fs/btrfs/file.c

   1 /*
   2  * Copyright (C) 2007 Oracle.  All rights reserved.
   3  *
   4  * This program is free software; you can redistribute it and/or
   5  * modify it under the terms of the GNU General Public
   6  * License v2 as published by the Free Software Foundation.
   7  *
   8  * This program is distributed in the hope that it will be useful,
   9  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  10  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  11  * General Public License for more details.
  12  *
  13  * You should have received a copy of the GNU General Public
  14  * License along with this program; if not, write to the
  15  * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  16  * Boston, MA 021110-1307, USA.
  17  */
  18
  19 #include <linux/fs.h>
  20 #include <linux/pagemap.h>
  21 #include <linux/highmem.h>
  22 #include <linux/time.h>
  23 #include <linux/init.h>
  24 #include <linux/string.h>
  25 #include <linux/backing-dev.h>
  26 #include <linux/mpage.h>
  27 #include <linux/falloc.h>
  28 #include <linux/swap.h>
  29 #include <linux/writeback.h>
  30 #include <linux/statfs.h>
  31 #include <linux/compat.h>
  32 #include <linux/slab.h>
  33 #include <linux/btrfs.h>
  34 #include <linux/uio.h>
  35 #include "ctree.h"
  36 #include "disk-io.h"
  37 #include "transaction.h"
  38 #include "btrfs_inode.h"
  39 #include "print-tree.h"
  40 #include "tree-log.h"
  41 #include "locking.h"
  42 #include "volumes.h"
  43 #include "qgroup.h"
  44
  45 static struct kmem_cache *btrfs_inode_defrag_cachep;
  46 /*
  47  * when auto defrag is enabled we
  48  * queue up these defrag structs to remember which
  49  * inodes need defragging passes
  50  */
  51 struct inode_defrag {
  52         struct rb_node rb_node;
  53         /* objectid */
  54         u64 ino;
  55         /*
  56          * transid where the defrag was added, we search for
  57          * extents newer than this
  58          */
  59         u64 transid;
  60
  61         /* root objectid */
  62         u64 root;
  63
  64         /* last offset we were able to defrag */
  65         u64 last_offset;
  66
  67         /* if we've wrapped around back to zero once already */
  68         int cycled;
  69 };
  70
  71 static int __compare_inode_defrag(struct inode_defrag *defrag1,
  72                                   struct inode_defrag *defrag2)
  73 {
  74         if (defrag1->root > defrag2->root)
  75                 return 1;
  76         else if (defrag1->root < defrag2->root)
  77                 return -1;
  78         else if (defrag1->ino > defrag2->ino)
  79                 return 1;
  80         else if (defrag1->ino < defrag2->ino)
  81                 return -1;
  82         else
  83                 return 0;
  84 }
  85
  86 /* pop a record for an inode into the defrag tree.  The lock
  87  * must be held already
  88  *
  89  * If you're inserting a record for an older transid than an
  90  * existing record, the transid already in the tree is lowered
  91  *
  92  * If an existing record is found the defrag item you
  93  * pass in is freed
  94  */
  95 static int __btrfs_add_inode_defrag(struct inode *inode,
  96                                     struct inode_defrag *defrag)
  97 {
  98         struct btrfs_root *root = BTRFS_I(inode)->root;
  99         struct inode_defrag *entry;
 100         struct rb_node **p;
 101         struct rb_node *parent = NULL;
 102         int ret;
 103
 104         p = &root->fs_info->defrag_inodes.rb_node;
 105         while (*p) {
 106                 parent = *p;
 107                 entry = rb_entry(parent, struct inode_defrag, rb_node);
 108
 109                 ret = __compare_inode_defrag(defrag, entry);
 110                 if (ret < 0)
 111                         p = &parent->rb_left;
 112                 else if (ret > 0)
 113                         p = &parent->rb_right;
 114                 else {
 115                         /* if we're reinserting an entry for
 116                          * an old defrag run, make sure to
 117                          * lower the transid of our existing record
 118                          */
 119                         if (defrag->transid < entry->transid)
 120                                 entry->transid = defrag->transid;
 121                         if (defrag->last_offset > entry->last_offset)
 122                                 entry->last_offset = defrag->last_offset;
 123                         return -EEXIST;
 124                 }
 125         }
 126         set_bit(BTRFS_INODE_IN_DEFRAG, &BTRFS_I(inode)->runtime_flags);
 127         rb_link_node(&defrag->rb_node, parent, p);
 128         rb_insert_color(&defrag->rb_node, &root->fs_info->defrag_inodes);
 129         return 0;
 130 }
 131
 132 static inline int __need_auto_defrag(struct btrfs_root *root)
 133 {
 134         if (!btrfs_test_opt(root, AUTO_DEFRAG))
 135                 return 0;
 136
 137         if (btrfs_fs_closing(root->fs_info))
 138                 return 0;
 139
 140         return 1;
 141 }
 142
 143 /*
 144  * insert a defrag record for this inode if auto defrag is
 145  * enabled
 146  */
 147 int btrfs_add_inode_defrag(struct btrfs_trans_handle *trans,
 148                            struct inode *inode)
 149 {
 150         struct btrfs_root *root = BTRFS_I(inode)->root;
 151         struct inode_defrag *defrag;
 152         u64 transid;
 153         int ret;
 154
 155         if (!__need_auto_defrag(root))
 156                 return 0;
 157
 158         if (test_bit(BTRFS_INODE_IN_DEFRAG, &BTRFS_I(inode)->runtime_flags))
 159                 return 0;
 160
 161         if (trans)
 162                 transid = trans->transid;
 163         else
 164                 transid = BTRFS_I(inode)->root->last_trans;
 165
 166         defrag = kmem_cache_zalloc(btrfs_inode_defrag_cachep, GFP_NOFS);
 167         if (!defrag)
 168                 return -ENOMEM;
 169
 170         defrag->ino = btrfs_ino(inode);
 171         defrag->transid = transid;
 172         defrag->root = root->root_key.objectid;
 173
 174         spin_lock(&root->fs_info->defrag_inodes_lock);
 175         if (!test_bit(BTRFS_INODE_IN_DEFRAG, &BTRFS_I(inode)->runtime_flags)) {
 176                 /*
 177                  * If we set IN_DEFRAG flag and evict the inode from memory,
 178                  * and then re-read this inode, this new inode doesn't have
 179                  * IN_DEFRAG flag. At the case, we may find the existed defrag.
 180                  */
 181                 ret = __btrfs_add_inode_defrag(inode, defrag);
 182                 if (ret)
 183                         kmem_cache_free(btrfs_inode_defrag_cachep, defrag);
 184         } else {
 185                 kmem_cache_free(btrfs_inode_defrag_cachep, defrag);
 186         }
 187         spin_unlock(&root->fs_info->defrag_inodes_lock);
 188         return 0;
 189 }
 190
 191 /*
 192  * Requeue the defrag object. If there is a defrag object that points to
 193  * the same inode in the tree, we will merge them together (by
 194  * __btrfs_add_inode_defrag()) and free the one that we want to requeue.
 195  */
 196 static void btrfs_requeue_inode_defrag(struct inode *inode,
 197                                        struct inode_defrag *defrag)
 198 {
 199         struct btrfs_root *root = BTRFS_I(inode)->root;
 200         int ret;
 201
 202         if (!__need_auto_defrag(root))
 203                 goto out;
 204
 205         /*
 206          * Here we don't check the IN_DEFRAG flag, because we need merge
 207          * them together.
 208          */
 209         spin_lock(&root->fs_info->defrag_inodes_lock);
 210         ret = __btrfs_add_inode_defrag(inode, defrag);
 211         spin_unlock(&root->fs_info->defrag_inodes_lock);
 212         if (ret)
 213                 goto out;
 214         return;
 215 out:
 216         kmem_cache_free(btrfs_inode_defrag_cachep, defrag);
 217 }
 218
 219 /*
 220  * pick the defragable inode that we want, if it doesn't exist, we will get
 221  * the next one.
 222  */
 223 static struct inode_defrag *
 224 btrfs_pick_defrag_inode(struct btrfs_fs_info *fs_info, u64 root, u64 ino)
 225 {
 226         struct inode_defrag *entry = NULL;
 227         struct inode_defrag tmp;
 228         struct rb_node *p;
 229         struct rb_node *parent = NULL;
 230         int ret;
 231
 232         tmp.ino = ino;
 233         tmp.root = root;
 234
 235         spin_lock(&fs_info->defrag_inodes_lock);
 236         p = fs_info->defrag_inodes.rb_node;
 237         while (p) {
 238                 parent = p;
 239                 entry = rb_entry(parent, struct inode_defrag, rb_node);
 240
 241                 ret = __compare_inode_defrag(&tmp, entry);
 242                 if (ret < 0)
 243                         p = parent->rb_left;
 244                 else if (ret > 0)
 245                         p = parent->rb_right;
 246                 else
 247                         goto out;
 248         }
 249
 250         if (parent && __compare_inode_defrag(&tmp, entry) > 0) {
 251                 parent = rb_next(parent);
 252                 if (parent)
 253                         entry = rb_entry(parent, struct inode_defrag, rb_node);
 254                 else
 255                         entry = NULL;
 256         }
 257 out:
 258         if (entry)
 259                 rb_erase(parent, &fs_info->defrag_inodes);
 260         spin_unlock(&fs_info->defrag_inodes_lock);
 261         return entry;
 262 }
 263
 264 void btrfs_cleanup_defrag_inodes(struct btrfs_fs_info *fs_info)
 265 {
 266         struct inode_defrag *defrag;
 267         struct rb_node *node;
 268
 269         spin_lock(&fs_info->defrag_inodes_lock);
 270         node = rb_first(&fs_info->defrag_inodes);
 271         while (node) {
 272                 rb_erase(node, &fs_info->defrag_inodes);
 273                 defrag = rb_entry(node, struct inode_defrag, rb_node);
 274                 kmem_cache_free(btrfs_inode_defrag_cachep, defrag);
 275
 276                 cond_resched_lock(&fs_info->defrag_inodes_lock);
 277
 278                 node = rb_first(&fs_info->defrag_inodes);
 279         }
 280         spin_unlock(&fs_info->defrag_inodes_lock);
 281 }
 282
 283 #define BTRFS_DEFRAG_BATCH      1024
 284
 285 static int __btrfs_run_defrag_inode(struct btrfs_fs_info *fs_info,
 286                                     struct inode_defrag *defrag)
 287 {
 288         struct btrfs_root *inode_root;
 289         struct inode *inode;
 290         struct btrfs_key key;
 291         struct btrfs_ioctl_defrag_range_args range;
 292         int num_defrag;
 293         int index;
 294         int ret;
 295
 296         /* get the inode */
 297         key.objectid = defrag->root;
 298         key.type = BTRFS_ROOT_ITEM_KEY;
 299         key.offset = (u64)-1;
 300
 301         index = srcu_read_lock(&fs_info->subvol_srcu);
 302
 303         inode_root = btrfs_read_fs_root_no_name(fs_info, &key);
 304         if (IS_ERR(inode_root)) {
 305                 ret = PTR_ERR(inode_root);
 306                 goto cleanup;
 307         }
 308
 309         key.objectid = defrag->ino;
 310         key.type = BTRFS_INODE_ITEM_KEY;
 311         key.offset = 0;
 312         inode = btrfs_iget(fs_info->sb, &key, inode_root, NULL);
 313         if (IS_ERR(inode)) {
 314                 ret = PTR_ERR(inode);
 315                 goto cleanup;
 316         }
 317         srcu_read_unlock(&fs_info->subvol_srcu, index);
 318
 319         /* do a chunk of defrag */
 320         clear_bit(BTRFS_INODE_IN_DEFRAG, &BTRFS_I(inode)->runtime_flags);
 321         memset(&range, 0, sizeof(range));
 322         range.len = (u64)-1;
 323         range.start = defrag->last_offset;
 324
 325         sb_start_write(fs_info->sb);
 326         num_defrag = btrfs_defrag_file(inode, NULL, &range, defrag->transid,
 327                                        BTRFS_DEFRAG_BATCH);
 328         sb_end_write(fs_info->sb);
 329         /*
 330          * if we filled the whole defrag batch, there
 331          * must be more work to do.  Queue this defrag
 332          * again
 333          */
 334         if (num_defrag == BTRFS_DEFRAG_BATCH) {
 335                 defrag->last_offset = range.start;
 336                 btrfs_requeue_inode_defrag(inode, defrag);
 337         } else if (defrag->last_offset && !defrag->cycled) {
 338                 /*
 339                  * we didn't fill our defrag batch, but
 340                  * we didn't start at zero.  Make sure we loop
 341                  * around to the start of the file.
 342                  */
 343                 defrag->last_offset = 0;
 344                 defrag->cycled = 1;
 345                 btrfs_requeue_inode_defrag(inode, defrag);
 346         } else {
 347                 kmem_cache_free(btrfs_inode_defrag_cachep, defrag);
 348         }
 349
 350         iput(inode);
 351         return 0;
 352 cleanup:
 353         srcu_read_unlock(&fs_info->subvol_srcu, index);
 354         kmem_cache_free(btrfs_inode_defrag_cachep, defrag);
 355         return ret;
 356 }
 357
 358 /*
 359  * run through the list of inodes in the FS that need
 360  * defragging
 361  */
 362 int btrfs_run_defrag_inodes(struct btrfs_fs_info *fs_info)
 363 {
 364         struct inode_defrag *defrag;
 365         u64 first_ino = 0;
 366         u64 root_objectid = 0;
 367
 368         atomic_inc(&fs_info->defrag_running);
 369         while (1) {
 370                 /* Pause the auto defragger. */
 371                 if (test_bit(BTRFS_FS_STATE_REMOUNTING,
 372                              &fs_info->fs_state))
 373                         break;
 374
 375                 if (!__need_auto_defrag(fs_info->tree_root))
 376                         break;
 377
 378                 /* find an inode to defrag */
 379                 defrag = btrfs_pick_defrag_inode(fs_info, root_objectid,
 380                                                  first_ino);
 381                 if (!defrag) {
 382                         if (root_objectid || first_ino) {
 383                                 root_objectid = 0;
 384                                 first_ino = 0;
 385                                 continue;
 386                         } else {
 387                                 break;
 388                         }
 389                 }
 390
 391                 first_ino = defrag->ino + 1;
 392                 root_objectid = defrag->root;
 393
 394                 __btrfs_run_defrag_inode(fs_info, defrag);
 395         }
 396         atomic_dec(&fs_info->defrag_running);
 397
 398         /*
 399          * during unmount, we use the transaction_wait queue to
 400          * wait for the defragger to stop
 401          */
 402         wake_up(&fs_info->transaction_wait);
 403         return 0;
 404 }
 405
 406 /* simple helper to fault in pages and copy.  This should go away
 407  * and be replaced with calls into generic code.
 408  */
 409 static noinline int btrfs_copy_from_user(loff_t pos, int num_pages,
 410                                          size_t write_bytes,
 411                                          struct page **prepared_pages,
 412                                          struct iov_iter *i)
 413 {
 414         size_t copied = 0;
 415         size_t total_copied = 0;
 416         int pg = 0;
 417         int offset = pos & (PAGE_CACHE_SIZE - 1);
 418
 419         while (write_bytes > 0) {
 420                 size_t count = min_t(size_t,
 421                                      PAGE_CACHE_SIZE - offset, write_bytes);
 422                 struct page *page = prepared_pages[pg];
 423                 /*
 424                  * Copy data from userspace to the current page
 425                  */
 426                 copied = iov_iter_copy_from_user_atomic(page, i, offset, count);
 427
 428                 /* Flush processor's dcache for this page */
 429                 flush_dcache_page(page);
 430
 431                 /*
 432                  * if we get a partial write, we can end up with
 433                  * partially up to date pages.  These add
 434                  * a lot of complexity, so make sure they don't
 435                  * happen by forcing this copy to be retried.
 436                  *
 437                  * The rest of the btrfs_file_write code will fall
 438                  * back to page at a time copies after we return 0.
 439                  */
 440                 if (!PageUptodate(page) && copied < count)
 441                         copied = 0;
 442
 443                 iov_iter_advance(i, copied);
 444                 write_bytes -= copied;
 445                 total_copied += copied;
 446
 447                 /* Return to btrfs_file_write_iter to fault page */
 448                 if (unlikely(copied == 0))
 449                         break;
 450
 451                 if (copied < PAGE_CACHE_SIZE - offset) {
 452                         offset += copied;
 453                 } else {
 454                         pg++;
 455                         offset = 0;
 456                 }
 457         }
 458         return total_copied;
 459 }
 460
 461 /*
 462  * unlocks pages after btrfs_file_write is done with them
 463  */
 464 static void btrfs_drop_pages(struct page **pages, size_t num_pages)
 465 {
 466         size_t i;
 467         for (i = 0; i < num_pages; i++) {
 468                 /* page checked is some magic around finding pages that
 469                  * have been modified without going through btrfs_set_page_dirty
 470                  * clear it here. There should be no need to mark the pages
 471                  * accessed as prepare_pages should have marked them accessed
 472                  * in prepare_pages via find_or_create_page()
 473                  */
 474                 ClearPageChecked(pages[i]);
 475                 unlock_page(pages[i]);
 476                 page_cache_release(pages[i]);
 477         }
 478 }
 479
 480 /*
 481  * after copy_from_user, pages need to be dirtied and we need to make
 482  * sure holes are created between the current EOF and the start of
 483  * any next extents (if required).
 484  *
 485  * this also makes the decision about creating an inline extent vs
 486  * doing real data extents, marking pages dirty and delalloc as required.
 487  */
 488 int btrfs_dirty_pages(struct btrfs_root *root, struct inode *inode,
 489                              struct page **pages, size_t num_pages,
 490                              loff_t pos, size_t write_bytes,
 491                              struct extent_state **cached)
 492 {
 493         int err = 0;
 494         int i;
 495         u64 num_bytes;
 496         u64 start_pos;
 497         u64 end_of_last_block;
 498         u64 end_pos = pos + write_bytes;
 499         loff_t isize = i_size_read(inode);
 500
 501         start_pos = pos & ~((u64)root->sectorsize - 1);
 502         num_bytes = ALIGN(write_bytes + pos - start_pos, root->sectorsize);
 503
 504         end_of_last_block = start_pos + num_bytes - 1;
 505         err = btrfs_set_extent_delalloc(inode, start_pos, end_of_last_block,
 506                                         cached);
 507         if (err)
 508                 return err;
 509
 510         for (i = 0; i < num_pages; i++) {
 511                 struct page *p = pages[i];
 512                 SetPageUptodate(p);
 513                 ClearPageChecked(p);
 514                 set_page_dirty(p);
 515         }
 516
 517         /*
 518          * we've only changed i_size in ram, and we haven't updated
 519          * the disk i_size.  There is no need to log the inode
 520          * at this time.
 521          */
 522         if (end_pos > isize)
 523                 i_size_write(inode, end_pos);
 524         return 0;
 525 }
 526
 527 /*
 528  * this drops all the extents in the cache that intersect the range
 529  * [start, end].  Existing extents are split as required.
 530  */
 531 void btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end,
 532                              int skip_pinned)
 533 {
 534         struct extent_map *em;
 535         struct extent_map *split = NULL;
 536         struct extent_map *split2 = NULL;
 537         struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
 538         u64 len = end - start + 1;
 539         u64 gen;
 540         int ret;
 541         int testend = 1;
 542         unsigned long flags;
 543         int compressed = 0;
 544         bool modified;
 545
 546         WARN_ON(end < start);
 547         if (end == (u64)-1) {
 548                 len = (u64)-1;
 549                 testend = 0;
 550         }
 551         while (1) {
 552                 int no_splits = 0;
 553
 554                 modified = false;
 555                 if (!split)
 556                         split = alloc_extent_map();
 557                 if (!split2)
 558                         split2 = alloc_extent_map();
 559                 if (!split || !split2)
 560                         no_splits = 1;
 561
 562                 write_lock(&em_tree->lock);
 563                 em = lookup_extent_mapping(em_tree, start, len);
 564                 if (!em) {
 565                         write_unlock(&em_tree->lock);
 566                         break;
 567                 }
 568                 flags = em->flags;
 569                 gen = em->generation;
 570                 if (skip_pinned && test_bit(EXTENT_FLAG_PINNED, &em->flags)) {
 571                         if (testend && em->start + em->len >= start + len) {
 572                                 free_extent_map(em);
 573                                 write_unlock(&em_tree->lock);
 574                                 break;
 575                         }
 576                         start = em->start + em->len;
 577                         if (testend)
 578                                 len = start + len - (em->start + em->len);
 579                         free_extent_map(em);
 580                         write_unlock(&em_tree->lock);
 581                         continue;
 582                 }
 583                 compressed = test_bit(EXTENT_FLAG_COMPRESSED, &em->flags);
 584                 clear_bit(EXTENT_FLAG_PINNED, &em->flags);
 585                 clear_bit(EXTENT_FLAG_LOGGING, &flags);
 586                 modified = !list_empty(&em->list);
 587                 if (no_splits)
 588                         goto next;
 589
 590                 if (em->start < start) {
 591                         split->start = em->start;
 592                         split->len = start - em->start;
 593
 594                         if (em->block_start < EXTENT_MAP_LAST_BYTE) {
 595                                 split->orig_start = em->orig_start;
 596                                 split->block_start = em->block_start;
 597
 598                                 if (compressed)
 599                                         split->block_len = em->block_len;
 600                                 else
 601                                         split->block_len = split->len;
 602                                 split->orig_block_len = max(split->block_len,
 603                                                 em->orig_block_len);
 604                                 split->ram_bytes = em->ram_bytes;
 605                         } else {
 606                                 split->orig_start = split->start;
 607                                 split->block_len = 0;
 608                                 split->block_start = em->block_start;
 609                                 split->orig_block_len = 0;
 610                                 split->ram_bytes = split->len;
 611                         }
 612
 613                         split->generation = gen;
 614                         split->bdev = em->bdev;
 615                         split->flags = flags;
 616                         split->compress_type = em->compress_type;
 617                         replace_extent_mapping(em_tree, em, split, modified);
 618                         free_extent_map(split);
 619                         split = split2;
 620                         split2 = NULL;
 621                 }
 622                 if (testend && em->start + em->len > start + len) {
 623                         u64 diff = start + len - em->start;
 624
 625                         split->start = start + len;
 626                         split->len = em->start + em->len - (start + len);
 627                         split->bdev = em->bdev;
 628                         split->flags = flags;
 629                         split->compress_type = em->compress_type;
 630                         split->generation = gen;
 631
 632                         if (em->block_start < EXTENT_MAP_LAST_BYTE) {
 633                                 split->orig_block_len = max(em->block_len,
 634                                                     em->orig_block_len);
 635
 636                                 split->ram_bytes = em->ram_bytes;
 637                                 if (compressed) {
 638                                         split->block_len = em->block_len;
 639                                         split->block_start = em->block_start;
 640                                         split->orig_start = em->orig_start;
 641                                 } else {
 642                                         split->block_len = split->len;
 643                                         split->block_start = em->block_start
 644                                                 + diff;
 645                                         split->orig_start = em->orig_start;
 646                                 }
 647                         } else {
 648                                 split->ram_bytes = split->len;
 649                                 split->orig_start = split->start;
 650                                 split->block_len = 0;
 651                                 split->block_start = em->block_start;
 652                                 split->orig_block_len = 0;
 653                         }
 654
 655                         if (extent_map_in_tree(em)) {
 656                                 replace_extent_mapping(em_tree, em, split,
 657                                                        modified);
 658                         } else {
 659                                 ret = add_extent_mapping(em_tree, split,
 660                                                          modified);
 661                                 ASSERT(ret == 0); /* Logic error */
 662                         }
 663                         free_extent_map(split);
 664                         split = NULL;
 665                 }
 666 next:
 667                 if (extent_map_in_tree(em))
 668                         remove_extent_mapping(em_tree, em);
 669                 write_unlock(&em_tree->lock);
 670
 671                 /* once for us */
 672                 free_extent_map(em);
 673                 /* once for the tree*/
 674                 free_extent_map(em);
 675         }
 676         if (split)
 677                 free_extent_map(split);
 678         if (split2)
 679                 free_extent_map(split2);
 680 }
 681
 682 /*
 683  * this is very complex, but the basic idea is to drop all extents
 684  * in the range start - end.  hint_block is filled in with a block number
 685  * that would be a good hint to the block allocator for this file.
 686  *
 687  * If an extent intersects the range but is not entirely inside the range
 688  * it is either truncated or split.  Anything entirely inside the range
 689  * is deleted from the tree.
 690  */
 691 int __btrfs_drop_extents(struct btrfs_trans_handle *trans,
 692                          struct btrfs_root *root, struct inode *inode,
 693                          struct btrfs_path *path, u64 start, u64 end,
 694                          u64 *drop_end, int drop_cache,
 695                          int replace_extent,
 696                          u32 extent_item_size,
 697                          int *key_inserted)
 698 {
 699         struct extent_buffer *leaf;
 700         struct btrfs_file_extent_item *fi;
 701         struct btrfs_key key;
 702         struct btrfs_key new_key;
 703         u64 ino = btrfs_ino(inode);
 704         u64 search_start = start;
 705         u64 disk_bytenr = 0;
 706         u64 num_bytes = 0;
 707         u64 extent_offset = 0;
 708         u64 extent_end = 0;
 709         int del_nr = 0;
 710         int del_slot = 0;
 711         int extent_type;
 712         int recow;
 713         int ret;
 714         int modify_tree = -1;
 715         int update_refs;
 716         int found = 0;
 717         int leafs_visited = 0;
 718
 719         if (drop_cache)
 720                 btrfs_drop_extent_cache(inode, start, end - 1, 0);
 721
 722         if (start >= BTRFS_I(inode)->disk_i_size && !replace_extent)
 723                 modify_tree = 0;
 724
 725         update_refs = (test_bit(BTRFS_ROOT_REF_COWS, &root->state) ||
 726                        root == root->fs_info->tree_root);
 727         while (1) {
 728                 recow = 0;
 729                 ret = btrfs_lookup_file_extent(trans, root, path, ino,
 730                                                search_start, modify_tree);
 731                 if (ret < 0)
 732                         break;
 733                 if (ret > 0 && path->slots[0] > 0 && search_start == start) {
 734                         leaf = path->nodes[0];
 735                         btrfs_item_key_to_cpu(leaf, &key, path->slots[0] - 1);
 736                         if (key.objectid == ino &&
 737                             key.type == BTRFS_EXTENT_DATA_KEY)
 738                                 path->slots[0]--;
 739                 }
 740                 ret = 0;
 741                 leafs_visited++;
 742 next_slot:
 743                 leaf = path->nodes[0];
 744                 if (path->slots[0] >= btrfs_header_nritems(leaf)) {
 745                         BUG_ON(del_nr > 0);
 746                         ret = btrfs_next_leaf(root, path);
 747                         if (ret < 0)
 748                                 break;
 749                         if (ret > 0) {
 750                                 ret = 0;
 751                                 break;
 752                         }
 753                         leafs_visited++;
 754                         leaf = path->nodes[0];
 755                         recow = 1;
 756                 }
 757
 758                 btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
 759
 760                 if (key.objectid > ino)
 761                         break;
 762                 if (WARN_ON_ONCE(key.objectid < ino) ||
 763                     key.type < BTRFS_EXTENT_DATA_KEY) {
 764                         ASSERT(del_nr == 0);
 765                         path->slots[0]++;
 766                         goto next_slot;
 767                 }
 768                 if (key.type > BTRFS_EXTENT_DATA_KEY || key.offset >= end)
 769                         break;
 770
 771                 fi = btrfs_item_ptr(leaf, path->slots[0],
 772                                     struct btrfs_file_extent_item);
 773                 extent_type = btrfs_file_extent_type(leaf, fi);
 774
 775                 if (extent_type == BTRFS_FILE_EXTENT_REG ||
 776                     extent_type == BTRFS_FILE_EXTENT_PREALLOC) {
 777                         disk_bytenr = btrfs_file_extent_disk_bytenr(leaf, fi);
 778                         num_bytes = btrfs_file_extent_disk_num_bytes(leaf, fi);
 779                         extent_offset = btrfs_file_extent_offset(leaf, fi);
 780                         extent_end = key.offset +
 781                                 btrfs_file_extent_num_bytes(leaf, fi);
 782                 } else if (extent_type == BTRFS_FILE_EXTENT_INLINE) {
 783                         extent_end = key.offset +
 784                                 btrfs_file_extent_inline_len(leaf,
 785                                                      path->slots[0], fi);
 786                 } else {
 787                         /* can't happen */
 788                         BUG();
 789                 }
 790
 791                 /*
 792                  * Don't skip extent items representing 0 byte lengths. They
 793                  * used to be created (bug) if while punching holes we hit
 794                  * -ENOSPC condition. So if we find one here, just ensure we
 795                  * delete it, otherwise we would insert a new file extent item
 796                  * with the same key (offset) as that 0 bytes length file
 797                  * extent item in the call to setup_items_for_insert() later
 798                  * in this function.
 799                  */
 800                 if (extent_end == key.offset && extent_end >= search_start)
 801                         goto delete_extent_item;
 802
 803                 if (extent_end <= search_start) {
 804                         path->slots[0]++;
 805                         goto next_slot;
 806                 }
 807
 808                 found = 1;
 809                 search_start = max(key.offset, start);
 810                 if (recow || !modify_tree) {
 811                         modify_tree = -1;
 812                         btrfs_release_path(path);
 813                         continue;
 814                 }
 815
 816                 /*
 817                  *     | - range to drop - |
 818                  *  | -------- extent -------- |
 819                  */
 820                 if (start > key.offset && end < extent_end) {
 821                         BUG_ON(del_nr > 0);
 822                         if (extent_type == BTRFS_FILE_EXTENT_INLINE) {
 823                                 ret = -EOPNOTSUPP;
 824                                 break;
 825                         }
 826
 827                         memcpy(&new_key, &key, sizeof(new_key));
 828                         new_key.offset = start;
 829                         ret = btrfs_duplicate_item(trans, root, path,
 830                                                    &new_key);
 831                         if (ret == -EAGAIN) {
 832                                 btrfs_release_path(path);
 833                                 continue;
 834                         }
 835                         if (ret < 0)
 836                                 break;
 837
 838                         leaf = path->nodes[0];
 839                         fi = btrfs_item_ptr(leaf, path->slots[0] - 1,
 840                                             struct btrfs_file_extent_item);
 841                         btrfs_set_file_extent_num_bytes(leaf, fi,
 842                                                         start - key.offset);
 843
 844                         fi = btrfs_item_ptr(leaf, path->slots[0],
 845                                             struct btrfs_file_extent_item);
 846
 847                         extent_offset += start - key.offset;
 848                         btrfs_set_file_extent_offset(leaf, fi, extent_offset);
 849                         btrfs_set_file_extent_num_bytes(leaf, fi,
 850                                                         extent_end - start);
 851                         btrfs_mark_buffer_dirty(leaf);
 852
 853                         if (update_refs && disk_bytenr > 0) {
 854                                 ret = btrfs_inc_extent_ref(trans, root,
 855                                                 disk_bytenr, num_bytes, 0,
 856                                                 root->root_key.objectid,
 857                                                 new_key.objectid,
 858                                                 start - extent_offset);
 859                                 BUG_ON(ret); /* -ENOMEM */
 860                         }
 861                         key.offset = start;
 862                 }
 863                 /*
 864                  *  | ---- range to drop ----- |
 865                  *      | -------- extent -------- |
 866                  */
 867                 if (start <= key.offset && end < extent_end) {
 868                         if (extent_type == BTRFS_FILE_EXTENT_INLINE) {
 869                                 ret = -EOPNOTSUPP;
 870                                 break;
 871                         }
 872
 873                         memcpy(&new_key, &key, sizeof(new_key));
 874                         new_key.offset = end;
 875                         btrfs_set_item_key_safe(root->fs_info, path, &new_key);
 876
 877                         extent_offset += end - key.offset;
 878                         btrfs_set_file_extent_offset(leaf, fi, extent_offset);
 879                         btrfs_set_file_extent_num_bytes(leaf, fi,
 880                                                         extent_end - end);
 881                         btrfs_mark_buffer_dirty(leaf);
 882                         if (update_refs && disk_bytenr > 0)
 883                                 inode_sub_bytes(inode, end - key.offset);
 884                         break;
 885                 }
 886
 887                 search_start = extent_end;
 888                 /*
 889                  *       | ---- range to drop ----- |
 890                  *  | -------- extent -------- |
 891                  */
 892                 if (start > key.offset && end >= extent_end) {
 893                         BUG_ON(del_nr > 0);
 894                         if (extent_type == BTRFS_FILE_EXTENT_INLINE) {
 895                                 ret = -EOPNOTSUPP;
 896                                 break;
 897                         }
 898
 899                         btrfs_set_file_extent_num_bytes(leaf, fi,
 900                                                         start - key.offset);
 901                         btrfs_mark_buffer_dirty(leaf);
 902                         if (update_refs && disk_bytenr > 0)
 903                                 inode_sub_bytes(inode, extent_end - start);
 904                         if (end == extent_end)
 905                                 break;
 906
 907                         path->slots[0]++;
 908                         goto next_slot;
 909                 }
 910
 911                 /*
 912                  *  | ---- range to drop ----- |
 913                  *    | ------ extent ------ |
 914                  */
 915                 if (start <= key.offset && end >= extent_end) {
 916 delete_extent_item:
 917                         if (del_nr == 0) {
 918                                 del_slot = path->slots[0];
 919                                 del_nr = 1;
 920                         } else {
 921                                 BUG_ON(del_slot + del_nr != path->slots[0]);
 922                                 del_nr++;
 923                         }
 924
 925                         if (update_refs &&
 926                             extent_type == BTRFS_FILE_EXTENT_INLINE) {
 927                                 inode_sub_bytes(inode,
 928                                                 extent_end - key.offset);
 929                                 extent_end = ALIGN(extent_end,
 930                                                    root->sectorsize);
 931                         } else if (update_refs && disk_bytenr > 0) {
 932                                 ret = btrfs_free_extent(trans, root,
 933                                                 disk_bytenr, num_bytes, 0,
 934                                                 root->root_key.objectid,
 935                                                 key.objectid, key.offset -
 936                                                 extent_offset);
 937                                 BUG_ON(ret); /* -ENOMEM */
 938                                 inode_sub_bytes(inode,
 939                                                 extent_end - key.offset);
 940                         }
 941
 942                         if (end == extent_end)
 943                                 break;
 944
 945                         if (path->slots[0] + 1 < btrfs_header_nritems(leaf)) {
 946                                 path->slots[0]++;
 947                                 goto next_slot;
 948                         }
 949
 950                         ret = btrfs_del_items(trans, root, path, del_slot,
 951                                               del_nr);
 952                         if (ret) {
 953                                 btrfs_abort_transaction(trans, root, ret);
 954                                 break;
 955                         }
 956
 957                         del_nr = 0;
 958                         del_slot = 0;
 959
 960                         btrfs_release_path(path);
 961                         continue;
 962                 }
 963
 964                 BUG_ON(1);
 965         }
 966
 967         if (!ret && del_nr > 0) {
 968                 /*
 969                  * Set path->slots[0] to first slot, so that after the delete
 970                  * if items are move off from our leaf to its immediate left or
 971                  * right neighbor leafs, we end up with a correct and adjusted
 972                  * path->slots[0] for our insertion (if replace_extent != 0).
 973                  */
 974                 path->slots[0] = del_slot;
 975                 ret = btrfs_del_items(trans, root, path, del_slot, del_nr);
 976                 if (ret)
 977                         btrfs_abort_transaction(trans, root, ret);
 978         }
 979
 980         leaf = path->nodes[0];
 981         /*
 982          * If btrfs_del_items() was called, it might have deleted a leaf, in
 983          * which case it unlocked our path, so check path->locks[0] matches a
 984          * write lock.
 985          */
 986         if (!ret && replace_extent && leafs_visited == 1 &&
 987             (path->locks[0] == BTRFS_WRITE_LOCK_BLOCKING ||
 988              path->locks[0] == BTRFS_WRITE_LOCK) &&
 989             btrfs_leaf_free_space(root, leaf) >=
 990             sizeof(struct btrfs_item) + extent_item_size) {
 991
 992                 key.objectid = ino;
 993                 key.type = BTRFS_EXTENT_DATA_KEY;
 994                 key.offset = start;
 995                 if (!del_nr && path->slots[0] < btrfs_header_nritems(leaf)) {
 996                         struct btrfs_key slot_key;
 997
 998                         btrfs_item_key_to_cpu(leaf, &slot_key, path->slots[0]);
 999                         if (btrfs_comp_cpu_keys(&key, &slot_key) > 0)
1000                                 path->slots[0]++;
1001                 }
1002                 setup_items_for_insert(root, path, &key,
1003                                        &extent_item_size,
1004                                        extent_item_size,
1005                                        sizeof(struct btrfs_item) +
1006                                        extent_item_size, 1);
1007                 *key_inserted = 1;
1008         }
1009
1010         if (!replace_extent || !(*key_inserted))
1011                 btrfs_release_path(path);
1012         if (drop_end)
1013                 *drop_end = found ? min(end, extent_end) : end;
1014         return ret;
1015 }
1016
1017 int btrfs_drop_extents(struct btrfs_trans_handle *trans,
1018                        struct btrfs_root *root, struct inode *inode, u64 start,
1019                        u64 end, int drop_cache)
1020 {
1021         struct btrfs_path *path;
1022         int ret;
1023
1024         path = btrfs_alloc_path();
1025         if (!path)
1026                 return -ENOMEM;
1027         ret = __btrfs_drop_extents(trans, root, inode, path, start, end, NULL,
1028                                    drop_cache, 0, 0, NULL);
1029         btrfs_free_path(path);
1030         return ret;
1031 }
1032
1033 static int extent_mergeable(struct extent_buffer *leaf, int slot,
1034                             u64 objectid, u64 bytenr, u64 orig_offset,
1035                             u64 *start, u64 *end)
1036 {
1037         struct btrfs_file_extent_item *fi;
1038         struct btrfs_key key;
1039         u64 extent_end;
1040
1041         if (slot < 0 || slot >= btrfs_header_nritems(leaf))
1042                 return 0;
1043
1044         btrfs_item_key_to_cpu(leaf, &key, slot);
1045         if (key.objectid != objectid || key.type != BTRFS_EXTENT_DATA_KEY)
1046                 return 0;
1047
1048         fi = btrfs_item_ptr(leaf, slot, struct btrfs_file_extent_item);
1049         if (btrfs_file_extent_type(leaf, fi) != BTRFS_FILE_EXTENT_REG ||
1050             btrfs_file_extent_disk_bytenr(leaf, fi) != bytenr ||
1051             btrfs_file_extent_offset(leaf, fi) != key.offset - orig_offset ||
1052             btrfs_file_extent_compression(leaf, fi) ||
1053             btrfs_file_extent_encryption(leaf, fi) ||
1054             btrfs_file_extent_other_encoding(leaf, fi))
1055                 return 0;
1056
1057         extent_end = key.offset + btrfs_file_extent_num_bytes(leaf, fi);
1058         if ((*start && *start != key.offset) || (*end && *end != extent_end))
1059                 return 0;
1060
1061         *start = key.offset;
1062         *end = extent_end;
1063         return 1;
1064 }
1065
1066 /*
1067  * Mark extent in the range start - end as written.
1068  *
1069  * This changes extent type from 'pre-allocated' to 'regular'. If only
1070  * part of extent is marked as written, the extent will be split into
1071  * two or three.
1072  */
1073 int btrfs_mark_extent_written(struct btrfs_trans_handle *trans,
1074                               struct inode *inode, u64 start, u64 end)
1075 {
1076         struct btrfs_root *root = BTRFS_I(inode)->root;
1077         struct extent_buffer *leaf;
1078         struct btrfs_path *path;
1079         struct btrfs_file_extent_item *fi;
1080         struct btrfs_key key;
1081         struct btrfs_key new_key;
1082         u64 bytenr;
1083         u64 num_bytes;
1084         u64 extent_end;
1085         u64 orig_offset;
1086         u64 other_start;
1087         u64 other_end;
1088         u64 split;
1089         int del_nr = 0;
1090         int del_slot = 0;
1091         int recow;
1092         int ret;
1093         u64 ino = btrfs_ino(inode);
1094
1095         path = btrfs_alloc_path();
1096         if (!path)
1097                 return -ENOMEM;
1098 again:
1099         recow = 0;
1100         split = start;
1101         key.objectid = ino;
1102         key.type = BTRFS_EXTENT_DATA_KEY;
1103         key.offset = split;
1104
1105         ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
1106         if (ret < 0)
1107                 goto out;
1108         if (ret > 0 && path->slots[0] > 0)
1109                 path->slots[0]--;
1110
1111         leaf = path->nodes[0];
1112         btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
1113         BUG_ON(key.objectid != ino || key.type != BTRFS_EXTENT_DATA_KEY);
1114         fi = btrfs_item_ptr(leaf, path->slots[0],
1115                             struct btrfs_file_extent_item);
1116         BUG_ON(btrfs_file_extent_type(leaf, fi) !=
1117                BTRFS_FILE_EXTENT_PREALLOC);
1118         extent_end = key.offset + btrfs_file_extent_num_bytes(leaf, fi);
1119         BUG_ON(key.offset > start || extent_end < end);
1120
1121         bytenr = btrfs_file_extent_disk_bytenr(leaf, fi);
1122         num_bytes = btrfs_file_extent_disk_num_bytes(leaf, fi);
1123         orig_offset = key.offset - btrfs_file_extent_offset(leaf, fi);
1124         memcpy(&new_key, &key, sizeof(new_key));
1125
1126         if (start == key.offset && end < extent_end) {
1127                 other_start = 0;
1128                 other_end = start;
1129                 if (extent_mergeable(leaf, path->slots[0] - 1,
1130                                      ino, bytenr, orig_offset,
1131                                      &other_start, &other_end)) {
1132                         new_key.offset = end;
1133                         btrfs_set_item_key_safe(root->fs_info, path, &new_key);
1134                         fi = btrfs_item_ptr(leaf, path->slots[0],
1135                                             struct btrfs_file_extent_item);
1136                         btrfs_set_file_extent_generation(leaf, fi,
1137                                                          trans->transid);
1138                         btrfs_set_file_extent_num_bytes(leaf, fi,
1139                                                         extent_end - end);
1140                         btrfs_set_file_extent_offset(leaf, fi,
1141                                                      end - orig_offset);
1142                         fi = btrfs_item_ptr(leaf, path->slots[0] - 1,
1143                                             struct btrfs_file_extent_item);
1144                         btrfs_set_file_extent_generation(leaf, fi,
1145                                                          trans->transid);
1146                         btrfs_set_file_extent_num_bytes(leaf, fi,
1147                                                         end - other_start);
1148                         btrfs_mark_buffer_dirty(leaf);
1149                         goto out;
1150                 }
1151         }
1152
1153         if (start > key.offset && end == extent_end) {
1154                 other_start = end;
1155                 other_end = 0;
1156                 if (extent_mergeable(leaf, path->slots[0] + 1,
1157                                      ino, bytenr, orig_offset,
1158                                      &other_start, &other_end)) {
1159                         fi = btrfs_item_ptr(leaf, path->slots[0],
1160                                             struct btrfs_file_extent_item);
1161                         btrfs_set_file_extent_num_bytes(leaf, fi,
1162                                                         start - key.offset);
1163                         btrfs_set_file_extent_generation(leaf, fi,
1164                                                          trans->transid);
1165                         path->slots[0]++;
1166                         new_key.offset = start;
1167                         btrfs_set_item_key_safe(root->fs_info, path, &new_key);
1168
1169                         fi = btrfs_item_ptr(leaf, path->slots[0],
1170                                             struct btrfs_file_extent_item);
1171                         btrfs_set_file_extent_generation(leaf, fi,
1172                                                          trans->transid);
1173                         btrfs_set_file_extent_num_bytes(leaf, fi,
1174                                                         other_end - start);
1175                         btrfs_set_file_extent_offset(leaf, fi,
1176                                                      start - orig_offset);
1177                         btrfs_mark_buffer_dirty(leaf);
1178                         goto out;
1179                 }
1180         }
1181
1182         while (start > key.offset || end < extent_end) {
1183                 if (key.offset == start)
1184                         split = end;
1185
1186                 new_key.offset = split;
1187                 ret = btrfs_duplicate_item(trans, root, path, &new_key);
1188                 if (ret == -EAGAIN) {
1189                         btrfs_release_path(path);
1190                         goto again;
1191                 }
1192                 if (ret < 0) {
1193                         btrfs_abort_transaction(trans, root, ret);
1194                         goto out;
1195                 }
1196
1197                 leaf = path->nodes[0];
1198                 fi = btrfs_item_ptr(leaf, path->slots[0] - 1,
1199                                     struct btrfs_file_extent_item);
1200                 btrfs_set_file_extent_generation(leaf, fi, trans->transid);
1201                 btrfs_set_file_extent_num_bytes(leaf, fi,
1202                                                 split - key.offset);
1203
1204                 fi = btrfs_item_ptr(leaf, path->slots[0],
1205                                     struct btrfs_file_extent_item);
1206
1207                 btrfs_set_file_extent_generation(leaf, fi, trans->transid);
1208                 btrfs_set_file_extent_offset(leaf, fi, split - orig_offset);
1209                 btrfs_set_file_extent_num_bytes(leaf, fi,
1210                                                 extent_end - split);
1211                 btrfs_mark_buffer_dirty(leaf);
1212
1213                 ret = btrfs_inc_extent_ref(trans, root, bytenr, num_bytes, 0,
1214                                            root->root_key.objectid,
1215                                            ino, orig_offset);
1216                 BUG_ON(ret); /* -ENOMEM */
1217
1218                 if (split == start) {
1219                         key.offset = start;
1220                 } else {
1221                         BUG_ON(start != key.offset);
1222                         path->slots[0]--;
1223                         extent_end = end;
1224                 }
1225                 recow = 1;
1226         }
1227
1228         other_start = end;
1229         other_end = 0;
1230         if (extent_mergeable(leaf, path->slots[0] + 1,
1231                              ino, bytenr, orig_offset,
1232                              &other_start, &other_end)) {
1233                 if (recow) {
1234                         btrfs_release_path(path);
1235                         goto again;
1236                 }
1237                 extent_end = other_end;
1238                 del_slot = path->slots[0] + 1;
1239                 del_nr++;
1240                 ret = btrfs_free_extent(trans, root, bytenr, num_bytes,
1241                                         0, root->root_key.objectid,
1242                                         ino, orig_offset);
1243                 BUG_ON(ret); /* -ENOMEM */
1244         }
1245         other_start = 0;
1246         other_end = start;
1247         if (extent_mergeable(leaf, path->slots[0] - 1,
1248                              ino, bytenr, orig_offset,
1249                              &other_start, &other_end)) {
1250                 if (recow) {
1251                         btrfs_release_path(path);
1252                         goto again;
1253                 }
1254                 key.offset = other_start;
1255                 del_slot = path->slots[0];
1256                 del_nr++;
1257                 ret = btrfs_free_extent(trans, root, bytenr, num_bytes,
1258                                         0, root->root_key.objectid,
1259                                         ino, orig_offset);
1260                 BUG_ON(ret); /* -ENOMEM */
1261         }
1262         if (del_nr == 0) {
1263                 fi = btrfs_item_ptr(leaf, path->slots[0],
1264                            struct btrfs_file_extent_item);
1265                 btrfs_set_file_extent_type(leaf, fi,
1266                                            BTRFS_FILE_EXTENT_REG);
1267                 btrfs_set_file_extent_generation(leaf, fi, trans->transid);
1268                 btrfs_mark_buffer_dirty(leaf);
1269         } else {
1270                 fi = btrfs_item_ptr(leaf, del_slot - 1,
1271                            struct btrfs_file_extent_item);
1272                 btrfs_set_file_extent_type(leaf, fi,
1273                                            BTRFS_FILE_EXTENT_REG);
1274                 btrfs_set_file_extent_generation(leaf, fi, trans->transid);
1275                 btrfs_set_file_extent_num_bytes(leaf, fi,
1276                                                 extent_end - key.offset);
1277                 btrfs_mark_buffer_dirty(leaf);
1278
1279                 ret = btrfs_del_items(trans, root, path, del_slot, del_nr);
1280                 if (ret < 0) {
1281                         btrfs_abort_transaction(trans, root, ret);
1282                         goto out;
1283                 }
1284         }
1285 out:
1286         btrfs_free_path(path);
1287         return 0;
1288 }
1289
1290 /*
1291  * on error we return an unlocked page and the error value
1292  * on success we return a locked page and 0
1293  */
1294 static int prepare_uptodate_page(struct inode *inode,
1295                                  struct page *page, u64 pos,
1296                                  bool force_uptodate)
1297 {
1298         int ret = 0;
1299
1300         if (((pos & (PAGE_CACHE_SIZE - 1)) || force_uptodate) &&
1301             !PageUptodate(page)) {
1302                 ret = btrfs_readpage(NULL, page);
1303                 if (ret)
1304                         return ret;
1305                 lock_page(page);
1306                 if (!PageUptodate(page)) {
1307                         unlock_page(page);
1308                         return -EIO;
1309                 }
1310                 if (page->mapping != inode->i_mapping) {
1311                         unlock_page(page);
1312                         return -EAGAIN;
1313                 }
1314         }
1315         return 0;
1316 }
1317
1318 /*
1319  * this just gets pages into the page cache and locks them down.
1320  */
1321 static noinline int prepare_pages(struct inode *inode, struct page **pages,
1322                                   size_t num_pages, loff_t pos,
1323                                   size_t write_bytes, bool force_uptodate)
1324 {
1325         int i;
1326         unsigned long index = pos >> PAGE_CACHE_SHIFT;
1327         gfp_t mask = btrfs_alloc_write_mask(inode->i_mapping);
1328         int err = 0;
1329         int faili;
1330
1331         for (i = 0; i < num_pages; i++) {
1332 again:
1333                 pages[i] = find_or_create_page(inode->i_mapping, index + i,
1334                                                mask | __GFP_WRITE);
1335                 if (!pages[i]) {
1336                         faili = i - 1;
1337                         err = -ENOMEM;
1338                         goto fail;
1339                 }
1340
1341                 if (i == 0)
1342                         err = prepare_uptodate_page(inode, pages[i], pos,
1343                                                     force_uptodate);
1344                 if (!err && i == num_pages - 1)
1345                         err = prepare_uptodate_page(inode, pages[i],
1346                                                     pos + write_bytes, false);
1347                 if (err) {
1348                         page_cache_release(pages[i]);
1349                         if (err == -EAGAIN) {
1350                                 err = 0;
1351                                 goto again;
1352                         }
1353                         faili = i - 1;
1354                         goto fail;
1355                 }
1356                 wait_on_page_writeback(pages[i]);
1357         }
1358
1359         return 0;
1360 fail:
1361         while (faili >= 0) {
1362                 unlock_page(pages[faili]);
1363                 page_cache_release(pages[faili]);
1364                 faili--;
1365         }
1366         return err;
1367
1368 }
1369
1370 /*
1371  * This function locks the extent and properly waits for data=ordered extents
1372  * to finish before allowing the pages to be modified if need.
1373  *
1374  * The return value:
1375  * 1 - the extent is locked
1376  * 0 - the extent is not locked, and everything is OK
1377  * -EAGAIN - need re-prepare the pages
1378  * the other < 0 number - Something wrong happens
1379  */
1380 static noinline int
1381 lock_and_cleanup_extent_if_need(struct inode *inode, struct page **pages,
1382                                 size_t num_pages, loff_t pos,
1383                                 u64 *lockstart, u64 *lockend,
1384                                 struct extent_state **cached_state)
1385 {
1386         u64 start_pos;
1387         u64 last_pos;
1388         int i;
1389         int ret = 0;
1390
1391         start_pos = pos & ~((u64)PAGE_CACHE_SIZE - 1);
1392         last_pos = start_pos + ((u64)num_pages << PAGE_CACHE_SHIFT) - 1;
1393
1394         if (start_pos < inode->i_size) {
1395                 struct btrfs_ordered_extent *ordered;
1396                 lock_extent_bits(&BTRFS_I(inode)->io_tree,
1397                                  start_pos, last_pos, cached_state);
1398                 ordered = btrfs_lookup_ordered_range(inode, start_pos,
1399                                                      last_pos - start_pos + 1);
1400                 if (ordered &&
1401                     ordered->file_offset + ordered->len > start_pos &&
1402                     ordered->file_offset <= last_pos) {
1403                         unlock_extent_cached(&BTRFS_I(inode)->io_tree,
1404                                              start_pos, last_pos,
1405                                              cached_state, GFP_NOFS);
1406                         for (i = 0; i < num_pages; i++) {
1407                                 unlock_page(pages[i]);
1408                                 page_cache_release(pages[i]);
1409                         }
1410                         btrfs_start_ordered_extent(inode, ordered, 1);
1411                         btrfs_put_ordered_extent(ordered);
1412                         return -EAGAIN;
1413                 }
1414                 if (ordered)
1415                         btrfs_put_ordered_extent(ordered);
1416
1417                 clear_extent_bit(&BTRFS_I(inode)->io_tree, start_pos,
1418                                   last_pos, EXTENT_DIRTY | EXTENT_DELALLOC |
1419                                   EXTENT_DO_ACCOUNTING | EXTENT_DEFRAG,
1420                                   0, 0, cached_state, GFP_NOFS);
1421                 *lockstart = start_pos;
1422                 *lockend = last_pos;
1423                 ret = 1;
1424         }
1425
1426         for (i = 0; i < num_pages; i++) {
1427                 if (clear_page_dirty_for_io(pages[i]))
1428                         account_page_redirty(pages[i]);
1429                 set_page_extent_mapped(pages[i]);
1430                 WARN_ON(!PageLocked(pages[i]));
1431         }
1432
1433         return ret;
1434 }
1435
1436 static noinline int check_can_nocow(struct inode *inode, loff_t pos,
1437                                     size_t *write_bytes)
1438 {
1439         struct btrfs_root *root = BTRFS_I(inode)->root;
1440         struct btrfs_ordered_extent *ordered;
1441         u64 lockstart, lockend;
1442         u64 num_bytes;
1443         int ret;
1444
1445         ret = btrfs_start_write_no_snapshoting(root);
1446         if (!ret)
1447                 return -ENOSPC;
1448
1449         lockstart = round_down(pos, root->sectorsize);
1450         lockend = round_up(pos + *write_bytes, root->sectorsize) - 1;
1451
1452         while (1) {
1453                 lock_extent(&BTRFS_I(inode)->io_tree, lockstart, lockend);
1454                 ordered = btrfs_lookup_ordered_range(inode, lockstart,
1455                                                      lockend - lockstart + 1);
1456                 if (!ordered) {
1457                         break;
1458                 }
1459                 unlock_extent(&BTRFS_I(inode)->io_tree, lockstart, lockend);
1460                 btrfs_start_ordered_extent(inode, ordered, 1);
1461                 btrfs_put_ordered_extent(ordered);
1462         }
1463
1464         num_bytes = lockend - lockstart + 1;
1465         ret = can_nocow_extent(inode, lockstart, &num_bytes, NULL, NULL, NULL);
1466         if (ret <= 0) {
1467                 ret = 0;
1468                 btrfs_end_write_no_snapshoting(root);
1469         } else {
1470                 *write_bytes = min_t(size_t, *write_bytes ,
1471                                      num_bytes - pos + lockstart);
1472         }
1473
1474         unlock_extent(&BTRFS_I(inode)->io_tree, lockstart, lockend);
1475
1476         return ret;
1477 }
1478
1479 static noinline ssize_t __btrfs_buffered_write(struct file *file,
1480                                                struct iov_iter *i,
1481                                                loff_t pos)
1482 {
1483         struct inode *inode = file_inode(file);
1484         struct btrfs_root *root = BTRFS_I(inode)->root;
1485         struct page **pages = NULL;
1486         struct extent_state *cached_state = NULL;
1487         u64 release_bytes = 0;
1488         u64 lockstart;
1489         u64 lockend;
1490         size_t num_written = 0;
1491         int nrptrs;
1492         int ret = 0;
1493         bool only_release_metadata = false;
1494         bool force_page_uptodate = false;
1495         bool need_unlock;
1496
1497         nrptrs = min(DIV_ROUND_UP(iov_iter_count(i), PAGE_CACHE_SIZE),
1498                         PAGE_CACHE_SIZE / (sizeof(struct page *)));
1499         nrptrs = min(nrptrs, current->nr_dirtied_pause - current->nr_dirtied);
1500         nrptrs = max(nrptrs, 8);
1501         pages = kmalloc_array(nrptrs, sizeof(struct page *), GFP_KERNEL);
1502         if (!pages)
1503                 return -ENOMEM;
1504
1505         while (iov_iter_count(i) > 0) {
1506                 size_t offset = pos & (PAGE_CACHE_SIZE - 1);
1507                 size_t write_bytes = min(iov_iter_count(i),
1508                                          nrptrs * (size_t)PAGE_CACHE_SIZE -
1509                                          offset);
1510                 size_t num_pages = DIV_ROUND_UP(write_bytes + offset,
1511                                                 PAGE_CACHE_SIZE);
1512                 size_t reserve_bytes;
1513                 size_t dirty_pages;
1514                 size_t copied;
1515
1516                 WARN_ON(num_pages > nrptrs);
1517
1518                 /*
1519                  * Fault pages before locking them in prepare_pages
1520                  * to avoid recursive lock
1521                  */
1522                 if (unlikely(iov_iter_fault_in_readable(i, write_bytes))) {
1523                         ret = -EFAULT;
1524                         break;
1525                 }
1526
1527                 reserve_bytes = num_pages << PAGE_CACHE_SHIFT;
1528
1529                 if (BTRFS_I(inode)->flags & (BTRFS_INODE_NODATACOW |
1530                                              BTRFS_INODE_PREALLOC)) {
1531                         ret = check_can_nocow(inode, pos, &write_bytes);
1532                         if (ret < 0)
1533                                 break;
1534                         if (ret > 0) {
1535                                 /*
1536                                  * For nodata cow case, no need to reserve
1537                                  * data space.
1538                                  */
1539                                 only_release_metadata = true;
1540                                 /*
1541                                  * our prealloc extent may be smaller than
1542                                  * write_bytes, so scale down.
1543                                  */
1544                                 num_pages = DIV_ROUND_UP(write_bytes + offset,
1545                                                          PAGE_CACHE_SIZE);
1546                                 reserve_bytes = num_pages << PAGE_CACHE_SHIFT;
1547                                 goto reserve_metadata;
1548                         }
1549                 }
1550                 ret = btrfs_check_data_free_space(inode, pos, write_bytes);
1551                 if (ret < 0)
1552                         break;
1553
1554 reserve_metadata:
1555                 ret = btrfs_delalloc_reserve_metadata(inode, reserve_bytes);
1556                 if (ret) {
1557                         if (!only_release_metadata)
1558                                 btrfs_free_reserved_data_space(inode, pos,
1559                                                                write_bytes);
1560                         else
1561                                 btrfs_end_write_no_snapshoting(root);
1562                         break;
1563                 }
1564
1565                 release_bytes = reserve_bytes;
1566                 need_unlock = false;
1567 again:
1568                 /*
1569                  * This is going to setup the pages array with the number of
1570                  * pages we want, so we don't really need to worry about the
1571                  * contents of pages from loop to loop
1572                  */
1573                 ret = prepare_pages(inode, pages, num_pages,
1574                                     pos, write_bytes,
1575                                     force_page_uptodate);
1576                 if (ret)
1577                         break;
1578
1579                 ret = lock_and_cleanup_extent_if_need(inode, pages, num_pages,
1580                                                       pos, &lockstart, &lockend,
1581                                                       &cached_state);
1582                 if (ret < 0) {
1583                         if (ret == -EAGAIN)
1584                                 goto again;
1585                         break;
1586                 } else if (ret > 0) {
1587                         need_unlock = true;
1588                         ret = 0;
1589                 }
1590
1591                 copied = btrfs_copy_from_user(pos, num_pages,
1592                                            write_bytes, pages, i);
1593
1594                 /*
1595                  * if we have trouble faulting in the pages, fall
1596                  * back to one page at a time
1597                  */
1598                 if (copied < write_bytes)
1599                         nrptrs = 1;
1600
1601                 if (copied == 0) {
1602                         force_page_uptodate = true;
1603                         dirty_pages = 0;
1604                 } else {
1605                         force_page_uptodate = false;
1606                         dirty_pages = DIV_ROUND_UP(copied + offset,
1607                                                    PAGE_CACHE_SIZE);
1608                 }
1609
1610                 /*
1611                  * If we had a short copy we need to release the excess delaloc
1612                  * bytes we reserved.  We need to increment outstanding_extents
1613                  * because btrfs_delalloc_release_space will decrement it, but
1614                  * we still have an outstanding extent for the chunk we actually
1615                  * managed to copy.
1616                  */
1617                 if (num_pages > dirty_pages) {
1618                         release_bytes = (num_pages - dirty_pages) <<
1619                                 PAGE_CACHE_SHIFT;
1620                         if (copied > 0) {
1621                                 spin_lock(&BTRFS_I(inode)->lock);
1622                                 BTRFS_I(inode)->outstanding_extents++;
1623                                 spin_unlock(&BTRFS_I(inode)->lock);
1624                         }
1625                         if (only_release_metadata) {
1626                                 btrfs_delalloc_release_metadata(inode,
1627                                                                 release_bytes);
1628                         } else {
1629                                 u64 __pos;
1630
1631                                 __pos = round_down(pos, root->sectorsize) +
1632                                         (dirty_pages << PAGE_CACHE_SHIFT);
1633                                 btrfs_delalloc_release_space(inode, __pos,
1634                                                              release_bytes);
1635                         }
1636                 }
1637
1638                 release_bytes = dirty_pages << PAGE_CACHE_SHIFT;
1639
1640                 if (copied > 0)
1641                         ret = btrfs_dirty_pages(root, inode, pages,
1642                                                 dirty_pages, pos, copied,
1643                                                 NULL);
1644                 if (need_unlock)
1645                         unlock_extent_cached(&BTRFS_I(inode)->io_tree,
1646                                              lockstart, lockend, &cached_state,
1647                                              GFP_NOFS);
1648                 if (ret) {
1649                         btrfs_drop_pages(pages, num_pages);
1650                         break;
1651                 }
1652
1653                 release_bytes = 0;
1654                 if (only_release_metadata)
1655                         btrfs_end_write_no_snapshoting(root);
1656
1657                 if (only_release_metadata && copied > 0) {
1658                         lockstart = round_down(pos, root->sectorsize);
1659                         lockend = lockstart +
1660                                 (dirty_pages << PAGE_CACHE_SHIFT) - 1;
1661
1662                         set_extent_bit(&BTRFS_I(inode)->io_tree, lockstart,
1663                                        lockend, EXTENT_NORESERVE, NULL,
1664                                        NULL, GFP_NOFS);
1665                         only_release_metadata = false;
1666                 }
1667
1668                 btrfs_drop_pages(pages, num_pages);
1669
1670                 cond_resched();
1671
1672                 balance_dirty_pages_ratelimited(inode->i_mapping);
1673                 if (dirty_pages < (root->nodesize >> PAGE_CACHE_SHIFT) + 1)
1674                         btrfs_btree_balance_dirty(root);
1675
1676                 pos += copied;
1677                 num_written += copied;
1678         }
1679
1680         kfree(pages);
1681
1682         if (release_bytes) {
1683                 if (only_release_metadata) {
1684                         btrfs_end_write_no_snapshoting(root);
1685                         btrfs_delalloc_release_metadata(inode, release_bytes);
1686                 } else {
1687                         btrfs_delalloc_release_space(inode, pos, release_bytes);
1688                 }
1689         }
1690
1691         return num_written ? num_written : ret;
1692 }
1693
1694 static ssize_t __btrfs_direct_write(struct kiocb *iocb,
1695                                     struct iov_iter *from,
1696                                     loff_t pos)
1697 {
1698         struct file *file = iocb->ki_filp;
1699         struct inode *inode = file_inode(file);
1700         ssize_t written;
1701         ssize_t written_buffered;
1702         loff_t endbyte;
1703         int err;
1704
1705         written = generic_file_direct_write(iocb, from, pos);
1706
1707         if (written < 0 || !iov_iter_count(from))
1708                 return written;
1709
1710         pos += written;
1711         written_buffered = __btrfs_buffered_write(file, from, pos);
1712         if (written_buffered < 0) {
1713                 err = written_buffered;
1714                 goto out;
1715         }
1716         /*
1717          * Ensure all data is persisted. We want the next direct IO read to be
1718          * able to read what was just written.
1719          */
1720         endbyte = pos + written_buffered - 1;
1721         err = btrfs_fdatawrite_range(inode, pos, endbyte);
1722         if (err)
1723                 goto out;
1724         err = filemap_fdatawait_range(inode->i_mapping, pos, endbyte);
1725         if (err)
1726                 goto out;
1727         written += written_buffered;
1728         iocb->ki_pos = pos + written_buffered;
1729         invalidate_mapping_pages(file->f_mapping, pos >> PAGE_CACHE_SHIFT,
1730                                  endbyte >> PAGE_CACHE_SHIFT);
1731 out:
1732         return written ? written : err;
1733 }
1734
1735 static void update_time_for_write(struct inode *inode)
1736 {
1737         struct timespec now;
1738
1739         if (IS_NOCMTIME(inode))
1740                 return;
1741
1742         now = current_fs_time(inode->i_sb);
1743         if (!timespec_equal(&inode->i_mtime, &now))
1744                 inode->i_mtime = now;
1745
1746         if (!timespec_equal(&inode->i_ctime, &now))
1747                 inode->i_ctime = now;
1748
1749         if (IS_I_VERSION(inode))
1750                 inode_inc_iversion(inode);
1751 }
1752
1753 static ssize_t btrfs_file_write_iter(struct kiocb *iocb,
1754                                     struct iov_iter *from)
1755 {
1756         struct file *file = iocb->ki_filp;
1757         struct inode *inode = file_inode(file);
1758         struct btrfs_root *root = BTRFS_I(inode)->root;
1759         u64 start_pos;
1760         u64 end_pos;
1761         ssize_t num_written = 0;
1762         bool sync = (file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host);
1763         ssize_t err;
1764         loff_t pos;
1765         size_t count;
1766
1767         mutex_lock(&inode->i_mutex);
1768         err = generic_write_checks(iocb, from);
1769         if (err <= 0) {
1770                 mutex_unlock(&inode->i_mutex);
1771                 return err;
1772         }
1773
1774         current->backing_dev_info = inode_to_bdi(inode);
1775         err = file_remove_privs(file);
1776         if (err) {
1777                 mutex_unlock(&inode->i_mutex);
1778                 goto out;
1779         }
1780
1781         /*
1782          * If BTRFS flips readonly due to some impossible error
1783          * (fs_info->fs_state now has BTRFS_SUPER_FLAG_ERROR),
1784          * although we have opened a file as writable, we have
1785          * to stop this write operation to ensure FS consistency.
1786          */
1787         if (test_bit(BTRFS_FS_STATE_ERROR, &root->fs_info->fs_state)) {
1788                 mutex_unlock(&inode->i_mutex);
1789                 err = -EROFS;
1790                 goto out;
1791         }
1792
1793         /*
1794          * We reserve space for updating the inode when we reserve space for the
1795          * extent we are going to write, so we will enospc out there.  We don't
1796          * need to start yet another transaction to update the inode as we will
1797          * update the inode when we finish writing whatever data we write.
1798          */
1799         update_time_for_write(inode);
1800
1801         pos = iocb->ki_pos;
1802         count = iov_iter_count(from);
1803         start_pos = round_down(pos, root->sectorsize);
1804         if (start_pos > i_size_read(inode)) {
1805                 /* Expand hole size to cover write data, preventing empty gap */
1806                 end_pos = round_up(pos + count, root->sectorsize);
1807                 err = btrfs_cont_expand(inode, i_size_read(inode), end_pos);
1808                 if (err) {
1809                         mutex_unlock(&inode->i_mutex);
1810                         goto out;
1811                 }
1812         }
1813
1814         if (sync)
1815                 atomic_inc(&BTRFS_I(inode)->sync_writers);
1816
1817         if (iocb->ki_flags & IOCB_DIRECT) {
1818                 num_written = __btrfs_direct_write(iocb, from, pos);
1819         } else {
1820                 num_written = __btrfs_buffered_write(file, from, pos);
1821                 if (num_written > 0)
1822                         iocb->ki_pos = pos + num_written;
1823         }
1824
1825         mutex_unlock(&inode->i_mutex);
1826
1827         /*
1828          * We also have to set last_sub_trans to the current log transid,
1829          * otherwise subsequent syncs to a file that's been synced in this
1830          * transaction will appear to have already occured.
1831          */
1832         spin_lock(&BTRFS_I(inode)->lock);
1833         BTRFS_I(inode)->last_sub_trans = root->log_transid;
1834         spin_unlock(&BTRFS_I(inode)->lock);
1835         if (num_written > 0) {
1836                 err = generic_write_sync(file, pos, num_written);
1837                 if (err < 0)
1838                         num_written = err;
1839         }
1840
1841         if (sync)
1842                 atomic_dec(&BTRFS_I(inode)->sync_writers);
1843 out:
1844         current->backing_dev_info = NULL;
1845         return num_written ? num_written : err;
1846 }
1847
1848 int btrfs_release_file(struct inode *inode, struct file *filp)
1849 {
1850         if (filp->private_data)
1851                 btrfs_ioctl_trans_end(filp);
1852         /*
1853          * ordered_data_close is set by settattr when we are about to truncate
1854          * a file from a non-zero size to a zero size.  This tries to
1855          * flush down new bytes that may have been written if the
1856          * application were using truncate to replace a file in place.
1857          */
1858         if (test_and_clear_bit(BTRFS_INODE_ORDERED_DATA_CLOSE,
1859                                &BTRFS_I(inode)->runtime_flags))
1860                         filemap_flush(inode->i_mapping);
1861         return 0;
1862 }
1863
1864 static int start_ordered_ops(struct inode *inode, loff_t start, loff_t end)
1865 {
1866         int ret;
1867
1868         atomic_inc(&BTRFS_I(inode)->sync_writers);
1869         ret = btrfs_fdatawrite_range(inode, start, end);
1870         atomic_dec(&BTRFS_I(inode)->sync_writers);
1871
1872         return ret;
1873 }
1874
1875 /*
1876  * fsync call for both files and directories.  This logs the inode into
1877  * the tree log instead of forcing full commits whenever possible.
1878  *
1879  * It needs to call filemap_fdatawait so that all ordered extent updates are
1880  * in the metadata btree are up to date for copying to the log.
1881  *
1882  * It drops the inode mutex before doing the tree log commit.  This is an
1883  * important optimization for directories because holding the mutex prevents
1884  * new operations on the dir while we write to disk.
1885  */
1886 int btrfs_sync_file(struct file *file, loff_t start, loff_t end, int datasync)
1887 {
1888         struct dentry *dentry = file->f_path.dentry;
1889         struct inode *inode = d_inode(dentry);
1890         struct btrfs_root *root = BTRFS_I(inode)->root;
1891         struct btrfs_trans_handle *trans;
1892         struct btrfs_log_ctx ctx;
1893         int ret = 0;
1894         bool full_sync = 0;
1895         u64 len;
1896
1897         /*
1898          * The range length can be represented by u64, we have to do the typecasts
1899          * to avoid signed overflow if it's [0, LLONG_MAX] eg. from fsync()
1900          */
1901         len = (u64)end - (u64)start + 1;
1902         trace_btrfs_sync_file(file, datasync);
1903
1904         /*
1905          * We write the dirty pages in the range and wait until they complete
1906          * out of the ->i_mutex. If so, we can flush the dirty pages by
1907          * multi-task, and make the performance up.  See
1908          * btrfs_wait_ordered_range for an explanation of the ASYNC check.
1909          */
1910         ret = start_ordered_ops(inode, start, end);
1911         if (ret)
1912                 return ret;
1913
1914         mutex_lock(&inode->i_mutex);
1915         atomic_inc(&root->log_batch);
1916         full_sync = test_bit(BTRFS_INODE_NEEDS_FULL_SYNC,
1917                              &BTRFS_I(inode)->runtime_flags);
1918         /*
1919          * We might have have had more pages made dirty after calling
1920          * start_ordered_ops and before acquiring the inode's i_mutex.
1921          */
1922         if (full_sync) {
1923                 /*
1924                  * For a full sync, we need to make sure any ordered operations
1925                  * start and finish before we start logging the inode, so that
1926                  * all extents are persisted and the respective file extent
1927                  * items are in the fs/subvol btree.
1928                  */
1929                 ret = btrfs_wait_ordered_range(inode, start, len);
1930         } else {
1931                 /*
1932                  * Start any new ordered operations before starting to log the
1933                  * inode. We will wait for them to finish in btrfs_sync_log().
1934                  *
1935                  * Right before acquiring the inode's mutex, we might have new
1936                  * writes dirtying pages, which won't immediately start the
1937                  * respective ordered operations - that is done through the
1938                  * fill_delalloc callbacks invoked from the writepage and
1939                  * writepages address space operations. So make sure we start
1940                  * all ordered operations before starting to log our inode. Not
1941                  * doing this means that while logging the inode, writeback
1942                  * could start and invoke writepage/writepages, which would call
1943                  * the fill_delalloc callbacks (cow_file_range,
1944                  * submit_compressed_extents). These callbacks add first an
1945                  * extent map to the modified list of extents and then create
1946                  * the respective ordered operation, which means in
1947                  * tree-log.c:btrfs_log_inode() we might capture all existing
1948                  * ordered operations (with btrfs_get_logged_extents()) before
1949                  * the fill_delalloc callback adds its ordered operation, and by
1950                  * the time we visit the modified list of extent maps (with
1951                  * btrfs_log_changed_extents()), we see and process the extent
1952                  * map they created. We then use the extent map to construct a
1953                  * file extent item for logging without waiting for the
1954                  * respective ordered operation to finish - this file extent
1955                  * item points to a disk location that might not have yet been
1956                  * written to, containing random data - so after a crash a log
1957                  * replay will make our inode have file extent items that point
1958                  * to disk locations containing invalid data, as we returned
1959                  * success to userspace without waiting for the respective
1960                  * ordered operation to finish, because it wasn't captured by
1961                  * btrfs_get_logged_extents().
1962                  */
1963                 ret = start_ordered_ops(inode, start, end);
1964         }
1965         if (ret) {
1966                 mutex_unlock(&inode->i_mutex);
1967                 goto out;
1968         }
1969         atomic_inc(&root->log_batch);
1970
1971         /*
1972          * If the last transaction that changed this file was before the current
1973          * transaction and we have the full sync flag set in our inode, we can
1974          * bail out now without any syncing.
1975          *
1976          * Note that we can't bail out if the full sync flag isn't set. This is
1977          * because when the full sync flag is set we start all ordered extents
1978          * and wait for them to fully complete - when they complete they update
1979          * the inode's last_trans field through:
1980          *
1981          *     btrfs_finish_ordered_io() ->
1982          *         btrfs_update_inode_fallback() ->
1983          *             btrfs_update_inode() ->
1984          *                 btrfs_set_inode_last_trans()
1985          *
1986          * So we are sure that last_trans is up to date and can do this check to
1987          * bail out safely. For the fast path, when the full sync flag is not
1988          * set in our inode, we can not do it because we start only our ordered
1989          * extents and don't wait for them to complete (that is when
1990          * btrfs_finish_ordered_io runs), so here at this point their last_trans
1991          * value might be less than or equals to fs_info->last_trans_committed,
1992          * and setting a speculative last_trans for an inode when a buffered
1993          * write is made (such as fs_info->generation + 1 for example) would not
1994          * be reliable since after setting the value and before fsync is called
1995          * any number of transactions can start and commit (transaction kthread
1996          * commits the current transaction periodically), and a transaction
1997          * commit does not start nor waits for ordered extents to complete.
1998          */
1999         smp_mb();
2000         if (btrfs_inode_in_log(inode, root->fs_info->generation) ||
2001             (BTRFS_I(inode)->last_trans <=
2002              root->fs_info->last_trans_committed &&
2003              (full_sync ||
2004               !btrfs_have_ordered_extents_in_range(inode, start, len)))) {
2005                 /*
2006                  * We'v had everything committed since the last time we were
2007                  * modified so clear this flag in case it was set for whatever
2008                  * reason, it's no longer relevant.
2009                  */
2010                 clear_bit(BTRFS_INODE_NEEDS_FULL_SYNC,
2011                           &BTRFS_I(inode)->runtime_flags);
2012                 mutex_unlock(&inode->i_mutex);
2013                 goto out;
2014         }
2015
2016         /*
2017          * ok we haven't committed the transaction yet, lets do a commit
2018          */
2019         if (file->private_data)
2020                 btrfs_ioctl_trans_end(file);
2021
2022         /*
2023          * We use start here because we will need to wait on the IO to complete
2024          * in btrfs_sync_log, which could require joining a transaction (for
2025          * example checking cross references in the nocow path).  If we use join
2026          * here we could get into a situation where we're waiting on IO to
2027          * happen that is blocked on a transaction trying to commit.  With start
2028          * we inc the extwriter counter, so we wait for all extwriters to exit
2029          * before we start blocking join'ers.  This comment is to keep somebody
2030          * from thinking they are super smart and changing this to
2031          * btrfs_join_transaction *cough*Josef*cough*.
2032          */
2033         trans = btrfs_start_transaction(root, 0);
2034         if (IS_ERR(trans)) {
2035                 ret = PTR_ERR(trans);
2036                 mutex_unlock(&inode->i_mutex);
2037                 goto out;
2038         }
2039         trans->sync = true;
2040
2041         btrfs_init_log_ctx(&ctx);
2042
2043         ret = btrfs_log_dentry_safe(trans, root, dentry, start, end, &ctx);
2044         if (ret < 0) {
2045                 /* Fallthrough and commit/free transaction. */
2046                 ret = 1;
2047         }
2048
2049         /* we've logged all the items and now have a consistent
2050          * version of the file in the log.  It is possible that
2051          * someone will come in and modify the file, but that's
2052          * fine because the log is consistent on disk, and we
2053          * have references to all of the file's extents
2054          *
2055          * It is possible that someone will come in and log the
2056          * file again, but that will end up using the synchronization
2057          * inside btrfs_sync_log to keep things safe.
2058          */
2059         mutex_unlock(&inode->i_mutex);
2060
2061         /*
2062          * If any of the ordered extents had an error, just return it to user
2063          * space, so that the application knows some writes didn't succeed and
2064          * can take proper action (retry for e.g.). Blindly committing the
2065          * transaction in this case, would fool userspace that everything was
2066          * successful. And we also want to make sure our log doesn't contain
2067          * file extent items pointing to extents that weren't fully written to -
2068          * just like in the non fast fsync path, where we check for the ordered
2069          * operation's error flag before writing to the log tree and return -EIO
2070          * if any of them had this flag set (btrfs_wait_ordered_range) -
2071          * therefore we need to check for errors in the ordered operations,
2072          * which are indicated by ctx.io_err.
2073          */
2074         if (ctx.io_err) {
2075                 btrfs_end_transaction(trans, root);
2076                 ret = ctx.io_err;
2077                 goto out;
2078         }
2079
2080         if (ret != BTRFS_NO_LOG_SYNC) {
2081                 if (!ret) {
2082                         ret = btrfs_sync_log(trans, root, &ctx);
2083                         if (!ret) {
2084                                 ret = btrfs_end_transaction(trans, root);
2085                                 goto out;
2086                         }
2087                 }
2088                 if (!full_sync) {
2089                         ret = btrfs_wait_ordered_range(inode, start, len);
2090                         if (ret) {
2091                                 btrfs_end_transaction(trans, root);
2092                                 goto out;
2093                         }
2094                 }
2095                 ret = btrfs_commit_transaction(trans, root);
2096         } else {
2097                 ret = btrfs_end_transaction(trans, root);
2098         }
2099 out:
2100         return ret > 0 ? -EIO : ret;
2101 }
2102
2103 static const struct vm_operations_struct btrfs_file_vm_ops = {
2104         .fault          = filemap_fault,
2105         .map_pages      = filemap_map_pages,
2106         .page_mkwrite   = btrfs_page_mkwrite,
2107 };
2108
2109 static int btrfs_file_mmap(struct file  *filp, struct vm_area_struct *vma)
2110 {
2111         struct address_space *mapping = filp->f_mapping;
2112
2113         if (!mapping->a_ops->readpage)
2114                 return -ENOEXEC;
2115
2116         file_accessed(filp);
2117         vma->vm_ops = &btrfs_file_vm_ops;
2118
2119         return 0;
2120 }
2121
2122 static int hole_mergeable(struct inode *inode, struct extent_buffer *leaf,
2123                           int slot, u64 start, u64 end)
2124 {
2125         struct btrfs_file_extent_item *fi;
2126         struct btrfs_key key;
2127
2128         if (slot < 0 || slot >= btrfs_header_nritems(leaf))
2129                 return 0;
2130
2131         btrfs_item_key_to_cpu(leaf, &key, slot);
2132         if (key.objectid != btrfs_ino(inode) ||
2133             key.type != BTRFS_EXTENT_DATA_KEY)
2134                 return 0;
2135
2136         fi = btrfs_item_ptr(leaf, slot, struct btrfs_file_extent_item);
2137
2138         if (btrfs_file_extent_type(leaf, fi) != BTRFS_FILE_EXTENT_REG)
2139                 return 0;
2140
2141         if (btrfs_file_extent_disk_bytenr(leaf, fi))
2142                 return 0;
2143
2144         if (key.offset == end)
2145                 return 1;
2146         if (key.offset + btrfs_file_extent_num_bytes(leaf, fi) == start)
2147                 return 1;
2148         return 0;
2149 }
2150
2151 static int fill_holes(struct btrfs_trans_handle *trans, struct inode *inode,
2152                       struct btrfs_path *path, u64 offset, u64 end)
2153 {
2154         struct btrfs_root *root = BTRFS_I(inode)->root;
2155         struct extent_buffer *leaf;
2156         struct btrfs_file_extent_item *fi;
2157         struct extent_map *hole_em;
2158         struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
2159         struct btrfs_key key;
2160         int ret;
2161
2162         if (btrfs_fs_incompat(root->fs_info, NO_HOLES))
2163                 goto out;
2164
2165         key.objectid = btrfs_ino(inode);
2166         key.type = BTRFS_EXTENT_DATA_KEY;
2167         key.offset = offset;
2168
2169         ret = btrfs_search_slot(trans, root, &key, path, 0, 1);
2170         if (ret < 0)
2171                 return ret;
2172         BUG_ON(!ret);
2173
2174         leaf = path->nodes[0];
2175         if (hole_mergeable(inode, leaf, path->slots[0]-1, offset, end)) {
2176                 u64 num_bytes;
2177
2178                 path->slots[0]--;
2179                 fi = btrfs_item_ptr(leaf, path->slots[0],
2180                                     struct btrfs_file_extent_item);
2181                 num_bytes = btrfs_file_extent_num_bytes(leaf, fi) +
2182                         end - offset;
2183                 btrfs_set_file_extent_num_bytes(leaf, fi, num_bytes);
2184                 btrfs_set_file_extent_ram_bytes(leaf, fi, num_bytes);
2185                 btrfs_set_file_extent_offset(leaf, fi, 0);
2186                 btrfs_mark_buffer_dirty(leaf);
2187                 goto out;
2188         }
2189
2190         if (hole_mergeable(inode, leaf, path->slots[0], offset, end)) {
2191                 u64 num_bytes;
2192
2193                 key.offset = offset;
2194                 btrfs_set_item_key_safe(root->fs_info, path, &key);
2195                 fi = btrfs_item_ptr(leaf, path->slots[0],
2196                                     struct btrfs_file_extent_item);
2197                 num_bytes = btrfs_file_extent_num_bytes(leaf, fi) + end -
2198                         offset;
2199                 btrfs_set_file_extent_num_bytes(leaf, fi, num_bytes);
2200                 btrfs_set_file_extent_ram_bytes(leaf, fi, num_bytes);
2201                 btrfs_set_file_extent_offset(leaf, fi, 0);
2202                 btrfs_mark_buffer_dirty(leaf);
2203                 goto out;
2204         }
2205         btrfs_release_path(path);
2206
2207         ret = btrfs_insert_file_extent(trans, root, btrfs_ino(inode), offset,
2208                                        0, 0, end - offset, 0, end - offset,
2209                                        0, 0, 0);
2210         if (ret)
2211                 return ret;
2212
2213 out:
2214         btrfs_release_path(path);
2215
2216         hole_em = alloc_extent_map();
2217         if (!hole_em) {
2218                 btrfs_drop_extent_cache(inode, offset, end - 1, 0);
2219                 set_bit(BTRFS_INODE_NEEDS_FULL_SYNC,
2220                         &BTRFS_I(inode)->runtime_flags);
2221         } else {
2222                 hole_em->start = offset;
2223                 hole_em->len = end - offset;
2224                 hole_em->ram_bytes = hole_em->len;
2225                 hole_em->orig_start = offset;
2226
2227                 hole_em->block_start = EXTENT_MAP_HOLE;
2228                 hole_em->block_len = 0;
2229                 hole_em->orig_block_len = 0;
2230                 hole_em->bdev = root->fs_info->fs_devices->latest_bdev;
2231                 hole_em->compress_type = BTRFS_COMPRESS_NONE;
2232                 hole_em->generation = trans->transid;
2233
2234                 do {
2235                         btrfs_drop_extent_cache(inode, offset, end - 1, 0);
2236                         write_lock(&em_tree->lock);
2237                         ret = add_extent_mapping(em_tree, hole_em, 1);
2238                         write_unlock(&em_tree->lock);
2239                 } while (ret == -EEXIST);
2240                 free_extent_map(hole_em);
2241                 if (ret)
2242                         set_bit(BTRFS_INODE_NEEDS_FULL_SYNC,
2243                                 &BTRFS_I(inode)->runtime_flags);
2244         }
2245
2246         return 0;
2247 }
2248
2249 /*
2250  * Find a hole extent on given inode and change start/len to the end of hole
2251  * extent.(hole/vacuum extent whose em->start <= start &&
2252  *         em->start + em->len > start)
2253  * When a hole extent is found, return 1 and modify start/len.
2254  */
2255 static int find_first_non_hole(struct inode *inode, u64 *start, u64 *len)
2256 {
2257         struct extent_map *em;
2258         int ret = 0;
2259
2260         em = btrfs_get_extent(inode, NULL, 0, *start, *len, 0);
2261         if (IS_ERR_OR_NULL(em)) {
2262                 if (!em)
2263                         ret = -ENOMEM;
2264                 else
2265                         ret = PTR_ERR(em);
2266                 return ret;
2267         }
2268
2269         /* Hole or vacuum extent(only exists in no-hole mode) */
2270         if (em->block_start == EXTENT_MAP_HOLE) {
2271                 ret = 1;
2272                 *len = em->start + em->len > *start + *len ?
2273                        0 : *start + *len - em->start - em->len;
2274                 *start = em->start + em->len;
2275         }
2276         free_extent_map(em);
2277         return ret;
2278 }
2279
2280 static int btrfs_punch_hole(struct inode *inode, loff_t offset, loff_t len)
2281 {
2282         struct btrfs_root *root = BTRFS_I(inode)->root;
2283         struct extent_state *cached_state = NULL;
2284         struct btrfs_path *path;
2285         struct btrfs_block_rsv *rsv;
2286         struct btrfs_trans_handle *trans;
2287         u64 lockstart;
2288         u64 lockend;
2289         u64 tail_start;
2290         u64 tail_len;
2291         u64 orig_start = offset;
2292         u64 cur_offset;
2293         u64 min_size = btrfs_calc_trunc_metadata_size(root, 1);
2294         u64 drop_end;
2295         int ret = 0;
2296         int err = 0;
2297         unsigned int rsv_count;
2298         bool same_page;
2299         bool no_holes = btrfs_fs_incompat(root->fs_info, NO_HOLES);
2300         u64 ino_size;
2301         bool truncated_page = false;
2302         bool updated_inode = false;
2303
2304         ret = btrfs_wait_ordered_range(inode, offset, len);
2305         if (ret)
2306                 return ret;
2307
2308         mutex_lock(&inode->i_mutex);
2309         ino_size = round_up(inode->i_size, PAGE_CACHE_SIZE);
2310         ret = find_first_non_hole(inode, &offset, &len);
2311         if (ret < 0)
2312                 goto out_only_mutex;
2313         if (ret && !len) {
2314                 /* Already in a large hole */
2315                 ret = 0;
2316                 goto out_only_mutex;
2317         }
2318
2319         lockstart = round_up(offset, BTRFS_I(inode)->root->sectorsize);
2320         lockend = round_down(offset + len,
2321                              BTRFS_I(inode)->root->sectorsize) - 1;
2322         same_page = ((offset >> PAGE_CACHE_SHIFT) ==
2323                     ((offset + len - 1) >> PAGE_CACHE_SHIFT));
2324
2325         /*
2326          * We needn't truncate any page which is beyond the end of the file
2327          * because we are sure there is no data there.
2328          */
2329         /*
2330          * Only do this if we are in the same page and we aren't doing the
2331          * entire page.
2332          */
2333         if (same_page && len < PAGE_CACHE_SIZE) {
2334                 if (offset < ino_size) {
2335                         truncated_page = true;
2336                         ret = btrfs_truncate_page(inode, offset, len, 0);
2337                 } else {
2338                         ret = 0;
2339                 }
2340                 goto out_only_mutex;
2341         }
2342
2343         /* zero back part of the first page */
2344         if (offset < ino_size) {
2345                 truncated_page = true;
2346                 ret = btrfs_truncate_page(inode, offset, 0, 0);
2347                 if (ret) {
2348                         mutex_unlock(&inode->i_mutex);
2349                         return ret;
2350                 }
2351         }
2352
2353         /* Check the aligned pages after the first unaligned page,
2354          * if offset != orig_start, which means the first unaligned page
2355          * including serveral following pages are already in holes,
2356          * the extra check can be skipped */
2357         if (offset == orig_start) {
2358                 /* after truncate page, check hole again */
2359                 len = offset + len - lockstart;
2360                 offset = lockstart;
2361                 ret = find_first_non_hole(inode, &offset, &len);
2362                 if (ret < 0)
2363                         goto out_only_mutex;
2364                 if (ret && !len) {
2365                         ret = 0;
2366                         goto out_only_mutex;
2367                 }
2368                 lockstart = offset;
2369         }
2370
2371         /* Check the tail unaligned part is in a hole */
2372         tail_start = lockend + 1;
2373         tail_len = offset + len - tail_start;
2374         if (tail_len) {
2375                 ret = find_first_non_hole(inode, &tail_start, &tail_len);
2376                 if (unlikely(ret < 0))
2377                         goto out_only_mutex;
2378                 if (!ret) {
2379                         /* zero the front end of the last page */
2380                         if (tail_start + tail_len < ino_size) {
2381                                 truncated_page = true;
2382                                 ret = btrfs_truncate_page(inode,
2383                                                 tail_start + tail_len, 0, 1);
2384                                 if (ret)
2385                                         goto out_only_mutex;
2386                         }
2387                 }
2388         }
2389
2390         if (lockend < lockstart) {
2391                 ret = 0;
2392                 goto out_only_mutex;
2393         }
2394
2395         while (1) {
2396                 struct btrfs_ordered_extent *ordered;
2397
2398                 truncate_pagecache_range(inode, lockstart, lockend);
2399
2400                 lock_extent_bits(&BTRFS_I(inode)->io_tree, lockstart, lockend,
2401                                  &cached_state);
2402                 ordered = btrfs_lookup_first_ordered_extent(inode, lockend);
2403
2404                 /*
2405                  * We need to make sure we have no ordered extents in this range
2406                  * and nobody raced in and read a page in this range, if we did
2407                  * we need to try again.
2408                  */
2409                 if ((!ordered ||
2410                     (ordered->file_offset + ordered->len <= lockstart ||
2411                      ordered->file_offset > lockend)) &&
2412                      !btrfs_page_exists_in_range(inode, lockstart, lockend)) {
2413                         if (ordered)
2414                                 btrfs_put_ordered_extent(ordered);
2415                         break;
2416                 }
2417                 if (ordered)
2418                         btrfs_put_ordered_extent(ordered);
2419                 unlock_extent_cached(&BTRFS_I(inode)->io_tree, lockstart,
2420                                      lockend, &cached_state, GFP_NOFS);
2421                 ret = btrfs_wait_ordered_range(inode, lockstart,
2422                                                lockend - lockstart + 1);
2423                 if (ret) {
2424                         mutex_unlock(&inode->i_mutex);
2425                         return ret;
2426                 }
2427         }
2428
2429         path = btrfs_alloc_path();
2430         if (!path) {
2431                 ret = -ENOMEM;
2432                 goto out;
2433         }
2434
2435         rsv = btrfs_alloc_block_rsv(root, BTRFS_BLOCK_RSV_TEMP);
2436         if (!rsv) {
2437                 ret = -ENOMEM;
2438                 goto out_free;
2439         }
2440         rsv->size = btrfs_calc_trunc_metadata_size(root, 1);
2441         rsv->failfast = 1;
2442
2443         /*
2444          * 1 - update the inode
2445          * 1 - removing the extents in the range
2446          * 1 - adding the hole extent if no_holes isn't set
2447          */
2448         rsv_count = no_holes ? 2 : 3;
2449         trans = btrfs_start_transaction(root, rsv_count);
2450         if (IS_ERR(trans)) {
2451                 err = PTR_ERR(trans);
2452                 goto out_free;
2453         }
2454
2455         ret = btrfs_block_rsv_migrate(&root->fs_info->trans_block_rsv, rsv,
2456                                       min_size);
2457         BUG_ON(ret);
2458         trans->block_rsv = rsv;
2459
2460         cur_offset = lockstart;
2461         len = lockend - cur_offset;
2462         while (cur_offset < lockend) {
2463                 ret = __btrfs_drop_extents(trans, root, inode, path,
2464                                            cur_offset, lockend + 1,
2465                                            &drop_end, 1, 0, 0, NULL);
2466                 if (ret != -ENOSPC)
2467                         break;
2468
2469                 trans->block_rsv = &root->fs_info->trans_block_rsv;
2470
2471                 if (cur_offset < ino_size) {
2472                         ret = fill_holes(trans, inode, path, cur_offset,
2473                                          drop_end);
2474                         if (ret) {
2475                                 err = ret;
2476                                 break;
2477                         }
2478                 }
2479
2480                 cur_offset = drop_end;
2481
2482                 ret = btrfs_update_inode(trans, root, inode);
2483                 if (ret) {
2484                         err = ret;
2485                         break;
2486                 }
2487
2488                 btrfs_end_transaction(trans, root);
2489                 btrfs_btree_balance_dirty(root);
2490
2491                 trans = btrfs_start_transaction(root, rsv_count);
2492                 if (IS_ERR(trans)) {
2493                         ret = PTR_ERR(trans);
2494                         trans = NULL;
2495                         break;
2496                 }
2497
2498                 ret = btrfs_block_rsv_migrate(&root->fs_info->trans_block_rsv,
2499                                               rsv, min_size);
2500                 BUG_ON(ret);    /* shouldn't happen */
2501                 trans->block_rsv = rsv;
2502
2503                 ret = find_first_non_hole(inode, &cur_offset, &len);
2504                 if (unlikely(ret < 0))
2505                         break;
2506                 if (ret && !len) {
2507                         ret = 0;
2508                         break;
2509                 }
2510         }
2511
2512         if (ret) {
2513                 err = ret;
2514                 goto out_trans;
2515         }
2516
2517         trans->block_rsv = &root->fs_info->trans_block_rsv;
2518         /*
2519          * If we are using the NO_HOLES feature we might have had already an
2520          * hole that overlaps a part of the region [lockstart, lockend] and
2521          * ends at (or beyond) lockend. Since we have no file extent items to
2522          * represent holes, drop_end can be less than lockend and so we must
2523          * make sure we have an extent map representing the existing hole (the
2524          * call to __btrfs_drop_extents() might have dropped the existing extent
2525          * map representing the existing hole), otherwise the fast fsync path
2526          * will not record the existence of the hole region
2527          * [existing_hole_start, lockend].
2528          */
2529         if (drop_end <= lockend)
2530                 drop_end = lockend + 1;
2531         /*
2532          * Don't insert file hole extent item if it's for a range beyond eof
2533          * (because it's useless) or if it represents a 0 bytes range (when
2534          * cur_offset == drop_end).
2535          */
2536         if (cur_offset < ino_size && cur_offset < drop_end) {
2537                 ret = fill_holes(trans, inode, path, cur_offset, drop_end);
2538                 if (ret) {
2539                         err = ret;
2540                         goto out_trans;
2541                 }
2542         }
2543
2544 out_trans:
2545         if (!trans)
2546                 goto out_free;
2547
2548         inode_inc_iversion(inode);
2549         inode->i_mtime = inode->i_ctime = CURRENT_TIME;
2550
2551         trans->block_rsv = &root->fs_info->trans_block_rsv;
2552         ret = btrfs_update_inode(trans, root, inode);
2553         updated_inode = true;
2554         btrfs_end_transaction(trans, root);
2555         btrfs_btree_balance_dirty(root);
2556 out_free:
2557         btrfs_free_path(path);
2558         btrfs_free_block_rsv(root, rsv);
2559 out:
2560         unlock_extent_cached(&BTRFS_I(inode)->io_tree, lockstart, lockend,
2561                              &cached_state, GFP_NOFS);
2562 out_only_mutex:
2563         if (!updated_inode && truncated_page && !ret && !err) {
2564                 /*
2565                  * If we only end up zeroing part of a page, we still need to
2566                  * update the inode item, so that all the time fields are
2567                  * updated as well as the necessary btrfs inode in memory fields
2568                  * for detecting, at fsync time, if the inode isn't yet in the
2569                  * log tree or it's there but not up to date.
2570                  */
2571                 trans = btrfs_start_transaction(root, 1);
2572                 if (IS_ERR(trans)) {
2573                         err = PTR_ERR(trans);
2574                 } else {
2575                         err = btrfs_update_inode(trans, root, inode);
2576                         ret = btrfs_end_transaction(trans, root);
2577                 }
2578         }
2579         mutex_unlock(&inode->i_mutex);
2580         if (ret && !err)
2581                 err = ret;
2582         return err;
2583 }
2584
2585 /* Helper structure to record which range is already reserved */
2586 struct falloc_range {
2587         struct list_head list;
2588         u64 start;
2589         u64 len;
2590 };
2591
2592 /*
2593  * Helper function to add falloc range
2594  *
2595  * Caller should have locked the larger range of extent containing
2596  * [start, len)
2597  */
2598 static int add_falloc_range(struct list_head *head, u64 start, u64 len)
2599 {
2600         struct falloc_range *prev = NULL;
2601         struct falloc_range *range = NULL;
2602
2603         if (list_empty(head))
2604                 goto insert;
2605
2606         /*
2607          * As fallocate iterate by bytenr order, we only need to check
2608          * the last range.
2609          */
2610         prev = list_entry(head->prev, struct falloc_range, list);
2611         if (prev->start + prev->len == start) {
2612                 prev->len += len;
2613                 return 0;
2614         }
2615 insert:
2616         range = kmalloc(sizeof(*range), GFP_NOFS);
2617         if (!range)
2618                 return -ENOMEM;
2619         range->start = start;
2620         range->len = len;
2621         list_add_tail(&range->list, head);
2622         return 0;
2623 }
2624
2625 static long btrfs_fallocate(struct file *file, int mode,
2626                             loff_t offset, loff_t len)
2627 {
2628         struct inode *inode = file_inode(file);
2629         struct extent_state *cached_state = NULL;
2630         struct falloc_range *range;
2631         struct falloc_range *tmp;
2632         struct list_head reserve_list;
2633         u64 cur_offset;
2634         u64 last_byte;
2635         u64 alloc_start;
2636         u64 alloc_end;
2637         u64 alloc_hint = 0;
2638         u64 locked_end;
2639         u64 actual_end = 0;
2640         struct extent_map *em;
2641         int blocksize = BTRFS_I(inode)->root->sectorsize;
2642         int ret;
2643
2644         alloc_start = round_down(offset, blocksize);
2645         alloc_end = round_up(offset + len, blocksize);
2646
2647         /* Make sure we aren't being give some crap mode */
2648         if (mode & ~(FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE))
2649                 return -EOPNOTSUPP;
2650
2651         if (mode & FALLOC_FL_PUNCH_HOLE)
2652                 return btrfs_punch_hole(inode, offset, len);
2653
2654         /*
2655          * Only trigger disk allocation, don't trigger qgroup reserve
2656          *
2657          * For qgroup space, it will be checked later.
2658          */
2659         ret = btrfs_alloc_data_chunk_ondemand(inode, alloc_end - alloc_start);
2660         if (ret < 0)
2661                 return ret;
2662
2663         mutex_lock(&inode->i_mutex);
2664         ret = inode_newsize_ok(inode, alloc_end);
2665         if (ret)
2666                 goto out;
2667
2668         /*
2669          * TODO: Move these two operations after we have checked
2670          * accurate reserved space, or fallocate can still fail but
2671          * with page truncated or size expanded.
2672          *
2673          * But that's a minor problem and won't do much harm BTW.
2674          */
2675         if (alloc_start > inode->i_size) {
2676                 ret = btrfs_cont_expand(inode, i_size_read(inode),
2677                                         alloc_start);
2678                 if (ret)
2679                         goto out;
2680         } else if (offset + len > inode->i_size) {
2681                 /*
2682                  * If we are fallocating from the end of the file onward we
2683                  * need to zero out the end of the page if i_size lands in the
2684                  * middle of a page.
2685                  */
2686                 ret = btrfs_truncate_page(inode, inode->i_size, 0, 0);
2687                 if (ret)
2688                         goto out;
2689         }
2690
2691         /*
2692          * wait for ordered IO before we have any locks.  We'll loop again
2693          * below with the locks held.
2694          */
2695         ret = btrfs_wait_ordered_range(inode, alloc_start,
2696                                        alloc_end - alloc_start);
2697         if (ret)
2698                 goto out;
2699
2700         locked_end = alloc_end - 1;
2701         while (1) {
2702                 struct btrfs_ordered_extent *ordered;
2703
2704                 /* the extent lock is ordered inside the running
2705                  * transaction
2706                  */
2707                 lock_extent_bits(&BTRFS_I(inode)->io_tree, alloc_start,
2708                                  locked_end, &cached_state);
2709                 ordered = btrfs_lookup_first_ordered_extent(inode,
2710                                                             alloc_end - 1);
2711                 if (ordered &&
2712                     ordered->file_offset + ordered->len > alloc_start &&
2713                     ordered->file_offset < alloc_end) {
2714                         btrfs_put_ordered_extent(ordered);
2715                         unlock_extent_cached(&BTRFS_I(inode)->io_tree,
2716                                              alloc_start, locked_end,
2717                                              &cached_state, GFP_NOFS);
2718                         /*
2719                          * we can't wait on the range with the transaction
2720                          * running or with the extent lock held
2721                          */
2722                         ret = btrfs_wait_ordered_range(inode, alloc_start,
2723                                                        alloc_end - alloc_start);
2724                         if (ret)
2725                                 goto out;
2726                 } else {
2727                         if (ordered)
2728                                 btrfs_put_ordered_extent(ordered);
2729                         break;
2730                 }
2731         }
2732
2733         /* First, check if we exceed the qgroup limit */
2734         INIT_LIST_HEAD(&reserve_list);
2735         cur_offset = alloc_start;
2736         while (1) {
2737                 em = btrfs_get_extent(inode, NULL, 0, cur_offset,
2738                                       alloc_end - cur_offset, 0);
2739                 if (IS_ERR_OR_NULL(em)) {
2740                         if (!em)
2741                                 ret = -ENOMEM;
2742                         else
2743                                 ret = PTR_ERR(em);
2744                         break;
2745                 }
2746                 last_byte = min(extent_map_end(em), alloc_end);
2747                 actual_end = min_t(u64, extent_map_end(em), offset + len);
2748                 last_byte = ALIGN(last_byte, blocksize);
2749                 if (em->block_start == EXTENT_MAP_HOLE ||
2750                     (cur_offset >= inode->i_size &&
2751                      !test_bit(EXTENT_FLAG_PREALLOC, &em->flags))) {
2752                         ret = add_falloc_range(&reserve_list, cur_offset,
2753                                                last_byte - cur_offset);
2754                         if (ret < 0) {
2755                                 free_extent_map(em);
2756                                 break;
2757                         }
2758                         ret = btrfs_qgroup_reserve_data(inode, cur_offset,
2759                                         last_byte - cur_offset);
2760                         if (ret < 0)
2761                                 break;
2762                 }
2763                 free_extent_map(em);
2764                 cur_offset = last_byte;
2765                 if (cur_offset >= alloc_end)
2766                         break;
2767         }
2768
2769         /*
2770          * If ret is still 0, means we're OK to fallocate.
2771          * Or just cleanup the list and exit.
2772          */
2773         list_for_each_entry_safe(range, tmp, &reserve_list, list) {
2774                 if (!ret)
2775                         ret = btrfs_prealloc_file_range(inode, mode,
2776                                         range->start,
2777                                         range->len, 1 << inode->i_blkbits,
2778                                         offset + len, &alloc_hint);
2779                 list_del(&range->list);
2780                 kfree(range);
2781         }
2782         if (ret < 0)
2783                 goto out_unlock;
2784
2785         if (actual_end > inode->i_size &&
2786             !(mode & FALLOC_FL_KEEP_SIZE)) {
2787                 struct btrfs_trans_handle *trans;
2788                 struct btrfs_root *root = BTRFS_I(inode)->root;
2789
2790                 /*
2791                  * We didn't need to allocate any more space, but we
2792                  * still extended the size of the file so we need to
2793                  * update i_size and the inode item.
2794                  */
2795                 trans = btrfs_start_transaction(root, 1);
2796                 if (IS_ERR(trans)) {
2797                         ret = PTR_ERR(trans);
2798                 } else {
2799                         inode->i_ctime = CURRENT_TIME;
2800                         i_size_write(inode, actual_end);
2801                         btrfs_ordered_update_i_size(inode, actual_end, NULL);
2802                         ret = btrfs_update_inode(trans, root, inode);
2803                         if (ret)
2804                                 btrfs_end_transaction(trans, root);
2805                         else
2806                                 ret = btrfs_end_transaction(trans, root);
2807                 }
2808         }
2809 out_unlock:
2810         unlock_extent_cached(&BTRFS_I(inode)->io_tree, alloc_start, locked_end,
2811                              &cached_state, GFP_NOFS);
2812 out:
2813         /*
2814          * As we waited the extent range, the data_rsv_map must be empty
2815          * in the range, as written data range will be released from it.
2816          * And for prealloacted extent, it will also be released when
2817          * its metadata is written.
2818          * So this is completely used as cleanup.
2819          */
2820         btrfs_qgroup_free_data(inode, alloc_start, alloc_end - alloc_start);
2821         mutex_unlock(&inode->i_mutex);
2822         /* Let go of our reservation. */
2823         btrfs_free_reserved_data_space(inode, alloc_start,
2824                                        alloc_end - alloc_start);
2825         return ret;
2826 }
2827
2828 static int find_desired_extent(struct inode *inode, loff_t *offset, int whence)
2829 {
2830         struct btrfs_root *root = BTRFS_I(inode)->root;
2831         struct extent_map *em = NULL;
2832         struct extent_state *cached_state = NULL;
2833         u64 lockstart;
2834         u64 lockend;
2835         u64 start;
2836         u64 len;
2837         int ret = 0;
2838
2839         if (inode->i_size == 0)
2840                 return -ENXIO;
2841
2842         /*
2843          * *offset can be negative, in this case we start finding DATA/HOLE from
2844          * the very start of the file.
2845          */
2846         start = max_t(loff_t, 0, *offset);
2847
2848         lockstart = round_down(start, root->sectorsize);
2849         lockend = round_up(i_size_read(inode), root->sectorsize);
2850         if (lockend <= lockstart)
2851                 lockend = lockstart + root->sectorsize;
2852         lockend--;
2853         len = lockend - lockstart + 1;
2854
2855         lock_extent_bits(&BTRFS_I(inode)->io_tree, lockstart, lockend,
2856                          &cached_state);
2857
2858         while (start < inode->i_size) {
2859                 em = btrfs_get_extent_fiemap(inode, NULL, 0, start, len, 0);
2860                 if (IS_ERR(em)) {
2861                         ret = PTR_ERR(em);
2862                         em = NULL;
2863                         break;
2864                 }
2865
2866                 if (whence == SEEK_HOLE &&
2867                     (em->block_start == EXTENT_MAP_HOLE ||
2868                      test_bit(EXTENT_FLAG_PREALLOC, &em->flags)))
2869                         break;
2870                 else if (whence == SEEK_DATA &&
2871                            (em->block_start != EXTENT_MAP_HOLE &&
2872                             !test_bit(EXTENT_FLAG_PREALLOC, &em->flags)))
2873                         break;
2874
2875                 start = em->start + em->len;
2876                 free_extent_map(em);
2877                 em = NULL;
2878                 cond_resched();
2879         }
2880         free_extent_map(em);
2881         if (!ret) {
2882                 if (whence == SEEK_DATA && start >= inode->i_size)
2883                         ret = -ENXIO;
2884                 else
2885                         *offset = min_t(loff_t, start, inode->i_size);
2886         }
2887         unlock_extent_cached(&BTRFS_I(inode)->io_tree, lockstart, lockend,
2888                              &cached_state, GFP_NOFS);
2889         return ret;
2890 }
2891
2892 static loff_t btrfs_file_llseek(struct file *file, loff_t offset, int whence)
2893 {
2894         struct inode *inode = file->f_mapping->host;
2895         int ret;
2896
2897         mutex_lock(&inode->i_mutex);
2898         switch (whence) {
2899         case SEEK_END:
2900         case SEEK_CUR:
2901                 offset = generic_file_llseek(file, offset, whence);
2902                 goto out;
2903         case SEEK_DATA:
2904         case SEEK_HOLE:
2905                 if (offset >= i_size_read(inode)) {
2906                         mutex_unlock(&inode->i_mutex);
2907                         return -ENXIO;
2908                 }
2909
2910                 ret = find_desired_extent(inode, &offset, whence);
2911                 if (ret) {
2912                         mutex_unlock(&inode->i_mutex);
2913                         return ret;
2914                 }
2915         }
2916
2917         offset = vfs_setpos(file, offset, inode->i_sb->s_maxbytes);
2918 out:
2919         mutex_unlock(&inode->i_mutex);
2920         return offset;
2921 }
2922
2923 const struct file_operations btrfs_file_operations = {
2924         .llseek         = btrfs_file_llseek,
2925         .read_iter      = generic_file_read_iter,
2926         .splice_read    = generic_file_splice_read,
2927         .write_iter     = btrfs_file_write_iter,
2928         .mmap           = btrfs_file_mmap,
2929         .open           = generic_file_open,
2930         .release        = btrfs_release_file,
2931         .fsync          = btrfs_sync_file,
2932         .fallocate      = btrfs_fallocate,
2933         .unlocked_ioctl = btrfs_ioctl,
2934 #ifdef CONFIG_COMPAT
2935         .compat_ioctl   = btrfs_ioctl,
2936 #endif
2937         .copy_file_range = btrfs_copy_file_range,
2938         .clone_file_range = btrfs_clone_file_range,
2939         .dedupe_file_range = btrfs_dedupe_file_range,
2940 };
2941
2942 void btrfs_auto_defrag_exit(void)
2943 {
2944         if (btrfs_inode_defrag_cachep)
2945                 kmem_cache_destroy(btrfs_inode_defrag_cachep);
2946 }
2947
2948 int btrfs_auto_defrag_init(void)
2949 {
2950         btrfs_inode_defrag_cachep = kmem_cache_create("btrfs_inode_defrag",
2951                                         sizeof(struct inode_defrag), 0,
2952                                         SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD,
2953                                         NULL);
2954         if (!btrfs_inode_defrag_cachep)
2955                 return -ENOMEM;
2956
2957         return 0;
2958 }
2959
2960 int btrfs_fdatawrite_range(struct inode *inode, loff_t start, loff_t end)
2961 {
2962         int ret;
2963
2964         /*
2965          * So with compression we will find and lock a dirty page and clear the
2966          * first one as dirty, setup an async extent, and immediately return
2967          * with the entire range locked but with nobody actually marked with
2968          * writeback.  So we can't just filemap_write_and_wait_range() and
2969          * expect it to work since it will just kick off a thread to do the
2970          * actual work.  So we need to call filemap_fdatawrite_range _again_
2971          * since it will wait on the page lock, which won't be unlocked until
2972          * after the pages have been marked as writeback and so we're good to go
2973          * from there.  We have to do this otherwise we'll miss the ordered
2974          * extents and that results in badness.  Please Josef, do not think you
2975          * know better and pull this out at some point in the future, it is
2976          * right and you are wrong.
2977          */
2978         ret = filemap_fdatawrite_range(inode->i_mapping, start, end);
2979         if (!ret && test_bit(BTRFS_INODE_HAS_ASYNC_EXTENT,
2980                              &BTRFS_I(inode)->runtime_flags))
2981                 ret = filemap_fdatawrite_range(inode->i_mapping, start, end);
2982
2983         return ret;
2984 }