git.oblomov.eu Git - linux-2.6/blob - 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/gfp.h>
   6 #include <linux/pagemap.h>
   7 #include <linux/page-flags.h>
   8 #include <linux/module.h>
   9 #include <linux/spinlock.h>
  10 #include <linux/blkdev.h>
  11 #include <linux/swap.h>
  12 #include <linux/writeback.h>
  13 #include <linux/pagevec.h>
  14 #include "extent_io.h"
  15 #include "extent_map.h"
  16 #include "compat.h"
  17 #include "ctree.h"
  18 #include "btrfs_inode.h"
  19
  20 /* temporary define until extent_map moves out of btrfs */
  21 struct kmem_cache *btrfs_cache_create(const char *name, size_t size,
  22                                        unsigned long extra_flags,
  23                                        void (*ctor)(void *, struct kmem_cache *,
  24                                                     unsigned long));
  25
  26 static struct kmem_cache *extent_state_cache;
  27 static struct kmem_cache *extent_buffer_cache;
  28
  29 static LIST_HEAD(buffers);
  30 static LIST_HEAD(states);
  31
  32 #define LEAK_DEBUG 0
  33 #if LEAK_DEBUG
  34 static DEFINE_SPINLOCK(leak_lock);
  35 #endif
  36
  37 #define BUFFER_LRU_MAX 64
  38
  39 struct tree_entry {
  40         u64 start;
  41         u64 end;
  42         struct rb_node rb_node;
  43 };
  44
  45 struct extent_page_data {
  46         struct bio *bio;
  47         struct extent_io_tree *tree;
  48         get_extent_t *get_extent;
  49
  50         /* tells writepage not to lock the state bits for this range
  51          * it still does the unlocking
  52          */
  53         int extent_locked;
  54 };
  55
  56 int __init extent_io_init(void)
  57 {
  58         extent_state_cache = btrfs_cache_create("extent_state",
  59                                             sizeof(struct extent_state), 0,
  60                                             NULL);
  61         if (!extent_state_cache)
  62                 return -ENOMEM;
  63
  64         extent_buffer_cache = btrfs_cache_create("extent_buffers",
  65                                             sizeof(struct extent_buffer), 0,
  66                                             NULL);
  67         if (!extent_buffer_cache)
  68                 goto free_state_cache;
  69         return 0;
  70
  71 free_state_cache:
  72         kmem_cache_destroy(extent_state_cache);
  73         return -ENOMEM;
  74 }
  75
  76 void extent_io_exit(void)
  77 {
  78         struct extent_state *state;
  79         struct extent_buffer *eb;
  80
  81         while (!list_empty(&states)) {
  82                 state = list_entry(states.next, struct extent_state, leak_list);
  83                 printk(KERN_ERR "btrfs state leak: start %llu end %llu "
  84                        "state %lu in tree %p refs %d\n",
  85                        (unsigned long long)state->start,
  86                        (unsigned long long)state->end,
  87                        state->state, state->tree, atomic_read(&state->refs));
  88                 list_del(&state->leak_list);
  89                 kmem_cache_free(extent_state_cache, state);
  90
  91         }
  92
  93         while (!list_empty(&buffers)) {
  94                 eb = list_entry(buffers.next, struct extent_buffer, leak_list);
  95                 printk(KERN_ERR "btrfs buffer leak start %llu len %lu "
  96                        "refs %d\n", (unsigned long long)eb->start,
  97                        eb->len, atomic_read(&eb->refs));
  98                 list_del(&eb->leak_list);
  99                 kmem_cache_free(extent_buffer_cache, eb);
 100         }
 101         if (extent_state_cache)
 102                 kmem_cache_destroy(extent_state_cache);
 103         if (extent_buffer_cache)
 104                 kmem_cache_destroy(extent_buffer_cache);
 105 }
 106
 107 void extent_io_tree_init(struct extent_io_tree *tree,
 108                           struct address_space *mapping, gfp_t mask)
 109 {
 110         tree->state.rb_node = NULL;
 111         tree->buffer.rb_node = NULL;
 112         tree->ops = NULL;
 113         tree->dirty_bytes = 0;
 114         spin_lock_init(&tree->lock);
 115         spin_lock_init(&tree->buffer_lock);
 116         tree->mapping = mapping;
 117 }
 118
 119 static struct extent_state *alloc_extent_state(gfp_t mask)
 120 {
 121         struct extent_state *state;
 122 #if LEAK_DEBUG
 123         unsigned long flags;
 124 #endif
 125
 126         state = kmem_cache_alloc(extent_state_cache, mask);
 127         if (!state)
 128                 return state;
 129         state->state = 0;
 130         state->private = 0;
 131         state->tree = NULL;
 132 #if LEAK_DEBUG
 133         spin_lock_irqsave(&leak_lock, flags);
 134         list_add(&state->leak_list, &states);
 135         spin_unlock_irqrestore(&leak_lock, flags);
 136 #endif
 137         atomic_set(&state->refs, 1);
 138         init_waitqueue_head(&state->wq);
 139         return state;
 140 }
 141
 142 static void free_extent_state(struct extent_state *state)
 143 {
 144         if (!state)
 145                 return;
 146         if (atomic_dec_and_test(&state->refs)) {
 147 #if LEAK_DEBUG
 148                 unsigned long flags;
 149 #endif
 150                 WARN_ON(state->tree);
 151 #if LEAK_DEBUG
 152                 spin_lock_irqsave(&leak_lock, flags);
 153                 list_del(&state->leak_list);
 154                 spin_unlock_irqrestore(&leak_lock, flags);
 155 #endif
 156                 kmem_cache_free(extent_state_cache, state);
 157         }
 158 }
 159
 160 static struct rb_node *tree_insert(struct rb_root *root, u64 offset,
 161                                    struct rb_node *node)
 162 {
 163         struct rb_node **p = &root->rb_node;
 164         struct rb_node *parent = NULL;
 165         struct tree_entry *entry;
 166
 167         while (*p) {
 168                 parent = *p;
 169                 entry = rb_entry(parent, struct tree_entry, rb_node);
 170
 171                 if (offset < entry->start)
 172                         p = &(*p)->rb_left;
 173                 else if (offset > entry->end)
 174                         p = &(*p)->rb_right;
 175                 else
 176                         return parent;
 177         }
 178
 179         entry = rb_entry(node, struct tree_entry, rb_node);
 180         rb_link_node(node, parent, p);
 181         rb_insert_color(node, root);
 182         return NULL;
 183 }
 184
 185 static struct rb_node *__etree_search(struct extent_io_tree *tree, u64 offset,
 186                                      struct rb_node **prev_ret,
 187                                      struct rb_node **next_ret)
 188 {
 189         struct rb_root *root = &tree->state;
 190         struct rb_node *n = root->rb_node;
 191         struct rb_node *prev = NULL;
 192         struct rb_node *orig_prev = NULL;
 193         struct tree_entry *entry;
 194         struct tree_entry *prev_entry = NULL;
 195
 196         while (n) {
 197                 entry = rb_entry(n, struct tree_entry, rb_node);
 198                 prev = n;
 199                 prev_entry = entry;
 200
 201                 if (offset < entry->start)
 202                         n = n->rb_left;
 203                 else if (offset > entry->end)
 204                         n = n->rb_right;
 205                 else
 206                         return n;
 207         }
 208
 209         if (prev_ret) {
 210                 orig_prev = prev;
 211                 while (prev && offset > prev_entry->end) {
 212                         prev = rb_next(prev);
 213                         prev_entry = rb_entry(prev, struct tree_entry, rb_node);
 214                 }
 215                 *prev_ret = prev;
 216                 prev = orig_prev;
 217         }
 218
 219         if (next_ret) {
 220                 prev_entry = rb_entry(prev, struct tree_entry, rb_node);
 221                 while (prev && offset < prev_entry->start) {
 222                         prev = rb_prev(prev);
 223                         prev_entry = rb_entry(prev, struct tree_entry, rb_node);
 224                 }
 225                 *next_ret = prev;
 226         }
 227         return NULL;
 228 }
 229
 230 static inline struct rb_node *tree_search(struct extent_io_tree *tree,
 231                                           u64 offset)
 232 {
 233         struct rb_node *prev = NULL;
 234         struct rb_node *ret;
 235
 236         ret = __etree_search(tree, offset, &prev, NULL);
 237         if (!ret)
 238                 return prev;
 239         return ret;
 240 }
 241
 242 static struct extent_buffer *buffer_tree_insert(struct extent_io_tree *tree,
 243                                           u64 offset, struct rb_node *node)
 244 {
 245         struct rb_root *root = &tree->buffer;
 246         struct rb_node **p = &root->rb_node;
 247         struct rb_node *parent = NULL;
 248         struct extent_buffer *eb;
 249
 250         while (*p) {
 251                 parent = *p;
 252                 eb = rb_entry(parent, struct extent_buffer, rb_node);
 253
 254                 if (offset < eb->start)
 255                         p = &(*p)->rb_left;
 256                 else if (offset > eb->start)
 257                         p = &(*p)->rb_right;
 258                 else
 259                         return eb;
 260         }
 261
 262         rb_link_node(node, parent, p);
 263         rb_insert_color(node, root);
 264         return NULL;
 265 }
 266
 267 static struct extent_buffer *buffer_search(struct extent_io_tree *tree,
 268                                            u64 offset)
 269 {
 270         struct rb_root *root = &tree->buffer;
 271         struct rb_node *n = root->rb_node;
 272         struct extent_buffer *eb;
 273
 274         while (n) {
 275                 eb = rb_entry(n, struct extent_buffer, rb_node);
 276                 if (offset < eb->start)
 277                         n = n->rb_left;
 278                 else if (offset > eb->start)
 279                         n = n->rb_right;
 280                 else
 281                         return eb;
 282         }
 283         return NULL;
 284 }
 285
 286 /*
 287  * utility function to look for merge candidates inside a given range.
 288  * Any extents with matching state are merged together into a single
 289  * extent in the tree.  Extents with EXTENT_IO in their state field
 290  * are not merged because the end_io handlers need to be able to do
 291  * operations on them without sleeping (or doing allocations/splits).
 292  *
 293  * This should be called with the tree lock held.
 294  */
 295 static int merge_state(struct extent_io_tree *tree,
 296                        struct extent_state *state)
 297 {
 298         struct extent_state *other;
 299         struct rb_node *other_node;
 300
 301         if (state->state & (EXTENT_IOBITS | EXTENT_BOUNDARY))
 302                 return 0;
 303
 304         other_node = rb_prev(&state->rb_node);
 305         if (other_node) {
 306                 other = rb_entry(other_node, struct extent_state, rb_node);
 307                 if (other->end == state->start - 1 &&
 308                     other->state == state->state) {
 309                         state->start = other->start;
 310                         other->tree = NULL;
 311                         rb_erase(&other->rb_node, &tree->state);
 312                         free_extent_state(other);
 313                 }
 314         }
 315         other_node = rb_next(&state->rb_node);
 316         if (other_node) {
 317                 other = rb_entry(other_node, struct extent_state, rb_node);
 318                 if (other->start == state->end + 1 &&
 319                     other->state == state->state) {
 320                         other->start = state->start;
 321                         state->tree = NULL;
 322                         rb_erase(&state->rb_node, &tree->state);
 323                         free_extent_state(state);
 324                 }
 325         }
 326         return 0;
 327 }
 328
 329 static void set_state_cb(struct extent_io_tree *tree,
 330                          struct extent_state *state,
 331                          unsigned long bits)
 332 {
 333         if (tree->ops && tree->ops->set_bit_hook) {
 334                 tree->ops->set_bit_hook(tree->mapping->host, state->start,
 335                                         state->end, state->state, bits);
 336         }
 337 }
 338
 339 static void clear_state_cb(struct extent_io_tree *tree,
 340                            struct extent_state *state,
 341                            unsigned long bits)
 342 {
 343         if (tree->ops && tree->ops->clear_bit_hook) {
 344                 tree->ops->clear_bit_hook(tree->mapping->host, state->start,
 345                                           state->end, state->state, bits);
 346         }
 347 }
 348
 349 /*
 350  * insert an extent_state struct into the tree.  'bits' are set on the
 351  * struct before it is inserted.
 352  *
 353  * This may return -EEXIST if the extent is already there, in which case the
 354  * state struct is freed.
 355  *
 356  * The tree lock is not taken internally.  This is a utility function and
 357  * probably isn't what you want to call (see set/clear_extent_bit).
 358  */
 359 static int insert_state(struct extent_io_tree *tree,
 360                         struct extent_state *state, u64 start, u64 end,
 361                         int bits)
 362 {
 363         struct rb_node *node;
 364
 365         if (end < start) {
 366                 printk(KERN_ERR "btrfs end < start %llu %llu\n",
 367                        (unsigned long long)end,
 368                        (unsigned long long)start);
 369                 WARN_ON(1);
 370         }
 371         if (bits & EXTENT_DIRTY)
 372                 tree->dirty_bytes += end - start + 1;
 373         set_state_cb(tree, state, bits);
 374         state->state |= bits;
 375         state->start = start;
 376         state->end = end;
 377         node = tree_insert(&tree->state, end, &state->rb_node);
 378         if (node) {
 379                 struct extent_state *found;
 380                 found = rb_entry(node, struct extent_state, rb_node);
 381                 printk(KERN_ERR "btrfs found node %llu %llu on insert of "
 382                        "%llu %llu\n", (unsigned long long)found->start,
 383                        (unsigned long long)found->end,
 384                        (unsigned long long)start, (unsigned long long)end);
 385                 free_extent_state(state);
 386                 return -EEXIST;
 387         }
 388         state->tree = tree;
 389         merge_state(tree, state);
 390         return 0;
 391 }
 392
 393 /*
 394  * split a given extent state struct in two, inserting the preallocated
 395  * struct 'prealloc' as the newly created second half.  'split' indicates an
 396  * offset inside 'orig' where it should be split.
 397  *
 398  * Before calling,
 399  * the tree has 'orig' at [orig->start, orig->end].  After calling, there
 400  * are two extent state structs in the tree:
 401  * prealloc: [orig->start, split - 1]
 402  * orig: [ split, orig->end ]
 403  *
 404  * The tree locks are not taken by this function. They need to be held
 405  * by the caller.
 406  */
 407 static int split_state(struct extent_io_tree *tree, struct extent_state *orig,
 408                        struct extent_state *prealloc, u64 split)
 409 {
 410         struct rb_node *node;
 411         prealloc->start = orig->start;
 412         prealloc->end = split - 1;
 413         prealloc->state = orig->state;
 414         orig->start = split;
 415
 416         node = tree_insert(&tree->state, prealloc->end, &prealloc->rb_node);
 417         if (node) {
 418                 free_extent_state(prealloc);
 419                 return -EEXIST;
 420         }
 421         prealloc->tree = tree;
 422         return 0;
 423 }
 424
 425 /*
 426  * utility function to clear some bits in an extent state struct.
 427  * it will optionally wake up any one waiting on this state (wake == 1), or
 428  * forcibly remove the state from the tree (delete == 1).
 429  *
 430  * If no bits are set on the state struct after clearing things, the
 431  * struct is freed and removed from the tree
 432  */
 433 static int clear_state_bit(struct extent_io_tree *tree,
 434                             struct extent_state *state, int bits, int wake,
 435                             int delete)
 436 {
 437         int ret = state->state & bits;
 438
 439         if ((bits & EXTENT_DIRTY) && (state->state & EXTENT_DIRTY)) {
 440                 u64 range = state->end - state->start + 1;
 441                 WARN_ON(range > tree->dirty_bytes);
 442                 tree->dirty_bytes -= range;
 443         }
 444         clear_state_cb(tree, state, bits);
 445         state->state &= ~bits;
 446         if (wake)
 447                 wake_up(&state->wq);
 448         if (delete || state->state == 0) {
 449                 if (state->tree) {
 450                         clear_state_cb(tree, state, state->state);
 451                         rb_erase(&state->rb_node, &tree->state);
 452                         state->tree = NULL;
 453                         free_extent_state(state);
 454                 } else {
 455                         WARN_ON(1);
 456                 }
 457         } else {
 458                 merge_state(tree, state);
 459         }
 460         return ret;
 461 }
 462
 463 /*
 464  * clear some bits on a range in the tree.  This may require splitting
 465  * or inserting elements in the tree, so the gfp mask is used to
 466  * indicate which allocations or sleeping are allowed.
 467  *
 468  * pass 'wake' == 1 to kick any sleepers, and 'delete' == 1 to remove
 469  * the given range from the tree regardless of state (ie for truncate).
 470  *
 471  * the range [start, end] is inclusive.
 472  *
 473  * This takes the tree lock, and returns < 0 on error, > 0 if any of the
 474  * bits were already set, or zero if none of the bits were already set.
 475  */
 476 int clear_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
 477                      int bits, int wake, int delete, gfp_t mask)
 478 {
 479         struct extent_state *state;
 480         struct extent_state *prealloc = NULL;
 481         struct rb_node *node;
 482         int err;
 483         int set = 0;
 484
 485 again:
 486         if (!prealloc && (mask & __GFP_WAIT)) {
 487                 prealloc = alloc_extent_state(mask);
 488                 if (!prealloc)
 489                         return -ENOMEM;
 490         }
 491
 492         spin_lock(&tree->lock);
 493         /*
 494          * this search will find the extents that end after
 495          * our range starts
 496          */
 497         node = tree_search(tree, start);
 498         if (!node)
 499                 goto out;
 500         state = rb_entry(node, struct extent_state, rb_node);
 501         if (state->start > end)
 502                 goto out;
 503         WARN_ON(state->end < start);
 504
 505         /*
 506          *     | ---- desired range ---- |
 507          *  | state | or
 508          *  | ------------- state -------------- |
 509          *
 510          * We need to split the extent we found, and may flip
 511          * bits on second half.
 512          *
 513          * If the extent we found extends past our range, we
 514          * just split and search again.  It'll get split again
 515          * the next time though.
 516          *
 517          * If the extent we found is inside our range, we clear
 518          * the desired bit on it.
 519          */
 520
 521         if (state->start < start) {
 522                 if (!prealloc)
 523                         prealloc = alloc_extent_state(GFP_ATOMIC);
 524                 err = split_state(tree, state, prealloc, start);
 525                 BUG_ON(err == -EEXIST);
 526                 prealloc = NULL;
 527                 if (err)
 528                         goto out;
 529                 if (state->end <= end) {
 530                         start = state->end + 1;
 531                         set |= clear_state_bit(tree, state, bits,
 532                                         wake, delete);
 533                 } else {
 534                         start = state->start;
 535                 }
 536                 goto search_again;
 537         }
 538         /*
 539          * | ---- desired range ---- |
 540          *                        | state |
 541          * We need to split the extent, and clear the bit
 542          * on the first half
 543          */
 544         if (state->start <= end && state->end > end) {
 545                 if (!prealloc)
 546                         prealloc = alloc_extent_state(GFP_ATOMIC);
 547                 err = split_state(tree, state, prealloc, end + 1);
 548                 BUG_ON(err == -EEXIST);
 549
 550                 if (wake)
 551                         wake_up(&state->wq);
 552                 set |= clear_state_bit(tree, prealloc, bits,
 553                                        wake, delete);
 554                 prealloc = NULL;
 555                 goto out;
 556         }
 557
 558         start = state->end + 1;
 559         set |= clear_state_bit(tree, state, bits, wake, delete);
 560         goto search_again;
 561
 562 out:
 563         spin_unlock(&tree->lock);
 564         if (prealloc)
 565                 free_extent_state(prealloc);
 566
 567         return set;
 568
 569 search_again:
 570         if (start > end)
 571                 goto out;
 572         spin_unlock(&tree->lock);
 573         if (mask & __GFP_WAIT)
 574                 cond_resched();
 575         goto again;
 576 }
 577
 578 static int wait_on_state(struct extent_io_tree *tree,
 579                          struct extent_state *state)
 580                 __releases(tree->lock)
 581                 __acquires(tree->lock)
 582 {
 583         DEFINE_WAIT(wait);
 584         prepare_to_wait(&state->wq, &wait, TASK_UNINTERRUPTIBLE);
 585         spin_unlock(&tree->lock);
 586         schedule();
 587         spin_lock(&tree->lock);
 588         finish_wait(&state->wq, &wait);
 589         return 0;
 590 }
 591
 592 /*
 593  * waits for one or more bits to clear on a range in the state tree.
 594  * The range [start, end] is inclusive.
 595  * The tree lock is taken by this function
 596  */
 597 int wait_extent_bit(struct extent_io_tree *tree, u64 start, u64 end, int bits)
 598 {
 599         struct extent_state *state;
 600         struct rb_node *node;
 601
 602         spin_lock(&tree->lock);
 603 again:
 604         while (1) {
 605                 /*
 606                  * this search will find all the extents that end after
 607                  * our range starts
 608                  */
 609                 node = tree_search(tree, start);
 610                 if (!node)
 611                         break;
 612
 613                 state = rb_entry(node, struct extent_state, rb_node);
 614
 615                 if (state->start > end)
 616                         goto out;
 617
 618                 if (state->state & bits) {
 619                         start = state->start;
 620                         atomic_inc(&state->refs);
 621                         wait_on_state(tree, state);
 622                         free_extent_state(state);
 623                         goto again;
 624                 }
 625                 start = state->end + 1;
 626
 627                 if (start > end)
 628                         break;
 629
 630                 if (need_resched()) {
 631                         spin_unlock(&tree->lock);
 632                         cond_resched();
 633                         spin_lock(&tree->lock);
 634                 }
 635         }
 636 out:
 637         spin_unlock(&tree->lock);
 638         return 0;
 639 }
 640
 641 static void set_state_bits(struct extent_io_tree *tree,
 642                            struct extent_state *state,
 643                            int bits)
 644 {
 645         if ((bits & EXTENT_DIRTY) && !(state->state & EXTENT_DIRTY)) {
 646                 u64 range = state->end - state->start + 1;
 647                 tree->dirty_bytes += range;
 648         }
 649         set_state_cb(tree, state, bits);
 650         state->state |= bits;
 651 }
 652
 653 /*
 654  * set some bits on a range in the tree.  This may require allocations
 655  * or sleeping, so the gfp mask is used to indicate what is allowed.
 656  *
 657  * If 'exclusive' == 1, this will fail with -EEXIST if some part of the
 658  * range already has the desired bits set.  The start of the existing
 659  * range is returned in failed_start in this case.
 660  *
 661  * [start, end] is inclusive
 662  * This takes the tree lock.
 663  */
 664 static int set_extent_bit(struct extent_io_tree *tree, u64 start, u64 end,
 665                           int bits, int exclusive, u64 *failed_start,
 666                           gfp_t mask)
 667 {
 668         struct extent_state *state;
 669         struct extent_state *prealloc = NULL;
 670         struct rb_node *node;
 671         int err = 0;
 672         int set;
 673         u64 last_start;
 674         u64 last_end;
 675 again:
 676         if (!prealloc && (mask & __GFP_WAIT)) {
 677                 prealloc = alloc_extent_state(mask);
 678                 if (!prealloc)
 679                         return -ENOMEM;
 680         }
 681
 682         spin_lock(&tree->lock);
 683         /*
 684          * this search will find all the extents that end after
 685          * our range starts.
 686          */
 687         node = tree_search(tree, start);
 688         if (!node) {
 689                 err = insert_state(tree, prealloc, start, end, bits);
 690                 prealloc = NULL;
 691                 BUG_ON(err == -EEXIST);
 692                 goto out;
 693         }
 694
 695         state = rb_entry(node, struct extent_state, rb_node);
 696         last_start = state->start;
 697         last_end = state->end;
 698
 699         /*
 700          * | ---- desired range ---- |
 701          * | state |
 702          *
 703          * Just lock what we found and keep going
 704          */
 705         if (state->start == start && state->end <= end) {
 706                 set = state->state & bits;
 707                 if (set && exclusive) {
 708                         *failed_start = state->start;
 709                         err = -EEXIST;
 710                         goto out;
 711                 }
 712                 set_state_bits(tree, state, bits);
 713                 start = state->end + 1;
 714                 merge_state(tree, state);
 715                 goto search_again;
 716         }
 717
 718         /*
 719          *     | ---- desired range ---- |
 720          * | state |
 721          *   or
 722          * | ------------- state -------------- |
 723          *
 724          * We need to split the extent we found, and may flip bits on
 725          * second half.
 726          *
 727          * If the extent we found extends past our
 728          * range, we just split and search again.  It'll get split
 729          * again the next time though.
 730          *
 731          * If the extent we found is inside our range, we set the
 732          * desired bit on it.
 733          */
 734         if (state->start < start) {
 735                 set = state->state & bits;
 736                 if (exclusive && set) {
 737                         *failed_start = start;
 738                         err = -EEXIST;
 739                         goto out;
 740                 }
 741                 err = split_state(tree, state, prealloc, start);
 742                 BUG_ON(err == -EEXIST);
 743                 prealloc = NULL;
 744                 if (err)
 745                         goto out;
 746                 if (state->end <= end) {
 747                         set_state_bits(tree, state, bits);
 748                         start = state->end + 1;
 749                         merge_state(tree, state);
 750                 } else {
 751                         start = state->start;
 752                 }
 753                 goto search_again;
 754         }
 755         /*
 756          * | ---- desired range ---- |
 757          *     | state | or               | state |
 758          *
 759          * There's a hole, we need to insert something in it and
 760          * ignore the extent we found.
 761          */
 762         if (state->start > start) {
 763                 u64 this_end;
 764                 if (end < last_start)
 765                         this_end = end;
 766                 else
 767                         this_end = last_start - 1;
 768                 err = insert_state(tree, prealloc, start, this_end,
 769                                    bits);
 770                 prealloc = NULL;
 771                 BUG_ON(err == -EEXIST);
 772                 if (err)
 773                         goto out;
 774                 start = this_end + 1;
 775                 goto search_again;
 776         }
 777         /*
 778          * | ---- desired range ---- |
 779          *                        | state |
 780          * We need to split the extent, and set the bit
 781          * on the first half
 782          */
 783         if (state->start <= end && state->end > end) {
 784                 set = state->state & bits;
 785                 if (exclusive && set) {
 786                         *failed_start = start;
 787                         err = -EEXIST;
 788                         goto out;
 789                 }
 790                 err = split_state(tree, state, prealloc, end + 1);
 791                 BUG_ON(err == -EEXIST);
 792
 793                 set_state_bits(tree, prealloc, bits);
 794                 merge_state(tree, prealloc);
 795                 prealloc = NULL;
 796                 goto out;
 797         }
 798
 799         goto search_again;
 800
 801 out:
 802         spin_unlock(&tree->lock);
 803         if (prealloc)
 804                 free_extent_state(prealloc);
 805
 806         return err;
 807
 808 search_again:
 809         if (start > end)
 810                 goto out;
 811         spin_unlock(&tree->lock);
 812         if (mask & __GFP_WAIT)
 813                 cond_resched();
 814         goto again;
 815 }
 816
 817 /* wrappers around set/clear extent bit */
 818 int set_extent_dirty(struct extent_io_tree *tree, u64 start, u64 end,
 819                      gfp_t mask)
 820 {
 821         return set_extent_bit(tree, start, end, EXTENT_DIRTY, 0, NULL,
 822                               mask);
 823 }
 824
 825 int set_extent_ordered(struct extent_io_tree *tree, u64 start, u64 end,
 826                        gfp_t mask)
 827 {
 828         return set_extent_bit(tree, start, end, EXTENT_ORDERED, 0, NULL, mask);
 829 }
 830
 831 int set_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
 832                     int bits, gfp_t mask)
 833 {
 834         return set_extent_bit(tree, start, end, bits, 0, NULL,
 835                               mask);
 836 }
 837
 838 int clear_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
 839                       int bits, gfp_t mask)
 840 {
 841         return clear_extent_bit(tree, start, end, bits, 0, 0, mask);
 842 }
 843
 844 int set_extent_delalloc(struct extent_io_tree *tree, u64 start, u64 end,
 845                      gfp_t mask)
 846 {
 847         return set_extent_bit(tree, start, end,
 848                               EXTENT_DELALLOC | EXTENT_DIRTY,
 849                               0, NULL, mask);
 850 }
 851
 852 int clear_extent_dirty(struct extent_io_tree *tree, u64 start, u64 end,
 853                        gfp_t mask)
 854 {
 855         return clear_extent_bit(tree, start, end,
 856                                 EXTENT_DIRTY | EXTENT_DELALLOC, 0, 0, mask);
 857 }
 858
 859 int clear_extent_ordered(struct extent_io_tree *tree, u64 start, u64 end,
 860                          gfp_t mask)
 861 {
 862         return clear_extent_bit(tree, start, end, EXTENT_ORDERED, 1, 0, mask);
 863 }
 864
 865 int set_extent_new(struct extent_io_tree *tree, u64 start, u64 end,
 866                      gfp_t mask)
 867 {
 868         return set_extent_bit(tree, start, end, EXTENT_NEW, 0, NULL,
 869                               mask);
 870 }
 871
 872 static int clear_extent_new(struct extent_io_tree *tree, u64 start, u64 end,
 873                        gfp_t mask)
 874 {
 875         return clear_extent_bit(tree, start, end, EXTENT_NEW, 0, 0, mask);
 876 }
 877
 878 int set_extent_uptodate(struct extent_io_tree *tree, u64 start, u64 end,
 879                         gfp_t mask)
 880 {
 881         return set_extent_bit(tree, start, end, EXTENT_UPTODATE, 0, NULL,
 882                               mask);
 883 }
 884
 885 static int clear_extent_uptodate(struct extent_io_tree *tree, u64 start,
 886                                  u64 end, gfp_t mask)
 887 {
 888         return clear_extent_bit(tree, start, end, EXTENT_UPTODATE, 0, 0, mask);
 889 }
 890
 891 static int set_extent_writeback(struct extent_io_tree *tree, u64 start, u64 end,
 892                          gfp_t mask)
 893 {
 894         return set_extent_bit(tree, start, end, EXTENT_WRITEBACK,
 895                               0, NULL, mask);
 896 }
 897
 898 static int clear_extent_writeback(struct extent_io_tree *tree, u64 start,
 899                                   u64 end, gfp_t mask)
 900 {
 901         return clear_extent_bit(tree, start, end, EXTENT_WRITEBACK, 1, 0, mask);
 902 }
 903
 904 int wait_on_extent_writeback(struct extent_io_tree *tree, u64 start, u64 end)
 905 {
 906         return wait_extent_bit(tree, start, end, EXTENT_WRITEBACK);
 907 }
 908
 909 /*
 910  * either insert or lock state struct between start and end use mask to tell
 911  * us if waiting is desired.
 912  */
 913 int lock_extent(struct extent_io_tree *tree, u64 start, u64 end, gfp_t mask)
 914 {
 915         int err;
 916         u64 failed_start;
 917         while (1) {
 918                 err = set_extent_bit(tree, start, end, EXTENT_LOCKED, 1,
 919                                      &failed_start, mask);
 920                 if (err == -EEXIST && (mask & __GFP_WAIT)) {
 921                         wait_extent_bit(tree, failed_start, end, EXTENT_LOCKED);
 922                         start = failed_start;
 923                 } else {
 924                         break;
 925                 }
 926                 WARN_ON(start > end);
 927         }
 928         return err;
 929 }
 930
 931 int try_lock_extent(struct extent_io_tree *tree, u64 start, u64 end,
 932                     gfp_t mask)
 933 {
 934         int err;
 935         u64 failed_start;
 936
 937         err = set_extent_bit(tree, start, end, EXTENT_LOCKED, 1,
 938                              &failed_start, mask);
 939         if (err == -EEXIST) {
 940                 if (failed_start > start)
 941                         clear_extent_bit(tree, start, failed_start - 1,
 942                                          EXTENT_LOCKED, 1, 0, mask);
 943                 return 0;
 944         }
 945         return 1;
 946 }
 947
 948 int unlock_extent(struct extent_io_tree *tree, u64 start, u64 end,
 949                   gfp_t mask)
 950 {
 951         return clear_extent_bit(tree, start, end, EXTENT_LOCKED, 1, 0, mask);
 952 }
 953
 954 /*
 955  * helper function to set pages and extents in the tree dirty
 956  */
 957 int set_range_dirty(struct extent_io_tree *tree, u64 start, u64 end)
 958 {
 959         unsigned long index = start >> PAGE_CACHE_SHIFT;
 960         unsigned long end_index = end >> PAGE_CACHE_SHIFT;
 961         struct page *page;
 962
 963         while (index <= end_index) {
 964                 page = find_get_page(tree->mapping, index);
 965                 BUG_ON(!page);
 966                 __set_page_dirty_nobuffers(page);
 967                 page_cache_release(page);
 968                 index++;
 969         }
 970         set_extent_dirty(tree, start, end, GFP_NOFS);
 971         return 0;
 972 }
 973
 974 /*
 975  * helper function to set both pages and extents in the tree writeback
 976  */
 977 static int set_range_writeback(struct extent_io_tree *tree, u64 start, u64 end)
 978 {
 979         unsigned long index = start >> PAGE_CACHE_SHIFT;
 980         unsigned long end_index = end >> PAGE_CACHE_SHIFT;
 981         struct page *page;
 982
 983         while (index <= end_index) {
 984                 page = find_get_page(tree->mapping, index);
 985                 BUG_ON(!page);
 986                 set_page_writeback(page);
 987                 page_cache_release(page);
 988                 index++;
 989         }
 990         set_extent_writeback(tree, start, end, GFP_NOFS);
 991         return 0;
 992 }
 993
 994 /*
 995  * find the first offset in the io tree with 'bits' set. zero is
 996  * returned if we find something, and *start_ret and *end_ret are
 997  * set to reflect the state struct that was found.
 998  *
 999  * If nothing was found, 1 is returned, < 0 on error
1000  */
1001 int find_first_extent_bit(struct extent_io_tree *tree, u64 start,
1002                           u64 *start_ret, u64 *end_ret, int bits)
1003 {
1004         struct rb_node *node;
1005         struct extent_state *state;
1006         int ret = 1;
1007
1008         spin_lock(&tree->lock);
1009         /*
1010          * this search will find all the extents that end after
1011          * our range starts.
1012          */
1013         node = tree_search(tree, start);
1014         if (!node)
1015                 goto out;
1016
1017         while (1) {
1018                 state = rb_entry(node, struct extent_state, rb_node);
1019                 if (state->end >= start && (state->state & bits)) {
1020                         *start_ret = state->start;
1021                         *end_ret = state->end;
1022                         ret = 0;
1023                         break;
1024                 }
1025                 node = rb_next(node);
1026                 if (!node)
1027                         break;
1028         }
1029 out:
1030         spin_unlock(&tree->lock);
1031         return ret;
1032 }
1033
1034 /* find the first state struct with 'bits' set after 'start', and
1035  * return it.  tree->lock must be held.  NULL will returned if
1036  * nothing was found after 'start'
1037  */
1038 struct extent_state *find_first_extent_bit_state(struct extent_io_tree *tree,
1039                                                  u64 start, int bits)
1040 {
1041         struct rb_node *node;
1042         struct extent_state *state;
1043
1044         /*
1045          * this search will find all the extents that end after
1046          * our range starts.
1047          */
1048         node = tree_search(tree, start);
1049         if (!node)
1050                 goto out;
1051
1052         while (1) {
1053                 state = rb_entry(node, struct extent_state, rb_node);
1054                 if (state->end >= start && (state->state & bits))
1055                         return state;
1056
1057                 node = rb_next(node);
1058                 if (!node)
1059                         break;
1060         }
1061 out:
1062         return NULL;
1063 }
1064
1065 /*
1066  * find a contiguous range of bytes in the file marked as delalloc, not
1067  * more than 'max_bytes'.  start and end are used to return the range,
1068  *
1069  * 1 is returned if we find something, 0 if nothing was in the tree
1070  */
1071 static noinline u64 find_delalloc_range(struct extent_io_tree *tree,
1072                                         u64 *start, u64 *end, u64 max_bytes)
1073 {
1074         struct rb_node *node;
1075         struct extent_state *state;
1076         u64 cur_start = *start;
1077         u64 found = 0;
1078         u64 total_bytes = 0;
1079
1080         spin_lock(&tree->lock);
1081
1082         /*
1083          * this search will find all the extents that end after
1084          * our range starts.
1085          */
1086         node = tree_search(tree, cur_start);
1087         if (!node) {
1088                 if (!found)
1089                         *end = (u64)-1;
1090                 goto out;
1091         }
1092
1093         while (1) {
1094                 state = rb_entry(node, struct extent_state, rb_node);
1095                 if (found && (state->start != cur_start ||
1096                               (state->state & EXTENT_BOUNDARY))) {
1097                         goto out;
1098                 }
1099                 if (!(state->state & EXTENT_DELALLOC)) {
1100                         if (!found)
1101                                 *end = state->end;
1102                         goto out;
1103                 }
1104                 if (!found)
1105                         *start = state->start;
1106                 found++;
1107                 *end = state->end;
1108                 cur_start = state->end + 1;
1109                 node = rb_next(node);
1110                 if (!node)
1111                         break;
1112                 total_bytes += state->end - state->start + 1;
1113                 if (total_bytes >= max_bytes)
1114                         break;
1115         }
1116 out:
1117         spin_unlock(&tree->lock);
1118         return found;
1119 }
1120
1121 static noinline int __unlock_for_delalloc(struct inode *inode,
1122                                           struct page *locked_page,
1123                                           u64 start, u64 end)
1124 {
1125         int ret;
1126         struct page *pages[16];
1127         unsigned long index = start >> PAGE_CACHE_SHIFT;
1128         unsigned long end_index = end >> PAGE_CACHE_SHIFT;
1129         unsigned long nr_pages = end_index - index + 1;
1130         int i;
1131
1132         if (index == locked_page->index && end_index == index)
1133                 return 0;
1134
1135         while (nr_pages > 0) {
1136                 ret = find_get_pages_contig(inode->i_mapping, index,
1137                                      min_t(unsigned long, nr_pages,
1138                                      ARRAY_SIZE(pages)), pages);
1139                 for (i = 0; i < ret; i++) {
1140                         if (pages[i] != locked_page)
1141                                 unlock_page(pages[i]);
1142                         page_cache_release(pages[i]);
1143                 }
1144                 nr_pages -= ret;
1145                 index += ret;
1146                 cond_resched();
1147         }
1148         return 0;
1149 }
1150
1151 static noinline int lock_delalloc_pages(struct inode *inode,
1152                                         struct page *locked_page,
1153                                         u64 delalloc_start,
1154                                         u64 delalloc_end)
1155 {
1156         unsigned long index = delalloc_start >> PAGE_CACHE_SHIFT;
1157         unsigned long start_index = index;
1158         unsigned long end_index = delalloc_end >> PAGE_CACHE_SHIFT;
1159         unsigned long pages_locked = 0;
1160         struct page *pages[16];
1161         unsigned long nrpages;
1162         int ret;
1163         int i;
1164
1165         /* the caller is responsible for locking the start index */
1166         if (index == locked_page->index && index == end_index)
1167                 return 0;
1168
1169         /* skip the page at the start index */
1170         nrpages = end_index - index + 1;
1171         while (nrpages > 0) {
1172                 ret = find_get_pages_contig(inode->i_mapping, index,
1173                                      min_t(unsigned long,
1174                                      nrpages, ARRAY_SIZE(pages)), pages);
1175                 if (ret == 0) {
1176                         ret = -EAGAIN;
1177                         goto done;
1178                 }
1179                 /* now we have an array of pages, lock them all */
1180                 for (i = 0; i < ret; i++) {
1181                         /*
1182                          * the caller is taking responsibility for
1183                          * locked_page
1184                          */
1185                         if (pages[i] != locked_page) {
1186                                 lock_page(pages[i]);
1187                                 if (!PageDirty(pages[i]) ||
1188                                     pages[i]->mapping != inode->i_mapping) {
1189                                         ret = -EAGAIN;
1190                                         unlock_page(pages[i]);
1191                                         page_cache_release(pages[i]);
1192                                         goto done;
1193                                 }
1194                         }
1195                         page_cache_release(pages[i]);
1196                         pages_locked++;
1197                 }
1198                 nrpages -= ret;
1199                 index += ret;
1200                 cond_resched();
1201         }
1202         ret = 0;
1203 done:
1204         if (ret && pages_locked) {
1205                 __unlock_for_delalloc(inode, locked_page,
1206                               delalloc_start,
1207                               ((u64)(start_index + pages_locked - 1)) <<
1208                               PAGE_CACHE_SHIFT);
1209         }
1210         return ret;
1211 }
1212
1213 /*
1214  * find a contiguous range of bytes in the file marked as delalloc, not
1215  * more than 'max_bytes'.  start and end are used to return the range,
1216  *
1217  * 1 is returned if we find something, 0 if nothing was in the tree
1218  */
1219 static noinline u64 find_lock_delalloc_range(struct inode *inode,
1220                                              struct extent_io_tree *tree,
1221                                              struct page *locked_page,
1222                                              u64 *start, u64 *end,
1223                                              u64 max_bytes)
1224 {
1225         u64 delalloc_start;
1226         u64 delalloc_end;
1227         u64 found;
1228         int ret;
1229         int loops = 0;
1230
1231 again:
1232         /* step one, find a bunch of delalloc bytes starting at start */
1233         delalloc_start = *start;
1234         delalloc_end = 0;
1235         found = find_delalloc_range(tree, &delalloc_start, &delalloc_end,
1236                                     max_bytes);
1237         if (!found || delalloc_end <= *start) {
1238                 *start = delalloc_start;
1239                 *end = delalloc_end;
1240                 return found;
1241         }
1242
1243         /*
1244          * start comes from the offset of locked_page.  We have to lock
1245          * pages in order, so we can't process delalloc bytes before
1246          * locked_page
1247          */
1248         if (delalloc_start < *start)
1249                 delalloc_start = *start;
1250
1251         /*
1252          * make sure to limit the number of pages we try to lock down
1253          * if we're looping.
1254          */
1255         if (delalloc_end + 1 - delalloc_start > max_bytes && loops)
1256                 delalloc_end = delalloc_start + PAGE_CACHE_SIZE - 1;
1257
1258         /* step two, lock all the pages after the page that has start */
1259         ret = lock_delalloc_pages(inode, locked_page,
1260                                   delalloc_start, delalloc_end);
1261         if (ret == -EAGAIN) {
1262                 /* some of the pages are gone, lets avoid looping by
1263                  * shortening the size of the delalloc range we're searching
1264                  */
1265                 if (!loops) {
1266                         unsigned long offset = (*start) & (PAGE_CACHE_SIZE - 1);
1267                         max_bytes = PAGE_CACHE_SIZE - offset;
1268                         loops = 1;
1269                         goto again;
1270                 } else {
1271                         found = 0;
1272                         goto out_failed;
1273                 }
1274         }
1275         BUG_ON(ret);
1276
1277         /* step three, lock the state bits for the whole range */
1278         lock_extent(tree, delalloc_start, delalloc_end, GFP_NOFS);
1279
1280         /* then test to make sure it is all still delalloc */
1281         ret = test_range_bit(tree, delalloc_start, delalloc_end,
1282                              EXTENT_DELALLOC, 1);
1283         if (!ret) {
1284                 unlock_extent(tree, delalloc_start, delalloc_end, GFP_NOFS);
1285                 __unlock_for_delalloc(inode, locked_page,
1286                               delalloc_start, delalloc_end);
1287                 cond_resched();
1288                 goto again;
1289         }
1290         *start = delalloc_start;
1291         *end = delalloc_end;
1292 out_failed:
1293         return found;
1294 }
1295
1296 int extent_clear_unlock_delalloc(struct inode *inode,
1297                                 struct extent_io_tree *tree,
1298                                 u64 start, u64 end, struct page *locked_page,
1299                                 int unlock_pages,
1300                                 int clear_unlock,
1301                                 int clear_delalloc, int clear_dirty,
1302                                 int set_writeback,
1303                                 int end_writeback)
1304 {
1305         int ret;
1306         struct page *pages[16];
1307         unsigned long index = start >> PAGE_CACHE_SHIFT;
1308         unsigned long end_index = end >> PAGE_CACHE_SHIFT;
1309         unsigned long nr_pages = end_index - index + 1;
1310         int i;
1311         int clear_bits = 0;
1312
1313         if (clear_unlock)
1314                 clear_bits |= EXTENT_LOCKED;
1315         if (clear_dirty)
1316                 clear_bits |= EXTENT_DIRTY;
1317
1318         if (clear_delalloc)
1319                 clear_bits |= EXTENT_DELALLOC;
1320
1321         clear_extent_bit(tree, start, end, clear_bits, 1, 0, GFP_NOFS);
1322         if (!(unlock_pages || clear_dirty || set_writeback || end_writeback))
1323                 return 0;
1324
1325         while (nr_pages > 0) {
1326                 ret = find_get_pages_contig(inode->i_mapping, index,
1327                                      min_t(unsigned long,
1328                                      nr_pages, ARRAY_SIZE(pages)), pages);
1329                 for (i = 0; i < ret; i++) {
1330                         if (pages[i] == locked_page) {
1331                                 page_cache_release(pages[i]);
1332                                 continue;
1333                         }
1334                         if (clear_dirty)
1335                                 clear_page_dirty_for_io(pages[i]);
1336                         if (set_writeback)
1337                                 set_page_writeback(pages[i]);
1338                         if (end_writeback)
1339                                 end_page_writeback(pages[i]);
1340                         if (unlock_pages)
1341                                 unlock_page(pages[i]);
1342                         page_cache_release(pages[i]);
1343                 }
1344                 nr_pages -= ret;
1345                 index += ret;
1346                 cond_resched();
1347         }
1348         return 0;
1349 }
1350
1351 /*
1352  * count the number of bytes in the tree that have a given bit(s)
1353  * set.  This can be fairly slow, except for EXTENT_DIRTY which is
1354  * cached.  The total number found is returned.
1355  */
1356 u64 count_range_bits(struct extent_io_tree *tree,
1357                      u64 *start, u64 search_end, u64 max_bytes,
1358                      unsigned long bits)
1359 {
1360         struct rb_node *node;
1361         struct extent_state *state;
1362         u64 cur_start = *start;
1363         u64 total_bytes = 0;
1364         int found = 0;
1365
1366         if (search_end <= cur_start) {
1367                 WARN_ON(1);
1368                 return 0;
1369         }
1370
1371         spin_lock(&tree->lock);
1372         if (cur_start == 0 && bits == EXTENT_DIRTY) {
1373                 total_bytes = tree->dirty_bytes;
1374                 goto out;
1375         }
1376         /*
1377          * this search will find all the extents that end after
1378          * our range starts.
1379          */
1380         node = tree_search(tree, cur_start);
1381         if (!node)
1382                 goto out;
1383
1384         while (1) {
1385                 state = rb_entry(node, struct extent_state, rb_node);
1386                 if (state->start > search_end)
1387                         break;
1388                 if (state->end >= cur_start && (state->state & bits)) {
1389                         total_bytes += min(search_end, state->end) + 1 -
1390                                        max(cur_start, state->start);
1391                         if (total_bytes >= max_bytes)
1392                                 break;
1393                         if (!found) {
1394                                 *start = state->start;
1395                                 found = 1;
1396                         }
1397                 }
1398                 node = rb_next(node);
1399                 if (!node)
1400                         break;
1401         }
1402 out:
1403         spin_unlock(&tree->lock);
1404         return total_bytes;
1405 }
1406
1407 #if 0
1408 /*
1409  * helper function to lock both pages and extents in the tree.
1410  * pages must be locked first.
1411  */
1412 static int lock_range(struct extent_io_tree *tree, u64 start, u64 end)
1413 {
1414         unsigned long index = start >> PAGE_CACHE_SHIFT;
1415         unsigned long end_index = end >> PAGE_CACHE_SHIFT;
1416         struct page *page;
1417         int err;
1418
1419         while (index <= end_index) {
1420                 page = grab_cache_page(tree->mapping, index);
1421                 if (!page) {
1422                         err = -ENOMEM;
1423                         goto failed;
1424                 }
1425                 if (IS_ERR(page)) {
1426                         err = PTR_ERR(page);
1427                         goto failed;
1428                 }
1429                 index++;
1430         }
1431         lock_extent(tree, start, end, GFP_NOFS);
1432         return 0;
1433
1434 failed:
1435         /*
1436          * we failed above in getting the page at 'index', so we undo here
1437          * up to but not including the page at 'index'
1438          */
1439         end_index = index;
1440         index = start >> PAGE_CACHE_SHIFT;
1441         while (index < end_index) {
1442                 page = find_get_page(tree->mapping, index);
1443                 unlock_page(page);
1444                 page_cache_release(page);
1445                 index++;
1446         }
1447         return err;
1448 }
1449
1450 /*
1451  * helper function to unlock both pages and extents in the tree.
1452  */
1453 static int unlock_range(struct extent_io_tree *tree, u64 start, u64 end)
1454 {
1455         unsigned long index = start >> PAGE_CACHE_SHIFT;
1456         unsigned long end_index = end >> PAGE_CACHE_SHIFT;
1457         struct page *page;
1458
1459         while (index <= end_index) {
1460                 page = find_get_page(tree->mapping, index);
1461                 unlock_page(page);
1462                 page_cache_release(page);
1463                 index++;
1464         }
1465         unlock_extent(tree, start, end, GFP_NOFS);
1466         return 0;
1467 }
1468 #endif
1469
1470 /*
1471  * set the private field for a given byte offset in the tree.  If there isn't
1472  * an extent_state there already, this does nothing.
1473  */
1474 int set_state_private(struct extent_io_tree *tree, u64 start, u64 private)
1475 {
1476         struct rb_node *node;
1477         struct extent_state *state;
1478         int ret = 0;
1479
1480         spin_lock(&tree->lock);
1481         /*
1482          * this search will find all the extents that end after
1483          * our range starts.
1484          */
1485         node = tree_search(tree, start);
1486         if (!node) {
1487                 ret = -ENOENT;
1488                 goto out;
1489         }
1490         state = rb_entry(node, struct extent_state, rb_node);
1491         if (state->start != start) {
1492                 ret = -ENOENT;
1493                 goto out;
1494         }
1495         state->private = private;
1496 out:
1497         spin_unlock(&tree->lock);
1498         return ret;
1499 }
1500
1501 int get_state_private(struct extent_io_tree *tree, u64 start, u64 *private)
1502 {
1503         struct rb_node *node;
1504         struct extent_state *state;
1505         int ret = 0;
1506
1507         spin_lock(&tree->lock);
1508         /*
1509          * this search will find all the extents that end after
1510          * our range starts.
1511          */
1512         node = tree_search(tree, start);
1513         if (!node) {
1514                 ret = -ENOENT;
1515                 goto out;
1516         }
1517         state = rb_entry(node, struct extent_state, rb_node);
1518         if (state->start != start) {
1519                 ret = -ENOENT;
1520                 goto out;
1521         }
1522         *private = state->private;
1523 out:
1524         spin_unlock(&tree->lock);
1525         return ret;
1526 }
1527
1528 /*
1529  * searches a range in the state tree for a given mask.
1530  * If 'filled' == 1, this returns 1 only if every extent in the tree
1531  * has the bits set.  Otherwise, 1 is returned if any bit in the
1532  * range is found set.
1533  */
1534 int test_range_bit(struct extent_io_tree *tree, u64 start, u64 end,
1535                    int bits, int filled)
1536 {
1537         struct extent_state *state = NULL;
1538         struct rb_node *node;
1539         int bitset = 0;
1540
1541         spin_lock(&tree->lock);
1542         node = tree_search(tree, start);
1543         while (node && start <= end) {
1544                 state = rb_entry(node, struct extent_state, rb_node);
1545
1546                 if (filled && state->start > start) {
1547                         bitset = 0;
1548                         break;
1549                 }
1550
1551                 if (state->start > end)
1552                         break;
1553
1554                 if (state->state & bits) {
1555                         bitset = 1;
1556                         if (!filled)
1557                                 break;
1558                 } else if (filled) {
1559                         bitset = 0;
1560                         break;
1561                 }
1562                 start = state->end + 1;
1563                 if (start > end)
1564                         break;
1565                 node = rb_next(node);
1566                 if (!node) {
1567                         if (filled)
1568                                 bitset = 0;
1569                         break;
1570                 }
1571         }
1572         spin_unlock(&tree->lock);
1573         return bitset;
1574 }
1575
1576 /*
1577  * helper function to set a given page up to date if all the
1578  * extents in the tree for that page are up to date
1579  */
1580 static int check_page_uptodate(struct extent_io_tree *tree,
1581                                struct page *page)
1582 {
1583         u64 start = (u64)page->index << PAGE_CACHE_SHIFT;
1584         u64 end = start + PAGE_CACHE_SIZE - 1;
1585         if (test_range_bit(tree, start, end, EXTENT_UPTODATE, 1))
1586                 SetPageUptodate(page);
1587         return 0;
1588 }
1589
1590 /*
1591  * helper function to unlock a page if all the extents in the tree
1592  * for that page are unlocked
1593  */
1594 static int check_page_locked(struct extent_io_tree *tree,
1595                              struct page *page)
1596 {
1597         u64 start = (u64)page->index << PAGE_CACHE_SHIFT;
1598         u64 end = start + PAGE_CACHE_SIZE - 1;
1599         if (!test_range_bit(tree, start, end, EXTENT_LOCKED, 0))
1600                 unlock_page(page);
1601         return 0;
1602 }
1603
1604 /*
1605  * helper function to end page writeback if all the extents
1606  * in the tree for that page are done with writeback
1607  */
1608 static int check_page_writeback(struct extent_io_tree *tree,
1609                              struct page *page)
1610 {
1611         u64 start = (u64)page->index << PAGE_CACHE_SHIFT;
1612         u64 end = start + PAGE_CACHE_SIZE - 1;
1613         if (!test_range_bit(tree, start, end, EXTENT_WRITEBACK, 0))
1614                 end_page_writeback(page);
1615         return 0;
1616 }
1617
1618 /* lots and lots of room for performance fixes in the end_bio funcs */
1619
1620 /*
1621  * after a writepage IO is done, we need to:
1622  * clear the uptodate bits on error
1623  * clear the writeback bits in the extent tree for this IO
1624  * end_page_writeback if the page has no more pending IO
1625  *
1626  * Scheduling is not allowed, so the extent state tree is expected
1627  * to have one and only one object corresponding to this IO.
1628  */
1629 static void end_bio_extent_writepage(struct bio *bio, int err)
1630 {
1631         int uptodate = err == 0;
1632         struct bio_vec *bvec = bio->bi_io_vec + bio->bi_vcnt - 1;
1633         struct extent_io_tree *tree;
1634         u64 start;
1635         u64 end;
1636         int whole_page;
1637         int ret;
1638
1639         do {
1640                 struct page *page = bvec->bv_page;
1641                 tree = &BTRFS_I(page->mapping->host)->io_tree;
1642
1643                 start = ((u64)page->index << PAGE_CACHE_SHIFT) +
1644                          bvec->bv_offset;
1645                 end = start + bvec->bv_len - 1;
1646
1647                 if (bvec->bv_offset == 0 && bvec->bv_len == PAGE_CACHE_SIZE)
1648                         whole_page = 1;
1649                 else
1650                         whole_page = 0;
1651
1652                 if (--bvec >= bio->bi_io_vec)
1653                         prefetchw(&bvec->bv_page->flags);
1654                 if (tree->ops && tree->ops->writepage_end_io_hook) {
1655                         ret = tree->ops->writepage_end_io_hook(page, start,
1656                                                        end, NULL, uptodate);
1657                         if (ret)
1658                                 uptodate = 0;
1659                 }
1660
1661                 if (!uptodate && tree->ops &&
1662                     tree->ops->writepage_io_failed_hook) {
1663                         ret = tree->ops->writepage_io_failed_hook(bio, page,
1664                                                          start, end, NULL);
1665                         if (ret == 0) {
1666                                 uptodate = (err == 0);
1667                                 continue;
1668                         }
1669                 }
1670
1671                 if (!uptodate) {
1672                         clear_extent_uptodate(tree, start, end, GFP_ATOMIC);
1673                         ClearPageUptodate(page);
1674                         SetPageError(page);
1675                 }
1676
1677                 clear_extent_writeback(tree, start, end, GFP_ATOMIC);
1678
1679                 if (whole_page)
1680                         end_page_writeback(page);
1681                 else
1682                         check_page_writeback(tree, page);
1683         } while (bvec >= bio->bi_io_vec);
1684
1685         bio_put(bio);
1686 }
1687
1688 /*
1689  * after a readpage IO is done, we need to:
1690  * clear the uptodate bits on error
1691  * set the uptodate bits if things worked
1692  * set the page up to date if all extents in the tree are uptodate
1693  * clear the lock bit in the extent tree
1694  * unlock the page if there are no other extents locked for it
1695  *
1696  * Scheduling is not allowed, so the extent state tree is expected
1697  * to have one and only one object corresponding to this IO.
1698  */
1699 static void end_bio_extent_readpage(struct bio *bio, int err)
1700 {
1701         int uptodate = test_bit(BIO_UPTODATE, &bio->bi_flags);
1702         struct bio_vec *bvec = bio->bi_io_vec + bio->bi_vcnt - 1;
1703         struct extent_io_tree *tree;
1704         u64 start;
1705         u64 end;
1706         int whole_page;
1707         int ret;
1708
1709         if (err)
1710                 uptodate = 0;
1711
1712         do {
1713                 struct page *page = bvec->bv_page;
1714                 tree = &BTRFS_I(page->mapping->host)->io_tree;
1715
1716                 start = ((u64)page->index << PAGE_CACHE_SHIFT) +
1717                         bvec->bv_offset;
1718                 end = start + bvec->bv_len - 1;
1719
1720                 if (bvec->bv_offset == 0 && bvec->bv_len == PAGE_CACHE_SIZE)
1721                         whole_page = 1;
1722                 else
1723                         whole_page = 0;
1724
1725                 if (--bvec >= bio->bi_io_vec)
1726                         prefetchw(&bvec->bv_page->flags);
1727
1728                 if (uptodate && tree->ops && tree->ops->readpage_end_io_hook) {
1729                         ret = tree->ops->readpage_end_io_hook(page, start, end,
1730                                                               NULL);
1731                         if (ret)
1732                                 uptodate = 0;
1733                 }
1734                 if (!uptodate && tree->ops &&
1735                     tree->ops->readpage_io_failed_hook) {
1736                         ret = tree->ops->readpage_io_failed_hook(bio, page,
1737                                                          start, end, NULL);
1738                         if (ret == 0) {
1739                                 uptodate =
1740                                         test_bit(BIO_UPTODATE, &bio->bi_flags);
1741                                 if (err)
1742                                         uptodate = 0;
1743                                 continue;
1744                         }
1745                 }
1746
1747                 if (uptodate) {
1748                         set_extent_uptodate(tree, start, end,
1749                                             GFP_ATOMIC);
1750                 }
1751                 unlock_extent(tree, start, end, GFP_ATOMIC);
1752
1753                 if (whole_page) {
1754                         if (uptodate) {
1755                                 SetPageUptodate(page);
1756                         } else {
1757                                 ClearPageUptodate(page);
1758                                 SetPageError(page);
1759                         }
1760                         unlock_page(page);
1761                 } else {
1762                         if (uptodate) {
1763                                 check_page_uptodate(tree, page);
1764                         } else {
1765                                 ClearPageUptodate(page);
1766                                 SetPageError(page);
1767                         }
1768                         check_page_locked(tree, page);
1769                 }
1770         } while (bvec >= bio->bi_io_vec);
1771
1772         bio_put(bio);
1773 }
1774
1775 /*
1776  * IO done from prepare_write is pretty simple, we just unlock
1777  * the structs in the extent tree when done, and set the uptodate bits
1778  * as appropriate.
1779  */
1780 static void end_bio_extent_preparewrite(struct bio *bio, int err)
1781 {
1782         const int uptodate = test_bit(BIO_UPTODATE, &bio->bi_flags);
1783         struct bio_vec *bvec = bio->bi_io_vec + bio->bi_vcnt - 1;
1784         struct extent_io_tree *tree;
1785         u64 start;
1786         u64 end;
1787
1788         do {
1789                 struct page *page = bvec->bv_page;
1790                 tree = &BTRFS_I(page->mapping->host)->io_tree;
1791
1792                 start = ((u64)page->index << PAGE_CACHE_SHIFT) +
1793                         bvec->bv_offset;
1794                 end = start + bvec->bv_len - 1;
1795
1796                 if (--bvec >= bio->bi_io_vec)
1797                         prefetchw(&bvec->bv_page->flags);
1798
1799                 if (uptodate) {
1800                         set_extent_uptodate(tree, start, end, GFP_ATOMIC);
1801                 } else {
1802                         ClearPageUptodate(page);
1803                         SetPageError(page);
1804                 }
1805
1806                 unlock_extent(tree, start, end, GFP_ATOMIC);
1807
1808         } while (bvec >= bio->bi_io_vec);
1809
1810         bio_put(bio);
1811 }
1812
1813 static struct bio *
1814 extent_bio_alloc(struct block_device *bdev, u64 first_sector, int nr_vecs,
1815                  gfp_t gfp_flags)
1816 {
1817         struct bio *bio;
1818
1819         bio = bio_alloc(gfp_flags, nr_vecs);
1820
1821         if (bio == NULL && (current->flags & PF_MEMALLOC)) {
1822                 while (!bio && (nr_vecs /= 2))
1823                         bio = bio_alloc(gfp_flags, nr_vecs);
1824         }
1825
1826         if (bio) {
1827                 bio->bi_size = 0;
1828                 bio->bi_bdev = bdev;
1829                 bio->bi_sector = first_sector;
1830         }
1831         return bio;
1832 }
1833
1834 static int submit_one_bio(int rw, struct bio *bio, int mirror_num,
1835                           unsigned long bio_flags)
1836 {
1837         int ret = 0;
1838         struct bio_vec *bvec = bio->bi_io_vec + bio->bi_vcnt - 1;
1839         struct page *page = bvec->bv_page;
1840         struct extent_io_tree *tree = bio->bi_private;
1841         u64 start;
1842         u64 end;
1843
1844         start = ((u64)page->index << PAGE_CACHE_SHIFT) + bvec->bv_offset;
1845         end = start + bvec->bv_len - 1;
1846
1847         bio->bi_private = NULL;
1848
1849         bio_get(bio);
1850
1851         if (tree->ops && tree->ops->submit_bio_hook)
1852                 tree->ops->submit_bio_hook(page->mapping->host, rw, bio,
1853                                            mirror_num, bio_flags);
1854         else
1855                 submit_bio(rw, bio);
1856         if (bio_flagged(bio, BIO_EOPNOTSUPP))
1857                 ret = -EOPNOTSUPP;
1858         bio_put(bio);
1859         return ret;
1860 }
1861
1862 static int submit_extent_page(int rw, struct extent_io_tree *tree,
1863                               struct page *page, sector_t sector,
1864                               size_t size, unsigned long offset,
1865                               struct block_device *bdev,
1866                               struct bio **bio_ret,
1867                               unsigned long max_pages,
1868                               bio_end_io_t end_io_func,
1869                               int mirror_num,
1870                               unsigned long prev_bio_flags,
1871                               unsigned long bio_flags)
1872 {
1873         int ret = 0;
1874         struct bio *bio;
1875         int nr;
1876         int contig = 0;
1877         int this_compressed = bio_flags & EXTENT_BIO_COMPRESSED;
1878         int old_compressed = prev_bio_flags & EXTENT_BIO_COMPRESSED;
1879         size_t page_size = min_t(size_t, size, PAGE_CACHE_SIZE);
1880
1881         if (bio_ret && *bio_ret) {
1882                 bio = *bio_ret;
1883                 if (old_compressed)
1884                         contig = bio->bi_sector == sector;
1885                 else
1886                         contig = bio->bi_sector + (bio->bi_size >> 9) ==
1887                                 sector;
1888
1889                 if (prev_bio_flags != bio_flags || !contig ||
1890                     (tree->ops && tree->ops->merge_bio_hook &&
1891                      tree->ops->merge_bio_hook(page, offset, page_size, bio,
1892                                                bio_flags)) ||
1893                     bio_add_page(bio, page, page_size, offset) < page_size) {
1894                         ret = submit_one_bio(rw, bio, mirror_num,
1895                                              prev_bio_flags);
1896                         bio = NULL;
1897                 } else {
1898                         return 0;
1899                 }
1900         }
1901         if (this_compressed)
1902                 nr = BIO_MAX_PAGES;
1903         else
1904                 nr = bio_get_nr_vecs(bdev);
1905
1906         bio = extent_bio_alloc(bdev, sector, nr, GFP_NOFS | __GFP_HIGH);
1907
1908         bio_add_page(bio, page, page_size, offset);
1909         bio->bi_end_io = end_io_func;
1910         bio->bi_private = tree;
1911
1912         if (bio_ret)
1913                 *bio_ret = bio;
1914         else
1915                 ret = submit_one_bio(rw, bio, mirror_num, bio_flags);
1916
1917         return ret;
1918 }
1919
1920 void set_page_extent_mapped(struct page *page)
1921 {
1922         if (!PagePrivate(page)) {
1923                 SetPagePrivate(page);
1924                 page_cache_get(page);
1925                 set_page_private(page, EXTENT_PAGE_PRIVATE);
1926         }
1927 }
1928
1929 static void set_page_extent_head(struct page *page, unsigned long len)
1930 {
1931         set_page_private(page, EXTENT_PAGE_PRIVATE_FIRST_PAGE | len << 2);
1932 }
1933
1934 /*
1935  * basic readpage implementation.  Locked extent state structs are inserted
1936  * into the tree that are removed when the IO is done (by the end_io
1937  * handlers)
1938  */
1939 static int __extent_read_full_page(struct extent_io_tree *tree,
1940                                    struct page *page,
1941                                    get_extent_t *get_extent,
1942                                    struct bio **bio, int mirror_num,
1943                                    unsigned long *bio_flags)
1944 {
1945         struct inode *inode = page->mapping->host;
1946         u64 start = (u64)page->index << PAGE_CACHE_SHIFT;
1947         u64 page_end = start + PAGE_CACHE_SIZE - 1;
1948         u64 end;
1949         u64 cur = start;
1950         u64 extent_offset;
1951         u64 last_byte = i_size_read(inode);
1952         u64 block_start;
1953         u64 cur_end;
1954         sector_t sector;
1955         struct extent_map *em;
1956         struct block_device *bdev;
1957         int ret;
1958         int nr = 0;
1959         size_t page_offset = 0;
1960         size_t iosize;
1961         size_t disk_io_size;
1962         size_t blocksize = inode->i_sb->s_blocksize;
1963         unsigned long this_bio_flag = 0;
1964
1965         set_page_extent_mapped(page);
1966
1967         end = page_end;
1968         lock_extent(tree, start, end, GFP_NOFS);
1969
1970         if (page->index == last_byte >> PAGE_CACHE_SHIFT) {
1971                 char *userpage;
1972                 size_t zero_offset = last_byte & (PAGE_CACHE_SIZE - 1);
1973
1974                 if (zero_offset) {
1975                         iosize = PAGE_CACHE_SIZE - zero_offset;
1976                         userpage = kmap_atomic(page, KM_USER0);
1977                         memset(userpage + zero_offset, 0, iosize);
1978                         flush_dcache_page(page);
1979                         kunmap_atomic(userpage, KM_USER0);
1980                 }
1981         }
1982         while (cur <= end) {
1983                 if (cur >= last_byte) {
1984                         char *userpage;
1985                         iosize = PAGE_CACHE_SIZE - page_offset;
1986                         userpage = kmap_atomic(page, KM_USER0);
1987                         memset(userpage + page_offset, 0, iosize);
1988                         flush_dcache_page(page);
1989                         kunmap_atomic(userpage, KM_USER0);
1990                         set_extent_uptodate(tree, cur, cur + iosize - 1,
1991                                             GFP_NOFS);
1992                         unlock_extent(tree, cur, cur + iosize - 1, GFP_NOFS);
1993                         break;
1994                 }
1995                 em = get_extent(inode, page, page_offset, cur,
1996                                 end - cur + 1, 0);
1997                 if (IS_ERR(em) || !em) {
1998                         SetPageError(page);
1999                         unlock_extent(tree, cur, end, GFP_NOFS);
2000                         break;
2001                 }
2002                 extent_offset = cur - em->start;
2003                 BUG_ON(extent_map_end(em) <= cur);
2004                 BUG_ON(end < cur);
2005
2006                 if (test_bit(EXTENT_FLAG_COMPRESSED, &em->flags))
2007                         this_bio_flag = EXTENT_BIO_COMPRESSED;
2008
2009                 iosize = min(extent_map_end(em) - cur, end - cur + 1);
2010                 cur_end = min(extent_map_end(em) - 1, end);
2011                 iosize = (iosize + blocksize - 1) & ~((u64)blocksize - 1);
2012                 if (this_bio_flag & EXTENT_BIO_COMPRESSED) {
2013                         disk_io_size = em->block_len;
2014                         sector = em->block_start >> 9;
2015                 } else {
2016                         sector = (em->block_start + extent_offset) >> 9;
2017                         disk_io_size = iosize;
2018                 }
2019                 bdev = em->bdev;
2020                 block_start = em->block_start;
2021                 if (test_bit(EXTENT_FLAG_PREALLOC, &em->flags))
2022                         block_start = EXTENT_MAP_HOLE;
2023                 free_extent_map(em);
2024                 em = NULL;
2025
2026                 /* we've found a hole, just zero and go on */
2027                 if (block_start == EXTENT_MAP_HOLE) {
2028                         char *userpage;
2029                         userpage = kmap_atomic(page, KM_USER0);
2030                         memset(userpage + page_offset, 0, iosize);
2031                         flush_dcache_page(page);
2032                         kunmap_atomic(userpage, KM_USER0);
2033
2034                         set_extent_uptodate(tree, cur, cur + iosize - 1,
2035                                             GFP_NOFS);
2036                         unlock_extent(tree, cur, cur + iosize - 1, GFP_NOFS);
2037                         cur = cur + iosize;
2038                         page_offset += iosize;
2039                         continue;
2040                 }
2041                 /* the get_extent function already copied into the page */
2042                 if (test_range_bit(tree, cur, cur_end, EXTENT_UPTODATE, 1)) {
2043                         check_page_uptodate(tree, page);
2044                         unlock_extent(tree, cur, cur + iosize - 1, GFP_NOFS);
2045                         cur = cur + iosize;
2046                         page_offset += iosize;
2047                         continue;
2048                 }
2049                 /* we have an inline extent but it didn't get marked up
2050                  * to date.  Error out
2051                  */
2052                 if (block_start == EXTENT_MAP_INLINE) {
2053                         SetPageError(page);
2054                         unlock_extent(tree, cur, cur + iosize - 1, GFP_NOFS);
2055                         cur = cur + iosize;
2056                         page_offset += iosize;
2057                         continue;
2058                 }
2059
2060                 ret = 0;
2061                 if (tree->ops && tree->ops->readpage_io_hook) {
2062                         ret = tree->ops->readpage_io_hook(page, cur,
2063                                                           cur + iosize - 1);
2064                 }
2065                 if (!ret) {
2066                         unsigned long pnr = (last_byte >> PAGE_CACHE_SHIFT) + 1;
2067                         pnr -= page->index;
2068                         ret = submit_extent_page(READ, tree, page,
2069                                          sector, disk_io_size, page_offset,
2070                                          bdev, bio, pnr,
2071                                          end_bio_extent_readpage, mirror_num,
2072                                          *bio_flags,
2073                                          this_bio_flag);
2074                         nr++;
2075                         *bio_flags = this_bio_flag;
2076                 }
2077                 if (ret)
2078                         SetPageError(page);
2079                 cur = cur + iosize;
2080                 page_offset += iosize;
2081         }
2082         if (!nr) {
2083                 if (!PageError(page))
2084                         SetPageUptodate(page);
2085                 unlock_page(page);
2086         }
2087         return 0;
2088 }
2089
2090 int extent_read_full_page(struct extent_io_tree *tree, struct page *page,
2091                             get_extent_t *get_extent)
2092 {
2093         struct bio *bio = NULL;
2094         unsigned long bio_flags = 0;
2095         int ret;
2096
2097         ret = __extent_read_full_page(tree, page, get_extent, &bio, 0,
2098                                       &bio_flags);
2099         if (bio)
2100                 submit_one_bio(READ, bio, 0, bio_flags);
2101         return ret;
2102 }
2103
2104 /*
2105  * the writepage semantics are similar to regular writepage.  extent
2106  * records are inserted to lock ranges in the tree, and as dirty areas
2107  * are found, they are marked writeback.  Then the lock bits are removed
2108  * and the end_io handler clears the writeback ranges
2109  */
2110 static int __extent_writepage(struct page *page, struct writeback_control *wbc,
2111                               void *data)
2112 {
2113         struct inode *inode = page->mapping->host;
2114         struct extent_page_data *epd = data;
2115         struct extent_io_tree *tree = epd->tree;
2116         u64 start = (u64)page->index << PAGE_CACHE_SHIFT;
2117         u64 delalloc_start;
2118         u64 page_end = start + PAGE_CACHE_SIZE - 1;
2119         u64 end;
2120         u64 cur = start;
2121         u64 extent_offset;
2122         u64 last_byte = i_size_read(inode);
2123         u64 block_start;
2124         u64 iosize;
2125         u64 unlock_start;
2126         sector_t sector;
2127         struct extent_map *em;
2128         struct block_device *bdev;
2129         int ret;
2130         int nr = 0;
2131         size_t pg_offset = 0;
2132         size_t blocksize;
2133         loff_t i_size = i_size_read(inode);
2134         unsigned long end_index = i_size >> PAGE_CACHE_SHIFT;
2135         u64 nr_delalloc;
2136         u64 delalloc_end;
2137         int page_started;
2138         int compressed;
2139         unsigned long nr_written = 0;
2140
2141         WARN_ON(!PageLocked(page));
2142         pg_offset = i_size & (PAGE_CACHE_SIZE - 1);
2143         if (page->index > end_index ||
2144            (page->index == end_index && !pg_offset)) {
2145                 page->mapping->a_ops->invalidatepage(page, 0);
2146                 unlock_page(page);
2147                 return 0;
2148         }
2149
2150         if (page->index == end_index) {
2151                 char *userpage;
2152
2153                 userpage = kmap_atomic(page, KM_USER0);
2154                 memset(userpage + pg_offset, 0,
2155                        PAGE_CACHE_SIZE - pg_offset);
2156                 kunmap_atomic(userpage, KM_USER0);
2157                 flush_dcache_page(page);
2158         }
2159         pg_offset = 0;
2160
2161         set_page_extent_mapped(page);
2162
2163         delalloc_start = start;
2164         delalloc_end = 0;
2165         page_started = 0;
2166         if (!epd->extent_locked) {
2167                 while (delalloc_end < page_end) {
2168                         nr_delalloc = find_lock_delalloc_range(inode, tree,
2169                                                        page,
2170                                                        &delalloc_start,
2171                                                        &delalloc_end,
2172                                                        128 * 1024 * 1024);
2173                         if (nr_delalloc == 0) {
2174                                 delalloc_start = delalloc_end + 1;
2175                                 continue;
2176                         }
2177                         tree->ops->fill_delalloc(inode, page, delalloc_start,
2178                                                  delalloc_end, &page_started,
2179                                                  &nr_written);
2180                         delalloc_start = delalloc_end + 1;
2181                 }
2182
2183                 /* did the fill delalloc function already unlock and start
2184                  * the IO?
2185                  */
2186                 if (page_started) {
2187                         ret = 0;
2188                         goto update_nr_written;
2189                 }
2190         }
2191         lock_extent(tree, start, page_end, GFP_NOFS);
2192
2193         unlock_start = start;
2194
2195         if (tree->ops && tree->ops->writepage_start_hook) {
2196                 ret = tree->ops->writepage_start_hook(page, start,
2197                                                       page_end);
2198                 if (ret == -EAGAIN) {
2199                         unlock_extent(tree, start, page_end, GFP_NOFS);
2200                         redirty_page_for_writepage(wbc, page);
2201                         unlock_page(page);
2202                         ret = 0;
2203                         goto update_nr_written;
2204                 }
2205         }
2206
2207         nr_written++;
2208
2209         end = page_end;
2210         if (test_range_bit(tree, start, page_end, EXTENT_DELALLOC, 0))
2211                 printk(KERN_ERR "btrfs delalloc bits after lock_extent\n");
2212
2213         if (last_byte <= start) {
2214                 clear_extent_dirty(tree, start, page_end, GFP_NOFS);
2215                 unlock_extent(tree, start, page_end, GFP_NOFS);
2216                 if (tree->ops && tree->ops->writepage_end_io_hook)
2217                         tree->ops->writepage_end_io_hook(page, start,
2218                                                          page_end, NULL, 1);
2219                 unlock_start = page_end + 1;
2220                 goto done;
2221         }
2222
2223         set_extent_uptodate(tree, start, page_end, GFP_NOFS);
2224         blocksize = inode->i_sb->s_blocksize;
2225
2226         while (cur <= end) {
2227                 if (cur >= last_byte) {
2228                         clear_extent_dirty(tree, cur, page_end, GFP_NOFS);
2229                         unlock_extent(tree, unlock_start, page_end, GFP_NOFS);
2230                         if (tree->ops && tree->ops->writepage_end_io_hook)
2231                                 tree->ops->writepage_end_io_hook(page, cur,
2232                                                          page_end, NULL, 1);
2233                         unlock_start = page_end + 1;
2234                         break;
2235                 }
2236                 em = epd->get_extent(inode, page, pg_offset, cur,
2237                                      end - cur + 1, 1);
2238                 if (IS_ERR(em) || !em) {
2239                         SetPageError(page);
2240                         break;
2241                 }
2242
2243                 extent_offset = cur - em->start;
2244                 BUG_ON(extent_map_end(em) <= cur);
2245                 BUG_ON(end < cur);
2246                 iosize = min(extent_map_end(em) - cur, end - cur + 1);
2247                 iosize = (iosize + blocksize - 1) & ~((u64)blocksize - 1);
2248                 sector = (em->block_start + extent_offset) >> 9;
2249                 bdev = em->bdev;
2250                 block_start = em->block_start;
2251                 compressed = test_bit(EXTENT_FLAG_COMPRESSED, &em->flags);
2252                 free_extent_map(em);
2253                 em = NULL;
2254
2255                 /*
2256                  * compressed and inline extents are written through other
2257                  * paths in the FS
2258                  */
2259                 if (compressed || block_start == EXTENT_MAP_HOLE ||
2260                     block_start == EXTENT_MAP_INLINE) {
2261                         clear_extent_dirty(tree, cur,
2262                                            cur + iosize - 1, GFP_NOFS);
2263
2264                         unlock_extent(tree, unlock_start, cur + iosize - 1,
2265                                       GFP_NOFS);
2266
2267                         /*
2268                          * end_io notification does not happen here for
2269                          * compressed extents
2270                          */
2271                         if (!compressed && tree->ops &&
2272                             tree->ops->writepage_end_io_hook)
2273                                 tree->ops->writepage_end_io_hook(page, cur,
2274                                                          cur + iosize - 1,
2275                                                          NULL, 1);
2276                         else if (compressed) {
2277                                 /* we don't want to end_page_writeback on
2278                                  * a compressed extent.  this happens
2279                                  * elsewhere
2280                                  */
2281                                 nr++;
2282                         }
2283
2284                         cur += iosize;
2285                         pg_offset += iosize;
2286                         unlock_start = cur;
2287                         continue;
2288                 }
2289                 /* leave this out until we have a page_mkwrite call */
2290                 if (0 && !test_range_bit(tree, cur, cur + iosize - 1,
2291                                    EXTENT_DIRTY, 0)) {
2292                         cur = cur + iosize;
2293                         pg_offset += iosize;
2294                         continue;
2295                 }
2296
2297                 clear_extent_dirty(tree, cur, cur + iosize - 1, GFP_NOFS);
2298                 if (tree->ops && tree->ops->writepage_io_hook) {
2299                         ret = tree->ops->writepage_io_hook(page, cur,
2300                                                 cur + iosize - 1);
2301                 } else {
2302                         ret = 0;
2303                 }
2304                 if (ret) {
2305                         SetPageError(page);
2306                 } else {
2307                         unsigned long max_nr = end_index + 1;
2308
2309                         set_range_writeback(tree, cur, cur + iosize - 1);
2310                         if (!PageWriteback(page)) {
2311                                 printk(KERN_ERR "btrfs warning page %lu not "
2312                                        "writeback, cur %llu end %llu\n",
2313                                        page->index, (unsigned long long)cur,
2314                                        (unsigned long long)end);
2315                         }
2316
2317                         ret = submit_extent_page(WRITE, tree, page, sector,
2318                                                  iosize, pg_offset, bdev,
2319                                                  &epd->bio, max_nr,
2320                                                  end_bio_extent_writepage,
2321                                                  0, 0, 0);
2322                         if (ret)
2323                                 SetPageError(page);
2324                 }
2325                 cur = cur + iosize;
2326                 pg_offset += iosize;
2327                 nr++;
2328         }
2329 done:
2330         if (nr == 0) {
2331                 /* make sure the mapping tag for page dirty gets cleared */
2332                 set_page_writeback(page);
2333                 end_page_writeback(page);
2334         }
2335         if (unlock_start <= page_end)
2336                 unlock_extent(tree, unlock_start, page_end, GFP_NOFS);
2337         unlock_page(page);
2338
2339 update_nr_written:
2340         wbc->nr_to_write -= nr_written;
2341         if (wbc->range_cyclic || (wbc->nr_to_write > 0 &&
2342             wbc->range_start == 0 && wbc->range_end == LLONG_MAX))
2343                 page->mapping->writeback_index = page->index + nr_written;
2344         return 0;
2345 }
2346
2347 /**
2348  * write_cache_pages - walk the list of dirty pages of the given address space and write all of them.
2349  * @mapping: address space structure to write
2350  * @wbc: subtract the number of written pages from *@wbc->nr_to_write
2351  * @writepage: function called for each page
2352  * @data: data passed to writepage function
2353  *
2354  * If a page is already under I/O, write_cache_pages() skips it, even
2355  * if it's dirty.  This is desirable behaviour for memory-cleaning writeback,
2356  * but it is INCORRECT for data-integrity system calls such as fsync().  fsync()
2357  * and msync() need to guarantee that all the data which was dirty at the time
2358  * the call was made get new I/O started against them.  If wbc->sync_mode is
2359  * WB_SYNC_ALL then we were called for data integrity and we must wait for
2360  * existing IO to complete.
2361  */
2362 static int extent_write_cache_pages(struct extent_io_tree *tree,
2363                              struct address_space *mapping,
2364                              struct writeback_control *wbc,
2365                              writepage_t writepage, void *data,
2366                              void (*flush_fn)(void *))
2367 {
2368         struct backing_dev_info *bdi = mapping->backing_dev_info;
2369         int ret = 0;
2370         int done = 0;
2371         struct pagevec pvec;
2372         int nr_pages;
2373         pgoff_t index;
2374         pgoff_t end;            /* Inclusive */
2375         int scanned = 0;
2376         int range_whole = 0;
2377
2378         pagevec_init(&pvec, 0);
2379         if (wbc->range_cyclic) {
2380                 index = mapping->writeback_index; /* Start from prev offset */
2381                 end = -1;
2382         } else {
2383                 index = wbc->range_start >> PAGE_CACHE_SHIFT;
2384                 end = wbc->range_end >> PAGE_CACHE_SHIFT;
2385                 if (wbc->range_start == 0 && wbc->range_end == LLONG_MAX)
2386                         range_whole = 1;
2387                 scanned = 1;
2388         }
2389 retry:
2390         while (!done && (index <= end) &&
2391                (nr_pages = pagevec_lookup_tag(&pvec, mapping, &index,
2392                               PAGECACHE_TAG_DIRTY, min(end - index,
2393                                   (pgoff_t)PAGEVEC_SIZE-1) + 1))) {
2394                 unsigned i;
2395
2396                 scanned = 1;
2397                 for (i = 0; i < nr_pages; i++) {
2398                         struct page *page = pvec.pages[i];
2399
2400                         /*
2401                          * At this point we hold neither mapping->tree_lock nor
2402                          * lock on the page itself: the page may be truncated or
2403                          * invalidated (changing page->mapping to NULL), or even
2404                          * swizzled back from swapper_space to tmpfs file
2405                          * mapping
2406                          */
2407                         if (tree->ops && tree->ops->write_cache_pages_lock_hook)
2408                                 tree->ops->write_cache_pages_lock_hook(page);
2409                         else
2410                                 lock_page(page);
2411
2412                         if (unlikely(page->mapping != mapping)) {
2413                                 unlock_page(page);
2414                                 continue;
2415                         }
2416
2417                         if (!wbc->range_cyclic && page->index > end) {
2418                                 done = 1;
2419                                 unlock_page(page);
2420                                 continue;
2421                         }
2422
2423                         if (wbc->sync_mode != WB_SYNC_NONE) {
2424                                 if (PageWriteback(page))
2425                                         flush_fn(data);
2426                                 wait_on_page_writeback(page);
2427                         }
2428
2429                         if (PageWriteback(page) ||
2430                             !clear_page_dirty_for_io(page)) {
2431                                 unlock_page(page);
2432                                 continue;
2433                         }
2434
2435                         ret = (*writepage)(page, wbc, data);
2436
2437                         if (unlikely(ret == AOP_WRITEPAGE_ACTIVATE)) {
2438                                 unlock_page(page);
2439                                 ret = 0;
2440                         }
2441                         if (ret || wbc->nr_to_write <= 0)
2442                                 done = 1;
2443                         if (wbc->nonblocking && bdi_write_congested(bdi)) {
2444                                 wbc->encountered_congestion = 1;
2445                                 done = 1;
2446                         }
2447                 }
2448                 pagevec_release(&pvec);
2449                 cond_resched();
2450         }
2451         if (!scanned && !done) {
2452                 /*
2453                  * We hit the last page and there is more work to be done: wrap
2454                  * back to the start of the file
2455                  */
2456                 scanned = 1;
2457                 index = 0;
2458                 goto retry;
2459         }
2460         return ret;
2461 }
2462
2463 static noinline void flush_write_bio(void *data)
2464 {
2465         struct extent_page_data *epd = data;
2466         if (epd->bio) {
2467                 submit_one_bio(WRITE, epd->bio, 0, 0);
2468                 epd->bio = NULL;
2469         }
2470 }
2471
2472 int extent_write_full_page(struct extent_io_tree *tree, struct page *page,
2473                           get_extent_t *get_extent,
2474                           struct writeback_control *wbc)
2475 {
2476         int ret;
2477         struct address_space *mapping = page->mapping;
2478         struct extent_page_data epd = {
2479                 .bio = NULL,
2480                 .tree = tree,
2481                 .get_extent = get_extent,
2482                 .extent_locked = 0,
2483         };
2484         struct writeback_control wbc_writepages = {
2485                 .bdi            = wbc->bdi,
2486                 .sync_mode      = WB_SYNC_NONE,
2487                 .older_than_this = NULL,
2488                 .nr_to_write    = 64,
2489                 .range_start    = page_offset(page) + PAGE_CACHE_SIZE,
2490                 .range_end      = (loff_t)-1,
2491         };
2492
2493
2494         ret = __extent_writepage(page, wbc, &epd);
2495
2496         extent_write_cache_pages(tree, mapping, &wbc_writepages,
2497                                  __extent_writepage, &epd, flush_write_bio);
2498         if (epd.bio)
2499                 submit_one_bio(WRITE, epd.bio, 0, 0);
2500         return ret;
2501 }
2502
2503 int extent_write_locked_range(struct extent_io_tree *tree, struct inode *inode,
2504                               u64 start, u64 end, get_extent_t *get_extent,
2505                               int mode)
2506 {
2507         int ret = 0;
2508         struct address_space *mapping = inode->i_mapping;
2509         struct page *page;
2510         unsigned long nr_pages = (end - start + PAGE_CACHE_SIZE) >>
2511                 PAGE_CACHE_SHIFT;
2512
2513         struct extent_page_data epd = {
2514                 .bio = NULL,
2515                 .tree = tree,
2516                 .get_extent = get_extent,
2517                 .extent_locked = 1,
2518         };
2519         struct writeback_control wbc_writepages = {
2520                 .bdi            = inode->i_mapping->backing_dev_info,
2521                 .sync_mode      = mode,
2522                 .older_than_this = NULL,
2523                 .nr_to_write    = nr_pages * 2,
2524                 .range_start    = start,
2525                 .range_end      = end + 1,
2526         };
2527
2528         while (start <= end) {
2529                 page = find_get_page(mapping, start >> PAGE_CACHE_SHIFT);
2530                 if (clear_page_dirty_for_io(page))
2531                         ret = __extent_writepage(page, &wbc_writepages, &epd);
2532                 else {
2533                         if (tree->ops && tree->ops->writepage_end_io_hook)
2534                                 tree->ops->writepage_end_io_hook(page, start,
2535                                                  start + PAGE_CACHE_SIZE - 1,
2536                                                  NULL, 1);
2537                         unlock_page(page);
2538                 }
2539                 page_cache_release(page);
2540                 start += PAGE_CACHE_SIZE;
2541         }
2542
2543         if (epd.bio)
2544                 submit_one_bio(WRITE, epd.bio, 0, 0);
2545         return ret;
2546 }
2547
2548 int extent_writepages(struct extent_io_tree *tree,
2549                       struct address_space *mapping,
2550                       get_extent_t *get_extent,
2551                       struct writeback_control *wbc)
2552 {
2553         int ret = 0;
2554         struct extent_page_data epd = {
2555                 .bio = NULL,
2556                 .tree = tree,
2557                 .get_extent = get_extent,
2558                 .extent_locked = 0,
2559         };
2560
2561         ret = extent_write_cache_pages(tree, mapping, wbc,
2562                                        __extent_writepage, &epd,
2563                                        flush_write_bio);
2564         if (epd.bio)
2565                 submit_one_bio(WRITE, epd.bio, 0, 0);
2566         return ret;
2567 }
2568
2569 int extent_readpages(struct extent_io_tree *tree,
2570                      struct address_space *mapping,
2571                      struct list_head *pages, unsigned nr_pages,
2572                      get_extent_t get_extent)
2573 {
2574         struct bio *bio = NULL;
2575         unsigned page_idx;
2576         struct pagevec pvec;
2577         unsigned long bio_flags = 0;
2578
2579         pagevec_init(&pvec, 0);
2580         for (page_idx = 0; page_idx < nr_pages; page_idx++) {
2581                 struct page *page = list_entry(pages->prev, struct page, lru);
2582
2583                 prefetchw(&page->flags);
2584                 list_del(&page->lru);
2585                 /*
2586                  * what we want to do here is call add_to_page_cache_lru,
2587                  * but that isn't exported, so we reproduce it here
2588                  */
2589                 if (!add_to_page_cache(page, mapping,
2590                                         page->index, GFP_KERNEL)) {
2591
2592                         /* open coding of lru_cache_add, also not exported */
2593                         page_cache_get(page);
2594                         if (!pagevec_add(&pvec, page))
2595                                 __pagevec_lru_add_file(&pvec);
2596                         __extent_read_full_page(tree, page, get_extent,
2597                                                 &bio, 0, &bio_flags);
2598                 }
2599                 page_cache_release(page);
2600         }
2601         if (pagevec_count(&pvec))
2602                 __pagevec_lru_add_file(&pvec);
2603         BUG_ON(!list_empty(pages));
2604         if (bio)
2605                 submit_one_bio(READ, bio, 0, bio_flags);
2606         return 0;
2607 }
2608
2609 /*
2610  * basic invalidatepage code, this waits on any locked or writeback
2611  * ranges corresponding to the page, and then deletes any extent state
2612  * records from the tree
2613  */
2614 int extent_invalidatepage(struct extent_io_tree *tree,
2615                           struct page *page, unsigned long offset)
2616 {
2617         u64 start = ((u64)page->index << PAGE_CACHE_SHIFT);
2618         u64 end = start + PAGE_CACHE_SIZE - 1;
2619         size_t blocksize = page->mapping->host->i_sb->s_blocksize;
2620
2621         start += (offset + blocksize - 1) & ~(blocksize - 1);
2622         if (start > end)
2623                 return 0;
2624
2625         lock_extent(tree, start, end, GFP_NOFS);
2626         wait_on_extent_writeback(tree, start, end);
2627         clear_extent_bit(tree, start, end,
2628                          EXTENT_LOCKED | EXTENT_DIRTY | EXTENT_DELALLOC,
2629                          1, 1, GFP_NOFS);
2630         return 0;
2631 }
2632
2633 /*
2634  * simple commit_write call, set_range_dirty is used to mark both
2635  * the pages and the extent records as dirty
2636  */
2637 int extent_commit_write(struct extent_io_tree *tree,
2638                         struct inode *inode, struct page *page,
2639                         unsigned from, unsigned to)
2640 {
2641         loff_t pos = ((loff_t)page->index << PAGE_CACHE_SHIFT) + to;
2642
2643         set_page_extent_mapped(page);
2644         set_page_dirty(page);
2645
2646         if (pos > inode->i_size) {
2647                 i_size_write(inode, pos);
2648                 mark_inode_dirty(inode);
2649         }
2650         return 0;
2651 }
2652
2653 int extent_prepare_write(struct extent_io_tree *tree,
2654                          struct inode *inode, struct page *page,
2655                          unsigned from, unsigned to, get_extent_t *get_extent)
2656 {
2657         u64 page_start = (u64)page->index << PAGE_CACHE_SHIFT;
2658         u64 page_end = page_start + PAGE_CACHE_SIZE - 1;
2659         u64 block_start;
2660         u64 orig_block_start;
2661         u64 block_end;
2662         u64 cur_end;
2663         struct extent_map *em;
2664         unsigned blocksize = 1 << inode->i_blkbits;
2665         size_t page_offset = 0;
2666         size_t block_off_start;
2667         size_t block_off_end;
2668         int err = 0;
2669         int iocount = 0;
2670         int ret = 0;
2671         int isnew;
2672
2673         set_page_extent_mapped(page);
2674
2675         block_start = (page_start + from) & ~((u64)blocksize - 1);
2676         block_end = (page_start + to - 1) | (blocksize - 1);
2677         orig_block_start = block_start;
2678
2679         lock_extent(tree, page_start, page_end, GFP_NOFS);
2680         while (block_start <= block_end) {
2681                 em = get_extent(inode, page, page_offset, block_start,
2682                                 block_end - block_start + 1, 1);
2683                 if (IS_ERR(em) || !em)
2684                         goto err;
2685
2686                 cur_end = min(block_end, extent_map_end(em) - 1);
2687                 block_off_start = block_start & (PAGE_CACHE_SIZE - 1);
2688                 block_off_end = block_off_start + blocksize;
2689                 isnew = clear_extent_new(tree, block_start, cur_end, GFP_NOFS);
2690
2691                 if (!PageUptodate(page) && isnew &&
2692                     (block_off_end > to || block_off_start < from)) {
2693                         void *kaddr;
2694
2695                         kaddr = kmap_atomic(page, KM_USER0);
2696                         if (block_off_end > to)
2697                                 memset(kaddr + to, 0, block_off_end - to);
2698                         if (block_off_start < from)
2699                                 memset(kaddr + block_off_start, 0,
2700                                        from - block_off_start);
2701                         flush_dcache_page(page);
2702                         kunmap_atomic(kaddr, KM_USER0);
2703                 }
2704                 if ((em->block_start != EXTENT_MAP_HOLE &&
2705                      em->block_start != EXTENT_MAP_INLINE) &&
2706                     !isnew && !PageUptodate(page) &&
2707                     (block_off_end > to || block_off_start < from) &&
2708                     !test_range_bit(tree, block_start, cur_end,
2709                                     EXTENT_UPTODATE, 1)) {
2710                         u64 sector;
2711                         u64 extent_offset = block_start - em->start;
2712                         size_t iosize;
2713                         sector = (em->block_start + extent_offset) >> 9;
2714                         iosize = (cur_end - block_start + blocksize) &
2715                                 ~((u64)blocksize - 1);
2716                         /*
2717                          * we've already got the extent locked, but we
2718                          * need to split the state such that our end_bio
2719                          * handler can clear the lock.
2720                          */
2721                         set_extent_bit(tree, block_start,
2722                                        block_start + iosize - 1,
2723                                        EXTENT_LOCKED, 0, NULL, GFP_NOFS);
2724                         ret = submit_extent_page(READ, tree, page,
2725                                          sector, iosize, page_offset, em->bdev,
2726                                          NULL, 1,
2727                                          end_bio_extent_preparewrite, 0,
2728                                          0, 0);
2729                         iocount++;
2730                         block_start = block_start + iosize;
2731                 } else {
2732                         set_extent_uptodate(tree, block_start, cur_end,
2733                                             GFP_NOFS);
2734                         unlock_extent(tree, block_start, cur_end, GFP_NOFS);
2735                         block_start = cur_end + 1;
2736                 }
2737                 page_offset = block_start & (PAGE_CACHE_SIZE - 1);
2738                 free_extent_map(em);
2739         }
2740         if (iocount) {
2741                 wait_extent_bit(tree, orig_block_start,
2742                                 block_end, EXTENT_LOCKED);
2743         }
2744         check_page_uptodate(tree, page);
2745 err:
2746         /* FIXME, zero out newly allocated blocks on error */
2747         return err;
2748 }
2749
2750 /*
2751  * a helper for releasepage, this tests for areas of the page that
2752  * are locked or under IO and drops the related state bits if it is safe
2753  * to drop the page.
2754  */
2755 int try_release_extent_state(struct extent_map_tree *map,
2756                              struct extent_io_tree *tree, struct page *page,
2757                              gfp_t mask)
2758 {
2759         u64 start = (u64)page->index << PAGE_CACHE_SHIFT;
2760         u64 end = start + PAGE_CACHE_SIZE - 1;
2761         int ret = 1;
2762
2763         if (test_range_bit(tree, start, end,
2764                            EXTENT_IOBITS | EXTENT_ORDERED, 0))
2765                 ret = 0;
2766         else {
2767                 if ((mask & GFP_NOFS) == GFP_NOFS)
2768                         mask = GFP_NOFS;
2769                 clear_extent_bit(tree, start, end, EXTENT_UPTODATE,
2770                                  1, 1, mask);
2771         }
2772         return ret;
2773 }
2774
2775 /*
2776  * a helper for releasepage.  As long as there are no locked extents
2777  * in the range corresponding to the page, both state records and extent
2778  * map records are removed
2779  */
2780 int try_release_extent_mapping(struct extent_map_tree *map,
2781                                struct extent_io_tree *tree, struct page *page,
2782                                gfp_t mask)
2783 {
2784         struct extent_map *em;
2785         u64 start = (u64)page->index << PAGE_CACHE_SHIFT;
2786         u64 end = start + PAGE_CACHE_SIZE - 1;
2787
2788         if ((mask & __GFP_WAIT) &&
2789             page->mapping->host->i_size > 16 * 1024 * 1024) {
2790                 u64 len;
2791                 while (start <= end) {
2792                         len = end - start + 1;
2793                         spin_lock(&map->lock);
2794                         em = lookup_extent_mapping(map, start, len);
2795                         if (!em || IS_ERR(em)) {
2796                                 spin_unlock(&map->lock);
2797                                 break;
2798                         }
2799                         if (test_bit(EXTENT_FLAG_PINNED, &em->flags) ||
2800                             em->start != start) {
2801                                 spin_unlock(&map->lock);
2802                                 free_extent_map(em);
2803                                 break;
2804                         }
2805                         if (!test_range_bit(tree, em->start,
2806                                             extent_map_end(em) - 1,
2807                                             EXTENT_LOCKED | EXTENT_WRITEBACK |
2808                                             EXTENT_ORDERED,
2809                                             0)) {
2810                                 remove_extent_mapping(map, em);
2811                                 /* once for the rb tree */
2812                                 free_extent_map(em);
2813                         }
2814                         start = extent_map_end(em);
2815                         spin_unlock(&map->lock);
2816
2817                         /* once for us */
2818                         free_extent_map(em);
2819                 }
2820         }
2821         return try_release_extent_state(map, tree, page, mask);
2822 }
2823
2824 sector_t extent_bmap(struct address_space *mapping, sector_t iblock,
2825                 get_extent_t *get_extent)
2826 {
2827         struct inode *inode = mapping->host;
2828         u64 start = iblock << inode->i_blkbits;
2829         sector_t sector = 0;
2830         size_t blksize = (1 << inode->i_blkbits);
2831         struct extent_map *em;
2832
2833         lock_extent(&BTRFS_I(inode)->io_tree, start, start + blksize - 1,
2834                     GFP_NOFS);
2835         em = get_extent(inode, NULL, 0, start, blksize, 0);
2836         unlock_extent(&BTRFS_I(inode)->io_tree, start, start + blksize - 1,
2837                       GFP_NOFS);
2838         if (!em || IS_ERR(em))
2839                 return 0;
2840
2841         if (em->block_start > EXTENT_MAP_LAST_BYTE)
2842                 goto out;
2843
2844         sector = (em->block_start + start - em->start) >> inode->i_blkbits;
2845 out:
2846         free_extent_map(em);
2847         return sector;
2848 }
2849
2850 int extent_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
2851                 __u64 start, __u64 len, get_extent_t *get_extent)
2852 {
2853         int ret;
2854         u64 off = start;
2855         u64 max = start + len;
2856         u32 flags = 0;
2857         u64 disko = 0;
2858         struct extent_map *em = NULL;
2859         int end = 0;
2860         u64 em_start = 0, em_len = 0;
2861         unsigned long emflags;
2862         ret = 0;
2863
2864         if (len == 0)
2865                 return -EINVAL;
2866
2867         lock_extent(&BTRFS_I(inode)->io_tree, start, start + len,
2868                 GFP_NOFS);
2869         em = get_extent(inode, NULL, 0, off, max - off, 0);
2870         if (!em)
2871                 goto out;
2872         if (IS_ERR(em)) {
2873                 ret = PTR_ERR(em);
2874                 goto out;
2875         }
2876         while (!end) {
2877                 off = em->start + em->len;
2878                 if (off >= max)
2879                         end = 1;
2880
2881                 em_start = em->start;
2882                 em_len = em->len;
2883
2884                 disko = 0;
2885                 flags = 0;
2886
2887                 switch (em->block_start) {
2888                 case EXTENT_MAP_LAST_BYTE:
2889                         end = 1;
2890                         flags |= FIEMAP_EXTENT_LAST;
2891                         break;
2892                 case EXTENT_MAP_HOLE:
2893                         flags |= FIEMAP_EXTENT_UNWRITTEN;
2894                         break;
2895                 case EXTENT_MAP_INLINE:
2896                         flags |= (FIEMAP_EXTENT_DATA_INLINE |
2897                                   FIEMAP_EXTENT_NOT_ALIGNED);
2898                         break;
2899                 case EXTENT_MAP_DELALLOC:
2900                         flags |= (FIEMAP_EXTENT_DELALLOC |
2901                                   FIEMAP_EXTENT_UNKNOWN);
2902                         break;
2903                 default:
2904                         disko = em->block_start;
2905                         break;
2906                 }
2907                 if (test_bit(EXTENT_FLAG_COMPRESSED, &em->flags))
2908                         flags |= FIEMAP_EXTENT_ENCODED;
2909
2910                 emflags = em->flags;
2911                 free_extent_map(em);
2912                 em = NULL;
2913
2914                 if (!end) {
2915                         em = get_extent(inode, NULL, 0, off, max - off, 0);
2916                         if (!em)
2917                                 goto out;
2918                         if (IS_ERR(em)) {
2919                                 ret = PTR_ERR(em);
2920                                 goto out;
2921                         }
2922                         emflags = em->flags;
2923                 }
2924                 if (test_bit(EXTENT_FLAG_VACANCY, &emflags)) {
2925                         flags |= FIEMAP_EXTENT_LAST;
2926                         end = 1;
2927                 }
2928
2929                 ret = fiemap_fill_next_extent(fieinfo, em_start, disko,
2930                                         em_len, flags);
2931                 if (ret)
2932                         goto out_free;
2933         }
2934 out_free:
2935         free_extent_map(em);
2936 out:
2937         unlock_extent(&BTRFS_I(inode)->io_tree, start, start + len,
2938                         GFP_NOFS);
2939         return ret;
2940 }
2941
2942 static inline struct page *extent_buffer_page(struct extent_buffer *eb,
2943                                               unsigned long i)
2944 {
2945         struct page *p;
2946         struct address_space *mapping;
2947
2948         if (i == 0)
2949                 return eb->first_page;
2950         i += eb->start >> PAGE_CACHE_SHIFT;
2951         mapping = eb->first_page->mapping;
2952         if (!mapping)
2953                 return NULL;
2954
2955         /*
2956          * extent_buffer_page is only called after pinning the page
2957          * by increasing the reference count.  So we know the page must
2958          * be in the radix tree.
2959          */
2960         rcu_read_lock();
2961         p = radix_tree_lookup(&mapping->page_tree, i);
2962         rcu_read_unlock();
2963
2964         return p;
2965 }
2966
2967 static inline unsigned long num_extent_pages(u64 start, u64 len)
2968 {
2969         return ((start + len + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT) -
2970                 (start >> PAGE_CACHE_SHIFT);
2971 }
2972
2973 static struct extent_buffer *__alloc_extent_buffer(struct extent_io_tree *tree,
2974                                                    u64 start,
2975                                                    unsigned long len,
2976                                                    gfp_t mask)
2977 {
2978         struct extent_buffer *eb = NULL;
2979 #if LEAK_DEBUG
2980         unsigned long flags;
2981 #endif
2982
2983         eb = kmem_cache_zalloc(extent_buffer_cache, mask);
2984         eb->start = start;
2985         eb->len = len;
2986         spin_lock_init(&eb->lock);
2987         init_waitqueue_head(&eb->lock_wq);
2988
2989 #if LEAK_DEBUG
2990         spin_lock_irqsave(&leak_lock, flags);
2991         list_add(&eb->leak_list, &buffers);
2992         spin_unlock_irqrestore(&leak_lock, flags);
2993 #endif
2994         atomic_set(&eb->refs, 1);
2995
2996         return eb;
2997 }
2998
2999 static void __free_extent_buffer(struct extent_buffer *eb)
3000 {
3001 #if LEAK_DEBUG
3002         unsigned long flags;
3003         spin_lock_irqsave(&leak_lock, flags);
3004         list_del(&eb->leak_list);
3005         spin_unlock_irqrestore(&leak_lock, flags);
3006 #endif
3007         kmem_cache_free(extent_buffer_cache, eb);
3008 }
3009
3010 struct extent_buffer *alloc_extent_buffer(struct extent_io_tree *tree,
3011                                           u64 start, unsigned long len,
3012                                           struct page *page0,
3013                                           gfp_t mask)
3014 {
3015         unsigned long num_pages = num_extent_pages(start, len);
3016         unsigned long i;
3017         unsigned long index = start >> PAGE_CACHE_SHIFT;
3018         struct extent_buffer *eb;
3019         struct extent_buffer *exists = NULL;
3020         struct page *p;
3021         struct address_space *mapping = tree->mapping;
3022         int uptodate = 1;
3023
3024         spin_lock(&tree->buffer_lock);
3025         eb = buffer_search(tree, start);
3026         if (eb) {
3027                 atomic_inc(&eb->refs);
3028                 spin_unlock(&tree->buffer_lock);
3029                 mark_page_accessed(eb->first_page);
3030                 return eb;
3031         }
3032         spin_unlock(&tree->buffer_lock);
3033
3034         eb = __alloc_extent_buffer(tree, start, len, mask);
3035         if (!eb)
3036                 return NULL;
3037
3038         if (page0) {
3039                 eb->first_page = page0;
3040                 i = 1;
3041                 index++;
3042                 page_cache_get(page0);
3043                 mark_page_accessed(page0);
3044                 set_page_extent_mapped(page0);
3045                 set_page_extent_head(page0, len);
3046                 uptodate = PageUptodate(page0);
3047         } else {
3048                 i = 0;
3049         }
3050         for (; i < num_pages; i++, index++) {
3051                 p = find_or_create_page(mapping, index, mask | __GFP_HIGHMEM);
3052                 if (!p) {
3053                         WARN_ON(1);
3054                         goto free_eb;
3055                 }
3056                 set_page_extent_mapped(p);
3057                 mark_page_accessed(p);
3058                 if (i == 0) {
3059                         eb->first_page = p;
3060                         set_page_extent_head(p, len);
3061                 } else {
3062                         set_page_private(p, EXTENT_PAGE_PRIVATE);
3063                 }
3064                 if (!PageUptodate(p))
3065                         uptodate = 0;
3066                 unlock_page(p);
3067         }
3068         if (uptodate)
3069                 set_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags);
3070
3071         spin_lock(&tree->buffer_lock);
3072         exists = buffer_tree_insert(tree, start, &eb->rb_node);
3073         if (exists) {
3074                 /* add one reference for the caller */
3075                 atomic_inc(&exists->refs);
3076                 spin_unlock(&tree->buffer_lock);
3077                 goto free_eb;
3078         }
3079         spin_unlock(&tree->buffer_lock);
3080
3081         /* add one reference for the tree */
3082         atomic_inc(&eb->refs);
3083         return eb;
3084
3085 free_eb:
3086         if (!atomic_dec_and_test(&eb->refs))
3087                 return exists;
3088         for (index = 1; index < i; index++)
3089                 page_cache_release(extent_buffer_page(eb, index));
3090         page_cache_release(extent_buffer_page(eb, 0));
3091         __free_extent_buffer(eb);
3092         return exists;
3093 }
3094
3095 struct extent_buffer *find_extent_buffer(struct extent_io_tree *tree,
3096                                          u64 start, unsigned long len,
3097                                           gfp_t mask)
3098 {
3099         struct extent_buffer *eb;
3100
3101         spin_lock(&tree->buffer_lock);
3102         eb = buffer_search(tree, start);
3103         if (eb)
3104                 atomic_inc(&eb->refs);
3105         spin_unlock(&tree->buffer_lock);
3106
3107         if (eb)
3108                 mark_page_accessed(eb->first_page);
3109
3110         return eb;
3111 }
3112
3113 void free_extent_buffer(struct extent_buffer *eb)
3114 {
3115         if (!eb)
3116                 return;
3117
3118         if (!atomic_dec_and_test(&eb->refs))
3119                 return;
3120
3121         WARN_ON(1);
3122 }
3123
3124 int clear_extent_buffer_dirty(struct extent_io_tree *tree,
3125                               struct extent_buffer *eb)
3126 {
3127         int set;
3128         unsigned long i;
3129         unsigned long num_pages;
3130         struct page *page;
3131
3132         u64 start = eb->start;
3133         u64 end = start + eb->len - 1;
3134
3135         set = clear_extent_dirty(tree, start, end, GFP_NOFS);
3136         num_pages = num_extent_pages(eb->start, eb->len);
3137
3138         for (i = 0; i < num_pages; i++) {
3139                 page = extent_buffer_page(eb, i);
3140                 if (!set && !PageDirty(page))
3141                         continue;
3142
3143                 lock_page(page);
3144                 if (i == 0)
3145                         set_page_extent_head(page, eb->len);
3146                 else
3147                         set_page_private(page, EXTENT_PAGE_PRIVATE);
3148
3149                 /*
3150                  * if we're on the last page or the first page and the
3151                  * block isn't aligned on a page boundary, do extra checks
3152                  * to make sure we don't clean page that is partially dirty
3153                  */
3154                 if ((i == 0 && (eb->start & (PAGE_CACHE_SIZE - 1))) ||
3155                     ((i == num_pages - 1) &&
3156                      ((eb->start + eb->len) & (PAGE_CACHE_SIZE - 1)))) {
3157                         start = (u64)page->index << PAGE_CACHE_SHIFT;
3158                         end  = start + PAGE_CACHE_SIZE - 1;
3159                         if (test_range_bit(tree, start, end,
3160                                            EXTENT_DIRTY, 0)) {
3161                                 unlock_page(page);
3162                                 continue;
3163                         }
3164                 }
3165                 clear_page_dirty_for_io(page);
3166                 spin_lock_irq(&page->mapping->tree_lock);
3167                 if (!PageDirty(page)) {
3168                         radix_tree_tag_clear(&page->mapping->page_tree,
3169                                                 page_index(page),
3170                                                 PAGECACHE_TAG_DIRTY);
3171                 }
3172                 spin_unlock_irq(&page->mapping->tree_lock);
3173                 unlock_page(page);
3174         }
3175         return 0;
3176 }
3177
3178 int wait_on_extent_buffer_writeback(struct extent_io_tree *tree,
3179                                     struct extent_buffer *eb)
3180 {
3181         return wait_on_extent_writeback(tree, eb->start,
3182                                         eb->start + eb->len - 1);
3183 }
3184
3185 int set_extent_buffer_dirty(struct extent_io_tree *tree,
3186                              struct extent_buffer *eb)
3187 {
3188         unsigned long i;
3189         unsigned long num_pages;
3190
3191         num_pages = num_extent_pages(eb->start, eb->len);
3192         for (i = 0; i < num_pages; i++) {
3193                 struct page *page = extent_buffer_page(eb, i);
3194                 /* writepage may need to do something special for the
3195                  * first page, we have to make sure page->private is
3196                  * properly set.  releasepage may drop page->private
3197                  * on us if the page isn't already dirty.
3198                  */
3199                 lock_page(page);
3200                 if (i == 0) {
3201                         set_page_extent_head(page, eb->len);
3202                 } else if (PagePrivate(page) &&
3203                            page->private != EXTENT_PAGE_PRIVATE) {
3204                         set_page_extent_mapped(page);
3205                 }
3206                 __set_page_dirty_nobuffers(extent_buffer_page(eb, i));
3207                 set_extent_dirty(tree, page_offset(page),
3208                                  page_offset(page) + PAGE_CACHE_SIZE - 1,
3209                                  GFP_NOFS);
3210                 unlock_page(page);
3211         }
3212         return 0;
3213 }
3214
3215 int clear_extent_buffer_uptodate(struct extent_io_tree *tree,
3216                                 struct extent_buffer *eb)
3217 {
3218         unsigned long i;
3219         struct page *page;
3220         unsigned long num_pages;
3221
3222         num_pages = num_extent_pages(eb->start, eb->len);
3223         clear_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags);
3224
3225         clear_extent_uptodate(tree, eb->start, eb->start + eb->len - 1,
3226                               GFP_NOFS);
3227         for (i = 0; i < num_pages; i++) {
3228                 page = extent_buffer_page(eb, i);
3229                 if (page)
3230                         ClearPageUptodate(page);
3231         }
3232         return 0;
3233 }
3234
3235 int set_extent_buffer_uptodate(struct extent_io_tree *tree,
3236                                 struct extent_buffer *eb)
3237 {
3238         unsigned long i;
3239         struct page *page;
3240         unsigned long num_pages;
3241
3242         num_pages = num_extent_pages(eb->start, eb->len);
3243
3244         set_extent_uptodate(tree, eb->start, eb->start + eb->len - 1,
3245                             GFP_NOFS);
3246         for (i = 0; i < num_pages; i++) {
3247                 page = extent_buffer_page(eb, i);
3248                 if ((i == 0 && (eb->start & (PAGE_CACHE_SIZE - 1))) ||
3249                     ((i == num_pages - 1) &&
3250                      ((eb->start + eb->len) & (PAGE_CACHE_SIZE - 1)))) {
3251                         check_page_uptodate(tree, page);
3252                         continue;
3253                 }
3254                 SetPageUptodate(page);
3255         }
3256         return 0;
3257 }
3258
3259 int extent_range_uptodate(struct extent_io_tree *tree,
3260                           u64 start, u64 end)
3261 {
3262         struct page *page;
3263         int ret;
3264         int pg_uptodate = 1;
3265         int uptodate;
3266         unsigned long index;
3267
3268         ret = test_range_bit(tree, start, end, EXTENT_UPTODATE, 1);
3269         if (ret)
3270                 return 1;
3271         while (start <= end) {
3272                 index = start >> PAGE_CACHE_SHIFT;
3273                 page = find_get_page(tree->mapping, index);
3274                 uptodate = PageUptodate(page);
3275                 page_cache_release(page);
3276                 if (!uptodate) {
3277                         pg_uptodate = 0;
3278                         break;
3279                 }
3280                 start += PAGE_CACHE_SIZE;
3281         }
3282         return pg_uptodate;
3283 }
3284
3285 int extent_buffer_uptodate(struct extent_io_tree *tree,
3286                            struct extent_buffer *eb)
3287 {
3288         int ret = 0;
3289         unsigned long num_pages;
3290         unsigned long i;
3291         struct page *page;
3292         int pg_uptodate = 1;
3293
3294         if (test_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags))
3295                 return 1;
3296
3297         ret = test_range_bit(tree, eb->start, eb->start + eb->len - 1,
3298                            EXTENT_UPTODATE, 1);
3299         if (ret)
3300                 return ret;
3301
3302         num_pages = num_extent_pages(eb->start, eb->len);
3303         for (i = 0; i < num_pages; i++) {
3304                 page = extent_buffer_page(eb, i);
3305                 if (!PageUptodate(page)) {
3306                         pg_uptodate = 0;
3307                         break;
3308                 }
3309         }
3310         return pg_uptodate;
3311 }
3312
3313 int read_extent_buffer_pages(struct extent_io_tree *tree,
3314                              struct extent_buffer *eb,
3315                              u64 start, int wait,
3316                              get_extent_t *get_extent, int mirror_num)
3317 {
3318         unsigned long i;
3319         unsigned long start_i;
3320         struct page *page;
3321         int err;
3322         int ret = 0;
3323         int locked_pages = 0;
3324         int all_uptodate = 1;
3325         int inc_all_pages = 0;
3326         unsigned long num_pages;
3327         struct bio *bio = NULL;
3328         unsigned long bio_flags = 0;
3329
3330         if (test_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags))
3331                 return 0;
3332
3333         if (test_range_bit(tree, eb->start, eb->start + eb->len - 1,
3334                            EXTENT_UPTODATE, 1)) {
3335                 return 0;
3336         }
3337
3338         if (start) {
3339                 WARN_ON(start < eb->start);
3340                 start_i = (start >> PAGE_CACHE_SHIFT) -
3341                         (eb->start >> PAGE_CACHE_SHIFT);
3342         } else {
3343                 start_i = 0;
3344         }
3345
3346         num_pages = num_extent_pages(eb->start, eb->len);
3347         for (i = start_i; i < num_pages; i++) {
3348                 page = extent_buffer_page(eb, i);
3349                 if (!wait) {
3350                         if (!trylock_page(page))
3351                                 goto unlock_exit;
3352                 } else {
3353                         lock_page(page);
3354                 }
3355                 locked_pages++;
3356                 if (!PageUptodate(page))
3357                         all_uptodate = 0;
3358         }
3359         if (all_uptodate) {
3360                 if (start_i == 0)
3361                         set_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags);
3362                 goto unlock_exit;
3363         }
3364
3365         for (i = start_i; i < num_pages; i++) {
3366                 page = extent_buffer_page(eb, i);
3367                 if (inc_all_pages)
3368                         page_cache_get(page);
3369                 if (!PageUptodate(page)) {
3370                         if (start_i == 0)
3371                                 inc_all_pages = 1;
3372                         ClearPageError(page);
3373                         err = __extent_read_full_page(tree, page,
3374                                                       get_extent, &bio,
3375                                                       mirror_num, &bio_flags);
3376                         if (err)
3377                                 ret = err;
3378                 } else {
3379                         unlock_page(page);
3380                 }
3381         }
3382
3383         if (bio)
3384                 submit_one_bio(READ, bio, mirror_num, bio_flags);
3385
3386         if (ret || !wait)
3387                 return ret;
3388
3389         for (i = start_i; i < num_pages; i++) {
3390                 page = extent_buffer_page(eb, i);
3391                 wait_on_page_locked(page);
3392                 if (!PageUptodate(page))
3393                         ret = -EIO;
3394         }
3395
3396         if (!ret)
3397                 set_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags);
3398         return ret;
3399
3400 unlock_exit:
3401         i = start_i;
3402         while (locked_pages > 0) {
3403                 page = extent_buffer_page(eb, i);
3404                 i++;
3405                 unlock_page(page);
3406                 locked_pages--;
3407         }
3408         return ret;
3409 }
3410
3411 void read_extent_buffer(struct extent_buffer *eb, void *dstv,
3412                         unsigned long start,
3413                         unsigned long len)
3414 {
3415         size_t cur;
3416         size_t offset;
3417         struct page *page;
3418         char *kaddr;
3419         char *dst = (char *)dstv;
3420         size_t start_offset = eb->start & ((u64)PAGE_CACHE_SIZE - 1);
3421         unsigned long i = (start_offset + start) >> PAGE_CACHE_SHIFT;
3422
3423         WARN_ON(start > eb->len);
3424         WARN_ON(start + len > eb->start + eb->len);
3425
3426         offset = (start_offset + start) & ((unsigned long)PAGE_CACHE_SIZE - 1);
3427
3428         while (len > 0) {
3429                 page = extent_buffer_page(eb, i);
3430
3431                 cur = min(len, (PAGE_CACHE_SIZE - offset));
3432                 kaddr = kmap_atomic(page, KM_USER1);
3433                 memcpy(dst, kaddr + offset, cur);
3434                 kunmap_atomic(kaddr, KM_USER1);
3435
3436                 dst += cur;
3437                 len -= cur;
3438                 offset = 0;
3439                 i++;
3440         }
3441 }
3442
3443 int map_private_extent_buffer(struct extent_buffer *eb, unsigned long start,
3444                                unsigned long min_len, char **token, char **map,
3445                                unsigned long *map_start,
3446                                unsigned long *map_len, int km)
3447 {
3448         size_t offset = start & (PAGE_CACHE_SIZE - 1);
3449         char *kaddr;
3450         struct page *p;
3451         size_t start_offset = eb->start & ((u64)PAGE_CACHE_SIZE - 1);
3452         unsigned long i = (start_offset + start) >> PAGE_CACHE_SHIFT;
3453         unsigned long end_i = (start_offset + start + min_len - 1) >>
3454                 PAGE_CACHE_SHIFT;
3455
3456         if (i != end_i)
3457                 return -EINVAL;
3458
3459         if (i == 0) {
3460                 offset = start_offset;
3461                 *map_start = 0;
3462         } else {
3463                 offset = 0;
3464                 *map_start = ((u64)i << PAGE_CACHE_SHIFT) - start_offset;
3465         }
3466
3467         if (start + min_len > eb->len) {
3468                 printk(KERN_ERR "btrfs bad mapping eb start %llu len %lu, "
3469                        "wanted %lu %lu\n", (unsigned long long)eb->start,
3470                        eb->len, start, min_len);
3471                 WARN_ON(1);
3472         }
3473
3474         p = extent_buffer_page(eb, i);
3475         kaddr = kmap_atomic(p, km);
3476         *token = kaddr;
3477         *map = kaddr + offset;
3478         *map_len = PAGE_CACHE_SIZE - offset;
3479         return 0;
3480 }
3481
3482 int map_extent_buffer(struct extent_buffer *eb, unsigned long start,
3483                       unsigned long min_len,
3484                       char **token, char **map,
3485                       unsigned long *map_start,
3486                       unsigned long *map_len, int km)
3487 {
3488         int err;
3489         int save = 0;
3490         if (eb->map_token) {
3491                 unmap_extent_buffer(eb, eb->map_token, km);
3492                 eb->map_token = NULL;
3493                 save = 1;
3494         }
3495         err = map_private_extent_buffer(eb, start, min_len, token, map,
3496                                        map_start, map_len, km);
3497         if (!err && save) {
3498                 eb->map_token = *token;
3499                 eb->kaddr = *map;
3500                 eb->map_start = *map_start;
3501                 eb->map_len = *map_len;
3502         }
3503         return err;
3504 }
3505
3506 void unmap_extent_buffer(struct extent_buffer *eb, char *token, int km)
3507 {
3508         kunmap_atomic(token, km);
3509 }
3510
3511 int memcmp_extent_buffer(struct extent_buffer *eb, const void *ptrv,
3512                           unsigned long start,
3513                           unsigned long len)
3514 {
3515         size_t cur;
3516         size_t offset;
3517         struct page *page;
3518         char *kaddr;
3519         char *ptr = (char *)ptrv;
3520         size_t start_offset = eb->start & ((u64)PAGE_CACHE_SIZE - 1);
3521         unsigned long i = (start_offset + start) >> PAGE_CACHE_SHIFT;
3522         int ret = 0;
3523
3524         WARN_ON(start > eb->len);
3525         WARN_ON(start + len > eb->start + eb->len);
3526
3527         offset = (start_offset + start) & ((unsigned long)PAGE_CACHE_SIZE - 1);
3528
3529         while (len > 0) {
3530                 page = extent_buffer_page(eb, i);
3531
3532                 cur = min(len, (PAGE_CACHE_SIZE - offset));
3533
3534                 kaddr = kmap_atomic(page, KM_USER0);
3535                 ret = memcmp(ptr, kaddr + offset, cur);
3536                 kunmap_atomic(kaddr, KM_USER0);
3537                 if (ret)
3538                         break;
3539
3540                 ptr += cur;
3541                 len -= cur;
3542                 offset = 0;
3543                 i++;
3544         }
3545         return ret;
3546 }
3547
3548 void write_extent_buffer(struct extent_buffer *eb, const void *srcv,
3549                          unsigned long start, unsigned long len)
3550 {
3551         size_t cur;
3552         size_t offset;
3553         struct page *page;
3554         char *kaddr;
3555         char *src = (char *)srcv;
3556         size_t start_offset = eb->start & ((u64)PAGE_CACHE_SIZE - 1);
3557         unsigned long i = (start_offset + start) >> PAGE_CACHE_SHIFT;
3558
3559         WARN_ON(start > eb->len);
3560         WARN_ON(start + len > eb->start + eb->len);
3561
3562         offset = (start_offset + start) & ((unsigned long)PAGE_CACHE_SIZE - 1);
3563
3564         while (len > 0) {
3565                 page = extent_buffer_page(eb, i);
3566                 WARN_ON(!PageUptodate(page));
3567
3568                 cur = min(len, PAGE_CACHE_SIZE - offset);
3569                 kaddr = kmap_atomic(page, KM_USER1);
3570                 memcpy(kaddr + offset, src, cur);
3571                 kunmap_atomic(kaddr, KM_USER1);
3572
3573                 src += cur;
3574                 len -= cur;
3575                 offset = 0;
3576                 i++;
3577         }
3578 }
3579
3580 void memset_extent_buffer(struct extent_buffer *eb, char c,
3581                           unsigned long start, unsigned long len)
3582 {
3583         size_t cur;
3584         size_t offset;
3585         struct page *page;
3586         char *kaddr;
3587         size_t start_offset = eb->start & ((u64)PAGE_CACHE_SIZE - 1);
3588         unsigned long i = (start_offset + start) >> PAGE_CACHE_SHIFT;
3589
3590         WARN_ON(start > eb->len);
3591         WARN_ON(start + len > eb->start + eb->len);
3592
3593         offset = (start_offset + start) & ((unsigned long)PAGE_CACHE_SIZE - 1);
3594
3595         while (len > 0) {
3596                 page = extent_buffer_page(eb, i);
3597                 WARN_ON(!PageUptodate(page));
3598
3599                 cur = min(len, PAGE_CACHE_SIZE - offset);
3600                 kaddr = kmap_atomic(page, KM_USER0);
3601                 memset(kaddr + offset, c, cur);
3602                 kunmap_atomic(kaddr, KM_USER0);
3603
3604                 len -= cur;
3605                 offset = 0;
3606                 i++;
3607         }
3608 }
3609
3610 void copy_extent_buffer(struct extent_buffer *dst, struct extent_buffer *src,
3611                         unsigned long dst_offset, unsigned long src_offset,
3612                         unsigned long len)
3613 {
3614         u64 dst_len = dst->len;
3615         size_t cur;
3616         size_t offset;
3617         struct page *page;
3618         char *kaddr;
3619         size_t start_offset = dst->start & ((u64)PAGE_CACHE_SIZE - 1);
3620         unsigned long i = (start_offset + dst_offset) >> PAGE_CACHE_SHIFT;
3621
3622         WARN_ON(src->len != dst_len);
3623
3624         offset = (start_offset + dst_offset) &
3625                 ((unsigned long)PAGE_CACHE_SIZE - 1);
3626
3627         while (len > 0) {
3628                 page = extent_buffer_page(dst, i);
3629                 WARN_ON(!PageUptodate(page));
3630
3631                 cur = min(len, (unsigned long)(PAGE_CACHE_SIZE - offset));
3632
3633                 kaddr = kmap_atomic(page, KM_USER0);
3634                 read_extent_buffer(src, kaddr + offset, src_offset, cur);
3635                 kunmap_atomic(kaddr, KM_USER0);
3636
3637                 src_offset += cur;
3638                 len -= cur;
3639                 offset = 0;
3640                 i++;
3641         }
3642 }
3643
3644 static void move_pages(struct page *dst_page, struct page *src_page,
3645                        unsigned long dst_off, unsigned long src_off,
3646                        unsigned long len)
3647 {
3648         char *dst_kaddr = kmap_atomic(dst_page, KM_USER0);
3649         if (dst_page == src_page) {
3650                 memmove(dst_kaddr + dst_off, dst_kaddr + src_off, len);
3651         } else {
3652                 char *src_kaddr = kmap_atomic(src_page, KM_USER1);
3653                 char *p = dst_kaddr + dst_off + len;
3654                 char *s = src_kaddr + src_off + len;
3655
3656                 while (len--)
3657                         *--p = *--s;
3658
3659                 kunmap_atomic(src_kaddr, KM_USER1);
3660         }
3661         kunmap_atomic(dst_kaddr, KM_USER0);
3662 }
3663
3664 static void copy_pages(struct page *dst_page, struct page *src_page,
3665                        unsigned long dst_off, unsigned long src_off,
3666                        unsigned long len)
3667 {
3668         char *dst_kaddr = kmap_atomic(dst_page, KM_USER0);
3669         char *src_kaddr;
3670
3671         if (dst_page != src_page)
3672                 src_kaddr = kmap_atomic(src_page, KM_USER1);
3673         else
3674                 src_kaddr = dst_kaddr;
3675
3676         memcpy(dst_kaddr + dst_off, src_kaddr + src_off, len);
3677         kunmap_atomic(dst_kaddr, KM_USER0);
3678         if (dst_page != src_page)
3679                 kunmap_atomic(src_kaddr, KM_USER1);
3680 }
3681
3682 void memcpy_extent_buffer(struct extent_buffer *dst, unsigned long dst_offset,
3683                            unsigned long src_offset, unsigned long len)
3684 {
3685         size_t cur;
3686         size_t dst_off_in_page;
3687         size_t src_off_in_page;
3688         size_t start_offset = dst->start & ((u64)PAGE_CACHE_SIZE - 1);
3689         unsigned long dst_i;
3690         unsigned long src_i;
3691
3692         if (src_offset + len > dst->len) {
3693                 printk(KERN_ERR "btrfs memmove bogus src_offset %lu move "
3694                        "len %lu dst len %lu\n", src_offset, len, dst->len);
3695                 BUG_ON(1);
3696         }
3697         if (dst_offset + len > dst->len) {
3698                 printk(KERN_ERR "btrfs memmove bogus dst_offset %lu move "
3699                        "len %lu dst len %lu\n", dst_offset, len, dst->len);
3700                 BUG_ON(1);
3701         }
3702
3703         while (len > 0) {
3704                 dst_off_in_page = (start_offset + dst_offset) &
3705                         ((unsigned long)PAGE_CACHE_SIZE - 1);
3706                 src_off_in_page = (start_offset + src_offset) &
3707                         ((unsigned long)PAGE_CACHE_SIZE - 1);
3708
3709                 dst_i = (start_offset + dst_offset) >> PAGE_CACHE_SHIFT;
3710                 src_i = (start_offset + src_offset) >> PAGE_CACHE_SHIFT;
3711
3712                 cur = min(len, (unsigned long)(PAGE_CACHE_SIZE -
3713                                                src_off_in_page));
3714                 cur = min_t(unsigned long, cur,
3715                         (unsigned long)(PAGE_CACHE_SIZE - dst_off_in_page));
3716
3717                 copy_pages(extent_buffer_page(dst, dst_i),
3718                            extent_buffer_page(dst, src_i),
3719                            dst_off_in_page, src_off_in_page, cur);
3720
3721                 src_offset += cur;
3722                 dst_offset += cur;
3723                 len -= cur;
3724         }
3725 }
3726
3727 void memmove_extent_buffer(struct extent_buffer *dst, unsigned long dst_offset,
3728                            unsigned long src_offset, unsigned long len)
3729 {
3730         size_t cur;
3731         size_t dst_off_in_page;
3732         size_t src_off_in_page;
3733         unsigned long dst_end = dst_offset + len - 1;
3734         unsigned long src_end = src_offset + len - 1;
3735         size_t start_offset = dst->start & ((u64)PAGE_CACHE_SIZE - 1);
3736         unsigned long dst_i;
3737         unsigned long src_i;
3738
3739         if (src_offset + len > dst->len) {
3740                 printk(KERN_ERR "btrfs memmove bogus src_offset %lu move "
3741                        "len %lu len %lu\n", src_offset, len, dst->len);
3742                 BUG_ON(1);
3743         }
3744         if (dst_offset + len > dst->len) {
3745                 printk(KERN_ERR "btrfs memmove bogus dst_offset %lu move "
3746                        "len %lu len %lu\n", dst_offset, len, dst->len);
3747                 BUG_ON(1);
3748         }
3749         if (dst_offset < src_offset) {
3750                 memcpy_extent_buffer(dst, dst_offset, src_offset, len);
3751                 return;
3752         }
3753         while (len > 0) {
3754                 dst_i = (start_offset + dst_end) >> PAGE_CACHE_SHIFT;
3755                 src_i = (start_offset + src_end) >> PAGE_CACHE_SHIFT;
3756
3757                 dst_off_in_page = (start_offset + dst_end) &
3758                         ((unsigned long)PAGE_CACHE_SIZE - 1);
3759                 src_off_in_page = (start_offset + src_end) &
3760                         ((unsigned long)PAGE_CACHE_SIZE - 1);
3761
3762                 cur = min_t(unsigned long, len, src_off_in_page + 1);
3763                 cur = min(cur, dst_off_in_page + 1);
3764                 move_pages(extent_buffer_page(dst, dst_i),
3765                            extent_buffer_page(dst, src_i),
3766                            dst_off_in_page - cur + 1,
3767                            src_off_in_page - cur + 1, cur);
3768
3769                 dst_end -= cur;
3770                 src_end -= cur;
3771                 len -= cur;
3772         }
3773 }
3774
3775 int try_release_extent_buffer(struct extent_io_tree *tree, struct page *page)
3776 {
3777         u64 start = page_offset(page);
3778         struct extent_buffer *eb;
3779         int ret = 1;
3780         unsigned long i;
3781         unsigned long num_pages;
3782
3783         spin_lock(&tree->buffer_lock);
3784         eb = buffer_search(tree, start);
3785         if (!eb)
3786                 goto out;
3787
3788         if (atomic_read(&eb->refs) > 1) {
3789                 ret = 0;
3790                 goto out;
3791         }
3792         /* at this point we can safely release the extent buffer */
3793         num_pages = num_extent_pages(eb->start, eb->len);
3794         for (i = 0; i < num_pages; i++)
3795                 page_cache_release(extent_buffer_page(eb, i));
3796         rb_erase(&eb->rb_node, &tree->buffer);
3797         __free_extent_buffer(eb);
3798 out:
3799         spin_unlock(&tree->buffer_lock);
3800         return ret;
3801 }