fs/btrfs/free-space-cache.c

   1 // SPDX-License-Identifier: GPL-2.0
   2 /*
   3  * Copyright (C) 2008 Red Hat.  All rights reserved.
   4  */
   5
   6 #include <linux/pagemap.h>
   7 #include <linux/sched.h>
   8 #include <linux/sched/signal.h>
   9 #include <linux/slab.h>
  10 #include <linux/math64.h>
  11 #include <linux/ratelimit.h>
  12 #include <linux/error-injection.h>
  13 #include <linux/sched/mm.h>
  14 #include "ctree.h"
  15 #include "fs.h"
  16 #include "messages.h"
  17 #include "misc.h"
  18 #include "free-space-cache.h"
  19 #include "transaction.h"
  20 #include "disk-io.h"
  21 #include "extent_io.h"
  22 #include "volumes.h"
  23 #include "space-info.h"
  24 #include "delalloc-space.h"
  25 #include "block-group.h"
  26 #include "discard.h"
  27 #include "subpage.h"
  28 #include "inode-item.h"
  29 #include "accessors.h"
  30 #include "file-item.h"
  31 #include "file.h"
  32 #include "super.h"
  33
  34 #define BITS_PER_BITMAP         (PAGE_SIZE * 8UL)
  35 #define MAX_CACHE_BYTES_PER_GIG SZ_64K
  36 #define FORCE_EXTENT_THRESHOLD  SZ_1M
  37
  38 static struct kmem_cache *btrfs_free_space_cachep;
  39 static struct kmem_cache *btrfs_free_space_bitmap_cachep;
  40
  41 struct btrfs_trim_range {
  42         u64 start;
  43         u64 bytes;
  44         struct list_head list;
  45 };
  46
  47 static int link_free_space(struct btrfs_free_space_ctl *ctl,
  48                            struct btrfs_free_space *info);
  49 static void unlink_free_space(struct btrfs_free_space_ctl *ctl,
  50                               struct btrfs_free_space *info, bool update_stat);
  51 static int search_bitmap(struct btrfs_free_space_ctl *ctl,
  52                          struct btrfs_free_space *bitmap_info, u64 *offset,
  53                          u64 *bytes, bool for_alloc);
  54 static void free_bitmap(struct btrfs_free_space_ctl *ctl,
  55                         struct btrfs_free_space *bitmap_info);
  56 static void bitmap_clear_bits(struct btrfs_free_space_ctl *ctl,
  57                               struct btrfs_free_space *info, u64 offset,
  58                               u64 bytes, bool update_stats);
  59
  60 static void __btrfs_remove_free_space_cache(struct btrfs_free_space_ctl *ctl)
  61 {
  62         struct btrfs_free_space *info;
  63         struct rb_node *node;
  64
  65         while ((node = rb_last(&ctl->free_space_offset)) != NULL) {
  66                 info = rb_entry(node, struct btrfs_free_space, offset_index);
  67                 if (!info->bitmap) {
  68                         unlink_free_space(ctl, info, true);
  69                         kmem_cache_free(btrfs_free_space_cachep, info);
  70                 } else {
  71                         free_bitmap(ctl, info);
  72                 }
  73
  74                 cond_resched_lock(&ctl->tree_lock);
  75         }
  76 }
  77
  78 static struct inode *__lookup_free_space_inode(struct btrfs_root *root,
  79                                                struct btrfs_path *path,
  80                                                u64 offset)
  81 {
  82         struct btrfs_fs_info *fs_info = root->fs_info;
  83         struct btrfs_key key;
  84         struct btrfs_key location;
  85         struct btrfs_disk_key disk_key;
  86         struct btrfs_free_space_header *header;
  87         struct extent_buffer *leaf;
  88         struct inode *inode = NULL;
  89         unsigned nofs_flag;
  90         int ret;
  91
  92         key.objectid = BTRFS_FREE_SPACE_OBJECTID;
  93         key.offset = offset;
  94         key.type = 0;
  95
  96         ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
  97         if (ret < 0)
  98                 return ERR_PTR(ret);
  99         if (ret > 0) {
 100                 btrfs_release_path(path);
 101                 return ERR_PTR(-ENOENT);
 102         }
 103
 104         leaf = path->nodes[0];
 105         header = btrfs_item_ptr(leaf, path->slots[0],
 106                                 struct btrfs_free_space_header);
 107         btrfs_free_space_key(leaf, header, &disk_key);
 108         btrfs_disk_key_to_cpu(&location, &disk_key);
 109         btrfs_release_path(path);
 110
 111         /*
 112          * We are often under a trans handle at this point, so we need to make
 113          * sure NOFS is set to keep us from deadlocking.
 114          */
 115         nofs_flag = memalloc_nofs_save();
 116         inode = btrfs_iget_path(fs_info->sb, location.objectid, root, path);
 117         btrfs_release_path(path);
 118         memalloc_nofs_restore(nofs_flag);
 119         if (IS_ERR(inode))
 120                 return inode;
 121
 122         mapping_set_gfp_mask(inode->i_mapping,
 123                         mapping_gfp_constraint(inode->i_mapping,
 124                         ~(__GFP_FS | __GFP_HIGHMEM)));
 125
 126         return inode;
 127 }
 128
 129 struct inode *lookup_free_space_inode(struct btrfs_block_group *block_group,
 130                 struct btrfs_path *path)
 131 {
 132         struct btrfs_fs_info *fs_info = block_group->fs_info;
 133         struct inode *inode = NULL;
 134         u32 flags = BTRFS_INODE_NODATASUM | BTRFS_INODE_NODATACOW;
 135
 136         spin_lock(&block_group->lock);
 137         if (block_group->inode)
 138                 inode = igrab(block_group->inode);
 139         spin_unlock(&block_group->lock);
 140         if (inode)
 141                 return inode;
 142
 143         inode = __lookup_free_space_inode(fs_info->tree_root, path,
 144                                           block_group->start);
 145         if (IS_ERR(inode))
 146                 return inode;
 147
 148         spin_lock(&block_group->lock);
 149         if (!((BTRFS_I(inode)->flags & flags) == flags)) {
 150                 btrfs_info(fs_info, "Old style space inode found, converting.");
 151                 BTRFS_I(inode)->flags |= BTRFS_INODE_NODATASUM |
 152                         BTRFS_INODE_NODATACOW;
 153                 block_group->disk_cache_state = BTRFS_DC_CLEAR;
 154         }
 155
 156         if (!test_and_set_bit(BLOCK_GROUP_FLAG_IREF, &block_group->runtime_flags))
 157                 block_group->inode = igrab(inode);
 158         spin_unlock(&block_group->lock);
 159
 160         return inode;
 161 }
 162
 163 static int __create_free_space_inode(struct btrfs_root *root,
 164                                      struct btrfs_trans_handle *trans,
 165                                      struct btrfs_path *path,
 166                                      u64 ino, u64 offset)
 167 {
 168         struct btrfs_key key;
 169         struct btrfs_disk_key disk_key;
 170         struct btrfs_free_space_header *header;
 171         struct btrfs_inode_item *inode_item;
 172         struct extent_buffer *leaf;
 173         /* We inline CRCs for the free disk space cache */
 174         const u64 flags = BTRFS_INODE_NOCOMPRESS | BTRFS_INODE_PREALLOC |
 175                           BTRFS_INODE_NODATASUM | BTRFS_INODE_NODATACOW;
 176         int ret;
 177
 178         ret = btrfs_insert_empty_inode(trans, root, path, ino);
 179         if (ret)
 180                 return ret;
 181
 182         leaf = path->nodes[0];
 183         inode_item = btrfs_item_ptr(leaf, path->slots[0],
 184                                     struct btrfs_inode_item);
 185         btrfs_item_key(leaf, &disk_key, path->slots[0]);
 186         memzero_extent_buffer(leaf, (unsigned long)inode_item,
 187                              sizeof(*inode_item));
 188         btrfs_set_inode_generation(leaf, inode_item, trans->transid);
 189         btrfs_set_inode_size(leaf, inode_item, 0);
 190         btrfs_set_inode_nbytes(leaf, inode_item, 0);
 191         btrfs_set_inode_uid(leaf, inode_item, 0);
 192         btrfs_set_inode_gid(leaf, inode_item, 0);
 193         btrfs_set_inode_mode(leaf, inode_item, S_IFREG | 0600);
 194         btrfs_set_inode_flags(leaf, inode_item, flags);
 195         btrfs_set_inode_nlink(leaf, inode_item, 1);
 196         btrfs_set_inode_transid(leaf, inode_item, trans->transid);
 197         btrfs_set_inode_block_group(leaf, inode_item, offset);
 198         btrfs_mark_buffer_dirty(leaf);
 199         btrfs_release_path(path);
 200
 201         key.objectid = BTRFS_FREE_SPACE_OBJECTID;
 202         key.offset = offset;
 203         key.type = 0;
 204         ret = btrfs_insert_empty_item(trans, root, path, &key,
 205                                       sizeof(struct btrfs_free_space_header));
 206         if (ret < 0) {
 207                 btrfs_release_path(path);
 208                 return ret;
 209         }
 210
 211         leaf = path->nodes[0];
 212         header = btrfs_item_ptr(leaf, path->slots[0],
 213                                 struct btrfs_free_space_header);
 214         memzero_extent_buffer(leaf, (unsigned long)header, sizeof(*header));
 215         btrfs_set_free_space_key(leaf, header, &disk_key);
 216         btrfs_mark_buffer_dirty(leaf);
 217         btrfs_release_path(path);
 218
 219         return 0;
 220 }
 221
 222 int create_free_space_inode(struct btrfs_trans_handle *trans,
 223                             struct btrfs_block_group *block_group,
 224                             struct btrfs_path *path)
 225 {
 226         int ret;
 227         u64 ino;
 228
 229         ret = btrfs_get_free_objectid(trans->fs_info->tree_root, &ino);
 230         if (ret < 0)
 231                 return ret;
 232
 233         return __create_free_space_inode(trans->fs_info->tree_root, trans, path,
 234                                          ino, block_group->start);
 235 }
 236
 237 /*
 238  * inode is an optional sink: if it is NULL, btrfs_remove_free_space_inode
 239  * handles lookup, otherwise it takes ownership and iputs the inode.
 240  * Don't reuse an inode pointer after passing it into this function.
 241  */
 242 int btrfs_remove_free_space_inode(struct btrfs_trans_handle *trans,
 243                                   struct inode *inode,
 244                                   struct btrfs_block_group *block_group)
 245 {
 246         struct btrfs_path *path;
 247         struct btrfs_key key;
 248         int ret = 0;
 249
 250         path = btrfs_alloc_path();
 251         if (!path)
 252                 return -ENOMEM;
 253
 254         if (!inode)
 255                 inode = lookup_free_space_inode(block_group, path);
 256         if (IS_ERR(inode)) {
 257                 if (PTR_ERR(inode) != -ENOENT)
 258                         ret = PTR_ERR(inode);
 259                 goto out;
 260         }
 261         ret = btrfs_orphan_add(trans, BTRFS_I(inode));
 262         if (ret) {
 263                 btrfs_add_delayed_iput(BTRFS_I(inode));
 264                 goto out;
 265         }
 266         clear_nlink(inode);
 267         /* One for the block groups ref */
 268         spin_lock(&block_group->lock);
 269         if (test_and_clear_bit(BLOCK_GROUP_FLAG_IREF, &block_group->runtime_flags)) {
 270                 block_group->inode = NULL;
 271                 spin_unlock(&block_group->lock);
 272                 iput(inode);
 273         } else {
 274                 spin_unlock(&block_group->lock);
 275         }
 276         /* One for the lookup ref */
 277         btrfs_add_delayed_iput(BTRFS_I(inode));
 278
 279         key.objectid = BTRFS_FREE_SPACE_OBJECTID;
 280         key.type = 0;
 281         key.offset = block_group->start;
 282         ret = btrfs_search_slot(trans, trans->fs_info->tree_root, &key, path,
 283                                 -1, 1);
 284         if (ret) {
 285                 if (ret > 0)
 286                         ret = 0;
 287                 goto out;
 288         }
 289         ret = btrfs_del_item(trans, trans->fs_info->tree_root, path);
 290 out:
 291         btrfs_free_path(path);
 292         return ret;
 293 }
 294
 295 int btrfs_truncate_free_space_cache(struct btrfs_trans_handle *trans,
 296                                     struct btrfs_block_group *block_group,
 297                                     struct inode *vfs_inode)
 298 {
 299         struct btrfs_truncate_control control = {
 300                 .inode = BTRFS_I(vfs_inode),
 301                 .new_size = 0,
 302                 .ino = btrfs_ino(BTRFS_I(vfs_inode)),
 303                 .min_type = BTRFS_EXTENT_DATA_KEY,
 304                 .clear_extent_range = true,
 305         };
 306         struct btrfs_inode *inode = BTRFS_I(vfs_inode);
 307         struct btrfs_root *root = inode->root;
 308         struct extent_state *cached_state = NULL;
 309         int ret = 0;
 310         bool locked = false;
 311
 312         if (block_group) {
 313                 struct btrfs_path *path = btrfs_alloc_path();
 314
 315                 if (!path) {
 316                         ret = -ENOMEM;
 317                         goto fail;
 318                 }
 319                 locked = true;
 320                 mutex_lock(&trans->transaction->cache_write_mutex);
 321                 if (!list_empty(&block_group->io_list)) {
 322                         list_del_init(&block_group->io_list);
 323
 324                         btrfs_wait_cache_io(trans, block_group, path);
 325                         btrfs_put_block_group(block_group);
 326                 }
 327
 328                 /*
 329                  * now that we've truncated the cache away, its no longer
 330                  * setup or written
 331                  */
 332                 spin_lock(&block_group->lock);
 333                 block_group->disk_cache_state = BTRFS_DC_CLEAR;
 334                 spin_unlock(&block_group->lock);
 335                 btrfs_free_path(path);
 336         }
 337
 338         btrfs_i_size_write(inode, 0);
 339         truncate_pagecache(vfs_inode, 0);
 340
 341         lock_extent(&inode->io_tree, 0, (u64)-1, &cached_state);
 342         btrfs_drop_extent_map_range(inode, 0, (u64)-1, false);
 343
 344         /*
 345          * We skip the throttling logic for free space cache inodes, so we don't
 346          * need to check for -EAGAIN.
 347          */
 348         ret = btrfs_truncate_inode_items(trans, root, &control);
 349
 350         inode_sub_bytes(&inode->vfs_inode, control.sub_bytes);
 351         btrfs_inode_safe_disk_i_size_write(inode, control.last_size);
 352
 353         unlock_extent(&inode->io_tree, 0, (u64)-1, &cached_state);
 354         if (ret)
 355                 goto fail;
 356
 357         ret = btrfs_update_inode(trans, root, inode);
 358
 359 fail:
 360         if (locked)
 361                 mutex_unlock(&trans->transaction->cache_write_mutex);
 362         if (ret)
 363                 btrfs_abort_transaction(trans, ret);
 364
 365         return ret;
 366 }
 367
 368 static void readahead_cache(struct inode *inode)
 369 {
 370         struct file_ra_state ra;
 371         unsigned long last_index;
 372
 373         file_ra_state_init(&ra, inode->i_mapping);
 374         last_index = (i_size_read(inode) - 1) >> PAGE_SHIFT;
 375
 376         page_cache_sync_readahead(inode->i_mapping, &ra, NULL, 0, last_index);
 377 }
 378
 379 static int io_ctl_init(struct btrfs_io_ctl *io_ctl, struct inode *inode,
 380                        int write)
 381 {
 382         int num_pages;
 383
 384         num_pages = DIV_ROUND_UP(i_size_read(inode), PAGE_SIZE);
 385
 386         /* Make sure we can fit our crcs and generation into the first page */
 387         if (write && (num_pages * sizeof(u32) + sizeof(u64)) > PAGE_SIZE)
 388                 return -ENOSPC;
 389
 390         memset(io_ctl, 0, sizeof(struct btrfs_io_ctl));
 391
 392         io_ctl->pages = kcalloc(num_pages, sizeof(struct page *), GFP_NOFS);
 393         if (!io_ctl->pages)
 394                 return -ENOMEM;
 395
 396         io_ctl->num_pages = num_pages;
 397         io_ctl->fs_info = btrfs_sb(inode->i_sb);
 398         io_ctl->inode = inode;
 399
 400         return 0;
 401 }
 402 ALLOW_ERROR_INJECTION(io_ctl_init, ERRNO);
 403
 404 static void io_ctl_free(struct btrfs_io_ctl *io_ctl)
 405 {
 406         kfree(io_ctl->pages);
 407         io_ctl->pages = NULL;
 408 }
 409
 410 static void io_ctl_unmap_page(struct btrfs_io_ctl *io_ctl)
 411 {
 412         if (io_ctl->cur) {
 413                 io_ctl->cur = NULL;
 414                 io_ctl->orig = NULL;
 415         }
 416 }
 417
 418 static void io_ctl_map_page(struct btrfs_io_ctl *io_ctl, int clear)
 419 {
 420         ASSERT(io_ctl->index < io_ctl->num_pages);
 421         io_ctl->page = io_ctl->pages[io_ctl->index++];
 422         io_ctl->cur = page_address(io_ctl->page);
 423         io_ctl->orig = io_ctl->cur;
 424         io_ctl->size = PAGE_SIZE;
 425         if (clear)
 426                 clear_page(io_ctl->cur);
 427 }
 428
 429 static void io_ctl_drop_pages(struct btrfs_io_ctl *io_ctl)
 430 {
 431         int i;
 432
 433         io_ctl_unmap_page(io_ctl);
 434
 435         for (i = 0; i < io_ctl->num_pages; i++) {
 436                 if (io_ctl->pages[i]) {
 437                         btrfs_page_clear_checked(io_ctl->fs_info,
 438                                         io_ctl->pages[i],
 439                                         page_offset(io_ctl->pages[i]),
 440                                         PAGE_SIZE);
 441                         unlock_page(io_ctl->pages[i]);
 442                         put_page(io_ctl->pages[i]);
 443                 }
 444         }
 445 }
 446
 447 static int io_ctl_prepare_pages(struct btrfs_io_ctl *io_ctl, bool uptodate)
 448 {
 449         struct page *page;
 450         struct inode *inode = io_ctl->inode;
 451         gfp_t mask = btrfs_alloc_write_mask(inode->i_mapping);
 452         int i;
 453
 454         for (i = 0; i < io_ctl->num_pages; i++) {
 455                 int ret;
 456
 457                 page = find_or_create_page(inode->i_mapping, i, mask);
 458                 if (!page) {
 459                         io_ctl_drop_pages(io_ctl);
 460                         return -ENOMEM;
 461                 }
 462
 463                 ret = set_page_extent_mapped(page);
 464                 if (ret < 0) {
 465                         unlock_page(page);
 466                         put_page(page);
 467                         io_ctl_drop_pages(io_ctl);
 468                         return ret;
 469                 }
 470
 471                 io_ctl->pages[i] = page;
 472                 if (uptodate && !PageUptodate(page)) {
 473                         btrfs_read_folio(NULL, page_folio(page));
 474                         lock_page(page);
 475                         if (page->mapping != inode->i_mapping) {
 476                                 btrfs_err(BTRFS_I(inode)->root->fs_info,
 477                                           "free space cache page truncated");
 478                                 io_ctl_drop_pages(io_ctl);
 479                                 return -EIO;
 480                         }
 481                         if (!PageUptodate(page)) {
 482                                 btrfs_err(BTRFS_I(inode)->root->fs_info,
 483                                            "error reading free space cache");
 484                                 io_ctl_drop_pages(io_ctl);
 485                                 return -EIO;
 486                         }
 487                 }
 488         }
 489
 490         for (i = 0; i < io_ctl->num_pages; i++)
 491                 clear_page_dirty_for_io(io_ctl->pages[i]);
 492
 493         return 0;
 494 }
 495
 496 static void io_ctl_set_generation(struct btrfs_io_ctl *io_ctl, u64 generation)
 497 {
 498         io_ctl_map_page(io_ctl, 1);
 499
 500         /*
 501          * Skip the csum areas.  If we don't check crcs then we just have a
 502          * 64bit chunk at the front of the first page.
 503          */
 504         io_ctl->cur += (sizeof(u32) * io_ctl->num_pages);
 505         io_ctl->size -= sizeof(u64) + (sizeof(u32) * io_ctl->num_pages);
 506
 507         put_unaligned_le64(generation, io_ctl->cur);
 508         io_ctl->cur += sizeof(u64);
 509 }
 510
 511 static int io_ctl_check_generation(struct btrfs_io_ctl *io_ctl, u64 generation)
 512 {
 513         u64 cache_gen;
 514
 515         /*
 516          * Skip the crc area.  If we don't check crcs then we just have a 64bit
 517          * chunk at the front of the first page.
 518          */
 519         io_ctl->cur += sizeof(u32) * io_ctl->num_pages;
 520         io_ctl->size -= sizeof(u64) + (sizeof(u32) * io_ctl->num_pages);
 521
 522         cache_gen = get_unaligned_le64(io_ctl->cur);
 523         if (cache_gen != generation) {
 524                 btrfs_err_rl(io_ctl->fs_info,
 525                         "space cache generation (%llu) does not match inode (%llu)",
 526                                 cache_gen, generation);
 527                 io_ctl_unmap_page(io_ctl);
 528                 return -EIO;
 529         }
 530         io_ctl->cur += sizeof(u64);
 531         return 0;
 532 }
 533
 534 static void io_ctl_set_crc(struct btrfs_io_ctl *io_ctl, int index)
 535 {
 536         u32 *tmp;
 537         u32 crc = ~(u32)0;
 538         unsigned offset = 0;
 539
 540         if (index == 0)
 541                 offset = sizeof(u32) * io_ctl->num_pages;
 542
 543         crc = btrfs_crc32c(crc, io_ctl->orig + offset, PAGE_SIZE - offset);
 544         btrfs_crc32c_final(crc, (u8 *)&crc);
 545         io_ctl_unmap_page(io_ctl);
 546         tmp = page_address(io_ctl->pages[0]);
 547         tmp += index;
 548         *tmp = crc;
 549 }
 550
 551 static int io_ctl_check_crc(struct btrfs_io_ctl *io_ctl, int index)
 552 {
 553         u32 *tmp, val;
 554         u32 crc = ~(u32)0;
 555         unsigned offset = 0;
 556
 557         if (index == 0)
 558                 offset = sizeof(u32) * io_ctl->num_pages;
 559
 560         tmp = page_address(io_ctl->pages[0]);
 561         tmp += index;
 562         val = *tmp;
 563
 564         io_ctl_map_page(io_ctl, 0);
 565         crc = btrfs_crc32c(crc, io_ctl->orig + offset, PAGE_SIZE - offset);
 566         btrfs_crc32c_final(crc, (u8 *)&crc);
 567         if (val != crc) {
 568                 btrfs_err_rl(io_ctl->fs_info,
 569                         "csum mismatch on free space cache");
 570                 io_ctl_unmap_page(io_ctl);
 571                 return -EIO;
 572         }
 573
 574         return 0;
 575 }
 576
 577 static int io_ctl_add_entry(struct btrfs_io_ctl *io_ctl, u64 offset, u64 bytes,
 578                             void *bitmap)
 579 {
 580         struct btrfs_free_space_entry *entry;
 581
 582         if (!io_ctl->cur)
 583                 return -ENOSPC;
 584
 585         entry = io_ctl->cur;
 586         put_unaligned_le64(offset, &entry->offset);
 587         put_unaligned_le64(bytes, &entry->bytes);
 588         entry->type = (bitmap) ? BTRFS_FREE_SPACE_BITMAP :
 589                 BTRFS_FREE_SPACE_EXTENT;
 590         io_ctl->cur += sizeof(struct btrfs_free_space_entry);
 591         io_ctl->size -= sizeof(struct btrfs_free_space_entry);
 592
 593         if (io_ctl->size >= sizeof(struct btrfs_free_space_entry))
 594                 return 0;
 595
 596         io_ctl_set_crc(io_ctl, io_ctl->index - 1);
 597
 598         /* No more pages to map */
 599         if (io_ctl->index >= io_ctl->num_pages)
 600                 return 0;
 601
 602         /* map the next page */
 603         io_ctl_map_page(io_ctl, 1);
 604         return 0;
 605 }
 606
 607 static int io_ctl_add_bitmap(struct btrfs_io_ctl *io_ctl, void *bitmap)
 608 {
 609         if (!io_ctl->cur)
 610                 return -ENOSPC;
 611
 612         /*
 613          * If we aren't at the start of the current page, unmap this one and
 614          * map the next one if there is any left.
 615          */
 616         if (io_ctl->cur != io_ctl->orig) {
 617                 io_ctl_set_crc(io_ctl, io_ctl->index - 1);
 618                 if (io_ctl->index >= io_ctl->num_pages)
 619                         return -ENOSPC;
 620                 io_ctl_map_page(io_ctl, 0);
 621         }
 622
 623         copy_page(io_ctl->cur, bitmap);
 624         io_ctl_set_crc(io_ctl, io_ctl->index - 1);
 625         if (io_ctl->index < io_ctl->num_pages)
 626                 io_ctl_map_page(io_ctl, 0);
 627         return 0;
 628 }
 629
 630 static void io_ctl_zero_remaining_pages(struct btrfs_io_ctl *io_ctl)
 631 {
 632         /*
 633          * If we're not on the boundary we know we've modified the page and we
 634          * need to crc the page.
 635          */
 636         if (io_ctl->cur != io_ctl->orig)
 637                 io_ctl_set_crc(io_ctl, io_ctl->index - 1);
 638         else
 639                 io_ctl_unmap_page(io_ctl);
 640
 641         while (io_ctl->index < io_ctl->num_pages) {
 642                 io_ctl_map_page(io_ctl, 1);
 643                 io_ctl_set_crc(io_ctl, io_ctl->index - 1);
 644         }
 645 }
 646
 647 static int io_ctl_read_entry(struct btrfs_io_ctl *io_ctl,
 648                             struct btrfs_free_space *entry, u8 *type)
 649 {
 650         struct btrfs_free_space_entry *e;
 651         int ret;
 652
 653         if (!io_ctl->cur) {
 654                 ret = io_ctl_check_crc(io_ctl, io_ctl->index);
 655                 if (ret)
 656                         return ret;
 657         }
 658
 659         e = io_ctl->cur;
 660         entry->offset = get_unaligned_le64(&e->offset);
 661         entry->bytes = get_unaligned_le64(&e->bytes);
 662         *type = e->type;
 663         io_ctl->cur += sizeof(struct btrfs_free_space_entry);
 664         io_ctl->size -= sizeof(struct btrfs_free_space_entry);
 665
 666         if (io_ctl->size >= sizeof(struct btrfs_free_space_entry))
 667                 return 0;
 668
 669         io_ctl_unmap_page(io_ctl);
 670
 671         return 0;
 672 }
 673
 674 static int io_ctl_read_bitmap(struct btrfs_io_ctl *io_ctl,
 675                               struct btrfs_free_space *entry)
 676 {
 677         int ret;
 678
 679         ret = io_ctl_check_crc(io_ctl, io_ctl->index);
 680         if (ret)
 681                 return ret;
 682
 683         copy_page(entry->bitmap, io_ctl->cur);
 684         io_ctl_unmap_page(io_ctl);
 685
 686         return 0;
 687 }
 688
 689 static void recalculate_thresholds(struct btrfs_free_space_ctl *ctl)
 690 {
 691         struct btrfs_block_group *block_group = ctl->block_group;
 692         u64 max_bytes;
 693         u64 bitmap_bytes;
 694         u64 extent_bytes;
 695         u64 size = block_group->length;
 696         u64 bytes_per_bg = BITS_PER_BITMAP * ctl->unit;
 697         u64 max_bitmaps = div64_u64(size + bytes_per_bg - 1, bytes_per_bg);
 698
 699         max_bitmaps = max_t(u64, max_bitmaps, 1);
 700
 701         if (ctl->total_bitmaps > max_bitmaps)
 702                 btrfs_err(block_group->fs_info,
 703 "invalid free space control: bg start=%llu len=%llu total_bitmaps=%u unit=%u max_bitmaps=%llu bytes_per_bg=%llu",
 704                           block_group->start, block_group->length,
 705                           ctl->total_bitmaps, ctl->unit, max_bitmaps,
 706                           bytes_per_bg);
 707         ASSERT(ctl->total_bitmaps <= max_bitmaps);
 708
 709         /*
 710          * We are trying to keep the total amount of memory used per 1GiB of
 711          * space to be MAX_CACHE_BYTES_PER_GIG.  However, with a reclamation
 712          * mechanism of pulling extents >= FORCE_EXTENT_THRESHOLD out of
 713          * bitmaps, we may end up using more memory than this.
 714          */
 715         if (size < SZ_1G)
 716                 max_bytes = MAX_CACHE_BYTES_PER_GIG;
 717         else
 718                 max_bytes = MAX_CACHE_BYTES_PER_GIG * div_u64(size, SZ_1G);
 719
 720         bitmap_bytes = ctl->total_bitmaps * ctl->unit;
 721
 722         /*
 723          * we want the extent entry threshold to always be at most 1/2 the max
 724          * bytes we can have, or whatever is less than that.
 725          */
 726         extent_bytes = max_bytes - bitmap_bytes;
 727         extent_bytes = min_t(u64, extent_bytes, max_bytes >> 1);
 728
 729         ctl->extents_thresh =
 730                 div_u64(extent_bytes, sizeof(struct btrfs_free_space));
 731 }
 732
 733 static int __load_free_space_cache(struct btrfs_root *root, struct inode *inode,
 734                                    struct btrfs_free_space_ctl *ctl,
 735                                    struct btrfs_path *path, u64 offset)
 736 {
 737         struct btrfs_fs_info *fs_info = root->fs_info;
 738         struct btrfs_free_space_header *header;
 739         struct extent_buffer *leaf;
 740         struct btrfs_io_ctl io_ctl;
 741         struct btrfs_key key;
 742         struct btrfs_free_space *e, *n;
 743         LIST_HEAD(bitmaps);
 744         u64 num_entries;
 745         u64 num_bitmaps;
 746         u64 generation;
 747         u8 type;
 748         int ret = 0;
 749
 750         /* Nothing in the space cache, goodbye */
 751         if (!i_size_read(inode))
 752                 return 0;
 753
 754         key.objectid = BTRFS_FREE_SPACE_OBJECTID;
 755         key.offset = offset;
 756         key.type = 0;
 757
 758         ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
 759         if (ret < 0)
 760                 return 0;
 761         else if (ret > 0) {
 762                 btrfs_release_path(path);
 763                 return 0;
 764         }
 765
 766         ret = -1;
 767
 768         leaf = path->nodes[0];
 769         header = btrfs_item_ptr(leaf, path->slots[0],
 770                                 struct btrfs_free_space_header);
 771         num_entries = btrfs_free_space_entries(leaf, header);
 772         num_bitmaps = btrfs_free_space_bitmaps(leaf, header);
 773         generation = btrfs_free_space_generation(leaf, header);
 774         btrfs_release_path(path);
 775
 776         if (!BTRFS_I(inode)->generation) {
 777                 btrfs_info(fs_info,
 778                            "the free space cache file (%llu) is invalid, skip it",
 779                            offset);
 780                 return 0;
 781         }
 782
 783         if (BTRFS_I(inode)->generation != generation) {
 784                 btrfs_err(fs_info,
 785                           "free space inode generation (%llu) did not match free space cache generation (%llu)",
 786                           BTRFS_I(inode)->generation, generation);
 787                 return 0;
 788         }
 789
 790         if (!num_entries)
 791                 return 0;
 792
 793         ret = io_ctl_init(&io_ctl, inode, 0);
 794         if (ret)
 795                 return ret;
 796
 797         readahead_cache(inode);
 798
 799         ret = io_ctl_prepare_pages(&io_ctl, true);
 800         if (ret)
 801                 goto out;
 802
 803         ret = io_ctl_check_crc(&io_ctl, 0);
 804         if (ret)
 805                 goto free_cache;
 806
 807         ret = io_ctl_check_generation(&io_ctl, generation);
 808         if (ret)
 809                 goto free_cache;
 810
 811         while (num_entries) {
 812                 e = kmem_cache_zalloc(btrfs_free_space_cachep,
 813                                       GFP_NOFS);
 814                 if (!e) {
 815                         ret = -ENOMEM;
 816                         goto free_cache;
 817                 }
 818
 819                 ret = io_ctl_read_entry(&io_ctl, e, &type);
 820                 if (ret) {
 821                         kmem_cache_free(btrfs_free_space_cachep, e);
 822                         goto free_cache;
 823                 }
 824
 825                 if (!e->bytes) {
 826                         ret = -1;
 827                         kmem_cache_free(btrfs_free_space_cachep, e);
 828                         goto free_cache;
 829                 }
 830
 831                 if (type == BTRFS_FREE_SPACE_EXTENT) {
 832                         spin_lock(&ctl->tree_lock);
 833                         ret = link_free_space(ctl, e);
 834                         spin_unlock(&ctl->tree_lock);
 835                         if (ret) {
 836                                 btrfs_err(fs_info,
 837                                         "Duplicate entries in free space cache, dumping");
 838                                 kmem_cache_free(btrfs_free_space_cachep, e);
 839                                 goto free_cache;
 840                         }
 841                 } else {
 842                         ASSERT(num_bitmaps);
 843                         num_bitmaps--;
 844                         e->bitmap = kmem_cache_zalloc(
 845                                         btrfs_free_space_bitmap_cachep, GFP_NOFS);
 846                         if (!e->bitmap) {
 847                                 ret = -ENOMEM;
 848                                 kmem_cache_free(
 849                                         btrfs_free_space_cachep, e);
 850                                 goto free_cache;
 851                         }
 852                         spin_lock(&ctl->tree_lock);
 853                         ret = link_free_space(ctl, e);
 854                         if (ret) {
 855                                 spin_unlock(&ctl->tree_lock);
 856                                 btrfs_err(fs_info,
 857                                         "Duplicate entries in free space cache, dumping");
 858                                 kmem_cache_free(btrfs_free_space_cachep, e);
 859                                 goto free_cache;
 860                         }
 861                         ctl->total_bitmaps++;
 862                         recalculate_thresholds(ctl);
 863                         spin_unlock(&ctl->tree_lock);
 864                         list_add_tail(&e->list, &bitmaps);
 865                 }
 866
 867                 num_entries--;
 868         }
 869
 870         io_ctl_unmap_page(&io_ctl);
 871
 872         /*
 873          * We add the bitmaps at the end of the entries in order that
 874          * the bitmap entries are added to the cache.
 875          */
 876         list_for_each_entry_safe(e, n, &bitmaps, list) {
 877                 list_del_init(&e->list);
 878                 ret = io_ctl_read_bitmap(&io_ctl, e);
 879                 if (ret)
 880                         goto free_cache;
 881         }
 882
 883         io_ctl_drop_pages(&io_ctl);
 884         ret = 1;
 885 out:
 886         io_ctl_free(&io_ctl);
 887         return ret;
 888 free_cache:
 889         io_ctl_drop_pages(&io_ctl);
 890
 891         spin_lock(&ctl->tree_lock);
 892         __btrfs_remove_free_space_cache(ctl);
 893         spin_unlock(&ctl->tree_lock);
 894         goto out;
 895 }
 896
 897 static int copy_free_space_cache(struct btrfs_block_group *block_group,
 898                                  struct btrfs_free_space_ctl *ctl)
 899 {
 900         struct btrfs_free_space *info;
 901         struct rb_node *n;
 902         int ret = 0;
 903
 904         while (!ret && (n = rb_first(&ctl->free_space_offset)) != NULL) {
 905                 info = rb_entry(n, struct btrfs_free_space, offset_index);
 906                 if (!info->bitmap) {
 907                         const u64 offset = info->offset;
 908                         const u64 bytes = info->bytes;
 909
 910                         unlink_free_space(ctl, info, true);
 911                         spin_unlock(&ctl->tree_lock);
 912                         kmem_cache_free(btrfs_free_space_cachep, info);
 913                         ret = btrfs_add_free_space(block_group, offset, bytes);
 914                         spin_lock(&ctl->tree_lock);
 915                 } else {
 916                         u64 offset = info->offset;
 917                         u64 bytes = ctl->unit;
 918
 919                         ret = search_bitmap(ctl, info, &offset, &bytes, false);
 920                         if (ret == 0) {
 921                                 bitmap_clear_bits(ctl, info, offset, bytes, true);
 922                                 spin_unlock(&ctl->tree_lock);
 923                                 ret = btrfs_add_free_space(block_group, offset,
 924                                                            bytes);
 925                                 spin_lock(&ctl->tree_lock);
 926                         } else {
 927                                 free_bitmap(ctl, info);
 928                                 ret = 0;
 929                         }
 930                 }
 931                 cond_resched_lock(&ctl->tree_lock);
 932         }
 933         return ret;
 934 }
 935
 936 static struct lock_class_key btrfs_free_space_inode_key;
 937
 938 int load_free_space_cache(struct btrfs_block_group *block_group)
 939 {
 940         struct btrfs_fs_info *fs_info = block_group->fs_info;
 941         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
 942         struct btrfs_free_space_ctl tmp_ctl = {};
 943         struct inode *inode;
 944         struct btrfs_path *path;
 945         int ret = 0;
 946         bool matched;
 947         u64 used = block_group->used;
 948
 949         /*
 950          * Because we could potentially discard our loaded free space, we want
 951          * to load everything into a temporary structure first, and then if it's
 952          * valid copy it all into the actual free space ctl.
 953          */
 954         btrfs_init_free_space_ctl(block_group, &tmp_ctl);
 955
 956         /*
 957          * If this block group has been marked to be cleared for one reason or
 958          * another then we can't trust the on disk cache, so just return.
 959          */
 960         spin_lock(&block_group->lock);
 961         if (block_group->disk_cache_state != BTRFS_DC_WRITTEN) {
 962                 spin_unlock(&block_group->lock);
 963                 return 0;
 964         }
 965         spin_unlock(&block_group->lock);
 966
 967         path = btrfs_alloc_path();
 968         if (!path)
 969                 return 0;
 970         path->search_commit_root = 1;
 971         path->skip_locking = 1;
 972
 973         /*
 974          * We must pass a path with search_commit_root set to btrfs_iget in
 975          * order to avoid a deadlock when allocating extents for the tree root.
 976          *
 977          * When we are COWing an extent buffer from the tree root, when looking
 978          * for a free extent, at extent-tree.c:find_free_extent(), we can find
 979          * block group without its free space cache loaded. When we find one
 980          * we must load its space cache which requires reading its free space
 981          * cache's inode item from the root tree. If this inode item is located
 982          * in the same leaf that we started COWing before, then we end up in
 983          * deadlock on the extent buffer (trying to read lock it when we
 984          * previously write locked it).
 985          *
 986          * It's safe to read the inode item using the commit root because
 987          * block groups, once loaded, stay in memory forever (until they are
 988          * removed) as well as their space caches once loaded. New block groups
 989          * once created get their ->cached field set to BTRFS_CACHE_FINISHED so
 990          * we will never try to read their inode item while the fs is mounted.
 991          */
 992         inode = lookup_free_space_inode(block_group, path);
 993         if (IS_ERR(inode)) {
 994                 btrfs_free_path(path);
 995                 return 0;
 996         }
 997
 998         /* We may have converted the inode and made the cache invalid. */
 999         spin_lock(&block_group->lock);
1000         if (block_group->disk_cache_state != BTRFS_DC_WRITTEN) {
1001                 spin_unlock(&block_group->lock);
1002                 btrfs_free_path(path);
1003                 goto out;
1004         }
1005         spin_unlock(&block_group->lock);
1006
1007         /*
1008          * Reinitialize the class of struct inode's mapping->invalidate_lock for
1009          * free space inodes to prevent false positives related to locks for normal
1010          * inodes.
1011          */
1012         lockdep_set_class(&(&inode->i_data)->invalidate_lock,
1013                           &btrfs_free_space_inode_key);
1014
1015         ret = __load_free_space_cache(fs_info->tree_root, inode, &tmp_ctl,
1016                                       path, block_group->start);
1017         btrfs_free_path(path);
1018         if (ret <= 0)
1019                 goto out;
1020
1021         matched = (tmp_ctl.free_space == (block_group->length - used -
1022                                           block_group->bytes_super));
1023
1024         if (matched) {
1025                 spin_lock(&tmp_ctl.tree_lock);
1026                 ret = copy_free_space_cache(block_group, &tmp_ctl);
1027                 spin_unlock(&tmp_ctl.tree_lock);
1028                 /*
1029                  * ret == 1 means we successfully loaded the free space cache,
1030                  * so we need to re-set it here.
1031                  */
1032                 if (ret == 0)
1033                         ret = 1;
1034         } else {
1035                 /*
1036                  * We need to call the _locked variant so we don't try to update
1037                  * the discard counters.
1038                  */
1039                 spin_lock(&tmp_ctl.tree_lock);
1040                 __btrfs_remove_free_space_cache(&tmp_ctl);
1041                 spin_unlock(&tmp_ctl.tree_lock);
1042                 btrfs_warn(fs_info,
1043                            "block group %llu has wrong amount of free space",
1044                            block_group->start);
1045                 ret = -1;
1046         }
1047 out:
1048         if (ret < 0) {
1049                 /* This cache is bogus, make sure it gets cleared */
1050                 spin_lock(&block_group->lock);
1051                 block_group->disk_cache_state = BTRFS_DC_CLEAR;
1052                 spin_unlock(&block_group->lock);
1053                 ret = 0;
1054
1055                 btrfs_warn(fs_info,
1056                            "failed to load free space cache for block group %llu, rebuilding it now",
1057                            block_group->start);
1058         }
1059
1060         spin_lock(&ctl->tree_lock);
1061         btrfs_discard_update_discardable(block_group);
1062         spin_unlock(&ctl->tree_lock);
1063         iput(inode);
1064         return ret;
1065 }
1066
1067 static noinline_for_stack
1068 int write_cache_extent_entries(struct btrfs_io_ctl *io_ctl,
1069                               struct btrfs_free_space_ctl *ctl,
1070                               struct btrfs_block_group *block_group,
1071                               int *entries, int *bitmaps,
1072                               struct list_head *bitmap_list)
1073 {
1074         int ret;
1075         struct btrfs_free_cluster *cluster = NULL;
1076         struct btrfs_free_cluster *cluster_locked = NULL;
1077         struct rb_node *node = rb_first(&ctl->free_space_offset);
1078         struct btrfs_trim_range *trim_entry;
1079
1080         /* Get the cluster for this block_group if it exists */
1081         if (block_group && !list_empty(&block_group->cluster_list)) {
1082                 cluster = list_entry(block_group->cluster_list.next,
1083                                      struct btrfs_free_cluster,
1084                                      block_group_list);
1085         }
1086
1087         if (!node && cluster) {
1088                 cluster_locked = cluster;
1089                 spin_lock(&cluster_locked->lock);
1090                 node = rb_first(&cluster->root);
1091                 cluster = NULL;
1092         }
1093
1094         /* Write out the extent entries */
1095         while (node) {
1096                 struct btrfs_free_space *e;
1097
1098                 e = rb_entry(node, struct btrfs_free_space, offset_index);
1099                 *entries += 1;
1100
1101                 ret = io_ctl_add_entry(io_ctl, e->offset, e->bytes,
1102                                        e->bitmap);
1103                 if (ret)
1104                         goto fail;
1105
1106                 if (e->bitmap) {
1107                         list_add_tail(&e->list, bitmap_list);
1108                         *bitmaps += 1;
1109                 }
1110                 node = rb_next(node);
1111                 if (!node && cluster) {
1112                         node = rb_first(&cluster->root);
1113                         cluster_locked = cluster;
1114                         spin_lock(&cluster_locked->lock);
1115                         cluster = NULL;
1116                 }
1117         }
1118         if (cluster_locked) {
1119                 spin_unlock(&cluster_locked->lock);
1120                 cluster_locked = NULL;
1121         }
1122
1123         /*
1124          * Make sure we don't miss any range that was removed from our rbtree
1125          * because trimming is running. Otherwise after a umount+mount (or crash
1126          * after committing the transaction) we would leak free space and get
1127          * an inconsistent free space cache report from fsck.
1128          */
1129         list_for_each_entry(trim_entry, &ctl->trimming_ranges, list) {
1130                 ret = io_ctl_add_entry(io_ctl, trim_entry->start,
1131                                        trim_entry->bytes, NULL);
1132                 if (ret)
1133                         goto fail;
1134                 *entries += 1;
1135         }
1136
1137         return 0;
1138 fail:
1139         if (cluster_locked)
1140                 spin_unlock(&cluster_locked->lock);
1141         return -ENOSPC;
1142 }
1143
1144 static noinline_for_stack int
1145 update_cache_item(struct btrfs_trans_handle *trans,
1146                   struct btrfs_root *root,
1147                   struct inode *inode,
1148                   struct btrfs_path *path, u64 offset,
1149                   int entries, int bitmaps)
1150 {
1151         struct btrfs_key key;
1152         struct btrfs_free_space_header *header;
1153         struct extent_buffer *leaf;
1154         int ret;
1155
1156         key.objectid = BTRFS_FREE_SPACE_OBJECTID;
1157         key.offset = offset;
1158         key.type = 0;
1159
1160         ret = btrfs_search_slot(trans, root, &key, path, 0, 1);
1161         if (ret < 0) {
1162                 clear_extent_bit(&BTRFS_I(inode)->io_tree, 0, inode->i_size - 1,
1163                                  EXTENT_DELALLOC, NULL);
1164                 goto fail;
1165         }
1166         leaf = path->nodes[0];
1167         if (ret > 0) {
1168                 struct btrfs_key found_key;
1169                 ASSERT(path->slots[0]);
1170                 path->slots[0]--;
1171                 btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
1172                 if (found_key.objectid != BTRFS_FREE_SPACE_OBJECTID ||
1173                     found_key.offset != offset) {
1174                         clear_extent_bit(&BTRFS_I(inode)->io_tree, 0,
1175                                          inode->i_size - 1, EXTENT_DELALLOC,
1176                                          NULL);
1177                         btrfs_release_path(path);
1178                         goto fail;
1179                 }
1180         }
1181
1182         BTRFS_I(inode)->generation = trans->transid;
1183         header = btrfs_item_ptr(leaf, path->slots[0],
1184                                 struct btrfs_free_space_header);
1185         btrfs_set_free_space_entries(leaf, header, entries);
1186         btrfs_set_free_space_bitmaps(leaf, header, bitmaps);
1187         btrfs_set_free_space_generation(leaf, header, trans->transid);
1188         btrfs_mark_buffer_dirty(leaf);
1189         btrfs_release_path(path);
1190
1191         return 0;
1192
1193 fail:
1194         return -1;
1195 }
1196
1197 static noinline_for_stack int write_pinned_extent_entries(
1198                             struct btrfs_trans_handle *trans,
1199                             struct btrfs_block_group *block_group,
1200                             struct btrfs_io_ctl *io_ctl,
1201                             int *entries)
1202 {
1203         u64 start, extent_start, extent_end, len;
1204         struct extent_io_tree *unpin = NULL;
1205         int ret;
1206
1207         if (!block_group)
1208                 return 0;
1209
1210         /*
1211          * We want to add any pinned extents to our free space cache
1212          * so we don't leak the space
1213          *
1214          * We shouldn't have switched the pinned extents yet so this is the
1215          * right one
1216          */
1217         unpin = &trans->transaction->pinned_extents;
1218
1219         start = block_group->start;
1220
1221         while (start < block_group->start + block_group->length) {
1222                 if (!find_first_extent_bit(unpin, start,
1223                                            &extent_start, &extent_end,
1224                                            EXTENT_DIRTY, NULL))
1225                         return 0;
1226
1227                 /* This pinned extent is out of our range */
1228                 if (extent_start >= block_group->start + block_group->length)
1229                         return 0;
1230
1231                 extent_start = max(extent_start, start);
1232                 extent_end = min(block_group->start + block_group->length,
1233                                  extent_end + 1);
1234                 len = extent_end - extent_start;
1235
1236                 *entries += 1;
1237                 ret = io_ctl_add_entry(io_ctl, extent_start, len, NULL);
1238                 if (ret)
1239                         return -ENOSPC;
1240
1241                 start = extent_end;
1242         }
1243
1244         return 0;
1245 }
1246
1247 static noinline_for_stack int
1248 write_bitmap_entries(struct btrfs_io_ctl *io_ctl, struct list_head *bitmap_list)
1249 {
1250         struct btrfs_free_space *entry, *next;
1251         int ret;
1252
1253         /* Write out the bitmaps */
1254         list_for_each_entry_safe(entry, next, bitmap_list, list) {
1255                 ret = io_ctl_add_bitmap(io_ctl, entry->bitmap);
1256                 if (ret)
1257                         return -ENOSPC;
1258                 list_del_init(&entry->list);
1259         }
1260
1261         return 0;
1262 }
1263
1264 static int flush_dirty_cache(struct inode *inode)
1265 {
1266         int ret;
1267
1268         ret = btrfs_wait_ordered_range(inode, 0, (u64)-1);
1269         if (ret)
1270                 clear_extent_bit(&BTRFS_I(inode)->io_tree, 0, inode->i_size - 1,
1271                                  EXTENT_DELALLOC, NULL);
1272
1273         return ret;
1274 }
1275
1276 static void noinline_for_stack
1277 cleanup_bitmap_list(struct list_head *bitmap_list)
1278 {
1279         struct btrfs_free_space *entry, *next;
1280
1281         list_for_each_entry_safe(entry, next, bitmap_list, list)
1282                 list_del_init(&entry->list);
1283 }
1284
1285 static void noinline_for_stack
1286 cleanup_write_cache_enospc(struct inode *inode,
1287                            struct btrfs_io_ctl *io_ctl,
1288                            struct extent_state **cached_state)
1289 {
1290         io_ctl_drop_pages(io_ctl);
1291         unlock_extent(&BTRFS_I(inode)->io_tree, 0, i_size_read(inode) - 1,
1292                       cached_state);
1293 }
1294
1295 static int __btrfs_wait_cache_io(struct btrfs_root *root,
1296                                  struct btrfs_trans_handle *trans,
1297                                  struct btrfs_block_group *block_group,
1298                                  struct btrfs_io_ctl *io_ctl,
1299                                  struct btrfs_path *path, u64 offset)
1300 {
1301         int ret;
1302         struct inode *inode = io_ctl->inode;
1303
1304         if (!inode)
1305                 return 0;
1306
1307         /* Flush the dirty pages in the cache file. */
1308         ret = flush_dirty_cache(inode);
1309         if (ret)
1310                 goto out;
1311
1312         /* Update the cache item to tell everyone this cache file is valid. */
1313         ret = update_cache_item(trans, root, inode, path, offset,
1314                                 io_ctl->entries, io_ctl->bitmaps);
1315 out:
1316         if (ret) {
1317                 invalidate_inode_pages2(inode->i_mapping);
1318                 BTRFS_I(inode)->generation = 0;
1319                 if (block_group)
1320                         btrfs_debug(root->fs_info,
1321           "failed to write free space cache for block group %llu error %d",
1322                                   block_group->start, ret);
1323         }
1324         btrfs_update_inode(trans, root, BTRFS_I(inode));
1325
1326         if (block_group) {
1327                 /* the dirty list is protected by the dirty_bgs_lock */
1328                 spin_lock(&trans->transaction->dirty_bgs_lock);
1329
1330                 /* the disk_cache_state is protected by the block group lock */
1331                 spin_lock(&block_group->lock);
1332
1333                 /*
1334                  * only mark this as written if we didn't get put back on
1335                  * the dirty list while waiting for IO.   Otherwise our
1336                  * cache state won't be right, and we won't get written again
1337                  */
1338                 if (!ret && list_empty(&block_group->dirty_list))
1339                         block_group->disk_cache_state = BTRFS_DC_WRITTEN;
1340                 else if (ret)
1341                         block_group->disk_cache_state = BTRFS_DC_ERROR;
1342
1343                 spin_unlock(&block_group->lock);
1344                 spin_unlock(&trans->transaction->dirty_bgs_lock);
1345                 io_ctl->inode = NULL;
1346                 iput(inode);
1347         }
1348
1349         return ret;
1350
1351 }
1352
1353 int btrfs_wait_cache_io(struct btrfs_trans_handle *trans,
1354                         struct btrfs_block_group *block_group,
1355                         struct btrfs_path *path)
1356 {
1357         return __btrfs_wait_cache_io(block_group->fs_info->tree_root, trans,
1358                                      block_group, &block_group->io_ctl,
1359                                      path, block_group->start);
1360 }
1361
1362 /*
1363  * Write out cached info to an inode.
1364  *
1365  * @root:        root the inode belongs to
1366  * @inode:       freespace inode we are writing out
1367  * @ctl:         free space cache we are going to write out
1368  * @block_group: block_group for this cache if it belongs to a block_group
1369  * @io_ctl:      holds context for the io
1370  * @trans:       the trans handle
1371  *
1372  * This function writes out a free space cache struct to disk for quick recovery
1373  * on mount.  This will return 0 if it was successful in writing the cache out,
1374  * or an errno if it was not.
1375  */
1376 static int __btrfs_write_out_cache(struct btrfs_root *root, struct inode *inode,
1377                                    struct btrfs_free_space_ctl *ctl,
1378                                    struct btrfs_block_group *block_group,
1379                                    struct btrfs_io_ctl *io_ctl,
1380                                    struct btrfs_trans_handle *trans)
1381 {
1382         struct extent_state *cached_state = NULL;
1383         LIST_HEAD(bitmap_list);
1384         int entries = 0;
1385         int bitmaps = 0;
1386         int ret;
1387         int must_iput = 0;
1388
1389         if (!i_size_read(inode))
1390                 return -EIO;
1391
1392         WARN_ON(io_ctl->pages);
1393         ret = io_ctl_init(io_ctl, inode, 1);
1394         if (ret)
1395                 return ret;
1396
1397         if (block_group && (block_group->flags & BTRFS_BLOCK_GROUP_DATA)) {
1398                 down_write(&block_group->data_rwsem);
1399                 spin_lock(&block_group->lock);
1400                 if (block_group->delalloc_bytes) {
1401                         block_group->disk_cache_state = BTRFS_DC_WRITTEN;
1402                         spin_unlock(&block_group->lock);
1403                         up_write(&block_group->data_rwsem);
1404                         BTRFS_I(inode)->generation = 0;
1405                         ret = 0;
1406                         must_iput = 1;
1407                         goto out;
1408                 }
1409                 spin_unlock(&block_group->lock);
1410         }
1411
1412         /* Lock all pages first so we can lock the extent safely. */
1413         ret = io_ctl_prepare_pages(io_ctl, false);
1414         if (ret)
1415                 goto out_unlock;
1416
1417         lock_extent(&BTRFS_I(inode)->io_tree, 0, i_size_read(inode) - 1,
1418                     &cached_state);
1419
1420         io_ctl_set_generation(io_ctl, trans->transid);
1421
1422         mutex_lock(&ctl->cache_writeout_mutex);
1423         /* Write out the extent entries in the free space cache */
1424         spin_lock(&ctl->tree_lock);
1425         ret = write_cache_extent_entries(io_ctl, ctl,
1426                                          block_group, &entries, &bitmaps,
1427                                          &bitmap_list);
1428         if (ret)
1429                 goto out_nospc_locked;
1430
1431         /*
1432          * Some spaces that are freed in the current transaction are pinned,
1433          * they will be added into free space cache after the transaction is
1434          * committed, we shouldn't lose them.
1435          *
1436          * If this changes while we are working we'll get added back to
1437          * the dirty list and redo it.  No locking needed
1438          */
1439         ret = write_pinned_extent_entries(trans, block_group, io_ctl, &entries);
1440         if (ret)
1441                 goto out_nospc_locked;
1442
1443         /*
1444          * At last, we write out all the bitmaps and keep cache_writeout_mutex
1445          * locked while doing it because a concurrent trim can be manipulating
1446          * or freeing the bitmap.
1447          */
1448         ret = write_bitmap_entries(io_ctl, &bitmap_list);
1449         spin_unlock(&ctl->tree_lock);
1450         mutex_unlock(&ctl->cache_writeout_mutex);
1451         if (ret)
1452                 goto out_nospc;
1453
1454         /* Zero out the rest of the pages just to make sure */
1455         io_ctl_zero_remaining_pages(io_ctl);
1456
1457         /* Everything is written out, now we dirty the pages in the file. */
1458         ret = btrfs_dirty_pages(BTRFS_I(inode), io_ctl->pages,
1459                                 io_ctl->num_pages, 0, i_size_read(inode),
1460                                 &cached_state, false);
1461         if (ret)
1462                 goto out_nospc;
1463
1464         if (block_group && (block_group->flags & BTRFS_BLOCK_GROUP_DATA))
1465                 up_write(&block_group->data_rwsem);
1466         /*
1467          * Release the pages and unlock the extent, we will flush
1468          * them out later
1469          */
1470         io_ctl_drop_pages(io_ctl);
1471         io_ctl_free(io_ctl);
1472
1473         unlock_extent(&BTRFS_I(inode)->io_tree, 0, i_size_read(inode) - 1,
1474                       &cached_state);
1475
1476         /*
1477          * at this point the pages are under IO and we're happy,
1478          * The caller is responsible for waiting on them and updating
1479          * the cache and the inode
1480          */
1481         io_ctl->entries = entries;
1482         io_ctl->bitmaps = bitmaps;
1483
1484         ret = btrfs_fdatawrite_range(inode, 0, (u64)-1);
1485         if (ret)
1486                 goto out;
1487
1488         return 0;
1489
1490 out_nospc_locked:
1491         cleanup_bitmap_list(&bitmap_list);
1492         spin_unlock(&ctl->tree_lock);
1493         mutex_unlock(&ctl->cache_writeout_mutex);
1494
1495 out_nospc:
1496         cleanup_write_cache_enospc(inode, io_ctl, &cached_state);
1497
1498 out_unlock:
1499         if (block_group && (block_group->flags & BTRFS_BLOCK_GROUP_DATA))
1500                 up_write(&block_group->data_rwsem);
1501
1502 out:
1503         io_ctl->inode = NULL;
1504         io_ctl_free(io_ctl);
1505         if (ret) {
1506                 invalidate_inode_pages2(inode->i_mapping);
1507                 BTRFS_I(inode)->generation = 0;
1508         }
1509         btrfs_update_inode(trans, root, BTRFS_I(inode));
1510         if (must_iput)
1511                 iput(inode);
1512         return ret;
1513 }
1514
1515 int btrfs_write_out_cache(struct btrfs_trans_handle *trans,
1516                           struct btrfs_block_group *block_group,
1517                           struct btrfs_path *path)
1518 {
1519         struct btrfs_fs_info *fs_info = trans->fs_info;
1520         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
1521         struct inode *inode;
1522         int ret = 0;
1523
1524         spin_lock(&block_group->lock);
1525         if (block_group->disk_cache_state < BTRFS_DC_SETUP) {
1526                 spin_unlock(&block_group->lock);
1527                 return 0;
1528         }
1529         spin_unlock(&block_group->lock);
1530
1531         inode = lookup_free_space_inode(block_group, path);
1532         if (IS_ERR(inode))
1533                 return 0;
1534
1535         ret = __btrfs_write_out_cache(fs_info->tree_root, inode, ctl,
1536                                 block_group, &block_group->io_ctl, trans);
1537         if (ret) {
1538                 btrfs_debug(fs_info,
1539           "failed to write free space cache for block group %llu error %d",
1540                           block_group->start, ret);
1541                 spin_lock(&block_group->lock);
1542                 block_group->disk_cache_state = BTRFS_DC_ERROR;
1543                 spin_unlock(&block_group->lock);
1544
1545                 block_group->io_ctl.inode = NULL;
1546                 iput(inode);
1547         }
1548
1549         /*
1550          * if ret == 0 the caller is expected to call btrfs_wait_cache_io
1551          * to wait for IO and put the inode
1552          */
1553
1554         return ret;
1555 }
1556
1557 static inline unsigned long offset_to_bit(u64 bitmap_start, u32 unit,
1558                                           u64 offset)
1559 {
1560         ASSERT(offset >= bitmap_start);
1561         offset -= bitmap_start;
1562         return (unsigned long)(div_u64(offset, unit));
1563 }
1564
1565 static inline unsigned long bytes_to_bits(u64 bytes, u32 unit)
1566 {
1567         return (unsigned long)(div_u64(bytes, unit));
1568 }
1569
1570 static inline u64 offset_to_bitmap(struct btrfs_free_space_ctl *ctl,
1571                                    u64 offset)
1572 {
1573         u64 bitmap_start;
1574         u64 bytes_per_bitmap;
1575
1576         bytes_per_bitmap = BITS_PER_BITMAP * ctl->unit;
1577         bitmap_start = offset - ctl->start;
1578         bitmap_start = div64_u64(bitmap_start, bytes_per_bitmap);
1579         bitmap_start *= bytes_per_bitmap;
1580         bitmap_start += ctl->start;
1581
1582         return bitmap_start;
1583 }
1584
1585 static int tree_insert_offset(struct btrfs_free_space_ctl *ctl,
1586                               struct btrfs_free_cluster *cluster,
1587                               struct btrfs_free_space *new_entry)
1588 {
1589         struct rb_root *root;
1590         struct rb_node **p;
1591         struct rb_node *parent = NULL;
1592
1593         lockdep_assert_held(&ctl->tree_lock);
1594
1595         if (cluster) {
1596                 lockdep_assert_held(&cluster->lock);
1597                 root = &cluster->root;
1598         } else {
1599                 root = &ctl->free_space_offset;
1600         }
1601
1602         p = &root->rb_node;
1603
1604         while (*p) {
1605                 struct btrfs_free_space *info;
1606
1607                 parent = *p;
1608                 info = rb_entry(parent, struct btrfs_free_space, offset_index);
1609
1610                 if (new_entry->offset < info->offset) {
1611                         p = &(*p)->rb_left;
1612                 } else if (new_entry->offset > info->offset) {
1613                         p = &(*p)->rb_right;
1614                 } else {
1615                         /*
1616                          * we could have a bitmap entry and an extent entry
1617                          * share the same offset.  If this is the case, we want
1618                          * the extent entry to always be found first if we do a
1619                          * linear search through the tree, since we want to have
1620                          * the quickest allocation time, and allocating from an
1621                          * extent is faster than allocating from a bitmap.  So
1622                          * if we're inserting a bitmap and we find an entry at
1623                          * this offset, we want to go right, or after this entry
1624                          * logically.  If we are inserting an extent and we've
1625                          * found a bitmap, we want to go left, or before
1626                          * logically.
1627                          */
1628                         if (new_entry->bitmap) {
1629                                 if (info->bitmap) {
1630                                         WARN_ON_ONCE(1);
1631                                         return -EEXIST;
1632                                 }
1633                                 p = &(*p)->rb_right;
1634                         } else {
1635                                 if (!info->bitmap) {
1636                                         WARN_ON_ONCE(1);
1637                                         return -EEXIST;
1638                                 }
1639                                 p = &(*p)->rb_left;
1640                         }
1641                 }
1642         }
1643
1644         rb_link_node(&new_entry->offset_index, parent, p);
1645         rb_insert_color(&new_entry->offset_index, root);
1646
1647         return 0;
1648 }
1649
1650 /*
1651  * This is a little subtle.  We *only* have ->max_extent_size set if we actually
1652  * searched through the bitmap and figured out the largest ->max_extent_size,
1653  * otherwise it's 0.  In the case that it's 0 we don't want to tell the
1654  * allocator the wrong thing, we want to use the actual real max_extent_size
1655  * we've found already if it's larger, or we want to use ->bytes.
1656  *
1657  * This matters because find_free_space() will skip entries who's ->bytes is
1658  * less than the required bytes.  So if we didn't search down this bitmap, we
1659  * may pick some previous entry that has a smaller ->max_extent_size than we
1660  * have.  For example, assume we have two entries, one that has
1661  * ->max_extent_size set to 4K and ->bytes set to 1M.  A second entry hasn't set
1662  * ->max_extent_size yet, has ->bytes set to 8K and it's contiguous.  We will
1663  *  call into find_free_space(), and return with max_extent_size == 4K, because
1664  *  that first bitmap entry had ->max_extent_size set, but the second one did
1665  *  not.  If instead we returned 8K we'd come in searching for 8K, and find the
1666  *  8K contiguous range.
1667  *
1668  *  Consider the other case, we have 2 8K chunks in that second entry and still
1669  *  don't have ->max_extent_size set.  We'll return 16K, and the next time the
1670  *  allocator comes in it'll fully search our second bitmap, and this time it'll
1671  *  get an uptodate value of 8K as the maximum chunk size.  Then we'll get the
1672  *  right allocation the next loop through.
1673  */
1674 static inline u64 get_max_extent_size(const struct btrfs_free_space *entry)
1675 {
1676         if (entry->bitmap && entry->max_extent_size)
1677                 return entry->max_extent_size;
1678         return entry->bytes;
1679 }
1680
1681 /*
1682  * We want the largest entry to be leftmost, so this is inverted from what you'd
1683  * normally expect.
1684  */
1685 static bool entry_less(struct rb_node *node, const struct rb_node *parent)
1686 {
1687         const struct btrfs_free_space *entry, *exist;
1688
1689         entry = rb_entry(node, struct btrfs_free_space, bytes_index);
1690         exist = rb_entry(parent, struct btrfs_free_space, bytes_index);
1691         return get_max_extent_size(exist) < get_max_extent_size(entry);
1692 }
1693
1694 /*
1695  * searches the tree for the given offset.
1696  *
1697  * fuzzy - If this is set, then we are trying to make an allocation, and we just
1698  * want a section that has at least bytes size and comes at or after the given
1699  * offset.
1700  */
1701 static struct btrfs_free_space *
1702 tree_search_offset(struct btrfs_free_space_ctl *ctl,
1703                    u64 offset, int bitmap_only, int fuzzy)
1704 {
1705         struct rb_node *n = ctl->free_space_offset.rb_node;
1706         struct btrfs_free_space *entry = NULL, *prev = NULL;
1707
1708         lockdep_assert_held(&ctl->tree_lock);
1709
1710         /* find entry that is closest to the 'offset' */
1711         while (n) {
1712                 entry = rb_entry(n, struct btrfs_free_space, offset_index);
1713                 prev = entry;
1714
1715                 if (offset < entry->offset)
1716                         n = n->rb_left;
1717                 else if (offset > entry->offset)
1718                         n = n->rb_right;
1719                 else
1720                         break;
1721
1722                 entry = NULL;
1723         }
1724
1725         if (bitmap_only) {
1726                 if (!entry)
1727                         return NULL;
1728                 if (entry->bitmap)
1729                         return entry;
1730
1731                 /*
1732                  * bitmap entry and extent entry may share same offset,
1733                  * in that case, bitmap entry comes after extent entry.
1734                  */
1735                 n = rb_next(n);
1736                 if (!n)
1737                         return NULL;
1738                 entry = rb_entry(n, struct btrfs_free_space, offset_index);
1739                 if (entry->offset != offset)
1740                         return NULL;
1741
1742                 WARN_ON(!entry->bitmap);
1743                 return entry;
1744         } else if (entry) {
1745                 if (entry->bitmap) {
1746                         /*
1747                          * if previous extent entry covers the offset,
1748                          * we should return it instead of the bitmap entry
1749                          */
1750                         n = rb_prev(&entry->offset_index);
1751                         if (n) {
1752                                 prev = rb_entry(n, struct btrfs_free_space,
1753                                                 offset_index);
1754                                 if (!prev->bitmap &&
1755                                     prev->offset + prev->bytes > offset)
1756                                         entry = prev;
1757                         }
1758                 }
1759                 return entry;
1760         }
1761
1762         if (!prev)
1763                 return NULL;
1764
1765         /* find last entry before the 'offset' */
1766         entry = prev;
1767         if (entry->offset > offset) {
1768                 n = rb_prev(&entry->offset_index);
1769                 if (n) {
1770                         entry = rb_entry(n, struct btrfs_free_space,
1771                                         offset_index);
1772                         ASSERT(entry->offset <= offset);
1773                 } else {
1774                         if (fuzzy)
1775                                 return entry;
1776                         else
1777                                 return NULL;
1778                 }
1779         }
1780
1781         if (entry->bitmap) {
1782                 n = rb_prev(&entry->offset_index);
1783                 if (n) {
1784                         prev = rb_entry(n, struct btrfs_free_space,
1785                                         offset_index);
1786                         if (!prev->bitmap &&
1787                             prev->offset + prev->bytes > offset)
1788                                 return prev;
1789                 }
1790                 if (entry->offset + BITS_PER_BITMAP * ctl->unit > offset)
1791                         return entry;
1792         } else if (entry->offset + entry->bytes > offset)
1793                 return entry;
1794
1795         if (!fuzzy)
1796                 return NULL;
1797
1798         while (1) {
1799                 n = rb_next(&entry->offset_index);
1800                 if (!n)
1801                         return NULL;
1802                 entry = rb_entry(n, struct btrfs_free_space, offset_index);
1803                 if (entry->bitmap) {
1804                         if (entry->offset + BITS_PER_BITMAP *
1805                             ctl->unit > offset)
1806                                 break;
1807                 } else {
1808                         if (entry->offset + entry->bytes > offset)
1809                                 break;
1810                 }
1811         }
1812         return entry;
1813 }
1814
1815 static inline void unlink_free_space(struct btrfs_free_space_ctl *ctl,
1816                                      struct btrfs_free_space *info,
1817                                      bool update_stat)
1818 {
1819         lockdep_assert_held(&ctl->tree_lock);
1820
1821         rb_erase(&info->offset_index, &ctl->free_space_offset);
1822         rb_erase_cached(&info->bytes_index, &ctl->free_space_bytes);
1823         ctl->free_extents--;
1824
1825         if (!info->bitmap && !btrfs_free_space_trimmed(info)) {
1826                 ctl->discardable_extents[BTRFS_STAT_CURR]--;
1827                 ctl->discardable_bytes[BTRFS_STAT_CURR] -= info->bytes;
1828         }
1829
1830         if (update_stat)
1831                 ctl->free_space -= info->bytes;
1832 }
1833
1834 static int link_free_space(struct btrfs_free_space_ctl *ctl,
1835                            struct btrfs_free_space *info)
1836 {
1837         int ret = 0;
1838
1839         lockdep_assert_held(&ctl->tree_lock);
1840
1841         ASSERT(info->bytes || info->bitmap);
1842         ret = tree_insert_offset(ctl, NULL, info);
1843         if (ret)
1844                 return ret;
1845
1846         rb_add_cached(&info->bytes_index, &ctl->free_space_bytes, entry_less);
1847
1848         if (!info->bitmap && !btrfs_free_space_trimmed(info)) {
1849                 ctl->discardable_extents[BTRFS_STAT_CURR]++;
1850                 ctl->discardable_bytes[BTRFS_STAT_CURR] += info->bytes;
1851         }
1852
1853         ctl->free_space += info->bytes;
1854         ctl->free_extents++;
1855         return ret;
1856 }
1857
1858 static void relink_bitmap_entry(struct btrfs_free_space_ctl *ctl,
1859                                 struct btrfs_free_space *info)
1860 {
1861         ASSERT(info->bitmap);
1862
1863         /*
1864          * If our entry is empty it's because we're on a cluster and we don't
1865          * want to re-link it into our ctl bytes index.
1866          */
1867         if (RB_EMPTY_NODE(&info->bytes_index))
1868                 return;
1869
1870         lockdep_assert_held(&ctl->tree_lock);
1871
1872         rb_erase_cached(&info->bytes_index, &ctl->free_space_bytes);
1873         rb_add_cached(&info->bytes_index, &ctl->free_space_bytes, entry_less);
1874 }
1875
1876 static inline void bitmap_clear_bits(struct btrfs_free_space_ctl *ctl,
1877                                      struct btrfs_free_space *info,
1878                                      u64 offset, u64 bytes, bool update_stat)
1879 {
1880         unsigned long start, count, end;
1881         int extent_delta = -1;
1882
1883         start = offset_to_bit(info->offset, ctl->unit, offset);
1884         count = bytes_to_bits(bytes, ctl->unit);
1885         end = start + count;
1886         ASSERT(end <= BITS_PER_BITMAP);
1887
1888         bitmap_clear(info->bitmap, start, count);
1889
1890         info->bytes -= bytes;
1891         if (info->max_extent_size > ctl->unit)
1892                 info->max_extent_size = 0;
1893
1894         relink_bitmap_entry(ctl, info);
1895
1896         if (start && test_bit(start - 1, info->bitmap))
1897                 extent_delta++;
1898
1899         if (end < BITS_PER_BITMAP && test_bit(end, info->bitmap))
1900                 extent_delta++;
1901
1902         info->bitmap_extents += extent_delta;
1903         if (!btrfs_free_space_trimmed(info)) {
1904                 ctl->discardable_extents[BTRFS_STAT_CURR] += extent_delta;
1905                 ctl->discardable_bytes[BTRFS_STAT_CURR] -= bytes;
1906         }
1907
1908         if (update_stat)
1909                 ctl->free_space -= bytes;
1910 }
1911
1912 static void bitmap_set_bits(struct btrfs_free_space_ctl *ctl,
1913                             struct btrfs_free_space *info, u64 offset,
1914                             u64 bytes)
1915 {
1916         unsigned long start, count, end;
1917         int extent_delta = 1;
1918
1919         start = offset_to_bit(info->offset, ctl->unit, offset);
1920         count = bytes_to_bits(bytes, ctl->unit);
1921         end = start + count;
1922         ASSERT(end <= BITS_PER_BITMAP);
1923
1924         bitmap_set(info->bitmap, start, count);
1925
1926         /*
1927          * We set some bytes, we have no idea what the max extent size is
1928          * anymore.
1929          */
1930         info->max_extent_size = 0;
1931         info->bytes += bytes;
1932         ctl->free_space += bytes;
1933
1934         relink_bitmap_entry(ctl, info);
1935
1936         if (start && test_bit(start - 1, info->bitmap))
1937                 extent_delta--;
1938
1939         if (end < BITS_PER_BITMAP && test_bit(end, info->bitmap))
1940                 extent_delta--;
1941
1942         info->bitmap_extents += extent_delta;
1943         if (!btrfs_free_space_trimmed(info)) {
1944                 ctl->discardable_extents[BTRFS_STAT_CURR] += extent_delta;
1945                 ctl->discardable_bytes[BTRFS_STAT_CURR] += bytes;
1946         }
1947 }
1948
1949 /*
1950  * If we can not find suitable extent, we will use bytes to record
1951  * the size of the max extent.
1952  */
1953 static int search_bitmap(struct btrfs_free_space_ctl *ctl,
1954                          struct btrfs_free_space *bitmap_info, u64 *offset,
1955                          u64 *bytes, bool for_alloc)
1956 {
1957         unsigned long found_bits = 0;
1958         unsigned long max_bits = 0;
1959         unsigned long bits, i;
1960         unsigned long next_zero;
1961         unsigned long extent_bits;
1962
1963         /*
1964          * Skip searching the bitmap if we don't have a contiguous section that
1965          * is large enough for this allocation.
1966          */
1967         if (for_alloc &&
1968             bitmap_info->max_extent_size &&
1969             bitmap_info->max_extent_size < *bytes) {
1970                 *bytes = bitmap_info->max_extent_size;
1971                 return -1;
1972         }
1973
1974         i = offset_to_bit(bitmap_info->offset, ctl->unit,
1975                           max_t(u64, *offset, bitmap_info->offset));
1976         bits = bytes_to_bits(*bytes, ctl->unit);
1977
1978         for_each_set_bit_from(i, bitmap_info->bitmap, BITS_PER_BITMAP) {
1979                 if (for_alloc && bits == 1) {
1980                         found_bits = 1;
1981                         break;
1982                 }
1983                 next_zero = find_next_zero_bit(bitmap_info->bitmap,
1984                                                BITS_PER_BITMAP, i);
1985                 extent_bits = next_zero - i;
1986                 if (extent_bits >= bits) {
1987                         found_bits = extent_bits;
1988                         break;
1989                 } else if (extent_bits > max_bits) {
1990                         max_bits = extent_bits;
1991                 }
1992                 i = next_zero;
1993         }
1994
1995         if (found_bits) {
1996                 *offset = (u64)(i * ctl->unit) + bitmap_info->offset;
1997                 *bytes = (u64)(found_bits) * ctl->unit;
1998                 return 0;
1999         }
2000
2001         *bytes = (u64)(max_bits) * ctl->unit;
2002         bitmap_info->max_extent_size = *bytes;
2003         relink_bitmap_entry(ctl, bitmap_info);
2004         return -1;
2005 }
2006
2007 /* Cache the size of the max extent in bytes */
2008 static struct btrfs_free_space *
2009 find_free_space(struct btrfs_free_space_ctl *ctl, u64 *offset, u64 *bytes,
2010                 unsigned long align, u64 *max_extent_size, bool use_bytes_index)
2011 {
2012         struct btrfs_free_space *entry;
2013         struct rb_node *node;
2014         u64 tmp;
2015         u64 align_off;
2016         int ret;
2017
2018         if (!ctl->free_space_offset.rb_node)
2019                 goto out;
2020 again:
2021         if (use_bytes_index) {
2022                 node = rb_first_cached(&ctl->free_space_bytes);
2023         } else {
2024                 entry = tree_search_offset(ctl, offset_to_bitmap(ctl, *offset),
2025                                            0, 1);
2026                 if (!entry)
2027                         goto out;
2028                 node = &entry->offset_index;
2029         }
2030
2031         for (; node; node = rb_next(node)) {
2032                 if (use_bytes_index)
2033                         entry = rb_entry(node, struct btrfs_free_space,
2034                                          bytes_index);
2035                 else
2036                         entry = rb_entry(node, struct btrfs_free_space,
2037                                          offset_index);
2038
2039                 /*
2040                  * If we are using the bytes index then all subsequent entries
2041                  * in this tree are going to be < bytes, so simply set the max
2042                  * extent size and exit the loop.
2043                  *
2044                  * If we're using the offset index then we need to keep going
2045                  * through the rest of the tree.
2046                  */
2047                 if (entry->bytes < *bytes) {
2048                         *max_extent_size = max(get_max_extent_size(entry),
2049                                                *max_extent_size);
2050                         if (use_bytes_index)
2051                                 break;
2052                         continue;
2053                 }
2054
2055                 /* make sure the space returned is big enough
2056                  * to match our requested alignment
2057                  */
2058                 if (*bytes >= align) {
2059                         tmp = entry->offset - ctl->start + align - 1;
2060                         tmp = div64_u64(tmp, align);
2061                         tmp = tmp * align + ctl->start;
2062                         align_off = tmp - entry->offset;
2063                 } else {
2064                         align_off = 0;
2065                         tmp = entry->offset;
2066                 }
2067
2068                 /*
2069                  * We don't break here if we're using the bytes index because we
2070                  * may have another entry that has the correct alignment that is
2071                  * the right size, so we don't want to miss that possibility.
2072                  * At worst this adds another loop through the logic, but if we
2073                  * broke here we could prematurely ENOSPC.
2074                  */
2075                 if (entry->bytes < *bytes + align_off) {
2076                         *max_extent_size = max(get_max_extent_size(entry),
2077                                                *max_extent_size);
2078                         continue;
2079                 }
2080
2081                 if (entry->bitmap) {
2082                         struct rb_node *old_next = rb_next(node);
2083                         u64 size = *bytes;
2084
2085                         ret = search_bitmap(ctl, entry, &tmp, &size, true);
2086                         if (!ret) {
2087                                 *offset = tmp;
2088                                 *bytes = size;
2089                                 return entry;
2090                         } else {
2091                                 *max_extent_size =
2092                                         max(get_max_extent_size(entry),
2093                                             *max_extent_size);
2094                         }
2095
2096                         /*
2097                          * The bitmap may have gotten re-arranged in the space
2098                          * index here because the max_extent_size may have been
2099                          * updated.  Start from the beginning again if this
2100                          * happened.
2101                          */
2102                         if (use_bytes_index && old_next != rb_next(node))
2103                                 goto again;
2104                         continue;
2105                 }
2106
2107                 *offset = tmp;
2108                 *bytes = entry->bytes - align_off;
2109                 return entry;
2110         }
2111 out:
2112         return NULL;
2113 }
2114
2115 static void add_new_bitmap(struct btrfs_free_space_ctl *ctl,
2116                            struct btrfs_free_space *info, u64 offset)
2117 {
2118         info->offset = offset_to_bitmap(ctl, offset);
2119         info->bytes = 0;
2120         info->bitmap_extents = 0;
2121         INIT_LIST_HEAD(&info->list);
2122         link_free_space(ctl, info);
2123         ctl->total_bitmaps++;
2124         recalculate_thresholds(ctl);
2125 }
2126
2127 static void free_bitmap(struct btrfs_free_space_ctl *ctl,
2128                         struct btrfs_free_space *bitmap_info)
2129 {
2130         /*
2131          * Normally when this is called, the bitmap is completely empty. However,
2132          * if we are blowing up the free space cache for one reason or another
2133          * via __btrfs_remove_free_space_cache(), then it may not be freed and
2134          * we may leave stats on the table.
2135          */
2136         if (bitmap_info->bytes && !btrfs_free_space_trimmed(bitmap_info)) {
2137                 ctl->discardable_extents[BTRFS_STAT_CURR] -=
2138                         bitmap_info->bitmap_extents;
2139                 ctl->discardable_bytes[BTRFS_STAT_CURR] -= bitmap_info->bytes;
2140
2141         }
2142         unlink_free_space(ctl, bitmap_info, true);
2143         kmem_cache_free(btrfs_free_space_bitmap_cachep, bitmap_info->bitmap);
2144         kmem_cache_free(btrfs_free_space_cachep, bitmap_info);
2145         ctl->total_bitmaps--;
2146         recalculate_thresholds(ctl);
2147 }
2148
2149 static noinline int remove_from_bitmap(struct btrfs_free_space_ctl *ctl,
2150                               struct btrfs_free_space *bitmap_info,
2151                               u64 *offset, u64 *bytes)
2152 {
2153         u64 end;
2154         u64 search_start, search_bytes;
2155         int ret;
2156
2157 again:
2158         end = bitmap_info->offset + (u64)(BITS_PER_BITMAP * ctl->unit) - 1;
2159
2160         /*
2161          * We need to search for bits in this bitmap.  We could only cover some
2162          * of the extent in this bitmap thanks to how we add space, so we need
2163          * to search for as much as it as we can and clear that amount, and then
2164          * go searching for the next bit.
2165          */
2166         search_start = *offset;
2167         search_bytes = ctl->unit;
2168         search_bytes = min(search_bytes, end - search_start + 1);
2169         ret = search_bitmap(ctl, bitmap_info, &search_start, &search_bytes,
2170                             false);
2171         if (ret < 0 || search_start != *offset)
2172                 return -EINVAL;
2173
2174         /* We may have found more bits than what we need */
2175         search_bytes = min(search_bytes, *bytes);
2176
2177         /* Cannot clear past the end of the bitmap */
2178         search_bytes = min(search_bytes, end - search_start + 1);
2179
2180         bitmap_clear_bits(ctl, bitmap_info, search_start, search_bytes, true);
2181         *offset += search_bytes;
2182         *bytes -= search_bytes;
2183
2184         if (*bytes) {
2185                 struct rb_node *next = rb_next(&bitmap_info->offset_index);
2186                 if (!bitmap_info->bytes)
2187                         free_bitmap(ctl, bitmap_info);
2188
2189                 /*
2190                  * no entry after this bitmap, but we still have bytes to
2191                  * remove, so something has gone wrong.
2192                  */
2193                 if (!next)
2194                         return -EINVAL;
2195
2196                 bitmap_info = rb_entry(next, struct btrfs_free_space,
2197                                        offset_index);
2198
2199                 /*
2200                  * if the next entry isn't a bitmap we need to return to let the
2201                  * extent stuff do its work.
2202                  */
2203                 if (!bitmap_info->bitmap)
2204                         return -EAGAIN;
2205
2206                 /*
2207                  * Ok the next item is a bitmap, but it may not actually hold
2208                  * the information for the rest of this free space stuff, so
2209                  * look for it, and if we don't find it return so we can try
2210                  * everything over again.
2211                  */
2212                 search_start = *offset;
2213                 search_bytes = ctl->unit;
2214                 ret = search_bitmap(ctl, bitmap_info, &search_start,
2215                                     &search_bytes, false);
2216                 if (ret < 0 || search_start != *offset)
2217                         return -EAGAIN;
2218
2219                 goto again;
2220         } else if (!bitmap_info->bytes)
2221                 free_bitmap(ctl, bitmap_info);
2222
2223         return 0;
2224 }
2225
2226 static u64 add_bytes_to_bitmap(struct btrfs_free_space_ctl *ctl,
2227                                struct btrfs_free_space *info, u64 offset,
2228                                u64 bytes, enum btrfs_trim_state trim_state)
2229 {
2230         u64 bytes_to_set = 0;
2231         u64 end;
2232
2233         /*
2234          * This is a tradeoff to make bitmap trim state minimal.  We mark the
2235          * whole bitmap untrimmed if at any point we add untrimmed regions.
2236          */
2237         if (trim_state == BTRFS_TRIM_STATE_UNTRIMMED) {
2238                 if (btrfs_free_space_trimmed(info)) {
2239                         ctl->discardable_extents[BTRFS_STAT_CURR] +=
2240                                 info->bitmap_extents;
2241                         ctl->discardable_bytes[BTRFS_STAT_CURR] += info->bytes;
2242                 }
2243                 info->trim_state = BTRFS_TRIM_STATE_UNTRIMMED;
2244         }
2245
2246         end = info->offset + (u64)(BITS_PER_BITMAP * ctl->unit);
2247
2248         bytes_to_set = min(end - offset, bytes);
2249
2250         bitmap_set_bits(ctl, info, offset, bytes_to_set);
2251
2252         return bytes_to_set;
2253
2254 }
2255
2256 static bool use_bitmap(struct btrfs_free_space_ctl *ctl,
2257                       struct btrfs_free_space *info)
2258 {
2259         struct btrfs_block_group *block_group = ctl->block_group;
2260         struct btrfs_fs_info *fs_info = block_group->fs_info;
2261         bool forced = false;
2262
2263 #ifdef CONFIG_BTRFS_DEBUG
2264         if (btrfs_should_fragment_free_space(block_group))
2265                 forced = true;
2266 #endif
2267
2268         /* This is a way to reclaim large regions from the bitmaps. */
2269         if (!forced && info->bytes >= FORCE_EXTENT_THRESHOLD)
2270                 return false;
2271
2272         /*
2273          * If we are below the extents threshold then we can add this as an
2274          * extent, and don't have to deal with the bitmap
2275          */
2276         if (!forced && ctl->free_extents < ctl->extents_thresh) {
2277                 /*
2278                  * If this block group has some small extents we don't want to
2279                  * use up all of our free slots in the cache with them, we want
2280                  * to reserve them to larger extents, however if we have plenty
2281                  * of cache left then go ahead an dadd them, no sense in adding
2282                  * the overhead of a bitmap if we don't have to.
2283                  */
2284                 if (info->bytes <= fs_info->sectorsize * 8) {
2285                         if (ctl->free_extents * 3 <= ctl->extents_thresh)
2286                                 return false;
2287                 } else {
2288                         return false;
2289                 }
2290         }
2291
2292         /*
2293          * The original block groups from mkfs can be really small, like 8
2294          * megabytes, so don't bother with a bitmap for those entries.  However
2295          * some block groups can be smaller than what a bitmap would cover but
2296          * are still large enough that they could overflow the 32k memory limit,
2297          * so allow those block groups to still be allowed to have a bitmap
2298          * entry.
2299          */
2300         if (((BITS_PER_BITMAP * ctl->unit) >> 1) > block_group->length)
2301                 return false;
2302
2303         return true;
2304 }
2305
2306 static const struct btrfs_free_space_op free_space_op = {
2307         .use_bitmap             = use_bitmap,
2308 };
2309
2310 static int insert_into_bitmap(struct btrfs_free_space_ctl *ctl,
2311                               struct btrfs_free_space *info)
2312 {
2313         struct btrfs_free_space *bitmap_info;
2314         struct btrfs_block_group *block_group = NULL;
2315         int added = 0;
2316         u64 bytes, offset, bytes_added;
2317         enum btrfs_trim_state trim_state;
2318         int ret;
2319
2320         bytes = info->bytes;
2321         offset = info->offset;
2322         trim_state = info->trim_state;
2323
2324         if (!ctl->op->use_bitmap(ctl, info))
2325                 return 0;
2326
2327         if (ctl->op == &free_space_op)
2328                 block_group = ctl->block_group;
2329 again:
2330         /*
2331          * Since we link bitmaps right into the cluster we need to see if we
2332          * have a cluster here, and if so and it has our bitmap we need to add
2333          * the free space to that bitmap.
2334          */
2335         if (block_group && !list_empty(&block_group->cluster_list)) {
2336                 struct btrfs_free_cluster *cluster;
2337                 struct rb_node *node;
2338                 struct btrfs_free_space *entry;
2339
2340                 cluster = list_entry(block_group->cluster_list.next,
2341                                      struct btrfs_free_cluster,
2342                                      block_group_list);
2343                 spin_lock(&cluster->lock);
2344                 node = rb_first(&cluster->root);
2345                 if (!node) {
2346                         spin_unlock(&cluster->lock);
2347                         goto no_cluster_bitmap;
2348                 }
2349
2350                 entry = rb_entry(node, struct btrfs_free_space, offset_index);
2351                 if (!entry->bitmap) {
2352                         spin_unlock(&cluster->lock);
2353                         goto no_cluster_bitmap;
2354                 }
2355
2356                 if (entry->offset == offset_to_bitmap(ctl, offset)) {
2357                         bytes_added = add_bytes_to_bitmap(ctl, entry, offset,
2358                                                           bytes, trim_state);
2359                         bytes -= bytes_added;
2360                         offset += bytes_added;
2361                 }
2362                 spin_unlock(&cluster->lock);
2363                 if (!bytes) {
2364                         ret = 1;
2365                         goto out;
2366                 }
2367         }
2368
2369 no_cluster_bitmap:
2370         bitmap_info = tree_search_offset(ctl, offset_to_bitmap(ctl, offset),
2371                                          1, 0);
2372         if (!bitmap_info) {
2373                 ASSERT(added == 0);
2374                 goto new_bitmap;
2375         }
2376
2377         bytes_added = add_bytes_to_bitmap(ctl, bitmap_info, offset, bytes,
2378                                           trim_state);
2379         bytes -= bytes_added;
2380         offset += bytes_added;
2381         added = 0;
2382
2383         if (!bytes) {
2384                 ret = 1;
2385                 goto out;
2386         } else
2387                 goto again;
2388
2389 new_bitmap:
2390         if (info && info->bitmap) {
2391                 add_new_bitmap(ctl, info, offset);
2392                 added = 1;
2393                 info = NULL;
2394                 goto again;
2395         } else {
2396                 spin_unlock(&ctl->tree_lock);
2397
2398                 /* no pre-allocated info, allocate a new one */
2399                 if (!info) {
2400                         info = kmem_cache_zalloc(btrfs_free_space_cachep,
2401                                                  GFP_NOFS);
2402                         if (!info) {
2403                                 spin_lock(&ctl->tree_lock);
2404                                 ret = -ENOMEM;
2405                                 goto out;
2406                         }
2407                 }
2408
2409                 /* allocate the bitmap */
2410                 info->bitmap = kmem_cache_zalloc(btrfs_free_space_bitmap_cachep,
2411                                                  GFP_NOFS);
2412                 info->trim_state = BTRFS_TRIM_STATE_TRIMMED;
2413                 spin_lock(&ctl->tree_lock);
2414                 if (!info->bitmap) {
2415                         ret = -ENOMEM;
2416                         goto out;
2417                 }
2418                 goto again;
2419         }
2420
2421 out:
2422         if (info) {
2423                 if (info->bitmap)
2424                         kmem_cache_free(btrfs_free_space_bitmap_cachep,
2425                                         info->bitmap);
2426                 kmem_cache_free(btrfs_free_space_cachep, info);
2427         }
2428
2429         return ret;
2430 }
2431
2432 /*
2433  * Free space merging rules:
2434  *  1) Merge trimmed areas together
2435  *  2) Let untrimmed areas coalesce with trimmed areas
2436  *  3) Always pull neighboring regions from bitmaps
2437  *
2438  * The above rules are for when we merge free space based on btrfs_trim_state.
2439  * Rules 2 and 3 are subtle because they are suboptimal, but are done for the
2440  * same reason: to promote larger extent regions which makes life easier for
2441  * find_free_extent().  Rule 2 enables coalescing based on the common path
2442  * being returning free space from btrfs_finish_extent_commit().  So when free
2443  * space is trimmed, it will prevent aggregating trimmed new region and
2444  * untrimmed regions in the rb_tree.  Rule 3 is purely to obtain larger extents
2445  * and provide find_free_extent() with the largest extents possible hoping for
2446  * the reuse path.
2447  */
2448 static bool try_merge_free_space(struct btrfs_free_space_ctl *ctl,
2449                           struct btrfs_free_space *info, bool update_stat)
2450 {
2451         struct btrfs_free_space *left_info = NULL;
2452         struct btrfs_free_space *right_info;
2453         bool merged = false;
2454         u64 offset = info->offset;
2455         u64 bytes = info->bytes;
2456         const bool is_trimmed = btrfs_free_space_trimmed(info);
2457         struct rb_node *right_prev = NULL;
2458
2459         /*
2460          * first we want to see if there is free space adjacent to the range we
2461          * are adding, if there is remove that struct and add a new one to
2462          * cover the entire range
2463          */
2464         right_info = tree_search_offset(ctl, offset + bytes, 0, 0);
2465         if (right_info)
2466                 right_prev = rb_prev(&right_info->offset_index);
2467
2468         if (right_prev)
2469                 left_info = rb_entry(right_prev, struct btrfs_free_space, offset_index);
2470         else if (!right_info)
2471                 left_info = tree_search_offset(ctl, offset - 1, 0, 0);
2472
2473         /* See try_merge_free_space() comment. */
2474         if (right_info && !right_info->bitmap &&
2475             (!is_trimmed || btrfs_free_space_trimmed(right_info))) {
2476                 unlink_free_space(ctl, right_info, update_stat);
2477                 info->bytes += right_info->bytes;
2478                 kmem_cache_free(btrfs_free_space_cachep, right_info);
2479                 merged = true;
2480         }
2481
2482         /* See try_merge_free_space() comment. */
2483         if (left_info && !left_info->bitmap &&
2484             left_info->offset + left_info->bytes == offset &&
2485             (!is_trimmed || btrfs_free_space_trimmed(left_info))) {
2486                 unlink_free_space(ctl, left_info, update_stat);
2487                 info->offset = left_info->offset;
2488                 info->bytes += left_info->bytes;
2489                 kmem_cache_free(btrfs_free_space_cachep, left_info);
2490                 merged = true;
2491         }
2492
2493         return merged;
2494 }
2495
2496 static bool steal_from_bitmap_to_end(struct btrfs_free_space_ctl *ctl,
2497                                      struct btrfs_free_space *info,
2498                                      bool update_stat)
2499 {
2500         struct btrfs_free_space *bitmap;
2501         unsigned long i;
2502         unsigned long j;
2503         const u64 end = info->offset + info->bytes;
2504         const u64 bitmap_offset = offset_to_bitmap(ctl, end);
2505         u64 bytes;
2506
2507         bitmap = tree_search_offset(ctl, bitmap_offset, 1, 0);
2508         if (!bitmap)
2509                 return false;
2510
2511         i = offset_to_bit(bitmap->offset, ctl->unit, end);
2512         j = find_next_zero_bit(bitmap->bitmap, BITS_PER_BITMAP, i);
2513         if (j == i)
2514                 return false;
2515         bytes = (j - i) * ctl->unit;
2516         info->bytes += bytes;
2517
2518         /* See try_merge_free_space() comment. */
2519         if (!btrfs_free_space_trimmed(bitmap))
2520                 info->trim_state = BTRFS_TRIM_STATE_UNTRIMMED;
2521
2522         bitmap_clear_bits(ctl, bitmap, end, bytes, update_stat);
2523
2524         if (!bitmap->bytes)
2525                 free_bitmap(ctl, bitmap);
2526
2527         return true;
2528 }
2529
2530 static bool steal_from_bitmap_to_front(struct btrfs_free_space_ctl *ctl,
2531                                        struct btrfs_free_space *info,
2532                                        bool update_stat)
2533 {
2534         struct btrfs_free_space *bitmap;
2535         u64 bitmap_offset;
2536         unsigned long i;
2537         unsigned long j;
2538         unsigned long prev_j;
2539         u64 bytes;
2540
2541         bitmap_offset = offset_to_bitmap(ctl, info->offset);
2542         /* If we're on a boundary, try the previous logical bitmap. */
2543         if (bitmap_offset == info->offset) {
2544                 if (info->offset == 0)
2545                         return false;
2546                 bitmap_offset = offset_to_bitmap(ctl, info->offset - 1);
2547         }
2548
2549         bitmap = tree_search_offset(ctl, bitmap_offset, 1, 0);
2550         if (!bitmap)
2551                 return false;
2552
2553         i = offset_to_bit(bitmap->offset, ctl->unit, info->offset) - 1;
2554         j = 0;
2555         prev_j = (unsigned long)-1;
2556         for_each_clear_bit_from(j, bitmap->bitmap, BITS_PER_BITMAP) {
2557                 if (j > i)
2558                         break;
2559                 prev_j = j;
2560         }
2561         if (prev_j == i)
2562                 return false;
2563
2564         if (prev_j == (unsigned long)-1)
2565                 bytes = (i + 1) * ctl->unit;
2566         else
2567                 bytes = (i - prev_j) * ctl->unit;
2568
2569         info->offset -= bytes;
2570         info->bytes += bytes;
2571
2572         /* See try_merge_free_space() comment. */
2573         if (!btrfs_free_space_trimmed(bitmap))
2574                 info->trim_state = BTRFS_TRIM_STATE_UNTRIMMED;
2575
2576         bitmap_clear_bits(ctl, bitmap, info->offset, bytes, update_stat);
2577
2578         if (!bitmap->bytes)
2579                 free_bitmap(ctl, bitmap);
2580
2581         return true;
2582 }
2583
2584 /*
2585  * We prefer always to allocate from extent entries, both for clustered and
2586  * non-clustered allocation requests. So when attempting to add a new extent
2587  * entry, try to see if there's adjacent free space in bitmap entries, and if
2588  * there is, migrate that space from the bitmaps to the extent.
2589  * Like this we get better chances of satisfying space allocation requests
2590  * because we attempt to satisfy them based on a single cache entry, and never
2591  * on 2 or more entries - even if the entries represent a contiguous free space
2592  * region (e.g. 1 extent entry + 1 bitmap entry starting where the extent entry
2593  * ends).
2594  */
2595 static void steal_from_bitmap(struct btrfs_free_space_ctl *ctl,
2596                               struct btrfs_free_space *info,
2597                               bool update_stat)
2598 {
2599         /*
2600          * Only work with disconnected entries, as we can change their offset,
2601          * and must be extent entries.
2602          */
2603         ASSERT(!info->bitmap);
2604         ASSERT(RB_EMPTY_NODE(&info->offset_index));
2605
2606         if (ctl->total_bitmaps > 0) {
2607                 bool stole_end;
2608                 bool stole_front = false;
2609
2610                 stole_end = steal_from_bitmap_to_end(ctl, info, update_stat);
2611                 if (ctl->total_bitmaps > 0)
2612                         stole_front = steal_from_bitmap_to_front(ctl, info,
2613                                                                  update_stat);
2614
2615                 if (stole_end || stole_front)
2616                         try_merge_free_space(ctl, info, update_stat);
2617         }
2618 }
2619
2620 int __btrfs_add_free_space(struct btrfs_block_group *block_group,
2621                            u64 offset, u64 bytes,
2622                            enum btrfs_trim_state trim_state)
2623 {
2624         struct btrfs_fs_info *fs_info = block_group->fs_info;
2625         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2626         struct btrfs_free_space *info;
2627         int ret = 0;
2628         u64 filter_bytes = bytes;
2629
2630         ASSERT(!btrfs_is_zoned(fs_info));
2631
2632         info = kmem_cache_zalloc(btrfs_free_space_cachep, GFP_NOFS);
2633         if (!info)
2634                 return -ENOMEM;
2635
2636         info->offset = offset;
2637         info->bytes = bytes;
2638         info->trim_state = trim_state;
2639         RB_CLEAR_NODE(&info->offset_index);
2640         RB_CLEAR_NODE(&info->bytes_index);
2641
2642         spin_lock(&ctl->tree_lock);
2643
2644         if (try_merge_free_space(ctl, info, true))
2645                 goto link;
2646
2647         /*
2648          * There was no extent directly to the left or right of this new
2649          * extent then we know we're going to have to allocate a new extent, so
2650          * before we do that see if we need to drop this into a bitmap
2651          */
2652         ret = insert_into_bitmap(ctl, info);
2653         if (ret < 0) {
2654                 goto out;
2655         } else if (ret) {
2656                 ret = 0;
2657                 goto out;
2658         }
2659 link:
2660         /*
2661          * Only steal free space from adjacent bitmaps if we're sure we're not
2662          * going to add the new free space to existing bitmap entries - because
2663          * that would mean unnecessary work that would be reverted. Therefore
2664          * attempt to steal space from bitmaps if we're adding an extent entry.
2665          */
2666         steal_from_bitmap(ctl, info, true);
2667
2668         filter_bytes = max(filter_bytes, info->bytes);
2669
2670         ret = link_free_space(ctl, info);
2671         if (ret)
2672                 kmem_cache_free(btrfs_free_space_cachep, info);
2673 out:
2674         btrfs_discard_update_discardable(block_group);
2675         spin_unlock(&ctl->tree_lock);
2676
2677         if (ret) {
2678                 btrfs_crit(fs_info, "unable to add free space :%d", ret);
2679                 ASSERT(ret != -EEXIST);
2680         }
2681
2682         if (trim_state != BTRFS_TRIM_STATE_TRIMMED) {
2683                 btrfs_discard_check_filter(block_group, filter_bytes);
2684                 btrfs_discard_queue_work(&fs_info->discard_ctl, block_group);
2685         }
2686
2687         return ret;
2688 }
2689
2690 static int __btrfs_add_free_space_zoned(struct btrfs_block_group *block_group,
2691                                         u64 bytenr, u64 size, bool used)
2692 {
2693         struct btrfs_space_info *sinfo = block_group->space_info;
2694         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2695         u64 offset = bytenr - block_group->start;
2696         u64 to_free, to_unusable;
2697         int bg_reclaim_threshold = 0;
2698         bool initial = (size == block_group->length);
2699         u64 reclaimable_unusable;
2700
2701         WARN_ON(!initial && offset + size > block_group->zone_capacity);
2702
2703         if (!initial)
2704                 bg_reclaim_threshold = READ_ONCE(sinfo->bg_reclaim_threshold);
2705
2706         spin_lock(&ctl->tree_lock);
2707         if (!used)
2708                 to_free = size;
2709         else if (initial)
2710                 to_free = block_group->zone_capacity;
2711         else if (offset >= block_group->alloc_offset)
2712                 to_free = size;
2713         else if (offset + size <= block_group->alloc_offset)
2714                 to_free = 0;
2715         else
2716                 to_free = offset + size - block_group->alloc_offset;
2717         to_unusable = size - to_free;
2718
2719         ctl->free_space += to_free;
2720         /*
2721          * If the block group is read-only, we should account freed space into
2722          * bytes_readonly.
2723          */
2724         if (!block_group->ro)
2725                 block_group->zone_unusable += to_unusable;
2726         spin_unlock(&ctl->tree_lock);
2727         if (!used) {
2728                 spin_lock(&block_group->lock);
2729                 block_group->alloc_offset -= size;
2730                 spin_unlock(&block_group->lock);
2731         }
2732
2733         reclaimable_unusable = block_group->zone_unusable -
2734                                (block_group->length - block_group->zone_capacity);
2735         /* All the region is now unusable. Mark it as unused and reclaim */
2736         if (block_group->zone_unusable == block_group->length) {
2737                 btrfs_mark_bg_unused(block_group);
2738         } else if (bg_reclaim_threshold &&
2739                    reclaimable_unusable >=
2740                    mult_perc(block_group->zone_capacity, bg_reclaim_threshold)) {
2741                 btrfs_mark_bg_to_reclaim(block_group);
2742         }
2743
2744         return 0;
2745 }
2746
2747 int btrfs_add_free_space(struct btrfs_block_group *block_group,
2748                          u64 bytenr, u64 size)
2749 {
2750         enum btrfs_trim_state trim_state = BTRFS_TRIM_STATE_UNTRIMMED;
2751
2752         if (btrfs_is_zoned(block_group->fs_info))
2753                 return __btrfs_add_free_space_zoned(block_group, bytenr, size,
2754                                                     true);
2755
2756         if (btrfs_test_opt(block_group->fs_info, DISCARD_SYNC))
2757                 trim_state = BTRFS_TRIM_STATE_TRIMMED;
2758
2759         return __btrfs_add_free_space(block_group, bytenr, size, trim_state);
2760 }
2761
2762 int btrfs_add_free_space_unused(struct btrfs_block_group *block_group,
2763                                 u64 bytenr, u64 size)
2764 {
2765         if (btrfs_is_zoned(block_group->fs_info))
2766                 return __btrfs_add_free_space_zoned(block_group, bytenr, size,
2767                                                     false);
2768
2769         return btrfs_add_free_space(block_group, bytenr, size);
2770 }
2771
2772 /*
2773  * This is a subtle distinction because when adding free space back in general,
2774  * we want it to be added as untrimmed for async. But in the case where we add
2775  * it on loading of a block group, we want to consider it trimmed.
2776  */
2777 int btrfs_add_free_space_async_trimmed(struct btrfs_block_group *block_group,
2778                                        u64 bytenr, u64 size)
2779 {
2780         enum btrfs_trim_state trim_state = BTRFS_TRIM_STATE_UNTRIMMED;
2781
2782         if (btrfs_is_zoned(block_group->fs_info))
2783                 return __btrfs_add_free_space_zoned(block_group, bytenr, size,
2784                                                     true);
2785
2786         if (btrfs_test_opt(block_group->fs_info, DISCARD_SYNC) ||
2787             btrfs_test_opt(block_group->fs_info, DISCARD_ASYNC))
2788                 trim_state = BTRFS_TRIM_STATE_TRIMMED;
2789
2790         return __btrfs_add_free_space(block_group, bytenr, size, trim_state);
2791 }
2792
2793 int btrfs_remove_free_space(struct btrfs_block_group *block_group,
2794                             u64 offset, u64 bytes)
2795 {
2796         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2797         struct btrfs_free_space *info;
2798         int ret;
2799         bool re_search = false;
2800
2801         if (btrfs_is_zoned(block_group->fs_info)) {
2802                 /*
2803                  * This can happen with conventional zones when replaying log.
2804                  * Since the allocation info of tree-log nodes are not recorded
2805                  * to the extent-tree, calculate_alloc_pointer() failed to
2806                  * advance the allocation pointer after last allocated tree log
2807                  * node blocks.
2808                  *
2809                  * This function is called from
2810                  * btrfs_pin_extent_for_log_replay() when replaying the log.
2811                  * Advance the pointer not to overwrite the tree-log nodes.
2812                  */
2813                 if (block_group->start + block_group->alloc_offset <
2814                     offset + bytes) {
2815                         block_group->alloc_offset =
2816                                 offset + bytes - block_group->start;
2817                 }
2818                 return 0;
2819         }
2820
2821         spin_lock(&ctl->tree_lock);
2822
2823 again:
2824         ret = 0;
2825         if (!bytes)
2826                 goto out_lock;
2827
2828         info = tree_search_offset(ctl, offset, 0, 0);
2829         if (!info) {
2830                 /*
2831                  * oops didn't find an extent that matched the space we wanted
2832                  * to remove, look for a bitmap instead
2833                  */
2834                 info = tree_search_offset(ctl, offset_to_bitmap(ctl, offset),
2835                                           1, 0);
2836                 if (!info) {
2837                         /*
2838                          * If we found a partial bit of our free space in a
2839                          * bitmap but then couldn't find the other part this may
2840                          * be a problem, so WARN about it.
2841                          */
2842                         WARN_ON(re_search);
2843                         goto out_lock;
2844                 }
2845         }
2846
2847         re_search = false;
2848         if (!info->bitmap) {
2849                 unlink_free_space(ctl, info, true);
2850                 if (offset == info->offset) {
2851                         u64 to_free = min(bytes, info->bytes);
2852
2853                         info->bytes -= to_free;
2854                         info->offset += to_free;
2855                         if (info->bytes) {
2856                                 ret = link_free_space(ctl, info);
2857                                 WARN_ON(ret);
2858                         } else {
2859                                 kmem_cache_free(btrfs_free_space_cachep, info);
2860                         }
2861
2862                         offset += to_free;
2863                         bytes -= to_free;
2864                         goto again;
2865                 } else {
2866                         u64 old_end = info->bytes + info->offset;
2867
2868                         info->bytes = offset - info->offset;
2869                         ret = link_free_space(ctl, info);
2870                         WARN_ON(ret);
2871                         if (ret)
2872                                 goto out_lock;
2873
2874                         /* Not enough bytes in this entry to satisfy us */
2875                         if (old_end < offset + bytes) {
2876                                 bytes -= old_end - offset;
2877                                 offset = old_end;
2878                                 goto again;
2879                         } else if (old_end == offset + bytes) {
2880                                 /* all done */
2881                                 goto out_lock;
2882                         }
2883                         spin_unlock(&ctl->tree_lock);
2884
2885                         ret = __btrfs_add_free_space(block_group,
2886                                                      offset + bytes,
2887                                                      old_end - (offset + bytes),
2888                                                      info->trim_state);
2889                         WARN_ON(ret);
2890                         goto out;
2891                 }
2892         }
2893
2894         ret = remove_from_bitmap(ctl, info, &offset, &bytes);
2895         if (ret == -EAGAIN) {
2896                 re_search = true;
2897                 goto again;
2898         }
2899 out_lock:
2900         btrfs_discard_update_discardable(block_group);
2901         spin_unlock(&ctl->tree_lock);
2902 out:
2903         return ret;
2904 }
2905
2906 void btrfs_dump_free_space(struct btrfs_block_group *block_group,
2907                            u64 bytes)
2908 {
2909         struct btrfs_fs_info *fs_info = block_group->fs_info;
2910         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2911         struct btrfs_free_space *info;
2912         struct rb_node *n;
2913         int count = 0;
2914
2915         /*
2916          * Zoned btrfs does not use free space tree and cluster. Just print
2917          * out the free space after the allocation offset.
2918          */
2919         if (btrfs_is_zoned(fs_info)) {
2920                 btrfs_info(fs_info, "free space %llu active %d",
2921                            block_group->zone_capacity - block_group->alloc_offset,
2922                            test_bit(BLOCK_GROUP_FLAG_ZONE_IS_ACTIVE,
2923                                     &block_group->runtime_flags));
2924                 return;
2925         }
2926
2927         spin_lock(&ctl->tree_lock);
2928         for (n = rb_first(&ctl->free_space_offset); n; n = rb_next(n)) {
2929                 info = rb_entry(n, struct btrfs_free_space, offset_index);
2930                 if (info->bytes >= bytes && !block_group->ro)
2931                         count++;
2932                 btrfs_crit(fs_info, "entry offset %llu, bytes %llu, bitmap %s",
2933                            info->offset, info->bytes,
2934                        (info->bitmap) ? "yes" : "no");
2935         }
2936         spin_unlock(&ctl->tree_lock);
2937         btrfs_info(fs_info, "block group has cluster?: %s",
2938                list_empty(&block_group->cluster_list) ? "no" : "yes");
2939         btrfs_info(fs_info,
2940                    "%d free space entries at or bigger than %llu bytes",
2941                    count, bytes);
2942 }
2943
2944 void btrfs_init_free_space_ctl(struct btrfs_block_group *block_group,
2945                                struct btrfs_free_space_ctl *ctl)
2946 {
2947         struct btrfs_fs_info *fs_info = block_group->fs_info;
2948
2949         spin_lock_init(&ctl->tree_lock);
2950         ctl->unit = fs_info->sectorsize;
2951         ctl->start = block_group->start;
2952         ctl->block_group = block_group;
2953         ctl->op = &free_space_op;
2954         ctl->free_space_bytes = RB_ROOT_CACHED;
2955         INIT_LIST_HEAD(&ctl->trimming_ranges);
2956         mutex_init(&ctl->cache_writeout_mutex);
2957
2958         /*
2959          * we only want to have 32k of ram per block group for keeping
2960          * track of free space, and if we pass 1/2 of that we want to
2961          * start converting things over to using bitmaps
2962          */
2963         ctl->extents_thresh = (SZ_32K / 2) / sizeof(struct btrfs_free_space);
2964 }
2965
2966 /*
2967  * for a given cluster, put all of its extents back into the free
2968  * space cache.  If the block group passed doesn't match the block group
2969  * pointed to by the cluster, someone else raced in and freed the
2970  * cluster already.  In that case, we just return without changing anything
2971  */
2972 static void __btrfs_return_cluster_to_free_space(
2973                              struct btrfs_block_group *block_group,
2974                              struct btrfs_free_cluster *cluster)
2975 {
2976         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
2977         struct rb_node *node;
2978
2979         lockdep_assert_held(&ctl->tree_lock);
2980
2981         spin_lock(&cluster->lock);
2982         if (cluster->block_group != block_group) {
2983                 spin_unlock(&cluster->lock);
2984                 return;
2985         }
2986
2987         cluster->block_group = NULL;
2988         cluster->window_start = 0;
2989         list_del_init(&cluster->block_group_list);
2990
2991         node = rb_first(&cluster->root);
2992         while (node) {
2993                 struct btrfs_free_space *entry;
2994
2995                 entry = rb_entry(node, struct btrfs_free_space, offset_index);
2996                 node = rb_next(&entry->offset_index);
2997                 rb_erase(&entry->offset_index, &cluster->root);
2998                 RB_CLEAR_NODE(&entry->offset_index);
2999
3000                 if (!entry->bitmap) {
3001                         /* Merging treats extents as if they were new */
3002                         if (!btrfs_free_space_trimmed(entry)) {
3003                                 ctl->discardable_extents[BTRFS_STAT_CURR]--;
3004                                 ctl->discardable_bytes[BTRFS_STAT_CURR] -=
3005                                         entry->bytes;
3006                         }
3007
3008                         try_merge_free_space(ctl, entry, false);
3009                         steal_from_bitmap(ctl, entry, false);
3010
3011                         /* As we insert directly, update these statistics */
3012                         if (!btrfs_free_space_trimmed(entry)) {
3013                                 ctl->discardable_extents[BTRFS_STAT_CURR]++;
3014                                 ctl->discardable_bytes[BTRFS_STAT_CURR] +=
3015                                         entry->bytes;
3016                         }
3017                 }
3018                 tree_insert_offset(ctl, NULL, entry);
3019                 rb_add_cached(&entry->bytes_index, &ctl->free_space_bytes,
3020                               entry_less);
3021         }
3022         cluster->root = RB_ROOT;
3023         spin_unlock(&cluster->lock);
3024         btrfs_put_block_group(block_group);
3025 }
3026
3027 void btrfs_remove_free_space_cache(struct btrfs_block_group *block_group)
3028 {
3029         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3030         struct btrfs_free_cluster *cluster;
3031         struct list_head *head;
3032
3033         spin_lock(&ctl->tree_lock);
3034         while ((head = block_group->cluster_list.next) !=
3035                &block_group->cluster_list) {
3036                 cluster = list_entry(head, struct btrfs_free_cluster,
3037                                      block_group_list);
3038
3039                 WARN_ON(cluster->block_group != block_group);
3040                 __btrfs_return_cluster_to_free_space(block_group, cluster);
3041
3042                 cond_resched_lock(&ctl->tree_lock);
3043         }
3044         __btrfs_remove_free_space_cache(ctl);
3045         btrfs_discard_update_discardable(block_group);
3046         spin_unlock(&ctl->tree_lock);
3047
3048 }
3049
3050 /*
3051  * Walk @block_group's free space rb_tree to determine if everything is trimmed.
3052  */
3053 bool btrfs_is_free_space_trimmed(struct btrfs_block_group *block_group)
3054 {
3055         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3056         struct btrfs_free_space *info;
3057         struct rb_node *node;
3058         bool ret = true;
3059
3060         spin_lock(&ctl->tree_lock);
3061         node = rb_first(&ctl->free_space_offset);
3062
3063         while (node) {
3064                 info = rb_entry(node, struct btrfs_free_space, offset_index);
3065
3066                 if (!btrfs_free_space_trimmed(info)) {
3067                         ret = false;
3068                         break;
3069                 }
3070
3071                 node = rb_next(node);
3072         }
3073
3074         spin_unlock(&ctl->tree_lock);
3075         return ret;
3076 }
3077
3078 u64 btrfs_find_space_for_alloc(struct btrfs_block_group *block_group,
3079                                u64 offset, u64 bytes, u64 empty_size,
3080                                u64 *max_extent_size)
3081 {
3082         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3083         struct btrfs_discard_ctl *discard_ctl =
3084                                         &block_group->fs_info->discard_ctl;
3085         struct btrfs_free_space *entry = NULL;
3086         u64 bytes_search = bytes + empty_size;
3087         u64 ret = 0;
3088         u64 align_gap = 0;
3089         u64 align_gap_len = 0;
3090         enum btrfs_trim_state align_gap_trim_state = BTRFS_TRIM_STATE_UNTRIMMED;
3091         bool use_bytes_index = (offset == block_group->start);
3092
3093         ASSERT(!btrfs_is_zoned(block_group->fs_info));
3094
3095         spin_lock(&ctl->tree_lock);
3096         entry = find_free_space(ctl, &offset, &bytes_search,
3097                                 block_group->full_stripe_len, max_extent_size,
3098                                 use_bytes_index);
3099         if (!entry)
3100                 goto out;
3101
3102         ret = offset;
3103         if (entry->bitmap) {
3104                 bitmap_clear_bits(ctl, entry, offset, bytes, true);
3105
3106                 if (!btrfs_free_space_trimmed(entry))
3107                         atomic64_add(bytes, &discard_ctl->discard_bytes_saved);
3108
3109                 if (!entry->bytes)
3110                         free_bitmap(ctl, entry);
3111         } else {
3112                 unlink_free_space(ctl, entry, true);
3113                 align_gap_len = offset - entry->offset;
3114                 align_gap = entry->offset;
3115                 align_gap_trim_state = entry->trim_state;
3116
3117                 if (!btrfs_free_space_trimmed(entry))
3118                         atomic64_add(bytes, &discard_ctl->discard_bytes_saved);
3119
3120                 entry->offset = offset + bytes;
3121                 WARN_ON(entry->bytes < bytes + align_gap_len);
3122
3123                 entry->bytes -= bytes + align_gap_len;
3124                 if (!entry->bytes)
3125                         kmem_cache_free(btrfs_free_space_cachep, entry);
3126                 else
3127                         link_free_space(ctl, entry);
3128         }
3129 out:
3130         btrfs_discard_update_discardable(block_group);
3131         spin_unlock(&ctl->tree_lock);
3132
3133         if (align_gap_len)
3134                 __btrfs_add_free_space(block_group, align_gap, align_gap_len,
3135                                        align_gap_trim_state);
3136         return ret;
3137 }
3138
3139 /*
3140  * given a cluster, put all of its extents back into the free space
3141  * cache.  If a block group is passed, this function will only free
3142  * a cluster that belongs to the passed block group.
3143  *
3144  * Otherwise, it'll get a reference on the block group pointed to by the
3145  * cluster and remove the cluster from it.
3146  */
3147 void btrfs_return_cluster_to_free_space(
3148                                struct btrfs_block_group *block_group,
3149                                struct btrfs_free_cluster *cluster)
3150 {
3151         struct btrfs_free_space_ctl *ctl;
3152
3153         /* first, get a safe pointer to the block group */
3154         spin_lock(&cluster->lock);
3155         if (!block_group) {
3156                 block_group = cluster->block_group;
3157                 if (!block_group) {
3158                         spin_unlock(&cluster->lock);
3159                         return;
3160                 }
3161         } else if (cluster->block_group != block_group) {
3162                 /* someone else has already freed it don't redo their work */
3163                 spin_unlock(&cluster->lock);
3164                 return;
3165         }
3166         btrfs_get_block_group(block_group);
3167         spin_unlock(&cluster->lock);
3168
3169         ctl = block_group->free_space_ctl;
3170
3171         /* now return any extents the cluster had on it */
3172         spin_lock(&ctl->tree_lock);
3173         __btrfs_return_cluster_to_free_space(block_group, cluster);
3174         spin_unlock(&ctl->tree_lock);
3175
3176         btrfs_discard_queue_work(&block_group->fs_info->discard_ctl, block_group);
3177
3178         /* finally drop our ref */
3179         btrfs_put_block_group(block_group);
3180 }
3181
3182 static u64 btrfs_alloc_from_bitmap(struct btrfs_block_group *block_group,
3183                                    struct btrfs_free_cluster *cluster,
3184                                    struct btrfs_free_space *entry,
3185                                    u64 bytes, u64 min_start,
3186                                    u64 *max_extent_size)
3187 {
3188         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3189         int err;
3190         u64 search_start = cluster->window_start;
3191         u64 search_bytes = bytes;
3192         u64 ret = 0;
3193
3194         search_start = min_start;
3195         search_bytes = bytes;
3196
3197         err = search_bitmap(ctl, entry, &search_start, &search_bytes, true);
3198         if (err) {
3199                 *max_extent_size = max(get_max_extent_size(entry),
3200                                        *max_extent_size);
3201                 return 0;
3202         }
3203
3204         ret = search_start;
3205         bitmap_clear_bits(ctl, entry, ret, bytes, false);
3206
3207         return ret;
3208 }
3209
3210 /*
3211  * given a cluster, try to allocate 'bytes' from it, returns 0
3212  * if it couldn't find anything suitably large, or a logical disk offset
3213  * if things worked out
3214  */
3215 u64 btrfs_alloc_from_cluster(struct btrfs_block_group *block_group,
3216                              struct btrfs_free_cluster *cluster, u64 bytes,
3217                              u64 min_start, u64 *max_extent_size)
3218 {
3219         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3220         struct btrfs_discard_ctl *discard_ctl =
3221                                         &block_group->fs_info->discard_ctl;
3222         struct btrfs_free_space *entry = NULL;
3223         struct rb_node *node;
3224         u64 ret = 0;
3225
3226         ASSERT(!btrfs_is_zoned(block_group->fs_info));
3227
3228         spin_lock(&cluster->lock);
3229         if (bytes > cluster->max_size)
3230                 goto out;
3231
3232         if (cluster->block_group != block_group)
3233                 goto out;
3234
3235         node = rb_first(&cluster->root);
3236         if (!node)
3237                 goto out;
3238
3239         entry = rb_entry(node, struct btrfs_free_space, offset_index);
3240         while (1) {
3241                 if (entry->bytes < bytes)
3242                         *max_extent_size = max(get_max_extent_size(entry),
3243                                                *max_extent_size);
3244
3245                 if (entry->bytes < bytes ||
3246                     (!entry->bitmap && entry->offset < min_start)) {
3247                         node = rb_next(&entry->offset_index);
3248                         if (!node)
3249                                 break;
3250                         entry = rb_entry(node, struct btrfs_free_space,
3251                                          offset_index);
3252                         continue;
3253                 }
3254
3255                 if (entry->bitmap) {
3256                         ret = btrfs_alloc_from_bitmap(block_group,
3257                                                       cluster, entry, bytes,
3258                                                       cluster->window_start,
3259                                                       max_extent_size);
3260                         if (ret == 0) {
3261                                 node = rb_next(&entry->offset_index);
3262                                 if (!node)
3263                                         break;
3264                                 entry = rb_entry(node, struct btrfs_free_space,
3265                                                  offset_index);
3266                                 continue;
3267                         }
3268                         cluster->window_start += bytes;
3269                 } else {
3270                         ret = entry->offset;
3271
3272                         entry->offset += bytes;
3273                         entry->bytes -= bytes;
3274                 }
3275
3276                 break;
3277         }
3278 out:
3279         spin_unlock(&cluster->lock);
3280
3281         if (!ret)
3282                 return 0;
3283
3284         spin_lock(&ctl->tree_lock);
3285
3286         if (!btrfs_free_space_trimmed(entry))
3287                 atomic64_add(bytes, &discard_ctl->discard_bytes_saved);
3288
3289         ctl->free_space -= bytes;
3290         if (!entry->bitmap && !btrfs_free_space_trimmed(entry))
3291                 ctl->discardable_bytes[BTRFS_STAT_CURR] -= bytes;
3292
3293         spin_lock(&cluster->lock);
3294         if (entry->bytes == 0) {
3295                 rb_erase(&entry->offset_index, &cluster->root);
3296                 ctl->free_extents--;
3297                 if (entry->bitmap) {
3298                         kmem_cache_free(btrfs_free_space_bitmap_cachep,
3299                                         entry->bitmap);
3300                         ctl->total_bitmaps--;
3301                         recalculate_thresholds(ctl);
3302                 } else if (!btrfs_free_space_trimmed(entry)) {
3303                         ctl->discardable_extents[BTRFS_STAT_CURR]--;
3304                 }
3305                 kmem_cache_free(btrfs_free_space_cachep, entry);
3306         }
3307
3308         spin_unlock(&cluster->lock);
3309         spin_unlock(&ctl->tree_lock);
3310
3311         return ret;
3312 }
3313
3314 static int btrfs_bitmap_cluster(struct btrfs_block_group *block_group,
3315                                 struct btrfs_free_space *entry,
3316                                 struct btrfs_free_cluster *cluster,
3317                                 u64 offset, u64 bytes,
3318                                 u64 cont1_bytes, u64 min_bytes)
3319 {
3320         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3321         unsigned long next_zero;
3322         unsigned long i;
3323         unsigned long want_bits;
3324         unsigned long min_bits;
3325         unsigned long found_bits;
3326         unsigned long max_bits = 0;
3327         unsigned long start = 0;
3328         unsigned long total_found = 0;
3329         int ret;
3330
3331         lockdep_assert_held(&ctl->tree_lock);
3332
3333         i = offset_to_bit(entry->offset, ctl->unit,
3334                           max_t(u64, offset, entry->offset));
3335         want_bits = bytes_to_bits(bytes, ctl->unit);
3336         min_bits = bytes_to_bits(min_bytes, ctl->unit);
3337
3338         /*
3339          * Don't bother looking for a cluster in this bitmap if it's heavily
3340          * fragmented.
3341          */
3342         if (entry->max_extent_size &&
3343             entry->max_extent_size < cont1_bytes)
3344                 return -ENOSPC;
3345 again:
3346         found_bits = 0;
3347         for_each_set_bit_from(i, entry->bitmap, BITS_PER_BITMAP) {
3348                 next_zero = find_next_zero_bit(entry->bitmap,
3349                                                BITS_PER_BITMAP, i);
3350                 if (next_zero - i >= min_bits) {
3351                         found_bits = next_zero - i;
3352                         if (found_bits > max_bits)
3353                                 max_bits = found_bits;
3354                         break;
3355                 }
3356                 if (next_zero - i > max_bits)
3357                         max_bits = next_zero - i;
3358                 i = next_zero;
3359         }
3360
3361         if (!found_bits) {
3362                 entry->max_extent_size = (u64)max_bits * ctl->unit;
3363                 return -ENOSPC;
3364         }
3365
3366         if (!total_found) {
3367                 start = i;
3368                 cluster->max_size = 0;
3369         }
3370
3371         total_found += found_bits;
3372
3373         if (cluster->max_size < found_bits * ctl->unit)
3374                 cluster->max_size = found_bits * ctl->unit;
3375
3376         if (total_found < want_bits || cluster->max_size < cont1_bytes) {
3377                 i = next_zero + 1;
3378                 goto again;
3379         }
3380
3381         cluster->window_start = start * ctl->unit + entry->offset;
3382         rb_erase(&entry->offset_index, &ctl->free_space_offset);
3383         rb_erase_cached(&entry->bytes_index, &ctl->free_space_bytes);
3384
3385         /*
3386          * We need to know if we're currently on the normal space index when we
3387          * manipulate the bitmap so that we know we need to remove and re-insert
3388          * it into the space_index tree.  Clear the bytes_index node here so the
3389          * bitmap manipulation helpers know not to mess with the space_index
3390          * until this bitmap entry is added back into the normal cache.
3391          */
3392         RB_CLEAR_NODE(&entry->bytes_index);
3393
3394         ret = tree_insert_offset(ctl, cluster, entry);
3395         ASSERT(!ret); /* -EEXIST; Logic error */
3396
3397         trace_btrfs_setup_cluster(block_group, cluster,
3398                                   total_found * ctl->unit, 1);
3399         return 0;
3400 }
3401
3402 /*
3403  * This searches the block group for just extents to fill the cluster with.
3404  * Try to find a cluster with at least bytes total bytes, at least one
3405  * extent of cont1_bytes, and other clusters of at least min_bytes.
3406  */
3407 static noinline int
3408 setup_cluster_no_bitmap(struct btrfs_block_group *block_group,
3409                         struct btrfs_free_cluster *cluster,
3410                         struct list_head *bitmaps, u64 offset, u64 bytes,
3411                         u64 cont1_bytes, u64 min_bytes)
3412 {
3413         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3414         struct btrfs_free_space *first = NULL;
3415         struct btrfs_free_space *entry = NULL;
3416         struct btrfs_free_space *last;
3417         struct rb_node *node;
3418         u64 window_free;
3419         u64 max_extent;
3420         u64 total_size = 0;
3421
3422         lockdep_assert_held(&ctl->tree_lock);
3423
3424         entry = tree_search_offset(ctl, offset, 0, 1);
3425         if (!entry)
3426                 return -ENOSPC;
3427
3428         /*
3429          * We don't want bitmaps, so just move along until we find a normal
3430          * extent entry.
3431          */
3432         while (entry->bitmap || entry->bytes < min_bytes) {
3433                 if (entry->bitmap && list_empty(&entry->list))
3434                         list_add_tail(&entry->list, bitmaps);
3435                 node = rb_next(&entry->offset_index);
3436                 if (!node)
3437                         return -ENOSPC;
3438                 entry = rb_entry(node, struct btrfs_free_space, offset_index);
3439         }
3440
3441         window_free = entry->bytes;
3442         max_extent = entry->bytes;
3443         first = entry;
3444         last = entry;
3445
3446         for (node = rb_next(&entry->offset_index); node;
3447              node = rb_next(&entry->offset_index)) {
3448                 entry = rb_entry(node, struct btrfs_free_space, offset_index);
3449
3450                 if (entry->bitmap) {
3451                         if (list_empty(&entry->list))
3452                                 list_add_tail(&entry->list, bitmaps);
3453                         continue;
3454                 }
3455
3456                 if (entry->bytes < min_bytes)
3457                         continue;
3458
3459                 last = entry;
3460                 window_free += entry->bytes;
3461                 if (entry->bytes > max_extent)
3462                         max_extent = entry->bytes;
3463         }
3464
3465         if (window_free < bytes || max_extent < cont1_bytes)
3466                 return -ENOSPC;
3467
3468         cluster->window_start = first->offset;
3469
3470         node = &first->offset_index;
3471
3472         /*
3473          * now we've found our entries, pull them out of the free space
3474          * cache and put them into the cluster rbtree
3475          */
3476         do {
3477                 int ret;
3478
3479                 entry = rb_entry(node, struct btrfs_free_space, offset_index);
3480                 node = rb_next(&entry->offset_index);
3481                 if (entry->bitmap || entry->bytes < min_bytes)
3482                         continue;
3483
3484                 rb_erase(&entry->offset_index, &ctl->free_space_offset);
3485                 rb_erase_cached(&entry->bytes_index, &ctl->free_space_bytes);
3486                 ret = tree_insert_offset(ctl, cluster, entry);
3487                 total_size += entry->bytes;
3488                 ASSERT(!ret); /* -EEXIST; Logic error */
3489         } while (node && entry != last);
3490
3491         cluster->max_size = max_extent;
3492         trace_btrfs_setup_cluster(block_group, cluster, total_size, 0);
3493         return 0;
3494 }
3495
3496 /*
3497  * This specifically looks for bitmaps that may work in the cluster, we assume
3498  * that we have already failed to find extents that will work.
3499  */
3500 static noinline int
3501 setup_cluster_bitmap(struct btrfs_block_group *block_group,
3502                      struct btrfs_free_cluster *cluster,
3503                      struct list_head *bitmaps, u64 offset, u64 bytes,
3504                      u64 cont1_bytes, u64 min_bytes)
3505 {
3506         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3507         struct btrfs_free_space *entry = NULL;
3508         int ret = -ENOSPC;
3509         u64 bitmap_offset = offset_to_bitmap(ctl, offset);
3510
3511         if (ctl->total_bitmaps == 0)
3512                 return -ENOSPC;
3513
3514         /*
3515          * The bitmap that covers offset won't be in the list unless offset
3516          * is just its start offset.
3517          */
3518         if (!list_empty(bitmaps))
3519                 entry = list_first_entry(bitmaps, struct btrfs_free_space, list);
3520
3521         if (!entry || entry->offset != bitmap_offset) {
3522                 entry = tree_search_offset(ctl, bitmap_offset, 1, 0);
3523                 if (entry && list_empty(&entry->list))
3524                         list_add(&entry->list, bitmaps);
3525         }
3526
3527         list_for_each_entry(entry, bitmaps, list) {
3528                 if (entry->bytes < bytes)
3529                         continue;
3530                 ret = btrfs_bitmap_cluster(block_group, entry, cluster, offset,
3531                                            bytes, cont1_bytes, min_bytes);
3532                 if (!ret)
3533                         return 0;
3534         }
3535
3536         /*
3537          * The bitmaps list has all the bitmaps that record free space
3538          * starting after offset, so no more search is required.
3539          */
3540         return -ENOSPC;
3541 }
3542
3543 /*
3544  * here we try to find a cluster of blocks in a block group.  The goal
3545  * is to find at least bytes+empty_size.
3546  * We might not find them all in one contiguous area.
3547  *
3548  * returns zero and sets up cluster if things worked out, otherwise
3549  * it returns -enospc
3550  */
3551 int btrfs_find_space_cluster(struct btrfs_block_group *block_group,
3552                              struct btrfs_free_cluster *cluster,
3553                              u64 offset, u64 bytes, u64 empty_size)
3554 {
3555         struct btrfs_fs_info *fs_info = block_group->fs_info;
3556         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3557         struct btrfs_free_space *entry, *tmp;
3558         LIST_HEAD(bitmaps);
3559         u64 min_bytes;
3560         u64 cont1_bytes;
3561         int ret;
3562
3563         /*
3564          * Choose the minimum extent size we'll require for this
3565          * cluster.  For SSD_SPREAD, don't allow any fragmentation.
3566          * For metadata, allow allocates with smaller extents.  For
3567          * data, keep it dense.
3568          */
3569         if (btrfs_test_opt(fs_info, SSD_SPREAD)) {
3570                 cont1_bytes = bytes + empty_size;
3571                 min_bytes = cont1_bytes;
3572         } else if (block_group->flags & BTRFS_BLOCK_GROUP_METADATA) {
3573                 cont1_bytes = bytes;
3574                 min_bytes = fs_info->sectorsize;
3575         } else {
3576                 cont1_bytes = max(bytes, (bytes + empty_size) >> 2);
3577                 min_bytes = fs_info->sectorsize;
3578         }
3579
3580         spin_lock(&ctl->tree_lock);
3581
3582         /*
3583          * If we know we don't have enough space to make a cluster don't even
3584          * bother doing all the work to try and find one.
3585          */
3586         if (ctl->free_space < bytes) {
3587                 spin_unlock(&ctl->tree_lock);
3588                 return -ENOSPC;
3589         }
3590
3591         spin_lock(&cluster->lock);
3592
3593         /* someone already found a cluster, hooray */
3594         if (cluster->block_group) {
3595                 ret = 0;
3596                 goto out;
3597         }
3598
3599         trace_btrfs_find_cluster(block_group, offset, bytes, empty_size,
3600                                  min_bytes);
3601
3602         ret = setup_cluster_no_bitmap(block_group, cluster, &bitmaps, offset,
3603                                       bytes + empty_size,
3604                                       cont1_bytes, min_bytes);
3605         if (ret)
3606                 ret = setup_cluster_bitmap(block_group, cluster, &bitmaps,
3607                                            offset, bytes + empty_size,
3608                                            cont1_bytes, min_bytes);
3609
3610         /* Clear our temporary list */
3611         list_for_each_entry_safe(entry, tmp, &bitmaps, list)
3612                 list_del_init(&entry->list);
3613
3614         if (!ret) {
3615                 btrfs_get_block_group(block_group);
3616                 list_add_tail(&cluster->block_group_list,
3617                               &block_group->cluster_list);
3618                 cluster->block_group = block_group;
3619         } else {
3620                 trace_btrfs_failed_cluster_setup(block_group);
3621         }
3622 out:
3623         spin_unlock(&cluster->lock);
3624         spin_unlock(&ctl->tree_lock);
3625
3626         return ret;
3627 }
3628
3629 /*
3630  * simple code to zero out a cluster
3631  */
3632 void btrfs_init_free_cluster(struct btrfs_free_cluster *cluster)
3633 {
3634         spin_lock_init(&cluster->lock);
3635         spin_lock_init(&cluster->refill_lock);
3636         cluster->root = RB_ROOT;
3637         cluster->max_size = 0;
3638         cluster->fragmented = false;
3639         INIT_LIST_HEAD(&cluster->block_group_list);
3640         cluster->block_group = NULL;
3641 }
3642
3643 static int do_trimming(struct btrfs_block_group *block_group,
3644                        u64 *total_trimmed, u64 start, u64 bytes,
3645                        u64 reserved_start, u64 reserved_bytes,
3646                        enum btrfs_trim_state reserved_trim_state,
3647                        struct btrfs_trim_range *trim_entry)
3648 {
3649         struct btrfs_space_info *space_info = block_group->space_info;
3650         struct btrfs_fs_info *fs_info = block_group->fs_info;
3651         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3652         int ret;
3653         int update = 0;
3654         const u64 end = start + bytes;
3655         const u64 reserved_end = reserved_start + reserved_bytes;
3656         enum btrfs_trim_state trim_state = BTRFS_TRIM_STATE_UNTRIMMED;
3657         u64 trimmed = 0;
3658
3659         spin_lock(&space_info->lock);
3660         spin_lock(&block_group->lock);
3661         if (!block_group->ro) {
3662                 block_group->reserved += reserved_bytes;
3663                 space_info->bytes_reserved += reserved_bytes;
3664                 update = 1;
3665         }
3666         spin_unlock(&block_group->lock);
3667         spin_unlock(&space_info->lock);
3668
3669         ret = btrfs_discard_extent(fs_info, start, bytes, &trimmed);
3670         if (!ret) {
3671                 *total_trimmed += trimmed;
3672                 trim_state = BTRFS_TRIM_STATE_TRIMMED;
3673         }
3674
3675         mutex_lock(&ctl->cache_writeout_mutex);
3676         if (reserved_start < start)
3677                 __btrfs_add_free_space(block_group, reserved_start,
3678                                        start - reserved_start,
3679                                        reserved_trim_state);
3680         if (end < reserved_end)
3681                 __btrfs_add_free_space(block_group, end, reserved_end - end,
3682                                        reserved_trim_state);
3683         __btrfs_add_free_space(block_group, start, bytes, trim_state);
3684         list_del(&trim_entry->list);
3685         mutex_unlock(&ctl->cache_writeout_mutex);
3686
3687         if (update) {
3688                 spin_lock(&space_info->lock);
3689                 spin_lock(&block_group->lock);
3690                 if (block_group->ro)
3691                         space_info->bytes_readonly += reserved_bytes;
3692                 block_group->reserved -= reserved_bytes;
3693                 space_info->bytes_reserved -= reserved_bytes;
3694                 spin_unlock(&block_group->lock);
3695                 spin_unlock(&space_info->lock);
3696         }
3697
3698         return ret;
3699 }
3700
3701 /*
3702  * If @async is set, then we will trim 1 region and return.
3703  */
3704 static int trim_no_bitmap(struct btrfs_block_group *block_group,
3705                           u64 *total_trimmed, u64 start, u64 end, u64 minlen,
3706                           bool async)
3707 {
3708         struct btrfs_discard_ctl *discard_ctl =
3709                                         &block_group->fs_info->discard_ctl;
3710         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3711         struct btrfs_free_space *entry;
3712         struct rb_node *node;
3713         int ret = 0;
3714         u64 extent_start;
3715         u64 extent_bytes;
3716         enum btrfs_trim_state extent_trim_state;
3717         u64 bytes;
3718         const u64 max_discard_size = READ_ONCE(discard_ctl->max_discard_size);
3719
3720         while (start < end) {
3721                 struct btrfs_trim_range trim_entry;
3722
3723                 mutex_lock(&ctl->cache_writeout_mutex);
3724                 spin_lock(&ctl->tree_lock);
3725
3726                 if (ctl->free_space < minlen)
3727                         goto out_unlock;
3728
3729                 entry = tree_search_offset(ctl, start, 0, 1);
3730                 if (!entry)
3731                         goto out_unlock;
3732
3733                 /* Skip bitmaps and if async, already trimmed entries */
3734                 while (entry->bitmap ||
3735                        (async && btrfs_free_space_trimmed(entry))) {
3736                         node = rb_next(&entry->offset_index);
3737                         if (!node)
3738                                 goto out_unlock;
3739                         entry = rb_entry(node, struct btrfs_free_space,
3740                                          offset_index);
3741                 }
3742
3743                 if (entry->offset >= end)
3744                         goto out_unlock;
3745
3746                 extent_start = entry->offset;
3747                 extent_bytes = entry->bytes;
3748                 extent_trim_state = entry->trim_state;
3749                 if (async) {
3750                         start = entry->offset;
3751                         bytes = entry->bytes;
3752                         if (bytes < minlen) {
3753                                 spin_unlock(&ctl->tree_lock);
3754                                 mutex_unlock(&ctl->cache_writeout_mutex);
3755                                 goto next;
3756                         }
3757                         unlink_free_space(ctl, entry, true);
3758                         /*
3759                          * Let bytes = BTRFS_MAX_DISCARD_SIZE + X.
3760                          * If X < BTRFS_ASYNC_DISCARD_MIN_FILTER, we won't trim
3761                          * X when we come back around.  So trim it now.
3762                          */
3763                         if (max_discard_size &&
3764                             bytes >= (max_discard_size +
3765                                       BTRFS_ASYNC_DISCARD_MIN_FILTER)) {
3766                                 bytes = max_discard_size;
3767                                 extent_bytes = max_discard_size;
3768                                 entry->offset += max_discard_size;
3769                                 entry->bytes -= max_discard_size;
3770                                 link_free_space(ctl, entry);
3771                         } else {
3772                                 kmem_cache_free(btrfs_free_space_cachep, entry);
3773                         }
3774                 } else {
3775                         start = max(start, extent_start);
3776                         bytes = min(extent_start + extent_bytes, end) - start;
3777                         if (bytes < minlen) {
3778                                 spin_unlock(&ctl->tree_lock);
3779                                 mutex_unlock(&ctl->cache_writeout_mutex);
3780                                 goto next;
3781                         }
3782
3783                         unlink_free_space(ctl, entry, true);
3784                         kmem_cache_free(btrfs_free_space_cachep, entry);
3785                 }
3786
3787                 spin_unlock(&ctl->tree_lock);
3788                 trim_entry.start = extent_start;
3789                 trim_entry.bytes = extent_bytes;
3790                 list_add_tail(&trim_entry.list, &ctl->trimming_ranges);
3791                 mutex_unlock(&ctl->cache_writeout_mutex);
3792
3793                 ret = do_trimming(block_group, total_trimmed, start, bytes,
3794                                   extent_start, extent_bytes, extent_trim_state,
3795                                   &trim_entry);
3796                 if (ret) {
3797                         block_group->discard_cursor = start + bytes;
3798                         break;
3799                 }
3800 next:
3801                 start += bytes;
3802                 block_group->discard_cursor = start;
3803                 if (async && *total_trimmed)
3804                         break;
3805
3806                 if (fatal_signal_pending(current)) {
3807                         ret = -ERESTARTSYS;
3808                         break;
3809                 }
3810
3811                 cond_resched();
3812         }
3813
3814         return ret;
3815
3816 out_unlock:
3817         block_group->discard_cursor = btrfs_block_group_end(block_group);
3818         spin_unlock(&ctl->tree_lock);
3819         mutex_unlock(&ctl->cache_writeout_mutex);
3820
3821         return ret;
3822 }
3823
3824 /*
3825  * If we break out of trimming a bitmap prematurely, we should reset the
3826  * trimming bit.  In a rather contrieved case, it's possible to race here so
3827  * reset the state to BTRFS_TRIM_STATE_UNTRIMMED.
3828  *
3829  * start = start of bitmap
3830  * end = near end of bitmap
3831  *
3832  * Thread 1:                    Thread 2:
3833  * trim_bitmaps(start)
3834  *                              trim_bitmaps(end)
3835  *                              end_trimming_bitmap()
3836  * reset_trimming_bitmap()
3837  */
3838 static void reset_trimming_bitmap(struct btrfs_free_space_ctl *ctl, u64 offset)
3839 {
3840         struct btrfs_free_space *entry;
3841
3842         spin_lock(&ctl->tree_lock);
3843         entry = tree_search_offset(ctl, offset, 1, 0);
3844         if (entry) {
3845                 if (btrfs_free_space_trimmed(entry)) {
3846                         ctl->discardable_extents[BTRFS_STAT_CURR] +=
3847                                 entry->bitmap_extents;
3848                         ctl->discardable_bytes[BTRFS_STAT_CURR] += entry->bytes;
3849                 }
3850                 entry->trim_state = BTRFS_TRIM_STATE_UNTRIMMED;
3851         }
3852
3853         spin_unlock(&ctl->tree_lock);
3854 }
3855
3856 static void end_trimming_bitmap(struct btrfs_free_space_ctl *ctl,
3857                                 struct btrfs_free_space *entry)
3858 {
3859         if (btrfs_free_space_trimming_bitmap(entry)) {
3860                 entry->trim_state = BTRFS_TRIM_STATE_TRIMMED;
3861                 ctl->discardable_extents[BTRFS_STAT_CURR] -=
3862                         entry->bitmap_extents;
3863                 ctl->discardable_bytes[BTRFS_STAT_CURR] -= entry->bytes;
3864         }
3865 }
3866
3867 /*
3868  * If @async is set, then we will trim 1 region and return.
3869  */
3870 static int trim_bitmaps(struct btrfs_block_group *block_group,
3871                         u64 *total_trimmed, u64 start, u64 end, u64 minlen,
3872                         u64 maxlen, bool async)
3873 {
3874         struct btrfs_discard_ctl *discard_ctl =
3875                                         &block_group->fs_info->discard_ctl;
3876         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
3877         struct btrfs_free_space *entry;
3878         int ret = 0;
3879         int ret2;
3880         u64 bytes;
3881         u64 offset = offset_to_bitmap(ctl, start);
3882         const u64 max_discard_size = READ_ONCE(discard_ctl->max_discard_size);
3883
3884         while (offset < end) {
3885                 bool next_bitmap = false;
3886                 struct btrfs_trim_range trim_entry;
3887
3888                 mutex_lock(&ctl->cache_writeout_mutex);
3889                 spin_lock(&ctl->tree_lock);
3890
3891                 if (ctl->free_space < minlen) {
3892                         block_group->discard_cursor =
3893                                 btrfs_block_group_end(block_group);
3894                         spin_unlock(&ctl->tree_lock);
3895                         mutex_unlock(&ctl->cache_writeout_mutex);
3896                         break;
3897                 }
3898
3899                 entry = tree_search_offset(ctl, offset, 1, 0);
3900                 /*
3901                  * Bitmaps are marked trimmed lossily now to prevent constant
3902                  * discarding of the same bitmap (the reason why we are bound
3903                  * by the filters).  So, retrim the block group bitmaps when we
3904                  * are preparing to punt to the unused_bgs list.  This uses
3905                  * @minlen to determine if we are in BTRFS_DISCARD_INDEX_UNUSED
3906                  * which is the only discard index which sets minlen to 0.
3907                  */
3908                 if (!entry || (async && minlen && start == offset &&
3909                                btrfs_free_space_trimmed(entry))) {
3910                         spin_unlock(&ctl->tree_lock);
3911                         mutex_unlock(&ctl->cache_writeout_mutex);
3912                         next_bitmap = true;
3913                         goto next;
3914                 }
3915
3916                 /*
3917                  * Async discard bitmap trimming begins at by setting the start
3918                  * to be key.objectid and the offset_to_bitmap() aligns to the
3919                  * start of the bitmap.  This lets us know we are fully
3920                  * scanning the bitmap rather than only some portion of it.
3921                  */
3922                 if (start == offset)
3923                         entry->trim_state = BTRFS_TRIM_STATE_TRIMMING;
3924
3925                 bytes = minlen;
3926                 ret2 = search_bitmap(ctl, entry, &start, &bytes, false);
3927                 if (ret2 || start >= end) {
3928                         /*
3929                          * We lossily consider a bitmap trimmed if we only skip
3930                          * over regions <= BTRFS_ASYNC_DISCARD_MIN_FILTER.
3931                          */
3932                         if (ret2 && minlen <= BTRFS_ASYNC_DISCARD_MIN_FILTER)
3933                                 end_trimming_bitmap(ctl, entry);
3934                         else
3935                                 entry->trim_state = BTRFS_TRIM_STATE_UNTRIMMED;
3936                         spin_unlock(&ctl->tree_lock);
3937                         mutex_unlock(&ctl->cache_writeout_mutex);
3938                         next_bitmap = true;
3939                         goto next;
3940                 }
3941
3942                 /*
3943                  * We already trimmed a region, but are using the locking above
3944                  * to reset the trim_state.
3945                  */
3946                 if (async && *total_trimmed) {
3947                         spin_unlock(&ctl->tree_lock);
3948                         mutex_unlock(&ctl->cache_writeout_mutex);
3949                         goto out;
3950                 }
3951
3952                 bytes = min(bytes, end - start);
3953                 if (bytes < minlen || (async && maxlen && bytes > maxlen)) {
3954                         spin_unlock(&ctl->tree_lock);
3955                         mutex_unlock(&ctl->cache_writeout_mutex);
3956                         goto next;
3957                 }
3958
3959                 /*
3960                  * Let bytes = BTRFS_MAX_DISCARD_SIZE + X.
3961                  * If X < @minlen, we won't trim X when we come back around.
3962                  * So trim it now.  We differ here from trimming extents as we
3963                  * don't keep individual state per bit.
3964                  */
3965                 if (async &&
3966                     max_discard_size &&
3967                     bytes > (max_discard_size + minlen))
3968                         bytes = max_discard_size;
3969
3970                 bitmap_clear_bits(ctl, entry, start, bytes, true);
3971                 if (entry->bytes == 0)
3972                         free_bitmap(ctl, entry);
3973
3974                 spin_unlock(&ctl->tree_lock);
3975                 trim_entry.start = start;
3976                 trim_entry.bytes = bytes;
3977                 list_add_tail(&trim_entry.list, &ctl->trimming_ranges);
3978                 mutex_unlock(&ctl->cache_writeout_mutex);
3979
3980                 ret = do_trimming(block_group, total_trimmed, start, bytes,
3981                                   start, bytes, 0, &trim_entry);
3982                 if (ret) {
3983                         reset_trimming_bitmap(ctl, offset);
3984                         block_group->discard_cursor =
3985                                 btrfs_block_group_end(block_group);
3986                         break;
3987                 }
3988 next:
3989                 if (next_bitmap) {
3990                         offset += BITS_PER_BITMAP * ctl->unit;
3991                         start = offset;
3992                 } else {
3993                         start += bytes;
3994                 }
3995                 block_group->discard_cursor = start;
3996
3997                 if (fatal_signal_pending(current)) {
3998                         if (start != offset)
3999                                 reset_trimming_bitmap(ctl, offset);
4000                         ret = -ERESTARTSYS;
4001                         break;
4002                 }
4003
4004                 cond_resched();
4005         }
4006
4007         if (offset >= end)
4008                 block_group->discard_cursor = end;
4009
4010 out:
4011         return ret;
4012 }
4013
4014 int btrfs_trim_block_group(struct btrfs_block_group *block_group,
4015                            u64 *trimmed, u64 start, u64 end, u64 minlen)
4016 {
4017         struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
4018         int ret;
4019         u64 rem = 0;
4020
4021         ASSERT(!btrfs_is_zoned(block_group->fs_info));
4022
4023         *trimmed = 0;
4024
4025         spin_lock(&block_group->lock);
4026         if (test_bit(BLOCK_GROUP_FLAG_REMOVED, &block_group->runtime_flags)) {
4027                 spin_unlock(&block_group->lock);
4028                 return 0;
4029         }
4030         btrfs_freeze_block_group(block_group);
4031         spin_unlock(&block_group->lock);
4032
4033         ret = trim_no_bitmap(block_group, trimmed, start, end, minlen, false);
4034         if (ret)
4035                 goto out;
4036
4037         ret = trim_bitmaps(block_group, trimmed, start, end, minlen, 0, false);
4038         div64_u64_rem(end, BITS_PER_BITMAP * ctl->unit, &rem);
4039         /* If we ended in the middle of a bitmap, reset the trimming flag */
4040         if (rem)
4041                 reset_trimming_bitmap(ctl, offset_to_bitmap(ctl, end));
4042 out:
4043         btrfs_unfreeze_block_group(block_group);
4044         return ret;
4045 }
4046
4047 int btrfs_trim_block_group_extents(struct btrfs_block_group *block_group,
4048                                    u64 *trimmed, u64 start, u64 end, u64 minlen,
4049                                    bool async)
4050 {
4051         int ret;
4052
4053         *trimmed = 0;
4054
4055         spin_lock(&block_group->lock);
4056         if (test_bit(BLOCK_GROUP_FLAG_REMOVED, &block_group->runtime_flags)) {
4057                 spin_unlock(&block_group->lock);
4058                 return 0;
4059         }
4060         btrfs_freeze_block_group(block_group);
4061         spin_unlock(&block_group->lock);
4062
4063         ret = trim_no_bitmap(block_group, trimmed, start, end, minlen, async);
4064         btrfs_unfreeze_block_group(block_group);
4065
4066         return ret;
4067 }
4068
4069 int btrfs_trim_block_group_bitmaps(struct btrfs_block_group *block_group,
4070                                    u64 *trimmed, u64 start, u64 end, u64 minlen,
4071                                    u64 maxlen, bool async)
4072 {
4073         int ret;
4074
4075         *trimmed = 0;
4076
4077         spin_lock(&block_group->lock);
4078         if (test_bit(BLOCK_GROUP_FLAG_REMOVED, &block_group->runtime_flags)) {
4079                 spin_unlock(&block_group->lock);
4080                 return 0;
4081         }
4082         btrfs_freeze_block_group(block_group);
4083         spin_unlock(&block_group->lock);
4084
4085         ret = trim_bitmaps(block_group, trimmed, start, end, minlen, maxlen,
4086                            async);
4087
4088         btrfs_unfreeze_block_group(block_group);
4089
4090         return ret;
4091 }
4092
4093 bool btrfs_free_space_cache_v1_active(struct btrfs_fs_info *fs_info)
4094 {
4095         return btrfs_super_cache_generation(fs_info->super_copy);
4096 }
4097
4098 static int cleanup_free_space_cache_v1(struct btrfs_fs_info *fs_info,
4099                                        struct btrfs_trans_handle *trans)
4100 {
4101         struct btrfs_block_group *block_group;
4102         struct rb_node *node;
4103         int ret = 0;
4104
4105         btrfs_info(fs_info, "cleaning free space cache v1");
4106
4107         node = rb_first_cached(&fs_info->block_group_cache_tree);
4108         while (node) {
4109                 block_group = rb_entry(node, struct btrfs_block_group, cache_node);
4110                 ret = btrfs_remove_free_space_inode(trans, NULL, block_group);
4111                 if (ret)
4112                         goto out;
4113                 node = rb_next(node);
4114         }
4115 out:
4116         return ret;
4117 }
4118
4119 int btrfs_set_free_space_cache_v1_active(struct btrfs_fs_info *fs_info, bool active)
4120 {
4121         struct btrfs_trans_handle *trans;
4122         int ret;
4123
4124         /*
4125          * update_super_roots will appropriately set or unset
4126          * super_copy->cache_generation based on SPACE_CACHE and
4127          * BTRFS_FS_CLEANUP_SPACE_CACHE_V1. For this reason, we need a
4128          * transaction commit whether we are enabling space cache v1 and don't
4129          * have any other work to do, or are disabling it and removing free
4130          * space inodes.
4131          */
4132         trans = btrfs_start_transaction(fs_info->tree_root, 0);
4133         if (IS_ERR(trans))
4134                 return PTR_ERR(trans);
4135
4136         if (!active) {
4137                 set_bit(BTRFS_FS_CLEANUP_SPACE_CACHE_V1, &fs_info->flags);
4138                 ret = cleanup_free_space_cache_v1(fs_info, trans);
4139                 if (ret) {
4140                         btrfs_abort_transaction(trans, ret);
4141                         btrfs_end_transaction(trans);
4142                         goto out;
4143                 }
4144         }
4145
4146         ret = btrfs_commit_transaction(trans);
4147 out:
4148         clear_bit(BTRFS_FS_CLEANUP_SPACE_CACHE_V1, &fs_info->flags);
4149
4150         return ret;
4151 }
4152
4153 int __init btrfs_free_space_init(void)
4154 {
4155         btrfs_free_space_cachep = kmem_cache_create("btrfs_free_space",
4156                         sizeof(struct btrfs_free_space), 0,
4157                         SLAB_MEM_SPREAD, NULL);
4158         if (!btrfs_free_space_cachep)
4159                 return -ENOMEM;
4160
4161         btrfs_free_space_bitmap_cachep = kmem_cache_create("btrfs_free_space_bitmap",
4162                                                         PAGE_SIZE, PAGE_SIZE,
4163                                                         SLAB_MEM_SPREAD, NULL);
4164         if (!btrfs_free_space_bitmap_cachep) {
4165                 kmem_cache_destroy(btrfs_free_space_cachep);
4166                 return -ENOMEM;
4167         }
4168
4169         return 0;
4170 }
4171
4172 void __cold btrfs_free_space_exit(void)
4173 {
4174         kmem_cache_destroy(btrfs_free_space_cachep);
4175         kmem_cache_destroy(btrfs_free_space_bitmap_cachep);
4176 }
4177
4178 #ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
4179 /*
4180  * Use this if you need to make a bitmap or extent entry specifically, it
4181  * doesn't do any of the merging that add_free_space does, this acts a lot like
4182  * how the free space cache loading stuff works, so you can get really weird
4183  * configurations.
4184  */
4185 int test_add_free_space_entry(struct btrfs_block_group *cache,
4186                               u64 offset, u64 bytes, bool bitmap)
4187 {
4188         struct btrfs_free_space_ctl *ctl = cache->free_space_ctl;
4189         struct btrfs_free_space *info = NULL, *bitmap_info;
4190         void *map = NULL;
4191         enum btrfs_trim_state trim_state = BTRFS_TRIM_STATE_TRIMMED;
4192         u64 bytes_added;
4193         int ret;
4194
4195 again:
4196         if (!info) {
4197                 info = kmem_cache_zalloc(btrfs_free_space_cachep, GFP_NOFS);
4198                 if (!info)
4199                         return -ENOMEM;
4200         }
4201
4202         if (!bitmap) {
4203                 spin_lock(&ctl->tree_lock);
4204                 info->offset = offset;
4205                 info->bytes = bytes;
4206                 info->max_extent_size = 0;
4207                 ret = link_free_space(ctl, info);
4208                 spin_unlock(&ctl->tree_lock);
4209                 if (ret)
4210                         kmem_cache_free(btrfs_free_space_cachep, info);
4211                 return ret;
4212         }
4213
4214         if (!map) {
4215                 map = kmem_cache_zalloc(btrfs_free_space_bitmap_cachep, GFP_NOFS);
4216                 if (!map) {
4217                         kmem_cache_free(btrfs_free_space_cachep, info);
4218                         return -ENOMEM;
4219                 }
4220         }
4221
4222         spin_lock(&ctl->tree_lock);
4223         bitmap_info = tree_search_offset(ctl, offset_to_bitmap(ctl, offset),
4224                                          1, 0);
4225         if (!bitmap_info) {
4226                 info->bitmap = map;
4227                 map = NULL;
4228                 add_new_bitmap(ctl, info, offset);
4229                 bitmap_info = info;
4230                 info = NULL;
4231         }
4232
4233         bytes_added = add_bytes_to_bitmap(ctl, bitmap_info, offset, bytes,
4234                                           trim_state);
4235
4236         bytes -= bytes_added;
4237         offset += bytes_added;
4238         spin_unlock(&ctl->tree_lock);
4239
4240         if (bytes)
4241                 goto again;
4242
4243         if (info)
4244                 kmem_cache_free(btrfs_free_space_cachep, info);
4245         if (map)
4246                 kmem_cache_free(btrfs_free_space_bitmap_cachep, map);
4247         return 0;
4248 }
4249
4250 /*
4251  * Checks to see if the given range is in the free space cache.  This is really
4252  * just used to check the absence of space, so if there is free space in the
4253  * range at all we will return 1.
4254  */
4255 int test_check_exists(struct btrfs_block_group *cache,
4256                       u64 offset, u64 bytes)
4257 {
4258         struct btrfs_free_space_ctl *ctl = cache->free_space_ctl;
4259         struct btrfs_free_space *info;
4260         int ret = 0;
4261
4262         spin_lock(&ctl->tree_lock);
4263         info = tree_search_offset(ctl, offset, 0, 0);
4264         if (!info) {
4265                 info = tree_search_offset(ctl, offset_to_bitmap(ctl, offset),
4266                                           1, 0);
4267                 if (!info)
4268                         goto out;
4269         }
4270
4271 have_info:
4272         if (info->bitmap) {
4273                 u64 bit_off, bit_bytes;
4274                 struct rb_node *n;
4275                 struct btrfs_free_space *tmp;
4276
4277                 bit_off = offset;
4278                 bit_bytes = ctl->unit;
4279                 ret = search_bitmap(ctl, info, &bit_off, &bit_bytes, false);
4280                 if (!ret) {
4281                         if (bit_off == offset) {
4282                                 ret = 1;
4283                                 goto out;
4284                         } else if (bit_off > offset &&
4285                                    offset + bytes > bit_off) {
4286                                 ret = 1;
4287                                 goto out;
4288                         }
4289                 }
4290
4291                 n = rb_prev(&info->offset_index);
4292                 while (n) {
4293                         tmp = rb_entry(n, struct btrfs_free_space,
4294                                        offset_index);
4295                         if (tmp->offset + tmp->bytes < offset)
4296                                 break;
4297                         if (offset + bytes < tmp->offset) {
4298                                 n = rb_prev(&tmp->offset_index);
4299                                 continue;
4300                         }
4301                         info = tmp;
4302                         goto have_info;
4303                 }
4304
4305                 n = rb_next(&info->offset_index);
4306                 while (n) {
4307                         tmp = rb_entry(n, struct btrfs_free_space,
4308                                        offset_index);
4309                         if (offset + bytes < tmp->offset)
4310                                 break;
4311                         if (tmp->offset + tmp->bytes < offset) {
4312                                 n = rb_next(&tmp->offset_index);
4313                                 continue;
4314                         }
4315                         info = tmp;
4316                         goto have_info;
4317                 }
4318
4319                 ret = 0;
4320                 goto out;
4321         }
4322
4323         if (info->offset == offset) {
4324                 ret = 1;
4325                 goto out;
4326         }
4327
4328         if (offset > info->offset && offset < info->offset + info->bytes)
4329                 ret = 1;
4330 out:
4331         spin_unlock(&ctl->tree_lock);
4332         return ret;
4333 }
4334 #endif /* CONFIG_BTRFS_FS_RUN_SANITY_TESTS */