kernel/drivers/md/bcache/btree.c

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