fs/btrfs/extent_io.c

   1 #include <linux/bitops.h>
   2 #include <linux/slab.h>
   3 #include <linux/bio.h>
   4 #include <linux/mm.h>
   5 #include <linux/pagemap.h>
   6 #include <linux/page-flags.h>
   7 #include <linux/module.h>
   8 #include <linux/spinlock.h>
   9 #include <linux/blkdev.h>
  10 #include <linux/swap.h>
  11 #include <linux/writeback.h>
  12 #include <linux/pagevec.h>
  13 #include <linux/prefetch.h>
  14 #include <linux/cleancache.h>
  15 #include "extent_io.h"
  16 #include "extent_map.h"
  17 #include "compat.h"
  18 #include "ctree.h"
  19 #include "btrfs_inode.h"
  20
  21 static struct kmem_cache *extent_state_cache;
  22 static struct kmem_cache *extent_buffer_cache;
  23
  24 static LIST_HEAD(buffers);
  25 static LIST_HEAD(states);
  26
  27 #define LEAK_DEBUG 0
  28 #if LEAK_DEBUG
  29 static DEFINE_SPINLOCK(leak_lock);
  30 #endif
  31
  32 #define BUFFER_LRU_MAX 64
  33
  34 struct tree_entry {
  35         u64 start;
  36         u64 end;
  37         struct rb_node rb_node;
  38 };
  39
  40 struct extent_page_data {
  41         struct bio *bio;
  42         struct extent_io_tree *tree;
  43         get_extent_t *get_extent;
  44
  45         /* tells writepage not to lock the state bits for this range
  46          * it still does the unlocking
  47          */
  48         unsigned int extent_locked:1;
  49
  50         /* tells the submit_bio code to use a WRITE_SYNC */
  51         unsigned int sync_io:1;
  52 };
  53
  54 int __init extent_io_init(void)
  55 {
  56         extent_state_cache = kmem_cache_create("extent_state",
  57                         sizeof(struct extent_state), 0,
  58                         SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD, NULL);
  59         if (!extent_state_cache)
  60                 return -ENOMEM;
  61
  62         extent_buffer_cache = kmem_cache_create("extent_buffers",
  63                         sizeof(struct extent_buffer), 0,
  64                         SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD, NULL);
  65         if (!extent_buffer_cache)
  66                 goto free_state_cache;
  67         return 0;
  68
  69 free_state_cache:
  70         kmem_cache_destroy(extent_state_cache);
  71         return -ENOMEM;
  72 }
  73
  74 void extent_io_exit(void)
  75 {
  76         struct extent_state *state;
  77         struct extent_buffer *eb;
  78
  79         while (!list_empty(&states)) {
  80                 state = list_entry(states.next, struct extent_state, leak_list);
  81                 printk(KERN_ERR "btrfs state leak: start %llu end %llu "
  82                        "state %lu in tree %p refs %d\n",
  83                        (unsigned long long)state->start,
  84                        (unsigned long long)state->end,
  85                        state->state, state->tree, atomic_read(&state->refs));
  86                 list_del(&state->leak_list);
  87                 kmem_cache_free(extent_state_cache, state);
  88
  89         }
  90
  91         while (!list_empty(&buffers)) {
  92                 eb = list_entry(buffers.next, struct extent_buffer, leak_list);
  93                 printk(KERN_ERR "btrfs buffer leak start %llu len %lu "
  94                        "refs %d\n", (unsigned long long)eb->start,
  95                        eb->len, atomic_read(&eb->refs));
  96                 list_del(&eb->leak_list);
  97                 kmem_cache_free(extent_buffer_cache, eb);
  98         }
  99         if (extent_state_cache)
 100                 kmem_cache_destroy(extent_state_cache);
 101         if (extent_buffer_cache)
 102                 kmem_cache_destroy(extent_buffer_cache);
 103 }
 104
 105 void extent_io_tree_init(struct extent_io_tree *tree,
 106                          struct address_space *mapping)
 107 {
 108         tree->state = RB_ROOT;
 109         INIT_RADIX_TREE(&tree->buffer, GFP_ATOMIC);
 110         tree->ops = NULL;
 111         tree->dirty_bytes = 0;
 112         spin_lock_init(&tree->lock);
 113         spin_lock_init(&tree->buffer_lock);
 114         tree->mapping = mapping;
 115 }
 116
 117 static struct extent_state *alloc_extent_state(gfp_t mask)
 118 {
 119         struct extent_state *state;
 120 #if LEAK_DEBUG
 121         unsigned long flags;
 122 #endif
 123
 124         state = kmem_cache_alloc(extent_state_cache, mask);
 125         if (!state)
 126                 return state;
 127         state->state = 0;
 128         state->private = 0;
 129         state->tree = NULL;
 130 #if LEAK_DEBUG
 131         spin_lock_irqsave(&leak_lock, flags);
 132         list_add(&state->leak_list, &states);
 133         spin_unlock_irqrestore(&leak_lock, flags);
 134 #endif
 135         atomic_set(&state->refs, 1);
 136         init_waitqueue_head(&state->wq);
 137         return state;
 138 }
 139
 140 void free_extent_state(struct extent_state *state)
 141 {
 142         if (!state)
 143                 return;
 144         if (atomic_dec_and_test(&state->refs)) {
 145 #if LEAK_DEBUG
 146                 unsigned long flags;
 147 #endif
 148                 WARN_ON(state->tree);
 149 #if LEAK_DEBUG
 150                 spin_lock_irqsave(&leak_lock, flags);
 151                 list_del(&state->leak_list);
 152                 spin_unlock_irqrestore(&leak_lock, flags);
 153 #endif
 154                 kmem_cache_free(extent_state_cache, state);
 155         }
 156 }
 157
 158 static struct rb_node *tree_insert(struct rb_root *root, u64 offset,
 159                                    struct rb_node *node)
 160 {
 161         struct rb_node **p = &root->rb_node;
 162         struct rb_node *parent = NULL;
 163         struct tree_entry *entry;
 164
 165         while (*p) {
 166                 parent = *p;
 167                 entry = rb_entry(parent, struct tree_entry, rb_node);
 168
 169                 if (offset < entry->start)
 170                         p = &(*p)->rb_left;
 171                 else if (offset > entry->end)
 172                         p = &(*p)->rb_right;
 173                 else
 174                         return parent;
 175         }
 176
 177         entry = rb_entry(node, struct tree_entry, rb_node);
 178         rb_link_node(node, parent, p);
 179         rb_insert_color(node, root);
 180         return NULL;
 181 }
 182
 183 static struct rb_node *__etree_search(struct extent_io_tree *tree, u64 offset,
 184                                      struct rb_node **prev_ret,
 185                                      struct rb_node **next_ret)
 186 {
 187         struct rb_root *root = &tree->state;
 188         struct rb_node *n = root->rb_node;
 189         struct rb_node *prev = NULL;
 190         struct rb_node *orig_prev = NULL;
 191         struct tree_entry *entry;
 192         struct tree_entry *prev_entry = NULL;
 193
 194         while (n) {
 195                 entry = rb_entry(n, struct tree_entry, rb_node);
 196                 prev = n;
 197                 prev_entry = entry;
 198
 199                 if (offset < entry->start)
 200                         n = n->rb_left;
 201                 else if (offset > entry->end)
 202                         n = n->rb_right;
 203                 else
 204                         return n;
 205         }
 206
 207         if (prev_ret) {
 208                 orig_prev = prev;
 209                 while (prev && offset > prev_entry->end) {
 210                         prev = rb_next(prev);
 211                         prev_entry = rb_entry(prev, struct tree_entry, rb_node);
 212                 }
 213                 *prev_ret = prev;
 214                 prev = orig_prev;
 215         }
 216
 217         if (next_ret) {
 218                 prev_entry = rb_entry(prev, struct tree_entry, rb_node);
 219                 while (prev && offset < prev_entry->start) {
 220                         prev = rb_prev(prev);
 221                         prev_entry = rb_entry(prev, struct tree_entry, rb_node);
 222                 }
 223                 *next_ret = prev;
 224         }
 225         return NULL;
 226 }
 227
 228 static inline struct rb_node *tree_search(struct extent_io_tree *tree,
 229                                           u64 offset)
 230 {
 231         struct rb_node *prev = NULL;
 232         struct rb_node *ret;
 233
 234         ret = __etree_search(tree, offset, &prev, NULL);
 235         if (!ret)
 236                 return prev;
 237         return ret;
 238 }
 239
 240 static void merge_cb(struct extent_io_tree *tree, struct extent_state *new,
 241                      struct extent_state *other)
 242 {
 243         if (tree->ops && tree->ops->merge_extent_hook)
 244                 tree->ops->merge_extent_hook(tree->mapping->host, new,
 245                                              other);
 246 }
 247
 248 /*
 249  * utility function to look for merge candidates inside a given range.
 250  * Any extents with matching state are merged together into a single
 251  * extent in the tree.  Extents with EXTENT_IO in their state field
 252  * are not merged because the end_io handlers need to be able to do
 253  * operations on them without sleeping (or doing allocations/splits).
 254  *
 255  * This should be called with the tree lock held.
 256  */
 257 static void merge_state(struct extent_io_tree *tree,
 258                         struct extent_state *state)
 259 {
 260         struct extent_state *other;
 261         struct rb_node *other_node;
 262
 263         if (state->state & (EXTENT_IOBITS | EXTENT_BOUNDARY))
 264                 return;
 265
 266         other_node = rb_prev(&state->rb_node);
 267         if (other_node) {
 268                 other = rb_entry(other_node, struct extent_state, rb_node);
 269                 if (other->end == state->start - 1 &&
 270                     other->state == state->state) {
 271                         merge_cb(tree, state, other);
 272                         state->start = other->start;
 273                         other->tree = NULL;
 274                         rb_erase(&other->rb_node, &tree->state);
 275                         free_extent_state(other);
 276                 }
 277         }
 278         other_node = rb_next(&state->rb_node);
 279         if (other_node) {
 280                 other = rb_entry(other_node, struct extent_state, rb_node);
 281                 if (other->start == state->end + 1 &&
 282                     other->state == state->state) {
 283                         merge_cb(tree, state, other);
 284                         state->end = other->end;
 285                         other->tree = NULL;
 286                         rb_erase(&other->rb_node, &tree->state);
 287                         free_extent_state(other);
 288                 }
 289         }
 290 }
 291
 292 static void set_state_cb(struct extent_io_tree *tree,
 293                          struct extent_state *state, int *bits)
 294 {
 295         if (tree->ops && tree->ops->set_bit_hook)
 296                 tree->ops->set_bit_hook(tree->mapping->host, state, bits);
 297 }
 298
 299 static void clear_state_cb(struct extent_io_tree *tree,
 300                            struct extent_state *state, int *bits)
 301 {
 302         if (tree->ops && tree->ops->clear_bit_hook)
 303                 tree->ops->clear_bit_hook(tree->mapping->host, state, bits);
 304 }
 305
 306 static void set_state_bits(struct extent_io_tree *tree,
 307                            struct extent_state *state, int *bits);
 308
 309 /*
 310  * insert an extent_state struct into the tree.  'bits' are set on the
 311  * struct before it is inserted.
 312  *
 313  * This may return -EEXIST if the extent is already there, in which case the
 314  * state struct is freed.
 315  *
 316  * The tree lock is not taken internally.  This is a utility function and
 317  * probably isn't what you want to call (see set/clear_extent_bit).
 318  */
 319 static int insert_state(struct extent_io_tree *tree,
 320                         struct extent_state *state, u64 start, u64 end,
 321                         int *bits)
 322 {
 323         struct rb_node *node;
 324
 325         if (end < start) {
 326                 printk(KERN_ERR "btrfs end < start %llu %llu\n",
 327                        (unsigned long long)end,
 328                        (unsigned long long)start);
 329                 WARN_ON(1);
 330         }
 331         state->start = start;
 332         state->end = end;
 333
 334         set_state_bits(tree, state, bits);
 335
 336         node = tree_insert(&tree->state, end, &state->rb_node);
 337         if (node) {
 338                 struct extent_state *found;
 339                 found = rb_entry(node, struct extent_state, rb_node);
 340                 printk(KERN_ERR "btrfs found node %llu %llu on insert of "
 341                        "%llu %llu\n", (unsigned long long)found->start,
 342                        (unsigned long long)found->end,
 343                        (unsigned long long)start, (unsigned long long)end);
 344                 return -EEXIST;
 345         }
 346         state->tree = tree;
 347         merge_state(tree, state);
 348         return 0;
 349 }
 350
 351 static void split_cb(struct extent_io_tree *tree, struct extent_state *orig,
 352                      u64 split)
 353 {
 354         if (tree->ops && tree->ops->split_extent_hook)
 355                 tree->ops->split_extent_hook(tree->mapping->host, orig, split);
 356 }
 357
 358 /*
 359  * split a given extent state struct in two, inserting the preallocated
 360  * struct 'prealloc' as the newly created second half.  'split' indicates an
 361  * offset inside 'orig' where it should be split.
 362  *
 363  * Before calling,
 364  * the tree has 'orig' at [orig->start, orig->end].  After calling, there
 365  * are two extent state structs in the tree:
 366  * prealloc: [orig->start, split - 1]
 367  * orig: [ split, orig->end ]
 368  *
 369  * The tree locks are not taken by this function. They need to be held
 370  * by the caller.
 371  */
 372 static int split_state(struct extent_io_tree *tree, struct extent_state *orig,
 373                        struct extent_state *prealloc, u64 split)
 374 {
 375         struct rb_node *node;
 376
 377         split_cb(tree, orig, split);
 378
 379         prealloc->start = orig->start;
 380         prealloc->end = split - 1;
 381         prealloc->state = orig->state;
 382         orig->start = split;
 383
 384         node = tree_insert(&tree->state, prealloc->end, &prealloc->rb_node);
 385         if (node) {
 386                 free_extent_state(prealloc);
 387                 return -EEXIST;
 388         }
 389         prealloc->tree = tree;
 390         return 0;
 391 }
 392
 393 /*
 394  * utility function to clear some bits in an extent state struct.
 395  * it will optionally wake up any one waiting on this state (wake == 1), or
 396  * forcibly remove the state from the tree (delete == 1).
 397  *
 398  * If no bits are set on the state struct after clearing things, the
 399  * struct is freed and removed from the tree
 400  */
 401 static int clear_state_bit(struct extent_io_tree *tree,
 402                             struct extent_state *state,
 403                             int *bits, int wake)
 404 {
 405         int bits_to_clear = *bits & ~EXTENT_CTLBITS;
 406         int ret = state->state & bits_to_clear;
 407
 408         if ((bits_to_clear & EXTENT_DIRTY) && (state->state & EXTENT_DIRTY)) {
 409                 u64 range = state->end - state->start + 1;
 410                 WARN_ON(range > tree->dirty_bytes);
 411                 tree->dirty_bytes -= range;
 412         }
 413         clear_state_cb(tree, state, bits);
 414         state->state &= ~bits_to_clear;
 415         if (wake)
 416                 wake_up(&state->wq);
 417         if (state->state == 0) {
 418                 if (state->tree) {
 419                         rb_erase(&state->rb_node, &tree->state);
 420                         state->tree = NULL;
 421                         free_extent_state(state);
 422                 } else {
 423                         WARN_ON(1);
 424                 }
 425         } else {
 426                 merge_state(tree, state);
 427         }
 428         return ret;
 429 }
 430
 431 static struct extent_state *
 432 alloc_extent_state_atomic(struct extent_state *prealloc)
 433 {
 434         if (!prealloc)
 435                 prealloc = alloc_extent_state(GFP_ATOMIC);
 436
 437         return prealloc;
 438 }
 439
 440 /*
 441  * clear some bits on a range in the tree.  This may require splitting
 442  * or inserting elements in the tree, so the gfp mask is used to
 443  * indicate which allocations or sleeping are allowed.
 444  *
 445  * pass 'wake' == 1 to kick any sleepers, and 'delete' == 1 to remove
 446  * the given range from the tree regardless of state (ie for truncate).
 447  *
 448  * the range [start, end] is inclusive.
 449  *
 450  * This takes the tree lock, and returns < 0 on error, > 0 if any of the
 451  * bits were already set, or zero if none of the bits were already set.
 452  */
 453 int clear_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
 454                      int bits, int wake, int delete,
 455                      struct extent_state **cached_state,
 456                      gfp_t mask)
 457 {
 458         struct extent_state *state;
 459         struct extent_state *cached;
 460         struct extent_state *prealloc = NULL;
 461         struct rb_node *next_node;
 462         struct rb_node *node;
 463         u64 last_end;
 464         int err;
 465         int set = 0;
 466         int clear = 0;
 467
 468         if (delete)
 469                 bits |= ~EXTENT_CTLBITS;
 470         bits |= EXTENT_FIRST_DELALLOC;
 471
 472         if (bits & (EXTENT_IOBITS | EXTENT_BOUNDARY))
 473                 clear = 1;
 474 again:
 475         if (!prealloc && (mask & __GFP_WAIT)) {
 476                 prealloc = alloc_extent_state(mask);
 477                 if (!prealloc)
 478                         return -ENOMEM;
 479         }
 480
 481         spin_lock(&tree->lock);
 482         if (cached_state) {
 483                 cached = *cached_state;
 484
 485                 if (clear) {
 486                         *cached_state = NULL;
 487                         cached_state = NULL;
 488                 }
 489
 490                 if (cached && cached->tree && cached->start <= start &&
 491                     cached->end > start) {
 492                         if (clear)
 493                                 atomic_dec(&cached->refs);
 494                         state = cached;
 495                         goto hit_next;
 496                 }
 497                 if (clear)
 498                         free_extent_state(cached);
 499         }
 500         /*
 501          * this search will find the extents that end after
 502          * our range starts
 503          */
 504         node = tree_search(tree, start);
 505         if (!node)
 506                 goto out;
 507         state = rb_entry(node, struct extent_state, rb_node);
 508 hit_next:
 509         if (state->start > end)
 510                 goto out;
 511         WARN_ON(state->end < start);
 512         last_end = state->end;
 513
 514         /*
 515          *     | ---- desired range ---- |
 516          *  | state | or
 517          *  | ------------- state -------------- |
 518          *
 519          * We need to split the extent we found, and may flip
 520          * bits on second half.
 521          *
 522          * If the extent we found extends past our range, we
 523          * just split and search again.  It'll get split again
 524          * the next time though.
 525          *
 526          * If the extent we found is inside our range, we clear
 527          * the desired bit on it.
 528          */
 529
 530         if (state->start < start) {
 531                 prealloc = alloc_extent_state_atomic(prealloc);
 532                 BUG_ON(!prealloc);
 533                 err = split_state(tree, state, prealloc, start);
 534                 BUG_ON(err == -EEXIST);
 535                 prealloc = NULL;
 536                 if (err)
 537                         goto out;
 538                 if (state->end <= end) {
 539                         set |= clear_state_bit(tree, state, &bits, wake);
 540                         if (last_end == (u64)-1)
 541                                 goto out;
 542                         start = last_end + 1;
 543                 }
 544                 goto search_again;
 545         }
 546         /*
 547          * | ---- desired range ---- |
 548          *                        | state |
 549          * We need to split the extent, and clear the bit
 550          * on the first half
 551          */
 552         if (state->start <= end && state->end > end) {
 553                 prealloc = alloc_extent_state_atomic(prealloc);
 554                 BUG_ON(!prealloc);
 555                 err = split_state(tree, state, prealloc, end + 1);
 556                 BUG_ON(err == -EEXIST);
 557                 if (wake)
 558                         wake_up(&state->wq);
 559
 560                 set |= clear_state_bit(tree, prealloc, &bits, wake);
 561
 562                 prealloc = NULL;
 563                 goto out;
 564         }
 565
 566         if (state->end < end && prealloc && !need_resched())
 567                 next_node = rb_next(&state->rb_node);
 568         else
 569                 next_node = NULL;
 570
 571         set |= clear_state_bit(tree, state, &bits, wake);
 572         if (last_end == (u64)-1)
 573                 goto out;
 574         start = last_end + 1;
 575         if (start <= end && next_node) {
 576                 state = rb_entry(next_node, struct extent_state,
 577                                  rb_node);
 578                 if (state->start == start)
 579                         goto hit_next;
 580         }
 581         goto search_again;
 582
 583 out:
 584         spin_unlock(&tree->lock);
 585         if (prealloc)
 586                 free_extent_state(prealloc);
 587
 588         return set;
 589
 590 search_again:
 591         if (start > end)
 592                 goto out;
 593         spin_unlock(&tree->lock);
 594         if (mask & __GFP_WAIT)
 595                 cond_resched();
 596         goto again;
 597 }
 598
 599 static int wait_on_state(struct extent_io_tree *tree,
 600                          struct extent_state *state)
 601                 __releases(tree->lock)
 602                 __acquires(tree->lock)
 603 {
 604         DEFINE_WAIT(wait);
 605         prepare_to_wait(&state->wq, &wait, TASK_UNINTERRUPTIBLE);
 606         spin_unlock(&tree->lock);
 607         schedule();
 608         spin_lock(&tree->lock);
 609         finish_wait(&state->wq, &wait);
 610         return 0;
 611 }
 612
 613 /*
 614  * waits for one or more bits to clear on a range in the state tree.
 615  * The range [start, end] is inclusive.
 616  * The tree lock is taken by this function
 617  */
 618 int wait_extent_bit(struct extent_io_tree *tree, u64 start, u64 end, int bits)
 619 {
 620         struct extent_state *state;
 621         struct rb_node *node;
 622
 623         spin_lock(&tree->lock);
 624 again:
 625         while (1) {
 626                 /*
 627                  * this search will find all the extents that end after
 628                  * our range starts
 629                  */
 630                 node = tree_search(tree, start);
 631                 if (!node)
 632                         break;
 633
 634                 state = rb_entry(node, struct extent_state, rb_node);
 635
 636                 if (state->start > end)
 637                         goto out;
 638
 639                 if (state->state & bits) {
 640                         start = state->start;
 641                         atomic_inc(&state->refs);
 642                         wait_on_state(tree, state);
 643                         free_extent_state(state);
 644                         goto again;
 645                 }
 646                 start = state->end + 1;
 647
 648                 if (start > end)
 649                         break;
 650
 651                 cond_resched_lock(&tree->lock);
 652         }
 653 out:
 654         spin_unlock(&tree->lock);
 655         return 0;
 656 }
 657
 658 static void set_state_bits(struct extent_io_tree *tree,
 659                            struct extent_state *state,
 660                            int *bits)
 661 {
 662         int bits_to_set = *bits & ~EXTENT_CTLBITS;
 663
 664         set_state_cb(tree, state, bits);
 665         if ((bits_to_set & EXTENT_DIRTY) && !(state->state & EXTENT_DIRTY)) {
 666                 u64 range = state->end - state->start + 1;
 667                 tree->dirty_bytes += range;
 668         }
 669         state->state |= bits_to_set;
 670 }
 671
 672 static void cache_state(struct extent_state *state,
 673                         struct extent_state **cached_ptr)
 674 {
 675         if (cached_ptr && !(*cached_ptr)) {
 676                 if (state->state & (EXTENT_IOBITS | EXTENT_BOUNDARY)) {
 677                         *cached_ptr = state;
 678                         atomic_inc(&state->refs);
 679                 }
 680         }
 681 }
 682
 683 static void uncache_state(struct extent_state **cached_ptr)
 684 {
 685         if (cached_ptr && (*cached_ptr)) {
 686                 struct extent_state *state = *cached_ptr;
 687                 *cached_ptr = NULL;
 688                 free_extent_state(state);
 689         }
 690 }
 691
 692 /*
 693  * set some bits on a range in the tree.  This may require allocations or
 694  * sleeping, so the gfp mask is used to indicate what is allowed.
 695  *
 696  * If any of the exclusive bits are set, this will fail with -EEXIST if some
 697  * part of the range already has the desired bits set.  The start of the
 698  * existing range is returned in failed_start in this case.
 699  *
 700  * [start, end] is inclusive This takes the tree lock.
 701  */
 702
 703 int set_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
 704                    int bits, int exclusive_bits, u64 *failed_start,
 705                    struct extent_state **cached_state, gfp_t mask)
 706 {
 707         struct extent_state *state;
 708         struct extent_state *prealloc = NULL;
 709         struct rb_node *node;
 710         int err = 0;
 711         u64 last_start;
 712         u64 last_end;
 713
 714         bits |= EXTENT_FIRST_DELALLOC;
 715 again:
 716         if (!prealloc && (mask & __GFP_WAIT)) {
 717                 prealloc = alloc_extent_state(mask);
 718                 BUG_ON(!prealloc);
 719         }
 720
 721         spin_lock(&tree->lock);
 722         if (cached_state && *cached_state) {
 723                 state = *cached_state;
 724                 if (state->start <= start && state->end > start &&
 725                     state->tree) {
 726                         node = &state->rb_node;
 727                         goto hit_next;
 728                 }
 729         }
 730         /*
 731          * this search will find all the extents that end after
 732          * our range starts.
 733          */
 734         node = tree_search(tree, start);
 735         if (!node) {
 736                 prealloc = alloc_extent_state_atomic(prealloc);
 737                 BUG_ON(!prealloc);
 738                 err = insert_state(tree, prealloc, start, end, &bits);
 739                 prealloc = NULL;
 740                 BUG_ON(err == -EEXIST);
 741                 goto out;
 742         }
 743         state = rb_entry(node, struct extent_state, rb_node);
 744 hit_next:
 745         last_start = state->start;
 746         last_end = state->end;
 747
 748         /*
 749          * | ---- desired range ---- |
 750          * | state |
 751          *
 752          * Just lock what we found and keep going
 753          */
 754         if (state->start == start && state->end <= end) {
 755                 struct rb_node *next_node;
 756                 if (state->state & exclusive_bits) {
 757                         *failed_start = state->start;
 758                         err = -EEXIST;
 759                         goto out;
 760                 }
 761
 762                 set_state_bits(tree, state, &bits);
 763
 764                 cache_state(state, cached_state);
 765                 merge_state(tree, state);
 766                 if (last_end == (u64)-1)
 767                         goto out;
 768
 769                 start = last_end + 1;
 770                 next_node = rb_next(&state->rb_node);
 771                 if (next_node && start < end && prealloc && !need_resched()) {
 772                         state = rb_entry(next_node, struct extent_state,
 773                                          rb_node);
 774                         if (state->start == start)
 775                                 goto hit_next;
 776                 }
 777                 goto search_again;
 778         }
 779
 780         /*
 781          *     | ---- desired range ---- |
 782          * | state |
 783          *   or
 784          * | ------------- state -------------- |
 785          *
 786          * We need to split the extent we found, and may flip bits on
 787          * second half.
 788          *
 789          * If the extent we found extends past our
 790          * range, we just split and search again.  It'll get split
 791          * again the next time though.
 792          *
 793          * If the extent we found is inside our range, we set the
 794          * desired bit on it.
 795          */
 796         if (state->start < start) {
 797                 if (state->state & exclusive_bits) {
 798                         *failed_start = start;
 799                         err = -EEXIST;
 800                         goto out;
 801                 }
 802
 803                 prealloc = alloc_extent_state_atomic(prealloc);
 804                 BUG_ON(!prealloc);
 805                 err = split_state(tree, state, prealloc, start);
 806                 BUG_ON(err == -EEXIST);
 807                 prealloc = NULL;
 808                 if (err)
 809                         goto out;
 810                 if (state->end <= end) {
 811                         set_state_bits(tree, state, &bits);
 812                         cache_state(state, cached_state);
 813                         merge_state(tree, state);
 814                         if (last_end == (u64)-1)
 815                                 goto out;
 816                         start = last_end + 1;
 817                 }
 818                 goto search_again;
 819         }
 820         /*
 821          * | ---- desired range ---- |
 822          *     | state | or               | state |
 823          *
 824          * There's a hole, we need to insert something in it and
 825          * ignore the extent we found.
 826          */
 827         if (state->start > start) {
 828                 u64 this_end;
 829                 if (end < last_start)
 830                         this_end = end;
 831                 else
 832                         this_end = last_start - 1;
 833
 834                 prealloc = alloc_extent_state_atomic(prealloc);
 835                 BUG_ON(!prealloc);
 836
 837                 /*
 838                  * Avoid to free 'prealloc' if it can be merged with
 839                  * the later extent.
 840                  */
 841                 err = insert_state(tree, prealloc, start, this_end,
 842                                    &bits);
 843                 BUG_ON(err == -EEXIST);
 844                 if (err) {
 845                         free_extent_state(prealloc);
 846                         prealloc = NULL;
 847                         goto out;
 848                 }
 849                 cache_state(prealloc, cached_state);
 850                 prealloc = NULL;
 851                 start = this_end + 1;
 852                 goto search_again;
 853         }
 854         /*
 855          * | ---- desired range ---- |
 856          *                        | state |
 857          * We need to split the extent, and set the bit
 858          * on the first half
 859          */
 860         if (state->start <= end && state->end > end) {
 861                 if (state->state & exclusive_bits) {
 862                         *failed_start = start;
 863                         err = -EEXIST;
 864                         goto out;
 865                 }
 866
 867                 prealloc = alloc_extent_state_atomic(prealloc);
 868                 BUG_ON(!prealloc);
 869                 err = split_state(tree, state, prealloc, end + 1);
 870                 BUG_ON(err == -EEXIST);
 871
 872                 set_state_bits(tree, prealloc, &bits);
 873                 cache_state(prealloc, cached_state);
 874                 merge_state(tree, prealloc);
 875                 prealloc = NULL;
 876                 goto out;
 877         }
 878
 879         goto search_again;
 880
 881 out:
 882         spin_unlock(&tree->lock);
 883         if (prealloc)
 884                 free_extent_state(prealloc);
 885
 886         return err;
 887
 888 search_again:
 889         if (start > end)
 890                 goto out;
 891         spin_unlock(&tree->lock);
 892         if (mask & __GFP_WAIT)
 893                 cond_resched();
 894         goto again;
 895 }
 896
 897 /* wrappers around set/clear extent bit */
 898 int set_extent_dirty(struct extent_io_tree *tree, u64 start, u64 end,
 899                      gfp_t mask)
 900 {
 901         return set_extent_bit(tree, start, end, EXTENT_DIRTY, 0, NULL,
 902                               NULL, mask);
 903 }
 904
 905 int set_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
 906                     int bits, gfp_t mask)
 907 {
 908         return set_extent_bit(tree, start, end, bits, 0, NULL,
 909                               NULL, mask);
 910 }
 911
 912 int clear_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
 913                       int bits, gfp_t mask)
 914 {
 915         return clear_extent_bit(tree, start, end, bits, 0, 0, NULL, mask);
 916 }
 917
 918 int set_extent_delalloc(struct extent_io_tree *tree, u64 start, u64 end,
 919                         struct extent_state **cached_state, gfp_t mask)
 920 {
 921         return set_extent_bit(tree, start, end,
 922                               EXTENT_DELALLOC | EXTENT_DIRTY | EXTENT_UPTODATE,
 923                               0, NULL, cached_state, mask);
 924 }
 925
 926 int clear_extent_dirty(struct extent_io_tree *tree, u64 start, u64 end,
 927                        gfp_t mask)
 928 {
 929         return clear_extent_bit(tree, start, end,
 930                                 EXTENT_DIRTY | EXTENT_DELALLOC |
 931                                 EXTENT_DO_ACCOUNTING, 0, 0, NULL, mask);
 932 }
 933
 934 int set_extent_new(struct extent_io_tree *tree, u64 start, u64 end,
 935                      gfp_t mask)
 936 {
 937         return set_extent_bit(tree, start, end, EXTENT_NEW, 0, NULL,
 938                               NULL, mask);
 939 }
 940
 941 int set_extent_uptodate(struct extent_io_tree *tree, u64 start, u64 end,
 942                         struct extent_state **cached_state, gfp_t mask)
 943 {
 944         return set_extent_bit(tree, start, end, EXTENT_UPTODATE, 0,
 945                               NULL, cached_state, mask);
 946 }
 947
 948 static int clear_extent_uptodate(struct extent_io_tree *tree, u64 start,
 949                                  u64 end, struct extent_state **cached_state,
 950                                  gfp_t mask)
 951 {
 952         return clear_extent_bit(tree, start, end, EXTENT_UPTODATE, 0, 0,
 953                                 cached_state, mask);
 954 }
 955
 956 /*
 957  * either insert or lock state struct between start and end use mask to tell
 958  * us if waiting is desired.
 959  */
 960 int lock_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
 961                      int bits, struct extent_state **cached_state, gfp_t mask)
 962 {
 963         int err;
 964         u64 failed_start;
 965         while (1) {
 966                 err = set_extent_bit(tree, start, end, EXTENT_LOCKED | bits,
 967                                      EXTENT_LOCKED, &failed_start,
 968                                      cached_state, mask);
 969                 if (err == -EEXIST && (mask & __GFP_WAIT)) {
 970                         wait_extent_bit(tree, failed_start, end, EXTENT_LOCKED);
 971                         start = failed_start;
 972                 } else {
 973                         break;
 974                 }
 975                 WARN_ON(start > end);
 976         }
 977         return err;
 978 }
 979
 980 int lock_extent(struct extent_io_tree *tree, u64 start, u64 end, gfp_t mask)
 981 {
 982         return lock_extent_bits(tree, start, end, 0, NULL, mask);
 983 }
 984
 985 int try_lock_extent(struct extent_io_tree *tree, u64 start, u64 end,
 986                     gfp_t mask)
 987 {
 988         int err;
 989         u64 failed_start;
 990
 991         err = set_extent_bit(tree, start, end, EXTENT_LOCKED, EXTENT_LOCKED,
 992                              &failed_start, NULL, mask);
 993         if (err == -EEXIST) {
 994                 if (failed_start > start)
 995                         clear_extent_bit(tree, start, failed_start - 1,
 996                                          EXTENT_LOCKED, 1, 0, NULL, mask);
 997                 return 0;
 998         }
 999         return 1;
1000 }
1001
1002 int unlock_extent_cached(struct extent_io_tree *tree, u64 start, u64 end,
1003                          struct extent_state **cached, gfp_t mask)
1004 {
1005         return clear_extent_bit(tree, start, end, EXTENT_LOCKED, 1, 0, cached,
1006                                 mask);
1007 }
1008
1009 int unlock_extent(struct extent_io_tree *tree, u64 start, u64 end, gfp_t mask)
1010 {
1011         return clear_extent_bit(tree, start, end, EXTENT_LOCKED, 1, 0, NULL,
1012                                 mask);
1013 }
1014
1015 /*
1016  * helper function to set both pages and extents in the tree writeback
1017  */
1018 static int set_range_writeback(struct extent_io_tree *tree, u64 start, u64 end)
1019 {
1020         unsigned long index = start >> PAGE_CACHE_SHIFT;
1021         unsigned long end_index = end >> PAGE_CACHE_SHIFT;
1022         struct page *page;
1023
1024         while (index <= end_index) {
1025                 page = find_get_page(tree->mapping, index);
1026                 BUG_ON(!page);
1027                 set_page_writeback(page);
1028                 page_cache_release(page);
1029                 index++;
1030         }
1031         return 0;
1032 }
1033
1034 /* find the first state struct with 'bits' set after 'start', and
1035  * return it.  tree->lock must be held.  NULL will returned if
1036  * nothing was found after 'start'
1037  */
1038 struct extent_state *find_first_extent_bit_state(struct extent_io_tree *tree,
1039                                                  u64 start, int bits)
1040 {
1041         struct rb_node *node;
1042         struct extent_state *state;
1043
1044         /*
1045          * this search will find all the extents that end after
1046          * our range starts.
1047          */
1048         node = tree_search(tree, start);
1049         if (!node)
1050                 goto out;
1051
1052         while (1) {
1053                 state = rb_entry(node, struct extent_state, rb_node);
1054                 if (state->end >= start && (state->state & bits))
1055                         return state;
1056
1057                 node = rb_next(node);
1058                 if (!node)
1059                         break;
1060         }
1061 out:
1062         return NULL;
1063 }
1064
1065 /*
1066  * find the first offset in the io tree with 'bits' set. zero is
1067  * returned if we find something, and *start_ret and *end_ret are
1068  * set to reflect the state struct that was found.
1069  *
1070  * If nothing was found, 1 is returned, < 0 on error
1071  */
1072 int find_first_extent_bit(struct extent_io_tree *tree, u64 start,
1073                           u64 *start_ret, u64 *end_ret, int bits)
1074 {
1075         struct extent_state *state;
1076         int ret = 1;
1077
1078         spin_lock(&tree->lock);
1079         state = find_first_extent_bit_state(tree, start, bits);
1080         if (state) {
1081                 *start_ret = state->start;
1082                 *end_ret = state->end;
1083                 ret = 0;
1084         }
1085         spin_unlock(&tree->lock);
1086         return ret;
1087 }
1088
1089 /*
1090  * find a contiguous range of bytes in the file marked as delalloc, not
1091  * more than 'max_bytes'.  start and end are used to return the range,
1092  *
1093  * 1 is returned if we find something, 0 if nothing was in the tree
1094  */
1095 static noinline u64 find_delalloc_range(struct extent_io_tree *tree,
1096                                         u64 *start, u64 *end, u64 max_bytes,
1097                                         struct extent_state **cached_state)
1098 {
1099         struct rb_node *node;
1100         struct extent_state *state;
1101         u64 cur_start = *start;
1102         u64 found = 0;
1103         u64 total_bytes = 0;
1104
1105         spin_lock(&tree->lock);
1106
1107         /*
1108          * this search will find all the extents that end after
1109          * our range starts.
1110          */
1111         node = tree_search(tree, cur_start);
1112         if (!node) {
1113                 if (!found)
1114                         *end = (u64)-1;
1115                 goto out;
1116         }
1117
1118         while (1) {
1119                 state = rb_entry(node, struct extent_state, rb_node);
1120                 if (found && (state->start != cur_start ||
1121                               (state->state & EXTENT_BOUNDARY))) {
1122                         goto out;
1123                 }
1124                 if (!(state->state & EXTENT_DELALLOC)) {
1125                         if (!found)
1126                                 *end = state->end;
1127                         goto out;
1128                 }
1129                 if (!found) {
1130                         *start = state->start;
1131                         *cached_state = state;
1132                         atomic_inc(&state->refs);
1133                 }
1134                 found++;
1135                 *end = state->end;
1136                 cur_start = state->end + 1;
1137                 node = rb_next(node);
1138                 if (!node)
1139                         break;
1140                 total_bytes += state->end - state->start + 1;
1141                 if (total_bytes >= max_bytes)
1142                         break;
1143         }
1144 out:
1145         spin_unlock(&tree->lock);
1146         return found;
1147 }
1148
1149 static noinline int __unlock_for_delalloc(struct inode *inode,
1150                                           struct page *locked_page,
1151                                           u64 start, u64 end)
1152 {
1153         int ret;
1154         struct page *pages[16];
1155         unsigned long index = start >> PAGE_CACHE_SHIFT;
1156         unsigned long end_index = end >> PAGE_CACHE_SHIFT;
1157         unsigned long nr_pages = end_index - index + 1;
1158         int i;
1159
1160         if (index == locked_page->index && end_index == index)
1161                 return 0;
1162
1163         while (nr_pages > 0) {
1164                 ret = find_get_pages_contig(inode->i_mapping, index,
1165                                      min_t(unsigned long, nr_pages,
1166                                      ARRAY_SIZE(pages)), pages);
1167                 for (i = 0; i < ret; i++) {
1168                         if (pages[i] != locked_page)
1169                                 unlock_page(pages[i]);
1170                         page_cache_release(pages[i]);
1171                 }
1172                 nr_pages -= ret;
1173                 index += ret;
1174                 cond_resched();
1175         }
1176         return 0;
1177 }
1178
1179 static noinline int lock_delalloc_pages(struct inode *inode,
1180                                         struct page *locked_page,
1181                                         u64 delalloc_start,
1182                                         u64 delalloc_end)
1183 {
1184         unsigned long index = delalloc_start >> PAGE_CACHE_SHIFT;
1185         unsigned long start_index = index;
1186         unsigned long end_index = delalloc_end >> PAGE_CACHE_SHIFT;
1187         unsigned long pages_locked = 0;
1188         struct page *pages[16];
1189         unsigned long nrpages;
1190         int ret;
1191         int i;
1192
1193         /* the caller is responsible for locking the start index */
1194         if (index == locked_page->index && index == end_index)
1195                 return 0;
1196
1197         /* skip the page at the start index */
1198         nrpages = end_index - index + 1;
1199         while (nrpages > 0) {
1200                 ret = find_get_pages_contig(inode->i_mapping, index,
1201                                      min_t(unsigned long,
1202                                      nrpages, ARRAY_SIZE(pages)), pages);
1203                 if (ret == 0) {
1204                         ret = -EAGAIN;
1205                         goto done;
1206                 }
1207                 /* now we have an array of pages, lock them all */
1208                 for (i = 0; i < ret; i++) {
1209                         /*
1210                          * the caller is taking responsibility for
1211                          * locked_page
1212                          */
1213                         if (pages[i] != locked_page) {
1214                                 lock_page(pages[i]);
1215                                 if (!PageDirty(pages[i]) ||
1216                                     pages[i]->mapping != inode->i_mapping) {
1217                                         ret = -EAGAIN;
1218                                         unlock_page(pages[i]);
1219                                         page_cache_release(pages[i]);
1220                                         goto done;
1221                                 }
1222                         }
1223                         page_cache_release(pages[i]);
1224                         pages_locked++;
1225                 }
1226                 nrpages -= ret;
1227                 index += ret;
1228                 cond_resched();
1229         }
1230         ret = 0;
1231 done:
1232         if (ret && pages_locked) {
1233                 __unlock_for_delalloc(inode, locked_page,
1234                               delalloc_start,
1235                               ((u64)(start_index + pages_locked - 1)) <<
1236                               PAGE_CACHE_SHIFT);
1237         }
1238         return ret;
1239 }
1240
1241 /*
1242  * find a contiguous range of bytes in the file marked as delalloc, not
1243  * more than 'max_bytes'.  start and end are used to return the range,
1244  *
1245  * 1 is returned if we find something, 0 if nothing was in the tree
1246  */
1247 static noinline u64 find_lock_delalloc_range(struct inode *inode,
1248                                              struct extent_io_tree *tree,
1249                                              struct page *locked_page,
1250                                              u64 *start, u64 *end,
1251                                              u64 max_bytes)
1252 {
1253         u64 delalloc_start;
1254         u64 delalloc_end;
1255         u64 found;
1256         struct extent_state *cached_state = NULL;
1257         int ret;
1258         int loops = 0;
1259
1260 again:
1261         /* step one, find a bunch of delalloc bytes starting at start */
1262         delalloc_start = *start;
1263         delalloc_end = 0;
1264         found = find_delalloc_range(tree, &delalloc_start, &delalloc_end,
1265                                     max_bytes, &cached_state);
1266         if (!found || delalloc_end <= *start) {
1267                 *start = delalloc_start;
1268                 *end = delalloc_end;
1269                 free_extent_state(cached_state);
1270                 return found;
1271         }
1272
1273         /*
1274          * start comes from the offset of locked_page.  We have to lock
1275          * pages in order, so we can't process delalloc bytes before
1276          * locked_page
1277          */
1278         if (delalloc_start < *start)
1279                 delalloc_start = *start;
1280
1281         /*
1282          * make sure to limit the number of pages we try to lock down
1283          * if we're looping.
1284          */
1285         if (delalloc_end + 1 - delalloc_start > max_bytes && loops)
1286                 delalloc_end = delalloc_start + PAGE_CACHE_SIZE - 1;
1287
1288         /* step two, lock all the pages after the page that has start */
1289         ret = lock_delalloc_pages(inode, locked_page,
1290                                   delalloc_start, delalloc_end);
1291         if (ret == -EAGAIN) {
1292                 /* some of the pages are gone, lets avoid looping by
1293                  * shortening the size of the delalloc range we're searching
1294                  */
1295                 free_extent_state(cached_state);
1296                 if (!loops) {
1297                         unsigned long offset = (*start) & (PAGE_CACHE_SIZE - 1);
1298                         max_bytes = PAGE_CACHE_SIZE - offset;
1299                         loops = 1;
1300                         goto again;
1301                 } else {
1302                         found = 0;
1303                         goto out_failed;
1304                 }
1305         }
1306         BUG_ON(ret);
1307
1308         /* step three, lock the state bits for the whole range */
1309         lock_extent_bits(tree, delalloc_start, delalloc_end,
1310                          0, &cached_state, GFP_NOFS);
1311
1312         /* then test to make sure it is all still delalloc */
1313         ret = test_range_bit(tree, delalloc_start, delalloc_end,
1314                              EXTENT_DELALLOC, 1, cached_state);
1315         if (!ret) {
1316                 unlock_extent_cached(tree, delalloc_start, delalloc_end,
1317                                      &cached_state, GFP_NOFS);
1318                 __unlock_for_delalloc(inode, locked_page,
1319                               delalloc_start, delalloc_end);
1320                 cond_resched();
1321                 goto again;
1322         }
1323         free_extent_state(cached_state);
1324         *start = delalloc_start;
1325         *end = delalloc_end;
1326 out_failed:
1327         return found;
1328 }
1329
1330 int extent_clear_unlock_delalloc(struct inode *inode,
1331                                 struct extent_io_tree *tree,
1332                                 u64 start, u64 end, struct page *locked_page,
1333                                 unsigned long op)
1334 {
1335         int ret;
1336         struct page *pages[16];
1337         unsigned long index = start >> PAGE_CACHE_SHIFT;
1338         unsigned long end_index = end >> PAGE_CACHE_SHIFT;
1339         unsigned long nr_pages = end_index - index + 1;
1340         int i;
1341         int clear_bits = 0;
1342
1343         if (op & EXTENT_CLEAR_UNLOCK)
1344                 clear_bits |= EXTENT_LOCKED;
1345         if (op & EXTENT_CLEAR_DIRTY)
1346                 clear_bits |= EXTENT_DIRTY;
1347
1348         if (op & EXTENT_CLEAR_DELALLOC)
1349                 clear_bits |= EXTENT_DELALLOC;
1350
1351         clear_extent_bit(tree, start, end, clear_bits, 1, 0, NULL, GFP_NOFS);
1352         if (!(op & (EXTENT_CLEAR_UNLOCK_PAGE | EXTENT_CLEAR_DIRTY |
1353                     EXTENT_SET_WRITEBACK | EXTENT_END_WRITEBACK |
1354                     EXTENT_SET_PRIVATE2)))
1355                 return 0;
1356
1357         while (nr_pages > 0) {
1358                 ret = find_get_pages_contig(inode->i_mapping, index,
1359                                      min_t(unsigned long,
1360                                      nr_pages, ARRAY_SIZE(pages)), pages);
1361                 for (i = 0; i < ret; i++) {
1362
1363                         if (op & EXTENT_SET_PRIVATE2)
1364                                 SetPagePrivate2(pages[i]);
1365
1366                         if (pages[i] == locked_page) {
1367                                 page_cache_release(pages[i]);
1368                                 continue;
1369                         }
1370                         if (op & EXTENT_CLEAR_DIRTY)
1371                                 clear_page_dirty_for_io(pages[i]);
1372                         if (op & EXTENT_SET_WRITEBACK)
1373                                 set_page_writeback(pages[i]);
1374                         if (op & EXTENT_END_WRITEBACK)
1375                                 end_page_writeback(pages[i]);
1376                         if (op & EXTENT_CLEAR_UNLOCK_PAGE)
1377                                 unlock_page(pages[i]);
1378                         page_cache_release(pages[i]);
1379                 }
1380                 nr_pages -= ret;
1381                 index += ret;
1382                 cond_resched();
1383         }
1384         return 0;
1385 }
1386
1387 /*
1388  * count the number of bytes in the tree that have a given bit(s)
1389  * set.  This can be fairly slow, except for EXTENT_DIRTY which is
1390  * cached.  The total number found is returned.
1391  */
1392 u64 count_range_bits(struct extent_io_tree *tree,
1393                      u64 *start, u64 search_end, u64 max_bytes,
1394                      unsigned long bits, int contig)
1395 {
1396         struct rb_node *node;
1397         struct extent_state *state;
1398         u64 cur_start = *start;
1399         u64 total_bytes = 0;
1400         u64 last = 0;
1401         int found = 0;
1402
1403         if (search_end <= cur_start) {
1404                 WARN_ON(1);
1405                 return 0;
1406         }
1407
1408         spin_lock(&tree->lock);
1409         if (cur_start == 0 && bits == EXTENT_DIRTY) {
1410                 total_bytes = tree->dirty_bytes;
1411                 goto out;
1412         }
1413         /*
1414          * this search will find all the extents that end after
1415          * our range starts.
1416          */
1417         node = tree_search(tree, cur_start);
1418         if (!node)
1419                 goto out;
1420
1421         while (1) {
1422                 state = rb_entry(node, struct extent_state, rb_node);
1423                 if (state->start > search_end)
1424                         break;
1425                 if (contig && found && state->start > last + 1)
1426                         break;
1427                 if (state->end >= cur_start && (state->state & bits) == bits) {
1428                         total_bytes += min(search_end, state->end) + 1 -
1429                                        max(cur_start, state->start);
1430                         if (total_bytes >= max_bytes)
1431                                 break;
1432                         if (!found) {
1433                                 *start = max(cur_start, state->start);
1434                                 found = 1;
1435                         }
1436                         last = state->end;
1437                 } else if (contig && found) {
1438                         break;
1439                 }
1440                 node = rb_next(node);
1441                 if (!node)
1442                         break;
1443         }
1444 out:
1445         spin_unlock(&tree->lock);
1446         return total_bytes;
1447 }
1448
1449 /*
1450  * set the private field for a given byte offset in the tree.  If there isn't
1451  * an extent_state there already, this does nothing.
1452  */
1453 int set_state_private(struct extent_io_tree *tree, u64 start, u64 private)
1454 {
1455         struct rb_node *node;
1456         struct extent_state *state;
1457         int ret = 0;
1458
1459         spin_lock(&tree->lock);
1460         /*
1461          * this search will find all the extents that end after
1462          * our range starts.
1463          */
1464         node = tree_search(tree, start);
1465         if (!node) {
1466                 ret = -ENOENT;
1467                 goto out;
1468         }
1469         state = rb_entry(node, struct extent_state, rb_node);
1470         if (state->start != start) {
1471                 ret = -ENOENT;
1472                 goto out;
1473         }
1474         state->private = private;
1475 out:
1476         spin_unlock(&tree->lock);
1477         return ret;
1478 }
1479
1480 int get_state_private(struct extent_io_tree *tree, u64 start, u64 *private)
1481 {
1482         struct rb_node *node;
1483         struct extent_state *state;
1484         int ret = 0;
1485
1486         spin_lock(&tree->lock);
1487         /*
1488          * this search will find all the extents that end after
1489          * our range starts.
1490          */
1491         node = tree_search(tree, start);
1492         if (!node) {
1493                 ret = -ENOENT;
1494                 goto out;
1495         }
1496         state = rb_entry(node, struct extent_state, rb_node);
1497         if (state->start != start) {
1498                 ret = -ENOENT;
1499                 goto out;
1500         }
1501         *private = state->private;
1502 out:
1503         spin_unlock(&tree->lock);
1504         return ret;
1505 }
1506
1507 /*
1508  * searches a range in the state tree for a given mask.
1509  * If 'filled' == 1, this returns 1 only if every extent in the tree
1510  * has the bits set.  Otherwise, 1 is returned if any bit in the
1511  * range is found set.
1512  */
1513 int test_range_bit(struct extent_io_tree *tree, u64 start, u64 end,
1514                    int bits, int filled, struct extent_state *cached)
1515 {
1516         struct extent_state *state = NULL;
1517         struct rb_node *node;
1518         int bitset = 0;
1519
1520         spin_lock(&tree->lock);
1521         if (cached && cached->tree && cached->start <= start &&
1522             cached->end > start)
1523                 node = &cached->rb_node;
1524         else
1525                 node = tree_search(tree, start);
1526         while (node && start <= end) {
1527                 state = rb_entry(node, struct extent_state, rb_node);
1528
1529                 if (filled && state->start > start) {
1530                         bitset = 0;
1531                         break;
1532                 }
1533
1534                 if (state->start > end)
1535                         break;
1536
1537                 if (state->state & bits) {
1538                         bitset = 1;
1539                         if (!filled)
1540                                 break;
1541                 } else if (filled) {
1542                         bitset = 0;
1543                         break;
1544                 }
1545
1546                 if (state->end == (u64)-1)
1547                         break;
1548
1549                 start = state->end + 1;
1550                 if (start > end)
1551                         break;
1552                 node = rb_next(node);
1553                 if (!node) {
1554                         if (filled)
1555                                 bitset = 0;
1556                         break;
1557                 }
1558         }
1559         spin_unlock(&tree->lock);
1560         return bitset;
1561 }
1562
1563 /*
1564  * helper function to set a given page up to date if all the
1565  * extents in the tree for that page are up to date
1566  */
1567 static int check_page_uptodate(struct extent_io_tree *tree,
1568                                struct page *page)
1569 {
1570         u64 start = (u64)page->index << PAGE_CACHE_SHIFT;
1571         u64 end = start + PAGE_CACHE_SIZE - 1;
1572         if (test_range_bit(tree, start, end, EXTENT_UPTODATE, 1, NULL))
1573                 SetPageUptodate(page);
1574         return 0;
1575 }
1576
1577 /*
1578  * helper function to unlock a page if all the extents in the tree
1579  * for that page are unlocked
1580  */
1581 static int check_page_locked(struct extent_io_tree *tree,
1582                              struct page *page)
1583 {
1584         u64 start = (u64)page->index << PAGE_CACHE_SHIFT;
1585         u64 end = start + PAGE_CACHE_SIZE - 1;
1586         if (!test_range_bit(tree, start, end, EXTENT_LOCKED, 0, NULL))
1587                 unlock_page(page);
1588         return 0;
1589 }
1590
1591 /*
1592  * helper function to end page writeback if all the extents
1593  * in the tree for that page are done with writeback
1594  */
1595 static int check_page_writeback(struct extent_io_tree *tree,
1596                              struct page *page)
1597 {
1598         end_page_writeback(page);
1599         return 0;
1600 }
1601
1602 /* lots and lots of room for performance fixes in the end_bio funcs */
1603
1604 /*
1605  * after a writepage IO is done, we need to:
1606  * clear the uptodate bits on error
1607  * clear the writeback bits in the extent tree for this IO
1608  * end_page_writeback if the page has no more pending IO
1609  *
1610  * Scheduling is not allowed, so the extent state tree is expected
1611  * to have one and only one object corresponding to this IO.
1612  */
1613 static void end_bio_extent_writepage(struct bio *bio, int err)
1614 {
1615         int uptodate = err == 0;
1616         struct bio_vec *bvec = bio->bi_io_vec + bio->bi_vcnt - 1;
1617         struct extent_io_tree *tree;
1618         u64 start;
1619         u64 end;
1620         int whole_page;
1621         int ret;
1622
1623         do {
1624                 struct page *page = bvec->bv_page;
1625                 tree = &BTRFS_I(page->mapping->host)->io_tree;
1626
1627                 start = ((u64)page->index << PAGE_CACHE_SHIFT) +
1628                          bvec->bv_offset;
1629                 end = start + bvec->bv_len - 1;
1630
1631                 if (bvec->bv_offset == 0 && bvec->bv_len == PAGE_CACHE_SIZE)
1632                         whole_page = 1;
1633                 else
1634                         whole_page = 0;
1635
1636                 if (--bvec >= bio->bi_io_vec)
1637                         prefetchw(&bvec->bv_page->flags);
1638                 if (tree->ops && tree->ops->writepage_end_io_hook) {
1639                         ret = tree->ops->writepage_end_io_hook(page, start,
1640                                                        end, NULL, uptodate);
1641                         if (ret)
1642                                 uptodate = 0;
1643                 }
1644
1645                 if (!uptodate && tree->ops &&
1646                     tree->ops->writepage_io_failed_hook) {
1647                         ret = tree->ops->writepage_io_failed_hook(bio, page,
1648                                                          start, end, NULL);
1649                         if (ret == 0) {
1650                                 uptodate = (err == 0);
1651                                 continue;
1652                         }
1653                 }
1654
1655                 if (!uptodate) {
1656                         clear_extent_uptodate(tree, start, end, NULL, GFP_NOFS);
1657                         ClearPageUptodate(page);
1658                         SetPageError(page);
1659                 }
1660
1661                 if (whole_page)
1662                         end_page_writeback(page);
1663                 else
1664                         check_page_writeback(tree, page);
1665         } while (bvec >= bio->bi_io_vec);
1666
1667         bio_put(bio);
1668 }
1669
1670 /*
1671  * after a readpage IO is done, we need to:
1672  * clear the uptodate bits on error
1673  * set the uptodate bits if things worked
1674  * set the page up to date if all extents in the tree are uptodate
1675  * clear the lock bit in the extent tree
1676  * unlock the page if there are no other extents locked for it
1677  *
1678  * Scheduling is not allowed, so the extent state tree is expected
1679  * to have one and only one object corresponding to this IO.
1680  */
1681 static void end_bio_extent_readpage(struct bio *bio, int err)
1682 {
1683         int uptodate = test_bit(BIO_UPTODATE, &bio->bi_flags);
1684         struct bio_vec *bvec_end = bio->bi_io_vec + bio->bi_vcnt - 1;
1685         struct bio_vec *bvec = bio->bi_io_vec;
1686         struct extent_io_tree *tree;
1687         u64 start;
1688         u64 end;
1689         int whole_page;
1690         int ret;
1691
1692         if (err)
1693                 uptodate = 0;
1694
1695         do {
1696                 struct page *page = bvec->bv_page;
1697                 struct extent_state *cached = NULL;
1698                 struct extent_state *state;
1699
1700                 tree = &BTRFS_I(page->mapping->host)->io_tree;
1701
1702                 start = ((u64)page->index << PAGE_CACHE_SHIFT) +
1703                         bvec->bv_offset;
1704                 end = start + bvec->bv_len - 1;
1705
1706                 if (bvec->bv_offset == 0 && bvec->bv_len == PAGE_CACHE_SIZE)
1707                         whole_page = 1;
1708                 else
1709                         whole_page = 0;
1710
1711                 if (++bvec <= bvec_end)
1712                         prefetchw(&bvec->bv_page->flags);
1713
1714                 spin_lock(&tree->lock);
1715                 state = find_first_extent_bit_state(tree, start, EXTENT_LOCKED);
1716                 if (state && state->start == start) {
1717                         /*
1718                          * take a reference on the state, unlock will drop
1719                          * the ref
1720                          */
1721                         cache_state(state, &cached);
1722                 }
1723                 spin_unlock(&tree->lock);
1724
1725                 if (uptodate && tree->ops && tree->ops->readpage_end_io_hook) {
1726                         ret = tree->ops->readpage_end_io_hook(page, start, end,
1727                                                               state);
1728                         if (ret)
1729                                 uptodate = 0;
1730                 }
1731                 if (!uptodate && tree->ops &&
1732                     tree->ops->readpage_io_failed_hook) {
1733                         ret = tree->ops->readpage_io_failed_hook(bio, page,
1734                                                          start, end, state);
1735                         if (ret == 0) {
1736                                 uptodate =
1737                                         test_bit(BIO_UPTODATE, &bio->bi_flags);
1738                                 if (err)
1739                                         uptodate = 0;
1740                                 uncache_state(&cached);
1741                                 continue;
1742                         }
1743                 }
1744
1745                 if (uptodate) {
1746                         set_extent_uptodate(tree, start, end, &cached,
1747                                             GFP_ATOMIC);
1748                 }
1749                 unlock_extent_cached(tree, start, end, &cached, GFP_ATOMIC);
1750
1751                 if (whole_page) {
1752                         if (uptodate) {
1753                                 SetPageUptodate(page);
1754                         } else {
1755                                 ClearPageUptodate(page);
1756                                 SetPageError(page);
1757                         }
1758                         unlock_page(page);
1759                 } else {
1760                         if (uptodate) {
1761                                 check_page_uptodate(tree, page);
1762                         } else {
1763                                 ClearPageUptodate(page);
1764                                 SetPageError(page);
1765                         }
1766                         check_page_locked(tree, page);
1767                 }
1768         } while (bvec <= bvec_end);
1769
1770         bio_put(bio);
1771 }
1772
1773 struct bio *
1774 btrfs_bio_alloc(struct block_device *bdev, u64 first_sector, int nr_vecs,
1775                 gfp_t gfp_flags)
1776 {
1777         struct bio *bio;
1778
1779         bio = bio_alloc(gfp_flags, nr_vecs);
1780
1781         if (bio == NULL && (current->flags & PF_MEMALLOC)) {
1782                 while (!bio && (nr_vecs /= 2))
1783                         bio = bio_alloc(gfp_flags, nr_vecs);
1784         }
1785
1786         if (bio) {
1787                 bio->bi_size = 0;
1788                 bio->bi_bdev = bdev;
1789                 bio->bi_sector = first_sector;
1790         }
1791         return bio;
1792 }
1793
1794 static int submit_one_bio(int rw, struct bio *bio, int mirror_num,
1795                           unsigned long bio_flags)
1796 {
1797         int ret = 0;
1798         struct bio_vec *bvec = bio->bi_io_vec + bio->bi_vcnt - 1;
1799         struct page *page = bvec->bv_page;
1800         struct extent_io_tree *tree = bio->bi_private;
1801         u64 start;
1802
1803         start = ((u64)page->index << PAGE_CACHE_SHIFT) + bvec->bv_offset;
1804
1805         bio->bi_private = NULL;
1806
1807         bio_get(bio);
1808
1809         if (tree->ops && tree->ops->submit_bio_hook)
1810                 ret = tree->ops->submit_bio_hook(page->mapping->host, rw, bio,
1811                                            mirror_num, bio_flags, start);
1812         else
1813                 submit_bio(rw, bio);
1814         if (bio_flagged(bio, BIO_EOPNOTSUPP))
1815                 ret = -EOPNOTSUPP;
1816         bio_put(bio);
1817         return ret;
1818 }
1819
1820 static int submit_extent_page(int rw, struct extent_io_tree *tree,
1821                               struct page *page, sector_t sector,
1822                               size_t size, unsigned long offset,
1823                               struct block_device *bdev,
1824                               struct bio **bio_ret,
1825                               unsigned long max_pages,
1826                               bio_end_io_t end_io_func,
1827                               int mirror_num,
1828                               unsigned long prev_bio_flags,
1829                               unsigned long bio_flags)
1830 {
1831         int ret = 0;
1832         struct bio *bio;
1833         int nr;
1834         int contig = 0;
1835         int this_compressed = bio_flags & EXTENT_BIO_COMPRESSED;
1836         int old_compressed = prev_bio_flags & EXTENT_BIO_COMPRESSED;
1837         size_t page_size = min_t(size_t, size, PAGE_CACHE_SIZE);
1838
1839         if (bio_ret && *bio_ret) {
1840                 bio = *bio_ret;
1841                 if (old_compressed)
1842                         contig = bio->bi_sector == sector;
1843                 else
1844                         contig = bio->bi_sector + (bio->bi_size >> 9) ==
1845                                 sector;
1846
1847                 if (prev_bio_flags != bio_flags || !contig ||
1848                     (tree->ops && tree->ops->merge_bio_hook &&
1849                      tree->ops->merge_bio_hook(page, offset, page_size, bio,
1850                                                bio_flags)) ||
1851                     bio_add_page(bio, page, page_size, offset) < page_size) {
1852                         ret = submit_one_bio(rw, bio, mirror_num,
1853                                              prev_bio_flags);
1854                         bio = NULL;
1855                 } else {
1856                         return 0;
1857                 }
1858         }
1859         if (this_compressed)
1860                 nr = BIO_MAX_PAGES;
1861         else
1862                 nr = bio_get_nr_vecs(bdev);
1863
1864         bio = btrfs_bio_alloc(bdev, sector, nr, GFP_NOFS | __GFP_HIGH);
1865         if (!bio)
1866                 return -ENOMEM;
1867
1868         bio_add_page(bio, page, page_size, offset);
1869         bio->bi_end_io = end_io_func;
1870         bio->bi_private = tree;
1871
1872         if (bio_ret)
1873                 *bio_ret = bio;
1874         else
1875                 ret = submit_one_bio(rw, bio, mirror_num, bio_flags);
1876
1877         return ret;
1878 }
1879
1880 void set_page_extent_mapped(struct page *page)
1881 {
1882         if (!PagePrivate(page)) {
1883                 SetPagePrivate(page);
1884                 page_cache_get(page);
1885                 set_page_private(page, EXTENT_PAGE_PRIVATE);
1886         }
1887 }
1888
1889 static void set_page_extent_head(struct page *page, unsigned long len)
1890 {
1891         WARN_ON(!PagePrivate(page));
1892         set_page_private(page, EXTENT_PAGE_PRIVATE_FIRST_PAGE | len << 2);
1893 }
1894
1895 /*
1896  * basic readpage implementation.  Locked extent state structs are inserted
1897  * into the tree that are removed when the IO is done (by the end_io
1898  * handlers)
1899  */
1900 static int __extent_read_full_page(struct extent_io_tree *tree,
1901                                    struct page *page,
1902                                    get_extent_t *get_extent,
1903                                    struct bio **bio, int mirror_num,
1904                                    unsigned long *bio_flags)
1905 {
1906         struct inode *inode = page->mapping->host;
1907         u64 start = (u64)page->index << PAGE_CACHE_SHIFT;
1908         u64 page_end = start + PAGE_CACHE_SIZE - 1;
1909         u64 end;
1910         u64 cur = start;
1911         u64 extent_offset;
1912         u64 last_byte = i_size_read(inode);
1913         u64 block_start;
1914         u64 cur_end;
1915         sector_t sector;
1916         struct extent_map *em;
1917         struct block_device *bdev;
1918         struct btrfs_ordered_extent *ordered;
1919         int ret;
1920         int nr = 0;
1921         size_t pg_offset = 0;
1922         size_t iosize;
1923         size_t disk_io_size;
1924         size_t blocksize = inode->i_sb->s_blocksize;
1925         unsigned long this_bio_flag = 0;
1926
1927         set_page_extent_mapped(page);
1928
1929         if (!PageUptodate(page)) {
1930                 if (cleancache_get_page(page) == 0) {
1931                         BUG_ON(blocksize != PAGE_SIZE);
1932                         goto out;
1933                 }
1934         }
1935
1936         end = page_end;
1937         while (1) {
1938                 lock_extent(tree, start, end, GFP_NOFS);
1939                 ordered = btrfs_lookup_ordered_extent(inode, start);
1940                 if (!ordered)
1941                         break;
1942                 unlock_extent(tree, start, end, GFP_NOFS);
1943                 btrfs_start_ordered_extent(inode, ordered, 1);
1944                 btrfs_put_ordered_extent(ordered);
1945         }
1946
1947         if (page->index == last_byte >> PAGE_CACHE_SHIFT) {
1948                 char *userpage;
1949                 size_t zero_offset = last_byte & (PAGE_CACHE_SIZE - 1);
1950
1951                 if (zero_offset) {
1952                         iosize = PAGE_CACHE_SIZE - zero_offset;
1953                         userpage = kmap_atomic(page, KM_USER0);
1954                         memset(userpage + zero_offset, 0, iosize);
1955                         flush_dcache_page(page);
1956                         kunmap_atomic(userpage, KM_USER0);
1957                 }
1958         }
1959         while (cur <= end) {
1960                 if (cur >= last_byte) {
1961                         char *userpage;
1962                         struct extent_state *cached = NULL;
1963
1964                         iosize = PAGE_CACHE_SIZE - pg_offset;
1965                         userpage = kmap_atomic(page, KM_USER0);
1966                         memset(userpage + pg_offset, 0, iosize);
1967                         flush_dcache_page(page);
1968                         kunmap_atomic(userpage, KM_USER0);
1969                         set_extent_uptodate(tree, cur, cur + iosize - 1,
1970                                             &cached, GFP_NOFS);
1971                         unlock_extent_cached(tree, cur, cur + iosize - 1,
1972                                              &cached, GFP_NOFS);
1973                         break;
1974                 }
1975                 em = get_extent(inode, page, pg_offset, cur,
1976                                 end - cur + 1, 0);
1977                 if (IS_ERR_OR_NULL(em)) {
1978                         SetPageError(page);
1979                         unlock_extent(tree, cur, end, GFP_NOFS);
1980                         break;
1981                 }
1982                 extent_offset = cur - em->start;
1983                 BUG_ON(extent_map_end(em) <= cur);
1984                 BUG_ON(end < cur);
1985
1986                 if (test_bit(EXTENT_FLAG_COMPRESSED, &em->flags)) {
1987                         this_bio_flag = EXTENT_BIO_COMPRESSED;
1988                         extent_set_compress_type(&this_bio_flag,
1989                                                  em->compress_type);
1990                 }
1991
1992                 iosize = min(extent_map_end(em) - cur, end - cur + 1);
1993                 cur_end = min(extent_map_end(em) - 1, end);
1994                 iosize = (iosize + blocksize - 1) & ~((u64)blocksize - 1);
1995                 if (this_bio_flag & EXTENT_BIO_COMPRESSED) {
1996                         disk_io_size = em->block_len;
1997                         sector = em->block_start >> 9;
1998                 } else {
1999                         sector = (em->block_start + extent_offset) >> 9;
2000                         disk_io_size = iosize;
2001                 }
2002                 bdev = em->bdev;
2003                 block_start = em->block_start;
2004                 if (test_bit(EXTENT_FLAG_PREALLOC, &em->flags))
2005                         block_start = EXTENT_MAP_HOLE;
2006                 free_extent_map(em);
2007                 em = NULL;
2008
2009                 /* we've found a hole, just zero and go on */
2010                 if (block_start == EXTENT_MAP_HOLE) {
2011                         char *userpage;
2012                         struct extent_state *cached = NULL;
2013
2014                         userpage = kmap_atomic(page, KM_USER0);
2015                         memset(userpage + pg_offset, 0, iosize);
2016                         flush_dcache_page(page);
2017                         kunmap_atomic(userpage, KM_USER0);
2018
2019                         set_extent_uptodate(tree, cur, cur + iosize - 1,
2020                                             &cached, GFP_NOFS);
2021                         unlock_extent_cached(tree, cur, cur + iosize - 1,
2022                                              &cached, GFP_NOFS);
2023                         cur = cur + iosize;
2024                         pg_offset += iosize;
2025                         continue;
2026                 }
2027                 /* the get_extent function already copied into the page */
2028                 if (test_range_bit(tree, cur, cur_end,
2029                                    EXTENT_UPTODATE, 1, NULL)) {
2030                         check_page_uptodate(tree, page);
2031                         unlock_extent(tree, cur, cur + iosize - 1, GFP_NOFS);
2032                         cur = cur + iosize;
2033                         pg_offset += iosize;
2034                         continue;
2035                 }
2036                 /* we have an inline extent but it didn't get marked up
2037                  * to date.  Error out
2038                  */
2039                 if (block_start == EXTENT_MAP_INLINE) {
2040                         SetPageError(page);
2041                         unlock_extent(tree, cur, cur + iosize - 1, GFP_NOFS);
2042                         cur = cur + iosize;
2043                         pg_offset += iosize;
2044                         continue;
2045                 }
2046
2047                 ret = 0;
2048                 if (tree->ops && tree->ops->readpage_io_hook) {
2049                         ret = tree->ops->readpage_io_hook(page, cur,
2050                                                           cur + iosize - 1);
2051                 }
2052                 if (!ret) {
2053                         unsigned long pnr = (last_byte >> PAGE_CACHE_SHIFT) + 1;
2054                         pnr -= page->index;
2055                         ret = submit_extent_page(READ, tree, page,
2056                                          sector, disk_io_size, pg_offset,
2057                                          bdev, bio, pnr,
2058                                          end_bio_extent_readpage, mirror_num,
2059                                          *bio_flags,
2060                                          this_bio_flag);
2061                         nr++;
2062                         *bio_flags = this_bio_flag;
2063                 }
2064                 if (ret)
2065                         SetPageError(page);
2066                 cur = cur + iosize;
2067                 pg_offset += iosize;
2068         }
2069 out:
2070         if (!nr) {
2071                 if (!PageError(page))
2072                         SetPageUptodate(page);
2073                 unlock_page(page);
2074         }
2075         return 0;
2076 }
2077
2078 int extent_read_full_page(struct extent_io_tree *tree, struct page *page,
2079                             get_extent_t *get_extent)
2080 {
2081         struct bio *bio = NULL;
2082         unsigned long bio_flags = 0;
2083         int ret;
2084
2085         ret = __extent_read_full_page(tree, page, get_extent, &bio, 0,
2086                                       &bio_flags);
2087         if (bio)
2088                 ret = submit_one_bio(READ, bio, 0, bio_flags);
2089         return ret;
2090 }
2091
2092 static noinline void update_nr_written(struct page *page,
2093                                       struct writeback_control *wbc,
2094                                       unsigned long nr_written)
2095 {
2096         wbc->nr_to_write -= nr_written;
2097         if (wbc->range_cyclic || (wbc->nr_to_write > 0 &&
2098             wbc->range_start == 0 && wbc->range_end == LLONG_MAX))
2099                 page->mapping->writeback_index = page->index + nr_written;
2100 }
2101
2102 /*
2103  * the writepage semantics are similar to regular writepage.  extent
2104  * records are inserted to lock ranges in the tree, and as dirty areas
2105  * are found, they are marked writeback.  Then the lock bits are removed
2106  * and the end_io handler clears the writeback ranges
2107  */
2108 static int __extent_writepage(struct page *page, struct writeback_control *wbc,
2109                               void *data)
2110 {
2111         struct inode *inode = page->mapping->host;
2112         struct extent_page_data *epd = data;
2113         struct extent_io_tree *tree = epd->tree;
2114         u64 start = (u64)page->index << PAGE_CACHE_SHIFT;
2115         u64 delalloc_start;
2116         u64 page_end = start + PAGE_CACHE_SIZE - 1;
2117         u64 end;
2118         u64 cur = start;
2119         u64 extent_offset;
2120         u64 last_byte = i_size_read(inode);
2121         u64 block_start;
2122         u64 iosize;
2123         sector_t sector;
2124         struct extent_state *cached_state = NULL;
2125         struct extent_map *em;
2126         struct block_device *bdev;
2127         int ret;
2128         int nr = 0;
2129         size_t pg_offset = 0;
2130         size_t blocksize;
2131         loff_t i_size = i_size_read(inode);
2132         unsigned long end_index = i_size >> PAGE_CACHE_SHIFT;
2133         u64 nr_delalloc;
2134         u64 delalloc_end;
2135         int page_started;
2136         int compressed;
2137         int write_flags;
2138         unsigned long nr_written = 0;
2139
2140         if (wbc->sync_mode == WB_SYNC_ALL)
2141                 write_flags = WRITE_SYNC;
2142         else
2143                 write_flags = WRITE;
2144
2145         trace___extent_writepage(page, inode, wbc);
2146
2147         WARN_ON(!PageLocked(page));
2148         pg_offset = i_size & (PAGE_CACHE_SIZE - 1);
2149         if (page->index > end_index ||
2150            (page->index == end_index && !pg_offset)) {
2151                 page->mapping->a_ops->invalidatepage(page, 0);
2152                 unlock_page(page);
2153                 return 0;
2154         }
2155
2156         if (page->index == end_index) {
2157                 char *userpage;
2158
2159                 userpage = kmap_atomic(page, KM_USER0);
2160                 memset(userpage + pg_offset, 0,
2161                        PAGE_CACHE_SIZE - pg_offset);
2162                 kunmap_atomic(userpage, KM_USER0);
2163                 flush_dcache_page(page);
2164         }
2165         pg_offset = 0;
2166
2167         set_page_extent_mapped(page);
2168
2169         delalloc_start = start;
2170         delalloc_end = 0;
2171         page_started = 0;
2172         if (!epd->extent_locked) {
2173                 u64 delalloc_to_write = 0;
2174                 /*
2175                  * make sure the wbc mapping index is at least updated
2176                  * to this page.
2177                  */
2178                 update_nr_written(page, wbc, 0);
2179
2180                 while (delalloc_end < page_end) {
2181                         nr_delalloc = find_lock_delalloc_range(inode, tree,
2182                                                        page,
2183                                                        &delalloc_start,
2184                                                        &delalloc_end,
2185                                                        128 * 1024 * 1024);
2186                         if (nr_delalloc == 0) {
2187                                 delalloc_start = delalloc_end + 1;
2188                                 continue;
2189                         }
2190                         tree->ops->fill_delalloc(inode, page, delalloc_start,
2191                                                  delalloc_end, &page_started,
2192                                                  &nr_written);
2193                         /*
2194                          * delalloc_end is already one less than the total
2195                          * length, so we don't subtract one from
2196                          * PAGE_CACHE_SIZE
2197                          */
2198                         delalloc_to_write += (delalloc_end - delalloc_start +
2199                                               PAGE_CACHE_SIZE) >>
2200                                               PAGE_CACHE_SHIFT;
2201                         delalloc_start = delalloc_end + 1;
2202                 }
2203                 if (wbc->nr_to_write < delalloc_to_write) {
2204                         int thresh = 8192;
2205
2206                         if (delalloc_to_write < thresh * 2)
2207                                 thresh = delalloc_to_write;
2208                         wbc->nr_to_write = min_t(u64, delalloc_to_write,
2209                                                  thresh);
2210                 }
2211
2212                 /* did the fill delalloc function already unlock and start
2213                  * the IO?
2214                  */
2215                 if (page_started) {
2216                         ret = 0;
2217                         /*
2218                          * we've unlocked the page, so we can't update
2219                          * the mapping's writeback index, just update
2220                          * nr_to_write.
2221                          */
2222                         wbc->nr_to_write -= nr_written;
2223                         goto done_unlocked;
2224                 }
2225         }
2226         if (tree->ops && tree->ops->writepage_start_hook) {
2227                 ret = tree->ops->writepage_start_hook(page, start,
2228                                                       page_end);
2229                 if (ret == -EAGAIN) {
2230                         redirty_page_for_writepage(wbc, page);
2231                         update_nr_written(page, wbc, nr_written);
2232                         unlock_page(page);
2233                         ret = 0;
2234                         goto done_unlocked;
2235                 }
2236         }
2237
2238         /*
2239          * we don't want to touch the inode after unlocking the page,
2240          * so we update the mapping writeback index now
2241          */
2242         update_nr_written(page, wbc, nr_written + 1);
2243
2244         end = page_end;
2245         if (last_byte <= start) {
2246                 if (tree->ops && tree->ops->writepage_end_io_hook)
2247                         tree->ops->writepage_end_io_hook(page, start,
2248                                                          page_end, NULL, 1);
2249                 goto done;
2250         }
2251
2252         blocksize = inode->i_sb->s_blocksize;
2253
2254         while (cur <= end) {
2255                 if (cur >= last_byte) {
2256                         if (tree->ops && tree->ops->writepage_end_io_hook)
2257                                 tree->ops->writepage_end_io_hook(page, cur,
2258                                                          page_end, NULL, 1);
2259                         break;
2260                 }
2261                 em = epd->get_extent(inode, page, pg_offset, cur,
2262                                      end - cur + 1, 1);
2263                 if (IS_ERR_OR_NULL(em)) {
2264                         SetPageError(page);
2265                         break;
2266                 }
2267
2268                 extent_offset = cur - em->start;
2269                 BUG_ON(extent_map_end(em) <= cur);
2270                 BUG_ON(end < cur);
2271                 iosize = min(extent_map_end(em) - cur, end - cur + 1);
2272                 iosize = (iosize + blocksize - 1) & ~((u64)blocksize - 1);
2273                 sector = (em->block_start + extent_offset) >> 9;
2274                 bdev = em->bdev;
2275                 block_start = em->block_start;
2276                 compressed = test_bit(EXTENT_FLAG_COMPRESSED, &em->flags);
2277                 free_extent_map(em);
2278                 em = NULL;
2279
2280                 /*
2281                  * compressed and inline extents are written through other
2282                  * paths in the FS
2283                  */
2284                 if (compressed || block_start == EXTENT_MAP_HOLE ||
2285                     block_start == EXTENT_MAP_INLINE) {
2286                         /*
2287                          * end_io notification does not happen here for
2288                          * compressed extents
2289                          */
2290                         if (!compressed && tree->ops &&
2291                             tree->ops->writepage_end_io_hook)
2292                                 tree->ops->writepage_end_io_hook(page, cur,
2293                                                          cur + iosize - 1,
2294                                                          NULL, 1);
2295                         else if (compressed) {
2296                                 /* we don't want to end_page_writeback on
2297                                  * a compressed extent.  this happens
2298                                  * elsewhere
2299                                  */
2300                                 nr++;
2301                         }
2302
2303                         cur += iosize;
2304                         pg_offset += iosize;
2305                         continue;
2306                 }
2307                 /* leave this out until we have a page_mkwrite call */
2308                 if (0 && !test_range_bit(tree, cur, cur + iosize - 1,
2309                                    EXTENT_DIRTY, 0, NULL)) {
2310                         cur = cur + iosize;
2311                         pg_offset += iosize;
2312                         continue;
2313                 }
2314
2315                 if (tree->ops && tree->ops->writepage_io_hook) {
2316                         ret = tree->ops->writepage_io_hook(page, cur,
2317                                                 cur + iosize - 1);
2318                 } else {
2319                         ret = 0;
2320                 }
2321                 if (ret) {
2322                         SetPageError(page);
2323                 } else {
2324                         unsigned long max_nr = end_index + 1;
2325
2326                         set_range_writeback(tree, cur, cur + iosize - 1);
2327                         if (!PageWriteback(page)) {
2328                                 printk(KERN_ERR "btrfs warning page %lu not "
2329                                        "writeback, cur %llu end %llu\n",
2330                                        page->index, (unsigned long long)cur,
2331                                        (unsigned long long)end);
2332                         }
2333
2334                         ret = submit_extent_page(write_flags, tree, page,
2335                                                  sector, iosize, pg_offset,
2336                                                  bdev, &epd->bio, max_nr,
2337                                                  end_bio_extent_writepage,
2338                                                  0, 0, 0);
2339                         if (ret)
2340                                 SetPageError(page);
2341                 }
2342                 cur = cur + iosize;
2343                 pg_offset += iosize;
2344                 nr++;
2345         }
2346 done:
2347         if (nr == 0) {
2348                 /* make sure the mapping tag for page dirty gets cleared */
2349                 set_page_writeback(page);
2350                 end_page_writeback(page);
2351         }
2352         unlock_page(page);
2353
2354 done_unlocked:
2355
2356         /* drop our reference on any cached states */
2357         free_extent_state(cached_state);
2358         return 0;
2359 }
2360
2361 /**
2362  * write_cache_pages - walk the list of dirty pages of the given address space and write all of them.
2363  * @mapping: address space structure to write
2364  * @wbc: subtract the number of written pages from *@wbc->nr_to_write
2365  * @writepage: function called for each page
2366  * @data: data passed to writepage function
2367  *
2368  * If a page is already under I/O, write_cache_pages() skips it, even
2369  * if it's dirty.  This is desirable behaviour for memory-cleaning writeback,
2370  * but it is INCORRECT for data-integrity system calls such as fsync().  fsync()
2371  * and msync() need to guarantee that all the data which was dirty at the time
2372  * the call was made get new I/O started against them.  If wbc->sync_mode is
2373  * WB_SYNC_ALL then we were called for data integrity and we must wait for
2374  * existing IO to complete.
2375  */
2376 static int extent_write_cache_pages(struct extent_io_tree *tree,
2377                              struct address_space *mapping,
2378                              struct writeback_control *wbc,
2379                              writepage_t writepage, void *data,
2380                              void (*flush_fn)(void *))
2381 {
2382         int ret = 0;
2383         int done = 0;
2384         int nr_to_write_done = 0;
2385         struct pagevec pvec;
2386         int nr_pages;
2387         pgoff_t index;
2388         pgoff_t end;            /* Inclusive */
2389         int scanned = 0;
2390         int tag;
2391
2392         pagevec_init(&pvec, 0);
2393         if (wbc->range_cyclic) {
2394                 index = mapping->writeback_index; /* Start from prev offset */
2395                 end = -1;
2396         } else {
2397                 index = wbc->range_start >> PAGE_CACHE_SHIFT;
2398                 end = wbc->range_end >> PAGE_CACHE_SHIFT;
2399                 scanned = 1;
2400         }
2401         if (wbc->sync_mode == WB_SYNC_ALL)
2402                 tag = PAGECACHE_TAG_TOWRITE;
2403         else
2404                 tag = PAGECACHE_TAG_DIRTY;
2405 retry:
2406         if (wbc->sync_mode == WB_SYNC_ALL)
2407                 tag_pages_for_writeback(mapping, index, end);
2408         while (!done && !nr_to_write_done && (index <= end) &&
2409                (nr_pages = pagevec_lookup_tag(&pvec, mapping, &index, tag,
2410                         min(end - index, (pgoff_t)PAGEVEC_SIZE-1) + 1))) {
2411                 unsigned i;
2412
2413                 scanned = 1;
2414                 for (i = 0; i < nr_pages; i++) {
2415                         struct page *page = pvec.pages[i];
2416
2417                         /*
2418                          * At this point we hold neither mapping->tree_lock nor
2419                          * lock on the page itself: the page may be truncated or
2420                          * invalidated (changing page->mapping to NULL), or even
2421                          * swizzled back from swapper_space to tmpfs file
2422                          * mapping
2423                          */
2424                         if (tree->ops && tree->ops->write_cache_pages_lock_hook)
2425                                 tree->ops->write_cache_pages_lock_hook(page);
2426                         else
2427                                 lock_page(page);
2428
2429                         if (unlikely(page->mapping != mapping)) {
2430                                 unlock_page(page);
2431                                 continue;
2432                         }
2433
2434                         if (!wbc->range_cyclic && page->index > end) {
2435                                 done = 1;
2436                                 unlock_page(page);
2437                                 continue;
2438                         }
2439
2440                         if (wbc->sync_mode != WB_SYNC_NONE) {
2441                                 if (PageWriteback(page))
2442                                         flush_fn(data);
2443                                 wait_on_page_writeback(page);
2444                         }
2445
2446                         if (PageWriteback(page) ||
2447                             !clear_page_dirty_for_io(page)) {
2448                                 unlock_page(page);
2449                                 continue;
2450                         }
2451
2452                         ret = (*writepage)(page, wbc, data);
2453
2454                         if (unlikely(ret == AOP_WRITEPAGE_ACTIVATE)) {
2455                                 unlock_page(page);
2456                                 ret = 0;
2457                         }
2458                         if (ret)
2459                                 done = 1;
2460
2461                         /*
2462                          * the filesystem may choose to bump up nr_to_write.
2463                          * We have to make sure to honor the new nr_to_write
2464                          * at any time
2465                          */
2466                         nr_to_write_done = wbc->nr_to_write <= 0;
2467                 }
2468                 pagevec_release(&pvec);
2469                 cond_resched();
2470         }
2471         if (!scanned && !done) {
2472                 /*
2473                  * We hit the last page and there is more work to be done: wrap
2474                  * back to the start of the file
2475                  */
2476                 scanned = 1;
2477                 index = 0;
2478                 goto retry;
2479         }
2480         return ret;
2481 }
2482
2483 static void flush_epd_write_bio(struct extent_page_data *epd)
2484 {
2485         if (epd->bio) {
2486                 if (epd->sync_io)
2487                         submit_one_bio(WRITE_SYNC, epd->bio, 0, 0);
2488                 else
2489                         submit_one_bio(WRITE, epd->bio, 0, 0);
2490                 epd->bio = NULL;
2491         }
2492 }
2493
2494 static noinline void flush_write_bio(void *data)
2495 {
2496         struct extent_page_data *epd = data;
2497         flush_epd_write_bio(epd);
2498 }
2499
2500 int extent_write_full_page(struct extent_io_tree *tree, struct page *page,
2501                           get_extent_t *get_extent,
2502                           struct writeback_control *wbc)
2503 {
2504         int ret;
2505         struct extent_page_data epd = {
2506                 .bio = NULL,
2507                 .tree = tree,
2508                 .get_extent = get_extent,
2509                 .extent_locked = 0,
2510                 .sync_io = wbc->sync_mode == WB_SYNC_ALL,
2511         };
2512
2513         ret = __extent_writepage(page, wbc, &epd);
2514
2515         flush_epd_write_bio(&epd);
2516         return ret;
2517 }
2518
2519 int extent_write_locked_range(struct extent_io_tree *tree, struct inode *inode,
2520                               u64 start, u64 end, get_extent_t *get_extent,
2521                               int mode)
2522 {
2523         int ret = 0;
2524         struct address_space *mapping = inode->i_mapping;
2525         struct page *page;
2526         unsigned long nr_pages = (end - start + PAGE_CACHE_SIZE) >>
2527                 PAGE_CACHE_SHIFT;
2528
2529         struct extent_page_data epd = {
2530                 .bio = NULL,
2531                 .tree = tree,
2532                 .get_extent = get_extent,
2533                 .extent_locked = 1,
2534                 .sync_io = mode == WB_SYNC_ALL,
2535         };
2536         struct writeback_control wbc_writepages = {
2537                 .sync_mode      = mode,
2538                 .nr_to_write    = nr_pages * 2,
2539                 .range_start    = start,
2540                 .range_end      = end + 1,
2541         };
2542
2543         while (start <= end) {
2544                 page = find_get_page(mapping, start >> PAGE_CACHE_SHIFT);
2545                 if (clear_page_dirty_for_io(page))
2546                         ret = __extent_writepage(page, &wbc_writepages, &epd);
2547                 else {
2548                         if (tree->ops && tree->ops->writepage_end_io_hook)
2549                                 tree->ops->writepage_end_io_hook(page, start,
2550                                                  start + PAGE_CACHE_SIZE - 1,
2551                                                  NULL, 1);
2552                         unlock_page(page);
2553                 }
2554                 page_cache_release(page);
2555                 start += PAGE_CACHE_SIZE;
2556         }
2557
2558         flush_epd_write_bio(&epd);
2559         return ret;
2560 }
2561
2562 int extent_writepages(struct extent_io_tree *tree,
2563                       struct address_space *mapping,
2564                       get_extent_t *get_extent,
2565                       struct writeback_control *wbc)
2566 {
2567         int ret = 0;
2568         struct extent_page_data epd = {
2569                 .bio = NULL,
2570                 .tree = tree,
2571                 .get_extent = get_extent,
2572                 .extent_locked = 0,
2573                 .sync_io = wbc->sync_mode == WB_SYNC_ALL,
2574         };
2575
2576         ret = extent_write_cache_pages(tree, mapping, wbc,
2577                                        __extent_writepage, &epd,
2578                                        flush_write_bio);
2579         flush_epd_write_bio(&epd);
2580         return ret;
2581 }
2582
2583 int extent_readpages(struct extent_io_tree *tree,
2584                      struct address_space *mapping,
2585                      struct list_head *pages, unsigned nr_pages,
2586                      get_extent_t get_extent)
2587 {
2588         struct bio *bio = NULL;
2589         unsigned page_idx;
2590         unsigned long bio_flags = 0;
2591
2592         for (page_idx = 0; page_idx < nr_pages; page_idx++) {
2593                 struct page *page = list_entry(pages->prev, struct page, lru);
2594
2595                 prefetchw(&page->flags);
2596                 list_del(&page->lru);
2597                 if (!add_to_page_cache_lru(page, mapping,
2598                                         page->index, GFP_NOFS)) {
2599                         __extent_read_full_page(tree, page, get_extent,
2600                                                 &bio, 0, &bio_flags);
2601                 }
2602                 page_cache_release(page);
2603         }
2604         BUG_ON(!list_empty(pages));
2605         if (bio)
2606                 submit_one_bio(READ, bio, 0, bio_flags);
2607         return 0;
2608 }
2609
2610 /*
2611  * basic invalidatepage code, this waits on any locked or writeback
2612  * ranges corresponding to the page, and then deletes any extent state
2613  * records from the tree
2614  */
2615 int extent_invalidatepage(struct extent_io_tree *tree,
2616                           struct page *page, unsigned long offset)
2617 {
2618         struct extent_state *cached_state = NULL;
2619         u64 start = ((u64)page->index << PAGE_CACHE_SHIFT);
2620         u64 end = start + PAGE_CACHE_SIZE - 1;
2621         size_t blocksize = page->mapping->host->i_sb->s_blocksize;
2622
2623         start += (offset + blocksize - 1) & ~(blocksize - 1);
2624         if (start > end)
2625                 return 0;
2626
2627         lock_extent_bits(tree, start, end, 0, &cached_state, GFP_NOFS);
2628         wait_on_page_writeback(page);
2629         clear_extent_bit(tree, start, end,
2630                          EXTENT_LOCKED | EXTENT_DIRTY | EXTENT_DELALLOC |
2631                          EXTENT_DO_ACCOUNTING,
2632                          1, 1, &cached_state, GFP_NOFS);
2633         return 0;
2634 }
2635
2636 /*
2637  * a helper for releasepage, this tests for areas of the page that
2638  * are locked or under IO and drops the related state bits if it is safe
2639  * to drop the page.
2640  */
2641 int try_release_extent_state(struct extent_map_tree *map,
2642                              struct extent_io_tree *tree, struct page *page,
2643                              gfp_t mask)
2644 {
2645         u64 start = (u64)page->index << PAGE_CACHE_SHIFT;
2646         u64 end = start + PAGE_CACHE_SIZE - 1;
2647         int ret = 1;
2648
2649         if (test_range_bit(tree, start, end,
2650                            EXTENT_IOBITS, 0, NULL))
2651                 ret = 0;
2652         else {
2653                 if ((mask & GFP_NOFS) == GFP_NOFS)
2654                         mask = GFP_NOFS;
2655                 /*
2656                  * at this point we can safely clear everything except the
2657                  * locked bit and the nodatasum bit
2658                  */
2659                 ret = clear_extent_bit(tree, start, end,
2660                                  ~(EXTENT_LOCKED | EXTENT_NODATASUM),
2661                                  0, 0, NULL, mask);
2662
2663                 /* if clear_extent_bit failed for enomem reasons,
2664                  * we can't allow the release to continue.
2665                  */
2666                 if (ret < 0)
2667                         ret = 0;
2668                 else
2669                         ret = 1;
2670         }
2671         return ret;
2672 }
2673
2674 /*
2675  * a helper for releasepage.  As long as there are no locked extents
2676  * in the range corresponding to the page, both state records and extent
2677  * map records are removed
2678  */
2679 int try_release_extent_mapping(struct extent_map_tree *map,
2680                                struct extent_io_tree *tree, struct page *page,
2681                                gfp_t mask)
2682 {
2683         struct extent_map *em;
2684         u64 start = (u64)page->index << PAGE_CACHE_SHIFT;
2685         u64 end = start + PAGE_CACHE_SIZE - 1;
2686
2687         if ((mask & __GFP_WAIT) &&
2688             page->mapping->host->i_size > 16 * 1024 * 1024) {
2689                 u64 len;
2690                 while (start <= end) {
2691                         len = end - start + 1;
2692                         write_lock(&map->lock);
2693                         em = lookup_extent_mapping(map, start, len);
2694                         if (IS_ERR_OR_NULL(em)) {
2695                                 write_unlock(&map->lock);
2696                                 break;
2697                         }
2698                         if (test_bit(EXTENT_FLAG_PINNED, &em->flags) ||
2699                             em->start != start) {
2700                                 write_unlock(&map->lock);
2701                                 free_extent_map(em);
2702                                 break;
2703                         }
2704                         if (!test_range_bit(tree, em->start,
2705                                             extent_map_end(em) - 1,
2706                                             EXTENT_LOCKED | EXTENT_WRITEBACK,
2707                                             0, NULL)) {
2708                                 remove_extent_mapping(map, em);
2709                                 /* once for the rb tree */
2710                                 free_extent_map(em);
2711                         }
2712                         start = extent_map_end(em);
2713                         write_unlock(&map->lock);
2714
2715                         /* once for us */
2716                         free_extent_map(em);
2717                 }
2718         }
2719         return try_release_extent_state(map, tree, page, mask);
2720 }
2721
2722 /*
2723  * helper function for fiemap, which doesn't want to see any holes.
2724  * This maps until we find something past 'last'
2725  */
2726 static struct extent_map *get_extent_skip_holes(struct inode *inode,
2727                                                 u64 offset,
2728                                                 u64 last,
2729                                                 get_extent_t *get_extent)
2730 {
2731         u64 sectorsize = BTRFS_I(inode)->root->sectorsize;
2732         struct extent_map *em;
2733         u64 len;
2734
2735         if (offset >= last)
2736                 return NULL;
2737
2738         while(1) {
2739                 len = last - offset;
2740                 if (len == 0)
2741                         break;
2742                 len = (len + sectorsize - 1) & ~(sectorsize - 1);
2743                 em = get_extent(inode, NULL, 0, offset, len, 0);
2744                 if (IS_ERR_OR_NULL(em))
2745                         return em;
2746
2747                 /* if this isn't a hole return it */
2748                 if (!test_bit(EXTENT_FLAG_VACANCY, &em->flags) &&
2749                     em->block_start != EXTENT_MAP_HOLE) {
2750                         return em;
2751                 }
2752
2753                 /* this is a hole, advance to the next extent */
2754                 offset = extent_map_end(em);
2755                 free_extent_map(em);
2756                 if (offset >= last)
2757                         break;
2758         }
2759         return NULL;
2760 }
2761
2762 int extent_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
2763                 __u64 start, __u64 len, get_extent_t *get_extent)
2764 {
2765         int ret = 0;
2766         u64 off = start;
2767         u64 max = start + len;
2768         u32 flags = 0;
2769         u32 found_type;
2770         u64 last;
2771         u64 last_for_get_extent = 0;
2772         u64 disko = 0;
2773         u64 isize = i_size_read(inode);
2774         struct btrfs_key found_key;
2775         struct extent_map *em = NULL;
2776         struct extent_state *cached_state = NULL;
2777         struct btrfs_path *path;
2778         struct btrfs_file_extent_item *item;
2779         int end = 0;
2780         u64 em_start = 0;
2781         u64 em_len = 0;
2782         u64 em_end = 0;
2783         unsigned long emflags;
2784
2785         if (len == 0)
2786                 return -EINVAL;
2787
2788         path = btrfs_alloc_path();
2789         if (!path)
2790                 return -ENOMEM;
2791         path->leave_spinning = 1;
2792
2793         /*
2794          * lookup the last file extent.  We're not using i_size here
2795          * because there might be preallocation past i_size
2796          */
2797         ret = btrfs_lookup_file_extent(NULL, BTRFS_I(inode)->root,
2798                                        path, btrfs_ino(inode), -1, 0);
2799         if (ret < 0) {
2800                 btrfs_free_path(path);
2801                 return ret;
2802         }
2803         WARN_ON(!ret);
2804         path->slots[0]--;
2805         item = btrfs_item_ptr(path->nodes[0], path->slots[0],
2806                               struct btrfs_file_extent_item);
2807         btrfs_item_key_to_cpu(path->nodes[0], &found_key, path->slots[0]);
2808         found_type = btrfs_key_type(&found_key);
2809
2810         /* No extents, but there might be delalloc bits */
2811         if (found_key.objectid != btrfs_ino(inode) ||
2812             found_type != BTRFS_EXTENT_DATA_KEY) {
2813                 /* have to trust i_size as the end */
2814                 last = (u64)-1;
2815                 last_for_get_extent = isize;
2816         } else {
2817                 /*
2818                  * remember the start of the last extent.  There are a
2819                  * bunch of different factors that go into the length of the
2820                  * extent, so its much less complex to remember where it started
2821                  */
2822                 last = found_key.offset;
2823                 last_for_get_extent = last + 1;
2824         }
2825         btrfs_free_path(path);
2826
2827         /*
2828          * we might have some extents allocated but more delalloc past those
2829          * extents.  so, we trust isize unless the start of the last extent is
2830          * beyond isize
2831          */
2832         if (last < isize) {
2833                 last = (u64)-1;
2834                 last_for_get_extent = isize;
2835         }
2836
2837         lock_extent_bits(&BTRFS_I(inode)->io_tree, start, start + len, 0,
2838                          &cached_state, GFP_NOFS);
2839
2840         em = get_extent_skip_holes(inode, off, last_for_get_extent,
2841                                    get_extent);
2842         if (!em)
2843                 goto out;
2844         if (IS_ERR(em)) {
2845                 ret = PTR_ERR(em);
2846                 goto out;
2847         }
2848
2849         while (!end) {
2850                 u64 offset_in_extent;
2851
2852                 /* break if the extent we found is outside the range */
2853                 if (em->start >= max || extent_map_end(em) < off)
2854                         break;
2855
2856                 /*
2857                  * get_extent may return an extent that starts before our
2858                  * requested range.  We have to make sure the ranges
2859                  * we return to fiemap always move forward and don't
2860                  * overlap, so adjust the offsets here
2861                  */
2862                 em_start = max(em->start, off);
2863
2864                 /*
2865                  * record the offset from the start of the extent
2866                  * for adjusting the disk offset below
2867                  */
2868                 offset_in_extent = em_start - em->start;
2869                 em_end = extent_map_end(em);
2870                 em_len = em_end - em_start;
2871                 emflags = em->flags;
2872                 disko = 0;
2873                 flags = 0;
2874
2875                 /*
2876                  * bump off for our next call to get_extent
2877                  */
2878                 off = extent_map_end(em);
2879                 if (off >= max)
2880                         end = 1;
2881
2882                 if (em->block_start == EXTENT_MAP_LAST_BYTE) {
2883                         end = 1;
2884                         flags |= FIEMAP_EXTENT_LAST;
2885                 } else if (em->block_start == EXTENT_MAP_INLINE) {
2886                         flags |= (FIEMAP_EXTENT_DATA_INLINE |
2887                                   FIEMAP_EXTENT_NOT_ALIGNED);
2888                 } else if (em->block_start == EXTENT_MAP_DELALLOC) {
2889                         flags |= (FIEMAP_EXTENT_DELALLOC |
2890                                   FIEMAP_EXTENT_UNKNOWN);
2891                 } else {
2892                         disko = em->block_start + offset_in_extent;
2893                 }
2894                 if (test_bit(EXTENT_FLAG_COMPRESSED, &em->flags))
2895                         flags |= FIEMAP_EXTENT_ENCODED;
2896
2897                 free_extent_map(em);
2898                 em = NULL;
2899                 if ((em_start >= last) || em_len == (u64)-1 ||
2900                    (last == (u64)-1 && isize <= em_end)) {
2901                         flags |= FIEMAP_EXTENT_LAST;
2902                         end = 1;
2903                 }
2904
2905                 /* now scan forward to see if this is really the last extent. */
2906                 em = get_extent_skip_holes(inode, off, last_for_get_extent,
2907                                            get_extent);
2908                 if (IS_ERR(em)) {
2909                         ret = PTR_ERR(em);
2910                         goto out;
2911                 }
2912                 if (!em) {
2913                         flags |= FIEMAP_EXTENT_LAST;
2914                         end = 1;
2915                 }
2916                 ret = fiemap_fill_next_extent(fieinfo, em_start, disko,
2917                                               em_len, flags);
2918                 if (ret)
2919                         goto out_free;
2920         }
2921 out_free:
2922         free_extent_map(em);
2923 out:
2924         unlock_extent_cached(&BTRFS_I(inode)->io_tree, start, start + len,
2925                              &cached_state, GFP_NOFS);
2926         return ret;
2927 }
2928
2929 static inline struct page *extent_buffer_page(struct extent_buffer *eb,
2930                                               unsigned long i)
2931 {
2932         struct page *p;
2933         struct address_space *mapping;
2934
2935         if (i == 0)
2936                 return eb->first_page;
2937         i += eb->start >> PAGE_CACHE_SHIFT;
2938         mapping = eb->first_page->mapping;
2939         if (!mapping)
2940                 return NULL;
2941
2942         /*
2943          * extent_buffer_page is only called after pinning the page
2944          * by increasing the reference count.  So we know the page must
2945          * be in the radix tree.
2946          */
2947         rcu_read_lock();
2948         p = radix_tree_lookup(&mapping->page_tree, i);
2949         rcu_read_unlock();
2950
2951         return p;
2952 }
2953
2954 static inline unsigned long num_extent_pages(u64 start, u64 len)
2955 {
2956         return ((start + len + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT) -
2957                 (start >> PAGE_CACHE_SHIFT);
2958 }
2959
2960 static struct extent_buffer *__alloc_extent_buffer(struct extent_io_tree *tree,
2961                                                    u64 start,
2962                                                    unsigned long len,
2963                                                    gfp_t mask)
2964 {
2965         struct extent_buffer *eb = NULL;
2966 #if LEAK_DEBUG
2967         unsigned long flags;
2968 #endif
2969
2970         eb = kmem_cache_zalloc(extent_buffer_cache, mask);
2971         if (eb == NULL)
2972                 return NULL;
2973         eb->start = start;
2974         eb->len = len;
2975         rwlock_init(&eb->lock);
2976         atomic_set(&eb->write_locks, 0);
2977         atomic_set(&eb->read_locks, 0);
2978         atomic_set(&eb->blocking_readers, 0);
2979         atomic_set(&eb->blocking_writers, 0);
2980         atomic_set(&eb->spinning_readers, 0);
2981         atomic_set(&eb->spinning_writers, 0);
2982         init_waitqueue_head(&eb->write_lock_wq);
2983         init_waitqueue_head(&eb->read_lock_wq);
2984
2985 #if LEAK_DEBUG
2986         spin_lock_irqsave(&leak_lock, flags);
2987         list_add(&eb->leak_list, &buffers);
2988         spin_unlock_irqrestore(&leak_lock, flags);
2989 #endif
2990         atomic_set(&eb->refs, 1);
2991
2992         return eb;
2993 }
2994
2995 static void __free_extent_buffer(struct extent_buffer *eb)
2996 {
2997 #if LEAK_DEBUG
2998         unsigned long flags;
2999         spin_lock_irqsave(&leak_lock, flags);
3000         list_del(&eb->leak_list);
3001         spin_unlock_irqrestore(&leak_lock, flags);
3002 #endif
3003         kmem_cache_free(extent_buffer_cache, eb);
3004 }
3005
3006 /*
3007  * Helper for releasing extent buffer page.
3008  */
3009 static void btrfs_release_extent_buffer_page(struct extent_buffer *eb,
3010                                                 unsigned long start_idx)
3011 {
3012         unsigned long index;
3013         struct page *page;
3014
3015         if (!eb->first_page)
3016                 return;
3017
3018         index = num_extent_pages(eb->start, eb->len);
3019         if (start_idx >= index)
3020                 return;
3021
3022         do {
3023                 index--;
3024                 page = extent_buffer_page(eb, index);
3025                 if (page)
3026                         page_cache_release(page);
3027         } while (index != start_idx);
3028 }
3029
3030 /*
3031  * Helper for releasing the extent buffer.
3032  */
3033 static inline void btrfs_release_extent_buffer(struct extent_buffer *eb)
3034 {
3035         btrfs_release_extent_buffer_page(eb, 0);
3036         __free_extent_buffer(eb);
3037 }
3038
3039 struct extent_buffer *alloc_extent_buffer(struct extent_io_tree *tree,
3040                                           u64 start, unsigned long len,
3041                                           struct page *page0)
3042 {
3043         unsigned long num_pages = num_extent_pages(start, len);
3044         unsigned long i;
3045         unsigned long index = start >> PAGE_CACHE_SHIFT;
3046         struct extent_buffer *eb;
3047         struct extent_buffer *exists = NULL;
3048         struct page *p;
3049         struct address_space *mapping = tree->mapping;
3050         int uptodate = 1;
3051         int ret;
3052
3053         rcu_read_lock();
3054         eb = radix_tree_lookup(&tree->buffer, start >> PAGE_CACHE_SHIFT);
3055         if (eb && atomic_inc_not_zero(&eb->refs)) {
3056                 rcu_read_unlock();
3057                 mark_page_accessed(eb->first_page);
3058                 return eb;
3059         }
3060         rcu_read_unlock();
3061
3062         eb = __alloc_extent_buffer(tree, start, len, GFP_NOFS);
3063         if (!eb)
3064                 return NULL;
3065
3066         if (page0) {
3067                 eb->first_page = page0;
3068                 i = 1;
3069                 index++;
3070                 page_cache_get(page0);
3071                 mark_page_accessed(page0);
3072                 set_page_extent_mapped(page0);
3073                 set_page_extent_head(page0, len);
3074                 uptodate = PageUptodate(page0);
3075         } else {
3076                 i = 0;
3077         }
3078         for (; i < num_pages; i++, index++) {
3079                 p = find_or_create_page(mapping, index, GFP_NOFS);
3080                 if (!p) {
3081                         WARN_ON(1);
3082                         goto free_eb;
3083                 }
3084                 set_page_extent_mapped(p);
3085                 mark_page_accessed(p);
3086                 if (i == 0) {
3087                         eb->first_page = p;
3088                         set_page_extent_head(p, len);
3089                 } else {
3090                         set_page_private(p, EXTENT_PAGE_PRIVATE);
3091                 }
3092                 if (!PageUptodate(p))
3093                         uptodate = 0;
3094
3095                 /*
3096                  * see below about how we avoid a nasty race with release page
3097                  * and why we unlock later
3098                  */
3099                 if (i != 0)
3100                         unlock_page(p);
3101         }
3102         if (uptodate)
3103                 set_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags);
3104
3105         ret = radix_tree_preload(GFP_NOFS & ~__GFP_HIGHMEM);
3106         if (ret)
3107                 goto free_eb;
3108
3109         spin_lock(&tree->buffer_lock);
3110         ret = radix_tree_insert(&tree->buffer, start >> PAGE_CACHE_SHIFT, eb);
3111         if (ret == -EEXIST) {
3112                 exists = radix_tree_lookup(&tree->buffer,
3113                                                 start >> PAGE_CACHE_SHIFT);
3114                 /* add one reference for the caller */
3115                 atomic_inc(&exists->refs);
3116                 spin_unlock(&tree->buffer_lock);
3117                 radix_tree_preload_end();
3118                 goto free_eb;
3119         }
3120         /* add one reference for the tree */
3121         atomic_inc(&eb->refs);
3122         spin_unlock(&tree->buffer_lock);
3123         radix_tree_preload_end();
3124
3125         /*
3126          * there is a race where release page may have
3127          * tried to find this extent buffer in the radix
3128          * but failed.  It will tell the VM it is safe to
3129          * reclaim the, and it will clear the page private bit.
3130          * We must make sure to set the page private bit properly
3131          * after the extent buffer is in the radix tree so
3132          * it doesn't get lost
3133          */
3134         set_page_extent_mapped(eb->first_page);
3135         set_page_extent_head(eb->first_page, eb->len);
3136         if (!page0)
3137                 unlock_page(eb->first_page);
3138         return eb;
3139
3140 free_eb:
3141         if (eb->first_page && !page0)
3142                 unlock_page(eb->first_page);
3143
3144         if (!atomic_dec_and_test(&eb->refs))
3145                 return exists;
3146         btrfs_release_extent_buffer(eb);
3147         return exists;
3148 }
3149
3150 struct extent_buffer *find_extent_buffer(struct extent_io_tree *tree,
3151                                          u64 start, unsigned long len)
3152 {
3153         struct extent_buffer *eb;
3154
3155         rcu_read_lock();
3156         eb = radix_tree_lookup(&tree->buffer, start >> PAGE_CACHE_SHIFT);
3157         if (eb && atomic_inc_not_zero(&eb->refs)) {
3158                 rcu_read_unlock();
3159                 mark_page_accessed(eb->first_page);
3160                 return eb;
3161         }
3162         rcu_read_unlock();
3163
3164         return NULL;
3165 }
3166
3167 void free_extent_buffer(struct extent_buffer *eb)
3168 {
3169         if (!eb)
3170                 return;
3171
3172         if (!atomic_dec_and_test(&eb->refs))
3173                 return;
3174
3175         WARN_ON(1);
3176 }
3177
3178 int clear_extent_buffer_dirty(struct extent_io_tree *tree,
3179                               struct extent_buffer *eb)
3180 {
3181         unsigned long i;
3182         unsigned long num_pages;
3183         struct page *page;
3184
3185         num_pages = num_extent_pages(eb->start, eb->len);
3186
3187         for (i = 0; i < num_pages; i++) {
3188                 page = extent_buffer_page(eb, i);
3189                 if (!PageDirty(page))
3190                         continue;
3191
3192                 lock_page(page);
3193                 WARN_ON(!PagePrivate(page));
3194
3195                 set_page_extent_mapped(page);
3196                 if (i == 0)
3197                         set_page_extent_head(page, eb->len);
3198
3199                 clear_page_dirty_for_io(page);
3200                 spin_lock_irq(&page->mapping->tree_lock);
3201                 if (!PageDirty(page)) {
3202                         radix_tree_tag_clear(&page->mapping->page_tree,
3203                                                 page_index(page),
3204                                                 PAGECACHE_TAG_DIRTY);
3205                 }
3206                 spin_unlock_irq(&page->mapping->tree_lock);
3207                 unlock_page(page);
3208         }
3209         return 0;
3210 }
3211
3212 int set_extent_buffer_dirty(struct extent_io_tree *tree,
3213                              struct extent_buffer *eb)
3214 {
3215         unsigned long i;
3216         unsigned long num_pages;
3217         int was_dirty = 0;
3218
3219         was_dirty = test_and_set_bit(EXTENT_BUFFER_DIRTY, &eb->bflags);
3220         num_pages = num_extent_pages(eb->start, eb->len);
3221         for (i = 0; i < num_pages; i++)
3222                 __set_page_dirty_nobuffers(extent_buffer_page(eb, i));
3223         return was_dirty;
3224 }
3225
3226 static int __eb_straddles_pages(u64 start, u64 len)
3227 {
3228         if (len < PAGE_CACHE_SIZE)
3229                 return 1;
3230         if (start & (PAGE_CACHE_SIZE - 1))
3231                 return 1;
3232         if ((start + len) & (PAGE_CACHE_SIZE - 1))
3233                 return 1;
3234         return 0;
3235 }
3236
3237 static int eb_straddles_pages(struct extent_buffer *eb)
3238 {
3239         return __eb_straddles_pages(eb->start, eb->len);
3240 }
3241
3242 int clear_extent_buffer_uptodate(struct extent_io_tree *tree,
3243                                 struct extent_buffer *eb,
3244                                 struct extent_state **cached_state)
3245 {
3246         unsigned long i;
3247         struct page *page;
3248         unsigned long num_pages;
3249
3250         num_pages = num_extent_pages(eb->start, eb->len);
3251         clear_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags);
3252
3253         if (eb_straddles_pages(eb)) {
3254                 clear_extent_uptodate(tree, eb->start, eb->start + eb->len - 1,
3255                                       cached_state, GFP_NOFS);
3256         }
3257         for (i = 0; i < num_pages; i++) {
3258                 page = extent_buffer_page(eb, i);
3259                 if (page)
3260                         ClearPageUptodate(page);
3261         }
3262         return 0;
3263 }
3264
3265 int set_extent_buffer_uptodate(struct extent_io_tree *tree,
3266                                 struct extent_buffer *eb)
3267 {
3268         unsigned long i;
3269         struct page *page;
3270         unsigned long num_pages;
3271
3272         num_pages = num_extent_pages(eb->start, eb->len);
3273
3274         if (eb_straddles_pages(eb)) {
3275                 set_extent_uptodate(tree, eb->start, eb->start + eb->len - 1,
3276                                     NULL, GFP_NOFS);
3277         }
3278         for (i = 0; i < num_pages; i++) {
3279                 page = extent_buffer_page(eb, i);
3280                 if ((i == 0 && (eb->start & (PAGE_CACHE_SIZE - 1))) ||
3281                     ((i == num_pages - 1) &&
3282                      ((eb->start + eb->len) & (PAGE_CACHE_SIZE - 1)))) {
3283                         check_page_uptodate(tree, page);
3284                         continue;
3285                 }
3286                 SetPageUptodate(page);
3287         }
3288         return 0;
3289 }
3290
3291 int extent_range_uptodate(struct extent_io_tree *tree,
3292                           u64 start, u64 end)
3293 {
3294         struct page *page;
3295         int ret;
3296         int pg_uptodate = 1;
3297         int uptodate;
3298         unsigned long index;
3299
3300         if (__eb_straddles_pages(start, end - start + 1)) {
3301                 ret = test_range_bit(tree, start, end,
3302                                      EXTENT_UPTODATE, 1, NULL);
3303                 if (ret)
3304                         return 1;
3305         }
3306         while (start <= end) {
3307                 index = start >> PAGE_CACHE_SHIFT;
3308                 page = find_get_page(tree->mapping, index);
3309                 uptodate = PageUptodate(page);
3310                 page_cache_release(page);
3311                 if (!uptodate) {
3312                         pg_uptodate = 0;
3313                         break;
3314                 }
3315                 start += PAGE_CACHE_SIZE;
3316         }
3317         return pg_uptodate;
3318 }
3319
3320 int extent_buffer_uptodate(struct extent_io_tree *tree,
3321                            struct extent_buffer *eb,
3322                            struct extent_state *cached_state)
3323 {
3324         int ret = 0;
3325         unsigned long num_pages;
3326         unsigned long i;
3327         struct page *page;
3328         int pg_uptodate = 1;
3329
3330         if (test_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags))
3331                 return 1;
3332
3333         if (eb_straddles_pages(eb)) {
3334                 ret = test_range_bit(tree, eb->start, eb->start + eb->len - 1,
3335                                    EXTENT_UPTODATE, 1, cached_state);
3336                 if (ret)
3337                         return ret;
3338         }
3339
3340         num_pages = num_extent_pages(eb->start, eb->len);
3341         for (i = 0; i < num_pages; i++) {
3342                 page = extent_buffer_page(eb, i);
3343                 if (!PageUptodate(page)) {
3344                         pg_uptodate = 0;
3345                         break;
3346                 }
3347         }
3348         return pg_uptodate;
3349 }
3350
3351 int read_extent_buffer_pages(struct extent_io_tree *tree,
3352                              struct extent_buffer *eb, u64 start, int wait,
3353                              get_extent_t *get_extent, int mirror_num)
3354 {
3355         unsigned long i;
3356         unsigned long start_i;
3357         struct page *page;
3358         int err;
3359         int ret = 0;
3360         int locked_pages = 0;
3361         int all_uptodate = 1;
3362         int inc_all_pages = 0;
3363         unsigned long num_pages;
3364         struct bio *bio = NULL;
3365         unsigned long bio_flags = 0;
3366
3367         if (test_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags))
3368                 return 0;
3369
3370         if (eb_straddles_pages(eb)) {
3371                 if (test_range_bit(tree, eb->start, eb->start + eb->len - 1,
3372                                    EXTENT_UPTODATE, 1, NULL)) {
3373                         return 0;
3374                 }
3375         }
3376
3377         if (start) {
3378                 WARN_ON(start < eb->start);
3379                 start_i = (start >> PAGE_CACHE_SHIFT) -
3380                         (eb->start >> PAGE_CACHE_SHIFT);
3381         } else {
3382                 start_i = 0;
3383         }
3384
3385         num_pages = num_extent_pages(eb->start, eb->len);
3386         for (i = start_i; i < num_pages; i++) {
3387                 page = extent_buffer_page(eb, i);
3388                 if (wait == WAIT_NONE) {
3389                         if (!trylock_page(page))
3390                                 goto unlock_exit;
3391                 } else {
3392                         lock_page(page);
3393                 }
3394                 locked_pages++;
3395                 if (!PageUptodate(page))
3396                         all_uptodate = 0;
3397         }
3398         if (all_uptodate) {
3399                 if (start_i == 0)
3400                         set_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags);
3401                 goto unlock_exit;
3402         }
3403
3404         for (i = start_i; i < num_pages; i++) {
3405                 page = extent_buffer_page(eb, i);
3406
3407                 WARN_ON(!PagePrivate(page));
3408
3409                 set_page_extent_mapped(page);
3410                 if (i == 0)
3411                         set_page_extent_head(page, eb->len);
3412
3413                 if (inc_all_pages)
3414                         page_cache_get(page);
3415                 if (!PageUptodate(page)) {
3416                         if (start_i == 0)
3417                                 inc_all_pages = 1;
3418                         ClearPageError(page);
3419                         err = __extent_read_full_page(tree, page,
3420                                                       get_extent, &bio,
3421                                                       mirror_num, &bio_flags);
3422                         if (err)
3423                                 ret = err;
3424                 } else {
3425                         unlock_page(page);
3426                 }
3427         }
3428
3429         if (bio)
3430                 submit_one_bio(READ, bio, mirror_num, bio_flags);
3431
3432         if (ret || wait != WAIT_COMPLETE)
3433                 return ret;
3434
3435         for (i = start_i; i < num_pages; i++) {
3436                 page = extent_buffer_page(eb, i);
3437                 wait_on_page_locked(page);
3438                 if (!PageUptodate(page))
3439                         ret = -EIO;
3440         }
3441
3442         if (!ret)
3443                 set_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags);
3444         return ret;
3445
3446 unlock_exit:
3447         i = start_i;
3448         while (locked_pages > 0) {
3449                 page = extent_buffer_page(eb, i);
3450                 i++;
3451                 unlock_page(page);
3452                 locked_pages--;
3453         }
3454         return ret;
3455 }
3456
3457 void read_extent_buffer(struct extent_buffer *eb, void *dstv,
3458                         unsigned long start,
3459                         unsigned long len)
3460 {
3461         size_t cur;
3462         size_t offset;
3463         struct page *page;
3464         char *kaddr;
3465         char *dst = (char *)dstv;
3466         size_t start_offset = eb->start & ((u64)PAGE_CACHE_SIZE - 1);
3467         unsigned long i = (start_offset + start) >> PAGE_CACHE_SHIFT;
3468
3469         WARN_ON(start > eb->len);
3470         WARN_ON(start + len > eb->start + eb->len);
3471
3472         offset = (start_offset + start) & ((unsigned long)PAGE_CACHE_SIZE - 1);
3473
3474         while (len > 0) {
3475                 page = extent_buffer_page(eb, i);
3476
3477                 cur = min(len, (PAGE_CACHE_SIZE - offset));
3478                 kaddr = page_address(page);
3479                 memcpy(dst, kaddr + offset, cur);
3480
3481                 dst += cur;
3482                 len -= cur;
3483                 offset = 0;
3484                 i++;
3485         }
3486 }
3487
3488 int map_private_extent_buffer(struct extent_buffer *eb, unsigned long start,
3489                                unsigned long min_len, char **map,
3490                                unsigned long *map_start,
3491                                unsigned long *map_len)
3492 {
3493         size_t offset = start & (PAGE_CACHE_SIZE - 1);
3494         char *kaddr;
3495         struct page *p;
3496         size_t start_offset = eb->start & ((u64)PAGE_CACHE_SIZE - 1);
3497         unsigned long i = (start_offset + start) >> PAGE_CACHE_SHIFT;
3498         unsigned long end_i = (start_offset + start + min_len - 1) >>
3499                 PAGE_CACHE_SHIFT;
3500
3501         if (i != end_i)
3502                 return -EINVAL;
3503
3504         if (i == 0) {
3505                 offset = start_offset;
3506                 *map_start = 0;
3507         } else {
3508                 offset = 0;
3509                 *map_start = ((u64)i << PAGE_CACHE_SHIFT) - start_offset;
3510         }
3511
3512         if (start + min_len > eb->len) {
3513                 printk(KERN_ERR "btrfs bad mapping eb start %llu len %lu, "
3514                        "wanted %lu %lu\n", (unsigned long long)eb->start,
3515                        eb->len, start, min_len);
3516                 WARN_ON(1);
3517                 return -EINVAL;
3518         }
3519
3520         p = extent_buffer_page(eb, i);
3521         kaddr = page_address(p);
3522         *map = kaddr + offset;
3523         *map_len = PAGE_CACHE_SIZE - offset;
3524         return 0;
3525 }
3526
3527 int memcmp_extent_buffer(struct extent_buffer *eb, const void *ptrv,
3528                           unsigned long start,
3529                           unsigned long len)
3530 {
3531         size_t cur;
3532         size_t offset;
3533         struct page *page;
3534         char *kaddr;
3535         char *ptr = (char *)ptrv;
3536         size_t start_offset = eb->start & ((u64)PAGE_CACHE_SIZE - 1);
3537         unsigned long i = (start_offset + start) >> PAGE_CACHE_SHIFT;
3538         int ret = 0;
3539
3540         WARN_ON(start > eb->len);
3541         WARN_ON(start + len > eb->start + eb->len);
3542
3543         offset = (start_offset + start) & ((unsigned long)PAGE_CACHE_SIZE - 1);
3544
3545         while (len > 0) {
3546                 page = extent_buffer_page(eb, i);
3547
3548                 cur = min(len, (PAGE_CACHE_SIZE - offset));
3549
3550                 kaddr = page_address(page);
3551                 ret = memcmp(ptr, kaddr + offset, cur);
3552                 if (ret)
3553                         break;
3554
3555                 ptr += cur;
3556                 len -= cur;
3557                 offset = 0;
3558                 i++;
3559         }
3560         return ret;
3561 }
3562
3563 void write_extent_buffer(struct extent_buffer *eb, const void *srcv,
3564                          unsigned long start, unsigned long len)
3565 {
3566         size_t cur;
3567         size_t offset;
3568         struct page *page;
3569         char *kaddr;
3570         char *src = (char *)srcv;
3571         size_t start_offset = eb->start & ((u64)PAGE_CACHE_SIZE - 1);
3572         unsigned long i = (start_offset + start) >> PAGE_CACHE_SHIFT;
3573
3574         WARN_ON(start > eb->len);
3575         WARN_ON(start + len > eb->start + eb->len);
3576
3577         offset = (start_offset + start) & ((unsigned long)PAGE_CACHE_SIZE - 1);
3578
3579         while (len > 0) {
3580                 page = extent_buffer_page(eb, i);
3581                 WARN_ON(!PageUptodate(page));
3582
3583                 cur = min(len, PAGE_CACHE_SIZE - offset);
3584                 kaddr = page_address(page);
3585                 memcpy(kaddr + offset, src, cur);
3586
3587                 src += cur;
3588                 len -= cur;
3589                 offset = 0;
3590                 i++;
3591         }
3592 }
3593
3594 void memset_extent_buffer(struct extent_buffer *eb, char c,
3595                           unsigned long start, unsigned long len)
3596 {
3597         size_t cur;
3598         size_t offset;
3599         struct page *page;
3600         char *kaddr;
3601         size_t start_offset = eb->start & ((u64)PAGE_CACHE_SIZE - 1);
3602         unsigned long i = (start_offset + start) >> PAGE_CACHE_SHIFT;
3603
3604         WARN_ON(start > eb->len);
3605         WARN_ON(start + len > eb->start + eb->len);
3606
3607         offset = (start_offset + start) & ((unsigned long)PAGE_CACHE_SIZE - 1);
3608
3609         while (len > 0) {
3610                 page = extent_buffer_page(eb, i);
3611                 WARN_ON(!PageUptodate(page));
3612
3613                 cur = min(len, PAGE_CACHE_SIZE - offset);
3614                 kaddr = page_address(page);
3615                 memset(kaddr + offset, c, cur);
3616
3617                 len -= cur;
3618                 offset = 0;
3619                 i++;
3620         }
3621 }
3622
3623 void copy_extent_buffer(struct extent_buffer *dst, struct extent_buffer *src,
3624                         unsigned long dst_offset, unsigned long src_offset,
3625                         unsigned long len)
3626 {
3627         u64 dst_len = dst->len;
3628         size_t cur;
3629         size_t offset;
3630         struct page *page;
3631         char *kaddr;
3632         size_t start_offset = dst->start & ((u64)PAGE_CACHE_SIZE - 1);
3633         unsigned long i = (start_offset + dst_offset) >> PAGE_CACHE_SHIFT;
3634
3635         WARN_ON(src->len != dst_len);
3636
3637         offset = (start_offset + dst_offset) &
3638                 ((unsigned long)PAGE_CACHE_SIZE - 1);
3639
3640         while (len > 0) {
3641                 page = extent_buffer_page(dst, i);
3642                 WARN_ON(!PageUptodate(page));
3643
3644                 cur = min(len, (unsigned long)(PAGE_CACHE_SIZE - offset));
3645
3646                 kaddr = page_address(page);
3647                 read_extent_buffer(src, kaddr + offset, src_offset, cur);
3648
3649                 src_offset += cur;
3650                 len -= cur;
3651                 offset = 0;
3652                 i++;
3653         }
3654 }
3655
3656 static void move_pages(struct page *dst_page, struct page *src_page,
3657                        unsigned long dst_off, unsigned long src_off,
3658                        unsigned long len)
3659 {
3660         char *dst_kaddr = page_address(dst_page);
3661         if (dst_page == src_page) {
3662                 memmove(dst_kaddr + dst_off, dst_kaddr + src_off, len);
3663         } else {
3664                 char *src_kaddr = page_address(src_page);
3665                 char *p = dst_kaddr + dst_off + len;
3666                 char *s = src_kaddr + src_off + len;
3667
3668                 while (len--)
3669                         *--p = *--s;
3670         }
3671 }
3672
3673 static inline bool areas_overlap(unsigned long src, unsigned long dst, unsigned long len)
3674 {
3675         unsigned long distance = (src > dst) ? src - dst : dst - src;
3676         return distance < len;
3677 }
3678
3679 static void copy_pages(struct page *dst_page, struct page *src_page,
3680                        unsigned long dst_off, unsigned long src_off,
3681                        unsigned long len)
3682 {
3683         char *dst_kaddr = page_address(dst_page);
3684         char *src_kaddr;
3685
3686         if (dst_page != src_page) {
3687                 src_kaddr = page_address(src_page);
3688         } else {
3689                 src_kaddr = dst_kaddr;
3690                 BUG_ON(areas_overlap(src_off, dst_off, len));
3691         }
3692
3693         memcpy(dst_kaddr + dst_off, src_kaddr + src_off, len);
3694 }
3695
3696 void memcpy_extent_buffer(struct extent_buffer *dst, unsigned long dst_offset,
3697                            unsigned long src_offset, unsigned long len)
3698 {
3699         size_t cur;
3700         size_t dst_off_in_page;
3701         size_t src_off_in_page;
3702         size_t start_offset = dst->start & ((u64)PAGE_CACHE_SIZE - 1);
3703         unsigned long dst_i;
3704         unsigned long src_i;
3705
3706         if (src_offset + len > dst->len) {
3707                 printk(KERN_ERR "btrfs memmove bogus src_offset %lu move "
3708                        "len %lu dst len %lu\n", src_offset, len, dst->len);
3709                 BUG_ON(1);
3710         }
3711         if (dst_offset + len > dst->len) {
3712                 printk(KERN_ERR "btrfs memmove bogus dst_offset %lu move "
3713                        "len %lu dst len %lu\n", dst_offset, len, dst->len);
3714                 BUG_ON(1);
3715         }
3716
3717         while (len > 0) {
3718                 dst_off_in_page = (start_offset + dst_offset) &
3719                         ((unsigned long)PAGE_CACHE_SIZE - 1);
3720                 src_off_in_page = (start_offset + src_offset) &
3721                         ((unsigned long)PAGE_CACHE_SIZE - 1);
3722
3723                 dst_i = (start_offset + dst_offset) >> PAGE_CACHE_SHIFT;
3724                 src_i = (start_offset + src_offset) >> PAGE_CACHE_SHIFT;
3725
3726                 cur = min(len, (unsigned long)(PAGE_CACHE_SIZE -
3727                                                src_off_in_page));
3728                 cur = min_t(unsigned long, cur,
3729                         (unsigned long)(PAGE_CACHE_SIZE - dst_off_in_page));
3730
3731                 copy_pages(extent_buffer_page(dst, dst_i),
3732                            extent_buffer_page(dst, src_i),
3733                            dst_off_in_page, src_off_in_page, cur);
3734
3735                 src_offset += cur;
3736                 dst_offset += cur;
3737                 len -= cur;
3738         }
3739 }
3740
3741 void memmove_extent_buffer(struct extent_buffer *dst, unsigned long dst_offset,
3742                            unsigned long src_offset, unsigned long len)
3743 {
3744         size_t cur;
3745         size_t dst_off_in_page;
3746         size_t src_off_in_page;
3747         unsigned long dst_end = dst_offset + len - 1;
3748         unsigned long src_end = src_offset + len - 1;
3749         size_t start_offset = dst->start & ((u64)PAGE_CACHE_SIZE - 1);
3750         unsigned long dst_i;
3751         unsigned long src_i;
3752
3753         if (src_offset + len > dst->len) {
3754                 printk(KERN_ERR "btrfs memmove bogus src_offset %lu move "
3755                        "len %lu len %lu\n", src_offset, len, dst->len);
3756                 BUG_ON(1);
3757         }
3758         if (dst_offset + len > dst->len) {
3759                 printk(KERN_ERR "btrfs memmove bogus dst_offset %lu move "
3760                        "len %lu len %lu\n", dst_offset, len, dst->len);
3761                 BUG_ON(1);
3762         }
3763         if (!areas_overlap(src_offset, dst_offset, len)) {
3764                 memcpy_extent_buffer(dst, dst_offset, src_offset, len);
3765                 return;
3766         }
3767         while (len > 0) {
3768                 dst_i = (start_offset + dst_end) >> PAGE_CACHE_SHIFT;
3769                 src_i = (start_offset + src_end) >> PAGE_CACHE_SHIFT;
3770
3771                 dst_off_in_page = (start_offset + dst_end) &
3772                         ((unsigned long)PAGE_CACHE_SIZE - 1);
3773                 src_off_in_page = (start_offset + src_end) &
3774                         ((unsigned long)PAGE_CACHE_SIZE - 1);
3775
3776                 cur = min_t(unsigned long, len, src_off_in_page + 1);
3777                 cur = min(cur, dst_off_in_page + 1);
3778                 move_pages(extent_buffer_page(dst, dst_i),
3779                            extent_buffer_page(dst, src_i),
3780                            dst_off_in_page - cur + 1,
3781                            src_off_in_page - cur + 1, cur);
3782
3783                 dst_end -= cur;
3784                 src_end -= cur;
3785                 len -= cur;
3786         }
3787 }
3788
3789 static inline void btrfs_release_extent_buffer_rcu(struct rcu_head *head)
3790 {
3791         struct extent_buffer *eb =
3792                         container_of(head, struct extent_buffer, rcu_head);
3793
3794         btrfs_release_extent_buffer(eb);
3795 }
3796
3797 int try_release_extent_buffer(struct extent_io_tree *tree, struct page *page)
3798 {
3799         u64 start = page_offset(page);
3800         struct extent_buffer *eb;
3801         int ret = 1;
3802
3803         spin_lock(&tree->buffer_lock);
3804         eb = radix_tree_lookup(&tree->buffer, start >> PAGE_CACHE_SHIFT);
3805         if (!eb) {
3806                 spin_unlock(&tree->buffer_lock);
3807                 return ret;
3808         }
3809
3810         if (test_bit(EXTENT_BUFFER_DIRTY, &eb->bflags)) {
3811                 ret = 0;
3812                 goto out;
3813         }
3814
3815         /*
3816          * set @eb->refs to 0 if it is already 1, and then release the @eb.
3817          * Or go back.
3818          */
3819         if (atomic_cmpxchg(&eb->refs, 1, 0) != 1) {
3820                 ret = 0;
3821                 goto out;
3822         }
3823
3824         radix_tree_delete(&tree->buffer, start >> PAGE_CACHE_SHIFT);
3825 out:
3826         spin_unlock(&tree->buffer_lock);
3827
3828         /* at this point we can safely release the extent buffer */
3829         if (atomic_read(&eb->refs) == 0)
3830                 call_rcu(&eb->rcu_head, btrfs_release_extent_buffer_rcu);
3831         return ret;
3832 }