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 "extent_io.h"
  14 #include "extent_map.h"
  15 #include "compat.h"
  16 #include "ctree.h"
  17 #include "btrfs_inode.h"
  18
  19 static struct kmem_cache *extent_state_cache;
  20 static struct kmem_cache *extent_buffer_cache;
  21
  22 static LIST_HEAD(buffers);
  23 static LIST_HEAD(states);
  24
  25 #define LEAK_DEBUG 0
  26 #if LEAK_DEBUG
  27 static DEFINE_SPINLOCK(leak_lock);
  28 #endif
  29
  30 #define BUFFER_LRU_MAX 64
  31
  32 struct tree_entry {
  33         u64 start;
  34         u64 end;
  35         struct rb_node rb_node;
  36 };
  37
  38 struct extent_page_data {
  39         struct bio *bio;
  40         struct extent_io_tree *tree;
  41         get_extent_t *get_extent;
  42
  43         /* tells writepage not to lock the state bits for this range
  44          * it still does the unlocking
  45          */
  46         unsigned int extent_locked:1;
  47
  48         /* tells the submit_bio code to use a WRITE_SYNC */
  49         unsigned int sync_io:1;
  50 };
  51
  52 int __init extent_io_init(void)
  53 {
  54         extent_state_cache = kmem_cache_create("extent_state",
  55                         sizeof(struct extent_state), 0,
  56                         SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD, NULL);
  57         if (!extent_state_cache)
  58                 return -ENOMEM;
  59
  60         extent_buffer_cache = kmem_cache_create("extent_buffers",
  61                         sizeof(struct extent_buffer), 0,
  62                         SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD, NULL);
  63         if (!extent_buffer_cache)
  64                 goto free_state_cache;
  65         return 0;
  66
  67 free_state_cache:
  68         kmem_cache_destroy(extent_state_cache);
  69         return -ENOMEM;
  70 }
  71
  72 void extent_io_exit(void)
  73 {
  74         struct extent_state *state;
  75         struct extent_buffer *eb;
  76
  77         while (!list_empty(&states)) {
  78                 state = list_entry(states.next, struct extent_state, leak_list);
  79                 printk(KERN_ERR "btrfs state leak: start %llu end %llu "
  80                        "state %lu in tree %p refs %d\n",
  81                        (unsigned long long)state->start,
  82                        (unsigned long long)state->end,
  83                        state->state, state->tree, atomic_read(&state->refs));
  84                 list_del(&state->leak_list);
  85                 kmem_cache_free(extent_state_cache, state);
  86
  87         }
  88
  89         while (!list_empty(&buffers)) {
  90                 eb = list_entry(buffers.next, struct extent_buffer, leak_list);
  91                 printk(KERN_ERR "btrfs buffer leak start %llu len %lu "
  92                        "refs %d\n", (unsigned long long)eb->start,
  93                        eb->len, atomic_read(&eb->refs));
  94                 list_del(&eb->leak_list);
  95                 kmem_cache_free(extent_buffer_cache, eb);
  96         }
  97         if (extent_state_cache)
  98                 kmem_cache_destroy(extent_state_cache);
  99         if (extent_buffer_cache)
 100                 kmem_cache_destroy(extent_buffer_cache);
 101 }
 102
 103 void extent_io_tree_init(struct extent_io_tree *tree,
 104                           struct address_space *mapping, gfp_t mask)
 105 {
 106         tree->state = RB_ROOT;
 107         tree->buffer = RB_ROOT;
 108         tree->ops = NULL;
 109         tree->dirty_bytes = 0;
 110         spin_lock_init(&tree->lock);
 111         spin_lock_init(&tree->buffer_lock);
 112         tree->mapping = mapping;
 113 }
 114
 115 static struct extent_state *alloc_extent_state(gfp_t mask)
 116 {
 117         struct extent_state *state;
 118 #if LEAK_DEBUG
 119         unsigned long flags;
 120 #endif
 121
 122         state = kmem_cache_alloc(extent_state_cache, mask);
 123         if (!state)
 124                 return state;
 125         state->state = 0;
 126         state->private = 0;
 127         state->tree = NULL;
 128 #if LEAK_DEBUG
 129         spin_lock_irqsave(&leak_lock, flags);
 130         list_add(&state->leak_list, &states);
 131         spin_unlock_irqrestore(&leak_lock, flags);
 132 #endif
 133         atomic_set(&state->refs, 1);
 134         init_waitqueue_head(&state->wq);
 135         return state;
 136 }
 137
 138 void free_extent_state(struct extent_state *state)
 139 {
 140         if (!state)
 141                 return;
 142         if (atomic_dec_and_test(&state->refs)) {
 143 #if LEAK_DEBUG
 144                 unsigned long flags;
 145 #endif
 146                 WARN_ON(state->tree);
 147 #if LEAK_DEBUG
 148                 spin_lock_irqsave(&leak_lock, flags);
 149                 list_del(&state->leak_list);
 150                 spin_unlock_irqrestore(&leak_lock, flags);
 151 #endif
 152                 kmem_cache_free(extent_state_cache, state);
 153         }
 154 }
 155
 156 static struct rb_node *tree_insert(struct rb_root *root, u64 offset,
 157                                    struct rb_node *node)
 158 {
 159         struct rb_node **p = &root->rb_node;
 160         struct rb_node *parent = NULL;
 161         struct tree_entry *entry;
 162
 163         while (*p) {
 164                 parent = *p;
 165                 entry = rb_entry(parent, struct tree_entry, rb_node);
 166
 167                 if (offset < entry->start)
 168                         p = &(*p)->rb_left;
 169                 else if (offset > entry->end)
 170                         p = &(*p)->rb_right;
 171                 else
 172                         return parent;
 173         }
 174
 175         entry = rb_entry(node, struct tree_entry, rb_node);
 176         rb_link_node(node, parent, p);
 177         rb_insert_color(node, root);
 178         return NULL;
 179 }
 180
 181 static struct rb_node *__etree_search(struct extent_io_tree *tree, u64 offset,
 182                                      struct rb_node **prev_ret,
 183                                      struct rb_node **next_ret)
 184 {
 185         struct rb_root *root = &tree->state;
 186         struct rb_node *n = root->rb_node;
 187         struct rb_node *prev = NULL;
 188         struct rb_node *orig_prev = NULL;
 189         struct tree_entry *entry;
 190         struct tree_entry *prev_entry = NULL;
 191
 192         while (n) {
 193                 entry = rb_entry(n, struct tree_entry, rb_node);
 194                 prev = n;
 195                 prev_entry = entry;
 196
 197                 if (offset < entry->start)
 198                         n = n->rb_left;
 199                 else if (offset > entry->end)
 200                         n = n->rb_right;
 201                 else
 202                         return n;
 203         }
 204
 205         if (prev_ret) {
 206                 orig_prev = prev;
 207                 while (prev && offset > prev_entry->end) {
 208                         prev = rb_next(prev);
 209                         prev_entry = rb_entry(prev, struct tree_entry, rb_node);
 210                 }
 211                 *prev_ret = prev;
 212                 prev = orig_prev;
 213         }
 214
 215         if (next_ret) {
 216                 prev_entry = rb_entry(prev, struct tree_entry, rb_node);
 217                 while (prev && offset < prev_entry->start) {
 218                         prev = rb_prev(prev);
 219                         prev_entry = rb_entry(prev, struct tree_entry, rb_node);
 220                 }
 221                 *next_ret = prev;
 222         }
 223         return NULL;
 224 }
 225
 226 static inline struct rb_node *tree_search(struct extent_io_tree *tree,
 227                                           u64 offset)
 228 {
 229         struct rb_node *prev = NULL;
 230         struct rb_node *ret;
 231
 232         ret = __etree_search(tree, offset, &prev, NULL);
 233         if (!ret)
 234                 return prev;
 235         return ret;
 236 }
 237
 238 static struct extent_buffer *buffer_tree_insert(struct extent_io_tree *tree,
 239                                           u64 offset, struct rb_node *node)
 240 {
 241         struct rb_root *root = &tree->buffer;
 242         struct rb_node **p = &root->rb_node;
 243         struct rb_node *parent = NULL;
 244         struct extent_buffer *eb;
 245
 246         while (*p) {
 247                 parent = *p;
 248                 eb = rb_entry(parent, struct extent_buffer, rb_node);
 249
 250                 if (offset < eb->start)
 251                         p = &(*p)->rb_left;
 252                 else if (offset > eb->start)
 253                         p = &(*p)->rb_right;
 254                 else
 255                         return eb;
 256         }
 257
 258         rb_link_node(node, parent, p);
 259         rb_insert_color(node, root);
 260         return NULL;
 261 }
 262
 263 static struct extent_buffer *buffer_search(struct extent_io_tree *tree,
 264                                            u64 offset)
 265 {
 266         struct rb_root *root = &tree->buffer;
 267         struct rb_node *n = root->rb_node;
 268         struct extent_buffer *eb;
 269
 270         while (n) {
 271                 eb = rb_entry(n, struct extent_buffer, rb_node);
 272                 if (offset < eb->start)
 273                         n = n->rb_left;
 274                 else if (offset > eb->start)
 275                         n = n->rb_right;
 276                 else
 277                         return eb;
 278         }
 279         return NULL;
 280 }
 281
 282 static void merge_cb(struct extent_io_tree *tree, struct extent_state *new,
 283                      struct extent_state *other)
 284 {
 285         if (tree->ops && tree->ops->merge_extent_hook)
 286                 tree->ops->merge_extent_hook(tree->mapping->host, new,
 287                                              other);
 288 }
 289
 290 /*
 291  * utility function to look for merge candidates inside a given range.
 292  * Any extents with matching state are merged together into a single
 293  * extent in the tree.  Extents with EXTENT_IO in their state field
 294  * are not merged because the end_io handlers need to be able to do
 295  * operations on them without sleeping (or doing allocations/splits).
 296  *
 297  * This should be called with the tree lock held.
 298  */
 299 static int merge_state(struct extent_io_tree *tree,
 300                        struct extent_state *state)
 301 {
 302         struct extent_state *other;
 303         struct rb_node *other_node;
 304
 305         if (state->state & (EXTENT_IOBITS | EXTENT_BOUNDARY))
 306                 return 0;
 307
 308         other_node = rb_prev(&state->rb_node);
 309         if (other_node) {
 310                 other = rb_entry(other_node, struct extent_state, rb_node);
 311                 if (other->end == state->start - 1 &&
 312                     other->state == state->state) {
 313                         merge_cb(tree, state, other);
 314                         state->start = other->start;
 315                         other->tree = NULL;
 316                         rb_erase(&other->rb_node, &tree->state);
 317                         free_extent_state(other);
 318                 }
 319         }
 320         other_node = rb_next(&state->rb_node);
 321         if (other_node) {
 322                 other = rb_entry(other_node, struct extent_state, rb_node);
 323                 if (other->start == state->end + 1 &&
 324                     other->state == state->state) {
 325                         merge_cb(tree, state, other);
 326                         other->start = state->start;
 327                         state->tree = NULL;
 328                         rb_erase(&state->rb_node, &tree->state);
 329                         free_extent_state(state);
 330                         state = NULL;
 331                 }
 332         }
 333
 334         return 0;
 335 }
 336
 337 static int set_state_cb(struct extent_io_tree *tree,
 338                          struct extent_state *state, int *bits)
 339 {
 340         if (tree->ops && tree->ops->set_bit_hook) {
 341                 return tree->ops->set_bit_hook(tree->mapping->host,
 342                                                state, bits);
 343         }
 344
 345         return 0;
 346 }
 347
 348 static void clear_state_cb(struct extent_io_tree *tree,
 349                            struct extent_state *state, int *bits)
 350 {
 351         if (tree->ops && tree->ops->clear_bit_hook)
 352                 tree->ops->clear_bit_hook(tree->mapping->host, state, bits);
 353 }
 354
 355 /*
 356  * insert an extent_state struct into the tree.  'bits' are set on the
 357  * struct before it is inserted.
 358  *
 359  * This may return -EEXIST if the extent is already there, in which case the
 360  * state struct is freed.
 361  *
 362  * The tree lock is not taken internally.  This is a utility function and
 363  * probably isn't what you want to call (see set/clear_extent_bit).
 364  */
 365 static int insert_state(struct extent_io_tree *tree,
 366                         struct extent_state *state, u64 start, u64 end,
 367                         int *bits)
 368 {
 369         struct rb_node *node;
 370         int bits_to_set = *bits & ~EXTENT_CTLBITS;
 371         int ret;
 372
 373         if (end < start) {
 374                 printk(KERN_ERR "btrfs end < start %llu %llu\n",
 375                        (unsigned long long)end,
 376                        (unsigned long long)start);
 377                 WARN_ON(1);
 378         }
 379         state->start = start;
 380         state->end = end;
 381         ret = set_state_cb(tree, state, bits);
 382         if (ret)
 383                 return ret;
 384
 385         if (bits_to_set & EXTENT_DIRTY)
 386                 tree->dirty_bytes += end - start + 1;
 387         state->state |= bits_to_set;
 388         node = tree_insert(&tree->state, end, &state->rb_node);
 389         if (node) {
 390                 struct extent_state *found;
 391                 found = rb_entry(node, struct extent_state, rb_node);
 392                 printk(KERN_ERR "btrfs found node %llu %llu on insert of "
 393                        "%llu %llu\n", (unsigned long long)found->start,
 394                        (unsigned long long)found->end,
 395                        (unsigned long long)start, (unsigned long long)end);
 396                 free_extent_state(state);
 397                 return -EEXIST;
 398         }
 399         state->tree = tree;
 400         merge_state(tree, state);
 401         return 0;
 402 }
 403
 404 static int split_cb(struct extent_io_tree *tree, struct extent_state *orig,
 405                      u64 split)
 406 {
 407         if (tree->ops && tree->ops->split_extent_hook)
 408                 return tree->ops->split_extent_hook(tree->mapping->host,
 409                                                     orig, split);
 410         return 0;
 411 }
 412
 413 /*
 414  * split a given extent state struct in two, inserting the preallocated
 415  * struct 'prealloc' as the newly created second half.  'split' indicates an
 416  * offset inside 'orig' where it should be split.
 417  *
 418  * Before calling,
 419  * the tree has 'orig' at [orig->start, orig->end].  After calling, there
 420  * are two extent state structs in the tree:
 421  * prealloc: [orig->start, split - 1]
 422  * orig: [ split, orig->end ]
 423  *
 424  * The tree locks are not taken by this function. They need to be held
 425  * by the caller.
 426  */
 427 static int split_state(struct extent_io_tree *tree, struct extent_state *orig,
 428                        struct extent_state *prealloc, u64 split)
 429 {
 430         struct rb_node *node;
 431
 432         split_cb(tree, orig, split);
 433
 434         prealloc->start = orig->start;
 435         prealloc->end = split - 1;
 436         prealloc->state = orig->state;
 437         orig->start = split;
 438
 439         node = tree_insert(&tree->state, prealloc->end, &prealloc->rb_node);
 440         if (node) {
 441                 free_extent_state(prealloc);
 442                 return -EEXIST;
 443         }
 444         prealloc->tree = tree;
 445         return 0;
 446 }
 447
 448 /*
 449  * utility function to clear some bits in an extent state struct.
 450  * it will optionally wake up any one waiting on this state (wake == 1), or
 451  * forcibly remove the state from the tree (delete == 1).
 452  *
 453  * If no bits are set on the state struct after clearing things, the
 454  * struct is freed and removed from the tree
 455  */
 456 static int clear_state_bit(struct extent_io_tree *tree,
 457                             struct extent_state *state,
 458                             int *bits, int wake)
 459 {
 460         int bits_to_clear = *bits & ~EXTENT_CTLBITS;
 461         int ret = state->state & bits_to_clear;
 462
 463         if ((bits_to_clear & EXTENT_DIRTY) && (state->state & EXTENT_DIRTY)) {
 464                 u64 range = state->end - state->start + 1;
 465                 WARN_ON(range > tree->dirty_bytes);
 466                 tree->dirty_bytes -= range;
 467         }
 468         clear_state_cb(tree, state, bits);
 469         state->state &= ~bits_to_clear;
 470         if (wake)
 471                 wake_up(&state->wq);
 472         if (state->state == 0) {
 473                 if (state->tree) {
 474                         rb_erase(&state->rb_node, &tree->state);
 475                         state->tree = NULL;
 476                         free_extent_state(state);
 477                 } else {
 478                         WARN_ON(1);
 479                 }
 480         } else {
 481                 merge_state(tree, state);
 482         }
 483         return ret;
 484 }
 485
 486 /*
 487  * clear some bits on a range in the tree.  This may require splitting
 488  * or inserting elements in the tree, so the gfp mask is used to
 489  * indicate which allocations or sleeping are allowed.
 490  *
 491  * pass 'wake' == 1 to kick any sleepers, and 'delete' == 1 to remove
 492  * the given range from the tree regardless of state (ie for truncate).
 493  *
 494  * the range [start, end] is inclusive.
 495  *
 496  * This takes the tree lock, and returns < 0 on error, > 0 if any of the
 497  * bits were already set, or zero if none of the bits were already set.
 498  */
 499 int clear_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
 500                      int bits, int wake, int delete,
 501                      struct extent_state **cached_state,
 502                      gfp_t mask)
 503 {
 504         struct extent_state *state;
 505         struct extent_state *cached;
 506         struct extent_state *prealloc = NULL;
 507         struct rb_node *next_node;
 508         struct rb_node *node;
 509         u64 last_end;
 510         int err;
 511         int set = 0;
 512         int clear = 0;
 513
 514         if (delete)
 515                 bits |= ~EXTENT_CTLBITS;
 516         bits |= EXTENT_FIRST_DELALLOC;
 517
 518         if (bits & (EXTENT_IOBITS | EXTENT_BOUNDARY))
 519                 clear = 1;
 520 again:
 521         if (!prealloc && (mask & __GFP_WAIT)) {
 522                 prealloc = alloc_extent_state(mask);
 523                 if (!prealloc)
 524                         return -ENOMEM;
 525         }
 526
 527         spin_lock(&tree->lock);
 528         if (cached_state) {
 529                 cached = *cached_state;
 530
 531                 if (clear) {
 532                         *cached_state = NULL;
 533                         cached_state = NULL;
 534                 }
 535
 536                 if (cached && cached->tree && cached->start == start) {
 537                         if (clear)
 538                                 atomic_dec(&cached->refs);
 539                         state = cached;
 540                         goto hit_next;
 541                 }
 542                 if (clear)
 543                         free_extent_state(cached);
 544         }
 545         /*
 546          * this search will find the extents that end after
 547          * our range starts
 548          */
 549         node = tree_search(tree, start);
 550         if (!node)
 551                 goto out;
 552         state = rb_entry(node, struct extent_state, rb_node);
 553 hit_next:
 554         if (state->start > end)
 555                 goto out;
 556         WARN_ON(state->end < start);
 557         last_end = state->end;
 558
 559         /*
 560          *     | ---- desired range ---- |
 561          *  | state | or
 562          *  | ------------- state -------------- |
 563          *
 564          * We need to split the extent we found, and may flip
 565          * bits on second half.
 566          *
 567          * If the extent we found extends past our range, we
 568          * just split and search again.  It'll get split again
 569          * the next time though.
 570          *
 571          * If the extent we found is inside our range, we clear
 572          * the desired bit on it.
 573          */
 574
 575         if (state->start < start) {
 576                 if (!prealloc)
 577                         prealloc = alloc_extent_state(GFP_ATOMIC);
 578                 err = split_state(tree, state, prealloc, start);
 579                 BUG_ON(err == -EEXIST);
 580                 prealloc = NULL;
 581                 if (err)
 582                         goto out;
 583                 if (state->end <= end) {
 584                         set |= clear_state_bit(tree, state, &bits, wake);
 585                         if (last_end == (u64)-1)
 586                                 goto out;
 587                         start = last_end + 1;
 588                 }
 589                 goto search_again;
 590         }
 591         /*
 592          * | ---- desired range ---- |
 593          *                        | state |
 594          * We need to split the extent, and clear the bit
 595          * on the first half
 596          */
 597         if (state->start <= end && state->end > end) {
 598                 if (!prealloc)
 599                         prealloc = alloc_extent_state(GFP_ATOMIC);
 600                 err = split_state(tree, state, prealloc, end + 1);
 601                 BUG_ON(err == -EEXIST);
 602                 if (wake)
 603                         wake_up(&state->wq);
 604
 605                 set |= clear_state_bit(tree, prealloc, &bits, wake);
 606
 607                 prealloc = NULL;
 608                 goto out;
 609         }
 610
 611         if (state->end < end && prealloc && !need_resched())
 612                 next_node = rb_next(&state->rb_node);
 613         else
 614                 next_node = NULL;
 615
 616         set |= clear_state_bit(tree, state, &bits, wake);
 617         if (last_end == (u64)-1)
 618                 goto out;
 619         start = last_end + 1;
 620         if (start <= end && next_node) {
 621                 state = rb_entry(next_node, struct extent_state,
 622                                  rb_node);
 623                 if (state->start == start)
 624                         goto hit_next;
 625         }
 626         goto search_again;
 627
 628 out:
 629         spin_unlock(&tree->lock);
 630         if (prealloc)
 631                 free_extent_state(prealloc);
 632
 633         return set;
 634
 635 search_again:
 636         if (start > end)
 637                 goto out;
 638         spin_unlock(&tree->lock);
 639         if (mask & __GFP_WAIT)
 640                 cond_resched();
 641         goto again;
 642 }
 643
 644 static int wait_on_state(struct extent_io_tree *tree,
 645                          struct extent_state *state)
 646                 __releases(tree->lock)
 647                 __acquires(tree->lock)
 648 {
 649         DEFINE_WAIT(wait);
 650         prepare_to_wait(&state->wq, &wait, TASK_UNINTERRUPTIBLE);
 651         spin_unlock(&tree->lock);
 652         schedule();
 653         spin_lock(&tree->lock);
 654         finish_wait(&state->wq, &wait);
 655         return 0;
 656 }
 657
 658 /*
 659  * waits for one or more bits to clear on a range in the state tree.
 660  * The range [start, end] is inclusive.
 661  * The tree lock is taken by this function
 662  */
 663 int wait_extent_bit(struct extent_io_tree *tree, u64 start, u64 end, int bits)
 664 {
 665         struct extent_state *state;
 666         struct rb_node *node;
 667
 668         spin_lock(&tree->lock);
 669 again:
 670         while (1) {
 671                 /*
 672                  * this search will find all the extents that end after
 673                  * our range starts
 674                  */
 675                 node = tree_search(tree, start);
 676                 if (!node)
 677                         break;
 678
 679                 state = rb_entry(node, struct extent_state, rb_node);
 680
 681                 if (state->start > end)
 682                         goto out;
 683
 684                 if (state->state & bits) {
 685                         start = state->start;
 686                         atomic_inc(&state->refs);
 687                         wait_on_state(tree, state);
 688                         free_extent_state(state);
 689                         goto again;
 690                 }
 691                 start = state->end + 1;
 692
 693                 if (start > end)
 694                         break;
 695
 696                 if (need_resched()) {
 697                         spin_unlock(&tree->lock);
 698                         cond_resched();
 699                         spin_lock(&tree->lock);
 700                 }
 701         }
 702 out:
 703         spin_unlock(&tree->lock);
 704         return 0;
 705 }
 706
 707 static int set_state_bits(struct extent_io_tree *tree,
 708                            struct extent_state *state,
 709                            int *bits)
 710 {
 711         int ret;
 712         int bits_to_set = *bits & ~EXTENT_CTLBITS;
 713
 714         ret = set_state_cb(tree, state, bits);
 715         if (ret)
 716                 return ret;
 717         if ((bits_to_set & EXTENT_DIRTY) && !(state->state & EXTENT_DIRTY)) {
 718                 u64 range = state->end - state->start + 1;
 719                 tree->dirty_bytes += range;
 720         }
 721         state->state |= bits_to_set;
 722
 723         return 0;
 724 }
 725
 726 static void cache_state(struct extent_state *state,
 727                         struct extent_state **cached_ptr)
 728 {
 729         if (cached_ptr && !(*cached_ptr)) {
 730                 if (state->state & (EXTENT_IOBITS | EXTENT_BOUNDARY)) {
 731                         *cached_ptr = state;
 732                         atomic_inc(&state->refs);
 733                 }
 734         }
 735 }
 736
 737 /*
 738  * set some bits on a range in the tree.  This may require allocations or
 739  * sleeping, so the gfp mask is used to indicate what is allowed.
 740  *
 741  * If any of the exclusive bits are set, this will fail with -EEXIST if some
 742  * part of the range already has the desired bits set.  The start of the
 743  * existing range is returned in failed_start in this case.
 744  *
 745  * [start, end] is inclusive This takes the tree lock.
 746  */
 747
 748 int set_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
 749                    int bits, int exclusive_bits, u64 *failed_start,
 750                    struct extent_state **cached_state, gfp_t mask)
 751 {
 752         struct extent_state *state;
 753         struct extent_state *prealloc = NULL;
 754         struct rb_node *node;
 755         int err = 0;
 756         u64 last_start;
 757         u64 last_end;
 758
 759         bits |= EXTENT_FIRST_DELALLOC;
 760 again:
 761         if (!prealloc && (mask & __GFP_WAIT)) {
 762                 prealloc = alloc_extent_state(mask);
 763                 if (!prealloc)
 764                         return -ENOMEM;
 765         }
 766
 767         spin_lock(&tree->lock);
 768         if (cached_state && *cached_state) {
 769                 state = *cached_state;
 770                 if (state->start == start && state->tree) {
 771                         node = &state->rb_node;
 772                         goto hit_next;
 773                 }
 774         }
 775         /*
 776          * this search will find all the extents that end after
 777          * our range starts.
 778          */
 779         node = tree_search(tree, start);
 780         if (!node) {
 781                 err = insert_state(tree, prealloc, start, end, &bits);
 782                 prealloc = NULL;
 783                 BUG_ON(err == -EEXIST);
 784                 goto out;
 785         }
 786         state = rb_entry(node, struct extent_state, rb_node);
 787 hit_next:
 788         last_start = state->start;
 789         last_end = state->end;
 790
 791         /*
 792          * | ---- desired range ---- |
 793          * | state |
 794          *
 795          * Just lock what we found and keep going
 796          */
 797         if (state->start == start && state->end <= end) {
 798                 struct rb_node *next_node;
 799                 if (state->state & exclusive_bits) {
 800                         *failed_start = state->start;
 801                         err = -EEXIST;
 802                         goto out;
 803                 }
 804
 805                 err = set_state_bits(tree, state, &bits);
 806                 if (err)
 807                         goto out;
 808
 809                 cache_state(state, cached_state);
 810                 merge_state(tree, state);
 811                 if (last_end == (u64)-1)
 812                         goto out;
 813
 814                 start = last_end + 1;
 815                 if (start < end && prealloc && !need_resched()) {
 816                         next_node = rb_next(node);
 817                         if (next_node) {
 818                                 state = rb_entry(next_node, struct extent_state,
 819                                                  rb_node);
 820                                 if (state->start == start)
 821                                         goto hit_next;
 822                         }
 823                 }
 824                 goto search_again;
 825         }
 826
 827         /*
 828          *     | ---- desired range ---- |
 829          * | state |
 830          *   or
 831          * | ------------- state -------------- |
 832          *
 833          * We need to split the extent we found, and may flip bits on
 834          * second half.
 835          *
 836          * If the extent we found extends past our
 837          * range, we just split and search again.  It'll get split
 838          * again the next time though.
 839          *
 840          * If the extent we found is inside our range, we set the
 841          * desired bit on it.
 842          */
 843         if (state->start < start) {
 844                 if (state->state & exclusive_bits) {
 845                         *failed_start = start;
 846                         err = -EEXIST;
 847                         goto out;
 848                 }
 849                 err = split_state(tree, state, prealloc, start);
 850                 BUG_ON(err == -EEXIST);
 851                 prealloc = NULL;
 852                 if (err)
 853                         goto out;
 854                 if (state->end <= end) {
 855                         err = set_state_bits(tree, state, &bits);
 856                         if (err)
 857                                 goto out;
 858                         cache_state(state, cached_state);
 859                         merge_state(tree, state);
 860                         if (last_end == (u64)-1)
 861                                 goto out;
 862                         start = last_end + 1;
 863                 }
 864                 goto search_again;
 865         }
 866         /*
 867          * | ---- desired range ---- |
 868          *     | state | or               | state |
 869          *
 870          * There's a hole, we need to insert something in it and
 871          * ignore the extent we found.
 872          */
 873         if (state->start > start) {
 874                 u64 this_end;
 875                 if (end < last_start)
 876                         this_end = end;
 877                 else
 878                         this_end = last_start - 1;
 879                 err = insert_state(tree, prealloc, start, this_end,
 880                                    &bits);
 881                 BUG_ON(err == -EEXIST);
 882                 if (err) {
 883                         prealloc = NULL;
 884                         goto out;
 885                 }
 886                 cache_state(prealloc, cached_state);
 887                 prealloc = NULL;
 888                 start = this_end + 1;
 889                 goto search_again;
 890         }
 891         /*
 892          * | ---- desired range ---- |
 893          *                        | state |
 894          * We need to split the extent, and set the bit
 895          * on the first half
 896          */
 897         if (state->start <= end && state->end > end) {
 898                 if (state->state & exclusive_bits) {
 899                         *failed_start = start;
 900                         err = -EEXIST;
 901                         goto out;
 902                 }
 903                 err = split_state(tree, state, prealloc, end + 1);
 904                 BUG_ON(err == -EEXIST);
 905
 906                 err = set_state_bits(tree, prealloc, &bits);
 907                 if (err) {
 908                         prealloc = NULL;
 909                         goto out;
 910                 }
 911                 cache_state(prealloc, cached_state);
 912                 merge_state(tree, prealloc);
 913                 prealloc = NULL;
 914                 goto out;
 915         }
 916
 917         goto search_again;
 918
 919 out:
 920         spin_unlock(&tree->lock);
 921         if (prealloc)
 922                 free_extent_state(prealloc);
 923
 924         return err;
 925
 926 search_again:
 927         if (start > end)
 928                 goto out;
 929         spin_unlock(&tree->lock);
 930         if (mask & __GFP_WAIT)
 931                 cond_resched();
 932         goto again;
 933 }
 934
 935 /* wrappers around set/clear extent bit */
 936 int set_extent_dirty(struct extent_io_tree *tree, u64 start, u64 end,
 937                      gfp_t mask)
 938 {
 939         return set_extent_bit(tree, start, end, EXTENT_DIRTY, 0, NULL,
 940                               NULL, mask);
 941 }
 942
 943 int set_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
 944                     int bits, gfp_t mask)
 945 {
 946         return set_extent_bit(tree, start, end, bits, 0, NULL,
 947                               NULL, mask);
 948 }
 949
 950 int clear_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
 951                       int bits, gfp_t mask)
 952 {
 953         return clear_extent_bit(tree, start, end, bits, 0, 0, NULL, mask);
 954 }
 955
 956 int set_extent_delalloc(struct extent_io_tree *tree, u64 start, u64 end,
 957                         struct extent_state **cached_state, gfp_t mask)
 958 {
 959         return set_extent_bit(tree, start, end,
 960                               EXTENT_DELALLOC | EXTENT_DIRTY | EXTENT_UPTODATE,
 961                               0, NULL, cached_state, mask);
 962 }
 963
 964 int clear_extent_dirty(struct extent_io_tree *tree, u64 start, u64 end,
 965                        gfp_t mask)
 966 {
 967         return clear_extent_bit(tree, start, end,
 968                                 EXTENT_DIRTY | EXTENT_DELALLOC |
 969                                 EXTENT_DO_ACCOUNTING, 0, 0, NULL, mask);
 970 }
 971
 972 int set_extent_new(struct extent_io_tree *tree, u64 start, u64 end,
 973                      gfp_t mask)
 974 {
 975         return set_extent_bit(tree, start, end, EXTENT_NEW, 0, NULL,
 976                               NULL, mask);
 977 }
 978
 979 static int clear_extent_new(struct extent_io_tree *tree, u64 start, u64 end,
 980                        gfp_t mask)
 981 {
 982         return clear_extent_bit(tree, start, end, EXTENT_NEW, 0, 0,
 983                                 NULL, mask);
 984 }
 985
 986 int set_extent_uptodate(struct extent_io_tree *tree, u64 start, u64 end,
 987                         gfp_t mask)
 988 {
 989         return set_extent_bit(tree, start, end, EXTENT_UPTODATE, 0, NULL,
 990                               NULL, mask);
 991 }
 992
 993 static int clear_extent_uptodate(struct extent_io_tree *tree, u64 start,
 994                                  u64 end, struct extent_state **cached_state,
 995                                  gfp_t mask)
 996 {
 997         return clear_extent_bit(tree, start, end, EXTENT_UPTODATE, 0, 0,
 998                                 cached_state, mask);
 999 }
1000
1001 int wait_on_extent_writeback(struct extent_io_tree *tree, u64 start, u64 end)
1002 {
1003         return wait_extent_bit(tree, start, end, EXTENT_WRITEBACK);
1004 }
1005
1006 /*
1007  * either insert or lock state struct between start and end use mask to tell
1008  * us if waiting is desired.
1009  */
1010 int lock_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
1011                      int bits, struct extent_state **cached_state, gfp_t mask)
1012 {
1013         int err;
1014         u64 failed_start;
1015         while (1) {
1016                 err = set_extent_bit(tree, start, end, EXTENT_LOCKED | bits,
1017                                      EXTENT_LOCKED, &failed_start,
1018                                      cached_state, mask);
1019                 if (err == -EEXIST && (mask & __GFP_WAIT)) {
1020                         wait_extent_bit(tree, failed_start, end, EXTENT_LOCKED);
1021                         start = failed_start;
1022                 } else {
1023                         break;
1024                 }
1025                 WARN_ON(start > end);
1026         }
1027         return err;
1028 }
1029
1030 int lock_extent(struct extent_io_tree *tree, u64 start, u64 end, gfp_t mask)
1031 {
1032         return lock_extent_bits(tree, start, end, 0, NULL, mask);
1033 }
1034
1035 int try_lock_extent(struct extent_io_tree *tree, u64 start, u64 end,
1036                     gfp_t mask)
1037 {
1038         int err;
1039         u64 failed_start;
1040
1041         err = set_extent_bit(tree, start, end, EXTENT_LOCKED, EXTENT_LOCKED,
1042                              &failed_start, NULL, mask);
1043         if (err == -EEXIST) {
1044                 if (failed_start > start)
1045                         clear_extent_bit(tree, start, failed_start - 1,
1046                                          EXTENT_LOCKED, 1, 0, NULL, mask);
1047                 return 0;
1048         }
1049         return 1;
1050 }
1051
1052 int unlock_extent_cached(struct extent_io_tree *tree, u64 start, u64 end,
1053                          struct extent_state **cached, gfp_t mask)
1054 {
1055         return clear_extent_bit(tree, start, end, EXTENT_LOCKED, 1, 0, cached,
1056                                 mask);
1057 }
1058
1059 int unlock_extent(struct extent_io_tree *tree, u64 start, u64 end,
1060                   gfp_t mask)
1061 {
1062         return clear_extent_bit(tree, start, end, EXTENT_LOCKED, 1, 0, NULL,
1063                                 mask);
1064 }
1065
1066 /*
1067  * helper function to set pages and extents in the tree dirty
1068  */
1069 int set_range_dirty(struct extent_io_tree *tree, u64 start, u64 end)
1070 {
1071         unsigned long index = start >> PAGE_CACHE_SHIFT;
1072         unsigned long end_index = end >> PAGE_CACHE_SHIFT;
1073         struct page *page;
1074
1075         while (index <= end_index) {
1076                 page = find_get_page(tree->mapping, index);
1077                 BUG_ON(!page);
1078                 __set_page_dirty_nobuffers(page);
1079                 page_cache_release(page);
1080                 index++;
1081         }
1082         return 0;
1083 }
1084
1085 /*
1086  * helper function to set both pages and extents in the tree writeback
1087  */
1088 static int set_range_writeback(struct extent_io_tree *tree, u64 start, u64 end)
1089 {
1090         unsigned long index = start >> PAGE_CACHE_SHIFT;
1091         unsigned long end_index = end >> PAGE_CACHE_SHIFT;
1092         struct page *page;
1093
1094         while (index <= end_index) {
1095                 page = find_get_page(tree->mapping, index);
1096                 BUG_ON(!page);
1097                 set_page_writeback(page);
1098                 page_cache_release(page);
1099                 index++;
1100         }
1101         return 0;
1102 }
1103
1104 /*
1105  * find the first offset in the io tree with 'bits' set. zero is
1106  * returned if we find something, and *start_ret and *end_ret are
1107  * set to reflect the state struct that was found.
1108  *
1109  * If nothing was found, 1 is returned, < 0 on error
1110  */
1111 int find_first_extent_bit(struct extent_io_tree *tree, u64 start,
1112                           u64 *start_ret, u64 *end_ret, int bits)
1113 {
1114         struct rb_node *node;
1115         struct extent_state *state;
1116         int ret = 1;
1117
1118         spin_lock(&tree->lock);
1119         /*
1120          * this search will find all the extents that end after
1121          * our range starts.
1122          */
1123         node = tree_search(tree, start);
1124         if (!node)
1125                 goto out;
1126
1127         while (1) {
1128                 state = rb_entry(node, struct extent_state, rb_node);
1129                 if (state->end >= start && (state->state & bits)) {
1130                         *start_ret = state->start;
1131                         *end_ret = state->end;
1132                         ret = 0;
1133                         break;
1134                 }
1135                 node = rb_next(node);
1136                 if (!node)
1137                         break;
1138         }
1139 out:
1140         spin_unlock(&tree->lock);
1141         return ret;
1142 }
1143
1144 /* find the first state struct with 'bits' set after 'start', and
1145  * return it.  tree->lock must be held.  NULL will returned if
1146  * nothing was found after 'start'
1147  */
1148 struct extent_state *find_first_extent_bit_state(struct extent_io_tree *tree,
1149                                                  u64 start, int bits)
1150 {
1151         struct rb_node *node;
1152         struct extent_state *state;
1153
1154         /*
1155          * this search will find all the extents that end after
1156          * our range starts.
1157          */
1158         node = tree_search(tree, start);
1159         if (!node)
1160                 goto out;
1161
1162         while (1) {
1163                 state = rb_entry(node, struct extent_state, rb_node);
1164                 if (state->end >= start && (state->state & bits))
1165                         return state;
1166
1167                 node = rb_next(node);
1168                 if (!node)
1169                         break;
1170         }
1171 out:
1172         return NULL;
1173 }
1174
1175 /*
1176  * find a contiguous range of bytes in the file marked as delalloc, not
1177  * more than 'max_bytes'.  start and end are used to return the range,
1178  *
1179  * 1 is returned if we find something, 0 if nothing was in the tree
1180  */
1181 static noinline u64 find_delalloc_range(struct extent_io_tree *tree,
1182                                         u64 *start, u64 *end, u64 max_bytes,
1183                                         struct extent_state **cached_state)
1184 {
1185         struct rb_node *node;
1186         struct extent_state *state;
1187         u64 cur_start = *start;
1188         u64 found = 0;
1189         u64 total_bytes = 0;
1190
1191         spin_lock(&tree->lock);
1192
1193         /*
1194          * this search will find all the extents that end after
1195          * our range starts.
1196          */
1197         node = tree_search(tree, cur_start);
1198         if (!node) {
1199                 if (!found)
1200                         *end = (u64)-1;
1201                 goto out;
1202         }
1203
1204         while (1) {
1205                 state = rb_entry(node, struct extent_state, rb_node);
1206                 if (found && (state->start != cur_start ||
1207                               (state->state & EXTENT_BOUNDARY))) {
1208                         goto out;
1209                 }
1210                 if (!(state->state & EXTENT_DELALLOC)) {
1211                         if (!found)
1212                                 *end = state->end;
1213                         goto out;
1214                 }
1215                 if (!found) {
1216                         *start = state->start;
1217                         *cached_state = state;
1218                         atomic_inc(&state->refs);
1219                 }
1220                 found++;
1221                 *end = state->end;
1222                 cur_start = state->end + 1;
1223                 node = rb_next(node);
1224                 if (!node)
1225                         break;
1226                 total_bytes += state->end - state->start + 1;
1227                 if (total_bytes >= max_bytes)
1228                         break;
1229         }
1230 out:
1231         spin_unlock(&tree->lock);
1232         return found;
1233 }
1234
1235 static noinline int __unlock_for_delalloc(struct inode *inode,
1236                                           struct page *locked_page,
1237                                           u64 start, u64 end)
1238 {
1239         int ret;
1240         struct page *pages[16];
1241         unsigned long index = start >> PAGE_CACHE_SHIFT;
1242         unsigned long end_index = end >> PAGE_CACHE_SHIFT;
1243         unsigned long nr_pages = end_index - index + 1;
1244         int i;
1245
1246         if (index == locked_page->index && end_index == index)
1247                 return 0;
1248
1249         while (nr_pages > 0) {
1250                 ret = find_get_pages_contig(inode->i_mapping, index,
1251                                      min_t(unsigned long, nr_pages,
1252                                      ARRAY_SIZE(pages)), pages);
1253                 for (i = 0; i < ret; i++) {
1254                         if (pages[i] != locked_page)
1255                                 unlock_page(pages[i]);
1256                         page_cache_release(pages[i]);
1257                 }
1258                 nr_pages -= ret;
1259                 index += ret;
1260                 cond_resched();
1261         }
1262         return 0;
1263 }
1264
1265 static noinline int lock_delalloc_pages(struct inode *inode,
1266                                         struct page *locked_page,
1267                                         u64 delalloc_start,
1268                                         u64 delalloc_end)
1269 {
1270         unsigned long index = delalloc_start >> PAGE_CACHE_SHIFT;
1271         unsigned long start_index = index;
1272         unsigned long end_index = delalloc_end >> PAGE_CACHE_SHIFT;
1273         unsigned long pages_locked = 0;
1274         struct page *pages[16];
1275         unsigned long nrpages;
1276         int ret;
1277         int i;
1278
1279         /* the caller is responsible for locking the start index */
1280         if (index == locked_page->index && index == end_index)
1281                 return 0;
1282
1283         /* skip the page at the start index */
1284         nrpages = end_index - index + 1;
1285         while (nrpages > 0) {
1286                 ret = find_get_pages_contig(inode->i_mapping, index,
1287                                      min_t(unsigned long,
1288                                      nrpages, ARRAY_SIZE(pages)), pages);
1289                 if (ret == 0) {
1290                         ret = -EAGAIN;
1291                         goto done;
1292                 }
1293                 /* now we have an array of pages, lock them all */
1294                 for (i = 0; i < ret; i++) {
1295                         /*
1296                          * the caller is taking responsibility for
1297                          * locked_page
1298                          */
1299                         if (pages[i] != locked_page) {
1300                                 lock_page(pages[i]);
1301                                 if (!PageDirty(pages[i]) ||
1302                                     pages[i]->mapping != inode->i_mapping) {
1303                                         ret = -EAGAIN;
1304                                         unlock_page(pages[i]);
1305                                         page_cache_release(pages[i]);
1306                                         goto done;
1307                                 }
1308                         }
1309                         page_cache_release(pages[i]);
1310                         pages_locked++;
1311                 }
1312                 nrpages -= ret;
1313                 index += ret;
1314                 cond_resched();
1315         }
1316         ret = 0;
1317 done:
1318         if (ret && pages_locked) {
1319                 __unlock_for_delalloc(inode, locked_page,
1320                               delalloc_start,
1321                               ((u64)(start_index + pages_locked - 1)) <<
1322                               PAGE_CACHE_SHIFT);
1323         }
1324         return ret;
1325 }
1326
1327 /*
1328  * find a contiguous range of bytes in the file marked as delalloc, not
1329  * more than 'max_bytes'.  start and end are used to return the range,
1330  *
1331  * 1 is returned if we find something, 0 if nothing was in the tree
1332  */
1333 static noinline u64 find_lock_delalloc_range(struct inode *inode,
1334                                              struct extent_io_tree *tree,
1335                                              struct page *locked_page,
1336                                              u64 *start, u64 *end,
1337                                              u64 max_bytes)
1338 {
1339         u64 delalloc_start;
1340         u64 delalloc_end;
1341         u64 found;
1342         struct extent_state *cached_state = NULL;
1343         int ret;
1344         int loops = 0;
1345
1346 again:
1347         /* step one, find a bunch of delalloc bytes starting at start */
1348         delalloc_start = *start;
1349         delalloc_end = 0;
1350         found = find_delalloc_range(tree, &delalloc_start, &delalloc_end,
1351                                     max_bytes, &cached_state);
1352         if (!found || delalloc_end <= *start) {
1353                 *start = delalloc_start;
1354                 *end = delalloc_end;
1355                 free_extent_state(cached_state);
1356                 return found;
1357         }
1358
1359         /*
1360          * start comes from the offset of locked_page.  We have to lock
1361          * pages in order, so we can't process delalloc bytes before
1362          * locked_page
1363          */
1364         if (delalloc_start < *start)
1365                 delalloc_start = *start;
1366
1367         /*
1368          * make sure to limit the number of pages we try to lock down
1369          * if we're looping.
1370          */
1371         if (delalloc_end + 1 - delalloc_start > max_bytes && loops)
1372                 delalloc_end = delalloc_start + PAGE_CACHE_SIZE - 1;
1373
1374         /* step two, lock all the pages after the page that has start */
1375         ret = lock_delalloc_pages(inode, locked_page,
1376                                   delalloc_start, delalloc_end);
1377         if (ret == -EAGAIN) {
1378                 /* some of the pages are gone, lets avoid looping by
1379                  * shortening the size of the delalloc range we're searching
1380                  */
1381                 free_extent_state(cached_state);
1382                 if (!loops) {
1383                         unsigned long offset = (*start) & (PAGE_CACHE_SIZE - 1);
1384                         max_bytes = PAGE_CACHE_SIZE - offset;
1385                         loops = 1;
1386                         goto again;
1387                 } else {
1388                         found = 0;
1389                         goto out_failed;
1390                 }
1391         }
1392         BUG_ON(ret);
1393
1394         /* step three, lock the state bits for the whole range */
1395         lock_extent_bits(tree, delalloc_start, delalloc_end,
1396                          0, &cached_state, GFP_NOFS);
1397
1398         /* then test to make sure it is all still delalloc */
1399         ret = test_range_bit(tree, delalloc_start, delalloc_end,
1400                              EXTENT_DELALLOC, 1, cached_state);
1401         if (!ret) {
1402                 unlock_extent_cached(tree, delalloc_start, delalloc_end,
1403                                      &cached_state, GFP_NOFS);
1404                 __unlock_for_delalloc(inode, locked_page,
1405                               delalloc_start, delalloc_end);
1406                 cond_resched();
1407                 goto again;
1408         }
1409         free_extent_state(cached_state);
1410         *start = delalloc_start;
1411         *end = delalloc_end;
1412 out_failed:
1413         return found;
1414 }
1415
1416 int extent_clear_unlock_delalloc(struct inode *inode,
1417                                 struct extent_io_tree *tree,
1418                                 u64 start, u64 end, struct page *locked_page,
1419                                 unsigned long op)
1420 {
1421         int ret;
1422         struct page *pages[16];
1423         unsigned long index = start >> PAGE_CACHE_SHIFT;
1424         unsigned long end_index = end >> PAGE_CACHE_SHIFT;
1425         unsigned long nr_pages = end_index - index + 1;
1426         int i;
1427         int clear_bits = 0;
1428
1429         if (op & EXTENT_CLEAR_UNLOCK)
1430                 clear_bits |= EXTENT_LOCKED;
1431         if (op & EXTENT_CLEAR_DIRTY)
1432                 clear_bits |= EXTENT_DIRTY;
1433
1434         if (op & EXTENT_CLEAR_DELALLOC)
1435                 clear_bits |= EXTENT_DELALLOC;
1436
1437         clear_extent_bit(tree, start, end, clear_bits, 1, 0, NULL, GFP_NOFS);
1438         if (!(op & (EXTENT_CLEAR_UNLOCK_PAGE | EXTENT_CLEAR_DIRTY |
1439                     EXTENT_SET_WRITEBACK | EXTENT_END_WRITEBACK |
1440                     EXTENT_SET_PRIVATE2)))
1441                 return 0;
1442
1443         while (nr_pages > 0) {
1444                 ret = find_get_pages_contig(inode->i_mapping, index,
1445                                      min_t(unsigned long,
1446                                      nr_pages, ARRAY_SIZE(pages)), pages);
1447                 for (i = 0; i < ret; i++) {
1448
1449                         if (op & EXTENT_SET_PRIVATE2)
1450                                 SetPagePrivate2(pages[i]);
1451
1452                         if (pages[i] == locked_page) {
1453                                 page_cache_release(pages[i]);
1454                                 continue;
1455                         }
1456                         if (op & EXTENT_CLEAR_DIRTY)
1457                                 clear_page_dirty_for_io(pages[i]);
1458                         if (op & EXTENT_SET_WRITEBACK)
1459                                 set_page_writeback(pages[i]);
1460                         if (op & EXTENT_END_WRITEBACK)
1461                                 end_page_writeback(pages[i]);
1462                         if (op & EXTENT_CLEAR_UNLOCK_PAGE)
1463                                 unlock_page(pages[i]);
1464                         page_cache_release(pages[i]);
1465                 }
1466                 nr_pages -= ret;
1467                 index += ret;
1468                 cond_resched();
1469         }
1470         return 0;
1471 }
1472
1473 /*
1474  * count the number of bytes in the tree that have a given bit(s)
1475  * set.  This can be fairly slow, except for EXTENT_DIRTY which is
1476  * cached.  The total number found is returned.
1477  */
1478 u64 count_range_bits(struct extent_io_tree *tree,
1479                      u64 *start, u64 search_end, u64 max_bytes,
1480                      unsigned long bits)
1481 {
1482         struct rb_node *node;
1483         struct extent_state *state;
1484         u64 cur_start = *start;
1485         u64 total_bytes = 0;
1486         int found = 0;
1487
1488         if (search_end <= cur_start) {
1489                 WARN_ON(1);
1490                 return 0;
1491         }
1492
1493         spin_lock(&tree->lock);
1494         if (cur_start == 0 && bits == EXTENT_DIRTY) {
1495                 total_bytes = tree->dirty_bytes;
1496                 goto out;
1497         }
1498         /*
1499          * this search will find all the extents that end after
1500          * our range starts.
1501          */
1502         node = tree_search(tree, cur_start);
1503         if (!node)
1504                 goto out;
1505
1506         while (1) {
1507                 state = rb_entry(node, struct extent_state, rb_node);
1508                 if (state->start > search_end)
1509                         break;
1510                 if (state->end >= cur_start && (state->state & bits)) {
1511                         total_bytes += min(search_end, state->end) + 1 -
1512                                        max(cur_start, state->start);
1513                         if (total_bytes >= max_bytes)
1514                                 break;
1515                         if (!found) {
1516                                 *start = state->start;
1517                                 found = 1;
1518                         }
1519                 }
1520                 node = rb_next(node);
1521                 if (!node)
1522                         break;
1523         }
1524 out:
1525         spin_unlock(&tree->lock);
1526         return total_bytes;
1527 }
1528
1529 /*
1530  * set the private field for a given byte offset in the tree.  If there isn't
1531  * an extent_state there already, this does nothing.
1532  */
1533 int set_state_private(struct extent_io_tree *tree, u64 start, u64 private)
1534 {
1535         struct rb_node *node;
1536         struct extent_state *state;
1537         int ret = 0;
1538
1539         spin_lock(&tree->lock);
1540         /*
1541          * this search will find all the extents that end after
1542          * our range starts.
1543          */
1544         node = tree_search(tree, start);
1545         if (!node) {
1546                 ret = -ENOENT;
1547                 goto out;
1548         }
1549         state = rb_entry(node, struct extent_state, rb_node);
1550         if (state->start != start) {
1551                 ret = -ENOENT;
1552                 goto out;
1553         }
1554         state->private = private;
1555 out:
1556         spin_unlock(&tree->lock);
1557         return ret;
1558 }
1559
1560 int get_state_private(struct extent_io_tree *tree, u64 start, u64 *private)
1561 {
1562         struct rb_node *node;
1563         struct extent_state *state;
1564         int ret = 0;
1565
1566         spin_lock(&tree->lock);
1567         /*
1568          * this search will find all the extents that end after
1569          * our range starts.
1570          */
1571         node = tree_search(tree, start);
1572         if (!node) {
1573                 ret = -ENOENT;
1574                 goto out;
1575         }
1576         state = rb_entry(node, struct extent_state, rb_node);
1577         if (state->start != start) {
1578                 ret = -ENOENT;
1579                 goto out;
1580         }
1581         *private = state->private;
1582 out:
1583         spin_unlock(&tree->lock);
1584         return ret;
1585 }
1586
1587 /*
1588  * searches a range in the state tree for a given mask.
1589  * If 'filled' == 1, this returns 1 only if every extent in the tree
1590  * has the bits set.  Otherwise, 1 is returned if any bit in the
1591  * range is found set.
1592  */
1593 int test_range_bit(struct extent_io_tree *tree, u64 start, u64 end,
1594                    int bits, int filled, struct extent_state *cached)
1595 {
1596         struct extent_state *state = NULL;
1597         struct rb_node *node;
1598         int bitset = 0;
1599
1600         spin_lock(&tree->lock);
1601         if (cached && cached->tree && cached->start == start)
1602                 node = &cached->rb_node;
1603         else
1604                 node = tree_search(tree, start);
1605         while (node && start <= end) {
1606                 state = rb_entry(node, struct extent_state, rb_node);
1607
1608                 if (filled && state->start > start) {
1609                         bitset = 0;
1610                         break;
1611                 }
1612
1613                 if (state->start > end)
1614                         break;
1615
1616                 if (state->state & bits) {
1617                         bitset = 1;
1618                         if (!filled)
1619                                 break;
1620                 } else if (filled) {
1621                         bitset = 0;
1622                         break;
1623                 }
1624
1625                 if (state->end == (u64)-1)
1626                         break;
1627
1628                 start = state->end + 1;
1629                 if (start > end)
1630                         break;
1631                 node = rb_next(node);
1632                 if (!node) {
1633                         if (filled)
1634                                 bitset = 0;
1635                         break;
1636                 }
1637         }
1638         spin_unlock(&tree->lock);
1639         return bitset;
1640 }
1641
1642 /*
1643  * helper function to set a given page up to date if all the
1644  * extents in the tree for that page are up to date
1645  */
1646 static int check_page_uptodate(struct extent_io_tree *tree,
1647                                struct page *page)
1648 {
1649         u64 start = (u64)page->index << PAGE_CACHE_SHIFT;
1650         u64 end = start + PAGE_CACHE_SIZE - 1;
1651         if (test_range_bit(tree, start, end, EXTENT_UPTODATE, 1, NULL))
1652                 SetPageUptodate(page);
1653         return 0;
1654 }
1655
1656 /*
1657  * helper function to unlock a page if all the extents in the tree
1658  * for that page are unlocked
1659  */
1660 static int check_page_locked(struct extent_io_tree *tree,
1661                              struct page *page)
1662 {
1663         u64 start = (u64)page->index << PAGE_CACHE_SHIFT;
1664         u64 end = start + PAGE_CACHE_SIZE - 1;
1665         if (!test_range_bit(tree, start, end, EXTENT_LOCKED, 0, NULL))
1666                 unlock_page(page);
1667         return 0;
1668 }
1669
1670 /*
1671  * helper function to end page writeback if all the extents
1672  * in the tree for that page are done with writeback
1673  */
1674 static int check_page_writeback(struct extent_io_tree *tree,
1675                              struct page *page)
1676 {
1677         end_page_writeback(page);
1678         return 0;
1679 }
1680
1681 /* lots and lots of room for performance fixes in the end_bio funcs */
1682
1683 /*
1684  * after a writepage IO is done, we need to:
1685  * clear the uptodate bits on error
1686  * clear the writeback bits in the extent tree for this IO
1687  * end_page_writeback if the page has no more pending IO
1688  *
1689  * Scheduling is not allowed, so the extent state tree is expected
1690  * to have one and only one object corresponding to this IO.
1691  */
1692 static void end_bio_extent_writepage(struct bio *bio, int err)
1693 {
1694         int uptodate = err == 0;
1695         struct bio_vec *bvec = bio->bi_io_vec + bio->bi_vcnt - 1;
1696         struct extent_io_tree *tree;
1697         u64 start;
1698         u64 end;
1699         int whole_page;
1700         int ret;
1701
1702         do {
1703                 struct page *page = bvec->bv_page;
1704                 tree = &BTRFS_I(page->mapping->host)->io_tree;
1705
1706                 start = ((u64)page->index << PAGE_CACHE_SHIFT) +
1707                          bvec->bv_offset;
1708                 end = start + bvec->bv_len - 1;
1709
1710                 if (bvec->bv_offset == 0 && bvec->bv_len == PAGE_CACHE_SIZE)
1711                         whole_page = 1;
1712                 else
1713                         whole_page = 0;
1714
1715                 if (--bvec >= bio->bi_io_vec)
1716                         prefetchw(&bvec->bv_page->flags);
1717                 if (tree->ops && tree->ops->writepage_end_io_hook) {
1718                         ret = tree->ops->writepage_end_io_hook(page, start,
1719                                                        end, NULL, uptodate);
1720                         if (ret)
1721                                 uptodate = 0;
1722                 }
1723
1724                 if (!uptodate && tree->ops &&
1725                     tree->ops->writepage_io_failed_hook) {
1726                         ret = tree->ops->writepage_io_failed_hook(bio, page,
1727                                                          start, end, NULL);
1728                         if (ret == 0) {
1729                                 uptodate = (err == 0);
1730                                 continue;
1731                         }
1732                 }
1733
1734                 if (!uptodate) {
1735                         clear_extent_uptodate(tree, start, end, NULL, GFP_NOFS);
1736                         ClearPageUptodate(page);
1737                         SetPageError(page);
1738                 }
1739
1740                 if (whole_page)
1741                         end_page_writeback(page);
1742                 else
1743                         check_page_writeback(tree, page);
1744         } while (bvec >= bio->bi_io_vec);
1745
1746         bio_put(bio);
1747 }
1748
1749 /*
1750  * after a readpage IO is done, we need to:
1751  * clear the uptodate bits on error
1752  * set the uptodate bits if things worked
1753  * set the page up to date if all extents in the tree are uptodate
1754  * clear the lock bit in the extent tree
1755  * unlock the page if there are no other extents locked for it
1756  *
1757  * Scheduling is not allowed, so the extent state tree is expected
1758  * to have one and only one object corresponding to this IO.
1759  */
1760 static void end_bio_extent_readpage(struct bio *bio, int err)
1761 {
1762         int uptodate = test_bit(BIO_UPTODATE, &bio->bi_flags);
1763         struct bio_vec *bvec_end = bio->bi_io_vec + bio->bi_vcnt - 1;
1764         struct bio_vec *bvec = bio->bi_io_vec;
1765         struct extent_io_tree *tree;
1766         u64 start;
1767         u64 end;
1768         int whole_page;
1769         int ret;
1770
1771         if (err)
1772                 uptodate = 0;
1773
1774         do {
1775                 struct page *page = bvec->bv_page;
1776                 tree = &BTRFS_I(page->mapping->host)->io_tree;
1777
1778                 start = ((u64)page->index << PAGE_CACHE_SHIFT) +
1779                         bvec->bv_offset;
1780                 end = start + bvec->bv_len - 1;
1781
1782                 if (bvec->bv_offset == 0 && bvec->bv_len == PAGE_CACHE_SIZE)
1783                         whole_page = 1;
1784                 else
1785                         whole_page = 0;
1786
1787                 if (++bvec <= bvec_end)
1788                         prefetchw(&bvec->bv_page->flags);
1789
1790                 if (uptodate && tree->ops && tree->ops->readpage_end_io_hook) {
1791                         ret = tree->ops->readpage_end_io_hook(page, start, end,
1792                                                               NULL);
1793                         if (ret)
1794                                 uptodate = 0;
1795                 }
1796                 if (!uptodate && tree->ops &&
1797                     tree->ops->readpage_io_failed_hook) {
1798                         ret = tree->ops->readpage_io_failed_hook(bio, page,
1799                                                          start, end, NULL);
1800                         if (ret == 0) {
1801                                 uptodate =
1802                                         test_bit(BIO_UPTODATE, &bio->bi_flags);
1803                                 if (err)
1804                                         uptodate = 0;
1805                                 continue;
1806                         }
1807                 }
1808
1809                 if (uptodate) {
1810                         set_extent_uptodate(tree, start, end,
1811                                             GFP_ATOMIC);
1812                 }
1813                 unlock_extent(tree, start, end, GFP_ATOMIC);
1814
1815                 if (whole_page) {
1816                         if (uptodate) {
1817                                 SetPageUptodate(page);
1818                         } else {
1819                                 ClearPageUptodate(page);
1820                                 SetPageError(page);
1821                         }
1822                         unlock_page(page);
1823                 } else {
1824                         if (uptodate) {
1825                                 check_page_uptodate(tree, page);
1826                         } else {
1827                                 ClearPageUptodate(page);
1828                                 SetPageError(page);
1829                         }
1830                         check_page_locked(tree, page);
1831                 }
1832         } while (bvec <= bvec_end);
1833
1834         bio_put(bio);
1835 }
1836
1837 /*
1838  * IO done from prepare_write is pretty simple, we just unlock
1839  * the structs in the extent tree when done, and set the uptodate bits
1840  * as appropriate.
1841  */
1842 static void end_bio_extent_preparewrite(struct bio *bio, int err)
1843 {
1844         const int uptodate = test_bit(BIO_UPTODATE, &bio->bi_flags);
1845         struct bio_vec *bvec = bio->bi_io_vec + bio->bi_vcnt - 1;
1846         struct extent_io_tree *tree;
1847         u64 start;
1848         u64 end;
1849
1850         do {
1851                 struct page *page = bvec->bv_page;
1852                 tree = &BTRFS_I(page->mapping->host)->io_tree;
1853
1854                 start = ((u64)page->index << PAGE_CACHE_SHIFT) +
1855                         bvec->bv_offset;
1856                 end = start + bvec->bv_len - 1;
1857
1858                 if (--bvec >= bio->bi_io_vec)
1859                         prefetchw(&bvec->bv_page->flags);
1860
1861                 if (uptodate) {
1862                         set_extent_uptodate(tree, start, end, GFP_ATOMIC);
1863                 } else {
1864                         ClearPageUptodate(page);
1865                         SetPageError(page);
1866                 }
1867
1868                 unlock_extent(tree, start, end, GFP_ATOMIC);
1869
1870         } while (bvec >= bio->bi_io_vec);
1871
1872         bio_put(bio);
1873 }
1874
1875 static struct bio *
1876 extent_bio_alloc(struct block_device *bdev, u64 first_sector, int nr_vecs,
1877                  gfp_t gfp_flags)
1878 {
1879         struct bio *bio;
1880
1881         bio = bio_alloc(gfp_flags, nr_vecs);
1882
1883         if (bio == NULL && (current->flags & PF_MEMALLOC)) {
1884                 while (!bio && (nr_vecs /= 2))
1885                         bio = bio_alloc(gfp_flags, nr_vecs);
1886         }
1887
1888         if (bio) {
1889                 bio->bi_size = 0;
1890                 bio->bi_bdev = bdev;
1891                 bio->bi_sector = first_sector;
1892         }
1893         return bio;
1894 }
1895
1896 static int submit_one_bio(int rw, struct bio *bio, int mirror_num,
1897                           unsigned long bio_flags)
1898 {
1899         int ret = 0;
1900         struct bio_vec *bvec = bio->bi_io_vec + bio->bi_vcnt - 1;
1901         struct page *page = bvec->bv_page;
1902         struct extent_io_tree *tree = bio->bi_private;
1903         u64 start;
1904         u64 end;
1905
1906         start = ((u64)page->index << PAGE_CACHE_SHIFT) + bvec->bv_offset;
1907         end = start + bvec->bv_len - 1;
1908
1909         bio->bi_private = NULL;
1910
1911         bio_get(bio);
1912
1913         if (tree->ops && tree->ops->submit_bio_hook)
1914                 tree->ops->submit_bio_hook(page->mapping->host, rw, bio,
1915                                            mirror_num, bio_flags, start);
1916         else
1917                 submit_bio(rw, bio);
1918         if (bio_flagged(bio, BIO_EOPNOTSUPP))
1919                 ret = -EOPNOTSUPP;
1920         bio_put(bio);
1921         return ret;
1922 }
1923
1924 static int submit_extent_page(int rw, struct extent_io_tree *tree,
1925                               struct page *page, sector_t sector,
1926                               size_t size, unsigned long offset,
1927                               struct block_device *bdev,
1928                               struct bio **bio_ret,
1929                               unsigned long max_pages,
1930                               bio_end_io_t end_io_func,
1931                               int mirror_num,
1932                               unsigned long prev_bio_flags,
1933                               unsigned long bio_flags)
1934 {
1935         int ret = 0;
1936         struct bio *bio;
1937         int nr;
1938         int contig = 0;
1939         int this_compressed = bio_flags & EXTENT_BIO_COMPRESSED;
1940         int old_compressed = prev_bio_flags & EXTENT_BIO_COMPRESSED;
1941         size_t page_size = min_t(size_t, size, PAGE_CACHE_SIZE);
1942
1943         if (bio_ret && *bio_ret) {
1944                 bio = *bio_ret;
1945                 if (old_compressed)
1946                         contig = bio->bi_sector == sector;
1947                 else
1948                         contig = bio->bi_sector + (bio->bi_size >> 9) ==
1949                                 sector;
1950
1951                 if (prev_bio_flags != bio_flags || !contig ||
1952                     (tree->ops && tree->ops->merge_bio_hook &&
1953                      tree->ops->merge_bio_hook(page, offset, page_size, bio,
1954                                                bio_flags)) ||
1955                     bio_add_page(bio, page, page_size, offset) < page_size) {
1956                         ret = submit_one_bio(rw, bio, mirror_num,
1957                                              prev_bio_flags);
1958                         bio = NULL;
1959                 } else {
1960                         return 0;
1961                 }
1962         }
1963         if (this_compressed)
1964                 nr = BIO_MAX_PAGES;
1965         else
1966                 nr = bio_get_nr_vecs(bdev);
1967
1968         bio = extent_bio_alloc(bdev, sector, nr, GFP_NOFS | __GFP_HIGH);
1969
1970         bio_add_page(bio, page, page_size, offset);
1971         bio->bi_end_io = end_io_func;
1972         bio->bi_private = tree;
1973
1974         if (bio_ret)
1975                 *bio_ret = bio;
1976         else
1977                 ret = submit_one_bio(rw, bio, mirror_num, bio_flags);
1978
1979         return ret;
1980 }
1981
1982 void set_page_extent_mapped(struct page *page)
1983 {
1984         if (!PagePrivate(page)) {
1985                 SetPagePrivate(page);
1986                 page_cache_get(page);
1987                 set_page_private(page, EXTENT_PAGE_PRIVATE);
1988         }
1989 }
1990
1991 static void set_page_extent_head(struct page *page, unsigned long len)
1992 {
1993         set_page_private(page, EXTENT_PAGE_PRIVATE_FIRST_PAGE | len << 2);
1994 }
1995
1996 /*
1997  * basic readpage implementation.  Locked extent state structs are inserted
1998  * into the tree that are removed when the IO is done (by the end_io
1999  * handlers)
2000  */
2001 static int __extent_read_full_page(struct extent_io_tree *tree,
2002                                    struct page *page,
2003                                    get_extent_t *get_extent,
2004                                    struct bio **bio, int mirror_num,
2005                                    unsigned long *bio_flags)
2006 {
2007         struct inode *inode = page->mapping->host;
2008         u64 start = (u64)page->index << PAGE_CACHE_SHIFT;
2009         u64 page_end = start + PAGE_CACHE_SIZE - 1;
2010         u64 end;
2011         u64 cur = start;
2012         u64 extent_offset;
2013         u64 last_byte = i_size_read(inode);
2014         u64 block_start;
2015         u64 cur_end;
2016         sector_t sector;
2017         struct extent_map *em;
2018         struct block_device *bdev;
2019         struct btrfs_ordered_extent *ordered;
2020         int ret;
2021         int nr = 0;
2022         size_t page_offset = 0;
2023         size_t iosize;
2024         size_t disk_io_size;
2025         size_t blocksize = inode->i_sb->s_blocksize;
2026         unsigned long this_bio_flag = 0;
2027
2028         set_page_extent_mapped(page);
2029
2030         end = page_end;
2031         while (1) {
2032                 lock_extent(tree, start, end, GFP_NOFS);
2033                 ordered = btrfs_lookup_ordered_extent(inode, start);
2034                 if (!ordered)
2035                         break;
2036                 unlock_extent(tree, start, end, GFP_NOFS);
2037                 btrfs_start_ordered_extent(inode, ordered, 1);
2038                 btrfs_put_ordered_extent(ordered);
2039         }
2040
2041         if (page->index == last_byte >> PAGE_CACHE_SHIFT) {
2042                 char *userpage;
2043                 size_t zero_offset = last_byte & (PAGE_CACHE_SIZE - 1);
2044
2045                 if (zero_offset) {
2046                         iosize = PAGE_CACHE_SIZE - zero_offset;
2047                         userpage = kmap_atomic(page, KM_USER0);
2048                         memset(userpage + zero_offset, 0, iosize);
2049                         flush_dcache_page(page);
2050                         kunmap_atomic(userpage, KM_USER0);
2051                 }
2052         }
2053         while (cur <= end) {
2054                 if (cur >= last_byte) {
2055                         char *userpage;
2056                         iosize = PAGE_CACHE_SIZE - page_offset;
2057                         userpage = kmap_atomic(page, KM_USER0);
2058                         memset(userpage + page_offset, 0, iosize);
2059                         flush_dcache_page(page);
2060                         kunmap_atomic(userpage, KM_USER0);
2061                         set_extent_uptodate(tree, cur, cur + iosize - 1,
2062                                             GFP_NOFS);
2063                         unlock_extent(tree, cur, cur + iosize - 1, GFP_NOFS);
2064                         break;
2065                 }
2066                 em = get_extent(inode, page, page_offset, cur,
2067                                 end - cur + 1, 0);
2068                 if (IS_ERR(em) || !em) {
2069                         SetPageError(page);
2070                         unlock_extent(tree, cur, end, GFP_NOFS);
2071                         break;
2072                 }
2073                 extent_offset = cur - em->start;
2074                 BUG_ON(extent_map_end(em) <= cur);
2075                 BUG_ON(end < cur);
2076
2077                 if (test_bit(EXTENT_FLAG_COMPRESSED, &em->flags))
2078                         this_bio_flag = EXTENT_BIO_COMPRESSED;
2079
2080                 iosize = min(extent_map_end(em) - cur, end - cur + 1);
2081                 cur_end = min(extent_map_end(em) - 1, end);
2082                 iosize = (iosize + blocksize - 1) & ~((u64)blocksize - 1);
2083                 if (this_bio_flag & EXTENT_BIO_COMPRESSED) {
2084                         disk_io_size = em->block_len;
2085                         sector = em->block_start >> 9;
2086                 } else {
2087                         sector = (em->block_start + extent_offset) >> 9;
2088                         disk_io_size = iosize;
2089                 }
2090                 bdev = em->bdev;
2091                 block_start = em->block_start;
2092                 if (test_bit(EXTENT_FLAG_PREALLOC, &em->flags))
2093                         block_start = EXTENT_MAP_HOLE;
2094                 free_extent_map(em);
2095                 em = NULL;
2096
2097                 /* we've found a hole, just zero and go on */
2098                 if (block_start == EXTENT_MAP_HOLE) {
2099                         char *userpage;
2100                         userpage = kmap_atomic(page, KM_USER0);
2101                         memset(userpage + page_offset, 0, iosize);
2102                         flush_dcache_page(page);
2103                         kunmap_atomic(userpage, KM_USER0);
2104
2105                         set_extent_uptodate(tree, cur, cur + iosize - 1,
2106                                             GFP_NOFS);
2107                         unlock_extent(tree, cur, cur + iosize - 1, GFP_NOFS);
2108                         cur = cur + iosize;
2109                         page_offset += iosize;
2110                         continue;
2111                 }
2112                 /* the get_extent function already copied into the page */
2113                 if (test_range_bit(tree, cur, cur_end,
2114                                    EXTENT_UPTODATE, 1, NULL)) {
2115                         check_page_uptodate(tree, page);
2116                         unlock_extent(tree, cur, cur + iosize - 1, GFP_NOFS);
2117                         cur = cur + iosize;
2118                         page_offset += iosize;
2119                         continue;
2120                 }
2121                 /* we have an inline extent but it didn't get marked up
2122                  * to date.  Error out
2123                  */
2124                 if (block_start == EXTENT_MAP_INLINE) {
2125                         SetPageError(page);
2126                         unlock_extent(tree, cur, cur + iosize - 1, GFP_NOFS);
2127                         cur = cur + iosize;
2128                         page_offset += iosize;
2129                         continue;
2130                 }
2131
2132                 ret = 0;
2133                 if (tree->ops && tree->ops->readpage_io_hook) {
2134                         ret = tree->ops->readpage_io_hook(page, cur,
2135                                                           cur + iosize - 1);
2136                 }
2137                 if (!ret) {
2138                         unsigned long pnr = (last_byte >> PAGE_CACHE_SHIFT) + 1;
2139                         pnr -= page->index;
2140                         ret = submit_extent_page(READ, tree, page,
2141                                          sector, disk_io_size, page_offset,
2142                                          bdev, bio, pnr,
2143                                          end_bio_extent_readpage, mirror_num,
2144                                          *bio_flags,
2145                                          this_bio_flag);
2146                         nr++;
2147                         *bio_flags = this_bio_flag;
2148                 }
2149                 if (ret)
2150                         SetPageError(page);
2151                 cur = cur + iosize;
2152                 page_offset += iosize;
2153         }
2154         if (!nr) {
2155                 if (!PageError(page))
2156                         SetPageUptodate(page);
2157                 unlock_page(page);
2158         }
2159         return 0;
2160 }
2161
2162 int extent_read_full_page(struct extent_io_tree *tree, struct page *page,
2163                             get_extent_t *get_extent)
2164 {
2165         struct bio *bio = NULL;
2166         unsigned long bio_flags = 0;
2167         int ret;
2168
2169         ret = __extent_read_full_page(tree, page, get_extent, &bio, 0,
2170                                       &bio_flags);
2171         if (bio)
2172                 submit_one_bio(READ, bio, 0, bio_flags);
2173         return ret;
2174 }
2175
2176 static noinline void update_nr_written(struct page *page,
2177                                       struct writeback_control *wbc,
2178                                       unsigned long nr_written)
2179 {
2180         wbc->nr_to_write -= nr_written;
2181         if (wbc->range_cyclic || (wbc->nr_to_write > 0 &&
2182             wbc->range_start == 0 && wbc->range_end == LLONG_MAX))
2183                 page->mapping->writeback_index = page->index + nr_written;
2184 }
2185
2186 /*
2187  * the writepage semantics are similar to regular writepage.  extent
2188  * records are inserted to lock ranges in the tree, and as dirty areas
2189  * are found, they are marked writeback.  Then the lock bits are removed
2190  * and the end_io handler clears the writeback ranges
2191  */
2192 static int __extent_writepage(struct page *page, struct writeback_control *wbc,
2193                               void *data)
2194 {
2195         struct inode *inode = page->mapping->host;
2196         struct extent_page_data *epd = data;
2197         struct extent_io_tree *tree = epd->tree;
2198         u64 start = (u64)page->index << PAGE_CACHE_SHIFT;
2199         u64 delalloc_start;
2200         u64 page_end = start + PAGE_CACHE_SIZE - 1;
2201         u64 end;
2202         u64 cur = start;
2203         u64 extent_offset;
2204         u64 last_byte = i_size_read(inode);
2205         u64 block_start;
2206         u64 iosize;
2207         u64 unlock_start;
2208         sector_t sector;
2209         struct extent_state *cached_state = NULL;
2210         struct extent_map *em;
2211         struct block_device *bdev;
2212         int ret;
2213         int nr = 0;
2214         size_t pg_offset = 0;
2215         size_t blocksize;
2216         loff_t i_size = i_size_read(inode);
2217         unsigned long end_index = i_size >> PAGE_CACHE_SHIFT;
2218         u64 nr_delalloc;
2219         u64 delalloc_end;
2220         int page_started;
2221         int compressed;
2222         int write_flags;
2223         unsigned long nr_written = 0;
2224
2225         if (wbc->sync_mode == WB_SYNC_ALL)
2226                 write_flags = WRITE_SYNC_PLUG;
2227         else
2228                 write_flags = WRITE;
2229
2230         WARN_ON(!PageLocked(page));
2231         pg_offset = i_size & (PAGE_CACHE_SIZE - 1);
2232         if (page->index > end_index ||
2233            (page->index == end_index && !pg_offset)) {
2234                 page->mapping->a_ops->invalidatepage(page, 0);
2235                 unlock_page(page);
2236                 return 0;
2237         }
2238
2239         if (page->index == end_index) {
2240                 char *userpage;
2241
2242                 userpage = kmap_atomic(page, KM_USER0);
2243                 memset(userpage + pg_offset, 0,
2244                        PAGE_CACHE_SIZE - pg_offset);
2245                 kunmap_atomic(userpage, KM_USER0);
2246                 flush_dcache_page(page);
2247         }
2248         pg_offset = 0;
2249
2250         set_page_extent_mapped(page);
2251
2252         delalloc_start = start;
2253         delalloc_end = 0;
2254         page_started = 0;
2255         if (!epd->extent_locked) {
2256                 u64 delalloc_to_write = 0;
2257                 /*
2258                  * make sure the wbc mapping index is at least updated
2259                  * to this page.
2260                  */
2261                 update_nr_written(page, wbc, 0);
2262
2263                 while (delalloc_end < page_end) {
2264                         nr_delalloc = find_lock_delalloc_range(inode, tree,
2265                                                        page,
2266                                                        &delalloc_start,
2267                                                        &delalloc_end,
2268                                                        128 * 1024 * 1024);
2269                         if (nr_delalloc == 0) {
2270                                 delalloc_start = delalloc_end + 1;
2271                                 continue;
2272                         }
2273                         tree->ops->fill_delalloc(inode, page, delalloc_start,
2274                                                  delalloc_end, &page_started,
2275                                                  &nr_written);
2276                         /*
2277                          * delalloc_end is already one less than the total
2278                          * length, so we don't subtract one from
2279                          * PAGE_CACHE_SIZE
2280                          */
2281                         delalloc_to_write += (delalloc_end - delalloc_start +
2282                                               PAGE_CACHE_SIZE) >>
2283                                               PAGE_CACHE_SHIFT;
2284                         delalloc_start = delalloc_end + 1;
2285                 }
2286                 if (wbc->nr_to_write < delalloc_to_write) {
2287                         int thresh = 8192;
2288
2289                         if (delalloc_to_write < thresh * 2)
2290                                 thresh = delalloc_to_write;
2291                         wbc->nr_to_write = min_t(u64, delalloc_to_write,
2292                                                  thresh);
2293                 }
2294
2295                 /* did the fill delalloc function already unlock and start
2296                  * the IO?
2297                  */
2298                 if (page_started) {
2299                         ret = 0;
2300                         /*
2301                          * we've unlocked the page, so we can't update
2302                          * the mapping's writeback index, just update
2303                          * nr_to_write.
2304                          */
2305                         wbc->nr_to_write -= nr_written;
2306                         goto done_unlocked;
2307                 }
2308         }
2309         if (tree->ops && tree->ops->writepage_start_hook) {
2310                 ret = tree->ops->writepage_start_hook(page, start,
2311                                                       page_end);
2312                 if (ret == -EAGAIN) {
2313                         redirty_page_for_writepage(wbc, page);
2314                         update_nr_written(page, wbc, nr_written);
2315                         unlock_page(page);
2316                         ret = 0;
2317                         goto done_unlocked;
2318                 }
2319         }
2320
2321         /*
2322          * we don't want to touch the inode after unlocking the page,
2323          * so we update the mapping writeback index now
2324          */
2325         update_nr_written(page, wbc, nr_written + 1);
2326
2327         end = page_end;
2328         if (last_byte <= start) {
2329                 if (tree->ops && tree->ops->writepage_end_io_hook)
2330                         tree->ops->writepage_end_io_hook(page, start,
2331                                                          page_end, NULL, 1);
2332                 unlock_start = page_end + 1;
2333                 goto done;
2334         }
2335
2336         blocksize = inode->i_sb->s_blocksize;
2337
2338         while (cur <= end) {
2339                 if (cur >= last_byte) {
2340                         if (tree->ops && tree->ops->writepage_end_io_hook)
2341                                 tree->ops->writepage_end_io_hook(page, cur,
2342                                                          page_end, NULL, 1);
2343                         unlock_start = page_end + 1;
2344                         break;
2345                 }
2346                 em = epd->get_extent(inode, page, pg_offset, cur,
2347                                      end - cur + 1, 1);
2348                 if (IS_ERR(em) || !em) {
2349                         SetPageError(page);
2350                         break;
2351                 }
2352
2353                 extent_offset = cur - em->start;
2354                 BUG_ON(extent_map_end(em) <= cur);
2355                 BUG_ON(end < cur);
2356                 iosize = min(extent_map_end(em) - cur, end - cur + 1);
2357                 iosize = (iosize + blocksize - 1) & ~((u64)blocksize - 1);
2358                 sector = (em->block_start + extent_offset) >> 9;
2359                 bdev = em->bdev;
2360                 block_start = em->block_start;
2361                 compressed = test_bit(EXTENT_FLAG_COMPRESSED, &em->flags);
2362                 free_extent_map(em);
2363                 em = NULL;
2364
2365                 /*
2366                  * compressed and inline extents are written through other
2367                  * paths in the FS
2368                  */
2369                 if (compressed || block_start == EXTENT_MAP_HOLE ||
2370                     block_start == EXTENT_MAP_INLINE) {
2371                         /*
2372                          * end_io notification does not happen here for
2373                          * compressed extents
2374                          */
2375                         if (!compressed && tree->ops &&
2376                             tree->ops->writepage_end_io_hook)
2377                                 tree->ops->writepage_end_io_hook(page, cur,
2378                                                          cur + iosize - 1,
2379                                                          NULL, 1);
2380                         else if (compressed) {
2381                                 /* we don't want to end_page_writeback on
2382                                  * a compressed extent.  this happens
2383                                  * elsewhere
2384                                  */
2385                                 nr++;
2386                         }
2387
2388                         cur += iosize;
2389                         pg_offset += iosize;
2390                         unlock_start = cur;
2391                         continue;
2392                 }
2393                 /* leave this out until we have a page_mkwrite call */
2394                 if (0 && !test_range_bit(tree, cur, cur + iosize - 1,
2395                                    EXTENT_DIRTY, 0, NULL)) {
2396                         cur = cur + iosize;
2397                         pg_offset += iosize;
2398                         continue;
2399                 }
2400
2401                 if (tree->ops && tree->ops->writepage_io_hook) {
2402                         ret = tree->ops->writepage_io_hook(page, cur,
2403                                                 cur + iosize - 1);
2404                 } else {
2405                         ret = 0;
2406                 }
2407                 if (ret) {
2408                         SetPageError(page);
2409                 } else {
2410                         unsigned long max_nr = end_index + 1;
2411
2412                         set_range_writeback(tree, cur, cur + iosize - 1);
2413                         if (!PageWriteback(page)) {
2414                                 printk(KERN_ERR "btrfs warning page %lu not "
2415                                        "writeback, cur %llu end %llu\n",
2416                                        page->index, (unsigned long long)cur,
2417                                        (unsigned long long)end);
2418                         }
2419
2420                         ret = submit_extent_page(write_flags, tree, page,
2421                                                  sector, iosize, pg_offset,
2422                                                  bdev, &epd->bio, max_nr,
2423                                                  end_bio_extent_writepage,
2424                                                  0, 0, 0);
2425                         if (ret)
2426                                 SetPageError(page);
2427                 }
2428                 cur = cur + iosize;
2429                 pg_offset += iosize;
2430                 nr++;
2431         }
2432 done:
2433         if (nr == 0) {
2434                 /* make sure the mapping tag for page dirty gets cleared */
2435                 set_page_writeback(page);
2436                 end_page_writeback(page);
2437         }
2438         unlock_page(page);
2439
2440 done_unlocked:
2441
2442         /* drop our reference on any cached states */
2443         free_extent_state(cached_state);
2444         return 0;
2445 }
2446
2447 /**
2448  * write_cache_pages - walk the list of dirty pages of the given address space and write all of them.
2449  * @mapping: address space structure to write
2450  * @wbc: subtract the number of written pages from *@wbc->nr_to_write
2451  * @writepage: function called for each page
2452  * @data: data passed to writepage function
2453  *
2454  * If a page is already under I/O, write_cache_pages() skips it, even
2455  * if it's dirty.  This is desirable behaviour for memory-cleaning writeback,
2456  * but it is INCORRECT for data-integrity system calls such as fsync().  fsync()
2457  * and msync() need to guarantee that all the data which was dirty at the time
2458  * the call was made get new I/O started against them.  If wbc->sync_mode is
2459  * WB_SYNC_ALL then we were called for data integrity and we must wait for
2460  * existing IO to complete.
2461  */
2462 static int extent_write_cache_pages(struct extent_io_tree *tree,
2463                              struct address_space *mapping,
2464                              struct writeback_control *wbc,
2465                              writepage_t writepage, void *data,
2466                              void (*flush_fn)(void *))
2467 {
2468         int ret = 0;
2469         int done = 0;
2470         int nr_to_write_done = 0;
2471         struct pagevec pvec;
2472         int nr_pages;
2473         pgoff_t index;
2474         pgoff_t end;            /* Inclusive */
2475         int scanned = 0;
2476         int range_whole = 0;
2477
2478         pagevec_init(&pvec, 0);
2479         if (wbc->range_cyclic) {
2480                 index = mapping->writeback_index; /* Start from prev offset */
2481                 end = -1;
2482         } else {
2483                 index = wbc->range_start >> PAGE_CACHE_SHIFT;
2484                 end = wbc->range_end >> PAGE_CACHE_SHIFT;
2485                 if (wbc->range_start == 0 && wbc->range_end == LLONG_MAX)
2486                         range_whole = 1;
2487                 scanned = 1;
2488         }
2489 retry:
2490         while (!done && !nr_to_write_done && (index <= end) &&
2491                (nr_pages = pagevec_lookup_tag(&pvec, mapping, &index,
2492                               PAGECACHE_TAG_DIRTY, min(end - index,
2493                                   (pgoff_t)PAGEVEC_SIZE-1) + 1))) {
2494                 unsigned i;
2495
2496                 scanned = 1;
2497                 for (i = 0; i < nr_pages; i++) {
2498                         struct page *page = pvec.pages[i];
2499
2500                         /*
2501                          * At this point we hold neither mapping->tree_lock nor
2502                          * lock on the page itself: the page may be truncated or
2503                          * invalidated (changing page->mapping to NULL), or even
2504                          * swizzled back from swapper_space to tmpfs file
2505                          * mapping
2506                          */
2507                         if (tree->ops && tree->ops->write_cache_pages_lock_hook)
2508                                 tree->ops->write_cache_pages_lock_hook(page);
2509                         else
2510                                 lock_page(page);
2511
2512                         if (unlikely(page->mapping != mapping)) {
2513                                 unlock_page(page);
2514                                 continue;
2515                         }
2516
2517                         if (!wbc->range_cyclic && page->index > end) {
2518                                 done = 1;
2519                                 unlock_page(page);
2520                                 continue;
2521                         }
2522
2523                         if (wbc->sync_mode != WB_SYNC_NONE) {
2524                                 if (PageWriteback(page))
2525                                         flush_fn(data);
2526                                 wait_on_page_writeback(page);
2527                         }
2528
2529                         if (PageWriteback(page) ||
2530                             !clear_page_dirty_for_io(page)) {
2531                                 unlock_page(page);
2532                                 continue;
2533                         }
2534
2535                         ret = (*writepage)(page, wbc, data);
2536
2537                         if (unlikely(ret == AOP_WRITEPAGE_ACTIVATE)) {
2538                                 unlock_page(page);
2539                                 ret = 0;
2540                         }
2541                         if (ret)
2542                                 done = 1;
2543
2544                         /*
2545                          * the filesystem may choose to bump up nr_to_write.
2546                          * We have to make sure to honor the new nr_to_write
2547                          * at any time
2548                          */
2549                         nr_to_write_done = wbc->nr_to_write <= 0;
2550                 }
2551                 pagevec_release(&pvec);
2552                 cond_resched();
2553         }
2554         if (!scanned && !done) {
2555                 /*
2556                  * We hit the last page and there is more work to be done: wrap
2557                  * back to the start of the file
2558                  */
2559                 scanned = 1;
2560                 index = 0;
2561                 goto retry;
2562         }
2563         return ret;
2564 }
2565
2566 static void flush_epd_write_bio(struct extent_page_data *epd)
2567 {
2568         if (epd->bio) {
2569                 if (epd->sync_io)
2570                         submit_one_bio(WRITE_SYNC, epd->bio, 0, 0);
2571                 else
2572                         submit_one_bio(WRITE, epd->bio, 0, 0);
2573                 epd->bio = NULL;
2574         }
2575 }
2576
2577 static noinline void flush_write_bio(void *data)
2578 {
2579         struct extent_page_data *epd = data;
2580         flush_epd_write_bio(epd);
2581 }
2582
2583 int extent_write_full_page(struct extent_io_tree *tree, struct page *page,
2584                           get_extent_t *get_extent,
2585                           struct writeback_control *wbc)
2586 {
2587         int ret;
2588         struct address_space *mapping = page->mapping;
2589         struct extent_page_data epd = {
2590                 .bio = NULL,
2591                 .tree = tree,
2592                 .get_extent = get_extent,
2593                 .extent_locked = 0,
2594                 .sync_io = wbc->sync_mode == WB_SYNC_ALL,
2595         };
2596         struct writeback_control wbc_writepages = {
2597                 .sync_mode      = wbc->sync_mode,
2598                 .older_than_this = NULL,
2599                 .nr_to_write    = 64,
2600                 .range_start    = page_offset(page) + PAGE_CACHE_SIZE,
2601                 .range_end      = (loff_t)-1,
2602         };
2603
2604         ret = __extent_writepage(page, wbc, &epd);
2605
2606         extent_write_cache_pages(tree, mapping, &wbc_writepages,
2607                                  __extent_writepage, &epd, flush_write_bio);
2608         flush_epd_write_bio(&epd);
2609         return ret;
2610 }
2611
2612 int extent_write_locked_range(struct extent_io_tree *tree, struct inode *inode,
2613                               u64 start, u64 end, get_extent_t *get_extent,
2614                               int mode)
2615 {
2616         int ret = 0;
2617         struct address_space *mapping = inode->i_mapping;
2618         struct page *page;
2619         unsigned long nr_pages = (end - start + PAGE_CACHE_SIZE) >>
2620                 PAGE_CACHE_SHIFT;
2621
2622         struct extent_page_data epd = {
2623                 .bio = NULL,
2624                 .tree = tree,
2625                 .get_extent = get_extent,
2626                 .extent_locked = 1,
2627                 .sync_io = mode == WB_SYNC_ALL,
2628         };
2629         struct writeback_control wbc_writepages = {
2630                 .sync_mode      = mode,
2631                 .older_than_this = NULL,
2632                 .nr_to_write    = nr_pages * 2,
2633                 .range_start    = start,
2634                 .range_end      = end + 1,
2635         };
2636
2637         while (start <= end) {
2638                 page = find_get_page(mapping, start >> PAGE_CACHE_SHIFT);
2639                 if (clear_page_dirty_for_io(page))
2640                         ret = __extent_writepage(page, &wbc_writepages, &epd);
2641                 else {
2642                         if (tree->ops && tree->ops->writepage_end_io_hook)
2643                                 tree->ops->writepage_end_io_hook(page, start,
2644                                                  start + PAGE_CACHE_SIZE - 1,
2645                                                  NULL, 1);
2646                         unlock_page(page);
2647                 }
2648                 page_cache_release(page);
2649                 start += PAGE_CACHE_SIZE;
2650         }
2651
2652         flush_epd_write_bio(&epd);
2653         return ret;
2654 }
2655
2656 int extent_writepages(struct extent_io_tree *tree,
2657                       struct address_space *mapping,
2658                       get_extent_t *get_extent,
2659                       struct writeback_control *wbc)
2660 {
2661         int ret = 0;
2662         struct extent_page_data epd = {
2663                 .bio = NULL,
2664                 .tree = tree,
2665                 .get_extent = get_extent,
2666                 .extent_locked = 0,
2667                 .sync_io = wbc->sync_mode == WB_SYNC_ALL,
2668         };
2669
2670         ret = extent_write_cache_pages(tree, mapping, wbc,
2671                                        __extent_writepage, &epd,
2672                                        flush_write_bio);
2673         flush_epd_write_bio(&epd);
2674         return ret;
2675 }
2676
2677 int extent_readpages(struct extent_io_tree *tree,
2678                      struct address_space *mapping,
2679                      struct list_head *pages, unsigned nr_pages,
2680                      get_extent_t get_extent)
2681 {
2682         struct bio *bio = NULL;
2683         unsigned page_idx;
2684         unsigned long bio_flags = 0;
2685
2686         for (page_idx = 0; page_idx < nr_pages; page_idx++) {
2687                 struct page *page = list_entry(pages->prev, struct page, lru);
2688
2689                 prefetchw(&page->flags);
2690                 list_del(&page->lru);
2691                 if (!add_to_page_cache_lru(page, mapping,
2692                                         page->index, GFP_KERNEL)) {
2693                         __extent_read_full_page(tree, page, get_extent,
2694                                                 &bio, 0, &bio_flags);
2695                 }
2696                 page_cache_release(page);
2697         }
2698         BUG_ON(!list_empty(pages));
2699         if (bio)
2700                 submit_one_bio(READ, bio, 0, bio_flags);
2701         return 0;
2702 }
2703
2704 /*
2705  * basic invalidatepage code, this waits on any locked or writeback
2706  * ranges corresponding to the page, and then deletes any extent state
2707  * records from the tree
2708  */
2709 int extent_invalidatepage(struct extent_io_tree *tree,
2710                           struct page *page, unsigned long offset)
2711 {
2712         struct extent_state *cached_state = NULL;
2713         u64 start = ((u64)page->index << PAGE_CACHE_SHIFT);
2714         u64 end = start + PAGE_CACHE_SIZE - 1;
2715         size_t blocksize = page->mapping->host->i_sb->s_blocksize;
2716
2717         start += (offset + blocksize - 1) & ~(blocksize - 1);
2718         if (start > end)
2719                 return 0;
2720
2721         lock_extent_bits(tree, start, end, 0, &cached_state, GFP_NOFS);
2722         wait_on_page_writeback(page);
2723         clear_extent_bit(tree, start, end,
2724                          EXTENT_LOCKED | EXTENT_DIRTY | EXTENT_DELALLOC |
2725                          EXTENT_DO_ACCOUNTING,
2726                          1, 1, &cached_state, GFP_NOFS);
2727         return 0;
2728 }
2729
2730 /*
2731  * simple commit_write call, set_range_dirty is used to mark both
2732  * the pages and the extent records as dirty
2733  */
2734 int extent_commit_write(struct extent_io_tree *tree,
2735                         struct inode *inode, struct page *page,
2736                         unsigned from, unsigned to)
2737 {
2738         loff_t pos = ((loff_t)page->index << PAGE_CACHE_SHIFT) + to;
2739
2740         set_page_extent_mapped(page);
2741         set_page_dirty(page);
2742
2743         if (pos > inode->i_size) {
2744                 i_size_write(inode, pos);
2745                 mark_inode_dirty(inode);
2746         }
2747         return 0;
2748 }
2749
2750 int extent_prepare_write(struct extent_io_tree *tree,
2751                          struct inode *inode, struct page *page,
2752                          unsigned from, unsigned to, get_extent_t *get_extent)
2753 {
2754         u64 page_start = (u64)page->index << PAGE_CACHE_SHIFT;
2755         u64 page_end = page_start + PAGE_CACHE_SIZE - 1;
2756         u64 block_start;
2757         u64 orig_block_start;
2758         u64 block_end;
2759         u64 cur_end;
2760         struct extent_map *em;
2761         unsigned blocksize = 1 << inode->i_blkbits;
2762         size_t page_offset = 0;
2763         size_t block_off_start;
2764         size_t block_off_end;
2765         int err = 0;
2766         int iocount = 0;
2767         int ret = 0;
2768         int isnew;
2769
2770         set_page_extent_mapped(page);
2771
2772         block_start = (page_start + from) & ~((u64)blocksize - 1);
2773         block_end = (page_start + to - 1) | (blocksize - 1);
2774         orig_block_start = block_start;
2775
2776         lock_extent(tree, page_start, page_end, GFP_NOFS);
2777         while (block_start <= block_end) {
2778                 em = get_extent(inode, page, page_offset, block_start,
2779                                 block_end - block_start + 1, 1);
2780                 if (IS_ERR(em) || !em)
2781                         goto err;
2782
2783                 cur_end = min(block_end, extent_map_end(em) - 1);
2784                 block_off_start = block_start & (PAGE_CACHE_SIZE - 1);
2785                 block_off_end = block_off_start + blocksize;
2786                 isnew = clear_extent_new(tree, block_start, cur_end, GFP_NOFS);
2787
2788                 if (!PageUptodate(page) && isnew &&
2789                     (block_off_end > to || block_off_start < from)) {
2790                         void *kaddr;
2791
2792                         kaddr = kmap_atomic(page, KM_USER0);
2793                         if (block_off_end > to)
2794                                 memset(kaddr + to, 0, block_off_end - to);
2795                         if (block_off_start < from)
2796                                 memset(kaddr + block_off_start, 0,
2797                                        from - block_off_start);
2798                         flush_dcache_page(page);
2799                         kunmap_atomic(kaddr, KM_USER0);
2800                 }
2801                 if ((em->block_start != EXTENT_MAP_HOLE &&
2802                      em->block_start != EXTENT_MAP_INLINE) &&
2803                     !isnew && !PageUptodate(page) &&
2804                     (block_off_end > to || block_off_start < from) &&
2805                     !test_range_bit(tree, block_start, cur_end,
2806                                     EXTENT_UPTODATE, 1, NULL)) {
2807                         u64 sector;
2808                         u64 extent_offset = block_start - em->start;
2809                         size_t iosize;
2810                         sector = (em->block_start + extent_offset) >> 9;
2811                         iosize = (cur_end - block_start + blocksize) &
2812                                 ~((u64)blocksize - 1);
2813                         /*
2814                          * we've already got the extent locked, but we
2815                          * need to split the state such that our end_bio
2816                          * handler can clear the lock.
2817                          */
2818                         set_extent_bit(tree, block_start,
2819                                        block_start + iosize - 1,
2820                                        EXTENT_LOCKED, 0, NULL, NULL, GFP_NOFS);
2821                         ret = submit_extent_page(READ, tree, page,
2822                                          sector, iosize, page_offset, em->bdev,
2823                                          NULL, 1,
2824                                          end_bio_extent_preparewrite, 0,
2825                                          0, 0);
2826                         iocount++;
2827                         block_start = block_start + iosize;
2828                 } else {
2829                         set_extent_uptodate(tree, block_start, cur_end,
2830                                             GFP_NOFS);
2831                         unlock_extent(tree, block_start, cur_end, GFP_NOFS);
2832                         block_start = cur_end + 1;
2833                 }
2834                 page_offset = block_start & (PAGE_CACHE_SIZE - 1);
2835                 free_extent_map(em);
2836         }
2837         if (iocount) {
2838                 wait_extent_bit(tree, orig_block_start,
2839                                 block_end, EXTENT_LOCKED);
2840         }
2841         check_page_uptodate(tree, page);
2842 err:
2843         /* FIXME, zero out newly allocated blocks on error */
2844         return err;
2845 }
2846
2847 /*
2848  * a helper for releasepage, this tests for areas of the page that
2849  * are locked or under IO and drops the related state bits if it is safe
2850  * to drop the page.
2851  */
2852 int try_release_extent_state(struct extent_map_tree *map,
2853                              struct extent_io_tree *tree, struct page *page,
2854                              gfp_t mask)
2855 {
2856         u64 start = (u64)page->index << PAGE_CACHE_SHIFT;
2857         u64 end = start + PAGE_CACHE_SIZE - 1;
2858         int ret = 1;
2859
2860         if (test_range_bit(tree, start, end,
2861                            EXTENT_IOBITS, 0, NULL))
2862                 ret = 0;
2863         else {
2864                 if ((mask & GFP_NOFS) == GFP_NOFS)
2865                         mask = GFP_NOFS;
2866                 /*
2867                  * at this point we can safely clear everything except the
2868                  * locked bit and the nodatasum bit
2869                  */
2870                 clear_extent_bit(tree, start, end,
2871                                  ~(EXTENT_LOCKED | EXTENT_NODATASUM),
2872                                  0, 0, NULL, mask);
2873         }
2874         return ret;
2875 }
2876
2877 /*
2878  * a helper for releasepage.  As long as there are no locked extents
2879  * in the range corresponding to the page, both state records and extent
2880  * map records are removed
2881  */
2882 int try_release_extent_mapping(struct extent_map_tree *map,
2883                                struct extent_io_tree *tree, struct page *page,
2884                                gfp_t mask)
2885 {
2886         struct extent_map *em;
2887         u64 start = (u64)page->index << PAGE_CACHE_SHIFT;
2888         u64 end = start + PAGE_CACHE_SIZE - 1;
2889
2890         if ((mask & __GFP_WAIT) &&
2891             page->mapping->host->i_size > 16 * 1024 * 1024) {
2892                 u64 len;
2893                 while (start <= end) {
2894                         len = end - start + 1;
2895                         write_lock(&map->lock);
2896                         em = lookup_extent_mapping(map, start, len);
2897                         if (!em || IS_ERR(em)) {
2898                                 write_unlock(&map->lock);
2899                                 break;
2900                         }
2901                         if (test_bit(EXTENT_FLAG_PINNED, &em->flags) ||
2902                             em->start != start) {
2903                                 write_unlock(&map->lock);
2904                                 free_extent_map(em);
2905                                 break;
2906                         }
2907                         if (!test_range_bit(tree, em->start,
2908                                             extent_map_end(em) - 1,
2909                                             EXTENT_LOCKED | EXTENT_WRITEBACK,
2910                                             0, NULL)) {
2911                                 remove_extent_mapping(map, em);
2912                                 /* once for the rb tree */
2913                                 free_extent_map(em);
2914                         }
2915                         start = extent_map_end(em);
2916                         write_unlock(&map->lock);
2917
2918                         /* once for us */
2919                         free_extent_map(em);
2920                 }
2921         }
2922         return try_release_extent_state(map, tree, page, mask);
2923 }
2924
2925 sector_t extent_bmap(struct address_space *mapping, sector_t iblock,
2926                 get_extent_t *get_extent)
2927 {
2928         struct inode *inode = mapping->host;
2929         struct extent_state *cached_state = NULL;
2930         u64 start = iblock << inode->i_blkbits;
2931         sector_t sector = 0;
2932         size_t blksize = (1 << inode->i_blkbits);
2933         struct extent_map *em;
2934
2935         lock_extent_bits(&BTRFS_I(inode)->io_tree, start, start + blksize - 1,
2936                          0, &cached_state, GFP_NOFS);
2937         em = get_extent(inode, NULL, 0, start, blksize, 0);
2938         unlock_extent_cached(&BTRFS_I(inode)->io_tree, start,
2939                              start + blksize - 1, &cached_state, GFP_NOFS);
2940         if (!em || IS_ERR(em))
2941                 return 0;
2942
2943         if (em->block_start > EXTENT_MAP_LAST_BYTE)
2944                 goto out;
2945
2946         sector = (em->block_start + start - em->start) >> inode->i_blkbits;
2947 out:
2948         free_extent_map(em);
2949         return sector;
2950 }
2951
2952 int extent_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
2953                 __u64 start, __u64 len, get_extent_t *get_extent)
2954 {
2955         int ret;
2956         u64 off = start;
2957         u64 max = start + len;
2958         u32 flags = 0;
2959         u64 disko = 0;
2960         struct extent_map *em = NULL;
2961         struct extent_state *cached_state = NULL;
2962         int end = 0;
2963         u64 em_start = 0, em_len = 0;
2964         unsigned long emflags;
2965         ret = 0;
2966
2967         if (len == 0)
2968                 return -EINVAL;
2969
2970         lock_extent_bits(&BTRFS_I(inode)->io_tree, start, start + len, 0,
2971                          &cached_state, GFP_NOFS);
2972         em = get_extent(inode, NULL, 0, off, max - off, 0);
2973         if (!em)
2974                 goto out;
2975         if (IS_ERR(em)) {
2976                 ret = PTR_ERR(em);
2977                 goto out;
2978         }
2979         while (!end) {
2980                 off = em->start + em->len;
2981                 if (off >= max)
2982                         end = 1;
2983
2984                 em_start = em->start;
2985                 em_len = em->len;
2986
2987                 disko = 0;
2988                 flags = 0;
2989
2990                 if (em->block_start == EXTENT_MAP_LAST_BYTE) {
2991                         end = 1;
2992                         flags |= FIEMAP_EXTENT_LAST;
2993                 } else if (em->block_start == EXTENT_MAP_HOLE) {
2994                         flags |= FIEMAP_EXTENT_UNWRITTEN;
2995                 } else if (em->block_start == EXTENT_MAP_INLINE) {
2996                         flags |= (FIEMAP_EXTENT_DATA_INLINE |
2997                                   FIEMAP_EXTENT_NOT_ALIGNED);
2998                 } else if (em->block_start == EXTENT_MAP_DELALLOC) {
2999                         flags |= (FIEMAP_EXTENT_DELALLOC |
3000                                   FIEMAP_EXTENT_UNKNOWN);
3001                 } else {
3002                         disko = em->block_start;
3003                 }
3004                 if (test_bit(EXTENT_FLAG_COMPRESSED, &em->flags))
3005                         flags |= FIEMAP_EXTENT_ENCODED;
3006
3007                 emflags = em->flags;
3008                 free_extent_map(em);
3009                 em = NULL;
3010
3011                 if (!end) {
3012                         em = get_extent(inode, NULL, 0, off, max - off, 0);
3013                         if (!em)
3014                                 goto out;
3015                         if (IS_ERR(em)) {
3016                                 ret = PTR_ERR(em);
3017                                 goto out;
3018                         }
3019                         emflags = em->flags;
3020                 }
3021                 if (test_bit(EXTENT_FLAG_VACANCY, &emflags)) {
3022                         flags |= FIEMAP_EXTENT_LAST;
3023                         end = 1;
3024                 }
3025
3026                 ret = fiemap_fill_next_extent(fieinfo, em_start, disko,
3027                                         em_len, flags);
3028                 if (ret)
3029                         goto out_free;
3030         }
3031 out_free:
3032         free_extent_map(em);
3033 out:
3034         unlock_extent_cached(&BTRFS_I(inode)->io_tree, start, start + len,
3035                              &cached_state, GFP_NOFS);
3036         return ret;
3037 }
3038
3039 static inline struct page *extent_buffer_page(struct extent_buffer *eb,
3040                                               unsigned long i)
3041 {
3042         struct page *p;
3043         struct address_space *mapping;
3044
3045         if (i == 0)
3046                 return eb->first_page;
3047         i += eb->start >> PAGE_CACHE_SHIFT;
3048         mapping = eb->first_page->mapping;
3049         if (!mapping)
3050                 return NULL;
3051
3052         /*
3053          * extent_buffer_page is only called after pinning the page
3054          * by increasing the reference count.  So we know the page must
3055          * be in the radix tree.
3056          */
3057         rcu_read_lock();
3058         p = radix_tree_lookup(&mapping->page_tree, i);
3059         rcu_read_unlock();
3060
3061         return p;
3062 }
3063
3064 static inline unsigned long num_extent_pages(u64 start, u64 len)
3065 {
3066         return ((start + len + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT) -
3067                 (start >> PAGE_CACHE_SHIFT);
3068 }
3069
3070 static struct extent_buffer *__alloc_extent_buffer(struct extent_io_tree *tree,
3071                                                    u64 start,
3072                                                    unsigned long len,
3073                                                    gfp_t mask)
3074 {
3075         struct extent_buffer *eb = NULL;
3076 #if LEAK_DEBUG
3077         unsigned long flags;
3078 #endif
3079
3080         eb = kmem_cache_zalloc(extent_buffer_cache, mask);
3081         eb->start = start;
3082         eb->len = len;
3083         spin_lock_init(&eb->lock);
3084         init_waitqueue_head(&eb->lock_wq);
3085
3086 #if LEAK_DEBUG
3087         spin_lock_irqsave(&leak_lock, flags);
3088         list_add(&eb->leak_list, &buffers);
3089         spin_unlock_irqrestore(&leak_lock, flags);
3090 #endif
3091         atomic_set(&eb->refs, 1);
3092
3093         return eb;
3094 }
3095
3096 static void __free_extent_buffer(struct extent_buffer *eb)
3097 {
3098 #if LEAK_DEBUG
3099         unsigned long flags;
3100         spin_lock_irqsave(&leak_lock, flags);
3101         list_del(&eb->leak_list);
3102         spin_unlock_irqrestore(&leak_lock, flags);
3103 #endif
3104         kmem_cache_free(extent_buffer_cache, eb);
3105 }
3106
3107 struct extent_buffer *alloc_extent_buffer(struct extent_io_tree *tree,
3108                                           u64 start, unsigned long len,
3109                                           struct page *page0,
3110                                           gfp_t mask)
3111 {
3112         unsigned long num_pages = num_extent_pages(start, len);
3113         unsigned long i;
3114         unsigned long index = start >> PAGE_CACHE_SHIFT;
3115         struct extent_buffer *eb;
3116         struct extent_buffer *exists = NULL;
3117         struct page *p;
3118         struct address_space *mapping = tree->mapping;
3119         int uptodate = 1;
3120
3121         spin_lock(&tree->buffer_lock);
3122         eb = buffer_search(tree, start);
3123         if (eb) {
3124                 atomic_inc(&eb->refs);
3125                 spin_unlock(&tree->buffer_lock);
3126                 mark_page_accessed(eb->first_page);
3127                 return eb;
3128         }
3129         spin_unlock(&tree->buffer_lock);
3130
3131         eb = __alloc_extent_buffer(tree, start, len, mask);
3132         if (!eb)
3133                 return NULL;
3134
3135         if (page0) {
3136                 eb->first_page = page0;
3137                 i = 1;
3138                 index++;
3139                 page_cache_get(page0);
3140                 mark_page_accessed(page0);
3141                 set_page_extent_mapped(page0);
3142                 set_page_extent_head(page0, len);
3143                 uptodate = PageUptodate(page0);
3144         } else {
3145                 i = 0;
3146         }
3147         for (; i < num_pages; i++, index++) {
3148                 p = find_or_create_page(mapping, index, mask | __GFP_HIGHMEM);
3149                 if (!p) {
3150                         WARN_ON(1);
3151                         goto free_eb;
3152                 }
3153                 set_page_extent_mapped(p);
3154                 mark_page_accessed(p);
3155                 if (i == 0) {
3156                         eb->first_page = p;
3157                         set_page_extent_head(p, len);
3158                 } else {
3159                         set_page_private(p, EXTENT_PAGE_PRIVATE);
3160                 }
3161                 if (!PageUptodate(p))
3162                         uptodate = 0;
3163                 unlock_page(p);
3164         }
3165         if (uptodate)
3166                 set_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags);
3167
3168         spin_lock(&tree->buffer_lock);
3169         exists = buffer_tree_insert(tree, start, &eb->rb_node);
3170         if (exists) {
3171                 /* add one reference for the caller */
3172                 atomic_inc(&exists->refs);
3173                 spin_unlock(&tree->buffer_lock);
3174                 goto free_eb;
3175         }
3176         /* add one reference for the tree */
3177         atomic_inc(&eb->refs);
3178         spin_unlock(&tree->buffer_lock);
3179         return eb;
3180
3181 free_eb:
3182         if (!atomic_dec_and_test(&eb->refs))
3183                 return exists;
3184         for (index = 1; index < i; index++)
3185                 page_cache_release(extent_buffer_page(eb, index));
3186         page_cache_release(extent_buffer_page(eb, 0));
3187         __free_extent_buffer(eb);
3188         return exists;
3189 }
3190
3191 struct extent_buffer *find_extent_buffer(struct extent_io_tree *tree,
3192                                          u64 start, unsigned long len,
3193                                           gfp_t mask)
3194 {
3195         struct extent_buffer *eb;
3196
3197         spin_lock(&tree->buffer_lock);
3198         eb = buffer_search(tree, start);
3199         if (eb)
3200                 atomic_inc(&eb->refs);
3201         spin_unlock(&tree->buffer_lock);
3202
3203         if (eb)
3204                 mark_page_accessed(eb->first_page);
3205
3206         return eb;
3207 }
3208
3209 void free_extent_buffer(struct extent_buffer *eb)
3210 {
3211         if (!eb)
3212                 return;
3213
3214         if (!atomic_dec_and_test(&eb->refs))
3215                 return;
3216
3217         WARN_ON(1);
3218 }
3219
3220 int clear_extent_buffer_dirty(struct extent_io_tree *tree,
3221                               struct extent_buffer *eb)
3222 {
3223         unsigned long i;
3224         unsigned long num_pages;
3225         struct page *page;
3226
3227         num_pages = num_extent_pages(eb->start, eb->len);
3228
3229         for (i = 0; i < num_pages; i++) {
3230                 page = extent_buffer_page(eb, i);
3231                 if (!PageDirty(page))
3232                         continue;
3233
3234                 lock_page(page);
3235                 if (i == 0)
3236                         set_page_extent_head(page, eb->len);
3237                 else
3238                         set_page_private(page, EXTENT_PAGE_PRIVATE);
3239
3240                 clear_page_dirty_for_io(page);
3241                 spin_lock_irq(&page->mapping->tree_lock);
3242                 if (!PageDirty(page)) {
3243                         radix_tree_tag_clear(&page->mapping->page_tree,
3244                                                 page_index(page),
3245                                                 PAGECACHE_TAG_DIRTY);
3246                 }
3247                 spin_unlock_irq(&page->mapping->tree_lock);
3248                 unlock_page(page);
3249         }
3250         return 0;
3251 }
3252
3253 int wait_on_extent_buffer_writeback(struct extent_io_tree *tree,
3254                                     struct extent_buffer *eb)
3255 {
3256         return wait_on_extent_writeback(tree, eb->start,
3257                                         eb->start + eb->len - 1);
3258 }
3259
3260 int set_extent_buffer_dirty(struct extent_io_tree *tree,
3261                              struct extent_buffer *eb)
3262 {
3263         unsigned long i;
3264         unsigned long num_pages;
3265         int was_dirty = 0;
3266
3267         was_dirty = test_and_set_bit(EXTENT_BUFFER_DIRTY, &eb->bflags);
3268         num_pages = num_extent_pages(eb->start, eb->len);
3269         for (i = 0; i < num_pages; i++)
3270                 __set_page_dirty_nobuffers(extent_buffer_page(eb, i));
3271         return was_dirty;
3272 }
3273
3274 int clear_extent_buffer_uptodate(struct extent_io_tree *tree,
3275                                 struct extent_buffer *eb,
3276                                 struct extent_state **cached_state)
3277 {
3278         unsigned long i;
3279         struct page *page;
3280         unsigned long num_pages;
3281
3282         num_pages = num_extent_pages(eb->start, eb->len);
3283         clear_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags);
3284
3285         clear_extent_uptodate(tree, eb->start, eb->start + eb->len - 1,
3286                               cached_state, GFP_NOFS);
3287         for (i = 0; i < num_pages; i++) {
3288                 page = extent_buffer_page(eb, i);
3289                 if (page)
3290                         ClearPageUptodate(page);
3291         }
3292         return 0;
3293 }
3294
3295 int set_extent_buffer_uptodate(struct extent_io_tree *tree,
3296                                 struct extent_buffer *eb)
3297 {
3298         unsigned long i;
3299         struct page *page;
3300         unsigned long num_pages;
3301
3302         num_pages = num_extent_pages(eb->start, eb->len);
3303
3304         set_extent_uptodate(tree, eb->start, eb->start + eb->len - 1,
3305                             GFP_NOFS);
3306         for (i = 0; i < num_pages; i++) {
3307                 page = extent_buffer_page(eb, i);
3308                 if ((i == 0 && (eb->start & (PAGE_CACHE_SIZE - 1))) ||
3309                     ((i == num_pages - 1) &&
3310                      ((eb->start + eb->len) & (PAGE_CACHE_SIZE - 1)))) {
3311                         check_page_uptodate(tree, page);
3312                         continue;
3313                 }
3314                 SetPageUptodate(page);
3315         }
3316         return 0;
3317 }
3318
3319 int extent_range_uptodate(struct extent_io_tree *tree,
3320                           u64 start, u64 end)
3321 {
3322         struct page *page;
3323         int ret;
3324         int pg_uptodate = 1;
3325         int uptodate;
3326         unsigned long index;
3327
3328         ret = test_range_bit(tree, start, end, EXTENT_UPTODATE, 1, NULL);
3329         if (ret)
3330                 return 1;
3331         while (start <= end) {
3332                 index = start >> PAGE_CACHE_SHIFT;
3333                 page = find_get_page(tree->mapping, index);
3334                 uptodate = PageUptodate(page);
3335                 page_cache_release(page);
3336                 if (!uptodate) {
3337                         pg_uptodate = 0;
3338                         break;
3339                 }
3340                 start += PAGE_CACHE_SIZE;
3341         }
3342         return pg_uptodate;
3343 }
3344
3345 int extent_buffer_uptodate(struct extent_io_tree *tree,
3346                            struct extent_buffer *eb,
3347                            struct extent_state *cached_state)
3348 {
3349         int ret = 0;
3350         unsigned long num_pages;
3351         unsigned long i;
3352         struct page *page;
3353         int pg_uptodate = 1;
3354
3355         if (test_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags))
3356                 return 1;
3357
3358         ret = test_range_bit(tree, eb->start, eb->start + eb->len - 1,
3359                            EXTENT_UPTODATE, 1, cached_state);
3360         if (ret)
3361                 return ret;
3362
3363         num_pages = num_extent_pages(eb->start, eb->len);
3364         for (i = 0; i < num_pages; i++) {
3365                 page = extent_buffer_page(eb, i);
3366                 if (!PageUptodate(page)) {
3367                         pg_uptodate = 0;
3368                         break;
3369                 }
3370         }
3371         return pg_uptodate;
3372 }
3373
3374 int read_extent_buffer_pages(struct extent_io_tree *tree,
3375                              struct extent_buffer *eb,
3376                              u64 start, int wait,
3377                              get_extent_t *get_extent, int mirror_num)
3378 {
3379         unsigned long i;
3380         unsigned long start_i;
3381         struct page *page;
3382         int err;
3383         int ret = 0;
3384         int locked_pages = 0;
3385         int all_uptodate = 1;
3386         int inc_all_pages = 0;
3387         unsigned long num_pages;
3388         struct bio *bio = NULL;
3389         unsigned long bio_flags = 0;
3390
3391         if (test_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags))
3392                 return 0;
3393
3394         if (test_range_bit(tree, eb->start, eb->start + eb->len - 1,
3395                            EXTENT_UPTODATE, 1, NULL)) {
3396                 return 0;
3397         }
3398
3399         if (start) {
3400                 WARN_ON(start < eb->start);
3401                 start_i = (start >> PAGE_CACHE_SHIFT) -
3402                         (eb->start >> PAGE_CACHE_SHIFT);
3403         } else {
3404                 start_i = 0;
3405         }
3406
3407         num_pages = num_extent_pages(eb->start, eb->len);
3408         for (i = start_i; i < num_pages; i++) {
3409                 page = extent_buffer_page(eb, i);
3410                 if (!wait) {
3411                         if (!trylock_page(page))
3412                                 goto unlock_exit;
3413                 } else {
3414                         lock_page(page);
3415                 }
3416                 locked_pages++;
3417                 if (!PageUptodate(page))
3418                         all_uptodate = 0;
3419         }
3420         if (all_uptodate) {
3421                 if (start_i == 0)
3422                         set_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags);
3423                 goto unlock_exit;
3424         }
3425
3426         for (i = start_i; i < num_pages; i++) {
3427                 page = extent_buffer_page(eb, i);
3428                 if (inc_all_pages)
3429                         page_cache_get(page);
3430                 if (!PageUptodate(page)) {
3431                         if (start_i == 0)
3432                                 inc_all_pages = 1;
3433                         ClearPageError(page);
3434                         err = __extent_read_full_page(tree, page,
3435                                                       get_extent, &bio,
3436                                                       mirror_num, &bio_flags);
3437                         if (err)
3438                                 ret = err;
3439                 } else {
3440                         unlock_page(page);
3441                 }
3442         }
3443
3444         if (bio)
3445                 submit_one_bio(READ, bio, mirror_num, bio_flags);
3446
3447         if (ret || !wait)
3448                 return ret;
3449
3450         for (i = start_i; i < num_pages; i++) {
3451                 page = extent_buffer_page(eb, i);
3452                 wait_on_page_locked(page);
3453                 if (!PageUptodate(page))
3454                         ret = -EIO;
3455         }
3456
3457         if (!ret)
3458                 set_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags);
3459         return ret;
3460
3461 unlock_exit:
3462         i = start_i;
3463         while (locked_pages > 0) {
3464                 page = extent_buffer_page(eb, i);
3465                 i++;
3466                 unlock_page(page);
3467                 locked_pages--;
3468         }
3469         return ret;
3470 }
3471
3472 void read_extent_buffer(struct extent_buffer *eb, void *dstv,
3473                         unsigned long start,
3474                         unsigned long len)
3475 {
3476         size_t cur;
3477         size_t offset;
3478         struct page *page;
3479         char *kaddr;
3480         char *dst = (char *)dstv;
3481         size_t start_offset = eb->start & ((u64)PAGE_CACHE_SIZE - 1);
3482         unsigned long i = (start_offset + start) >> PAGE_CACHE_SHIFT;
3483
3484         WARN_ON(start > eb->len);
3485         WARN_ON(start + len > eb->start + eb->len);
3486
3487         offset = (start_offset + start) & ((unsigned long)PAGE_CACHE_SIZE - 1);
3488
3489         while (len > 0) {
3490                 page = extent_buffer_page(eb, i);
3491
3492                 cur = min(len, (PAGE_CACHE_SIZE - offset));
3493                 kaddr = kmap_atomic(page, KM_USER1);
3494                 memcpy(dst, kaddr + offset, cur);
3495                 kunmap_atomic(kaddr, KM_USER1);
3496
3497                 dst += cur;
3498                 len -= cur;
3499                 offset = 0;
3500                 i++;
3501         }
3502 }
3503
3504 int map_private_extent_buffer(struct extent_buffer *eb, unsigned long start,
3505                                unsigned long min_len, char **token, char **map,
3506                                unsigned long *map_start,
3507                                unsigned long *map_len, int km)
3508 {
3509         size_t offset = start & (PAGE_CACHE_SIZE - 1);
3510         char *kaddr;
3511         struct page *p;
3512         size_t start_offset = eb->start & ((u64)PAGE_CACHE_SIZE - 1);
3513         unsigned long i = (start_offset + start) >> PAGE_CACHE_SHIFT;
3514         unsigned long end_i = (start_offset + start + min_len - 1) >>
3515                 PAGE_CACHE_SHIFT;
3516
3517         if (i != end_i)
3518                 return -EINVAL;
3519
3520         if (i == 0) {
3521                 offset = start_offset;
3522                 *map_start = 0;
3523         } else {
3524                 offset = 0;
3525                 *map_start = ((u64)i << PAGE_CACHE_SHIFT) - start_offset;
3526         }
3527
3528         if (start + min_len > eb->len) {
3529                 printk(KERN_ERR "btrfs bad mapping eb start %llu len %lu, "
3530                        "wanted %lu %lu\n", (unsigned long long)eb->start,
3531                        eb->len, start, min_len);
3532                 WARN_ON(1);
3533         }
3534
3535         p = extent_buffer_page(eb, i);
3536         kaddr = kmap_atomic(p, km);
3537         *token = kaddr;
3538         *map = kaddr + offset;
3539         *map_len = PAGE_CACHE_SIZE - offset;
3540         return 0;
3541 }
3542
3543 int map_extent_buffer(struct extent_buffer *eb, unsigned long start,
3544                       unsigned long min_len,
3545                       char **token, char **map,
3546                       unsigned long *map_start,
3547                       unsigned long *map_len, int km)
3548 {
3549         int err;
3550         int save = 0;
3551         if (eb->map_token) {
3552                 unmap_extent_buffer(eb, eb->map_token, km);
3553                 eb->map_token = NULL;
3554                 save = 1;
3555         }
3556         err = map_private_extent_buffer(eb, start, min_len, token, map,
3557                                        map_start, map_len, km);
3558         if (!err && save) {
3559                 eb->map_token = *token;
3560                 eb->kaddr = *map;
3561                 eb->map_start = *map_start;
3562                 eb->map_len = *map_len;
3563         }
3564         return err;
3565 }
3566
3567 void unmap_extent_buffer(struct extent_buffer *eb, char *token, int km)
3568 {
3569         kunmap_atomic(token, km);
3570 }
3571
3572 int memcmp_extent_buffer(struct extent_buffer *eb, const void *ptrv,
3573                           unsigned long start,
3574                           unsigned long len)
3575 {
3576         size_t cur;
3577         size_t offset;
3578         struct page *page;
3579         char *kaddr;
3580         char *ptr = (char *)ptrv;
3581         size_t start_offset = eb->start & ((u64)PAGE_CACHE_SIZE - 1);
3582         unsigned long i = (start_offset + start) >> PAGE_CACHE_SHIFT;
3583         int ret = 0;
3584
3585         WARN_ON(start > eb->len);
3586         WARN_ON(start + len > eb->start + eb->len);
3587
3588         offset = (start_offset + start) & ((unsigned long)PAGE_CACHE_SIZE - 1);
3589
3590         while (len > 0) {
3591                 page = extent_buffer_page(eb, i);
3592
3593                 cur = min(len, (PAGE_CACHE_SIZE - offset));
3594
3595                 kaddr = kmap_atomic(page, KM_USER0);
3596                 ret = memcmp(ptr, kaddr + offset, cur);
3597                 kunmap_atomic(kaddr, KM_USER0);
3598                 if (ret)
3599                         break;
3600
3601                 ptr += cur;
3602                 len -= cur;
3603                 offset = 0;
3604                 i++;
3605         }
3606         return ret;
3607 }
3608
3609 void write_extent_buffer(struct extent_buffer *eb, const void *srcv,
3610                          unsigned long start, unsigned long len)
3611 {
3612         size_t cur;
3613         size_t offset;
3614         struct page *page;
3615         char *kaddr;
3616         char *src = (char *)srcv;
3617         size_t start_offset = eb->start & ((u64)PAGE_CACHE_SIZE - 1);
3618         unsigned long i = (start_offset + start) >> PAGE_CACHE_SHIFT;
3619
3620         WARN_ON(start > eb->len);
3621         WARN_ON(start + len > eb->start + eb->len);
3622
3623         offset = (start_offset + start) & ((unsigned long)PAGE_CACHE_SIZE - 1);
3624
3625         while (len > 0) {
3626                 page = extent_buffer_page(eb, i);
3627                 WARN_ON(!PageUptodate(page));
3628
3629                 cur = min(len, PAGE_CACHE_SIZE - offset);
3630                 kaddr = kmap_atomic(page, KM_USER1);
3631                 memcpy(kaddr + offset, src, cur);
3632                 kunmap_atomic(kaddr, KM_USER1);
3633
3634                 src += cur;
3635                 len -= cur;
3636                 offset = 0;
3637                 i++;
3638         }
3639 }
3640
3641 void memset_extent_buffer(struct extent_buffer *eb, char c,
3642                           unsigned long start, unsigned long len)
3643 {
3644         size_t cur;
3645         size_t offset;
3646         struct page *page;
3647         char *kaddr;
3648         size_t start_offset = eb->start & ((u64)PAGE_CACHE_SIZE - 1);
3649         unsigned long i = (start_offset + start) >> PAGE_CACHE_SHIFT;
3650
3651         WARN_ON(start > eb->len);
3652         WARN_ON(start + len > eb->start + eb->len);
3653
3654         offset = (start_offset + start) & ((unsigned long)PAGE_CACHE_SIZE - 1);
3655
3656         while (len > 0) {
3657                 page = extent_buffer_page(eb, i);
3658                 WARN_ON(!PageUptodate(page));
3659
3660                 cur = min(len, PAGE_CACHE_SIZE - offset);
3661                 kaddr = kmap_atomic(page, KM_USER0);
3662                 memset(kaddr + offset, c, cur);
3663                 kunmap_atomic(kaddr, KM_USER0);
3664
3665                 len -= cur;
3666                 offset = 0;
3667                 i++;
3668         }
3669 }
3670
3671 void copy_extent_buffer(struct extent_buffer *dst, struct extent_buffer *src,
3672                         unsigned long dst_offset, unsigned long src_offset,
3673                         unsigned long len)
3674 {
3675         u64 dst_len = dst->len;
3676         size_t cur;
3677         size_t offset;
3678         struct page *page;
3679         char *kaddr;
3680         size_t start_offset = dst->start & ((u64)PAGE_CACHE_SIZE - 1);
3681         unsigned long i = (start_offset + dst_offset) >> PAGE_CACHE_SHIFT;
3682
3683         WARN_ON(src->len != dst_len);
3684
3685         offset = (start_offset + dst_offset) &
3686                 ((unsigned long)PAGE_CACHE_SIZE - 1);
3687
3688         while (len > 0) {
3689                 page = extent_buffer_page(dst, i);
3690                 WARN_ON(!PageUptodate(page));
3691
3692                 cur = min(len, (unsigned long)(PAGE_CACHE_SIZE - offset));
3693
3694                 kaddr = kmap_atomic(page, KM_USER0);
3695                 read_extent_buffer(src, kaddr + offset, src_offset, cur);
3696                 kunmap_atomic(kaddr, KM_USER0);
3697
3698                 src_offset += cur;
3699                 len -= cur;
3700                 offset = 0;
3701                 i++;
3702         }
3703 }
3704
3705 static void move_pages(struct page *dst_page, struct page *src_page,
3706                        unsigned long dst_off, unsigned long src_off,
3707                        unsigned long len)
3708 {
3709         char *dst_kaddr = kmap_atomic(dst_page, KM_USER0);
3710         if (dst_page == src_page) {
3711                 memmove(dst_kaddr + dst_off, dst_kaddr + src_off, len);
3712         } else {
3713                 char *src_kaddr = kmap_atomic(src_page, KM_USER1);
3714                 char *p = dst_kaddr + dst_off + len;
3715                 char *s = src_kaddr + src_off + len;
3716
3717                 while (len--)
3718                         *--p = *--s;
3719
3720                 kunmap_atomic(src_kaddr, KM_USER1);
3721         }
3722         kunmap_atomic(dst_kaddr, KM_USER0);
3723 }
3724
3725 static void copy_pages(struct page *dst_page, struct page *src_page,
3726                        unsigned long dst_off, unsigned long src_off,
3727                        unsigned long len)
3728 {
3729         char *dst_kaddr = kmap_atomic(dst_page, KM_USER0);
3730         char *src_kaddr;
3731
3732         if (dst_page != src_page)
3733                 src_kaddr = kmap_atomic(src_page, KM_USER1);
3734         else
3735                 src_kaddr = dst_kaddr;
3736
3737         memcpy(dst_kaddr + dst_off, src_kaddr + src_off, len);
3738         kunmap_atomic(dst_kaddr, KM_USER0);
3739         if (dst_page != src_page)
3740                 kunmap_atomic(src_kaddr, KM_USER1);
3741 }
3742
3743 void memcpy_extent_buffer(struct extent_buffer *dst, unsigned long dst_offset,
3744                            unsigned long src_offset, unsigned long len)
3745 {
3746         size_t cur;
3747         size_t dst_off_in_page;
3748         size_t src_off_in_page;
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 dst 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 dst len %lu\n", dst_offset, len, dst->len);
3761                 BUG_ON(1);
3762         }
3763
3764         while (len > 0) {
3765                 dst_off_in_page = (start_offset + dst_offset) &
3766                         ((unsigned long)PAGE_CACHE_SIZE - 1);
3767                 src_off_in_page = (start_offset + src_offset) &
3768                         ((unsigned long)PAGE_CACHE_SIZE - 1);
3769
3770                 dst_i = (start_offset + dst_offset) >> PAGE_CACHE_SHIFT;
3771                 src_i = (start_offset + src_offset) >> PAGE_CACHE_SHIFT;
3772
3773                 cur = min(len, (unsigned long)(PAGE_CACHE_SIZE -
3774                                                src_off_in_page));
3775                 cur = min_t(unsigned long, cur,
3776                         (unsigned long)(PAGE_CACHE_SIZE - dst_off_in_page));
3777
3778                 copy_pages(extent_buffer_page(dst, dst_i),
3779                            extent_buffer_page(dst, src_i),
3780                            dst_off_in_page, src_off_in_page, cur);
3781
3782                 src_offset += cur;
3783                 dst_offset += cur;
3784                 len -= cur;
3785         }
3786 }
3787
3788 void memmove_extent_buffer(struct extent_buffer *dst, unsigned long dst_offset,
3789                            unsigned long src_offset, unsigned long len)
3790 {
3791         size_t cur;
3792         size_t dst_off_in_page;
3793         size_t src_off_in_page;
3794         unsigned long dst_end = dst_offset + len - 1;
3795         unsigned long src_end = src_offset + len - 1;
3796         size_t start_offset = dst->start & ((u64)PAGE_CACHE_SIZE - 1);
3797         unsigned long dst_i;
3798         unsigned long src_i;
3799
3800         if (src_offset + len > dst->len) {
3801                 printk(KERN_ERR "btrfs memmove bogus src_offset %lu move "
3802                        "len %lu len %lu\n", src_offset, len, dst->len);
3803                 BUG_ON(1);
3804         }
3805         if (dst_offset + len > dst->len) {
3806                 printk(KERN_ERR "btrfs memmove bogus dst_offset %lu move "
3807                        "len %lu len %lu\n", dst_offset, len, dst->len);
3808                 BUG_ON(1);
3809         }
3810         if (dst_offset < src_offset) {
3811                 memcpy_extent_buffer(dst, dst_offset, src_offset, len);
3812                 return;
3813         }
3814         while (len > 0) {
3815                 dst_i = (start_offset + dst_end) >> PAGE_CACHE_SHIFT;
3816                 src_i = (start_offset + src_end) >> PAGE_CACHE_SHIFT;
3817
3818                 dst_off_in_page = (start_offset + dst_end) &
3819                         ((unsigned long)PAGE_CACHE_SIZE - 1);
3820                 src_off_in_page = (start_offset + src_end) &
3821                         ((unsigned long)PAGE_CACHE_SIZE - 1);
3822
3823                 cur = min_t(unsigned long, len, src_off_in_page + 1);
3824                 cur = min(cur, dst_off_in_page + 1);
3825                 move_pages(extent_buffer_page(dst, dst_i),
3826                            extent_buffer_page(dst, src_i),
3827                            dst_off_in_page - cur + 1,
3828                            src_off_in_page - cur + 1, cur);
3829
3830                 dst_end -= cur;
3831                 src_end -= cur;
3832                 len -= cur;
3833         }
3834 }
3835
3836 int try_release_extent_buffer(struct extent_io_tree *tree, struct page *page)
3837 {
3838         u64 start = page_offset(page);
3839         struct extent_buffer *eb;
3840         int ret = 1;
3841         unsigned long i;
3842         unsigned long num_pages;
3843
3844         spin_lock(&tree->buffer_lock);
3845         eb = buffer_search(tree, start);
3846         if (!eb)
3847                 goto out;
3848
3849         if (atomic_read(&eb->refs) > 1) {
3850                 ret = 0;
3851                 goto out;
3852         }
3853         if (test_bit(EXTENT_BUFFER_DIRTY, &eb->bflags)) {
3854                 ret = 0;
3855                 goto out;
3856         }
3857         /* at this point we can safely release the extent buffer */
3858         num_pages = num_extent_pages(eb->start, eb->len);
3859         for (i = 0; i < num_pages; i++)
3860                 page_cache_release(extent_buffer_page(eb, i));
3861         rb_erase(&eb->rb_node, &tree->buffer);
3862         __free_extent_buffer(eb);
3863 out:
3864         spin_unlock(&tree->buffer_lock);
3865         return ret;
3866 }