info->bitmap);
1882         kmem_cache_free(btrfs_free_space_cachep, bitmap_info);
1883         ctl->total_bitmaps--;
1884         ctl->op->recalc_thresholds(ctl);
1885 }
1886
1887 static noinline int remove_from_bitmap(struct btrfs_free_space_ctl *ctl,
1888                               struct btrfs_free_space *bitmap_info,
1889                               u64 *offset, u64 *bytes)
1890 {
1891         u64 end;
1892         u64 search_start, search_bytes;
1893         int ret;
1894
1895 again:
1896         end = bitmap_info->offset + (u64)(BITS_PER_BITMAP * ctl->unit) - 1;
1897
1898         /*
1899          * We need to search for bits in this bitmap.  We could only cover some
1900          * of the extent in this bitmap thanks to how we add space, so we need
1901          * to search for as much as it as we can and clear that amount, and then
1902          * go searching for the next bit.
1903          */
1904         search_start = *offset;
1905         search_bytes = ctl->unit;
1906         search_bytes = min(search_bytes, end - search_start + 1);
1907         ret = search_bitmap(ctl, bitmap_info, &search_start, &search_bytes,
1908                             false);
1909         if (ret < 0 || search_start != *offset)
1910                 return -EINVAL;
1911
1912         /* We may have found more bits than what we need */
1913         search_bytes = min(search_bytes, *bytes);
1914
1915         /* Cannot clear past the end of the bitmap */
1916         search_bytes = min(search_bytes, end - search_start + 1);
1917
1918         bitmap_clear_bits(ctl, bitmap_info, search_start, search_bytes);
1919         *offset += search_bytes;
1920         *bytes -= search_bytes;
1921
1922         if (*bytes) {
1923                 struct rb_node *next = rb_next(&bitmap_info->offset_index);
1924                 if (!bitmap_info->bytes)
1925                         free_bitmap(ctl, bitmap_info);
1926
1927                 /*
1928                  * no entry after this bitmap, but we still have bytes to
1929                  * remove, so something has gone wrong.
1930                  */
1931                 if (!next)
1932                         return -EINVAL;
1933
1934                 bitmap_info = rb_entry(next, struct btrfs_free_space,
1935                                        offset_index);
1936
1937                 /*
1938                  * if the next entry isn't a bitmap we need to return to let the
1939                  * extent stuff do its work.
1940                  */
1941                 if (!bitmap_info->bitmap)
1942                         return -EAGAIN;
1943
1944                 /*
1945                  * Ok the next item is a bitmap, but it may not actually hold
1946                  * the information for the rest of this free space stuff, so
1947                  * look for it, and if we don't find it return so we can try
1948                  * everything over again.
1949                  */
1950                 search_start = *offset;
1951                 search_bytes = ctl->unit;
1952                 ret = search_bitmap(ctl, bitmap_info, &search_start,
1953                                     &search_bytes, false);
1954                 if (ret < 0 || search_start != *offset)
1955                         return -EAGAIN;
1956
1957                 goto again;
1958         } else if (!bitmap_info->bytes)
1959                 free_bitmap(ctl, bitmap_info);
1960
1961         return 0;
1962 }
1963
1964 static u64 add_bytes_to_bitmap(struct btrfs_free_space_ctl *ctl,
1965                                struct btrfs_free_space *info, u64 offset,
1966                                u64 bytes)
1967 {
1968         u64 bytes_to_set = 0;
1969         u64 end;
1970
1971         end = info->offset + (u64)(BITS_PER_BITMAP * ctl->unit);
1972
1973         bytes_to_set = min(end - offset, bytes);
1974
1975         bitmap_set_bits(ctl, info, offset, bytes_to_set);
1976
1977         /*
1978          * We set some bytes, we have no idea what the max extent size is
1979          * anymore.
1980          */
1981         info->max_extent_size = 0;
1982
1983         return bytes_to_set;
1984
1985 }
1986
1987 static bool use_bitmap(struct btrfs_free_space_ctl *ctl,
1988                       struct btrfs_free_space *info)
1989 {
1990         struct btrfs_block_group_cache *block_group = ctl->private;
1991         struct btrfs_fs_info *fs_info = block_group->fs_info;
1992         bool forced = false;
1993
1994 #ifdef CONFIG_BTRFS_DEBUG
1995         if (btrfs_should_fragment_free_space(block_group))
1996                 forced = true;
1997 #endif
1998
1999         /*
2000          * If we are below the extents threshold then we can add this as an
2001          * extent, and don't have to deal with the bitmap
2002          */
2003         if (!forced && ctl->free_extents < ctl->extents_thresh) {
2004                 /*
2005                  * If this block group has some small extents we don't want to
2006                  * use up all of our free slots in the cache with them, we want
2007                  * to reserve them to larger extents, however if we have plenty
2008                  * of cache left then go ahead an dadd them, no sense in adding
2009                  * the overhead of a bitmap if we don't have to.
2010                  */
2011                 if (info->bytes <= fs_info->sectorsize * 4) {
2012                         if (ctl->free_extents * 2 <= ctl->extents_thresh)
2013                                 return false;
2014                 } else {
2015                         return false;
2016                 }
2017         }
2018
2019         /*
2020          * The original block groups from mkfs can be really small, like 8
2021          * megabytes, so don't bother with a bitmap for those entries.  However
2022          * some block groups can be smaller than what a bitmap would cover but
2023          * are still large enough that they could overflow the 32k memory limit,
2024          * so allow those block groups to still be allowed to have a bitmap
2025          * entry.
2026          */
2027         if (((BITS_PER_BITMAP * ctl->unit) >> 1) > block_group->key.offset)
2028                 return false;
2029
2030         return true;
2031 }
2032
2033 static const struct btrfs_free_space_op free_space_op = {
2034         .recalc_thresholds      = recalculate_thresholds,
2035         .use_bitmap             = use_bitmap,
2036 };
2037
2038 static int insert_into_bitmap(struct btrfs_free_space_ctl *ctl,
2039                               struct btrfs_free_space *info)
2040 {
2041         struct btrfs_free_space *bitmap_info;
2042         struct btrfs_block_group_cache *block_group = NULL;
2043         int added = 0;
2044         u64 bytes, offset, bytes_added;
2045         int ret;
2046
2047         bytes = info->bytes;
2048         offset = info->offset;
2049
2050         if (!ctl->op->use_bitmap(ctl, info))
2051                 return 0;
2052
2053         if (ctl->op == &free_space_op)
2054                 block_group = ctl->private;
2055 again:
2056         /*
2057          * Since we link bitmaps right into the cluster we need to see if we
2058          * have a cluster here, and if so and it has our bitmap we need to add
2059          * the free space to that bitmap.
2060          */
2061         if (block_group && !list_empty(&block_group->cluster_list)) {
2062                 struct btrfs_free_cluster *cluster;
2063                 struct rb_node *node;
2064                 struct btrfs_free_space *entry;
2065
2066                 cluster = list_entry(block_group->cluster_list.next,
2067                                      struct btrfs_free_cluster,
2068                                      block_group_list);
2069                 spin_lock(&cluster->lock);
2070                 node = rb_first(&cluster->root);
2071                 if (!node) {
2072                         spin_unlock(&cluster->lock);
2073                         goto no_cluster_bitmap;
2074                 }
2075
2076                 entry = rb_entry(node, struct btrfs_free_space, offset_index);
2077                 if (!entry->bitmap) {
2078                         spin_unlock(&cluster->lock);
2079                         goto no_cluster_bitmap;
2080                 }
2081
2082                 if (entry->offset == offset_to_bitmap(ctl, offset)) {
2083                         bytes_added = add_bytes_to_bitmap(ctl, entry,
2084                                                           offset, bytes);
2085                         bytes -= bytes_added;
2086                         offset += bytes_added;
2087                 }
2088                 spin_unlock(&cluster->lock);
2089                 if (!bytes) {
2090                         ret = 1;
2091                         goto out;
2092                 }
2093         }
2094
2095 no_cluster_bitmap:
2096         bitmap_info = tree_search_offset(ctl, offset_to_bitmap(ctl, offset),
2097                                          1, 0);
2098         if (!bitmap_info) {
2099                 ASSERT(added == 0);
2100                 goto new_bitmap;
2101         }
2102
2103         bytes_added = add_bytes_to_bitmap(ctl, bitmap_info, offset, bytes);
2104         bytes -= bytes_added;
2105         offset += bytes_added;
2106         added = 0;
2107
2108         if (!bytes) {
2109                 ret = 1;
2110                 goto out;
2111         } else
2112                 goto again;
2113
2114 new_bitmap:
2115         if (info && info->bitmap) {
2116                 add_new_bitmap(ctl, info, offset);
2117                 added = 1;
2118                 info = NULL;
2119                 goto again;
2120         } else {
2121                 spin_unlock(&ctl->tree_lock);
2122
2123                 /* no pre-allocated info, allocate a new one */
2124                 if (!info) {
2125                         info = kmem_cache_zalloc(btrfs_free_space_cachep,
2126                                                  GFP_NOFS);
2127                         if (!info) {
2128                                 spin_lock(&ctl->tree_lock);
2129                                 ret = -ENOMEM;
2130                                 goto out;
2131                         }
2132                 }
2133
2134                 /* allocate the bitmap */
2135                 info->bitmap = kzalloc(PAGE_SIZE, GFP_NOFS);
2136                 spin_lock(&ctl->tree_lock);
2137                 if (!info->bitmap) {
2138                         ret = -ENOMEM;
2139                         goto out;
2140                 }
2141                 goto again;
2142         }
2143
2144 out:
2145         if (info) {
2146                 if (info->bitmap)
2147                         kfree(info->bitmap);
2148                 kmem_cache_free(btrfs_free_space_cachep, info);
2149         }
2150
2151         return ret;
2152 }
2153
2154 static bool try_merge_free_space(struct btrfs_free_space_ctl *ctl,
2155                           struct btrfs_free_space *info, bool update_stat)
2156 {
2157         struct btrfs_free_space *left_info;
2158         struct btrfs_free_space *right_info;
2159         bool merged = false;
2160         u64 offset = info->offset;
2161         u64 bytes = info->bytes;
2162
2163         /*
2164          * first we want to see if there is free space adjacent to the range we
2165          * are adding, if there is remove that struct and add a new one to
2166          * cover the entire range
2167          */
2168         right_info = tree_search_offset(ctl, offset + bytes, 0, 0);
2169         if (right_info && rb_prev(&right_info->offset_index))
2170                 left_info = rb_entry(rb_prev(&right_info->offset_index),
2171                                      struct btrfs_free_space, offset_index);
2172         else
2173                 left_info = tree_search_offset(ctl, offset - 1, 0, 0);
2174
2175         if (right_info && !right_info->bitmap) {
2176                 if (update_stat)
2177                         unlink_free_space(ctl, right_info);
2178                 else
2179                         __unlink_free_space(ctl, right_info);
2180                 info->bytes += right_info->bytes;
2181                 kmem_cache_free(btrfs_free_space_cachep, right_info);
2182                 merged = true;
2183         }
2184
2185         if (left_info && !left_info->bitmap &&
2186             left_info->offset + left_info->bytes == offset) {
2187                 if (update_stat)
2188                         unlink_free_space(ctl, left_info);
2189                 else
2190                         __unlink_free_space(ctl, left_info);
2191                 info->offset = left_info->offset;
2192                 info->bytes += left_info->bytes;
2193                 kmem_cache_free(btrfs_free_space_cachep, left_info);
2194                 merged = true;
2195         }
2196
2197         return merged;
2198 }
2199
2200 static bool steal_from_bitmap_to_end(struct btrfs_free_space_ctl *ctl,
2201                                      struct btrfs_free_space *info,
2202                                      bool update_stat)
2203 {
2204         struct btrfs_free_space *bitmap;
2205         unsigned long i;
2206         unsigned long j;
2207         const u64 end = info->offset + info->bytes;
2208         const u64 bitmap_offset = offset_to_bitmap(ctl, end);
2209         u64 bytes;
2210
2211         bitmap = tree_search_offset(ctl, bitmap_offset, 1, 0);
2212         if (!bitmap)
2213                 return false;
2214
2215         i = offset_to_bit(bitmap->offset, ctl->unit, end);
2216         j = find_next_zero_bit(bitmap->bitmap, BITS_PER_BITMAP, i);
2217         if (j == i)
2218                 return false;
2219         bytes = (j - i) * ctl->unit;
2220         info->bytes += bytes;
2221
2222         if (update_stat)
2223                 bitmap_clear_bits(ctl, bitmap, end, bytes);
2224         else
2225                 __bitmap_clear_bits(ctl, bitmap, end, bytes);
2226
2227         if (!bitmap->bytes)
2228                 free_bitmap(ctl, bitmap);
2229
2230         return true;
2231 }
2232
2233 static bool steal_from_bitmap_to_front(struct btrfs_free_space_ctl *ctl,
2234                                        struct btrfs_free_space *info,
2235                                        bool update_stat)
2236 {
2237         struct btrfs_free_space *bitmap;
2238         u64 bitmap_offset;
2239         unsigned long i;
2240         unsigned long j;
2241         unsigned long prev_j;
2242         u64 bytes;
2243
2244         bitmap_offset = offset_to_bitmap(ctl, info->offset);
2245         /* If we're on a boundary, try the previous logical bitmap. */
2246         if (bitmap_offset == info->offset) {
2247                 if (info->offset == 0)
2248                         return false;
2249                 bitmap_offset = offset_to_bitmap(ctl, info->offset - 1);
2250         }
2251
2252         bitmap = tree_search_offset(ctl, bitmap_offset, 1, 0);
2253         if (!bitmap)
2254                 return false;
2255
2256         i = offset_to_bit(bitmap->offset, ctl->unit, info->offset) - 1;
2257         j = 0;
2258         prev_j = (unsigned long)-1;
2259         for_each_clear_bit_from(j, bitmap->bitmap, BITS_PER_BITMAP) {
2260                 if (j > i)
2261                         break;
2262                 prev_j = j;
2263         }
2264         if (prev_j == i)
2265                 return false;
2266
2267         if (prev_j == (unsigned long)-1)
2268                 bytes = (i + 1) * ctl->unit;
2269         else
2270                 bytes = (i - prev_j) * ctl->unit;
2271
2272         info->offset -= bytes;
2273         info->bytes += bytes;
2274
2275         if (update_stat)
2276                 bitmap_clear_bits(ctl, bitmap, info->offset, bytes);
2277         else
2278                 __bitmap_clear_bits(ctl, bitmap, info->offset, bytes);
2279
2280         if (!bitmap->bytes)
2281                 free_bitmap(ctl, bitmap);
2282
2283         return true;
2284 }
2285
2286 /*
2287  * We prefer always to allocate from extent entries, both for clustered and
2288  * non-clustered allocation requests. So when attempting to add a new extent
2289  * entry, try to see if there's adjacent free space in bitmap entries, and if
2290  * there is, migrate that space from the bitmaps to the extent.
2291  * Like this we get better chances of satisfying space allocation requests
2292  * because we attempt to satisfy them based on a single cache entry, and never
2293  * on 2 or more entries - even if the entries represent a contiguous free space
2294  * region (e.g. 1 extent entry + 1 bitmap entry starting where the extent entry
2295  * ends).
2296  */
2297 static void steal_from_bitmap(struct btrfs_free_space_ctl *ctl,
2298                               struct btrfs_free_space *info,
2299                               bool update_stat)
2300 {
2301         /*
2302          * Only work with disconnected entries, as we can change their offset,
2303          * and must be extent entries.
2304          */
2305         ASSERT(!info->bitmap);
2306         ASSERT(RB_EMPTY_NODE(&info->offset_index));
2307
2308         if (ctl->total_bitmaps > 0) {
2309                 bool stole_end;
2310                 bool stole_front = false;
2311
2312                 stole_end = steal_from_bitmap_to_end(ctl, info, update_stat);
2313                 if (ctl->total_bitmaps > 0)
2314                         stole_front = steal_from_bitmap_to_front(ctl, info,
2315                                                                  update_stat);
2316
2317                 if (stole_end || stole_front)
2318                         try_merge_free_space(ctl, info, update_stat);
2319         }
2320 }
2321
2322 int __btrfs_add_free_space(struct btrfs_fs_info *fs_info,
2323                            struct btrfs_free_space_ctl *ctl,
2324                            u64 offset, u64 bytes)
2325 {
2326         struct btrfs_free_space *info;
2327         int ret = 0;
2328
2329         info = kmem_cache_zalloc(btrfs_free_space_cachep, GFP_NOFS);
2330         if (!info)
2331                 return -ENOMEM;
2332
2333         info->offset = offset;
2334         info->bytes = bytes;
2335         RB_CLEAR_NODE(&info->offset_index);
2336
2337         spin_lock(&ctl->tree_lock);
2338
2339         if (try_merge_free_space(ctl, info, true))
2340                 goto link;
2341
2342         /*
2343          * There was no extent directly to the left or right of this new
2344          * extent then we know we're going to have to allocate a new extent, so
2345          * before we do that see if we need to drop this into a bitmap
2346          */
2347         ret = insert_into_bitmap(ctl, info);
2348         if (ret < 0) {
2349                 goto out;
2350         } else if (ret) {
2351                 ret = 0;
2352                 goto out;
2353         }
2354 link:
2355         /*
2356          * Only steal free space from adjacent bitmaps if we're sure we're not
2357          * going to add the new free space to existing bitmap entries - because
2358          * that would mean unnecessary work that would be reverted. Therefore
2359          * attempt to steal space from bitmaps if we're adding an extent entry.
2360          */
2361         steal_from_bitmap(ctl, info, true);
2362
2363         ret = link_free_space(ctl, info);
2364         if (ret)
2365                 kmem_cache_free(btrfs_free_space_cachep, info);
2366 out:
2367         spin_unlock(&ctl->tree_lock);
2368
2369         if (ret) {
2370                 btrfs_crit(fs_info, "unable to add free space :%d", ret);
2371                 ASSERT(ret != -EEXIST);
2372         }
2373
2374         return ret;
2375 }
2376
2377 int btrfs_remove_free_space(struct btrfs_block_group_cache *block_group,
2378                             u64 offset, u64 bytes)
2379 {
2380         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2381         struct btrfs_free_space *info;
2382         int ret;
2383         bool re_search = false;
2384
2385         spin_lock(&ctl->tree_lock);
2386
2387 again:
2388         ret = 0;
2389         if (!bytes)
2390                 goto out_lock;
2391
2392         info = tree_search_offset(ctl, offset, 0, 0);
2393         if (!info) {
2394                 /*
2395                  * oops didn't find an extent that matched the space we wanted
2396                  * to remove, look for a bitmap instead
2397                  */
2398                 info = tree_search_offset(ctl, offset_to_bitmap(ctl, offset),
2399                                           1, 0);
2400                 if (!info) {
2401                         /*
2402                          * If we found a partial bit of our free space in a
2403                          * bitmap but then couldn't find the other part this may
2404                          * be a problem, so WARN about it.
2405                          */
2406                         WARN_ON(re_search);
2407                         goto out_lock;
2408                 }
2409         }
2410
2411         re_search = false;
2412         if (!info->bitmap) {
2413                 unlink_free_space(ctl, info);
2414                 if (offset == info->offset) {
2415                         u64 to_free = min(bytes, info->bytes);
2416
2417                         info->bytes -= to_free;
2418                         info->offset += to_free;
2419                         if (info->bytes) {
2420                                 ret = link_free_space(ctl, info);
2421                                 WARN_ON(ret);
2422                         } else {
2423                                 kmem_cache_free(btrfs_free_space_cachep, info);
2424                         }
2425
2426                         offset += to_free;
2427                         bytes -= to_free;
2428                         goto again;
2429                 } else {
2430                         u64 old_end = info->bytes + info->offset;
2431
2432                         info->bytes = offset - info->offset;
2433                         ret = link_free_space(ctl, info);
2434                         WARN_ON(ret);
2435                         if (ret)
2436                                 goto out_lock;
2437
2438                         /* Not enough bytes in this entry to satisfy us */
2439                         if (old_end < offset + bytes) {
2440                                 bytes -= old_end - offset;
2441                                 offset = old_end;
2442                                 goto again;
2443                         } else if (old_end == offset + bytes) {
2444                                 /* all done */
2445                                 goto out_lock;
2446                         }
2447                         spin_unlock(&ctl->tree_lock);
2448
2449                         ret = btrfs_add_free_space(block_group, offset + bytes,
2450                                                    old_end - (offset + bytes));
2451                         WARN_ON(ret);
2452                         goto out;
2453                 }
2454         }
2455
2456         ret = remove_from_bitmap(ctl, info, &offset, &bytes);
2457         if (ret == -EAGAIN) {
2458                 re_search = true;
2459                 goto again;
2460         }
2461 out_lock:
2462         spin_unlock(&ctl->tree_lock);
2463 out:
2464         return ret;
2465 }
2466
2467 void btrfs_dump_free_space(struct btrfs_block_group_cache *block_group,
2468                            u64 bytes)
2469 {
2470         struct btrfs_fs_info *fs_info = block_group->fs_info;
2471         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2472         struct btrfs_free_space *info;
2473         struct rb_node *n;
2474         int count = 0;
2475
2476         for (n = rb_first(&ctl->free_space_offset); n; n = rb_next(n)) {
2477                 info = rb_entry(n, struct btrfs_free_space, offset_index);
2478                 if (info->bytes >= bytes && !block_group->ro)
2479                         count++;
2480                 btrfs_crit(fs_info, "entry offset %llu, bytes %llu, bitmap %s",
2481                            info->offset, info->bytes,
2482                        (info->bitmap) ? "yes" : "no");
2483         }
2484         btrfs_info(fs_info, "block group has cluster?: %s",
2485                list_empty(&block_group->cluster_list) ? "no" : "yes");
2486         btrfs_info(fs_info,
2487                    "%d blocks of free space at or bigger than bytes is", count);
2488 }
2489
2490 void btrfs_init_free_space_ctl(struct btrfs_block_group_cache *block_group)
2491 {
2492         struct btrfs_fs_info *fs_info = block_group->fs_info;
2493         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2494
2495         spin_lock_init(&ctl->tree_lock);
2496         ctl->unit = fs_info->sectorsize;
2497         ctl->start = block_group->key.objectid;
2498         ctl->private = block_group;
2499         ctl->op = &free_space_op;
2500         INIT_LIST_HEAD(&ctl->trimming_ranges);
2501         mutex_init(&ctl->cache_writeout_mutex);
2502
2503         /*
2504          * we only want to have 32k of ram per block group for keeping
2505          * track of free space, and if we pass 1/2 of that we want to
2506          * start converting things over to using bitmaps
2507          */
2508         ctl->extents_thresh = (SZ_32K / 2) / sizeof(struct btrfs_free_space);
2509 }
2510
2511 /*
2512  * for a given cluster, put all of its extents back into the free
2513  * space cache.  If the block group passed doesn't match the block group
2514  * pointed to by the cluster, someone else raced in and freed the
2515  * cluster already.  In that case, we just return without changing anything
2516  */
2517 static int
2518 __btrfs_return_cluster_to_free_space(
2519                              struct btrfs_block_group_cache *block_group,
2520                              struct btrfs_free_cluster *cluster)
2521 {
2522         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2523         struct btrfs_free_space *entry;
2524         struct rb_node *node;
2525
2526         spin_lock(&cluster->lock);
2527         if (cluster->block_group != block_group)
2528                 goto out;
2529
2530         cluster->block_group = NULL;
2531         cluster->window_start = 0;
2532         list_del_init(&cluster->block_group_list);
2533
2534         node = rb_first(&cluster->root);
2535         while (node) {
2536                 bool bitmap;
2537
2538                 entry = rb_entry(node, struct btrfs_free_space, offset_index);
2539                 node = rb_next(&entry->offset_index);
2540                 rb_erase(&entry->offset_index, &cluster->root);
2541                 RB_CLEAR_NODE(&entry->offset_index);
2542
2543                 bitmap = (entry->bitmap != NULL);
2544                 if (!bitmap) {
2545                         try_merge_free_space(ctl, entry, false);
2546                         steal_from_bitmap(ctl, entry, false);
2547                 }
2548                 tree_insert_offset(&ctl->free_space_offset,
2549                                    entry->offset, &entry->offset_index, bitmap);
2550         }
2551         cluster->root = RB_ROOT;
2552
2553 out:
2554         spin_unlock(&cluster->lock);
2555         btrfs_put_block_group(block_group);
2556         return 0;
2557 }
2558
2559 static void __btrfs_remove_free_space_cache_locked(
2560                                 struct btrfs_free_space_ctl *ctl)
2561 {
2562         struct btrfs_free_space *info;
2563         struct rb_node *node;
2564
2565         while ((node = rb_last(&ctl->free_space_offset)) != NULL) {
2566                 info = rb_entry(node, struct btrfs_free_space, offset_index);
2567                 if (!info->bitmap) {
2568                         unlink_free_space(ctl, info);
2569                         kmem_cache_free(btrfs_free_space_cachep, info);
2570                 } else {
2571                         free_bitmap(ctl, info);
2572                 }
2573
2574                 cond_resched_lock(&ctl->tree_lock);
2575         }
2576 }
2577
2578 void __btrfs_remove_free_space_cache(struct btrfs_free_space_ctl *ctl)
2579 {
2580         spin_lock(&ctl->tree_lock);
2581         __btrfs_remove_free_space_cache_locked(ctl);
2582         spin_unlock(&ctl->tree_lock);
2583 }
2584
2585 void btrfs_remove_free_space_cache(struct btrfs_block_group_cache *block_group)
2586 {
2587         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2588         struct btrfs_free_cluster *cluster;
2589         struct list_head *head;
2590
2591         spin_lock(&ctl->tree_lock);
2592         while ((head = block_group->cluster_list.next) !=
2593                &block_group->cluster_list) {
2594                 cluster = list_entry(head, struct btrfs_free_cluster,
2595                                      block_group_list);
2596
2597                 WARN_ON(cluster->block_group != block_group);
2598                 __btrfs_return_cluster_to_free_space(block_group, cluster);
2599
2600                 cond_resched_lock(&ctl->tree_lock);
2601         }
2602         __btrfs_remove_free_space_cache_locked(ctl);
2603         spin_unlock(&ctl->tree_lock);
2604
2605 }
2606
2607 u64 btrfs_find_space_for_alloc(struct btrfs_block_group_cache *block_group,
2608                                u64 offset, u64 bytes, u64 empty_size,
2609                                u64 *max_extent_size)
2610 {
2611         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2612         struct btrfs_free_space *entry = NULL;
2613         u64 bytes_search = bytes + empty_size;
2614         u64 ret = 0;
2615         u64 align_gap = 0;
2616         u64 align_gap_len = 0;
2617
2618         spin_lock(&ctl->tree_lock);
2619         entry = find_free_space(ctl, &offset, &bytes_search,
2620                                 block_group->full_stripe_len, max_extent_size);
2621         if (!entry)
2622                 goto out;
2623
2624         ret = offset;
2625         if (entry->bitmap) {
2626                 bitmap_clear_bits(ctl, entry, offset, bytes);
2627                 if (!entry->bytes)
2628                         free_bitmap(ctl, entry);
2629         } else {
2630                 unlink_free_space(ctl, entry);
2631                 align_gap_len = offset - entry->offset;
2632                 align_gap = entry->offset;
2633
2634                 entry->offset = offset + bytes;
2635                 WARN_ON(entry->bytes < bytes + align_gap_len);
2636
2637                 entry->bytes -= bytes + align_gap_len;
2638                 if (!entry->bytes)
2639                         kmem_cache_free(btrfs_free_space_cachep, entry);
2640                 else
2641                         link_free_space(ctl, entry);
2642         }
2643 out:
2644         spin_unlock(&ctl->tree_lock);
2645
2646         if (align_gap_len)
2647                 __btrfs_add_free_space(block_group->fs_info, ctl,
2648                                        align_gap, align_gap_len);
2649         return ret;
2650 }
2651
2652 /*
2653  * given a cluster, put all of its extents back into the free space
2654  * cache.  If a block group is passed, this function will only free
2655  * a cluster that belongs to the passed block group.
2656  *
2657  * Otherwise, it'll get a reference on the block group pointed to by the
2658  * cluster and remove the cluster from it.
2659  */
2660 int btrfs_return_cluster_to_free_space(
2661                                struct btrfs_block_group_cache *block_group,
2662                                struct btrfs_free_cluster *cluster)
2663 {
2664         struct btrfs_free_space_ctl *ctl;
2665         int ret;
2666
2667         /* first, get a safe pointer to the block group */
2668         spin_lock(&cluster->lock);
2669         if (!block_group) {
2670                 block_group = cluster->block_group;
2671                 if (!block_group) {
2672                         spin_unlock(&cluster->lock);
2673                         return 0;
2674                 }
2675         } else if (cluster->block_group != block_group) {
2676                 /* someone else has already freed it don't redo their work */
2677                 spin_unlock(&cluster->lock);
2678                 return 0;
2679         }
2680         atomic_inc(&block_group->count);
2681         spin_unlock(&cluster->lock);
2682
2683         ctl = block_group->free_space_ctl;
2684
2685         /* now return any extents the cluster had on it */
2686         spin_lock(&ctl->tree_lock);
2687         ret = __btrfs_return_cluster_to_free_space(block_group, cluster);
2688         spin_unlock(&ctl->tree_lock);
2689
2690         /* finally drop our ref */
2691         btrfs_put_block_group(block_group);
2692         return ret;
2693 }
2694
2695 static u64 btrfs_alloc_from_bitmap(struct btrfs_block_group_cache *block_group,
2696                                    struct btrfs_free_cluster *cluster,
2697                                    struct btrfs_free_space *entry,
2698                                    u64 bytes, u64 min_start,
2699                                    u64 *max_extent_size)
2700 {
2701         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2702         int err;
2703         u64 search_start = cluster->window_start;
2704         u64 search_bytes = bytes;
2705         u64 ret = 0;
2706
2707         search_start = min_start;
2708         search_bytes = bytes;
2709
2710         err = search_bitmap(ctl, entry, &search_start, &search_bytes, true);
2711         if (err) {
2712                 if (search_bytes > *max_extent_size)
2713                         *max_extent_size = search_bytes;
2714                 return 0;
2715         }
2716
2717         ret = search_start;
2718         __bitmap_clear_bits(ctl, entry, ret, bytes);
2719
2720         return ret;
2721 }
2722
2723 /*
2724  * given a cluster, try to allocate 'bytes' from it, returns 0
2725  * if it couldn't find anything suitably large, or a logical disk offset
2726  * if things worked out
2727  */
2728 u64 btrfs_alloc_from_cluster(struct btrfs_block_group_cache *block_group,
2729                              struct btrfs_free_cluster *cluster, u64 bytes,
2730                              u64 min_start, u64 *max_extent_size)
2731 {
2732         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2733         struct btrfs_free_space *entry = NULL;
2734         struct rb_node *node;
2735         u64 ret = 0;
2736
2737         spin_lock(&cluster->lock);
2738         if (bytes > cluster->max_size)
2739                 goto out;
2740
2741         if (cluster->block_group != block_group)
2742                 goto out;
2743
2744         node = rb_first(&cluster->root);
2745         if (!node)
2746                 goto out;
2747
2748         entry = rb_entry(node, struct btrfs_free_space, offset_index);
2749         while (1) {
2750                 if (entry->bytes < bytes && entry->bytes > *max_extent_size)
2751                         *max_extent_size = entry->bytes;
2752
2753                 if (entry->bytes < bytes ||
2754                     (!entry->bitmap && entry->offset < min_start)) {
2755                         node = rb_next(&entry->offset_index);
2756                         if (!node)
2757                                 break;
2758                         entry = rb_entry(node, struct btrfs_free_space,
2759                                          offset_index);
2760                         continue;
2761                 }
2762
2763                 if (entry->bitmap) {
2764                         ret = btrfs_alloc_from_bitmap(block_group,
2765                                                       cluster, entry, bytes,
2766                                                       cluster->window_start,
2767                                                       max_extent_size);
2768                         if (ret == 0) {
2769                                 node = rb_next(&entry->offset_index);
2770                                 if (!node)
2771                                         break;
2772                                 entry = rb_entry(node, struct btrfs_free_space,
2773                                                  offset_index);
2774                                 continue;
2775                         }
2776                         cluster->window_start += bytes;
2777                 } else {
2778                         ret = entry->offset;
2779
2780                         entry->offset += bytes;
2781                         entry->bytes -= bytes;
2782                 }
2783
2784                 if (entry->bytes == 0)
2785                         rb_erase(&entry->offset_index, &cluster->root);
2786                 break;
2787         }
2788 out:
2789         spin_unlock(&cluster->lock);
2790
2791         if (!ret)
2792                 return 0;
2793
2794         spin_lock(&ctl->tree_lock);
2795
2796         ctl->free_space -= bytes;
2797         if (entry->bytes == 0) {
2798                 ctl->free_extents--;
2799                 if (entry->bitmap) {
2800                         kfree(entry->bitmap);
2801                         ctl->total_bitmaps--;
2802                         ctl->op->recalc_thresholds(ctl);
2803                 }
2804                 kmem_cache_free(btrfs_free_space_cachep, entry);
2805         }
2806
2807         spin_unlock(&ctl->tree_lock);
2808
2809         return ret;
2810 }
2811
2812 static int btrfs_bitmap_cluster(struct btrfs_block_group_cache *block_group,
2813                                 struct btrfs_free_space *entry,
2814                                 struct btrfs_free_cluster *cluster,
2815                                 u64 offset, u64 bytes,
2816                                 u64 cont1_bytes, u64 min_bytes)
2817 {
2818         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2819         unsigned long next_zero;
2820         unsigned long i;
2821         unsigned long want_bits;
2822         unsigned long min_bits;
2823         unsigned long found_bits;
2824         unsigned long max_bits = 0;
2825         unsigned long start = 0;
2826         unsigned long total_found = 0;
2827         int ret;
2828
2829         i = offset_to_bit(entry->offset, ctl->unit,
2830                           max_t(u64, offset, entry->offset));
2831         want_bits = bytes_to_bits(bytes, ctl->unit);
2832         min_bits = bytes_to_bits(min_bytes, ctl->unit);
2833
2834         /*
2835          * Don't bother looking for a cluster in this bitmap if it's heavily
2836          * fragmented.
2837          */
2838         if (entry->max_extent_size &&
2839             entry->max_extent_size < cont1_bytes)
2840                 return -ENOSPC;
2841 again:
2842         found_bits = 0;
2843         for_each_set_bit_from(i, entry->bitmap, BITS_PER_BITMAP) {
2844                 next_zero = find_next_zero_bit(entry->bitmap,
2845                                                BITS_PER_BITMAP, i);
2846                 if (next_zero - i >= min_bits) {
2847                         found_bits = next_zero - i;
2848                         if (found_bits > max_bits)
2849                                 max_bits = found_bits;
2850                         break;
2851                 }
2852                 if (next_zero - i > max_bits)
2853                         max_bits = next_zero - i;
2854                 i = next_zero;
2855         }
2856
2857         if (!found_bits) {
2858                 entry->max_extent_size = (u64)max_bits * ctl->unit;
2859                 return -ENOSPC;
2860         }
2861
2862         if (!total_found) {
2863                 start = i;
2864                 cluster->max_size = 0;
2865         }
2866
2867         total_found += found_bits;
2868
2869         if (cluster->max_size < found_bits * ctl->unit)
2870                 cluster->max_size = found_bits * ctl->unit;
2871
2872         if (total_found < want_bits || cluster->max_size < cont1_bytes) {
2873                 i = next_zero + 1;
2874                 goto again;
2875         }
2876
2877         cluster->window_start = start * ctl->unit + entry->offset;
2878         rb_erase(&entry->offset_index, &ctl->free_space_offset);
2879         ret = tree_insert_offset(&cluster->root, entry->offset,
2880                                  &entry->offset_index, 1);
2881         ASSERT(!ret); /* -EEXIST; Logic error */
2882
2883         trace_btrfs_setup_cluster(block_group, cluster,
2884                                   total_found * ctl->unit, 1);
2885         return 0;
2886 }
2887
2888 /*
2889  * This searches the block group for just extents to fill the cluster with.
2890  * Try to find a cluster with at least bytes total bytes, at least one
2891  * extent of cont1_bytes, and other clusters of at least min_bytes.
2892  */
2893 static noinline int
2894 setup_cluster_no_bitmap(struct btrfs_block_group_cache *block_group,
2895                         struct btrfs_free_cluster *cluster,
2896                         struct list_head *bitmaps, u64 offset, u64 bytes,
2897                         u64 cont1_bytes, u64 min_bytes)
2898 {
2899         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2900         struct btrfs_free_space *first = NULL;
2901         struct btrfs_free_space *entry = NULL;
2902         struct btrfs_free_space *last;
2903         struct rb_node *node;
2904         u64 window_free;
2905         u64 max_extent;
2906         u64 total_size = 0;
2907
2908         entry = tree_search_offset(ctl, offset, 0, 1);
2909         if (!entry)
2910                 return -ENOSPC;
2911
2912         /*
2913          * We don't want bitmaps, so just move along until we find a normal
2914          * extent entry.
2915          */
2916         while (entry->bitmap || entry->bytes < min_bytes) {
2917                 if (entry->bitmap && list_empty(&entry->list))
2918                         list_add_tail(&entry->list, bitmaps);
2919                 node = rb_next(&entry->offset_index);
2920                 if (!node)
2921                         return -ENOSPC;
2922                 entry = rb_entry(node, struct btrfs_free_space, offset_index);
2923         }
2924
2925         window_free = entry->bytes;
2926         max_extent = entry->bytes;
2927         first = entry;
2928         last = entry;
2929
2930         for (node = rb_next(&entry->offset_index); node;
2931              node = rb_next(&entry->offset_index)) {
2932                 entry = rb_entry(node, struct btrfs_free_space, offset_index);
2933
2934                 if (entry->bitmap) {
2935                         if (list_empty(&entry->list))
2936                                 list_add_tail(&entry->list, bitmaps);
2937                         continue;
2938                 }
2939
2940                 if (entry->bytes < min_bytes)
2941                         continue;
2942
2943                 last = entry;
2944                 window_free += entry->bytes;
2945                 if (entry->bytes > max_extent)
2946                         max_extent = entry->bytes;
2947         }
2948
2949         if (window_free < bytes || max_extent < cont1_bytes)
2950                 return -ENOSPC;
2951
2952         cluster->window_start = first->offset;
2953
2954         node = &first->offset_index;
2955
2956         /*
2957          * now we've found our entries, pull them out of the free space
2958          * cache and put them into the cluster rbtree
2959          */
2960         do {
2961                 int ret;
2962
2963                 entry = rb_entry(node, struct btrfs_free_space, offset_index);
2964                 node = rb_next(&entry->offset_index);
2965                 if (entry->bitmap || entry->bytes < min_bytes)
2966                         continue;
2967
2968                 rb_erase(&entry->offset_index, &ctl->free_space_offset);
2969                 ret = tree_insert_offset(&cluster->root, entry->offset,
2970                                          &entry->offset_index, 0);
2971                 total_size += entry->bytes;
2972                 ASSERT(!ret); /* -EEXIST; Logic error */
2973         } while (node && entry != last);
2974
2975         cluster->max_size = max_extent;
2976         trace_btrfs_setup_cluster(block_group, cluster, total_size, 0);
2977         return 0;
2978 }
2979
2980 /*
2981  * This specifically looks for bitmaps that may work in the cluster, we assume
2982  * that we have already failed to find extents that will work.
2983  */
2984 static noinline int
2985 setup_cluster_bitmap(struct btrfs_block_group_cache *block_group,
2986                      struct btrfs_free_cluster *cluster,
2987                      struct list_head *bitmaps, u64 offset, u64 bytes,
2988                      u64 cont1_bytes, u64 min_bytes)
2989 {
2990         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2991         struct btrfs_free_space *entry = NULL;
2992         int ret = -ENOSPC;
2993         u64 bitmap_offset = offset_to_bitmap(ctl, offset);
2994
2995         if (ctl->total_bitmaps == 0)
2996                 return -ENOSPC;
2997
2998         /*
2999          * The bitmap that covers offset won't be in the list unless offset
3000          * is just its start offset.
3001          */
3002         if (!list_empty(bitmaps))
3003                 entry = list_first_entry(bitmaps, struct btrfs_free_space, list);
3004
3005         if (!entry || entry->offset != bitmap_offset) {
3006                 entry = tree_search_offset(ctl, bitmap_offset, 1, 0);
3007                 if (entry && list_empty(&entry->list))
3008                         list_add(&entry->list, bitmaps);
3009         }
3010
3011         list_for_each_entry(entry, bitmaps, list) {
3012                 if (entry->bytes < bytes)
3013                         continue;
3014                 ret = btrfs_bitmap_cluster(block_group, entry, cluster, offset,
3015                                            bytes, cont1_bytes, min_bytes);
3016                 if (!ret)
3017                         return 0;
3018         }
3019
3020         /*
3021          * The bitmaps list has all the bitmaps that record free space
3022          * starting after offset, so no more search is required.
3023          */
3024         return -ENOSPC;
3025 }
3026
3027 /*
3028  * here we try to find a cluster of blocks in a block group.  The goal
3029  * is to find at least bytes+empty_size.
3030  * We might not find them all in one contiguous area.
3031  *
3032  * returns zero and sets up cluster if things worked out, otherwise
3033  * it returns -enospc
3034  */
3035 int btrfs_find_space_cluster(struct btrfs_fs_info *fs_info,
3036                              struct btrfs_block_group_cache *block_group,
3037                              struct btrfs_free_cluster *cluster,
3038                              u64 offset, u64 bytes, u64 empty_size)
3039 {
3040         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3041         struct btrfs_free_space *entry, *tmp;
3042         LIST_HEAD(bitmaps);
3043         u64 min_bytes;
3044         u64 cont1_bytes;
3045         int ret;
3046
3047         /*
3048          * Choose the minimum extent size we'll require for this
3049          * cluster.  For SSD_SPREAD, don't allow any fragmentation.
3050          * For metadata, allow allocates with smaller extents.  For
3051          * data, keep it dense.
3052          */
3053         if (btrfs_test_opt(fs_info, SSD_SPREAD)) {
3054                 cont1_bytes = min_bytes = bytes + empty_size;
3055         } else if (block_group->flags & BTRFS_BLOCK_GROUP_METADATA) {
3056                 cont1_bytes = bytes;
3057                 min_bytes = fs_info->sectorsize;
3058         } else {
3059                 cont1_bytes = max(bytes, (bytes + empty_size) >> 2);
3060                 min_bytes = fs_info->sectorsize;
3061         }
3062
3063         spin_lock(&ctl->tree_lock);
3064
3065         /*
3066          * If we know we don't have enough space to make a cluster don't even
3067          * bother doing all the work to try and find one.
3068          */
3069         if (ctl->free_space < bytes) {
3070                 spin_unlock(&ctl->tree_lock);
3071                 return -ENOSPC;
3072         }
3073
3074         spin_lock(&cluster->lock);
3075
3076         /* someone already found a cluster, hooray */
3077         if (cluster->block_group) {
3078                 ret = 0;
3079                 goto out;
3080         }
3081
3082         trace_btrfs_find_cluster(block_group, offset, bytes, empty_size,
3083                                  min_bytes);
3084
3085         ret = setup_cluster_no_bitmap(block_group, cluster, &bitmaps, offset,
3086                                       bytes + empty_size,
3087                                       cont1_bytes, min_bytes);
3088         if (ret)
3089                 ret = setup_cluster_bitmap(block_group, cluster, &bitmaps,
3090                                            offset, bytes + empty_size,
3091                                            cont1_bytes, min_bytes);
3092
3093         /* Clear our temporary list */
3094         list_for_each_entry_safe(entry, tmp, &bitmaps, list)
3095                 list_del_init(&entry->list);
3096
3097         if (!ret) {
3098                 atomic_inc(&block_group->count);
3099                 list_add_tail(&cluster->block_group_list,
3100                               &block_group->cluster_list);
3101                 cluster->block_group = block_group;
3102         } else {
3103                 trace_btrfs_failed_cluster_setup(block_group);
3104         }
3105 out:
3106         spin_unlock(&cluster->lock);
3107         spin_unlock(&ctl->tree_lock);
3108
3109         return ret;
3110 }
3111
3112 /*
3113  * simple code to zero out a cluster
3114  */
3115 void btrfs_init_free_cluster(struct btrfs_free_cluster *cluster)
3116 {
3117         spin_lock_init(&cluster->lock);
3118         spin_lock_init(&cluster->refill_lock);
3119         cluster->root = RB_ROOT;
3120         cluster->max_size = 0;
3121         cluster->fragmented = false;
3122         INIT_LIST_HEAD(&cluster->block_group_list);
3123         cluster->block_group = NULL;
3124 }
3125
3126 static int do_trimming(struct btrfs_block_group_cache *block_group,
3127                        u64 *total_trimmed, u64 start, u64 bytes,
3128                        u64 reserved_start, u64 reserved_bytes,
3129                        struct btrfs_trim_range *trim_entry)
3130 {
3131         struct btrfs_space_info *space_info = block_group->space_info;
3132         struct btrfs_fs_info *fs_info = block_group->fs_info;
3133         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3134         int ret;
3135         int update = 0;
3136         u64 trimmed = 0;
3137
3138         spin_lock(&space_info->lock);
3139         spin_lock(&block_group->lock);
3140         if (!block_group->ro) {
3141                 block_group->reserved += reserved_bytes;
3142                 space_info->bytes_reserved += reserved_bytes;
3143                 update = 1;
3144         }
3145         spin_unlock(&block_group->lock);
3146         spin_unlock(&space_info->lock);
3147
3148         ret = btrfs_discard_extent(fs_info, start, bytes, &trimmed);
3149         if (!ret)
3150                 *total_trimmed += trimmed;
3151
3152         mutex_lock(&ctl->cache_writeout_mutex);
3153         btrfs_add_free_space(block_group, reserved_start, reserved_bytes);
3154         list_del(&trim_entry->list);
3155         mutex_unlock(&ctl->cache_writeout_mutex);
3156
3157         if (update) {
3158                 spin_lock(&space_info->lock);
3159                 spin_lock(&block_group->lock);
3160                 if (block_group->ro)
3161                         space_info->bytes_readonly += reserved_bytes;
3162                 block_group->reserved -= reserved_bytes;
3163                 space_info->bytes_reserved -= reserved_bytes;
3164                 spin_unlock(&space_info->lock);
3165                 spin_unlock(&block_group->lock);
3166         }
3167
3168         return ret;
3169 }
3170
3171 static int trim_no_bitmap(struct btrfs_block_group_cache *block_group,
3172                           u64 *total_trimmed, u64 start, u64 end, u64 minlen)
3173 {
3174         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3175         struct btrfs_free_space *entry;
3176         struct rb_node *node;
3177         int ret = 0;
3178         u64 extent_start;
3179         u64 extent_bytes;
3180         u64 bytes;
3181
3182         while (start < end) {
3183                 struct btrfs_trim_range trim_entry;
3184
3185                 mutex_lock(&ctl->cache_writeout_mutex);
3186                 spin_lock(&ctl->tree_lock);
3187
3188                 if (ctl->free_space < minlen) {
3189                         spin_unlock(&ctl->tree_lock);
3190                         mutex_unlock(&ctl->cache_writeout_mutex);
3191                         break;
3192                 }
3193
3194                 entry = tree_search_offset(ctl, start, 0, 1);
3195                 if (!entry) {
3196                         spin_unlock(&ctl->tree_lock);
3197                         mutex_unlock(&ctl->cache_writeout_mutex);
3198                         break;
3199                 }
3200
3201                 /* skip bitmaps */
3202                 while (entry->bitmap) {
3203                         node = rb_next(&entry->offset_index);
3204                         if (!node) {
3205                                 spin_unlock(&ctl->tree_lock);
3206                                 mutex_unlock(&ctl->cache_writeout_mutex);
3207                                 goto out;
3208                         }
3209                         entry = rb_entry(node, struct btrfs_free_space,
3210                                          offset_index);
3211                 }
3212
3213                 if (entry->offset >= end) {
3214                         spin_unlock(&ctl->tree_lock);
3215                         mutex_unlock(&ctl->cache_writeout_mutex);
3216                         break;
3217                 }
3218
3219                 extent_start = entry->offset;
3220                 extent_bytes = entry->bytes;
3221                 start = max(start, extent_start);
3222                 bytes = min(extent_start + extent_bytes, end) - start;
3223                 if (bytes < minlen) {
3224                         spin_unlock(&ctl->tree_lock);
3225                         mutex_unlock(&ctl->cache_writeout_mutex);
3226                         goto next;
3227                 }
3228
3229                 unlink_free_space(ctl, entry);
3230                 kmem_cache_free(btrfs_free_space_cachep, entry);
3231
3232                 spin_unlock(&ctl->tree_lock);
3233                 trim_entry.start = extent_start;
3234                 trim_entry.bytes = extent_bytes;
3235                 list_add_tail(&trim_entry.list, &ctl->trimming_ranges);
3236                 mutex_unlock(&ctl->cache_writeout_mutex);
3237
3238                 ret = do_trimming(block_group, total_trimmed, start, bytes,
3239                                   extent_start, extent_bytes, &trim_entry);
3240                 if (ret)
3241                         break;
3242 next:
3243                 start += bytes;
3244
3245                 if (fatal_signal_pending(current)) {
3246                         ret = -ERESTARTSYS;
3247                         break;
3248                 }
3249
3250                 cond_resched();
3251         }
3252 out:
3253         return ret;
3254 }
3255
3256 static int trim_bitmaps(struct btrfs_block_group_cache *block_group,
3257                         u64 *total_trimmed, u64 start, u64 end, u64 minlen)
3258 {
3259         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3260         struct btrfs_free_space *entry;
3261         int ret = 0;
3262         int ret2;
3263         u64 bytes;
3264         u64 offset = offset_to_bitmap(ctl, start);
3265
3266         while (offset < end) {
3267                 bool next_bitmap = false;
3268                 struct btrfs_trim_range trim_entry;
3269
3270                 mutex_lock(&ctl->cache_writeout_mutex);
3271                 spin_lock(&ctl->tree_lock);
3272
3273                 if (ctl->free_space < minlen) {
3274                         spin_unlock(&ctl->tree_lock);
3275                         mutex_unlock(&ctl->cache_writeout_mutex);
3276                         break;
3277                 }
3278
3279                 entry = tree_search_offset(ctl, offset, 1, 0);
3280                 if (!entry) {
3281                         spin_unlock(&ctl->tree_lock);
3282                         mutex_unlock(&ctl->cache_writeout_mutex);
3283                         next_bitmap = true;
3284                         goto next;
3285                 }
3286
3287                 bytes = minlen;
3288                 ret2 = search_bitmap(ctl, entry, &start, &bytes, false);
3289                 if (ret2 || start >= end) {
3290                         spin_unlock(&ctl->tree_lock);
3291                         mutex_unlock(&ctl->cache_writeout_mutex);
3292                         next_bitmap = true;
3293                         goto next;
3294                 }
3295
3296                 bytes = min(bytes, end - start);
3297                 if (bytes < minlen) {
3298                         spin_unlock(&ctl->tree_lock);
3299                         mutex_unlock(&ctl->cache_writeout_mutex);
3300                         goto next;
3301                 }
3302
3303                 bitmap_clear_bits(ctl, entry, start, bytes);
3304                 if (entry->bytes == 0)
3305                         free_bitmap(ctl, entry);
3306
3307                 spin_unlock(&ctl->tree_lock);
3308                 trim_entry.start = start;
3309                 trim_entry.bytes = bytes;
3310                 list_add_tail(&trim_entry.list, &ctl->trimming_ranges);
3311                 mutex_unlock(&ctl->cache_writeout_mutex);
3312
3313                 ret = do_trimming(block_group, total_trimmed, start, bytes,
3314                                   start, bytes, &trim_entry);
3315                 if (ret)
3316                         break;
3317 next:
3318                 if (next_bitmap) {
3319                         offset += BITS_PER_BITMAP * ctl->unit;
3320                 } else {
3321                         start += bytes;
3322                         if (start >= offset + BITS_PER_BITMAP * ctl->unit)
3323                                 offset += BITS_PER_BITMAP * ctl->unit;
3324                 }
3325
3326                 if (fatal_signal_pending(current)) {
3327                         ret = -ERESTARTSYS;
3328                         break;
3329                 }
3330
3331                 cond_resched();
3332         }
3333
3334         return ret;
3335 }
3336
3337 void btrfs_get_block_group_trimming(struct btrfs_block_group_cache *cache)
3338 {
3339         atomic_inc(&cache->trimming);
3340 }
3341
3342 void btrfs_put_block_group_trimming(struct btrfs_block_group_cache *block_group)
3343 {
3344         struct btrfs_fs_info *fs_info = block_group->fs_info;
3345         struct extent_map_tree *em_tree;
3346         struct extent_map *em;
3347         bool cleanup;
3348
3349         spin_lock(&block_group->lock);
3350         cleanup = (atomic_dec_and_test(&block_group->trimming) &&
3351                    block_group->removed);
3352         spin_unlock(&block_group->lock);
3353
3354         if (cleanup) {
3355                 mutex_lock(&fs_info->chunk_mutex);
3356                 em_tree = &fs_info->mapping_tree.map_tree;
3357                 write_lock(&em_tree->lock);
3358                 em = lookup_extent_mapping(em_tree, block_group->key.objectid,
3359                                            1);
3360                 BUG_ON(!em); /* logic error, can't happen */
3361                 /*
3362                  * remove_extent_mapping() will delete us from the pinned_chunks
3363                  * list, which is protected by the chunk mutex.
3364                  */
3365                 remove_extent_mapping(em_tree, em);
3366                 write_unlock(&em_tree->lock);
3367                 mutex_unlock(&fs_info->chunk_mutex);
3368
3369                 /* once for us and once for the tree */
3370                 free_extent_map(em);
3371                 free_extent_map(em);
3372
3373                 /*
3374                  * We've left one free space entry and other tasks trimming
3375                  * this block group have left 1 entry each one. Free them.
3376                  */
3377                 __btrfs_remove_free_space_cache(block_group->free_space_ctl);
3378         }
3379 }
3380
3381 int btrfs_trim_block_group(struct btrfs_block_group_cache *block_group,
3382                            u64 *trimmed, u64 start, u64 end, u64 minlen)
3383 {
3384         int ret;
3385
3386         *trimmed = 0;
3387
3388         spin_lock(&block_group->lock);
3389         if (block_group->removed) {
3390                 spin_unlock(&block_group->lock);
3391                 return 0;
3392         }
3393         btrfs_get_block_group_trimming(block_group);
3394         spin_unlock(&block_group->lock);
3395
3396         ret = trim_no_bitmap(block_group, trimmed, start, end, minlen);
3397         if (ret)
3398                 goto out;
3399
3400         ret = trim_bitmaps(block_group, trimmed, start, end, minlen);
3401 out:
3402         btrfs_put_block_group_trimming(block_group);
3403         return ret;
3404 }
3405
3406 /*
3407  * Find the left-most item in the cache tree, and then return the
3408  * smallest inode number in the item.
3409  *
3410  * Note: the returned inode number may not be the smallest one in
3411  * the tree, if the left-most item is a bitmap.
3412  */
3413 u64 btrfs_find_ino_for_alloc(struct btrfs_root *fs_root)
3414 {
3415         struct btrfs_free_space_ctl *ctl = fs_root->free_ino_ctl;
3416         struct btrfs_free_space *entry = NULL;
3417         u64 ino = 0;
3418
3419         spin_lock(&ctl->tree_lock);
3420
3421         if (RB_EMPTY_ROOT(&ctl->free_space_offset))
3422                 goto out;
3423
3424         entry = rb_entry(rb_first(&ctl->free_space_offset),
3425                          struct btrfs_free_space, offset_index);
3426
3427         if (!entry->bitmap) {
3428                 ino = entry->offset;
3429
3430                 unlink_free_space(ctl, entry);
3431                 entry->offset++;
3432                 entry->bytes--;
3433                 if (!entry->bytes)
3434                         kmem_cache_free(btrfs_free_space_cachep, entry);
3435                 else
3436                         link_free_space(ctl, entry);
3437         } else {
3438                 u64 offset = 0;
3439                 u64 count = 1;
3440                 int ret;
3441
3442                 ret = search_bitmap(ctl, entry, &offset, &count, true);
3443                 /* Logic error; Should be empty if it can't find anything */
3444                 ASSERT(!ret);
3445
3446                 ino = offset;
3447                 bitmap_clear_bits(ctl, entry, offset, 1);
3448                 if (entry->bytes == 0)
3449                         free_bitmap(ctl, entry);
3450         }
3451 out:
3452         spin_unlock(&ctl->tree_lock);
3453
3454         return ino;
3455 }
3456
3457 struct inode *lookup_free_ino_inode(struct btrfs_root *root,
3458                                     struct btrfs_path *path)
3459 {
3460         struct inode *inode = NULL;
3461
3462         spin_lock(&root->ino_cache_lock);
3463         if (root->ino_cache_inode)
3464                 inode = igrab(root->ino_cache_inode);
3465         spin_unlock(&root->ino_cache_lock);
3466         if (inode)
3467                 return inode;
3468
3469         inode = __lookup_free_space_inode(root, path, 0);
3470         if (IS_ERR(inode))
3471                 return inode;
3472
3473         spin_lock(&root->ino_cache_lock);
3474         if (!btrfs_fs_closing(root->fs_info))
3475                 root->ino_cache_inode = igrab(inode);
3476         spin_unlock(&root->ino_cache_lock);
3477
3478         return inode;
3479 }
3480
3481 int create_free_ino_inode(struct btrfs_root *root,
3482                           struct btrfs_trans_handle *trans,
3483                           struct btrfs_path *path)
3484 {
3485         return __create_free_space_inode(root, trans, path,
3486                                          BTRFS_FREE_INO_OBJECTID, 0);
3487 }
3488
3489 int load_free_ino_cache(struct btrfs_fs_info *fs_info, struct btrfs_root *root)
3490 {
3491         struct btrfs_free_space_ctl *ctl = root->free_ino_ctl;
3492         struct btrfs_path *path;
3493         struct inode *inode;
3494         int ret = 0;
3495         u64 root_gen = btrfs_root_generation(&root->root_item);
3496
3497         if (!btrfs_test_opt(fs_info, INODE_MAP_CACHE))
3498                 return 0;
3499
3500         /*
3501          * If we're unmounting then just return, since this does a search on the
3502          * normal root and not the commit root and we could deadlock.
3503          */
3504         if (btrfs_fs_closing(fs_info))
3505                 return 0;
3506
3507         path = btrfs_alloc_path();
3508         if (!path)
3509                 return 0;
3510
3511         inode = lookup_free_ino_inode(root, path);
3512         if (IS_ERR(inode))
3513                 goto out;
3514
3515         if (root_gen != BTRFS_I(inode)->generation)
3516                 goto out_put;
3517
3518         ret = __load_free_space_cache(root, inode, ctl, path, 0);
3519
3520         if (ret < 0)
3521                 btrfs_err(fs_info,
3522                         "failed to load free ino cache for root %llu",
3523                         root->root_key.objectid);
3524 out_put:
3525         iput(inode);
3526 out:
3527         btrfs_free_path(path);
3528         return ret;
3529 }
3530
3531 int btrfs_write_out_ino_cache(struct btrfs_root *root,
3532                               struct btrfs_trans_handle *trans,
3533                               struct btrfs_path *path,
3534                               struct inode *inode)
3535 {
3536         struct btrfs_fs_info *fs_info = root->fs_info;
3537         struct btrfs_free_space_ctl *ctl = root->free_ino_ctl;
3538         int ret;
3539         struct btrfs_io_ctl io_ctl;
3540         bool release_metadata = true;
3541
3542         if (!btrfs_test_opt(fs_info, INODE_MAP_CACHE))
3543                 return 0;
3544
3545         memset(&io_ctl, 0, sizeof(io_ctl));
3546         ret = __btrfs_write_out_cache(root, inode, ctl, NULL, &io_ctl,
3547                                       trans, path, 0);
3548         if (!ret) {
3549                 /*
3550                  * At this point writepages() didn't error out, so our metadata
3551                  * reservation is released when the writeback finishes, at
3552                  * inode.c:btrfs_finish_ordered_io(), regardless of it finishing
3553                  * with or without an error.
3554                  */
3555                 release_metadata = false;
3556                 ret = btrfs_wait_cache_io_root(root, trans, &io_ctl, path);
3557         }
3558
3559         if (ret) {
3560                 if (release_metadata)
3561                         btrfs_delalloc_release_metadata(inode, inode->i_size);
3562 #ifdef DEBUG
3563                 btrfs_err(fs_info,
3564                           "failed to write free ino cache for root %llu",
3565                           root->root_key.objectid);
3566 #endif
3567         }
3568
3569         return ret;
3570 }
3571
3572 #ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
3573 /*
3574  * Use this if you need to make a bitmap or extent entry specifically, it
3575  * doesn't do any of the merging that add_free_space does, this acts a lot like
3576  * how the free space cache loading stuff works, so you can get really weird
3577  * configurations.
3578  */
3579 int test_add_free_space_entry(struct btrfs_block_group_cache *cache,
3580                               u64 offset, u64 bytes, bool bitmap)
3581 {
3582         struct btrfs_free_space_ctl *ctl = cache->free_space_ctl;
3583         struct btrfs_free_space *info = NULL, *bitmap_info;
3584         void *map = NULL;
3585         u64 bytes_added;
3586         int ret;
3587
3588 again:
3589         if (!info) {
3590                 info = kmem_cache_zalloc(btrfs_free_space_cachep, GFP_NOFS);
3591                 if (!info)
3592                         return -ENOMEM;
3593         }
3594
3595         if (!bitmap) {
3596                 spin_lock(&ctl->tree_lock);
3597                 info->offset = offset;
3598                 info->bytes = bytes;
3599                 info->max_extent_size = 0;
3600                 ret = link_free_space(ctl, info);
3601                 spin_unlock(&ctl->tree_lock);
3602                 if (ret)
3603                         kmem_cache_free(btrfs_free_space_cachep, info);
3604                 return ret;
3605         }
3606
3607         if (!map) {
3608                 map = kzalloc(PAGE_SIZE, GFP_NOFS);
3609                 if (!map) {
3610                         kmem_cache_free(btrfs_free_space_cachep, info);
3611                         return -ENOMEM;
3612                 }
3613         }
3614
3615         spin_lock(&ctl->tree_lock);
3616         bitmap_info = tree_search_offset(ctl, offset_to_bitmap(ctl, offset),
3617                                          1, 0);
3618         if (!bitmap_info) {
3619                 info->bitmap = map;
3620                 map = NULL;
3621                 add_new_bitmap(ctl, info, offset);
3622                 bitmap_info = info;
3623                 info = NULL;
3624         }
3625
3626         bytes_added = add_bytes_to_bitmap(ctl, bitmap_info, offset, bytes);
3627
3628         bytes -= bytes_added;
3629         offset += bytes_added;
3630         spin_unlock(&ctl->tree_lock);
3631
3632         if (bytes)
3633                 goto again;
3634
3635         if (info)
3636                 kmem_cache_free(btrfs_free_space_cachep, info);
3637         if (map)
3638                 kfree(map);
3639         return 0;
3640 }
3641
3642 /*
3643  * Checks to see if the given range is in the free space cache.  This is really
3644  * just used to check the absence of space, so if there is free space in the
3645  * range at all we will return 1.
3646  */
3647 int test_check_exists(struct btrfs_block_group_cache *cache,
3648                       u64 offset, u64 bytes)
3649 {
3650         struct btrfs_free_space_ctl *ctl = cache->free_space_ctl;
3651         struct btrfs_free_space *info;
3652         int ret = 0;
3653
3654         spin_lock(&ctl->tree_lock);
3655         info = tree_search_offset(ctl, offset, 0, 0);
3656         if (!info) {
3657                 info = tree_search_offset(ctl, offset_to_bitmap(ctl, offset),
3658                                           1, 0);
3659                 if (!info)
3660                         goto out;
3661         }
3662
3663 have_info:
3664         if (info->bitmap) {
3665                 u64 bit_off, bit_bytes;
3666                 struct rb_node *n;
3667                 struct btrfs_free_space *tmp;
3668
3669                 bit_off = offset;
3670                 bit_bytes = ctl->unit;
3671                 ret = search_bitmap(ctl, info, &bit_off, &bit_bytes, false);
3672                 if (!ret) {
3673                         if (bit_off == offset) {
3674                                 ret = 1;
3675                                 goto out;
3676                         } else if (bit_off > offset &&
3677                                    offset + bytes > bit_off) {
3678                                 ret = 1;
3679                                 goto out;
3680                         }
3681                 }
3682
3683                 n = rb_prev(&info->offset_index);
3684                 while (n) {
3685                         tmp = rb_entry(n, struct btrfs_free_space,
3686                                        offset_index);
3687                         if (tmp->offset + tmp->bytes < offset)
3688                                 break;
3689                         if (offset + bytes < tmp->offset) {
3690                                 n = rb_prev(&tmp->offset_index);
3691                                 continue;
3692                         }
3693                         info = tmp;
3694                         goto have_info;
3695                 }
3696
3697                 n = rb_next(&info->offset_index);
3698                 while (n) {
3699                         tmp = rb_entry(n, struct btrfs_free_space,
3700                                        offset_index);
3701                         if (offset + bytes < tmp->offset)
3702                                 break;
3703                         if (tmp->offset + tmp->bytes < offset) {
3704                                 n = rb_next(&tmp->offset_index);
3705                                 continue;
3706                         }
3707                         info = tmp;
3708                         goto have_info;
3709                 }
3710
3711                 ret = 0;
3712                 goto out;
3713         }
3714
3715         if (info->offset == offset) {
3716                 ret = 1;
3717                 goto out;
3718         }
3719
3720         if (offset > info->offset && offset < info->offset + info->bytes)
3721                 ret = 1;
3722 out:
3723         spin_unlock(&ctl->tree_lock);
3724         return ret;
3725 }
3726 #endif /* CONFIG_BTRFS_FS_RUN_SANITY_TESTS */