1 // SPDX-License-Identifier: GPL-2.0 2 /* 3 * Copyright (C) 2008 Oracle. All rights reserved. 4 */ 5 6 #include <linux/sched.h> 7 #include <linux/pagemap.h> 8 #include <linux/spinlock.h> 9 #include <linux/page-flags.h> 10 #include <asm/bug.h> 11 #include "misc.h" 12 #include "ctree.h" 13 #include "extent_io.h" 14 #include "locking.h" 15 16 /* 17 * Extent buffer locking 18 * ===================== 19 * 20 * The locks use a custom scheme that allows to do more operations than are 21 * available fromt current locking primitives. The building blocks are still 22 * rwlock and wait queues. 23 * 24 * Required semantics: 25 * 26 * - reader/writer exclusion 27 * - writer/writer exclusion 28 * - reader/reader sharing 29 * - spinning lock semantics 30 * - blocking lock semantics 31 * - try-lock semantics for readers and writers 32 * - one level nesting, allowing read lock to be taken by the same thread that 33 * already has write lock 34 * 35 * The extent buffer locks (also called tree locks) manage access to eb data 36 * related to the storage in the b-tree (keys, items, but not the individual 37 * members of eb). 38 * We want concurrency of many readers and safe updates. The underlying locking 39 * is done by read-write spinlock and the blocking part is implemented using 40 * counters and wait queues. 41 * 42 * spinning semantics - the low-level rwlock is held so all other threads that 43 * want to take it are spinning on it. 44 * 45 * blocking semantics - the low-level rwlock is not held but the counter 46 * denotes how many times the blocking lock was held; 47 * sleeping is possible 48 * 49 * Write lock always allows only one thread to access the data. 50 * 51 * 52 * Debugging 53 * --------- 54 * 55 * There are additional state counters that are asserted in various contexts, 56 * removed from non-debug build to reduce extent_buffer size and for 57 * performance reasons. 58 * 59 * 60 * Lock recursion 61 * -------------- 62 * 63 * A write operation on a tree might indirectly start a look up on the same 64 * tree. This can happen when btrfs_cow_block locks the tree and needs to 65 * lookup free extents. 66 * 67 * btrfs_cow_block 68 * .. 69 * alloc_tree_block_no_bg_flush 70 * btrfs_alloc_tree_block 71 * btrfs_reserve_extent 72 * .. 73 * load_free_space_cache 74 * .. 75 * btrfs_lookup_file_extent 76 * btrfs_search_slot 77 * 78 * 79 * Locking pattern - spinning 80 * -------------------------- 81 * 82 * The simple locking scenario, the +--+ denotes the spinning section. 83 * 84 * +- btrfs_tree_lock 85 * | - extent_buffer::rwlock is held 86 * | - no heavy operations should happen, eg. IO, memory allocations, large 87 * | structure traversals 88 * +- btrfs_tree_unock 89 * 90 * 91 * Locking pattern - blocking 92 * -------------------------- 93 * 94 * The blocking write uses the following scheme. The +--+ denotes the spinning 95 * section. 96 * 97 * +- btrfs_tree_lock 98 * | 99 * +- btrfs_set_lock_blocking_write 100 * 101 * - allowed: IO, memory allocations, etc. 102 * 103 * -- btrfs_tree_unlock - note, no explicit unblocking necessary 104 * 105 * 106 * Blocking read is similar. 107 * 108 * +- btrfs_tree_read_lock 109 * | 110 * +- btrfs_set_lock_blocking_read 111 * 112 * - heavy operations allowed 113 * 114 * +- btrfs_tree_read_unlock_blocking 115 * | 116 * +- btrfs_tree_read_unlock 117 * 118 */ 119 120 #ifdef CONFIG_BTRFS_DEBUG 121 static inline void btrfs_assert_spinning_writers_get(struct extent_buffer *eb) 122 { 123 WARN_ON(eb->spinning_writers); 124 eb->spinning_writers++; 125 } 126 127 static inline void btrfs_assert_spinning_writers_put(struct extent_buffer *eb) 128 { 129 WARN_ON(eb->spinning_writers != 1); 130 eb->spinning_writers--; 131 } 132 133 static inline void btrfs_assert_no_spinning_writers(struct extent_buffer *eb) 134 { 135 WARN_ON(eb->spinning_writers); 136 } 137 138 static inline void btrfs_assert_spinning_readers_get(struct extent_buffer *eb) 139 { 140 atomic_inc(&eb->spinning_readers); 141 } 142 143 static inline void btrfs_assert_spinning_readers_put(struct extent_buffer *eb) 144 { 145 WARN_ON(atomic_read(&eb->spinning_readers) == 0); 146 atomic_dec(&eb->spinning_readers); 147 } 148 149 static inline void btrfs_assert_tree_read_locks_get(struct extent_buffer *eb) 150 { 151 atomic_inc(&eb->read_locks); 152 } 153 154 static inline void btrfs_assert_tree_read_locks_put(struct extent_buffer *eb) 155 { 156 atomic_dec(&eb->read_locks); 157 } 158 159 static inline void btrfs_assert_tree_read_locked(struct extent_buffer *eb) 160 { 161 BUG_ON(!atomic_read(&eb->read_locks)); 162 } 163 164 static inline void btrfs_assert_tree_write_locks_get(struct extent_buffer *eb) 165 { 166 eb->write_locks++; 167 } 168 169 static inline void btrfs_assert_tree_write_locks_put(struct extent_buffer *eb) 170 { 171 eb->write_locks--; 172 } 173 174 #else 175 static void btrfs_assert_spinning_writers_get(struct extent_buffer *eb) { } 176 static void btrfs_assert_spinning_writers_put(struct extent_buffer *eb) { } 177 static void btrfs_assert_no_spinning_writers(struct extent_buffer *eb) { } 178 static void btrfs_assert_spinning_readers_put(struct extent_buffer *eb) { } 179 static void btrfs_assert_spinning_readers_get(struct extent_buffer *eb) { } 180 static void btrfs_assert_tree_read_locked(struct extent_buffer *eb) { } 181 static void btrfs_assert_tree_read_locks_get(struct extent_buffer *eb) { } 182 static void btrfs_assert_tree_read_locks_put(struct extent_buffer *eb) { } 183 static void btrfs_assert_tree_write_locks_get(struct extent_buffer *eb) { } 184 static void btrfs_assert_tree_write_locks_put(struct extent_buffer *eb) { } 185 #endif 186 187 /* 188 * Mark already held read lock as blocking. Can be nested in write lock by the 189 * same thread. 190 * 191 * Use when there are potentially long operations ahead so other thread waiting 192 * on the lock will not actively spin but sleep instead. 193 * 194 * The rwlock is released and blocking reader counter is increased. 195 */ 196 void btrfs_set_lock_blocking_read(struct extent_buffer *eb) 197 { 198 trace_btrfs_set_lock_blocking_read(eb); 199 /* 200 * No lock is required. The lock owner may change if we have a read 201 * lock, but it won't change to or away from us. If we have the write 202 * lock, we are the owner and it'll never change. 203 */ 204 if (eb->lock_recursed && current->pid == eb->lock_owner) 205 return; 206 btrfs_assert_tree_read_locked(eb); 207 atomic_inc(&eb->blocking_readers); 208 btrfs_assert_spinning_readers_put(eb); 209 read_unlock(&eb->lock); 210 } 211 212 /* 213 * Mark already held write lock as blocking. 214 * 215 * Use when there are potentially long operations ahead so other threads 216 * waiting on the lock will not actively spin but sleep instead. 217 * 218 * The rwlock is released and blocking writers is set. 219 */ 220 void btrfs_set_lock_blocking_write(struct extent_buffer *eb) 221 { 222 trace_btrfs_set_lock_blocking_write(eb); 223 /* 224 * No lock is required. The lock owner may change if we have a read 225 * lock, but it won't change to or away from us. If we have the write 226 * lock, we are the owner and it'll never change. 227 */ 228 if (eb->lock_recursed && current->pid == eb->lock_owner) 229 return; 230 if (eb->blocking_writers == 0) { 231 btrfs_assert_spinning_writers_put(eb); 232 btrfs_assert_tree_locked(eb); 233 WRITE_ONCE(eb->blocking_writers, 1); 234 write_unlock(&eb->lock); 235 } 236 } 237 238 /* 239 * Lock the extent buffer for read. Wait for any writers (spinning or blocking). 240 * Can be nested in write lock by the same thread. 241 * 242 * Use when the locked section does only lightweight actions and busy waiting 243 * would be cheaper than making other threads do the wait/wake loop. 244 * 245 * The rwlock is held upon exit. 246 */ 247 void __btrfs_tree_read_lock(struct extent_buffer *eb, enum btrfs_lock_nesting nest, 248 bool recurse) 249 { 250 u64 start_ns = 0; 251 252 if (trace_btrfs_tree_read_lock_enabled()) 253 start_ns = ktime_get_ns(); 254 again: 255 read_lock(&eb->lock); 256 BUG_ON(eb->blocking_writers == 0 && 257 current->pid == eb->lock_owner); 258 if (eb->blocking_writers) { 259 if (current->pid == eb->lock_owner) { 260 /* 261 * This extent is already write-locked by our thread. 262 * We allow an additional read lock to be added because 263 * it's for the same thread. btrfs_find_all_roots() 264 * depends on this as it may be called on a partly 265 * (write-)locked tree. 266 */ 267 WARN_ON(!recurse); 268 BUG_ON(eb->lock_recursed); 269 eb->lock_recursed = true; 270 read_unlock(&eb->lock); 271 trace_btrfs_tree_read_lock(eb, start_ns); 272 return; 273 } 274 read_unlock(&eb->lock); 275 wait_event(eb->write_lock_wq, 276 READ_ONCE(eb->blocking_writers) == 0); 277 goto again; 278 } 279 btrfs_assert_tree_read_locks_get(eb); 280 btrfs_assert_spinning_readers_get(eb); 281 trace_btrfs_tree_read_lock(eb, start_ns); 282 } 283 284 void btrfs_tree_read_lock(struct extent_buffer *eb) 285 { 286 __btrfs_tree_read_lock(eb, BTRFS_NESTING_NORMAL, false); 287 } 288 289 /* 290 * Lock extent buffer for read, optimistically expecting that there are no 291 * contending blocking writers. If there are, don't wait. 292 * 293 * Return 1 if the rwlock has been taken, 0 otherwise 294 */ 295 int btrfs_tree_read_lock_atomic(struct extent_buffer *eb) 296 { 297 if (READ_ONCE(eb->blocking_writers)) 298 return 0; 299 300 read_lock(&eb->lock); 301 /* Refetch value after lock */ 302 if (READ_ONCE(eb->blocking_writers)) { 303 read_unlock(&eb->lock); 304 return 0; 305 } 306 btrfs_assert_tree_read_locks_get(eb); 307 btrfs_assert_spinning_readers_get(eb); 308 trace_btrfs_tree_read_lock_atomic(eb); 309 return 1; 310 } 311 312 /* 313 * Try-lock for read. Don't block or wait for contending writers. 314 * 315 * Retrun 1 if the rwlock has been taken, 0 otherwise 316 */ 317 int btrfs_try_tree_read_lock(struct extent_buffer *eb) 318 { 319 if (READ_ONCE(eb->blocking_writers)) 320 return 0; 321 322 if (!read_trylock(&eb->lock)) 323 return 0; 324 325 /* Refetch value after lock */ 326 if (READ_ONCE(eb->blocking_writers)) { 327 read_unlock(&eb->lock); 328 return 0; 329 } 330 btrfs_assert_tree_read_locks_get(eb); 331 btrfs_assert_spinning_readers_get(eb); 332 trace_btrfs_try_tree_read_lock(eb); 333 return 1; 334 } 335 336 /* 337 * Try-lock for write. May block until the lock is uncontended, but does not 338 * wait until it is free. 339 * 340 * Retrun 1 if the rwlock has been taken, 0 otherwise 341 */ 342 int btrfs_try_tree_write_lock(struct extent_buffer *eb) 343 { 344 if (READ_ONCE(eb->blocking_writers) || atomic_read(&eb->blocking_readers)) 345 return 0; 346 347 write_lock(&eb->lock); 348 /* Refetch value after lock */ 349 if (READ_ONCE(eb->blocking_writers) || atomic_read(&eb->blocking_readers)) { 350 write_unlock(&eb->lock); 351 return 0; 352 } 353 btrfs_assert_tree_write_locks_get(eb); 354 btrfs_assert_spinning_writers_get(eb); 355 eb->lock_owner = current->pid; 356 trace_btrfs_try_tree_write_lock(eb); 357 return 1; 358 } 359 360 /* 361 * Release read lock. Must be used only if the lock is in spinning mode. If 362 * the read lock is nested, must pair with read lock before the write unlock. 363 * 364 * The rwlock is not held upon exit. 365 */ 366 void btrfs_tree_read_unlock(struct extent_buffer *eb) 367 { 368 trace_btrfs_tree_read_unlock(eb); 369 /* 370 * if we're nested, we have the write lock. No new locking 371 * is needed as long as we are the lock owner. 372 * The write unlock will do a barrier for us, and the lock_recursed 373 * field only matters to the lock owner. 374 */ 375 if (eb->lock_recursed && current->pid == eb->lock_owner) { 376 eb->lock_recursed = false; 377 return; 378 } 379 btrfs_assert_tree_read_locked(eb); 380 btrfs_assert_spinning_readers_put(eb); 381 btrfs_assert_tree_read_locks_put(eb); 382 read_unlock(&eb->lock); 383 } 384 385 /* 386 * Release read lock, previously set to blocking by a pairing call to 387 * btrfs_set_lock_blocking_read(). Can be nested in write lock by the same 388 * thread. 389 * 390 * State of rwlock is unchanged, last reader wakes waiting threads. 391 */ 392 void btrfs_tree_read_unlock_blocking(struct extent_buffer *eb) 393 { 394 trace_btrfs_tree_read_unlock_blocking(eb); 395 /* 396 * if we're nested, we have the write lock. No new locking 397 * is needed as long as we are the lock owner. 398 * The write unlock will do a barrier for us, and the lock_recursed 399 * field only matters to the lock owner. 400 */ 401 if (eb->lock_recursed && current->pid == eb->lock_owner) { 402 eb->lock_recursed = false; 403 return; 404 } 405 btrfs_assert_tree_read_locked(eb); 406 WARN_ON(atomic_read(&eb->blocking_readers) == 0); 407 /* atomic_dec_and_test implies a barrier */ 408 if (atomic_dec_and_test(&eb->blocking_readers)) 409 cond_wake_up_nomb(&eb->read_lock_wq); 410 btrfs_assert_tree_read_locks_put(eb); 411 } 412 413 /* 414 * Lock for write. Wait for all blocking and spinning readers and writers. This 415 * starts context where reader lock could be nested by the same thread. 416 * 417 * The rwlock is held for write upon exit. 418 */ 419 void __btrfs_tree_lock(struct extent_buffer *eb, enum btrfs_lock_nesting nest) 420 __acquires(&eb->lock) 421 { 422 u64 start_ns = 0; 423 424 if (trace_btrfs_tree_lock_enabled()) 425 start_ns = ktime_get_ns(); 426 427 WARN_ON(eb->lock_owner == current->pid); 428 again: 429 wait_event(eb->read_lock_wq, atomic_read(&eb->blocking_readers) == 0); 430 wait_event(eb->write_lock_wq, READ_ONCE(eb->blocking_writers) == 0); 431 write_lock(&eb->lock); 432 /* Refetch value after lock */ 433 if (atomic_read(&eb->blocking_readers) || 434 READ_ONCE(eb->blocking_writers)) { 435 write_unlock(&eb->lock); 436 goto again; 437 } 438 btrfs_assert_spinning_writers_get(eb); 439 btrfs_assert_tree_write_locks_get(eb); 440 eb->lock_owner = current->pid; 441 trace_btrfs_tree_lock(eb, start_ns); 442 } 443 444 void btrfs_tree_lock(struct extent_buffer *eb) 445 { 446 __btrfs_tree_lock(eb, BTRFS_NESTING_NORMAL); 447 } 448 449 /* 450 * Release the write lock, either blocking or spinning (ie. there's no need 451 * for an explicit blocking unlock, like btrfs_tree_read_unlock_blocking). 452 * This also ends the context for nesting, the read lock must have been 453 * released already. 454 * 455 * Tasks blocked and waiting are woken, rwlock is not held upon exit. 456 */ 457 void btrfs_tree_unlock(struct extent_buffer *eb) 458 { 459 /* 460 * This is read both locked and unlocked but always by the same thread 461 * that already owns the lock so we don't need to use READ_ONCE 462 */ 463 int blockers = eb->blocking_writers; 464 465 BUG_ON(blockers > 1); 466 467 btrfs_assert_tree_locked(eb); 468 trace_btrfs_tree_unlock(eb); 469 eb->lock_owner = 0; 470 btrfs_assert_tree_write_locks_put(eb); 471 472 if (blockers) { 473 btrfs_assert_no_spinning_writers(eb); 474 /* Unlocked write */ 475 WRITE_ONCE(eb->blocking_writers, 0); 476 /* 477 * We need to order modifying blocking_writers above with 478 * actually waking up the sleepers to ensure they see the 479 * updated value of blocking_writers 480 */ 481 cond_wake_up(&eb->write_lock_wq); 482 } else { 483 btrfs_assert_spinning_writers_put(eb); 484 write_unlock(&eb->lock); 485 } 486 } 487 488 /* 489 * Set all locked nodes in the path to blocking locks. This should be done 490 * before scheduling 491 */ 492 void btrfs_set_path_blocking(struct btrfs_path *p) 493 { 494 int i; 495 496 for (i = 0; i < BTRFS_MAX_LEVEL; i++) { 497 if (!p->nodes[i] || !p->locks[i]) 498 continue; 499 /* 500 * If we currently have a spinning reader or writer lock this 501 * will bump the count of blocking holders and drop the 502 * spinlock. 503 */ 504 if (p->locks[i] == BTRFS_READ_LOCK) { 505 btrfs_set_lock_blocking_read(p->nodes[i]); 506 p->locks[i] = BTRFS_READ_LOCK_BLOCKING; 507 } else if (p->locks[i] == BTRFS_WRITE_LOCK) { 508 btrfs_set_lock_blocking_write(p->nodes[i]); 509 p->locks[i] = BTRFS_WRITE_LOCK_BLOCKING; 510 } 511 } 512 } 513 514 /* 515 * This releases any locks held in the path starting at level and going all the 516 * way up to the root. 517 * 518 * btrfs_search_slot will keep the lock held on higher nodes in a few corner 519 * cases, such as COW of the block at slot zero in the node. This ignores 520 * those rules, and it should only be called when there are no more updates to 521 * be done higher up in the tree. 522 */ 523 void btrfs_unlock_up_safe(struct btrfs_path *path, int level) 524 { 525 int i; 526 527 if (path->keep_locks) 528 return; 529 530 for (i = level; i < BTRFS_MAX_LEVEL; i++) { 531 if (!path->nodes[i]) 532 continue; 533 if (!path->locks[i]) 534 continue; 535 btrfs_tree_unlock_rw(path->nodes[i], path->locks[i]); 536 path->locks[i] = 0; 537 } 538 } 539 540 /* 541 * Loop around taking references on and locking the root node of the tree until 542 * we end up with a lock on the root node. 543 * 544 * Return: root extent buffer with write lock held 545 */ 546 struct extent_buffer *btrfs_lock_root_node(struct btrfs_root *root) 547 { 548 struct extent_buffer *eb; 549 550 while (1) { 551 eb = btrfs_root_node(root); 552 btrfs_tree_lock(eb); 553 if (eb == root->node) 554 break; 555 btrfs_tree_unlock(eb); 556 free_extent_buffer(eb); 557 } 558 return eb; 559 } 560 561 /* 562 * Loop around taking references on and locking the root node of the tree until 563 * we end up with a lock on the root node. 564 * 565 * Return: root extent buffer with read lock held 566 */ 567 struct extent_buffer *__btrfs_read_lock_root_node(struct btrfs_root *root, 568 bool recurse) 569 { 570 struct extent_buffer *eb; 571 572 while (1) { 573 eb = btrfs_root_node(root); 574 __btrfs_tree_read_lock(eb, BTRFS_NESTING_NORMAL, recurse); 575 if (eb == root->node) 576 break; 577 btrfs_tree_read_unlock(eb); 578 free_extent_buffer(eb); 579 } 580 return eb; 581 } 582 583 /* 584 * DREW locks 585 * ========== 586 * 587 * DREW stands for double-reader-writer-exclusion lock. It's used in situation 588 * where you want to provide A-B exclusion but not AA or BB. 589 * 590 * Currently implementation gives more priority to reader. If a reader and a 591 * writer both race to acquire their respective sides of the lock the writer 592 * would yield its lock as soon as it detects a concurrent reader. Additionally 593 * if there are pending readers no new writers would be allowed to come in and 594 * acquire the lock. 595 */ 596 597 int btrfs_drew_lock_init(struct btrfs_drew_lock *lock) 598 { 599 int ret; 600 601 ret = percpu_counter_init(&lock->writers, 0, GFP_KERNEL); 602 if (ret) 603 return ret; 604 605 atomic_set(&lock->readers, 0); 606 init_waitqueue_head(&lock->pending_readers); 607 init_waitqueue_head(&lock->pending_writers); 608 609 return 0; 610 } 611 612 void btrfs_drew_lock_destroy(struct btrfs_drew_lock *lock) 613 { 614 percpu_counter_destroy(&lock->writers); 615 } 616 617 /* Return true if acquisition is successful, false otherwise */ 618 bool btrfs_drew_try_write_lock(struct btrfs_drew_lock *lock) 619 { 620 if (atomic_read(&lock->readers)) 621 return false; 622 623 percpu_counter_inc(&lock->writers); 624 625 /* Ensure writers count is updated before we check for pending readers */ 626 smp_mb(); 627 if (atomic_read(&lock->readers)) { 628 btrfs_drew_write_unlock(lock); 629 return false; 630 } 631 632 return true; 633 } 634 635 void btrfs_drew_write_lock(struct btrfs_drew_lock *lock) 636 { 637 while (true) { 638 if (btrfs_drew_try_write_lock(lock)) 639 return; 640 wait_event(lock->pending_writers, !atomic_read(&lock->readers)); 641 } 642 } 643 644 void btrfs_drew_write_unlock(struct btrfs_drew_lock *lock) 645 { 646 percpu_counter_dec(&lock->writers); 647 cond_wake_up(&lock->pending_readers); 648 } 649 650 void btrfs_drew_read_lock(struct btrfs_drew_lock *lock) 651 { 652 atomic_inc(&lock->readers); 653 654 /* 655 * Ensure the pending reader count is perceieved BEFORE this reader 656 * goes to sleep in case of active writers. This guarantees new writers 657 * won't be allowed and that the current reader will be woken up when 658 * the last active writer finishes its jobs. 659 */ 660 smp_mb__after_atomic(); 661 662 wait_event(lock->pending_readers, 663 percpu_counter_sum(&lock->writers) == 0); 664 } 665 666 void btrfs_drew_read_unlock(struct btrfs_drew_lock *lock) 667 { 668 /* 669 * atomic_dec_and_test implies a full barrier, so woken up writers 670 * are guaranteed to see the decrement 671 */ 672 if (atomic_dec_and_test(&lock->readers)) 673 wake_up(&lock->pending_writers); 674 } 675