| 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 */ |