Linux Audio

Check our new training course

Loading...
Note: File does not exist in v3.1.
   1// SPDX-License-Identifier: GPL-2.0
   2/*
   3 * Copyright (C) 2010 Kent Overstreet <kent.overstreet@gmail.com>
   4 *
   5 * Uses a block device as cache for other block devices; optimized for SSDs.
   6 * All allocation is done in buckets, which should match the erase block size
   7 * of the device.
   8 *
   9 * Buckets containing cached data are kept on a heap sorted by priority;
  10 * bucket priority is increased on cache hit, and periodically all the buckets
  11 * on the heap have their priority scaled down. This currently is just used as
  12 * an LRU but in the future should allow for more intelligent heuristics.
  13 *
  14 * Buckets have an 8 bit counter; freeing is accomplished by incrementing the
  15 * counter. Garbage collection is used to remove stale pointers.
  16 *
  17 * Indexing is done via a btree; nodes are not necessarily fully sorted, rather
  18 * as keys are inserted we only sort the pages that have not yet been written.
  19 * When garbage collection is run, we resort the entire node.
  20 *
  21 * All configuration is done via sysfs; see Documentation/admin-guide/bcache.rst.
  22 */
  23
  24#include "bcache.h"
  25#include "btree.h"
  26#include "debug.h"
  27#include "extents.h"
  28
  29#include <linux/slab.h>
  30#include <linux/bitops.h>
  31#include <linux/hash.h>
  32#include <linux/kthread.h>
  33#include <linux/prefetch.h>
  34#include <linux/random.h>
  35#include <linux/rcupdate.h>
  36#include <linux/sched/clock.h>
  37#include <linux/rculist.h>
  38#include <linux/delay.h>
  39#include <trace/events/bcache.h>
  40
  41/*
  42 * Todo:
  43 * register_bcache: Return errors out to userspace correctly
  44 *
  45 * Writeback: don't undirty key until after a cache flush
  46 *
  47 * Create an iterator for key pointers
  48 *
  49 * On btree write error, mark bucket such that it won't be freed from the cache
  50 *
  51 * Journalling:
  52 *   Check for bad keys in replay
  53 *   Propagate barriers
  54 *   Refcount journal entries in journal_replay
  55 *
  56 * Garbage collection:
  57 *   Finish incremental gc
  58 *   Gc should free old UUIDs, data for invalid UUIDs
  59 *
  60 * Provide a way to list backing device UUIDs we have data cached for, and
  61 * probably how long it's been since we've seen them, and a way to invalidate
  62 * dirty data for devices that will never be attached again
  63 *
  64 * Keep 1 min/5 min/15 min statistics of how busy a block device has been, so
  65 * that based on that and how much dirty data we have we can keep writeback
  66 * from being starved
  67 *
  68 * Add a tracepoint or somesuch to watch for writeback starvation
  69 *
  70 * When btree depth > 1 and splitting an interior node, we have to make sure
  71 * alloc_bucket() cannot fail. This should be true but is not completely
  72 * obvious.
  73 *
  74 * Plugging?
  75 *
  76 * If data write is less than hard sector size of ssd, round up offset in open
  77 * bucket to the next whole sector
  78 *
  79 * Superblock needs to be fleshed out for multiple cache devices
  80 *
  81 * Add a sysfs tunable for the number of writeback IOs in flight
  82 *
  83 * Add a sysfs tunable for the number of open data buckets
  84 *
  85 * IO tracking: Can we track when one process is doing io on behalf of another?
  86 * IO tracking: Don't use just an average, weigh more recent stuff higher
  87 *
  88 * Test module load/unload
  89 */
  90
  91#define MAX_NEED_GC		64
  92#define MAX_SAVE_PRIO		72
  93#define MAX_GC_TIMES		100
  94#define MIN_GC_NODES		100
  95#define GC_SLEEP_MS		100
  96
  97#define PTR_DIRTY_BIT		(((uint64_t) 1 << 36))
  98
  99#define PTR_HASH(c, k)							\
 100	(((k)->ptr[0] >> c->bucket_bits) | PTR_GEN(k, 0))
 101
 102#define insert_lock(s, b)	((b)->level <= (s)->lock)
 103
 104
 105static inline struct bset *write_block(struct btree *b)
 106{
 107	return ((void *) btree_bset_first(b)) + b->written * block_bytes(b->c);
 108}
 109
 110static void bch_btree_init_next(struct btree *b)
 111{
 112	/* If not a leaf node, always sort */
 113	if (b->level && b->keys.nsets)
 114		bch_btree_sort(&b->keys, &b->c->sort);
 115	else
 116		bch_btree_sort_lazy(&b->keys, &b->c->sort);
 117
 118	if (b->written < btree_blocks(b))
 119		bch_bset_init_next(&b->keys, write_block(b),
 120				   bset_magic(&b->c->sb));
 121
 122}
 123
 124/* Btree key manipulation */
 125
 126void bkey_put(struct cache_set *c, struct bkey *k)
 127{
 128	unsigned int i;
 129
 130	for (i = 0; i < KEY_PTRS(k); i++)
 131		if (ptr_available(c, k, i))
 132			atomic_dec_bug(&PTR_BUCKET(c, k, i)->pin);
 133}
 134
 135/* Btree IO */
 136
 137static uint64_t btree_csum_set(struct btree *b, struct bset *i)
 138{
 139	uint64_t crc = b->key.ptr[0];
 140	void *data = (void *) i + 8, *end = bset_bkey_last(i);
 141
 142	crc = bch_crc64_update(crc, data, end - data);
 143	return crc ^ 0xffffffffffffffffULL;
 144}
 145
 146void bch_btree_node_read_done(struct btree *b)
 147{
 148	const char *err = "bad btree header";
 149	struct bset *i = btree_bset_first(b);
 150	struct btree_iter *iter;
 151
 152	/*
 153	 * c->fill_iter can allocate an iterator with more memory space
 154	 * than static MAX_BSETS.
 155	 * See the comment arount cache_set->fill_iter.
 156	 */
 157	iter = mempool_alloc(&b->c->fill_iter, GFP_NOIO);
 158	iter->size = b->c->sb.bucket_size / b->c->sb.block_size;
 159	iter->used = 0;
 160
 161#ifdef CONFIG_BCACHE_DEBUG
 162	iter->b = &b->keys;
 163#endif
 164
 165	if (!i->seq)
 166		goto err;
 167
 168	for (;
 169	     b->written < btree_blocks(b) && i->seq == b->keys.set[0].data->seq;
 170	     i = write_block(b)) {
 171		err = "unsupported bset version";
 172		if (i->version > BCACHE_BSET_VERSION)
 173			goto err;
 174
 175		err = "bad btree header";
 176		if (b->written + set_blocks(i, block_bytes(b->c)) >
 177		    btree_blocks(b))
 178			goto err;
 179
 180		err = "bad magic";
 181		if (i->magic != bset_magic(&b->c->sb))
 182			goto err;
 183
 184		err = "bad checksum";
 185		switch (i->version) {
 186		case 0:
 187			if (i->csum != csum_set(i))
 188				goto err;
 189			break;
 190		case BCACHE_BSET_VERSION:
 191			if (i->csum != btree_csum_set(b, i))
 192				goto err;
 193			break;
 194		}
 195
 196		err = "empty set";
 197		if (i != b->keys.set[0].data && !i->keys)
 198			goto err;
 199
 200		bch_btree_iter_push(iter, i->start, bset_bkey_last(i));
 201
 202		b->written += set_blocks(i, block_bytes(b->c));
 203	}
 204
 205	err = "corrupted btree";
 206	for (i = write_block(b);
 207	     bset_sector_offset(&b->keys, i) < KEY_SIZE(&b->key);
 208	     i = ((void *) i) + block_bytes(b->c))
 209		if (i->seq == b->keys.set[0].data->seq)
 210			goto err;
 211
 212	bch_btree_sort_and_fix_extents(&b->keys, iter, &b->c->sort);
 213
 214	i = b->keys.set[0].data;
 215	err = "short btree key";
 216	if (b->keys.set[0].size &&
 217	    bkey_cmp(&b->key, &b->keys.set[0].end) < 0)
 218		goto err;
 219
 220	if (b->written < btree_blocks(b))
 221		bch_bset_init_next(&b->keys, write_block(b),
 222				   bset_magic(&b->c->sb));
 223out:
 224	mempool_free(iter, &b->c->fill_iter);
 225	return;
 226err:
 227	set_btree_node_io_error(b);
 228	bch_cache_set_error(b->c, "%s at bucket %zu, block %u, %u keys",
 229			    err, PTR_BUCKET_NR(b->c, &b->key, 0),
 230			    bset_block_offset(b, i), i->keys);
 231	goto out;
 232}
 233
 234static void btree_node_read_endio(struct bio *bio)
 235{
 236	struct closure *cl = bio->bi_private;
 237
 238	closure_put(cl);
 239}
 240
 241static void bch_btree_node_read(struct btree *b)
 242{
 243	uint64_t start_time = local_clock();
 244	struct closure cl;
 245	struct bio *bio;
 246
 247	trace_bcache_btree_read(b);
 248
 249	closure_init_stack(&cl);
 250
 251	bio = bch_bbio_alloc(b->c);
 252	bio->bi_iter.bi_size = KEY_SIZE(&b->key) << 9;
 253	bio->bi_end_io	= btree_node_read_endio;
 254	bio->bi_private	= &cl;
 255	bio->bi_opf = REQ_OP_READ | REQ_META;
 256
 257	bch_bio_map(bio, b->keys.set[0].data);
 258
 259	bch_submit_bbio(bio, b->c, &b->key, 0);
 260	closure_sync(&cl);
 261
 262	if (bio->bi_status)
 263		set_btree_node_io_error(b);
 264
 265	bch_bbio_free(bio, b->c);
 266
 267	if (btree_node_io_error(b))
 268		goto err;
 269
 270	bch_btree_node_read_done(b);
 271	bch_time_stats_update(&b->c->btree_read_time, start_time);
 272
 273	return;
 274err:
 275	bch_cache_set_error(b->c, "io error reading bucket %zu",
 276			    PTR_BUCKET_NR(b->c, &b->key, 0));
 277}
 278
 279static void btree_complete_write(struct btree *b, struct btree_write *w)
 280{
 281	if (w->prio_blocked &&
 282	    !atomic_sub_return(w->prio_blocked, &b->c->prio_blocked))
 283		wake_up_allocators(b->c);
 284
 285	if (w->journal) {
 286		atomic_dec_bug(w->journal);
 287		__closure_wake_up(&b->c->journal.wait);
 288	}
 289
 290	w->prio_blocked	= 0;
 291	w->journal	= NULL;
 292}
 293
 294static void btree_node_write_unlock(struct closure *cl)
 295{
 296	struct btree *b = container_of(cl, struct btree, io);
 297
 298	up(&b->io_mutex);
 299}
 300
 301static void __btree_node_write_done(struct closure *cl)
 302{
 303	struct btree *b = container_of(cl, struct btree, io);
 304	struct btree_write *w = btree_prev_write(b);
 305
 306	bch_bbio_free(b->bio, b->c);
 307	b->bio = NULL;
 308	btree_complete_write(b, w);
 309
 310	if (btree_node_dirty(b))
 311		schedule_delayed_work(&b->work, 30 * HZ);
 312
 313	closure_return_with_destructor(cl, btree_node_write_unlock);
 314}
 315
 316static void btree_node_write_done(struct closure *cl)
 317{
 318	struct btree *b = container_of(cl, struct btree, io);
 319
 320	bio_free_pages(b->bio);
 321	__btree_node_write_done(cl);
 322}
 323
 324static void btree_node_write_endio(struct bio *bio)
 325{
 326	struct closure *cl = bio->bi_private;
 327	struct btree *b = container_of(cl, struct btree, io);
 328
 329	if (bio->bi_status)
 330		set_btree_node_io_error(b);
 331
 332	bch_bbio_count_io_errors(b->c, bio, bio->bi_status, "writing btree");
 333	closure_put(cl);
 334}
 335
 336static void do_btree_node_write(struct btree *b)
 337{
 338	struct closure *cl = &b->io;
 339	struct bset *i = btree_bset_last(b);
 340	BKEY_PADDED(key) k;
 341
 342	i->version	= BCACHE_BSET_VERSION;
 343	i->csum		= btree_csum_set(b, i);
 344
 345	BUG_ON(b->bio);
 346	b->bio = bch_bbio_alloc(b->c);
 347
 348	b->bio->bi_end_io	= btree_node_write_endio;
 349	b->bio->bi_private	= cl;
 350	b->bio->bi_iter.bi_size	= roundup(set_bytes(i), block_bytes(b->c));
 351	b->bio->bi_opf		= REQ_OP_WRITE | REQ_META | REQ_FUA;
 352	bch_bio_map(b->bio, i);
 353
 354	/*
 355	 * If we're appending to a leaf node, we don't technically need FUA -
 356	 * this write just needs to be persisted before the next journal write,
 357	 * which will be marked FLUSH|FUA.
 358	 *
 359	 * Similarly if we're writing a new btree root - the pointer is going to
 360	 * be in the next journal entry.
 361	 *
 362	 * But if we're writing a new btree node (that isn't a root) or
 363	 * appending to a non leaf btree node, we need either FUA or a flush
 364	 * when we write the parent with the new pointer. FUA is cheaper than a
 365	 * flush, and writes appending to leaf nodes aren't blocking anything so
 366	 * just make all btree node writes FUA to keep things sane.
 367	 */
 368
 369	bkey_copy(&k.key, &b->key);
 370	SET_PTR_OFFSET(&k.key, 0, PTR_OFFSET(&k.key, 0) +
 371		       bset_sector_offset(&b->keys, i));
 372
 373	if (!bch_bio_alloc_pages(b->bio, __GFP_NOWARN|GFP_NOWAIT)) {
 374		struct bio_vec *bv;
 375		void *addr = (void *) ((unsigned long) i & ~(PAGE_SIZE - 1));
 376		struct bvec_iter_all iter_all;
 377
 378		bio_for_each_segment_all(bv, b->bio, iter_all) {
 379			memcpy(page_address(bv->bv_page), addr, PAGE_SIZE);
 380			addr += PAGE_SIZE;
 381		}
 382
 383		bch_submit_bbio(b->bio, b->c, &k.key, 0);
 384
 385		continue_at(cl, btree_node_write_done, NULL);
 386	} else {
 387		/*
 388		 * No problem for multipage bvec since the bio is
 389		 * just allocated
 390		 */
 391		b->bio->bi_vcnt = 0;
 392		bch_bio_map(b->bio, i);
 393
 394		bch_submit_bbio(b->bio, b->c, &k.key, 0);
 395
 396		closure_sync(cl);
 397		continue_at_nobarrier(cl, __btree_node_write_done, NULL);
 398	}
 399}
 400
 401void __bch_btree_node_write(struct btree *b, struct closure *parent)
 402{
 403	struct bset *i = btree_bset_last(b);
 404
 405	lockdep_assert_held(&b->write_lock);
 406
 407	trace_bcache_btree_write(b);
 408
 409	BUG_ON(current->bio_list);
 410	BUG_ON(b->written >= btree_blocks(b));
 411	BUG_ON(b->written && !i->keys);
 412	BUG_ON(btree_bset_first(b)->seq != i->seq);
 413	bch_check_keys(&b->keys, "writing");
 414
 415	cancel_delayed_work(&b->work);
 416
 417	/* If caller isn't waiting for write, parent refcount is cache set */
 418	down(&b->io_mutex);
 419	closure_init(&b->io, parent ?: &b->c->cl);
 420
 421	clear_bit(BTREE_NODE_dirty,	 &b->flags);
 422	change_bit(BTREE_NODE_write_idx, &b->flags);
 423
 424	do_btree_node_write(b);
 425
 426	atomic_long_add(set_blocks(i, block_bytes(b->c)) * b->c->sb.block_size,
 427			&PTR_CACHE(b->c, &b->key, 0)->btree_sectors_written);
 428
 429	b->written += set_blocks(i, block_bytes(b->c));
 430}
 431
 432void bch_btree_node_write(struct btree *b, struct closure *parent)
 433{
 434	unsigned int nsets = b->keys.nsets;
 435
 436	lockdep_assert_held(&b->lock);
 437
 438	__bch_btree_node_write(b, parent);
 439
 440	/*
 441	 * do verify if there was more than one set initially (i.e. we did a
 442	 * sort) and we sorted down to a single set:
 443	 */
 444	if (nsets && !b->keys.nsets)
 445		bch_btree_verify(b);
 446
 447	bch_btree_init_next(b);
 448}
 449
 450static void bch_btree_node_write_sync(struct btree *b)
 451{
 452	struct closure cl;
 453
 454	closure_init_stack(&cl);
 455
 456	mutex_lock(&b->write_lock);
 457	bch_btree_node_write(b, &cl);
 458	mutex_unlock(&b->write_lock);
 459
 460	closure_sync(&cl);
 461}
 462
 463static void btree_node_write_work(struct work_struct *w)
 464{
 465	struct btree *b = container_of(to_delayed_work(w), struct btree, work);
 466
 467	mutex_lock(&b->write_lock);
 468	if (btree_node_dirty(b))
 469		__bch_btree_node_write(b, NULL);
 470	mutex_unlock(&b->write_lock);
 471}
 472
 473static void bch_btree_leaf_dirty(struct btree *b, atomic_t *journal_ref)
 474{
 475	struct bset *i = btree_bset_last(b);
 476	struct btree_write *w = btree_current_write(b);
 477
 478	lockdep_assert_held(&b->write_lock);
 479
 480	BUG_ON(!b->written);
 481	BUG_ON(!i->keys);
 482
 483	if (!btree_node_dirty(b))
 484		schedule_delayed_work(&b->work, 30 * HZ);
 485
 486	set_btree_node_dirty(b);
 487
 488	/*
 489	 * w->journal is always the oldest journal pin of all bkeys
 490	 * in the leaf node, to make sure the oldest jset seq won't
 491	 * be increased before this btree node is flushed.
 492	 */
 493	if (journal_ref) {
 494		if (w->journal &&
 495		    journal_pin_cmp(b->c, w->journal, journal_ref)) {
 496			atomic_dec_bug(w->journal);
 497			w->journal = NULL;
 498		}
 499
 500		if (!w->journal) {
 501			w->journal = journal_ref;
 502			atomic_inc(w->journal);
 503		}
 504	}
 505
 506	/* Force write if set is too big */
 507	if (set_bytes(i) > PAGE_SIZE - 48 &&
 508	    !current->bio_list)
 509		bch_btree_node_write(b, NULL);
 510}
 511
 512/*
 513 * Btree in memory cache - allocation/freeing
 514 * mca -> memory cache
 515 */
 516
 517#define mca_reserve(c)	(((c->root && c->root->level)		\
 518			  ? c->root->level : 1) * 8 + 16)
 519#define mca_can_free(c)						\
 520	max_t(int, 0, c->btree_cache_used - mca_reserve(c))
 521
 522static void mca_data_free(struct btree *b)
 523{
 524	BUG_ON(b->io_mutex.count != 1);
 525
 526	bch_btree_keys_free(&b->keys);
 527
 528	b->c->btree_cache_used--;
 529	list_move(&b->list, &b->c->btree_cache_freed);
 530}
 531
 532static void mca_bucket_free(struct btree *b)
 533{
 534	BUG_ON(btree_node_dirty(b));
 535
 536	b->key.ptr[0] = 0;
 537	hlist_del_init_rcu(&b->hash);
 538	list_move(&b->list, &b->c->btree_cache_freeable);
 539}
 540
 541static unsigned int btree_order(struct bkey *k)
 542{
 543	return ilog2(KEY_SIZE(k) / PAGE_SECTORS ?: 1);
 544}
 545
 546static void mca_data_alloc(struct btree *b, struct bkey *k, gfp_t gfp)
 547{
 548	if (!bch_btree_keys_alloc(&b->keys,
 549				  max_t(unsigned int,
 550					ilog2(b->c->btree_pages),
 551					btree_order(k)),
 552				  gfp)) {
 553		b->c->btree_cache_used++;
 554		list_move(&b->list, &b->c->btree_cache);
 555	} else {
 556		list_move(&b->list, &b->c->btree_cache_freed);
 557	}
 558}
 559
 560static struct btree *mca_bucket_alloc(struct cache_set *c,
 561				      struct bkey *k, gfp_t gfp)
 562{
 563	/*
 564	 * kzalloc() is necessary here for initialization,
 565	 * see code comments in bch_btree_keys_init().
 566	 */
 567	struct btree *b = kzalloc(sizeof(struct btree), gfp);
 568
 569	if (!b)
 570		return NULL;
 571
 572	init_rwsem(&b->lock);
 573	lockdep_set_novalidate_class(&b->lock);
 574	mutex_init(&b->write_lock);
 575	lockdep_set_novalidate_class(&b->write_lock);
 576	INIT_LIST_HEAD(&b->list);
 577	INIT_DELAYED_WORK(&b->work, btree_node_write_work);
 578	b->c = c;
 579	sema_init(&b->io_mutex, 1);
 580
 581	mca_data_alloc(b, k, gfp);
 582	return b;
 583}
 584
 585static int mca_reap(struct btree *b, unsigned int min_order, bool flush)
 586{
 587	struct closure cl;
 588
 589	closure_init_stack(&cl);
 590	lockdep_assert_held(&b->c->bucket_lock);
 591
 592	if (!down_write_trylock(&b->lock))
 593		return -ENOMEM;
 594
 595	BUG_ON(btree_node_dirty(b) && !b->keys.set[0].data);
 596
 597	if (b->keys.page_order < min_order)
 598		goto out_unlock;
 599
 600	if (!flush) {
 601		if (btree_node_dirty(b))
 602			goto out_unlock;
 603
 604		if (down_trylock(&b->io_mutex))
 605			goto out_unlock;
 606		up(&b->io_mutex);
 607	}
 608
 609retry:
 610	/*
 611	 * BTREE_NODE_dirty might be cleared in btree_flush_btree() by
 612	 * __bch_btree_node_write(). To avoid an extra flush, acquire
 613	 * b->write_lock before checking BTREE_NODE_dirty bit.
 614	 */
 615	mutex_lock(&b->write_lock);
 616	/*
 617	 * If this btree node is selected in btree_flush_write() by journal
 618	 * code, delay and retry until the node is flushed by journal code
 619	 * and BTREE_NODE_journal_flush bit cleared by btree_flush_write().
 620	 */
 621	if (btree_node_journal_flush(b)) {
 622		pr_debug("bnode %p is flushing by journal, retry\n", b);
 623		mutex_unlock(&b->write_lock);
 624		udelay(1);
 625		goto retry;
 626	}
 627
 628	if (btree_node_dirty(b))
 629		__bch_btree_node_write(b, &cl);
 630	mutex_unlock(&b->write_lock);
 631
 632	closure_sync(&cl);
 633
 634	/* wait for any in flight btree write */
 635	down(&b->io_mutex);
 636	up(&b->io_mutex);
 637
 638	return 0;
 639out_unlock:
 640	rw_unlock(true, b);
 641	return -ENOMEM;
 642}
 643
 644static unsigned long bch_mca_scan(struct shrinker *shrink,
 645				  struct shrink_control *sc)
 646{
 647	struct cache_set *c = container_of(shrink, struct cache_set, shrink);
 648	struct btree *b, *t;
 649	unsigned long i, nr = sc->nr_to_scan;
 650	unsigned long freed = 0;
 651	unsigned int btree_cache_used;
 652
 653	if (c->shrinker_disabled)
 654		return SHRINK_STOP;
 655
 656	if (c->btree_cache_alloc_lock)
 657		return SHRINK_STOP;
 658
 659	/* Return -1 if we can't do anything right now */
 660	if (sc->gfp_mask & __GFP_IO)
 661		mutex_lock(&c->bucket_lock);
 662	else if (!mutex_trylock(&c->bucket_lock))
 663		return -1;
 664
 665	/*
 666	 * It's _really_ critical that we don't free too many btree nodes - we
 667	 * have to always leave ourselves a reserve. The reserve is how we
 668	 * guarantee that allocating memory for a new btree node can always
 669	 * succeed, so that inserting keys into the btree can always succeed and
 670	 * IO can always make forward progress:
 671	 */
 672	nr /= c->btree_pages;
 673	if (nr == 0)
 674		nr = 1;
 675	nr = min_t(unsigned long, nr, mca_can_free(c));
 676
 677	i = 0;
 678	btree_cache_used = c->btree_cache_used;
 679	list_for_each_entry_safe_reverse(b, t, &c->btree_cache_freeable, list) {
 680		if (nr <= 0)
 681			goto out;
 682
 683		if (!mca_reap(b, 0, false)) {
 684			mca_data_free(b);
 685			rw_unlock(true, b);
 686			freed++;
 687		}
 688		nr--;
 689		i++;
 690	}
 691
 692	list_for_each_entry_safe_reverse(b, t, &c->btree_cache, list) {
 693		if (nr <= 0 || i >= btree_cache_used)
 694			goto out;
 695
 696		if (!mca_reap(b, 0, false)) {
 697			mca_bucket_free(b);
 698			mca_data_free(b);
 699			rw_unlock(true, b);
 700			freed++;
 701		}
 702
 703		nr--;
 704		i++;
 705	}
 706out:
 707	mutex_unlock(&c->bucket_lock);
 708	return freed * c->btree_pages;
 709}
 710
 711static unsigned long bch_mca_count(struct shrinker *shrink,
 712				   struct shrink_control *sc)
 713{
 714	struct cache_set *c = container_of(shrink, struct cache_set, shrink);
 715
 716	if (c->shrinker_disabled)
 717		return 0;
 718
 719	if (c->btree_cache_alloc_lock)
 720		return 0;
 721
 722	return mca_can_free(c) * c->btree_pages;
 723}
 724
 725void bch_btree_cache_free(struct cache_set *c)
 726{
 727	struct btree *b;
 728	struct closure cl;
 729
 730	closure_init_stack(&cl);
 731
 732	if (c->shrink.list.next)
 733		unregister_shrinker(&c->shrink);
 734
 735	mutex_lock(&c->bucket_lock);
 736
 737#ifdef CONFIG_BCACHE_DEBUG
 738	if (c->verify_data)
 739		list_move(&c->verify_data->list, &c->btree_cache);
 740
 741	free_pages((unsigned long) c->verify_ondisk, ilog2(meta_bucket_pages(&c->sb)));
 742#endif
 743
 744	list_splice(&c->btree_cache_freeable,
 745		    &c->btree_cache);
 746
 747	while (!list_empty(&c->btree_cache)) {
 748		b = list_first_entry(&c->btree_cache, struct btree, list);
 749
 750		/*
 751		 * This function is called by cache_set_free(), no I/O
 752		 * request on cache now, it is unnecessary to acquire
 753		 * b->write_lock before clearing BTREE_NODE_dirty anymore.
 754		 */
 755		if (btree_node_dirty(b)) {
 756			btree_complete_write(b, btree_current_write(b));
 757			clear_bit(BTREE_NODE_dirty, &b->flags);
 758		}
 759		mca_data_free(b);
 760	}
 761
 762	while (!list_empty(&c->btree_cache_freed)) {
 763		b = list_first_entry(&c->btree_cache_freed,
 764				     struct btree, list);
 765		list_del(&b->list);
 766		cancel_delayed_work_sync(&b->work);
 767		kfree(b);
 768	}
 769
 770	mutex_unlock(&c->bucket_lock);
 771}
 772
 773int bch_btree_cache_alloc(struct cache_set *c)
 774{
 775	unsigned int i;
 776
 777	for (i = 0; i < mca_reserve(c); i++)
 778		if (!mca_bucket_alloc(c, &ZERO_KEY, GFP_KERNEL))
 779			return -ENOMEM;
 780
 781	list_splice_init(&c->btree_cache,
 782			 &c->btree_cache_freeable);
 783
 784#ifdef CONFIG_BCACHE_DEBUG
 785	mutex_init(&c->verify_lock);
 786
 787	c->verify_ondisk = (void *)
 788		__get_free_pages(GFP_KERNEL|__GFP_COMP, ilog2(meta_bucket_pages(&c->sb)));
 789	if (!c->verify_ondisk) {
 790		/*
 791		 * Don't worry about the mca_rereserve buckets
 792		 * allocated in previous for-loop, they will be
 793		 * handled properly in bch_cache_set_unregister().
 794		 */
 795		return -ENOMEM;
 796	}
 797
 798	c->verify_data = mca_bucket_alloc(c, &ZERO_KEY, GFP_KERNEL);
 799
 800	if (c->verify_data &&
 801	    c->verify_data->keys.set->data)
 802		list_del_init(&c->verify_data->list);
 803	else
 804		c->verify_data = NULL;
 805#endif
 806
 807	c->shrink.count_objects = bch_mca_count;
 808	c->shrink.scan_objects = bch_mca_scan;
 809	c->shrink.seeks = 4;
 810	c->shrink.batch = c->btree_pages * 2;
 811
 812	if (register_shrinker(&c->shrink))
 813		pr_warn("bcache: %s: could not register shrinker\n",
 814				__func__);
 815
 816	return 0;
 817}
 818
 819/* Btree in memory cache - hash table */
 820
 821static struct hlist_head *mca_hash(struct cache_set *c, struct bkey *k)
 822{
 823	return &c->bucket_hash[hash_32(PTR_HASH(c, k), BUCKET_HASH_BITS)];
 824}
 825
 826static struct btree *mca_find(struct cache_set *c, struct bkey *k)
 827{
 828	struct btree *b;
 829
 830	rcu_read_lock();
 831	hlist_for_each_entry_rcu(b, mca_hash(c, k), hash)
 832		if (PTR_HASH(c, &b->key) == PTR_HASH(c, k))
 833			goto out;
 834	b = NULL;
 835out:
 836	rcu_read_unlock();
 837	return b;
 838}
 839
 840static int mca_cannibalize_lock(struct cache_set *c, struct btree_op *op)
 841{
 842	spin_lock(&c->btree_cannibalize_lock);
 843	if (likely(c->btree_cache_alloc_lock == NULL)) {
 844		c->btree_cache_alloc_lock = current;
 845	} else if (c->btree_cache_alloc_lock != current) {
 846		if (op)
 847			prepare_to_wait(&c->btree_cache_wait, &op->wait,
 848					TASK_UNINTERRUPTIBLE);
 849		spin_unlock(&c->btree_cannibalize_lock);
 850		return -EINTR;
 851	}
 852	spin_unlock(&c->btree_cannibalize_lock);
 853
 854	return 0;
 855}
 856
 857static struct btree *mca_cannibalize(struct cache_set *c, struct btree_op *op,
 858				     struct bkey *k)
 859{
 860	struct btree *b;
 861
 862	trace_bcache_btree_cache_cannibalize(c);
 863
 864	if (mca_cannibalize_lock(c, op))
 865		return ERR_PTR(-EINTR);
 866
 867	list_for_each_entry_reverse(b, &c->btree_cache, list)
 868		if (!mca_reap(b, btree_order(k), false))
 869			return b;
 870
 871	list_for_each_entry_reverse(b, &c->btree_cache, list)
 872		if (!mca_reap(b, btree_order(k), true))
 873			return b;
 874
 875	WARN(1, "btree cache cannibalize failed\n");
 876	return ERR_PTR(-ENOMEM);
 877}
 878
 879/*
 880 * We can only have one thread cannibalizing other cached btree nodes at a time,
 881 * or we'll deadlock. We use an open coded mutex to ensure that, which a
 882 * cannibalize_bucket() will take. This means every time we unlock the root of
 883 * the btree, we need to release this lock if we have it held.
 884 */
 885static void bch_cannibalize_unlock(struct cache_set *c)
 886{
 887	spin_lock(&c->btree_cannibalize_lock);
 888	if (c->btree_cache_alloc_lock == current) {
 889		c->btree_cache_alloc_lock = NULL;
 890		wake_up(&c->btree_cache_wait);
 891	}
 892	spin_unlock(&c->btree_cannibalize_lock);
 893}
 894
 895static struct btree *mca_alloc(struct cache_set *c, struct btree_op *op,
 896			       struct bkey *k, int level)
 897{
 898	struct btree *b;
 899
 900	BUG_ON(current->bio_list);
 901
 902	lockdep_assert_held(&c->bucket_lock);
 903
 904	if (mca_find(c, k))
 905		return NULL;
 906
 907	/* btree_free() doesn't free memory; it sticks the node on the end of
 908	 * the list. Check if there's any freed nodes there:
 909	 */
 910	list_for_each_entry(b, &c->btree_cache_freeable, list)
 911		if (!mca_reap(b, btree_order(k), false))
 912			goto out;
 913
 914	/* We never free struct btree itself, just the memory that holds the on
 915	 * disk node. Check the freed list before allocating a new one:
 916	 */
 917	list_for_each_entry(b, &c->btree_cache_freed, list)
 918		if (!mca_reap(b, 0, false)) {
 919			mca_data_alloc(b, k, __GFP_NOWARN|GFP_NOIO);
 920			if (!b->keys.set[0].data)
 921				goto err;
 922			else
 923				goto out;
 924		}
 925
 926	b = mca_bucket_alloc(c, k, __GFP_NOWARN|GFP_NOIO);
 927	if (!b)
 928		goto err;
 929
 930	BUG_ON(!down_write_trylock(&b->lock));
 931	if (!b->keys.set->data)
 932		goto err;
 933out:
 934	BUG_ON(b->io_mutex.count != 1);
 935
 936	bkey_copy(&b->key, k);
 937	list_move(&b->list, &c->btree_cache);
 938	hlist_del_init_rcu(&b->hash);
 939	hlist_add_head_rcu(&b->hash, mca_hash(c, k));
 940
 941	lock_set_subclass(&b->lock.dep_map, level + 1, _THIS_IP_);
 942	b->parent	= (void *) ~0UL;
 943	b->flags	= 0;
 944	b->written	= 0;
 945	b->level	= level;
 946
 947	if (!b->level)
 948		bch_btree_keys_init(&b->keys, &bch_extent_keys_ops,
 949				    &b->c->expensive_debug_checks);
 950	else
 951		bch_btree_keys_init(&b->keys, &bch_btree_keys_ops,
 952				    &b->c->expensive_debug_checks);
 953
 954	return b;
 955err:
 956	if (b)
 957		rw_unlock(true, b);
 958
 959	b = mca_cannibalize(c, op, k);
 960	if (!IS_ERR(b))
 961		goto out;
 962
 963	return b;
 964}
 965
 966/*
 967 * bch_btree_node_get - find a btree node in the cache and lock it, reading it
 968 * in from disk if necessary.
 969 *
 970 * If IO is necessary and running under submit_bio_noacct, returns -EAGAIN.
 971 *
 972 * The btree node will have either a read or a write lock held, depending on
 973 * level and op->lock.
 974 */
 975struct btree *bch_btree_node_get(struct cache_set *c, struct btree_op *op,
 976				 struct bkey *k, int level, bool write,
 977				 struct btree *parent)
 978{
 979	int i = 0;
 980	struct btree *b;
 981
 982	BUG_ON(level < 0);
 983retry:
 984	b = mca_find(c, k);
 985
 986	if (!b) {
 987		if (current->bio_list)
 988			return ERR_PTR(-EAGAIN);
 989
 990		mutex_lock(&c->bucket_lock);
 991		b = mca_alloc(c, op, k, level);
 992		mutex_unlock(&c->bucket_lock);
 993
 994		if (!b)
 995			goto retry;
 996		if (IS_ERR(b))
 997			return b;
 998
 999		bch_btree_node_read(b);
1000
1001		if (!write)
1002			downgrade_write(&b->lock);
1003	} else {
1004		rw_lock(write, b, level);
1005		if (PTR_HASH(c, &b->key) != PTR_HASH(c, k)) {
1006			rw_unlock(write, b);
1007			goto retry;
1008		}
1009		BUG_ON(b->level != level);
1010	}
1011
1012	if (btree_node_io_error(b)) {
1013		rw_unlock(write, b);
1014		return ERR_PTR(-EIO);
1015	}
1016
1017	BUG_ON(!b->written);
1018
1019	b->parent = parent;
1020
1021	for (; i <= b->keys.nsets && b->keys.set[i].size; i++) {
1022		prefetch(b->keys.set[i].tree);
1023		prefetch(b->keys.set[i].data);
1024	}
1025
1026	for (; i <= b->keys.nsets; i++)
1027		prefetch(b->keys.set[i].data);
1028
1029	return b;
1030}
1031
1032static void btree_node_prefetch(struct btree *parent, struct bkey *k)
1033{
1034	struct btree *b;
1035
1036	mutex_lock(&parent->c->bucket_lock);
1037	b = mca_alloc(parent->c, NULL, k, parent->level - 1);
1038	mutex_unlock(&parent->c->bucket_lock);
1039
1040	if (!IS_ERR_OR_NULL(b)) {
1041		b->parent = parent;
1042		bch_btree_node_read(b);
1043		rw_unlock(true, b);
1044	}
1045}
1046
1047/* Btree alloc */
1048
1049static void btree_node_free(struct btree *b)
1050{
1051	trace_bcache_btree_node_free(b);
1052
1053	BUG_ON(b == b->c->root);
1054
1055retry:
1056	mutex_lock(&b->write_lock);
1057	/*
1058	 * If the btree node is selected and flushing in btree_flush_write(),
1059	 * delay and retry until the BTREE_NODE_journal_flush bit cleared,
1060	 * then it is safe to free the btree node here. Otherwise this btree
1061	 * node will be in race condition.
1062	 */
1063	if (btree_node_journal_flush(b)) {
1064		mutex_unlock(&b->write_lock);
1065		pr_debug("bnode %p journal_flush set, retry\n", b);
1066		udelay(1);
1067		goto retry;
1068	}
1069
1070	if (btree_node_dirty(b)) {
1071		btree_complete_write(b, btree_current_write(b));
1072		clear_bit(BTREE_NODE_dirty, &b->flags);
1073	}
1074
1075	mutex_unlock(&b->write_lock);
1076
1077	cancel_delayed_work(&b->work);
1078
1079	mutex_lock(&b->c->bucket_lock);
1080	bch_bucket_free(b->c, &b->key);
1081	mca_bucket_free(b);
1082	mutex_unlock(&b->c->bucket_lock);
1083}
1084
1085struct btree *__bch_btree_node_alloc(struct cache_set *c, struct btree_op *op,
1086				     int level, bool wait,
1087				     struct btree *parent)
1088{
1089	BKEY_PADDED(key) k;
1090	struct btree *b = ERR_PTR(-EAGAIN);
1091
1092	mutex_lock(&c->bucket_lock);
1093retry:
1094	if (__bch_bucket_alloc_set(c, RESERVE_BTREE, &k.key, 1, wait))
1095		goto err;
1096
1097	bkey_put(c, &k.key);
1098	SET_KEY_SIZE(&k.key, c->btree_pages * PAGE_SECTORS);
1099
1100	b = mca_alloc(c, op, &k.key, level);
1101	if (IS_ERR(b))
1102		goto err_free;
1103
1104	if (!b) {
1105		cache_bug(c,
1106			"Tried to allocate bucket that was in btree cache");
1107		goto retry;
1108	}
1109
1110	b->parent = parent;
1111	bch_bset_init_next(&b->keys, b->keys.set->data, bset_magic(&b->c->sb));
1112
1113	mutex_unlock(&c->bucket_lock);
1114
1115	trace_bcache_btree_node_alloc(b);
1116	return b;
1117err_free:
1118	bch_bucket_free(c, &k.key);
1119err:
1120	mutex_unlock(&c->bucket_lock);
1121
1122	trace_bcache_btree_node_alloc_fail(c);
1123	return b;
1124}
1125
1126static struct btree *bch_btree_node_alloc(struct cache_set *c,
1127					  struct btree_op *op, int level,
1128					  struct btree *parent)
1129{
1130	return __bch_btree_node_alloc(c, op, level, op != NULL, parent);
1131}
1132
1133static struct btree *btree_node_alloc_replacement(struct btree *b,
1134						  struct btree_op *op)
1135{
1136	struct btree *n = bch_btree_node_alloc(b->c, op, b->level, b->parent);
1137
1138	if (!IS_ERR_OR_NULL(n)) {
1139		mutex_lock(&n->write_lock);
1140		bch_btree_sort_into(&b->keys, &n->keys, &b->c->sort);
1141		bkey_copy_key(&n->key, &b->key);
1142		mutex_unlock(&n->write_lock);
1143	}
1144
1145	return n;
1146}
1147
1148static void make_btree_freeing_key(struct btree *b, struct bkey *k)
1149{
1150	unsigned int i;
1151
1152	mutex_lock(&b->c->bucket_lock);
1153
1154	atomic_inc(&b->c->prio_blocked);
1155
1156	bkey_copy(k, &b->key);
1157	bkey_copy_key(k, &ZERO_KEY);
1158
1159	for (i = 0; i < KEY_PTRS(k); i++)
1160		SET_PTR_GEN(k, i,
1161			    bch_inc_gen(PTR_CACHE(b->c, &b->key, i),
1162					PTR_BUCKET(b->c, &b->key, i)));
1163
1164	mutex_unlock(&b->c->bucket_lock);
1165}
1166
1167static int btree_check_reserve(struct btree *b, struct btree_op *op)
1168{
1169	struct cache_set *c = b->c;
1170	struct cache *ca;
1171	unsigned int i, reserve = (c->root->level - b->level) * 2 + 1;
1172
1173	mutex_lock(&c->bucket_lock);
1174
1175	for_each_cache(ca, c, i)
1176		if (fifo_used(&ca->free[RESERVE_BTREE]) < reserve) {
1177			if (op)
1178				prepare_to_wait(&c->btree_cache_wait, &op->wait,
1179						TASK_UNINTERRUPTIBLE);
1180			mutex_unlock(&c->bucket_lock);
1181			return -EINTR;
1182		}
1183
1184	mutex_unlock(&c->bucket_lock);
1185
1186	return mca_cannibalize_lock(b->c, op);
1187}
1188
1189/* Garbage collection */
1190
1191static uint8_t __bch_btree_mark_key(struct cache_set *c, int level,
1192				    struct bkey *k)
1193{
1194	uint8_t stale = 0;
1195	unsigned int i;
1196	struct bucket *g;
1197
1198	/*
1199	 * ptr_invalid() can't return true for the keys that mark btree nodes as
1200	 * freed, but since ptr_bad() returns true we'll never actually use them
1201	 * for anything and thus we don't want mark their pointers here
1202	 */
1203	if (!bkey_cmp(k, &ZERO_KEY))
1204		return stale;
1205
1206	for (i = 0; i < KEY_PTRS(k); i++) {
1207		if (!ptr_available(c, k, i))
1208			continue;
1209
1210		g = PTR_BUCKET(c, k, i);
1211
1212		if (gen_after(g->last_gc, PTR_GEN(k, i)))
1213			g->last_gc = PTR_GEN(k, i);
1214
1215		if (ptr_stale(c, k, i)) {
1216			stale = max(stale, ptr_stale(c, k, i));
1217			continue;
1218		}
1219
1220		cache_bug_on(GC_MARK(g) &&
1221			     (GC_MARK(g) == GC_MARK_METADATA) != (level != 0),
1222			     c, "inconsistent ptrs: mark = %llu, level = %i",
1223			     GC_MARK(g), level);
1224
1225		if (level)
1226			SET_GC_MARK(g, GC_MARK_METADATA);
1227		else if (KEY_DIRTY(k))
1228			SET_GC_MARK(g, GC_MARK_DIRTY);
1229		else if (!GC_MARK(g))
1230			SET_GC_MARK(g, GC_MARK_RECLAIMABLE);
1231
1232		/* guard against overflow */
1233		SET_GC_SECTORS_USED(g, min_t(unsigned int,
1234					     GC_SECTORS_USED(g) + KEY_SIZE(k),
1235					     MAX_GC_SECTORS_USED));
1236
1237		BUG_ON(!GC_SECTORS_USED(g));
1238	}
1239
1240	return stale;
1241}
1242
1243#define btree_mark_key(b, k)	__bch_btree_mark_key(b->c, b->level, k)
1244
1245void bch_initial_mark_key(struct cache_set *c, int level, struct bkey *k)
1246{
1247	unsigned int i;
1248
1249	for (i = 0; i < KEY_PTRS(k); i++)
1250		if (ptr_available(c, k, i) &&
1251		    !ptr_stale(c, k, i)) {
1252			struct bucket *b = PTR_BUCKET(c, k, i);
1253
1254			b->gen = PTR_GEN(k, i);
1255
1256			if (level && bkey_cmp(k, &ZERO_KEY))
1257				b->prio = BTREE_PRIO;
1258			else if (!level && b->prio == BTREE_PRIO)
1259				b->prio = INITIAL_PRIO;
1260		}
1261
1262	__bch_btree_mark_key(c, level, k);
1263}
1264
1265void bch_update_bucket_in_use(struct cache_set *c, struct gc_stat *stats)
1266{
1267	stats->in_use = (c->nbuckets - c->avail_nbuckets) * 100 / c->nbuckets;
1268}
1269
1270static bool btree_gc_mark_node(struct btree *b, struct gc_stat *gc)
1271{
1272	uint8_t stale = 0;
1273	unsigned int keys = 0, good_keys = 0;
1274	struct bkey *k;
1275	struct btree_iter iter;
1276	struct bset_tree *t;
1277
1278	gc->nodes++;
1279
1280	for_each_key_filter(&b->keys, k, &iter, bch_ptr_invalid) {
1281		stale = max(stale, btree_mark_key(b, k));
1282		keys++;
1283
1284		if (bch_ptr_bad(&b->keys, k))
1285			continue;
1286
1287		gc->key_bytes += bkey_u64s(k);
1288		gc->nkeys++;
1289		good_keys++;
1290
1291		gc->data += KEY_SIZE(k);
1292	}
1293
1294	for (t = b->keys.set; t <= &b->keys.set[b->keys.nsets]; t++)
1295		btree_bug_on(t->size &&
1296			     bset_written(&b->keys, t) &&
1297			     bkey_cmp(&b->key, &t->end) < 0,
1298			     b, "found short btree key in gc");
1299
1300	if (b->c->gc_always_rewrite)
1301		return true;
1302
1303	if (stale > 10)
1304		return true;
1305
1306	if ((keys - good_keys) * 2 > keys)
1307		return true;
1308
1309	return false;
1310}
1311
1312#define GC_MERGE_NODES	4U
1313
1314struct gc_merge_info {
1315	struct btree	*b;
1316	unsigned int	keys;
1317};
1318
1319static int bch_btree_insert_node(struct btree *b, struct btree_op *op,
1320				 struct keylist *insert_keys,
1321				 atomic_t *journal_ref,
1322				 struct bkey *replace_key);
1323
1324static int btree_gc_coalesce(struct btree *b, struct btree_op *op,
1325			     struct gc_stat *gc, struct gc_merge_info *r)
1326{
1327	unsigned int i, nodes = 0, keys = 0, blocks;
1328	struct btree *new_nodes[GC_MERGE_NODES];
1329	struct keylist keylist;
1330	struct closure cl;
1331	struct bkey *k;
1332
1333	bch_keylist_init(&keylist);
1334
1335	if (btree_check_reserve(b, NULL))
1336		return 0;
1337
1338	memset(new_nodes, 0, sizeof(new_nodes));
1339	closure_init_stack(&cl);
1340
1341	while (nodes < GC_MERGE_NODES && !IS_ERR_OR_NULL(r[nodes].b))
1342		keys += r[nodes++].keys;
1343
1344	blocks = btree_default_blocks(b->c) * 2 / 3;
1345
1346	if (nodes < 2 ||
1347	    __set_blocks(b->keys.set[0].data, keys,
1348			 block_bytes(b->c)) > blocks * (nodes - 1))
1349		return 0;
1350
1351	for (i = 0; i < nodes; i++) {
1352		new_nodes[i] = btree_node_alloc_replacement(r[i].b, NULL);
1353		if (IS_ERR_OR_NULL(new_nodes[i]))
1354			goto out_nocoalesce;
1355	}
1356
1357	/*
1358	 * We have to check the reserve here, after we've allocated our new
1359	 * nodes, to make sure the insert below will succeed - we also check
1360	 * before as an optimization to potentially avoid a bunch of expensive
1361	 * allocs/sorts
1362	 */
1363	if (btree_check_reserve(b, NULL))
1364		goto out_nocoalesce;
1365
1366	for (i = 0; i < nodes; i++)
1367		mutex_lock(&new_nodes[i]->write_lock);
1368
1369	for (i = nodes - 1; i > 0; --i) {
1370		struct bset *n1 = btree_bset_first(new_nodes[i]);
1371		struct bset *n2 = btree_bset_first(new_nodes[i - 1]);
1372		struct bkey *k, *last = NULL;
1373
1374		keys = 0;
1375
1376		if (i > 1) {
1377			for (k = n2->start;
1378			     k < bset_bkey_last(n2);
1379			     k = bkey_next(k)) {
1380				if (__set_blocks(n1, n1->keys + keys +
1381						 bkey_u64s(k),
1382						 block_bytes(b->c)) > blocks)
1383					break;
1384
1385				last = k;
1386				keys += bkey_u64s(k);
1387			}
1388		} else {
1389			/*
1390			 * Last node we're not getting rid of - we're getting
1391			 * rid of the node at r[0]. Have to try and fit all of
1392			 * the remaining keys into this node; we can't ensure
1393			 * they will always fit due to rounding and variable
1394			 * length keys (shouldn't be possible in practice,
1395			 * though)
1396			 */
1397			if (__set_blocks(n1, n1->keys + n2->keys,
1398					 block_bytes(b->c)) >
1399			    btree_blocks(new_nodes[i]))
1400				goto out_unlock_nocoalesce;
1401
1402			keys = n2->keys;
1403			/* Take the key of the node we're getting rid of */
1404			last = &r->b->key;
1405		}
1406
1407		BUG_ON(__set_blocks(n1, n1->keys + keys, block_bytes(b->c)) >
1408		       btree_blocks(new_nodes[i]));
1409
1410		if (last)
1411			bkey_copy_key(&new_nodes[i]->key, last);
1412
1413		memcpy(bset_bkey_last(n1),
1414		       n2->start,
1415		       (void *) bset_bkey_idx(n2, keys) - (void *) n2->start);
1416
1417		n1->keys += keys;
1418		r[i].keys = n1->keys;
1419
1420		memmove(n2->start,
1421			bset_bkey_idx(n2, keys),
1422			(void *) bset_bkey_last(n2) -
1423			(void *) bset_bkey_idx(n2, keys));
1424
1425		n2->keys -= keys;
1426
1427		if (__bch_keylist_realloc(&keylist,
1428					  bkey_u64s(&new_nodes[i]->key)))
1429			goto out_unlock_nocoalesce;
1430
1431		bch_btree_node_write(new_nodes[i], &cl);
1432		bch_keylist_add(&keylist, &new_nodes[i]->key);
1433	}
1434
1435	for (i = 0; i < nodes; i++)
1436		mutex_unlock(&new_nodes[i]->write_lock);
1437
1438	closure_sync(&cl);
1439
1440	/* We emptied out this node */
1441	BUG_ON(btree_bset_first(new_nodes[0])->keys);
1442	btree_node_free(new_nodes[0]);
1443	rw_unlock(true, new_nodes[0]);
1444	new_nodes[0] = NULL;
1445
1446	for (i = 0; i < nodes; i++) {
1447		if (__bch_keylist_realloc(&keylist, bkey_u64s(&r[i].b->key)))
1448			goto out_nocoalesce;
1449
1450		make_btree_freeing_key(r[i].b, keylist.top);
1451		bch_keylist_push(&keylist);
1452	}
1453
1454	bch_btree_insert_node(b, op, &keylist, NULL, NULL);
1455	BUG_ON(!bch_keylist_empty(&keylist));
1456
1457	for (i = 0; i < nodes; i++) {
1458		btree_node_free(r[i].b);
1459		rw_unlock(true, r[i].b);
1460
1461		r[i].b = new_nodes[i];
1462	}
1463
1464	memmove(r, r + 1, sizeof(r[0]) * (nodes - 1));
1465	r[nodes - 1].b = ERR_PTR(-EINTR);
1466
1467	trace_bcache_btree_gc_coalesce(nodes);
1468	gc->nodes--;
1469
1470	bch_keylist_free(&keylist);
1471
1472	/* Invalidated our iterator */
1473	return -EINTR;
1474
1475out_unlock_nocoalesce:
1476	for (i = 0; i < nodes; i++)
1477		mutex_unlock(&new_nodes[i]->write_lock);
1478
1479out_nocoalesce:
1480	closure_sync(&cl);
1481
1482	while ((k = bch_keylist_pop(&keylist)))
1483		if (!bkey_cmp(k, &ZERO_KEY))
1484			atomic_dec(&b->c->prio_blocked);
1485	bch_keylist_free(&keylist);
1486
1487	for (i = 0; i < nodes; i++)
1488		if (!IS_ERR_OR_NULL(new_nodes[i])) {
1489			btree_node_free(new_nodes[i]);
1490			rw_unlock(true, new_nodes[i]);
1491		}
1492	return 0;
1493}
1494
1495static int btree_gc_rewrite_node(struct btree *b, struct btree_op *op,
1496				 struct btree *replace)
1497{
1498	struct keylist keys;
1499	struct btree *n;
1500
1501	if (btree_check_reserve(b, NULL))
1502		return 0;
1503
1504	n = btree_node_alloc_replacement(replace, NULL);
1505
1506	/* recheck reserve after allocating replacement node */
1507	if (btree_check_reserve(b, NULL)) {
1508		btree_node_free(n);
1509		rw_unlock(true, n);
1510		return 0;
1511	}
1512
1513	bch_btree_node_write_sync(n);
1514
1515	bch_keylist_init(&keys);
1516	bch_keylist_add(&keys, &n->key);
1517
1518	make_btree_freeing_key(replace, keys.top);
1519	bch_keylist_push(&keys);
1520
1521	bch_btree_insert_node(b, op, &keys, NULL, NULL);
1522	BUG_ON(!bch_keylist_empty(&keys));
1523
1524	btree_node_free(replace);
1525	rw_unlock(true, n);
1526
1527	/* Invalidated our iterator */
1528	return -EINTR;
1529}
1530
1531static unsigned int btree_gc_count_keys(struct btree *b)
1532{
1533	struct bkey *k;
1534	struct btree_iter iter;
1535	unsigned int ret = 0;
1536
1537	for_each_key_filter(&b->keys, k, &iter, bch_ptr_bad)
1538		ret += bkey_u64s(k);
1539
1540	return ret;
1541}
1542
1543static size_t btree_gc_min_nodes(struct cache_set *c)
1544{
1545	size_t min_nodes;
1546
1547	/*
1548	 * Since incremental GC would stop 100ms when front
1549	 * side I/O comes, so when there are many btree nodes,
1550	 * if GC only processes constant (100) nodes each time,
1551	 * GC would last a long time, and the front side I/Os
1552	 * would run out of the buckets (since no new bucket
1553	 * can be allocated during GC), and be blocked again.
1554	 * So GC should not process constant nodes, but varied
1555	 * nodes according to the number of btree nodes, which
1556	 * realized by dividing GC into constant(100) times,
1557	 * so when there are many btree nodes, GC can process
1558	 * more nodes each time, otherwise, GC will process less
1559	 * nodes each time (but no less than MIN_GC_NODES)
1560	 */
1561	min_nodes = c->gc_stats.nodes / MAX_GC_TIMES;
1562	if (min_nodes < MIN_GC_NODES)
1563		min_nodes = MIN_GC_NODES;
1564
1565	return min_nodes;
1566}
1567
1568
1569static int btree_gc_recurse(struct btree *b, struct btree_op *op,
1570			    struct closure *writes, struct gc_stat *gc)
1571{
1572	int ret = 0;
1573	bool should_rewrite;
1574	struct bkey *k;
1575	struct btree_iter iter;
1576	struct gc_merge_info r[GC_MERGE_NODES];
1577	struct gc_merge_info *i, *last = r + ARRAY_SIZE(r) - 1;
1578
1579	bch_btree_iter_init(&b->keys, &iter, &b->c->gc_done);
1580
1581	for (i = r; i < r + ARRAY_SIZE(r); i++)
1582		i->b = ERR_PTR(-EINTR);
1583
1584	while (1) {
1585		k = bch_btree_iter_next_filter(&iter, &b->keys, bch_ptr_bad);
1586		if (k) {
1587			r->b = bch_btree_node_get(b->c, op, k, b->level - 1,
1588						  true, b);
1589			if (IS_ERR(r->b)) {
1590				ret = PTR_ERR(r->b);
1591				break;
1592			}
1593
1594			r->keys = btree_gc_count_keys(r->b);
1595
1596			ret = btree_gc_coalesce(b, op, gc, r);
1597			if (ret)
1598				break;
1599		}
1600
1601		if (!last->b)
1602			break;
1603
1604		if (!IS_ERR(last->b)) {
1605			should_rewrite = btree_gc_mark_node(last->b, gc);
1606			if (should_rewrite) {
1607				ret = btree_gc_rewrite_node(b, op, last->b);
1608				if (ret)
1609					break;
1610			}
1611
1612			if (last->b->level) {
1613				ret = btree_gc_recurse(last->b, op, writes, gc);
1614				if (ret)
1615					break;
1616			}
1617
1618			bkey_copy_key(&b->c->gc_done, &last->b->key);
1619
1620			/*
1621			 * Must flush leaf nodes before gc ends, since replace
1622			 * operations aren't journalled
1623			 */
1624			mutex_lock(&last->b->write_lock);
1625			if (btree_node_dirty(last->b))
1626				bch_btree_node_write(last->b, writes);
1627			mutex_unlock(&last->b->write_lock);
1628			rw_unlock(true, last->b);
1629		}
1630
1631		memmove(r + 1, r, sizeof(r[0]) * (GC_MERGE_NODES - 1));
1632		r->b = NULL;
1633
1634		if (atomic_read(&b->c->search_inflight) &&
1635		    gc->nodes >= gc->nodes_pre + btree_gc_min_nodes(b->c)) {
1636			gc->nodes_pre =  gc->nodes;
1637			ret = -EAGAIN;
1638			break;
1639		}
1640
1641		if (need_resched()) {
1642			ret = -EAGAIN;
1643			break;
1644		}
1645	}
1646
1647	for (i = r; i < r + ARRAY_SIZE(r); i++)
1648		if (!IS_ERR_OR_NULL(i->b)) {
1649			mutex_lock(&i->b->write_lock);
1650			if (btree_node_dirty(i->b))
1651				bch_btree_node_write(i->b, writes);
1652			mutex_unlock(&i->b->write_lock);
1653			rw_unlock(true, i->b);
1654		}
1655
1656	return ret;
1657}
1658
1659static int bch_btree_gc_root(struct btree *b, struct btree_op *op,
1660			     struct closure *writes, struct gc_stat *gc)
1661{
1662	struct btree *n = NULL;
1663	int ret = 0;
1664	bool should_rewrite;
1665
1666	should_rewrite = btree_gc_mark_node(b, gc);
1667	if (should_rewrite) {
1668		n = btree_node_alloc_replacement(b, NULL);
1669
1670		if (!IS_ERR_OR_NULL(n)) {
1671			bch_btree_node_write_sync(n);
1672
1673			bch_btree_set_root(n);
1674			btree_node_free(b);
1675			rw_unlock(true, n);
1676
1677			return -EINTR;
1678		}
1679	}
1680
1681	__bch_btree_mark_key(b->c, b->level + 1, &b->key);
1682
1683	if (b->level) {
1684		ret = btree_gc_recurse(b, op, writes, gc);
1685		if (ret)
1686			return ret;
1687	}
1688
1689	bkey_copy_key(&b->c->gc_done, &b->key);
1690
1691	return ret;
1692}
1693
1694static void btree_gc_start(struct cache_set *c)
1695{
1696	struct cache *ca;
1697	struct bucket *b;
1698	unsigned int i;
1699
1700	if (!c->gc_mark_valid)
1701		return;
1702
1703	mutex_lock(&c->bucket_lock);
1704
1705	c->gc_mark_valid = 0;
1706	c->gc_done = ZERO_KEY;
1707
1708	for_each_cache(ca, c, i)
1709		for_each_bucket(b, ca) {
1710			b->last_gc = b->gen;
1711			if (!atomic_read(&b->pin)) {
1712				SET_GC_MARK(b, 0);
1713				SET_GC_SECTORS_USED(b, 0);
1714			}
1715		}
1716
1717	mutex_unlock(&c->bucket_lock);
1718}
1719
1720static void bch_btree_gc_finish(struct cache_set *c)
1721{
1722	struct bucket *b;
1723	struct cache *ca;
1724	unsigned int i;
1725
1726	mutex_lock(&c->bucket_lock);
1727
1728	set_gc_sectors(c);
1729	c->gc_mark_valid = 1;
1730	c->need_gc	= 0;
1731
1732	for (i = 0; i < KEY_PTRS(&c->uuid_bucket); i++)
1733		SET_GC_MARK(PTR_BUCKET(c, &c->uuid_bucket, i),
1734			    GC_MARK_METADATA);
1735
1736	/* don't reclaim buckets to which writeback keys point */
1737	rcu_read_lock();
1738	for (i = 0; i < c->devices_max_used; i++) {
1739		struct bcache_device *d = c->devices[i];
1740		struct cached_dev *dc;
1741		struct keybuf_key *w, *n;
1742		unsigned int j;
1743
1744		if (!d || UUID_FLASH_ONLY(&c->uuids[i]))
1745			continue;
1746		dc = container_of(d, struct cached_dev, disk);
1747
1748		spin_lock(&dc->writeback_keys.lock);
1749		rbtree_postorder_for_each_entry_safe(w, n,
1750					&dc->writeback_keys.keys, node)
1751			for (j = 0; j < KEY_PTRS(&w->key); j++)
1752				SET_GC_MARK(PTR_BUCKET(c, &w->key, j),
1753					    GC_MARK_DIRTY);
1754		spin_unlock(&dc->writeback_keys.lock);
1755	}
1756	rcu_read_unlock();
1757
1758	c->avail_nbuckets = 0;
1759	for_each_cache(ca, c, i) {
1760		uint64_t *i;
1761
1762		ca->invalidate_needs_gc = 0;
1763
1764		for (i = ca->sb.d; i < ca->sb.d + ca->sb.keys; i++)
1765			SET_GC_MARK(ca->buckets + *i, GC_MARK_METADATA);
1766
1767		for (i = ca->prio_buckets;
1768		     i < ca->prio_buckets + prio_buckets(ca) * 2; i++)
1769			SET_GC_MARK(ca->buckets + *i, GC_MARK_METADATA);
1770
1771		for_each_bucket(b, ca) {
1772			c->need_gc	= max(c->need_gc, bucket_gc_gen(b));
1773
1774			if (atomic_read(&b->pin))
1775				continue;
1776
1777			BUG_ON(!GC_MARK(b) && GC_SECTORS_USED(b));
1778
1779			if (!GC_MARK(b) || GC_MARK(b) == GC_MARK_RECLAIMABLE)
1780				c->avail_nbuckets++;
1781		}
1782	}
1783
1784	mutex_unlock(&c->bucket_lock);
1785}
1786
1787static void bch_btree_gc(struct cache_set *c)
1788{
1789	int ret;
1790	struct gc_stat stats;
1791	struct closure writes;
1792	struct btree_op op;
1793	uint64_t start_time = local_clock();
1794
1795	trace_bcache_gc_start(c);
1796
1797	memset(&stats, 0, sizeof(struct gc_stat));
1798	closure_init_stack(&writes);
1799	bch_btree_op_init(&op, SHRT_MAX);
1800
1801	btree_gc_start(c);
1802
1803	/* if CACHE_SET_IO_DISABLE set, gc thread should stop too */
1804	do {
1805		ret = bcache_btree_root(gc_root, c, &op, &writes, &stats);
1806		closure_sync(&writes);
1807		cond_resched();
1808
1809		if (ret == -EAGAIN)
1810			schedule_timeout_interruptible(msecs_to_jiffies
1811						       (GC_SLEEP_MS));
1812		else if (ret)
1813			pr_warn("gc failed!\n");
1814	} while (ret && !test_bit(CACHE_SET_IO_DISABLE, &c->flags));
1815
1816	bch_btree_gc_finish(c);
1817	wake_up_allocators(c);
1818
1819	bch_time_stats_update(&c->btree_gc_time, start_time);
1820
1821	stats.key_bytes *= sizeof(uint64_t);
1822	stats.data	<<= 9;
1823	bch_update_bucket_in_use(c, &stats);
1824	memcpy(&c->gc_stats, &stats, sizeof(struct gc_stat));
1825
1826	trace_bcache_gc_end(c);
1827
1828	bch_moving_gc(c);
1829}
1830
1831static bool gc_should_run(struct cache_set *c)
1832{
1833	struct cache *ca;
1834	unsigned int i;
1835
1836	for_each_cache(ca, c, i)
1837		if (ca->invalidate_needs_gc)
1838			return true;
1839
1840	if (atomic_read(&c->sectors_to_gc) < 0)
1841		return true;
1842
1843	return false;
1844}
1845
1846static int bch_gc_thread(void *arg)
1847{
1848	struct cache_set *c = arg;
1849
1850	while (1) {
1851		wait_event_interruptible(c->gc_wait,
1852			   kthread_should_stop() ||
1853			   test_bit(CACHE_SET_IO_DISABLE, &c->flags) ||
1854			   gc_should_run(c));
1855
1856		if (kthread_should_stop() ||
1857		    test_bit(CACHE_SET_IO_DISABLE, &c->flags))
1858			break;
1859
1860		set_gc_sectors(c);
1861		bch_btree_gc(c);
1862	}
1863
1864	wait_for_kthread_stop();
1865	return 0;
1866}
1867
1868int bch_gc_thread_start(struct cache_set *c)
1869{
1870	c->gc_thread = kthread_run(bch_gc_thread, c, "bcache_gc");
1871	return PTR_ERR_OR_ZERO(c->gc_thread);
1872}
1873
1874/* Initial partial gc */
1875
1876static int bch_btree_check_recurse(struct btree *b, struct btree_op *op)
1877{
1878	int ret = 0;
1879	struct bkey *k, *p = NULL;
1880	struct btree_iter iter;
1881
1882	for_each_key_filter(&b->keys, k, &iter, bch_ptr_invalid)
1883		bch_initial_mark_key(b->c, b->level, k);
1884
1885	bch_initial_mark_key(b->c, b->level + 1, &b->key);
1886
1887	if (b->level) {
1888		bch_btree_iter_init(&b->keys, &iter, NULL);
1889
1890		do {
1891			k = bch_btree_iter_next_filter(&iter, &b->keys,
1892						       bch_ptr_bad);
1893			if (k) {
1894				btree_node_prefetch(b, k);
1895				/*
1896				 * initiallize c->gc_stats.nodes
1897				 * for incremental GC
1898				 */
1899				b->c->gc_stats.nodes++;
1900			}
1901
1902			if (p)
1903				ret = bcache_btree(check_recurse, p, b, op);
1904
1905			p = k;
1906		} while (p && !ret);
1907	}
1908
1909	return ret;
1910}
1911
1912
1913static int bch_btree_check_thread(void *arg)
1914{
1915	int ret;
1916	struct btree_check_info *info = arg;
1917	struct btree_check_state *check_state = info->state;
1918	struct cache_set *c = check_state->c;
1919	struct btree_iter iter;
1920	struct bkey *k, *p;
1921	int cur_idx, prev_idx, skip_nr;
1922
1923	k = p = NULL;
1924	cur_idx = prev_idx = 0;
1925	ret = 0;
1926
1927	/* root node keys are checked before thread created */
1928	bch_btree_iter_init(&c->root->keys, &iter, NULL);
1929	k = bch_btree_iter_next_filter(&iter, &c->root->keys, bch_ptr_bad);
1930	BUG_ON(!k);
1931
1932	p = k;
1933	while (k) {
1934		/*
1935		 * Fetch a root node key index, skip the keys which
1936		 * should be fetched by other threads, then check the
1937		 * sub-tree indexed by the fetched key.
1938		 */
1939		spin_lock(&check_state->idx_lock);
1940		cur_idx = check_state->key_idx;
1941		check_state->key_idx++;
1942		spin_unlock(&check_state->idx_lock);
1943
1944		skip_nr = cur_idx - prev_idx;
1945
1946		while (skip_nr) {
1947			k = bch_btree_iter_next_filter(&iter,
1948						       &c->root->keys,
1949						       bch_ptr_bad);
1950			if (k)
1951				p = k;
1952			else {
1953				/*
1954				 * No more keys to check in root node,
1955				 * current checking threads are enough,
1956				 * stop creating more.
1957				 */
1958				atomic_set(&check_state->enough, 1);
1959				/* Update check_state->enough earlier */
1960				smp_mb__after_atomic();
1961				goto out;
1962			}
1963			skip_nr--;
1964			cond_resched();
1965		}
1966
1967		if (p) {
1968			struct btree_op op;
1969
1970			btree_node_prefetch(c->root, p);
1971			c->gc_stats.nodes++;
1972			bch_btree_op_init(&op, 0);
1973			ret = bcache_btree(check_recurse, p, c->root, &op);
1974			if (ret)
1975				goto out;
1976		}
1977		p = NULL;
1978		prev_idx = cur_idx;
1979		cond_resched();
1980	}
1981
1982out:
1983	info->result = ret;
1984	/* update check_state->started among all CPUs */
1985	smp_mb__before_atomic();
1986	if (atomic_dec_and_test(&check_state->started))
1987		wake_up(&check_state->wait);
1988
1989	return ret;
1990}
1991
1992
1993
1994static int bch_btree_chkthread_nr(void)
1995{
1996	int n = num_online_cpus()/2;
1997
1998	if (n == 0)
1999		n = 1;
2000	else if (n > BCH_BTR_CHKTHREAD_MAX)
2001		n = BCH_BTR_CHKTHREAD_MAX;
2002
2003	return n;
2004}
2005
2006int bch_btree_check(struct cache_set *c)
2007{
2008	int ret = 0;
2009	int i;
2010	struct bkey *k = NULL;
2011	struct btree_iter iter;
2012	struct btree_check_state *check_state;
2013	char name[32];
2014
2015	/* check and mark root node keys */
2016	for_each_key_filter(&c->root->keys, k, &iter, bch_ptr_invalid)
2017		bch_initial_mark_key(c, c->root->level, k);
2018
2019	bch_initial_mark_key(c, c->root->level + 1, &c->root->key);
2020
2021	if (c->root->level == 0)
2022		return 0;
2023
2024	check_state = kzalloc(sizeof(struct btree_check_state), GFP_KERNEL);
2025	if (!check_state)
2026		return -ENOMEM;
2027
2028	check_state->c = c;
2029	check_state->total_threads = bch_btree_chkthread_nr();
2030	check_state->key_idx = 0;
2031	spin_lock_init(&check_state->idx_lock);
2032	atomic_set(&check_state->started, 0);
2033	atomic_set(&check_state->enough, 0);
2034	init_waitqueue_head(&check_state->wait);
2035
2036	/*
2037	 * Run multiple threads to check btree nodes in parallel,
2038	 * if check_state->enough is non-zero, it means current
2039	 * running check threads are enough, unncessary to create
2040	 * more.
2041	 */
2042	for (i = 0; i < check_state->total_threads; i++) {
2043		/* fetch latest check_state->enough earlier */
2044		smp_mb__before_atomic();
2045		if (atomic_read(&check_state->enough))
2046			break;
2047
2048		check_state->infos[i].result = 0;
2049		check_state->infos[i].state = check_state;
2050		snprintf(name, sizeof(name), "bch_btrchk[%u]", i);
2051		atomic_inc(&check_state->started);
2052
2053		check_state->infos[i].thread =
2054			kthread_run(bch_btree_check_thread,
2055				    &check_state->infos[i],
2056				    name);
2057		if (IS_ERR(check_state->infos[i].thread)) {
2058			pr_err("fails to run thread bch_btrchk[%d]\n", i);
2059			for (--i; i >= 0; i--)
2060				kthread_stop(check_state->infos[i].thread);
2061			ret = -ENOMEM;
2062			goto out;
2063		}
2064	}
2065
2066	wait_event_interruptible(check_state->wait,
2067				 atomic_read(&check_state->started) == 0 ||
2068				  test_bit(CACHE_SET_IO_DISABLE, &c->flags));
2069
2070	for (i = 0; i < check_state->total_threads; i++) {
2071		if (check_state->infos[i].result) {
2072			ret = check_state->infos[i].result;
2073			goto out;
2074		}
2075	}
2076
2077out:
2078	kfree(check_state);
2079	return ret;
2080}
2081
2082void bch_initial_gc_finish(struct cache_set *c)
2083{
2084	struct cache *ca;
2085	struct bucket *b;
2086	unsigned int i;
2087
2088	bch_btree_gc_finish(c);
2089
2090	mutex_lock(&c->bucket_lock);
2091
2092	/*
2093	 * We need to put some unused buckets directly on the prio freelist in
2094	 * order to get the allocator thread started - it needs freed buckets in
2095	 * order to rewrite the prios and gens, and it needs to rewrite prios
2096	 * and gens in order to free buckets.
2097	 *
2098	 * This is only safe for buckets that have no live data in them, which
2099	 * there should always be some of.
2100	 */
2101	for_each_cache(ca, c, i) {
2102		for_each_bucket(b, ca) {
2103			if (fifo_full(&ca->free[RESERVE_PRIO]) &&
2104			    fifo_full(&ca->free[RESERVE_BTREE]))
2105				break;
2106
2107			if (bch_can_invalidate_bucket(ca, b) &&
2108			    !GC_MARK(b)) {
2109				__bch_invalidate_one_bucket(ca, b);
2110				if (!fifo_push(&ca->free[RESERVE_PRIO],
2111				   b - ca->buckets))
2112					fifo_push(&ca->free[RESERVE_BTREE],
2113						  b - ca->buckets);
2114			}
2115		}
2116	}
2117
2118	mutex_unlock(&c->bucket_lock);
2119}
2120
2121/* Btree insertion */
2122
2123static bool btree_insert_key(struct btree *b, struct bkey *k,
2124			     struct bkey *replace_key)
2125{
2126	unsigned int status;
2127
2128	BUG_ON(bkey_cmp(k, &b->key) > 0);
2129
2130	status = bch_btree_insert_key(&b->keys, k, replace_key);
2131	if (status != BTREE_INSERT_STATUS_NO_INSERT) {
2132		bch_check_keys(&b->keys, "%u for %s", status,
2133			       replace_key ? "replace" : "insert");
2134
2135		trace_bcache_btree_insert_key(b, k, replace_key != NULL,
2136					      status);
2137		return true;
2138	} else
2139		return false;
2140}
2141
2142static size_t insert_u64s_remaining(struct btree *b)
2143{
2144	long ret = bch_btree_keys_u64s_remaining(&b->keys);
2145
2146	/*
2147	 * Might land in the middle of an existing extent and have to split it
2148	 */
2149	if (b->keys.ops->is_extents)
2150		ret -= KEY_MAX_U64S;
2151
2152	return max(ret, 0L);
2153}
2154
2155static bool bch_btree_insert_keys(struct btree *b, struct btree_op *op,
2156				  struct keylist *insert_keys,
2157				  struct bkey *replace_key)
2158{
2159	bool ret = false;
2160	int oldsize = bch_count_data(&b->keys);
2161
2162	while (!bch_keylist_empty(insert_keys)) {
2163		struct bkey *k = insert_keys->keys;
2164
2165		if (bkey_u64s(k) > insert_u64s_remaining(b))
2166			break;
2167
2168		if (bkey_cmp(k, &b->key) <= 0) {
2169			if (!b->level)
2170				bkey_put(b->c, k);
2171
2172			ret |= btree_insert_key(b, k, replace_key);
2173			bch_keylist_pop_front(insert_keys);
2174		} else if (bkey_cmp(&START_KEY(k), &b->key) < 0) {
2175			BKEY_PADDED(key) temp;
2176			bkey_copy(&temp.key, insert_keys->keys);
2177
2178			bch_cut_back(&b->key, &temp.key);
2179			bch_cut_front(&b->key, insert_keys->keys);
2180
2181			ret |= btree_insert_key(b, &temp.key, replace_key);
2182			break;
2183		} else {
2184			break;
2185		}
2186	}
2187
2188	if (!ret)
2189		op->insert_collision = true;
2190
2191	BUG_ON(!bch_keylist_empty(insert_keys) && b->level);
2192
2193	BUG_ON(bch_count_data(&b->keys) < oldsize);
2194	return ret;
2195}
2196
2197static int btree_split(struct btree *b, struct btree_op *op,
2198		       struct keylist *insert_keys,
2199		       struct bkey *replace_key)
2200{
2201	bool split;
2202	struct btree *n1, *n2 = NULL, *n3 = NULL;
2203	uint64_t start_time = local_clock();
2204	struct closure cl;
2205	struct keylist parent_keys;
2206
2207	closure_init_stack(&cl);
2208	bch_keylist_init(&parent_keys);
2209
2210	if (btree_check_reserve(b, op)) {
2211		if (!b->level)
2212			return -EINTR;
2213		else
2214			WARN(1, "insufficient reserve for split\n");
2215	}
2216
2217	n1 = btree_node_alloc_replacement(b, op);
2218	if (IS_ERR(n1))
2219		goto err;
2220
2221	split = set_blocks(btree_bset_first(n1),
2222			   block_bytes(n1->c)) > (btree_blocks(b) * 4) / 5;
2223
2224	if (split) {
2225		unsigned int keys = 0;
2226
2227		trace_bcache_btree_node_split(b, btree_bset_first(n1)->keys);
2228
2229		n2 = bch_btree_node_alloc(b->c, op, b->level, b->parent);
2230		if (IS_ERR(n2))
2231			goto err_free1;
2232
2233		if (!b->parent) {
2234			n3 = bch_btree_node_alloc(b->c, op, b->level + 1, NULL);
2235			if (IS_ERR(n3))
2236				goto err_free2;
2237		}
2238
2239		mutex_lock(&n1->write_lock);
2240		mutex_lock(&n2->write_lock);
2241
2242		bch_btree_insert_keys(n1, op, insert_keys, replace_key);
2243
2244		/*
2245		 * Has to be a linear search because we don't have an auxiliary
2246		 * search tree yet
2247		 */
2248
2249		while (keys < (btree_bset_first(n1)->keys * 3) / 5)
2250			keys += bkey_u64s(bset_bkey_idx(btree_bset_first(n1),
2251							keys));
2252
2253		bkey_copy_key(&n1->key,
2254			      bset_bkey_idx(btree_bset_first(n1), keys));
2255		keys += bkey_u64s(bset_bkey_idx(btree_bset_first(n1), keys));
2256
2257		btree_bset_first(n2)->keys = btree_bset_first(n1)->keys - keys;
2258		btree_bset_first(n1)->keys = keys;
2259
2260		memcpy(btree_bset_first(n2)->start,
2261		       bset_bkey_last(btree_bset_first(n1)),
2262		       btree_bset_first(n2)->keys * sizeof(uint64_t));
2263
2264		bkey_copy_key(&n2->key, &b->key);
2265
2266		bch_keylist_add(&parent_keys, &n2->key);
2267		bch_btree_node_write(n2, &cl);
2268		mutex_unlock(&n2->write_lock);
2269		rw_unlock(true, n2);
2270	} else {
2271		trace_bcache_btree_node_compact(b, btree_bset_first(n1)->keys);
2272
2273		mutex_lock(&n1->write_lock);
2274		bch_btree_insert_keys(n1, op, insert_keys, replace_key);
2275	}
2276
2277	bch_keylist_add(&parent_keys, &n1->key);
2278	bch_btree_node_write(n1, &cl);
2279	mutex_unlock(&n1->write_lock);
2280
2281	if (n3) {
2282		/* Depth increases, make a new root */
2283		mutex_lock(&n3->write_lock);
2284		bkey_copy_key(&n3->key, &MAX_KEY);
2285		bch_btree_insert_keys(n3, op, &parent_keys, NULL);
2286		bch_btree_node_write(n3, &cl);
2287		mutex_unlock(&n3->write_lock);
2288
2289		closure_sync(&cl);
2290		bch_btree_set_root(n3);
2291		rw_unlock(true, n3);
2292	} else if (!b->parent) {
2293		/* Root filled up but didn't need to be split */
2294		closure_sync(&cl);
2295		bch_btree_set_root(n1);
2296	} else {
2297		/* Split a non root node */
2298		closure_sync(&cl);
2299		make_btree_freeing_key(b, parent_keys.top);
2300		bch_keylist_push(&parent_keys);
2301
2302		bch_btree_insert_node(b->parent, op, &parent_keys, NULL, NULL);
2303		BUG_ON(!bch_keylist_empty(&parent_keys));
2304	}
2305
2306	btree_node_free(b);
2307	rw_unlock(true, n1);
2308
2309	bch_time_stats_update(&b->c->btree_split_time, start_time);
2310
2311	return 0;
2312err_free2:
2313	bkey_put(b->c, &n2->key);
2314	btree_node_free(n2);
2315	rw_unlock(true, n2);
2316err_free1:
2317	bkey_put(b->c, &n1->key);
2318	btree_node_free(n1);
2319	rw_unlock(true, n1);
2320err:
2321	WARN(1, "bcache: btree split failed (level %u)", b->level);
2322
2323	if (n3 == ERR_PTR(-EAGAIN) ||
2324	    n2 == ERR_PTR(-EAGAIN) ||
2325	    n1 == ERR_PTR(-EAGAIN))
2326		return -EAGAIN;
2327
2328	return -ENOMEM;
2329}
2330
2331static int bch_btree_insert_node(struct btree *b, struct btree_op *op,
2332				 struct keylist *insert_keys,
2333				 atomic_t *journal_ref,
2334				 struct bkey *replace_key)
2335{
2336	struct closure cl;
2337
2338	BUG_ON(b->level && replace_key);
2339
2340	closure_init_stack(&cl);
2341
2342	mutex_lock(&b->write_lock);
2343
2344	if (write_block(b) != btree_bset_last(b) &&
2345	    b->keys.last_set_unwritten)
2346		bch_btree_init_next(b); /* just wrote a set */
2347
2348	if (bch_keylist_nkeys(insert_keys) > insert_u64s_remaining(b)) {
2349		mutex_unlock(&b->write_lock);
2350		goto split;
2351	}
2352
2353	BUG_ON(write_block(b) != btree_bset_last(b));
2354
2355	if (bch_btree_insert_keys(b, op, insert_keys, replace_key)) {
2356		if (!b->level)
2357			bch_btree_leaf_dirty(b, journal_ref);
2358		else
2359			bch_btree_node_write(b, &cl);
2360	}
2361
2362	mutex_unlock(&b->write_lock);
2363
2364	/* wait for btree node write if necessary, after unlock */
2365	closure_sync(&cl);
2366
2367	return 0;
2368split:
2369	if (current->bio_list) {
2370		op->lock = b->c->root->level + 1;
2371		return -EAGAIN;
2372	} else if (op->lock <= b->c->root->level) {
2373		op->lock = b->c->root->level + 1;
2374		return -EINTR;
2375	} else {
2376		/* Invalidated all iterators */
2377		int ret = btree_split(b, op, insert_keys, replace_key);
2378
2379		if (bch_keylist_empty(insert_keys))
2380			return 0;
2381		else if (!ret)
2382			return -EINTR;
2383		return ret;
2384	}
2385}
2386
2387int bch_btree_insert_check_key(struct btree *b, struct btree_op *op,
2388			       struct bkey *check_key)
2389{
2390	int ret = -EINTR;
2391	uint64_t btree_ptr = b->key.ptr[0];
2392	unsigned long seq = b->seq;
2393	struct keylist insert;
2394	bool upgrade = op->lock == -1;
2395
2396	bch_keylist_init(&insert);
2397
2398	if (upgrade) {
2399		rw_unlock(false, b);
2400		rw_lock(true, b, b->level);
2401
2402		if (b->key.ptr[0] != btree_ptr ||
2403		    b->seq != seq + 1) {
2404			op->lock = b->level;
2405			goto out;
2406		}
2407	}
2408
2409	SET_KEY_PTRS(check_key, 1);
2410	get_random_bytes(&check_key->ptr[0], sizeof(uint64_t));
2411
2412	SET_PTR_DEV(check_key, 0, PTR_CHECK_DEV);
2413
2414	bch_keylist_add(&insert, check_key);
2415
2416	ret = bch_btree_insert_node(b, op, &insert, NULL, NULL);
2417
2418	BUG_ON(!ret && !bch_keylist_empty(&insert));
2419out:
2420	if (upgrade)
2421		downgrade_write(&b->lock);
2422	return ret;
2423}
2424
2425struct btree_insert_op {
2426	struct btree_op	op;
2427	struct keylist	*keys;
2428	atomic_t	*journal_ref;
2429	struct bkey	*replace_key;
2430};
2431
2432static int btree_insert_fn(struct btree_op *b_op, struct btree *b)
2433{
2434	struct btree_insert_op *op = container_of(b_op,
2435					struct btree_insert_op, op);
2436
2437	int ret = bch_btree_insert_node(b, &op->op, op->keys,
2438					op->journal_ref, op->replace_key);
2439	if (ret && !bch_keylist_empty(op->keys))
2440		return ret;
2441	else
2442		return MAP_DONE;
2443}
2444
2445int bch_btree_insert(struct cache_set *c, struct keylist *keys,
2446		     atomic_t *journal_ref, struct bkey *replace_key)
2447{
2448	struct btree_insert_op op;
2449	int ret = 0;
2450
2451	BUG_ON(current->bio_list);
2452	BUG_ON(bch_keylist_empty(keys));
2453
2454	bch_btree_op_init(&op.op, 0);
2455	op.keys		= keys;
2456	op.journal_ref	= journal_ref;
2457	op.replace_key	= replace_key;
2458
2459	while (!ret && !bch_keylist_empty(keys)) {
2460		op.op.lock = 0;
2461		ret = bch_btree_map_leaf_nodes(&op.op, c,
2462					       &START_KEY(keys->keys),
2463					       btree_insert_fn);
2464	}
2465
2466	if (ret) {
2467		struct bkey *k;
2468
2469		pr_err("error %i\n", ret);
2470
2471		while ((k = bch_keylist_pop(keys)))
2472			bkey_put(c, k);
2473	} else if (op.op.insert_collision)
2474		ret = -ESRCH;
2475
2476	return ret;
2477}
2478
2479void bch_btree_set_root(struct btree *b)
2480{
2481	unsigned int i;
2482	struct closure cl;
2483
2484	closure_init_stack(&cl);
2485
2486	trace_bcache_btree_set_root(b);
2487
2488	BUG_ON(!b->written);
2489
2490	for (i = 0; i < KEY_PTRS(&b->key); i++)
2491		BUG_ON(PTR_BUCKET(b->c, &b->key, i)->prio != BTREE_PRIO);
2492
2493	mutex_lock(&b->c->bucket_lock);
2494	list_del_init(&b->list);
2495	mutex_unlock(&b->c->bucket_lock);
2496
2497	b->c->root = b;
2498
2499	bch_journal_meta(b->c, &cl);
2500	closure_sync(&cl);
2501}
2502
2503/* Map across nodes or keys */
2504
2505static int bch_btree_map_nodes_recurse(struct btree *b, struct btree_op *op,
2506				       struct bkey *from,
2507				       btree_map_nodes_fn *fn, int flags)
2508{
2509	int ret = MAP_CONTINUE;
2510
2511	if (b->level) {
2512		struct bkey *k;
2513		struct btree_iter iter;
2514
2515		bch_btree_iter_init(&b->keys, &iter, from);
2516
2517		while ((k = bch_btree_iter_next_filter(&iter, &b->keys,
2518						       bch_ptr_bad))) {
2519			ret = bcache_btree(map_nodes_recurse, k, b,
2520				    op, from, fn, flags);
2521			from = NULL;
2522
2523			if (ret != MAP_CONTINUE)
2524				return ret;
2525		}
2526	}
2527
2528	if (!b->level || flags == MAP_ALL_NODES)
2529		ret = fn(op, b);
2530
2531	return ret;
2532}
2533
2534int __bch_btree_map_nodes(struct btree_op *op, struct cache_set *c,
2535			  struct bkey *from, btree_map_nodes_fn *fn, int flags)
2536{
2537	return bcache_btree_root(map_nodes_recurse, c, op, from, fn, flags);
2538}
2539
2540int bch_btree_map_keys_recurse(struct btree *b, struct btree_op *op,
2541				      struct bkey *from, btree_map_keys_fn *fn,
2542				      int flags)
2543{
2544	int ret = MAP_CONTINUE;
2545	struct bkey *k;
2546	struct btree_iter iter;
2547
2548	bch_btree_iter_init(&b->keys, &iter, from);
2549
2550	while ((k = bch_btree_iter_next_filter(&iter, &b->keys, bch_ptr_bad))) {
2551		ret = !b->level
2552			? fn(op, b, k)
2553			: bcache_btree(map_keys_recurse, k,
2554				       b, op, from, fn, flags);
2555		from = NULL;
2556
2557		if (ret != MAP_CONTINUE)
2558			return ret;
2559	}
2560
2561	if (!b->level && (flags & MAP_END_KEY))
2562		ret = fn(op, b, &KEY(KEY_INODE(&b->key),
2563				     KEY_OFFSET(&b->key), 0));
2564
2565	return ret;
2566}
2567
2568int bch_btree_map_keys(struct btree_op *op, struct cache_set *c,
2569		       struct bkey *from, btree_map_keys_fn *fn, int flags)
2570{
2571	return bcache_btree_root(map_keys_recurse, c, op, from, fn, flags);
2572}
2573
2574/* Keybuf code */
2575
2576static inline int keybuf_cmp(struct keybuf_key *l, struct keybuf_key *r)
2577{
2578	/* Overlapping keys compare equal */
2579	if (bkey_cmp(&l->key, &START_KEY(&r->key)) <= 0)
2580		return -1;
2581	if (bkey_cmp(&START_KEY(&l->key), &r->key) >= 0)
2582		return 1;
2583	return 0;
2584}
2585
2586static inline int keybuf_nonoverlapping_cmp(struct keybuf_key *l,
2587					    struct keybuf_key *r)
2588{
2589	return clamp_t(int64_t, bkey_cmp(&l->key, &r->key), -1, 1);
2590}
2591
2592struct refill {
2593	struct btree_op	op;
2594	unsigned int	nr_found;
2595	struct keybuf	*buf;
2596	struct bkey	*end;
2597	keybuf_pred_fn	*pred;
2598};
2599
2600static int refill_keybuf_fn(struct btree_op *op, struct btree *b,
2601			    struct bkey *k)
2602{
2603	struct refill *refill = container_of(op, struct refill, op);
2604	struct keybuf *buf = refill->buf;
2605	int ret = MAP_CONTINUE;
2606
2607	if (bkey_cmp(k, refill->end) > 0) {
2608		ret = MAP_DONE;
2609		goto out;
2610	}
2611
2612	if (!KEY_SIZE(k)) /* end key */
2613		goto out;
2614
2615	if (refill->pred(buf, k)) {
2616		struct keybuf_key *w;
2617
2618		spin_lock(&buf->lock);
2619
2620		w = array_alloc(&buf->freelist);
2621		if (!w) {
2622			spin_unlock(&buf->lock);
2623			return MAP_DONE;
2624		}
2625
2626		w->private = NULL;
2627		bkey_copy(&w->key, k);
2628
2629		if (RB_INSERT(&buf->keys, w, node, keybuf_cmp))
2630			array_free(&buf->freelist, w);
2631		else
2632			refill->nr_found++;
2633
2634		if (array_freelist_empty(&buf->freelist))
2635			ret = MAP_DONE;
2636
2637		spin_unlock(&buf->lock);
2638	}
2639out:
2640	buf->last_scanned = *k;
2641	return ret;
2642}
2643
2644void bch_refill_keybuf(struct cache_set *c, struct keybuf *buf,
2645		       struct bkey *end, keybuf_pred_fn *pred)
2646{
2647	struct bkey start = buf->last_scanned;
2648	struct refill refill;
2649
2650	cond_resched();
2651
2652	bch_btree_op_init(&refill.op, -1);
2653	refill.nr_found	= 0;
2654	refill.buf	= buf;
2655	refill.end	= end;
2656	refill.pred	= pred;
2657
2658	bch_btree_map_keys(&refill.op, c, &buf->last_scanned,
2659			   refill_keybuf_fn, MAP_END_KEY);
2660
2661	trace_bcache_keyscan(refill.nr_found,
2662			     KEY_INODE(&start), KEY_OFFSET(&start),
2663			     KEY_INODE(&buf->last_scanned),
2664			     KEY_OFFSET(&buf->last_scanned));
2665
2666	spin_lock(&buf->lock);
2667
2668	if (!RB_EMPTY_ROOT(&buf->keys)) {
2669		struct keybuf_key *w;
2670
2671		w = RB_FIRST(&buf->keys, struct keybuf_key, node);
2672		buf->start	= START_KEY(&w->key);
2673
2674		w = RB_LAST(&buf->keys, struct keybuf_key, node);
2675		buf->end	= w->key;
2676	} else {
2677		buf->start	= MAX_KEY;
2678		buf->end	= MAX_KEY;
2679	}
2680
2681	spin_unlock(&buf->lock);
2682}
2683
2684static void __bch_keybuf_del(struct keybuf *buf, struct keybuf_key *w)
2685{
2686	rb_erase(&w->node, &buf->keys);
2687	array_free(&buf->freelist, w);
2688}
2689
2690void bch_keybuf_del(struct keybuf *buf, struct keybuf_key *w)
2691{
2692	spin_lock(&buf->lock);
2693	__bch_keybuf_del(buf, w);
2694	spin_unlock(&buf->lock);
2695}
2696
2697bool bch_keybuf_check_overlapping(struct keybuf *buf, struct bkey *start,
2698				  struct bkey *end)
2699{
2700	bool ret = false;
2701	struct keybuf_key *p, *w, s;
2702
2703	s.key = *start;
2704
2705	if (bkey_cmp(end, &buf->start) <= 0 ||
2706	    bkey_cmp(start, &buf->end) >= 0)
2707		return false;
2708
2709	spin_lock(&buf->lock);
2710	w = RB_GREATER(&buf->keys, s, node, keybuf_nonoverlapping_cmp);
2711
2712	while (w && bkey_cmp(&START_KEY(&w->key), end) < 0) {
2713		p = w;
2714		w = RB_NEXT(w, node);
2715
2716		if (p->private)
2717			ret = true;
2718		else
2719			__bch_keybuf_del(buf, p);
2720	}
2721
2722	spin_unlock(&buf->lock);
2723	return ret;
2724}
2725
2726struct keybuf_key *bch_keybuf_next(struct keybuf *buf)
2727{
2728	struct keybuf_key *w;
2729
2730	spin_lock(&buf->lock);
2731
2732	w = RB_FIRST(&buf->keys, struct keybuf_key, node);
2733
2734	while (w && w->private)
2735		w = RB_NEXT(w, node);
2736
2737	if (w)
2738		w->private = ERR_PTR(-EINTR);
2739
2740	spin_unlock(&buf->lock);
2741	return w;
2742}
2743
2744struct keybuf_key *bch_keybuf_next_rescan(struct cache_set *c,
2745					  struct keybuf *buf,
2746					  struct bkey *end,
2747					  keybuf_pred_fn *pred)
2748{
2749	struct keybuf_key *ret;
2750
2751	while (1) {
2752		ret = bch_keybuf_next(buf);
2753		if (ret)
2754			break;
2755
2756		if (bkey_cmp(&buf->last_scanned, end) >= 0) {
2757			pr_debug("scan finished\n");
2758			break;
2759		}
2760
2761		bch_refill_keybuf(c, buf, end, pred);
2762	}
2763
2764	return ret;
2765}
2766
2767void bch_keybuf_init(struct keybuf *buf)
2768{
2769	buf->last_scanned	= MAX_KEY;
2770	buf->keys		= RB_ROOT;
2771
2772	spin_lock_init(&buf->lock);
2773	array_allocator_init(&buf->freelist);
2774}